Selective heating

Why does a microwave oven heat the food but leave the plate cold?

Put a bowl of soup on a ceramic plate in a microwave oven and run it for two minutes. The soup is scalding; the plate is merely warm, and warm only where the soup touched it. Yet the microwaves flooded the cavity indiscriminately — the plate was bathed in exactly the same field. Something about the soup accepts energy from that field and something about the plate refuses it. What is the difference, and does it have anything to do with the oven being "tuned" to water, as almost everyone says?

Begin with what a microwave field is: an electric field reversing direction billions of times a second. A domestic oven runs at 2.45 gigahertz, so the field flips about five thousand million times per second. What could a molecule do about that?

If the molecule has a permanent electric dipole — a positive end and a negative end, as water does, its oxygen holding electrons more tightly than its hydrogens — then an electric field exerts a torque on it, trying to line it up. Reverse the field and it is torqued the other way. So water molecules are being wrenched back and forth.

Now the crucial step. Do they turn freely? Not remotely. In liquid water each molecule is hydrogen-bonded to its neighbours and packed shoulder to shoulder, so turning means shoving neighbours aside. The molecule cannot keep up with the field; it lags behind, and that lag is exactly where energy is lost — the work done against the surrounding molecules ends up as random molecular motion, which is to say heat. The field does not deposit energy in individual molecules and leave; it stirs a crowd, and the crowd's internal friction warms it.

Note what has not been invoked: no resonance. This is worth dwelling on, because the resonance story is repeated almost universally and is wrong. If 2.45 gigahertz were a resonant absorption line of water, absorption would be enormously strong at that frequency and negligible either side of it — and an oven designed that way would be useless, dumping all its energy in the outermost millimetre and never reaching the middle of a potato. The opposite is true: water's loss at microwave frequencies is a broad, featureless hump. It peaks somewhere around 19 or 20 gigahertz at room temperature, and 2.45 sits well down the low-frequency flank, absorbing weakly enough to penetrate a few centimetres. Weak absorption is a feature, not a compromise.

So why 2.45 gigahertz at all? Because it lies in a band internationally reserved for industrial, scientific and medical use, because magnetrons at that frequency are cheap, and because the penetration depth suits food-sized objects. The number is an engineering and licensing choice, not a molecular one.

With the mechanism straight, the plate explains itself. Ceramic is a rigid oxide lattice; its charges are not free to reorient, so there is no population of mobile dipoles to be spun and the field passes through doing almost no work. The same is broadly true of most glass, many plastics, and dry paper. Any warmth in the plate arrives by ordinary conduction from the hot food sitting on it — which is why it is hot under the bowl and cool at the rim.

Then consider ice, where the model earns its keep by predicting something counterintuitive. Ice is water and the dipoles are still there, but in the crystal each molecule is locked into a lattice and reorienting requires a defect to migrate past; the characteristic reorientation time is measured in microseconds rather than picoseconds. At five thousand million reversals a second the molecules are hopelessly slow to respond, so they barely turn and barely dissipate. Ice absorbs microwave energy far more weakly than liquid water at the same temperature. That is the whole reason defrosting is difficult, and why it fails the way it does: the first patch to melt absorbs strongly, heats fast, melts its neighbours and runs away, giving a boiling rim around a frozen core. Defrost cycles pulse the magnetron on and off precisely so conduction can even things out between bursts.

Three refinements.

First, dipole rotation is not the only channel. Dissolved ions — the salt in soup, the minerals in gravy — are pushed bodily back and forth by the field and dissipate energy through ordinary conduction. Salty food heats faster than plain water for that reason, not because it holds more water.

Second, "leaves the plate cold" is a simplification. Some ceramics and glasses contain enough mobile ions or absorbed moisture to warm noticeably, and a few glazes carry metallic pigment that arcs. Materials sit on a continuum of loss, not in two boxes.

Third, metal is the opposite failure. A conductor's electrons respond so freely that the field is reflected rather than absorbed; the problem with foil is not that it heats but that charge concentrates at sharp edges until the air breaks down, which is arcing, not cooking.

• Why penetration depth, not power, sets the maximum practical thickness of food a microwave can cook evenly.
• How industrial radio-frequency and dielectric heating pick quite different frequencies for quite different materials.

Every term the collection defines is gathered in the glossary.