Spherical Harmonics

The last three articles built a machine from pure algebra: operators $\hat{\mathbf L}$ that generate rotations, commutators that lock their spectrum, a ladder that climbs the $m$ -rungs. But an algebra does not tell you what an electron's cloud looks like. This article cashes the machinery into functions you can plot: the spherical harmonics $Y_\ell^m(\theta,\phi)$ , the simultaneous eigenfunctions of $\hat{\mathbf L}^2$ and $\hat L_z$ on the unit sphere. They are the sphere's standing waves — the spherical analogue of the square well's sinusoids. Each $Y_\ell^m$ is a fixed pattern of nodal lines: $\ell$ counts the total angular wiggles (hence the angular kinetic energy), $m$ counts how many wrap the polar axis. The $s$ , $p$ , $d$ shapes of chemistry are nothing but $|Y_\ell^m|^2$ drawn in polar ink. The reason they matter beyond portraiture is universality: every central potential — hydrogen, a tumbling molecule, a nucleon in a mean field — separates into a radial factor times these same functions. Solve the angular problem once and it is solved for every spherically symmetric system. Here you will grow the whole family from one first-order ODE plus the ladder, prove orthonormality and completeness, extract the parity $(-1)^\ell$ , and use the results on real numbers: why $p$ -orbitals hug their axis to within $39^\circ$ on average, and why carbon monoxide broadcasts at 115.8 GHz.

Key takeaways

$Y_\ell^m(\theta,\phi)$ are the joint eigenfunctions on the sphere: $\hat{\mathbf L}^2Y_\ell^m=\ell(\ell+1)\hbar^2\,Y_\ell^m$ and $\hat L_zY_\ell^m=m\hbar\,Y_\ell^m$ .
Orbital angular momentum keeps only $\ell=0,1,2,\dots$ and $m=-\ell,\dots,+\ell$ — $2\ell+1$ states per $\ell$ ; the half-integer ladders never make it onto the sphere.
They factor as $Y_\ell^m = N_\ell^m\,P_\ell^m(\cos\theta)\,e^{im\phi}$ , and the whole family grows from one seed: $\hat L_+Y_\ell^\ell=0$ gives $Y_\ell^\ell\propto\sin^\ell\theta\,e^{i\ell\phi}$ , then $\hat L_-$ descends the rungs.
They are orthonormal and complete: any function on the sphere expands as $\sum_{\ell m}c_{\ell m}Y_\ell^m$ — the Fourier analysis of angular shape.
Parity is $(-1)^\ell$ ; the closure $\sum_m|Y_\ell^m|^2=(2\ell+1)/4\pi$ makes every filled shell exactly spherical.
Key idea: the spherical harmonics package the entire angular content of every central-potential problem into one rotation-covariant basis labelled $(\ell,m)$ .

The core insight

Spherical harmonics are the joint eigenfunctions of $\hat{\mathbf L}^2$ and $\hat L_z$ on the unit sphere; they package the entire angular structure of any central-potential problem into a complete, orthonormal, rotation-covariant basis labelled by $(\ell,m)$ .

Where this lives today

The largest physics dataset ever assembled is a spherical-harmonic expansion: cosmic-microwave-background maps are reported as coefficients $a_{\ell m}$ , and the CMB power spectrum is literally $\sum_m|a_{\ell m}|^2/(2\ell+1)$ plotted against $\ell$ — cosmology's headline numbers are Laplace coefficients of the sky. The same expansion compresses Earth's gravity field (geodesy models carry $\ell$ into the thousands), lighting in computer graphics, and the angular part of every orbital in every quantum-chemistry code. And radio telescopes track galactic star-forming gas through the CO $\ell=1\to0$ line at 115 GHz computed in Example 2 — the rotational ladder, observed daily at industrial scale.

The physical picture

Think of the sphere as a drumhead with no rim. A drum's standing waves are labelled by their nodal circles and diameters; the sphere's are labelled the same way. $Y_\ell^m$ carries exactly $\ell$ nodal lines: $|m|$ great circles through the poles, and $\ell-|m|$ circles of constant latitude. More nodes means a more wrinkled function and — since $\hat{\mathbf L}^2$ is the angular part of kinetic energy — a higher rotational energy $\ell(\ell+1)\hbar^2/2I$ for anything that turns.

Figure

The gallery — cross-sections of |Yₗᵐ|², the angular probability lobes of s, p and d states

Two numbers, two questions. $\ell$ answers how much total angular momentum — $|\mathbf L|=\sqrt{\ell(\ell+1)}\,\hbar$ — and sets the nodal count; $m$ answers how the pattern is oriented: the projection $m\hbar$ , equivalently the number of nodal meridians. The pair $(\ell,m)$ is the maximal address rotational symmetry permits — $\hat L_x,\hat L_y$ refuse to be pinned simultaneously, as the commutation relations dictated.

Figure

Reading a harmonic's address — nodal meridians count m, nodal latitudes count ℓ − |m|

One more picture: the ladder now has a face. The top rung $Y_\ell^\ell$ is a band hugging the equator ( $\sin^\ell\theta$ — maximal circulation about $z$ ); each $\hat L_-$ tilts the pattern, trading azimuthal wraps for latitude nodes, until the bottom rung wraps the other way. The gallery above is not a zoo — it is one algebra photographed rung by rung.

The eigenproblem, inherited whole

Nothing new is postulated. Orbital Angular Momentum Operators derived the differential forms on the sphere — $\hat L_z=-i\hbar\,\partial_\phi$ and the angular Laplacian shape of $\hat{\mathbf L}^2$ — consumed here as given:

\hat{\mathbf L}^2 = -\hbar^2\left[\frac{1}{\sin\theta}\,\partial_\theta\big(\sin\theta\,\partial_\theta\big)+\frac{1}{\sin^2\theta}\,\partial_\phi^2\right]

where: the operator touches only angles — no $r$ anywhere — so its eigenfunctions live on the unit sphere and multiply any radial profile: that $r$ -blindness is why one angular solution serves every central potential. The eigenproblem is the pair

\hat{\mathbf L}^2\,Y_\ell^m=\ell(\ell+1)\hbar^2\,Y_\ell^m, \qquad \hat L_z\,Y_\ell^m=m\hbar\,Y_\ell^m

where: the eigenvalue shapes $\ell(\ell+1)\hbar^2$ and $m\hbar$ are not assumptions but the output of Ladder Operators and the Angular Momentum Spectrum, imported wholesale. What remains is to find the functions. The $\hat L_z$ equation solves itself: $-i\hbar\,\partial_\phi Y = m\hbar Y$ forces $Y_\ell^m=\Theta(\theta)\,e^{im\phi}$ . Here the sphere levies its one tax: $\phi$ and $\phi+2\pi$ label the same point of space, so $e^{im\phi}$ must be single-valued and $m$ is forced into the integers — that argument and its fine print are owned by Eigenfunctions of L_z and the Azimuthal Quantization; here $m\in\mathbb Z$ is consumed as fact. The ladder finishes the job: $m$ runs $-\ell..+\ell$ in unit steps, so $\ell$ is a non-negative integer. The half-integer ladders are real physics — but they belong to spin, and no function on the sphere will ever realize them. The polar factor is conventionally written with the associated Legendre functions $P_\ell^m$ , giving the standard closed form

Y_\ell^m(\theta,\phi)=(-1)^m\sqrt{\frac{2\ell+1}{4\pi}\,\frac{(\ell-m)!}{(\ell+m)!}}\;P_\ell^m(\cos\theta)\,e^{im\phi}

where: the square root normalizes on the sphere, the $(-1)^m$ is the Condon–Shortley phase (the same convention fixed in the ladder article — mismatch and signs break), and $P_\ell^m$ is quoted purely as the polar factor. Its own theory — Rodrigues' formula, recurrences, normalization — is owned by Associated Legendre Functions; the ladder builds every $Y_\ell^m$ without opening that toolbox.

Derivation AThe highest-weight state and the descent of the multiplet

The goal is to construct every $Y_\ell^m$ explicitly, using only the differential forms of $\hat L_\pm$ and the ladder machinery of the previous article. The strategy is the standard one for any ladder: find the top rung by demanding annihilation — which collapses a second-order eigenvalue problem into a first-order ODE — then descend with $\hat L_-$ , whose coefficients are already known. Watch two things: how cheap the top-rung equation is, and how the Condon–Shortley convention rides along automatically once the coefficients $\hbar\sqrt{\ell(\ell+1)-m(m\mp1)}$ are taken real and non-negative.

Step 1. The raising operator in spherical coordinates (from the operator article's forms) is

\hat L_+=\hbar\,e^{i\phi}\big(\partial_\theta+i\cot\theta\,\partial_\phi\big).

Step 2. Impose annihilation on the top rung. Write $Y_\ell^\ell=\Theta_\ell(\theta)\,e^{i\ell\phi}$ and set $\hat L_+Y_\ell^\ell=0$ . The $\partial_\phi$ acting on $e^{i\ell\phi}$ yields $i\ell$ , so the condition becomes

\left(\frac{\mathrm d}{\mathrm d\theta}-\ell\cot\theta\right)\Theta_\ell=0.

Why first-order: annihilation is one condition on one unknown function of one variable — the entire second-order Sturm–Liouville problem in $\theta$ has been traded for the statement "there is nothing above the top."

Step 3. Integrate: $\mathrm d\Theta/\Theta=\ell\cot\theta\,\mathrm d\theta$ gives $\ln\Theta=\ell\ln\sin\theta+\text{const}$ , i.e.

Y_\ell^\ell = c_\ell\,\sin^\ell\theta\;e^{i\ell\phi}.

The stretched state is an equatorial band — maximal circulation about $z$ , probability squeezed toward the equator, sharper with every power of $\sin\theta$ .

Step 4. Normalize on the sphere. For $\ell=1$ : $|c_1|^2\int_0^{2\pi}\mathrm d\phi\int_0^\pi\sin^2\theta\,\sin\theta\,\mathrm d\theta=|c_1|^2\cdot2\pi\cdot\tfrac43=1$ , so $|c_1|^2=3/8\pi$ — and with the conventional sign, $Y_1^1=-\sqrt{3/8\pi}\,\sin\theta\,e^{i\phi}$ .

Step 5. Descend. The previous article's master formula fixes every step with no further integration:

Y_\ell^{m-1}=\frac{\hat L_-\,Y_\ell^m}{\hbar\sqrt{\ell(\ell+1)-m(m-1)}},

with $\hat L_-=\hbar e^{-i\phi}(-\partial_\theta+i\cot\theta\,\partial_\phi)$ . One step from the top of $\ell=1$ : $\hat L_-Y_1^1=\hbar\sqrt2\,Y_1^0$ with $Y_1^0=\sqrt{3/4\pi}\cos\theta$ ; one more gives $Y_1^{-1}$ , and the next annihilates — the coefficient $\sqrt{\ell(\ell+1)-m(m-1)}$ hits its exact zero after $2\ell$ steps, terminating the family at $2\ell+1$ members. Verified: symbolically with sympy — $\hat L_+Y_1^1=0$ and $\hat L_+Y_2^2=0$ exactly; $\hat L_-Y_1^1=\hbar\sqrt2\,Y_1^0$ and $\hat L_-Y_1^0=\hbar\sqrt2\,Y_1^{-1}$ and $\hat L_-Y_1^{-1}=0$ ; the $\ell=1$ normalization integral gives $3/8\pi$ ; every constructed function satisfies both eigenvalue equations for $\ell\le3$ . Matches Griffiths §4.3, Sakurai §3.6.

The catalogue

The low family, written out once — these six functions run most of chemistry and a large share of nuclear and atomic physics:

$Y_\ell^m$	explicit form	shape
$Y_0^0$	$\dfrac{1}{\sqrt{4\pi}}$	perfect sphere — the only isotropic harmonic
$Y_1^0$	$\sqrt{\dfrac{3}{4\pi}}\cos\theta$	two lobes along $\pm z$ ( $p_z$ ); node at the equator
$Y_1^{\pm1}$	$\mp\sqrt{\dfrac{3}{8\pi}}\sin\theta\,e^{\pm i\phi}$	equatorial doughnut; node at both poles
$Y_2^0$	$\sqrt{\dfrac{5}{16\pi}}\,(3\cos^2\theta-1)$	lobes at the poles + small equatorial belt ( $d_{z^2}$ )
$Y_2^{\pm1}$	$\mp\sqrt{\dfrac{15}{8\pi}}\sin\theta\cos\theta\,e^{\pm i\phi}$	four-leaf clover at $45^\circ$
$Y_2^{\pm2}$	$\sqrt{\dfrac{15}{32\pi}}\sin^2\theta\,e^{\pm2i\phi}$	flat equatorial band, double azimuthal wrap

Read the table with the drumhead rule: $Y_1^0$ places its one node at the equator, $Y_2^{\pm1}$ splits its two nodes one each way — the clover. The odd- $m$ signs are Condon–Shortley bookkeeping; their squares, the shapes chemistry plots, never see them.

Orthonormality and completeness

Two different harmonics never overlap. For different $\ell$ the proof needs one line: $\hat{\mathbf L}^2$ is Hermitian on the sphere (a closed surface has no boundary to leak terms), so sandwiching it two ways gives

\hbar^2\,\ell(\ell+1)\,\langle Y_{\ell'}^{m'}|Y_\ell^m\rangle=\hbar^2\,\ell'(\ell'+1)\,\langle Y_{\ell'}^{m'}|Y_\ell^m\rangle

where: unequal prefactors force a vanishing bracket — Hermitian operators do the orthogonality work free, the theorem that separated square-well eigenstates. Different $m$ at equal $\ell$ is quicker still: the $\phi$ -integral of $e^{i(m-m')\phi}$ vanishes. With the construction card's normalization,

\int_0^{2\pi}\!\!\int_0^\pi {Y_{\ell'}^{m'}}^*\,Y_\ell^m\;\sin\theta\,\mathrm d\theta\,\mathrm d\phi=\delta_{\ell\ell'}\,\delta_{mm'}

where: the measure $\sin\theta\,\mathrm d\theta\,\mathrm d\phi=\mathrm d\Omega$ is the honest area element of the sphere — forget the $\sin\theta$ and orthogonality silently fails. Completeness upgrades the family from "orthonormal set" to "basis": any square-integrable function of direction expands as

f(\theta,\phi)=\sum_{\ell=0}^{\infty}\sum_{m=-\ell}^{\ell}c_{\ell m}\,Y_\ell^m(\theta,\phi), \qquad c_{\ell m}=\int {Y_\ell^m}^*\,f\,\mathrm d\Omega

where: coefficient extraction, uniqueness, and the Bessel–Parseval accounting are exactly the machinery of Complete Orthonormal Bases, transplanted to the sphere — Fourier analysis whose "frequencies" are shapes. A discrete sum suffices because the sphere is compact: no Dirac-delta apparatus needed.

Go deeper · Derivation B · 1 minParity from the top rung — why every harmonic answers a mirror with a sign

Derivation BParity from the top rung — why every harmonic answers a mirror with a sign

The goal is the parity theorem $Y_\ell^m(\pi-\theta,\phi+\pi)=(-1)^\ell\,Y_\ell^m(\theta,\phi)$ — the antipodal map multiplies every harmonic by a sign that depends only on $\ell$ . The strategy avoids Legendre-function identities entirely: check the claim on the top rung, where the function is a one-liner, then ride the ladder down, using the fact that parity commutes with the ladder operators. That commutation is the honest content: it descends from $\hat{\mathbf L}=\hat{\mathbf r}\times\hat{\mathbf p}$ being an axial vector.

Step 1. The antipodal map $\hat{\mathbf r}\to-\hat{\mathbf r}$ in angles is $(\theta,\phi)\to(\pi-\theta,\,\phi+\pi)$ : reflect the latitude, rotate the longitude by half a turn.

Step 2. Parity commutes with every component of $\hat{\mathbf L}$ . Under $\hat\Pi$ both $\hat{\mathbf r}$ and $\hat{\mathbf p}$ flip sign, so the cross product $\hat{\mathbf L}=\hat{\mathbf r}\times\hat{\mathbf p}$ is untouched: $[\hat\Pi,\hat L_i]=0$ , hence $[\hat\Pi,\hat L_\pm]=0$ . Why this is the load-bearing step: if parity commutes with the ladder, then applying $\hat L_-$ can never change a state's parity eigenvalue — one check at the top rung settles the whole multiplet.

Step 3. Check the top rung. $Y_\ell^\ell\propto\sin^\ell\theta\,e^{i\ell\phi}$ ; under the map, $\sin(\pi-\theta)=\sin\theta$ is blind, while $e^{i\ell(\phi+\pi)}=e^{i\ell\pi}e^{i\ell\phi}=(-1)^\ell e^{i\ell\phi}$ . The top rung has parity $(-1)^\ell$ , full stop.

Step 4. Descend: $Y_\ell^{m}$ is $\hat L_-$ applied $(\ell-m)$ times to $Y_\ell^\ell$ (up to positive constants), and each application preserves the parity eigenvalue by Step 2. Every member of the multiplet shares the top rung's sign:

\hat\Pi\,Y_\ell^m = (-1)^\ell\,Y_\ell^m.

Why $m$ drops out: parity is a rotationally invariant question, and $m$ is pure orientation bookkeeping — a label that rotations shuffle. Only the rotation-proof label $\ell$ can decide a rotation-proof property. Verified: sympy substitution $(\theta,\phi)\to(\pi-\theta,\phi+\pi)$ on all $Y_\ell^m$ for $\ell\le3$ , every $m$ — each returns exactly $(-1)^\ell$ times itself. Matches Griffiths §4.3 and the 1D parity logic of the earlier track articles.

Parity is spectroscopy's quiet workhorse: an electric-dipole transition inserts one power of $\hat{\mathbf r}$ (odd), so initial and final harmonics must differ in parity — the seed of the $\Delta\ell=\pm1$ selection rule later articles harvest. The 1D ancestor of the logic is Parity Symmetry of One-Dimensional Eigenstates; the sphere replaces the mirror with the antipode.

Go deeper · Derivation C · 1 minUnsöld's closure — why filled shells are perfectly round

Derivation CUnsöld's closure — why filled shells are perfectly round

The goal is the closure identity $\sum_{m=-\ell}^{\ell}|Y_\ell^m(\theta,\phi)|^2=\dfrac{2\ell+1}{4\pi}$ — a constant, the same in every direction. The strategy is symmetry rather than computation: show the left side is invariant under all rotations, conclude it must be constant on the sphere, then fix the constant by one integral. This three-line argument explains a chemical fact: closed shells and half-filled shells have spherical charge clouds.

Step 1. The sum $S(\theta,\phi)=\sum_m|Y_\ell^m|^2$ is the diagonal of the projector onto the $2\ell+1$ -dimensional $\ell$ -multiplet, evaluated at the direction $(\theta,\phi)$ .

Step 2. Rotate the sphere. A rotation mixes the $Y_\ell^m$ of fixed $\ell$ among themselves unitarily (they are a closed multiplet — the ladder built them that way). A unitary mix leaves $\sum_m|Y_\ell^m|^2$ unchanged, term reshuffling notwithstanding — the same reason $|c_1|^2+|c_2|^2$ survives a basis change. Why this forces constancy: any direction can be rotated into any other; a rotation-invariant function of direction has no choice but to be constant.

Step 3. Fix the constant by integrating: $\int S\,\mathrm d\Omega=\sum_m\int|Y_\ell^m|^2\mathrm d\Omega=2\ell+1$ by normalization, and the sphere has area $4\pi$ , so $S=(2\ell+1)/4\pi$ everywhere. Verified: sympy sums the explicit families at $\ell=1$ ( $3/4\pi$ ) and $\ell=2$ ( $5/4\pi$ ) — both reduce to exact constants. Unsöld proved the general statement in 1927.

Limiting cases

case	result	meaning
$\ell=0$	$Y_0^0=1/\sqrt{4\pi}$	the unique isotropic harmonic: no nodes, no angular momentum, no direction preferred
$\ell=1,m=0$	$\sqrt{3/4\pi}\cos\theta$	the $p_z$ shape: one equatorial node, lobes along the axis
stretched $m=\pm\ell$	$Y_\ell^{\pm\ell}\propto\sin^\ell\theta\,e^{\pm i\ell\phi}$	equatorial band, ever thinner with $\ell$ — circulation maximally tipped along $\pm z$
filled shell	$\sum_m\vert Y_\ell^m\vert^2=(2\ell+1)/4\pi$	closed shells are exactly spherical — Unsöld's theorem
$\ell\to\infty$	$\sqrt{\ell(\ell+1)}\,\hbar\to\ell\hbar$	eigenvalue ratio $m/\sqrt{\ell(\ell+1)}\to\cos\theta$ : the classical top recovered, cones narrowing

Worked examples

Example 1 — how well does a $p$ -orbital hug its axis? Task: for the states $Y_1^0$ and $Y_1^{\pm1}$ , compute $\langle\cos^2\theta\rangle$ and the corresponding r.m.s. polar angle.

Step 1. Set up the average as a sphere integral: $\langle\cos^2\theta\rangle=\int|Y_1^m|^2\cos^2\theta\,\mathrm d\Omega$ .

Step 2. For $Y_1^0$ : the integrand is $\tfrac{3}{4\pi}\cos^4\theta$ , and the integral evaluates exactly to $\boldsymbol{\langle\cos^2\theta\rangle=\tfrac35}$ . The r.m.s. direction cosine is $\sqrt{3/5}$ , i.e. a typical polar angle of $\arccos\sqrt{0.6}\approx\boldsymbol{39.2^\circ}$ — the $p_z$ cloud leans strongly toward its axis.

Step 3. For $Y_1^{\pm1}$ : the integrand is $\tfrac{3}{8\pi}\sin^2\theta\cos^2\theta$ , giving exactly $\boldsymbol{\tfrac15}$ — typical angle $\arccos\sqrt{0.2}\approx\boldsymbol{63.4^\circ}$ , a cloud flattened toward the equator.

Step 4. Check against Unsöld: $\tfrac13(\tfrac35+\tfrac15+\tfrac15)=\tfrac13$ , the isotropic value, as a filled $p$ -shell demands. $\tfrac35$ and $\tfrac15$ are the quantitative content of " $p_z$ points, $p_{\pm1}$ wraps."

Example 2 — carbon monoxide's rotational broadcast. Task: CO has bond length $r=0.1128$ nm. Find the rotational constant $B$ and the frequency of the $\ell=1\to0$ line.

Step 1. Reduced mass: $\mu=\dfrac{m_Cm_O}{m_C+m_O}=\dfrac{12.000\times15.9949}{27.9949}\,\mathrm u=6.8562\,\mathrm u=1.1385\times10^{-26}$ kg.

Step 2. Moment of inertia: $I=\mu r^2=1.1385\times10^{-26}\times(1.128\times10^{-10})^2=\boldsymbol{1.449\times10^{-46}\ \mathrm{kg\,m^2}}$ .

Step 3. Rotational constant: $B=\dfrac{\hbar}{4\pi cI}=\boldsymbol{1.93\ \mathrm{cm^{-1}}}$ — the spectroscopist's unit; equivalently the level ladder is $E_\ell=\ell(\ell+1)\,\hbar^2/2I$ with $E_1=2\times\tfrac{\hbar^2}{2I}=0.479$ meV.

Step 4. The lowest line sits at $E_1-E_0=2B$ in wavenumbers: $2\times1.93=3.86\ \mathrm{cm^{-1}}$ , i.e. $\nu=2Bc\approx\boldsymbol{115.8\ \mathrm{GHz}}$ — millimetre waves. Radio astronomers map galaxies in this line. $B=1.93\ \mathrm{cm^{-1}}$ , line at 115.8 GHz.

Origins · 1 minPierre-Simon Laplace and how this was discovered

Pierre-Simon Laplace

1749–1827 · Paris — Académie des Sciences

The functions in this article are older than quantum mechanics by 140 years, and for most of that time they were called Laplace's functions. In the 1782–85 memoirs on the attraction of spheroids, Laplace — building directly on Legendre's coefficient polynomials from the same years — expanded the gravitational potential of a lumpy body in what we now write as $\sum c_{\ell m}Y_\ell^m$ , and established the workhorse facts: the expansion, the orthogonality that fishes out each coefficient by an integral, the term-by-term solution of his equation. He was measuring the Earth's oblateness and the Moon's pull; the same mathematics, unchanged, now measures the shape of an electron cloud. The name's migration — "Laplace's functions" to Thomson and Tait's 1867 "spherical harmonics" to the quantum $Y_\ell^m$ — traces the physics' migration from celestial mechanics to the hydrogen atom: one basis for every science that ever asked "what shape is this field on a sphere?" (His celestial-mechanics program and its stability theorems are a different band, told elsewhere in this wiki.)

History

1782

— Legendre introduces his coefficient polynomials for the attraction of spheroids; Laplace extends them to the full two-angle functions on the sphere

1785

— Laplace's memoir expands planetary potentials in the functions that will carry his name for a century

1867

— Thomson and Tait's Treatise on Natural Philosophy coins "spherical harmonics"

1926

— Schrödinger's hydrogen solution reveals the same functions as the angular factor of every atomic orbital

1927

— Unsöld proves the closure theorem: filled shells are spherical

Going further

References. Griffiths, Introduction to Quantum Mechanics, 3rd ed., Ch. 4 · Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., Ch. 3 · 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, Ch. 9.

Provenance. Eigenvalue equations, explicit forms, normalization, orthonormality, ladder coefficients, parity, and the addition/multipole theorems are standard and can be verified against Griffiths Ch. 4 (Eqs. 4.18–4.32, 4.120–4.121), Sakurai §3.6, Shankar §12.5–12.6, Cohen-Tannoudji Ch. VI and Complements, and Townsend Ch. 9. A verifier should: (1) confirm $Y_0^0=1/\sqrt{4\pi}$ and $Y_1^0=\sqrt{3/4\pi}\cos\theta$ ; (2) check the ladder coefficient $\sqrt{l(l+1)-m(m\pm1)}$ via $L_-L_+=\mathbf{L}^2-L_z^2-\hbar L_z$ ; (3) verify the Legendre normalization integral $\int_{-1}^1[P_l^m]^2\mathrm{d}x=\tfrac{2}{2l+1}\tfrac{(l+m)!}{(l-m)!}$ ; (4) confirm the rigid-rotor spectrum $E_l=\hbar^2 l(l+1)/2I$ . No nonstandard numbers are claimed.

Upstream: Orbital Angular Momentum Operators (the differential forms consumed here) · Angular Momentum Commutation Relations (the algebra) · Ladder Operators and the Angular Momentum Spectrum (the spectrum and the descent machinery) · Complete Orthonormal Bases (the expansion grammar). Downstream: Eigenfunctions of L_z and the Azimuthal Quantization (the single-valuedness fine print) · Associated Legendre Functions (the polar factor's own theory, including the addition-theorem proof) · the hydrogen chapter, where $\psi=R(r)Y_\ell^m$ becomes the atom · Wigner D-Matrices (how rotations mix a multiplet). One pointer across tracks: the multipole expansions of electrodynamics run on this same basis.

Do the spherical harmonics depend on the radial coordinate at all?

No — and that is their superpower. The operators differentiate only angles, so any central-potential eigenfunction factorizes as $R(r)\,Y_\ell^m(\theta,\phi)$ with all potential-dependence quarantined in $R(r)$ . Hydrogen, a 3D oscillator, and a nucleon in a Woods–Saxon well share identical angular factors; only the radial files differ. That factorization is the next chapter's starting point.

Chemistry draws p_x and p_y orbitals, but the table only has Y₁⁰ and Y₁^±1. Who is right?

Both — they are two bases for the same two-dimensional space. $p_x=(Y_1^{-1}-Y_1^{+1})/\sqrt2\propto\sin\theta\cos\phi$ and $p_y=i(Y_1^{-1}+Y_1^{+1})/\sqrt2\propto\sin\theta\sin\phi$ are real combinations, convenient for drawing and for directional bonding. The price: $p_x$ and $p_y$ are not eigenfunctions of $\hat L_z$ (they mix $m=\pm1$ ), so they are the wrong basis the moment a magnetic field or any $z$ -selecting physics enters. Same span, different bookkeeping.

Is m telling me the particle literally orbits around the z-axis?

The honest version is about probability current, not trajectories. $m\ne0$ states carry the phase winding $e^{im\phi}$ , which drives a real circulating current (hence orbital magnetic moments) — but $|Y_\ell^m|^2$ is time-independent: nothing moves in the density. Think "standing pattern with internal circulation," not "planet on a track." The sign of $m$ sets the sense; $m=0$ states have none.

Why does the normalization carry that strange √((2ℓ+1)/4π) factor?

Count states against area. The $4\pi$ is the sphere's area — the isotropic harmonic must be $1/\sqrt{4\pi}$ to normalize. The $2\ell+1$ enters because the multiplet tiles the sphere at density $(2\ell+1)/4\pi$ (Unsöld), so each member's normalization carries that weight; the factorial ratio balances the $m$ -dependence of the polar factor so every rung normalizes identically.

What is the addition theorem the article states but does not prove?

$P_\ell(\cos\gamma)=\dfrac{4\pi}{2\ell+1}\sum_m {Y_\ell^m}^*(\hat n_1)\,Y_\ell^m(\hat n_2)$ , with $\gamma$ the angle between the two directions. Read it as the $\ell$ -multiplet projector written in coordinates: it depends only on the relative angle by the same rotation-invariance as Unsöld's theorem (the special case $\hat n_1=\hat n_2$ ), and it powers multipole expansions of $1/|\mathbf r-\mathbf r'|$ . The proof belongs to Associated Legendre Functions; a numerical spot-check at $\ell=1$ sits in this article's verification script.

The Condon-Shortley (−1)^m looks cosmetic. When would I actually notice it?

The moment you combine harmonics rather than square them. Clebsch–Gordan coefficients, cross-orbital matrix elements, and the real combinations $p_x,p_y$ all involve sums of harmonics, where relative signs decide constructive versus destructive interference. The convention here is deliberately the one fixed for the ladder coefficients in the previous article — one choice, threaded through $\hat L_\pm$ , the $Y_\ell^m$ , and every table downstream. Mixing sources with different conventions is the classic way to a wrong sign.

How do I actually expand a given function f(θ,φ) in spherical harmonics?

Exactly like Fourier analysis, with the sphere's measure: $c_{\ell m}=\int {Y_\ell^m}^*f\,\mathrm d\Omega$ (remember the $\sin\theta$ ), then $f=\sum c_{\ell m}Y_\ell^m$ . Practical aids: symmetry kills coefficients first (antipode-even $f$ has only even $\ell$ ; axial symmetry kills $m\ne0$ ), and Parseval $\int|f|^2\mathrm d\Omega=\sum|c_{\ell m}|^2$ audits truncation, exactly as in Complete Orthonormal Bases.