There seem to be two distinct modes of global star formation. An
inefficient, quiescent mode that is self-regulated by feedback occurs in
galactic disks. A violent, efficient starburst mode is triggered and
fueled by merger-induced
tidal torques, and accounts for the formation of the massive stellar
spheroids. The disk mode of star formation most likely also accounts for the
low mass spheroids, which are generated dynamically via secular evolution of
Semi-analytic galaxy formation modelling has not yet incorporated the full
richness of star formation phenomenology. There are five critical issues
currently confronting semi-analytic theory, and improvements in star
formation modelling and the dynamical coupling of baryonic and non-baryonic
matter will be necessary to address most of them.
- Perhaps the most innocuous of the problems is the overproduction of
substructure, and in particular the predicted abundance of dwarf galaxies.
This is likely to be resolved at least in part by early photo-ionization at
redshifts above 6, before the reionization of the intergalactic medium was
complete. Early photo-ionization of dwarfs with velocity dispersion less
about 30 km/s suffices to eject most of the gas before the bulk of the star
formation has occurred. Because this happens early, there is a proximity
effect: only the most widely separated dwarfs, which experience a lower UV
flux, retain their gas and survive as stellar systems. One can both account
for the paucity of observed dwarfs and their mean separations from the
galaxies. What is not yet clear is whether the number of SMC-like systems
predicted is consistant with observations, nor more seriously perhaps, is
whether reconciling the low dwarf abundance in the Local Group allows one to
also understand the apparent surfeit of dwarfs observed in some nearby
- The baryon overcooling problem is more serious. The factor of about 2
excess predicted for disk baryons can be resolved either by ejection or by
hiding the gas. Winds can occur if outflows from massive galaxies are
stronger than predicted by simple disk modelling and simulations. The Lyman
break galaxies possibly indicate the effectiveness of early winds, but it is
likely that these galaxies end up as E or S0 galaxies, as inferred from
spatial clustering. Substantial amounts of gas could be hidden in the
form of dense cold clumps of H2, or even in diffuse
H2 in the outer disk,
where H2 may be more readily hidden.
- Disk scale lengths can possibly be explained via turbulent
feedback. Winds may play a role here too, if the predominantly low angular
momentum gas, which ends up in the central core, is preferentially
expelled. A supermassive black hole-driven outflow may provide a possible
driver for the nuclear wind.
- Disk colours, and indeed spheroid colours, present another difficulty,
as exemplified by the colour-magnitude relation. Disks, whose global colours
are characteristic of intermediate age populations, have stellar populations
that are systematically too red, and too old, at large masses. For
ellipticals, the problem is somewhat different. The colour-magnitude
is driven primarily by metallicity, with there being little evidence for any
significant age spread in cluster galaxies
[Vazdekis et al.2001].
The stellar populations are predominantly old. Both massive and low mass
similar ages. Either the low mass systems formed stars earlier but less
efficiently, hence reducing their effective age, or gas accretion
produce a similar effect. The alpha element ratio enhancement for the cores
of massive ellipticals supports an early and efficient duration of star
formation for these systems. Field ellipticals and S0s however show a
substantial age spread.
- The issue of a cusp in the dark matter profile represents a contentious
issue around which the observations have not yet converged. It may be that
dynamical effects, such as black hole mergers, provide the best means of
scouring out the cusp. The related matter of
dark matter concentration seems to vary between galaxies,
and there are hints that it varies between barred and unbarred galaxies.
This suggests again that a complex dynamical history may be partly to blame.
I am indebted to my group at Oxford for many discussions of galaxy
formation, and in particular for conversations pertinent to this review with
Julien Devriendt, Ignacio Ferreras, Adrianne Slyz and
James Taylor. Part of this review was completed at the Institut
d'Astrophysique de Paris, where I am grateful to the Director, Bernard Fort,
for his kind hospitality. I also thank N. Prantzos for useful discussions.