The effect of non-gravitational gas heating in groups and clusters of galaxies

We present a detailed study of a set of gas-dynamical simulations of galaxy groups and clusters in a flat, Lambda-cold dark matter (LambdaCDM) model with Omega(m) = 0.3, aimed at exploring the effect of non-gravitational heating on the observable properties of the intracluster medium (ICM). We use gasoline, a version of the code pkdgrav that includes a smoothed particle hydrodynamics description of hydrodynamics to simulate the formation of four haloes with virial temperatures in the range 0.5 less than or similar toT less than or similar to 8 keV. These simulations resolve the structure and properties of the ICM down to a small fraction of the virial radius, R (vir). At our resolution X-ray luminosities, L (X), of runs with gravitational heating only are in good agreement with analytical predictions, which assume a universal profile for CDM haloes, over almost two orders of magnitude in mass. For each simulated structure, non-gravitational heating of the ICM is implemented in two different ways: (i) by imposing a minimum-entropy floor, S-fl, at a given redshift, which we take in the range 1less than or equal toz less than or equal to 5; (ii) by gradually heating gas within collapsed regions, proportionally to the supernova rate expected from semi-analytical modelling of galaxy formation in haloes having mass equal to that of the simulated systems. Our main results are the following. (i) An extra heating energy E (h) greater than or similar to 1 keV per gas particle within R (vir) at z= 0 is required to reproduce the observed L-X- T relation, independent of whether it is provided in an impulsive way to create an entropy floor S (fl) = 50-100 keV cm(2), or is modulated in redshift according to the star formation rate; our supernova (SN) feedback recipe provides at most E (h) similar or equal to 1/3 keV particle(-1) and, therefore, its effect on the L (X) -T relation is too small to account for the observed L (X) -T relation. (ii) The required heating implies, in small groups with Tsimilar to 0.5 keV, a baryon fraction as low as less than or similar to40 per cent of the cosmic value at R (vir)/2; this fraction increases to about 80 per cent for a T similar or equal to 3 keV cluster. (iii) Temperature profiles are almost scale-free across the whole explored mass range, with T decreasing by a factor of 3 at the virial radius. (iv) The mass-temperature relation is almost unaffected by non-gravitational heating and follows quite closely the Mproportional toT (3/2) scaling; however, when compared with data on the M-500-T-ew relation, it has a similar to40 per cent higher normalization. This discrepancy is independent of the heating scheme adopted. The inclusion of cooling in a run of a small group steepens the central profile of the potential well while removing gas from the diffuse phase. This has the effects of increasing T-ew by similar to30 per cent, possibly reconciling the simulated and the observed M (500)- T-ew relations, and of decreasing L (X) by similar to40 per cent. However, in spite of the inclusion of SN feedback energy, almost 40 per cent of the gas drops out from the hot diffuse phase, in excess of current observational estimates of the number of cold baryons in galaxy systems. It is likely that only a combination of different heating sources (SNe and active galactic nuclei) and cooling will be able to reproduce both the L (X) -T (ew) and M (500)- T (ew) relations, as observed in groups and clusters, while balancing the cooling runaway.