Dielectrics, polarization P, and the electric displacement D

Intuition Beginner

A dielectric is an insulator — a material whose electrons stay bound to their parent atoms or molecules and do not flow freely. Glass, paper, plastic, water, ceramic, and even air are dielectrics. Apply an electric field to one of these materials and the charges cannot drift across the sample, but the field still tugs on them: positive nuclei lean a little in the direction of the field, and the electron clouds lean a little the opposite way. Each atom or molecule becomes a tiny electric dipole, and the whole sample becomes polarized.

The picture is simple. Inside the dielectric, all the small induced dipoles point along the external field, like a forest of tiny arrows. Inside the bulk, every dipole's tail of negative charge cancels the next dipole's head of positive charge, so there is no net charge in the interior. But the two outer faces of the material now carry uncompensated bound charge — positive on the face where the field points outward, negative on the face where it points inward. Those bound surface charges produce an internal field that pushes back against the applied field, reducing the total field inside the sample.

So a dielectric does not block the field, but it weakens it. The fraction by which the field is reduced is captured by a single number for the material called the dielectric constant $\kappa$. Vacuum has $\kappa =1$ by definition. Air has $\kappa \approx 1.0006$, so close to vacuum that you can ignore the difference for most purposes. Paper sits near $\kappa \approx 4$, water is $\kappa \approx 80$ (a remarkably large value because of its rotating polar molecules), and engineered ceramics such as barium titanate reach $\kappa \approx 1200$.

The practical pay-off: slide a dielectric into the gap of a capacitor and the capacitance jumps from $C$ to $\kappa C$. The same voltage now stores $\kappa$ times more charge, because the bound surface charges on the dielectric partially cancel the free charges on the plates and let the plates accept more before the voltage saturates. This is why every commercial capacitor — from the picofarad ceramic discs in your radio to the millifarad electrolytic cans in a power supply — uses a dielectric and never empty vacuum.

Visual Beginner

Picture two metal plates with a slab of glass between them. Before you charge the plates, the glass molecules are randomly oriented and the slab is electrically neutral. Now charge the upper plate positive and the lower plate negative. Inside the glass, every molecule polarizes: a sliver of negative charge appears at its top side and a sliver of positive at its bottom side. In the bulk, neighbouring slivers cancel. But on the top face of the glass, a thin layer of negative bound charge survives, and on the bottom face, a thin layer of positive bound charge survives.

The picture captures three rules every dielectric obeys at this introductory level. First, polarization is a response, not an action: the dipoles only appear once you apply a field. Second, the bound surface charges live only on faces of the dielectric, not in its interior. Third, the bound charges always sit opposite the free charges on the nearby conductor, partially screening them and weakening the field within the dielectric by the factor $1/\kappa$.

Worked example Beginner

A parallel-plate capacitor has square plates of side $1\mathrm{cm}$, separated by a gap of $d=1\mathrm{mm}$. With vacuum between the plates, its capacitance was $C_0=0.88\mathrm{pF}$ (from the previous unit). Now slide a piece of paper of dielectric constant $\kappa =4$ into the gap, filling it completely. What is the new capacitance, and if you charge it to $100\mathrm{V}$, how much charge does it now hold?

Step 1. The new capacitance with the dielectric present is $C=\kappa C_0=4\times 0.88\mathrm{pF}=3.5\mathrm{pF}$.

Step 2. At voltage $V=100\mathrm{V}$, the charge on each plate is $Q=CV=(3.5\times 10^{-12})(100)=3.5\times 10^{-10}\mathrm{C}$, or $350\mathrm{pC}$.

Step 3. Compare with the vacuum case: at the same voltage the vacuum capacitor held only $Q_0=C_{0}V=88\mathrm{pC}$. The paper-filled capacitor holds $4\times$ as much charge at the same voltage.

Step 4. Where does the extra charge go? The plates accept it because the bound surface charges on the dielectric partially cancel the field of the free charges; the plates can pile on more free charge before the voltage between them reaches $100\mathrm{V}$.

What this tells us: a dielectric multiplies a capacitor's storage by exactly its dielectric constant, with no other change to the geometry. This is why ceramic capacitors with $\kappa \sim 1000$ can be a thousand times smaller than vacuum capacitors of the same rating.

Check your understanding Beginner

Formal definition Intermediate+

A dielectric is a region of space $\Omega \subset {\mathbb{R}}^3$ filled with bound charges whose response to an applied electric field is to polarize. The local polarization state is described by the polarization vector field $\mathbf{P}(\mathbf{r})$, defined as the electric dipole moment per unit volume:

$$\mathbf{P}(\mathbf{r})=\underset{\Delta V\to 0}{\lim }\frac{1}{\Delta V}\sum _{i\in \Delta V}{\mathbf{p}}_i,$$

where the sum runs over the molecular dipoles ${\mathbf{p}}_i$ inside a small parcel $\Delta V$ centred on $\mathbf{r}$. The limit is taken in the macroscopic sense — $\Delta V$ small on a laboratory scale but containing many molecules.

Bound charge densities. A polarized medium is electrically equivalent to a distribution of bound charges. The volume bound-charge density is

$${\rho }_b(\mathbf{r})=-\nabla \cdot \mathbf{P}(\mathbf{r}),$$

and the surface bound-charge density on a face with outward normal $\hat{\mathbf{n}}$ is

$${\sigma }_b=\mathbf{P}\cdot \hat{\mathbf{n}}.$$

These identities are exact consequences of the dipole moment expansion of the electrostatic potential.

Electric displacement. Combining Gauss's law $\nabla \cdot \mathbf{E}={\rho }_{\text{total}}/{ϵ}_0$ with the decomposition ${\rho }_{\text{total}}={\rho }_{\text{free}}+{\rho }_b$ and the bound-charge identity gives

$$\nabla \cdot ({ϵ}_0\mathbf{E}+\mathbf{P})={\rho }_{\text{free}}.$$

This motivates the definition of the electric displacement $\mathbf{D}$:

$$\mathbf{D}={ϵ}_0\mathbf{E}+\mathbf{P},\nabla \cdot \mathbf{D}={\rho }_{\text{free}}.$$

Only the free charge sources $\mathbf{D}$. The displacement field hides the bound-charge bookkeeping inside the constitutive relation $\mathbf{P}(\mathbf{E})$.

Linear dielectrics. In a linear, isotropic, homogeneous dielectric,

$$\mathbf{P}={ϵ}_0{\chi }_e\mathbf{E},$$

where ${\chi }_e$ is the electric susceptibility (a dimensionless material constant). Then $\mathbf{D}={ϵ}_0(1+{\chi }_e)\mathbf{E}=ϵ\mathbf{E}$, where $ϵ={ϵ}_0(1+{\chi }_e)$ is the permittivity and $\kappa =ϵ/{ϵ}_0=1+{\chi }_e$ is the relative permittivity (or dielectric constant).

Boundary conditions at a dielectric interface. Let $S$ be a surface separating two dielectrics with permittivities ${ϵ}_1$ and ${ϵ}_2$, with outward normal $\hat{\mathbf{n}}$ pointing from $1$ to $2$, and let ${\sigma }_f$ be any free surface charge on $S$. Pillbox and stokes arguments give

$${\mathbf{D}}_2\cdot \hat{\mathbf{n}}-{\mathbf{D}}_1\cdot \hat{\mathbf{n}}={\sigma }_f,{\mathbf{E}}_{2,\parallel }={\mathbf{E}}_{1,\parallel }.$$

In words: the normal component of $\mathbf{D}$ jumps by ${\sigma }_f$, and the tangential component of $\mathbf{E}$ is continuous. When ${\sigma }_f=0$, $D_n$ is continuous across the interface; $E_n$ is not.

Energy in a linear dielectric. The electrostatic energy stored in a linear dielectric is

$$U=\frac{1}{2}{\int }_{\Omega }\mathbf{D}\cdot \mathbf{E}{d}^{3}r=\frac{{ϵ}_0}{2}{\int }_{\Omega }\kappa (\mathbf{r})|\mathbf{E}{|}^2{d}^{3}r.$$

This generalises the vacuum formula $U=({ϵ}_0/2)\int |\mathbf{E}{|}^2{d}^{3}r$ from unit 10.01.03 by replacing ${ϵ}_0$ with $ϵ(\mathbf{r})$.

Counterexamples to common slips Intermediate+

• The field lines of $\mathbf{D}$ and $\mathbf{E}$ generally point in different directions inside a dielectric. They coincide only in linear isotropic media; in an anisotropic crystal, $\mathbf{P}$ need not be parallel to $\mathbf{E}$, so $\mathbf{D}$ and $\mathbf{E}$ differ in direction and the permittivity becomes a tensor ${ϵ}_{ij}$.
• $\mathbf{D}$ is not "the field" — it is the bookkeeping device for free charges. The physical force per unit charge is $\mathbf{E}$, not $\mathbf{D}$; a test charge in a dielectric feels $\mathbf{E}$, not $\mathbf{D}$. $\mathbf{D}$ is useful because its divergence sees only free charges, but it does not have a Coulomb interpretation.
• Bound surface charge density and free surface charge density are physically distinct. Bound surface charge ${\sigma }_b=\mathbf{P}\cdot \hat{\mathbf{n}}$ comes from molecular polarization and disappears when the field is removed; free surface charge ${\sigma }_f$ comes from charges placed on the conductor and persists. The two add to give the total surface charge that produces the discontinuity in $\mathbf{E}$.

Key theorem with proof Intermediate+

Theorem (dielectric sphere in a uniform external field). A linear isotropic dielectric sphere of radius $R$ and dielectric constant $\kappa$ placed in a uniform external electric field ${\mathbf{E}}_{\infty }=E_{\infty }\hat{\mathbf{z}}$ acquires a uniform polarization

$$\mathbf{P}=\frac{3(\kappa -1)}{\kappa +2}{ϵ}_0{\mathbf{E}}_{\infty },$$

and an induced dipole moment

$$\mathbf{p}=4\pi {ϵ}_0{R}^3\frac{\kappa -1}{\kappa +2}{\mathbf{E}}_{\infty }.$$

The field inside the sphere is uniform and equal to ${\mathbf{E}}_{\text{in}}=3{\mathbf{E}}_{\infty }/(\kappa +2)$.

Proof. Choose the $z$-axis along ${\mathbf{E}}_{\infty }$ and place the sphere at the origin. The potential satisfies Laplace's equation ${\nabla }^{2}V=0$ both inside and outside the sphere, with boundary conditions: $V\to -E_{\infty }z=-E_{\infty }r\cos \theta$ as $r\to \infty$; $V$ continuous at $r=R$; and $D_r$ continuous at $r=R$ (no free surface charge there).

Inside the sphere, regularity at the origin restricts the harmonic solution to terms of the form $r^{\ell }{P}_{\ell }(\cos \theta )$. The asymptotic boundary condition couples to the $\ell =1$ Legendre polynomial only, so by superposition

$$V_{\text{in}}(r,\theta )=Ar\cos \theta ,r<R,$$

for some constant $A$ to be determined. Outside, the harmonic solution allowed by the asymptotic condition takes the form

$$V_{\text{out}}(r,\theta )=-E_{\infty }r\cos \theta +\frac{B\cos \theta }{r^2},r>R,$$

where the second term is the field of an induced dipole of strength $4\pi {ϵ}_{0}B$ located at the origin.

The continuity of $V$ at $r=R$ gives

$$AR\cos \theta =-E_{\infty }R\cos \theta +\frac{B\cos \theta }{R^2},B=R^3(A+E_{\infty }).$$

For the second boundary condition, compute the radial derivatives. Inside, ${\partial }_r{V}_{\text{in}}=A\cos \theta$. Outside, ${\partial }_r{V}_{\text{out}}=-E_{\infty }\cos \theta -2B\cos \theta /r^3$, evaluated at $r=R^{+}$ gives $-E_{\infty }\cos \theta -2B\cos \theta /R^3$. The radial component of $\mathbf{D}$ is $D_r=-ϵ{\partial }_{r}V$ inside and $D_r=-{ϵ}_0{\partial }_{r}V$ outside; continuity of $D_r$ at $r=R$ (no free surface charge) gives

$$-ϵA=-{ϵ}_0\left(-E_{\infty }-\frac{2B}{R^3}\right)={ϵ}_0{E}_{\infty }+\frac{2{ϵ}_{0}B}{R^3}.$$

Substituting $B=R^3(A+E_{\infty })$:

$$-ϵA={ϵ}_0{E}_{\infty }+2{ϵ}_0(A+E_{\infty })=3{ϵ}_0{E}_{\infty }+2{ϵ}_{0}A,$$ $$(ϵ+2{ϵ}_0)A=-3{ϵ}_0{E}_{\infty },A=-\frac{3{ϵ}_0}{ϵ+2{ϵ}_0}{E}_{\infty }=-\frac{3}{\kappa +2}{E}_{\infty }.$$

The interior field is ${\mathbf{E}}_{\text{in}}=-\nabla V_{\text{in}}=-A\hat{\mathbf{z}}=\frac{3}{\kappa +2}{E}_{\infty }\hat{\mathbf{z}}$. The polarization in the sphere is $\mathbf{P}={ϵ}_0(\kappa -1){\mathbf{E}}_{\text{in}}={ϵ}_0\cdot 3(\kappa -1)/(\kappa +2)\cdot E_{\infty }\hat{\mathbf{z}}$. The induced dipole moment is the polarization times the volume of the sphere:

$$\mathbf{p}=\mathbf{P}\cdot \frac{4}{3}\pi R^3=4\pi {ϵ}_0{R}^3\frac{\kappa -1}{\kappa +2}{\mathbf{E}}_{\infty },$$

in agreement with the dipole-field identification $B=p/(4\pi {ϵ}_0)$. $■$

Bridge. The dielectric-sphere result is the foundational reason that all isotropic dielectric BVPs reduce to the same algebraic pattern: the polarization scales with the Clausius-Mossotti combination $(\kappa -1)/(\kappa +2)$, and this is exactly the local-field correction that builds toward 10.01.05 multipole expansions, where the dipole moment $\mathbf{p}$ is the leading multipole of the induced response. The same algebraic combination appears again in the Clausius-Mossotti relation (Theorem 4 below) for the bulk dielectric constant in terms of molecular polarizabilities, where it tracks the depolarization field at the molecular site. Putting these together with the energy formula $U=\frac{1}{2}\int \mathbf{D}\cdot \mathbf{E}{d}^{3}r$, the bridge is between local microscopic response and macroscopic constitutive relations: every dielectric scenario from VLSI gate-dielectric design to colloidal force calculations reduces to specifying $\kappa$ and solving the same divergence-form elliptic BVP $\nabla \cdot (ϵ\nabla V)=-{\rho }_{\text{free}}$.

Exercises Intermediate+

Advanced results Master

Theorem 1 (Bound charge identities, Poisson 1812). Let $\mathbf{P}$ be a smooth polarization field on $\Omega \subset {\mathbb{R}}^3$, vanishing outside $\Omega$. The macroscopic electrostatic potential produced by $\mathbf{P}$ is identical to that of a charge distribution with volume density ${\rho }_b=-\nabla \cdot \mathbf{P}$ in $\Omega$ and surface density ${\sigma }_b=\mathbf{P}\cdot \hat{\mathbf{n}}$ on $\partial \Omega$. (Proof in Full Proof Set, Proposition 1.)

Theorem 2 (Gauss's law in matter). In a region containing both free and bound charges, the electric displacement $\mathbf{D}={ϵ}_0\mathbf{E}+\mathbf{P}$ satisfies $\nabla \cdot \mathbf{D}={\rho }_{\text{free}}$. Equivalently, ${\oint }_S\mathbf{D}\cdot d\mathbf{A}=Q_{\text{free, enc}}$ for any closed surface $S$.

Theorem 3 (Dielectric BVP — general statement, Stratton 1941). In a region $\Omega$ of ${\mathbb{R}}^3$ partitioned by smooth interfaces ${\Gamma }_k$ into subregions ${\Omega }_k$ each filled with linear isotropic dielectric of permittivity ${ϵ}_k(\mathbf{r})$, the electrostatic potential satisfies the divergence-form elliptic equation

$$-\nabla \cdot [ϵ(\mathbf{r})\nabla V]={\rho }_{\text{free}},\mathbf{r}\in \Omega ,$$

with transmission conditions $V$ continuous across each ${\Gamma }_k$, and $ϵ{\partial }_{n}V$ jumping by $-{\sigma }_f$ where there is free surface charge. With prescribed boundary data on $\partial \Omega$ (Dirichlet, Neumann, or mixed), the problem has a unique weak solution in $H^1(\Omega )$.

Theorem 4 (Clausius-Mossotti relation, Mossotti 1850, Clausius 1879). For a non-polar dielectric with molecular polarizability $\alpha$ and number density $n$, the dielectric constant satisfies

$$\frac{\kappa -1}{\kappa +2}=\frac{n\alpha }{3{ϵ}_0}.$$

The derivation uses the Lorentz local-field correction ${\mathbf{E}}_{\text{local}}=\mathbf{E}+\mathbf{P}/(3{ϵ}_0)$, arising from a spherical-cavity model of the field at a molecular site. (Derivation in Exercise 10.) The Lorentz local-field formula is exact for cubic lattices and approximate for amorphous media.

Theorem 5 (Onsager equation for polar dielectrics, Onsager 1936). For a polar fluid with isotropic molecular polarizability $\alpha$ and permanent molecular dipole moment $p$ at temperature $T$, the dielectric constant satisfies

$$\frac{(\kappa -1)(2\kappa +1)}{9\kappa }=\frac{n}{3{ϵ}_0}\left(\alpha +\frac{p^2}{3k_{B}T}\right).$$

The improvement over Debye 1929 (which used the Lorentz cavity unchanged) lies in treating the polarizable molecule as a point dipole inside a spherical cavity in the surrounding polarized fluid, accounting for the reaction field that the surroundings produce back on the dipole. Onsager's 1936 paper in the J. Am. Chem. Soc. corrected the long-standing failure of Clausius-Mossotti for polar liquids, where naive extrapolation of $n\alpha /(3{ϵ}_0)$ predicts a ferroelectric divergence well before any real liquid exhibits one. The Onsager formula correctly predicts the temperature dependence of water's dielectric constant down to a few percent over the liquid range.

Theorem 6 (Kirkwood correlation factor, Kirkwood 1939). For a polar fluid with short-range orientational correlations, Onsager's formula generalises to

$$\frac{(\kappa -1)(2\kappa +1)}{9\kappa }=\frac{n}{3{ϵ}_0}\left(\alpha +\frac{g{p}^2}{3k_{B}T}\right),$$

where $g=1+z⟨\cos \theta ⟩$ is the Kirkwood correlation factor counting how strongly a molecule's $z$ nearest neighbours align with it ($⟨\cos \theta ⟩$ is the mean cosine of the angle between dipoles). Liquid water has $g\approx 2.7$ at $25{}^{\circ }$C, reflecting hydrogen-bond alignment.

Theorem 7 (Ferroelectric transition — Landau theory, Devonshire 1949). In a ferroelectric crystal, the free energy near the Curie temperature $T_c$ expands as

$$F(P,T)=F_0(T)+\frac{1}{2}\alpha (T-T_c)P^2+\frac{1}{4}\beta P^4+\cdots ,$$

with $\alpha >0$. For $T>T_c$ the minimum is at $P=0$ (paraelectric phase); for $T<T_c$ the minimum is at $P^2=\alpha (T_c-T)/\beta$ (ferroelectric phase with spontaneous polarization). Above $T_c$ the susceptibility diverges as $\chi \sim 1/(T-T_c)$ (Curie-Weiss law); the dielectric constant of barium titanate reaches $\kappa \sim 10^4$ at the transition. Below $T_c$, the polarization is a hysteretic function of the applied field, exactly analogous to magnetization in a ferromagnet.

Theorem 8 (Maxwell relation: optical permittivity equals refractive index squared). For a non-conducting non-magnetic dielectric at optical frequencies, the relative permittivity $\kappa (\omega )$ and the refractive index $n(\omega )$ are related by

$$\kappa (\omega )=n(\omega {)}^2.$$

This is a corollary of Maxwell's equations: the phase velocity of an electromagnetic wave in a linear dielectric is $v=c/\sqrt{\kappa {\mu }_r}$, and with ${\mu }_r\approx 1$ the optical refractive index $n=c/v=\sqrt{\kappa }$. The frequency-dependence of $\kappa$ at optical frequencies is the origin of dispersion (the index varies with colour), which appears again in 10.04.05 pending as the Kramers-Kronig relations linking the real and imaginary parts of $ϵ(\omega )$.

Theorem 9 (Anisotropic dielectrics and the permittivity tensor). In a crystalline dielectric without isotropic symmetry, the constitutive relation is

$$D_i={ϵ}_{ij}{E}_j,{ϵ}_{ij}={ϵ}_{ji},$$

with ${ϵ}_{ij}$ a symmetric positive-definite tensor whose principal axes coincide with the crystal's principal optical axes. Diagonalising ${ϵ}_{ij}$ yields three principal permittivities ${ϵ}_1,{ϵ}_2,{ϵ}_3$; the crystal is biaxial when all three differ, uniaxial when two coincide, isotropic when all three coincide. Light propagation in an anisotropic dielectric splits into two orthogonally polarized modes with different phase velocities, producing the phenomenon of birefringence (Bartholin 1669 in Iceland spar, Huygens 1690 in his Traité de la Lumière). The Fresnel ellipsoid ${\sum }_i{x}_i^2/{ϵ}_i=1$ and the index ellipsoid (its dual) organise the geometry of light propagation through the crystal. Symmetry of ${ϵ}_{ij}$ is enforced by the energy condition $U=\frac{1}{2}{ϵ}_{ij}{E}_i{E}_j>0$: an antisymmetric part would produce no contribution to the energy and would have to vanish in equilibrium.

Theorem 10 (Piezoelectric coupling, Voigt 1910). In a non-centrosymmetric crystal, mechanical strain ${\epsilon }_{kl}$ and electric polarization $P_i$ are linearly coupled through the piezoelectric tensor $d_{ikl}$:

$$P_i=d_{ikl}{\sigma }_{kl}(\text{direct effect}),{\epsilon }_{kl}=d_{ikl}{E}_i(\text{converse effect}),$$

where ${\sigma }_{kl}$ is the mechanical stress tensor. The direct effect (mechanical stress producing polarization) was discovered by the Curie brothers 1880; the converse effect (electric field producing strain) was predicted thermodynamically by Lippmann 1881 and verified experimentally by the Curies the same year. The piezoelectric tensor $d_{ikl}$ vanishes identically in any centrosymmetric crystal — a corollary of inversion symmetry — restricting piezoelectric activity to 20 of the 32 crystallographic point groups. Modern applications include quartz frequency standards (Cady 1921, Pierce 1923), ultrasound transducers (Sokolov 1929, Firestone 1942), MEMS resonators in cell-phone gyroscopes, and energy-harvesting devices.

Theorem 11 (Dielectric breakdown). A dielectric subjected to an electric field above a material-specific critical value $E_c$ undergoes irreversible electrical breakdown: the bound electrons are ionised and the material becomes a momentary conductor. Typical critical fields: air ≈ 3 MV/m at standard pressure (the Paschen curve giving the breakdown voltage as a function of pressure-distance product, Paschen 1889); glass ≈ 10–25 MV/m; polystyrene and polyethylene ≈ 20–25 MV/m; mica ≈ 100 MV/m; thermally grown SiO${}_2$ ≈ 1 GV/m. The microscopic mechanism in solids is a Townsend avalanche: an initial free electron accelerated by the field gains enough energy between collisions to ionise lattice atoms, releasing more free electrons in an exponentially growing cascade. Breakdown sets an upper limit on the operating voltage of any capacitor, transformer, or transmission line; the product $\frac{1}{2}ϵE_c^2$ is the maximum stored energy density before catastrophic failure, ranging from $\sim 40{J/m}^3$ for air to $\sim 4\times 10^7{J/m}^3$ for mica.

Theorem 12 (Berry-phase polarization in crystalline solids, King-Smith-Vanderbilt 1993). In a periodic crystal, the macroscopic polarization is

$$P_i=\frac{e}{(2\pi {)}^3}\sum _n{\int }_{\text{BZ}}\mathrm{Im}⟨u_{n\mathbf{k}}|{\partial }_{k_i}{u}_{n\mathbf{k}}⟩d^{3}k,$$

where $u_{n\mathbf{k}}$ are the periodic parts of the Bloch wavefunctions, summed over occupied bands and integrated over the Brillouin zone. The 1993 King-Smith-Vanderbilt paper in Phys. Rev. B showed that the classical $\mathbf{P}=\int \rho \mathbf{r}{d}^{3}r$ definition is ill-defined for a periodic crystal (it depends on the choice of unit cell) and replaced it with the Berry-phase formula, modulo a polarization quantum $eR/V_{\text{cell}}$ where $R$ is a lattice vector. Differences in polarization across phase transitions (e.g., switching directions in a ferroelectric) are well-defined and Berry-phase-computable; the modern theory makes polarization a topological observable of the electronic band structure, with applications to ferroelectric switching, multiferroics, and topological insulators (Haldane 1988, Kane-Mele 2005, Fu-Kane-Mele 2007).

Synthesis. The displacement field $\mathbf{D}$ is the foundational reason that electrostatics in matter retains the structure of vacuum electrostatics: $\nabla \cdot \mathbf{D}={\rho }_{\text{free}}$ replaces $\nabla \cdot \mathbf{E}=\rho /{ϵ}_0$, and every theorem about vacuum capacitance, energy, and BVPs generalises by replacing ${ϵ}_0$ with $ϵ(\mathbf{r})$. The central insight is that the polarization $\mathbf{P}$ separates cleanly into bulk and surface contributions ${\rho }_b=-\nabla \cdot \mathbf{P}$ and ${\sigma }_b=\mathbf{P}\cdot \hat{\mathbf{n}}$, so that all the material physics is hidden inside the single constitutive relation $\mathbf{P}(\mathbf{E})$. Putting these together with the Clausius-Mossotti relation (Theorem 4), the bridge is between microscopic molecular polarizability and macroscopic dielectric constant: a single algebraic combination $(\kappa -1)/(\kappa +2)$ controls the connection.

The pattern recurs across scales. The Onsager refinement (Theorem 5) generalises the Clausius-Mossotti formula by accounting for the reaction field of the polarized surroundings, and appears again in 11.06.02 pending Landau theory of phase transitions as the mean-field treatment of order parameters coupled to their own response field. The ferroelectric transition (Theorem 7) is exactly the canonical Landau second-order transition with $P$ as the order parameter, dual to the ferromagnetic transition with magnetisation $M$ as the order parameter. The Maxwell relation $\kappa =n^2$ (Theorem 8) builds toward 10.04.05 pending dispersion theory, where the frequency-dependent permittivity $ϵ(\omega )$ — analytic in the upper half-plane by causality — relates absorption and dispersion through the Kramers-Kronig integrals. Across all these settings, the displacement-polarization formalism organises the constitutive physics into a single divergence-form elliptic operator $\nabla \cdot (ϵ\nabla V)$, the same operator that governs heat conduction, steady-state diffusion, and incompressible flow.

Full proof set Master

Proposition 1 (Macroscopic potential of a polarized medium). Let $\mathbf{P}$ be a smooth polarization field supported in $\Omega \subset {\mathbb{R}}^3$. The macroscopic electrostatic potential at $\mathbf{r}\notin \Omega$ is

$$V(\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\left[{\int }_{\Omega }\frac{{\rho }_b({\mathbf{r}}^{\prime })}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}{d}^3{r}^{\prime }+{\oint }_{\partial \Omega }\frac{{\sigma }_b({\mathbf{r}}^{\prime })}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}d{A}^{\prime }\right],$$

with ${\rho }_b=-\nabla \cdot \mathbf{P}$ and ${\sigma }_b=\mathbf{P}\cdot \hat{\mathbf{n}}$.

Proof. The macroscopic potential is the sum of contributions from each volume element $d^3{r}^{\prime }$ at ${\mathbf{r}}^{\prime }$, which carries dipole moment $\mathbf{P}({\mathbf{r}}^{\prime })d^3{r}^{\prime }$. A pure dipole $\mathbf{p}$ at ${\mathbf{r}}^{\prime }$ produces potential

$$\varphi (\mathbf{r})=\frac{\mathbf{p}\cdot (\mathbf{r}-{\mathbf{r}}^{\prime })}{4\pi {ϵ}_0|\mathbf{r}-{\mathbf{r}}^{\prime }{|}^3}=\frac{1}{4\pi {ϵ}_0}\mathbf{p}\cdot {\nabla }^{\prime }\frac{1}{|\mathbf{r}-{\mathbf{r}}^{\prime }|},$$

where ${\nabla }^{\prime }$ acts on the source variable ${\mathbf{r}}^{\prime }$. Summing:

$$V(\mathbf{r})=\frac{1}{4\pi {ϵ}_0}{\int }_{\Omega }\mathbf{P}({\mathbf{r}}^{\prime })\cdot {\nabla }^{\prime }\frac{1}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}{d}^3{r}^{\prime }.$$

Use $\mathbf{P}\cdot {\nabla }^{\prime }f={\nabla }^{\prime }\cdot (f\mathbf{P})-f{\nabla }^{\prime }\cdot \mathbf{P}$ with $f=1/|\mathbf{r}-{\mathbf{r}}^{\prime }|$:

$$V(\mathbf{r})=\frac{1}{4\pi {ϵ}_0}{\int }_{\Omega }{\nabla }^{\prime }\cdot \left(\frac{\mathbf{P}}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}\right)d^3{r}^{\prime }-\frac{1}{4\pi {ϵ}_0}{\int }_{\Omega }\frac{{\nabla }^{\prime }\cdot \mathbf{P}}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}{d}^3{r}^{\prime }.$$

The first integral converts by the divergence theorem to a surface integral over $\partial \Omega$ of $\mathbf{P}\cdot \hat{\mathbf{n}}/|\mathbf{r}-{\mathbf{r}}^{\prime }|$. Identifying the second integral as the volume contribution of $-{\nabla }^{\prime }\cdot \mathbf{P}$, the result follows:

$$V(\mathbf{r})=\frac{1}{4\pi {ϵ}_0}\left[{\oint }_{\partial \Omega }\frac{\mathbf{P}\cdot \hat{\mathbf{n}}}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}d{A}^{\prime }+{\int }_{\Omega }\frac{-{\nabla }^{\prime }\cdot \mathbf{P}}{|\mathbf{r}-{\mathbf{r}}^{\prime }|}{d}^3{r}^{\prime }\right].■$$

Proposition 2 (Energy of a linear dielectric system). For a linear dielectric configuration with $\mathbf{D}=ϵ\mathbf{E}$, the electrostatic energy is

$$U=\frac{1}{2}\int \mathbf{D}\cdot \mathbf{E}{d}^{3}r=\frac{1}{2}\int {\rho }_{\text{free}}V{d}^{3}r.$$

Proof. Charge up the system from $\mathbf{D}=0$ to its final state by parameter $\lambda \in [0,1]$, scaling ${\rho }_{\text{free}}\to \lambda {\rho }_{\text{free}}$. By linearity, $\mathbf{D}(\lambda )=\lambda \mathbf{D}$, $\mathbf{E}(\lambda )=\lambda \mathbf{E}$, $V(\lambda )=\lambda V$. The work done to add the increment $d\lambda$ of free charge $d\lambda {\rho }_{\text{free}}$ against the existing potential $\lambda V$ is

$$dW=\int d\lambda {\rho }_{\text{free}}(\mathbf{r})\lambda V(\mathbf{r})d^{3}r=\lambda d\lambda \int {\rho }_{\text{free}}V{d}^{3}r.$$

Integrating from $\lambda =0$ to $1$ gives $U=\frac{1}{2}\int {\rho }_{\text{free}}V{d}^{3}r$. To convert to the $\mathbf{D}\cdot \mathbf{E}$ form, use ${\rho }_{\text{free}}=\nabla \cdot \mathbf{D}$ and integrate by parts:

$$\int {\rho }_{\text{free}}V{d}^{3}r=\int V\nabla \cdot \mathbf{D}{d}^{3}r=-\int \nabla V\cdot \mathbf{D}{d}^{3}r+\text{boundary}=\int \mathbf{E}\cdot \mathbf{D}{d}^{3}r,$$

with boundary terms vanishing for localised systems. So $U=\frac{1}{2}\int \mathbf{D}\cdot \mathbf{E}{d}^{3}r$. $■$

Proposition 3 (Dielectric pulled into a capacitor at fixed voltage). A parallel-plate capacitor of plate area $A$ and gap $d$ is held at fixed voltage $V$ by a battery. A slab of dielectric constant $\kappa$ partially fills the gap to depth $x$ along the plate width $L$ (so $A=L\cdot w$ with width $w$). The force on the slab is

$$F=+\frac{1}{2}\frac{V^2{ϵ}_{0}w(\kappa -1)}{d},$$

directed to pull the slab further into the gap.

Proof. The capacitance with the slab inserted to depth $x$ is

$$C(x)=\frac{{ϵ}_{0}w}{d}\left[\kappa x+(L-x)\right]=\frac{{ϵ}_{0}w}{d}\left[L+(\kappa -1)x\right].$$

At fixed voltage $V$, the energy stored is $U_{\text{cap}}=\frac{1}{2}C(x)V^2$, but the battery also does work as it pushes charge onto the plates when $C$ increases. The total system energy (battery + capacitor) at fixed $V$ behaves with sign such that the generalised force on the slab is

$$F=+\frac{\partial U_{\text{cap}}}{\partial x}{|}_V=+\frac{1}{2}{V}^2\frac{\partial C}{\partial x}=+\frac{1}{2}{V}^2\cdot \frac{{ϵ}_{0}w(\kappa -1)}{d},$$

where the plus sign (instead of the minus sign familiar from fixed-charge problems) arises because the battery delivers energy $VdQ=V^{2}dC$ during the displacement, of which half goes into field energy and the other half drives the slab. The same force pulls the slab inward regardless of the sign convention; physically, the field-fringing region at the entry edge of the slab grabs onto the polarized bound charges and accelerates them toward higher capacitance. $■$

Proposition 4 (Microscopic origins of $\alpha$ for a non-polar dielectric). For a non-polar atom in a weak external field, the electronic polarizability ${\alpha }_e$ from a simple harmonic-oscillator model is

$${\alpha }_e=\frac{e^2}{m_e{\omega }_0^2},$$

where ${\omega }_0$ is the characteristic resonance frequency of the bound electron and $m_e$ is the electron mass.

Proof. Model the bound electron as a harmonic oscillator of natural frequency ${\omega }_0$ around the nucleus. Apply a static external field ${\mathbf{E}}_0$ in the $z$-direction. The equation of motion is

$$m_e\ddot{z}=-m_e{\omega }_0^{2}z-e{E}_0,$$

with the sign convention that the electron charge is $-e$ with $e>0$. In steady state, $\ddot{z}=0$ gives $z=-e{E}_0/(m_e{\omega }_0^2)$. The induced dipole moment is $p=(-e)z=e^2{E}_0/(m_e{\omega }_0^2)$, so ${\alpha }_e=p/E_0=e^2/(m_e{\omega }_0^2)$. $■$

Plugging in ${\omega }_0\sim 10^{16}rad/s$ (a typical valence-electron resonance, corresponding to UV transition energies) gives ${\alpha }_e\sim 10^{-40}{\mathrm{C}}^{2}m/N$, and via Clausius-Mossotti $\kappa \sim 1+n\alpha /{ϵ}_0\sim 1+0.1$ for $n\sim 3\times 10^{28}atoms/{\mathrm{m}}^3$ (atmospheric pressure-times-density) — within an order of magnitude of measured dielectric constants for non-polar gases. Adding ionic polarizability ${\alpha }_i=e^2/(M{\omega }_{\text{phonon}}^2)$ (an ionic oscillator with reduced mass $M$ and optical-phonon frequency) captures the lower-frequency response of ionic solids like NaCl, where the static $\kappa \approx 5.9$ exceeds the optical $n^2\approx 2.4$ because phonon modes respond at static frequencies but freeze out at optical frequencies.

Proposition 5 (Onsager's reaction-field derivation, sketch). The Onsager equation Theorem 5 follows from carefully accounting for two cavity-field contributions: (i) the cavity field $\mathbf{F}$ that the polarized surroundings produce at an empty spherical cavity in the dielectric, and (ii) the reaction field $\mathbf{R}$ that the surroundings produce in response to the dipole at the centre of the cavity.

Proof sketch. For a spherical cavity of radius $a$ carved from a dielectric of permittivity $ϵ$, with external macroscopic field $\mathbf{E}$ in the bulk, the field inside the empty cavity is $\mathbf{F}=3ϵ/(2ϵ+{ϵ}_0)\cdot \mathbf{E}$ (a standard BVP, dual to the dielectric-sphere problem of the Key Theorem). Now place a point dipole ${\mathbf{p}}_{\text{mol}}$ at the centre of the cavity. The dipole polarizes the surrounding dielectric, which in turn produces a reaction field $\mathbf{R}=[2(ϵ-{ϵ}_0)/(2ϵ+{ϵ}_0)]\cdot {\mathbf{p}}_{\text{mol}}/(4\pi {ϵ}_0{a}^3)$ back at the cavity centre, directed parallel to ${\mathbf{p}}_{\text{mol}}$. The total field at the dipole site is $\mathbf{F}+\mathbf{R}$, and the dipole responds to this total: ${\mathbf{p}}_{\text{mol}}=\alpha (\mathbf{F}+\mathbf{R})+{\mathbf{p}}_0$, where $\alpha$ is the molecular polarizability and ${\mathbf{p}}_0$ is the permanent dipole moment (with thermal averaging via Langevin's formula, $⟨{\mathbf{p}}_0{⟩}_{\text{thermal}}=p^2(\mathbf{F}+\mathbf{R})/(3k_{B}T)$ in the high-temperature limit). Solving the self-consistent system for $\mathbf{P}=n⟨{\mathbf{p}}_{\text{mol}}⟩$ and identifying $\mathbf{P}={ϵ}_0(\kappa -1)\mathbf{E}$, after algebraic rearrangement, gives the Onsager formula

$$\frac{(\kappa -1)(2\kappa +1)}{9\kappa }=\frac{n}{3{ϵ}_0}\left(\alpha +\frac{p^2}{3k_{B}T}\right).$$

Onsager's 1936 paper laid out the full derivation; modern treatments (Böttcher 1973 Theory of Electric Polarization) streamline the algebra but keep the same physical content. $■$

Connections Master

• Conductors, capacitance, and electrostatic energy 10.01.03. The capacitance-multiplication factor $\kappa$ is the central engineering use of dielectrics; the energy formula $U=\frac{1}{2}\int \mathbf{D}\cdot \mathbf{E}{d}^{3}r$ generalises the vacuum formula from 10.01.03 by replacing ${ϵ}_0$ with $ϵ(\mathbf{r})$, and the Dirichlet-to-Neumann capacitance-matrix framework of 10.01.03 carries over verbatim with the divergence-form operator $\nabla \cdot (ϵ\nabla V)$ in place of the constant-coefficient Laplacian.

• Laplace equation and BVPs 10.01.02. The dielectric BVP $\nabla \cdot (ϵ\nabla V)=-{\rho }_{\text{free}}$ with transmission conditions across dielectric interfaces is the divergence-form generalisation of the constant-coefficient Laplace problem treated in 10.01.02; the separation-of-variables techniques and method of images both extend to dielectrics, as exhibited by the dielectric-sphere problem (Key Theorem) and the dielectric-plane image problem (Exercise 8).

• Multipole expansion 10.01.05 (pending). The induced dipole moment of a polarized body is the leading multipole of its far-field potential, and the next-order quadrupole moment captures anisotropy of the polarization; the systematic multipole framework developed in 10.01.05 organises dielectric scattering, antenna response, and intermolecular forces.

• Dispersion and Kramers-Kronig relations 10.04.05 pending (pending). The frequency-dependent permittivity $ϵ(\omega )$ — analytic in the upper half-plane by causality — connects the real and imaginary parts through the Kramers-Kronig integrals. The static $\kappa$ of this unit is the $\omega \to 0$ limit; the high-frequency limit gives the squared refractive index $n^2$ via Maxwell's relation (Theorem 8).

• Landau theory of phase transitions 11.06.02 pending (pending). The ferroelectric transition (Theorem 7) is the canonical realisation of Landau's mean-field theory of second-order phase transitions, with polarization $P$ as the order parameter; the same expansion governs ferromagnetism, superconductivity (with the Ginzburg-Landau $\psi$), and the liquid-gas critical point.

• Magnetic polarization and the H-field 10.02.04 pending (pending). The magnetic analogue of the dielectric formalism replaces $\mathbf{P}$ with the magnetization $\mathbf{M}$, replaces $\mathbf{D}={ϵ}_0\mathbf{E}+\mathbf{P}$ with $\mathbf{H}=\mathbf{B}/{\mu }_0-\mathbf{M}$, and replaces $\kappa$ with the relative permeability ${\mu }_r$. The structural parallel is exact: only the free currents source $\mathbf{H}$ (as only the free charges source $\mathbf{D}$), boundary conditions transcribe by replacing electric quantities with magnetic ones, and the bulk constitutive theory (linear paramagnetism, diamagnetism, ferromagnetism) parallels the dielectric, polar-dielectric, and ferroelectric phenomenology.

Historical & philosophical context Master

Faraday's 1837 Experimental Researches in Electricity, Eleventh Series ^{[Faraday1837]}, introduced the term dielectric and the concept of specific inductive capacity — what is now $\kappa$ — through a careful experimental programme in which he placed different insulating materials between the plates of capacitors and measured the change in charge stored at fixed voltage. He found that the capacity ratio depended on the material in a way that could not be explained by action-at-a-distance Coulomb theory, and proposed instead that the medium itself participated in the electric action through what he called polarization. This was the conceptual bridge from Coulombic to field-theoretic electrostatics; Maxwell's later identification of the dielectric constant with the squared refractive index closed the loop from electricity to optics.

The macroscopic theory was systematised by Mossotti 1850 ^{[Mossotti1850]} in his Discussione analitica and by Clausius 1879 ^{[Clausius1879]} in Die Mechanische Behandlung der Electricität, who independently derived the relation now bearing both names. Their model treated the dielectric as a regular array of polarizable spheres, applied Lorentz's local-field correction ${\mathbf{E}}_{\text{local}}=\mathbf{E}+\mathbf{P}/(3{ϵ}_0)$ (the cavity field), and arrived at $(\kappa -1)/(\kappa +2)=n\alpha /(3{ϵ}_0)$. The piezoelectric effect was discovered by the Curie brothers in 1880 ^{[CurieCurie1880]} in tourmaline and quartz crystals, opening the era of electromechanical coupling and underpinning modern ultrasound transducers, MEMS resonators, and quartz oscillators.

The polar-dielectric correction is due to Debye 1929 ^[Debye1929] in Polar Molecules, who introduced the orientational polarizability ${\alpha }_o=p^2/(3k_{B}T)$ for permanent dipoles in thermal equilibrium. Debye's formula reproduced the temperature-dependence of polar gases but failed for polar liquids, where it predicted ferroelectric divergence well before any liquid exhibits one. Onsager 1936 ^{[Onsager1936]} in the J. Am. Chem. Soc. corrected this by replacing the Lorentz cavity by an Onsager cavity that included the reaction field of the polarized medium back on the central dipole, giving $(\kappa -1)(2\kappa +1)/(9\kappa )=n[\alpha +p^2/(3k_{B}T)]/(3{ϵ}_0)$. Kirkwood 1939 ^{[Kirkwood1939]} in the J. Chem. Phys. refined Onsager's treatment for short-range orientational correlations by introducing the correlation factor $g$. Devonshire 1949 ^{[Devonshire1949]} in the Phil. Mag. applied Landau's expansion of the free energy in the order parameter to ferroelectricity, treating barium titanate's Curie-temperature transition as a structural mean-field instability. Modern computational dielectric theory descends through the Marcus-Onsager response-field framework, density-functional methods (Kohn-Sham 1965, Born-Oppenheimer corrections, Berry-phase polarization theory of King-Smith and Vanderbilt 1993 Phys. Rev. B 47), and ab-initio molecular dynamics; the engineering side runs from capacitor design through high-$\kappa$ gate dielectrics (HfO${}_2$ in sub-nm CMOS), piezoelectric MEMS, electrocaloric coolers, and energy-harvesting transducers.

Bibliography Master

@article{Faraday1837,
  author    = {Faraday, Michael},
  title     = {Experimental Researches in Electricity. Eleventh Series. On Induction},
  journal   = {Philosophical Transactions of the Royal Society of London},
  volume    = {128},
  year      = {1837},
  pages     = {1--79},
}

@article{Mossotti1850,
  author    = {Mossotti, Ottaviano Fabrizio},
  title     = {Discussione analitica sull'influenza che l'azione di un mezzo dielettrico ha sulla distribuzione dell'elettricit{\`a} alla superficie di pi{\`u} corpi elettrici disseminati in esso},
  journal   = {Memorie di Matematica e di Fisica della Societ{\`a} Italiana delle Scienze, Modena},
  volume    = {24},
  year      = {1850},
  pages     = {49--74},
}

@book{Clausius1879,
  author    = {Clausius, Rudolf},
  title     = {Die Mechanische Behandlung der Electricit{\"a}t},
  publisher = {Vieweg, Braunschweig},
  year      = {1879},
}

@article{CurieCurie1880,
  author    = {Curie, Pierre and Curie, Jacques},
  title     = {D{\'e}veloppement par compression de l'{\'e}lectricit{\'e} polaire dans les cristaux h{\'e}mi{\`e}dres {\`a} faces inclin{\'e}es},
  journal   = {Bulletin de la Soci{\'e}t{\'e} Min{\'e}ralogique de France},
  volume    = {3},
  year      = {1880},
  pages     = {90--93},
}

@book{Debye1929,
  author    = {Debye, Peter},
  title     = {Polar Molecules},
  publisher = {Chemical Catalog Company},
  year      = {1929},
}

@article{Onsager1936,
  author    = {Onsager, Lars},
  title     = {Electric Moments of Molecules in Liquids},
  journal   = {Journal of the American Chemical Society},
  volume    = {58},
  year      = {1936},
  pages     = {1486--1493},
}

@article{Kirkwood1939,
  author    = {Kirkwood, John G.},
  title     = {The Dielectric Polarization of Polar Liquids},
  journal   = {The Journal of Chemical Physics},
  volume    = {7},
  year      = {1939},
  pages     = {911--919},
}

@article{Devonshire1949,
  author    = {Devonshire, A. F.},
  title     = {Theory of Barium Titanate: Part I},
  journal   = {Philosophical Magazine},
  volume    = {40},
  year      = {1949},
  pages     = {1040--1063},
}

@book{Jackson1999,
  author    = {Jackson, John David},
  title     = {Classical Electrodynamics},
  edition   = {3},
  publisher = {Wiley},
  year      = {1999},
}

@book{Griffiths2013,
  author    = {Griffiths, David J.},
  title     = {Introduction to Electrodynamics},
  edition   = {4},
  publisher = {Cambridge University Press},
  year      = {2013},
}

@book{LandauLifshitzVol8,
  author    = {Landau, Lev Davidovich and Lifshitz, Evgeny Mikhailovich},
  title     = {Electrodynamics of Continuous Media},
  series    = {Course of Theoretical Physics},
  volume    = {8},
  edition   = {2},
  publisher = {Pergamon},
  year      = {1984},
}

@article{KingSmithVanderbilt1993,
  author    = {King-Smith, R. D. and Vanderbilt, David},
  title     = {Theory of polarization of crystalline solids},
  journal   = {Physical Review B},
  volume    = {47},
  year      = {1993},
  pages     = {1651--1654},
}