Spherical Harmonics

Spherical harmonics are the single most recurrent angular object in physics: they solve the eigenvalue problem of the Laplace–Beltrami operator on the two-sphere, once and for all. Every time a problem carries rotational symmetry — a central potential, a charge distribution, the sky's temperature — separation of variables in spherical coordinates strips away the radius and leaves this one universal set of functions on the angles. They are to the surface of a sphere exactly what $e^{ikx}$ is to the line and $e^{in\varphi}$ is to the circle: the natural harmonic basis into which any field is resolved. Two viewpoints coincide in them, and the coincidence is the whole story. Analytically, $Y_\ell^m$ falls out of $\nabla^2$ in $(r,\theta,\varphi)$ — the azimuth gives $e^{im\varphi}$ with integer $m$ , the polar angle gives $P_\ell^m(\cos\theta)$ , and the radius decouples. Algebraically, the very same functions are the simultaneous eigenstates of $\hat L^2$ and $\hat L_z$ , built from a highest-weight state by lowering. Differential equation and Lie algebra hand you the identical multiplet, each explaining what the other merely asserts. What makes them indispensable is structure: they are orthonormal and complete on $L^2(S^2)$ , so any square-integrable function on the sphere expands uniquely in them — an expansion that is the multipole expansion, the partial-wave decomposition, and the CMB power spectrum all at once. Because rotations act only within each fixed $\ell$ , the degree labels an irreducible representation of $SO(3)$ , and the addition theorem folds its $2\ell+1$ members into a single invariant $P_\ell(\cos\gamma)$ . Master these and the central-force problem, electrostatic multipoles, scattering, and harmonic analysis on the sphere all open at once.

Key takeaways

Spherical harmonics $Y_\ell^m(\theta,\varphi)$ are the simultaneous eigenfunctions of $\hat L^2$ (eigenvalue $\ell(\ell+1)$ ) and $\hat L_z$ (eigenvalue $m$ ), with $\ell=0,1,2,\dots$ and integer $m$ in $-\ell\le m\le\ell$ .
Explicitly $Y_\ell^m=(-1)^m\sqrt{\tfrac{2\ell+1}{4\pi}\tfrac{(\ell-m)!}{(\ell+m)!}}\,P_\ell^m(\cos\theta)\,e^{im\varphi}$ — an associated Legendre function in latitude times a Fourier mode in longitude, normalized on the sphere.
They form a complete orthonormal basis of $L^2(S^2)$ : any field on the sphere expands as $f=\sum_{\ell m}a_{\ell m}Y_\ell^m$ , the spherical analogue of a Fourier series.
The addition theorem $P_\ell(\cos\gamma)=\tfrac{4\pi}{2\ell+1}\sum_m Y_\ell^m(\hat n_1)Y_\ell^{m*}(\hat n_2)$ turns the multiplet into one rotation-invariant Legendre polynomial and powers the multipole expansion of $1/|\mathbf r-\mathbf r'|$ .
Each degree $\ell$ carries the $(2\ell+1)$ -dimensional irrep of $SO(3)$ ; rotations mix only harmonics of the same $\ell$ , and parity acts as $(-1)^\ell$ independent of $m$ .
Key idea: $Y_\ell^m$ are the orthonormal eigenfunctions of $\hat L^2$ (eigenvalue $\ell(\ell+1)$ ) and $\hat L_z$ (eigenvalue $m$ ), equivalently the angular harmonic functions with $\nabla^2(r^\ell Y_\ell^m)=0$ ; they are a complete basis of $L^2(S^2)$ and span the $(2\ell+1)$ -dimensional irreps of $SO(3)$ .

The core insight

$Y_\ell^m(\theta,\varphi)$ are the orthonormal eigenfunctions of $\hat L^2$ (eigenvalue $\ell(\ell+1)$ ) and $\hat L_z$ (eigenvalue $m$ ), equivalently the angular harmonic functions $\nabla^2(r^\ell Y_\ell^m)=0$ ; they form a complete basis of $L^2(S^2)$ and span the $(2\ell+1)$ -dimensional irreducible representations of $SO(3)$ .

Where this lives today

The most spectacular modern use is the cosmic microwave background. The sky temperature is a field on a sphere, $\Delta T(\hat n)=\sum_{\ell m}a_{\ell m}Y_\ell^m(\hat n)$ , and statistical isotropy means all the cosmological information lives in the angular power spectrum $C_\ell=\langle|a_{\ell m}|^2\rangle$ . The Planck satellite measured it out to $\ell\approx2500$ , and its first acoustic peak near $\ell\approx220$ — an angular scale of about one degree — pins the universe to near-flatness and fixes the baryon density, compressing a whole-sky map into a handful of numbers that constrain the cosmos. The same mathematics runs geodesy: the Earth's gravitational potential is catalogued as coefficients $C_{\ell m},S_{\ell m}$ , with models such as EGM2008 resolving the field to degree and order $2190$ , fed by missions like GRACE that watch the coefficients drift as ice sheets melt. In atomic physics the $Y_\ell^m$ are the $s,p,d,f$ orbital shapes, the angular factor of every central-potential wavefunction $\psi=R_{n\ell}(r)Y_\ell^m$ ; in scattering they organize the partial-wave expansion, making phase shifts $\delta_\ell$ the observables. And in computation, HEALPix pixelization and fast transforms carry all of this to maps of millions of pixels — the discrete descendant of Laplace's expansion.

The physical picture

Think of $\hat L^2$ as the Laplacian confined to the surface of a sphere — the Laplace–Beltrami operator. Its eigenfunctions are the standing-wave patterns the sphere supports, and the two integers count nodes: $\ell$ is the total number of nodal circles wrinkling the surface, $m$ how many run around the polar axis (the rest are latitudes). Higher $\ell$ means finer structure, in exact parallel with higher $k$ for $e^{ikx}$ on a line. The lowest members are the shapes a chemist calls $s$ , $p$ , $d$ : a smooth ball, a two-lobed dumbbell, a four-lobed cloverleaf.

Figure

Standing waves on the sphere: the nodal patterns of the axial harmonics Yₗ⁰ for ℓ = 0, 1, 2 — the s, p, d shapes. The number of nodal circles equals ℓ, the eigenvalue of L̂² is ℓ(ℓ+1) = 0, 2, 6, and each degree holds 2ℓ+1 = 1, 3, 5 states.

Algebraically the same picture is a ladder. Fix $\ell$ ; then $m=-\ell,\dots,+\ell$ labels the $2\ell+1$ orientations of a fixed length of angular momentum, and the raising and lowering operators $\hat L_\pm$ step between neighbouring $m$ without touching $\ell$ . Rotating the physical sphere reshuffles the members of one multiplet but never connects different $\ell$ — that closure is why $\ell$ is a good quantum number. The factorized form $P_\ell^m(\cos\theta)\,e^{im\varphi}$ splits the symmetries: $e^{im\varphi}$ diagonalizes rotation about $z$ , $P_\ell^m$ carries the latitude profile, and the Condon–Shortley sign $(-1)^m$ is pure bookkeeping that makes the ladder act with clean, positive matrix elements.

Figure

The angular-momentum ladder for ℓ = 2: L̂₊ and L̂₋ step between the 2ℓ+1 = 5 states m = −2 to +2 without changing ℓ, with matrix elements √(ℓ(ℓ+1) − m(m±1)) = 2, √6, √6, 2. The top state is killed by L̂₊, the bottom by L̂₋.

The addition theorem records the deepest fact. Although the $2\ell+1$ harmonics of a given $\ell$ depend on your choice of axes, the combination $\sum_m Y_\ell^m(\hat n_1)Y_\ell^{m*}(\hat n_2)$ depends only on the angle $\gamma$ between the two directions: summed over its members, the multiplet becomes a rotation invariant, $P_\ell(\cos\gamma)$ . That invariant is the bridge from the abstract basis to the $1/|\mathbf r-\mathbf r'|$ that runs electrostatics and gravitation.

The core equations

Two eigenvalue equations define the harmonics and fix both quantum numbers at once: $\hat L^2$ is the negative Laplace–Beltrami operator, $\hat L_z$ generates rotations about the polar axis, and their common eigenfunctions, single-valued and regular at the poles, are the $Y_\ell^m$ .

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

where: $\hat L^2$ — squared orbital angular momentum, $=-\big[\tfrac{1}{\sin\theta}\partial_\theta(\sin\theta\,\partial_\theta)+\tfrac{1}{\sin^2\theta}\partial_\varphi^2\big]$ ; $\hat L_z$ — $=-i\,\partial_\varphi$ ; $\ell$ — non-negative integer, the degree (orbital quantum number); $m$ — integer with $-\ell\le m\le\ell$ , the order. Reading: the sphere admits standing waves only at the discrete eigenvalues $\ell(\ell+1)$ , each $(2\ell+1)$ -fold degenerate in $m$ . The explicit closed form ties the eigenfunction to the associated Legendre machinery: a latitude profile $P_\ell^m(\cos\theta)$ , a longitude Fourier mode, and a normalizing prefactor, in the Condon–Shortley convention with the sign on $Y_\ell^m$ .

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

where: $P_\ell^m$ — associated Legendre function of degree $\ell$ , order $m$ (no internal Condon–Shortley sign); $(-1)^m$ — the Condon–Shortley phase, placed on $Y_\ell^m$ ; $e^{im\varphi}$ — the $\hat L_z$ eigenmode. For $m<0$ , $Y_\ell^{-m}=(-1)^m Y_\ell^{m*}$ . Reading: harmonic $=$ normalization $\times$ latitude profile $\times$ longitude phase. Orthonormality is what makes the basis usable — distinct harmonics are orthogonal and each is unit-normalized over the full solid angle, which lets you project out a single coefficient by integration.

\int_{S^2} Y_{\ell'}^{m'*}(\theta,\varphi)\,Y_\ell^m(\theta,\varphi)\,\mathrm{d}\Omega=\delta_{\ell\ell'}\,\delta_{mm'} \qquad (3)

where: $\mathrm{d}\Omega=\sin\theta\,\mathrm{d}\theta\,\mathrm{d}\varphi$ with $\int\mathrm{d}\Omega=4\pi$ ; $\delta_{\ell\ell'},\delta_{mm'}$ — Kronecker deltas. Reading: an overlap of two harmonics collapses to a single surviving term. Completeness is the partner statement: the harmonics exhaust $L^2(S^2)$ , so their outer product reconstructs the identity — a Dirac delta on the sphere. Together (3) and (4) certify a genuine basis.

\sum_{\ell=0}^{\infty}\sum_{m=-\ell}^{\ell} Y_\ell^m(\hat n)\,Y_\ell^{m*}(\hat n')=\delta(\hat n-\hat n')=\frac{\delta(\theta-\theta')}{\sin\theta}\,\delta(\varphi-\varphi') \qquad (4)

where: $\hat n,\hat n'$ — unit directions; $\delta(\hat n-\hat n')$ — the solid-angle delta. Reading: no field on the sphere lies outside the span of the harmonics. The expansion is the immediate payoff: any square-integrable field on the sphere is a unique sum of harmonics, coefficients read off by orthonormality. This one line is the multipole expansion, the partial-wave series, and the CMB decomposition at once.

f(\theta,\varphi)=\sum_{\ell=0}^{\infty}\sum_{m=-\ell}^{\ell} a_{\ell m}\,Y_\ell^m(\theta,\varphi),\qquad a_{\ell m}=\int_{S^2} Y_\ell^{m*}\,f\,\mathrm{d}\Omega \qquad (5)

where: $f$ — any element of $L^2(S^2)$ ; $a_{\ell m}$ — expansion (multipole) coefficients. Reading: overlap $f$ with a harmonic and integrate to read off its coefficient. The addition theorem collapses a multiplet into one invariant: because $P_\ell(\cos\gamma)$ depends only on the angle between two directions, it must be expressible through the degree- $\ell$ harmonics of both, with the prefactor fixed by setting the directions equal.

P_\ell(\cos\gamma)=\frac{4\pi}{2\ell+1}\sum_{m=-\ell}^{\ell} Y_\ell^m(\theta_1,\varphi_1)\,Y_\ell^{m*}(\theta_2,\varphi_2) \qquad (6)

where: $\gamma$ — angle between $\hat n_1,\hat n_2$ , with $\cos\gamma=\cos\theta_1\cos\theta_2+\sin\theta_1\sin\theta_2\cos(\varphi_1-\varphi_2)$ ; $P_\ell$ — Legendre polynomial. Reading: sum a multiplet against itself and only the invariant angle $\gamma$ survives. The ladder action generates the whole multiplet from any one member; its square-root coefficient vanishes at $m=\pm\ell$ , which is why the ladder terminates at $2\ell+1$ rungs.

\hat L_\pm\,Y_\ell^m=\sqrt{\ell(\ell+1)-m(m\pm1)}\;Y_\ell^{m\pm1},\qquad \hat L_\pm=e^{\pm i\varphi}\big(\pm\partial_\theta+i\cot\theta\,\partial_\varphi\big) \qquad (7)

where: $\hat L_\pm=\hat L_x\pm i\hat L_y$ — raising/lowering operators. Reading: $\hat L_\pm$ steps $m$ by one, with a coefficient that dies at the top and bottom of the ladder. Finally, the three-dimensional face: multiplying a harmonic by the right power of $r$ makes it a solution of Laplace's equation in all space — a harmonic polynomial (regular) or its singular partner (irregular), the solid harmonics of every potential problem.

\nabla^2\big(r^\ell Y_\ell^m\big)=0,\qquad \nabla^2\big(r^{-\ell-1}Y_\ell^m\big)=0 \qquad (8)

where: $r^\ell Y_\ell^m$ — regular solid harmonic (a degree- $\ell$ harmonic polynomial in $x,y,z$ ); $r^{-\ell-1}Y_\ell^m$ — irregular solid harmonic, singular at the origin. Reading: the angular harmonics are the boundary data of genuine 3D potentials.

Derivation ASeparation of the Laplacian and why the integers appear

The goal is to show that demanding a regular, single-valued solution of Laplace's equation $\nabla^2\Phi=0$ in spherical coordinates forces the entire spherical-harmonic structure — the product form, the integer $m$ from the azimuth, the integer $\ell\ge|m|$ from the poles — with nothing put in by hand. We separate variables, extract the associated Legendre equation, and read the quantization off regularity, so both quantum numbers emerge as consequences rather than assumptions.

Step 1. Write the Laplacian in spherical coordinates, $\nabla^2\Phi=\tfrac{1}{r^2}\partial_r(r^2\partial_r\Phi)+\tfrac{1}{r^2\sin\theta}\partial_\theta(\sin\theta\,\partial_\theta\Phi)+\tfrac{1}{r^2\sin^2\theta}\partial_\varphi^2\Phi$ , and note the angular block is exactly $-\hat L^2/r^2$ . Why: the operator we want is literally the angular part of the everyday Laplacian, so a potential problem and an angular-momentum problem are the same problem.

Step 2. Take a separated ansatz $\Phi=R(r)\,\Theta(\theta)\,e^{im\varphi}$ . Single-valuedness under $\varphi\to\varphi+2\pi$ forces $m$ to be an integer. Why: the first quantum number is just the demand that the field return to itself after a full turn — topological, not dynamical.

Step 3. Insert the ansatz; the radial and angular pieces separate as $\tfrac{(r^2R')'}{R}=\tfrac{\hat L^2(\Theta e^{im\varphi})}{\Theta e^{im\varphi}}$ . Each side depends on disjoint variables, so both equal a constant, written $\ell(\ell+1)$ in anticipation. Why: separation is legitimate because the two sides share no variable, and the pole condition will vindicate naming the constant $\ell(\ell+1)$ .

Step 4. With $x=\cos\theta$ the polar factor obeys $\tfrac{\mathrm{d}}{\mathrm{d}x}\big[(1-x^2)\Theta'\big]+\big[\ell(\ell+1)-\tfrac{m^2}{1-x^2}\big]\Theta=0$ , the associated Legendre equation. Frobenius analysis at $x=\pm1$ shows the series diverges at the poles unless it terminates, which happens only for integer $\ell\ge|m|$ , giving $\Theta=P_\ell^m(\cos\theta)$ . Why: the second quantum number is the price of finiteness — regularity alone selects the integer spectrum.

Step 5. Reassemble: $\Phi=(A\,r^\ell+B\,r^{-\ell-1})\,P_\ell^m(\cos\theta)\,e^{im\varphi}$ . The normalized angular factor is $Y_\ell^m$ ; the two radial powers are the solid harmonics of (8). Why: the constant that quantized the angles also fixes the radial behaviour, so angular basis and 3D potentials are one package. *Verified: the $m=0$ case reduces to Legendre's equation with $\Theta=P_\ell(\cos\theta)$ ; the integer- $\ell$ termination is the standard result (Arfken–Weber §15.5, Hassani Ch. 12, Riley–Hobson–Bence §18.3).*

Go deeper · Derivation B · 1 minBuilding the multiplet algebraically from the top state

Derivation BBuilding the multiplet algebraically from the top state

The goal is to recover the entire structure — eigenvalue $\ell(\ell+1)$ , the $2\ell+1$ members, the ladder coefficients — using only the commutation algebra of $\hat L_\pm$ and $\hat L_z$ , with no differential equation solved beyond one first-order ODE. This operator route shows the integer spectrum a second, independent way and yields the explicit highest-weight harmonic that seeds all the rest.

Step 1. Define the top state by $\hat L_+ Y_\ell^\ell=0$ . With $Y_\ell^\ell=f(\theta)e^{i\ell\varphi}$ and $\hat L_+=e^{i\varphi}(\partial_\theta+i\cot\theta\,\partial_\varphi)$ , this becomes the first-order ODE $f'-\ell\cot\theta\,f=0$ . Why: the highest-weight state is the one that cannot be raised — the maximal projection $m=\ell$ — and that single condition fixes its shape.

Step 2. Integrate $f'/f=\ell\cot\theta$ to get $f\propto(\sin\theta)^\ell$ ; normalization gives $Y_\ell^\ell=\tfrac{(-1)^\ell}{2^\ell\ell!}\sqrt{\tfrac{(2\ell+1)!}{4\pi}}(\sin\theta)^\ell e^{i\ell\varphi}$ . Why: the top of every ladder is an elementary function, and the whole tower is built by lowering it.

Step 3. Use $\hat L^2=\hat L_-\hat L_++\hat L_z^2+\hat L_z$ . On $Y_\ell^\ell$ , $\hat L_+Y_\ell^\ell=0$ and $\hat L_z\to\ell$ , so $\hat L^2 Y_\ell^\ell=(\ell^2+\ell)Y_\ell^\ell=\ell(\ell+1)Y_\ell^\ell$ . Why: the eigenvalue drops out of pure algebra on the top state, never touching the Legendre equation.

Step 4. Lower repeatedly, $Y_\ell^{m-1}=\hat L_-Y_\ell^m/\sqrt{\ell(\ell+1)-m(m-1)}$ . The coefficient vanishes at $m=-\ell$ , so the process stops and the multiplet has exactly $2\ell+1$ states. Why: the ladder terminates where the square root dies, fixing the multiplet size and forbidding half-integer $\ell$ on the sphere. *Verified: the boxed $Y_\ell^\ell$ matches the analytic form from equation (2); the ladder coefficients reproduce $2,\sqrt6,\sqrt6,2$ for $\ell=2$ ; termination reproduces the integer- $\ell$ spectrum independently of the ODE method (Sakurai & Napolitano Ch. 3, Stone & Goldbart Ch. 8).*

Go deeper · Derivation C · 1 minThe Coulomb expansion through the addition theorem

Derivation CThe Coulomb expansion through the addition theorem

The goal is to derive the inverse-distance expansion of $1/|\mathbf r-\mathbf r'|$ in spherical harmonics — the identity underlying every electrostatic and gravitational multipole moment — by combining the Legendre generating function with the addition theorem. The result factorizes the geometry of source and field point, turning a two-body angular dependence into products of one-body harmonics and defining the multipole moments $q_{\ell m}$ that summarize a distribution.

Step 1. Start from the generating function $\tfrac{1}{|\mathbf r-\mathbf r'|}=\tfrac{1}{r_>}\sum_{\ell}\big(\tfrac{r_<}{r_>}\big)^\ell P_\ell(\cos\gamma)$ , with $r_<=\min(r,r')$ , $r_>=\max(r,r')$ and $\gamma$ the angle between $\mathbf r$ and $\mathbf r'$ . Why: this defining Legendre identity already isolates the radial ratio from the angular polynomial.

Step 2. Replace $P_\ell(\cos\gamma)$ using the addition theorem (6), $P_\ell(\cos\gamma)=\tfrac{4\pi}{2\ell+1}\sum_m Y_\ell^m(\hat r)Y_\ell^{m*}(\hat r')$ . Why: this is the crucial move — it splits the single variable $\gamma$ , which mixes the two directions, into products each depending on only one.

Step 3. Substituting gives $\tfrac{1}{|\mathbf r-\mathbf r'|}=\sum_{\ell m}\tfrac{4\pi}{2\ell+1}\tfrac{r_<^{\ell}}{r_>^{\ell+1}}Y_\ell^m(\hat r)Y_\ell^{m*}(\hat r')$ . For a localized $\rho$ observed outside its support ( $r>r'$ ), integrating against $\rho\,\mathrm{d}^3r'$ yields $\Phi(\mathbf r)=\tfrac{1}{4\pi\varepsilon_0}\sum_{\ell m}\tfrac{4\pi}{2\ell+1}\tfrac{q_{\ell m}}{r^{\ell+1}}Y_\ell^m(\hat r)$ with $q_{\ell m}=\int\rho(\mathbf r')\,r'^{\ell}Y_\ell^{m*}(\hat r')\,\mathrm{d}^3r'$ . Why: the source now enters only through the moments $q_{\ell m}$ , so a handful of coefficients captures the whole far field. *Verified: the $\ell=0$ term gives the monopole $Q/(4\pi\varepsilon_0 r)$ , the $\ell=1$ term reproduces the dipole field, and the worked quadrupole below checks $q_{20}=\sqrt{5/16\pi}\,Q_{zz}$ (Jackson §3.5–3.6, Byron–Fuller Ch. 5).*

Limiting cases

Special values of $\ell$ and $m$ collapse the harmonics onto elementary objects, each a useful check on conventions and normalization.

Case	Result	Meaning
$m=0$ (axial symmetry)	$Y_\ell^0=\sqrt{\tfrac{2\ell+1}{4\pi}}\,P_\ell(\cos\theta)$	no azimuthal dependence; the harmonic is an ordinary Legendre polynomial up to normalization
$\ell=0$ (monopole)	$Y_0^0=\tfrac{1}{\sqrt{4\pi}}$	a constant — the spherically symmetric mode whose square integrates to 1 over $S^2$
$\ell=1$ (dipole, $p$ -orbitals)	$Y_1^0=\sqrt{\tfrac{3}{4\pi}}\cos\theta,\ Y_1^{\pm1}=\mp\sqrt{\tfrac{3}{8\pi}}\sin\theta\,e^{\pm i\varphi}$	proportional to $z/r$ and $(x\pm iy)/r$ ; the simplest non-trivial multiplet
Top state $m=\pm\ell$	$Y_\ell^{\pm\ell}\propto(\sin\theta)^\ell e^{\pm i\ell\varphi}$	annihilated by $\hat L_\pm$ ; the seed from which the multiplet is lowered
Parity $\hat n\to-\hat n$	$Y_\ell^m(\pi-\theta,\varphi+\pi)=(-1)^\ell Y_\ell^m(\theta,\varphi)$	even $\ell$ parity-even, odd $\ell$ parity-odd, independent of $m$

Common trapApplying the Condon–Shortley phase twice, and expecting half-integer m

Conventions are the number-one source of sign errors here. The Condon–Shortley phase $(-1)^m$ may be carried inside $P_\ell^m$ or attached to $Y_\ell^m$ , but never both — different textbooks split it differently, and doubling it silently flips the sign of every odd- $m$ harmonic. Fix one convention (this article puts the sign on $Y_\ell^m$ , keeping $P_\ell^m$ phase-free) and check against a known case such as $Y_1^{+1}=-\sqrt{3/8\pi}\sin\theta\,e^{i\varphi}$ . A second trap: $m$ and $\ell$ are strictly integers. As functions on $S^2$ the harmonics realize only the integer-spin representations of $SO(3)$ ; the half-integer $j$ of electron spin lives in the double cover $SU(2)$ as spinors. If you write $Y_{1/2}^{1/2}$ , you have left the world of spherical harmonics.

Worked examples

1 · Expanding a field on the sphere (confidence builder)

Expand $f(\theta,\varphi)=\cos^2\theta$ in spherical harmonics. Because $f$ has no $\varphi$ -dependence, only $m=0$ terms can survive, so we need just the coefficients $a_{\ell 0}$ .

Step 1. Project onto $\ell=0$ : $a_{00}=\int Y_0^{0*}f\,\mathrm{d}\Omega=\tfrac{1}{\sqrt{4\pi}}\int\cos^2\theta\,\mathrm{d}\Omega$ . The azimuthal integral gives $2\pi$ and $\int_{-1}^1 x^2\,\mathrm{d}x=\tfrac23$ , so $a_{00}=\tfrac{1}{\sqrt{4\pi}}\cdot2\pi\cdot\tfrac23=\tfrac{2\sqrt\pi}{3}\approx\mathbf{1.1816}$ .

Step 2. Project onto $\ell=2$ with $Y_2^0=\sqrt{\tfrac{5}{16\pi}}(3\cos^2\theta-1)$ . Then $a_{20}=\sqrt{\tfrac{5}{16\pi}}\cdot2\pi\int_{-1}^1(3x^2-1)x^2\,\mathrm{d}x$ , and $\int_{-1}^1(3x^4-x^2)\mathrm{d}x=\tfrac65-\tfrac23=\tfrac{8}{15}$ , giving $a_{20}=\tfrac{4\sqrt{5\pi}}{15}\approx\mathbf{1.0569}$ . Every other coefficient vanishes: $\ell=1$ is odd in $\cos\theta$ while $f$ is even, and $\ell\ge3$ overlaps a quadratic to zero.

Step 3. Reassemble and verify: the $\ell=0$ piece is $a_{00}Y_0^0=\tfrac13$ , the $\ell=2$ piece is $a_{20}Y_2^0=\tfrac13(3\cos^2\theta-1)=\cos^2\theta-\tfrac13$ , and their sum is exactly $\cos^2\theta$ . The expansion is $\cos^2\theta=\tfrac{2\sqrt\pi}{3}\,Y_0^0+\tfrac{4\sqrt{5\pi}}{15}\,Y_2^0$ , i.e. $\tfrac13+\tfrac23 P_2(\cos\theta)$ — a two-term spectrum, power only at $\ell=0$ and $\ell=2$ .

2 · The quadrupole moment of a linear charge array

Three point charges lie on the $z$ -axis: $+q$ at $z=+d$ , $+q$ at $z=-d$ , and $-2q$ at the origin. Find the leading non-vanishing multipole moment $q_{\ell m}$ and the far-field potential.

Step 1. Check the low orders. Total charge $q+q-2q=0$ , so the monopole $q_{00}=0$ . The dipole $q_{10}=\sqrt{\tfrac{3}{4\pi}}\sum_i q_i r_i\cos\theta_i=\sqrt{\tfrac{3}{4\pi}}[qd(1)+qd(-1)]=0$ , and $q_{1,\pm1}\propto\sin\theta=0$ on the axis. The leading term is the quadrupole.

Step 2. Compute $q_{20}=\sqrt{\tfrac{5}{16\pi}}\sum_i q_i r_i^2(3\cos^2\theta_i-1)$ . The charges at $\theta=0$ and $\theta=\pi$ each give $q\,d^2(3-1)=2qd^2$ ; the origin contributes nothing. So $\sum_i=4qd^2$ and $q_{20}=\sqrt{\tfrac{5}{16\pi}}\,(4qd^2)\approx\mathbf{1.2616\,qd^2}$ , with all other $q_{2m}=0$ by axial symmetry.

Step 3. Relate to the Cartesian quadrupole and read off the field. The trace-form component is $Q_{zz}=\sum_i q_i(3z_i^2-r_i^2)=4qd^2$ , so $q_{20}=\sqrt{\tfrac{5}{16\pi}}\,Q_{zz}$ — the standard bridge between spherical and Cartesian moments. Feeding $q_{20}$ into Derivation C, the $\tfrac{4\pi}{5}$ and two factors of $\sqrt{5/16\pi}$ combine to $qd^2$ , giving $\Phi(\mathbf r)\approx\dfrac{1}{4\pi\varepsilon_0}\,\dfrac{qd^2\,(3\cos^2\theta-1)}{r^3}$ — a pure quadrupole field falling as $1/r^3$ , with the $3\cos^2\theta-1$ shape of $Y_2^0$ .

Physics of the portrait

Spherical harmonics were born not in quantum mechanics but in eighteenth-century celestial mechanics, in the effort to describe the gravitational potential of a not-quite-spherical planet.

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

Pierre-Simon Laplace

1749–1827 · potential theory and the harmonics of the sphere

Pierre-Simon Laplace introduced what we now call spherical harmonics while attacking the shape and attraction of the planets in his Mécanique Céleste. The equation that bears his name, $\nabla^2\Phi=0$ , governs the gravitational potential in empty space, and around 1782–1785 Laplace found that its angular solutions form a discrete family of functions on the sphere — the coefficients in an expansion of a spheroid's potential. His contemporary Adrien-Marie Legendre had, in a 1782 memoir on the same problem, already introduced the polynomials $P_\ell$ that are the axially symmetric ( $m=0$ ) case; the associated functions $P_\ell^m$ were systematized later, notably by Ferrers. The name "spherical harmonics" is Victorian — Thomson (Lord Kelvin) and Tait popularized "harmonics" in their Treatise on Natural Philosophy (1867), by analogy with a vibrating string. The functions then found a second life: with quantum mechanics in the 1920s they proved to be exactly the eigenstates of orbital angular momentum and the shapes of atomic orbitals, and through Wigner's representation theory the very embodiment of $SO(3)$ acting on physics, with Condon and Shortley fixing the phase convention that made the ladder operators behave. What began as bookkeeping for the oblateness of the Earth became the universal language of angular momentum.

History

1782

— Legendre introduces the polynomials Pₗ in his memoir on the attraction of spheroids

1785

— Laplace develops the spherical harmonics as angular solutions of ∇²Φ = 0 in celestial mechanics

1867

— Thomson and Tait popularize the name spherical harmonics in their Treatise on Natural Philosophy

1927

— spherical harmonics are recognized as the eigenstates of orbital angular momentum in the new quantum mechanics

1935

— Condon and Shortley fix the phase convention in The Theory of Atomic Spectra

Going further

Texts. Arfken & Weber, Mathematical Methods for Physicists, 7th ed., Ch. 15, §15.4–15.6 · Hassani, Mathematical Physics, 2nd ed., Ch. 12–13 · Stone & Goldbart, Mathematics for Physics, Ch. 8–9 · Nakahara, Geometry, Topology and Physics, §3–4 · Byron & Fuller, Mathematics of Classical and Quantum Physics, Ch. 5–7 · Jackson, Classical Electrodynamics, 3rd ed., §3.5–3.6 · Sakurai & Napolitano, Modern Quantum Mechanics, Ch. 3.

Provenance. A verifier should recompute: (1) the normalization $\int_{-1}^1[P_\ell^m]^2\mathrm{d}x=\tfrac{2}{2\ell+1}\tfrac{(\ell+m)!}{(\ell-m)!}$ and the $Y_\ell^m$ prefactor; (2) the explicit $\ell=0,1$ harmonics against Arfken §15.6 / Jackson §3.5; (3) the addition-theorem prefactor $4\pi/(2\ell+1)$ by setting $\hat n_1=\hat n_2$ ; (4) the $\ell=2$ ladder coefficients $2,\sqrt6,\sqrt6,2$ ; (5) the expansion $\cos^2\theta=\tfrac13+\tfrac23 P_2(\cos\theta)$ and quadrupole $q_{20}=\sqrt{5/16\pi}\,Q_{zz}=1.2616\,qd^2$ ; (6) parity $(-1)^\ell$ .

Prerequisites. Associated Legendre Functions · Legendre Polynomials · Orthogonal Polynomials: General Theory · Recurrence Relations for Special Functions. Leads to. Spherical Bessel Functions · Hankel Functions and Asymptotic Behavior of Bessel Functions · Generating Functions for Special Functions.

Ask the tutor: Ask me to walk through building the $\ell=2$ multiplet by hand with the ladder operators, or to expand your own field on the sphere and read off its power spectrum.

Why do spherical harmonics show up in so many unrelated problems?

Because they solve one problem hidden inside all of them: the angular part of the Laplacian. Whenever a law is written with $\nabla^2$ — Laplace and Poisson for potentials, Helmholtz for waves, Schrödinger for a central potential — and the setting is rotationally symmetric, separating variables in spherical coordinates leaves the identical angular equation whose solutions are $Y_\ell^m$ . The radial part remembers the specific physics; the angular part is universal. So electrostatics, gravitation, quantum orbitals, and heat flow on a ball all inherit the same angular basis — learn it once, use it everywhere.

What is the difference between the real and complex spherical harmonics?

The complex harmonics $Y_\ell^m\propto e^{im\varphi}$ are eigenstates of $\hat L_z$ , ideal when you care about angular momentum along an axis (Zeeman splitting, selection rules). For real fields — orbitals in chemistry, geophysical maps — it is cleaner to take the real combinations $\propto P_\ell^m\cos m\varphi$ and $\propto P_\ell^m\sin m\varphi$ , the familiar $p_x,p_y,p_z$ and $d_{xy},d_{x^2-y^2}$ shapes. They span the same $(2\ell+1)$ -dimensional space and are equally complete — just a different orthonormal basis within each multiplet, no longer $\hat L_z$ eigenstates but real-valued.

How is ℓ related to the number of nodes on the sphere?

A harmonic $Y_\ell^m$ has exactly $\ell$ nodal circles where it vanishes and changes sign. Of these, $|m|$ are meridians — lines of longitude from the zeros of $\cos m\varphi$ or $\sin m\varphi$ — and the remaining $\ell-|m|$ are parallels of latitude, from the zeros of the associated Legendre function. So $Y_5^0$ has five latitude circles and no meridians, while $Y_5^5$ has five meridians and no latitude nodes. Counting nodes is the fastest way to sketch a harmonic: $\ell$ sets total complexity, $m$ splits it between longitude and latitude.

Why does the addition theorem matter physically?

It converts an abstract basis into a usable expansion of $1/|\mathbf r-\mathbf r'|$ . The two-point function's natural variable $\gamma$ , the angle between source and field directions, entangles both points. The addition theorem factorizes that invariant into products $Y_\ell^m(\hat r)Y_\ell^{m*}(\hat r')$ , each depending on one direction, so source integral and field evaluation decouple. That decoupling defines the multipole moments $q_{\ell m}$ : a whole distribution collapses into a short list of numbers reproducing its far field.

Do spherical harmonics form a basis for every function on the sphere?

For every square-integrable function, yes — that is the completeness relation, equation (4). Any $f\in L^2(S^2)$ has a unique expansion $f=\sum_{\ell m}a_{\ell m}Y_\ell^m$ converging in mean square, just as a Fourier series represents any $L^2$ function on a circle. Pointwise convergence needs mild smoothness (a jump produces Gibbs ringing), but the spectral picture is exact: a rougher field carries more power at high $\ell$ . This is why band-limiting to $\ell\le\ell_{\max}$ is the natural way to smooth data on a sphere, from CMB maps to planetary topography.

Why can't spherical harmonics describe electron spin?

Because spin is not a function on the sphere. Spherical harmonics realize the integer- $\ell$ representations of $SO(3)$ , and under a full $2\pi$ rotation every $Y_\ell^m$ returns to itself. A spin- $\tfrac12$ state, by contrast, picks up a minus sign under $2\pi$ and only returns after $4\pi$ — it transforms under the double cover $SU(2)$ , a half-integer representation with no realization as a single-valued function of angle. Orbital angular momentum, built from motion in real space, lives in the $Y_\ell^m$ ; spin needs two-component spinors, not functions on $S^2$ .