A supernova factory in Mrk 273?

We report on 1.6- and 5.0-GHz observations of the ultraluminous infrared galaxy (ULIRG) Mrk 273, using the European VLBI Network and the Multi-Element Radio-Linked Interferometer Network (MERLIN). We also make use of published 1.4-GHz Very Long Baseline Array observations of Mrk 273 by Carilli & Taylor (2000). Our 5-GHz images have a maximum resolution of 5-10mas, which corresponds to linear resolutions of 3.5-7pc at the distance of Mrk 273, and are the most sensitive high-resolution radio observations yet made of this ULIRG. Component N1, often pinpointed as a possible active galactic nucleus, displays a steep spectral index (α= 1.2 +/- 0.1; S_ν~ν^-α) hence it is very difficult to reconcile with N1 being an active galactic nucleus (AGN), and rather suggests that the compact non-thermal radio emission is produced by an extremely high luminous individual radio supernova (RSN), or a combination of unresolved emission from nested supernova remnants (SNRs), luminous RSNe, or both. Component N2 is partly resolved out into several compact radio sources - none of which clearly dominates - and a region of extended emission about 30pc in size. The integrated spectral index of this region is flat (α= 0.15 +/- 0.1), which can be interpreted as due to a superposition of several unresolved components, e.g. RSNe or SNRs, whose radio emission peaks at different frequencies and is partly free-free absorbed. Is it also possible that one of the compact components detected in this region is the radio counterpart of the AGN. The overall extended radio emission from component N is typical of non-thermal, optically thin radio emission (α= 0.8 +/- 0.1), and its 1.4-GHz luminosity (L_1.4 GHz = (2.2 +/- 0.1) × 10²³ W Hz^-1) is consistent with being produced by relativistic electrons diffused away from SNRs in an outburst. The southern component, SE, shows also a very steep spectrum (α= 1.4 +/- 0.2), and extended radio emission whose origin and physical interpretation is not straightforward.