Estimating turbulent velocities in the elliptical galaxies NGC 5044 and NGC 5813

Context. The interstellar and intra-cluster medium (ICM) in giant elliptical galaxies and clusters of galaxies is often assumed to be in hydrostatic equilibrium. Numerical simulations, however, show that about 5-30% of the pressure in a cluster is provided by turbulence induced by, for example, the central active galactic nucleus (AGN) and merger activity. Aims. We aim to put constraints on the turbulent velocities and the turbulent pressure in the ICM of the giant elliptical galaxies NGC 5044 and NGC 5813 using XMM-Newton reflection grating spectrometer (RGS) observations. Methods. The magnitude of the turbulence is estimated using the Fe XVII lines at 15.01 angstrom, 17.05 angstrom, and 17.10 angstrom in the RGS spectra. At low turbulent velocities, the gas becomes optically thick in the 15.01 angstrom line due to resonant scattering, while the 17 angstrom lines remain optically thin. By comparing the (I-17.05 + I-17.10)/I-15.01 line ratio from RGS with simulated line ratios for different Mach numbers, the level of turbulence is constrained. The measurement is, however, limited by the systematic uncertainty in the line ratio for an optically thin plasma, which is about 20-30%. Results. We find that the (I-17.05 + I-17.10)/I-15.01 line ratio in NGC 5813 is significantly higher than in NGC 5044. This difference can be explained by a higher level of turbulence in NGC 5044. The best estimates for the turbulent velocities using resonant scattering and upper limits from the line widths, are 320 < V-turb < 720 km s(-1) for NGC 5044 and 140 < V-turb < 540 km s(-1) for NGC 5813 at the 90% confidence limit. Conclusions. The high turbulent velocities and the fraction of the turbulent pressure support of > 40% in NGC 5044, assuming isotropic turbulence, confirm that it is a highly disturbed system, probably due to an off-axis merger. The turbulent pressure support in NGC 5813 is more modest at 15-45%. The (I-17.05 + I-17.10)/I-15.01 line ratio in an optically thin plasma, calculated using AtomDB v2.0.1, is 2s above the ratio measured in NGC 5044, which cannot be explained by resonant scattering. This shows that the discrepancies between theoretical, laboratory, and astrophysical data on Fe xvii lines need to be reduced to improve the accuracy of the determination of turbulent velocities using resonant scattering.