Let us now interpret the correlation between CO linewidth and starburst age. We restrict the discussion to the evolution of starbursts and CO distributions in the central regions of galaxies, since the starburst ages of our sample are those of nuclear HII regions.
The first explanation that comes to mind is that the molecular clouds are initially concentrated right near the nucleus ( 100 pc). The lines are broadened by galactic rotation, assuming a standard rotation curve (Sofue 1996) and/or by the velocity dispersion among individual clouds due to the kinetical energy from supernovae and stellar winds associated with the starburst (Irwin & Sofue 1996). As the starburst ages, the gas expands outward ( 500 pc) and the lines become wider because the clouds take part in the galactic rotation. The problems with this scenario are that most of the gas must stay in the disk and that there is no physical mechanism by which the clouds can acquire enough angular momentum to participate to the galactic rotation. This is moreover in contradiction with the fact that the gas on the contrary tends to lose angular momentum (in the inflow scenario).
A classical gas inflow cannot explain the correlation either, since it predicts that the starbursts may be ignited in a circum-nuclear ring before the gas reaches the center, and young starbursts with broad CO lines are expected.
Another possibility is that the clouds are pushed outward, preferentially out of the plane of the galaxy, by powerful outflowing winds generated by supernova explosions (Heckman et al. 1990). The broadening of the lines is then due to expansion rather than to rotation. The problem here is that the line profiles should be double peaked, which is generally not the case in our sample.
Finally, the correlation could be spurious, if galaxies with young starbursts are nearly face-on or have a low mass, but this is not the case here.
None of the above scenarii, outflow of molecular gas inside or out of the disk, are completely satisfactory. This correlation remains a puzzling problem which deserves more work. On one hand, the numerical simulations must increase both their spatial and temporal resolutions and include an energetic starburst in the nucleus of galaxies. On the other hand, we must acquire higher resolution ( 100 pc) CO observations in the nuclei of galaxies at different stages of starburst evolution.
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998