The freezer door that will not open twice
Assumes: Pressure is a rate of arrival, and the gas law falls out of counting · The speeds in a still room
Open a chest freezer, take something out, close the lid, then realise something else is needed and lift the lid again. For the first several seconds it does not want to come. Upright freezers do the same, sometimes so firmly that the whole cabinet rocks forward on its feet before the seal breaks. Wait a minute and the door opens as easily as it did the first time.
It feels like a latch or like ice, and it is neither. It is the ideal gas law, applied to the air the first opening let in. That air was at room temperature when the door shut, and a few tens of seconds later it is at freezer temperature. The same amount of gas, in the same volume, at a lower temperature, is at a lower pressure, and a door is a large area. A deficit of one per cent of an atmosphere across a freezer door is a tonne.
What cooling does to a gas that cannot get more
Pressure is a rate of arrival derived the gas law by counting: pressure is the rate at which molecules deliver momentum to a wall, and that rate is proportional to how many molecules there are in each cubic metre and to the mean of the square of their speeds. The mean square speed is proportional to the absolute temperature. Cool a gas in a closed box and the number of molecules stays fixed, but each one arrives less often and strikes more gently, and the pressure falls in proportion to the temperature.
The two curves are the speed distributions the speeds in a still room drew for ordinary air at room temperature and at the temperature of a domestic freezer. The difference between them looks small, and on the scale of molecular speeds it is: the typical molecule slows by seven per cent. But pressure goes as the square of the speed, and kelvin to is a fall of thirteen and a half per cent. A freezer full of room air, sealed and cooled, would end up thirteen and a half per cent of an atmosphere below the room, 13.7 kilopascals.
Across a door of nine-tenths of a square metre that deficit is 12 kilonewtons, the weight of a car, and it is not something anyone opens. Opening a door does not usually exchange all the air: the cold, dense air inside slumps out of the bottom of an upright door as warm air flows in at the top, and how much is exchanged depends on how long the door was open and how wide. A quick opening might replace a third. The deficit is then smaller in proportion, but even a third of the full exchange gives four and a half kilopascals, and four kilonewtons on the door.
Doors do not stay that hard to open, so something else must be going on. The sealed calculation is the right starting point and the wrong answer, because a freezer is not sealed.
Two clocks, one running against the other
Two things happen after the door shuts, and each has its own timescale.
The warm air cools. Air has very little heat capacity next to the walls, the shelves and the frozen food. A freezer of three hundred litres holds about 0.36 kilograms of air, and warming it by forty degrees takes fourteen kilojoules; twenty kilograms of frozen food would absorb that and warm by less than half a degree. So the walls and contents act as a reservoir at fixed temperature, and the air cools towards them at a rate set by how fast heat crosses from air to surface. Still air against a cold surface transfers heat at roughly five to ten watts per square metre per kelvin, and a freezer’s interior offers a few square metres of wall, shelf and packaging, so the time constant, the air’s heat capacity over that conductance, comes out at
from ten seconds to a few tens, depending on how full the freezer is and whether a fan is stirring the air. Thirty seconds is used in the figures. As the air cools, its pressure falls.
The gas law used here is the ideal one, and it is accurate. The first correction to the gas law measured how far real air departs from it, and at an atmosphere and −18 °C the departure is about a tenth of a per cent, a correction to the correction. The heat capacity that matters is the one at constant volume, since the door holds the volume fixed, which the ball that weighs how a gas stores heat measured for air by timing a ball bouncing on a trapped column of it; using the heat capacity at constant pressure instead would lengthen by two fifths and change the sealed deficit not at all.
Room air leaks back in. The gasket is a soft magnetic strip pressed against the cabinet, and it is not airtight; air seeps through the tiny gaps where the strip does not quite meet the frame, and through the drain and the cable passages. Whatever the pressure difference, the leak works to reduce it, and it does so on a timescale of its own: the time it would take to equalise a pressure difference if nothing else were changing.
The figure integrates the gas law with both processes running. Writing for the pressure inside, for the temperature of the air, for the room’s pressure and for the leak’s time constant,
The first term is cooling at fixed amount of gas; the second is the leak. With no leak, the deficit climbs as the air cools and stays at the sealed value. With a leak, the deficit rises only while the air is cooling faster than the leak can refill it. Early on, the air is warmest and cooling fastest and the deficit is smallest, so the cooling wins and the deficit grows. Later, the cooling slows as the air approaches the freezer’s temperature, while the leak, driven by an ever larger deficit, speeds up. They cross at the peak, and after it the leak wins and the pressure returns to the room’s.
The result is a pulse, not a step. With a gasket that would equalise a pressure difference in three seconds, the deficit peaks at about 350 pascals eight seconds after the door was shut and has fallen to a few tens of pascals within a minute. A tighter gasket, taking ten seconds, gives a peak nearly three times higher that lasts longer. A leakier one, one second, barely registers.
The ratio that decides it
The shape of the pulse depends on the two time constants, and the height on only their ratio. For small deficits the equation can be linearised and solved exactly. Writing for the sealed deficit, for the cooling time and , the deficit is
the difference of two exponentials, and its peak is
When the seal equalises much faster than the air cools, is small, the peak is about times the sealed deficit, and the door hardly sticks. When the times are equal the peak is of the sealed deficit. When the seal is much slower than the cooling, nearly the whole sealed deficit builds before the leak can do anything, and the door would be as hard to open as the sealed calculation says, until the leak slowly lets it go.
The same two-exponential shape turns up wherever one process fills a reservoir and a slower or faster one drains it. It is the shape of a drug’s concentration in the blood after a tablet is swallowed, absorption racing elimination, and of the daughter nucleus in a decay chain, which the chain that runs at its slowest member’s rate followed through radioactive families. The freezer is a decay chain in which the parent is the excess heat of the air and the daughter is the pressure deficit, and the deficit can only build if the heat leaves faster than the deficit does.
None of this is particular to freezers among cold machines. A fridge with no work going into it described an absorption refrigerator, the kind found in caravans and hotel minibars, and its door sticks in exactly the same way: the mechanism that makes the box cold has no part in the effect, which needs only a cold box, a door, and a leak.
This is the reason good seals make the effect worse. An energy-efficient freezer has a gasket designed to let as little warm, moist air in as possible, and that means a long , a large and a large peak. Many freezers now carry a small pressure-equalising valve, a deliberate leak that opens only under a pressure difference, to set small during the seconds after the door closes without letting air in the rest of the time. Laboratory freezers at −80 °C, where the sealed deficit after a full exchange is a third of an atmosphere, have such valves as a matter of course, and some have a heated port that must be opened by hand before the door.
What the handle has to supply
A pressure deficit pushes uniformly on the whole door. The door is hinged on one side, so the resultant force acts at the door’s centre, halfway between the hinge and the handle, and the hinge takes half of it.
Half of 350 pascals times 0.9 square metres is 160 newtons at the handle, about the weight of sixteen kilograms, on top of the forty or so newtons the magnetic gasket always needs. That is a hard pull, enough to tip an empty upright freezer forward on its feet before the seal breaks, and it matches the experience. The fridge, cooling its air through eighteen degrees rather than forty, peaks at less than half that. The chest freezer’s lid, with half the area, needs about half the force of the upright door at the same pressure, and gravity holds the cold air in when it is opened, so less air is exchanged in the first place, which is one reason chest freezers stick less.
The pulse also explains why the sticking is worst for a second opening shortly after the first and not for the first opening of the day: the door has to have been open recently, long enough to let warm air in, and shut for long enough for the air to start cooling but not for the leak to catch up. Opened again after a minute, the freezer is back at room pressure and opens normally. Opened again immediately, before any cooling, it is also at room pressure. The trap is the interval of a few seconds to a minute after closing.
The moisture that freezes out
Room air carries water vapour, and the freezer removes it. At 22 °C and half saturation, room air has a water-vapour partial pressure of about 1.3 kilopascals. Over ice at −18 °C the vapour pressure is a tenth of a kilopascal. As the warm air cools, nearly all its water vapour freezes onto the coldest surfaces as frost, and those molecules are no longer part of the gas pushing on the door. This removes gas, not just speed, and it adds to the deficit: for a third of the air exchanged, about 0.4 kilopascals, roughly a tenth of the thermal deficit. In a humid kitchen it is more. The frost that builds up on a freezer’s walls over months of openings is the record of exactly this contribution. The vapour freezes out quickly: the puddle that could dry in a second found that molecules cross between a surface and its vapour at kilograms per square metre per second when nothing slows them, and what limits frost formation here is how fast the vapour diffuses through the air to the cold surfaces, on the same timescale as the cooling. The two contributions to the deficit therefore rise together.
This is the one place the freezer departs from a sealed jar, and a sealed jar is the cleanest version of the effect. When jam is poured hot into a jar and the lid screwed on, the air and steam in the headspace are at near boiling; as they cool, the air’s pressure falls by the ratio of the temperatures and the steam condenses almost completely, and the lid is pulled down hard enough to click and to stay sealed for years. No leak competes, so the full sealed deficit, most of an atmosphere for a headspace that was mostly steam, holds the lid on. The freezer is the same physics with a leaky lid.
The molecules that come back in
The gas leaking back through the gasket is room air, and it does not arrive as an average sample of the room. The gas that leaves is not the gas inside found that gas escaping through a hole small compared with the distance its molecules travel between collisions favours fast molecules, because they hit the hole more often. A gasket’s gaps are wider than that, typically tens of micrometres against the free path how far a molecule gets found for air, sixty-eight nanometres, so the inflow is a viscous flow, driven by the pressure difference and proportional to it, which is why the leak can be described by a single time constant. Only at very narrow gaps, or at much lower pressures, would the leak rate stop being proportional to the deficit and start depending on the molecules’ speeds.
The flow through the gasket’s gaps also explains why the leak time constant depends so strongly on the gasket’s condition. Viscous flow through a slot goes as the cube of the slot’s width, so a gasket whose gaps are half as wide leaks eight times more slowly. A new gasket, or one that has been cleaned and has settled snugly against the frame, can multiply several times and turn a door that never sticks into one that does.
What a single cooling time leaves out
The calculation treats the air inside as one parcel with one temperature, cooling exponentially. Real air in a freezer is stratified, colder at the bottom and warmer at the top, and the warm air let in at the top cools as it sinks past the shelves; the cooling is better described by a spread of time constants than by one. The leak is treated as linear in the pressure difference, which is right for a viscous leak and approximately right for a gasket that does not deform; a soft gasket pulled against the frame by the deficit may tighten as the deficit grows, which steepens the peak. The door is treated as rigid and the cabinet as unmoving. And the amount of air exchanged, which sets the whole scale, is the least predictable number of all, varying several-fold with how long the door was open, how full the freezer is and whether anyone stood in the doorway.
What does not depend on any of those details is the shape: a pulse whose height, as a share of what a sealed freezer would reach, is set by the ratio of two times and nothing else.
Still open: how much air an opening exchanges
The volume of air that an opened door exchanges has been measured with tracer gases and modelled as a gravity current, the cold air spilling out of the bottom of the opening while warm air enters at the top. The flow through a vertical opening between two air masses at different temperatures is reasonably understood when the opening is held steady. A real opening is not: the door swings, which pumps air, someone reaches in, and the contents of the freezer change how the cold air drains. How much of a freezer’s air is replaced by an ordinary opening, and how much energy that costs over a year, are estimated with uncertainties of a factor of two. That number, more than any property of the gasket, decides how hard the door sticks on any particular occasion.
The habit worth carrying away is to ask which of two clocks is faster. A freezer that takes in a third of its volume of room air would, if sealed, fall 4.6 kPa below the room as the air cooled, since a fixed amount of gas loses pressure in proportion to its absolute temperature; a gasket that equalises in 3 s against air that cools in 30 s turns that into a pulse peaking near 350 Pa after eight seconds — 160 N at the handle — because the peak is the sealed deficit times with the ratio of the two times. A door that sticks is a thermometer that a leak is in the middle of resetting.
Part 13 of 13
This essay is one argument about Kinetic theory. The others:
The objects named here
The third axis, after the field and the reading path: the things themselves, and every essay that touches each one.
The ideal gas lawLeak rateMaxwell–Boltzmann distributionPartial pressurePressureTime constant
- The balloon a cold morning lifts the ideal gas law, time constant