From simulations of the structure formation of the Universe, large accretion shocks around clusters of galaxies are expected. If magnetic fields are present at these locations, particle acceleration should take place. Protons might be accelerated to the highest observed energies, as demonstrated by Kang et al. (1997), because of their weak energy losses and their possible injection out of a thermal population. If a population of relativistic low-energy electrons is also present, they would be accelerated, too. High-energy ultra-relativistic electrons lose energy by inverse-Compton scattering and synchrotron emission. This allows to see them in the radio or X-ray regime and therefore verify the existence of the large-scale formation shock structures.
At peripheral positions with respect to the cluster center regions of diffuse radio emission are found, the so-called cluster radio relics (Figs. 1 and 2). They are assumed to be remnants of radio galaxies. Their energy supply was a long lasting outstanding problem, since their electron cooling times are usually too short to allow any of the nearby possible former parent galaxies to have moved from the relic to its present position within that time. Electron acceleration has to take place within them, but turbulent acceleration from galactic wakes fails to explain this, due to the low galaxy densities in these regions.
If these relics are close to the accretion shock, a shock efficiency of only is sufficient to power them. The accretion shock itself should collect relics since every volume filled with radio plasma, being injected into the inter galactic medium above such a shock by a radio galaxy, would be dragged into the shock by the fast motion of the infalling gas. After passing the shock, the relic remains close to it for some time, since the growth rate of the shock radius is much slower than the accretion velocity. Buoyant motion of the light radio plasma embedded in the heavier surrounding gas might help to keep it longer there.
The relic is proposed to be radio illuminated by relativistic electrons (re)accelerated at the shock and therefore tracing its position. The radio spectrum of such relics is steeper than expected for electron acceleration within strong shocks, indicating that the temperature of the infalling matter should be 0.5 - 1 keV (see Fig. 4). Thus, preheating of this gas should have taken place. The accreting gas has a typical density of . 1706+78 in A2256 of course does not fit into this temperature scheme, due to its different physical nature, being located at a merger shock front in the interior of the cluster. Matter which has fallen onto the large-scale cosmic sheets should be shock-heated to about 0.1 keV and have a density of far away of clusters. Adiabatic compression and internal shocks of the converging flow within these sheets onto clusters of galaxies should heat the gas further to the above temperatures of keV.
During the passage through the shock, the magnetic fields are amplified and aligned with the shock plane. The radio emission of a relic should therefore exhibit polarization properties depending on the viewing angle. Assuming a spherical accretion shock with a radius depending on the depth of the cluster potential, the viewing angle and the resulting polarization can be estimated (Fig. 3). These agree well with measured polarizations for relics in two clusters: 1253+275 in Coma, and 1712+64 in A2255. For 0917+75 in A786, the large projected radius would imply a deep gravitational potential and therefore a very high cluster temperature, which is not yet measured. 0917+75 might also trace an accretion shock of the stream onto the supercluster Rood #27, which A786 is a member of. This could also explain the large apparent radius of the relic.
In case of 1706+78 in A2256 the measured polarization exceeds the expected one significantly, indicating that the viewing angle is much higher than if the relic is located at the accretion shock. A consistent shock radius would be 1 Mpc , which means that an internal shock of the cluster powers the relic, which should result from a highly developed merger event. Roettiger et al. (1995) discuss several apparent signs of such an event in A2256, and fit simulations to the X-ray data. Their best-fit value of the angle between the merger axis and the line-of-sight within our theory of field compression predicts exactly the observed polarization.
The directions of the projected magnetic fields are roughly perpendicular in all above examples to the direction pointing to the cluster center, as is expected.
For the other five relics of our sample no polarization measurements are available. We predict polarization for 2006-56 in A3667, for 2010-57 in A3667, and less than 5% for 0038-096 in A85, 1140+203 in A1367, and 1401-33 in S753. Polarization measurements of these sources could verify the existence of cluster accretion shocks and help us to reveal an important piece of the cosmological structure formation puzzle.
The time elapsed since the magnetized plasma has started to pass the shock can be calculated in two independent ways. Both estimates give ages of the order of some yr, and are therefore a further justification of our model.
Since our sample contains mostly relics seen under viewing angles below , we expect that more relics might be discovered in the outer regions of clusters. Any radio search within these large areas should be guided by the expected correlation of the large-scale structure filaments and typical inflow directions into clusters.
We conclude that the properties of cluster radio relics are naturally explained if they are understood to trace some of the giant shock fronts of the cosmological large-scale motion of the on-going structure formation. This demonstrates for the first time the existence of these theoretically predicted shocks. Estimated magnetic fields, temperatures and densities of the accreting matter fit into the structure formation scenario.
© European Southern Observatory (ESO) 1998
Online publication: March 23, 1998