Coffee Cooling: Newton's Law and the Physics of the Perfect Sip

Newton's law of cooling

Isaac Newton's simplest model of heat loss says the rate at which an object cools is proportional to the temperature difference between it and its surroundings — the bigger the gap, the faster the heat leaves, and the closer the object gets to ambient temperature, the slower it cools. Written as a differential equation and solved, it gives a clean exponential decay:

dT/dt = -k * (T - T_ambient)
T(t) = T_ambient + (T0 - T_ambient) * exp(-k * t)   // k = cooling constant, depends on cup + environment

That single constant k quietly bundles together every physical detail of the situation: the surface area of the coffee exposed to air, the convective heat-transfer coefficient of the surrounding air currents, the thermal conductivity and thickness of the cup wall, and even a bit of radiative loss. Everything the simulation lets you change — the cup material, whether there's a lid — is really just a way of dialling k up or down.

Why the cup material changes the curve so much

Heat leaves hot coffee two different ways: through the liquid surface into the air (evaporation and convection) and through the cup wall by conduction, then convection off the outside of the cup. A ceramic mug is a mediocre thermal insulator (thermal conductivity of typical ceramic ≈ 1–1.5 W/m·K), so it loses a meaningful amount of heat through its walls in addition to the open top. A metal cup (thermal conductivity of stainless steel ≈ 15 W/m·K, aluminium ≈ 200 W/m·K) conducts heat through its walls far more readily, which sounds like faster cooling, but a thin metal cup also has very little thermal mass of its own, so it heats up almost instantly to match the coffee and then radiates and convects that heat away quickly through its now-hot outer surface. A double-walled vacuum thermos essentially removes the conductive pathway altogether — the vacuum gap between the walls has almost no medium to conduct or convect heat through — leaving only radiative loss and whatever heat escapes through the small opening at the top, which is why a thermos can keep coffee hot for hours where a ceramic mug cannot.

The lid matters more than people expect

A large fraction of a hot drink's heat loss is not from the liquid surface conducting to the air directly, but from evaporative cooling: water molecules escaping as vapour carry away a disproportionate amount of energy per molecule (the latent heat of vaporisation), and evaporation also drives convective air currents right at the surface that continually sweep away the warmed boundary layer of air, refreshing it with cooler air. Simply covering the cup with a lid cuts off most of the surface exposed to open air, suppressing both the evaporation and the convective air exchange, and can noticeably slow the exponential decay — often more than switching cup material alone.

Finding the best moment to drink it

Coffee tastes best in a fairly narrow temperature window — often cited as somewhere around 55–65°C, hot enough to carry aroma compounds but below the point where heat itself dulls the tongue's sensitivity to flavour and, at the very top of the range, risks mild burns. Because the temperature curve is a decaying exponential, the coffee spends comparatively little time in any particular narrow window near the start (temperature is changing fastest right after pouring) and progressively more time per degree as it approaches room temperature. Reading the k value off a particular cup and lid combination — which the simulation effectively lets you do by watching the curve — tells you almost exactly how long to wait after pouring before the coffee sits in the ideal range, and for how long it stays there before drifting past it.

A caveat: cooling isn't always purely exponential

Newton's law assumes the whole object is at a uniform, single temperature and that the ambient temperature stays fixed — reasonable for a small, well-stirred cup of coffee in a large room, but the model breaks down for very large heat differences, where radiative loss (which scales with the fourth power of absolute temperature, per the Stefan-Boltzmann law) contributes more than the pure linear-in-ΔT assumption predicts, and this is a meaningfully faster early cooling rate than pure Newtonian cooling would suggest, right after pouring boiling water.

Frequently asked questions

Try it live

Everything above runs in your browser — open Coffee Cooling and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.