Separation of Variables in Spherical Coordinates

Point a coordinate system at a centre of symmetry — a nucleus, a charged sphere, a resonating cavity, a ball cooling in still air — and the whole apparatus of mathematical physics reorganizes around the rotation group. In spherical coordinates $(r,\theta,\phi)$ the Laplacian, which appears in nearly every linear field equation, splits cleanly into a radial piece and an angular piece acting on disjoint variables. That single algebraic fact makes separation of variables work here: a PDE in three variables collapses into three ODEs tied by two separation constants. The angular machinery — the spherical harmonics $Y_\ell^m(\theta,\phi)$ , built from associated Legendre functions and complex exponentials — is universal, the same for Laplace, Poisson, Helmholtz, heat, wave, and Schrödinger equations. What distinguishes one problem from another lives entirely in the radial equation. This article is the spherical companion to Cartesian and cylindrical separation. It takes the ansatz $u=R(r)\,\Theta(\theta)\,\Phi(\phi)$ and shows how single-valuedness in $\phi$ quantizes the azimuthal constant to an integer $m$ , how regularity at the poles quantizes the polar constant to exactly $\ell(\ell+1)$ with $\ell\ge|m|$ , and how the radial ODE is Euler for Laplace but spherical Bessel for Helmholtz. The deeper reason the method never fails is symmetry: the Laplacian commutes with $SO(3)$ , the spherical harmonics carry its irreducible representations, and separation is nothing but the simultaneous diagonalization of $L^2$ and $L_z$ — the structure that organizes the hydrogen atom and the multipole expansion.

Key takeaways

Separation of variables seeks $u(r,\theta,\phi)=R(r)\,\Theta(\theta)\,\Phi(\phi)$ , turning one PDE on a spherical domain into three ODEs linked by two separation constants.
The Laplacian splits as $\nabla^2=\frac{1}{r^2}\partial_r(r^2\partial_r)+\frac{1}{r^2}\nabla^2_{S^2}$ ; the angular part $\nabla^2_{S^2}$ carries no $r$ , so its eigenfunctions are the same for every field equation.
Single-valuedness gives $\Phi=e^{im\phi}$ with integer $m$ ; regularity at the poles forces the associated Legendre solution $P_\ell^m(\cos\theta)$ with eigenvalue exactly $\ell(\ell+1)$ , $\ell=0,1,2,\dots$ , $|m|\le\ell$ .
The angular factors combine into the spherical harmonics $Y_\ell^m$ , a complete orthonormal basis on the sphere $S^2$ and the joint eigenfunctions of $L^2$ and $L_z$ .
The radial ODE holds the physics: $r^\ell$ and $r^{-\ell-1}$ for Laplace, spherical Bessel functions $j_\ell(kr),y_\ell(kr)$ for Helmholtz; regularity and boundary conditions select the spectrum.
Key idea: the Laplacian separates in spherical coordinates because it decomposes into commuting radial and angular operators; demanding single-valuedness in φ and regularity at the poles quantizes the angular eigenvalue to exactly ℓ(ℓ+1), producing the spherical harmonics, and the PDE type then fixes only the radial equation.

The core insight

The Laplacian in spherical coordinates separates because $\nabla^2=\frac{1}{r^2}\partial_r(r^2\partial_r)+\frac{1}{r^2}\nabla^2_{S^2}$ : the radial and angular operators act on disjoint variables. Single-valuedness in $\phi$ and regularity at the poles force the angular eigenvalues to be exactly $\ell(\ell+1)$ with integer $\ell$ , producing the spherical harmonics; the PDE type then fixes only the radial ODE.

Where this lives today

Spherical separation is the working coordinate system for anything with a centre. Electronic-structure codes — density-functional theory, Hartree–Fock, quantum chemistry — expand atomic orbitals in radial functions times spherical harmonics, so the angular integrals reduce to tabulated Gaunt coefficients. Cosmologists expand the cosmic microwave background in $Y_\ell^m$ and read the early universe off the angular power spectrum $C_\ell$ , with the acoustic peaks near $\ell\approx200$ encoding the geometry and matter content of the cosmos. Geophysicists invert the Earth's free oscillations — the whole planet ringing after a great earthquake — into $j_\ell(kr)Y_\ell^m$ modes, exactly the spherical-cavity construction, to map the deep interior. In nuclear and particle physics the partial-wave expansion above turns every scattering experiment into a set of phase shifts $\delta_\ell$ , one per angular-momentum channel. And in computer graphics and machine learning, spherical-harmonic lighting and equivariant neural networks use the same $2\ell+1$ -dimensional representations of $SO(3)$ to encode rotational symmetry — the algebra Laplace built for planets now shading surfaces and predicting molecular energies.

The physical picture

Think of the unit sphere as a stage on which only certain standing-wave patterns can live. Walk once around the equator and you must return to the same value — that periodicity forces the azimuthal winding number $m$ to be a whole number. Travel from pole to pole and the field must stay smooth where all the meridians meet — that regularity at the poles forces the degree $\ell$ to be a non-negative integer and quantizes the angular eigenvalue to $\ell(\ell+1)$ . The spherical harmonics $Y_\ell^m$ are the patterns that survive both demands: $\ell$ counts the total angular structure (nodal circles), $m$ counts how much winds about the polar axis. Everything angular is settled by the geometry of the sphere, independent of which equation you started from — the first figure traces this cascade from the single PDE down to the three quantized ODEs.

Figure

The separation cascade for a spherical PDE — the Laplacian splits into a radial operator plus the angular Laplace–Beltrami operator, and the single equation becomes three ODEs: radial R(r), polar Θ(θ) with eigenvalue ℓ(ℓ+1), and azimuthal Φ(φ) with integer m, whose angular pair combines into the spherical harmonic Yₗᵐ

The radial equation is where the identity of the specific PDE enters. A static Laplace field has no length scale, so the radial solutions are scale-free powers, $r^\ell$ growing outward and $r^{-\ell-1}$ decaying — the monopole, dipole, and higher multipoles of electrostatics and gravitation. A wave or diffusion problem carries a wavenumber $k$ : the radial equation becomes the spherical Bessel equation, whose regular solution $j_\ell(kr)$ is a sphere-corrected sine finite at the origin, while the irregular $y_\ell(kr)$ is the cosine-like partner that blows up there. Each separation constant is an eigenvalue of an operator commuting with the Laplacian — $-\partial_\phi^2\to m^2$ from $L_z^2$ , and $-\nabla^2_{S^2}\to\ell(\ell+1)$ from $L^2$ — so separation is really the simultaneous diagonalization of commuting observables. The second figure shows the three lowest spherical Bessel functions and the zeros that set every spherical cavity resonance.

Figure

The spherical Bessel functions j₀, j₁, j₂ solving the Helmholtz radial equation — each finite at the origin and oscillating with a 1/x-decaying envelope; their first zeros x = 3.14159, 4.49341, 5.76346 fix the resonant wavenumbers ka of a spherical cavity, so a 10 cm air sphere with a pressure node at the wall rings at f = c/2a = 1715 Hz

The core equations

The starting point is the Laplacian in spherical coordinates. Every equation below follows from the way this operator distributes across the three coordinates, and the grouping of terms is exactly what makes separation possible.

\nabla^2=\frac{1}{r^2}\frac{\partial}{\partial r}\!\left(r^2\frac{\partial}{\partial r}\right)+\frac{1}{r^2\sin\theta}\frac{\partial}{\partial\theta}\!\left(\sin\theta\frac{\partial}{\partial\theta}\right)+\frac{1}{r^2\sin^2\theta}\frac{\partial^2}{\partial\phi^2} \qquad (1)

where: $r$ — radial distance, $r\ge0$ ; $\theta$ — polar (colatitude) angle, $0\le\theta\le\pi$ ; $\phi$ — azimuthal angle, $0\le\phi<2\pi$ . The first term is purely radial; the last two — the angular block — share an overall $1/r^2$ , the structural clue that they can be peeled off. Collecting the angular block gives the compact split the whole subject rests on. The operator $\nabla^2_{S^2}$ is the Laplace–Beltrami operator on the unit sphere and contains no $r$ .

\nabla^2=\frac{1}{r^2}\frac{\partial}{\partial r}\!\left(r^2\frac{\partial}{\partial r}\right)+\frac{1}{r^2}\nabla^2_{S^2},\qquad \nabla^2_{S^2}=\frac{1}{\sin\theta}\frac{\partial}{\partial\theta}\!\left(\sin\theta\frac{\partial}{\partial\theta}\right)+\frac{1}{\sin^2\theta}\frac{\partial^2}{\partial\phi^2} \qquad (2)

where: $\nabla^2_{S^2}$ — the angular Laplacian, whose eigenfunctions are the spherical harmonics. Because it carries no radial dependence, its eigenvalue problem is the same for Laplace, Helmholtz, heat, and wave equations — the angular answer is universal, and only the radial factor changes. We seek solutions of product form — an ansatz justified by the fact that it produces a complete basis for every solution.

u(r,\theta,\phi)=R(r)\,\Theta(\theta)\,\Phi(\phi) \qquad (3)

where: $R,\Theta,\Phi$ — functions of a single variable each. Substituting and dividing through isolates each variable on one side of an equation, which can only hold if each side equals a constant — the separation constants. The azimuthal coordinate is cyclic, so its equation is the simplest, with a complex-exponential solution quantized by periodicity.

\frac{\mathrm{d}^2\Phi}{\mathrm{d}\phi^2}=-m^2\Phi\ \Rightarrow\ \Phi_m(\phi)=e^{im\phi},\quad m\in\mathbb{Z} \qquad (4)

where: $m$ — azimuthal (magnetic) quantum number, forced integer by single-valuedness $\Phi(\phi+2\pi)=\Phi(\phi)$ . It counts how many full oscillations the field executes as you circle the polar axis once. The polar equation is the associated Legendre equation — the deep one, because its solvability condition quantizes the second separation constant.

\frac{1}{\sin\theta}\frac{\mathrm{d}}{\mathrm{d}\theta}\!\left(\sin\theta\frac{\mathrm{d}\Theta}{\mathrm{d}\theta}\right)+\left[\ell(\ell+1)-\frac{m^2}{\sin^2\theta}\right]\Theta=0 \qquad (5)

where: $\ell$ — degree, a non-negative integer fixed by regularity at the poles; $\ell(\ell+1)$ — the angular separation eigenvalue. Bounded solutions at $\theta=0,\pi$ exist only for $\ell\in\{0,1,2,\dots\}$ with $|m|\le\ell$ , giving $\Theta=P_\ell^m(\cos\theta)$ . The polar and azimuthal factors combine into the spherical harmonics, with the physics normalization and phase below.

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} \qquad (6)

where: $P_\ell^m$ — associated Legendre function of degree $\ell$ , order $m$ ; $(-1)^m$ — the Condon–Shortley phase. The prefactor makes the set orthonormal, and $2\ell+1$ is the number of $m$ values, the dimension of the $\ell$ -th representation of $SO(3)$ . That orthonormality makes the harmonics a usable basis: any function on the sphere expands in them with coefficients from a single integral.

\int_0^{2\pi}\!\!\int_0^\pi Y_\ell^{m\,*}(\theta,\phi)\,Y_{\ell'}^{m'}(\theta,\phi)\,\sin\theta\,\mathrm{d}\theta\,\mathrm{d}\phi=\delta_{\ell\ell'}\,\delta_{mm'} \qquad (7)

where: $\delta$ — Kronecker delta; $\sin\theta\,\mathrm{d}\theta\,\mathrm{d}\phi=\mathrm{d}\Omega$ — the solid-angle element. Distinct harmonics are orthogonal, each normalized to one, and the set is complete on $S^2$ . Only now does the specific PDE matter, through the radial equation. For Laplace's equation there is no length scale, so the radial ODE is equidimensional (Euler type).

\frac{\mathrm{d}}{\mathrm{d}r}\!\left(r^2\frac{\mathrm{d}R}{\mathrm{d}r}\right)-\ell(\ell+1)R=0\ \Rightarrow\ R(r)=A\,r^\ell+B\,r^{-\ell-1} \qquad (8)

where: $A,B$ — constants fixed by boundary conditions. $r^\ell$ is regular at the origin and grows outward (interior problems); $r^{-\ell-1}$ decays and is regular at infinity (exterior problems, multipoles). For the Helmholtz equation $\nabla^2u+k^2u=0$ — obtained after separating time from the wave or heat equation — the radial ODE acquires the wavenumber and becomes the spherical Bessel equation.

r^2R''+2rR'+\big[k^2r^2-\ell(\ell+1)\big]R=0\ \Rightarrow\ R(r)=A\,j_\ell(kr)+B\,y_\ell(kr) \qquad (9)

where: $k$ — wavenumber from separating time; $j_\ell,y_\ell$ — spherical Bessel functions of the first and second kind. $j_\ell$ is finite at $r=0$ and kept on any domain containing the origin; $y_\ell$ diverges there and is kept only for exterior or shell domains. Assembling all three factors and summing over the allowed labels gives the general solution every boundary-value problem specializes.

u(r,\theta,\phi)=\sum_{\ell=0}^{\infty}\sum_{m=-\ell}^{\ell}\big(A_{\ell m}\,r^\ell+B_{\ell m}\,r^{-\ell-1}\big)\,Y_\ell^m(\theta,\phi) \qquad (10)

where: $A_{\ell m},B_{\ell m}$ — coefficients set by boundary data via (7). This is the Laplace form; swap the radial factor for $j_\ell(kr),y_\ell(kr)$ to get the Helmholtz general solution.

Derivation AFull separation of Laplace's equation in spherical coordinates

The goal is to take $\nabla^2u=0$ , insert the ansatz $u=R\Theta\Phi$ , and reduce one partial differential equation to three ordinary ones, showing at each stage which physical demand quantizes each separation constant. We divide through to isolate the variables, quantize $m$ by single-valuedness, impose regularity at the poles to fix the polar constant at $\ell(\ell+1)$ , and solve the resulting Euler equation, reaching the boxed separated solution $u_{\ell m}=(A r^\ell+B r^{-\ell-1})P_\ell^m(\cos\theta)e^{im\phi}$ .

Step 1. Substitute $u=R(r)\Theta(\theta)\Phi(\phi)$ into $\nabla^2u=0$ using (1), giving three terms, one carrying $\Theta\Phi$ , one carrying $R\Phi$ , one carrying $R\Theta$ . Why: the product form is what lets every derivative act on only one factor, so the operator distributes into pieces that each depend on a single variable.

Step 2. Multiply the whole equation by $r^2/(R\Theta\Phi)$ . The radial term becomes $\frac{1}{R}\frac{\mathrm{d}}{\mathrm{d}r}(r^2 R')$ , a function of $r$ alone, and the two angular terms depend only on $\theta,\phi$ . Why: a function of $r$ equal to a function of $(\theta,\phi)$ for all values of the coordinates can only be a constant — this is the separation argument itself.

Step 3. Set the radial part equal to a constant $\lambda$ and the angular part to $-\lambda$ :

\frac{1}{R}\frac{\mathrm{d}}{\mathrm{d}r}\!\left(r^2\frac{\mathrm{d}R}{\mathrm{d}r}\right)=\lambda,\qquad \frac{1}{\Theta\sin\theta}\frac{\mathrm{d}}{\mathrm{d}\theta}\!\left(\sin\theta\frac{\mathrm{d}\Theta}{\mathrm{d}\theta}\right)+\frac{1}{\Phi\sin^2\theta}\frac{\mathrm{d}^2\Phi}{\mathrm{d}\phi^2}=-\lambda.

Why: this is the first separation; the value of $\lambda$ is not yet fixed — regularity in the next steps will pin it to $\ell(\ell+1)$ .

Step 4. Multiply the angular equation by $\sin^2\theta$ and rearrange so that all $\theta$ -dependence sits on one side and all $\phi$ -dependence on the other:

\frac{\sin\theta}{\Theta}\frac{\mathrm{d}}{\mathrm{d}\theta}\!\left(\sin\theta\frac{\mathrm{d}\Theta}{\mathrm{d}\theta}\right)+\lambda\sin^2\theta=-\frac{1}{\Phi}\frac{\mathrm{d}^2\Phi}{\mathrm{d}\phi^2}=m^2.

Why: this is the second separation, and it introduces the constant $m^2$ ; the sign is chosen so the $\phi$ -solution is oscillatory rather than exponential, as periodicity requires.

Step 5. Solve the azimuthal equation $\Phi''=-m^2\Phi$ to get $\Phi=e^{im\phi}$ , and impose single-valuedness $\Phi(\phi+2\pi)=\Phi(\phi)$ , which demands $e^{2\pi im}=1$ , hence $m\in\mathbb{Z}$ . Why: the field must have one value at each physical point, and $\phi$ and $\phi+2\pi$ label the same point — this is a genuine boundary condition disguised as periodicity.

Step 6. Write the polar equation and substitute $x=\cos\theta$ to obtain the associated Legendre equation $(1-x^2)\Theta''-2x\Theta'+[\lambda-m^2/(1-x^2)]\Theta=0$ , with regular singular points at $x=\pm1$ (the poles). A Frobenius analysis shows the series solution diverges at the poles unless it terminates, which happens exactly when $\lambda=\ell(\ell+1)$ with integer $\ell\ge|m|$ ; then $\Theta=P_\ell^m(\cos\theta)$ . Why: boundedness at the poles is the physical demand — a field cannot be infinite where the meridians meet — and it is what discretizes the angular spectrum to the integers.

Step 7. With $\lambda=\ell(\ell+1)$ the radial equation is $r^2R''+2rR'-\ell(\ell+1)R=0$ , an Euler equation; trying $R=r^s$ gives the indicial equation $s^2+s-\ell(\ell+1)=(s-\ell)(s+\ell+1)=0$ , so $s=\ell$ or $s=-\ell-1$ and $R=A r^\ell+B r^{-\ell-1}$ . Why: an equidimensional equation has power-law solutions, and the two roots are the growing and decaying multipole radial profiles selected later by the domain.

Go deeper · Derivation B · 1 minReduction of the Helmholtz radial equation to spherical Bessel form

Derivation BReduction of the Helmholtz radial equation to spherical Bessel form

The goal is to show that once the angular eigenvalue $\ell(\ell+1)$ is separated out of $\nabla^2u+k^2u=0$ , the radial equation is ordinary Bessel's equation of half-integer order in disguise, so its solutions are the spherical Bessel functions $j_\ell,y_\ell$ tied to $J_{\ell+1/2},Y_{\ell+1/2}$ . We write the radial ODE, remove the first-derivative term with $R=u/\sqrt r$ , read off the order, and check $\ell=0$ against elementary trigonometry.

Step 1. After angular separation the radial equation is

\frac{\mathrm{d}}{\mathrm{d}r}\!\left(r^2\frac{\mathrm{d}R}{\mathrm{d}r}\right)+\big[k^2r^2-\ell(\ell+1)\big]R=0,

the same structure as the Laplace radial equation but with the extra $k^2r^2$ term. Why: the wavenumber $k$ from separating time is what turns scale-free power laws into oscillatory functions with a natural length $1/k$ .

Step 2. Substitute $R(r)=u(r)/\sqrt{r}$ . This removes the awkward weighting of the first-derivative term and casts the equation into standard form. Why: multiplying a solution by $r^{-1/2}$ is the classic trick that converts a self-adjoint radial operator in three dimensions into the one-dimensional Bessel operator.

Step 3. The equation for $u$ becomes

r^2u''+r u'+\Big[k^2r^2-\big(\ell+\tfrac12\big)^2\Big]u=0,

which is Bessel's equation of order $\nu=\ell+\tfrac12$ in the variable $x=kr$ . Why: the shift $\ell(\ell+1)+\tfrac14=(\ell+\tfrac12)^2$ is exactly what makes the order half-integer, the signature of spherical (as opposed to cylindrical) Bessel functions.

Step 4. The two solutions are $u=J_{\ell+1/2}(kr)$ and $Y_{\ell+1/2}(kr)$ , so

R(r)=\sqrt{\frac{\pi}{2kr}}\,J_{\ell+1/2}(kr)\equiv j_\ell(kr),\qquad \sqrt{\frac{\pi}{2kr}}\,Y_{\ell+1/2}(kr)\equiv y_\ell(kr),

which are the definitions of the spherical Bessel functions. Why: the $\sqrt{\pi/2x}$ factor is precisely the $1/\sqrt r$ from Step 2 folded into the normalization, so the spherical functions inherit all half-integer Bessel identities.

Step 5. Check $\ell=0$ : the radial equation for $u$ is $u''+k^2u=0$ , so $u=\sin kr,\cos kr$ and $R=u/r$ , giving $j_0(x)=\sin x/x$ and $y_0(x)=-\cos x/x$ . Their small-argument forms $j_\ell(x)\to x^\ell/(2\ell+1)!!$ confirm $j_0\to1$ , $j_1\to x/3$ , matching the plotted curves. Why: the lowest order must reduce to elementary trigonometric waves divided by $r$ , the spherically symmetric standing wave, and it does.

Go deeper · Derivation C · 1 minPartial-wave expansion of a plane wave

Derivation CPartial-wave expansion of a plane wave

The goal is to expand a plane wave $e^{i\mathbf{k}\cdot\mathbf{r}}$ — the building block of scattering theory — in the spherical basis, producing the Rayleigh formula that opens the partial-wave method. Because a plane wave is finite everywhere only the regular $j_\ell$ appear, and because it is axisymmetric about $\mathbf{k}$ only $m=0$ harmonics (Legendre polynomials) enter. We write the axisymmetric ansatz, project onto $P_\ell$ by orthogonality, evaluate the integral, and collect the coefficient.

Step 1. Take $\mathbf{k}$ along the $z$ -axis so $\mathbf{k}\cdot\mathbf{r}=kr\cos\theta$ , and write the axisymmetric expansion

e^{ikr\cos\theta}=\sum_{\ell=0}^{\infty}c_\ell\,j_\ell(kr)\,P_\ell(\cos\theta).

Why: regularity at $r=0$ forbids $y_\ell$ , and independence of $\phi$ forbids $m\ne0$ , so the general solution collapses to this single sum over $\ell$ .

Step 2. Project onto $P_\ell$ using $\int_{-1}^1 P_\ell P_{\ell'}\,\mathrm{d}x=\frac{2}{2\ell+1}\delta_{\ell\ell'}$ with $x=\cos\theta$ :

c_\ell\,j_\ell(kr)=\frac{2\ell+1}{2}\int_{-1}^{1}e^{ikr x}P_\ell(x)\,\mathrm{d}x.

Why: orthogonality is the standard tool for extracting one coefficient at a time from a complete-basis expansion.

Step 3. The integral is a standard representation of the spherical Bessel function, $\int_{-1}^{1}e^{ikr x}P_\ell(x)\,\mathrm{d}x=2\,i^\ell\,j_\ell(kr)$ . Why: this identity is what ties the flat plane wave to the curved spherical basis, and the factor $i^\ell$ is the phase each angular momentum channel picks up.

Step 4. Substitute to find $c_\ell\,j_\ell(kr)=(2\ell+1)\,i^\ell\,j_\ell(kr)$ , so $c_\ell=(2\ell+1)\,i^\ell$ , giving the Rayleigh expansion

e^{i\mathbf{k}\cdot\mathbf{r}}=\sum_{\ell=0}^{\infty}(2\ell+1)\,i^\ell\,j_\ell(kr)\,P_\ell(\cos\gamma),

with $\gamma$ the angle between $\mathbf{k}$ and $\mathbf{r}$ in the general case. Why: the $j_\ell(kr)$ cancel, leaving pure numbers, and the $\ell=0$ leading term $j_0(kr)\to1$ recovers $e^{i\mathbf{k}\cdot\mathbf{r}}\to1$ as $kr\to0$ , a built-in consistency check.

Limiting cases

The radial behaviour and small- and large-argument limits are what distinguish the physical regimes, and they double as checks on the formulae:

Case	Result	Meaning
Laplace, interior	$R(r)=A_\ell\,r^\ell$	No length scale; regularity at $r=0$ drops $r^{-\ell-1}$ . Gives the multipole/Taylor structure of harmonic functions inside a source-free ball.
Laplace, exterior	$R(r)=B_\ell\,r^{-\ell-1}$	Regularity at infinity selects the decaying power; $\ell=0$ is the Coulomb/Newton $1/r$ monopole, $\ell=1$ the dipole.
Helmholtz, small $kr$	$j_\ell(kr)\to\dfrac{(kr)^\ell}{(2\ell+1)!!}$	Finite at the origin; the basis for interior cavity modes and the leading term of each partial wave. Numerically $j_1(0.05)=0.016662\approx0.05/3$ .
Helmholtz, large $kr$	$j_\ell(x)\to\dfrac{\sin\!\left(x-\ell\pi/2\right)}{x}$	Asymptotic standing wave with $1/x$ envelope; the outgoing Hankel form $h_\ell^{(1)}(x)\sim(-i)^{\ell+1}e^{ix}/x$ describes scattered waves.
Axisymmetric, $m=0$	$u=\sum_\ell\big(A_\ell r^\ell+B_\ell r^{-\ell-1}\big)P_\ell(\cos\theta)$	With no $\phi$ -dependence the harmonics reduce to Legendre polynomials $P_\ell(\cos\theta)$ — the workhorse of textbook boundary-value problems.
Monopole, $\ell=0$	$Y_0^0=1/\sqrt{4\pi}$ , $R=A+B/r$	The spherically symmetric channel; the constant plus $1/r$ potential, i.e. a point charge and a uniform offset.

Common trapKeeping the irregular solution, or forgetting that |m| ≤ ℓ

Two errors recur. The first is keeping the irregular radial solution on a domain that includes the origin: $r^{-\ell-1}$ (Laplace) and $y_\ell(kr)$ (Helmholtz) both diverge at $r=0$ , so on any solid ball the corresponding coefficient must be zero. Regularity at the origin is not bookkeeping — it is a genuine boundary condition that selects half the solution space, just as regularity at infinity selects the other half for exterior problems. The second is forgetting the order is bounded, $-\ell\le m\le\ell$ : $P_\ell^m$ vanishes when $|m|>\ell$ , so there are exactly $2\ell+1$ harmonics per degree. A related slip is dropping the $(-1)^m$ Condon–Shortley phase — harmless for one mode, but it scrambles the signs of matrix elements once harmonics are combined.

Worked examples

1 · Grounded conducting sphere in a uniform field (fully worked)

A grounded conducting sphere of radius $a$ sits in an originally uniform electric field $\mathbf{E}_0=E_0\hat{\mathbf z}$ . Find the potential outside the sphere, the induced dipole moment, and the surface charge density.

Step 1. The problem is axisymmetric, so use the $m=0$ exterior solution $\Phi=\sum_\ell(A_\ell r^\ell+B_\ell r^{-\ell-1})P_\ell(\cos\theta)$ . Far away the field is uniform, $\Phi\to-E_0 z=-E_0 r\,P_1(\cos\theta)$ , which fixes $A_1=-E_0$ and $A_\ell=0$ for $\ell\ne1$ — only the $\ell=1$ growing term survives, since a uniform field is pure $P_1$ .

Step 2. Apply the grounded condition $\Phi(a,\theta)=0$ for all $\theta$ . By orthogonality each coefficient vanishes separately: for $\ell\ne1$ , $B_\ell=0$ ; for $\ell=1$ , $A_1 a+B_1 a^{-2}=0\Rightarrow B_1=-A_1 a^3=E_0 a^3$ . The sphere responds only in the channel the field drives.

Step 3. Assemble the exterior potential:

\Phi(r,\theta)=-E_0\cos\theta\left(r-\frac{a^3}{r^2}\right).

The first term is the applied field, the second the induced response. The induced part is a pure dipole, $\Phi_{\text{ind}}=E_0 a^3\cos\theta/r^2$ .

Step 4. Comparing with $\Phi_{\text{dip}}=p\cos\theta/(4\pi\varepsilon_0 r^2)$ gives the induced dipole moment $p=4\pi\varepsilon_0 E_0 a^3$ , so the polarizability is $\alpha=4\pi\varepsilon_0 a^3$ — set purely by the sphere's volume, as if it were a dielectric of infinite permittivity.

Step 5. The surface charge is $\sigma=\varepsilon_0 E_r|_{a}=-\varepsilon_0\,\partial_r\Phi|_{a}$ . With $\partial_r\Phi=-E_0\cos\theta(1+2a^3/r^3)$ , at $r=a$ this is $-3E_0\cos\theta$ , so $\sigma(\theta)=3\varepsilon_0 E_0\cos\theta$ — maximal at the pole facing the field, opposite at the far pole, zero at the equator, integrating to zero net charge. The whole result lives in the single $\ell=1$ term.

2 · Fundamental acoustic mode of a spherical cavity (fully worked)

An air-filled spherical cavity of radius $a=0.10\ \mathrm{m}$ has a pressure-release wall (pressure node at $r=a$ ). With sound speed $c_s=343\ \mathrm{m/s}$ , find the lowest resonant frequency.

Step 1. The acoustic pressure obeys the Helmholtz equation, so the modes are $p=j_\ell(kr)\,Y_\ell^m(\theta,\phi)$ — only $j_\ell$ , since the cavity includes the origin. The pressure-release condition $p(a)=0$ demands a Bessel zero at the wall, $j_\ell(ka)=0$ .

Step 2. The allowed wavenumbers are $k_{\ell n}=x_{\ell n}/a$ with $x_{\ell n}$ the $n$ -th positive zero of $j_\ell$ ; the first zeros are $x_{01}=\pi$ , $x_{11}=4.49341$ , $x_{21}=5.76346$ . The smallest is $x_{01}=\pi$ , so the fundamental is the spherically symmetric $\ell=0$ breathing mode.

Step 3. For $\ell=0$ , $j_0(ka)=\sin(ka)/(ka)=0$ gives $ka=\pi$ , so

f=\frac{c_s k}{2\pi}=\frac{c_s}{2a}=\frac{343}{2(0.10)}=\mathbf{1715\ Hz}.

The fundamental sits at 1715 Hz — the $\ell=0$ zero at exactly $\pi$ makes it identical to a length- $a$ pipe.

Step 4. The next mode is $\ell=1$ (a dipole sloshing mode) at $ka=4.49341$ , giving $f=c_s x_{11}/(2\pi a)\approx\mathbf{2453\ Hz}$ . The two lowest resonances, 1715 and 2453 Hz, sit at a ratio $4.49341/\pi\approx1.43$ — inharmonic, because the zeros of $j_\ell$ , unlike a 1-D pipe's sinusoids, are not evenly spaced.

Physics of the portrait

Spherical harmonics are the thread that runs from the Newtonian potential of a squashed planet to the orbitals of an atom, and their story is a two-century collaboration between celestial mechanics, potential theory, and quantum physics.

Origins · 1 minspherical harmonics and how this was discovered

spherical harmonics

the angular eigenfunctions Yₗᵐ of the Laplacian on the sphere · potential theory and central-force problems

The spherical harmonics were born not in atomic physics but in the study of gravity. In 1782 Adrien-Marie Legendre, investigating the Newtonian attraction of spheroids, introduced the polynomials $P_\ell(\cos\theta)$ as the coefficients in the expansion of $1/|\mathbf r-\mathbf r'|$ — the generating function of the multipole expansion. Almost simultaneously Pierre-Simon Laplace, working on the figure of the planets, wrote $\nabla^2V=0$ and expanded its solutions in the full angular functions $Y_\ell^m$ (his "Laplace coefficients"); the associated Legendre functions $P_\ell^m$ followed for the $\phi$ -dependent terms. For most of the nineteenth century these were tools of geodesy and celestial mechanics, and William Thomson (Lord Kelvin) and Peter Guthrie Tait gave them their modern treatment as solid and surface harmonics in their 1867 Treatise on Natural Philosophy. The twist came with quantum mechanics: when Schrödinger separated hydrogen in 1926, the angular factor was exactly Laplace's $Y_\ell^m$ , and the integer $\ell$ became orbital angular momentum, the eigenvalue $\ell(\ell+1)\hbar^2$ of $L^2$ . Condon and Shortley fixed the $(-1)^m$ phase in 1935 that spectroscopists still use, and Wigner and Racah built the algebra of angular-momentum coupling on the same foundation. A device for computing the pull of a lumpy planet had become the alphabet of the atom.

History

1782

— Legendre introduces the polynomials Pₗ in the theory of gravitational attraction

1785

— Laplace expands the potential in the angular functions that become the spherical harmonics

1837

— Sturm and Liouville frame the separated ODEs as eigenvalue problems with orthogonal eigenfunctions

1867

— Kelvin and Tait systematize solid and surface harmonics in their Treatise on Natural Philosophy

1926

— Schrödinger separates the hydrogen atom, identifying Yₗᵐ as angular-momentum eigenstates

1935

— Condon and Shortley fix the phase convention standard ever since

Going further

Texts. Stone & Goldbart, Mathematics for Physics (2009), ch. 8–9 · Arfken & Weber, Mathematical Methods for Physicists, 7th ed., ch. 9, 11, 15 · Hassani, Mathematical Physics, 2nd ed., ch. 12–13 (Sturm–Liouville and special functions) · Byron & Fuller, Mathematics of Classical and Quantum Physics, ch. 6–7 · Jackson, Classical Electrodynamics, 3rd ed., ch. 3 · Riley, Hobson & Bence, Mathematical Methods for Physics and Engineering, 3rd ed., ch. 21 · Morse & Feshbach, Methods of Theoretical Physics, ch. 10. Key results to keep in mind: the indicial roots of the Laplace radial equation are $(s-\ell)(s+\ell+1)=0$ ; the substitution $R=u/\sqrt{r}$ reduces the Helmholtz radial ODE to Bessel's equation of order $\ell+\tfrac{1}{2}$ , with first zeros $x_{01}=\pi$ , $x_{11}=4.49341$ , $x_{21}=5.76346$ ; the Rayleigh coefficient is $c_\ell=(2\ell+1)i^\ell$ and $j_0=\sin x/x$ . The worked numbers — grounded-sphere dipole $p=4\pi\varepsilon_0 E_0 a^3$ and surface charge $\sigma=3\varepsilon_0 E_0\cos\theta$ ; cavity fundamental $f=c_s/2a=1715$ Hz for $a=0.10$ m, $c_s=343$ m/s, and the $\ell=1$ mode at $2453$ Hz — are all standard results.

Builds on. Classification of Second-Order Linear PDEs · Separation of Variables in Cartesian Coordinates · Sturm–Liouville Theory and Eigenfunction Expansions · Boundary Conditions and Well-Posedness. Leads to. The Helmholtz Equation · Laplace's and Poisson's Equations · Fundamental Solutions and Green's Functions for PDEs · The Wave Equation.

Ask the tutor: Ask me to walk through any spherical boundary-value problem step by step — a charged shell, a scattering phase shift, a cavity mode, or a hydrogen orbital — and watch the same three ODEs do all the work.

Why does separation of variables work at all in spherical coordinates?

Because the Laplacian decomposes into a radial and an angular operator that commute — they act on different variables — so the eigenvalue problem factorizes. Deeper still, $\nabla^2_{S^2}$ is just $-L^2/\hbar^2$ , the Casimir of $SO(3)$ , and separation is the simultaneous diagonalization of the commuting set $\{\nabla^2,\ L^2,\ L_z\}$ . Spherical coordinates are one of eleven systems in which the Laplacian separates, each tied to a symmetry: Cartesian to translations, cylindrical to axial rotations, spherical to full rotations. Separability is a shadow of the symmetry, not an accident of the coordinates.

Why is the angular eigenvalue exactly ℓ(ℓ+1) and not ℓ²?

Because $\ell(\ell+1)$ is the eigenvalue of the Casimir $L^2$ on the irreducible representations of $SO(3)$ . If $\ell$ is the top of the $m$ -ladder, acting with the lowering operator and the commutation relations forces $L^2\to\hbar^2\ell(\ell+1)$ , not $\hbar^2\ell^2$ — the extra $\ell$ comes from the non-commuting components of angular momentum. Analytically it is the value that makes the associated Legendre series terminate into a polynomial finite at the poles. The two viewpoints agree exactly, an early hint of the group-theoretic structure underneath.

What is the difference between spherical harmonics and Legendre polynomials?

Legendre polynomials $P_\ell(\cos\theta)$ are the axisymmetric ( $m=0$ ) special case; spherical harmonics $Y_\ell^m$ are the full set including $\phi$ -dependence. When a problem has no azimuthal dependence — a sphere in a uniform field, an on-axis point charge — only $P_\ell$ appear and the double sum collapses to a single sum over $\ell$ . The associated Legendre functions $P_\ell^m$ bridge the two: they are the $\theta$ -part of $Y_\ell^m$ , reducing to $P_\ell$ when $m=0$ . Reach for $P_\ell$ when the geometry is axially symmetric, and for the full harmonics only when the boundary data varies with $\phi$ .

When do I keep the irregular solutions r⁻ˡ⁻¹ and yₗ?

Whenever the origin is not part of your domain. For a solid ball both diverge at $r=0$ and must be dropped — only $r^\ell$ and $j_\ell$ survive. For an exterior region extending to infinity the decaying $r^{-\ell-1}$ is exactly what you want, since $r^\ell$ would blow up; these decaying terms are the multipole moments of the enclosed source. For a shell $a\le r\le b$ , which excludes the origin, you keep both solutions and fix the coefficients from the inner and outer boundaries. The rule: impose regularity wherever your domain reaches — the origin, infinity, or both.

How does this connect to the hydrogen atom and angular momentum?

The Schrödinger equation for any central potential $V(r)$ separates as $\psi=R_{n\ell}(r)Y_\ell^m(\theta,\phi)$ , with the exact angular factor derived here. The integer $\ell$ becomes orbital angular momentum, $Y_\ell^m$ is the joint eigenstate of $L^2$ and $L_z$ with eigenvalues $\hbar^2\ell(\ell+1)$ and $\hbar m$ , and the $2\ell+1$ values of $m$ are the degeneracy of a level with no preferred axis. Only the radial equation knows the potential — Coulomb for hydrogen, harmonic for the isotropic oscillator, a well for the particle in a sphere. That is why the s, p, d, f orbital shapes are universal: they are the spherical harmonics, fixed by rotational symmetry alone.

Why are the resonances of a spherical cavity inharmonic?

Because the radial functions are spherical Bessel functions, whose zeros are not evenly spaced, unlike the sines of a 1-D pipe with zeros at exact multiples of $\pi$ . For the $\ell=0$ mode the zeros happen to be $n\pi$ , so that single channel is harmonic, $f=c_s/2a$ . But the $\ell=1$ zeros ( $4.49341,\dots$ ) and $\ell=2$ zeros ( $5.76346,\dots$ ) fall at irrational multiples of the fundamental, so the full spectrum is a dense, inharmonic set. This is why a struck sphere or bell lacks the clean harmonic series of a string or pipe — the three-dimensional geometry scrambles the overtones through the Bessel zeros.