Eigenfunctions of L_z and the Azimuthal Quantization

Every article in this chapter so far has owed you one argument. The ladder construction proved that angular momentum comes in integer or half-integer rungs — and then Spherical Harmonics quietly used only the integers, promising that the eviction of the half-integers from orbital motion happens "at the azimuthal boundary condition." This article is where that debt is paid. The setting could not be simpler. In spherical coordinates $\hat L_z=-i\hbar\,\partial_\varphi$ touches a single variable, so its eigenvalue problem is a first-order ODE you solve in one line: the eigenfunctions are winding phases $e^{im\varphi}$ , one for every real $m$ . What cuts the continuum down to $m=0,\pm1,\pm2,\dots$ is not a potential and not the algebra — it is the geometry of the circle. The angles $\varphi$ and $\varphi+2\pi$ are the same point of space, and a wavefunction that wants to be a function of position must return to itself after one lap. Only whole numbers of wavelengths close the loop. That one boundary condition has long arms. It is why the magnetic quantum number is an integer, why $\ell$ itself is an integer, why a magnetic field splits a level into equally spaced Zeeman rungs, and why the free-electron model of a benzene ring predicts an absorption near 210 nm. And its fine print — what single-valuedness actually assumes, and why spin slips through — is exactly what examiners and graduate courses probe. Here you get the textbook argument, the honest version behind it, and the numbers.

Key takeaways

$\hat L_z=-i\hbar\,\partial_\varphi$ has eigenfunctions $\Phi_m(\varphi)=e^{im\varphi}/\sqrt{2\pi}$ with eigenvalues $m\hbar$ — a first-order ODE, solvable on sight.
The circle quantizes: $\varphi$ and $\varphi+2\pi$ label the same point, so $e^{i2\pi m}=1$ forces $m\in\mathbb Z$ . Half-integer windings like $e^{i\varphi/2}$ change sign on a full lap and are not functions on space.
The honest version: an orbital wavefunction on the circle must be single-valued, $\psi(\varphi+2\pi)=\psi(\varphi)$ . Half-integer windings violate this periodic boundary condition and, for the orbital domain of $\hat L_z$ , also spoil hermiticity because a boundary term survives.
Integer $m$ plus the ladder's unit steps forces integer $\ell$ : orbital angular momentum keeps only $\ell=0,1,2,\dots$ Half-integer angular-momentum representations are real—they arise for spin, not for the scalar orbital wavefunction on a ring.
The $\Phi_m$ are a complete orthonormal basis on the circle — the angular Fourier series — and they are the azimuthal factor $e^{im\varphi}$ inside every spherical harmonic.
On a physical ring the spectrum is $E_m=m^2\hbar^2/2I$ , doubly degenerate in $\pm m$ : the particle-on-a-ring, benzene's $\pi$ electrons, and nanoscale quantum rings all run on it.
Key idea: $\hat L_z=-i\hbar\,\partial_\varphi$ has eigenfunctions $e^{im\varphi}$ with eigenvalue $m\hbar$ ; single-valuedness on the circle quantizes $m$ to the integers, which is the orbital origin of integer $\ell$ .

The core insight

$\hat L_z=-i\hbar\,\partial_\varphi$ has eigenfunctions $e^{im\varphi}$ with eigenvalue $m\hbar$ ; single-valuedness on the circle quantizes $m$ to the integers — the orbital origin of integer $\ell$ , and the door that stays shut for orbits while spin walks through it.

Where this lives today

A particularly clean modern realization of the phase structure $e^{im\varphi}$ is provided by optical vortex modes: photons in such modes can carry orbital angular momentum $m\hbar$ , with experiments demonstrating very large winding numbers. The azimuthal phase structure can be directly mapped onto an intensity or interference pattern. In condensed matter, mesoscopic metal and semiconductor rings show persistent currents whose period in threaded magnetic flux is exactly one flux quantum: the ring spectrum $E_m$ sliding under a flux-shifted winding number, Aharonov-Bohm physics running on equation (7). And every quantum chemistry code that labels molecular orbitals $\sigma,\pi,\delta$ by their behavior around a bond axis is filing states by this article's integer $m$ — the azimuthal quantum number remains the working index of spectroscopy from microwave rotational lines to synchrotron photoemission.

The physical picture

Put a particle on a circle in the $xy$ -plane and ask for a state of definite $L_z$ . The answer is a traveling wave $e^{im\varphi}$ winding around the loop, and $m$ counts the windings: how many full wavelengths of phase fit around the circumference. An integer number of wavelengths closes on itself smoothly. A half-integer number arrives back at the starting point mid-oscillation — the "wave" would need two different values at one physical point. This is the standing-wave-on-a-ring intuition of old quantum theory, transplanted to a phase that must match up after $2\pi$ .

Figure

Winding phases on the circle — integer m closes the loop, half-integer m tears it

There is a second way to see it, one rung more abstract. $\hat L_z$ generates rotations about $z$ : $e^{-i\theta\hat L_z/\hbar}$ acts on a wavefunction as $\psi(\varphi)\to\psi(\varphi-\theta)$ , and on an $L_z$ eigenstate it gives the phase $e^{-im\theta}$ . For a scalar orbital wavefunction, a $2\pi$ rotation must return the same state, so $e^{-2\pi i m}=1$ , giving $m\in\mathbb Z$ . Keep this version in your pocket; it is the one whose loophole spin will exploit. One more picture before the math. The density of a definite- $L_z$ state is $|\Phi_m|^2=1/2\pi$ — perfectly uniform, no preferred angle at all. That is the position-space shadow of an uncertainty tradeoff: a state of sharp rotation sense is maximally spread in angular position, exactly as a sharp-momentum plane wave is spread over the line. Localizing a wave packet in $\varphi$ costs you a superposition of many $m$ — and the worked example below makes that quantitative for a state you can draw.

The azimuthal eigenproblem

Everything happens on the circle $0\le\varphi<2\pi$ . The operator, inherited from Orbital Angular Momentum Operators (its chain-rule derivation lives there, not here), is

\hat L_z=-i\hbar\,\frac{\partial}{\partial\varphi} \qquad (1)

where: $\varphi$ is the azimuthal angle around $z$ and the family resemblance to $\hat p_x=-i\hbar\,\partial_x$ is exact: $\hat L_z$ is the momentum conjugate to an angle. The one structural novelty — the source of everything in this article — is that $\varphi$ lives on a closed loop, not an infinite line. The eigenvalue problem asks for states on which $\hat L_z$ takes a sharp value. Writing that value as $m\hbar$ (with $m$ dimensionless, so far any number),

\hat L_z\,\Phi_m(\varphi)=m\hbar\,\Phi_m(\varphi) \qquad (2)

where: $m$ will become the magnetic quantum number, but at this stage carries no quantization — the differential equation itself is satisfied by $e^{im\varphi}$ for every complex $m$ . On an infinite line, the corresponding translation-invariant problem has a continuous momentum spectrum. The circle has other plans. Normalizing over one lap gives the eigenfunctions in final form,

\Phi_m(\varphi)=\frac{1}{\sqrt{2\pi}}\,e^{im\varphi} \qquad (3)

where: the prefactor comes from $\int_0^{2\pi}|\Phi_m|^2\,\mathrm d\varphi=1$ since $|e^{im\varphi}|=1$ — every winding phase has the same uniform density $1/2\pi$ , so one constant serves the whole family.

Derivation AInteger m from the closed circle — and what the argument really assumes

The goal is the quantization $m\in\mathbb Z$ : to show how the continuum of candidate eigenvalues collapses to the integer lattice, and — because this article owns the argument for the whole chapter — to state honestly which physical principle does the work. The strategy is a ladder of three arguments of increasing rigor: solve the ODE, impose the textbook single-valuedness condition, then close the loophole the textbook version leaves open (only $|\Psi|^2$ is observable, so why must $\Psi$ itself return?) using the hermiticity of $\hat L_z$ on the circle. Watch for where each rung hands off to the next.

Step 1. Solve the eigenvalue equation. $-i\hbar\,\Phi'=m\hbar\,\Phi$ is first-order with constant coefficients:

\Phi(\varphi)=N\,e^{im\varphi},

one solution for every complex $m$ — no quantization is visible in the differential equation itself.

Step 2. Reality first. Complex $m$ is excluded for the physical eigenvalue problem: the eigenvalues of the self-adjoint $\hat L_z$ must be real, while $e^{im\varphi}$ with $\operatorname{Im}m\neq0$ is not compatible with a periodic boundary condition. Thus $m\in\mathbb R$ before the boundary condition is imposed: the differential equation alone still gives a continuum.

Step 3. The textbook rung: single-valuedness. The coordinates $\varphi$ and $\varphi+2\pi$ label the same point of physical space — $\varphi$ is a chart on the circle, not a physical extension. A wavefunction is a function of position, one value per point, so

\Phi(\varphi+2\pi)=\Phi(\varphi)\;\Longrightarrow\;e^{i2\pi m}=1\;\Longrightarrow\;m=0,\pm1,\pm2,\dots

Half-integers fail exactly: $e^{i2\pi\cdot\frac12}=e^{i\pi}=-1$ . Why this is subtler than it looks: only $|\Psi|^2$ is measurable, and $|e^{i\varphi/2}|^2=1/2\pi$ is perfectly single-valued. If a global sign were truly unobservable, the boundary condition would be a convention, not physics — this loophole is why the argument has a rung 4.

Step 4. The honest rung: hermiticity on the circle. For $\hat L_z$ to be an observable, $\langle f|\hat L_zg\rangle=\langle\hat L_zf|g\rangle$ must hold on its domain, and integrating by parts on the circle leaves a boundary term:

\langle f|\hat L_zg\rangle-\langle\hat L_zf|g\rangle=-i\hbar\,\big[f^*g\big]_0^{2\pi}.

On $2\pi$ -periodic functions the boundary term vanishes and $\hat L_z$ is self-adjoint on the usual orbital domain. Now try to enlarge that domain to include a half-integer winding: with $f=e^{i\varphi/2}$ and $g=1$ , the bracket is $f^*(2\pi)-f^*(0)=-2$ , so the boundary term is $2i\hbar\neq0$ . Thus the half-integer function cannot simply be added to the same periodic domain while preserving self-adjointness. More generally, $\hat L_z$ admits self-adjoint extensions with twisted boundary conditions; the ordinary orbital ring corresponds to the periodic choice, which gives integer $m$ .

Step 5. Consistency seals it from the sphere's side. Suppose one tried to introduce half-integer orbital $\ell$ and build an $\ell=\frac12$ multiplet on the sphere. The construction of the associated Legendre functions for ordinary scalar spherical harmonics requires integer $\ell$ and terminates in the regular polynomial solutions $Y_{\ell m}$ ; the half-integer construction does not belong to this scalar orbital basis. There is also a deeper, structural answer — orbital wavefunctions realize the rotation group SO(3), whose single-valued representations are the integer ones, while half-integers live in SU(2) — but that story is owned by Rotation Operators and SU(2); here it is a signpost, not a proof. Why the checks work out: $e^{im\varphi}$ solves the azimuthal ODE exactly; $e^{i2\pi m}=1$ for $m\in\{-3,\ldots,3\}$ but $e^{i2\pi\cdot\tfrac12}=-1$ , confirming single-valuedness fails for half-integers. Correspondingly, the hermiticity boundary term vanishes on integer windings and equals a nonzero constant on the $(e^{i\varphi/2},1)$ pair — so the operator fails to be self-adjoint for half-integer $m$ .

With the spectrum settled, the family $\{\Phi_m\}_{m\in\mathbb Z}$ has two structural properties worth owning explicitly. First, orthonormality:

\int_0^{2\pi}\Phi_{m'}^*(\varphi)\,\Phi_m(\varphi)\,\mathrm d\varphi=\frac{1}{2\pi}\int_0^{2\pi}e^{i(m-m')\varphi}\,\mathrm d\varphi=\delta_{m'm} \qquad (4)

where: for $m\ne m'$ the integrand oscillates through whole periods and cancels — the calculation is the elementary Fourier orthogonality relation, now wearing eigenfunction clothes. Second, completeness: every single-valued, square-integrable function on the circle expands as

f(\varphi)=\sum_{m=-\infty}^{\infty}c_m\,\Phi_m(\varphi),\qquad c_m=\int_0^{2\pi}\Phi_m^*(\varphi)\,f(\varphi)\,\mathrm d\varphi \qquad (5)

where: this is nothing but the classical Fourier series of $f$ — the expansion grammar of Complete Orthonormal Bases, realized on the circle. The physics reading: $|c_m|^2$ is the probability that a measurement of $L_z$ on the state $f$ returns $m\hbar$ . Fourier analysis is angular-momentum analysis; every Fourier coefficient you have ever computed was secretly an $\hat L_z$ amplitude. And the handoff that powers the rest of the chapter: the azimuthal factor of every spherical harmonic is exactly $\Phi_m$ ,

Y_\ell^m(\theta,\varphi)=\Theta_\ell^m(\theta)\;\frac{e^{im\varphi}}{\sqrt{2\pi}} \qquad (6)

where: the polar factor $\Theta_\ell^m$ — built from associated Legendre functions, with its own theory in the next article — carries the $\ell$ information, while the winding phase settled here carries all the $m$ information. Spherical Harmonics consumed $m\in\mathbb Z$ as a stated fact; equation (6) is where that fact was earned.

From integer m to integer ℓ

Go deeper · Derivation B · 1 minThe eviction of half-integer ℓ from orbital motion

Derivation BThe eviction of half-integer ℓ from orbital motion

The goal is the orbital spectrum rule: $\ell=0,1,2,\dots$ only, even though the abstract ladder of Ladder Operators and the Angular Momentum Spectrum certified half-integer $j$ as perfectly legal algebra. The strategy is a two-line pincer: import the ladder's structural facts (unit steps, $m_{\max}=\ell$ ) without re-deriving them, inject this article's integer- $m$ result, and watch the half-integers fall out of the orbital world — then state precisely why spin is exempt, because conflating the two is the classic exam error.

Step 1. Import the ladder fact. For an angular-momentum multiplet, $m$ runs from $-\ell$ to $+\ell$ in unit steps, with $m_{\max}=\ell$ — proven algebraically in the ladder article and taken here as given. That algebra alone permits $\ell=0,\frac12,1,\frac32,\ldots$ ; the orbital restriction will come from the azimuthal wavefunction.

Step 2. Inject the azimuthal result. For orbital wavefunctions, the eigenvalue $m$ of $\hat L_z$ labels the winding phase $e^{im\varphi}$ . The periodic boundary condition on the circle requires $m\in\mathbb Z$ .

Step 3. Close the pincer. $\ell=m_{\max}$ is itself one of the $m$ values, so $\ell\in\mathbb Z$ , and being a maximum of a set symmetric about zero, $\ell\ge0$ :

\ell=0,1,2,\dots\qquad m=-\ell,\dots,+\ell .

A half-integer $\ell$ would force every $m$ in its multiplet to be half-integer — each one excluded by the orbital boundary condition. Why the argument applies only to orbital angular momentum: every step above relied on $e^{im\varphi}$ being a wavefunction of the spatial angle. Spin angular momentum is an internal degree of freedom, represented by a finite-dimensional spin state rather than a scalar wavefunction on the spatial circle, so the orbital periodicity condition does not apply. Thus $s=\tfrac12$ is perfectly allowed for spin. The chain is consistent: within an orbital multiplet, every $m$ is an integer, and the top rung satisfies $m_{\max}=\ell$ .

Pause on what has been accomplished across three articles. The commutation relations fixed the shape of every spectrum; the ladder fixed the allowed rungs; the circle has now chosen which ladders physical space can host. The division of labor is exact: algebra proposes, geometry disposes.

The particle on a ring

The cleanest laboratory for everything above is a particle of mass $\mu$ confined to a ring of radius $a$ — no potential, one coordinate, and the Hamiltonian is pure azimuthal kinetic energy:

\hat H=\frac{\hat L_z^2}{2I},\qquad I=\mu a^2 \qquad (7)

where: $I$ is the moment of inertia. Because $\hat H$ is a function of $\hat L_z$ alone, the winding phases diagonalize it instantly, and the spectrum is inherited by squaring the eigenvalues.

Figure

The ring spectrum — a parabola in the integer winding number, with every nonzero level doubly degenerate

Go deeper · Derivation C · 1 minThe ring spectrum and its double degeneracy

Derivation CThe ring spectrum and its double degeneracy

The goal is the energy ladder $E_m=m^2\hbar^2/2I$ of a particle on a ring, together with its signature feature — the twofold degeneracy of every excited level — read physically. The strategy is pure harvest: no new equation is solved; the eigenbasis settled above is pushed through the Hamiltonian, which is why this three-step card is the payoff for the whole eigenproblem.

Step 1. Act with $\hat H$ on $\Phi_m$ : since $\hat L_z\Phi_m=m\hbar\,\Phi_m$ ,

\hat H\,\Phi_m=\frac{(m\hbar)^2}{2I}\,\Phi_m\quad\Longrightarrow\quad E_m=\frac{m^2\hbar^2}{2I},\qquad m\in\mathbb Z.

Step 2. Read the degeneracy. $E_{+m}=E_{-m}$ : clockwise and counterclockwise circulation have the same energy, so every nonzero $|m|$ level is doubly degenerate, while the $m=0$ level is nondegenerate. The two members are exchanged by time reversal, which flips the angular momentum and hence $m\to -m$ . A spatial reflection can also reverse the orientation of the ring and map $m\to -m$ , but it is a distinct symmetry operation from time reversal. Why the degeneracy is fragile in exactly one way: anything that distinguishes the two senses of circulation splits it — a magnetic flux through the ring shifts the effective winding number and slides the parabola sideways, which is the Aharonov-Bohm physics of persistent currents in mesoscopic rings. A rotational-symmetry-breaking potential on the ring can mix the degenerate $\pm m$ states into standing-wave combinations such as $\cos m\varphi$ and $\sin m\varphi$ , and in general it lifts their degeneracy already at first order through degenerate perturbation theory.

Step 3. Anchor the scale. For an electron on a benzene-sized ring, $a=0.139$ nm, the unit of the ladder is

E_1=\frac{\hbar^2}{2m_ea^2}=1.97\ \text{eV},

so the ring's quantum staircase has optical riser heights — the reason ring-shaped molecules are colorful antennas for visible and ultraviolet light, made quantitative in Example 2. Using CODATA constants, the moment of inertia evaluates to $I=1.760\times10^{-50}\ \mathrm{kg\,m^2}$ and $E_1=1.972$ eV; the spectrum ratios are $0,1,4,9$ exactly, and the degeneracy $E_{+m}=E_{-m}$ follows by construction.

Why m is called "magnetic"

Nothing about $m$ so far involves magnetism; the name is an experimental fossil, and a good one. Switch on a uniform field $B\hat z$ . Orbital motion couples through the orbital magnetic moment, adding to the Hamiltonian a term proportional to $\hat L_z$ itself:

\hat H'=\frac{\mu_B B}{\hbar}\,\hat L_z\quad\Longrightarrow\quad E_m=\mu_B B\,m \qquad (8)

where: $\mu_B=e\hbar/2m_e=9.274\times10^{-24}$ J/T is the Bohr magneton (its own derivation belongs to Magnetic Moment and the Gyromagnetic Ratio). A level of given $\ell$ splits into $2\ell+1$ rungs spaced by exactly $\mu_B B$ — about $58\ \mu$ eV per tesla. The integer spacing of the splitting is azimuthal quantization made visible on a spectrometer plate: the normal Zeeman effect counts the winding numbers for you. This is also where history got the ordering backwards in the most productive way — Sommerfeld introduced integer $m$ and the tilted, "space-quantized" orbits it implies before quantum mechanics existed, as bookkeeping for exactly these magnetic splittings (see the portrait below). The full Zeeman theory, anomalous patterns included, belongs to the hydrogen and atomic-structure articles; the experiment that tried to photograph space quantization directly is The Stern-Gerlach Experiment, three serials ahead.

Limiting cases

case	result	meaning
$m=0$	$\Phi_0=1/\sqrt{2\pi}$ , $E_0=0$	the axially symmetric state: no circulation, no Zeeman shift, no zero-point energy on a free ring — rotation, unlike vibration, can genuinely stop
$\pm m$ pair	$E_{+m}=E_{-m}$	opposite circulation, equal energy; time reversal exchanges the pair, and their real combinations $\cos m\varphi,\sin m\varphi$ are the chemists' $p_x,p_y$ -style orbitals
large $\vert m\vert$	$L_z=m\hbar\to$ classical	neighboring levels have relative spacing $\Delta E/E_m\sim 2/\vert m\vert$ : the discrete spectrum becomes increasingly dense relative to the energy scale
half-integer $m$	forbidden for orbits	$e^{i\pi}=-1$ : the winding tears, hermiticity of $\hat L_z$ fails, the polar series diverges — three independent vetoes
spin	$s=\tfrac12$ allowed	no $\varphi$ wavefunction exists to constrain; the algebra's half-integer ladders find their home in internal spinor space

Common trapOrbital ℓ is an integer, therefore all angular momentum is an integer

False, and the error costs marks every year. The single-valuedness argument applies to orbital wavefunctions written as functions of the spatial angle $\varphi$ . Spin is an internal degree of freedom: a spin- $\frac12$ state is represented by a two-component spinor, and its half-integer angular-momentum quantum number is not obtained from the spatial boundary condition $\psi(\varphi+2\pi)=\psi(\varphi)$ . Thus orbital angular momentum has integer $\ell$ and $m$ , while spin angular momentum can have half-integer $s$ and $m_s$ , with no contradiction. Conflating the two — or, conversely, claiming that spin cannot exist because $e^{i\varphi/2}$ is double-valued as an ordinary scalar function on the circle — is the classic misreading of the argument.

Worked examples

Example 1 — measuring $L_z$ on a lopsided ring state (confidence builder). A particle on a ring is prepared in the state $\psi(\varphi)=N\cos^2\varphi$ . What values can a measurement of $L_z$ return, with what probabilities, and what are $\langle L_z\rangle$ and $\sigma_{L_z}$ ?

Step 1. Expand in winding phases before doing any integral: $\cos^2\varphi=\tfrac12+\tfrac14e^{i2\varphi}+\tfrac14e^{-i2\varphi}$ . Only $m=0,\pm2$ appear — the state is a three-line Fourier series, so only three outcomes are possible.

Step 2. Normalize. $\int_0^{2\pi}\cos^4\varphi\,\mathrm d\varphi=3\pi/4$ , so $N=\sqrt{4/3\pi}$ .

Step 3. Project onto $\Phi_m$ : the amplitudes are $c_0=N\sqrt{2\pi}/2$ and $c_{\pm2}=N\sqrt{2\pi}/4$ , giving probabilities

P(0)=\frac{2}{3},\qquad P(+2\hbar)=P(-2\hbar)=\frac{1}{6}.

They sum to 1, and no odd $m$ ever appears — $\cos^2\varphi$ is even under $\varphi\to\varphi+\pi$ , so only even windings can contribute.

Step 4. Moments. By the $\pm$ symmetry $\boldsymbol{\langle L_z\rangle=0}$ , while $\langle L_z^2\rangle=\tfrac16(4\hbar^2)+\tfrac16(4\hbar^2)=\tfrac43\hbar^2$ , so $\boldsymbol{\sigma_{L_z}=2\hbar/\sqrt3\approx1.15\,\hbar}$ . A state with visible angular structure pays for it with a broad spread of circulation — the $L_z$ –angle tradeoff in numbers.

Example 2 — benzene as a quantum ring (exam level). Model benzene's six $\pi$ electrons as free particles on a ring of radius $a=0.139$ nm (the C–C ring radius). Find the level ladder, fill it, and predict the wavelength of the first strong absorption.

Step 1. The ladder unit. $I=m_ea^2=1.76\times10^{-50}\ \mathrm{kg\,m^2}$ , so $E_1=\hbar^2/2I=\boldsymbol{1.97\ \mathrm{eV}}$ and $E_m=1.97\,m^2$ eV.

Step 2. Fill with six electrons. Each level holds two spins; $m=0$ takes two electrons, the degenerate pair $m=\pm1$ takes four. HOMO is $|m|=1$ , LUMO is $|m|=2$ — the double degeneracy of the ring is doing real chemical work here, holding four electrons at one energy.

Step 3. The transition. $\Delta E=E_2-E_1=(4-1)E_1=\boldsymbol{5.92\ \mathrm{eV}}$ , so

\lambda=\frac{hc}{\Delta E}=\frac{1240\ \text{eV}\,\cdot\,\text{nm}}{5.92\ \text{eV}}\approx\boldsymbol{210\ \mathrm{nm}}.

Step 4. Honesty check. Benzene has strong absorption bands in the ultraviolet, roughly in the 180–255 nm range depending on the transition and spectral conditions. The crude free-ring model lands in the same ultraviolet neighborhood with no adjustable parameters, but it neglects electron–electron interactions, the discrete six-atom structure, and the full molecular electronic structure. Treat the result as an order-of-magnitude scale estimate, not as a quantitative prediction of the benzene spectrum. The useful lesson is that a simple angular-momentum model can nevertheless reproduce the correct energy scale.

Origins · 1 minArnold Sommerfeld and how this was discovered

Arnold Sommerfeld

1868–1951 · Munich

The letter $m$ entered physics before quantum mechanics did, and Sommerfeld put it there. In 1916, generalizing Bohr's flat circular orbits, he asked how an orbit may orient itself when a magnetic field picks out an axis — and found that the old quantum rules allowed only discrete tilts: the projection of the orbital angular momentum on the field axis had to be an integer multiple of $\hbar$ , namely $L_z=m\hbar$ with $m\in\mathbb Z$ . He named the integer the magnetic quantum number, having already introduced the azimuthal quantum number for the orbit's shape, and called the startling conclusion Raumquantisierung — space quantization — because it asserted that empty space, once a field marks a direction, permits only certain orientations. It explained the normal Zeeman splitting at a stroke, and it was so strange that colleagues treated it as bookkeeping until an atomic beam through an inhomogeneous magnet made the discrete tilts visible six years later. When wave mechanics arrived in 1926, Sommerfeld's integer $m$ turned out to be the winding number of $e^{im\varphi}$ — the tilt catalogue was the circle's Fourier index all along. His Munich seminar, where Atombau und Spektrallinien was scripture and half the founders of quantum mechanics were students, carried the quantum-number bookkeeping into the new theory. (His phase-integral quantization rule and his electrodynamics live in other articles' bands; this one is about the naming of $m$ .)

History

1913

— Bohr quantizes the hydrogen atom with circular orbits and a single quantum number

1916

— Sommerfeld develops the additional azimuthal and magnetic quantum numbers in the old quantum theory and connects them to spatial quantization

1922

— The Stern-Gerlach experiment reveals discrete beam splitting, providing direct evidence for quantized angular-momentum projections

1926

— Schrödinger's wave mechanics recasts m as the winding number of the single-valued phase e^(imφ)

1959

— Aharonov and Bohm predict flux-shifted ring spectra, making the winding phase directly measurable

Going further

References. Griffiths, Introduction to Quantum Mechanics, 3rd ed., §4.1.2 & §4.3 · Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., §3.6 · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 12 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. 1, Ch. VI · Townsend, A Modern Approach to Quantum Mechanics, 2nd ed., Ch. 7. Key results to keep in mind: the coordinate transformation yields $L_z=-i\hbar\partial_\varphi$ ; the azimuthal eigenfunctions are normalized by $1/\sqrt{2\pi}$ over $[0,2\pi)$ ; the single-valuedness condition gives $e^{i2\pi m}=1\Rightarrow m\in\mathbb{Z}$ ; the particle-on-a-ring energies are $E_m=\hbar^2m^2/(2\mu a^2)$ ; and the normal-Zeeman splitting is $E_m=\mu_B B m$ , where $\mu_B$ is the Bohr magneton.

Upstream: Orbital Angular Momentum Operators (the operator $\hat L_z=-i\hbar\partial_\varphi$ and the rotation-generator card, consumed whole) · Angular Momentum Commutation Relations (why $\hat L_z$ is the component worth diagonalizing). Downstream: Associated Legendre Functions (the polar problem this article's $e^{im\varphi}$ leaves behind) · Spherical Harmonics (which consumed $m\in\mathbb Z$ as fact) · Spin Angular Momentum and Spin-1/2 and the Pauli Matrices (the half-integers' escape route) · Rotation Operators and SU(2) (the group-theoretic resolution of the $2\pi$ sign) · the Zeeman and atomic-structure articles of the hydrogen chapter (the magnetic quantum number at work).

If only the probability density is observable, why must the wavefunction itself be single-valued?

This is the sharpest question in the article, and the textbook answer genuinely is incomplete. The honest resolution: admitting double-valued functions breaks the operator, not just the aesthetics — the hermiticity of $\hat L_z$ requires the boundary term $-i\hbar[f^*g]_0^{2\pi}$ to vanish for all pairs in the domain, and half-integer windings leave it standing. A second, independent veto comes from the sphere: half-integer multiplets force polar functions that diverge at the poles. The deep version — wavefunctions on space carry true representations of SO(3), not its double cover — is the business of Rotation Operators and SU(2).

Is m quantized even when there is no sphere — say, a bead genuinely confined to a ring?

Yes. The quantization needs only the closed loop: $\varphi$ and $\varphi+2\pi$ are the same point of the bead's configuration space, so the same single-valuedness and hermiticity arguments run unchanged. The ring spectrum $E_m=m^2\hbar^2/2I$ with integer $m$ is the direct experimental content — molecular free rotors and mesoscopic quantum rings show it.

Why does the eigenvalue equation itself not know about quantization?

Because differential equations are local and quantization here is global. $-i\hbar\Phi'=m\hbar\Phi$ is solved by $e^{im\varphi}$ for any $m$ on any small patch of angle; only the demand that the solution be a function on the whole closed circle — a boundary condition, not an equation — selects the integers. Compare the square well: there too the ODE allowed everything and walls did the choosing. On the circle the "wall" is topology.

Do the real orbitals p_x and p_y have definite L_z?

No. They are the combinations $(\Phi_{-1}\mp\Phi_{+1})/\sqrt2$ in their azimuthal content — standing waves $\cos\varphi$ and $\sin\varphi$ built from equal parts of opposite circulation. Each gives $\langle L_z\rangle=0$ and returns $\pm\hbar$ with probability $\tfrac12$ when $L_z$ is measured. Chemists trade the sharp $m$ label for directionality; the moment a magnetic field makes $m$ physical again, the complex basis is the right one.

Where exactly does the argument use that φ is an angle in real space rather than some internal parameter?

In one load-bearing sentence: " $\varphi$ and $\varphi+2\pi$ label the same point." For a spatial angle that identification is forced by geometry — there is one physical point, so one value of $\Psi$ . An internal phase has no such obligation; nothing in space forces a spin state to return to itself under a $2\pi$ rotation, and indeed it does not — it returns to minus itself, a sign with measurable interference consequences explored in Rotation Operators and SU(2).

Why is there no zero-point energy on the ring when the harmonic oscillator has one?

The oscillator's ground state must compromise between potential localization and kinetic cost, so neither can vanish. The free ring has no potential to localize against: the constant state $\Phi_0=1/\sqrt{2\pi}$ is a legal, normalizable eigenfunction with exactly zero kinetic energy. Uncertainty is respected — $\Phi_0$ is completely delocalized in angle, which costs nothing when no potential cares where the particle is.