Spherical Harmonics as Angular Momentum Eigenstates

Derivation

Spherical Harmonics as Angular Momentum Eigenstates

D-216 Home PU-301 Threads symmetry · waves · matter Depends on Angular Momentum Spectrum from Commutators

Statement

The simultaneous position-space eigenfunctions of the orbital angular-momentum operators \(\hat{L}^2\) and \(\hat{L}_z\), acting on single-valued functions on the unit sphere, are the spherical harmonics \(Y_\ell^m(\theta,\phi)\). They satisfy \(\hat{L}^2 Y_\ell^m = \hbar^2\,\ell(\ell+1)\,Y_\ell^m\) and \(\hat{L}_z Y_\ell^m = \hbar m\,Y_\ell^m\), where single-valuedness under \(\phi\to\phi+2\pi\) forces \(m\in\mathbb{Z}\) and hence \(\ell\in\{0,1,2,\dots\}\) with \(m=-\ell,\dots,\ell\).

Why it matters

The abstract ladder analysis fixes the spectrum of any angular momentum, allowing half-integer values, but says nothing about which values a coordinate-space particle can realize. This derivation supplies the missing physical input — single-valuedness of the wavefunction on the sphere — and shows it collapses the allowed orbital quantum numbers to integers, while simultaneously producing the explicit functions.

The resulting \(Y_\ell^m\) are the angular skeleton of every central-force problem: the hydrogen orbitals, the partial-wave expansion of scattering, multipole radiation, the rigid rotor, and the vibrational modes of a sphere all inherit their angular structure from these functions. They form a complete orthonormal basis for \(L^2(S^2)\), so any function on the sphere is one of their superpositions.

Assumptions

Orbital angular momentum only.We take \(\hat{\vec{L}}=\hat{\vec{r}}\times\hat{\vec{p}}=-i\hbar\,\vec{r}\times\nabla\), a differential operator on functions of position. If dropped and a general \(\hat{\vec{J}}\) (including spin) is allowed, the spectrum admits half-integer \(j\) that have no single-valued position representation at all.

Single-valued wavefunction.The state is an ordinary function on the sphere: \(\psi(\theta,\phi+2\pi)=\psi(\theta,\phi)\). If dropped (multivalued or anyonic wavefunctions in restricted geometries), \(m\) need not be an integer and the quantization below fails.

Radial and angular separation.We fix the radius (work on \(S^2\)); \(\hat{\vec L}\) contains no radial derivatives, so the angular eigenproblem decouples from the radial one. If dropped, \(\hat{L}^2\) still commutes with a rotationally invariant \(\hat H\), but one can no longer isolate a pure angular eigenfunction without carrying the radial factor.

Regularity at the poles.Eigenfunctions must be finite at \(\theta=0,\pi\), where the spherical coordinates are singular. If dropped, the second solution of the Legendre equation (\(Q_\ell^m\), logarithmically divergent) is admitted, and the eigenfunctions are no longer normalizable.

Derivation

\[ \hat{L}_z=-i\hbar\,\frac{\partial}{\partial\phi} \]

Cartesian \(\hat{L}_z=\hat{x}\hat{p}_y-\hat{y}\hat{p}_x\) transformed to spherical coordinates; only the azimuthal derivative survives because \(\hat{L}_z\) generates rotations about the \(z\)-axis, i.e. shifts in \(\phi\). A

\[ -i\hbar\,\frac{\partial}{\partial\phi}\,Y(\theta,\phi)=\hbar m\,Y(\theta,\phi)\ \Longrightarrow\ Y(\theta,\phi)=P(\theta)\,e^{im\phi} \]

Impose the \(\hat{L}_z\) eigenvalue equation; the \(\phi\)-dependence separates because the operator acts only on \(\phi\), leaving \(P(\theta)\) undetermined for now. A

\[ e^{im(\phi+2\pi)}=e^{im\phi}\ \Longrightarrow\ e^{2\pi i m}=1\ \Longrightarrow\ m\in\mathbb{Z} \]

Single-valuedness of the wavefunction on the sphere. This is the physical input the abstract ladder argument lacks: it rejects half-integer \(m\) for orbital motion. B

\[ \hat{L}_\pm=\hat{L}_x\pm i\hat{L}_y=\hbar\,e^{\pm i\phi}\!\left(\pm\frac{\partial}{\partial\theta}+i\cot\theta\,\frac{\partial}{\partial\phi}\right),\qquad \hat{L}^2=-\hbar^2\!\left[\frac{1}{\sin\theta}\frac{\partial}{\partial\theta}\!\left(\sin\theta\frac{\partial}{\partial\theta}\right)+\frac{1}{\sin^2\theta}\frac{\partial^2}{\partial\phi^2}\right] \]

Spherical-coordinate forms of the ladder and Casimir operators, obtained from \(\hat{\vec L}=-i\hbar\,\vec r\times\nabla\). These are the tools that turn algebra into differential equations. A

\[ \hat{L}_+\,Y_\ell^{\ell}=0 \]

From the assumed ladder spectrum: \(m\) is bounded above by \(\ell\), so raising the top state annihilates it. This turns the top rung into a solvable first-order ODE rather than a second-order one. B

\[ \hbar\,e^{i\phi}\!\left(\frac{\partial}{\partial\theta}+i\cot\theta\,\frac{\partial}{\partial\phi}\right)\!\left[P(\theta)\,e^{i\ell\phi}\right]=0 \ \Longrightarrow\ \frac{dP}{d\theta}-\ell\cot\theta\,P=0 \]

Insert \(Y_\ell^{\ell}=P(\theta)e^{i\ell\phi}\) into \(\hat{L}_+Y_\ell^\ell=0\); the \(\phi\)-derivative gives \(i\cot\theta\cdot i\ell=-\ell\cot\theta\), and the common factor \(\hbar e^{i(\ell+1)\phi}\) cancels. B

\[ \frac{dP}{P}=\ell\,\frac{\cos\theta}{\sin\theta}\,d\theta\ \Longrightarrow\ \ln P=\ell\ln\sin\theta+\text{const}\ \Longrightarrow\ Y_\ell^{\ell}(\theta,\phi)=c_\ell\,(\sin\theta)^{\ell}\,e^{i\ell\phi} \]

Separate and integrate the first-order ODE. Single-valuedness again requires \(\ell\in\mathbb{Z}_{\ge0}\); non-integer \(\ell\) would make \(e^{i\ell\phi}\) multivalued. B

\[ Y_\ell^{m}\ \propto\ \big(\hat{L}_-\big)^{\ell-m}Y_\ell^{\ell},\qquad \hat{L}_-\,Y_\ell^{m}=\hbar\sqrt{\ell(\ell+1)-m(m-1)}\;Y_\ell^{m-1} \]

Apply the lowering operator repeatedly. Each application lowers \(m\) by one and reintroduces \(\theta\)-structure; the outcome is \(e^{im\phi}\) times the associated Legendre function \(P_\ell^m(\cos\theta)\). The known ladder coefficient fixes each normalization ratio. B

\[ \hat{L}^2=\hat{L}_-\hat{L}_+ +\hat{L}_z^{\,2}+\hbar\hat{L}_z\ \Longrightarrow\ \hat{L}^2 Y_\ell^{\ell}=\big(\hbar^2\ell^2+\hbar^2\ell\big)Y_\ell^{\ell}=\hbar^2\ell(\ell+1)\,Y_\ell^{\ell} \]

Use the operator identity on the top state, where \(\hat{L}_+Y_\ell^\ell=0\). Since \([\hat{L}^2,\hat{L}_-]=0\), every \(Y_\ell^m\) generated in Step 8 shares this \(\hat{L}^2\) eigenvalue: the whole multiplet is degenerate in \(\ell(\ell+1)\). C

\[ \int_0^{2\pi}\!\!\int_0^{\pi}\big|Y_\ell^m\big|^2\sin\theta\,d\theta\,d\phi=1\ \Longrightarrow\ Y_\ell^{m}=(-1)^m\sqrt{\frac{2\ell+1}{4\pi}\frac{(\ell-m)!}{(\ell+m)!}}\;P_\ell^{m}(\cos\theta)\,e^{im\phi} \]

Normalize over the solid angle with the correct \(\sin\theta\,d\theta\,d\phi\) measure; the Condon–Shortley phase \((-1)^m\) is a conventional choice fixing the relative signs across the multiplet. B

Result

\[ \boxed{\ \hat{L}^2\,Y_\ell^{m}=\hbar^2\,\ell(\ell+1)\,Y_\ell^{m},\qquad \hat{L}_z\,Y_\ell^{m}=\hbar m\,Y_\ell^{m},\qquad Y_\ell^{m}(\theta,\phi)=(-1)^m\sqrt{\tfrac{2\ell+1}{4\pi}\tfrac{(\ell-m)!}{(\ell+m)!}}\;P_\ell^{m}(\cos\theta)\,e^{im\phi}\ } \]

Reading. The angular part of any central-force wavefunction is one of a discrete grid of functions labelled by two integers: \(\ell\ge0\) sets the total angular momentum magnitude \(\hbar\sqrt{\ell(\ell+1)}\), and \(m\) (with \(|m|\le\ell\)) sets its projection \(\hbar m\) on the chosen axis. The \(e^{im\phi}\) factor carries the azimuthal circulation; \(P_\ell^m(\cos\theta)\) shapes the polar profile and has exactly \(\ell-|m|\) nodes in \(\theta\).

Units check. \(Y_\ell^m\) is dimensionless: \(\int|Y_\ell^m|^2\,d\Omega=1\) with \(d\Omega\) in steradians (dimensionless), so \(|Y_\ell^m|^2\) is a probability per unit solid angle and \(Y_\ell^m\) is a pure number. The eigenvalue \(\hbar^2\ell(\ell+1)\) carries \([\text{J·s}]^2=[\text{angular momentum}]^2\), and \(\hbar m\) carries \([\text{J·s}]\), exactly as required for \(\hat{L}^2\) and \(\hat{L}_z\).

Limiting cases

\(\ell=0\): \(Y_0^0=\tfrac{1}{\sqrt{4\pi}}\), a constant — the isotropic \(s\)-state, no angular structure and no circulation.
\(m=0\): \(e^{im\phi}\to1\), \(Y_\ell^0=\sqrt{\tfrac{2\ell+1}{4\pi}}\,P_\ell(\cos\theta)\) is real and axially symmetric, built from ordinary Legendre polynomials.
\(m=\pm\ell\) (top/bottom rung): \(Y_\ell^{\pm\ell}\propto(\sin\theta)^\ell e^{\pm i\ell\phi}\), sharply peaked at the equator \(\theta=\tfrac{\pi}{2}\) as \(\ell\) grows — the state hugging the classical orbital plane.
Large \(\ell\): \(\sqrt{\ell(\ell+1)}\to\ell+\tfrac12\approx\ell\); the discrete spectrum crowds toward the classical continuum (correspondence principle), and high-\(\ell\), high-\(|m|\) states approach a definite classical orbit.

Breaks when

Half-integer angular momentum (spin). The ladder algebra permits \(j=\tfrac12,\tfrac32,\dots\), but \(e^{im\phi}\) with half-integer \(m\) is not single-valued on the sphere. Spin has no position-space eigenfunction of this form; it lives in an internal Hilbert space, not on \(S^2\). Orbital quantum numbers are therefore strictly integer.
Broken rotational symmetry / relativistic coupling. In a non-central potential, or once spin–orbit coupling is included, \(\hat{L}^2\) and \(\hat{L}_z\) no longer commute with \(\hat H\); the \(Y_\ell^m\) still exist as functions but cease to label stationary states — one must use \(\hat{J}^2,\hat{J}_z\) with \(\hat{\vec J}=\hat{\vec L}+\hat{\vec S}\).
Reduced dimensionality. On a plane or line the configuration space is not \(S^2\); angular momentum has a single component with continuous or differently quantized spectrum (in 2D, \(e^{im\phi}\) survives but there is no \(\hat{L}^2\) ladder), so this construction does not apply.

Failure modes

\(\ell^2\) vs \(\ell(\ell+1)\). Writing \(\hat{L}^2 Y=\hbar^2\ell^2 Y\); the correct eigenvalue is \(\hbar^2\ell(\ell+1)\), which is why \(|\vec L|>|L_z|_{\max}\) always.
"\(L_z\) can equal \(|\vec L|\)." Claiming a state with \(L_z=\hbar\ell\) points fully along \(z\). Since \(\hbar\sqrt{\ell(\ell+1)}>\hbar\ell\), the vector never aligns — a direct consequence of \(\hat{L}_x,\hat{L}_y\) not being simultaneously sharp.
Dropping the \(\sin\theta\) Jacobian. Normalizing with \(d\theta\,d\phi\) instead of \(\sin\theta\,d\theta\,d\phi\); this gives wrong prefactors and spurious non-orthogonality.
Allowing half-integer \(\ell\) for orbital motion. Forgetting that single-valuedness, not the algebra, is what forbids it.
Misplacing the Condon–Shortley phase. Applying \((-1)^m\) to both \(Y_\ell^m\) and \(P_\ell^m\), double-counting the sign; the phase convention lives in one place only.
Assigning units to \(Y_\ell^m\). Treating it as an amplitude with dimension of inverse length; it is dimensionless because the sphere's measure is dimensionless.

Discussion

The decisive step is not algebraic but topological. The ladder-spectrum result is representation theory of \(\mathfrak{su}(2)\) and knows nothing of coordinates; it happily produces spin-\(\tfrac12\). What selects integer \(\ell\) for orbital angular momentum is that the wavefunction must be a genuine function on the sphere \(S^2\), whose azimuthal loop closes after \(2\pi\). Half-integer representations require a double cover (the group \(SU(2)\) rather than \(SO(3)\)), which a single-valued position wavefunction cannot supply. This is the concrete meaning of the slogan "orbital angular momentum is quantized in integers."

The \(Y_\ell^m\) are a complete orthonormal basis for square-integrable functions on the sphere, \(\int Y_{\ell'}^{m'*}Y_\ell^m\,d\Omega=\delta_{\ell\ell'}\delta_{mm'}\). Completeness is why they appear everywhere: the multipole expansion, Laplace's equation in spherical regions, the CMB temperature map, and the addition theorem all rest on expanding an angular function in this basis. The degeneracy — \(2\ell+1\) states sharing one \(\hat{L}^2\) eigenvalue — is the fingerprint of the unbroken rotational symmetry \(SO(3)\), and any splitting of that degeneracy (a magnetic field, a crystal field) is a direct readout of how the symmetry is lowered.

A deeper viewpoint recognizes \(\hat{L}^2\) as (minus \(\hbar^2\) times) the Laplace–Beltrami operator on \(S^2\), so the \(Y_\ell^m\) are literally the vibrational eigenmodes of a sphere and the eigenvalue \(\ell(\ell+1)\) is the corresponding "frequency." The lowering operator \(\hat{L}_-\) realizes the intertwiner that maps one weight space to the next within an irreducible \(SO(3)\) representation; the fact that \((\hat L_-)^{\ell-m}\) applied to the highest weight exhausts the multiplet and then annihilates the lowest weight (\(\hat{L}_-Y_\ell^{-\ell}=0\)) is the statement that the representation is finite-dimensional and irreducible. The associated Legendre functions are precisely the matrix elements of finite rotations restricted to this representation.

Common misconceptions. \(m\) is not the angle of the angular-momentum vector — it is the sharp \(z\)-projection of an otherwise uncertain vector; \(\hat{L}_x\) and \(\hat{L}_y\) have zero mean but nonzero spread in a \(Y_\ell^m\) state. The real orbitals \(p_x,p_y,d_{xy},\dots\) used in chemistry are superpositions of \(Y_\ell^{\pm m}\), not eigenstates of \(\hat{L}_z\); they trade a definite \(m\) for a definite spatial shape. Finally, "more nodes means higher energy" is a radial statement — here the angular node count \(\ell-|m|\) tracks angular momentum, not energy directly.

Worked examples

Construct the \(\ell=1\) multiplet from the top state by lowering. Start from \(Y_1^{1}\) and generate \(Y_1^{0}\).

\[ Y_1^{1}=c_1(\sin\theta)e^{i\phi};\qquad 1=|c_1|^2\!\int_0^{2\pi}\!\!d\phi\int_0^{\pi}\!\sin^2\theta\,\sin\theta\,d\theta=|c_1|^2\,(2\pi)\!\left(\tfrac{4}{3}\right)\ \Rightarrow\ |c_1|^2=\tfrac{3}{8\pi} \]

\[ Y_1^{1}=-\sqrt{\tfrac{3}{8\pi}}\,\sin\theta\,e^{i\phi}\quad(\text{Condon–Shortley }(-1)^1) \]

\[ \hat{L}_-Y_1^{1}=\hbar\,e^{-i\phi}\!\left(-\frac{\partial}{\partial\theta}+i\cot\theta\frac{\partial}{\partial\phi}\right)\!\Big[-\sqrt{\tfrac{3}{8\pi}}\sin\theta\,e^{i\phi}\Big]=\hbar\,(2\cos\theta)\sqrt{\tfrac{3}{8\pi}} \]

\[ \hat{L}_-Y_1^{1}=\hbar\sqrt{1(1{+}1)-1(1{-}1)}\;Y_1^{0}=\hbar\sqrt{2}\;Y_1^{0}\ \Rightarrow\ Y_1^{0}=\frac{2\sqrt{3/8\pi}}{\sqrt2}\cos\theta=\sqrt{\tfrac{3}{4\pi}}\cos\theta \]

\[ Y_1^{1}=-\sqrt{\tfrac{3}{8\pi}}\sin\theta\,e^{i\phi},\quad Y_1^{0}=\sqrt{\tfrac{3}{4\pi}}\cos\theta,\quad Y_1^{-1}=+\sqrt{\tfrac{3}{8\pi}}\sin\theta\,e^{-i\phi} \]

Reading. Lowering once from the equatorial \(\sin\theta\) state produces the axial \(\cos\theta\) state; the ladder coefficient \(\sqrt2\) is exactly what makes the normalized \(Y_1^0\) come out with the standard \(\sqrt{3/4\pi}\). Applying \(\hat{L}_-\) again reaches \(Y_1^{-1}\), and a third application annihilates it.

Expectation values in a real \(p\)-orbital superposition. A particle on the sphere is in \(\psi=\tfrac{1}{\sqrt2}\big(Y_1^{1}+Y_1^{-1}\big)\). Find \(\langle\hat{L}_z\rangle\), \(\langle\hat{L}^2\rangle\), \(\langle\hat{L}_z^2\rangle\), and identify the state.

\[ \langle\hat{L}_z\rangle=\tfrac12\big[(+\hbar)+(-\hbar)\big]=0,\qquad \langle\hat{L}_z^2\rangle=\tfrac12\big[(\hbar)^2+(-\hbar)^2\big]=\hbar^2 \]

\[ \text{both terms have }\ell=1\ \Rightarrow\ \langle\hat{L}^2\rangle=\hbar^2\,\ell(\ell+1)=2\hbar^2 \]

\[ Y_1^{1}+Y_1^{-1}=\sqrt{\tfrac{3}{8\pi}}\sin\theta\big(e^{-i\phi}-e^{i\phi}\big)=-2i\sqrt{\tfrac{3}{8\pi}}\sin\theta\sin\phi\ \propto\ \frac{y}{r} \]

\[ \langle\hat{L}_z\rangle=0,\qquad \langle\hat{L}^2\rangle=2\hbar^2,\qquad \langle\hat{L}_z^2\rangle=\hbar^2 \]

Reading. Equal-weight mixing of \(m=+1\) and \(m=-1\) gives zero mean projection but a nonzero spread \(\sqrt{\langle L_z^2\rangle}=\hbar\); the state is the real \(p_y\) orbital (up to an overall phase), spatially oriented rather than circulating. A measurement of \(L_z\) yields \(+\hbar\) or \(-\hbar\) with probability \(\tfrac12\) each — never \(0\), despite the mean being \(0\).

Units check. \(\langle\hat{L}^2\rangle=2\hbar^2\) carries \([\text{J·s}]^2\); \(\langle\hat{L}_z\rangle=0\) and \(\sqrt{\langle L_z^2\rangle}=\hbar\) carry \([\text{J·s}]\), as required.

Problems

Show directly that \(\hat{L}_z\,e^{im\phi}=\hbar m\,e^{im\phi}\), and argue from single-valuedness that \(m\) must be an integer. What physical assumption would you have to abandon to allow \(m=\tfrac12\)?
Solution
\(\hat{L}_z=-i\hbar\,\partial_\phi\), so \(\hat{L}_z e^{im\phi}=-i\hbar\,(im)\,e^{im\phi}=\hbar m\,e^{im\phi}\); it is an eigenfunction with eigenvalue \(\hbar m\). Physical wavefunctions on the sphere must satisfy \(\psi(\phi+2\pi)=\psi(\phi)\), i.e. \(e^{im(\phi+2\pi)}=e^{im\phi}\Rightarrow e^{2\pi i m}=1\Rightarrow m\in\mathbb{Z}\). To allow \(m=\tfrac12\) one must abandon single-valuedness of the position-space wavefunction — permissible only for internal (spin) degrees of freedom, which are not functions on \(S^2\).
Solve the top-state condition \(\hat{L}_+Y_2^{2}=0\) for \(\ell=2\) and normalize to obtain \(Y_2^{2}\).
Solution
Writing \(Y_2^2=P(\theta)e^{2i\phi}\), the equation \(\hat{L}_+Y_2^2=0\) gives \(P'-2\cot\theta\,P=0\Rightarrow P\propto(\sin\theta)^2\), so \(Y_2^2=c\,(\sin\theta)^2 e^{2i\phi}\). Normalize: \(1=|c|^2(2\pi)\int_0^\pi\sin^5\theta\,d\theta=|c|^2(2\pi)\big(\tfrac{16}{15}\big)\Rightarrow|c|^2=\tfrac{15}{32\pi}\). With Condon–Shortley phase \((-1)^2=+1\), \(\ Y_2^{2}=\tfrac14\sqrt{\tfrac{15}{2\pi}}\,\sin^2\theta\,e^{2i\phi}\) (check: \(\tfrac{1}{16}\cdot\tfrac{15}{2\pi}=\tfrac{15}{32\pi}\), consistent).
Verify by direct differentiation that \(Y_1^{0}=\sqrt{\tfrac{3}{4\pi}}\cos\theta\) is an eigenfunction of \(\hat{L}^2\) and read off the eigenvalue.
Solution
With no \(\phi\)-dependence, \(\hat{L}^2 Y_1^0=-\hbar^2\,\tfrac{1}{\sin\theta}\tfrac{d}{d\theta}\!\big(\sin\theta\,\tfrac{d}{d\theta}\cos\theta\big)\sqrt{\tfrac{3}{4\pi}}\). Now \(\tfrac{d}{d\theta}\cos\theta=-\sin\theta\), so \(\sin\theta\cdot(-\sin\theta)=-\sin^2\theta\), and \(\tfrac{d}{d\theta}(-\sin^2\theta)=-2\sin\theta\cos\theta\). Dividing by \(\sin\theta\) gives \(-2\cos\theta\). Hence \(\hat{L}^2 Y_1^0=-\hbar^2(-2\cos\theta)\sqrt{\tfrac{3}{4\pi}}=2\hbar^2\,Y_1^0\). The eigenvalue is \(2\hbar^2=\hbar^2\ell(\ell+1)\) with \(\ell=1\).
For a particle in the state \(Y_1^{1}\), compute the probability of finding it in the upper hemisphere (\(0\le\theta\le\tfrac{\pi}{2}\)) and give \(\langle\hat{L}_z\rangle\).
Solution
\(|Y_1^1|^2=\tfrac{3}{8\pi}\sin^2\theta\). \(P_{\text{upper}}=\int_0^{2\pi}\!\!d\phi\int_0^{\pi/2}\tfrac{3}{8\pi}\sin^2\theta\,\sin\theta\,d\theta=\tfrac{3}{8\pi}(2\pi)\int_0^{\pi/2}\sin^3\theta\,d\theta=\tfrac34\cdot\tfrac23=\tfrac12\). The result is exactly \(\tfrac12\) because \(|Y_1^1|^2\) is symmetric under \(\theta\to\pi-\theta\). Since \(Y_1^1\) is a pure \(m=1\) eigenstate, \(\langle\hat{L}_z\rangle=\hbar\).
Show that under the parity operation \(\vec r\to-\vec r\) (i.e. \(\theta\to\pi-\theta,\ \phi\to\phi+\pi\)) the spherical harmonics obey \(Y_\ell^{m}\to(-1)^\ell Y_\ell^{m}\).
Solution
The azimuthal factor: \(e^{im(\phi+\pi)}=(-1)^m e^{im\phi}\). The polar factor uses \(\cos(\pi-\theta)=-\cos\theta\) and the Legendre parity \(P_\ell^m(-x)=(-1)^{\ell+m}P_\ell^m(x)\), so \(P_\ell^m(\cos\theta)\to(-1)^{\ell+m}P_\ell^m(\cos\theta)\). Multiplying, the total factor is \((-1)^m(-1)^{\ell+m}=(-1)^{\ell+2m}=(-1)^\ell\). Hence \(Y_\ell^m\) has definite parity \((-1)^\ell\), independent of \(m\) — the origin of the parity selection rule \(\Delta\ell=\pm1\) for electric-dipole transitions.