This is not really true. While the microwave absorption of water is significant at 2.4 GHz the peak is (a) quite broad and (b) centered at 10GHz or higher (higher if the temperature is raised); see plot in [0]. So ascribing some "specialness" to this frequency due to the abundance of water in living systems isn't an explanation.

[0] https://en.wikipedia.org/wiki/Electromagnetic_absorption_by_...

To expand, apparently microwaving food right at the absorption peak sucks because of low penetrance, heating only the top layer of the food. The reason 2.4 GHz was chosen was purely as that is where the ISM band already was at the time.

i’m trying to google a better answer, it seems ISM bands were chosen when microwaves were a nascent technology, so it’s a bit chicken and egg

2.6 was chosen based on empirical heat penetration testing for typical target food sizes, e.g. Potatoes.

GE lobbied for 2.4 instead so they could emit a noisy harmonic from a crystal based 1.2 system so the FCC agreed.

Source: national archives documents I researched

Cool! I came across some quoras that suggested the electronics were easy for 2.4, but didn't give further detail. Noisy harmonics does sound useful for cooking!

Not really an answer, but here's some context...

In 1952, a paper called Tubes for dielectric heating at 915 megacycles said this, "Dielectric heating can be much faster at microwave frequencies. Because leakage radiation is large enough to interfere with communications, heating is best done at assigned frequencies. The bands at 915 and 2,450 megacycles are most useful."

https://ieeexplore.ieee.org/abstract/document/6371890

Ah, OK, 1947.

"The plan which is, recommended by the United States provides for certain discrete frequencies, with appropriate guard bands, for those industrial scientific and medical devices which cannot be shielded effectively. This assures a minimum disruption of vital communication services. The allocation of frequencies for this purpose is justified because of the tremendous investments in equipment and the public demand for the various services performed, together with the harmful interference to radio communications which would result if such devices were scattered indiscriminately throughout the spectrum. The frequencies proposed are:

13660 kHz

27320 kHz

40980 kHz

2450 MHz"

http://handle.itu.int/11.1004/020.1000/4.62.51.en.108, page 628.

I did much of this archival research (and wrote the Wikipedia section) let me know if you have questions.

Yeah, I do not think I have a good source at hand, this is a fun fact from an E&M lecture we had.

I also wonder if anyone actually experimented with microwaving their dinner at 10GHz or if that was just an educated guess. X-band magnetrons for radars are widely available though, so I would not be surprised if that was coming from practical experience...

A review of the history of microwave cooking implies that the interaction of 10 GHz microwave radiation with water was understood by WWII.

"effective penetration and heating favors wavelengths in the cm range (1–30 GHz). Higher frequencies cannot penetrate sufficiently, and much lower frequencies require very high fields, and can have poor coupling unless making use of direct contact or inductive heating. In the years following WWII, emphasis was thus placed on frequencies in the low GHz region, where both the historic development of microwave power sources and serendipity converged."

https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9393469

A microwave oven is 10000 times more powerful than a typical WiFi signal and is focused on an object less than a foot away.

microwaves use it because water absorbs it + multiple wavelengths can fit inside the oven for even heating + easy to produce without fancy electronics

water can get shaken up by a wide band of frequencies

Microwaves could have used almost any frequency that was cost effective, 2.4 was a compromise to between competing companies and empirical testing.

From what I can find, it looks like the nearest absorption peak is around 20GHz, and while it is temperature-dependent, that peak doesn't shift all the way down to 2.45GHz.