Bound Charges: Surface and Volume

Here is the conceptual heart of the dielectrics chapter — the theorem that converts polarized matter back into something electrostatics already knows how to handle. A polarized object contains $10^{23}$ -odd molecular dipoles; computing their summed field dipole-by-dipole is unthinkable. The rescue is an exact bookkeeping identity: for the purpose of making fields, a polarization $\mathbf{P}(\mathbf{r})$ is completely equivalent to two ordinary charge densities — a surface layer $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ wherever the material ends, and a volume density $\rho_b = -\nabla\cdot\mathbf{P}$ wherever the polarization is non-uniform. No approximation, no far-field limit: replace the polarized object by these bound charges sitting in vacuum, and every macroscopic field, inside and out, comes out right. The physical picture deserves to be owned, not memorized. Polarization means chains of molecular dipoles, head to tail: deep in the bulk, each $+$ head sits against the next molecule's $-$ tail and cancels — a perfectly polarized interior is perfectly camouflaged. The camouflage fails in exactly two places. Where a chain terminates at a surface, its last head (or tail) stands exposed: that is $\sigma_b$ , largest where $\mathbf{P}$ punches straight out of the surface, zero where it runs parallel. And where the polarization varies, adjacent dipoles no longer match — the head shipped into a region is bigger or smaller than the tail shipped out — and the mismatch piles up as $\rho_b$ , with the minus sign recording that charge accumulates where the shipping converges. This article earns both densities by closing the transport ledger the previous article opened ( $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ = charge shipped across a patch), proves that the total bound charge of any object is exactly zero, and prosecutes the doctrine students resist most: bound charges are real — they make real fields, deflect real electrometers, and are "bound" only in the sense of being unavailable for conduction. The companion identity — deriving the same densities from the potential by integration by parts — belongs to Field of a Polarized Object, next.

Key takeaways

A polarized object is electrostatically equivalent to bound charges: $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ on its surfaces and $\rho_b = -\nabla\cdot\mathbf{P}$ in its bulk — exactly, for all field points.
The physical mechanism is chain bookkeeping: head-to-tail dipoles cancel in the bulk; exposed chain-ends make $\sigma_b$ ; mismatched neighbors (non-uniform $\mathbf{P}$ ) make $\rho_b$ .
The derivation is one step past the previous article: $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ is the charge shipped across a patch; close the surface, and what's left inside is $-\oint\mathbf{P}\cdot\mathrm{d}\mathbf{a} = \int(-\nabla\cdot\mathbf{P})\,\mathrm{d}\tau$ .
The minus sign is conservation: where shipping diverges (charge exported), a deficit remains; where it converges, a surplus.
Total bound charge is always zero — $\oint\sigma_b\,\mathrm{d}a + \int\rho_b\,\mathrm{d}\tau = 0$ identically — because polarization only rearranges charge.
Uniform $\mathbf{P}$ : all bound charge lives on the surfaces ( $\rho_b = 0$ ), the case that runs the Uniformly Polarized Sphere.
Bound charge is real charge: it sources $\mathbf{E}$ in Gauss's law exactly as free charge does; "bound" describes mobility, not reality.
KEY IDEA: $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ and $\rho_b = -\nabla\cdot\mathbf{P}$ — exposed chain-ends and shipping imbalances — replace polarized matter by ordinary charges in vacuum, exactly.

The core insight

When a dielectric is polarized, the net effect of all its molecular dipoles is captured exactly by two bound charge densities — $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ where dipole chains terminate at surfaces, and $\rho_b = -\nabla\cdot\mathbf{P}$ where non-uniform polarization makes neighboring dipoles mismatch — real charges that source fields like any others.

Where this lives today

Bound-charge engineering is quietly everywhere. Electret microphones — the slab of worked example 1, miniaturized — hold their $\sigma_b$ for decades and turn membrane vibrations into voltage in billions of devices. Piezoelectric sensors and energy harvesters run the logic dynamically: strain modulates $\mathbf{P}$ , the changing $\sigma_b$ drives charge through external circuits, and your gas-lighter's spark is a pulse of suddenly-exposed chain-ends. In reverse, ultrasound transducers push bound charge around megahertz-fast to image bodies. The most consequential modern arena is nitride electronics: GaN/AlGaN transistors (the fast chargers and radar amplifiers of the 2020s) exploit built-in polarization discontinuities — at the heterojunction, $\mathbf{P}$ jumps, and the interface bound charge $\sigma_b = (\mathbf{P}_1 - \mathbf{P}_2)\cdot\hat{\mathbf{n}}$ is so large it summons a two-dimensional electron gas with no doping at all: a conducting channel conjured from $\mathbf{P}\cdot\hat{\mathbf{n}}$ . The volume density has its own modern career. Flexoelectricity — polarization from strain gradients — makes $\rho_b = -\nabla\cdot\mathbf{P}$ a design variable in bent nanomembranes and is a suspect in how hearing works (bent hair-cell membranes polarizing). Graded ferroelectrics and compositionally graded films deliberately shape $\nabla\cdot\mathbf{P}$ to build internal fields — worked example 2 as device blueprint. And in every molecular-dynamics simulation of water at an interface, the measured "surface potential" is the direct descendant of this article's chain-end bookkeeping, computed molecule by molecule and still matching $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ when coarse-grained — the century-old ledger, validated one trajectory at a time.

Origins · 1 minSiméon Denis Poisson and how this was discovered

Siméon Denis Poisson

1781–1840 · École Polytechnique, Paris

Poisson owns the mathematics of this article twice over. His 1812 memoir on electrified conductors founded electrostatic potential theory (with the equation that bears his name), and his 1824–26 memoirs on magnetism did for magnetized matter precisely what this article does for polarized matter: he proved that a body with magnetization $\mathbf{M}$ produces, outside itself, the same field as an equivalent surface distribution plus an equivalent volume distribution — $\mathbf{M}\cdot\hat{\mathbf{n}}$ and $-\nabla\cdot\mathbf{M}$ in modern dress, derived with the same integration-by-parts and divergence-theorem machinery, decades before anyone could say what a molecular dipole was. The construction crossed into electrostatics intact when Faraday's dielectrics demanded it; Kelvin and Maxwell both built directly on "Poisson's equivalent distributions." A mathematician of legendary output (the Poisson distribution, Poisson brackets, elasticity's Poisson ratio), he was famously wrong exactly once in a useful way: his "absurd" bright-spot prediction, meant to refute Fresnel's wave optics, was confirmed and sealed the wave theory's victory — the equivalent-distribution theorem, meanwhile, has never needed correcting.

History

1812

— Poisson's electrostatics memoir founds potential theory for charged conductors

1824

— Poisson's magnetism memoirs prove the equivalent surface-plus-volume distribution theorem — sigma and rho for magnetized bodies

1837

— Faraday's dielectric experiments supply the polarized-matter physics the theorem was waiting for

1873

— Maxwell's Treatise systematizes bound charge within the full field theory of media

1999

— Polarization-induced 2D electron gases in GaN heterostructures turn interface bound charge into transistor channels

The physical picture

Build a polarized slab out of dipole chains and watch the cancellation do its work. Each molecule is a $-$ tail and a $+$ head; polarization lines them up head-to-tail like beads on strings threading the slab. Pick any interior molecule: its $+$ head nestles against the $-$ tail of the next molecule up the chain, charge for charge. An interior probe (averaging over the mesoscopic cell of the previous article) sees nothing — a uniformly polarized bulk is electrically invisible. All the drama is at the ends of the strings: at the top face a rank of $+$ heads stands exposed with no tails above to cancel them; at the bottom, orphaned $-$ tails. Two sheets of genuinely unbalanced charge, from nothing but honest pairing.

Figure

Dipole chains — bulk camouflage, exposed ends

Now un-uniform the polarization and watch the bulk lose its camouflage. Suppose the chains strengthen from left to right — molecules on the right displace more charge than their neighbors on the left. Draw a small box in the interior and audit it with the previous article's ledger: the charge shipped out through the right face, $P(\text{right})\,\mathrm{d}a$ , exceeds the charge shipped in through the left face, $P(\text{left})\,\mathrm{d}a$ . The box has been drained: a net negative charge remains where the shipping diverges. That is the whole content of $\rho_b = -\nabla\cdot\mathbf{P}$ , minus sign included — divergence measures net export, and what stays behind is the export's negative. No new physics entered after the transport identity; the volume charge is the closed-ledger corollary of " $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ = shipped charge."

Figure

The shipping audit — where div P is positive, charge is exported and a deficit remains

Before the formal derivation, fix the doctrine, because everything downstream depends on it: bound charges are real. The $+$ heads at the slab's top face are actual unbalanced molecular charge — about $625$ elementary charges per square micrometer at $\sigma_b = 10^{-4}\,\mathrm{C/m^2}$ . They produce fields by Coulomb's law, terminate field lines, push electrometers, and enter Gauss's law for $\mathbf{E}$ on exactly the same footing as charge sprayed on by a Van de Graaff. What they cannot do is leave: each is the end of a molecule that holds it, so they carry no current and can't be wiped off with a wire. "Bound" is an employment status, not a metaphysical discount — and the single most common dielectrics error, using Gauss's law for $\mathbf{E}$ with free charge only, is at bottom a refusal to take bound charge seriously.

The two densities

\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}, \qquad \rho_b = -\nabla\cdot\mathbf{P} \qquad (1)

where: $\hat{\mathbf{n}}$ — the outward unit normal of the dielectric's surface; $\sigma_b$ — bound charge per area where the material ends; $\rho_b$ — bound charge per volume in the bulk. Read the geometry in $\sigma_b$ : polarization punching perpendicularly out of a face deposits its full magnitude ( $\sigma_b = P$ ); running parallel to a face it deposits nothing ( $\mathbf{P}\cdot\hat{\mathbf{n}} = 0$ — chains skim the surface without terminating); the general tilt interpolates as the cosine. And read the sign in $\rho_b$ : charge remains where shipping converges — a polarization field pointing inward from all sides ( $\nabla\cdot\mathbf{P} < 0$ ) crowds positive heads into the middle. The bookkeeping closes globally:

Q_{b,\rm total} = \oint_{\mathcal{S}}\sigma_b\,\mathrm{d}a + \int_{\mathcal{V}}\rho_b\,\mathrm{d}\tau = 0 \qquad (2)

where: the integrals run over the object's full surface and volume. Polarization rearranges charge; it cannot mint any. Equation (2) is both a theorem (one line of the divergence theorem, in Derivation B) and the best error-detector in the chapter: any computed bound-charge inventory that does not total zero is wrong, no exceptions.

Derivation AClosing the ledger — surface and volume bound charge from the transport identity

The goal is both densities of (1) from a single principle: the previous article's transport identity, $\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ = charge shipped across a patch as the material polarized. The strategy is charge conservation applied to closed regions — first a box in the bulk, then a pillbox at the surface. This is the charge-counting route, chosen deliberately: it explains where the charges physically are and why the signs fall as they do. (The slicker route — extracting the same densities from the potential by integration by parts — is owned by Field of a Polarized Object, where it also proves the field-equivalence claim; the two routes landing on one answer is the chapter's consistency check.)

Step 1. The bulk. Take any volume $\mathcal{V}$ wholly inside the dielectric, boundary $\mathcal{S}$ . Before polarization the region was neutral. During polarization, the charge shipped out through $\mathcal{S}$ is $\oint_{\mathcal{S}}\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ (the transport identity, patch by patch). By conservation, the charge now residing in $\mathcal{V}$ is the negative of the export:

Q_b(\mathcal{V}) = -\oint_{\mathcal{S}}\mathbf{P}\cdot\mathrm{d}\mathbf{a} = -\int_{\mathcal{V}}\nabla\cdot\mathbf{P}\,\mathrm{d}\tau,

by the divergence theorem (E0's fundamental machinery). Since $\mathcal{V}$ was arbitrary, the integrands match pointwise:

\rho_b = -\nabla\cdot\mathbf{P}.

Why the localization step is legal: the standard shrinking-volume argument — if two continuous densities integrate equally over every region, they are equal everywhere — the same move that localized Gauss's law in E1.

Step 2. The surface. Where the material ends, $\mathbf{P}$ drops discontinuously to zero and $\nabla\cdot\mathbf{P}$ spikes — the signature, familiar from the conductor chapters, of a surface density hiding in a derivative. Resolve it with a thin pillbox straddling the boundary: faces of area $\mathrm{d}a$ parallel to the surface, negligible rim. The shipped-charge export through the pillbox has only two contributions: through the inner face (inside the material), $-\mathbf{P}\cdot\hat{\mathbf{n}}\,\mathrm{d}a$ (inward normal of the box is $-\hat{\mathbf{n}}$ ), and through the outer face, zero ( $\mathbf{P} = 0$ outside). Net charge accumulated in the box:

\mathrm{d}Q_b = -\left[0 - \mathbf{P}\cdot\hat{\mathbf{n}}\,\mathrm{d}a\right] = \mathbf{P}\cdot\hat{\mathbf{n}}\,\mathrm{d}a \qquad\Longrightarrow\qquad \sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}.

Why this is the chain-end picture in equations: the inner face intercepts the last link of every chain crossing it; the charge those chains shipped into the box has no outer face to leave through, so it stays — the exposed heads. Tilt the surface and only the perpendicular component of the shipping lands charge: the cosine in $\mathbf{P}\cdot\hat{\mathbf{n}}$ , geometrically inevitable.

Step 3. Uniform-polarization check. $\mathbf{P}$ constant: $\rho_b = 0$ (every interior box balances its shipping), and the entire effect is $\sigma_b = \pm P\cos\theta$ on the surfaces — the slab's two sheets, and, wrapped on a sphere, the $\sigma_b = P\cos\theta$ distribution whose beautiful consequences the Uniformly Polarized Sphere harvests.

Step 4. Where each density actually shows up. $\sigma_b$ : every polarized object with a boundary — which is all of them. $\rho_b$ : rarer in practice, needing genuinely non-uniform $\mathbf{P}$ — graded materials, thermal gradients, fields of non-trivial geometry (a linear dielectric in a point charge's field develops $\rho_b$ only where free charge sits, a slick result of the later Gauss-in-dielectrics machinery), and radially polarized devices like worked example 2's sphere. Sign check: polarization pointing toward a face gives positive $\sigma_b$ there, consistent with the slab picture. The uniform- $\mathbf{P}$ specialization reduces correctly to that same slab. Both densities are re-derived independently by the potential-identity route in Field of a Polarized Object — the two routes agree.

Go deeper · Derivation B · 1 minThe zero-sum theorem — and why bound charges are sources like any other

Derivation BThe zero-sum theorem — and why bound charges are sources like any other

The goal is two structural facts that discipline every dielectric calculation: total bound charge vanishes identically, and bound charge sources the electric field on equal terms with free charge. The strategy for the first is one application of the divergence theorem; for the second, an argument about what the mesoscopic average preserves. Both are short; both are load-bearing.

Step 1. The zero-sum. Integrate the two densities over the whole object (volume $\mathcal{V}$ , surface $\mathcal{S}$ ):

Q_{b,\rm total} = \oint_{\mathcal{S}}\mathbf{P}\cdot\hat{\mathbf{n}}\,\mathrm{d}a + \int_{\mathcal{V}}(-\nabla\cdot\mathbf{P})\,\mathrm{d}\tau = \oint_{\mathcal{S}}\mathbf{P}\cdot\mathrm{d}\mathbf{a} - \oint_{\mathcal{S}}\mathbf{P}\cdot\mathrm{d}\mathbf{a} = 0.

The two terms are the same integral with opposite signs — the theorem is the divergence theorem read twice. Physically: every chain has exactly one head and one tail; exposing ends and pooling mismatches only relocates the census, never changes it.

Step 2. The equal-footing claim. The macroscopic field is the mesoscopic average of the true microscopic field, and averaging is linear: it cannot know which microscopic charges were "free" and which sat at molecular chain-ends. Gauss's law for the averaged field therefore reads

\nabla\cdot\mathbf{E} = \frac{\rho_{\rm total}}{\varepsilon_0} = \frac{\rho_f + \rho_b}{\varepsilon_0},

with $\rho_b$ entering identically to $\rho_f$ . There is one electric field, sourced by all charge. The free/bound split is our ledger — profoundly convenient (it motivates the displacement field two articles hence) but invisible to $\mathbf{E}$ itself.

Step 3. The operational consequences. (i) Any Gauss's-law calculation for $\mathbf{E}$ in a dielectric problem must enclose bound charge too — forgetting it is the chapter's classic error. (ii) The equivalence is complete for fields: replace the polarized object by $(\sigma_b, \rho_b)$ in vacuum and compute with E1 methods — the license the next three articles cash. (iii) Zero-sum is the universal audit: worked example 2's inventory will total zero to the last coulomb, and so must yours. As a check: the zero-sum holds explicitly for the uniformly polarized slab (sheets $+PA$ and $-PA$ ) and for the radial sphere of worked example 2 ( $-5.0\times10^{-7}$ C volume charge, $+5.0\times10^{-7}$ C surface charge, summing to zero). The equal-footing statement — that bound charge sources $\mathbf{E}$ identically to free charge — follows directly from the linearity of the mesoscopic averaging.

Limiting cases

Case	Result	Meaning
uniform $\mathbf{P}$	$\rho_b = 0$ , $\sigma_b = \mathbf{P}\cdot\hat{\mathbf{n}}$ only	all action at the surfaces — slab, sphere classics
$\mathbf{P}$ parallel to a face	$\sigma_b = 0$ there	chains skim, never terminate: side walls of the slab
tilted face, angle $\theta$	$\sigma_b = P\cos\theta$	cosine projection of the shipping
converging $\mathbf{P}$ ( $\nabla\cdot\mathbf{P} < 0$ )	$\rho_b > 0$	heads crowd inward: positive pool
whole object	$Q_b = 0$ exactly	rearrangement, never creation — the audit
conductor comparison	induced $\sigma$ free, unbounded supply	dielectric $\sigma_b$ capped at $P$ : partial screening
$\mathbf{P} \to 0$	both densities vanish	unpolarized matter is charge-silent

Common trapBound charge is not a mathematical fiction — and E does not get to ignore it

The two-sided trap. Side one: treating $(\sigma_b, \rho_b)$ as formal devices "not really there." They are unbalanced molecular charge, measurable and field-producing; an electrometer at a polarized face reads them, and the Uniformly Polarized Sphere's entire interior field is made by nothing else. Side two, the operational version: applying $\nabla\cdot\mathbf{E} = \rho_f/\varepsilon_0$ — Gauss's law with only the charge you put there by hand. False in any dielectric: the honest law is $\nabla\cdot\mathbf{E} = (\rho_f + \rho_b)/\varepsilon_0$ , and omitting $\rho_b$ silently is the single most common error in this chapter's problem sets. (The legitimate way to bookkeep bound charge out of sight — defining $\mathbf{D}$ with $\nabla\cdot\mathbf{D} = \rho_f$ — costs the field its curl-free guarantee, a price tallied honestly two articles from now.)

Worked examples

1 · A uniformly polarized cylinder, faces and flanks (confidence builder)

A polystyrene cylinder, radius $R = 1.0\,\mathrm{cm}$ , length $10\,\mathrm{cm}$ , carries frozen-in uniform polarization $P = 1.0\times10^{-4}\,\mathrm{C/m^2}$ along its axis (a bar electret). Find all bound charges; then the charge on a face cut at $30°$ to the cross-section.

Step 1. Volume charge: $\mathbf{P}$ uniform $\Rightarrow \rho_b = -\nabla\cdot\mathbf{P} = 0$ . The bulk is camouflaged.

Step 2. The flat ends. Top face (normal along $\mathbf{P}$ ): $\sigma_b = +P = +1.0\times10^{-4}\,\mathrm{C/m^2}$ ; bottom face: $-P$ . Total per face: $Q = P\pi R^2 = (10^{-4})\pi(0.01)^2 = \pm31\,\mathrm{nC}$ — in exposed molecular chain-ends, about $625$ electrons' worth per square micrometer.

Step 3. The curved flank: $\mathbf{P}\cdot\hat{\mathbf{n}} = 0$ (polarization parallel to the surface) — zero bound charge on the sides, however long the cylinder. The chains run along the flank and terminate only at the ends.

Step 4. The oblique cut. A face tilted $30°$ from the cross-section has its normal $30°$ from $\mathbf{P}$ : $\sigma_b = P\cos30° = 8.7\times10^{-5}\,\mathrm{C/m^2}$ — but the tilted face's area is larger by $1/\cos30°$ , so the total face charge is still $P\pi R^2 = 31\,\mathrm{nC}$ . The cosine in the density and the secant in the area cancel: total chain-ends care only about how many chains the face intercepts. Audit: $+31 - 31 + 0 = 0$ ✓.

2 · A radially polarized sphere — volume charge in action (exam level)

A sphere of radius $R = 2.0\,\mathrm{cm}$ carries $\mathbf{P} = k\,r\,\hat{\mathbf{r}}$ with $k = 5.0\times10^{-3}\,\mathrm{C/m^3}$ (polarization growing linearly outward — a model graded electret). Find $\rho_b$ , $\sigma_b$ , verify the zero-sum, and get the interior field for free.

Step 1. Volume density, via the spherical divergence (E0 catalog): $\rho_b = -\nabla\cdot\mathbf{P} = -\frac{1}{r^2}\frac{\partial}{\partial r}\left(r^2\cdot kr\right) = -\frac{1}{r^2}(3kr^2) = -3k = -1.5\times10^{-2}\,\mathrm{C/m^3}$ , uniform and negative: shipping diverges everywhere, so every interior box is drained.

Step 2. Surface density: $\sigma_b = \mathbf{P}(R)\cdot\hat{\mathbf{r}} = kR = 1.0\times10^{-4}\,\mathrm{C/m^2}$ , positive: the exported heads pile up at the boundary.

Step 3. The audit. Volume: $\rho_b\cdot\tfrac43\pi R^3 = (-3k)\tfrac43\pi R^3 = -4\pi kR^3 = -5.0\times10^{-7}\,\mathrm{C}$ . Surface: $\sigma_b\cdot4\pi R^2 = kR\cdot4\pi R^2 = +4\pi kR^3 = +5.0\times10^{-7}\,\mathrm{C}$ . Sum: zero, exactly and symbolically — the ledger closes.

Step 4. The free field. Bound charge is real charge, so Gauss's law (E1, symmetric case) applies directly: for $r < R$ , $E(r)\,4\pi r^2 = \rho_b\cdot\tfrac43\pi r^3/\varepsilon_0$ , giving

E(r) = \frac{\rho_b\,r}{3\varepsilon_0} = -\frac{k\,r}{\varepsilon_0},

i.e. $\mathbf{E} = -\mathbf{P}/\varepsilon_0$ pointing inward, $5.6\times10^{8}\,\mathrm{V/m}$ per meter of radius. Outside: the total enclosed charge is zero, so $\mathbf{E} = 0$ identically — this object is electrostatically invisible from outside. Bound charges computed, then fed straight into vacuum electrostatics: the equivalence theorem in one worked motion.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §4.2 · Jackson, Classical Electrodynamics, 3rd ed., §4.3 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §10.7–10.8 · Zangwill, Modern Electrodynamics, §6.3. The value $\varepsilon_0=8.854\times10^{-12}$ F/m is the CODATA standard. The integration-by-parts route to the same densities is executed in Field of a Polarized Object; this article derives both densities by the charge-counting route, and the two are cross-checked. Worked numbers: cylinder — face charge $P\pi R^2 = 31$ nC, tilted-face density $8.7\times10^{-5}$ C/m² with invariant total; radial sphere — $\rho_b = -3k = -1.5\times10^{-2}$ C/m³, $\sigma_b = kR = 1.0\times10^{-4}$ C/m², totals $\mp5.0\times10^{-7}$ C summing to zero, interior $\mathbf{E} = -\mathbf{P}/\varepsilon_0$ , exterior field zero.

Builds on: Polarization Density P · Permanent Dipoles and Orientational Polarization · Dielectric Materials and Insulators · Gauss's Law in Differential Form

Leads to: Field of a Polarized Object · Uniformly Polarized Sphere · The Electric Displacement D · Gauss's Law in Dielectrics

If the bulk of a uniformly polarized object is neutral, why do we say the whole object is polarized?

Because neutrality and polarization are different questions. Every mesoscopic cell in the bulk has zero net charge but nonzero dipole moment — the charge is arranged, not absent. The arrangement is exactly what $\mathbf{P}$ records and what the surface sheets betray. An analogy: a battalion marching in step has zero average position change relative to its center, but you can still tell which way it faces. Bound charge is where the facing shows — the ends of the formation.

Can I measure sigma_b directly, or only infer it from fields?

Both, and the direct measurements are classics. Scrape or cleave a polarized electret and collect the charge; press a grounded probe against the face and integrate the induced current (the "thermally stimulated discharge" method heats an electret and literally collects its chain-ends as they unlock); Kelvin probes read the surface potential step $\sigma_b$ produces. In piezoelectrics, squeeze and measure the released charge — lighter sparks are the demonstration. What you cannot do is conduct $\sigma_b$ away while the polarization stands: touch a wire to the face and essentially nothing flows, the signature distinguishing bound from free charge in the lab.

Why does the volume density involve the divergence and not the gradient or curl of P?

Because the question "how much charge stays in a region?" is a flux-budget question, and flux budgets are what divergence is. The charge shipped out of a closed region is $\oint\mathbf{P}\cdot\mathrm{d}\mathbf{a}$ , and the divergence theorem converts precisely that closed-surface flux into $\int\nabla\cdot\mathbf{P}\,\mathrm{d}\tau$ . The gradient of $\mathbf{P}$ (a tensor) contains more information than the budget needs; the curl of $\mathbf{P}$ — circulation, not flux — moves no net charge across closed surfaces and hence makes no bound charge at all. (It is not physically idle, though: $\nabla\times\mathbf{P} \neq 0$ is exactly what obstructs the displacement field two articles from now.)

Do the bound charges of a dielectric in an applied field cancel the field inside, like a conductor's induced charges?

They oppose it but cannot finish the job. A conductor has an unbounded reservoir of mobile charge: induced $\sigma$ grows until the interior field is exactly zero. A dielectric's $\sigma_b$ is capped — it can never exceed $P$ , and $P$ is limited by how far molecules stretch or how fully they align. The interior field is reduced to $E_0/\kappa$ , the partial screening quantified in the chapter opener; the conductor is the $\kappa\to\infty$ limit where the cap disappears. Same physics, different budget — and the reason "dielectric" and "conductor" bracket a single continuum.

When does rho_b actually appear in real problems, rather than just sigma_b?

Less often than you'd guess, which is itself worth knowing. Uniform polarization gives none. More surprisingly, a homogeneous linear dielectric in any electrostatic situation develops $\rho_b$ only in proportion to the free charge density embedded in it (a clean result of the Gauss-in-dielectrics machinery) — so a charge-free linear slab, sphere, or shim has $\rho_b = 0$ everywhere no matter how non-uniform the field. Genuine volume charge needs inhomogeneity: graded composition (example 2), temperature gradients, radiation-damaged electrets, flexoelectric strain gradients — or nonlinear response. Exam problems that hand you $\mathbf{P}(\mathbf{r})$ explicitly are usually testing exactly this machinery.

The zero-sum theorem says total bound charge vanishes — but the two faces of a slab attract each other's charges. Isn't something unbalanced?

Nothing needs to be. Zero total charge is compatible with any amount of charge separation, and separation is what makes fields and forces — the slab's two sheets form a capacitor-like pair with a real interior field $\sigma_b/\varepsilon_0$ pointing against $\mathbf{P}$ (the "depolarizing field," the sphere version of which is the next-but-one article's punchline). The theorem's power is as an audit, not a physics-neutralizer: if your computed inventory doesn't total zero you've made an error, but a zero total still leaves all the electrostatics to do.

How do bound charges relate to the image charges of the conductor chapters — both are "induced," aren't they?

They share billing as response charge but differ in kind. Image charges are fictions — mathematical stand-ins, placed outside the physical region, that happen to reproduce the field of the conductor's real (free, mobile) surface charge. Bound charges are not stand-ins: they are the actual charge distribution present in the material, merely computed from $\mathbf{P}$ rather than posited. You can build an image-style solution for a dielectric problem (the half-space dielectric has a famous image construction), and then the image charge is again a fiction — one that summarizes the field of the perfectly real $\sigma_b$ at the interface. Keep the categories straight: real-but-immobile (bound) versus real-and-mobile (free) versus fictional-but-useful (image).