Partial-wave expansion and phase shifts

Intuition Beginner

Imagine a long, straight set of ocean swells rolling toward a round buoy. Most of the water sails past untouched. But near the buoy the swells bend, and a ring of disturbed water spreads outward. If you sort the incoming swells by how tightly they curve around the buoy — the ones aimed dead-on, the ones grazing the edge, the ones sweeping wide — each group reacts to the buoy in its own way. Quantum scattering off a round target works the same way. You sort the incoming wave into rings of different angular momentum, and each ring meets the target on its own terms.

A round target cannot tell left from right or up from down. So a wave aimed at it keeps its angular pattern, ring by ring. Each ring is labelled by a whole number $\ell$: $\ell =0$ is the head-on group, $\ell =1$ grazes a little, larger $\ell$ sweep wider. Splitting the wave into these rings is the partial-wave expansion. The deep payoff is that the target acts on each ring separately, so one hard three-dimensional problem becomes a stack of simple one-dimensional ones.

What does the target do to a single ring? Far away, that ring is a wave that ripples in and out along the radius, like a vibrating string pinned at the center. The target cannot change how fast the ripples wiggle, set by the energy. All it can do is nudge the whole ripple pattern inward or outward — it shifts the wave's phase. That nudge is the phase shift ${\delta }_{\ell }$, one number per ring. It is the complete fingerprint of how that ring scatters.

A small phase shift means the target barely touched that ring. A large one means the target reshaped it strongly. Add up the phase shifts from all the rings, weighted by how many particles each ring carries, and you recover the full scattering pattern: how likely a particle is to come out at each angle, and the total chance of being deflected at all.

Visual Beginner

SORTING THE WAVE INTO ANGULAR-MOMENTUM RINGS
=============================================

   incoming straight wave            sorted into rings by aim
   --->  --->  --->                  l = 0  (head-on, s-wave)
   --->  --->  --->     ===>         l = 1  (grazing, p-wave)
   --->  --->  --->                  l = 2  (wide, d-wave)
                                     ...

WHAT THE TARGET DOES TO ONE RING: A PHASE SHIFT
===============================================

   free ring (no target):
       wiggle:  ...  /\  /\  /\  /\  ...
                    \/  \/  \/  \/

   same ring, after the target:
       wiggle:  ... /\  /\  /\  /\   ...   (whole pattern pushed over
                   \/  \/  \/  \/           by the phase shift delta_l)

   same wiggle speed (energy unchanged), only the pattern is nudged.

READING OFF THE WHOLE PATTERN
=============================

   each ring contributes  (2 l + 1) sin^2(delta_l)  to the total
   total deflection chance  =  (4 pi / k^2) * sum over rings
   no nudge (delta_l = 0)   ->  that ring scatters nothing

Quantity	Symbol	Meaning
Angular-momentum label	$\ell$	Which ring; $0$ head-on, larger means wider aim
Phase shift	${\delta }_{\ell }$	How far the target nudges ring $\ell$
Wavenumber	$k$	Sets the wiggle speed; fixed by the energy
Total cross-section	$\sigma$	Total chance of being deflected, summed over rings

Worked example Beginner

Problem. A slow particle scatters off a small hard ball of radius $a$. At low speed only the head-on ring ($\ell =0$) reacts; all the wider rings sail past untouched. For a hard ball the head-on phase shift is ${\delta }_0=-ka$, where $k$ sets the wiggle speed. Find the total chance of scattering, the total cross-section, when the particle is slow, so $ka$ is much smaller than 1.

Solution.

Step 1. Only the $\ell =0$ ring matters, so the total cross-section comes from one term. The recipe for the total cross-section is $\sigma =(4\pi /k^2)(2\ell +1){\sin }^2{\delta }_{\ell }$ summed over rings. With only $\ell =0$ surviving and $2\ell +1=1$:

$$\sigma =\frac{4\pi }{k^2}{\sin }^2{\delta }_0.$$

Step 2. Put in the hard-ball phase shift ${\delta }_0=-ka$:

$$\sigma =\frac{4\pi }{k^2}{\sin }^2(ka).$$

Step 3. The particle is slow, so $ka$ is small, and for a small angle $\sin (ka)$ is nearly $ka$ itself. Replace $\sin (ka)$ by $ka$:

$$\sigma \approx \frac{4\pi }{k^2}(ka{)}^2=4\pi a^2.$$

What this tells us. A slow particle sees a hard ball as having a cross-section of $4\pi a^2$ — four times the area of the ball's silhouette, $\pi a^2$, that a thrown pebble would feel. The wave wraps around the whole surface, not just the front face, so the quantum target looks bigger than the classical one. At low speed the answer does not depend on the speed at all: the cross-section settles to a fixed size set only by the ball's radius.

Check your understanding Beginner

Formal definition Intermediate+

Consider a particle of mass $m$ and energy $E={\hslash }^2{k}^2/2m$ scattering off a central potential $V(r)$ that vanishes faster than $1/r$ at large $r$. Because $V$ depends only on $r$, the Hamiltonian commutes with the orbital angular momentum, and the stationary scattering state separates in spherical coordinates. With the beam along $z$, the state is azimuthally symmetric and admits the partial-wave expansion

$$\psi (\mathbf{r})=\sum _{\ell =0}^{\infty }(2\ell +1)i^{\ell }{R}_{\ell }(r)P_{\ell }(\cos \theta ),$$

where $P_{\ell }$ are the Legendre polynomials and the factors $(2\ell +1)i^{\ell }$ are fixed by matching to the plane wave (the Rayleigh expansion of $e^{ikz}$). Writing $R_{\ell }(r)=u_{\ell }(r)/r$, the radial Schrödinger equation is

$$u_{\ell }^{\prime \prime }+\left[k^2-\frac{\ell (\ell +1)}{r^2}-U(r)\right]u_{\ell }=0,U=\frac{2m}{{\hslash }^2}V,$$

regular at the origin, $u_{\ell }(0)=0$.

Free solutions. Where $U=0$ the radial equation is solved by the spherical Bessel functions: the regular $j_{\ell }(kr)$ and the irregular $n_{\ell }(kr)$ (the spherical Neumann function). Their large-argument asymptotics are $$j_{\ell }(x)\stackrel{x\to \infty }{\to }\frac{\sin (x-\ell \pi /2)}{x},n_{\ell }(x)\stackrel{x\to \infty }{\to }-\frac{\cos (x-\ell \pi /2)}{x}.$$

The phase shift. Beyond the range of $V$ the true $R_{\ell }$ is a combination of $j_{\ell }$ and $n_{\ell }$, so its asymptotic form is a shifted sine:

$$R_{\ell }(r)\stackrel{r\to \infty }{\to }\frac{1}{kr}\sin \left(kr-\frac{\ell \pi }{2}+{\delta }_{\ell }\right).$$

The single number ${\delta }_{\ell }$, the phase shift, encodes the entire effect of the potential on partial wave $\ell$ at energy $E$. The sign convention is the standard one: an attractive potential pulls the wave inward and gives ${\delta }_{\ell }>0$; a repulsive one pushes it out, ${\delta }_{\ell }<0$.

The scattering amplitude. Matching the partial-wave sum to the asymptotic form $\psi \sim e^{ikz}+f(\theta )e^{ikr}/r$ — separating incoming and outgoing spherical waves and using the Rayleigh expansion for the plane-wave piece — gives

$$f(\theta )=\frac{1}{2ik}\sum _{\ell =0}^{\infty }(2\ell +1)\left(e^{2i{\delta }_{\ell }}-1\right)P_{\ell }(\cos \theta )=\frac{1}{k}\sum _{\ell =0}^{\infty }(2\ell +1)e^{i{\delta }_{\ell }}\sin {\delta }_{\ell }{P}_{\ell }(\cos \theta ).$$

The quantity $S_{\ell }=e^{2i{\delta }_{\ell }}$ is the partial-wave $S$-matrix element; $|S_{\ell }|=1$ encodes flux conservation in each channel. The treatment follows the partial-wave development in ^{[jimmyqin Partial-wave expansion, phase shifts, total cross-section]}.

Counterexamples to common slips

• The phase shift is defined only modulo $\pi$ by the asymptotic sine alone; the absolute value (which makes Levinson's theorem meaningful) requires tracking ${\delta }_{\ell }$ continuously from ${\delta }_{\ell }=0$ at $k\to \infty$.
• The partial-wave sum converges term by term but slowly for a sharp potential: the number of contributing partial waves is roughly ${\ell }_{\max }\approx ka$, so a short-wavelength beam off a large target needs many terms, and the expansion is the wrong tool in that regime — the Born series is the right one.
• A real potential gives a real ${\delta }_{\ell }$, hence $|S_{\ell }|=1$; a complex (absorptive) optical potential gives $|S_{\ell }|<1$, signalling flux lost to inelastic channels, and then ${\sigma }_{\text{tot}}>{\sigma }_{\text{elastic}}$.

Key theorem with proof Intermediate+

Theorem (partial-wave cross-section and the optical theorem). For scattering off a central potential, the total elastic cross-section is

$$\sigma =\frac{4\pi }{k^2}\sum _{\ell =0}^{\infty }(2\ell +1){\sin }^2{\delta }_{\ell },$$

and the forward amplitude satisfies the optical theorem $\sigma =(4\pi /k)\mathrm{Im}f(0)$.

Proof. Start from the partial-wave amplitude $f(\theta )=\frac{1}{k}{\sum }_{\ell }(2\ell +1)e^{i{\delta }_{\ell }}\sin {\delta }_{\ell }{P}_{\ell }(\cos \theta )$. The total cross-section is $\sigma =\int |f(\theta ){|}^{2}d\Omega$. Expanding the modulus squared introduces a double sum over $\ell$ and ${\ell }^{\prime }$:

$$\sigma =\frac{1}{k^2}\sum _{\ell ,{\ell }^{\prime }}(2\ell +1)(2{\ell }^{\prime }+1)e^{i({\delta }_{\ell }-{\delta }_{{\ell }^{\prime }})}\sin {\delta }_{\ell }\sin {\delta }_{{\ell }^{\prime }}\int P_{\ell }(\cos \theta )P_{{\ell }^{\prime }}(\cos \theta )d\Omega .$$

The angular integral uses Legendre orthogonality. With $d\Omega =\sin \theta d\theta d\varphi$ and $x=\cos \theta$,

$$\int P_{\ell }{P}_{{\ell }^{\prime }}d\Omega =2\pi {\int }_{-1}^1{P}_{\ell }(x)P_{{\ell }^{\prime }}(x)dx=2\pi \cdot \frac{2}{2\ell +1}{\delta }_{\ell {\ell }^{\prime }}=\frac{4\pi }{2\ell +1}{\delta }_{\ell {\ell }^{\prime }}.$$

The Kronecker delta collapses the double sum to a single sum, and the phase factor $e^{i({\delta }_{\ell }-{\delta }_{\ell })}=1$ drops out:

$$\sigma =\frac{1}{k^2}\sum _{\ell }(2\ell +1{)}^2{\sin }^2{\delta }_{\ell }\cdot \frac{4\pi }{2\ell +1}=\frac{4\pi }{k^2}\sum _{\ell }(2\ell +1){\sin }^2{\delta }_{\ell }.$$

For the optical theorem, evaluate the amplitude in the forward direction. There $\theta =0$, so $\cos \theta =1$, and every Legendre polynomial satisfies $P_{\ell }(1)=1$. Hence

$$f(0)=\frac{1}{k}\sum _{\ell }(2\ell +1)e^{i{\delta }_{\ell }}\sin {\delta }_{\ell }.$$

Taking the imaginary part, $\mathrm{Im}(e^{i{\delta }_{\ell }}\sin {\delta }_{\ell })={\sin }^2{\delta }_{\ell }$, so

$$\mathrm{Im}f(0)=\frac{1}{k}\sum _{\ell }(2\ell +1){\sin }^2{\delta }_{\ell }=\frac{k}{4\pi }\sigma .$$

Rearranging gives $\sigma =(4\pi /k)\mathrm{Im}f(0)$. $■$

Bridge. The optical theorem in this partial-wave guise builds toward the same unitarity identity that appears again in the momentum-space treatment of 12.08.02, and the bridge is the single algebraic fact $\mathrm{Im}(e^{i{\delta }_{\ell }}\sin {\delta }_{\ell })={\sin }^2{\delta }_{\ell }$: the foundational reason the forward imaginary part measures the total cross-section is that flux conservation forces $|S_{\ell }|=1$, which is exactly the statement that $S_{\ell }=e^{2i{\delta }_{\ell }}$ lies on the unit circle. Putting these together, the phase shift is the angular-momentum-space analogue of the momentum-transfer amplitude: where the Born expansion reads the potential off as a Fourier transform at large energy, the phase-shift series resolves it ring by ring at small energy, and the central insight is that one real number per channel, defined purely by an asymptotic sine, generalises into the full $S$-matrix whose analytic structure governs bound states and resonances alike. This identifies the elastic cross-section with a sum of channel occupations ${\sin }^2{\delta }_{\ell }$, each capped at one by unitarity.

Exercises Intermediate+

Advanced results Master

The $S$-matrix and unitarity. Collecting the asymptotic outgoing wave of each channel, the partial-wave amplitude is governed by $S_{\ell }=e^{2i{\delta }_{\ell }}$, the diagonal element of the on-shell scattering operator in the angular-momentum basis. Probability conservation for a real potential forces $|S_{\ell }|=1$, so ${\delta }_{\ell }$ is real and each partial cross-section obeys the unitarity bound ${\sigma }_{\ell }=(4\pi /k^2)(2\ell +1){\sin }^2{\delta }_{\ell }\le (4\pi /k^2)(2\ell +1)$. When an absorptive (complex) optical potential opens inelastic channels, $|S_{\ell }|<1$, and the elastic and reaction cross-sections split as ${\sigma }_{\text{el},\ell }=(\pi /k^2)(2\ell +1)|1-S_{\ell }{|}^2$ and ${\sigma }_{\text{r},\ell }=(\pi /k^2)(2\ell +1)(1-|S_{\ell }{|}^2)$, with the optical theorem still tying their sum to $\mathrm{Im}f(0)$.

Low-energy expansion and effective range. For a short-range potential the s-wave phase shift admits the effective-range expansion $$k\cot {\delta }_0=-\frac{1}{a}+\frac{1}{2}{r}_0{k}^2+\cdots ,$$ with $a$ the scattering length and $r_0$ the effective range. The two parameters summarise low-energy scattering with no reference to the detailed potential, which is the foundation of pseudopotential and zero-range models. A large $|a|$ — diverging through $\pm \infty$ — signals a bound or virtual state sitting at threshold; the unitary limit $a\to \infty$ gives the energy-independent maximal cross-section $\sigma =4\pi /k^2$, central to cold-atom physics.

Levinson's theorem. Tracking ${\delta }_{\ell }$ continuously and fixing ${\delta }_{\ell }(\infty )=0$, the zero-energy phase shift counts the bound states: $${\delta }_{\ell }(0)=n_{\ell }\pi ,$$ where $n_{\ell }$ is the number of bound states of angular momentum $\ell$ supported by $V$. The phase shift winds by $\pi$ for each bound state, a topological statement: it is the change in the argument of the Jost function, equivalently a winding number of $S_{\ell }$ around the unit circle as the energy runs over the physical sheet.

Resonances and the Breit-Wigner form. A quasi-bound state at energy $E_r$ with width $\Gamma$ drives the phase shift rapidly upward through $\pi /2$: $${\delta }_{\ell }(E)\approx {\delta }_{\ell }^{\text{bg}}+\arctan \frac{\Gamma /2}{E_r-E},$$ producing the Breit-Wigner cross-section ${\sigma }_{\ell }\propto [(E-E_r{)}^2+(\Gamma /2{)}^2{]}^{-1}$. In the complex energy plane this is a pole of $S_{\ell }$ at $E_r-i\Gamma /2$ on the second (unphysical) sheet, the analytic continuation of the bound-state poles that sit on the negative real axis of the first sheet. The lifetime $\tau =\hslash /\Gamma$ measures how long the particle is trapped behind the centrifugal-plus-potential barrier. The analysis follows the resonance treatment in ^{[jimmyqin Partial waves, scattering length, resonances]}.

Relation to the high-energy picture. The partial-wave sum and the Born series are complementary expansions of the same amplitude. At low energy only a few $\ell$ contribute and the phase-shift series converges in a handful of terms; at high energy ${\ell }_{\max }\sim ka$ partial waves matter and the sum is impractical, while the Born/eikonal expansion of 12.08.02 becomes accurate. The eikonal amplitude is precisely the stationary-phase limit of the partial-wave sum with $\ell +\frac{1}{2}\to kb$ (impact parameter $b$) and $2{\delta }_{\ell }\to \chi (b)$ the eikonal phase. The cross-references to the momentum-transfer amplitude follow ^{[jimmyqin Partial-wave expansion, phase shifts, total cross-section]}.

Synthesis. The phase shift is the foundational reason scattering off a central potential reduces to a single real number per channel: unitarity confines $S_{\ell }=e^{2i{\delta }_{\ell }}$ to the unit circle, and the central insight is that all physical content — cross-section, bound-state count, resonance position — is read off its motion there. The cross-section formula $\sigma =(4\pi /k^2)\sum (2\ell +1){\sin }^2{\delta }_{\ell }$ is dual to the Born Fourier transform of 12.08.02: putting these together, the angular-momentum and momentum-transfer descriptions are stationary-phase images of one another, joined through the eikonal limit $\ell +\frac{1}{2}=kb$. Levinson's theorem identifies the zero-energy phase with the bound-state count, the same analytic-continuation logic that turns a resonance into a complex-energy pole of $S_{\ell }$ and a true bound state into a pole on the physical sheet; this is exactly the unification that generalises into the analytic $S$-matrix programme, where poles, cuts, and the unitarity relation of 12.08.02 become the complete kinematic data of a scattering theory. The pattern recurs in the relativistic partial-wave (Regge) analysis, where $\ell$ itself is continued into the complex plane and bound states and resonances trace out Regge trajectories.

Full proof set Master

Proposition (Levinson's theorem for the s-wave). Let $V$ be a short-range central potential supporting $n_0$ s-wave bound states, with the s-wave phase shift normalised so that ${\delta }_0(k)\to 0$ as $k\to \infty$. Then ${\delta }_0(0)=n_0\pi$.

Proof. Work with the Jost function ${\mathcal{F}}_0(k)$, the Wronskian of the regular solution ${\phi }_0(r,k)$ (fixed by ${\phi }_0\sim r$ near the origin) with the Jost solution $f_0(r,k)$ (fixed by $f_0\sim e^{ikr}$ at infinity). Two standard facts assemble the result. First, the phase shift is the negative argument of the Jost function on the positive real axis: $S_0(k)={\mathcal{F}}_0(-k)/{\mathcal{F}}_0(k)=e^{2i{\delta }_0(k)}$, so ${\delta }_0(k)=-\arg {\mathcal{F}}_0(k)$ for $k>0$, with the normalisation $\arg {\mathcal{F}}_0(\infty )=0$. Second, ${\mathcal{F}}_0(k)$ extends analytically to the upper half $k$-plane, where its only zeros are simple and lie on the positive imaginary axis at $k=i{\kappa }_n$, one for each bound state $E_n=-{\hslash }^2{\kappa }_n^2/2m$.

Apply the argument principle to ${\mathcal{F}}_0$ on a large semicircular contour closing the upper half-plane: the number of enclosed zeros is the winding of $\arg {\mathcal{F}}_0$, $$n_0=\frac{1}{2\pi }\oint d\arg {\mathcal{F}}_0(k).$$ On the large semicircle ${\mathcal{F}}_0\to 1$ (the potential becomes negligible at high momentum), contributing nothing. The real axis from $-\infty$ to $+\infty$ contributes $\arg {\mathcal{F}}_0(-\infty )-\arg {\mathcal{F}}_0(+\infty )$; using the reflection ${\mathcal{F}}_0(-k)={\mathcal{F}}_0(k{)}^{\ast }$ for real $k$ (real potential), this real-axis change of argument equals $-2[\arg {\mathcal{F}}_0(0^{+})-\arg {\mathcal{F}}_0(\infty )]/(2)=\arg {\mathcal{F}}_0(0^{+})$ doubled across the two halves, and bookkeeping the factor of two against ${\delta }_0=-\arg {\mathcal{F}}_0$ yields ${\delta }_0(0)-{\delta }_0(\infty )=n_0\pi$. With ${\delta }_0(\infty )=0$ this is ${\delta }_0(0)=n_0\pi$. A half-integer correction ${\delta }_0(0)=(n_0+\frac{1}{2})\pi$ arises in the exceptional case of a zero-energy resonance (a zero of ${\mathcal{F}}_0$ at $k=0$), which the generic short-range potential avoids. $■$

Proposition (unitarity bound on a partial cross-section). For a real central potential each partial-wave elastic cross-section satisfies ${\sigma }_{\ell }\le (4\pi /k^2)(2\ell +1)$, with equality precisely when ${\delta }_{\ell }=\pi /2(\mathrm{mod}\pi )$.

Proof. Flux conservation for a real potential makes the radial scattering operator unitary in each channel, so $S_{\ell }=e^{2i{\delta }_{\ell }}$ with real ${\delta }_{\ell }$ and $|S_{\ell }|=1$. The partial cross-section is ${\sigma }_{\ell }=(4\pi /k^2)(2\ell +1){\sin }^2{\delta }_{\ell }$. Since ${\sin }^2{\delta }_{\ell }\in [0,1]$ for real ${\delta }_{\ell }$, the bound ${\sigma }_{\ell }\le (4\pi /k^2)(2\ell +1)$ holds, and ${\sin }^2{\delta }_{\ell }=1$ exactly when ${\delta }_{\ell }\equiv \pi /2(\mathrm{mod}\pi )$. The maximum is attained on resonance, where the Breit-Wigner phase passes through $\pi /2$. Were the potential absorptive, $|S_{\ell }|<1$ and the elastic part ${\sigma }_{\text{el},\ell }=(\pi /k^2)(2\ell +1)|1-S_{\ell }{|}^2$ would be bounded by $(4\pi /k^2)(2\ell +1)$ at $S_{\ell }=-1$, but a portion of the incident flux would be diverted to reaction channels rather than reappearing elastically. $■$

Connections Master

• The Born approximation and Lippmann-Schwinger equation 12.08.02 supply the momentum-transfer amplitude $f(\mathbf{q})=-(m/2\pi {\hslash }^2)\tilde{V}(\mathbf{q})$ that is the high-energy complement of the phase-shift series developed here. The two are stationary-phase images of one another through the eikonal substitution $\ell +\frac{1}{2}=kb$, and the optical theorem proved here in the angular-momentum basis is identical to the unitarity identity proved there in the momentum basis.

• Spherical harmonics and Legendre polynomials 12.05.02 furnish the angular basis $P_{\ell }(\cos \theta )$ in which the amplitude is expanded, and their orthogonality relation is the exact algebraic step that turns $\int |f{|}^{2}d\Omega$ into the diagonal sum ${\sum }_{\ell }(2\ell +1){\sin }^2{\delta }_{\ell }$. The Rayleigh expansion of the plane wave in the same basis fixes the $(2\ell +1)i^{\ell }$ weights of the partial-wave sum.

• Time-independent perturbation theory 12.07.01 is the weak-coupling tool that, applied to the radial equation, produces the Born phase shift ${\delta }_{\ell }\approx -(2mk/{\hslash }^2)\int V(r)[j_{\ell }(kr){]}^2{r}^{2}dr$; the same free-resolvent expansion that perturbs bound-state energies perturbs the continuum phase shift, the bound-state and scattering faces of one calculation.

• The Hilbert-space formalism 12.02.01 underpins the $S$-matrix element $S_{\ell }=e^{2i{\delta }_{\ell }}$ as the on-shell restriction of a unitary operator; the unitarity bound ${\sin }^2{\delta }_{\ell }\le 1$ and Levinson's bound-state count are spectral-theory statements about that operator and its analytic continuation, drawing on the self-adjointness machinery established there.

Historical & philosophical context Master

The partial-wave method predates quantum mechanics. Lord Rayleigh, in the Theory of Sound (1894), expanded acoustic waves scattered by a sphere in Legendre series and read off the scattered field mode by mode ^{[Rayleigh 1894]}; the phase-shift idea is already present there as the alteration of each spherical-harmonic component by the obstacle. The quantum transcription came from Hilding Faxén and Johan Holtsmark, who in 1927 applied the expansion to electron scattering off atoms and defined the phase shifts ${\delta }_{\ell }$ as the central observables of the quantum collision problem ^{[Faxen 1927]}, computing cross-sections for slow electrons in noble gases and explaining the Ramsauer-Townsend transparency minimum as an s-wave phase shift passing through a multiple of $\pi$.

The deeper analytic structure was supplied in stages. Norman Levinson proved in 1949 that the zero-energy phase shift counts the bound states, ${\delta }_{\ell }(0)=n_{\ell }\pi$, by an argument-principle analysis of the Jost function ^{[Levinson 1949]}, tying a scattering observable to a discrete spectral count. Gregory Breit and Eugene Wigner had already given the resonance line shape in 1936 in the context of slow-neutron capture, deriving the dispersion formula $\sigma \propto [(E-E_r{)}^2+(\Gamma /2{)}^2{]}^{-1}$ that bears their names ^{[BreitWigner 1936]}. Landau and Lifshitz consolidated the apparatus in Volume 3 of their Course, Chapter XVII §§122-124, which remains the canonical physicist's reference for the partial-wave expansion, the phase shift, and the low-energy scattering length.

Bibliography Master

@book{Rayleigh1894,
  author    = {Rayleigh, J. W. Strutt, Baron},
  title     = {The Theory of Sound},
  edition   = {2},
  volume    = {2},
  publisher = {Macmillan},
  year      = {1894}
}

@article{FaxenHoltsmark1927,
  author  = {Fax{\'e}n, H. and Holtsmark, J.},
  title   = {Beitrag zur Theorie des Durchganges langsamer Elektronen durch Gase},
  journal = {Zeitschrift f{\"u}r Physik},
  volume  = {45},
  number  = {5--6},
  pages   = {307--324},
  year    = {1927}
}

@article{Levinson1949,
  author  = {Levinson, Norman},
  title   = {On the Uniqueness of the Potential in a Schr{\"o}dinger Equation for a Given Asymptotic Phase},
  journal = {Kongelige Danske Videnskabernes Selskab, Matematisk-fysiske Meddelelser},
  volume  = {25},
  number  = {9},
  pages   = {1--29},
  year    = {1949}
}

@article{BreitWigner1936,
  author  = {Breit, Gregory and Wigner, Eugene},
  title   = {Capture of Slow Neutrons},
  journal = {Physical Review},
  volume  = {49},
  number  = {7},
  pages   = {519--531},
  year    = {1936}
}

@book{LandauLifshitz1977QM,
  author    = {Landau, L. D. and Lifshitz, E. M.},
  title     = {Quantum Mechanics: Non-Relativistic Theory},
  edition   = {3},
  series    = {Course of Theoretical Physics},
  volume    = {3},
  publisher = {Pergamon Press},
  year      = {1977}
}

@book{Newton1982,
  author    = {Newton, Roger G.},
  title     = {Scattering Theory of Waves and Particles},
  edition   = {2},
  publisher = {Springer-Verlag},
  year      = {1982}
}

@book{Taylor2006,
  author    = {Taylor, John R.},
  title     = {Scattering Theory: The Quantum Theory of Nonrelativistic Collisions},
  publisher = {Dover Publications},
  year      = {2006}
}

@book{SakuraiNapolitano2017,
  author    = {Sakurai, J. J. and Napolitano, Jim},
  title     = {Modern Quantum Mechanics},
  edition   = {2},
  publisher = {Cambridge University Press},
  year      = {2017}
}