11.7.4. Central Hydrogen Depressions
As suggested by Roberts (1967) (see Section 11.4.5), it appears that a central depression in the HI surface brightness is a common feature in late-type spiral galaxies. M31 is the most obvious case where an HI "ring" is a good overall description. For the other galaxies so far observed the HI distribution is better described in terms of an asymmetry and a central depression. In no observed galaxy does the surface brightness resemble the overall light distribution, which is exponentially decreasing, or the total mass distribution (deduced from rotation curves or from an assumed mass-to-light ratio). The total mass and light distribution are centrally peaked; the HI surface brightness is rather flat or centrally depressed. There are several possible explanations:
The usual assumption that the gas is everywhere optically thin may not be correct. It is almost certainly true that small cold optically thick clouds occur locally in every galaxy. If the spin temperature or the cloudiness of the HI varies systematically across the galaxy, then the surface brightness of HI line radiation may not resemble the HI surface density distribution. This explanation requires lower spin temperatures or very clumpy HI in the central regions of galaxies.
The thickness of the HI layer may increase with radius, as observed in our galaxy (Kerr and Westerhout, 1965). The HI layer thickness in M33 can be estimated from the measured velocity dispersion in the HI line profiles. If we assume that the vertical velocity dispersion (c2) supports the gas against gravitational attraction, then the half-thickness () of the gas layer is given by Mestel (1963) as
where is the total mass surface density in the plane of the galaxy. has been estimated from a mass-model fitted to the rotation curve and decreases with radius. To first order the observed velocity dispersion in M33 is constant; therefore the HI layer thickness increases with radius. In M33 the HI surface density is roughly constant with the radius (see Figure 11.5) ; therefore the HI volume density decreases with radius. This is consistent with the observed decrease in the rate of star formation with radius in M33 (Madore et al., 1973). In M31 a ten-fold increase in layer thickness would make the HI volume density in the central regions equal to that at the position of the ring.
The gas in the central regions may be ionized. Monnet (1971) has observed a weak background of Ha radiation in the central regions of several galaxies. In the case of M31, however, the emission measure was less than 10 cm-6 pc, which is too small by an order of magnitude to give a surface density comparable with that of the HI peak.
The gas could be in molecular form without any observable consequences. Molecular hydrogen has been seen in the far-ultraviolet absorption spectra of the star Persei, which has a strong ultraviolet continuum emission (Carruthers, 1970). However, observation of H2 in external galaxies is far beyond present techniques.
Neutral hydrogen is expanding away from the central regions of the Milky Way (see Oort, 1964; Kerr, 1967). The 3-kpc-arm feature of our galaxy is expanding at a velocity of some 53 km sec-1, while asymmetry in the rotation curve for our galaxy can be interpreted as a 7 km sec-1 expansion in the vicinity of the Sun. The outflow of neutral gas from the central regions (about 1M per year) might be partially balanced by an inflow (under one interpretation of the high-velocity gas seen at high latitudes in the galaxy; see Oort, 1967), but is otherwise sufficient to deplete the central regions of HI in some 107 to 108 years.
Expansion motions have been observed in the central regions of some external galaxies (e.g., M101) but in M31 they are less than 10 km sec-1 (Gottesman and Davies, 1970) and are too small to have removed a significant mass of HI away from the central regions.
The gas in the central regions may have been used up by star formation. Star formation would initially be fastest in the central regions of galaxies where the gas density is highest. At the present epoch the rate of star formation in M31 is at its peak in the region of the ridge at 10-kpc radius.
The correct explanation of the deficit of HI in the central regions is probably a combination of explanations (2) and (6), although in some instances explanations (3) and (5) may be highly significant. Explanations (1) and (4) are unknown factors.