Supernova (SN) explosions and winds from massive stars provide input of
mechanical energy into the ISM in galaxies. In a starburst, the
collective action of many supernovae may lead to a galactic wind,
transporting material out from the plane of the galaxy. Winds could
affect the velocity field of the ionised gas and has to be taken
into account. Observationally, winds have been seen in some low luminosity,
hence low mass, BCGs (Meurer et al. 1992; Papaderos et al.
1994; Lequeux et al. 1995; Martin 1998).
Recent models which include also the restoring force provided by dark
matter have shown that feedback from star formation is not expected to
cause gas loss from galaxies more massive than
(MacLow &
Ferarra 1999; Ferarra & Tolstoy 2000). Thus for the galaxies studied in
this paper, it should be safe to
conclude that feedback has not affected the global velocity fields. However,
the calculations by MacLow & Ferarra (1999) and Ferarra & Tolstoy (2000)
are restricted to energy injection rates
corresponding to supernova frequencies of one SN every 30000 years or less.
In this sample, where the derived star formation rates span the range 0.25
to 20 
/year, we expect (assuming a Salpeter IMF with
,
,
and lower initial mass
limit for core-collapse supernova of 8 )
SN frequencies of 0.02 to
0.15 SN/year, hence 3 orders of magnitude larger! Extrapolating the calculations
by MacLow and Ferarra we still do not expect blow-away to occur, while blow-out
is possible. Of course, extrapolating by 3 orders of magnitude is dubious.
If gas gets blown out, it will stay in the halo and may later condense back
on the host galaxy.
Interestingly, deep H
images of ESO338-IG04 and ESO350-IG38 show that
the H
morphologies are extended along the minor axis, compared to
the broad band morphologies.
This is what is expected for disk galaxies: gas escapes perpendicular to
the plane while the gas in the plane is little affected. Thus the
rotation curves, which are derived along the major axis, are probably not
affected by ouflows and expanding superbubbles while the velocity fields
along the minor axis may well be (e.g. in ESO480-IG12).
Our sample of relatively luminous BCGs have irregular morphology at both bright and faint isophotal levels. This agrees with the findings of Telles et al. (1997) that BCGs with irregular morphology at faint isophotal levels are on average more luminous than those with regular morphology.
A galaxy in equilibrium should have a regular symmetric morphology and kinematics. The occurence of star formation, if not symmetrically distributed, may create apparent morphological irregularity. Nevertheless, the outer parts of a galaxy are essentially unaffected by the central activity, hence the outer isophotes of a galaxy mainly trace the distribution of stars. Therefore, significant large scale asymetries in the outer isophotes are signatures of an asymetric distribution of stars. Such asymetries could be the consequence of a interaction/merger but weaker distortions could also be produced by instabilities which may be inherent or tidally triggered. Mergers and interactions between galaxies lead to large scale morphological distortions (e.g. tidal tails and shells) and to perturbations in their rotation curves due to gas flows driven into the center. The light distributions of the 6 BCGs presented in this study show outer distortions to some extent which may suggest that the starbursts have been produced by interactions or mergers. In some cases, like ESO185-IG13 and ESO350-IG38, the merger signatures are very strong.
In principle, one of the best ways to investigate the effect of interaction between galaxies should be through the kinematics of the gas, since it traces well the dynamics of the burst tidally triggered trough the collision of gas-rich systems. From the reported results we have shown that on the top of a slow rotating disk system, important dynamical perturbations have been detected which prevent us in some cases from obtaining the gravitational mass of the galaxies. None of the BCGs show regular disk kinematics and different explanations have to be invoked for each particular case (see Sects. 3 and 4). It has to be noticed that the only truly regular kinematics occur in the companion galaxies or when we can disentangle quite well the different kinematical components, as in the cases of ESO185-IG13, ESO480-IG12 and Tololo0341-403. It is then not surprising that we cannot define a general trend for all the studied galaxies. Depending on the state in which we see the interaction/merger, we will detect more or less chaotic velocity maps.
Most investigations of the environments of BCGs have shown that these
kind of galaxies are rather isolated with respect to giant luminous
galaxies and preferentially occupy low and intermediate density
environments (see Introduction).
We have used NED![[*]](/icons/foot_motif.gif){kind=icon}
to
investigate the environment of the galaxies in
this study. Since NED was mainly composed from catalogues of bright
galaxies, such a study essentially provides information on the
presence of massive neighbouring galaxies. We searched a projected
radius of 1 Mpc around the galaxies, and 
kms-1 in velocity space.
The velocity range was chosen very generously so as not to reject possible neighbours
with high peculiar velocities. Travelling one Mpc with a
velocity of 1000 kms-1 would take one Gyr. Thus, if galaxies so far
apart have ever interacted, this must have happened more than 1 Gyr
ago, even for very high peculiar velocities. Hence, any signs of interaction
(e.g. a triggered starburst) should be long gone.
None of the galaxies in this study has any neighbour in
NED within 1 Mpc, except for ESO338-IG04 and ESO400-G43, the companions
of which are included in this study.
The analysis shows that the local density of catalogued galaxies is not enhanced in the locations where these galaxies are found; typically we find a density of catalogued galaxies of 0.005 Mpc-3, when averaged over a volume 2500 Mpc3 or greater. Thus, the sample studied is not located in particularly dense environments. In general, galaxy evolution is a strong function of the density of the environment. In low red-shift clusters, interactions are frequent but rarely lead to any significant star formation. The most active star-forming galaxies and star-forming mergers are found in the field or in sparse groups, and this is true also for BCGs, luminous or not.
A fundamental astrophysical problem is how to ignite a starburst. The
majority of galaxies are not involved in starbursts. For example, low
surface brightness galaxies (LSBGs) have very low star formation rates,
despite the generous supply of H I, which may be explained by
sub-critical H I column densities (van der Hulst
et al. 1993). Efficient star formation requires gas with high densities.
Here we consider as starbursts, galaxies which have gas depletion time-scales,
and time-scales for accumulating the observed photometric mass with
the current SFR much shorter than a Hubble-time, i.e.
Gyr or less.
A model where starbursts in dwarf galaxies are the consequense of
statistical fluctuations was presented by Gerola et al. (1980). Their
model predict that for galaxies with radii less than 1-2 kpc, star
formation will principally occur in short bursts, whereas larger
galaxies are predicted to have more continous star formation.
A similar argument is that only when
the star formation in a single giant molecular cloud would exceed the
time averaged SFR in a galaxy would we expect statistical fluctuations
to dominate. Hence for galaxies with a baryonic mass greater than
,
statistical fluctuations should not dominate the global SFR. The galaxies
presented in this paper are too large/massive for any significant
statistical fluctuations to be expected.
A popular scenario for BCGs has been one with cyclic bursts. Here, the
starburst is terminated by the expulsion of gas through supernova winds.
If the gas later accretes back to the galaxy, a new starburst could be
ignited. A potential problem is, however, the time-scale for the retrieval
of the gas. If the blown out gas eventually accretes back on the galaxy, one
expects this to occur over a time-scale much longer than the duration of
a starburst.
A constant inflow of gas at a slow rate is not likely to
produce a starburst, unless there is some threshold mechanism.
But even if there is a treshold, the accreted gas mass has to pass it coevally.
Since the star formation efficiency is less than 100%, the gas accretion
rate must be higher than the SFR. Such a high continous mass accretion rate
would not be compatible with the total gas+star masses of the galaxies.
For our BCGs which have burst masses of
or larger,
on the order of a few times 108 to
of gas would have to pass the treshold
within a few times 107 years. This implies a mass accretion time
scale of the same order as the free fall time.
Hence, gas falling back would have to come all at once in one lump,
contrary to the expected continous accretion. One could of course envision
that the expelled gas cools and clump together in a few giant massive
H I complexes, and if such an aggregation would fall into the centre,
we could get a starburst.
The time-scale for building up the observed photometric mass with the
current SFR is
Gyr (see Table 4), which indicates
that if the burst duration is 
100 Myr, the duty cycle between bursts
should be on the order of 1 Gyr. We would thus require that gas is
ejected on time scales shorter than 100 Myr, then the gas should stay away
for 1 Gyr, and finally, most of it should fall back within some 10 Myrs.
Supernova activity will begin after a few Myrs, whereas the bursts we observe
in these BCGs have already been active for several 10 Myrs. Hence, although we
in some cases see signatures of expanding bubbles, we see no evidence
for any superwinds capable of blowing out the ISM and terminating star formation.
Because of these arguments, we do not consider a self-regulated cyclic starburst
scenario very likely for these luminous BCGs. A cyclical burst scenario,
powered by winds or stochastic effects, is more likely in low mass BCGs.
Tidal forces generate gas flows that could funnel gas at large radii into the center of a galaxy and ignite a starburst, but generally on time scales that are too long to build up high central densities in competition with the destruction processes of molecular cloud complexes. Campos-Aguilar et al. (1993) showed that a dwarf galaxy is sensitive to tidal triggering only when the companion galaxy is very massive. With the lack of massive companions, interactions will only be strong when two dwarfs are actually merging or in physical contact. Similarly, Icke (1985) found that separations between galaxies for tidal gas flows to occur must be within a few galactic radii, but the typical time-scale is of the order of a galactic rotation period (here a few 100 Myrs). Thus, a weak tidal gas flow might well increase the star formation rate but likely not ignite a starburst. Mihos et al. (1997) used N-body simulations to study the response of LSBGs (possible progenitors of BCGs) to tidal interactions. They found that due to the low mass densities, LSBGs are very stable against tidal triggering, in contrast to normal high surface brightness giant disc galaxies. As there are no massive galaxies in the vicinity of our studied BCGs, and the two known companions are rather distant (more than 70 kpc), tidal interactions cannot be the main cause of the starbursts. However, tidal interactions may be responsible for the moderately high SFRs seen in the two companions.
Another scenario for BCGs is that the starburst is ignited through the merging between two dwarfs, or between a dwarf and a massive intergalactic gas cloud. In a merger, gas clouds lose a considerably fraction of their angular momentum in dissipative collisions and fall into the center of the galaxy/galaxies. The time-scale is sufficiently short for building up large overdensities that can start to form stars coevally. This is thus an attractive scenario for the origin of starbursts in BCGs. Massive isolated H I clouds seem to be very rare at the current epoch (see Kunth & Östlin 1999). However, any merger involving a sufficiently gas-rich component could do the job. Mergers involving at least one gas-rich component is the mechanism favoured for the BCGs in this sample.
HST observations of ESO338-IG04 has revealed that it contains a rich population of young globular clusters (Östlin et al. 1998). Similar rich populations of luminous star clusters were found in HST archive images of ESO350-IG38 and ESO185-IG13 (Östlin 2000). Numerous young globular cluster candidates have also been found in mergers, like the "Antennae" and NGC 7252 (Whitmore et al. 1993, 1995), and galactic bars (e.g. Barth et al. 1995). In the latter case, the star cluster formation probably is triggered by resonances enhancing the gas density. With respect to the host galaxy luminosity, luminous BCGs like ESO338-IG04, ESO350-IG38 and ESO185-IG13 contain higher numbers of compact star clusters than the classical mergers (Östlin 2000). A nearly coeval formation of a hundred globular clusters requires the build up of massive dense gas clouds on a time-scale of a few 10 Myrs. In dwarfs, where resonances that could help build-up large overdensities, are absent, mergers may be the only way to create the necessary conditions for cluster formation to occur in great numbers (see e.g. Elmegreen & Efremov 1997). Hence the richness of young massive star clusters in three of our galaxies is another support for the merger origin of their starbursts.
Given that the high SFRs seen in these luminous BCGs are transient, one might ask what came before, and what will follow: in between the burst phases a BCG will have a very different appeareance. It is however hazardeous to draw any general conclusions about the BCG pehomenon from such a small sample of luminous BCGs. The morphological diversity, notably the existence of BCGs with regular vs. irregular outer envelopes, suggests the possibility that different types of BCGs may have a different origin (Kunth & Östlin 1999).
The possible evolutionary links between BCGs and other types of galaxies have been widely discussed in the litterature (see e.g. Davies & Phillips 1988; Dekel & Silk 1986; Lin & Faber 1983; Thuan 1992; Babul & Rees 1992; Papaderos et al. 1996; Bergvall et al. 1999a, 1999b; Kunth & Östlin 1999). For the luminous BCGs in this study, a merger origin is favoured for most of the galaxies. The merger should include at least one gas-rich dwarf or alternatively massive intergalactic H I-cloud. This suggests that at least one dwarf irregular or LSBG is involved. LSBGs may be the most common type of galaxy in the universe, thus providing plenty of fuel for BCG activity. Mergers between spirals are believed to produce elliptical galaxies. Since several of our galaxies seem to rotate too slowly, one might speculate that they will evolve into low mass elliptical galaxies. However, to answer the question about the fate of these BCGs, information on stellar dynamics is required.
Studies of faint galaxies, e.g. in the Hubble Deep Field (HDF), has revealed the presence of a population of high red-shift (z = 0.4 to 1.4) compact emission line galaxies (Phillips et al. 1997), which may be responsible for up to 45% of the global cosmic SFR at 0.4 < z < 1.0(Guzmán et al. 1997). Comparing the mass averaged SFRs, line widths, line ratios etc., of our local BCGs to those derived for the compact galaxies in the HDF (Guzmán et al. 1997), we find them to have very similar properties, in particular when compared to the high (z>0.7) red-shift subsample. Glazebrook et al. (1995, 1998) identified the excess of faint blue galaxies in deep galaxy counts with peculiar/interacting galaxies, and showed that such galaxies are blue, compact and actively star-forming, with typical luminosities MB = -19, i.e. very similar to the galaxies discussed in this paper. In hierarchial galaxy formation scenarios, massive galaxies are succesively built up by galaxies of lower mass. Le Fèvre et al. (2000) show that mergers play a major role in galaxy evolution also at z<1. Thus, the local luminous BCGs may be the present day equivalents (NB not counterparts) of high red-shift BCGs and faint blue galaxies. The study of local luminous BCGs offer insights to the mechanisms operating at higher red-shifts and offers a way of studying hierarchial buildup in action.
Copyright ESO 2001