Tracing star formation with non-thermal radio emission

A key for understanding the evolution of galaxies and in particular their star formation history will be future ultra-deep radio surveys. While star formation rates (SFRs) are regularly estimated with phenomenological formulas based on the local FIR-radio correlation, we present here a physically motivated model to relate star formation with radio fluxes. Such a relation holds only in frequency ranges where the flux is dominated by synchrotron emission, as this radiation originates from cosmic rays produced in supernova remnants, therefore reflecting recent star formation. At low frequencies synchrotron emission can be absorbed by the free-free mechanism. This suppression becomes stronger with increasing number density of the gas, more precisely of the free electrons. We estimate the critical observing frequency below which radio emission is not tracing the SFR, and use the three well-studied local galaxies M 51, M 82, and Arp 220 as test cases for our model. If the observed galaxy is at high redshift, this critical frequency moves along with other spectral features to lower values in the observing frame. In the absence of systematic evolutionary effects, one would therefore expect that the method can be applied at lower observing frequencies for high redshift observations. However, in case of a strong increase of the typical gas column densities towards high redshift, the increasing free-free absorption may erase the star formation signatures at low frequencies. At high radio frequencies both, free-free emission and the thermal bump, can dominate the spectrum, also limiting the applicability of this method.