Cosmic Ray Air Showers (3D)

One cosmic ray hits the atmosphere; watch the cascade, its maximum Xmax and the muons reach a tank array.

Switch to 2D

Presets

Playback

Primary particle

Choose the particle type above, then its energy and arrival direction. Higher energy means a bigger, deeper shower; a larger zenith angle means a longer slant path through more air.

Hadronic physics

n_tot is how many pions one nuclear collision makes; E_dec is the energy below which charged pions decay to muons instead of colliding; E_c is where electrons lose energy mainly by ionisation.

Ground array

Water-Cherenkov tanks sit on a triangular grid. Move the shower core to see how many tanks are hit and how well the core is reconstructed.

Drawing

Real showers are about 100 m wide and 20 km long, so the lateral spread and the tank size are drawn larger than life. Physics numbers are not affected.

Camera

Layers

Quality

Model checks

The science

What an air shower is

A cosmic-ray proton, nucleus or gamma ray hits an air nucleus high in the atmosphere. The collision makes many secondary particles, which collide again or decay, and a cascade of billions of particles spreads down toward the ground, a few tens of metres wide at the core and many kilometres long. Detectors sample the particles that reach the ground (tanks) and the faint ultraviolet fluorescence light the cascade excites in nitrogen (telescopes).

1. Atmospheric depth

Depth is the mass of air above you, in g/cm2. In an isothermal atmosphere with scale height H = 8.4 km: X(h) = 1036 exp(-h/H) vertically and Xslant = X/cosθ along a path at zenith angle θ. The density is ρ = X/H. The sea-level column is 1036 g/cm2 (about 28 radiation lengths); an inclined shower crosses much more air, so it is old and electron-poor by the time it reaches the ground (Model check 2 integrates the density numerically).

2. Heitler electromagnetic cascade

An electron (or positron) radiates a photon, and a photon makes an electron-positron pair, after about one splitting length d = X0 ln 2 with radiation length X0 = 36.7 g/cm2 in air. After n splittings there are N = 2n particles of energy E0/2n. Multiplication stops when the energy falls to the critical energy Ec = 85 MeV, where ionisation loss takes over:

nc = ln(E0/Ec)/ln 2,   Nmax = E0/Ec,   Xmax = X1 + X0 ln(E0/Ec)

X1 is the depth of the first interaction. Each factor 10 in energy moves Xmax deeper by X0 ln 10 = 85 g/cm2: the elongation rate (Model check 4). The page runs this cascade as a Monte Carlo with Hillas thinning (a particle below a threshold energy is replaced by one carrying double weight, so energy is conserved exactly, Model check 1). The toy model counts all particles; the e± number at maximum is several times smaller than E0/Ec, which is why the profile peak is lower than the naive Nmax.

3. Hadronic showers: the Heitler-Matthews model

A proton collides after one interaction length (λp about 50 g/cm2 at 1018 eV, from λ = Aair mu / σ) and makes ntot pions, one third neutral and two thirds charged. Neutral pions decay at once into two photons and start electromagnetic sub-showers; charged pions collide again after 120 g/cm2 until their energy drops below Edec (about 20 GeV), when they decay into muons. After n generations the pion energy is E0/ntotn and the number of charged pions (2 ntot/3)n, so

Nμ = (E0/Edec)β,   β = ln(2 ntot/3) / ln ntot ≈ 0.82 for ntot = 10,   Xmax = X1 + X0 ln(E0/(2 A ntot Ec))

The exponent β below 1 is why the muon number grows more slowly than the energy. Because the generation number is an integer, the exact count rises in steps (dashed lines in the fourth graph). Model check 3 compares an independent ensemble of Monte Carlo showers with this closed form.

4. Nuclei: the superposition model

A nucleus of mass number A and energy E0 behaves like A independent nucleons of energy E0/A. Iron therefore has a shallower maximum, Xmax(A) = Xmax(p) - X0 ln A (about 150 g/cm2 for iron), smaller fluctuations, and more muons, Nμ(A) = A1-β Nμ(p). Its first interaction is also higher up because its cross-section is larger, σA scaling as (A1/3 + 2.44)2. Measuring Xmax and the muon number together is how observatories estimate the mass composition of cosmic rays.

5. Muon-to-electron ratio

Electrons are absorbed with a length of about 37 g/cm2 past the maximum, while muons barely interact. A muon of energy E that is created a distance s above the ground survives with probability exp(-s mμc / (E cτμ)) (cτ = 658 m) and loses about 2.2 MeV per g/cm2. So the ratio Nμ/Ne at the ground grows with zenith angle (fifth graph) and is larger for iron than for protons at the same energy.

6. Longitudinal and lateral profiles, tank signals

The smooth profile drawn on the first graph is the Gaisser-Hillas function N(X) = Nmax ((X - Xs)/(Xmax - Xs))(Xmax - Xs)/Λ exp((Xmax - X)/Λ), whose width Λ is fitted so its integral equals the Monte Carlo track length. At the ground the electron density follows the NKG function ρ(r) = Ne C(s) (r/rM)s-2 (1 + r/rM)s-4.5 with Molière radius rM (about 90 m at 1.4 km altitude) and shower age s = 3X/(X + 2Xmax); muons use a flatter power law. A tank of area 10.2 m2 records S = nμ + 0.5 ne vertical-equivalent-muon units (VEM) with Poisson fluctuations; three or more tanks above 3 VEM make an event, and the core is reconstructed as the signal-weighted centroid.

Model checks on this page

(1) Energy conservation of the displayed shower; (2) numerical integral of the air density against the closed-form depth; (3) muon number from independent Monte Carlo showers against the Heitler-Matthews closed form; (4) elongation rate of photon showers against X0 ln 10; (5) linearity of the integrated electron track length with energy. They re-run when the hadronic parameters change.

Honest simplifications

  • Heitler-type toy cascade: splitting is 1 into 2 with equal energy sharing; real showers have Bethe-Heitler cross-sections, multiple scattering, LPM suppression above 1019 eV for photons and geomagnetic effects (none included).
  • Hadronic physics uses a fixed multiplicity n_tot (default 10), a fixed charged fraction 2/3, a fixed pion interaction length and a fixed E_dec. Real multiplicity rises with energy, kaons and baryons matter, and the measured elongation rate (about 60 g/cm2 per decade) is smaller than the model's 85. The muon number of real showers is higher than this model predicts at the highest energies (the so-called muon puzzle).
  • At the highest energies the pure cascade has not finished decaying when it reaches the ground, so muons produced below the ground are not counted; the readouts show muons produced and muons at the ground separately.
  • Shower axis, lateral spread and tank size are drawn larger than life; positions of individual tracks are schematic and thinned. Counts, depths and signals come from the model, not from the drawing.
  • The atmosphere is isothermal (no seasonal profile); the lateral and tank-response laws are illustrative fits (NKG form, an assumed 0.5 VEM per electromagnetic particle), not calibrated to a real detector. Photon showers are almost muon-free here; real photon showers have roughly thirty times fewer muons than proton showers.
  • Primary energies, cross-sections and elongation numbers are of the right order, not catalogue values. No detector, institution or experiment is represented.

What to try

  • Press Photon 10^17 eV and then Proton 10^18 eV: compare the clean electromagnetic cascade with the shower that has a yellow hadron core and red muon tracks.
  • Switch between Proton and Iron at the same energy. Watch Xmax move up (shallower) by roughly 150 g/cm2 and the muon number rise.
  • Raise the zenith angle to 65 degrees: the ground depth grows, the electrons die out before they arrive and the muon-to-electron ratio on the fifth graph climbs.
  • Lower the array altitude to sea level: more atmosphere, so a vertical high-energy shower is already past its maximum, and the Molière radius shrinks.
  • Drag the core position away from the centre and watch the tank signals and the green reconstructed-core ring.
  • Change n_tot or E_dec and watch the exponent β and the muon count change, then see Model check 3 hold.
Loading model...
Loading Blender model...
Longitudinal profile N(X) of electrons and photons with the Gaisser-Hillas fit
Lateral distribution at the ground and the measured tank signals
Shower maximum Xmax against energy for photons, protons and iron
Muons produced against energy: beta law and integer generations
Muon-to-electron ratio at the ground against zenith angle
Ground array map: tank signals, true and reconstructed core
Made with 🟠 Blender · Three.js