Evidence for density bounding in the most luminous regions

H luminosity functions

The first clue to the presence of density bounding in the most luminous H II regions of spirals comes from their H luminosity functions (LF's).Following the early work by Kennicutt et al. (1989) using pre-CCD images, we showed with high resolution, high quality CCD frames (Rozas et al., 1996) that there is a break in the LF slope, accompanied by a jump in the function, at a luminosity which varies rather little from one galaxy to another (see Fig.2). Our tentative hypothesis for this (see Beckman et al., 1998) is that the relation between cumulative stellar mass and cloud mass in OB star-producing clouds causes the Lyc flux from OB associations to be a more than linearly rising function of the cloud mass. Thus at a critical mass the H II region just ionizes its cloud;clouds of higher mass will have a larger flux per unit mass and will therefore be density bounded. Although a model by McKee and Williams (1997), based on the transition from ionization by one star to more, gives rise to a change in LF slope, this occurs at a luminosity in Halpha lower by a factor 10 than our observed "glitch",and would not explain the jump in value which accompanies the break.

 
Figure 1: Observed internal surface brightness gradients of individual H II regions in the galaxies indicated, as a function of their luminosities. The break in this graph at L=10 erg s is as predicted for a change in regime from ionization bounding to density bounding. The gradients steepen and the change occurs over a restricted luminosity range.

Internal surface brightness gradients of H II regions

When we measure the radial gradients in the H surface brightnesses of individual H II regions, and plot these against their integrated luminosities, we find constant values up to a critical luminosity, followed by a steady rise with increasing luminosity to much steeper gradients at higher luminosities. This behaviour is shown in Fig.1 in a diagram which summarizes the results from a number of spirals. The value of the critical luminosity is just that at the jump observed in the H LF's (Fig.2). The hypothesis that the H II regions are ionization bounded below the critical value, and density bounded above it gives a model which explains this behaviour.It cannot be explained by assuming a jump,at this luminosity,in the number of stars ionizing the region, since the bjrightness gradient is independent of this number,as was in fact shown in Stromgren's original article. Only a geometrical property of the surrounding gas cloud can account for our observations. A simple corrollary of our hypothesis is that the brightest H II regions, those above the critical luminosity, should also be those with highest surface brightnesses, and this is clearly observed. This would not be caused by a change in the number of stars ionizing the regions, in the absence of the geometrical effect due to density bounding.

In a poster paper presented at this meeting (Sabalisck et al., 1997), we measure the internal velocity dispersions of the 200 most luminous regions in the grand-design spiral M100, via their H emission, using a Fabry-Perot mapping technique. Plotting these dispersions against luminosity we find an upper envelope which we attribute to virialized regions. We find a similar result, with an envelope having the same slope, 2.6 in the plane, for M101. This slope is predicted to take a value 4 (Terlevich & Melnick, 1981), but this value would be for ionization bounded regions. It is easy to show that if a fraction of the ionizing photons escape, the observed slope should be reduced. The value of 2.6 falls well within the predictions of a straightforward density bounded model for the regions lying on the envelope, and offers further evidence for our hypothesis.

 
Figure 2: H luminosity of the H II regions in four late-type spiral galaxies. The sample is complete to well below 10 erg s. The LF is obtained by projecting the number-diameter-luminosity surface into the log N-log L plane. Note the "glitch" (change in gradient accompained by peaks) in the LF near L=10 erg s (see text for details).

© Copyright Astronomical Society of Australia 1997