Orbital Magnetic Moments and Space Quantization

Electron paramagnetic resonance (EPR) spectroscopy directly measures $g_s$ and its shifts from the free-electron value to identify radical intermediates in enzymatic mechanisms, characterize transition-metal centers in metalloproteins (e.g., Cu(II), Mn(II), Fe(III) sites), and probe defect states in semiconductors and battery materials; the measured $g$-factor deviations from $2.0023$ report directly on spin–orbit mixing with nearby orbital states, a refinement built on exactly the $g_l$ vs. $g_s$ distinction developed here. Nuclear magnetic resonance (NMR), while governed by nuclear rather than electron moments, uses the identical Bohr-magneton-style scaling (with the nuclear magneton $\mu_N = e\hbar/2m_p$, a factor of roughly 1836 smaller because of the proton mass) and the same space-quantization logic for nuclear spin states in a field. Magnetochemistry of coordination complexes uses the effective magnetic moment $\mu_{\mathrm{eff}} = g_J\sqrt{J(J+1)}\mu_B$ (or, for many transition-metal ions where orbital angular momentum is quenched by the ligand field, the spin-only formula $\mu_{\mathrm{eff}} \approx 2\sqrt{S(S+1)}\mu_B$) to determine the number of unpaired electrons and hence oxidation state and spin state (high-spin vs. low-spin) from bulk susceptibility (SQUID) measurements — a routine diagnostic in inorganic and bioinorganic chemistry. Molecular magnets and single-molecule magnets, of interest for high-density data storage, are engineered by tuning total spin $S$ and magnetic anisotropy so that the space-quantized $M_S$ sublevels of a large-spin ground state develop a substantial energy barrier to reorientation. Finally, the theoretical machinery here — quantized $L_z$ and $S_z$, Bohr magneton scaling, and $g$-factors — is the direct precursor to the Zeeman Hamiltonian used throughout astrophysics (stellar magnetic field measurements via Zeeman line splitting) and in modern quantum sensing devices such as atomic magnetometers.

Ask the tutor: How does the Landé $g$-factor formula get derived from projecting $\vec L$ and $\vec S$ onto the total angular momentum $\vec J$?

Picture an electron in a hydrogenic orbital with orbital angular momentum quantum number $l$. Classically, an electron of charge $-e$ moving in a loop of current $I$ encloses area $A$ and generates a magnetic dipole moment $\mu = IA$, directed along the loop's axis by the right-hand rule (adjusted for the negative charge). If the electron of mass $m_e$ orbits with angular momentum $L = m_e v r$, a short classical calculation (below) shows

The moment is antiparallel to $\vec{L}$ because the electron's charge is negative — the orbital current circulates opposite to the sense in which the position vector sweeps out $\vec{L}$. Quantum mechanically $\vec{L}$ is not a classical vector with a definite direction; it is an operator triple $(\hat{L}_x, \hat{L}_y, \hat{L}_z)$ obeying $[\hat{L}_x,\hat{L}_y] = i\hbar\hat{L}_z$ (cyclic), so that $\hat{L}^2$ and only one Cartesian component (conventionally $\hat{L}_z$) can be simultaneously diagonalized. The joint eigenstates are the spherical harmonics $Y_l^{m_l}(\theta,\phi)$, satisfying

Because $\vec{\mu}_L \propto \vec{L}$, the magnetic moment operator shares these eigenstates. The magnitude of $\vec{\mu}_L$ is fixed by $l$, but only its $z$-projection is sharp; the $x$- and $y$-components are fundamentally indeterminate in a state of definite $m_l$. Geometrically, this is depicted as $\vec{\mu}$ (or $\vec{L}$) lying on a cone of half-angle $\theta$ about the $z$-axis, with

This is space quantization: not that the vector points in only a few directions in some absolute sense, but that its projection onto any chosen measurement axis takes only $2l+1$ discrete values, and correspondingly the polar angle relative to that axis takes only $2l+1$ discrete values.

The same logic, with a twist, applies to electron spin. Spin angular momentum $\hat{S}$ obeys the identical commutation algebra, but $s=1/2$ always, giving only two eigenvalues $m_s = \pm\tfrac12$. There is no classical current loop to derive the spin moment from — it is a purely relativistic/quantum property, correctly predicted only by the Dirac equation — but empirically and theoretically the spin magnetic moment obeys the same form with a different, larger proportionality constant, the spin g-factor $g_s \approx 2$.

Classical magnetic moment of orbiting charge. For a particle of charge $q$ and mass $m$ moving in a circular orbit,

where: $\vec{\mu}$ is the magnetic dipole moment, $q$ the particle's charge (signed), $m$ its mass, $\vec{L} = \vec{r}\times\vec{p}$ its orbital angular momentum about the same origin. Reading: the magnetic moment is always parallel or antiparallel to $\vec{L}$, with the sign of $q$ fixing the relative orientation, and the ratio $\mu/L = q/2m$ (the classical gyromagnetic ratio) is identical for every orbit regardless of size, shape, or speed — a consequence of the definitions of $I$, $A$, and $L$ canceling all orbit-specific parameters.

Orbital angular momentum quantization. From the eigenvalue equations for $\hat L^2$ and $\hat L_z$ (already assumed known),

where: $l$ is the orbital angular momentum (azimuthal) quantum number, $m_l$ the magnetic quantum number indexing the $2l+1$ orientations. Reading: $L$ itself (the magnitude) is not achievable as a $z$-projection except in the trivial case $l=0$ — even the maximal projection $m_l=l$ gives $L_z = l\hbar < \sqrt{l(l+1)}\hbar = L$, so the vector can never point exactly along $z$. This residual "tilt" is intrinsic and unavoidable, not a measurement artifact.

Orbital magnetic moment operator and the Bohr magneton. Substituting $q=-e$, $m=m_e$ into the classical relation and quantizing,

where: $e>0$ is the elementary charge, $\mu_B$ the Bohr magneton (SI value $9.2740\times10^{-24}\ \mathrm{J\,T^{-1}}$, equivalently $5.7884\times10^{-5}\ \mathrm{eV\,T^{-1}}$), $g_l$ the orbital $g$-factor. Reading: the $z$-component eigenvalue is $\mu_{L,z} = -g_l \mu_B m_l = -\mu_B m_l$; the magnitude of the moment vector is $|\vec\mu_L| = g_l\mu_B\sqrt{l(l+1)} = \mu_B\sqrt{l(l+1)}$, again never equal to the maximum measurable projection $\mu_B l$.

Spin magnetic moment and the anomalous $g$-factor. By exact analogy but with an empirically/relativistically determined prefactor,

where: $\hat{\vec S}$ is the spin angular momentum operator, $g_s$ the electron spin $g$-factor (the "$g_s \approx 2$, anomalous part $\approx 0.0023$" is the QED correction, $g_s/2 - 1 = \alpha/2\pi + \dots$). Reading: for a single electron, $m_s = \pm\frac12$, so $\mu_{S,z} = \mp g_s\mu_B/2 \approx \mp 1.001\mu_B$ — remarkably close in magnitude to one full Bohr magneton, even though the spin angular momentum itself is only $\hbar/2$, half of the orbital unit $\hbar$. This factor-of-two mismatch between the angular-momentum ratio ($\tfrac12$ vs. $1$) and the magnetic-moment ratio ($\approx 1$ vs. $1$) is precisely why $g_s\approx 2$ while $g_l=1$: spin is "twice as magnetically effective" per unit of angular momentum as orbital motion.

Total magnetic moment and the Landé g-factor (for coupled $\vec J = \vec L + \vec S$, foreshadowing spin–orbit coupling) combine as

where: $g_J$ is the Landé $g$-factor, reducing to $g_l=1$ when $S=0$ and to $g_s=2$ (approximately) when $L=0$; this expression is stated here only to show how the two elementary $g$-factors above generalize once orbital and spin momenta are vector-coupled — its full derivation belongs with spin–orbit coupling.

Example 1 (foundational): the $l=2$ ($d$) cone angles, exactly. Take an electron in a $3d$ orbital, $l=2$. The magnitude of the angular momentum is

The five allowed $m_l$ values are $-2,-1,0,+1,+2$, giving $L_z = m_l\hbar$ and cone angles from $\cos\theta = m_l/\sqrt{6}$:

and by symmetry $m_l=-1,-2$ give $\theta = 114.09^\circ, 144.74^\circ$. Notice that even the "most axial" orientation, $m_l=+2$, sits at $35.26^\circ$ from the $z$-axis, nowhere near $0^\circ$ — a direct numerical confirmation that exact alignment never occurs. The corresponding magnetic moment magnitude is $|\vec\mu_L| = \mu_B\sqrt6 \approx 2.449\,\mu_B$, while the maximum measurable $z$-projection is only $\mu_{L,z}^{\max} = 2\mu_B$ — smaller than the full magnitude, exactly as required by $\cos\theta_{\min}<1$. Limiting case check: as $l\to\infty$ (correspondence-principle limit), $\cos\theta_{\min} = l/\sqrt{l(l+1)}\to 1$, so $\theta_{\min}\to 0$ and the classical picture of a vector that can align with the axis is recovered smoothly — quantum discreteness washes out for macroscopic angular momenta, as it must.

Example 2 (applied): the free electron's spin moment in a 1.20 T field (Stern–Gerlach-style energetics). An unpaired electron (as in a paramagnetic complex or a radical, relevant to EPR spectroscopy) has $s=\tfrac12$, so $m_s = \pm\tfrac12$ only — a two-state (not five-state) system. Using $g_s = 2.0023$, the two $z$-projections of the spin magnetic moment are

In an external field $B = 1.20\ \mathrm{T}$ along $z$, the Zeeman energy of each spin state is $E = -\mu_{S,z}B$ (foreshadowing the full Zeeman treatment), giving two levels

so the splitting is

Converting to a resonance frequency via $\Delta E = h\nu$: $\nu = \Delta E/h = (2.229\times10^{-23})/(6.626\times10^{-34}) = 3.363\times10^{10}\ \mathrm{Hz} \approx 33.6\ \mathrm{GHz}$, squarely in the microwave range used for EPR spectroscopy (X-band EPR typically runs near 9.5 GHz at $B\approx0.34$ T; scaling that field up to 1.20 T shifts the resonance proportionally toward the Q-band region, consistent with this estimate). Limiting case check: setting $B\to 0$ collapses $\Delta E\to 0$ — the two spin states become degenerate in the absence of a field, as required (no spontaneous splitting without symmetry-breaking by an external axis).

Example 3 (exam/competition level): number of Stern–Gerlach spots and the total magnetic moment for a $^2D_{3/2}$ term. Consider a free atom in a term symbol $^{2S+1}L_J = {}^2D_{3/2}$, so $S=\tfrac12$, $L=2$ (D term), $J=\tfrac32$. This kind of term arises, e.g., from a single $d$-electron configuration outside closed shells (as a simplified model for a $d^1$ ion like Ti³⁺). (a) Degeneracy / number of beam spots. The number of allowed $M_J$ values is $2J+1 = 2(3/2)+1 = 4$, so a Stern–Gerlach experiment resolving this term would split the beam into 4 spots (not 2, because $J\neq\tfrac12$; not 5, because $J$ is not simply $L$). (b) Landé $g$-factor. Using the formula stated above,

Insert $J=3/2$ so $J(J+1) = \frac32\cdot\frac52=\frac{15}{4}$; $S=1/2$ so $S(S+1)=\frac34$; $L=2$ so $L(L+1)=6$:

(c) Effective magnetic moment. The conventional "effective moment" reported in magnetochemistry is

(d) Maximum measurable $z$-projection. The largest $|M_J|$ is $3/2$, so

again smaller than the full moment magnitude $g_J\sqrt{J(J+1)}\mu_B = 1.549\,\mu_B$ — the same never-quite-aligned pattern as in Example 1, now for a spin–orbit-coupled composite moment rather than a pure orbital one. Limiting case check: if this were a pure spin term with $L=0$ ($J=S$), the Landé formula gives $g_J = 1 + \frac{S(S+1)+S(S+1)-0}{2S(S+1)} = 1+1 = 2$, correctly recovering $g_s\approx2$; if instead $S=0$ ($J=L$), it gives $g_J = 1+\frac{L(L+1)-L(L+1)}{2L(L+1)}=1$, correctly recovering $g_l=1$. The formula used above interpolates smoothly between these two known limits, confirming its internal consistency.