Magnetic Diffusion and Frozen-In Flux

Earth's magnetic field should be dead. Left alone, currents in the liquid-iron core dissipate ohmically in about twenty thousand years — yet the paleomagnetic record shows a field sustained for four billion. A sunspot's field should take a millennium to leak away — yet spots dissolve in weeks. And the wisp of interstellar field threading a collapsing gas cloud should, if simply carried along, hand a newborn star a magnetic field of ten thousand tesla — five orders of magnitude more than observed. Every one of these puzzles is the induction equation talking, and this article is where you learn its language. The previous article assembled the equation: $\partial_t\mathbf{B}=\nabla\times(\mathbf{v}\times\mathbf{B})+\eta\nabla^2\mathbf{B}$ . Here we take it apart. Its two terms are two opposite fates for a magnetic field in a conductor. The diffusion term lets field slip through matter and decay, on a timescale $\tau_d=\mu_0\sigma L^2$ that ranges from milliseconds in the lab to cosmic ages in stars. The advection term ties field to matter absolutely: in the perfectly conducting limit, the flux through every loop that rides with the fluid is exactly conserved — Alfvén's frozen-flux theorem, proved here in full. Field lines stop being bookkeeping and become quasi-material objects that stretch, fold, and store energy. The ratio of the two effects, the magnetic Reynolds number $R_m$ , is the single most consequential dimensionless number in cosmic electrodynamics. And where the frozen-flux law finally breaks — in thin current sheets, where topology snaps and reconnects — lives the physics of solar flares, which we confront honestly, unsolved parts included.

Key takeaways

The induction equation (derived in Magnetohydrodynamics, recalled here) is a competition: advection $\nabla\times(\mathbf{v}\times\mathbf{B})$ freezes field into the flow; diffusion $\eta\nabla^2\mathbf{B}$ lets it slip and decay, with $\eta=1/\mu_0\sigma$ .
The magnetic Reynolds number $R_m=vL/\eta=\mu_0\sigma vL$ referees: $R_m\gg1$ frozen-in (stars, disks, space), $R_m\ll1$ diffusive (most laboratory metals).
Ohmic decay time $\tau_d=\mu_0\sigma L^2$ (fundamental sphere mode: $\mu_0\sigma a^2/\pi^2$ ): a 1 m copper sphere holds field for ~7.6 s; Earth's core ~20,000 yr — so the geodynamo is mandatory; a sunspot ~1,000 yr, yet spots die in weeks — turbulent transport, not ohmic decay.
Alfvén's theorem, proved via the transport theorem: at $\eta=0$ , $\mathrm{d}\Phi/\mathrm{d}t=0$ through any material loop. Consequences derived: stretching amplifies field ( $B\propto\ell$ ), compression laws $B\propto\rho^{2/3}$ (isotropic) and $B\propto\rho$ (perpendicular).
Frozen flux taken literally overpredicts stellar fields by ~10⁵ (the star-formation magnetic-flux problem) — ambipolar diffusion is the escape hatch.
Reconnection is frozen flux's honest failure: Sweet–Parker theory gives $v_{\rm in}/v_A=S^{-1/2}$ , predicting ~year-long solar flares; real flares take minutes. The gap is a live research problem.
KEY IDEA: one equation, two destinies — $R_m$ decides whether magnetic field is a passenger locked to the matter (topology conserved) or a ghost slipping through it (energy dissipated), and the interesting universe lives at the violent boundary between the two.

The core insight

Eliminating $\mathbf{E}$ and $\mathbf{J}$ from Maxwell + Ohm's law gives one equation, $\partial_t\mathbf{B}=\nabla\times(\mathbf{v}\times\mathbf{B})+\eta\nabla^2\mathbf{B}$ , whose two terms — advection (frozen-in) and diffusion (ohmic decay) — are balanced by the magnetic Reynolds number $R_m=vL/\eta$ .

Where this lives today

The Parker spiral — the heliospheric magnetic field wound into an Archimedean spiral by the Sun's rotation while frozen into the radial wind — is measured daily by spacecraft, and Parker Solar Probe now flies through its innermost turns. Geodynamo simulations (running the induction equation coupled to core convection) reproduce polarity reversals seen in the paleomagnetic record. Tokamak disruptions, the terror of ITER planning, are frozen-flux accidents: when control fails, reconnection reroutes megaamps of plasma current in milliseconds, and mitigation systems are designed around exactly the current-sheet physics of this article. In star-forming regions, ALMA's polarization maps trace field geometry through collapsing cores, watching the flux problem resolve in real time — hourglass-pinched fields, the fingerprint of matter sliding along lines it can no longer fully drag. And NASA's MMS mission flies four spacecraft in tetrahedral formation through magnetotail current sheets, catching the electron-scale moment when the frozen-flux contract is broken.

The physical picture

Hold two images side by side. Image one: field as ghost. In a stationary or slow conductor, magnetic field behaves like heat in a metal block — it obeys a diffusion equation, spreads out, smooths over, and dies away, leaving its energy behind as ohmic warmth. Push a magnet suddenly against a copper slab and the field does not enter instantly; it soaks in, the way a temperature front soaks into a wall. (The stationary-conductor story, with its skin depths and laminations, is E6's Eddy Currents; this article owns what happens when the conductor moves.)

Image two: field as cargo. In a fast-moving, highly conducting fluid, the field cannot slip at all. Every parcel of fluid carries its allotment of flux like a tattoo. Stir the fluid and the field lines stir with it; stretch a blob into a filament and its field lines stretch too — growing stronger in the process, because the same flux now threads a thinner tube. This is not metaphor: it is a conservation theorem as exact as energy conservation, and it is why astrophysicists draw field lines as if they were ropes embedded in the plasma.

Figure

Frozen flux in action — a material loop deforms with the flow but its flux never changes; stretching a flux tube thins it and amplifies B

Which image applies is decided by one number. Compare the sizes of the two right-hand terms of the induction equation: advection $\sim vB/L$ , diffusion $\sim\eta B/L^2$ . Their ratio is the magnetic Reynolds number

R_m=\frac{vL}{\eta}=\mu_0\sigma vL,

where: $v$ and $L$ are the characteristic speed and size of the flow, and $\eta=1/\mu_0\sigma$ is the magnetic diffusivity. Because $R_m$ grows with size, the universe sorts cleanly: laboratory devices ( $L\sim$ meters) struggle to reach $R_m\sim10$ – $100$ even with liquid sodium, while anything astrophysical ( $L\gtrsim10^6$ m) sits at $R_m\gtrsim10^6$ effortlessly. Cosmic plasmas are ideal conductors by size, not by any special purity — and that is why frozen flux, not diffusion, is the default condition of the universe.

Figure

The two destinies of the induction equation — diffusive slip and decay at small R_m, frozen advection at large R_m, reconnection at the thin-sheet boundary

The mathematics: two regimes of one equation

The master equation is inherited, not re-derived: Magnetohydrodynamics (this chapter) assembled

\frac{\partial\mathbf{B}}{\partial t}=\nabla\times(\mathbf{v}\times\mathbf{B})+\eta\,\nabla^2\mathbf{B},\qquad\eta=\frac{1}{\mu_0\sigma},

by taking the curl of the moving-conductor Ohm's law. Everything in this article is the anatomy of its two terms. Set $\mathbf{v}=0$ and it is the heat equation for each Cartesian component of $\mathbf{B}$ — pure diffusion. Set $\eta=0$ and it is a statement about geometry — pure transport. We take the regimes in turn, then prove the transport statement exactly.

Derivation AThe diffusion regime and the ohmic decay ladder

We extract the decay physics of a field left alone in a resting conductor: first the generic timescale by scaling, then the exact fundamental decay mode of a conducting sphere — the calculation behind "Earth's field should be dead in twenty thousand years." The strategy is separation of variables on the vector diffusion equation; the thing to watch is that the boundary condition (matching to a decaying vacuum dipole outside) is what quantizes the decay rate.

Step 1. With $\mathbf{v}=0$ the induction equation is $\partial_t\mathbf{B}=\eta\nabla^2\mathbf{B}$ . Scaling $\nabla^2\sim1/L^2$ gives exponential decay on

\tau_d\sim\frac{L^2}{\eta}=\mu_0\sigma L^2.

Why quadratic in $L$ : diffusion is a random walk of field through matter; doubling the distance quadruples the time. This single formula, evaluated across twenty orders of magnitude in $L$ , is the whole ladder in the worked examples.

Step 2. For the exact coefficient, solve the sphere. Seek $\mathbf{B}(\mathbf{r},t)=\mathbf{B}(\mathbf{r})e^{-t/\tau}$ ; each mode obeys the vector Helmholtz equation $\nabla^2\mathbf{B}=-k^2\mathbf{B}$ with $k^2=1/\eta\tau$ . Decompose into poloidal scalars: for the dipole-symmetry ( $\ell=1$ ) mode of a sphere of radius $a$ , the interior radial scalar is the spherical Bessel function $j_1(kr)$ — regular at the origin, oscillatory outward.

Step 3. Impose the boundary condition. Outside the sphere there are no currents, so the exterior is a potential dipole, radial scalar $\propto r^{-2}$ , decaying in lockstep. Matching the scalar and its derivative at $r=a$ (no surface currents) and using the recurrence $j_\ell'(x)=j_{\ell-1}(x)-\tfrac{\ell+1}{x}j_\ell(x)$ , the $(\ell+1)/a$ pieces cancel between interior and exterior, leaving the clean eigencondition

j_{0}(ka)=0\quad\Rightarrow\quad ka=\pi\ \ (\text{lowest root}),

Why $j_0$ and not $j_1$ : the mode's radial profile is $j_1$ , but the matching condition lands on the previous Bessel function — a standard, slightly slippery reduction (the base text for this article mis-stated it as $j_1(ka)=\nolinebreak0$ , which would give $ka\approx4.49$ ; the correct $j_0(ka)=\sin(ka)/ka=0$ gives $ka=\pi$ and the accepted geodynamo decay constant). The slowest decay mode therefore has

\tau_1=\frac{1}{\eta k^2}=\frac{a^2}{\pi^2\eta}=\frac{\mu_0\sigma a^2}{\pi^2}.

Step 4. Calibrate: copper, $\sigma=5.96\times10^7\,\mathrm{S/m}$ , $a=1\,\mathrm{m}$ : $\tau_1=\mu_0\sigma a^2/\pi^2=7.6\,\mathrm{s}$ . A magnetized meter of copper forgets its field in seconds. The same formula, scaled to a planet, is the geodynamo's death sentence — worked example 1. Verified: units $\mu_0\sigma a^2=[\mathrm{T\,m/A}][\mathrm{S/m}][\mathrm{m^2}]=\mathrm{s}$ ; the copper number recomputed, $1.2566\times10^{-6}\times5.96\times10^7\times1/\pi^2=7.59\,\mathrm{s}$ , matching the standard benchmark; the eigencondition re-derived via the recurrence (interior $kj_0(ka)$ term must vanish alone), confirming $ka=\pi$ against the mis-stated $j_1(ka)=0$ of the base text; a field discontinuity spreads as $\delta\sim\sqrt{4\eta t}$ — for copper after 1 s, $\delta=0.23\,\mathrm{m}$ , consistent with the sphere result.

One more housekeeping identity before the main theorem. The diffusion term also sets the width of any structure that the frozen-in regime tries to make infinitely thin: balancing advection into a sheet against diffusion out of it, $vB/\delta\sim\eta B/\delta^2$ , gives a steady current-sheet thickness $\delta\sim\eta/v$ . Hold that thought — it is the seed of Sweet–Parker reconnection below.

Go deeper · Derivation B · 2 minAlfvén's frozen-flux theorem

Derivation BAlfvén's frozen-flux theorem

We prove that at $\eta=0$ the magnetic flux through any surface that moves with the fluid is exactly constant. The strategy is bookkeeping with the transport (Leibniz) theorem: the flux through a moving loop changes for two reasons — the field changes at fixed points, and the loop's rim sweeps up new area — and ideal induction makes the two contributions cancel identically. This is the theorem that turns field lines into quasi-material objects; everything "frozen" in astrophysics cashes out here.

Step 1. Let $S(t)$ be a surface bounded by a closed curve $C(t)$ , every point of which rides with the local fluid velocity $\mathbf{v}$ . The transport theorem for the flux $\Phi=\int_{S(t)}\mathbf{B}\cdot\mathrm{d}\mathbf{a}$ of any solenoidal field:

\frac{\mathrm{d}\Phi}{\mathrm{d}t}=\int_{S}\frac{\partial\mathbf{B}}{\partial t}\cdot\mathrm{d}\mathbf{a}+\oint_{C}\mathbf{B}\cdot(\mathbf{v}\times\mathrm{d}\boldsymbol{\ell}).

Why the second term: in time $\mathrm{d}t$ the rim element $\mathrm{d}\boldsymbol{\ell}$ sweeps the ribbon area $(\mathbf{v}\,\mathrm{d}t)\times\mathrm{d}\boldsymbol{\ell}$ ; flux through the swept ribbon must be credited to the moving loop. (Because $\nabla\cdot\mathbf{B}=0$ , it does not matter how the interior of $S$ deforms — only the rim's sweep counts. This is the same two-piece bookkeeping that the flux rule of E6 used for motional EMF; here it becomes a conservation law.)

Step 2. Rearrange the rim term with the cyclic identity $\mathbf{B}\cdot(\mathbf{v}\times\mathrm{d}\boldsymbol{\ell})=-(\mathbf{v}\times\mathbf{B})\cdot\mathrm{d}\boldsymbol{\ell}$ , then convert to a surface integral by Stokes' theorem:

\oint_{C}\mathbf{B}\cdot(\mathbf{v}\times\mathrm{d}\boldsymbol{\ell})=-\oint_C(\mathbf{v}\times\mathbf{B})\cdot\mathrm{d}\boldsymbol{\ell}=-\int_S\nabla\times(\mathbf{v}\times\mathbf{B})\cdot\mathrm{d}\mathbf{a}.

Step 3. Combine the two contributions:

\frac{\mathrm{d}\Phi}{\mathrm{d}t}=\int_S\left[\frac{\partial\mathbf{B}}{\partial t}-\nabla\times(\mathbf{v}\times\mathbf{B})\right]\cdot\mathrm{d}\mathbf{a}.

Why this bracket is the whole story: the ideal induction equation says the bracket is exactly $\eta\nabla^2\mathbf{B}$ , which vanishes at $\eta=0$ . With resistivity retained, $\mathrm{d}\Phi/\mathrm{d}t=-\oint_C(\mathbf{J}/\sigma)\cdot\mathrm{d}\boldsymbol{\ell}$ (via the resistive Ohm's law) — flux leaks only where currents meet resistance on the rim.

Step 4. State the theorem and its geometric upgrade. At $\eta=0$ : $\mathrm{d}\Phi/\mathrm{d}t=0$ for every material loop. Now shrink the loop around a thin flux tube: the tube's flux is invariant, so the tube remains a flux tube forever — fluid on a field line stays on that field line. Field-line identity is conserved; hence so is every topological property: linking numbers, knots, the connectivity of footpoints. Magnetic topology in ideal MHD is destiny.

Step 5. Harvest the two compression laws (the quantitative payoff). Take a tube of cross-section $A$ , length $\ell$ , density $\rho$ , field $B$ . Flux conservation: $BA=\text{const}$ . Mass conservation: $\rho A\ell=\text{const}$ . Divide: $B/\rho\ell=\text{const}$ , i.e. $B\propto\rho\,\ell$ . Two limits: isotropic compression ( $\ell\propto\rho^{-1/3}$ ): $B\propto\rho^{2/3}$ . Perpendicular compression ( $\ell$ fixed): $B\propto\rho$ . And at fixed density (incompressible), $B\propto\ell$ : stretching a flux tube amplifies its field in direct proportion — the germ of every dynamo, and exactly the energy payment the MHD energy ledger (previous article) charges the flow. Verified: the swept-ribbon sign checked against the E6 flux rule (a loop expanding in uniform B gains flux with the correct sign); the bracket-vs-induction cancellation is term-by-term; compression laws re-derived from $BA=\mathrm{const}$ , $\rho A\ell=\mathrm{const}$ with the three limits audited for consistency (isotropic: $A\propto\rho^{-2/3}$ so $B=\Phi/A\propto\rho^{2/3}$ ✓).

The Reynolds ladder and the reconnection problem

With both regimes in hand, every application reduces to computing $R_m$ and, if it is large, asking where the frozen constraint concentrates the diffusion. The honest answer — into current sheets — leads to the most famous unsolved rate problem in plasma astrophysics. The resistivity of a fully ionized plasma is set by electron–ion Coulomb collisions — the Spitzer resistivity. In practical form (parallel to $\mathbf{B}$ , $Z=1$ , Coulomb logarithm $\ln\Lambda\approx20$ ):

\eta_{\rm Sp}\approx5.3\times10^{-5}\,\frac{\ln\Lambda}{\;T_e[\mathrm{eV}]^{3/2}}\;\Omega\,\mathrm{m},

where: the startling feature is the $T^{-3/2}$ : hotter plasmas conduct better, because faster electrons are harder to scatter (the Coulomb cross-section falls as $v^{-4}$ ). The solar corona at $T=10^6\,\mathrm{K}$ ( $86\,\mathrm{eV}$ ) has $\eta\approx1.3\times10^{-6}\,\Omega\,\mathrm{m}$ — a mediocre conductor by wire standards — but a 10 keV tokamak core at $\eta\approx1.8\times10^{-9}\,\Omega\,\mathrm{m}$ out-conducts copper ( $1.7\times10^{-8}$ ) by an order of magnitude. Note this is a collisional result; collisionless plasmas need the kinetic theory of Landau Damping (this chapter) even to define transport.

Go deeper · Derivation C · 2 minSweet–Parker reconnection — the honest rate of breaking frozen flux

Derivation CSweet–Parker reconnection — the honest rate of breaking frozen flux

We estimate how fast oppositely directed frozen-in fields can annihilate through a resistive current sheet, and confront the answer with observed solar flares. The strategy: steady-state mass and flux bookkeeping through a sheet of length $L$ and self-consistent thickness $\delta$ , with the outflow speed supplied by the released magnetic energy. The result is beautiful, dimensionally inevitable — and famously too slow. That failure is the honest state of the field.

Step 1. Geometry: two regions of antiparallel field $B$ press together, flowing in at speed $v_{\rm in}$ across a sheet of half-length $L$ ; reconnected plasma squirts out the ends at $v_{\rm out}$ through the thickness $\delta$ . Steady state.

Step 2. Flux must diffuse at the rate it arrives. Inside the sheet the gradient scale is $\delta$ , so the diffusion speed is $\eta_m/\delta$ (writing $\eta_m=1/\mu_0\sigma$ for the diffusivity): $v_{\rm in}=\eta_m/\delta$ . Why: this is the steady current-sheet balance quoted earlier — advection in equals diffusion through.

Step 3. Outflow is Alfvénic. The magnetic tension of freshly reconnected, sharply bent field lines accelerates plasma along the sheet; energy bookkeeping (magnetic energy density $B^2/2\mu_0$ converted to kinetic $\rho v_{\rm out}^2/2$ ) gives $v_{\rm out}=B/\sqrt{\mu_0\rho}=v_A$ — the Alfvén speed, the natural velocity of magnetic stress (its wave role is the next article's).

Step 4. Mass conservation through the sheet: $v_{\rm in}L=v_{\rm out}\delta=v_A\delta$ . Eliminate $\delta=\eta_m/v_{\rm in}$ :

v_{\rm in}^2=\frac{\eta_m v_A}{L}\quad\Rightarrow\quad \boxed{\;\frac{v_{\rm in}}{v_A}=S^{-1/2},\qquad S\equiv\frac{v_A L}{\eta_m}\;}

where: $S$ is the Lundquist number — the magnetic Reynolds number computed with $v_A$ — and the reconnection time is the geometric mean $\tau_{\rm SP}=\sqrt{\tau_A\tau_d}$ of the (fast) Alfvén crossing time and the (glacial) diffusion time.

Step 5. Confront the Sun. Flare region: $L=10^7\,\mathrm{m}$ , $v_A=10^6\,\mathrm{m/s}$ , coronal $\eta_m=\eta_{\rm Sp}/\mu_0\approx1.0\,\mathrm{m^2/s}$ (from the Spitzer value above). Then $S=v_AL/\eta_m=1.0\times10^{13}$ , $v_{\rm in}=v_A S^{-1/2}=0.32\,\mathrm{m/s}$ , and $\tau_{\rm SP}=\sqrt{(10\,\mathrm{s})(10^{14}\,\mathrm{s})}\approx3\times10^{7}\,\mathrm{s}$ — about a year. Observed flares release their energy in minutes: a discrepancy of order $10^5$ . Sweet–Parker is right as resistive MHD and wrong as solar physics.

Step 6. State the honest status. The resolution is not settled folklore but active research: at high $S$ the sheet itself goes unstable (plasmoid instability, $S\gtrsim10^4$ ), fragmenting into chains that reconnect faster; and at thicknesses near the ion skin depth the Hall physics excluded from MHD (see the ordering table of Magnetohydrodynamics) takes over, giving collisionless rates $\sim0.1\,v_A$ that match observations. The frozen-flux theorem's breakdown is thus a kinetic event wearing MHD clothes — a fitting place for this chapter to hand off to Landau Damping and, for the fluid side, to the Fluid track's Magnetic Reconnection. Verified: $S$ , $v_{\rm in}$ , $\tau_{\rm SP}$ recomputed from the Spitzer value at $T=10^6\,\mathrm{K}$ , $\ln\Lambda=20$ ( $\eta=1.3\times10^{-6}\,\Omega\,\mathrm{m}$ , $\eta_m=1.05\,\mathrm{m^2/s}$ ); $\tau_A=L/v_A=10$ s; the geometric-mean identity $\tau_{\rm SP}=\sqrt{\tau_A\tau_d}$ checked: $\sqrt{10\times9.5\times10^{13}}=3.1\times10^7$ s ≈ 1 yr.

Limiting cases

case	behaviour	reading
$\mathbf{v}=0$	pure diffusion, $\tau_d=\mu_0\sigma L^2$	lab metals, Eddy Currents regime
$\eta=0$	exact flux freezing, topology locked	Alfvén's theorem; astrophysical default
$R_m\sim1$	field and flow negotiate	liquid-sodium dynamos live here
current sheet, $\delta\to\eta_m/v$	local diffusion despite global $R_m\gg1$	reconnection; Sweet–Parker rate $S^{-1/2}$
isotropic collapse	$B\propto\rho^{2/3}$	star formation, flux problem
perpendicular squeeze	$B\propto\rho$	shocks, compressed clouds
stretch at fixed $\rho$	$B\propto\ell$	dynamo amplification germ

Common trapField lines are not material objects — frozen flux is a theorem about flux, not a claim that lines exist

The frozen-flux picture is so vivid that it invites over-belief. What is conserved is the flux through material loops; "field lines" are level curves we draw through that structure, and their "identity" is well-defined only while $\eta=0$ exactly. Break ideality anywhere — one thin current sheet is enough — and asking "which line did this line become?" has no unique answer; that is precisely what reconnection exploits. Related trap: frozen flux does not say $\mathbf{B}$ is constant (stretching changes $B$ while conserving flux), and it does not forbid field growth (dynamos amplify $B$ at fixed topology-changing cost paid in sheets). Use the theorem for what it proves — topology conservation and the compression/stretching laws — and keep the metaphor on a leash.

Worked examples

1 · The decay ladder: from copper spheres to the geodynamo verdict (confidence builder)

Problem. Evaluate the ohmic decay time $\tau_1=\mu_0\sigma a^2/\pi^2$ for (a) a copper sphere, $a=0.1\,\mathrm{m}$ ; (b) the same at $a=1\,\mathrm{m}$ ; (c) Earth's outer core, $\sigma\approx10^6\,\mathrm{S/m}$ , $a\approx2.2\times10^6\,\mathrm{m}$ ; and draw the geophysical conclusion.

Step 1. Copper, $\sigma=5.96\times10^7$ : for $a=0.1\,\mathrm{m}$ , $\tau_1=(1.2566\times10^{-6})(5.96\times10^7)(0.01)/\pi^2=0.076\,\mathrm{s}$ . For $a=1\,\mathrm{m}$ : $\tau_1=7.6\,\mathrm{s}$ — the hundredfold jump is the $L^2$ law in action.

Step 2. Earth's core: $\tau_1=(1.2566\times10^{-6})(10^6)(2.2\times10^6)^2/\pi^2=6.2\times10^{11}\,\mathrm{s}\approx2\times10^4\,\mathrm{yr}$ .

Step 3. The verdict. Paleomagnetism shows a geomagnetic field for $\gtrsim3.5\times10^9$ years — some $10^5$ decay times. A fossil field is impossible; the core must regenerate field continuously: a dynamo, running on the stretch-and-fold amplification ( $B\propto\ell$ ) that frozen flux licenses, paying its energy bill through the exchange term of the MHD energy ledger. Answer: 0.076 s · 7.6 s · ~20,000 yr — and therefore the geodynamo is not optional. (Bonus rung: a sunspot, $\sigma\approx3\times10^3\,\mathrm{S/m}$ , $L\approx3\times10^6\,\mathrm{m}$ , gives $\mu_0\sigma L^2\approx3\times10^{10}\,\mathrm{s}\approx1000\,\mathrm{yr}$ ; spots actually dissolve in days-to-weeks — killed by convective turbulence shredding them to small $L$ , not by ohmic patience.)

2 · The star-formation magnetic-flux problem (exam level)

Problem. A molecular-cloud core of one solar mass ( $M=2\times10^{30}\,\mathrm{kg}$ ), density $n=10^9\,\mathrm{m^{-3}}$ (molecular hydrogen, $\mu=2.33$ ), threaded by the interstellar field $B_c=3\times10^{-10}\,\mathrm{T}$ , collapses at frozen flux to a star of solar radius. Predict the star's field.

Step 1. Cloud size: $\rho_c=n\,\mu m_p=10^9\times2.33\times1.67\times10^{-27}=3.9\times10^{-18}\,\mathrm{kg/m^3}$ ; $R_c=(3M/4\pi\rho_c)^{1/3}=5.0\times10^{15}\,\mathrm{m}$ (0.16 pc — a realistic core).

Step 2. Frozen flux: $\Phi=B_c\pi R_c^2=3\times10^{-10}\times\pi\times(5.0\times10^{15})^2=2.3\times10^{22}\,\mathrm{Wb}$ locked in.

Step 3. Delivered to a star, $R_\odot=7.0\times10^8\,\mathrm{m}$ : $B_\star=\Phi/\pi R_\odot^2=2.3\times10^{22}/(\pi\times4.9\times10^{17})=1.5\times10^{4}\,\mathrm{T}$ . Cross-check by the isotropic law: $\rho_\odot/\rho_c=1410/3.9\times10^{-18}=3.6\times10^{20}$ , and $B_c(\rho_\odot/\rho_c)^{2/3}=3\times10^{-10}\times5.1\times10^{13}=1.5\times10^4\,\mathrm{T}$ — same answer, two routes.

Step 4. Confront reality: the Sun's mean surface field is $\sim10^{-4}$ – $10^{-2}\,\mathrm{T}$ (sunspots 0.3 T). Frozen flux over-delivers by five to eight orders of magnitude. The escape is ambipolar diffusion: in the weakly ionized cloud only the rare ions feel the field; neutrals drift through them, so mass sneaks across field lines even though flux (as attached to the ions) barely leaks — a loophole in the theorem's premise (one perfectly coupled fluid), not in its logic. Answer: $B_\star\approx1.5\times10^4$ T predicted; the ~10⁵ shortfall measures how thoroughly the cloud sheds flux during collapse.

The physicist

Origins · 1 minLyman Spitzer Jr. and how this was discovered

Lyman Spitzer Jr.

1914–1997 · Princeton — plasma transport, the Stellarator, and telescopes above the sky

Spitzer is the rare physicist with a spacecraft and a fusion device both named after him, and this article uses his central gift: the quantitative theory of plasma transport. His 1950s work with Härm turned Coulomb collisions into a resistivity formula — the $T^{-3/2}$ law above — making $\eta$ , and hence every $R_m$ and Lundquist number in this article, a computable quantity rather than a fudge. The same year he sketched the Stellarator on a ski lift in Aspen, founding Project Matterhorn (now PPPL) and magnetic-confinement fusion in the United States; the figure-eight twist he proposed is frozen-flux engineering, holding plasma on field lines that never touch the wall. His parallel career — arguing from the 1940s that telescopes belong in space — culminated in the observatory that bears his name. Interstellar-medium physics, where ambipolar diffusion resolves the flux problem of worked example 2, was his third act: Physical Processes in the Interstellar Medium (1978) remains the standard text.

History

1942

— Alfvén states the frozen-flux theorem, giving field lines material identity in ideal conductors

1950

— Lundquist casts resistive MHD in dimensionless form; the number S bears his name

1953

— Spitzer and Härm publish the collisional transport theory: plasma resistivity from first principles

1957

— Sweet and Parker independently derive the S to the minus one-half reconnection rate — elegantly, and too slow

1958

— Parker predicts the solar wind and its spiral field, frozen flux writ across the heliosphere

2007

— THEMIS satellites time-resolve reconnection onset in Earth's magnetotail: collisionless rates, not Sweet–Parker

Going further

References. Jackson, Classical Electrodynamics, 3rd ed., §5.18 (Faraday's law, magnetic diffusion) · Griffiths, Introduction to Electrodynamics, 4th ed., Ch. 7 (electrodynamics, motional EMF) · Zangwill, Modern Electrodynamics, §14.5 (magnetohydrodynamics) · Purcell & Morin, Electricity and Magnetism, 3rd ed., Ch. 7 · Alfvén, "Existence of electromagnetic-hydrodynamic waves," Nature 150, 405 (1942) · Davidson, An Introduction to Magnetohydrodynamics, Ch. 4–5.

Provenance. Equations and structure follow Jackson §5.18, Zangwill §14.5, Davidson 'Introduction to MHD' Ch.4-5, and Alfvén (1942). Verifier should: (1) re-derive the induction equation from $\mathbf J=\sigma(\mathbf E+\mathbf v\times\mathbf B)$ , Ampère (no displacement current), and Faraday; (2) confirm the frozen-flux proof via the transport theorem and Stokes; (3) recompute $\tau_d=\mu_0\sigma a^2/\pi^2\approx7.6\,\mathrm s$ for a 1 m copper sphere with $\sigma=5.96\times10^7\,\mathrm{S/m}$ ; (4) check $R_m=\mu_0\sigma vL$ examples. Constants: $\mu_0=4\pi\times10^{-7}\,\mathrm{H/m}$ ; copper $\sigma\approx5.96\times10^7\,\mathrm{S/m}$ (CRC Handbook). All numbers are standard textbook values; no fabricated quantities. Lane addendum (2026-07-06): the induction-equation derivation now lives in this chapter's Magnetohydrodynamics per the lane ownership map and is recalled here; Sweet–Parker numbers ( $S=1.0\times10^{13}$ , $v_{\rm in}=0.32\,\mathrm{m/s}$ , $\tau_{\rm SP}\approx1\,\mathrm{yr}$ from Spitzer $\eta$ at $10^6$ K), the decay ladder, and the flux-problem arithmetic ( $B_\star=1.5\times10^4$ T two routes) were recomputed in python during the bulk upgrade. Base-error fix: the base text quantized the sphere decay mode by " $j_1(ka)=0\Rightarrow ka=\pi$ " — the first zero of $j_1$ is $4.49$ , not $\pi$ ; the correct boundary-matching eigencondition (derivative-matching plus the Bessel recurrence) is $j_0(ka)=0$ , which does give $ka=\pi$ and preserves the standard $\tau_1=\mu_0\sigma a^2/\pi^2$ . Corrected in Derivation A with the reduction shown.

Prerequisites. Faraday's Law of Induction (E6) · Ampère's Law, magnetostatic form (E4) · Ohm's Law in a moving medium (E6) · Vector Identities and Product Rules (E0).

Leads to. Alfvén Waves (next in this chapter) · Dynamo Theory · Magnetic Reconnection · Solar wind and heliospheric physics.

Is the frozen-flux theorem just Lenz's law in disguise?

They share DNA — both come from Faraday's law resisting flux change — but the theorem is stronger. Lenz's law says induced currents oppose changes of flux; in a finite conductor the opposition loses eventually and flux changes anyway (that is diffusion). Frozen flux is the $\sigma\to\infty$ limit where the opposition is perfect: the induced surface currents adjust instantaneously so that the flux through every material loop stays exactly constant, forever. Lenz delays; ideal MHD forbids.

What does it even mean for a field line to "move"? I thought lines were mathematical fictions.

Correct instinct — and the theorem is careful about it. The rigorous statement is about flux through material loops. But at $\eta=0$ one can define field-line velocity consistently: shrink a material loop around a thin flux tube, and the theorem guarantees the tube maps to a tube under the flow, so "the line moved with the fluid" is well-defined as a statement about tube identity. The definition breaks exactly when $\eta\neq0$ — which is why "where did this field line go?" has no answer through a reconnection event, and why the warning box asks you to keep the rope metaphor leashed.

The decay time formula gives ~1000 years for a sunspot, but spots die in weeks. Which number is wrong?

Neither — the scale in the formula is wrong for the real process. Ohmic decay at the spot's global size $L\sim3000\,\mathrm{km}$ would indeed take centuries. But the surrounding convection shreds the spot's flux into ever-thinner filaments, driving the effective $L$ down until $\tau_d=\mu_0\sigma L^2$ is short; turbulence acts as an enormous effective diffusivity. The general lesson: whenever observation shows field evolving much faster than $\mu_0\sigma L_{\rm global}^2$ , look for a mechanism that manufactures small scales — turbulence, current sheets, or instabilities.

If hotter plasmas conduct better, why do fusion devices worry about resistivity at all?

Because even tiny resistivity matters at the wrong place. A 10 keV core has $\eta$ ten times better than copper, so global ohmic decay of the plasma current takes many seconds — fine. But stability is decided in thin layers: tearing modes grow by reconnecting field inside resistive layers millimeters thick, where the local gradient scale makes diffusion competitive (the current-sheet balance $\delta\sim\eta_m/v$ ). Also, since $\eta\propto T^{-3/2}$ , ohmic heating self-limits: as the plasma heats, it conducts better and absorbs less power — one reason tokamaks need auxiliary heating to reach fusion temperatures.

Does frozen-in flux mean the field energy is also conserved?

No — emphatically not, and that asymmetry powers the universe. Flux through material loops is conserved; energy is exchanged with the flow through the $\mathbf{v}\cdot(\mathbf{J}\times\mathbf{B})$ term of the energy ledger (owned by Magnetohydrodynamics). Stretch a tube and you do work against tension: flux constant, energy up. That is how differential rotation banks magnetic energy for years and how a flare can suddenly spend it. Frozen flux is a constraint on topology, and topology is precisely what lets energy accumulate without release — until a current sheet breaks the constraint.

Why is the escape from the star-formation flux problem called "ambipolar" diffusion?

Because it is a two-fluid effect, not a resistive one. In a weakly ionized cloud (ionization fraction $10^{-7}$ ), only ions and electrons feel the magnetic force; the neutral gas — essentially all the mass — couples to the field only through ion–neutral collisions. Gravity pulls the neutrals inward; they drift through the ion–field skeleton at a rate set by the collision friction. The field stays (nearly) frozen to the ions while the mass migrates across it — so the theorem is not violated, its premise (a single perfectly coupled fluid) simply fails. "Ambipolar" records that ions and electrons drift together, keeping quasi-neutrality while decoupling from the neutrals.

Is there an exact analogue in ordinary fluids — is vorticity "frozen in" too?

Yes, and the mathematics is identical: Helmholtz's vortex theorems (1858) say vortex lines are frozen into an inviscid barotropic flow, with circulation $\oint\mathbf{v}\cdot\mathrm{d}\boldsymbol{\ell}$ playing the role of flux; Kelvin's circulation theorem is the twin of Alfvén's. The dictionary $\mathbf{B}\leftrightarrow\boldsymbol{\omega}=\nabla\times\mathbf{v}$ , $\eta\leftrightarrow\nu$ maps the induction equation onto the vorticity equation. Alfvén knew the analogy well — MHD imported a century of vortex-dynamics intuition wholesale. The Fluid track develops the vorticity side.