Magnetic Moment and the Gyromagnetic Ratio

The dictionary is working infrastructure. Every MRI scanner is a 42.577-MHz-per-tesla proton radio; parts-per-million chemical shifts of that Larmor line are how NMR chemists read molecular structure. ESR g-factors near 2 diagnose defects in semiconductors; atomic magnetometers track alkali-atom precession to femtotesla sensitivity. And the anomaly is a two-front precision war: the electron's $a_e$ (Penning traps, $10^{-13}$) doubles as the best route to $\alpha$, while the muon's $g-2$, clocked at Fermilab against cyclotron rotation in a storage ring, is scrutinized for physics beyond the Standard Model. A century after the factor of two embarrassed everyone, the tenth decimal is where new physics is hunted.

Picture a charge $q$ circulating on a track: a current bounding an area is a magnet. Spin it twice as fast and the current doubles — the moment grows in lockstep with $\mathbf{L}$. That lockstep is the whole content of $\boldsymbol{\mu}\propto\mathbf{L}$; the constant carries a factor $\tfrac12$ because $L=mvr$ weighs one power of radius while $\mu=IA$ weighs area against a current falling as $1/r$. The geometry cancels; $q/2m$ survives for any orbit.

Spin wrecks the picture, informatively. A spinning charged ball, analysed shell by shell with the same loop argument, predicts $g_s=1$; nature says $g_s\approx2.00232$ — twice as magnetic per unit of angular momentum as any classical rotation allows. So spin is not mechanical rotation: it is the intrinsic attribute Spin Angular Momentum established, with a coupling fixed by relativistic quantum mechanics rather than by a radius times a velocity. The second half of the picture is dynamical. A moment in a uniform field does not swing into alignment like a compass needle: $\boldsymbol{\mu}$ is chained to $\hat{\mathbf J}$, and torque on an angular momentum changes direction, not magnitude. $\langle\hat{\mathbf J}\rangle$ sweeps a cone about $\mathbf{B}$ at the Larmor frequency $\omega_L$, forever, absent dissipation. Every magnetic-resonance technology listens to that rotation.

Everything in this article hangs on one operator equation. It answers the question: given that a system carries angular momentum $\hat{\mathbf J}$ — orbital, spin, or total — how strong a magnet is it? The relation asserts the answer is always a rigid proportionality, with the system's identity compressed into a single pure number:

where: $\boldsymbol{\mu}$ is the moment operator ($\hat{\mathbf J}$ rescaled, algebra and all), $\gamma=q/2m$ the classical gyromagnetic ratio carrying the units, and $g$ the dimensionless g-factor — the quantum correction to the classical bargain. Mind the sign: for the electron $q=-e<0$, so $\gamma<0$ and the moment points opposite to the angular momentum — a spin-up electron is a moment-down magnet. In a field, the coupling is the classical dipole energy promoted to an operator. With $\mathbf{B}=B\hat{\mathbf z}$,

where: the Hamiltonian is a rescaled $\hat J_z$ — its eigenstates are exactly the ladder states $|j,m\rangle$ of Ladder Operators and the Angular Momentum Spectrum. The spectrum of magnetism in a uniform field is the $m$-ladder, relabeled in energy units. The measurable component along the field inherits the ladder's discreteness:

where: $\mu_z$ takes exactly $2j+1$ equally spaced values and nothing in between — the field selects among allowed orientations, never tilts continuously. This is space quantization; The Stern-Gerlach Experiment is its photograph — a beam in a field gradient splits into $2j+1$ spots, one per $\mu_z$.

The table is the article in miniature: orbital motion obeys the classical rate, the electron doubles it, and the nucleons scatter wildly — flags of composite structure. A neutral particle with a moment (the neutron) is by itself proof of charged constituents circulating inside.

Insert the electron's numbers once and you get the currency of magnetic measurement. Orbital $\mu_z=-(e/2m_e)\,m\hbar$, so the natural unit is everything except the integer $m$:

where: $e$ is the (positive) elementary charge and $m_e$ the electron mass (value verified from CODATA). Orbital Angular Momentum Operators quoted this number; here it earns its derivation — one $\hbar$ of orbital angular momentum, converted at the rate $e/2m_e$. Conversions: $13.996$ GHz/T as a frequency ($\mu_B/h$), $0.6717$ K/T as a temperature ($\mu_B/k_B$). The proton mass gives the nuclear magneton $\mu_N=e\hbar/2m_p=5.051\times10^{-27}$ J/T — down from $\mu_B$ by exactly $m_p/m_e=1836.15$. Nuclear magnetism is three orders feebler than electronic, which is why NMR needs strong fields and why everyday magnetism is overwhelmingly electronic. The electron's spin moment, in this currency, is

where: $\hat{\mathbf S}=\tfrac\hbar2\boldsymbol{\sigma}$ for spin-$\tfrac12$ (anatomy: Spin-1/2 and the Pauli Matrices); the minus sign is the electron's charge again. Numerically $|\mu_z|=\tfrac{g_s}{2}\mu_B\approx1.001\,\mu_B$: half the angular momentum, one full magneton — an accident ($g_s\cdot\tfrac12\approx1$) that will matter historically in the next article.

Why $g_s\approx2$? Three tiers, each owned elsewhere and stated here. Experiment (1925, spectral doublets and the anomalous Zeeman effect): the spin moment is twice too large for its angular momentum. Dirac's equation (1928): $g_s=2$ exactly for a point spin-$\tfrac12$ particle — relativity's signature, derivation owned by the relativistic track. Then QED dresses the electron in virtual photons, shifting the value by the anomaly

where: $\alpha\approx1/137.04$ is the fine-structure constant; Schwinger's 1948 one-loop term $\alpha/2\pi=1.1614\times10^{-3}$ already sits within $0.2\%$ of the full anomaly. Measured (one electron in a Penning trap) and computed $g_s$ agree to about a part in $10^{12}$ — the most precise confrontation of theory and experiment in science. The moral: the g-factor is a fingerprint. $1$ says classical loop; $2$ says relativistic point spin; $5.586$ (proton) or $-3.826$ (neutron) says composite — look inside.

A static moment in a static field does something dynamical — and the dynamics is the measurement channel. The equation of motion comes from the expectation-value engine of Ehrenfest's Theorem, applied to $\hat{\mathbf J}$ itself.

Spin Angular Momentum solved the spin-$\tfrac12$ case explicitly — its 28.02 GHz/T electron figure is the $j=\tfrac12$ instance of this vector law: $g_s\mu_B/h=2.00232\times13.996=28.025$ GHz/T. For the proton, $\gamma_p=2.6752\times10^{8}$ rad s$^{-1}$T$^{-1}$ ($42.577$ MHz/T) and $\mu_p=2.793\,\mu_N$; the anomaly $\kappa_p=(g_p-2)/2=1.79$ above the Dirac value flags quark substructure.

An atomic electron usually carries both $\hat{\mathbf L}$ ($g_L=1$) and $\hat{\mathbf S}$ ($g_s\approx2$). The total moment $\boldsymbol{\mu}=-\tfrac{\mu_B}{\hbar}(\hat{\mathbf L}+2\hat{\mathbf S})$ is then not parallel to $\hat{\mathbf J}=\hat{\mathbf L}+\hat{\mathbf S}$ — the double spin weighting skews it. Spin–orbit coupling spins $\hat{\mathbf L}$ and $\hat{\mathbf S}$ rapidly about $\hat{\mathbf J}$; over a weak field's slow timescale the skewed part averages away, leaving only the $\hat{\mathbf J}$-projection — the Landé g-factor:

where: $l,s,j$ label the level, and the formula assumes $g_L=1$, $g_s=2$. The derivation — why "project onto $\hat{\mathbf J}$" is a theorem about vector operators, not a cartoon — is owned by Projection Theorem for Vector Operators and deliberately not reproduced here; this article states and stress-tests the result. The vector-model picture you can hold now:

Stress-test it where the answer is known — all four cases verified exactly. Pure spin ($l=0$, $j=s=\tfrac12$; the $^2S_{1/2}$ ground state of hydrogen or silver): $g_J=2$. Pure orbit ($s=0$): $g_J=1$ for any $l$. Sodium's doublet levels: $^2P_{3/2}\to\tfrac43$, $^2P_{1/2}\to\tfrac23$. Three g-factors in one atom ($2,\tfrac43,\tfrac23$) is exactly why the "anomalous" Zeeman effect looked anomalous — each level splits at its own $\Delta E=g_J\mu_B m_j B$. (Quantitative Zeeman theory belongs to the hydrogen chapter.)

Example 1 — the MRI chain (confidence builder). Task: a hospital scanner runs at $B=3$ T. Find the proton Larmor frequency, and the fraction of protons that actually contribute signal at body temperature $T=310$ K.

Step 1. Frequency. The proton entry is $\gamma_p/2\pi=42.577$ MHz/T, so at 3 T the scanner's radio dial sits at $\nu = 42.577\times3 = \boldsymbol{127.73\ \mathrm{MHz}}$.

Step 2. Zeeman gap. $\Delta E = h\nu = 8.46\times10^{-26}$ J $\approx 0.53\ \mu$eV — eight orders below bond energies; the field barely perturbs the matter.

Step 3. Polarization: $P=\tanh(\Delta E/2k_BT)$; at 310 K the argument is tiny and $\tanh$ is its argument:

Step 4. Consequence: only $\sim$10 protons per million are net-aligned. A 3 T scanner listens to a 127.73 MHz hum from a $10^{-5}$ polarization — which is why higher $B$ (better polarization and higher pickup frequency) is worth billions.

Example 2 — how big is a Bohr magneton, really? (exam level). Task: compare the magnetic energy of one Bohr magneton in a strong lab field ($B=1$ T) with thermal energy at room temperature, and find the electron-spin resonance frequency in that field.

Step 1. Magnetic scale: $\mu_B B = 9.274\times10^{-24}$ J $= \boldsymbol{57.88\ \mu\mathrm{eV}}$ at 1 T.

Step 2. Thermal scale: $k_BT = 25.85$ meV at 300 K. The ratio is

Thermal agitation out-shouts a tesla by a factor of 450 — why paramagnetism is weak and Curie-like ($\chi\propto\mu_B^2/k_BT$), and why quantum magnetic devices run in cryostats (at 20 mK the ratio exceeds 30 and spins freeze into the ground rung).

Step 3. Spin resonance. The 1 T electron-spin splitting is $g_s\mu_B B = \boldsymbol{115.9\ \mu\mathrm{eV}}$: a photon of $\nu = g_s\mu_B B/h = \boldsymbol{28.025}$ GHz — microwave territory, every ESR spectrometer's dial. Cross-check: $28.025/13.996 = 2.0023 = g_s$ — the dictionary's factor of two is the anomalous g-factor in plain sight.

References. Griffiths, 3rd ed., §4.4 · Sakurai & Napolitano, 2nd ed., §1.1, §3.2 · Shankar, 2nd ed., Ch. 14 · Cohen-Tannoudji, Diu & Laloë, Vol. 1, Ch. IV & Complement · Townsend, Ch. 1 · CODATA 2018 recommended values (Bohr magneton, electron g-factor).

Provenance. Equations are standard textbook results: defining relation and orbital derivation from Griffiths §4.4; Larmor precession from Sakurai §3.2; Landé projection from Cohen-Tannoudji. Numbers to verify: Bohr magneton $\mu_B\approx9.274\times10^{-24}\,\mathrm{J/T}$, electron anomaly $a_e\approx1.15965\times10^{-3}$ and $g_s\approx2.00232$ (CODATA 2018), Schwinger term $\alpha/2\pi$, proton $g_p\approx5.586$, and mass ratio $m_p/m_e\approx1836$. A verifier should recompute $g_J$ for sample states (e.g. $^2S_{1/2}\to2$, $^2P_{1/2}\to2/3$) and check the current-loop algebra cancels geometry to give $g_L=1$. For this edition every numeric claim was machine-verified (sympy/numpy, 42 checks): the loop algebra symbolically, $\mu_B$/$\mu_N$/all four dictionary frequencies from CODATA, all four Landé special cases exactly, the precession equations on explicit $j=\tfrac12$ and $j=1$ matrices, both worked-example chains. Ownership note: the base carries a Landé projection-theorem derivation; per the chapter map it belongs to Projection Theorem for Vector Operators and is deliberately not reproduced — this article states the formula and verifies its special cases.

Upstream: Orbital Angular Momentum Operators ($\hat{\mathbf L}$ machinery, first sight of $\mu_B$) · Angular Momentum Commutation Relations (the algebra the precession feeds on) · Spin Angular Momentum (the intrinsic carrier, two-state precession solve) · Ladder Operators and the Angular Momentum Spectrum (the $m$-census). Downstream: The Stern-Gerlach Experiment (the gradient force that photographs $\mu_z$) · Spin-1/2 and the Pauli Matrices ($\boldsymbol{\mu}$ in $\boldsymbol{\sigma}$ language) · Projection Theorem for Vector Operators (the Landé derivation, owed and paid there) · the hydrogen chapter's Zeeman treatment.