Torque and Force on a Dipole in a Field

Electric Dipole Moment closed with three teasers left deliberately unpaid: in an external field a dipole feels a torque $\mathbf{N} = \mathbf{p}\times\mathbf{E}$ , carries energy $U = -\mathbf{p}\cdot\mathbf{E}$ , and — if the field is non-uniform — a net force $\mathbf{F} = (\mathbf{p}\cdot\nabla)\mathbf{E}$ . This article pays all three, with derivations and their fine print. Together they are the mechanics of the dipole: the previous chapter asked what fields a dipole makes; this one asks what fields do to it, and the answers drive everything from the Langevin statistics of the next-door article to the reason ions wear coats of water. The physics is a division of labor. In a uniform field, the two ends of a dipole feel equal and opposite forces: no net pull, but a couple — a pure torque that twists the moment toward alignment, with strength $pE\sin\theta$ and an energy landscape $U = -pE\cos\theta$ whose minimum is alignment and whose unstable summit is anti-alignment. In a non-uniform field the cancellation between the ends is imperfect by exactly one Taylor-expansion step, leaving the residual force $(\mathbf{p}\cdot\nabla)\mathbf{E}$ : an aligned dipole is pulled toward stronger field, which is why neutral polar molecules migrate up field gradients. Two pieces of fine print earn their own derivations because they are where errors breed. First, $(\mathbf{p}\cdot\nabla)\mathbf{E}$ and $\nabla(\mathbf{p}\cdot\mathbf{E})$ — which textbooks use interchangeably — are equal only when $\mathbf{p}$ is constant and the field is curl-free; we prove the equivalence and flag its boundaries. Second, torque is referred to a point: about an arbitrary point the honest formula is $\mathbf{N} = \mathbf{p}\times\mathbf{E} + \mathbf{r}\times\mathbf{F}$ , with the pure couple $\mathbf{p}\times\mathbf{E}$ the origin-free part. The worked numbers end at the hydration shell: a water dipole a quarter-nanometer from a sodium ion is bound by tens of $k_BT$ — the torque-and-force article is, among other things, the physics of why salt dissolves.

Key takeaways

In a uniform field a dipole feels zero net force but a torque $\mathbf{N} = \mathbf{p}\times\mathbf{E}$ , magnitude $pE\sin\theta$ — a pure couple, the same about every point.
The orientation energy is $U = -\mathbf{p}\cdot\mathbf{E}$ : minimum aligned, maximum anti-aligned, barrier $2pE$ between them; the torque is the slope of this landscape.
In a non-uniform field the ends' forces cancel imperfectly, leaving $\mathbf{F} = (\mathbf{p}\cdot\nabla)\mathbf{E}$ — for an aligned dipole, a pull toward stronger field.
$(\mathbf{p}\cdot\nabla)\mathbf{E} = \nabla(\mathbf{p}\cdot\mathbf{E})$ holds when $\mathbf{p}$ is fixed and $\nabla\times\mathbf{E} = 0$ — true in electrostatics; the induced-dipole case replaces it by $\tfrac12\alpha\nabla(E^2)$ with its factor-of-half.
About an arbitrary reference point, $\mathbf{N} = \mathbf{p}\times\mathbf{E} + \mathbf{r}\times\mathbf{F}$ : the couple is intrinsic, the $\mathbf{r}\times\mathbf{F}$ part is origin bookkeeping.
Scale anchors: water in $10^6\,\mathrm{V/m}$ — torque $6\times10^{-24}\,\mathrm{N\cdot m}$ , libration near $90\,\mathrm{GHz}$ , flip barrier only $0.003\,k_BT$ ; water beside an ion — binding $\sim$ tens of $k_BT$ , force $\sim$ nanonewtons.
KEY IDEA: a dipole in a field minimizes $U = -\mathbf{p}\cdot\mathbf{E}$ — the torque $\mathbf{p}\times\mathbf{E}$ rotates it into alignment, and in a non-uniform field the residual force $(\mathbf{p}\cdot\nabla)\mathbf{E}$ drags it toward stronger field.

Where this lives today

The gradient force on dipoles is a working tool across the lab bench. Dielectrophoresis chips steer, sort, and trap living cells in microfluidic channels by pushing their induced dipoles with shaped AC field gradients — the $\tfrac12\alpha\nabla E^2$ variant of equation (3) — sorting cancerous from healthy cells by their dielectric contrast alone, no labels attached. Molecular-beam physics uses the permanent-dipole version directly: hexapole electrodes create field gradients that force-focus polar molecules state-by-state (the focusing that made the ammonia maser possible), and Stark decelerators fire timed gradient stages at supersonic beams to bring polar molecules to rest — electric-dipole mechanics as a brake. The torque half runs subtler machinery. Liquid-crystal displays operate entire industries on $\mathbf{N} = \mathbf{p}\times\mathbf{E}$ -type alignment physics: each pixel applies a small voltage to reorient anisotropic molecules, twisting the local optic axis and gating polarized light. Modern biomolecular force fields lean on the same mechanics in simulation — every water model in a protein simulation is a set of point charges whose $\mathbf{p}\times\mathbf{E}$ torques and $(\mathbf{p}\cdot\nabla)\mathbf{E}$ forces are integrated trillions of times per run — and the hydration-shell energetics of worked example 2 sets the scale for ion channels, battery electrolytes, and why your nerves conduct at all: selectivity filters in potassium channels work by out-competing exactly the $38\,k_BT$ water shell this article priced.

Origins · 1 minCharles-Augustin de Coulomb and how this was discovered

Charles-Augustin de Coulomb

1736–1806 · Paris, Académie des Sciences

Coulomb is filed in memory as the inverse-square man, but his instrument measured torque, and his deepest expertise was mechanical. A military engineer who had spent years on friction and torsion (his memoir on stiffness of ropes and torsion in thin wires founded the subject), he realized in 1784–85 that a filament's twist provides a restoring torque so feeble and so linear that it can weigh forces of $10^{-8}$ newtons: the torsion balance. With a charged pith ball on a suspended arm — a dipole-armed pendulum reading electrical action as angular deflection — he established the $1/r^2$ law for charges and, in companion memoirs, measured the alignment torques on magnetized needles, establishing the $\sin\theta$ law of couple on a magnetic moment that is this article's equation (1) in its original costume. The program — convert force questions into torque measurements via near-equilibrium mechanics — became the template for a century of precision physics: Cavendish weighed the Earth with it, and torsion fibers still hang test masses in gravitational-wave and fifth-force experiments. The unit of the charge his balance measured is, fittingly, the coulomb.

History

1785

— Coulomb's torsion balance turns electric and magnetic forces into measurable torques; the sin-theta couple law for needles

1912

— Debye builds orientational polarization on the dipole's U = −pE landscape and its Boltzmann statistics

1922

— Stern and Gerlach deflect silver atoms with the magnetic twin (m·grad)B, revealing space quantization

1955

— Townes' ammonia maser uses electric-field gradients to state-sort polar molecules — dipole force as a beam tool

1973

— Liquid-crystal displays commercialize field-induced reorientation of molecular dipole ensembles

The physical picture

Hold a compass needle in the Earth's field and you have the whole story in your hand. The needle feels no net pull toward the north pole — the field at the needle is as uniform as fields come — but twist it off north and it fights back, hardest at ninety degrees, not at all when aligned or exactly reversed. The electric dipole in a uniform $\mathbf{E}$ is the same machine: the $+q$ end is pushed along the field, the $-q$ end against it, and since the pushes are equal and opposite they form a couple — zero net force, pure twist. The twist is largest when the lever arm is largest ( $\theta = 90°$ ), vanishes at alignment (stable — displaced slightly, the torque restores) and at anti-alignment (unstable — the summit of the energy hill, where the slightest nudge sends the dipole swinging around).

Figure

The couple — equal and opposite forces, net twist toward alignment

Now let the field vary in space, and watch the cancellation develop a leak. The $+q$ end samples the field here, the $-q$ end samples it a displacement $\mathbf{d}$ away; if the field grows along the dipole's length, the end sitting in stronger field wins the tug, and the dipole as a whole is dragged — toward strong field if aligned, toward weak field if anti-aligned. The imbalance is first-order in $d$ , which is exactly why a dipole (whose $p = qd$ stays finite as $d \to 0$ ) feels it while a point charge does not: force on a neutral object is a derivative coupling. The hierarchy continues upward exactly as the multipole ladder suggests — a quadrupole needs the field's second derivative to feel a net force — and sideways into magnetism, where $(\mathbf{m}\cdot\nabla)\mathbf{B}$ deflecting silver atoms in a field gradient was how Stern and Gerlach split a beam in two and discovered that angular momentum is quantized. The energy landscape ties the whole article together. $U(\theta) = -pE\cos\theta$ is a single cosine valley: torque is its slope, equilibrium its stationary points, the flip barrier its $2pE$ height — and its Boltzmann average is precisely the input the Langevin statistics of Permanent Dipoles and Orientational Polarization consume. For water in a strong laboratory field the barrier is a thousandth of $k_BT$ : thermal chaos treats the valley as barely a dimple, which is the deep reason alignment fractions are parts-per-thousand. Beside an ion, where fields reach $10^{10}\,\mathrm{V/m}$ , the same valley is tens of $k_BT$ deep — no longer a dimple but a trap, and the water molecule locks on. One formula, two regimes, and most of solution chemistry in the gap between them.

Figure

The energy landscape — one cosine valley governs torque, stability, and statistics

The three laws

Torque, about the dipole's own center, in any field (uniform or not, evaluated at the dipole's location):

\mathbf{N} = \mathbf{p}\times\mathbf{E}, \qquad N = pE\sin\theta \qquad (1)

where: $\theta$ — the angle from $\mathbf{p}$ to $\mathbf{E}$ ; the direction of $\mathbf{N}$ (right-hand rule) is the axis about which $\mathbf{p}$ swings toward $\mathbf{E}$ . Maximum torque at $\theta = 90°$ ; zero at $0$ and $180°$ — but the zeros differ in kind, as the energy shows. For water in $10^6\,\mathrm{V/m}$ : $N_{\max} = pE = 6.2\times10^{-24}\,\mathrm{N\cdot m}$ — laughably small until divided by a molecular moment of inertia ( $\sim2\times10^{-47}\,\mathrm{kg\,m^2}$ ), whereupon it implies angular accelerations of $10^{23}\,\mathrm{rad/s^2}$ and libration frequencies near $90\,\mathrm{GHz}$ . Molecules are tiny; the mechanics is violent. Energy of orientation, from the work the torque does:

U = -\,\mathbf{p}\cdot\mathbf{E} = -pE\cos\theta \qquad (2)

where: the zero is conventionally at $\theta = 90°$ (moment perpendicular to field). Aligned: $U = -pE$ , the floor. Anti-aligned: $U = +pE$ , the summit — an equilibrium (zero torque) but an unstable one. The full flip costs $2pE$ . This is the $U$ whose Boltzmann factor $e^{pE\cos\theta/k_BT}$ ran the entire previous article; note it carries no factor $\tfrac12$ — the moment is permanent, pre-built, nothing is spent creating it (the induced-dipole half lives in Induced Dipoles and Atomic Polarizability, and the contrast is a standing exam trap). Force, when the field varies over the dipole's extent:

\mathbf{F} = (\mathbf{p}\cdot\nabla)\,\mathbf{E} \;=\; \nabla(\mathbf{p}\cdot\mathbf{E}) \quad \text{(fixed } \mathbf{p}\text{, electrostatic } \mathbf{E}\text{)} \qquad (3)

where: $(\mathbf{p}\cdot\nabla)\mathbf{E}$ — the directional derivative of the field along the dipole moment, times its magnitude: literally "how much the field changes from one end of the dipole to the other." In a uniform field it vanishes identically — no free lunch for neutral objects. The second form, gradient of the energy landscape with $\mathbf{p}$ held fixed, is often quicker to compute; Derivation B proves the two agree exactly when the moment is constant and the field curl-free, and patrols the boundary where they part company.

Derivation ATorque and energy — the couple, its landscape, and the reference-point theorem

The goal is equations (1) and (2) for a physical dipole (charges $\pm q$ separated by $\mathbf{d}$ , moment $\mathbf{p} = q\mathbf{d}$ ) in a uniform field, plus the bookkeeping theorem for torque about an arbitrary point — the piece that makes "the torque on a dipole" a well-posed phrase. The strategy is bare Newtonian mechanics on the two charges; the only electromagnetic input is $\mathbf{F} = q\mathbf{E}$ on each end. Watch how uniformity enters twice: once to kill the net force, once to make the torque origin-independent.

Step 1. Net force. The ends feel $+q\mathbf{E}$ and $-q\mathbf{E}$ ; in a uniform field these are equal and opposite:

\mathbf{F}_{\rm net} = q\mathbf{E} - q\mathbf{E} = 0.

A uniform field moves no neutral object, however polar. Whatever survives must be rotational.

Step 2. Torque about the dipole's center. Place the ends at $\pm\mathbf{d}/2$ from the center; sum $\mathbf{r}\times\mathbf{F}$ :

\mathbf{N} = \frac{\mathbf{d}}{2}\times(q\mathbf{E}) + \left(-\frac{\mathbf{d}}{2}\right)\times(-q\mathbf{E}) = q\,\mathbf{d}\times\mathbf{E} = \mathbf{p}\times\mathbf{E}.

Magnitude $pE\sin\theta$ , direction swinging $\mathbf{p}$ toward $\mathbf{E}$ . Nothing about the internal structure survives except the product $q\mathbf{d}$ — a pure dipole ( $d\to0$ , $p$ fixed) feels exactly the same torque, as the multipole philosophy demands.

Step 3. The reference-point theorem. About an arbitrary point, with the dipole's center at $\mathbf{r}$ from it, each position vector gains $\mathbf{r}$ : the added terms sum to $\mathbf{r}\times(q\mathbf{E} - q\mathbf{E}) = \mathbf{r}\times\mathbf{F}_{\rm net}$ . Hence in general

\mathbf{N}_{\rm about\ O} = \mathbf{p}\times\mathbf{E} + \mathbf{r}\times\mathbf{F}_{\rm net},

and in a uniform field, where $\mathbf{F}_{\rm net} = 0$ , the torque is the same about every point — the defining property of a couple. Only then is "the torque on the dipole" origin-free; in a non-uniform field one must name the reference point, and the clean split is intrinsic couple $\mathbf{p}\times\mathbf{E}$ plus the lever term $\mathbf{r}\times\mathbf{F}$ carrying all the origin dependence — the mechanical twin of the moment bookkeeping in Origin Dependence of Multipole Moments.

Step 4. The energy landscape. Rotate the dipole quasi-statically from the conventional zero at $\theta_0 = 90°$ to angle $\theta$ against the field's torque; the work stored is

U(\theta) = \int_{\pi/2}^{\theta} pE\sin\theta'\,\mathrm{d}\theta' = -pE\cos\theta = -\mathbf{p}\cdot\mathbf{E}.

Check the calculus closes: $N = -\mathrm{d}U/\mathrm{d}\theta = -pE\sin\theta$ — magnitude $pE\sin\theta$ , sign restoring toward $\theta = 0$ . Stable valley at alignment, unstable summit at anti-alignment, barrier $2pE$ . Verified: pure-dipole limit leaves both results unchanged (only p enters); the couple property matches the mechanics of force pairs; for water at 10⁶ V/m the barrier 2pE = 1.2×10⁻²³ J = 0.003 kBT, consistent with the ~0.05% alignment computed in the orientational-polarization article — the two articles' numbers interlock.

Go deeper · Derivation B · 2 minThe force in a non-uniform field — and the two formulas that are almost always equal

Derivation BThe force in a non-uniform field — and the two formulas that are almost always equal

The goal is $\mathbf{F} = (\mathbf{p}\cdot\nabla)\mathbf{E}$ from an honest Taylor expansion, then the identity connecting it to $\nabla(\mathbf{p}\cdot\mathbf{E})$ , with the conditions stated sharply — because the literature uses both, exams mix them, and the induced-dipole case genuinely differs. The strategy: expand the field between the dipole's ends to first order, cancel, and keep the residue; then a vector identity plus electrostatics for the equivalence. Watch the factor that does not appear: no $\tfrac12$ , because $\mathbf{p}$ here is permanent.

Step 1. Taylor-expand across the dipole. Ends at $\mathbf{r}_\pm = \mathbf{r}\pm\mathbf{d}/2$ :

\mathbf{F} = q\,\mathbf{E}(\mathbf{r} + \tfrac{\mathbf{d}}{2}) - q\,\mathbf{E}(\mathbf{r} - \tfrac{\mathbf{d}}{2}) = q\,\big[(\mathbf{d}\cdot\nabla)\mathbf{E} + O(d^3)\big] = (\mathbf{p}\cdot\nabla)\,\mathbf{E}.

The zeroth-order terms cancel (that was Step 1 of Derivation A); the first-order difference survives with coefficient $q\mathbf{d} = \mathbf{p}$ ; even-order terms cancel by symmetry, so the pure-dipole limit is exact. Why the derivative coupling matters: the force scale is $p\,|\partial E/\partial x|$ , not $pE$ . A huge but uniform field grips a dipole rotationally yet cannot move it; a modest field with a steep gradient — the flank of an ion, the focus of a beam — drags it hard.

Step 2. The equivalence. Use the identity $\nabla(\mathbf{p}\cdot\mathbf{E}) = (\mathbf{p}\cdot\nabla)\mathbf{E} + \mathbf{p}\times(\nabla\times\mathbf{E})$ for constant $\mathbf{p}$ . In electrostatics $\nabla\times\mathbf{E} = 0$ (the curl article of E1), so

\mathbf{F} = (\mathbf{p}\cdot\nabla)\mathbf{E} = \nabla(\mathbf{p}\cdot\mathbf{E}) = -\nabla U,

and the force is the downhill gradient of the same landscape whose angular slope is the torque — one potential governs both motions. The equivalence fails on two frontiers, both worth naming: time-varying fields, where $\nabla\times\mathbf{E} = -\partial\mathbf{B}/\partial t \neq 0$ and the extra term $\mathbf{p}\times(\nabla\times\mathbf{E})$ is real physics; and induced moments, where $\mathbf{p} = \alpha\mathbf{E}$ is not constant — there the correct energy is $-\tfrac12\alpha E^2$ and the force $\tfrac12\alpha\nabla(E^2)$ , the tweezer formula, with the half that permanent dipoles never pay.

Step 3. Orientation decides the sign. For $\mathbf{p}$ aligned with a field that strengthens along itself, $F > 0$ toward strong field; anti-aligned, the dipole is expelled toward weak field. In thermal ensembles alignment wins on Boltzmann average, so polar gases and polarizable matter migrate up-gradient — dielectrophoresis, and the reason a charged comb lifts neutral paper scraps: it first orients/induces their dipoles, then pulls on them with its own field gradient.

Step 4. A worked structure: dipole radial to a point charge $Q$ at distance $r$ , aligned outward-pointing along $-\hat{\mathbf{r}}$ ... take $\mathbf{p}$ pointing toward the charge (the bound orientation). The field magnitude is $E = Q/4\pi\varepsilon_0r^2$ , so $F = p\,|\mathrm{d}E/\mathrm{d}r| = 2Qp/4\pi\varepsilon_0r^3$ , attractive — the $1/r^3$ charge–dipole force law, one power steeper than Coulomb, exactly as the energy $U = -pE \propto 1/r^2$ demands. Worked example 2 feeds it real ionic numbers. Verified: the Taylor residue checked against exact two-charge summation for a finite dipole near a point charge (agreement to O(d²/r²)); the vector identity is standard (E0 product rules); the induced-dipole contrast matches the IPOL article's factor-of-half derivation; charge-dipole 1/r³ law matches Griffiths §4.1 problems.

Limiting cases

Case	Result	Meaning
uniform field	$\mathbf{F} = 0$ , $\mathbf{N} = \mathbf{p}\times\mathbf{E}$ same about every point	pure couple; alignment dynamics only
aligned, non-uniform	$\mathbf{F} = p\,\partial E/\partial s > 0$ toward strong field	up-gradient migration; comb and paper
anti-aligned	force reverses; equilibrium unstable	flips first, then migrates
$\theta = 0$ vs $\theta = \pi$	both $N = 0$ ; $U = -pE$ vs $+pE$	valley vs summit — stability differs
induced moment $\mathbf{p} = \alpha\mathbf{E}$	$U = -\tfrac12\alpha E^2$ , $\mathbf{F} = \tfrac12\alpha\nabla E^2$	the factor-half world (tweezers)
time-varying field	$\nabla(\mathbf{p}\cdot\mathbf{E}) \neq (\mathbf{p}\cdot\nabla)\mathbf{E}$	curl term wakes up; beyond statics
quadrupole in uniform or constant-gradient field	$F$ needs second derivatives	ladder of derivative couplings continues
charge–dipole, radial	$U \propto -1/r^2$ , $F \propto -1/r^3$	one power past Coulomb; hydration physics

Common trapZero torque does not mean equilibrium you can trust — and no force does not mean no interaction

Two traps, one geometric and one categorical. Geometric: the torque vanishes at both $\theta = 0$ and $\theta = 180°$ , but only alignment is stable — an anti-aligned dipole is balanced on the energy summit, and any problem asking "for what orientation is the dipole in equilibrium" deserves both answers with their stability labels. Categorical: "uniform field ⇒ no net force" is a statement about translation only — the field still torques the dipole, does work on its rotation, and stores $\pm pE$ of orientation energy. Conversely "the molecule is neutral" forbids nothing: derivative couplings give neutral matter both torque and (in gradients) force. The clean statement: charge couples to $\mathbf{E}$ , dipole to its first derivatives for force but to $\mathbf{E}$ itself for torque, quadrupole one derivative later throughout.

Worked examples

1 · Water in a laboratory field — torque, barrier, and the 90 GHz jiggle (confidence builder)

A water molecule ( $p = 6.2\times10^{-30}\,\mathrm{C\cdot m}$ , moment of inertia $I \approx 1.9\times10^{-47}\,\mathrm{kg\,m^2}$ ) sits in $E = 1.0\times10^6\,\mathrm{V/m}$ at $T = 293\,\mathrm{K}$ . Find the maximum torque, the flip barrier against $k_BT$ , and the small-oscillation frequency about alignment.

Step 1. Maximum torque (at $\theta = 90°$ ): $N_{\max} = pE = (6.2\times10^{-30})(10^6) = 6.2\times10^{-24}\,\mathrm{N\cdot m}$ .

Step 2. Flip barrier: $\Delta U = 2pE = 1.2\times10^{-23}\,\mathrm{J}$ . Against thermal energy: $\Delta U/k_BT = 1.2\times10^{-23}/4.0\times10^{-21} = 0.003$ . The full energy landscape is a 0.3%-of- $k_BT$ ripple — thermal motion barely notices it, which is precisely why the orientational-polarization article found alignment fractions of $10^{-4}$ : these two articles compute the same physics from opposite ends.

Step 3. Libration frequency. Small oscillations about alignment obey $I\ddot\theta = -pE\sin\theta \approx -pE\,\theta$ , a pendulum with $\omega = \sqrt{pE/I} = \sqrt{6.2\times10^{-24}/1.9\times10^{-47}} = 5.7\times10^{11}\,\mathrm{rad/s}$ , i.e. $f \approx 90\,\mathrm{GHz}$ .

Step 4. Read the numbers honestly. In a real liquid this "pendulum" never completes a swing — collisions arrive every $\sim0.1\,\mathrm{ps}$ , so the response is overdamped reorientational relaxation (Debye's picture, $\sim10\,\mathrm{ps}$ for water), not resonance. Your microwave oven at 2.45 GHz is not tuned to any molecular resonance — it drives the lazy tail of the relaxation, chosen for penetration depth into food, a persistent and pleasing myth-correction.

2 · The hydration shell — why salt dissolves (exam level)

A water molecule sits $r = 0.24\,\mathrm{nm}$ from a Na $^+$ ion (charge $+e$ ), dipole aligned toward the ion. Estimate the field at the molecule, the binding energy in units of $k_BT$ , and the attractive force. Then judge the estimate.

Step 1. Ion field at the molecule: $E = e/4\pi\varepsilon_0r^2 = (8.99\times10^9)(1.6\times10^{-19})/(0.24\times10^{-9})^2 = 2.5\times10^{10}\,\mathrm{V/m}$ — twenty-five thousand times any laboratory static field. Molecules living next to ions inhabit a different electrostatic universe.

Step 2. Binding energy: $U = -pE = -(6.2\times10^{-30})(2.5\times10^{10}) = -1.5\times10^{-19}\,\mathrm{J}$ , i.e. $|U| \approx 38\,k_BT$ at room temperature.

Step 3. Force, from the charge–dipole law: $F = 2ep/4\pi\varepsilon_0r^3 = 2(8.99\times10^9)(1.6\times10^{-19})(6.2\times10^{-30})/(0.24\times10^{-9})^3 = 1.3\times10^{-9}\,\mathrm{N}$ — about a nanonewton, enormous by molecular standards (covalent bonds rupture near a few nN).

Step 4. Judge it. The point-dipole-in-vacuum estimate overshoots — the real first-shell binding of water to Na $^+$ is roughly $15$ – $25\,k_BT$ once finite molecular size, neighboring-water screening, and imperfect alignment are paid for. But the verdict survives the corrections by a wide margin: ion–dipole binding is many times $k_BT$ , so ions in water wear ordered shells of oriented molecules — torque locks the orientation ( $u = pE/k_BT \approx 38 \gg 1$ : this is the saturated Langevin regime, the far plateau the previous article said static lab fields never reach), and the gradient force holds the shell radially. Dissolving salt is this example run $6\times10^{23}$ times.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §4.1.2–4.1.3 · Jackson, Classical Electrodynamics, 3rd ed., §4.2 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §10.6 · Zangwill, Modern Electrodynamics, §4.2.

Provenance. Equations and derivations follow Griffiths §4.1.2–4.1.3 (torque $\mathbf{p}\times\mathbf{E}$ , force $(\mathbf{p}\cdot\nabla)\mathbf{E}$ , energy $-\mathbf{p}\cdot\mathbf{E}$ ), Jackson §4.2 (origin dependence of torque), Purcell §10.6, and Zangwill §4.2. The water dipole moment $p\approx6.2\times10^{-30}$ C·m ( $\approx1.85$ D) is a standard tabulated value. A verifier should recheck the Taylor-expansion derivations of $\mathbf{F}$ and $\boldsymbol{\tau}$ , the vector identity used in the gradient form, and the factor of $\tfrac12$ in the induced-dipole energy against these texts. Worked numbers recomputed for this edition: $N_{\max} = 6.2\times10^{-24}$ N·m and barrier $2pE = 0.003\,k_BT$ at $10^6$ V/m; libration $\omega = 5.7\times10^{11}$ rad/s ( $\approx90$ GHz) with $I = 1.9\times10^{-47}$ kg·m²; ion example at 0.24 nm — $E = 2.5\times10^{10}$ V/m, $|U| = 1.5\times10^{-19}$ J $= 38\,k_BT$ , $F = 1.3$ nN.

Builds on: Electric Dipole Moment · Coulomb's Law · Curl of the Electric Field · Origin Dependence of Multipole Moments

Leads to: Permanent Dipoles and Orientational Polarization · Polarization Density P · Forces on Dielectrics · Dielectric Sphere in a Uniform Field

Why does the energy formula have no factor of one-half when the induced-dipole one does?

Because nothing here was built on credit. A permanent dipole's moment exists before the field arrives; bringing them together banks the full alignment payoff $-pE$ . An induced dipole must first be created by the field, and stretching the molecular spring costs exactly half the payoff, leaving $-\tfrac12\alpha E^2$ (the audit is in Induced Dipoles and Atomic Polarizability). The practical rule: moment fixed → no half; moment proportional to field → half. Molecules that are both polar and polarizable — most of them — take both terms, each under its own rule.

If the torque is the same about every point in a uniform field, why bother with the r cross F term at all?

Because fields are only ever locally uniform, and the moment you care about total angular momentum about a fixed axis — a molecule tumbling past an ion, a bead in a tweezer — the net force's lever arm contributes. The split $\mathbf{N} = \mathbf{p}\times\mathbf{E} + \mathbf{r}\times\mathbf{F}$ is the useful decomposition: the first term is intrinsic (origin-free, orientation physics), the second is pure bookkeeping that vanishes exactly when $\mathbf{F}$ does. It is the rotational twin of the origin-dependence theorem for multipole moments: intrinsic leading behavior, convention-dependent corrections fed by lower structure.

Which way does an anti-aligned dipole move in a non-uniform field — and does it stay anti-aligned?

Instantaneously, it is pushed toward weak field: $U = +pE$ makes strong field expensive. But the configuration is doubly doomed — anti-alignment sits on the torque's unstable summit, so any perturbation flips the dipole into the valley, whereupon the force reverses and drags it up-gradient. In thermal ensembles the flip happens essentially instantly, which is why polar and polarizable matter as a population always migrates toward strong field. The rare exceptions are quantum: molecules prepared in specific "low-field-seeking" states can be trapped at field minima before relaxing — exactly how Stark decelerators and some molecule traps choose to work.

Does the torque p cross E do work, and where does that energy come from?

Rotating from anti-aligned to aligned, the field's torque does work $2pE$ on the dipole — real energy, delivered to rotational kinetic energy (then typically to heat via collisions). It comes from the energy of the field-plus-sources system: strictly, the charges maintaining $\mathbf{E}$ (the battery on the capacitor plates) supply it. In a microwave oven this ledger runs continuously: the oscillating field torques water molecules, molecular friction converts the rotational work to heat, and the magnetron pays the bill — a few hundred watts of $\mathbf{N}\cdot\boldsymbol{\omega}$ , felt as hot soup.

How does the (p·grad)E force generalize to the interaction between two dipoles?

Put dipole 2 in the field of dipole 1 (the $1/r^3$ field owned by Electric Dipole Moment) and apply this article's laws: the energy is $U = [\mathbf{p}_1\cdot\mathbf{p}_2 - 3(\mathbf{p}_1\cdot\hat{\mathbf{r}})(\mathbf{p}_2\cdot\hat{\mathbf{r}})]/4\pi\varepsilon_0r^3$ , and the force falls as $1/r^4$ . Sign depends on geometry: head-to-tail attracts, side-by-side parallel repels — which is why polar molecules chain up. For water pairs at contact ( $r \approx 0.3$ nm) the energy is a few $k_BT$ : strong enough to structure the liquid, weak enough to keep it fluid. Thermal averaging of this interaction over orientations produces the $1/r^6$ Keesom contribution to van der Waals forces — this article's mechanics, twice Boltzmann-averaged.

Is there a magnetic version of every result here?

Almost line for line: $\mathbf{N} = \mathbf{m}\times\mathbf{B}$ , $U = -\mathbf{m}\cdot\mathbf{B}$ , $\mathbf{F} = \nabla(\mathbf{m}\cdot\mathbf{B})$ for a rigid moment — compass alignment, Stern–Gerlach deflection, MRI physics. The fine print differs instructively: magnetic fields are never curl-free where currents flow, so the $(\mathbf{m}\cdot\nabla)\mathbf{B}$ versus $\nabla(\mathbf{m}\cdot\mathbf{B})$ distinction has more bite, and "permanent" magnetic moments (spins) are quantum objects whose torque dynamics is precession rather than pendulum swing. The magnetostatics chapter rebuilds all of this properly; carrying over the electric intuition is legitimate and encouraged, fine print included.

The libration frequency came out at 90 GHz — why doesn't a 90 GHz beam heat water better than a microwave oven?

It would couple more strongly per molecule — but it would never reach most of the food. Absorption and penetration trade off: at 90 GHz the loss is so high that everything happens in the outer millimeter (crust, not cooking); at 2.45 GHz water absorbs weakly enough that fields penetrate centimeters, heating volumetrically. The oven frequency is an engineering optimum (plus regulatory allocation), not molecular tuning. In the liquid, moreover, the sharp 90 GHz pendulum line does not exist — collisions overdamp it into the broad Debye relaxation band — so there is no resonance to hit in the first place; librational features survive only as broad far-infrared bands.