3.6 Luminosity segregation
3.6.4 The cluster core
Several of these plots suggest that in the very center of the cluster,r <0.02R200, some of the observed trends reverse: the faint-to-luminous ratio is as low as in the field, and the typical mass of (red) galaxies increases by ∼ 60%. This is very intriguing in light of a prediction from the modeling by De Lucia et al. (2006): they combined the Millennium Simulation with a suite of semi-analytical models and found that the median mass of elliptical galaxies increases in the core of the cluster (Fig. 3.18). Their model implies that the stellar mass of ellipticals in the very core (r <0.02R200) is a factor of 6 larger than at the cluster outskirts. In their second bin, the stellar mass is approximately twice as large as at the cluster outskirts.
Several factors hinder us to directly associate our result with the prediction of De Lucia et al.:
• The prediction is for elliptical galaxies, whereas we have investigated red galaxies. These are not necessarily the same.
• As we have argued before, photometry is difficult in the vicinity of the BCG. This may work in two ways: on the one hand, faint (low-mass) galaxies are missed, either due to
0.01 0.1 1
Figure 3.17: The average mass of all galaxies (black points), red galaxies (red) and blue+green galaxies (blue), computed from galaxies above the mass limit of log(M⋆/M⊙)>9.6.
Figure 3.18: Fig. 8d) from De Lucia et al. (2006), depicting the median stellar mass of model elliptical galaxies in 51 clusters of the Millennium Simulation.
3.6 Luminosity segregation -2.0 < log(dist) <-1.8 -1.8 < log(dist) <-1.6 -1.6 < log(dist) <-1.4 -1.4 < log(dist) <-1.2 -1.2 < log(dist) <-1.0 -1.0 < log(dist) <-0.8 -0.8 < log(dist) <-0.6 -0.4 < log(dist) <-0.6 -0.2 < log(dist) <-0.4 0.0 < log(dist) <-0.2 10 10.5 11 11.5 0 0.1 0.2 0.3 0.4 0.5
Figure 3.19: The mass distributions of red galaxies in bins of distance from the BCG. The black, shaded histogram shows the distribution in the central bin, with √N error bars. The colored his- tograms show the distributions in the other radial bins (those bins at distances greater than 10−0.4R
200
are shown as dashed lines).
increased noise, or in fact, they might be “hidden” by the BCG. On the other hand, luminosities and stellar masses of BCG neighbors are likely underestimated (Sect. 2.2.2). Furthermore, the SDSS pipeline may “shred” large galaxies into multiple detections, thus creating “false detections”. We have verified by visual inspection that due to the rather bright magnitude limit, this is not a major influence on our sample - virtually all objects are individual galaxies, or secondary nuclei to the BCG.
• The dip in the faint-to-luminous ratio is almost entirely caused by the clusters atz >0.07 (Fig. 3.14). With increasing redshift, the BCG neighbors are at smaller angular separation from the BCG, i.e. they are more likely to be in the same “box” in which the local sky is estimated. This could cause fainter galaxies to miss the detection limit. On the other hand, this also causes the luminosity to be underestimated, and it is not clear which effect is stronger. A similar argumentation applies to the observation that the “dip” is stronger in clusters with lower velocity dispersions (where distances in units ofR200correspond to smaller physical distances).
It is difficult to estimate how much these points can contribute to masking the effect predicted by De Lucia et al., or whether they could create a false detection of such a trend. However, Fig. 3.11 shows an excess of galaxies in the central bin at all but the faintest magnitudes (where the number is compatible with the best-fit NFW). If we are indeed missing a substantial fraction of galaxies in this bin, then the excess of galaxies must be even stronger than observed.
Another clue to the reality of this result lies in the distribution of stellar mass within each bin in radial distance. This is shown in Fig. 3.19, which suggests that the mass distribution in the central bin is systematically different than in the other bins: there is both a deficiency of galaxies with masses lower than log(M⋆/M⊙)<10.4, and an excess of galaxies with masses
larger than log(M⋆/M⊙)>10.8. Thus, unless one argues that the shortcomings in the SDSS
photometric pipeline can cause an over-/underestimation of stellar masses of the order of 0.5 dex, or in fact, miss galaxies with masses of log(M⋆/M⊙) ∼ 10.2 altogether, this result
can be taken as an indication that the mass distribution in the very cluster core is indeed different than at larger radii.
A possible cause for this is dynamical friction. As Fig. 3.11 indicates, there is an excess of bright (red) galaxies of galaxies compared to the best-fit NFW profile. Dynamical fric- tion causes the more massive galaxies to sink to the cluster center. If they do not merge immediately with the BCG, then indeed an overdensity of massive galaxies is observed in the cluster center. The galaxy excess does not imply a steeper dark matter profile than NFW in the center. On a different scale, Goerdt et al. (2006) have argued that the presence of several globular clusters in the center of the Fornax dwarf spheroidal (i.e. an excess of test particles) suggests a cored mass profile, since the merging time for these in objects in a cuspy mass profile would be substantially shorter. If similar processes are at work on the much larger scales of galaxy clusters, then the excess of bright galaxies could indicate a flattening of the mass profile. Some strong lensing studies have indeed argued in favor of a cored mass profile in clusters. E.g. Rzepecki et al. (2007) find a core radius of 0.025R200 in the cluster RCS0224-0002, which is comparable to the scale where our data show a galaxy excess.
3.7 Summary and discussion
We have found that, on average, the number density profile of clusters is well described by an NFW profile. This applies to radii & 0.05R200, at smaller radii, there is a slight excess of galaxies. The distribution of blue galaxies is significantly more extended than that of red galaxies, which is a result of the color–density relation. For the red galaxies (and thus also for the overall galaxy population), the distribution of galaxies fainter than Mr0.1 >−20.5 is more concentrated than that of galaxies brighter than this. This is also apparent in the ratio of faint-to-luminous red galaxies, which decreases with increasing cluster radius. However, the average stellar mass of galaxies remains approximately constant. We have argued that this is predominantly due to the quenching of star formation in cluster galaxies, along with a subsequent fading and transition to the red sequence. Apart from the innermost bin, the galaxy mass distribution does not change significantly. This indicates that the process(es) which terminate star formation in cluster galaxies do not significantly alter their stellar mass. The excludes mergers from playing a dominant role within R200. If disruptive process such as tidal disruption play a role, then only at masses below .109.6M⊙ .
The mass distribution in the core, on the other hand, differs from that at other radii, in that there are more massive galaxies. With increasing mass, dynamical friction causes galaxies sink to the cluster center on shorter timescales. But if they were to merge on a short timescale with the BCG, then no galaxy excess would be seen. A flattening of the central slope of the mass profile could increase the merger timescale in the very center, so that a galaxy excess would be seen.
4
Population gradients in local clusters
In this chapter, I investigate the recent and current star formation activity of galaxies as function of distance from the cluster center, in the sample of SDSS clusters. To characterize the recent star formation history, I use the Principal Component Analysis of Wild et al. (2007), which is sensitive to the star formation history over the last∼2Gyr. I find a marked star formation–radius relation in that most galaxies in the cluster core are quiescent, i.e. have terminated star formation a few Gyr ago. This star formation–radius relation is most pronounced for low-mass galaxies. The fraction of galaxies with young stellar populations which have spectra indicative of a recent starburst, or truncation of star formation, is constant with cluster radius. The typical star formation rate of non-quiescent galaxies declines by approximately a factor of 2 towards the cluster center. These results are consistent with a scenario in which star formation is quenched on timescales similar to the cluster crossing time, i.e. a few Gyr. I furthermore find that the fraction of galaxies which host optical AGN declines towards the cluster center, largely due to a decline of AGN activity in quiescent galaxies. This suggests that the processes which trigger optical AGN activity are subject to similar environmental influences as star formation.4.1 Radial cluster profiles as density probes
In galaxy clusters, the star formation–density relation translates to a star formation–radius relation, since the density decreases with increasing (projected) clustercentric radius. Mea- surements of local density, such as the distance to the 5th neighbor have the disadvantage of strong shot noise, as well as selection and projection effects. In this sense, radial dis- tance presents a cleaner measurement. However, individual clusters are rarely axisymmetric systems, and thus the true strength of this method is investigate radial profiles in composite clusters. We have shown in Sect. 3.5 that the number density profiles of our composite cluster is well described by an NFW profile, and hence radial distance is directly related to average density. Furthermore, stacking individual clusters to one composite cluster allows for vastly improved number statistics than possible with any individual cluster.
Nevertheless, studying radial profiles of composite clusters has some obvious disadvantages. On the one hand, it neglects the influence of subclustering. A considerable fraction of galaxies arrive into the cluster as part of an infalling group of galaxies. The group environment may play an important role to “preprocessing” the galaxies (Zabludoff & Mulchaey 1998). Also, the properties of individual clusters differ significantly from the average cluster properties; even clusters with similar masses may have very different galaxy populations (Moran et al. 2007). This scatter seems to be mostly stochastic in nature: Goto (2005a) find that there is no systematic trend of varying galaxy populations with cluster mass. Furthermore, the radial distance is only the projected distance along the line-of-sight, i.e. for individual galaxies, it is only a lower limit to the true distance. This observational limitation necessarily dilutes the three-dimensional trend.
Clustercentric distance has a second interpretation apart from a local density measure: it also relates to the time since infall into the cluster (Gao et al. 2004). Of the galaxies at or beyond the virial radius, a considerable fraction has not (yet) experienced the dense cluster environment. Galaxies at the cluster core, on the other hand, were either born in a dense environment, or they must have traversed the cluster from the outskirts to the center at least once. The cluster crossing time is of the order of R200/σv, i.e. 2.5 Gyr for a cluster with σv ∼400 km/s, andR200∼1 Mpc. Therefore, clustercentric distance is also an approximate timescale, and is sensitive to processes occurring on timescales of the order of Gyr. This timescale is particularly interesting for the processes which have been proposed to quench star formation in galaxies entering the cluster environment (Sect. 1.5.5). Processes which quench star formation on similar timescales, such as strangulation or harassment, would induce gradients over the complete radial range. Processes which act on shorter timescales are likely to effect distinctive signatures at radii where they are most effective - e.g. ram- pressure stripping could be detectable as an increase in the (post-)starburst rate, presumably mostly in the cluster center, where the gas densities are highest. Mergers could induce (post-) starbursts primarily at outer cluster radii, where the relative galaxy velocities allow frequent mergers.
Using[Oii] as a tracer of star formation, Balogh et al. (1997) showed that the star forma- tion rate declines gradually in cluster galaxies between 2 and 0.3R200. Also, they find that a lower fraction of cluster galaxies displays strong emission lines indicative of a starburst than field galaxies. They conclude that the quenching of star formation is not accompanied by a brief enhancement of star formation, and that it is likely to proceed gradually, rather than instantaneously. In the same cluster sample, Balogh et al. (1999) find no enhancement of (post-)starburst galaxies, neither relative to the intermediate-redshift field, nor to low-redshift environments. This supports the conclusion of their previous work, that star formation is suppressed gradually, rather than truncated, and that a likely mechanism is strangulation. Interestingly, the results of Dressler et al. (1999) on the incidence of (post-)starburst galax- ies are markedly different. In a different intermediate-redshift cluster sample, Dressler et al. found that “k+a” galaxies, which they take to be post-starburst galaxies, are more frequent in intermediate-redshift clusters than the field. They find that these galaxies avoid the cluster center, but are more concentrated than the actively star-forming galaxies. Dressler et al. and Poggianti et al. (1999) argue that the increased rate of “k+a” galaxies in clusters is due to the cluster environment triggering (brief) episodes of star formation, before truncating star formation. They suggest that the most likely process quenching star formation in infalling galaxies is ram-pressure stripping. They further argue that the enhancement of star formation is a possible cause of the Butcher-Oemler effect, i.e. the higher incidence of blue galaxies in