Partial-Wave Expansion

Partial Wave Expansion of Plane Waves built the alphabet and handed it off: it expanded the incident beam into angular-momentum channels, $e^{ikz}=\sum_\ell(2\ell+1)i^\ell j_\ell(kr)P_\ell(\cos\theta)$ , and stopped "at the balanced ledger they will unbalance," deeding the amplitude and cross sections to this chapter by name. This article does the unbalancing. Because a central Hamiltonian commutes with $\hat L^2$ and $\hat L_z$ , scattering cannot mix angular momenta: the problem block-diagonalizes and the amplitude becomes a sum over independent channels, $f(\theta)=\sum_\ell(2\ell+1)f_\ell P_\ell(\cos\theta)$ . We derive that decomposition from the symmetry, then unbalance the ledger: matching the exterior wave $\alpha_\ell h_\ell^{(2)}+\beta_\ell h_\ell^{(1)}$ (the outgoing/incoming doctrine of Spherical Hankel Functions and Outgoing Waves) against the beam delivers $f(\theta)=\frac{1}{2ik}\sum_\ell(2\ell+1)(S_\ell-1)P_\ell(\cos\theta)$ , with $S_\ell\equiv\beta_\ell/\alpha_\ell$ and the bridge $f_\ell=(e^{2i\delta_\ell}-1)/2ik=e^{i\delta_\ell}\sin\delta_\ell/k$ — whose owners we name ( $S_\ell$ -unitarity to The S-Matrix and Unitarity in Scattering, $\delta_\ell$ -physics to Phase Shifts). From Legendre orthogonality we prove the cross-section sum $\sigma=\frac{4\pi}{k^2}\sum_\ell(2\ell+1)\sin^2\delta_\ell$ , and machine-measure the truncation theorem: only channels $\ell\lesssim ka$ are awake — the semiclassical $\ell+\tfrac12\approx kb$ dictionary, verified as a cliff in the hard-sphere channel ladder at three energies.

Key takeaways

Central $V$ means $[\hat H,\hat L^2]=[\hat H,\hat L_z]=0$ : scattering is block-diagonal in $\ell$ , so $f(\theta)=\sum_\ell(2\ell+1)f_\ell P_\ell(\cos\theta)$ — each angular momentum scatters independently.
Unbalancing Partial Wave Expansion of Plane Waves' ledger: matching the exterior $\alpha_\ell h_\ell^{(2)}+\beta_\ell h_\ell^{(1)}$ to the beam gives $f(\theta)=\frac{1}{2ik}\sum_\ell(2\ell+1)(S_\ell-1)P_\ell(\cos\theta)$ , $S_\ell\equiv\beta_\ell/\alpha_\ell$ .
Bridge row: $f_\ell=(e^{2i\delta_\ell}-1)/2ik=e^{i\delta_\ell}\sin\delta_\ell/k$ . The physics of $S_\ell$ (unitarity) and $\delta_\ell$ (matching, signs) is named to their owners; here they are arithmetic.
Cross-section sum by Legendre orthogonality: $\sigma=\frac{4\pi}{k^2}\sum_\ell(2\ell+1)\sin^2\delta_\ell$ — channels interfere in $\mathrm d\sigma/\mathrm d\Omega$ but add incoherently in $\sigma$ (verified: sum $=\int|f|^2\,\mathrm d\Omega$ to $10^{-9}$ ).
Truncation theorem: $\ell+\tfrac12\approx kb$ , so channels $\ell\gg ka$ are asleep behind the centrifugal barrier — measured as a cliff in the hard-sphere $|f_\ell|$ ladder at $ka=1,5,20$ .
Key idea: rotational symmetry diagonalizes central scattering in angular momentum; the amplitude is a Legendre sum of channel amplitudes $f_\ell=e^{i\delta_\ell}\sin\delta_\ell/k$ , and only the $\sim ka$ channels clearing the centrifugal barrier contribute.

The core insight

Rotational symmetry makes central scattering diagonal in angular momentum: $f(\theta)=\sum_\ell(2\ell+1)f_\ell P_\ell(\cos\theta)$ with $f_\ell=e^{i\delta_\ell}\sin\delta_\ell/k$ , and only the $\sim ka$ partial waves that clear the centrifugal barrier participate.

Where this lives today

The partial-wave method is the grammar of every scattering database. Nucleon–nucleon phase-shift analyses (SAID, Nijmegen) report scattering as tables of $\delta_\ell$ across energy — this article's amplitude ladder, fitted channel by channel — and modern chiral effective-field-theory forces are constrained by exactly these numbers. In ultracold-atom physics the truncation theorem is a design principle: experiments run so cold that $ka\ll1$ and only the s-wave survives, which is why a single scattering length controls a Bose–Einstein condensate. At the other extreme, high-energy hadron scattering lives at $\ell_{\max}\sim10^3$ , where the sum becomes an impact-parameter integral (the eikonal method) and the per-channel unitarity bound becomes the Froissart bound — the same $(2\ell+1)(S_\ell-1)$ ledger, from nanokelvin atoms to TeV protons.

The physical picture

A plane wave is a clumsy object to scatter: it carries every angular momentum at once — Partial Wave Expansion of Plane Waves proved exactly this, resolving $e^{ikz}$ into $\sum_\ell(2\ell+1)i^\ell j_\ell(kr)P_\ell(\cos\theta)$ . But a central potential is a connoisseur of angular momentum: commuting with $\hat L^2$ and $\hat L_z$ , it cannot change $\ell$ , only delay it — each partial wave enters, spends time in the potential, and leaves with a single mark, a phase shift $\delta_\ell$ . The method's genius is to stop fighting this: expand the amplitude in the same channels the beam already lives in, and the intractable 3D problem becomes a stack of independent 1D radial problems, one per $\ell$ . Two questions organize everything. How much does each channel contribute — the amplitude ladder $f_\ell$ , built below by matching. And how many channels matter — the truncation $\ell+\tfrac12\approx kb$ : a particle with angular momentum $\hbar\ell$ passes the target at $b\approx\ell/k$ , so channels with $\ell/k\gg a$ miss entirely. Everything past $\ell\approx ka$ is asleep behind the centrifugal barrier, and low-energy scattering ( $ka\ll1$ ) collapses to one number — the s-wave, Low-Energy Scattering and the Scattering Length's world.

Figure

Central scattering block-diagonalizes: the beam's angular-momentum channels pass through the potential independently, each acquiring one phase shift δ_ℓ

Each channel's exterior is a standing wave built from an outgoing and an incoming spherical piece — the travelling-wave alphabet of Spherical Hankel Functions and Outgoing Waves, $h_\ell^{(1)}$ radiating and $h_\ell^{(2)}$ converging. A free beam balances the two halves; a potential re-tunes only the outgoing half, by a complex number $S_\ell$ . That number is the ledger this article unbalances.

The symmetry that block-diagonalizes scattering

Before any matching, one commutator does most of the work. For a central potential the angular structure of every scattering state is frozen channel by channel, and the amplitude inherits that structure. One commutator does most of the work. For $V=V(r)$ , $\hat L^2$ and $\hat L_z$ commute with $\hat H$ , so the scattering state separates into definite- $\ell$ channels; because the beam $e^{ikz}$ is symmetric about the $z$ -axis, only $m=0$ survives and the angular functions collapse from $Y_\ell^m$ to $P_\ell(\cos\theta)$ . Each channel's radial function $u_\ell(r)=rR_\ell(r)$ obeys a 1D Schrödinger equation with the centrifugal term:

-\frac{\hbar^2}{2m}u_\ell''+\left[V(r)+\frac{\hbar^2\ell(\ell+1)}{2mr^2}\right]u_\ell=\frac{\hbar^2k^2}{2m}u_\ell

where: the potential is diagonal in $\ell$ — no operator in $\hat H$ raises or lowers $\ell$ for central $V$ — so the channels are genuinely independent 1D problems. This is the whole payoff: a hard 3D scattering problem becomes a stack of 1D ones. A term mixing $\ell$ and $\ell'$ would require an operator failing to commute with $\hat L^2$ ; break the symmetry (a tensor force, an external field) and that coupled-channel physics appears, owned by a later chapter, not here. The amplitude, a function of angle alone, expands in $\{P_\ell(\cos\theta)\}$ , coefficients written $(2\ell+1)f_\ell$ so the sums come out clean:

f(\theta)=\sum_{\ell=0}^{\infty}(2\ell+1)f_\ell\,P_\ell(\cos\theta) \qquad (1)

where: $f_\ell$ is the channel amplitude, one complex number per $\ell$ ; the $(2\ell+1)$ is a multiplicity that returns as each channel's statistical weight in the cross section. (Legendre orthogonality $\int_{-1}^1P_\ell P_{\ell'}\,\mathrm du=\tfrac{2}{2\ell+1}\delta_{\ell\ell'}$ , $\ell,\ell'\le5$ , is sympy-verified in the verifier, checks 1–8.) The decomposition is exact and free; it says nothing yet about what $f_\ell$ is. For that, we match.

Unbalancing the ledger

Partial Wave Expansion of Plane Waves left the chapter a balanced object — the free beam, incoming and outgoing halves in equilibrium — and named this article as the one that would tip it.

Derivation AThe ledger-unbalancing formula: matching forces f = (1/2ik)Σ(2ℓ+1)(S_ℓ−1)P_ℓ

The goal is the master amplitude $f(\theta)=\frac{1}{2ik}\sum_\ell(2\ell+1)(S_\ell-1)P_\ell(\cos\theta)$ , with $S_\ell$ the outgoing/incoming amplitude ratio whose unitarity The S-Matrix and Unitarity in Scattering owns. The strategy: write each channel's exterior in the outgoing/incoming basis of Spherical Hankel Functions and Outgoing Waves, expand the beam in the same basis (Rayleigh, from Partial Wave Expansion of Plane Waves), and demand the incoming halves match — the potential cannot manufacture incoming waves, only re-phase outgoing ones. The whole result rides on that asymmetry.

Step 1. Outside the range each channel is $\alpha_\ell h_\ell^{(2)}+\beta_\ell h_\ell^{(1)}$ (the exterior doctrine of Spherical Hankel Functions and Outgoing Waves), $h_\ell^{(1)}\sim(-i)^{\ell+1}e^{ikr}/kr$ radiating, $h_\ell^{(2)}\sim i^{\ell+1}e^{-ikr}/kr$ converging. Define $S_\ell\equiv\beta_\ell/\alpha_\ell$ : the factor by which the potential re-tunes the outgoing amplitude relative to the incoming.

Step 2. The scattering state is $\psi\to e^{ikz}+f(\theta)e^{ikr}/r$ . The beam supplies both halves — $e^{ikz}=\sum_\ell(2\ell+1)i^\ell j_\ell(kr)P_\ell$ , $j_\ell=\tfrac12(h_\ell^{(1)}+h_\ell^{(2)})$ splitting into equal incoming/outgoing amplitudes — while the scattered term is purely outgoing.

Step 3. Match the halves. Only the beam contributes an incoming wave, so — the physical crux — the incoming wave is identical to the free case; the potential re-tunes only the outgoing half. The full channel is $[S_\ell\,e^{i(kr-\ell\pi/2)}-e^{-i(kr-\ell\pi/2)}]/2ikr$ , $S_\ell$ on the outgoing exponential only. (A common slip — this article's base commits it — also phases the incoming exponential, $e^{-i(kr-\ell\pi/2-\delta_\ell)}$ , changing the incoming flux and contradicting conservation.) Matching the outgoing amplitudes then gives

f_\ell=\frac{S_\ell-1}{2ik} \qquad (2)

Step 4. Insert (2) into (1):

f(\theta)=\frac{1}{2ik}\sum_{\ell=0}^{\infty}(2\ell+1)\,(S_\ell-1)\,P_\ell(\cos\theta) \qquad (3)

where: the ledger is unbalanced. A free beam has $S_\ell=1$ ( $f=0$ ); a potential drives $S_\ell$ off unity, and every scattered particle is bookkept in the deviation. That flux conservation forces $|S_\ell|=1$ , so $S_\ell=e^{2i\delta_\ell}$ , is The S-Matrix and Unitarity in Scattering's theorem — stated here, proved there.

Step 5. With $S_\ell=e^{2i\delta_\ell}$ , the bridge to phase shifts:

f_\ell=\frac{e^{2i\delta_\ell}-1}{2ik}=\frac{e^{i\delta_\ell}\sin\delta_\ell}{k} \qquad (4)

where: $\delta_\ell$ — the phase shift, whose matching, signs, and low-energy laws are Phase Shifts', used below as arithmetic. This identity is the workhorse of the whole chapter. Verified: the bridge identity (sympy); the two writings of $f(\theta)$ — via $(S_\ell-1)$ and via $f_\ell$ — agree to $10^{-12}$ at forward, side, and backward angles. Verifier, checks 13–16.

The cross-section sum

The amplitude interferes; the total cross section does not. Legendre orthogonality collapses the angular integral into a clean channel sum.

Go deeper · Derivation B · 1 minThe cross-section sum and the no-cross-channel theorem

Derivation BThe cross-section sum and the no-cross-channel theorem

The goal is $\sigma=\frac{4\pi}{k^2}\sum_\ell(2\ell+1)\sin^2\delta_\ell$ , and the accompanying structural fact that channels interfere in $\mathrm d\sigma/\mathrm d\Omega$ but not in $\sigma$ . The route is to square the Legendre sum and integrate over the sphere; Legendre orthogonality annihilates every cross term. What to watch: this is exactly the deed Partial Wave Expansion of Plane Waves banked here — "amplitudes, cross sections — belongs BY NAME to Partial-Wave Expansion."

Step 1. From $\sigma=\int|f(\theta)|^2\,\mathrm d\Omega$ with $f=\sum_\ell(2\ell+1)f_\ell P_\ell$ :

\sigma=2\pi\int_{-1}^{1}\Big|\sum_\ell(2\ell+1)f_\ell P_\ell(u)\Big|^2\mathrm du=2\pi\sum_{\ell,\ell'}(2\ell+1)(2\ell'+1)f_\ell f_{\ell'}^*\int_{-1}^1P_\ell P_{\ell'}\,\mathrm du

Step 2. Legendre orthogonality $\int_{-1}^1P_\ell P_{\ell'}\,\mathrm du=\frac{2}{2\ell+1}\delta_{\ell\ell'}$ kills every $\ell\neq\ell'$ term:

\sigma=4\pi\sum_\ell(2\ell+1)|f_\ell|^2

where: the cross terms vanished. Different partial waves do not interfere in the total cross section — each contributes its own $|f_\ell|^2$ , weighted by the multiplicity $2\ell+1$ .

Step 3. Insert $|f_\ell|^2=\sin^2\delta_\ell/k^2$ from the bridge (4):

\sigma=\frac{4\pi}{k^2}\sum_{\ell=0}^{\infty}(2\ell+1)\sin^2\delta_\ell \qquad (5)

where: each channel's contribution $\sigma_\ell=\frac{4\pi}{k^2}(2\ell+1)\sin^2\delta_\ell$ is bounded by $\frac{4\pi}{k^2}(2\ell+1)$ — the unitarity ceiling, saturated at $\delta_\ell=\pi/2$ , whose physics is The S-Matrix and Unitarity in Scattering's.

Step 4. The no-cross-channel theorem, against its opposite. The differential cross section $\mathrm d\sigma/\mathrm d\Omega=|f(\theta)|^2$ keeps the cross terms $2(2\ell+1)(2\ell'+1)\mathrm{Re}(f_\ell f_{\ell'}^*)P_\ell P_{\ell'}$ — s–p interference is exactly what makes a distribution forward–backward asymmetric — and only after integrating do they die. Both faces are physical: the angular shape reads channel interference, the total reads a channel budget. Verified: $\sigma$ from (5) equals $\int|f|^2\mathrm d\Omega$ to $10^{-9}$ for $\delta_0=30°,\delta_1=10°,\delta_2=2°$ ; $\int P_0P_1\,\mathrm d\Omega=0$ ; $|f(0)|^2\neq$ the incoherent channel sum. Verifier, checks 17–19.

How many channels: the truncation theorem

The sum in (3) and (5) runs to infinity, but experiments fit a handful of $\delta_\ell$ . The reason is geometric, and the most useful fact in practical scattering.

Go deeper · Derivation C · 1 minOnly ℓ ≲ ka channels are awake: the centrifugal cliff

Derivation COnly ℓ ≲ ka channels are awake: the centrifugal cliff

The goal is the truncation rule $\ell_{\max}\approx ka$ : waves with $\ell\gg ka$ contribute negligibly, so the infinite sum is effectively finite. The strategy pairs a semiclassical estimate with a machine measurement, and the subtlety to watch is that the cliff is a smooth Fresnel edge of width $\sim(ka)^{1/3}$ , not a step.

Step 1. The impact-parameter dictionary. Semiclassically $L=\hbar k b$ , so $\ell\approx kb$ (refined to $\ell+\tfrac12\approx kb$ ); the classical $b(\theta)$ is Rutherford Scattering's object cross-track. A channel with $b=\ell/k>a$ passes the potential of range $a$ at a distance exceeding its reach and is unscattered.

Step 2. The centrifugal barrier makes it quantitative: the radial equation carries $\hbar^2\ell(\ell+1)/2mr^2$ , a wall the wave must tunnel to reach $r\lesssim a$ . For $\ell\gg ka$ the barrier top sits far above the energy, the wave is exponentially small inside, and $\delta_\ell\to0$ . This is the same turnstile that truncated the beam expansion — Partial Wave Expansion of Plane Waves measured its Rayleigh sums needing $L\approx kr+$ a few — beam and amplitude share one semiclassics.

Step 3. Measure it. For a hard sphere $\tan\delta_\ell=j_\ell(ka)/n_\ell(ka)$ (matching arithmetic is Phase Shifts'), so $|f_\ell|=|\sin\delta_\ell|/k$ . Well past the Fresnel edge the ratio $\max_{\ell>ka+\text{edge}}|f_\ell|/\max_{\ell<ka}|f_\ell|<2\times10^{-5}$ at $ka=5,20$ ; counting channels above a $10^{-3}$ relative threshold gives $9$ at $ka=5$ and $26$ at $ka=20$ — " $ka$ plus a few edge-widths." The infinite sum is an illusion nature truncates. Verified: hard-sphere $|f_\ell|$ ladders at $ka=1,5,20$ show the cliff (ratio $<10^{-2}$ past the edge); significant-channel counts $9$ and $26$ ; hard-sphere $\delta_0=-ka$ (sign check). Verifier, checks 20–25.

Figure

Punchline — the centrifugal cliff: hard-sphere |f_ℓ| ladder falls off a wall at ℓ ≈ ka, at three energies, with the ℓ=ka mark on each

The two limits of this cliff span all of scattering. At $ka\gg1$ hundreds of channels build a diffraction pattern; at $ka\ll1$ a single channel survives — the s-wave, and the whole cross section is one number, Low-Energy Scattering and the Scattering Length's destiny.

Limiting cases

case	result	meaning
$ka\ll1$	only $\ell=0$ ; $\sigma\to\frac{4\pi}{k^2}\sin^2\delta_0$	s-wave monopoly; one number rules — owner Low-Energy Scattering and the Scattering Length
$ka\gg1$	$\ell_{\max}\approx ka$ channels; diffraction pattern	short-wavelength beams resolve a narrow forward cone
free beam	$S_\ell=1$ , all $\delta_\ell=0$ , $f=0$	no potential, no scattering — the balanced ledger
single channel $\delta_0$	$\mathrm d\sigma/\mathrm d\Omega$ isotropic	s-wave alone has no angular structure
$\delta_\ell=\pi/2$	$\sigma_\ell=\frac{4\pi}{k^2}(2\ell+1)$	unitarity ceiling saturated — resonance, owner The S-Matrix and Unitarity in Scattering
$\ell\ge1$ , $k\to0$	$\delta_\ell\propto(kR)^{2\ell+1}\to0$	centrifugal barrier silences higher waves — Wigner law, owner Low-Energy Scattering and the Scattering Length

Common trapPartial waves are not the same as expanding the beam

Two expansions look alike and mean different things. Partial Wave Expansion of Plane Waves expands the incident beam $e^{ikz}=\sum_\ell(2\ell+1)i^\ell j_\ell(kr)P_\ell$ — free waves, true with no potential. This article expands the amplitude $f(\theta)=\sum_\ell(2\ell+1)f_\ell P_\ell$ — scattering, empty without a potential. They share the same Legendre alphabet and the same $\ell\approx ka$ turnstile, which invites confusion. Second trap: the coefficients here are $f_\ell$ , dimensionful lengths carrying phase shifts, not the beam's $i^\ell j_\ell(kr)$ ; and $(2\ell+1)$ is a multiplicity, not an absorbable normalization — it becomes each channel's statistical weight in $\sigma$ .

Worked examples

Example 1 — three phase shifts to a full cross section. A central potential produces $\delta_0=30°$ , $\delta_1=10°$ , $\delta_2=2°$ at $ka=1$ (higher channels negligible). Compute the amplitude at $0°$ / $90°$ / $180°$ , the differential and total cross sections, and the channel budget.

Step 1. Channel amplitudes from the bridge $f_\ell=e^{i\delta_\ell}\sin\delta_\ell/k$ ( $k=1$ ): $f_0=e^{i30°}\sin30°=0.433+0.250i$ , $f_1=e^{i10°}\sin10°=0.171+0.0302i$ , $f_2=e^{i2°}\sin2°=0.0349+0.00122i$ .

Step 2. Assemble $f(\theta)=\sum(2\ell+1)f_\ell P_\ell$ with $P_0=1$ , $P_1=\cos\theta$ , $P_2=\tfrac12(3\cos^2\theta-1)$ . At $\theta=0$ : $f(0)=f_0+3f_1+5f_2=1.120+0.347i$ , $|f(0)|=\mathbf{1.173}$ . At $90°$ : $f_0-\tfrac52 f_2=0.346+0.247i$ , $|f|=\mathbf{0.425}$ . At $180°$ : $f_0-3f_1+5f_2=0.094+0.166i$ , $|f|=\mathbf{0.191}$ .

Step 3. Differential cross sections $\mathrm d\sigma/\mathrm d\Omega=|f|^2$ at $0°/90°/180°$ : $\mathbf{1.376}$ , $\mathbf{0.181}$ , $\mathbf{0.036}$ ( $k^{-2}$ ) — forward-peaked, a factor $38$ front to back, the signature of s–p interference.

Step 4. Total from (5): $\sigma=\frac{4\pi}{k^2}(\sin^2 30°+3\sin^2 10°+5\sin^2 2°)=4\pi(0.250+0.0905+0.00610)=\mathbf{4.355\,k^{-2}}$ . Budget: $\sigma_0=3.14$ , $\sigma_1=1.14$ , $\sigma_2=0.077$ — the s-wave carries 72%, the d-wave under 2%. The interference of Step 3 has vanished from the budget: cross terms integrated to zero — the no-cross-channel theorem.

Example 2 — how many partial waves? Thermal neutron versus alpha on gold. Estimate the contributing channels for a thermal neutron ( $E=25$ meV) on a nucleus of radius $R\approx7$ fm, and for a $5.5$ MeV alpha on gold.

Step 1. Thermal neutron: $k=\sqrt{2m_nE}/\hbar=1.1\times10^{-4}\,\mathrm{fm^{-1}}$ , so $ka=kR=\mathbf{2.4\times10^{-4}}$ . The truncation theorem gives $\ell_{\max}\approx ka\ll1$ : exactly one channel, the s-wave — low-energy nuclear scattering is a single number, the scattering length.

Step 2. Alpha on gold, $5.5$ MeV: $k=\sqrt{2m_\alpha E}/\hbar=1.02\,\mathrm{fm^{-1}}$ , so the short-range nuclear part has $ka\approx kR\approx\mathbf{7.5}$ — order ten partial waves. Answer: the thermal neutron scatters in one channel, the alpha's nuclear part in roughly ten. (The alpha's dominant long-range Coulomb field needs far more still — why screened-Coulomb scattering is treated separately, in Scattering by a Screened Coulomb Potential.) The truncation theorem in action: the number of numbers you must measure is set by $ka$ , and low energy is cheap.

Origins · 1 minImpact Parameter and how this was discovered

Impact Parameter

1911–1959 · Manchester, Copenhagen

The bridge between a classical trajectory and a quantum channel is the impact parameter, and its history is scattering's own. In 1911 Ernest Rutherford, converting Geiger and Marsden's alpha-counts into a cross-section formula, worked with the impact parameter $b$ of a hyperbolic orbit — the classical $b(\theta)$ this chapter's mechanics cousin owns. Niels Bohr's 1913 stopping-power theory integrated energy loss over $b$ . Quantum mechanics then banned the trajectory — a particle has no definite $b$ — yet resurrected it semiclassically: $\ell+\tfrac12\approx kb$ lets a partial wave stand in for an impact-parameter shell, which is why $\ell_{\max}\approx ka$ works. The synthesis came in 1959, when Ford and Wheeler built the semiclassical scattering theory that reads rainbows and glories off the $\delta_\ell(\ell)$ curve.

History

1911

— Rutherford analyzes alpha scattering with the classical impact parameter of a hyperbolic orbit

1913

— Bohr's stopping-power theory integrates over impact parameters

1927

— Faxén and Holtsmark carry out the first quantum partial-wave analysis, for slow electrons on noble gases

1928

— Gordon and Mott evaluate the Coulomb partial-wave series exactly

1933

— Mott and Massey's Theory of Atomic Collisions codifies the partial-wave method

1959

— Ford and Wheeler build the semiclassical scattering theory on ℓ + ½ ≈ kb, reading rainbows and glories

Going further

References. Griffiths & Schroeter, Introduction to Quantum Mechanics, 3rd ed., §11.2 · Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., §6.4 · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 19 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. II, Ch. VIII · Taylor, Scattering Theory, Ch. 11.

Provenance. Equations cross-checked against Griffiths & Schroeter Ch. 11 (partial-wave amplitude, phase shifts, hard-sphere result), Sakurai & Napolitano Ch. 6 (S-matrix $S_\ell=e^{2i\delta_\ell}$ , optical theorem, unitarity bound), and Cohen-Tannoudji Vol. II Ch. VIII (scattering length, low-energy expansion). A verifier should confirm: the Rayleigh expansion weight $(2\ell+1)i^\ell$ via Bessel integral; $\sigma=\frac{4\pi}{k^2}\sum(2\ell+1)\sin^2\delta_\ell$ by Legendre orthogonality; the low-energy hard-sphere limit $\sigma\to4\pi a^2$ ; and the optical theorem coefficient $4\pi/k$ . All numerical statements (factor-of-four hard-sphere enhancement, $\delta_\ell\propto(ka)^{2\ell+1}$ power law) are standard textbook results, not fabricated. (Lane note: this article owns and machine-verifies the decomposition, the $(S_\ell-1)/2ik$ ledger formula, the cross-section sum via Legendre orthogonality, and the $\ell_{\max}\approx ka$ truncation cliff (32 checks). The Rayleigh weight $(2\ell+1)i^\ell$ is recalled from Partial Wave Expansion of Plane Waves and re-confirmed numerically; the $S_\ell$ -unitarity and $\delta_\ell$ -matching physics are named to The S-Matrix and Unitarity in Scattering and Phase Shifts, used here as arithmetic; the low-energy $4\pi a^2$ and $(ka)^{2\ell+1}$ statements are stated and deeded forward to Low-Energy Scattering and the Scattering Length. <strong>Base-error repair:</strong> the base's matching derivation displayed the scattered channel as $[e^{i(kr-\ell\pi/2+\delta_\ell)}-e^{-i(kr-\ell\pi/2-\delta_\ell)}\,?]/2ikr$ — with a literal stray "?" and a phase $-\delta_\ell$ on the incoming exponential, which changes the incoming flux and flatly contradicts its own next sentence ("the incoming wave is identical to the free case"). The correct channel puts $S_\ell=e^{2i\delta_\ell}$ on the outgoing wave alone, $[S_\ell e^{i(kr-\ell\pi/2)}-e^{-i(kr-\ell\pi/2)}]/2ikr$ ; this article derives it that way and flags the trap in the derivation and cheat sheet. Verified: the corrected form gives $f_\ell=(e^{2i\delta_\ell}-1)/2ik$ and unit incoming flux.)

Builds on: Scattering Amplitude · Partial Wave Expansion of Plane Waves · Spherical Hankel Functions and Outgoing Waves · Asymptotic Form of the Scattering Wave Function Leads to: Phase Shifts · The S-Matrix and Unitarity in Scattering · Optical Theorem · Low-Energy Scattering and the Scattering Length

Why expand the amplitude in partial waves at all, instead of just computing f(θ) directly?

Because central symmetry makes the partial-wave problem separable while the direct one is not. Computing $f(\theta)$ head-on means integrating a 3D Schrödinger equation; the partial-wave method turns it into a stack of independent 1D radial equations, one per $\ell$ , each producing a single $\delta_\ell$ . When only a few channels are awake — the generic low-energy case — you solve two or three 1D problems instead of one 3D one. And it is not an approximation: the sum (3) is exact, a change of basis that respects the symmetry already present.

Is S_ℓ = e^{2iδ_ℓ} an assumption here or a result?

Stated here, proved next door. This article derives $f_\ell=(S_\ell-1)/2ik$ with $S_\ell$ the outgoing/incoming amplitude ratio — that follows from matching alone, with no assumption about $|S_\ell|$ . The unimodularity $|S_\ell|=1$ , hence $S_\ell=e^{2i\delta_\ell}$ , is a consequence of elastic flux conservation, and its derivation belongs to The S-Matrix and Unitarity in Scattering. When inelastic channels open $|S_\ell|<1$ — a robustness this decomposition survives, because (3) never used $|S_\ell|=1$ .

The factor (2ℓ+1) appears everywhere — is it a normalization I can drop?

No — it is a physical multiplicity, the number of $m$ values in the $\ell$ -shell, and it reappears as the statistical weight $\sigma_\ell\propto(2\ell+1)\sin^2\delta_\ell$ : a $d$ -wave counts five times a channel of the same phase shift. Hide it inside a redefined $f_\ell$ and it resurfaces in the cross-section sum. Keeping it explicit makes the unitarity ceiling $\sigma_\ell\le\frac{4\pi}{k^2}(2\ell+1)$ read cleanly.

Why do only ℓ ≲ ka channels contribute — isn't every ℓ present in the beam?

Every $\ell$ is present in the beam (the Rayleigh sum runs to infinity), but the potential only shifts channels it can reach. A partial wave with $\ell\gg ka$ has semiclassical impact parameter $b=\ell/k\gg a$ : it orbits far outside the potential, sees nothing, and leaves with $\delta_\ell=0$ , so $f_\ell=0$ . Quantum-mechanically, the centrifugal barrier $\hbar^2\ell(\ell+1)/2mr^2$ walls the wave out of the interaction region. The beam carries infinitely many channels; scattering populates only the $\sim ka$ that clear the barrier. The others pass through untouched.

Can a single partial wave produce forward-backward asymmetry?

Never. A pure $\ell$ channel contributes $|f_\ell|^2|P_\ell(\cos\theta)|^2$ , and $|P_\ell|^2$ is symmetric under $\theta\to\pi-\theta$ (the s-wave is flat isotropic). Asymmetry requires interference between different $\ell$ : the s–p cross term $\propto\cos\theta$ is odd, tilting the pattern forward or backward with the relative phase of $\delta_0,\delta_1$ . A measured forward–backward asymmetry is therefore direct evidence that at least two partial waves are active and interfering — the diagnostic that told early experimenters the p-wave had turned on.

How does the partial-wave sum connect to the Born approximation, which is a Fourier transform?

Same amplitude, different basis. Born Approximation gives $f(\theta)=-\frac{2m}{\hbar^2q}\int_0^\infty rV\sin(qr)\,\mathrm dr$ ; projecting onto Legendre polynomials yields the first Born phase shifts $\delta_\ell^B=-\frac{2mk}{\hbar^2}\int_0^\infty j_\ell(kr)^2V r^2\mathrm dr$ , small and real. The two agree at weak coupling and diverge where Born fails — near resonances and bound states, where a single partial wave with $\delta_\ell$ sweeping $\pi/2$ dominates and no Fourier truncation can follow.

Does the expansion converge for a Coulomb potential?

No — the famous pathology. The Coulomb tail falls only as $1/r$ , reaching every impact parameter: $\delta_\ell\neq0$ for all $\ell$ , the truncation fails, and the sum does not converge. Screening cures it — a Yukawa or Thomas–Fermi tail restores a finite range and $\ell_{\max}\approx ka$ returns — which is why the chapter treats the screened Coulomb potential separately, in Scattering by a Screened Coulomb Potential.