THE ULTRAVIOLET-TO-RADIO CONTINUUM OF THE ULTRALUMINOUS GALAXY IRAS F10214+4724

We present new observations and theoretical models for the continuous ultraviolet-to-radio spectrum of the ultraluminous protogalaxy candidate IRAS F10214 + 4724. The radio spectrum at 1.49-8.44 GHz (consistent with a power law of slope 0.9 +/- 0.1) and the resolved nature of the radio source are compatible with a synchrotron origin of the radio radiation associated with a luminous starburst. The object has an exceptionally high ratio of far-infrared to radio emission, however, and this could imply that the starburst is unusually young. The submillimetre and far-infrared spectrum is modelled in terms of emission from both spherically symmetric dust clouds and an axially symmetric flared disc. A spherically symmetric starburst model proposed by Rowan-Robinson et al. (their model B) gives a good fit to the far-infrared and submillimetre data. The bolometric luminosity in this model is 4.7 x 10(14) L. (OMEGA0 = 1, H0 = 50 km s-1 Mpc-1). The model implies a high optical depth in dust (A(V) =20), and the observed optical and near-infrared continua would have no connection with the illuminating source. Axially symmetric models with high optical depth in the symmetry plane and low optical depth towards the poles can give an overall spectrum similar to that observed, provided our line of sight is suitably chosen. The mass of dust required is consistent with the molecular hydrogen mass inferred from the recently detected CO J = 3-2 line for a heavy element abundance approximately 0.001. The observed optical to near-infrared continuum derived from WHT-FOS spectroscopy and from UKIRT observations is interpreted in terms of reddened QSO and starburst models. This continuum is broadly consistent with emission from a QSO reddened by 1.5 mag of dust. The 2200-angstrom feature is absent, however, and, if an SMC extinction curve is used, the detailed fit to a reddened QSO model is poor. Moreover, the total bolometric power at 10(14)-10(18) Hz from such a quasar model would amount to only 9 per cent of the total far-infrared and submillimetre output. A better fit is obtained with an unreddened 9 x 10(8) yr old starburst model of the type proposed for high-redshift radio galaxies by Chambers & Charlot. In neither case can we be directly seeing the illuminating source for the far-infrared power, and this accords with the discovery that the optical continuum is strongly polarized. The quasar model would require a reflecting screen of dust or gas which scatters 9 per cent of the total power towards us. In the case of the starburst model, the absence of reddening implies that the reflecting screen would have to be gas. Thus the radio, submillimetre, far-infrared and CO observations of IRAS F10214 + 4724 are consistent with a powerful starburst early in the life of a massive galaxy, i.e. we are seeing a protogalaxy. The strong polarization and high-excitation emission lines, on the other hand, suggest the presence of an embedded quasar. There is no evidence, however, that this provides much of the total power, and there are difficulties in modelling the optical-UV continuum with a quasar model.