2000ApJ...544..347M

There are now observations of several emission lines from the ``warm ionized medium'' (WIM or, equivalently, the ``diffuse ionized gas'') of the local interstellar medium, from the Perseus arm in the Milky Way, and also in several other galaxies. Interesting features of these observations include the great strength of [N II] λ6563 (∼Hα in some cases) and the fact that [S II] λ6717/[N II] λ6583 is almost the same (∼0.6-0.7) in all locations and objects. Other line ratios (e.g., [O III] λ5007/Hβ) vary considerably.

This paper presents simple photoionization models that reproduce the observed spectra, providing extra heating beyond that supplied by photoionization is assumed. The same extra heating was used for models of all stellar temperatures being combined together, although it could easily depend on T_*.

With observed gas-phase abundances (not solar), the line ratios in the Local arm at b=0° are fitted with no extra heating and S/H=13 ppm, as opposed to solar (∼20 ppm). Local gas observed at b=-35° requires extra heating of about Γ_–25=0.75, where Γ_–25 is the extra heating in units of 10^–25 ergs.H^–1.s^–1. In the Perseus arm there are similar results: little extra heating is required at |z|=500 pc, and Γ_–25=3.0 is needed at |z|=1.2 kpc. To fit the observations, the gas-phase composition in the Perseus arm must be reduced as required by the Galactic abundance gradient observed for H II regions. The requirements for NGC 891 (the best observed other galaxy) at |z|=1 kpc and 2 kpc are similar to the Perseus arm: little or no extra heating near the plane (1 kpc in this case) and Γ_–25∼3 at |z|=2 kpc. In NGC 891 there is also an increase of λ5007/Hα with |z| that can only come about if most of the ionizing radiation is supplied by very hot stars (type O4: T_*∼50,000 K). Either their radiation must propagate from the plane to high |z| through very little intervening matter, or else the stars are located at high |z|. The total power requirement of the extra heating is ≲15% of the power to photoionize the WIM without extra heating.

Extra heating enhances [O II] λ3727/Hβ. Figure 1 shows that there is a spread in the predicted values, but the ratio can serve as a useful diagnostic of extra heating. The [S III] λλ9065, 9531 lines (see Fig. 2) are not useful in diagnosing extra heating.