A tray of unstable nuclei
The stage shows a sample as a square tray of 100, 400 or 1600 nuclei. Every glowing dot is a parent nucleus that has not decayed yet. In each tiny time step, each one has the same small chance of decaying, fixed by the isotope. When one goes, it flashes, sends out its radiation (orange alpha streaks, blue beta tracks or green gamma ripples, coloured as in the alpha, beta and gamma simulation) and dims to the daughter colour. Nothing about a nucleus warns you that it is next. The sample number makes the randomness repeatable: Reset replays exactly the same decays, while New sample deals a fresh set.
The staircase and the half-life line
The main graph plots the number of undecayed nuclei against time. Because nuclei decay one at a time, the line drops in small steps and wanders either side of the dashed theory curve, . Drag the orange line, or press its buttons, to set it at a half, a quarter, an eighth or a sixteenth of the starting number, and the time at which this sample first sank that low appears under the graph. Mark every halving adds a diamond at each halving and a bracket showing how long it took. The brackets come out roughly equal, and that is the whole idea of a half-life: the same fraction goes in each equal stretch of time, however many nuclei remain.
Small samples, big scatter
With 100 nuclei the brackets can differ by a quarter or more; with 1600 they agree far better. The scatter only grows as the square root of how many nuclei there are, so the relative wobble shrinks as samples grow. Real samples hold billions upon billions of nuclei, which is why measured decay curves look so smooth.
Counting with a GM tube
A Geiger–Müller tube beside the tray records a share of the decays, set by its efficiency. Its counter ticks and its lamp flashes for each one it catches. The Activity graph shows every decay in each counting interval, while Count rate shows only the recorded ones, so it sits lower. For the two lab isotopes, protactinium-234m and radon-220, and for a custom isotope, the tube also picks up background radiation at a rate you choose. Background does not decay, so the readings level off at it instead of falling towards zero. Halving uncorrected readings therefore gives a time that is too long, and subtracting the background from each reading first brings back the true half-life. A smooth best-fit curve through the points keeps the reading steady despite the noise.
Real isotopes on a time-lapse
Every run lasts roughly half a minute on screen, so the time axis is stretched or squeezed to suit the isotope, and the note under the menu states the scale: one second on screen is 15 s of protactinium time but 1250 years of carbon-14 time. The half-lives are rounded measured values, from 55.6 s for radon-220 to 5700 years for carbon-14. Each decay equation in the panel balances, with the mass numbers and the atomic numbers adding up on both sides.
Choosing an isotope for a job
The Choose an isotope for the job panel sets five jobs, from a smoke alarm to radiotherapy. Pick an isotope for each and the feedback explains, through its radiation and its half-life, why it would or would not do the job.