Stars influence the central distribution of dark matter in galaxy clusters

The Cosmic Microwave Background provides important information on how dark matter was distributed in the early Universe. Cosmological N-body simulations can be used to follow this distribution as it evolves forward in time, ultimately giving rise to today's cosmic web, made up of voids, filaments and the halos in which the galaxies live. It is an important task to characterise, both theoretically and observationally, the internal structure of these halos, since this constrains both the nature of the dark matter particle and the way galaxies form and evolve. Already in the 1990s, cosmological N-body simulations were able to characterise the density profiles of dark matter halos, showing that, to a good approximation, these have a universal shape from the scales of dwarf galaxies to those of galaxy clusters. The physical origin of this universal profile remains a mystery to this day. An important task in modern astronomy is to infer the distribution of dark matter in galaxies in order to test this prediction of the standard LCDM paradigm for halo structure.

Galaxy clusters are objects of prime interest to study dark matter because they give astronomers the largest number of independent probes of halo structure (stellar kinematics, strong gravitational lensing, weak gravitational lensing, X-ray emission from hot gas, galaxy motions). This helps considerably in obtaining robust and precise results which can put firm constraints on total mass profiles. Recent observations of galaxy clusters and of their central galaxies (Brightest Cluster Galaxies or BCGs) have combined a number of probes, revealing that the clusters' total density profiles are well described by the "universal" profile found in cosmological dark-matter-only simulations. However, their dark matter profiles are systematically shallower in the innermost regions (well inside the visible BCG).

As gas cools and condenses near the centre of a dark matter halo and begins to form stars, simple arguments suggest the dark matter should be pulled inwards, thus steepening its density profile. While this appears to contradict the observations, this is not the full story for BCGs because their growth can be more complicated than that of more typical galaxies. It was proposed in the 1970s that BCGs may grow through multiple mergers of preformed galaxies which will occur preferentially at the centres of clusters. This suggestion seems to hold up according to current detailed simulations of the formation of galaxies and clusters in the LCDM paradigm. However, previous work did not investigate whether this picture could explain the observed structural evolution of BCGs in detail (e.g. their stellar masses, sizes, shapes, surface brightness profiles and dark matter content, all as a function of redshift). A year ago, a team of scientists at the MPA and the National Astronomical Observatories in China have provided further support for this formation channel by comparing observations at low and high redshift with sophisticated methods for "painting" the stars onto cosmological dark matter N-body simulations of galaxy cluster formation.

More recently, MPA scientists conducted N-body simulations which explicitly and self-consistently followed the evolution of both stars and dark matter in clusters. These high-resolution simulations began with a dark matter distribution consistent with LCDM expectations and a galaxy population consistent with that observed in the z~2 universe (about 3 billion years after the Big Bang) and they followed evolution down to the present day. This required a new scheme to insert equilibrium galaxies of a prescribed structure into dark matter halos that had already formed in a cosmological simulation, while mimicking the contraction of the dark-matter halos induced by baryon condensation at their centres.

While the earlier conclusions on BCG evolution held up, the new simulations showed that the central mass re-distributes itself significantly as mergers proceed. By the present day, the mixture of dark and stellar matter in the BCGs had the same total mass density profiles as in test simulations which included dark matter alone. This demonstrated that evolution tends to drive the total mass density profile (stars and dark matter) towards the "universal" shape. Since the stars contribute most of the mass near the middle of the final BCGs, this meant that their dark matter density profiles were actually less centrally concentrated than in the dark-matter-only simulations, even though they started out more concentrated in the initial galaxies. As a result, the simulated BCGs appear to have dark matter profiles consistent with those inferred observationally.

The simulated BCGs typically experienced 6 or 7 mergers which, in real galaxies, would be accompanied by a merger of the central supermassive black holes. Such mergers pump energy into the innermost regions, causing the stars and dark matter to move outwards. Estimates of the size of this effect based on the simulations suggest that it might explain the large stellar cores often observed in BCGs. So far, the effects of supermassive black holes in BCGs cannot be directly simulated in a full cosmological context, so the current simulations offer realistic initial conditions for simplified numerical studies of supermassive black hole merging in the central regions of BCGs.

This study suggests that observations of the mass distribution in the centres of galaxy clusters can be understood if BCG evolution is primarily driven by dissipationless mergers. Within the standard LCDM paradigm, such an evolutionary path naturally explains a total density profile similar to those found in dark-matter-only simulations, together with a shallower dark matter density profile. There seems no need to appeal to the more radical explanations proposed in some recent papers such as new physics in the dark matter sector or dynamical effects driven by star and black hole formation which are much more violent than any observed.

Chervin Laporte and Simon White