Magnetic Dipole Moment of Atoms

The magnetization article filled matter with atomic dipoles and computed with their density; this article opens the atoms. Where do the moments come from, why do they arrive in a fixed universal size, and why is magnetism — alone among the classical subjects in this track — impossible without quantum mechanics? The answers all flow from one structural fact: an atomic magnetic moment is rigidly welded to angular momentum. A charge that goes around is simultaneously a current loop (a moment) and a mass in circulation (angular momentum), and the ratio of the two — the gyromagnetic ratio $\gamma$ — depends only on charge and mass: $\gamma_{\rm orb}=-e/2m_e$ for orbital electron motion, a number with no orbit radius, speed, or shape in it. That welding does two enormous things. First, it imports quantization: angular momentum comes in units of $\hbar$ (the one quantum input this article takes on credit), so moments come in units of $\gamma\hbar$ — the Bohr magneton $\mu_B=e\hbar/2m_e=9.274\times10^{-24}$ J/T, the atom of magnetism, with a nuclear counterpart $\mu_N$ suppressed by the proton's 1836-fold heavier mass. Iron's $2.2\,\mu_B$ per atom, the hydrogen hyperfine numbers, the entire moment catalog of this chapter: all denominated in these two coins. Second, it dictates dynamics: the mechanics article proved that a moment tied to angular momentum precesses rather than aligns, at $\omega=\gamma B$ — and with $\gamma$ now given its atomic values, that abstract law becomes numbers: 14 GHz/T for orbits, 28 GHz/T for electron spin, 42.6 MHz/T for protons — the operating frequencies of ESR, NMR, and MRI. Two honesty flags complete the opening. Spin has $g\approx2$ — twice the orbital ratio — and it is not a spinning ball (the classical picture fails by a factor of 170 times the speed of light, computed below); it is Dirac-equation physics wearing a classical costume. And deeper still: a clean theorem — Bohr–van Leeuwen — shows classical statistical mechanics predicts exactly zero equilibrium magnetization for any matter whatsoever. Every magnet you have ever touched is a quantum object.

Key takeaways

Universal welding: for any orbital motion of particles with one charge-to-mass ratio, $\vec m=\frac{q}{2m}\vec L$ — geometry drops out; only $q/2m$ survives. For electrons: $\vec m=-\frac{e}{2m_e}\vec L$ , moment antiparallel to angular momentum.
Quantized angular momentum ( $L\sim\hbar$ ) ⇒ quantized moments: Bohr magneton $\mu_B=e\hbar/2m_e=9.274\times10^{-24}$ J/T; nuclear magneton $\mu_N=\mu_B/1836=5.051\times10^{-27}$ J/T.
Spin breaks the classical ratio: $g_s\approx2$ (Dirac), measured $2.00232$ (QED, twelve digits) — and the spinning-ball picture requires surface speeds $\sim171c$ : spin is intrinsic, not mechanical rotation.
Precession frequencies follow by spending the mechanics article's $\omega=\gamma B$ : orbital 14 GHz/T, electron spin 28 GHz/T, proton 42.58 MHz/T — the dial settings of ESR/NMR/MRI.
Filled shells cancel: most atoms have small or zero net moment; unpaired electrons (and Hund's-rule bookkeeping) decide which atoms magnetize — the bridge to dia/para/ferro.
Bohr–van Leeuwen: classical equilibrium magnetization is identically zero — magnetism is quantum, full stop.
KEY IDEA: $\vec m=\gamma\vec J$ with $\gamma$ set by charge-to-mass ratio welds magnetism to angular momentum — quantizing one quantizes the other, in Bohr magnetons for electrons and 1836-fold smaller nuclear magnetons for nuclei.

The core insight

A magnetic moment is rigidly tied to angular momentum: $\vec m=\gamma\vec J$ with $\gamma=-e/2m_e$ for orbital motion and twice that for spin. Quantization of angular momentum in units of $\hbar$ therefore quantizes magnetic moments in units of the Bohr magneton $\mu_B$ .

Where this lives today

The gyromagnetic welding is now measured more precisely than almost anything humans measure. The electron's $g-2$ — the fractional excess of its spin moment over Dirac's prediction — is determined from single electrons in Penning traps to about one part in $10^{13}$ and matches the QED calculation (thousands of Feynman diagrams deep) — the sharpest agreement between theory and experiment in science, and a standing test: any new charged particle in nature's inventory would shift it. Its muon counterpart, measured at Fermilab through exactly this article's physics (spin precession versus cyclotron rotation of muons in a storage ring), spent the early 2020s as particle physics' most-watched anomaly before improved lattice-QCD theory largely closed the gap — the gyromagnetic ratio as frontier instrument. The applied descendants saturate technology. Atomic magnetometers read Larmor precession of optically pumped alkali vapors down to sub-femtotesla sensitivity — rivaling SQUIDs without cryogenics — and now map heart and brain fields in wearable form. NMR's chemical fingerprinting runs on parts-per-million shifts of equation (2)'s proton frequency caused by electron screening (this article's electrons quietly modifying this article's nuclear dial). And the moment menu drives materials design: engineering $g$ -factors and anisotropy in single-molecule magnets and rare-earth compounds is the current route to denser storage and, speculatively, molecular qubits — chemistry conducted in Bohr magnetons.

Origins · 1 minNiels Bohr and how this was discovered

Niels Bohr

1885–1962 · Copenhagen — the atom's legislator, and magnetism's first honest accountant

Bohr's doctoral dissertation (1911, on the electron theory of metals) contains a result he considered mostly negative: classical statistics makes the magnetic response of electron systems vanish — the theorem now shared with Hendrika van Leeuwen, who found it independently in 1919. Bohr understood the implication before anyone: magnetism was a standing contradiction inside classical physics, as sharp as blackbody radiation. Two years later his atomic model resolved both the atom's stability and, silently, magnetism's existence — quantized orbits mean quantized angular momentum means quantized moments, and the quantity $e\hbar/2m_e$ that falls out of his first orbit became magnetism's natural unit (the "Bohr magneton," computed in his framework by Procopiu in 1911 and named later). His institute in Copenhagen then midwifed the full quantum theory — Pauli's exclusion principle organizing which electrons pair and cancel, Heisenberg's exchange interaction explaining iron — so that the entire architecture of this chapter, from why most atoms are magnetically silent to why some conspire into permanent magnets, was drafted within walking distance of his office. The 1922 Nobel citation mentions atomic structure; the fine print includes every magnet ever understood.

History

1897

— Larmor derives the precession of orbital moments at eB/2m — the gyromagnetic ratio's first appearance, before the electron's role in atoms is clear.

1911

— Bohr's dissertation proves classical statistics yields zero magnetization; Procopiu computes the quantum moment unit; van Leeuwen's independent proof follows in 1919.

1913

— The Bohr model quantizes angular momentum in units of ħ: magnetic moments inherit the Bohr magneton.

1922

— Stern and Gerlach split the silver beam: moments are quantized in direction, one magneton at a time.

1925

— Uhlenbeck and Goudsmit propose electron spin; the anomalous factor g equals 2 is ratified by Dirac's equation in 1928.

1947

— Kusch measures g slightly above 2; the anomaly becomes QED's proving ground, today agreed to twelve digits.

The physical picture

Watch a single electron circle a nucleus and count what it carries. Once per orbit it transports its charge past any checkpoint: a current $I=-e/T$ around a loop of area $\pi r^2$ — a magnetic moment, by the last chapter's most-used formula $m=IA$ . The same circulation carries mass around: angular momentum $L=m_evr$ . Now compute the ratio and watch the geometry evaporate: $m/L=(-e/T)(\pi r^2)/(m_e v r)$ , and since $T=2\pi r/v$ , every factor of $r$ and $v$ cancels, leaving $-e/2m_e$ . Fast orbit or slow, wide or tight, circular or (as the derivation shows) any shape at all — the moment-to-angular-momentum exchange rate is a constant of nature's bookkeeping, fixed by charge and mass alone. Physics of this kind — a ratio into which the messy details cannot penetrate — is what earns a permanent name: the gyromagnetic ratio. Mind the sign, because it does daily work: the electron's charge is negative, so its moment points opposite its angular momentum — orbit counterclockwise (seen from above) and $\vec L$ points up while $\vec m$ points down. Every precession sense in ESR, every energy-level ordering in the Zeeman effect, inherits this flip.

Figure

The orbiting electron as a current loop — moment antiparallel to angular momentum, ratio fixed at minus e over 2m

The welding's consequences cascade. Because $L$ is quantized in $\hbar$ -sized steps (the quantum fact this article borrows, owned by the quantum track), moments arrive in $\mu_B$ -sized steps — magnetism has an atom. Because most electrons in most atoms pair up in filled shells with canceling $\vec L$ and spin, most atoms carry little or no net moment — the quiet default that the magnetization article's taxonomy needed: only unpaired electrons magnetize, and chemistry (which orbitals are half-filled) becomes destiny (which elements make magnets). And because moment rides on angular momentum, atomic moments precess in fields rather than aligning — the mechanics article's theorem, which this article can finally feed with numbers. One more cascade, subtler and deeper. If you try to build magnetism with classical mechanics alone — classical orbits, classical statistics, any Hamiltonian you like — a short theorem (Bohr's doctoral thesis, 1911; van Leeuwen, independently, 1919) proves the equilibrium magnetization of everything is exactly zero: in classical statistical mechanics, the magnetic field drops out of the partition function entirely. Diamagnetism, paramagnetism, iron itself — all classically impossible. The moments, their quantized sizes, and their survival against thermal chaos are quantum through and through; this chapter computes their consequences with classical field theory, but their existence is on quantum credit.

The gyromagnetic ladder

Derivation AThe gyromagnetic theorem — geometry drops out of m over L

The goal is the welding law in its general form: for any system of particles sharing one charge-to-mass ratio $q/m$ (an electron gas, a single orbiting electron, a rotating charged body with matched distributions),

\vec m=\frac{q}{2m}\,\vec L ,

with the circular orbit as warm-up and the general proof as the real content — because "no geometry in the ratio" is the theorem's entire point.

Step 1. The circular orbit, honestly. Period $T=2\pi r/v$ ; the circulating charge is a current $I=q/T=qv/2\pi r$ ; the moment is $m=I\cdot\pi r^2=qvr/2$ . The angular momentum is $L=mvr$ (mass $m$ , no confusion intended with the moment's symbol — context separates them). Ratio:

\frac{m_{\rm mag}}{L}=\frac{qvr/2}{mvr}=\frac{q}{2m}

— $r$ and $v$ gone, as promised. For an electron, $q=-e$ : moment antiparallel to $\vec L$ , magnitude $eL/2m_e$ .

Step 2. The general theorem. For particles at positions $\vec r_i$ , velocities $\vec v_i$ , the multipole article's moment (its volume dialect, with $\vec J=\sum_iq_i\vec v_i\,\delta^3(\vec r-\vec r_i)$ ) is

\vec m=\tfrac12\int\vec r\times\vec J\,\mathrm{d}\tau =\tfrac12\sum_i q_i\,\vec r_i\times\vec v_i .

If every particle shares one ratio $q_i/m_i=q/m$ , replace $q_i=(q/m)\,m_i$ :

\vec m=\frac{q}{2m}\sum_i m_i\,\vec r_i\times\vec v_i=\frac{q}{2m}\,\vec L .

Why the hypothesis matters — and when it fails: the step needed charge to ride mass in fixed proportion. A neutron (zero net charge, nonzero moment) violates it maximally: its quarks carry different $q/m$ , and no gyromagnetic shortcut applies — its moment must be computed from the actual internal currents, which is why $g_n=-3.83$ carries no classical explanation. Even for electrons the theorem covers only orbital motion; spin, below, breaks the ratio by its own factor of two.

Step 3. Time-averaging fine print. For non-circular orbits, $\vec r\times\vec v$ varies along the path, but its time average is what the mesoscopic moment means, and Kepler's second law (equal areas) makes the average work out identically — the swept-area reading of $\tfrac12\vec r\times\vec v$ is exactly the multipole article's fan-of-triangles, now swept in time. Verified: hydrogen Bohr orbit — $v=2.19\times10^6$ m/s, $r=a_0=5.29\times10^{-11}$ m: $T=1.52\times10^{-16}$ s, orbital current $I=e/T=1.05$ mA (a milliampere inside every hydrogen atom), moment $I\pi a_0^2=9.274\times10^{-24}$ J/T $=\mu_B$ exactly, and $L=m_eva_0=\hbar$ to six digits ✓ — the ratio checks as $-e/2m_e$ ; recomputed.

Go deeper · Derivation B · 2 minThe moment menu — Bohr magneton, nuclear magneton, and the g-factors

Derivation BThe moment menu — Bohr magneton, nuclear magneton, and the g-factors

The goal is the currency system of atomic magnetism: the natural unit for electrons, its nuclear counterpart, and the dimensionless $g$ -factors that price each angular-momentum type — including the honest accounting of spin's $g\approx2$ and the demolition of the spinning-ball picture. The strategy: feed quantized $L$ into the gyromagnetic theorem, then let measurement overrule classical expectation where it does.

Step 1. The Bohr magneton. The smallest orbital angular momentum scale is $\hbar$ . The theorem prices it:

\mu_B\equiv\frac{e\hbar}{2m_e} =\frac{1.602\times10^{-19}\times1.055\times10^{-34}}{2\times9.109\times10^{-31}} =9.274\times10^{-24}\ \mathrm{J/T} \qquad (1)

where: read the units as an energy scale per field — one Bohr magneton in one tesla is $9.3\times10^{-24}$ J $\approx58\ \mu$ eV $\approx0.67$ K: atomic magnetic energies are small compared to room temperature ( $k_BT\approx25$ meV), the single number behind paramagnetism's weakness and the Curie law's temperature war (the paramagnetism article's opening business).

Step 2. The nuclear magneton. Protons obey the same theorem with $m_p=1836\,m_e$ :

\mu_N=\frac{e\hbar}{2m_p}=\frac{\mu_B}{1836}=5.051\times10^{-27}\ \mathrm{J/T}.

Nuclear magnetism is electron magnetism taxed at 99.95%: same charge quantum, three orders heavier flywheel. That NMR and MRI work at all on these feeble moments is a triumph of resonance detection — and the suppression is why the magnetization of matter is an electron story, with nuclei as spectators (spectators worth imaging).

Step 3. The g-factor dictionary. Real moments are quoted as $\vec m=g\,\frac{q}{2m}\,\vec J$ , with $g$ measuring deviation from the orbital theorem. Orbital motion: $g_\ell=1$ by construction. Electron spin: $g_s\approx2$ — Dirac's relativistic equation predicts exactly 2 with no model of internal structure, and experiment refines it to $2.002319\ldots$ , the QED-computed anomaly agreed to twelve digits. Proton: $g_p=5.586$ ; neutron: $g_n=-3.826$ — wildly non-Dirac values that announced composite structure decades before quarks explained them.

Step 4. Kill the spinning ball. Could spin be literal rotation? Model the electron as a uniform ball of the classical electron radius $r_e=2.82\times10^{-15}$ m spinning with $L=\hbar/2$ : the required angular velocity gives an equatorial speed

v=\frac{(\hbar/2)\,r_e}{\tfrac25 m_er_e^2}=\frac{5\hbar}{4m_er_e} \approx5.1\times10^{10}\ \mathrm{m/s}\approx171\,c

— faster than light by two orders of magnitude, and shrinking the ball only makes it worse. Spin is intrinsic angular momentum: it enters the theorems of this chapter (precession, Stern–Gerlach forces, magnetization) exactly like $\vec L$ , carries a magnetic moment exactly as advertised, and has no classical mechanical picture whatsoever. The honest posture: use $\vec m=\gamma\vec J$ with measured $\gamma$ , and let the Dirac equation own the why. Verified: constants recomputed from CODATA inputs ( $\mu_B$ , $\mu_N$ , ratio 1836.15 ✓); $g_s$ anomaly value per the electron $g-2$ literature; the 171c figure recomputed from $r_e$ and $\hbar/2$ ✓.

With $\gamma$ now priced, the mechanics article's precession law pays out in laboratory frequencies. Spending $\omega=|\gamma|B$ :

f_{\rm orb}=\frac{eB}{4\pi m_e}=14.0\ \mathrm{GHz/T}, \qquad f_{\rm spin}=\frac{g_s\mu_BB}{h}=28.0\ \mathrm{GHz/T}, \qquad f_{\rm proton}=\frac{g_p\mu_NB}{h}=42.58\ \mathrm{MHz/T}. \qquad (2)

where: the first is Larmor's original 1897 frequency (half the cyclotron frequency — the factor $\tfrac12$ from the gyromagnetic $2m$ ), the second runs every electron-spin-resonance spectrometer, the third every NMR machine and MRI scanner. One law, three coins, and the entire radio-frequency architecture of magnetic resonance is equation (2): pick the moment, set the field, tune the dial.

Go deeper · Derivation C · 2 minBohr and van Leeuwen — classical physics forbids magnetism

Derivation CBohr and van Leeuwen — classical physics forbids magnetism

The goal is the chapter's deepest honesty flag, stated as the theorem it is: in classical statistical mechanics, the equilibrium magnetization of any system of charged particles is identically zero — no diamagnetism, no paramagnetism, no iron. The strategy is a sketch with the essential move shown honestly and the technical scaffolding (phase-space measure theory) deferred to statistical mechanics, where the theorem properly lives.

Step 1. Where B enters classical mechanics. The magnetic field enters a classical Hamiltonian in exactly one way — through the vector potential in the kinetic term (the canonical-momentum structure the vector-potential article established):

H=\sum_i\frac{\big(\vec p_i-q_i\vec A(\vec r_i)\big)^2}{2m_i}+V(\{\vec r_i\}) .

Step 2. The field vanishes from the partition function. Equilibrium physics flows from $Z=\int e^{-H/k_BT}\,\prod_i\mathrm{d}^3r_i\,\mathrm{d}^3p_i$ . In each momentum integral, substitute $\vec p_i'=\vec p_i-q_i\vec A(\vec r_i)$ — at fixed positions this is a rigid shift of the integration variable over all of momentum space, and shifts leave infinite integrals invariant:

\int_{-\infty}^{\infty}e^{-(\vec p-q\vec A)^2/2mk_BT}\,\mathrm{d}^3p =\int_{-\infty}^{\infty}e^{-p'^2/2mk_BT}\,\mathrm{d}^3p' .

Every trace of $\vec A$ — hence of $\vec B$ — disappears: $Z$ is field-independent, so the free energy is field-independent, so $M=-\partial F/\partial B=0$ . Exactly, at every temperature, for every classical charge system.

Step 3. Read the wreckage. The theorem is not a technicality; it kills each classical magnetism story in a recognizable way. Classical orbital diamagnetism? The circulating-orbit currents that seem to oppose the field are exactly canceled, in equilibrium, by the skipping trajectories of particles reflecting off the container walls — the bulk and the boundary conspire to zero. Classical paramagnetism from permanent moments? Classically there are no permanent moments: a classical orbit's $m$ can take any value, and Step 2 zeroes the average anyway. What quantum mechanics changes is structural: angular momentum (hence $m$ ) is quantized, energy levels are discrete, the momentum integral becomes a sum that the shift argument cannot touch — and magnetization is reborn. Landau's quantum treatment of free electrons (1930) recovers diamagnetism precisely where the classical theorem forbade it. Why this article hosts the theorem: it is the license for everything the next three articles do. Diamagnetism, paramagnetism, and ferromagnetism will each be derived with semiclassical pictures — orbits speeding up, moments Boltzmann-aligning — and every one of those pictures secretly leans on quantization (fixed moments, discrete states) at a critical step. Bohr–van Leeuwen marks exactly where the classical crutch would snap. *Verified: statement and shift argument per the standard literature (Bohr 1911 dissertation; van Leeuwen 1919; Van Vleck's classic discussion); consistency check — the argument requires unbounded momentum integration and classical statistics, and both loopholes (quantized levels; degenerate quantum gases) are exactly where real magnetism lives ✓.*

Limiting cases

case	result	meaning
single orbital electron, $L=\hbar$	$m=\mu_B$	hydrogen's Bohr orbit; the currency unit
spin only	$m\approx\mu_B$ again ( $g_s\approx2$ , $J=\hbar/2$ )	the two factors of 2 cancel — spin moment ≈ orbital unit
filled shell	$\sum\vec L=\sum\vec S=0$ , $m=0$	noble gases, most ions: magnetically silent → diamagnetism only
nucleus	$m\sim\mu_N=\mu_B/1836$	NMR's feeble, precious signal
iron atom in metal	$2.2\,\mu_B$ (spin-dominated, orbital quenched)	the magnetization article's input, now explained in kind
classical limit ( $\hbar\to0$ )	$\mu_B\to0$ and Bohr–van Leeuwen ⇒ $M\equiv0$	magnetism has no classical limit

Common trapSpin is not a spinning ball, and the moment is not along L for an electron

Two sign-and-picture errors that propagate silently. First: the electron's moment points opposite its angular momentum ( $q=-e$ in the gyromagnetic theorem) — draw $\vec m$ parallel to $\vec L$ and every Zeeman ordering, precession sense, and hyperfine sign downstream flips. Second: "spin" invites a mechanical picture that fails quantitatively and conceptually — a ball of any admissible classical radius must rotate superluminally (171c at the classical electron radius) to carry $\hbar/2$ , and no refinement rescues it; moreover the spin $g$ -factor is 2, not the orbital 1, so even the ratio refuses the mechanical analogy. Treat spin as what it is: intrinsic angular momentum with a measured gyromagnetic ratio, fully legitimate input to every theorem of this chapter (precession, forces, magnetization), with its explanation living in relativistic quantum mechanics. Related bookkeeping trap: $g$ -factors multiply different magnetons — $g_p=5.586$ prices $\mu_N$ , not $\mu_B$ ; forgetting which magneton underlies which $g$ produces thousand-fold errors in resonance predictions.

Worked examples

1 · The milliampere inside hydrogen (confidence builder)

For hydrogen's ground Bohr orbit ( $r=a_0=5.29\times10^{-11}$ m, $v=2.19\times10^6$ m/s), find the orbital current, the magnetic moment, and the field the orbit produces at the nucleus.

Step 1. Period and current. $T=2\pi a_0/v=1.52\times10^{-16}$ s; the electron passes any checkpoint $6.6\times10^{15}$ times per second, a current $I=e/T$ = 1.05 mA. A macroscopic current — the scale of an LED's — circulating inside every hydrogen atom.

Step 2. The moment. $m=I\pi a_0^2=1.05\times10^{-3}\times\pi\times(5.29\times10^{-11})^2=9.27\times10^{-24}$ J/T = exactly $\mu_B$ — as it must be: $L=m_eva_0=\hbar$ to six digits, and the gyromagnetic theorem prices $\hbar$ at one Bohr magneton.

Step 3. Field at the nucleus. The loop-center formula (Biot–Savart article's calibration case): $B=\mu_0I/2a_0=4\pi\times10^{-7}\times1.05\times10^{-3}/(2\times5.29\times10^{-11})$ = 12.5 T — the proton sits in a field stronger than most superconducting magnets, supplied by its own electron. Orbit-nucleus fields of this scale drive fine structure and the orbital part of hyperfine interactions; the 21-cm contact story (dipole article) is the spin part of the same conversation.

2 · Designing the dials: ESR versus NMR at one magnet (exam level)

A spectrometer magnet provides $B=0.35$ T. At what frequencies do electron spins and proton spins resonate, and what does the ratio say about the instruments?

Step 1. Electron spin. $f_e=g_s\mu_BB/h=2.00232\times9.274\times10^{-24}\times0.35/(6.626\times10^{-34})$ = 9.81 GHz — X-band microwaves, waveguides and cavities: the standard ESR spectrometer configuration, dictated entirely by equation (2).

Step 2. Protons, same field. $f_p=g_p\mu_NB/h=5.586\times5.051\times10^{-27}\times0.35/(6.626\times10^{-34})$ = 14.9 MHz — shortwave radio, coils and capacitors instead of waveguides.

Step 3. The ratio and its anatomy. $f_e/f_p=\mu_e/\mu_p=658$ : one factor $\sim1836$ from the mass-suppressed magneton, partially clawed back by the proton's anomalous $g_p/g_s\approx2.8$ . This single dimensionless number is why ESR and NMR are different engineering disciplines sharing one physics — and why dynamic nuclear polarization (borrowing the electron's 658-fold larger polarization to boost NMR signals) is worth its considerable trouble.

Step 4. Sanity anchor. Scale the proton line to a clinical 3 T scanner: $42.58\times3=127.7$ MHz — the mechanics article's MRI number, now derived from $g_p\mu_N$ rather than quoted: the chapter's dials all come from this article's coins.

Going further

References. Griffiths, Introduction to Electrodynamics, 4th ed., §6.1 · Jackson, Classical Electrodynamics, 3rd ed., §5.6–5.7 · Purcell & Morin, Electricity and Magnetism, 3rd ed., §11.2 · Van Vleck, The Theory of Electric and Magnetic Susceptibilities (1932), Ch. 4 (Bohr–van Leeuwen)

Provenance. Equations follow Griffiths §6.1 and Jackson §5.6–5.7. A verifier should recompute $\mu_B=e\hbar/2m_e=9.274\times10^{-24}\,\mathrm{J/T}$ and $\mu_N=e\hbar/2m_p=5.051\times10^{-27}\,\mathrm{J/T}$ from CODATA constants, check $g_s\approx2.0023$ , the mass ratio $m_p/m_e\approx1836$ , and verify the orbital-moment derivation $|\mathbf{m}|=eL/2m_e$ and the Larmor frequency $\omega_L=eB/2m_e$ . (Scope note: precession dynamics owned by the forces-and-torques article and spent here with atomic $\gamma$ values; Stern–Gerlach worked numbers live there too. Worked numbers: Bohr orbit $T=1.52\times10^{-16}$ s, $I=1.05$ mA, $m=\mu_B$ , $B_{\rm nucleus}=12.5$ T; dials 14.0/28.0 GHz/T and 42.58 MHz/T, 9.81 GHz and 14.9 MHz at 0.35 T, ratio 658; spinning-ball 171c — recomputed.)

Builds on: Magnetization · Forces and Torques on Magnetic Dipoles · Magnetic Dipole and Magnetic Moment · Multipole Expansion of the Vector Potential

Leads to: Diamagnetism · Paramagnetism · Ferromagnetism · Curie Law and Curie Temperature

If the gyromagnetic ratio has no geometry in it, why do different atoms have such different moments?

The ratio is universal; the angular momentum budget is not. An atom's moment is $\gamma$ times its net $\vec J$ , and chemistry sets the net: how many electrons sit unpaired (Pauli pairing cancels both spin and orbital contributions two by two), how their spins add (Hund's rules maximize alignment within a shell), and how much orbital momentum survives the crystal environment (often little — quenching). Helium: filled shell, zero. Sodium: one unpaired s-electron, essentially one spin moment. Iron in metal: roughly $2.2\,\mu_B$ of mostly spin. Gadolinium: seven unpaired 4f spins, $\sim8\,\mu_B$ . Same exchange rate everywhere; different balances in the account.

Why does orbital angular momentum get "quenched" in solids while spin survives?

Because the two couple to the environment differently. Orbital motion is charge circulating in space, and the crystal's electric fields grip it directly: the lattice environment selects real standing-wave orbitals (combinations with zero net circulation) as energy eigenstates, so $\langle\vec L\rangle\approx0$ — the circulation is frozen out, "quenched." Spin has no spatial wavefunction to grip; crystal fields touch it only indirectly through the relativistic spin–orbit coupling, weak in light elements. Hence iron-group magnetism is spin magnetism ( $g\approx2$ measured in ferromagnetic resonance confirms it), while rare earths — whose 4f orbitals hide deep inside filled shells, shielded from the lattice — keep their orbital moments and show the full $\vec L+\vec S$ complexity, plus the strong anisotropy that makes them premium magnet material.

How can the neutron have a magnetic moment with zero charge — doesn't the gyromagnetic theorem forbid it?

The theorem's hypothesis — one charge-to-mass ratio throughout — is exactly what the neutron violates, so no contradiction: the theorem simply doesn't apply, and the moment must come from internal structure. The neutron's quarks (two down at $-e/3$ , one up at $+2e/3$ ) circulate and spin with different charges riding different mass distributions; the net charge integrates to zero while the net current circulation does not. Its measured $g_n=-3.826$ (moment anti-parallel to spin) was, historically, the cleanest early evidence that "elementary" nucleons are composite — a pointlike Dirac neutral particle would have $g=0$ . The same logic in reverse makes the electron's $g=2.00232$ so powerful: any internal structure would show up as a bigger anomaly, and eleven digits of QED agreement say there is none down to extraordinary scales.

Is the Bohr model's success in example 1 a coincidence, given that the model is wrong?

Half-coincidence, half-structure. The model's picture (a literal circular orbit) is wrong: quantum hydrogen's ground state is an s-state with zero orbital angular momentum and zero orbital moment — the real ground-state magnetism of hydrogen is pure spin. But the ratio physics is exact: the gyromagnetic theorem is a theorem about any circulation, quantum included ( $\hat{\vec m}=-\frac{e}{2m_e}\hat{\vec L}$ as operators), so any state with $L_z=\hbar$ — say the real $2p$ , $m_\ell=1$ state — carries exactly one Bohr magneton of orbital moment. The Bohr model computes correct answers whenever the answer depends only on angular-momentum bookkeeping and not on orbital geometry; $\mu_B$ is the premier example, which is why the unit survived the model by a century.

Where's the loophole in Bohr-van Leeuwen — what exactly does quantum mechanics change?

The theorem's engine is the momentum-shift: $\vec A$ disappears from a classical partition function because $\int\mathrm{d}^3p$ over all momentum space is shift-invariant. Quantum mechanics breaks the engine in two ways. First, the trace over states is not a free momentum integral: energy levels are discrete (Landau levels, for free electrons in a field — spaced by $\hbar\omega_c$ , field-dependent), and the field cannot be shifted out of a discrete spectrum; Landau's 1930 calculation extracts precisely the diamagnetism classical physics forbade. Second, permanent quantized moments exist (spin, fixed atomic $J$ ), whose orientational statistics give paramagnetism — classically, moments could take any value and the average still vanished. Both loopholes are $\hbar$ -sized: send $\hbar\to0$ and the level spacing closes, the moments vanish, and van Leeuwen's zero returns. Magnetism is, rigorously, an observable quantum effect at everyday scale.

Why are magnetic energies so small compared to electric ones inside atoms, when the moments seem substantial?

Count the powers of the fine-structure constant. Atomic electric energies are $\sim e^2/4\pi\varepsilon_0a_0\approx27$ eV. A Bohr magneton in an atomic-scale field ( $\sim10$ T, example 1's own orbital field) is $\mu_BB\sim10^{-3}$ eV — ten-thousandfold smaller: magnetic interactions are relativistic corrections, suppressed by $v^2/c^2\sim\alpha^2\approx(1/137)^2$ relative to electrostatics. This ordering runs the whole subject: chemistry (electric) sets structures and pairings, magnetism decorates them; fine and hyperfine structure are small splittings atop Coulomb levels; and — the ferromagnetism article's crucial plot point — iron's strong alignment cannot come from magnetic dipole–dipole coupling (only $\sim10^{-4}$ eV $\sim1$ K) but rides the exchange interaction, an electrostatic-plus-Pauli quantum effect wearing magnetic clothes, at $\sim0.1$ eV $\sim1000$ K. Even ferromagnetism's strength is electric energy, steered by spin.

Do these moment formulas matter for anything at everyday scale, or only in spectrometers?

They price everyday magnetism directly. The magnetization article's iron benchmark — $M_s=1.7\times10^6$ A/m — is literally "count atoms, multiply by $2.2\,\mu_B$ ": your refrigerator magnet's strength is a Bohr-magneton census. Compass function: Earth's field exerts torque on the needle's net moment, itself $\sim10^{23}$ aligned atomic magnetons. Medical imaging: MRI's entire signal is the thermal-equilibrium excess of proton moments — about 10 per million at 3 T and body temperature, the ratio $\mu_NB/k_BT$ from this article's numbers — which is why scanners crave high field. Even the hard drive: each bit's stability against thermal flips is an energy barrier counted in magnetons times anisotropy, the "superparamagnetic limit" that ended one recording technology and forced the next. The magneton is a household unit that households have never heard of.