2010MNRAS.407..263A

The analysis of a sample of 52 clusters with precise and hypothesis-parsimonious measurements of mass, derived from caustics based on about 208 member velocities per cluster on average, shows that low-mass clusters and groups are not simple scaled-down versions of their massive cousins in terms of stellar content: lighter clusters have more stars per unit cluster mass. The same analysis also shows that the stellar content of clusters and groups displays an intrinsic spread at a given cluster mass, i.e. clusters are not similar to each other in the amount of stars they contain, not even at a fixed cluster mass. The stellar mass fraction depends on halo mass with (logarithmic) slope -0.55 ±0.08 and with 0.15±0.02 dex of intrinsic scatter at a fixed cluster mass. These results are confirmed by adopting masses derived from velocity dispersion. The intrinsic scatter at a fixed cluster mass we determine for gas mass fractions taken from literature is smaller, 0.06±0.01 dex. The intrinsic scatter in both the stellar and gas mass fractions is a distinctive signature that individual regions from which clusters and groups collected matter, a few tens of Mpc wide, are not yet representative of the mean gas and baryon content of the Universe. The observed stellar mass fraction values are in marked disagreement with gasdynamics simulations with cooling and star formation of clusters and groups. Instead, the amplitude and cluster mass dependency of observed stellar mass fractions are those required not to need any active galactic nuclei (AGN) feedback to describe gas and stellar mass fractions and X-ray scale relations in simple semi-analytic cluster models. By adding stellar and gas masses and accounting for the intrinsic variance of both quantities, we found that the baryon fraction is fairly constant for clusters and groups with masses between 10^13.7 and 10^15.0 M_☉ and it is offset from the WMAP-derived value by about 6σ. The offset is unlikely to be due to an underestimate of the stellar mass fraction, and could be related to the possible non-universality of the baryon fraction, pointed out by our measurements of the intrinsic scatter. Our analysis is the first that does not assume that clusters are identically equal at a given halo mass and it is also more accurate in many aspects. The data and code used for the stochastic computation are distributed with the paper.