Interstellar dust in galaxies in the Epoch of Reionization

We are living in the golden age of extra-galactic astrophysics. The construction of the submillimetre (sub-mm) interferometer ALMA, and the successful operation of the JWST, have marked a giant leap forward in terms of the amount and quality of extra-galactic data. Simultaneously, the rapidly increasing computational power of modern supercomputers allows us to develop and run increasingly detailed numerical simulations [1, 2], bridging the gap between cosmological [3, 4, 5, 6] and sub-galactic scale physics (e.g. [7, 8]). Before the advent of new generation sub-mm interferometers, the early Universe (z > 4) was believed to be nearly dust and metals free. Surprisingly, ALMA observations have revealed the presence of copious dust FIR continuum emission in galaxies when the Universe was only ∼ 600 Myr old [9, 10, 11, 12]. Dust thermal FIR emission is produced by interstellar dust grains heated by the UV and optical radiation coming from young stars [13, 14, 15]. Due to the stringent time constraints imposed by the age of the Universe and of the dominant stellar populations at z > 5, short-lived SN have been appointed as the dominant dust production mechanism at these high redshifts [16, 17, 18]. However, the amount of dust which is spared in a SN reverse shock is highly debated, and some theoretical works suggest SN dust yields which are in tension with the measured dust-to-stellar mass ratios [9, 19, 20, 21]. This tension between high-z FIR data and theory aggravated the so-called “dust budget crisis” in the early Universe [22]. Crucially, the poor understanding of the mechanisms responsible for the dust and metals build-up at high-z has been hindering the progress of galaxy evolution studies. The inferred large dust masses (Md) for high-z galaxies are often dependent on the cold dust temperatures (Td ∼ 25K, see [23] assumed in their FIR continuum fitting procedure. The lack of knowledge on Td (due to the few or even single available ALMA data) casts large uncertainties on the derived IR luminosity (LFIR) and obscured SFR of high-z galaxies. Several works based on sub-mm and radio serendipitous detections of UV dark systems have suggested that we might have been severely underestimating the contribution of the dust-obscured star formation to the cosmic SFRD [24] at z > 4 ([25, 26, 27]). The main uncertainties complicating this assessment come from i) the few and sparse observations of individual high-z galaxies FIR spectra and ii) our very biased view of star-forming galaxies at early epochs, mostly limited to UV-selected massive objects (which are hardly representative of the whole galaxy population). A consensus on the evolution of the obscured and total SFRD at high-z is yet to be reached. Other works based on high-z UVselected targets or the extrapolation from lower-z blank field sub-mm observations report a steep decrease in the obscured SFRD at z > 4 − 6 [28, 21, 29]. Crucially, the SFRD is a commonly used benchmark for galaxy formation and mass assembly models. Therefore, getting it right is paramount if we hope to constrain our models of galaxy formation. Open questions: The fundamental open questions motivating my research are: • How do the physical conditions (e.g. density, temperature, radiation intensity) in early galaxies SF regions impact interstellar dust properties? • What are the production mechanisms responsible for the dust/metals build-up in the early universe? • What is the contribution of the obscured sources/star formation to the cosmic SFRD beyond cosmic noon at z >∼ 5? A possible solution to the “dust budget crisis” in the early universe, is that high-z galaxies host warm/hot dust. Indeed, hotter Td can produce the same observed FIR flux without implying such large dust masses. This hypothesis seems to be supported by stateof- the-art cosmological zoom-in simulations at high-z (see e.g.[30, 31, 32, 33, 34, 6, 35, 8, 7]). Some of these works [30, 32] find that in early galaxies most of the FIR emission comes from a few UV-opaque GMC where dust temperatures as hot as ∼ 100K (∼ 4× higher than typically measured in the local universe) can be reached. Unfortunately, GMCs are not resolved in these simulations due to the demanding computational time required to run them. To further investigate these small scales, Sommovigo+20 (hereafter LS+20) zoomed in on the star-forming GMCs within a z ∼ 8 simulated galaxy from the state-of-the-art SERRA simulation suite [8]. We developed a sub-grid analytical model to predict gas and dust properties and model the FIR continuum emission. We found that in highz GMCs dust is warmer due to the compact structure (high internal pressure and gas column density) of the clouds. This compactness also delays the clouds dispersal by stellar feedback, thus ∼ 40% of the total UV radiation emitted by newly born stars remains obscured. This might question our understanding of dust obscuration at high-z, which is based on the simplistic assumption of the cospatiality of UV and IR emission. To overcome the observational limitations in constraining high-z galaxies dust properties (namely the temperature Td, mass Md, and IR luminosity LFIR), in LS+21 we pioneered a new method which relies on a single ALMA band observation per each source. We use the [C II] 158μm emission-line luminosity as a proxy for the dust mass, and the underlying continuum to constrain the dust temperature. In LS+22a,b we applied this new method to interpret the observations from two recent, ambitious ALMA large programs, which have ventured to trace the cold gas and dust emission in numerous galaxies approaching the EoR: ALPINE (z = 4 − 6, [36, 37]) and REBELS (z = 6 − 9, [38]). We stress that due to observation time constraints, most REBELS and ALPINE targets were observed in a single ALMA band around the [C II] emission. Thus, the dust properties of individual galaxies could not be inferred using traditional SED fitting techniques. We find that the dust-to-stellar mass ratios inferred for most ALPINE and REBELS sources are consistent with SN dust production constraints (assuming dust yields ∼ 0.1 M⊙ per SN event). This might hint that the “dust budget crisis” in the early universe is not as severe as initially thought. However, the large uncertainties on these galaxies stellar masses (> 1 dex, [39]) and the limited number (and biased nature) of sources, are currently preventing us from reaching definitive conclusions. We note that the obscured SFR fraction of the UV-bright galaxies targeted both in REBELS and ALPINE is on average >∼ 50%, reinforcing the importance of properly correcting for dust obscuration even at high-z at least in the most massive systems (M⋆ > 109 M⊙). Further, in LS+22a,b we find that the average dust temperature is nearly the same in REBELS and ALPINE galaxies (Td ∼ 45−50 K), and it is ∼ 2× warmer than that observed in most local main-sequence galaxies. A positive trend between Td and redshift did already appear in the analysis of stacked ALMA and Herschel data in the range z = 0 − 5 (e.g. [40, 41, 42]). However, different stacking procedures result in very different (empirically derived) Td−z relations (see [43]). The largest discrepancies arise at the high-z end (z > 4), where isolating low-z interlopers in the stacks and/or correcting for the bias towards the brightest sources becomes more challenging. To clarify this issue and address the lack of physical understanding for these reported Td − z trends, in LS+22a we produced a physical model for the cosmic evolution of Td. Using a simple energy conservation argument, we showed that the dust temperature anticorrelates with the total gas depletion time (tdep = Mgas/SFR). Due to the higher cosmological accretion rate, tdep decreases at early times (see also [23, 44]), thus implying a mild increase of Td with redshift, which changes our understanding of star formation conditions in the early universe. Finally, we conclude this by showing a preliminary analysis of early JWST pioneer observation of z > 10 galaxy candidates from Ziparo+23 [45]. JWST observations – albeit affected by several uncertainties due to calibration issues and uncertain photometric redshifts – seem to highlight an excess of UV bright, massive (stellar mass M⋆ ∼ 108−9) and very blue (UV slope βUV ∼ −2.6) systems at this extraordinarily high-z (e.g. [46, 47, 48, 49, 50, 51]). Our favoured explanation is that such extremely early galaxies undergo tumultuous star formation episodes, which due to their very compact sizes (∼ 100 pc), result in strong radiation pressure-driven winds which devoid the galaxies of a large fraction of their simultaneously produced dust and metals.