The ALMA REBELS Survey. Epoch of Reionization giants: Properties of dusty galaxies at z ≈ 7

We analyse FIR dust continuum measurements for 14 galaxies (redshift z ≈ 7) in the ALMA Reionization Era Bright Emission Line Survey (REBELS) Large Program to derive their physical properties. Our model uses three input data, i.e. (a) the UV spectral slope, β, (b) the observed UV continuum flux at 1500 Å, F₁₅₀₀, (c) the observed continuum flux at $\approx 158\, \mu$m, F₁₅₈, and considers Milky Way (MW) and SMC extinction curves, along with different dust geometries. We find that REBELS galaxies have 28-90.5 per cent of their star formation obscured; the total (UV+IR) star formation rates are in the range $31.5 \lt {\rm SFR}/({\rm M}_\odot \, {\rm yr}^{-1}) \lt 129.5$. The sample-averaged dust mass and temperature are $(1.3\pm 1.1)\times 10^7 \, \mathrm{M}_\odot$ and 52 ± 11 K, respectively. However, in some galaxies dust is particularly abundant (REBELS-14, $M^{\prime }_{\rm d} \approx 3.4 \times 10^7 \, \mathrm{M}_\odot$), or hot (REBELS-18, $T^{\prime }_{\rm d} \approx 67$ K). The dust distribution is compact (<0.3 kpc for 70 per cent of the galaxies). The inferred dust yield per supernova is $0.1 \le y_{\rm d}/\, \mathrm{M}_\odot \le 3.3$, with 70 per cent of the galaxies requiring $y_{\rm d} \lt 0.25 \, \mathrm{M}_\odot$. Three galaxies (REBELS-12, 14, 39) require $y_{\rm d} \gt 1 \, \mathrm{M}_\odot$, which is likely inconsistent with pure SN production, and might require dust growth via accretion of heavy elements from the interstellar medium. With the SFR predicted by the model and a MW extinction curve, REBELS galaxies detected in [C II] nicely follow the local L_CII-SFR relation, and are approximately located on the Kennicutt-Schmidt relation. The sample-averaged gas depletion time is $0.11\, y_{\rm P}^{-2}$ Gyr, where y_P is the ratio of the gas-to-stellar distribution radius. For some systems, a solution simultaneously matching the observed (β, F₁₅₀₀, F₁₅₈) values cannot be found. This occurs when the index I_m = (F₁₅₈/F₁₅₀₀)/(β - β_int), where β_int is the intrinsic UV slope, exceeds $I_m^{*}\approx 1120$ for an MW curve. For these objects, we argue that the FIR and UV emitting regions are not co-spatial, questioning the use of the IRX-β relation.