Why Pastry Blisters in a Microwave

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Because the pastry is making about four and a half times its own volume in steam every second, inside a skin that a microwave is physically unable to dry out. An oven makes the same steam twenty times more slowly and opens a door for it on the way. The microwave makes it fast and keeps the door shut.

That is the whole mechanism, and the useful part is that the numbers are not close. The pressure needed to inflate a blister in hot dough is a few thousand pascals. The pressure available is that much per second with ninety nine percent of the steam escaping freely. So blistering is not bad luck or bad pastry. It is the default outcome of the arithmetic.

Where the heat lands, and why the skin stays wet

An oven heats from the outside. The surface water boils off first, the surface then climbs well above 100 degrees, and it dries into a stiff, porous crust. That crust is the important part: it is permeable, so the steam still forming in the middle has a route out. The structure sets before the pressure builds.

A microwave does something else entirely. The field couples to water, so energy appears wherever the water is, throughout the interior at once. Two consequences follow from that.

The first is that the surface does not get ahead of the inside. A wet surface losing water is pinned at 100 degrees by evaporation, exactly as a wet thermometer bulb is. Our own page on why steam cannot get hotter than 100 degrees sets out the saturation argument properly and it applies here in reverse: while liquid water is leaving a surface, that surface cannot exceed the boiling point.

The second is more interesting, and it is the reason the first one never resolves. A microwave cannot heat a dry layer, because there is no water in it to absorb the field. So the usual escape from the pinned surface, which in an oven is that the surface dries and then gets hotter, does not happen. The skin stays wet, stays soft, stays impermeable, and stays exactly as elastic as a balloon.

The volume arithmetic, which is the number to remember

Take 100 grams of pastry in a TOSHIBA countertop microwave oven, which publishes 1000 watts of output and a 1.2 cubic foot cavity, and run it for 30 seconds. That is 30 kilojoules.

Heating the pastry itself from room temperature to boiling takes 100 grams times about 2.8 joules per gram kelvin times 80 degrees, which is 22.4 kilojoules. That leaves 7.6 kilojoules, and at 2,257 joules per gram of latent heat it makes 3.37 grams of steam.

Three and a third grams sounds trivial, and this is where the surprise is. Steam at 100 degrees and atmospheric pressure occupies 1.673 litres per gram. So 3.37 grams is 5.64 litres of gas. The pastry itself, at a density of around 0.6, is 167 cubic centimetres, or 0.167 litres. The steam is 34 times the entire volume of the object that produced it.

Per second is the figure that actually matters, because blistering is a rate problem. Once the pastry is at boiling, every watt makes 0.000443 grams of steam a second, so 1000 watts makes 0.741 litres a second. Divided by the pastry's 0.167 litres, that is 4.44 of its own volumes every second, continuously, needing to get out through an undried skin.

What an oven does instead

Put the same pastry in a 180 degree oven and work out the heat arriving.

Call the surface area 0.0184 square metres, which is a block roughly 10 by 6 by 2 centimetres. Convection at a representative coefficient of 20 watts per square metre kelvin across an 80 degree difference delivers 29.4 watts. Radiation from the cavity, at an emissivity of 0.9, adds another 21.4. Total, about 51 watts.

Fifty one watts is 0.0226 grams of steam a second, which is 0.0378 litres, which is 0.23 of the pastry's volume a second. Against 4.44 in the microwave. A factor of twenty in the generation rate, and the oven is simultaneously building the permeable crust that lets it out while the microwave is maintaining the skin that does not.

Those two differences compound, and between them they are the entire answer. The same dough, the same water, the same latent heat. Two completely different outcomes because one appliance delivers energy to the surface and the other delivers it to the middle.

Why the blister always wins

It is worth checking whether the available pressure is actually enough, because an argument about rates could still come out marginal. It does not.

A dome of radius R in a film of thickness t holds internal pressure as membrane tension, and it inflates once the pressure exceeds twice t times the film's yield stress divided by R. Hot, hydrated dough is very soft, with a yield stress somewhere around 10 kilopascals, and a pastry skin is perhaps 0.2 millimetres thick. Put those in and a one millimetre blister needs 4,000 pascals, a five millimetre blister needs 800, and a ten millimetre blister needs 400.

Notice the shape of that: small blisters need the most pressure, so once one starts it grows at a falling threshold. Which is why blisters appear suddenly and then balloon rather than swelling gently.

Now the supply. Suppose 99 percent of the steam escapes freely, which is generous. The remaining one percent of 0.741 litres a second accumulating in a 0.167 litre object is a 4.4 percent increase in volume every second, which at constant volume is a pressure rise of roughly 4.4 kilopascals a second. So the hardest threshold on the list, the one millimetre blister at 4,000 pascals, is crossed in about one second with ninety nine percent of the steam still getting out.

That is the sense in which this is not an accident. The pressure available exceeds the pressure required by orders of magnitude. A microwave does not sometimes blister pastry. It blisters pastry unless something specific stops it.

Turning the power down does not do what you think

The obvious response is to use a lower power level, and on most microwaves that does almost nothing for this problem. Our own page on why a microwave trips the breaker on defrost establishes the reason and it is worth crediting rather than re-deriving: a conventional oven at a reduced setting does not reduce its power, it switches the magnetron on and off at full power. Fifty percent is bursts of the full wattage with gaps between them.

Average power falls. Peak power does not. And blistering is driven by the peak, because the pressure threshold is crossed in about a second and the gaps between bursts are longer than that. So a 1000 watt oven on half power still delivers 4.44 pastry volumes a second while it is on, and merely does it in instalments.

A lower rated oven genuinely does reduce the peak. A Nostalgia retro microwave oven publishes 700 watts, which works out at 3.11 pastry volumes a second rather than 4.44, and its own drawbacks are honest that "700 watts is lower than many standard microwaves" and that it "May take longer to heat food compared to higher wattage models". Both true, and for this particular failure the lower wattage is the better machine, which is the one place in this argument where a slower appliance wins outright.

The appliance that can actually solve it is the one that can hold a reduced power continuously. A Panasonic NN-SN65KW microwave oven publishes 1200 watts with inverter technology and 11 power levels, and an inverter is the one design in this category that delivers a genuine fraction of its rating rather than a duty cycle. Run it at 30 percent and you get 360 watts continuously, which is 1.60 pastry volumes a second against 5.33 at full: a reduction of exactly the power ratio, 3.33 times, which a cycling oven cannot achieve at any setting. That single property earns the product its place on this page, and it is worth noticing that its record sells the inverter on defrosting speed rather than on the thing it is uniquely able to do.

The figure nobody in the category publishes

Every microwave on this site publishes an output wattage and a count of power levels. A TOSHIBA 0.9 cubic foot microwave oven publishes 900 watts and "Eleven power levels [that] allow precise control, from gentle defrosting to rapid [cooking]". Not one of them publishes what a power level actually does, which is the duty cycle period: how long the magnetron is on and how long it is off.

That one number decides everything above. A level that cycles in two second bursts and a level that cycles in twenty second bursts have the same average power and completely different effects on anything that inflates, scorches or splits. Eleven levels described as precise control, with no statement of whether the control is of power or of timing, is the single most consequential gap in these records.

The nearest thing to an honest disclosure is on the input side. The Toshiba 1.2 cubic foot model's drawbacks publish "Input power is high at 1600 watts", which is the only input figure in the category and which our breaker page rightly made the most of. The 0.9 cubic foot model's drawbacks say "Input power is higher than output, indicating energy use", which is true and contains no number. And the Panasonic, which publishes a 13.4 inch turntable, concedes in its own drawbacks that its "External dimensions [are] not [stated]", so the cavity is described by what will fit in it.

What to do about it, and the appliance that does the second stage

Three levers, in order of how much they help.

Open a path. Pierce or score the pastry before it goes in. The arithmetic above says 99 percent of the steam was already escaping and the last one percent was doing the damage, so a route for that last one percent is a disproportionate fix. This is the same reason a pie gets a vent.

Cut the rate. On a cycling oven that means a lower wattage machine or shorter bursts with real pauses, long enough for the pressure to fall rather than merely for the magnetron to rest. On an inverter it means an actual reduced power, which is a different and better thing.

And dry the surface, which no microwave can do. This is where a combination appliance earns its keep: a TOSHIBA 4-in-1 countertop microwave oven publishes convection, air fry and combination modes alongside 1000 watts of microwave power, and a convection stage is the only thing in the category that gets a surface above 100 degrees and turns it into the porous crust an oven would have made. Its own drawbacks are straight about the gap in its specification, "No specific air fryer wattage or [temperature]", so you cannot predict the drying rate from the listing, only the fact that it has one.

Our page on why an air fryer browns the crust before the middle cooks is the mirror image of this one and worth reading next to it. There, the heat is all at the surface and the limit is conduction inward. Here, the heat is all in the interior and the limit is permeability outward. The same pastry fails in opposite directions in the two machines, and a combination appliance exists precisely because neither direction is right on its own.

The honest limit

The constants are solid and the pastry is a worked example.

The latent heat, the specific volume of steam and the thin film pressure relation do not move. The 700, 1000 and 1200 watt figures are published by our own listings. Everything describing the food is representative: the 100 grams, the 2.8 joules per gram kelvin, the 0.6 density, the 0.0184 square metres of surface, the 0.2 millimetre skin and the 10 kilopascal hot dough yield. The oven's 51 watts rests on a convection coefficient of 20 and an emissivity of 0.9, neither measured.

One input deserves a specific warning, because it is the weakest and it is tempting. The penetration depth of microwaves into pastry is not something this page computed, and it cannot be, because fat is nearly transparent to the field while water absorbs it strongly. A laminated pastry is therefore an alternating stack of absorbing and transparent layers whose effective behaviour depends on a dielectric property that varies with temperature, fat content and water content, and that no listing publishes for any food. The layered heating that follows from it is real and is part of why blisters form between laminations rather than anywhere. It is not quantified here and should not be.

What survives all of it is the factor of twenty between an oven and a microwave, because it is a ratio of heat inputs, and the conclusion that the available pressure dwarfs the blister threshold, because that gap is three orders of magnitude wide and no plausible revision of a soft input closes it.