A radioactive atom does not know how old it is
Nothing accumulates inside an atom while it waits. That is why the half-life never changes.
Half of a carbon-14 sample is gone after 5,730 years. Half of what remains goes in the next 5,730, and half of that in the next.
The same fraction, at any age
Take an atom that has already sat there for a million years without decaying. Its chance of surviving the next 5,730 is 0.500000. Take one created this morning: 0.500000. Take one that has waited a single half-life: 0.500000.
Nothing has changed inside the old atom. It has no wear, no accumulated stress, no internal clock counting down. It is not closer to decaying than it was at the start.
This property is called memorylessness, and the exponential is the only continuous distribution that has it. If you require that survival for another interval be independent of age, exponential decay is not one possible answer but the answer.
Why a half-life exists at all
Most quantities do not work like this. Half of a population of people is gone by about age 80, but half of the survivors are not gone 80 years later, because people accumulate age and atoms do not.
So for people the phrase "half-life" would have to be recalculated at every age. For carbon-14 the same 5,730 works forever, and that constancy is the whole reason radiometric dating is possible. The ratio in a sample encodes elapsed time and nothing else.
Set Radioactive Decay to carbon-14's 5,730 years and step through half-lives. Five leaves 3.125% and ten leaves 0.098%, which is why carbon dating runs out around 50,000 years: not because the law changes, but because what remains stops being distinguishable from contamination.
The average life is not the half-life
The mean lifetime, written τ and spoken "tau", is τ = t½ / ln 2, so for carbon-14 it is 8,267 years, not 5,730.
The two differ because the distribution is skewed. Most atoms decay relatively early and a thin tail waits an enormously long time, dragging the average above the median. The median is the half-life by definition, since half is what median means.
And τ has a second meaning that follows from memorylessness: it is the expected remaining lifetime of any atom, at any age. An atom that has waited a million years still expects 8,267 more.
Where else this shows up
Exponential Growth & Decay is the same equation with the sign flipped, and the doubling time it prints plays the role the half-life does here. Anything whose rate of change is proportional to its current amount lands on this curve.
The memoryless property is also the assumption hiding inside the queue model in why a 99% busy system breaks. M/M/1 assumes exponential service times, which says a job that has already run for an hour is no closer to finishing than a fresh one.
For atoms that is a physical fact. For service times it is a modelling convenience that is frequently false, and knowing which of the two you are holding is the difference between a derivation and a guess.
Why nothing you do changes it
Heat the sample, freeze it, dissolve it, put it under pressure: the half-life does not move. This is not robustness, it is a mismatch of scale.
Chemistry rearranges outer electrons, which involves energies of a few electronvolts. Nuclear decay involves millions. Chemical bonds are the wrong size of lever by roughly a factor of a million, so they cannot reach the process that matters.
There is one honest exception. Isotopes that decay by capturing one of their own electrons, such as beryllium-7, depend slightly on the electron density at the nucleus, and that is chemistry's business. Measured shifts are a fraction of a percent, which is enough to confirm the mechanism and nowhere near enough to trouble a dating method.
The law that arrived before the mechanism
Rutherford and Soddy established the exponential law in 1902, working on thorium X, whose activity they watched halve in about four days. The name came later: Rutherford's half-life period, from 1907, shortened to half-life only in the 1950s.
What they had was a law with nothing underneath it. Every other rate in the physics of the day came from a cause: something got hotter, something was pushed, something wore out. Here was a process with a precise timetable for a population and no answer whatsoever for the individual, and no way to tell which atom would go next or why.
That unease was justified, and the mechanism took another generation. Gamow explained alpha decay as quantum tunnelling in 1928; carbon-14 is a beta emitter, and its route needed Fermi's theory of the weak interaction in 1934. Both answers have the same shape: a fixed probability per unit time, independent of how long the nucleus has already waited. The memorylessness that looked like a missing explanation turned out to be the explanation: there is no internal state to remember, so there is nothing to accumulate and nothing to wear out.
Watch
References (2)
- the exponential law, and the half-value period Rutherford & Soddy (1902). The cause and nature of radioactivity, Part I. Philosophical Magazine 4(21).
- alpha decay as tunnelling Gamow (1928). Zur Quantentheorie des Atomkernes. Zeitschrift für Physik 51.