The dusty high redshift universe: the dust content and its effects in the first galaxies

The advent of the new Atacama Large Millimeter/submillimeter Array (ALMA) has opened a new window onto the high redshift Universe, shedding light on the cold interstellar medium (ISM) of normal star forming galaxies at redshift z > 5 [Capak et al., 2015, Watson et al., 2015, Knudsen et al., 2017, Barisic et al., 2017, Laporte et al., 2017b]. The information collected so far through observations that map the rest-frame emission in the ultraviolet (UV) and infrared (IR) have started to paint a complex picture: while the ALMA view of the Hubble Ultra Deep Field (HUDF) has detected the most massive star forming galaxies [Dunlop, 2016], with only one source at z > 3.5, reflecting the rapid drop-out of high-mass galaxies with increasing redshift, these sources may be simply the tip of the iceberg of a larger population of fainter dusty systems. These systems are very interesting as their star formation rates are comparable to those of UV selected galaxies. The comparison between faint dusty galaxies and the unobscured population may be key to understand the factors that determine the dust content in galaxies with comparable properties. Faint dusty star forming galaxies are difficult to detect, particularly at high redshift, and the only two sources that have been detected in their rest-frame IR continuum at z > 6 so far are gravitationally lensed: A1689-zD1, a magnified galaxy at redshift around 7.5 with an estimated dust mass of the order of 107M [Watson et al., 2015, Knudsen et al., 2017], and the galaxy A2744 YD4 with z = 8.38 identified in the ALMA Frontier Fields, with an estimated dust mass of 6 times 10^6 solar masses [Laporte et al., 2017a]. These observations have shown that ALMA has the potential to detect dust emission at z > 6 and that future observations in conjunction with the upcoming James Webb Space Telescope will be able to trace the onset of chemical enrichment and the emergence of dust in the Universe. In this original work, we have attempted to improve our understanding of the dust content and its effects in z > 5 galaxies. To accomplish this goal, we have combined the information provided by multi-wavelength observations of high redshift galaxies with the results of cosmological hydrodynamical simulations [Maio et al., 2010, Dayal et al., 2014] coupled with a state-of-the-art chemical evolution model with dust [Valiante et al., 2009, de Bennassuti et al., 2014]. This semi-numerical model allows us to account for both dust production from stellar sources (Supernovae and Asymptotic Giant Branch stars) and for dust reprocessing in the ISM, including dust destruction in interstellar shock waves and grain growth in dense clouds. In its first application, the model has been used to investigate the origin of the observed dust mass in the z around 7.5 galaxy A1689-zD1 [Watson et al., 2015, Mancini et al., 2015]. We find that while stellar sources dominate the dust mass of small galaxies, the higher level of metal enrichment experienced by galaxies with stellar mass greater that 10^9 solar masses allows efficient grain growth, which provides the dominant contribution to the dust mass. Even assuming maximally efficient supernova dust production, the observed dust mass of the z = 7.5 galaxy A1689-zD1 requires very efficient grain growth. This, in turn, implies that in this galaxy the average density of the cold and dense gas, where grain growth occurs, is comparable to that inferred from observations of QSO host galaxies at similar redshifts [Valiante et al., 2009, 2012, 2014]. Although plausible, the upper limits on the dust continuum emission of galaxies at 6.5 < z < 7.5 show that these conditions must not apply to the bulk of the high-redshift galaxy population. Indeed, more recent and deeper ALMA observations of A1689-zD1 suggest that the thermal dust emission comes from two spatial components, and that the morphological structure is similar to what is observed with HST, pointing to a perturbed dynamical state, perhaps indicative of a major merger or a disc in early formation [Knudsen et al., 2017]. We then extended the analysis to investigate how dust properties affect the appearance of galaxies in the redshift range 5 < z < 8. Using a simple extinction model, we can relate the ISM dust content predicted for each galaxy by the model with direct observables [Bouwens et al., 2015, 2016], such as the number density of objects with a given UV magnitude (the UV Luminosity Functions, LF) and the magnitude dependence of their UV spectral slope (the Color Magnitude Relation, CMR). In addition, our simple model allows us to estimate the infrared luminosity due to dust thermal emission. This provides additional constraints on the mass and properties of dust, given the possibility to compare our predictions with the far infrared continuum emission from a sample of normal star forming galaxies at z around 5 [Capak et al., 2015, Barisic et al., 2017, Faisst et al., 2017]. We find that observations require a steep, Small Magellanic Cloud-like extinction curve and a clumpy dust distribution, where stellar populations younger than 15 Myr are still embedded in their dusty natal clouds. Investigating the scatter in the colour distribution and stellar mass, we find that the observed trends can be explained by the presence of two populations: younger, less massive galaxies where dust enrichment is mainly due to stellar sources, and massive, more chemically evolved ones, where ecient grain growth provides the dominant contribution to the total dust mass. Computing the IR/UV luminosity ratio (the so-called IRX) as a function of the UV colour , we find that all but the dustiest model galaxies follow a relation shallower than the Meurer et al. [1999] one, usually adopted to correct the observed UV luminosities of high-z galaxies for the effects of dust extinction. As a result, using the Meurer et al. [1999] relation to infer the dust correction from a given value of might lead to overestimate the star formation rate. Finally, we compare our predicted IRX- relation with observations of galaxies at 5.1 < z < 5.7 by Capak et al. [2015], which have been argued to be significantly more dust poor and less IR-luminous than lower z galaxies with comparable colours. We find that our simulated galaxies that follow a steep attenutation curve are marginally compatible with the ALMA detected sources by Capak et al. [2015], but that simulated galaxies with IRX compatible with the upper limits inferred for the ALMA undetected sources have significantly bluer colours than observed, consistent with their low dust content. Hence, our study confirms that it is dicult to explain the low IRX of the Capak et al. [2015] sources, unless their slopes have been overestimated or the dust temperature (hence the FIR flux) has been underestimated. Interestingly, both of these hypotheses have been recently confirmed by new observational works, that find systematically bluer colours [Barisic et al., 2017], and that normal high-redshift galaxies have a warmer infrared spectral energy distribution compared to average z < 4 galaxies that were used as prior in previous studies [Faisst et al., 2017]. These new data relieve some of the tension between theoretical predictions and observations [Mancini et al., 2016, Narayanan et al., 2017].