Does a microwave resonantly excite the rotational levels when cooking?

Wikipedia states there is no resonance absorption, but says at the same time that the molecules are oscillating like dipoles, which is kind of the same if you are exciting the rotational levels ? The spacing in the rotational levels will be almost independent of wavelength in the $\mathrm{MHz}\, -\, \mathrm{GHz}$ range?

The Wikipedia quote you are referring to is:

Sometimes, microwave heating is explained as a resonance of water molecules, but this is incorrect;[11] such resonances occur only at above 1 terahertz (THz).

The source cited is:

Schmitt, Ron (2002). Electromagnetics Explained: a handbook for wireless/RF, EMC, and high-speed electronics. Burlington, MA, USA: Elsevier. p. 343. ISBN 978-0-7506-7403-4. Retrieved 3 December 2012.

Often I've seen Microwave heating explained as the size (wavelength) of the microwaves as being the same size as water molecules, causing them to rotate. For example this image:

This is not correct. The electromagnetic wave would have a frequency on the order of $10^{18}\, \mathrm{Hz}$ if it were.

Microwave ovens operate at a frequency of 2.45 GHz (2.45x109 Hz) and this is NOT the resonant frequency of a water molecule. This frequency is much lower than the diatomic molecule resonant frequencies mentioned earlier. If 2.45 GHz were the resonant frequency of water molecules the microwaves would all be absorbed in the surface layer of a substance (liquid water or food) and so the interior of the food would not get cooked at all.

• Don't we not even expect discrete rotational bands to exist in the liquid phase?
– user4552
May 3, 2013 at 0:54
• @BenCrowell yes, with band center up at 30-50GHz. Don't forget that 27MHz medical diathermy is also heating water by dielectric absorption, just like microwaves do. It's the dipole rotation and ruptured bonds which heat the water, not an EM resonance. Aug 29, 2017 at 23:34

The resonance frequency is really the measure of the slips between your dipoles. Average formation time for - H····H - is close to ~10 +/-1 picoseconds.

Cube that for the xyz rotational axises and factor in a little disparty due to drag caused from longer range ordering, varience in liquid volume, localized # of H2O molecules and change in molecular rotaion freedom as a consequence of increased thermal energy.