2021ApJ...909...78D

We investigate what drives the redshift evolution of the typical electron density (n_e) in star-forming galaxies, using a sample of 140 galaxies drawn primarily from KMOS^3D (0.6 < z < 2.6) and 471 galaxies from SAMI (z < 0.113). We select galaxies that do not show evidence of active galactic nucleus activity or outflows to constrain the average conditions within H II regions. Measurements of the [S II]λ6716/[S II]λ6731 ratio in four redshift bins indicate that the local n_e in the line-emitting material decreases from 187_–132⁺¹⁴⁰ cm^–3 at z ∼ 2.2 to 32_–9⁺⁴ cm^–3 at z ∼ 0, consistent with previous results. We use the Hα luminosity to estimate the rms n_e averaged over the volumes of star-forming disks at each redshift. The local and volume-averaged n_e evolve at similar rates, hinting that the volume filling factor of the line-emitting gas may be approximately constant across 0 <= z <= 2.6. The KMOS^3D and SAMI galaxies follow a roughly monotonic trend between n_e and star formation rate, but the KMOS^3D galaxies have systematically higher n_e than the SAMI galaxies at a fixed offset from the star-forming main sequence, suggesting a link between the n_e evolution and the evolving main sequence normalization. We quantitatively test potential drivers of the density evolution and find that n_e(rms) ≃n_H2_, suggesting that the elevated n_e in high-z H II regions could plausibly be the direct result of higher densities in the parent molecular clouds. There is also tentative evidence that n_e could be influenced by the balance between stellar feedback, which drives the expansion of H II regions, and the ambient pressure, which resists their expansion.