5 H I and radio continuum of PSB and active galaxies

The physical mechanisms which trigger and quench the star formation activity, and the question if the star burst exhausts the H I reservoir is still a matter of debate. We address this problem pointing at 14 galaxies in Coma for which abnormal spectra are reported by Caldwell et al. (1993, 1997), in order to get H I and radio continuum information. The radio continuum at 1.4 GHz could be due to synchrotron emission from relativistic electrons accelerated by supernovae hence probing star formation, or it could come from AGN. At intermediate redshift Smail et al. (1999) detect in the cluster CL0939+4713 (z=0.41) several post starburst galaxies in radio continuum and these authors argue that the PSBs are most likely dust enshrouded starbursts. In the local universe Chang et al. (2001) observed in H I five of the spectroscopically selected sample of E+A galaxies (Zabludoff et al. 1996). They detected one in H I  and none in radio continuum. The implied low star formation rates and limits on the SFR for these galaxies rule out that any of these is in fact a dust enshrouded starburst.

Eleven of the 14 abnormal spectrum galaxies we pointed at in Coma are defined as PSBs because of their spectra with strong absorption lines and no emission at all; the three remaining are star forming systems (SBs), as indicated by their emission line spectrum superposed to absorption lines (C93). Their position relative to the X-ray emission is shown in Paper I (Fig. 6). We detected none of these objects in H I, and report very low H I mass upper limits between 3 and $7 \times 10^{7}$ $M_{\odot}$ (see Table 2). We do detect weak radio continuum emission in three PSB (IC3949, Mrk060 and RB042) and one SB galaxy (NGC4853, see Table 2). Our radio continuum levels in Coma imply even lower limits to the star formation rates than found by Miller & Owen (2001) and Chang et al. (2001) for the field E+A's. Thus in Coma, contrary to in CL 0939+4713, the post starburst galaxies have at best only very modest levels of star formation. Chang et al. (2001) detected one of five E+A's in H I, while our H I upper limits for the cluster PSBs are almost two orders of magnitude below this detection. This may indicate a real difference between cluster and field post starburst galaxies. Whatever triggers the starburst, whether collisions or ICM interaction, in a cluster the gas is cleaned out after the starburst, something that would not happen in the field. We show that the post starburst galaxies in Coma have very little, if any, H I.

Table 2 gives in the upper part the H I mass upper limits and radio continuum parameters for the sample of abnormal spectrum galaxies that we observed. For comparison, we show in the bottom the same results for the blue disks. RB042 is projected very close to the cD galaxy NGC4874 and was only marginally detected in radio. Three of four galaxies are new detections, the fourth, NGC4853, is catalogued in the FIRST Survey (Becker et al. 1995). Table 2 gives in Cols. 1 and 2 the galaxy identification, Col. 3 gives the morphological and spectral classification, following C93. In Col. 4 we list the H I upper limits, in Col. 5 the radio continuum flux, Col. 6 lists the rms noise in the continuum images. In Col. 7 we give the radio power, with upper limits set at 3 sigma. The associated SFR, listed in Col. 8, is calculated following Yun et al. (2001): ${\it SFR}_{1.4}=L_{1.4}$[ $5.9 \times 10^{\rm -22}$ WHz^-1]$M_{\odot}$yr^-1. In Col. 9 we give the relation between the SFR and the blue luminosity ($L_{\rm B}$) obtained from the blue total magnitude, using the Mean Data from the LEDA database for homogeneity.

Table 2 shows that the SFR in the PSBs is well below 1 $M_{\odot}$yr^-1. It comes as somewhat of a surprise that also the SB galaxies have SFR less than 1 $M_{\odot}$yr^-1. This contrasts sharply with the blue disk galaxies reported by Bothun & Dressler (1986), which were all detected in continuum (see Table 2) and for which the associated SFR are from 1.0 to 5.5 $M_{\odot}$yr^-1. We note that these rates are in good agreement with the rates reported from the H${\alpha}$ equivalent widths by Bothun & Dressler (1986). Thus the blue "starforming" disks have a significantly higher SFR than the "starbursting" galaxies identified by C93. This is also valid if we consider the factor SFR/$L_{\rm B}$ given in Col. 9 of Table 2, which is probably a better criteria than merely SFR if the sampled galaxies span a large range in size and luminosity. There is a consistent trend between Cols. 8 and 9, with blue disks showing systematically higher values of SFR/$L_{\rm B}$ than galaxies from C93. The explanation is simple: C93 only selected early type galaxies for their sample of abnormal spectrum galaxies. The emission lines typically have narrow H$\delta$ lines ($\leq$3 ${\rm\AA}$), while the blue disks have 3 ${\rm\AA}$ $\leq$ H$\delta$ $\leq$ 7 ${\rm\AA}$. Thus the SB in the C93 sample means a higher than usual SFR for an early type galaxy; the absolute levels are very low. It is interesting that the C93 sample is spatially distinct from the blue disks. The latter are located at the edge of the X-ray emission, while most of the C93 galaxies are in the cluster core or in the zone between the core and the SW group. The difference in environment may well account for the different star formation properties of these galaxies.

HST images (Caldwell et al. 1999) show that some of the PSB galaxies in Coma have retained their disk, even with a smooth spiral structure. D61 (Zw 160-104) shows two dust lanes forming an edge-on disk, and D216 (RB 160) shows clear spiral structure. Interestingly our limits to the SFR and H I content for these two galaxies are very low. Obviously the time scale for gas removal from these galaxies was shorter than for a morphological transformation. Caldwell et al. (1999) suggest that star formation activity in the PSBs of Coma could be triggered by galaxy harassment and by gravitational perturbations between the main cluster and the SW group. However, Moore et al. (1996) estimate a time-scale of ${\sim}$3 Gyr to produce considerable changes in the optical morphology by this process. This is too long for galaxies like D61 and D216 to explain the whole process of triggering a starburst, the loss of a big fraction of their gas reservoir, and the subsequent quenching of the activity. An alternative process to explain this scenario is the re-accretion of gas mass after the stripping, as the infalling gas colliding with the clouds remaining in the disk can trigger star formation within the disk. In a cluster like Virgo this process starts at ${\sim}$ $2 \times 10^8$ years after the closest passage of the cluster core, and ends at ${\sim}$ $5 \times 10^8$ years (Vollmer et al. 2001b). The same process in Coma should develop with at least the same time-scales (or shorter, if we consider the more hostile ICM conditions), suggesting that the interaction with the ICM and gas re-accretion may account for this scenario. This is reinforced by the fact that galaxies could reach their present condition during a single pass through the cluster core, on a time scale $\leq$1 Gyr, which is much shorter than the time needed by galaxy harassment to modify the optical morphology. The cases of the PSBs D61 and D216 support this scenario: the former presents a very young starburst age of 0.5 Gyr, and the latter, recently reclassified as a spiral (Caldwell et al. 1999), should have harbored a significant amount of gas in the recent past. Both need a faster process than galaxy harassment to account for the sudden removal of the H I reservoir and quenching of star forming activity.

The gas which has been accelerated to values below the escape velocity will accrete back to the galaxy at time scales of 2- $7 \times 10^8$ years (Vollmer et al. 2001b). As a first approach in Coma we apply a conservative value of $5 \times 10^8$ years for the re-accretion time scale, i.e. the same value found by Vollmer et al. in Virgo. If we consider a typical galaxy moving across the Coma cluster at $v_{\rm rel}$ ${\sim}$ 1000 kms^-1 (equivalent to the velocity dispersion) it will travel some 0.5 Mpc before the gas accretes back on to the disk and triggers a burst of star formation. It is very interesting to associate this with the fact that a considerable fraction (38%) of the star forming UV flux in Coma is produced in an annular region between 20$^{\prime}$ and 30$^{\prime}$, or ${\sim}$0.5 Mpc from the cluster center (Donas et al. 1995), and that it is in this zone where most of the blue galaxies in Coma are located (Paper I). Taking into account their position relative to the cluster, NGC4848 and most of the blue disk galaxies in Coma have had enough time after passing across the cluster core to re-accrete gas clouds and boost a major star formation event; this mechanism may in part account for the blue annulus observed in Coma which is also observed in high redshift clusters (Oemler 1992).