The problem: real time is too slow
A new product might need a two- or three-year shelf life, but the company cannot store it for three years before launch. The solution is accelerated stability testing: store samples hotter than normal, watch them degrade faster, and then mathematically translate that fast decay back to the rate at room temperature. Heat is the accelerator because nearly every reaction speeds up as temperature rises.
Internationally, the conditions are standardised by the ICH stability guidelines: a long-term shelf at 25 °C/60% relative humidity and an accelerated shelf at 40 °C/75% RH are the usual pair, with an intermediate condition if needed. Using the same conditions worldwide means a stability dataset travels across regulators.
Arrhenius: heat and rate, made precise
The link between temperature and the rate constant is the Arrhenius equation. In its useful form, plotting the logarithm of k against 1/T (temperature in kelvin) gives a straight line whose slope reveals the activation energy — the energy hill the reaction must climb. Measure k at two or three high temperatures, draw the line, and extrapolate it down to 25 °C to get the room-temperature k you could never afford to measure directly.
A friendlier rule of thumb is the Q10 factor: roughly how many times faster a reaction runs for every 10 °C rise. For many drug-degradation reactions Q10 is about 2–3 — a useful mental shortcut, though never a substitute for an actual Arrhenius fit when a real expiry date is at stake.
Q10 accelerated estimate (illustrative)
Accelerated study at 40 C gives:
t90 (40 C) = 6 months
We want t90 at the storage temperature 25 C.
Temperature drop = 40 - 25 = 15 C => 1.5 steps of 10 C
Assume Q10 = 3 (reaction 3x slower per 10 C cooling):
slowdown factor = Q10^(15/10) = 3^1.5 = 5.20
t90 (25 C) = 6 months x 5.20
t90 (25 C) = 31 months (~2.6 years)
=> Predicted shelf life ~ 30 months, to be CONFIRMED
by real-time 25 C data before it is final.Where the shortcut breaks
Arrhenius extrapolation assumes the same reaction runs at every temperature. Heat can break that assumption. If a solid melts, an API changes crystal form, a protein unfolds, or moisture behaves differently when hot, the high-temperature data no longer predict room temperature — and the straight line lies. This is why accelerated data are treated as a strong forecast, not a verdict.