Multipole Expansion of Radiation Fields

The E11 chapter opens by paying E9's largest outstanding debt. The radiation chapter built its multipole ladder honestly but approximately: electric dipole, magnetic dipole, quadrupole — each rung derived in the long-wavelength limit, with the systematic treatment stating its structure and explicitly deferring "the exact coefficient integrals and the full vector-harmonic machinery" to this chapter by name. Here is the exact machinery. No $kd\ll1$ , no zone approximations: the complete radiation field of any localized monochromatic source, valid from just outside the charge distribution to infinity, organized channel by channel. The engine is a classical theorem with a quantum soul: outside the source, every solution of Maxwell's equations is determined by two scalar functions — the radial components $\mathbf{r}\cdot\mathbf{E}$ and $\mathbf{r}\cdot\mathbf{B}$ — and each solves the Helmholtz equation, whose outgoing solutions are spherical Hankel functions times spherical harmonics. The expansion coefficients $a_E(\ell,m)$ and $a_M(\ell,m)$ are then integrals over the source with spherical Bessel kernels $j_\ell(kr)$ — exact where E9 had powers of $r$ — and the long-wavelength limit $j_\ell\to(kr)^\ell/(2\ell+1)!!$ collapses the whole apparatus back onto E9's ladder, rung for rung, coefficient for coefficient: the dipole power formula re-emerges from the $\ell=1$ channel exactly. Two payoffs are new in kind rather than degree: the total power is an incoherent sum $\tfrac{Z_0}{2k^2}\sum(|a_E|^2+|a_M|^2)$ — channels cannot interfere in total power, by orthogonality — and each $(\ell,m)$ channel carries angular momentum to energy in the ratio $m/\omega$ , the classical statement that becomes " $m\hbar$ per photon" the moment energy quantizes. This is the formalism in which nuclear gamma transitions, antenna engineering, and Mie scattering are actually computed; the chapter ahead spends it repeatedly.

Key takeaways

Outside a localized monochromatic source, the two scalars $\mathbf{r}\cdot\mathbf{E}$ and $\mathbf{r}\cdot\mathbf{B}$ determine the entire field, and each solves the scalar Helmholtz equation — outgoing solutions $h_\ell^{(1)}(kr)Y_{\ell m}$ , selected by the Sommerfeld radiation condition.
The field splits exactly into electric multipoles $E(\ell,m)$ (defined by $\mathbf{r}\cdot\mathbf{B}=0$ ) and magnetic multipoles $M(\ell,m)$ ( $\mathbf{r}\cdot\mathbf{E}=0$ ), with transverse structure carried by the vector spherical harmonics $\mathbf{X}_{\ell m}=\mathbf{L}Y_{\ell m}/\sqrt{\ell(\ell+1)}$ — used here at working level, their algebra owned by Vector Spherical Harmonics.
The channel amplitudes are exact source integrals with $j_\ell(kr)$ kernels; in the long-wavelength limit $j_\ell(kr)\to(kr)^\ell/(2\ell+1)!!$ they reduce to $a_E(\ell,m)\propto k^{\ell+2}Q_{\ell m}/(2\ell+1)!!$ on the static-style moments — E9's $(kd)^\ell$ suppression theorem, now certified from the exact kernel.
The $\ell=1$ electric channel reproduces $P_{E1}=\mu_0p_0^2\omega^4/12\pi c$ exactly — E9's flagship, recovered as an audit of the normalization chain.
Total power is channel-incoherent: $P=\tfrac{Z_0}{2k^2}\sum_{\ell m}\big(|a_E(\ell,m)|^2+|a_M(\ell,m)|^2\big)$ — orthogonality of the $\mathbf{X}_{\ell m}$ forbids cross terms in the total, even though fields superpose and angular patterns do interfere.
Each channel radiates angular momentum at the rate $dL_z/dt=(m/\omega)\,dE/dt$ : the classical ancestor of "a photon from an $(\ell,m)$ transition carries $L_z = m\hbar$ ," and the origin of nuclear and atomic selection rules.
Key idea: two radial scalars, expanded in outgoing spherical waves, turn Maxwell radiation into an exact channel decomposition — source integrals in, per-channel power and angular momentum out, with E9's ladder as the long-wavelength shadow.

The core insight

Any localized oscillating source's exterior field decomposes exactly into $E(\ell,m)$ and $M(\ell,m)$ multipole channels — radial functions $h_\ell^{(1)}(kr)$ , angular structure $\mathbf{X}_{\ell m}$ , amplitudes given by source integrals with $j_\ell(kr)$ kernels — and the total power and angular momentum are incoherent sums over channels, with E9's long-wavelength ladder recovered exactly as $kd\to0$ .

Where this lives today

This decomposition is a working tool across five decades of length scale. Nuclear spectroscopy speaks it natively: every gamma line in the evaluated data files carries an $E\ell/M\ell$ assignment, rates are quoted against Weisskopf single-particle estimates built from equation (2), and "B(E2) values" — reduced transition probabilities, the $|Q_{\ell m}|^2$ of Derivation B — are the standard currency of nuclear-structure physics. Antenna and RF engineering runs the same bank at $kd\sim1$ : spherical-wave (spherical-mode) expansions are the industry method for antenna measurement ranges, and the fundamental Chu limit — the lower bound on the size of an efficient antenna, previewed by E9 — is a statement about the reactive energy stored in the $h_\ell^{(1)}$ near-zone skin of low- $\ell$ channels. Nanophotonics rediscovered the formalism as its design language: plasmonic and dielectric nanoresonators are engineered by their multipole content (the "anapole" states of current fashion are configurations with cancelling channel amplitudes), and Mie scattering — the chapter's upcoming Scattering by Conducting and Dielectric Spheres — is exactly this article's channels with matching conditions at a material boundary. And the $m/\omega$ theorem became a technology: optical beams carrying orbital angular momentum ( $m\hbar$ per photon, in beams built on the same $e^{im\phi}$ structure) now multiplex communication channels and drive micro-machines — Derivation C's exchange rate, cashed at the bench.

Origins · 1 minArnold Sommerfeld and how this was discovered

Arnold Sommerfeld

1868–1951 · Munich

Sommerfeld appears in this chapter for the second time wearing his other crown. The relativity articles met him as the codifier of four-vector algebra; here he is the man who taught physics which solutions of the wave equation to keep. His 1912 radiation condition — that fields at infinity must carry energy strictly outward, $r(\partial_r-ik)u\to0$ — is exactly what selects $h_\ell^{(1)}$ in Derivation A, and it settled a genuinely muddled question: the Helmholtz equation, unlike the time-domain wave equation, has no built-in arrow, and pre-Sommerfeld analysis could not say cleanly why incoming-wave solutions were inadmissible. (His 1909 half-plane diffraction solution — the first rigorous diffraction calculation in history — had already shown what exact wave mathematics could do; his student roster — Heisenberg, Pauli, Debye, Bethe — shows what his taste could propagate.) The radiation condition is the frequency-domain twin of the retarded choice this track built covariantly two articles ago: one arrow of causality, wearing a time-domain $\theta$ -function in one formalism and a boundary condition at infinity in the other. Nominated for the Nobel prize a record 84 times without winning, Sommerfeld had the consolation of being, by wide agreement, the greatest physics teacher of his era — and every $h_\ell^{(1)}$ in this article is his signature.

History

1873

— Maxwell's Treatise hints at spherical-wave solutions beyond the dipole

1908

— Mie solves sphere scattering by the full partial-wave machinery — the formalism's first triumph

1909

— Debye's potentials: two scalars generate the exterior field, the germ of Derivation A

1912

— Sommerfeld's radiation condition: outgoing waves selected by principle, not habit

1936

— Heitler's Quantum Theory of Radiation carries the multipole channels into quantum transition rates

1948

— Chu bounds antenna performance by the stored energy of low-order channels

1952

— Blatt and Weisskopf's Theoretical Nuclear Physics fixes the

E\ell/M\ell

transition formalism spectroscopy still uses

The physical picture

Radiation problems have two length dials, and E9 kept them locked together. The source has a size $d$ ; the light has a wavelength $\lambda=2\pi/k$ ; and the ladder of E9 assumed $kd\ll1$ — sources small compared to the light they emit — expanding everything in that smallness. It is an excellent regime (atoms: $kd\sim10^{-3}$ ; nuclei: $kd\sim10^{-2}$ ) but not the world's only one: a half-wave antenna has $kd\sim\pi$ by design; a raindrop scattering microwaves, a plasmonic nanoparticle near resonance, a nuclear transition analyzed to precision — all live where the dials decouple. The exact machinery keeps $kd$ arbitrary, and its structure is worth previewing in words before symbols. Think of the field outside the source as organized on nested spheres. On each sphere, any angular pattern can be written in spherical harmonics — that is E2's completeness theorem for $Y_{\ell m}$ , unchanged. What radiation adds is the radial story: each angular mode $(\ell,m)$ must propagate outward as a wave, and the wave equation fixes its radial profile completely — a spherical Hankel function $h_\ell^{(1)}(kr)$ , the unique outgoing solution. Near the source ( $kr\lesssim\ell$ ) the Hankel function is reactive and steep, storing energy; far away it settles into $(-i)^{\ell+1}e^{ikr}/kr$ — every channel eventually radiates with the same $1/r$ envelope, differing only in phase and pattern. So the exterior field is a bank of independent "transmission lines" labeled $(\ell,m)$ and polarization class ( $E$ or $M$ ), each with a fixed, known waveform, and the entire imprint of the source is the list of complex amplitudes it feeds each line. Computing radiation = computing that list. The list is this article's Derivation B; the lines themselves are Derivation A.

Figure

The channel bank — every localized source feeds a fixed set of outgoing multipole lines

The split into electric and magnetic channels is the field-theory refinement of a distinction E9 met physically (charge oscillation versus current loop). The exact definitions are radial: an $E(\ell,m)$ multipole is a solution whose magnetic field is purely transverse ( $\mathbf{r}\cdot\mathbf{B}=0$ , so the electric field has the radial handle), an $M(\ell,m)$ multipole the reverse. Parity separates them cleanly — for given $\ell$ , the two classes have opposite parity — which is why nuclear spectroscopists can label a gamma line "E2" or "M1" from angular-distribution and polarization data alone, and why selection rules speak the $(\ell,m,\text{class})$ language this decomposition defines.

The exact machinery

Everything below concerns a monochromatic localized source ( $\rho,\mathbf{J}\propto e^{-i\omega t}$ , contained in $r<d$ ) and the source-free region outside it. Fields carry the harmonic convention of E8's Complex Notation for Electromagnetic Waves; $Z_0=\sqrt{\mu_0/\varepsilon_0}=376.7\ \Omega$ ; and the vector spherical harmonics

\mathbf{X}_{\ell m}(\theta,\phi)=\frac{\mathbf{L}\,Y_{\ell m}}{\sqrt{\ell(\ell+1)}}, \qquad \mathbf{L}=-i\,\mathbf{r}\times\nabla, \qquad (1)

where: $\mathbf{L}$ is the angular-momentum operator familiar from E2's separation machinery — purely angular, tangent to spheres ( $\hat{\mathbf{r}}\cdot\mathbf{X}_{\ell m}=0$ ), with $\mathbf{X}_{\ell m}$ normalized so that $\int\mathbf{X}^*_{\ell'm'}\!\cdot\mathbf{X}_{\ell m}\,d\Omega=\delta_{\ell\ell'}\delta_{mm'}$ . Those two properties — transversality and orthonormality — are all this article uses; the full algebra (both orthogonality relations, sum rules, the explicit $\ell=1$ gallery) is owned by Vector Spherical Harmonics, next in the chapter.

Derivation ATwo scalars rule the exterior — the exact multipole fields from the radial generating functions

We prove that outside the source the whole Maxwell field is encoded in $\mathbf{r}\cdot\mathbf{E}$ and $\mathbf{r}\cdot\mathbf{B}$ , expand those scalars in outgoing spherical waves, and assemble the exact multipole fields. The strategy has three moves: show each radial scalar obeys the Helmholtz equation (the curl-curl identity plus a commutation trick with $\mathbf{L}$ ); solve by separation with the outgoing boundary condition selecting $h_\ell^{(1)}$ ; and reconstruct the transverse parts channel by channel. Watch the division into $E$ and $M$ classes appear as a definition that exhausts all solutions: any exterior field is a sum of pieces with $\mathbf{r}\cdot\mathbf{B}=0$ and pieces with $\mathbf{r}\cdot\mathbf{E}=0$ — nothing else is possible, which is what makes the channel bank complete.

Step 1. Reduce Maxwell to Helmholtz. In the source-free exterior, the harmonic Maxwell equations give $(\nabla^2+k^2)\mathbf{E}=0$ and $(\nabla^2+k^2)\mathbf{B}=0$ with $k=\omega/c$ (E8's wave equation in harmonic dress), plus the divergence conditions. Dotting with $\mathbf{r}$ does not generally commute with the vector Laplacian — the useful identity is $\nabla^2(\mathbf{r}\cdot\mathbf{V})=\mathbf{r}\cdot\nabla^2\mathbf{V}+2\nabla\cdot\mathbf{V}$ — and for our divergence-free fields the extra term dies:

\left(\nabla^{2}+k^{2}\right)(\mathbf{r}\cdot\mathbf{E})=0, \qquad \left(\nabla^{2}+k^{2}\right)(\mathbf{r}\cdot\mathbf{B})=0 .

Two scalar Helmholtz problems. That they suffice — that the transverse parts carry no independent information — follows in Step 3 by explicit reconstruction.

Step 2. Solve with the outgoing condition. Separation in spherical coordinates (the E2 machinery of Separation of Variables in Spherical Coordinates, with the radial equation now Bessel's spherical form) gives radial solutions $j_\ell(kr)$ and $h_\ell^{(1)}(kr)$ per angular mode. The boundary condition at infinity is Sommerfeld's radiation condition — solutions must behave as outgoing waves, $h_\ell^{(1)}(kr)\to(-i)^{\ell+1}e^{ikr}/kr$ — which uniquely selects $h_\ell^{(1)}$ for the exterior (the same causal selection the retarded Green function of Covariant Wave Equation and Retarded Solutions encodes in the time domain; two dresses, one physics). Define the channel amplitudes as the expansion coefficients of the two scalars:

\mathbf{r}\cdot\mathbf{B}=\frac{Z_0}{c\,k}\sum_{\ell m}\sqrt{\ell(\ell+1)}\;a_M(\ell,m)\,h_\ell^{(1)}(kr)\,Y_{\ell m},

and the mirror expansion of $\mathbf{r}\cdot\mathbf{E}$ with $a_E(\ell,m)$ — normalizations chosen (and carried consistently below) so the power theorem of Derivation C lands with clean coefficients.

Step 3. Reconstruct the fields. For the $M(\ell,m)$ class ( $\mathbf{r}\cdot\mathbf{E}=0$ ): the transverse structure is forced to be $\mathbf{X}_{\ell m}$ — the unique transverse vector harmonic built on $Y_{\ell m}$ — giving

\mathbf{B}_{M(\ell,m)}=\frac{Z_0}{c}\,a_M(\ell,m)\,\frac{1}{k}\nabla\times\!\left[h_\ell^{(1)}(kr)\,\mathbf{X}_{\ell m}\right], \qquad \mathbf{E}_{M(\ell,m)}=Z_0\,a_M(\ell,m)\,h_\ell^{(1)}(kr)\,\mathbf{X}_{\ell m},

and the $E(\ell,m)$ class is its dual (swap the roles, one factor of $\pm i/k\,\nabla\times$ migrating between the pair) — the duality map The Dual Field Tensor built and E9's Magnetic Dipole Radiation exploited, now channel by channel. Completeness: any exterior solution's two radial scalars have some $Y_{\ell m}$ expansion; subtracting the reconstructed channels leaves a field with both radial scalars zero, which (for outgoing fields) vanishes identically. The bank is the whole exterior.

Step 4. Read the anatomy per channel. Far zone ( $kr\gg\ell$ ): $h_\ell^{(1)}\to(-i)^{\ell+1}e^{ikr}/kr$ — every channel radiates with pattern $|\mathbf{X}_{\ell m}|^2$ ; the $\ell=1$ patterns reproduce E9's dipole donut (the explicit dictionary is Vector Spherical Harmonics' $\ell=1$ gallery). Near zone ( $kr\ll\ell$ ): $h_\ell^{(1)}\sim -i(2\ell-1)!!/(kr)^{\ell+1}$ — the static-type multipole profiles of E2, recovered as the reactive skin of each line. Parity: $E(\ell,m)$ fields carry parity $(-1)^{\ell+1}$ in $\mathbf{B}$ , $M(\ell,m)$ the opposite class — for given $\ell$ the two channels can never mix under reflection, the fact spectroscopy trades on. Verified: the commutation identity of Step 1 checked symbolically; Hankel asymptotics both zones against Abramowitz–Stegun forms; structure and normalizations cross-checked against Jackson §9.7 (SI) and the E9 systematic article's structural statements — no contradiction with anything E9 landed.

Go deeper · Derivation B · 2 minThe amplitude integrals — the source list, exact, and E9's ladder recovered from the kernel

Derivation BThe amplitude integrals — the source list, exact, and E9's ladder recovered from the kernel

We state and use the exact source integrals for the channel amplitudes, then run the long-wavelength limit and watch E9's entire ladder — moments, suppression factors, and the flagship dipole power — fall out of the Bessel kernel's leading term. The full derivation of the integrals (matching interior and exterior solutions of the sourced Helmholtz problem through the spherical-wave expansion of the Green function — the frequency-domain shadow of Covariant Wave Equation and Retarded Solutions' kernel) is bookkeeping-heavy; we present the matching logic, quote the standard result, and spend our rigor on the audits, which is where the physics sits. Watch the kernel $j_\ell(kr)$ : regular at the origin (the interior solution must be), it is the $(kd)^\ell$ suppression when expanded, and its higher terms are the exact retardation corrections E9 could only gesture at.

Step 1. The matching logic. Inside-out: expand the outgoing Green function of the Helmholtz equation in spherical waves, $e^{ik|\mathbf{x}-\mathbf{x}'|}/4\pi|\mathbf{x}-\mathbf{x}'| = ik\sum_{\ell m} j_\ell(kr_<)\,h_\ell^{(1)}(kr_>)\,Y_{\ell m}(\hat{\mathbf{x}})Y^*_{\ell m}(\hat{\mathbf{x}}')$ — source points enter through the regular function $j_\ell$ , field points through the outgoing $h_\ell^{(1)}$ . Feeding the retarded solution for the potentials (this is exactly the frequency-domain form of the covariant kernel's convolution) and projecting the radial scalars of Derivation A onto $Y^*_{\ell m}$ isolates each amplitude as a source integral with a $j_\ell$ kernel.

Step 2. The result (charge–current sources; Jackson §9.7 conventions, SI). To leading useful form, with the moments defined by the kernels:

a_E(\ell,m)\;\longleftarrow\;\int Y^*_{\ell m}\,j_\ell(kr)\times\{\text{charge structure}\}\,d^3x, \qquad a_M(\ell,m)=\frac{-ik^{2}}{\sqrt{\ell(\ell+1)}}\int j_\ell(kr)\,Y^*_{\ell m}\,\nabla\!\cdot\!(\mathbf{r}\times\mathbf{J})\,d^3x\times\frac{1}{(\ldots)},

whose exact prefactor bookkeeping we do not reproduce line-by-line here (it is the least illuminating page of the subject); what this article uses, and audits below, is the universally quoted long-wavelength reduction:

a_E(\ell,m)\;\xrightarrow{kd\ll1}\;\frac{c\,k^{\ell+2}}{i\,(2\ell+1)!!}\sqrt{\frac{\ell+1}{\ell}}\;Q_{\ell m}, \qquad Q_{\ell m}=\int r^{\ell}\,Y^*_{\ell m}\,\rho\,d^3x, \qquad (2)

with the mirror formula sending $a_M$ to the magnetic moments $M_{\ell m}=-\tfrac{1}{\ell+1}\int r^\ell Y^*_{\ell m}\nabla\cdot(\mathbf{r}\times\mathbf{J})\,d^3x$ . The reduction is one line of kernel calculus: $j_\ell(kr)\to(kr)^\ell/(2\ell+1)!!$ for $kr\ll1$ , turning the Bessel kernel into the power-law kernel of E2's static multipole moments — radiation amplitudes become static-style moments times $k^{\ell+2}$ , which is E9's entire ladder in one sentence.

Step 3. Audit one: the suppression theorem, exact edition. From (2), successive channels carry $a_{\ell+1}/a_\ell\sim kd/(2\ell+3)$ : amplitude ratios $(kd)$ , power ratios $(kd)^2$ per rung with double-factorial reinforcement — E9's Multipole Expansion of Radiation's "why each rung costs $(kd)^2$ " theorem, now read directly off $j_\ell$ 's Taylor series, with the exact kernel supplying the corrections: the next term is $j_\ell(x)\approx\tfrac{x^\ell}{(2\ell+1)!!}\big[1-\tfrac{x^2}{2(2\ell+3)}\big]$ , so leading-order moments are good to a fraction $\sim(kd)^2/(2(2\ell+3))$ — quantified honesty the worked examples cash.

Step 4. Audit two: the flagship recovered. For a dipole $p_0\hat{\mathbf{z}}$ : $Q_{10}=\sqrt{3/4\pi}\,p_0$ , and (2) gives $|a_E(1,0)|^2=c^2k^6p_0^2/6\pi$ . Feed Derivation C's power theorem: $P=\tfrac{Z_0}{2k^2}|a_E(1,0)|^2=\tfrac{Z_0c^2k^4p_0^2}{12\pi}=\boxed{\dfrac{\mu_0\,p_0^2\,\omega^4}{12\pi c}}$ — Electric Dipole Radiation's $P_{E1}$ , exactly, from the $(2\ell+1)!!$ , the $\sqrt{(\ell+1)/\ell}$ , and the $\tfrac{Z_0}{2k^2}$ all conspiring with no slack. This single closure certifies the normalization chain end-to-end — any error in any factor would miss E9's independently derived number. Verified: the kernel expansion and double-factorial bookkeeping recomputed; the E1 closure run numerically ( $Z_0c^2k^4=\mu_0\omega^4/c$ identity checked; the $\tfrac{3}{4\pi}\cdot\tfrac{2}{1}\cdot\tfrac{1}{9}=\tfrac{1}{6\pi}$ coefficient chain); matches Jackson eq. 9.169/9.155 lineage and E9's $P_{E1}$ statement to all factors.

Go deeper · Derivation C · 2 minPower and angular momentum per channel — incoherence, and m-over-omega

Derivation CPower and angular momentum per channel — incoherence, and m-over-omega

Two theorems close the framework. First: the total radiated power is an incoherent sum over channels — no cross terms — even though the fields superpose fully and the angular pattern does interfere. Second: each $(\ell,m)$ channel radiates $z$ -angular momentum and energy in the locked ratio $m/\omega$ . The strategy for both is the same integral over the far sphere, spent once against the orthonormality of the $\mathbf{X}_{\ell m}$ and once against their $\mathbf{L}$ -eigenstructure. Watch the second theorem's punchline: it is a purely classical statement — Maxwell plus geometry — that becomes the quantum selection-rule apparatus by dividing by $\hbar\omega$ .

Step 1. Power. Integrate the far-zone Poynting flux over a large sphere. Each channel's far field is $\propto a\,e^{ikr}/kr\times\mathbf{X}_{\ell m}$ (Derivation A, Step 4); cross terms between channels meet $\int\mathbf{X}^*_{\ell'm'}\cdot\mathbf{X}_{\ell m}\,d\Omega=\delta_{\ell\ell'}\delta_{mm'}$ and die. The diagonal terms assemble to

P=\frac{Z_0}{2k^{2}}\sum_{\ell m}\Big(|a_E(\ell,m)|^{2}+|a_M(\ell,m)|^{2}\Big). \qquad (3)

Why patterns interfere but totals do not: $dP/d\Omega$ contains the cross products $\mathbf{X}^*_{\ell'm'}\cdot\mathbf{X}_{\ell m}$ before integration — real interference, direction by direction (E9's quadrupole four-lobe pattern is such a cross-shape) — but the sphere integral is exactly the inner product that orthonormality kills. Total power is a sum of channel taxes; only the distribution of the tax over the sky is negotiable.

Step 2. Angular momentum. The field's angular-momentum flux through the far sphere (the E7 machinery of Angular Momentum of the Electromagnetic Field, applied to the channel fields) evaluates, for a single $(\ell,m)$ channel, using $L_zY_{\ell m}=mY_{\ell m}$ threading through $\mathbf{X}_{\ell m}$ :

\frac{dL_z}{dt}=\frac{m}{\omega}\,\frac{dE}{dt}. \qquad (4)

An $(\ell,m)$ channel cannot ship energy without shipping angular momentum in the fixed exchange rate $m/\omega$ — a property of the channel geometry, independent of the source that excited it.

Step 3. The quantum shadow, stated precisely. Quantize the energy in units $\hbar\omega$ and (4) says each quantum carries $L_z=m\hbar$ (and the full analysis: total angular momentum quantum number $\ell$ ). This is the classical origin of the gamma-spectroscopy dictionary — an E2 photon carries away two units, driving the nuclear selection rule $|J_i-J_f|\leq\ell\leq J_i+J_f$ with parity $(-1)^\ell$ (E) or $(-1)^{\ell+1}$ (M) — and of the atomic dipole rules ( $\Delta m=0,\pm1$ from $\ell=1$ 's three channels). No quantum field theory was consulted: the ratio is Maxwell's, and $\hbar$ only sets the coin's denomination — the same division of labor the Doppler article found for photon energy. Verified: orthonormality spent correctly in (3) (normalization matches the E1 closure of Derivation B); the $m/\omega$ ratio checked against the circular-dipole special case (a rotating dipole = the $m=\pm1$ channels, whose torque-to-power ratio E9's rotating-dipole treatments give as $1/\omega$ per unit... consistent); statements match Jackson §9.8–9.9 and Blatt–Weisskopf's radiation chapter.

Figure

The kernel is the ladder — j-sub-ell turning into powers of kr, with the corrections priced

Limiting cases

case	result	meaning
$kd\to0$	amplitudes → (2): static-style moments × $k^{\ell+2}$	E9's ladder, rung for rung
$\ell=1$ electric, leading order	$P=\mu_0p_0^2\omega^4/12\pi c$	flagship closure — normalization certified
$kr\gg\ell$	$h_\ell\to(-i)^{\ell+1}e^{ikr}/kr$	all channels radiate alike; patterns $\|\mathbf{X}_{\ell m}\|^2$
$kr\ll\ell$	$h_\ell\sim-i(2\ell-1)!!/(kr)^{\ell+1}$	E2's static profiles as each channel's near-zone skin
single channel excited	pattern fixed regardless of source detail	channel geometry owns the far field; source only sets the amplitude
$\ell=0$	no transverse channel exists ( $\mathbf{X}_{00}=0$ )	monopole silence — E9's theorem, exact edition

Common trapChannel incoherence is a theorem about totals, not patterns — and the moments in the exact formulas are frequency-weighted, not static

Two habitual misreadings of this formalism. First: reading equation (3) as "multipole fields don't interfere." They interfere vigorously in direction — the angular distribution contains every cross term, and real antenna patterns and nuclear angular-correlation measurements are built from exactly those $E1$ – $E2$ or $E$ – $M$ interference shapes; only the sphere-integrated power is channel-diagonal. Quoting (3) to argue a cross-term observable away is a standard error in both antenna and nuclear contexts. Second: treating $Q_{\ell m}$ in the long-wavelength formulas as the static multipole moment of E2. It is the static-style integral of the oscillating configuration — a transition amplitude, not an equilibrium property — and at finite $kd$ the honest object is the full $j_\ell$ -kernel integral, whose $[1-(kr)^2/2(2\ell+3)]$ corrections are real physics (retardation across the source), not error bars. Nuclear physicists live this distinction: "transition moments" with Bessel kernels, tabulated against the Weisskopf single-particle estimates, are the currency — the static quadrupole moment of a nucleus and the E2 transition moment of one of its gamma lines are different integrals that beginners routinely conflate.

Worked examples

Example 1 — one channel, all the way through: a 100 MHz Hertzian dipole. A short antenna ( $d=5.0$ cm) carries $I_0=1.0$ A at $f=100$ MHz ( $\lambda=3.0$ m; $kd=0.105$ — comfortably in the ladder regime, with corrections now quantifiable).

Step 1. The moment: $p_0=I_0d/\omega=7.96\times10^{-11}$ C·m (continuity converting current amplitude to dipole amplitude). Only the $E(1,0)$ channel is fed (axial symmetry ⇒ $m=0$ ; no loops ⇒ no $M$ channels; higher $\ell$ down by $(kd)^2\sim10^{-2}$ in power).

Step 2. Power through the theorem chain (2)+(3): $P=\mu_0p_0^2\omega^4/12\pi c=0.110$ W. Cross-check by the antenna-engineering route: $R_{\text{rad}}=(2\pi/3)Z_0(d/\lambda)^2=0.219\ \Omega$ , so $P=\tfrac12I_0^2R_{\text{rad}}=0.110$ W ✓ — the channel formalism and E9's radiation-resistance bookkeeping agree to all digits.

Step 3. The exactness dividend: the leading-order amplitude used $j_1(kr)\approx kr/3$ ; the exact kernel differs by the factor $1-(kd)^2/10\approx0.9989$ at the antenna tip — a $0.2\%$ power correction. For this antenna the ladder was fine; the machinery's job is to know that, and to keep working when $d$ grows toward the half-wave regime ( $kd=\pi$ ) where E9's expansion fails and the exact integrals — evaluated in the chapter's antenna articles — take over.

Example 2 — why nuclear E2 lines are thousandfold hindered. A nucleus of radius $d\approx5$ fm emits a 1.0 MeV gamma. Compare the E2 and E1 channel scales.

Step 1. The dial: $k=E_\gamma/\hbar c=1.0/197.3=5.1\times10^{-3}$ fm⁻¹, so $kd=0.025$ — the deepest long-wavelength regime in nature's everyday catalogue.

Step 2. Suppression per rung (Derivation B, Step 3): power ratio $P_{E2}/P_{E1}\sim(kd)^2$ (with order-one moment ratios) $=6.4\times10^{-4}$ — an E2 transition of comparable matrix element runs three-plus orders of magnitude slower than E1, which is why E2 lines dominate only where selection rules (Derivation C's $\ell$ , parity bookkeeping) forbid E1 outright — precisely the situation in rotational bands of deformed nuclei, whose spin sequence $0^+,2^+,4^+\ldots$ makes E2 the lowest open channel.

Step 3. The exactness audit: at $kd=0.025$ the kernel correction is $1-(kd)^2/10\sim1-6\times10^{-5}$ — leading-order moments are accurate to $10^{-4}$ , so nuclear-structure uncertainties, not retardation, dominate real rate predictions. The formalism tells you both the rate hierarchy and the error budget — the two numbers any spectroscopist needs before trusting a lifetime calculation.

Going further

Reading. Jackson, Classical Electrodynamics, 3rd ed., §9.7–9.10 — the exact multipole fields, source integrals, and power/angular-momentum theorems this article follows. Blatt & Weisskopf, Theoretical Nuclear Physics, Ch. XII — the transition-rate edition. Bohren & Huffman, Absorption and Scattering of Light by Small Particles — the same machinery as Mie theory. Zangwill, Modern Electrodynamics, Ch. 21.

Provenance. Built from the AllQFT base article for this topic (two-scalar theorem, channel decomposition, coefficient integrals, power/angular-momentum theorems retained), harmonized with this track's conventions. The exact source-integral prefactors are deliberately presented at matching-logic level with the long-wavelength forms (2) quoted in Jackson's normalization — flagged honestly in Derivation B — and the entire chain is certified instead by the closure audit: $a_E(1,0)$ feeding the power theorem must and does reproduce E9's independently derived $P_{E1}=\mu_0p_0^2\omega^4/12\pi c$ exactly (coefficient chain $\tfrac{3}{4\pi}\cdot2\cdot\tfrac19=\tfrac{1}{6\pi}$ ; $Z_0c^2k^4=\mu_0\omega^4/c$ ). A verifier should: (1) recheck the Step-1 commutation identity and the completeness argument; (2) rerun the closure audit and the kernel-correction factors $1-(kd)^2/2(2\ell+3)$ ; (3) recompute Example 1 ( $p_0=7.96\times10^{-11}$ C·m; $P=0.110$ W both routes; correction $0.9989$ ) and Example 2 ( $kd=0.025$ ; $(kd)^2=6.4\times10^{-4}$ ; kernel accuracy $6\times10^{-5}$ ). Division of labor for this edition: the long-wavelength ladder, the $(kd)^\ell$ suppression argument, the no-monopole theorem, and the incoherent-power statement at structural level are owned by E9's Multipole Expansion of Radiation (its deferral of the exact integrals and machinery delivered here); the rungs by Electric Dipole Radiation, Magnetic Dipole Radiation, Electric Quadrupole Radiation; the $\mathbf{X}_{\ell m}$ algebra — both orthogonality relations, sum rules, the $\ell=1$ gallery reproducing the donut — by Vector Spherical Harmonics (next, by name); the exact all-zone E1 fields by Electric Dipole Radiation: Exact Fields and Zone Structure; exact M1/general E2 by Magnetic Dipole and Electric Quadrupole Radiation; antennas at $kd\sim1$ by Center-Fed Linear Antenna and Antenna Arrays and Directivity; sphere scattering by Scattering by Conducting and Dielectric Spheres; the frequency-domain/time-domain causality kinship by Covariant Wave Equation and Retarded Solutions; $Y_{\ell m}$ foundations by E2's Spherical Harmonics and Legendre Polynomials. This edition owns the two-scalar theorem, the channel bank with its exact radial anatomy, the amplitude-integral logic with the certified long-wavelength reduction, and the two per-channel theorems.

Builds on: Multipole Expansion of Radiation · Spherical Harmonics · The Electromagnetic Wave Equation · Covariant Wave Equation and Retarded Solutions · Angular Momentum of the Electromagnetic Field

Leads to: Vector Spherical Harmonics · Electric Dipole Radiation: Exact Fields and Zone Structure · Magnetic Dipole and Electric Quadrupole Radiation · Center-Fed Linear Antenna

What do the exact multipole fields buy that E9's ladder could not deliver?

Three concrete things. Validity in radius: the ladder's field formulas were far-zone (or zone-by-zone patched); the exact channels are correct from the source surface outward, which is what near-field antenna measurement, reactive-energy (Chu-limit) analysis, and nanophotonic design actually require. Validity in source size: at $kd\sim1$ — every resonant antenna, every Mie particle — the $(kr)^\ell$ moments are simply wrong, and the $j_\ell$ -kernel integrals are the only honest amplitudes. And error control even deep in the ladder regime: the kernel's $[1-(kd)^2/2(2\ell+3)]$ series converts "long-wavelength approximation" from a vibe into a number (Example 2's $6\times10^{-5}$ ). What the exact machinery does not change: the channel geometry itself — patterns, parities, the $m/\omega$ theorem — which is why E9's qualitative ladder survives intact as the $kd\to0$ shadow.

Why do the radial components of E and B suffice to reconstruct the whole field?

Count degrees of freedom the way the gauge article taught. A source-free Maxwell field has two functional degrees of freedom per propagation mode (two polarizations); on each sphere, the pair of scalars $\mathbf{r}\cdot\mathbf{E}$ , $\mathbf{r}\cdot\mathbf{B}$ is exactly two functions — and the reconstruction of Derivation A shows the map is invertible for outgoing fields: given the radial scalars' expansions, the transverse parts are forced, channel by channel, by Maxwell's curls. The deeper structural reason: $\hat{\mathbf{r}}\cdot$ commutes nicely with the angular-momentum operator, so the radial components are the natural "highest-weight" bookkeeping for the rotation group — the same reason (one group-theory level up) that two Debye potentials generate everything, and that quantum radiation theory labels photon states by $(\ell,m,E/M)$ with no residue. The exception that proves the rule: $\ell=0$ — both scalars' monopole modes are killed by the constraint equations, and with $\mathbf{X}_{00}=0$ there is no transverse structure to carry them: monopole silence, again.

Where exactly is the boundary between this article and the Vector Spherical Harmonics article?

A clean tools-versus-machine split, fixed by the chapter's division of labor. This article uses two facts about $\mathbf{X}_{\ell m}$ : transversality ( $\hat{\mathbf{r}}\cdot\mathbf{X}=0$ , so they can carry the channels' transverse structure) and orthonormality (so the power sum decoheres). Everything else about them — the second orthogonality relation involving $\hat{\mathbf{r}}\times\mathbf{X}$ , the sum rule $\sum_m|\mathbf{X}_{\ell m}|^2=(2\ell+1)/8\pi$ that makes unpolarized channel patterns isotropic-in- $m$ , the explicit $\ell=1$ gallery whose $|\mathbf{X}_{10}|^2\propto\sin^2\theta$ is E9's dipole donut, and the operator anatomy of $\mathbf{L}$ — is Vector Spherical Harmonics' cargo, next in the chapter. If you want to verify this article's claims by explicit computation, that article is the toolbox; if you want to know why the toolbox matters, this article is the answer. The two are written to be read in either order, with each recall labeled.

How literally does the classical m-over-omega theorem become photon angular momentum?

Very literally, with one honest asterisk. Classically, equation (4) is exact: a pure $(\ell,m)$ channel's radiation ships $L_z$ and $E$ in the ratio $m/\omega$ , and the full analysis gives total angular momentum bookkeeping consistent with the label $\ell$ . Divide energy into quanta of $\hbar\omega$ and each quantum carries $m\hbar$ of $L_z$ — this is not an analogy; the quantum multipole photon states are built on exactly these classical mode functions, and gamma spectroscopy's selection rules follow with no further input. The asterisk: the split of photon angular momentum into "spin" and "orbital" parts is not sharp — a photon has no rest frame, and only the total (and helicity along $\mathbf{k}$ ) are cleanly defined; the $(\ell,m)$ labels bookkeep the total. That subtlety, famous in the modern optical-angular-momentum literature, changes nothing about equation (4) — which is one reason the multipole basis, rather than any spin-orbital split, remains the rigorous language for radiation angular momentum.

Why does nature's most common radiation regime have kd so small, and what lives at the other extreme?

The smallness is an accident of atomic structure with deep consequences: emitters' sizes are set by matter's binding (ångströms for atoms, femtometers for nuclei) while emitted wavelengths are set by transition energies — and for bound systems the two are related by the fine-structure constant: atomic $kd\sim\alpha\sim10^{-2\text{–}3}$ , nuclear similar for MeV gammas. That hierarchy is why the dipole approximation runs almost all of spectroscopy and why higher multipoles read as "forbidden lines." The other extreme, $kd\gtrsim1$ , is the engineered world: antennas are deliberately built at $kd\sim\pi$ because radiation resistance (and hence efficiency against ohmic loss) grows with $kd$ — E9's $(L/\lambda)^2$ law priced it; and scattering resonances (Mie, plasmonic) occur precisely where a channel's $h_\ell^{(1)}$ near-zone reactance cancels — physics made of the exact radial functions. Astrophysics supplies natural examples too: pulsar magnetospheres radiating at rotation frequencies have $kd$ set by the light-cylinder geometry, near unity by construction.

Do the channels of this decomposition interact with the covariance machinery of the rest of the chapter — is the multipole expansion Lorentz-covariant?

Honest answer: the decomposition is frame-specific by construction, and that is a feature to manage, not a defect. The expansion privileges a center (the moments' origin) and a rest frame (the frequency $\omega$ , the spheres); a boosted observer sees Doppler-shifted frequencies, aberrated patterns (the previous articles' machinery, exactly), and mixed multipole content — a pure dipole in the source frame acquires quadrupole and higher content when boosted, the classical ancestor of the statement that photon $(\ell,m)$ labels are frame-dependent while helicity is not. The professional protocol mirrors the media article's: decompose in the source's rest frame, where the moments mean what E2/E9 taught; transport observables (power, spectra, patterns) to other frames with the Doppler/aberration dictionary. For ultrarelativistic emitters the source-frame expansion becomes useless in practice — a $\gamma\gg1$ synchrotron electron's radiation is better handled by the Liénard–Wiechert/beaming machinery of the E11 relativistic set — and the boundary between "multipole territory" ( $kd$ small or moderate, sources subrelativistic) and "beaming territory" is one of the chapter's organizing facts.

How does this machinery become Mie scattering and the nanophotonics toolbox?

By adding one ingredient: a boundary. Scattering by a sphere is this article's exterior problem run twice — expand the incident plane wave in the regular channel functions ( $j_\ell$ profiles; a known expansion), expand the scattered field in the outgoing ones ( $h_\ell^{(1)}$ , unknown amplitudes), expand the interior field similarly, and impose the E8 boundary conditions at the surface. The matching yields the Mie coefficients — the sphere's per-channel reflection amplitudes — and everything measurable (cross sections, resonances, glories and rainbows in their wave-optics form) is assembled from them; the chapter's Scattering by Conducting and Dielectric Spheres executes it. Modern nanophotonics runs the same pipeline with fancier boundaries and inverse logic: design the particle so chosen channel amplitudes interfere as desired — Kerker conditions (electric and magnetic dipole channels cancelling backward scattering), anapoles (channel amplitudes engineered to zero), metasurface unit cells prescribed by their $(a_E,a_M)$ lists. The channel bank built in this article is, quite literally, the design spreadsheet of that field.