Low-Energy Scattering and the Scattering Length

Cool a collision far enough and the projectile goes blind. A thermal neutron carries a de Broglie wavelength of $1.80$ Å — four orders of magnitude larger than the nucleus it strikes. At that resolution the potential's depth, shape, and wiggles are invisible; the only question the wave can still ask is where the target pushes its phase. The answer developed here: one real number — the scattering length $a_s$ — replaces the entire potential. This article owns the threshold machinery of the chapter. We derive the Wigner threshold law $\tan\delta_\ell\propto(kR)^{2\ell+1}$ from the matching formula of Phase Shifts — cashing the statement that Effective Potential and the Centrifugal Barrier explicitly banked to this chapter — which is why only the s-wave survives as $k\to0$ . We derive the zero-energy trio $\delta_0\to-ka_s$ , $f\to-a_s$ , $\sigma\to4\pi a_s^2$ , resolving the factor-of-four surprise that Differential and Total Cross Section stated and deeded here. We derive the effective range expansion $k\cot\delta_0=-1/a_s+\tfrac12 r_0k^2$ by Bethe's two-energy comparison and verify it numerically on an exactly solvable well. And we compute the universal curve $a_s(V_0)$ for the square well — divergences where bound states are born, zeros where the target turns transparent. The payoff is the most famous number in low-energy nuclear physics: assembling the spin-weighted neutron–proton cross section from two measured lengths, $\pi(3a_t^2+a_s^2)=20.5$ b — the anchor the chapter opener quoted, finally computed from first principles.

Key takeaways

At $ka\ll1$ the centrifugal barrier silences every $\ell\ge1$ channel: $\tan\delta_\ell\propto(ka)^{2\ell+1}$ (the Wigner threshold law, derived here from the matching formula). Low-energy scattering is an s-wave monopoly.
The s-wave phase shift vanishes linearly, $\delta_0\to-k a_s$ , defining the scattering length $a_s=-\lim_{k\to0}\tan\delta_0/k$ — geometrically, the $r$ -intercept of the straight-line zero-energy wavefunction $u_0\propto(r-a_s)$ .
Everything at threshold follows from $a_s$ : amplitude $f\to-a_s$ (isotropic), cross section $\sigma\to4\pi a_s^2$ — four times the classical silhouette, a pure wave effect.
The next correction is one more length: $k\cot\delta_0=-1/a_s+\tfrac12 r_0k^2$ , the effective range expansion — verified here on the square well to $0.1\%$ (analytic $a_s=11.13$ fm, $r_0=2.27$ fm against a numerical fit).
The square-well master curve $a_s=R[1-\tan(K_0R)/(K_0R)]$ diverges each time a new bound state arrives ( $K_0R=\pi/2,\,3\pi/2,\dots$ ) and crosses zero at $\tan(K_0R)=K_0R$ — near-threshold poles and transparencies, the doorway to Bound States and Zero-Energy Resonances.
Flagship: $a_t=+5.42$ fm and $a_s=-23.7$ fm assemble into $\sigma_{np}=\pi(3a_t^2+a_s^2)=20.5$ b — the thermal neutron–proton cross section, spin bookkeeping included.
Key idea: at low energy a short-range potential scatters only s-waves, and a single length — the scattering length $a_s$ — replaces the entire potential, giving an isotropic amplitude $f\to-a_s$ and a constant cross section $\sigma\to4\pi a_s^2$ .

Where this lives today

The scattering length may be the most used single number in quantum physics. Every cold-atom paper begins by quoting $a_s$ : the Gross–Pitaevskii coupling $g=4\pi\hbar^2a_s/m$ , healing length, loss rates, and the very sign of stability all flow from it, and near a Feshbach resonance a magnetic field sweeps $a_s$ through $\pm\infty$ — deliberately driving the system across the divergences of the universal curve to make molecules, unitary Fermi gases, and Efimov trimers to order. In neutron science the coherent spin-channel combination is the material constant of reflectometry and crystallography. And pionless effective field theory expands in $r_0/a_s$ , treating the deuteron and the np virtual state as two faces of one near-threshold pole — the unification the next article completes.

The physical picture

Picture a slow wave, wavelength meters long, washing over a pebble. It cannot resolve the pebble's shape — but it does respond. Outside the pebble the zero-energy s-wave is exact and trivial: $u_0''=0$ , whose complete solution is the straight line $u_0(r)\propto(r-a_s)$ . Whatever violence the potential does inside, the exterior remembers one number: where that line crosses zero. That intercept is the scattering length. The sign tells a story. A repulsive core pushes the wavefunction out: the line crosses at positive $r$ , and for a hard sphere the crossing sits exactly at the surface, $a_s=R$ . A weak attraction pulls the wavefunction in: the line now rises from a negative intercept — $a_s<0$ . Strengthen the attraction toward the point where a bound state is about to form and the exterior line tilts toward horizontal: its intercept races off to infinity. A diverging scattering length is a bound state announcing itself at threshold — the pole story that Bound States and Zero-Energy Resonances owns.

Figure

The zero-energy wavefunction is a straight line and its intercept is the scattering length

Why is the cross section $4\pi a_s^2$ and not $\pi a_s^2$ ? Because a $k\to0$ wave does not cast a shadow — it breathes. The scattered wave at threshold is a pure isotropic s-wave emitted coherently by the whole target; its strength is set by the intercept $a_s$ , and integrating the isotropic $|f|^2=a_s^2$ over all $4\pi$ steradians gives the full sphere's worth, not the disk's. Differential and Total Cross Section flagged exactly this — "four times the classical shadow $\pi R^2$ " — and named this article the owner of the derivation, delivered below. One more face, the experimentalist's: plot $k\cot\delta_0$ against $k^2$ and low-energy data collapse onto a straight line. The intercept reads off $-1/a_s$ ; the slope reads off half the effective range $r_0$ . Two numbers, measured from a line on graph paper, summarize any short-range potential — that is the shape independence that built low-energy nuclear physics.

The threshold laws

Everything rests on the matching formula of Phase Shifts: the interior fixes a logarithmic derivative at the potential's edge; comparison with free spherical waves yields $\tan\delta_\ell$ . We need only its low- $k$ limit.

Derivation AThe Wigner threshold law from one matching formula

We prove that $\tan\delta_\ell\propto k^{2\ell+1}$ as $k\to0$ for any potential of finite range $R$ — the law whose statement Effective Potential and the Centrifugal Barrier explicitly banked to this chapter ("owned with its derivation by the scattering chapter"). The strategy: take the master matching formula of Phase Shifts as given, insert the small-argument behavior of the free spherical waves, and watch the powers of $k$ organize themselves. The only subtlety worth guarding is that the interior logarithmic derivative tends to a finite, $k$ -independent constant as $k\to0$ — true for finite-range potentials because the interior equation at $E=0$ is perfectly regular.

Step 1. Start from the matching formula (recalled by name from Phase Shifts, not re-derived): with $\gamma_\ell$ the logarithmic derivative of the interior radial function at $r=R$ ,

\tan\delta_\ell=\frac{k\,j_\ell'(kR)-\gamma_\ell\, j_\ell(kR)}{k\,n_\ell'(kR)-\gamma_\ell\, n_\ell(kR)} \qquad (1)

where: $j_\ell,n_\ell$ — spherical Bessel and Neumann functions; primes are derivatives with respect to the argument. As $k\to0$ , $\gamma_\ell\to$ const.

Step 2. Insert the small-argument forms $j_\ell(x)\to x^\ell/(2\ell+1)!!$ and $n_\ell(x)\to-(2\ell-1)!!/x^{\ell+1}$ . The numerator scales as the regular family: every term carries $(kR)^\ell\cdot k$ or $(kR)^\ell\gamma_\ell$ — overall $k^\ell$ . The denominator is dominated by the irregular family: $n_\ell\sim k^{-(\ell+1)}$ . Why the powers cannot conspire away. The numerator's two terms could in principle cancel at leading order — that is precisely what happens at a zero-energy resonance, and it is non-generic. Generically the ratio scales as $k^{\ell}/k^{-(\ell+1)}=k^{2\ell+1}$ .

Step 3. Hence

\tan\delta_\ell \;\xrightarrow{k\to0}\; c_\ell\,(kR)^{2\ell+1},\qquad c_\ell=O(1) \qquad (2)

where: $c_\ell$ depends on the interior physics through $\gamma_\ell$ but the power is universal. Each unit of $\ell$ costs two powers of $ka$ — the centrifugal barrier's tunneling toll, now quantitative.

Step 4. Read the hierarchy: $\delta_0\sim kR$ survives; $\delta_1\sim(kR)^3$ and $\delta_2\sim(kR)^5$ die. For a thermal neutron on a nucleus $kR\sim10^{-4}$ : the p-wave is down eight orders of magnitude. One channel, hence — next derivation — one number. Verified: the sympy expansion of formula (1) with the exact $j_\ell,n_\ell$ for $\ell=0,1,2$ returns lowest powers $k^1,k^3,k^5$ exactly; the exponent for $\ell=1$ was also measured on a numerically integrated square well (log–log slope $3.000$ ). Article verifier, 26 checks.

The law explains the monopoly; the next card cashes it into observables.

Go deeper · Derivation B · 1 minOne number at threshold: the intercept, the amplitude, and four pi a squared

Derivation BOne number at threshold: the intercept, the amplitude, and four pi a squared

We define the scattering length three equivalent ways — slope of the phase shift, intercept of the zero-energy wavefunction, zero-energy amplitude — and prove the threshold cross section $\sigma\to4\pi a_s^2$ , resolving the hard-sphere factor-of-four that Differential and Total Cross Section deeded to this article. The route: linearize the s-wave phase shift, then let the exterior wavefunction's geometry speak.

Step 1. Define, from the threshold law's $\ell=0$ rung,

a_s\equiv-\lim_{k\to0}\frac{\tan\delta_0(k)}{k} \qquad (3)

where: the sign is the Sakurai–Shankar convention, chosen so a hard sphere gives $a_s=+R$ (some nuclear texts flip it — see the pitfall box). Dimensionally $a_s$ is a length: the only length the $k\to0$ limit can produce.

Step 2. Geometry. Outside the potential the s-wave is $u_0(r)\propto\sin(kr+\delta_0)$ ; as $k\to0$ the sine linearizes, $u_0\propto k r\cos\delta_0+\sin\delta_0\propto r+\tan\delta_0/k\to r-a_s$ . The zero-energy exterior is the straight line of the hero figure, and its $r$ -intercept is $a_s$ . Why this is the honest definition. Equation (3) references scattering data; the straight line references the wavefunction. They agree identically, which is what lets bound-state physics (wavefunctions) and scattering physics (phase shifts) trade information — the bridge Bound States and Zero-Energy Resonances will drive across.

Step 3. Amplitude. The exact s-wave amplitude from the partial-wave ladder of Partial-Wave Expansion is $f_0=1/(k\cot\delta_0-ik)$ . As $k\to0$ , $k\cot\delta_0\to-1/a_s$ while $ik\to0$ :

f\;\xrightarrow{k\to0}\;-a_s \qquad (4)

where: the amplitude is real, isotropic, and energy-independent — a point target with a single voice.

Step 4. Cross section. With only the s-wave alive, $\sigma=(4\pi/k^2)\sin^2\delta_0\to(4\pi/k^2)(ka_s)^2$ :

\sigma\;\xrightarrow{k\to0}\;4\pi a_s^2 \qquad (5)

where: for the hard sphere ( $a_s=R$ , from $\delta_0=-kR$ exactly) this is $4\pi R^2$ — the entire surface area, four times the geometric shadow $\pi R^2$ . The wave scatters coherently off the whole sphere, not off a silhouette. Verified: $f_0\to-a_s$ and $\sigma\to4\pi a_s^2$ symbolic; hard-sphere $\tan\delta_0=-\tan kR$ hence $a_s=R$ symbolic; hard-sphere $\tan\delta_1\to-(kR)^3/3$ (negative, as repulsion demands). Article verifier.

The effective range expansion

The scattering length is the $k=0$ snapshot. Real experiments run at small but finite $k$ , and the first correction turns out to be another length — not a new function. This robustness is why two numbers per channel organized all of low-energy nuclear physics.

Go deeper · Derivation C · 2 minThe effective range from a two-energy comparison

Derivation CThe effective range from a two-energy comparison

We derive $k\cot\delta_0=-1/a_s+\tfrac12r_0k^2+O(k^4)$ and identify $r_0$ as an integral over the region where the true wavefunction differs from its own asymptote — Bethe's 1949 comparison argument, the one he called the shape-independent approximation. The strategy: write the s-wave problem at two energies, cross-multiply the two Schrödinger equations, integrate, and let the boundary terms do the work. What to watch: the comparison functions must be normalized to the same asymptotic line, or the boundary terms lie.

Step 1. Let $u_k(r)$ solve the s-wave equation at energy $\hbar^2k^2/2m$ with $u_k(0)=0$ , and let $v_k(r)$ be the free solution matching $u_k$ exactly outside the range, normalized so $v_k(0)=1$ at every energy. Write the radial equations at $k_1$ and $k_2$ , cross-multiply, subtract, integrate $0\to\infty$ : the potential terms cancel identically, leaving boundary terms.

Step 2. The identical manipulation applied to the free pair $(v_{k_1},v_{k_2})$ gives boundary terms at both ends; for the true pair $(u_{k_1},u_{k_2})$ the origin contributes nothing since $u(0)=0$ . Subtracting the two identities:

k_2\cot\delta_0(k_2)-k_1\cot\delta_0(k_1)=(k_2^2-k_1^2)\int_0^\infty\!\big(v_{k_1}v_{k_2}-u_{k_1}u_{k_2}\big)\,\mathrm{d}r \qquad (6)

where: the integrand lives only inside the potential — outside, $u=v$ by construction and the integrand vanishes. All shape dependence is confined to a short-range integral.

Step 3. Set $k_1=0$ , $k_2=k$ , and expand the integral at small $k$ (its $k$ -dependence is analytic because the integrand has finite support):

k\cot\delta_0(k)=-\frac{1}{a_s}+\frac{1}{2}r_0k^2+O(k^4),\qquad r_0=2\int_0^\infty\!\big(v_0^2-u_0^2\big)\,\mathrm{d}r \qquad (7)

where: $r_0$ — the effective range — measures the spatial extent over which the true zero-energy wavefunction fails to be its own asymptote. It is a range, not a strength: potentials of wildly different depth but similar reach share $r_0$ .

Step 4. Instant payoff — the hard sphere. There $u_0=0$ inside ( $r<R$ ) and $u_0=v_0=1-r/R$ outside, so $r_0=2\int_0^R(1-r/R)^2\mathrm{d}r=2R/3$ . A hard sphere of radius $2$ fm has $a_s=2$ fm, $r_0=1.33$ fm — both lengths of order the geometry, as they should be.

Step 5. Machine check on a solvable well ( $V_0=20$ MeV, $R=2.5$ fm, np reduced mass). Analytic: $a_s=R-\tan(K_0R)/K_0=11.13$ fm, $r_0=R-1/(K_0^2a_s)-R^3/(3a_s^2)=2.272$ fm. Numerical: exact $\delta_0(k)$ at twelve momenta, fit $k\cot\delta_0$ vs $k^2$ — a line to $0.1\%$ , returning $a_s=11.13$ fm, $r_0=2.274$ fm. Two lengths exhaust the low-energy content. Verified: hard-sphere $r_0=2R/3$ symbolic (Bethe integral); square-well $a_s$ closed form symbolic; the $k\cot\delta_0$ linearity, fitted $(a_s,r_0)$ vs analytic values, all in the article verifier.

Figure

The experimentalist's straight line: k·cot δ₀ versus k² for the solvable well

The universal curve: where scattering lengths come from

For the attractive square well the closed form is worth staring at. With $K_0=\sqrt{2mV_0}/\hbar$ the zero-energy interior wavenumber,

a_s=R\left[1-\frac{\tan(K_0R)}{K_0R}\right] \qquad (8)

where: the whole personality of the well enters through the single dial $K_0R$ . Weak well ( $K_0R\ll1$ ): $a_s\approx-R(K_0R)^2/3<0$ — the generic weak-attraction signature. As $K_0R\to\pi/2$ the tangent blows up: $a_s\to-\infty$ , reappearing at $+\infty$ past threshold. Each divergence marks a bound state's birth; each zero (at $\tan K_0R=K_0R=4.493,\dots$ ) makes the well invisible to zero-energy projectiles.

Figure

The universal curve aₛ against well depth: divergences at bound-state thresholds, zeros at transparency

The two marked points are the whole of low-energy nuclear physics. The neutron–proton triplet channel sits just past the first divergence: one shallow bound state — the deuteron — and a large positive $a_t=+5.42$ fm. The singlet channel sits just before it: no bound state, but so nearly one that $a_s=-23.7$ fm — enormous and negative. What "almost bound" means precisely — an S-matrix pole on the wrong axis, a virtual state — is owned by Bound States and Zero-Energy Resonances. The deuteron connection is quantitative with tools already on the table. Spherical Hankel Functions and Outgoing Waves computed the deuteron's exterior decay length $1/\kappa=4.32$ fm from $E_B=2.2246$ MeV. That is not the scattering length — the effective range bridges the 20% gap: continuing $k\cot\delta_0$ to the pole gives $1/a_t=\kappa-\tfrac12 r_t\kappa^2$ , and $r_t=1.76$ fm predicts $a_t=5.42$ fm from the binding energy alone — matching the measured value. Two lengths, one channel, closed books.

Limiting cases

case	result	meaning
hard sphere, $k\to0$	$a_s=R$ , $r_0=2R/3$ , $\sigma=4\pi R^2$	intercept at the surface; full surface area, not the shadow
weak attraction ( $K_0R\ll1$ )	$a_s\approx-R(K_0R)^2/3<0$	small negative $a_s$ : generic signature of an unbinding attraction
binding threshold $K_0R\to\pi/2$	$a_s\to\pm\infty$ , $\sigma\to$ huge	zero-energy resonance; pole at $k=0$ (owner: Bound States and Zero-Energy Resonances)
transparency $\tan K_0R=K_0R$	$a_s=0$ : no $k\to0$ scattering	the well is invisible — the Ramsauer–Townsend mechanism (its article, by name)
$ka_s\ll1$ , $a_s$ fixed	$\sigma=4\pi a_s^2$ constant	threshold plateau; energy-independence is the s-wave fingerprint
$k\vert a_s\vert\gg1$ (huge $a_s$ )	$\sigma\to4\pi/k^2$	the unitarity ceiling of The S-Matrix and Unitarity in Scattering: $a_s$ drops out entirely
$\ell\ge1$ at threshold	$\tan\delta_\ell\propto(kR)^{2\ell+1}$	Wigner law (derived here): centrifugal barrier silences all but s-waves

Common trapThe scattering length is not the size of the target

Three traps in one word "length". First, $a_s$ measures interaction, not geometry: a well on the verge of binding has finite range but $|a_s|\to\infty$ — the np singlet's $23.7$ fm dwarfs the $\sim2$ fm force range. Second, the sign convention $a_s=-\lim\tan\delta_0/k$ (hard sphere $\Rightarrow+R$ ) is Sakurai/Shankar standard, but parts of the nuclear literature flip it — check before comparing tables. Third, $\sigma=4\pi a_s^2$ is a zero-energy statement: at finite $k$ use the unitarized $4\pi a_s^2/(1+k^2a_s^2)$ , or you will exceed the unitarity bound the moment $ka_s\sim1$ .

Worked examples

Example 1 — the 20.5 barns of the thermal neutron. Differential and Total Cross Section anchored its benchmark ladder with "n+p elastic: 20.4 b" while its geometric estimate was $\sim1$ b. Assemble that number from the two measured scattering lengths: $a_t=+5.424$ fm (spin-triplet), $a_s=-23.749$ fm (spin-singlet).

Step 1. Spin bookkeeping. An unpolarized neutron meets an unpolarized proton in a random relative spin state: of the four two-spin states, three are triplet and one is singlet, so the statistical weights are $3/4$ and $1/4$ .

Step 2. Each channel scatters with its own length: $\sigma_t=4\pi a_t^2=4\pi(5.424)^2=369.7\ \mathrm{fm}^2=3.70$ b and $\sigma_s=4\pi a_s^2=4\pi(23.749)^2=7087\ \mathrm{fm}^2=70.9$ b. The nearly-bound singlet channel scatters twenty times harder than the channel with the actual bound state — proximity to threshold, not binding, is what matters.

Step 3. Weight and add (cross sections add because triplet and singlet final states are distinguishable in principle):

\sigma_{np}=\tfrac34\sigma_t+\tfrac14\sigma_s=\pi\big(3a_t^2+a_s^2\big)=2049\ \mathrm{fm}^2

where: $1$ b $=100\ \mathrm{fm}^2$ , so $\sigma_{np}=20.5$ b — the chapter opener's anchor, recovered from two intercepts. Answer: $\sigma_{np}=20.5$ b.

Step 4. The instructive wrong answer: averaging lengths first — $\bar a=(3a_t+a_s)/4=-1.87$ fm, $4\pi\bar a^2=0.44$ b — misses by 47×. Probabilities average, amplitudes don't. (That coherent $\bar a$ is physical in neutron optics as the coherent scattering length — a different, interference, question.)

Example 2 — an ultracold gas is a scattering length with a thermometer. ⁸⁷Rb atoms have $a_s\approx100\,a_0=5.29$ nm. Compute the collision cross section at $T=100$ nK.

Step 1. Check the monopoly: relative momentum $k=\sqrt{2\mu k_BT}/\hbar$ , $\mu=m_{\mathrm{Rb}}/2$ , gives $ka_s=0.022\ll1$ — deep threshold, p-wave suppression $\sim(ka)^3\approx10^{-5}$ .

Step 2. Identical bosons in one internal state interfere constructively in the s-wave, doubling the cross section: $\sigma=8\pi a_s^2$ (symmetrization belongs to Identical-Particle Scattering; we import the factor).

Step 3. Numbers: $\sigma=8\pi(5.29\times10^{-9}\ \mathrm{m})^2=7.0\times10^{-16}\ \mathrm{m}^2$ — about $2500$ times the atom's geometric size. Answer: $\sigma\approx7.0\times10^{-16}\ \mathrm{m}^2$ , with $ka_s=0.022$ certifying pure threshold physics. Every BEC mean-field number — healing length, sound speed, stability — enters through this one length via $g=4\pi\hbar^2a_s/m$ .

Origins · 1 minScattering Length and how this was discovered

Scattering Length

1936–1949 · Rome, Ithaca

The scattering length was born twice in one decade. In Rome, Enrico Fermi's 1936 slow-neutron work introduced the pseudopotential — a point interaction tuned to reproduce one measured length — so neutron waves in hydrogenous materials could be computed without knowing the nuclear force; its regularized descendant $g\,\delta^3(\mathbf r)$ , $g=4\pi\hbar^2a_s/m$ , still runs cold-atom theory. After the war the systematics crystallized: Julian Schwinger's variational treatment of np scattering circulated in the late 1940s, and in 1949 Hans Bethe published the transparent two-energy derivation of the effective range expansion — the "shape-independent approximation" — with Blatt and Jackson fitting the np data the same year. Two lengths per channel carried nuclear physics for decades, then found a second life when ultracold-atom experiments learned to tune $a_s$ through Feshbach resonances, making the scattering length the most engineered number in many-body physics.

History

1936

— Fermi introduces the pseudopotential for slow neutrons in hydrogenous media: one length stands in for the nuclear force

1947

— Schwinger's variational effective-range results for np scattering circulate (Harvard notes; published abstracts)

1949

— Bethe's two-energy derivation of the effective range expansion (Phys. Rev. 76, 38); Blatt and Jackson fit the np lengths the same year

1957

— neutron interferometry and reactor physics standardize coherent scattering lengths as tabulated material constants

1998

— a magnetically tuned Feshbach resonance in a sodium BEC (Ketterle group) turns the scattering length into an experimental dial

Going further

Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., §6.4–6.5 — low-energy scattering, the scattering length, and the square-well curve.
Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 19 — phase shifts and threshold behavior.
Griffiths, Introduction to Quantum Mechanics, 3rd ed., §11.3 — the hard sphere and partial waves at low energy.
Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. II, Ch. VIII — the square-well scattering length and its divergences.
Taylor, Scattering Theory, Ch. 11; Newton, Scattering Theory of Waves and Particles, Ch. 11 — effective-range theory done rigorously.
Bethe, Phys. Rev. 76, 38 (1949) — the two-energy derivation; Blatt & Jackson, Phys. Rev. 76, 18 (1949) — the np fits.

Provenance. All standard results from low-energy scattering theory. A verifier should: (1) confirm $f_0\to -a_s$ and $\sigma\to4\pi a_s^2$ from the partial-wave amplitude (Sakurai §6.5, Shankar Ch.19); (2) reproduce the hard-sphere values $\delta_0=-kR$ , $a_s=R$ , $r_0=2R/3$ (Griffiths §11.3); (3) verify the square-well matching $a_s=R[1-\tan(K_0R)/(K_0R)]$ and its divergence at $K_0R=(n+\tfrac12)\pi$ (Cohen-Tannoudji Vol. II); (4) check the Bethe effective-range derivation and integral for $r_0$ (Newton, Taylor). Quoted np data ( $a_t\approx+5.4$ fm, $r_{0t}\approx1.75$ fm, $a_s\approx-23.7$ fm, deuteron $E_B\approx2.22$ MeV) are standard tabulated nuclear values (e.g. Krane, Introductory Nuclear Physics). Contact coupling $g=4\pi\hbar^2a_s/m$ is standard cold-atom theory (Pethick & Smith). — All four verifier items are executed in this article's verification script (26 checks), including the square-well fit and the np assembly; the threshold-law and $4\pi a_s^2$ derivations live in this article by chapter ownership, with the matching formula recalled from Phase Shifts and the $(kR)^{2\ell+1}$ statement cashed from Effective Potential and the Centrifugal Barrier.

Builds on: Partial-Wave Expansion · Phase Shifts · Asymptotic Form of the Scattering Wave Function · Scattering Amplitude Leads to: Optical Theorem · The S-Matrix and Unitarity in Scattering · Differential and Total Cross Section

Is the scattering length something you can measure directly, or only infer?

Both. In scattering you infer it from the low-energy plateau $\sigma=4\pi a_s^2$ or the intercept of the $k\cot\delta_0$ line. In neutron optics it is almost directly visible: the coherent scattering length sets a material's neutron refractive index, so total-reflection angles and interferometer phases read out scattering lengths to four significant figures — the np and nn lengths are known this precisely.

Why does one number suffice — where did all the information about the potential go?

Still there, but unresolvable. The exterior wavefunction at $k\to0$ is a straight line no matter what; the interior can only fix its intercept. Every other detail enters through terms that vanish with powers of $kR$ — the first survivor being the effective range at order $k^2$ . It is multipole logic: at long distance you see the monopole first, and long wavelength is the momentum-space version of far away.

Can the scattering length be complex?

For elastic scattering off a real potential, never — $\delta_0$ is real, so $a_s$ is real. With open inelastic channels (absorption, reactions) the s-wave "length" gains a negative imaginary part encoding loss, parallel to $|S_\ell|<1$ in Inelastic Scattering and Multichannel Theory — why tabulated lengths for cadmium, gadolinium, boron are complex.

If the singlet np channel has the bigger length, why is there no singlet deuteron?

Because $a_s=-23.7$ fm is negative: the near-threshold pole sits on the wrong side of $k=0$ — a virtual state. Scattering magnitude feels only $|a_s|$ , so the singlet channel scatters ferociously, but binding requires the pole on the positive imaginary axis. The pole taxonomy, and how experiment told the cases apart, is the business of Bound States and Zero-Energy Resonances.

Does the effective range expansion always converge?

It is an expansion of $k\cot\delta_0$ in $k^2$ , analytic for strictly finite-range potentials; convergence is limited by the nearest singularity — the potential's range scale, or for a Yukawa tail $e^{-\mu r}/r$ the branch point at $k=i\mu/2$ . Power-law tails degrade it: the van der Waals $-C_6/r^6$ tail of atomic physics allows only a modified expansion. Ask what tail your potential has before trusting the straight line far from $k=0$ .

Is a large scattering length fine-tuned or generic?

Fine-tuned: on the universal curve, $|a_s|\gg R$ occupies narrow windows around the divergences. Its natural occurrences are accidents — np sits near threshold by grace of QCD, helium-4 dimers by van der Waals luck. Feshbach resonances removed the luck: park the system anywhere on the curve, which is why large- $a_s$ universality went from curiosity to research field.

Low-Energy Scattering and the Scattering Length

Where this lives today

The physical picture

The threshold laws

The effective range expansion

The universal curve: where scattering lengths come from

Limiting cases

Worked examples

Going further

Before this

Where this leads