2011MNRAS.414..781M

Observations have revealed unexpectedly large amounts of dust in high-redshift galaxies and its origin is still much debated. Valiante et al. suggested the net stellar dust production of the quasar host galaxy SDSS J1148+5251 may be sufficient to explain the large dust mass detected in this galaxy, albeit under some very special assumptions (e.g. `closed-box' evolution and a rather high gas mass). Here it is shown that since accretion of essentially pristine material may lower the efficiency of dust formation significantly, and the observationally derived dust-to-gas ratios for these high-redshift galaxies are remarkably high, stellar dust production is likely insufficient. A model including metallicity-dependent, non-stellar dust formation (`secondary dust') is presented. The required contribution from this non-stellar dust component appears too large, however. If all observational constraints are to be met, the resultant dust-to-metals ratio is close to unity, which means that almost all interstellar metals exist in the form of dust. This is a very unlikely situation and suggests the large dust-to-gas ratios at high redshifts may be due to observational uncertainties and/or incorrect calibration of conversion factors for gas and dust tracers. Kepler observations of Am stars