The Method of Images

The problem it solves

Put a point charge +q near a grounded metal sheet and you have created a genuinely hard problem. The charge attracts the free electrons in the metal, they pile up on the near face, and that induced charge distribution — which you do not know in advance — produces a field of its own that adds to the field of q. To find the true field you would have to solve for the induced surface density σ(r) self-consistently, because σ depends on the total field and the total field depends on σ. This is a boundary-value problem for Poisson's equation with a nontrivial, unknown boundary charge.

The method of images sidesteps all of it. The insight: what physically enforces the answer is a single boundary condition — the conductor is an equipotential, and grounding pins it to V = 0. If we can find any arrangement of charges in the region outside the conductor that reproduces the same charge q there and makes V = 0 on the conductor surface, then by the uniqueness theorem that arrangement gives the correct field everywhere outside. So we throw away the conductor entirely and replace it with fictitious charges — the images — living in the region the conductor used to occupy.

Charge above a grounded plane

The canonical case: a charge +q at height d above an infinite grounded conducting plane occupying the z = 0 plane, with the charge on the axis at (0, 0, d). We want the field in the upper half-space z > 0.

The image: remove the plane and place a charge −q at the mirror-image point (0, 0, −d). Now check the old boundary: any point on the z = 0 plane is equidistant from +q and −q, so their potentials cancel exactly and V = 0 on the whole plane. The boundary condition is satisfied, the source charge +q in z > 0 is unchanged, so in the upper half-space the two-charge field is the answer.

The potential above the plane, with the charge at height d on the z-axis, is

V(x,y,z) = (1/4πε₀) [ q/√(x²+y²+(z−d)²) − q/√(x²+y²+(z+d)²) ]

where ε₀ = 8.854 × 10⁻¹² F/m is the permittivity of free space. The first term is the real charge; the second is the image. Set z = 0 and the two square roots are equal — V vanishes, as required. Below the plane (z < 0) the physical field is zero, because the conductor shields that region; the image potential there is fictitious and must be discarded.

Force and energy — and the factor-of-two trap

Because the field above the plane equals the field of the +q, −q pair, the force on the real charge equals the Coulomb force from its image. The two charges are separated by 2d (from +d to −d), so

F = − (1/4πε₀) · q² / (2d)²  ẑ  =  − q² / (16πε₀ d²)  ẑ

The minus sign means the force points toward the plane — the charge is always attracted, which is why a charge hovering over a metal surface is drawn in. Symbols: q is the charge (C), d the height above the plane (m), ε₀ the vacuum permittivity (F/m); F comes out in newtons.

The energy is the famous trap. You might expect the interaction energy to be the two-charge value −q²/(4πε₀·2d). It is not. The correct work to assemble this configuration — bringing the real charge from infinity to height d — is

W = − q² / (16πε₀ d)   (HALF the naive two-charge value)

The factor of ½ appears because the image is not an independent object. As you move the real charge, the image moves with it, and only the real charge's motion does work against the field. Integrating the true force F(d) from infinity to d gives the ½. Forgetting this is the single most common error with image problems.

Induced surface charge

The image trick even hands you the induced charge for free. The surface density on a conductor is σ = −ε₀ (∂V/∂n) evaluated just outside, where n is the outward normal. Differentiating the two-charge potential at z = 0 gives

σ(r) = − q d / [ 2π (r² + d²)^(3/2) ]

where r = √(x² + y²) is the distance from the foot of the perpendicular. Key features:

• Sign: negative everywhere — the plane draws opposite charge toward the +q, as expected.
• Peak: maximum magnitude directly under the charge, σ(0) = −q/(2πd²).
• Falloff: for r » d it decays as r⁻³, so the induced charge is concentrated in a spot of radius ~d.
• Total: integrating over the whole infinite plane, Q_ind = ∫σ dA = −q — exactly the value of the image. That is not a coincidence: the image charge is the collapsed representation of the entire induced distribution.

Why it is exact: the uniqueness theorem

The image is a guess. What upgrades it to a proof is the uniqueness theorem for Poisson's equation, ∇²V = −ρ/ε₀. The theorem says: in a region, the potential is fixed completely once you specify (a) the charge density ρ throughout the region and (b) the potential V on every bounding surface (a Dirichlet condition), or the normal derivative on each surface (Neumann). Any two solutions agreeing on both must be identical — their difference satisfies Laplace's equation with zero boundary values, and such a function is identically zero.

The image configuration matches both requirements in the physical region: the real charge q is the only real charge there, and V = 0 on the conductor boundary and at infinity. Therefore the image field cannot merely resemble the true field — it must be it, everywhere outside the conductor. This is the whole justification: you are allowed to invent charges outside the physical region as long as you do not touch ρ inside it and you reproduce the boundary values.

The grounded sphere (Kelvin's construction)

The plane is the easy case; Kelvin's real triumph in 1848 was the sphere. Take a point charge q a distance a from the center of a grounded conducting sphere of radius R, with a > R (charge outside). A single image charge does the job:

q′ = − q R / a        placed at   b = R² / a   (on the line from center to q, inside the sphere)

With this image the sphere surface becomes an equipotential at V = 0. Symbols: R is the sphere radius (m), a the charge's distance from center (m), b the image distance from center (m), q′ the image charge (C). The force on the real charge is again the Coulomb attraction to its image, giving

F = − (1/4πε₀) · q² R a / (a² − R²)²   (attractive)

If instead the sphere is isolated and neutral (not grounded), you superpose a second image: a charge +qR/a at the center, which restores the sphere's net charge to zero while keeping it an equipotential. If it carries a fixed charge Q, put Q − q′ at the center. This is how you handle a charged conducting sphere in an external field.

Plane vs sphere at a glance

Quantity	Grounded plane	Grounded sphere (radius R)
Source charge	+q at height d	+q at distance a from center (a > R)
Image charge	−q	−qR/a
Image location	−d (mirror point)	R²/a from center (inside)
Boundary made V = 0	z = 0 plane	sphere surface r = R
Force on charge	−q²/(16πε₀d²)	−q²Ra/[4πε₀(a²−R²)²]
Number of images	1	1 (grounded); 2 (neutral/charged)
Recovers plane as	—	R → ∞, a = R + d

Other geometries that work

Configuration	Image set
Charge above one grounded plane	1 image (−q at mirror point)
Charge between two parallel grounded planes	Infinite array of alternating ±q images (like facing mirrors)
Charge in a grounded corner (two planes at 90°)	3 images: −q, −q, +q (completes the reflections)
Charge in a wedge of angle π/n	2n − 1 images arranged around the vertex
Point charge near a grounded sphere	1 image −qR/a at R²/a
Point charge near a dielectric half-space (permittivity ε)	Image of strength −q(ε−ε₀)/(ε+ε₀) for the field in vacuum
Line charge near a grounded cylinder	Image line charge (2D analogue of the sphere)

The corner (two perpendicular grounded planes) is a nice sanity check: you need three images so that every point on both planes sits at V = 0. Fewer than three, and one plane fails; the +q image in the diagonally opposite quadrant is exactly what fixes the second plane.

Worked example

An electron sits d = 1.0 nm above a grounded gold surface. What is the attractive force?

Use F = q²/(16πε₀d²) with q = 1.602 × 10⁻¹⁹ C, d = 1.0 × 10⁻⁹ m, ε₀ = 8.854 × 10⁻¹² F/m:

F = (1.602e−19)² / (16π · 8.854e−12 · (1.0e−9)²)
  = 2.566e−38 / (4.449e−28)
  ≈ 5.8 × 10⁻¹¹ N  (≈ 58 pN), directed toward the surface

That is a substantial force at the nanoscale — the same physics governs the classical part of image-charge binding for electrons at metal surfaces, field emission, and scanning-tunneling geometries. (Quantum mechanically the image potential also produces bound “image states” a few tenths of an eV below the vacuum level, but the classical −q²/(16πε₀d) potential is the leading behavior far from the surface.)

JavaScript — image-charge calculators

const eps0 = 8.8541878128e-12; // F/m
const k    = 1 / (4 * Math.PI * eps0); // Coulomb constant ≈ 8.99e9

// Force on a charge q at height d above a grounded plane (attractive, magnitude)
function planeImageForce(q, d) {
  return q * q / (16 * Math.PI * eps0 * d * d);
}

// Work to assemble it (note the factor of 1/2 vs two real charges)
function planeImageEnergy(q, d) {
  return -q * q / (16 * Math.PI * eps0 * d);
}

// Induced surface charge density at radius r from the foot of the perpendicular
function inducedSigma(q, d, r) {
  return -q * d / (2 * Math.PI * Math.pow(r * r + d * d, 1.5));
}

const e = 1.602176634e-19, d = 1.0e-9;
console.log(`Force on e- at 1 nm: ${planeImageForce(e, d).toExponential(2)} N`); // ~5.77e-11
console.log(`Peak σ under charge: ${inducedSigma(e, d, 0).toExponential(2)} C/m²`);

// Grounded sphere: image charge and its location
function sphereImage(q, R, a) {
  return { qImage: -q * R / a, bDistance: (R * R) / a }; // a > R
}
console.log(sphereImage(1e-9, 0.05, 0.10)); // { qImage: -5e-10, bDistance: 0.025 }

// Force on a charge near a grounded sphere (attractive magnitude)
function sphereImageForce(q, R, a) {
  return k * q * q * R * a / Math.pow(a * a - R * R, 2);
}

Where the method of images shows up

• Electronics packaging. A signal trace over a ground plane is exactly the plane-image problem — the ground plane's return current mirrors the trace, setting the characteristic impedance of microstrip and stripline.
• Antennas. A monopole over a conducting ground plane behaves like a dipole because the ground provides the missing half via its image; a quarter-wave whip radiates like a half-wave dipole.
• Surface science. The classical image potential −e²/(16πε₀z) shapes the work function, field emission, and the ladder of Rydberg-like image states at metal surfaces.
• Colloids and electrochemistry. Ions near a charged electrode or membrane feel image forces that modify the double layer.
• Fluid dynamics (the exact analogue). A vortex or source near a wall is solved by an image vortex/source — the same reflection trick, because 2D potential flow obeys Laplace's equation too.
• Green's functions. The image solution is literally the Dirichlet Green's function for the half-space or sphere, written down by inspection instead of by summing eigenfunctions.

Common misconceptions

• Thinking the image charge is real. It is a mathematical fiction that lives in the region the conductor occupies. The field it “produces” there is not physical — inside/behind the conductor the true field is zero.
• Using the naive two-charge energy. The assembly energy is half the two-real-charge value: −q²/(16πε₀d), not −q²/(8πε₀d). The image moves with the charge, so only the real charge does work.
• Getting the separation wrong. The image sits at −d, so the charge–image distance is 2d, giving (2d)² = 4d² in the denominator. The height d and the separation 2d are not the same length.
• Applying it to a non-grounded, non-special shape. Only planes, spheres, cylinders, and wedges of angle π/n admit finite image sets. A cube or an ellipsoid does not — you need Green's functions or a numerical solver.
• Forgetting the second image for an isolated/charged sphere. A grounded sphere needs one image; an isolated neutral or fixed-charge sphere needs a second image at the center to fix the net charge.
• Confusing “grounded” with “isolated.” Grounded means V = 0 and the conductor can exchange charge with ground (net induced charge ≠ 0). Isolated means fixed total charge but a floating, nonzero potential. The image set differs.