Why your microwave runs at 2.45 GHz

Microwave spectroscopy is my research field, so this one is personal. In a spectrometer, "resonance" means something specific: a molecule absorbs at a sharp, fixed frequency set by its structure, and nowhere else. That is real, and water has such lines. They just are not where the oven is, and they do not belong to the water in your soup.

Water's sharp line belongs to the vapor

A water molecule floating free in a gas can spin. Quantum mechanics only allows certain spin energies, so the molecule absorbs microwaves only at the exact frequencies that bridge those energy levels. The best known of them sits at 22.235 GHz. The international standard for how the atmosphere absorbs radio signals, ITU-R Recommendation P.676, lists it first in its table of water-vapor lines, at 22.235080 GHz. (The same physics of sharp rotational lines is how molecules get identified in space, covered in Report 169.)

That is roughly nine times the frequency of your oven. So even the line people have in mind is nowhere near 2.45 GHz.

Liquid water does something else: relaxation

In a liquid, a water molecule cannot spin freely. It is locked into a shifting web of hydrogen bonds with its neighbors. What it can do is try to swing its electric dipole to follow the field, and it does so sluggishly, because the neighbors hold it back. Physicists call that relaxation, not resonance. Instead of a sharp line, you get a broad hump of absorption spread over a wide range of frequencies.

The standard reference for this is a 2007 review by William Ellison in the Journal of Physical and Chemical Reference Data, which combined all the experimental measurements available into one function for pure water from 0 to 25 terahertz and 0 to 100 °C. Its abstract describes three relaxations in the microwave region and two resonances only in the far infrared. At 25 °C, the dominant relaxation frequency is 18.56 GHz. An earlier lab study by Buchner, Barthel and Stauber (1999) likewise describes water's microwave spectrum as a superposition of two Debye relaxation processes, relates the dominant slow one to how fast water molecules break loose from their neighbors, and deduces hydrogen-bond probabilities from it.

So liquid water's absorption peaks near 19 GHz at room temperature, and it is a wide, gentle hump. At 2.45 GHz, the oven sits well down the low-frequency side of it. Water still absorbs there, plenty to heat food. There is just nothing special about 2.45 GHz as far as the water is concerned.

So where does 2.45 GHz come from?

From the radio regulators. Anything that heats with radio energy leaks some of it, so heating equipment gets its own bands, kept apart from communications. In the US these are the ISM (industrial, scientific and medical) frequencies in the FCC's rules, 47 CFR 18.301. The table includes 915 MHz with a tolerance of plus or minus 13 MHz, 2,450 MHz plus or minus 50 MHz, 5,800 MHz, and higher bands including 24.125 GHz. The 2,450 MHz band is the one home ovens use. A 2025 food-science paper on egg pasteurization describes the same split: 915 and 2,450 MHz are the frequencies allocated for microwave treatment.

Notice that the list includes a band at 24.125 GHz, close to liquid water's absorption peak. If matching water were the goal, that band was available. Ovens do not use it, and the reason is the most useful idea in this whole topic.

The trade-off: absorbing well versus reaching the middle

The more strongly a material absorbs, the faster the energy is used up as it travels inward. Food scientists measure this as penetration depth: how far in the power falls to about 37 percent (1/e) of what hit the surface. The egg study above gives the formula and measured it. For egg white at 20 °C, penetration depth was 5.96 cm at 27.12 MHz, a radio-frequency heating band, and 0.85 cm at 2,450 MHz. Higher frequency, stronger absorption, shallower reach.

Push the frequency up toward water's peak and, by the same formula, that depth shrinks further: the frequency rises and, up to the peak, so does water's absorption, and both cut the depth. My read of what that means in practice: an oven running near 20 GHz would cook a thin outer skin hard while the center stayed cold, which is the familiar microwave failure mode made worse. Running well below the peak trades some absorption strength for reach. By the same logic, the lower 915 MHz band reaches deeper still, which is the reason to choose it when the load is thick.

What to take from this

Your microwave heats food because liquid water, and other polar and salty things in food, absorb across a broad range of microwave frequencies. It does not need to hit a magic number, and it does not. The 2.45 GHz figure comes from a band regulators set aside, and it happens to give a reasonable balance between how hard the food absorbs and how deep the energy gets.

The general lesson carries well beyond the kitchen. "Resonance" is a word that sounds precise and gets used loosely. Whenever someone says a device is tuned to the resonant frequency of something, ask which phase of matter they mean, and whether anyone has measured a sharp line there at all.

The honest limits on this

I read the Ellison and Buchner papers only at the abstract level; both full texts are paywalled. The 18.56 GHz figure and the relaxation-versus-resonance description come from those abstracts. The penetration-depth numbers are for egg white, not for water or for any particular food, and real foods vary with salt, fat and temperature. The claim that a near-20 GHz oven would heat mostly the surface is my reasoning from the formula and the measurements, not a result any of these sources tested directly. If a source shows otherwise, I will correct this report in place.