2023MNRAS.525.5720J

Panchromatic analysis of galaxy spectral energy distributions, spanning from the ultraviolet to the far-infrared, probes not only the stellar population but also the properties of interstellar dust through its extinction and long-wavelength re-emission. However, little work has exploited the full power of such fitting to constrain the redshift evolution of dust temperature in galaxies. To do so, we simultaneously fit ultraviolet, optical, and infrared observations of stacked galaxy subsamples at a range of stellar masses and photometric redshifts at 0 < z < 5, using an energy-balance formalism. However, we find UV-emission beyond the Lyman limit in some photometric redshift selected galaxy subsamples, giving rise to the possibility of contaminated observations. We carefully define a robust clean subsample which extends to no further than z ∼ 2. This has consistently lower derived temperatures by $4.0^{+5.0}_{-1.9}$ K, relative to the full sample. We find a linear increase in dust temperature with redshift, with T_d (z) = (4.8 ± 1.5) × z + (26.2 ± 1.5) K. Our inferred temperature evolution is consistent with a modest rise in dust temperature with redshift, but inconsistent with some previous analyses. We also find a majority of photometrically selected subsamples at z > 4.5 underpredict the IR emission while giving reasonable fits to the UV-optical. This could be due to a spatial disconnect in the locations of the UV and IR emission peaks, suggesting that an energy-balance formalism may not always be applicable in the distant Universe.