Work and Energy of a Charge Configuration

Take four charges pinned at the corners of a square and ask a deceptively simple question: how much energy is locked in that arrangement? The answer matters far beyond the textbook — it is the binding energy of an ionic crystal, the energy budget of a protein folding against its own charged residues, the joules a capacitor will deliver when you close the switch. Electrostatics gives the question a precise operational meaning: the energy of a configuration is the work an external agent must do, moving quasi-statically, to assemble it from charges dispersed at infinite separation, where they do not interact. Because the electrostatic field is conservative — the curl vanished two articles ago — this work does not depend on the assembly route. Bring the charges in alphabetically, in reverse, or along spiral paths: the bill is identical. Assembly work is therefore a genuine state function of the final arrangement, and it can be written down without ever mentioning the history: a sum over pairs, $q_iq_j/4\pi\varepsilon_0r_{ij}$ , or more slyly as $\tfrac12\sum q_iV(\mathbf{r}_i)$ , where the factor of one-half quietly corrects a double count. The chapter's most consequential move comes at the continuous limit, $W=\tfrac12\int\rho V\,\mathrm{d}\tau$ — and with it a subtlety with a long afterlife: the continuous formula includes each charge's interaction with itself, while the discrete sum deliberately excludes it. Keeping track of which formula books what prevents real errors now, and sets up the question the next article answers: where, physically, does all this energy live?

Key takeaways

Configuration energy is defined operationally: $W$ = work to assemble the charges quasi-statically from infinite separation; path-independence (curl-free field) makes it a state function.
For point charges, $W=\frac{1}{4\pi\varepsilon_0}\sum_{i<j}\frac{q_iq_j}{r_{ij}}$ — one term per pair, self-interactions excluded by construction.
The symmetric rewrite $W=\tfrac12\sum_i q_iV(\mathbf{r}_i)$ counts each pair twice and halves the total; $V(\mathbf{r}_i)$ is the potential of all the other charges.
The continuous limit is $W=\tfrac12\int\rho V\,\mathrm{d}\tau$ — but here $V$ is the total potential, and the formula silently includes (finite, for smooth $\rho$ ) self-energy.
Signs carry physics: $W<0$ means the configuration is bound — energy must be supplied to take it apart; $W>0$ means it is spring-loaded.
Energy is quadratic in charge: double every charge and $W$ quadruples; energies of two configurations do not simply add.
Key idea: the energy of a charge configuration is the path-independent assembly work, bookable as a pair sum for point charges or as one-half the integral of ρV for continuous distributions — with self-energy the fine print separating the two.

The core insight

The energy of a charge configuration is the path-independent work of assembling it from infinity: $W=\frac{1}{4\pi\varepsilon_0}\sum_{i<j}\frac{q_iq_j}{r_{ij}}=\tfrac12\sum_iq_iV(\mathbf{r}_i)$ , becoming $\tfrac12\int\rho V\,\mathrm{d}\tau$ in the continuum — where the innocent-looking $\tfrac12$ fixes double counting and the continuum form quietly includes self-energy.

Where this lives today

Every molecular-dynamics simulation on the planet spends most of its arithmetic evaluating this article's equation (3). Simulating a solvated protein means summing pairwise Coulomb energies over hundreds of thousands of partial charges, every femtosecond step; the naive $N^2$ pair sum is hopeless, so production codes (GROMACS, AMBER, OpenMM) use particle-mesh Ewald — a fast-Fourier split of the Madelung-style lattice sum from the second worked example — to book the same $\tfrac12\sum q_iV$ in $N\log N$ time on GPUs. Drug-binding affinities, membrane electrostatics, the folding funnel of a protein: all are configuration-energy differences, computed at industrial scale with the sign conventions you just learned. The engineering face is the energy method: once $W$ is known as a function of a geometry parameter, its derivative is a force — how MEMS designers compute the pull-in of electrostatic actuators and comb drives in the accelerometer that just rotated your phone screen. And at the frontier of precision, trapped-ion quantum computers arrange their qubits as literal charge configurations: the ions' equilibrium spacings and normal modes — the very "phonon bus" that entangles them — come from minimizing exactly the pair-sum energy of mutually repelling charges in a confining potential.

The physical picture

Build the pile one pebble at a time. The first charge rides in free: space is empty, there is no field to fight, no work to do. The second charge must be pushed against (or eased along) the first one's field, and the work is exactly $q_2$ times the potential the first charge maintains at the destination. The third fights both residents; the fourth fights three. Every joule you spend is banked — release the charges and the field pays it all back as kinetic energy. That running total, tallied pair by pair as each newcomer meets each resident exactly once, is the configuration energy.

Figure

Assembly one charge at a time — the first is free, each newcomer pays against all residents

Now the accounting subtlety that decides the shape of every formula in this article. When the ledger is complete, each pair $(i,j)$ has been billed exactly once — the newcomer paid, the resident did not. But suppose instead you go around the finished configuration and ask every charge "what is the potential energy $q_iV(\mathbf{r}_i)$ of your position?" Each pair now gets reported twice — once from $i$ 's viewpoint, once from $j$ 's. The sum $\sum_iq_iV(\mathbf{r}_i)$ is therefore exactly double the assembly cost, and the honest total carries a factor of one-half. That innocent $\tfrac12$ is not decoration; it is the difference between an assembly ledger and a census of the finished state. One more picture to carry forward: energy is quadratic. The pair bill scales as charge times charge, so doubling every charge quadruples the energy, and the energy of two configurations placed together is not the sum of their separate energies — the pairs that straddle the two contribute an interaction term all their own. Linearity, the superpowered tool of fields and potentials, quietly abandons you at the energy level.

From the assembly ledger to the three formulas

The work to carry a single test charge $q$ between two points $a$ and $b$ , through anyone's field, was settled by The Electrostatic Potential:

W_{a\to b}=q\big[V(\mathbf{b})-V(\mathbf{a})\big] \qquad (1)

where: $V$ is the potential of all the other charges — the mover's own field does no net work on itself along the trip. Equation (1) is the atom of everything below: assembly is just (1) applied over and over, with the potential updating as residents accumulate. Bringing charge $q_i$ in from infinity (where $V=0$ for localized sources) to position $\mathbf{r}_i$ costs $q_iV_{\text{residents}}(\mathbf{r}_i)$ , and summing the tolls pair by pair gives the master formula for point charges:

W=\frac{1}{4\pi\varepsilon_0}\sum_{i=1}^{n}\sum_{j>i}\frac{q_iq_j}{r_{ij}} \qquad (2)

where: $r_{ij}=|\mathbf{r}_i-\mathbf{r}_j|$ and the restriction $j>i$ bills each pair exactly once — and, crucially, never bills a charge against itself. Signs ride along automatically: like-sign pairs contribute positive work (you pushed uphill), unlike pairs negative (the field pulled the newcomer in). A bound structure — an ionic crystal, an atom — has $W<0$ : you would have to pay to disperse it. The symmetric repackaging turns the pair list into a census:

W=\frac{1}{2}\sum_{i=1}^{n}q_i\,V(\mathbf{r}_i) \qquad (3)

where: $V(\mathbf{r}_i)$ is the potential at charge $i$ due to all the others, and the $\tfrac12$ undoes the double count described above. Equation (3) looks like a mere rewrite of (2) — it is — but it is the form that survives the continuum limit. Smear the charges into a density $\rho$ and the census becomes an integral:

W=\frac{1}{2}\int\rho\,V\,\mathrm{d}\tau \qquad (4)

where: $V$ is now the total potential of the distribution itself, computed as in Potential of a Localized Charge Distribution, and the integral runs over the charge (extending it to all space is free — the integrand vanishes where $\rho=0$ ). Read the fine print: in passing from (3) to (4) the phrase "all the others" silently disappeared. Each element $\rho\,\mathrm{d}\tau$ now feels the potential of the whole distribution, its own neighborhood included — equation (4) contains self-energy. For any smooth distribution the extra contribution is finite and harmless (the derivation below shows it vanishes cell by cell), but for point charges it is infinite, which is why (2) and (4) are different formulas, not two spellings of one. The full resolution of that tension — and the beautiful relocation of $W$ into the field itself — is the opening move of Energy Density of the Electric Field.

Which energy formula, when

situation	use	watch for
one test charge moved in a fixed field	$W=q\,\Delta V$ , equation (1)	sign: work done by you against the field
a handful of point charges	pair sum, equation (2)	every pair once; no self-terms exist
many charges, or symmetry helps	$\tfrac12\sum q_iV(\mathbf{r}_i)$ , equation (3)	$V$ excludes the charge it multiplies
continuous $\rho$	$\tfrac12\int\rho V\,\mathrm{d}\tau$ , equation (4)	$V$ is total; self-energy now included
want the energy's location	$\tfrac{\varepsilon_0}{2}\int E^2\,\mathrm{d}\tau$	the next article's subject — derived there, not here

Derivation AAssembling n charges and the origin of the one-half

The goal is the pair of point-charge formulas (2) and (3), built directly from the assembly definition of energy, with the factor of one-half earned honestly rather than asserted. The strategy is bookkeeping made visible: bring the charges in one at a time, record what each newcomer pays, and then re-sort the same ledger two ways. Watch how "each pair once" (assembly view) and "each charge's full bill, halved" (census view) are the identical number wearing different clothes.

Step 1. Charge 1 arrives in empty space: $W_1=0$ . Charge 2 pays against charge 1's potential: $W_2=q_2\cdot\frac{1}{4\pi\varepsilon_0}\frac{q_1}{r_{12}}$ . Charge 3 pays both residents: $W_3=\frac{q_3}{4\pi\varepsilon_0}\Big(\frac{q_1}{r_{13}}+\frac{q_2}{r_{23}}\Big)$ . Why the residents pay nothing: they are held fixed while the newcomer moves; only the newcomer's displacement threads work through equation (1).

Step 2. The pattern is clear: charge $k$ pays $W_k=\frac{q_k}{4\pi\varepsilon_0}\sum_{j<k}\frac{q_j}{r_{jk}}$ . Summing over all arrivals,

W=\sum_{k=2}^{n}W_k=\frac{1}{4\pi\varepsilon_0}\sum_{k}\sum_{j<k}\frac{q_jq_k}{r_{jk}}

which is exactly the pair sum (2): every unordered pair $(j,k)$ appears once, when its later member arrived. Why the order does not matter: each term depends only on the pair's final separation — conservativity in action — so any assembly sequence produces the same set of pair terms.

Step 3. Re-sort into the census. Write the pair sum symmetrically by summing over ordered pairs and halving: $\sum_{j<k}=\tfrac12\sum_{j\neq k}$ . Then

W=\frac{1}{2}\cdot\frac{1}{4\pi\varepsilon_0}\sum_{i}\sum_{j\neq i}\frac{q_iq_j}{r_{ij}}=\frac{1}{2}\sum_i q_i\Bigg(\frac{1}{4\pi\varepsilon_0}\sum_{j\neq i}\frac{q_j}{r_{ij}}\Bigg)=\frac{1}{2}\sum_iq_iV(\mathbf{r}_i)

The parenthesis is precisely the potential at $\mathbf{r}_i$ of all the other charges — equation (3). The $\tfrac12$ is now visibly the correction for counting each pair from both ends, nothing more mysterious.

Step 4. Guard the boundary of validity: the manipulation used $j\neq i$ throughout. Writing $\tfrac12\sum_{i,j}$ without the exclusion would add terms $q_i^2/r_{ii}$ — division by zero, the infinite self-energy of a point charge. The discrete formulas are self-energy-free by construction, not by accident. As a quick cross-check: for three equal charges $q$ at mutual distance $a$ , Step 2 gives $W=3\,kq^2/a$ directly, and Step 3 gives $\tfrac12\cdot3\cdot q\cdot(2kq/a)=3\,kq^2/a$ — the same result, confirming the double-count correction works as advertised.

Go deeper · Derivation B · 1 minThe continuum limit and where self-energy sneaks in

Derivation BThe continuum limit and where self-energy sneaks in

The goal is equation (4), the charge-integral form, together with an honest account of the fine print: why the continuum version includes self-energy, and why that inclusion is harmless (finite) for any smooth distribution. The strategy is to promote the census formula (3) to an integral and then interrogate the one step that is not innocent — dropping the "all the others" exclusion. That step is usually waved through; here we price it.

Step 1. Chop the distribution into cells of size $\delta$ , each carrying $q_{\text{cell}}=\rho\,\delta^3$ . The census formula (3) over cells reads $W=\tfrac12\sum_{\text{cells}}q_{\text{cell}}\,V_{\text{others}}(\mathbf{r}_{\text{cell}})$ , where $V_{\text{others}}$ omits the cell's own contribution.

Step 2. Compare $V_{\text{others}}$ with the full $V$ : they differ by the cell's own potential at its own center, of order $V_{\text{self}}\sim\frac{1}{4\pi\varepsilon_0}\frac{\rho\,\delta^3}{\delta}=\frac{\rho\,\delta^2}{4\pi\varepsilon_0}$ . As the mesh refines, $\delta\to0$ , each cell's self-potential dies like $\delta^2$ . Why this is the crux: for a smooth density, "a charge's own neighborhood" carries vanishing potential — so exclusion or inclusion changes nothing in the limit.

Step 3. Price the total self-energy directly as a cross-check: each cell's internal energy is $\sim kq_{\text{cell}}^2/\delta=k\rho^2\delta^5$ , and with $N\sim\mathcal{V}/\delta^3$ cells the total is $\sim k\rho^2\delta^2\,\mathcal{V}\to0$ . The self-energy contribution of smooth charge is not merely finite — cell by cell it vanishes, so both readings of $V$ give the same integral.

Step 4. Take the limit with the full $V$ (simpler, since it is the physical potential computed by Potential of a Localized Charge Distribution):

W=\frac{1}{2}\int\rho\,V\,\mathrm{d}\tau

with companions $\tfrac12\int\sigma V\,\mathrm{d}a$ and $\tfrac12\int\lambda V\,\mathrm{d}\ell$ for surface and line charge. Why point charges break the truce: a point charge is the $\delta\to0$ limit at fixed $q$ , not fixed $\rho$ — its self-energy $\sim kq^2/\delta$ diverges instead of vanishing. Feed point charges to (4) and it returns infinity; feed them to (2) and it politely ignores self-energy. Same physics, different bookkeeping contracts. *As a consistency note: applying equation (4) to the uniform sphere of the next derivation reproduces $\tfrac35kQ^2/R$ , the same value obtained there by the independent shell-assembly route — a reassuring agreement between the two methods.*

Figure

Building a charged sphere like an onion — each new shell pays the residents' surface potential

Go deeper · Derivation C · 1 minEnergy of a uniform sphere by shell-on-shell assembly

Derivation CEnergy of a uniform sphere by shell-on-shell assembly

The goal is the classic benchmark $W=\tfrac35\,kQ^2/R$ for a uniformly charged solid sphere — derived by pure assembly, the method this article owns: build the sphere like an onion, shell by shell, and integrate the tolls. (The same number will be recomputed from the field side in Energy Density of the Electric Field; agreement of the two routes is the point of the benchmark.) Watch the characteristic $q\,\mathrm{d}q$ structure — it is the continuum version of "newcomer pays all residents".

Step 1. Grow the sphere at fixed density $\rho=Q/(\tfrac43\pi R^3)$ . When its radius has reached $r$ , the resident charge is $q(r)=Q\,r^3/R^3$ , and the next shell brings $\mathrm{d}q=\rho\cdot4\pi r^2\mathrm{d}r=3Q\,r^2\mathrm{d}r/R^3$ .

Step 2. The incoming shell pays the residents' potential at radius $r$ . By the shell results of Applications of Gauss's Law and Symmetry, the assembled ball acts at its surface like a point charge: $V(r)=kq(r)/r$ . So $\mathrm{d}W=V(r)\,\mathrm{d}q=\frac{k\,q(r)}{r}\,\mathrm{d}q$ . Why the shell's own self-energy is ignorable: it is a smooth surface layer of vanishing charge, contributing at second order in $\mathrm{d}q$ — the continuum truce of the previous derivation at work.

Step 3. Substitute and integrate from empty to full:

W=\int_0^R\frac{k}{r}\cdot\frac{Qr^3}{R^3}\cdot\frac{3Qr^2}{R^3}\,\mathrm{d}r=\frac{3kQ^2}{R^6}\int_0^Rr^4\,\mathrm{d}r=\frac{3kQ^2}{R^6}\cdot\frac{R^5}{5}

W=\frac{3}{5}\,\frac{1}{4\pi\varepsilon_0}\frac{Q^2}{R}

Step 4. Read the result. The energy is positive (like charge compressed), quadratic in $Q$ (doubling the charge quadruples the bill), and diverges as $R\to0$ — the point-charge self-energy reappearing as the onion shrinks. Numbers for scale: $Q=1.0\ \mu\mathrm{C}$ , $R=10$ cm gives $W=0.6\times8.99\times10^9\times(10^{-6})^2/0.1=54\ \mathrm{mJ}$ — modest; but shrink to the classical electron radius scale and the same formula is the seed of the self-energy crisis the next article inherits. The field-form computation gives the identical $\tfrac35kQ^2/R$ , and the numerical value checks out: $0.6\times8.99\times10^9\times10^{-12}/0.1=5.39\times10^{-2}$ J.

Limiting cases

Case	Result	Meaning
two point charges	$W=kq_1q_2/r$	pure interaction; positive for like, negative for unlike
bound pair ( $q,-q$ )	$W=-kq^2/r<0$	energy must be supplied to separate — a bond
all charges doubled	$W\to4W$	quadratic scaling; energy never superposes linearly
two distant dipoles	$W_{\text{int}}\sim p_1p_2/4\pi\varepsilon_0r^3$	neutral systems still interact, one power faster in falloff
uniform sphere	$W=\tfrac35\,kQ^2/R$	the computable benchmark; diverges as $R\to0$
alternating 1D chain	$W=-2\ln2\,ke^2/a$ per ion pair	Madelung's convergent tug-of-war (worked example 2)

Common trapHalf of the sum with self-terms is not the energy — and energies do not add

Two traps share the factor of one-half. First: the symmetric formula is $\tfrac12\sum_iq_iV(\mathbf{r}_i)$ with $V$ the potential of the other charges — it is emphatically not $\tfrac12\sum_{i,j}$ over all pairs including $i=j$ . The self-terms are not small corrections; each one is infinite, and their exclusion is the entire difference between the discrete formula (finite, physical) and the continuum formula fed point charges (divergent). When in doubt, retreat to the pair sum $\sum_{i<j}$ , which cannot even express the mistake. Second: energy is quadratic in the sources, so it refuses to superpose. Two configurations with energies $W_1$ and $W_2$ , brought into each other's presence, store $W_1+W_2$ plus an interaction term — the pairs that straddle the two systems. Fields add; potentials add; energies do not. Budget an assembled system by adding sub-assembly energies and you will silently drop every cross-pair in the problem.

Worked examples

1 · Three charges on a triangle (confidence builder)

Three charges of $q=+2.0$ nC each sit at the corners of an equilateral triangle of side $a=5.0$ cm. (a) Find the configuration energy. (b) One charge is released while the other two stay pinned; find its kinetic energy far away, and the energy left in the pair. (c) Redo (a) with one charge flipped to $-2.0$ nC.

Step 1. One pair's worth of energy: $kq^2/a=8.99\times10^9\times(2.0\times10^{-9})^2/0.050=7.19\times10^{-7}$ J. Three identical pairs: $W=3\times7.19\times10^{-7}=\boxed{2.16\ \mu\mathrm{J}}$ . (Census check, equation (3): each corner sits at $V=2kq/a=719$ V, and $\tfrac12\times3\times2.0\times10^{-9}\times719=2.16\ \mu$ J ✓.)

Step 2. The escaper owns its two pairs with the pinned residents: it departs carrying $2\times7.19\times10^{-7}=\boxed{1.44\ \mu\mathrm{J}}$ of kinetic energy at infinity. The pinned pair keeps its single pair term, $\boxed{0.72\ \mu\mathrm{J}}$ . Audit: $1.44+0.72=2.16\ \mu$ J ✓ — the ledger balances because each pair's energy left with exactly one of its members.

Step 3. Flip one charge to $-2.0$ nC: now one like pair ( $+7.19\times10^{-7}$ J) and two unlike pairs ( $-7.19\times10^{-7}$ J each): $W=(1-2)\times7.19\times10^{-7}=\boxed{-0.72\ \mu\mathrm{J}}$ . Negative: the trio is bound — dispersing it costs 0.72 μJ of external work. Same geometry, same magnitudes, opposite thermodynamic character; signs in the pair sum are not optional decorations.

2 · An ionic chain and the Madelung factor (exam level)

Model a 1D ionic crystal as an infinite line of alternating charges $\pm e$ spaced $a=0.28$ nm apart. (a) Find the interaction energy of one isolated $\mathrm{Na^+Cl^-}$ pair at separation $a$ . (b) Show the chain's energy per ion pair is $-2\ln2\cdot ke^2/a$ and evaluate it. (c) Compare with the 3D NaCl value.

Step 1. One isolated pair: $W_{\text{pair}}=-ke^2/a=-8.99\times10^9\times(1.602\times10^{-19})^2/(0.28\times10^{-9})=-8.24\times10^{-19}\ \mathrm{J}=\boxed{-5.1\ \mathrm{eV}}$ . Attractive, bound, and already the right scale for chemistry — electron-volts, not joules.

Step 2. In the infinite chain, use the census formula (3). The potential at one ion from all the others is a sum over neighbors at distances $na$ with alternating signs: $V=\frac{2}{4\pi\varepsilon_0}\cdot\frac{-e}{a}\big(1-\tfrac12+\tfrac13-\tfrac14+\cdots\big)$ for a positive ion (factor 2 for both directions). The alternating series is $\ln2$ , so $q_iV(\mathbf{r}_i)=-2\ln2\,ke^2/a$ for every ion, of either sign.

Step 3. Halve for double counting and convert to "per ion pair": $W=\tfrac12\sum q_iV=N\cdot\tfrac12\cdot(-2\ln2\,ke^2/a)$ for $N$ ions, i.e. $N/2$ pairs — so per pair, $W_{\text{pair-in-chain}}=-2\ln2\cdot\frac{ke^2}{a}$ . The factor $2\ln2\approx1.386$ is the 1D Madelung constant: the bonus binding from the whole lattice relative to one isolated pair.

Step 4. Numbers: $W=-1.386\times8.24\times10^{-19}=-1.14\times10^{-18}\ \mathrm{J}=\boxed{-7.1\ \mathrm{eV}}$ per ion pair. In real 3D rock salt the geometric series re-sums to the Madelung constant $\alpha=1.7476$ , giving $-1.7476\times5.14\ \mathrm{eV}\approx-9.0$ eV of pure Coulomb binding; short-range quantum repulsion (the reason the lattice does not collapse) gives back roughly ten percent, landing near the measured $\sim8$ eV lattice energy of NaCl. A tabletop sum over pair energies, run to infinity with care about signs, prices a real crystal to within ten percent.

Origins · 1 minWilliam Thomson (Lord Kelvin) and how this was discovered

William Thomson (Lord Kelvin)

1824–1907 · Glasgow

Thomson is the physicist who made energy the currency of physics — and electrostatic energy was among his first exchanges. A Cambridge wrangler running his own Glasgow laboratory at twenty-two, he published in 1853 the "mechanical value" of a charged conductor system, $\tfrac12QV$ : this article's equation (3) in its earliest working clothes, derived when "energy" itself was a word he was busy standardizing (he and Rankine gave physics "kinetic" and "potential" energy; the absolute temperature scale bears his later title). Thomson thought in energies where his contemporaries thought in forces — his method of images (an E2 star turn) came from asking what charge arrangement stores the same field energy — and his energy accounting of signal propagation famously guided, and nearly scuttled, the Atlantic telegraph cable he helped drive to success in 1866, the feat that made him Lord Kelvin. His conviction that energy bookkeeping must balance to the last joule pushed electrostatics from a theory of forces between charges toward a theory of energy stored in systems — the exact pivot this article executes, and the one Maxwell completed by moving the energy into the field.

History

1785

— Coulomb quantifies the force between charges: the integrand of every electrostatic energy to come.

1828

— Green's potential function turns work-against-forces into scalar bookkeeping, making assembly energies computable.

1847

— Helmholtz's memoir on the conservation of force places electrical assembly work on the same ledger as heat and motion.

1853

— William Thomson computes the mechanical value of electricity in a charged system, one-half QV, founding electrostatic energetics.

1873

— Maxwell's Treatise relocates the energy from the charges into the surrounding field — the handoff the next article completes.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §2.4 · Jackson, Classical Electrodynamics, 3rd ed., §1.11 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §1.14-1.15 · Zangwill, Modern Electrodynamics, §3.6 The integration-by-parts route connecting $\rho=\varepsilon_0\nabla\cdot\mathbf{E}$ and $\mathbf{E}=-\nabla V$ to the field-energy form, and the classical electron radius $r_e\approx2.82\times10^{-15}\,\mathrm{m}$ , belong to Energy Density of the Electric Field, where they are executed; $\varepsilon_0=8.854\times10^{-12}\,\mathrm{F/m}$ . The key numerical results in this article are: triangle energies $2.16/1.44/0.72/-0.72\ \mu$ J; ionic pair $-5.1$ eV, 1D chain $-2\ln2\times5.14=-7.1$ eV per pair; uniform sphere $\tfrac35kQ^2/R=54$ mJ for $1\ \mu$ C at 10 cm.

Builds on: Coulomb's Law · The Electrostatic Potential · The Electric Field · Gauss's Law in Differential Form · Superposition Principle for Electric Fields Leads to: Capacitance and Stored Energy · Maxwell Stress Tensor · Energy Stored in Dielectric Systems · Classical Electron Self-Energy

Why is the assembly work independent of the order and the paths used?

Because every toll along the way is a line integral of a conservative field. Each newcomer's cost is $q\Delta V$ — endpoints only, by the curl-free property certified in The Curl of the Electrostatic Field — so the route of each trip is irrelevant. Order-independence is subtler but follows from the same fact: the final ledger contains one term per pair, $kq_iq_j/r_{ij}$ , depending only on the final separations, and every assembly sequence produces exactly that set of terms. This is what "the energy is a state function" means operationally: $W$ belongs to the configuration, not to its history. If the field were not curl-free — spin up a time-varying magnetic field — assembly work would become path-dependent and no configuration energy could be defined at all.

What does negative configuration energy actually mean physically?

That the assembled state lies below the dispersed state on the energy ladder: the field did net positive work during assembly, and an external agent must supply energy to take the configuration apart. Negative $W$ is the signature of binding — the $-7.1$ eV per ion pair of the worked chain is why salt is a solid, and the analogous (quantum-corrected) negative energies are why atoms and molecules exist. Signs trace to the pair sum: unlike pairs contribute negative terms, like pairs positive, and $W$ 's sign reports which camp wins. A configuration with $W>0$ is spring-loaded — released, it flies apart, converting all of $W$ to kinetic energy, as the triangle example's escaping charge did with its 1.44 μJ.

Why exactly does the continuum formula include self-energy when the pair sum does not?

Track the phrase "all the others". In $\tfrac12\sum q_iV(\mathbf{r}_i)$ , the potential multiplying $q_i$ explicitly omits $q_i$ 's own contribution. Passing to $\tfrac12\int\rho V\,\mathrm{d}\tau$ , one substitutes the total potential — including, at each point, the contribution of the charge in its own neighborhood. For smooth $\rho$ this is a free upgrade: a cell of size $\delta$ has self-potential of order $\rho\delta^2/4\pi\varepsilon_0\to0$ , so exclusion versus inclusion changes nothing in the limit (priced honestly in the second derivation). A point charge violates the smoothness assumption maximally: its self-energy $\sim kq^2/\delta$ diverges as $\delta\to0$ . Hence the working contract: discrete charges, pair sum; continuous charge, integral; never feed one formula the other's diet.

Where is the energy actually stored — in the charges or somewhere else?

This article deliberately cannot answer that, and the honesty is instructive. Equation (4) writes $W$ as an integral over the charge; but an integration by parts — using Gauss's law to trade $\rho$ for the field — rewrites the same $W$ as an integral over all space, weighted by $E^2$ , with no reference to where the charge sits. In statics the two bookkeepings are numerically identical and experiment cannot separate them: only energy differences are measured. The field version wins the argument later, decisively: when charges accelerate, energy demonstrably travels through empty space at finite speed (light warms your face across a vacuum), and only the field ledger can track it in transit. Building that ledger — energy density $\tfrac{\varepsilon_0}{2}E^2$ and its consequences — is precisely the business of Energy Density of the Electric Field, next.

The first charge came in free — but doesn't it carry its own self-energy?

It does, and the assembly definition quietly set that energy aside as an unpayable, unchanging deposit. "Assembling from infinity" moves pre-existing charges; each arrives already wearing whatever internal energy it took to make it (for a true point charge, formally infinite). Since that internal energy is the same in the dispersed and assembled states, it cancels in every difference the theory ever computes — forces, binding energies, released kinetic energy — and the pair sum legitimately ignores it. The bookkeeping only turns dangerous when a formula (the continuum integral fed a point charge) re-includes the deposit as if it were configuration energy. The divergence is a statement about the point-charge idealization, not about electrostatics; its afterlife, from the classical electron radius to renormalization, starts in the next article.

Do I integrate one-half rho V over the charge only, or over all space?

Either — they are the same number, and realizing why prevents a common worry. The integrand contains $\rho$ , which vanishes wherever there is no charge, so extending the region from "the charged lump" to "all space" adds exactly zero. (Contrast the field form waiting in the next article, where the integrand $E^2$ lives everywhere and truncating the region genuinely loses energy.) The practical guidance: integrate over whatever region makes the mathematics cleanest, but make sure $V$ inside the integrand is the full potential of the distribution — computed once, by the methods of Potential of a Localized Charge Distribution — not the potential of the piece you happen to be integrating over at that moment.

How does the capacitor formula one-half QV connect to this article?

It is equation (3), specialized to two conductors. A charged capacitor holds $+Q$ on one plate (an equipotential at $V_+$ ) and $-Q$ on the other (at $V_-$ ): the census sum collapses to $W=\tfrac12\big[QV_++(-Q)V_-\big]=\tfrac12Q(V_+-V_-)=\tfrac12QV$ , with $V$ the plate-to-plate voltage. The factor of one-half has the same double-counting pedigree — equivalently, it records that while charging, each increment $\mathrm{d}q$ crossed only the voltage built so far, averaging to half the final value. With the definition $C=Q/V$ this becomes the workhorse trio $\tfrac12QV=\tfrac12CV^2=Q^2/2C$ , the starting point of the capacitance-and-stored-energy story this article's leads-to list opens.