The LOFAR two-metre sky survey deep fields: the star-formation rate–radio luminosity relation at low frequencies

In this paper, we investigate the relationship between 150 MHz luminosity and the star-formation rate – the SFR-L150 MHz relation – using 150 MHz measurements for a near-infrared selected sample of 118 517 z < 1 galaxies. New radio survey data offer compelling advantages over previous generation surveys for studying star formation in galaxies, including huge increases in sensitivity, survey speed, and resolution, while remaining impervious to extinction. The LOFAR Surveys Key Science Project is transforming our understanding of the low-frequency radio sky, with the 150 MHz data over the European Large Area Infrared Space Observatory Survey-North 1 field reaching an rms sensitivity of 20 μJy beam−1 over 10 deg2 at 6 arcsec resolution. All of the galaxies studied have SFR and stellar mass estimates that were derived from energy balance spectral energy distribution fitting using redshifts and aperture-matched forced photometry from the LOFAR Two-metre Sky Survey (LoTSS) Deep Fields data release. The impact of active galactic nuclei (AGN) is minimised by leveraging the deep ancillary data in the LoTSS data release, alongside median-likelihood methods that we demonstrate are resistant to AGN contamination. We find a linear and non-evolving SFR-L150 MHz relation, apparently consistent with expectations based on calorimetric arguments, down to the lowest SFRs < 0.01M⊙ yr−1. However, we also recover compelling evidence for stellar mass dependence in line with previous work on this topic, in the sense that higher mass galaxies have a larger 150 MHz luminosity at a given SFR, suggesting that the overall agreement with calorimetric arguments may be a coincidence. We conclude that, in the absence of AGN, 150 MHz observations can be used to measure accurate galaxy SFRs out to z = 1 at least, but it is necessary to account for stellar mass in the estimation in order to obtain 150 MHz-derived SFRs accurate to better than 0.5 dex. Our best-fit relation is log10(L150 MHz ∕W Hz−1) = (0.90 ± 0.01)log10(ψ∕M⊙ yr−1) + (0.33 ± 0.04)log10(M∕1010M⊙) + 22.22 ± 0.02