loading...
running...
How to use
Drag to orbit, wheel or pinch to zoom. Move the detector arm with the slider (or the 1/2 keys) and watch the zinc-sulphide screen flash: Run angle scan repeats Geiger and Marsden's measurement automatically and plots counts against angle with the 1/sin⁴(θ/2) law. Tracers are drawn to show what happens; cyan tracers are hard scatters, sampled from the rare tail and shown far more often than they occur. Space pauses.
The apparatus was modelled and baked in Blender 5.2 (named parts, brushed metals, lead, gold foil, ZnS screen) and animated with Three.js r160. The physics is the same module as the 2D page.
The science
An alpha particle (charge +2e, kinetic energy E) meets a target nucleus (charge +Ze) and feels only the Coulomb force F = k·2Z e²/r². The orbit is a hyperbola. With the head-on closest approach
d0 = 2 Z k e² / E (k e² = 1.44 MeV·fm)
energy and angular-momentum conservation give the deflection for impact parameter b:
tan(θ/2) = d0 / (2 b) <=> b = (d0/2) cot(θ/2)
Averaging over a uniformly illuminated area gives Rutherford's cross-section
dσ/dΩ = (d0/4)² / sin⁴(θ/2)
so a foil with na = ρ NA t / M nuclei per unit area scatters a fraction P(>θ) = na π b(θ)² = na π (d0/2)² cot²(θ/2) of the beam beyond θ.
What the sampler does. Each alpha gets (1) a Gaussian multiple-scattering core from all encounters with b between bc (the impact parameter of a 10° deflection) and the atomic radius R, with 〈θ²〉 = 2π na d0² ln(R/bc), and (2) with probability na π bc² one hard single encounter whose b is drawn uniformly over the disc b < bc. For the plum-pudding model (Thomson, 1904; positive charge spread uniformly through an atom of radius R) the force is linear inside the atom, so the many small kicks add as a random walk: 〈θ²〉 = π na d0²/2 and the angle follows exp(−θ²/〈θ²〉). That Gaussian tail is why a 90° deflection is not merely rare but impossible in that model: this is what Geiger and Marsden (1909) saw fail, and what Rutherford (1911) explained with a nucleus.
Numerical checks run in the page (button below): (1) a velocity-Verlet integration of Newton's equations in the 1/r² force reproduces tan(θ/2) = d0/2b; (2) the head-on turning point of that integration is d0 (energy conservation); (3) numerical integration of dσ/dΩ reproduces the closed form π(d0/2)²cot²; (4) Monte Carlo counts beyond 30°, 60° and 90° agree with na π b² N within 3.5 standard deviations.
Honest simplifications. Non-relativistic point nucleus; no electron screening except a cut-off at the atomic radius; no energy loss in the foil; no recoil (the gold nucleus is 49 times heavier than an alpha); at most one hard encounter per particle; thin foil. When d0 becomes smaller than the nuclear radius 1.2·A1/3 fm (light targets such as aluminium at high energy) the alpha touches the nucleus and the pure Coulomb formula stops applying; the page flags this. The classic quotation that about 1 in 8000 alphas was turned back by a thin metal foil (Geiger & Marsden 1909, platinum) is the same order of magnitude as the formula, but their foil, source energy and geometry differ, so it is not a fit.