The Maxwell–Boltzmann Distribution Explained: Why Blowing on Tea Cools It
Pour a cup of tea and it's too hot to drink. Almost without thinking, you blow across the top, wait a moment, and try again. It works, and for most of us that is where the question ends.
A thermometer dipped into the cup would give you a single number, say 80 °C. The cup itself holds something like eight million billion billion water molecules, and not one of them is at 80 °C. Temperature belongs to the crowd; no single molecule carries it. Some are barely drifting, a tenth of them are moving faster than a kilometre per second, and the rest are spread out between, in a shape James Clerk Maxwell worked out in 1860 and Ludwig Boltzmann spent the following decade generalising and grounding.12
That shape is the Maxwell–Boltzmann distribution, and a cup of tea turns out to be a very good place to learn it. Everything the curve says shows up in the cup: why it steams, why it cools, why a puddle still dries on a cold day, and why blowing on it helps at all. The answer to that last one sits at the very edge of the crowd, in the handful of molecules fast enough to leave.
Temperature Belongs to the Crowd
Start with what the thermometer is actually reporting. Every molecule in the tea is moving, jostled by its neighbours, and every time it collides with one its speed changes. In liquid water that happens trillions of times a second, so the speed of any one molecule is a number that never sits still long enough to mean anything. What does sit still, once the cup has settled, is the average.
Temperature is a measure of that average kinetic energy. For the translational motion of a molecule (moving from place to place, as opposed to spinning or vibrating), each of the three directions it can move in gets an equal share, and that share is set by temperature alone:
Adding the three together gives the mean translational kinetic energy of a molecule:
Here
and the average molecule in the cup carries about one and a half times that, roughly 0.046 eV.
That is a small amount of energy, and it is only an average. The thermometer can't tell you how the energy is shared out, whether every molecule has close to 0.046 eV or whether a few have a great deal and most have very little. Two cups at the same temperature could, in principle, hide very different crowds. In practice they don't, because collisions keep shuffling energy around until the crowd settles into one particular shape, the same shape every time for a given temperature.2 Finding that shape is what Maxwell did.
The Shape of the Crowd
Maxwell's argument begins with a single direction of motion. Take the velocity of a molecule along one axis,
This is the Boltzmann factor, and it is the single most important expression in this post. It says that states cost energy, that the cost is measured in units of
which is a Gaussian, the familiar bell curve, centred on zero. The square root at the front is there only to make the total probability come out to one. The same curve applies to
That product tells you how likely each individual velocity is, direction included. Most of the time, though, the question is about speed, how fast a molecule is going regardless of where it's heading. Here the geometry does something the Gaussian alone doesn't hint at. Every velocity with the same speed
This is the Maxwell–Boltzmann speed distribution. The quantity
Its shape comes from two factors pulling in opposite directions. The
Three speeds are commonly used to summarise the curve, and because of the long tail they aren't the same. The most probable speed sits at the top of the hump, found by setting the derivative of
The mean speed is pulled to the right of it by the tail:
and the root-mean-square speed, the one tied directly to the average kinetic energy through
Whatever the gas or the temperature, the three always stand in the same ratio, roughly
Now put the tea in. A water molecule has a mass of 18.015 atomic mass units, about
which is roughly twice the speed of sound in air, for a liquid that is sitting perfectly still on the table. The spread is just as striking. Fewer than one molecule in two hundred is moving slower than 100 m/s. About one in ten is moving faster than 1 km/s, three in a thousand are faster than 1.5 km/s, and around two in every hundred thousand are going faster than 2 km/s. At that point the numbers are getting small, but in a cup holding eight million billion billion molecules, two in a hundred thousand is still around a hundred and sixty billion billion of them.
The fraction faster than any given speed has a closed form, which will be useful later. Writing the speed as a multiple of the most probable one,
where
One question is worth settling here, before it nags. Maxwell derived all of this for a gas, where molecules fly freely between collisions, and the tea is a liquid, where they're packed shoulder to shoulder and pulling on one another constantly. It turns out not to matter for the speeds. In classical physics a molecule's energy splits cleanly into a kinetic part that depends only on its velocity and a potential part that depends only on where it is relative to its neighbours, and the Boltzmann factor for the kinetic part is untouched by the potential part. However strongly the neighbours pull, the velocities in a classical liquid follow the same Maxwell–Boltzmann distribution as a gas at the same temperature.4 The pulling matters enormously for something else, which is what it takes for a molecule to leave.
Turning the Dials
Only two things set the shape of the curve: the temperature, and the mass of the molecule. Everything else drops out. The colour of the mug, the brand of tea and how long it has stewed make no difference at all (the milk does, though only because it changes the temperature).
Temperature first. Every characteristic speed goes as
The area under the curve has to stay at one, so as the hump moves right it also has to flatten and spread. The height of the peak falls as
The tail tells a different story. At 20 °C, about eight molecules in ten thousand are moving faster than 1.5 km/s; at 80 °C it's about thirty-two, four times as many. Go further out, to 2 km/s, and the count rises from under two in a million to about nineteen, more than ten times as many. The further out you look, the larger the change, and the rise in temperature that moved the hump by a tenth has multiplied the population at the far end several times over. That asymmetry, a bulk that barely moves and a tail that swings enormously, is the most important thing to carry into the rest of this post.
Mass works the other way. The speeds go as
The Edge of the Crowd
Speed is the natural way to draw the curve, but energy is the natural way to ask what a molecule can do. Escaping from a liquid, breaking a bond, starting a chemical reaction: each of these has a threshold, and the threshold is an energy. So it pays to rewrite the distribution in terms of kinetic energy,
and the mass has vanished. Measured in energy, every substance at the same temperature shares exactly one curve, whether it's water, nitrogen or helium. The shape comes from the same tug of war as before, with
There's a small trap here. The most probable energy is
What matters for the tea is the area under the tail, the fraction of molecules whose energy exceeds some threshold
which is the same expression as the speed tail from earlier, rewritten in energy. A few values give a feel for how quickly it falls:
| Threshold |
Fraction above threshold |
|---|---|
| 1 | 57% |
| 2 | 26% |
| 3 | 11% |
| 5 | 1.9% |
| 10 | 1.7 in ten thousand |
| 15 | 1.4 in a million |
| 20 | 1.1 in a hundred million |
| 30 | 5.9 in ten trillion |
Once the threshold is more than a few
Each extra
The exponential never reaches zero, either. On a logarithmic axis the tail runs down and away to the right forever, with no energy at which it stops. However much energy you name, some molecule could have it; past a certain point you would simply wait a very long time to see one. And the cup is a big crowd. Even at thirty
So the question for the tea is how far out its threshold sits: how much energy a water molecule needs before it can leave the surface, and how many
The Price of Leaving
A molecule at the surface of the tea is held there. Water molecules cling to one another through hydrogen bonds, each one tugging on its neighbours, and a molecule that wants to leave has to pull free of all of them at once. The energy it needs to do that is something we can measure directly, because it's the heat it takes to turn liquid water into vapour at the same temperature: the latent heat of vaporisation. Steam tables give it as about 2,309 kJ for every kilogram of water at 80 °C, or about 41.6 kJ per mole.5 Divided among the molecules, that's
per molecule, around nine times the 0.046 eV the average molecule in the cup is carrying. Measured in the unit the tail cares about,
and fourteen
This is also where the asymmetry from earlier stops being abstract. The same calculation at 20 °C, where the latent heat is slightly higher and
The fraction is small at 20 °C, though it isn't zero, and that matters more than its size. It's why a puddle dries on a cold day and washing dries on the line in winter, well below anything like boiling. Ice does the same thing without melting first. At 0 °C, the energy to pull a molecule straight out of the solid is around 22
This is the difference between evaporation and boiling, too. Evaporation is the tail leaking out through the surface, and it happens at any temperature. Boiling is what happens when the vapour can form bubbles inside the liquid and push the surroundings back, which needs the liquid hot enough for its vapour pressure to match the pressure of the air above it, 100 °C at sea level. The tea is nowhere near boiling, and it's evaporating anyway.
The Fastest Leave First
The molecules that escape come exclusively from the far end of the tail, and each one carries away around fourteen
It doesn't stop there, because the tail doesn't stay empty. Collisions in the liquid reshuffle energy so quickly that the gap left by each escaping molecule is filled almost at once, and the crowd settles straight back into a Maxwell–Boltzmann shape, at a fractionally lower temperature.2 The refilled tail is then skimmed again, and so on, for as long as the tea sits there. Evaporation is a continuous drain on the hottest part of the crowd, and the crowd's own collisions keep feeding the drain.
The size of the effect is easy to estimate. Each gram of water that evaporates at 80 °C takes about 2,309 J with it. Water needs about 4.19 J to change the temperature of a gram by one degree, so a 250 ml cup gives up about 1,050 J for every degree it cools. One gram evaporated, a little under half a percent of the cup, is enough on its own to cool it by about 2.2 °C. The cup loses heat in other ways as well, by conduction through the mug, by the air warming and rising off it and by radiation, but evaporation is the one with the exponential behind it, which is why it matters most when the tea is hottest and falls away fastest as the tea cools. It's also the same trick the body uses when it sweats.
Put that way, there's a puzzle. If molecules are leaving the tea all the time, at a rate set only by the temperature, then blowing on it shouldn't change anything: the tail is the same size whether or not you blow. And put a lid on the cup and it stays hot noticeably longer, though the tea underneath is just as warm and its tail just as long. The answer lies with the molecules coming back.
Traffic Both Ways
The air just above the tea is full of water vapour, molecules that have already escaped, and they're moving too. They have their own Maxwell–Boltzmann distribution, set by the temperature of the air they're in, and they bounce around at hundreds of metres per second like everything else. Some of them, inevitably, fly back down and hit the surface of the tea, where most are caught and pulled back into the liquid. So the surface is crossed by two streams at once: molecules leaving, at a rate set by the temperature of the tea, and molecules returning, at a rate set by how much vapour is sitting above it.
The returning stream is something Maxwell's distribution lets us count. For a gas with
The more vapour there is above the cup, the higher its pressure and the more molecules come back.
Now seal the space. Put a lid on the cup and the vapour has nowhere to go, so it builds up beneath the lid and the returning stream grows, until as many molecules come back each second as leave. At that point the liquid and its vapour are in equilibrium, and the pressure the vapour has reached is the saturated vapour pressure,
How large are these two streams? Taking the saturated vapour pressure at 80 °C and putting it into the expression above gives roughly
An open cup sits somewhere between the sealed case and the empty one. There's no lid, but the air immediately above the surface is still and thin, and it fills with vapour very quickly, so the layer hugging the tea sits close to saturation. Vapour can only leave that layer by drifting away slowly through the room's air, and the rate it drifts away at is what sets how much the net flow can be. The difference between the two streams is approximately proportional to the gap between the saturated pressure at the tea's surface and the actual pressure of the vapour just above it:
Close that gap and evaporation all but stops, however hot the tea. Open it, and the tail starts draining again.
Blowing on the Tea
This is all blowing does. A breath across the top of the cup sweeps away the layer of saturated air resting on the surface and replaces it with air carrying far less vapour. The pressure of vapour just above the tea drops, the returning stream drops with it, and the gap in the equation above opens up. The outgoing stream stays exactly as it was, since the tea's tail has no idea you're blowing, and the only thing to change is how many of the escaped molecules make it back. Every molecule that doesn't is another piece of the high-energy tail permanently taken out of the cup.
It's worth noticing how forgiving this is. Your breath is warm and damp, close to saturated at a little under body temperature, which still means a vapour pressure of only around 6 kPa against the 47 kPa an 80 °C surface would build up beneath it. Even humid breath is very dry air from the tea's point of view. Blowing also helps in the ordinary way, the moving air carrying heat off the surface directly, much as a breeze cools your face on a warm day, and both effects come from the same act of moving still air away from the surface.
This is the equilibrium blowing disturbs. Left to itself, the tea builds a thin blanket of its own vapour that very nearly balances its losses, and the blanket is what keeps the cooling slow. Every breath knocks that balance over, and the surface spends the next moment trying to rebuild it, losing its fastest molecules for good as it does. Keep blowing and the blanket never gets the chance to settle.
The other habits people have with hot drinks work through the same two streams. Pouring tea into a saucer, as people once commonly did, spreads it thin and multiplies the surface the tail can drain through. A wide mug cools faster than a narrow one for the same reason. Stirring helps as well, because evaporation cools the surface first, leaving a slightly cooler skin on top of hotter liquid, and stirring brings the hot liquid up to where the tail can be skimmed. Even the steam you can see rising from the cup belongs to this picture. Water vapour is invisible; the white wisps are vapour that has met the cooler room air and condensed back into tiny droplets, molecules that left the tea and have already started coming back together.
Where the Model Bends
Everything so far has rested on a few simplifications, and a reference should say where they are.
The first is classical physics itself. The Maxwell–Boltzmann distribution assumes molecules behave like tiny billiard balls with well-defined positions and speeds, and that holds as long as a molecule's quantum wavelength is much smaller than the distance between neighbours. For a water molecule at 80 °C, that wavelength is about 0.02 nm, against a spacing between molecules in the liquid of about 0.3 nm, so the tea is comfortably classical. The approximation fails for very light particles or very low temperatures. Electrons in a metal follow a different distribution entirely (Fermi–Dirac), as does light in a hot oven (Bose–Einstein), and helium cooled to within a couple of degrees of absolute zero stops behaving classically altogether.
The second is that a water molecule does more than move from place to place. It also rotates, and its atoms vibrate against one another, and both store energy. The rotations take their own equal share of
The third, and the one that matters most for the numbers, is using the latent heat as the energy a molecule needs to escape. Latent heat is measured for a whole liquid turning into vapour, and part of it, about
with
The last is equilibrium. The distribution describes a crowd that has settled, and a cup being blown on is being kept deliberately unsettled. It works anyway because the settling happens far faster than anything we can do to the cup. Collisions in the liquid restore the Maxwell–Boltzmann shape in around a trillionth of a second, so at every moment the tea is, to an excellent approximation, in equilibrium at a temperature that is slowly falling. The disequilibrium that matters is at the surface, between the liquid and the air, and that's exactly where blowing acts.
The Cup Again
The thermometer in the cup still reads one number, and it still describes nobody in particular. The crowd beneath it is mostly in no hurry. At 80 °C, the typical molecule in the cup is moving faster than a bullet leaves a pistol, getting nowhere, battered back and forth trillions of times a second by its neighbours. The cooling is happening at the edge, among a few in every million, molecules with enough energy to pull free of the rest and leave, and the overwhelming majority of those are being caught and brought back by the vapour they left behind.
All blowing does is move the air, so that fewer of them find their way home.
Footnotes
-
J. C. Maxwell, "Illustrations of the Dynamical Theory of Gases. Part I. On the Motions and Collisions of Perfectly Elastic Spheres", The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, Series 4, 19(124) (1860), pp. 19–32. Taylor & Francis ↑
-
L. Boltzmann, "Weitere Studien über das Wärmegleichgewicht unter Gasmolekülen" (Further Studies on the Thermal Equilibrium of Gas Molecules), Sitzungsberichte der Kaiserlichen Akademie der Wissenschaften in Wien, 66 (1872), pp. 275–370. This is the paper showing that collisions drive any starting distribution towards Maxwell's. Reprinted in his Wissenschaftliche Abhandlungen: Cambridge Core ↑ ↑2 ↑3
-
The 2019 revision of the SI fixed
at exactly , so the kelvin is now defined through it. Bureau International des Poids et Mesures, The International System of Units (SI), 9th edition (2019), bipm.org ↑ -
This holds as long as the motion can be treated classically, which is an excellent approximation for the translational motion of water molecules at everyday temperatures. The potential energy shapes where molecules sit; it has no say over how fast they move. Any standard statistical mechanics text covers it, e.g. D. Chandler, Introduction to Modern Statistical Mechanics (Oxford University Press, 1987). ↑
-
Latent heat of vaporisation of water at saturation: about 2,453.5 kJ/kg at 20 °C, 2,308.8 kJ/kg at 80 °C and 2,256.4 kJ/kg at 100 °C; the enthalpy of sublimation of ice at 0 °C is about 51 kJ/mol. W. Wagner & A. Kruse, Properties of Water and Steam: The Industrial Standard IAPWS-IF97 for the Thermodynamic Properties and Supplementary Equations for Other Properties (Springer, 1998); International Association for the Properties of Water and Steam, Revised Release on the Equation of State 2006 for H₂O Ice Ih, IAPWS R10-06(2009), iapws.org ↑
-
Saturated vapour pressure of water: about 2.339 kPa at 20 °C, 12.35 kPa at 50 °C, 47.41 kPa at 80 °C and 101.42 kPa at 100 °C (Wagner & Kruse, 1998, as above). ↑
-
The expression for the flux of molecules striking a surface goes back to H. Hertz, "Ueber die Verdunstung der Flüssigkeiten, insbesondere des Quecksilbers, im luftleeren Raume", Annalen der Physik 253(10) (1882), pp. 177–193 (NASA ADS), and M. Knudsen, "Die maximale Verdampfungsgeschwindigkeit des Quecksilbers", Annalen der Physik 352(13) (1915), pp. 697–708 (Wiley); the net-flux form is usually called the Hertz–Knudsen equation. In practice it is multiplied by an evaporation coefficient between 0 and 1, the fraction of molecules striking the surface that actually stick, and its measured value for water is still argued about: see R. Marek & J. Straub, "Analysis of the evaporation coefficient and the condensation coefficient of water", International Journal of Heat and Mass Transfer 44(1) (2001), pp. 39–53, and A. H. Persad & C. A. Ward, "Expressions for the Evaporation and Condensation Coefficients in the Hertz–Knudsen Relation", Chemical Reviews 116(14) (2016), pp. 7727–7767 (PubMed). A coefficient below one shrinks both streams by the same factor, so the outgoing flow stays enormous next to the net loss. ↑