Observations by Kennicutt suggest that significant star formation in galactic discs only occurs when a criterion for disc instability is satisfied and that, when star formation is occurring, departure from the marginal state is not great. We have studied galactic disc models in which there is no star formation until the disc has a threshold surface density at which an instability criterion is satisfied. We assume that the rate of the subsequent star formation is just such as to keep the criterion marginally satisfied. Our models contain many approximations and free parameters, but they do have properties which resemble those of real galactic discs. These include a slow variation of the star formation rate with time and a slow change in both space and time of the gas random velocity. There is good general agreement with the observed stellar age/metallicity relation. After a time, at any radial position, the rate of star formation is close to a power law in the gas surface density. Because of the threshold, no stars are produced until some time after the disc starts forming and, if the model were valid, there would therefore be no luminous discs at very high red shift. Moreover, there should not be a large density of hidden matter in the form of elementary particles in the solar neighbourhood. The threshold model cannot readily explain the G-dwarf distribution in the solar neighbourhood, and it is necessary to suppose that the disc starts with a non-zero initial metallicity. The model in which gas infall proceeds on the same time-scale at all radii does not produce abundance gradients as large as those observed. There is some improvement if the infall is biased, occurring more slowly at larger radii.
Read full abstract