Visual walkthrough — Applications — radiocarbon dating, medical (Tc-99m, I-131), RTG (Pu-238 in spacecraft)

Step 1 — Start with a pile of atoms and one honest coin

WHAT. Imagine a box holding $N$ radioactive atoms. Not a formula yet — just dots. Each atom has an identical, private chance of decaying in the next tiny slice of time. Nothing about its neighbours, its past, or how long it has already waited changes that chance.

WHY. This is the only physical assumption in the whole story. Radioactive decay is memoryless: an old atom is exactly as likely to decay next second as a fresh one. Once you accept "same chance per atom per second," everything else is arithmetic. We build the maths from this and nothing else.

PICTURE. Look at the box of white dots. The amber dots are the ones that will flip "decay" in this time-step. Notice the amber fraction is the same small slice whether the box is crowded or nearly empty — that constant slice is the seed of the whole law.

Step 2 — Turn "each atom flips a coin" into a rate

WHAT. In a short time $dt$, each atom decays with chance $\lambda dt$. If there are $N$ atoms, the number that decay is (chance per atom) × (number of atoms): $\underset{\text{how many vanish}}{\underbrace{dN}}=-\underset{\text{chance each}}{\underbrace{\lambda }}\underset{\text{how many present}}{\underbrace{N}}\underset{\text{tiny time slice}}{\underbrace{dt}}$

Read it left to right: a tiny change in population $dN$ equals minus $\lambda$ (each atom's chance) times $N$ (how many are available to take that chance) times $dt$ (the length of the moment).

WHY the minus sign. $N$ can only fall — atoms leave, none arrive. So $dN$ is negative. The minus makes the maths agree with reality.

WHY divide by $dt$. We want a rate — decays per unit time — so we can compare boxes fairly regardless of how long we watched. Dividing: $\frac{dN}{dt}=-\lambda N$ Here $\frac{dN}{dt}$ is "how fast the population is dropping," and the equation says that speed is proportional to how many are left. Crowded box → fast drop; nearly empty box → slow drop.

PICTURE. The graph shows population $N$ on the vertical axis, time on the horizontal. At each point an arrow shows the slope $\frac{dN}{dt}$. Where the curve is high the arrow points steeply down; where it is low the arrow is nearly flat. The steepness literally is $\lambda N$.

Step 3 — Sort the equation so we can add up the moments

WHAT. We have $\frac{dN}{dt}=-\lambda N$. This is a differential equation: it tells us the slope everywhere but not the actual curve. To get the curve we must "undo" the slope by summing all the tiny changes. First we separate the variables — put everything about $N$ on one side, everything about $t$ on the other: $\frac{dN}{N}=-\lambda dt$ Left side: the fractional change in population, $\frac{dN}{N}$ ("what slice of the pile just left"). Right side: minus $\lambda$ times the time slice.

WHY this tool — separation of variables. We chose it because the equation is a clean product of "a function of $N$" times "a function of $t$." When that happens, splitting the two sides lets us integrate each independently — the simplest possible attack. No fancier tool is needed, so we don't reach for one.

PICTURE. The figure shows the two sides as two stacks of thin slices. Every time-slice on the right (equal width) corresponds to an equal fractional bite out of the pile on the left. Equal fractions, not equal amounts — that's the fingerprint of exponential decay.

Step 4 — Add up all the slices: the integral gives a logarithm

WHAT. Summing infinitely many tiny slices is integration. We integrate the left side as $N$ runs from its start value $N_0$ (at $t=0$) to its current value $N$, and the right side as time runs from $0$ to $t$: ${\int }_{N_0}^N\frac{dN}{N}=-\lambda {\int }_0^{t}dt$

The left integral is a famous one: $\int \frac{dN}{N}=\ln N$. The natural logarithm $\ln$ is exactly "the running total of fractional changes," which is why it appears here and not, say, a square root. Evaluating both sides between their limits: $\underset{\ln (N/N_0)}{\underbrace{\ln N-\ln N_0}}=-\lambda t$

Using $\ln N-\ln N_0=\ln \frac{N}{N_0}$ (subtracting logs = dividing the insides), we land on: $\ln \frac{N}{N_0}=-\lambda t$ $\frac{N}{N_0}$ is the surviving fraction; its logarithm falls in a perfectly straight line as time passes, with slope $-\lambda$.

WHY $\ln$ and not something else. We didn't pick the logarithm — the integral of $1/N$ forces it. The moment decay became "proportional to what's left," the natural log became inevitable. That is the deep reason radioactive plots go straight when you use a log axis.

PICTURE. Two panels side by side: on a normal axis, $N$ curves down and flattens; on a log axis, $\ln (N/N_0)$ is a dead-straight ramp with slope $-\lambda$. Same data, straightened by the logarithm.

Step 5 — Undo the logarithm to reveal the exponential

WHAT. We have $\ln \frac{N}{N_0}=-\lambda t$. To free $N$ from inside the log we apply the log's inverse, the exponential $e^{(\cdots )}$, to both sides. Because $e^{\ln x}=x$, the log dissolves: $\frac{N}{N_0}=e^{-\lambda t}⟹\boxed{N=N_0{e}^{-\lambda t}}$

Term by term: $N$ = atoms left now; $N_0$ = atoms at the start; $e^{-\lambda t}$ = the surviving fraction, a number sliding from $1$ (at $t=0$, nothing decayed) toward $0$ (at $t=\infty$, all gone).

WHY $e$ and not $2$ or $10$. We are forced into base $e$ because $e$ is the unique base whose exponential is its own rate of change — the only base that plays perfectly with the $\ln$ our integral produced. Any other base would drag an ugly constant along. (Below we'll happily repackage it in base 2 for half-life intuition — but the machinery is born in base $e$.)

PICTURE. The full decay curve. Marked on it: the starting height $N_0$ at $t=0$; the value at one half-life dropping to $N_0/2$; the smooth approach to zero. The tangent at $t=0$ is drawn — its steepness is $\lambda N_0$, tying back to Step 2.

Step 6 — Squeeze out the half-life

WHAT. "Half-life" $t_{1/2}$ is the time for the pile to fall to half. Set $N=\frac{N_0}{2}$ in our result and solve for the time: $\frac{N_0}{2}=N_0{e}^{-\lambda t_{1/2}}\Rightarrow \frac{1}{2}=e^{-\lambda t_{1/2}}$ The $N_0$ cancels — the half-life doesn't depend on how many atoms you start with. Take $\ln$ of both sides ( $\ln \frac{1}{2}=-\ln 2$ ): $-\ln 2=-\lambda t_{1/2}\Rightarrow \boxed{t_{1/2}=\frac{\ln 2}{\lambda }}$ $\ln 2\approx 0.693$, so half-life and decay constant are just reciprocals scaled by $0.693$: a fast atom (big $\lambda$) has a short half-life, and vice versa.

WHY this matters for a clock. A fixed half-life independent of amount means the "melting speed" of our ice cube is a property of the isotope alone — the perfect ruler. C-14 always halves in 5730 years, whether it's a whole tree or a splinter.

PICTURE. The staircase of halvings: $N_0\to N_0/2\to N_0/4\to N_0/8$, each tread exactly $t_{1/2}$ wide. Equal time steps, equal ratios, never equal amounts.

Step 7 — From atoms to a stopwatch: the age formula

WHAT. We can't count atoms one by one, but we can count decays per minute — the activity $A$. Since every present atom carries the same decay chance, $A=\lambda N$, so activity is just a scaled copy of $N$ and obeys the same law: $A=A_0{e}^{-\lambda t}$ Now flip it to solve for $t$ (the whole point — we want time, not surviving fraction). Divide by $A_0$, take $\ln$, and rearrange: $\frac{A}{A_0}=e^{-\lambda t}\Rightarrow \ln \frac{A}{A_0}=-\lambda t\Rightarrow \boxed{t=\frac{1}{\lambda }\ln \frac{A_0}{A}=\frac{t_{1/2}}{\ln 2}\ln \frac{A_0}{A}}$ Here $A_0$ is the activity of fresh, living material (the modern reference) and $A$ is the old sample's activity. The ratio $\frac{A_0}{A}>1$ measures depletion; its logarithm, scaled by $\frac{t_{1/2}}{\ln 2}$, is the elapsed time since death.

WHY invert. In dating we observe $A$ and want $t$ — the unknown moved from the exponent to the answer, so we peel the exponent off with a logarithm, the exact inverse we used in Step 4.

PICTURE. The reading process drawn as a two-step arrow: measured activity $A$ on the curve → project up to find the surviving fraction → project across (via the log) to read off the age on the time axis.

Step 8 — The edge cases (so no scenario ambushes you)

WHAT & WHY. A formula you trust must survive its extremes. Walk the corners:

• $t=0$ (fresh sample): $e^{-\lambda \cdot 0}=e^0=1$, so $N=N_0$. Nothing decayed yet — the clock reads zero. ✔
• $t\to \infty$ (ancient sample): $e^{-\lambda t}\to 0$, so $N\to 0$. The curve approaches zero but never touches it — mathematically atoms linger forever, which is why in practice we hit a measurement floor.
• $A=A_0$ (no depletion): $\ln \frac{A_0}{A}=\ln 1=0$, giving $t=0$. A living sample dates to "just now." ✔ Any tiny counting error here can push the ratio below 1, giving a nonsensical negative age — the sign flips because $\ln (\text{less than }1)<0$. That warns you the sample is essentially modern.
• The practical dead-zone: after ~10 half-lives, $(1/2{)}^{10}\approx 0.001$ — only 0.1% of the C-14 is left, buried in background noise. This is the hard reason C-14 cannot date dinosaurs (65 Myr ≈ 11,000 half-lives — utterly gone). Long clocks like U-238 take over there.
• $\lambda =0$ (a stable atom): $e^0=1$ forever, $N$ never changes, $t_{1/2}=\frac{\ln 2}{0}=\infty$. A stable nucleus is a clock that never ticks — useless for dating, which is why we need a $\lambda$ matched to the age range.

PICTURE. One plot with all four regimes flagged: the $t=0$ start, the asymptotic tail hugging zero, the shaded "noise floor" band past 10 half-lives, and a flat line for $\lambda =0$.

The one-picture summary

Every arrow of the derivation, chained into a single diagram: coin-flip assumption → rate law → separate → integrate (log appears) → invert (exponential appears) → set half → invert for time. Each box names the tool that carried us across, and the amber trail shows the logic flowing from a single physical idea to a working stopwatch.

Flashcards