Scattering Amplitude

Amplitude analysis is a working industry. At LHCb and BESIII, multi-body decays are fitted amplitude-first — magnitudes and phases — because resonances reveal themselves through phase motion invisible to single-angle counting; the 2020s exotic-hadron claims stand or fall on such fits. In ultracold-atom labs the low-energy limit of $f$ is dialled: Feshbach resonances tune the scattering length through divergence — the amplitude as a control knob rather than an output. Theory has its own renaissance: the S-matrix bootstrap constrains amplitudes by unitarity and analyticity alone, and amplitude methods now drive collider QCD and even gravitational-wave source modelling. A century after Born's footnote, computing $f$ is still the field's shared verb.

Send a steady beam down the $z$-axis: a plane wave $e^{ikz}$, crests marching in lockstep. The target chips off a faint outgoing ripple, $f(\theta,\phi)\,e^{ikr}/r$, centred on itself. The amplitude $f$ is the ripple's directional voice — how loudly, and with what phase, the target re-radiates into each direction. A detector parked at $(\theta,\phi)$, far away, counts $|f(\theta,\phi)|^2$ per unit solid angle per unit incident flux. That is the whole measurement, and the rest of the chapter is about computing $f$. The $1/r$ is not decoration. Probability streaming outward through ever-larger spheres must be conserved, and only an amplitude falling exactly as $1/r$ makes the outward current through every sphere the same — Asymptotic Form of the Scattering Wave Function proves this is forced, not chosen. Squaring $1/r$ gives the $1/r^2$ dilution of any point-source flux, and the detector's area $r^2\mathrm{d}\Omega$ cancels it perfectly: the count rate is distance-independent, as any lab measurement requires. One subtlety hides in "plane wave plus ripple": the two coexist everywhere, so $|\psi|^2$ contains an interference term. Naively that should wreck the counting picture. It doesn't — the relative phase $kr(1-\cos\theta)$ changes furiously with angle everywhere except dead ahead. Off-axis the interference washes out under any finite resolution; only forward does it survive, permanently, at any distance. The beam leaves the target carrying a scar at $\theta=0$ — and that scar, quantified by the optical theorem in its own article, is how the amplitude pays for every particle it scatters away.

Every scattering experiment prepares the same far-field furniture. For a potential dying faster than $1/r$ (small print in the warning below), the stationary state at $E=\hbar^2k^2/2m$ approaches a universal shape. Asymptotic Form of the Scattering Wave Function proves the shape — the $e^{ikr}/r$ structure with exactly one power of $r$ — is forced by the free equation plus flux conservation plus the outgoing condition; we define our object on top of that result:

where: $k=\sqrt{2mE}/\hbar$ is the beam wavenumber; $e^{ikz}$ is the incident plane wave; $e^{ikr}/r$ is the outgoing spherical wave; $A$ is an overall normalization that cancels in every observable; and $f(\theta,\phi)$ — the scattering amplitude — is the one dynamical item on the page. Everything about the target that a distant detector can know is compressed into this single complex function of direction. Dimensional bookkeeping already teaches something: the plane-wave term is dimensionless while $e^{ikr}/r$ carries $1/\mathrm{length}$, so the amplitude is a length. Its square is an area per solid angle — the currency of $\mathrm{d}\sigma/\mathrm{d}\Omega$ established in Differential and Total Cross Section. Nuclear physics quotes $f$ in fm and $|f|^2$ in $\mathrm{fm^2/sr}$ ($1\,\mathrm{b}=100\,\mathrm{fm^2}$); atomic collisions, in $a_0$ and $a_0^2$. For a central potential, symmetry about the beam axis removes $\phi$: $f=f(\theta)$, the case of all numbers below. The amplitude also depends on energy; its behaviour near a quasi-bound state is the resonance story Resonances and the Breit-Wigner Formula owns.

The claim $\mathrm{d}\sigma/\mathrm{d}\Omega=|f|^2$ is used constantly and rarely derived with its interference small print. Here is the full statement, with the probability current $\mathbf{j}=(\hbar/m)\,\mathrm{Im}(\psi^*\nabla\psi)$ — whose derivation Probability Current and the Continuity Equation owns — doing all the work.

The cross term of Step 5 is this article's owned laboratory. Integrate it over the sphere, weighting each angle with a smooth angular profile $f(\theta)$ (substituting $u=\cos\theta$):

where: the two surviving contributions are endpoint terms — the forward direction $u=1$ ($\theta=0$) and the backward direction $u=-1$ ($\theta=\pi$); everything in between has self-cancelled. Physically: an oscillatory integral only remembers where its phase stops sweeping, and $kr(1-\cos\theta)$ is stationary in $u$ nowhere inside the interval — the boundary directions are all that survive, each already suppressed by $1/kr$. The two endpoints are not equals. The forward term $-f(1)/ikr$ carries no residual $r$-phase — locked to the beam at any distance. The backward term rides $e^{2ikr}$, which whirls with the slightest spread in $k$ or $r$: averaging the exact integral over a mere $2\%$ band around $kr=10^3$ leaves the forward term alone to better than $2\%$ (machine-measured). What survives all averaging is interference in the forward direction only — the plane wave permanently marked by the target. This is as far as this article goes, and the boundary is deliberate: the value of that forward scar — the theorem that its depleted flux equals $\sigma_{\mathrm{tot}}$, i.e. $\sigma_{\mathrm{tot}}=(4\pi/k)\,\mathrm{Im}f(0)$ — is the property of Optical Theorem. What we have proved is the mechanism: where the interference lives, why it survives, and at what rate ($\theta_1\propto(kr)^{-1/2}$) the rest of the sphere purges it.

Elastic scattering has two directions ($\mathbf{k}$ in, $\mathbf{k}'$ out) but only one dynamical variable beyond the energy. The natural compression is the momentum transfer, and every amplitude formula downstream — Born's Fourier transform above all — is a function of it.

Partial Wave Expansion of Plane Waves closed with a deed: "amplitude, $f(\theta)$, cross sections — belongs BY NAME to Scattering Amplitude, Partial-Wave Expansion and Phase Shifts." This article is the first heir: $f$ has three equivalent formulations, each owned downstream, stated here so the chapter's architecture is visible from its head.

The integral face. The Schrödinger equation with outgoing boundary conditions is equivalent to an integral equation — Lippmann-Schwinger Equation derives it — whose far-field limit, with the explicit kernel of Green's Function for Scattering, gives $f=-\frac{m}{2\pi\hbar^2}\int e^{-i\mathbf{k}'\cdot\mathbf{r}'}V\,\psi\,\mathrm{d}^3r'$: the potential integrated against the full scattering state. Exact, mostly unsolvable — the launching point of every approximation.

The Born face. Replace the full $\psi$ inside that integral by the incident plane wave and the amplitude collapses to a Fourier transform of the potential evaluated at the momentum transfer $\mathbf{q}$ of the previous section. When that surgery is legitimate, and what it yields for real potentials, is the property of Born Approximation and Born Series and Higher-Order Corrections.

The partial-wave face. For central potentials, rotational symmetry diagonalizes scattering in angular momentum: $f(\theta)=\sum_\ell(2\ell+1)f_\ell\,P_\ell(\cos\theta)$, with each channel amplitude $f_\ell$ carried by a single real phase shift $\delta_\ell$. The decomposition machinery is Partial-Wave Expansion's; the physics of $\delta_\ell$ is Phase Shifts'; the unitarity constraints tie into The S-Matrix and Unitarity in Scattering.

One structural fact belongs to the object itself: $f$ is complex. Detectors measure only $|f|^2$, so the phase at a single angle is invisible — but not idle: the forward phase is pinned by the interference proved above, and the phase's energy dependence is what sweeps through a resonance. Max Born's 1926 collision papers forced this probabilistic reading of a complex amplitude on physics — his story lives with Born Statistical Interpretation; here it is one credit line. The simplest inhabitant of all three faces: a constant amplitude $f_0$, giving isotropic $|f_0|^2$ and $\sigma_{\mathrm{tot}}=4\pi|f_0|^2$ (verified symbolically). Low-energy scattering approaches exactly this — one number summarizing the target — and Low-Energy Scattering and the Scattering Length owns that limit and its surprises.

Example 1 — from a model amplitude to detector currency. A low-energy nuclear amplitude is modelled as $f(\theta)=(a+b\cos\theta)$ with $a=2.0$ fm, $b=1.5$ fm (an s-wave term plus a small forward–backward asymmetry). Find the differential cross section at $0°$, $90°$, $180°$, and the total cross section in barns.

Step 1. Enumerate what the current theorem gives: $\mathrm{d}\sigma/\mathrm{d}\Omega=|f(\theta)|^2=(a+b\cos\theta)^2$ — real amplitude, so squaring is elementary.

Step 2. Apply at the three angles: at $\theta=0$: $(2.0+1.5)^2=\mathbf{12.25\;fm^2/sr}$; at $90°$: $(2.0)^2=\mathbf{4.00\;fm^2/sr}$; at $180°$: $(2.0-1.5)^2=\mathbf{0.25\;fm^2/sr}$. A $49{:}1$ forward/backward ratio from a modest $b/a=0.75$ — squaring exaggerates asymmetries.

Step 3. Integrate for the total. The cross term $2ab\cos\theta$ integrates to zero over the sphere (verified symbolically), leaving

Step 4. Convert to barns with $1\,\mathrm{b}=100\,\mathrm{fm^2}$: $\sigma_{\mathrm{tot}}=\mathbf{0.597\;b}$ — a typical few-hundred-millibarn nuclear-scale answer, and the boundary number a beam experiment would quote.

Example 2 — the inverse problem, and what it cannot tell you. A detector scan reports $\mathrm{d}\sigma/\mathrm{d}\Omega = 10.89$, $4.00$, $0.49\;\mathrm{fm^2/sr}$ at $\theta=30°$, $90°$, $150°$. Extract $|f|$ at each angle and state what remains unknown.

Step 1. Invert the current theorem pointwise: $|f(\theta)|=\sqrt{\mathrm{d}\sigma/\mathrm{d}\Omega}$, giving $|f|=\mathbf{3.30}$, $\mathbf{2.00}$, $\mathbf{0.70}$ fm at the three angles.

Step 2. Check consistency with a two-term model: $a+b\cos\theta$ with $a=2.0$, $b=1.5$ fm reproduces all three moduli (at $30°$: $2.0+1.5\times0.866=3.30$ ✓). The scan is consistent with Example 1's amplitude.

Step 3. State the blindness honestly: the data determine $|f|$, never the phase — $f$ and $fe^{i\alpha}$ fit equally well. Recovering phase needs interference: the forward theorem (Optical Theorem), unitarity (The S-Matrix and Unitarity in Scattering), or amplitude-mixing measurements. Answer: $|f| = 3.30/2.00/0.70$ fm; the phase is invisible at this level.

References. Griffiths, Introduction to Quantum Mechanics, 3rd ed., Ch. 11 · Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., Ch. 6 · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 19 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. II, Ch. VIII · Townsend, A Modern Approach to Quantum Mechanics, Ch. 13

Provenance. Equations cross-checked against Griffiths 3rd ed. Ch. 11 (asymptotic form, partial waves, Born), Sakurai & Napolitano 2nd ed. Ch. 6 (Lippmann–Schwinger, optical theorem), and Shankar 2nd ed. Ch. 19. A verifier should reconfirm: the prefactor $-m/2\pi\hbar^2$ in the exact/Born amplitude; $a_\ell=(e^{2i\delta_\ell}-1)/2ik$; the optical theorem $\sigma_{\mathrm{tot}}=(4\pi/k)\mathrm{Im}f(0)$; the Yukawa Born amplitude $-2m\beta/\hbar^2(q^2+\mu^2)$ and its Rutherford limit; and the unitarity bound $4\pi(2\ell+1)/k^2$. No nonstandard numerical constants are quoted. (Lane note: of the base's menu, this article retains and re-verifies the $|f|^2$ current theorem and kinematics; the exact-integral, partial-wave/phase-shift, optical-theorem and Born derivations are chapter property of serials 1009–1017 and are stated here with owners named — no physics error was found in the base, and its displayed formulas match the owners' standard forms.)

Builds on: Asymptotic Form of the Scattering Wave Function · Differential and Total Cross Section · Lippmann-Schwinger Equation · Green's Function for Scattering Leads to: Born Approximation · Partial-Wave Expansion · Optical Theorem · The S-Matrix and Unitarity in Scattering · Resonances and the Breit-Wigner Formula