Stark (2016) has reviewed our knowledge of early galaxies, in the first billion years after the Big-Bang. For z > 6, ALMA yields a good opportunity to detect the dust emission, provided that they are dusty enough. The peak of dust emission is indeed shifted towards λ > 0.7mm. For high-z objects, it becomes difficult to obtain spectroscopic redshifts optically, especially when obscured by dust. ALMA can then help to identify the objects, thanks to the [CII] line at 158 µm, redshifed to λ > 1.1mm. Models of the ISM had predicted that the main coolant would be through this [CII] line, however the observations reserved some surprises. The photoelectric heating efficiency of the dust, measured by the ratio L[CII] / LFIR, varies by about 2 orders of magnitude, and is decreasing at high LFIR, for strong starbursts. The main factor reducing this efficiency has been shown to be the dust temperature, and the strong UV field (Malhotra et al, 2017): indeed, the L[CII] / LFIR ratio is very well anti-correlated to the dust temperature, whatever the redshift. Figure 12 gathers a large fraction of the [CII] studies so far, and shows that the [CII] / FIR ratio is higher at high redshift, although still declining with LFIR. The high-z quasars detected reveal a wide range of properties, sometimes behaving like starbursts, while sometimes the quasar excitation may prevail (e.g. Venemans et al, 2016).
 |
Figure 12. The [CII] to FIR luminosity ratio versus the FIR luminosity. The low-redshift galaxies are plotted as black circles, from Malhotra et al (2001), Luhman (2013), and Díaz-Santos et al (2013). Various ULIRGs at high redshift (z = 1 to 6) detections from the literature are in green circles, and the Hello sources (z = 1-3) amplified by lenses are in blue circles (Malhotra et al, 2017). The red circles are the high-z SPT sources from Gullberg et al (2015), corrected for their amplification factor. Quasars at z > 4 are plotted as magenta stars (Venemans et al, 2012, Venemans et al, 2016). At high z, the [C II] / FIR ratio still declines with FIR luminosity, but takes higher values than at z = 0. |
Many searches have been made with ALMA, with some surprising failures, indicating that galaxies are really "primordial", with low metallicity (Z < 0.1) and little dust. The typical Lyα emitter Himiko was not detected in the continuum, nor in the [CII] line (Ouchi et al, 2013). Several other upper limits confirmed that most LBG between z = 6 and 8 are very difficult to detect, even with gravitational lensing (e.g. Schaerer et al, 2015). With more observations, Himiko is now detected in the [CII] line, but not in dust emission (Carniani et al, 2018), revealing a dust deficiency.
Evidence also exists of early galaxies, with their ISM extended over kpc sizes, but weak or undetected in dust emission, while revealing strong [CII] line emission (Capak et al, 2015). These show some similarity to low-metallicity dwarfs in the local Universe, like the LMC, which have a very high L[CII] / LFIR ratio. However, these high-z objects are much more massive, which implies on the contrary a rapid evolution in the ISM dust properties over redshift, due to metal deficiency. Selecting their objects with a relatively lower Lyα equivalent widths, indicating the presence of dust, at a given UV luminosity, Willott et al (2015) reports dust emission and [CII] line detections at z ∼ 6. The velocity redshift of the Lyα with respect to the [CII] line is very prominent at high-z, due to increased intergalactic gas (IGM) absorption of the blue wing of Lyα. The expected enhancement of IGM absorption in the EoR, is not always there (Pentericci et al, 2016), implying patchy reionization.
At high-z, galaxies are clumpy, and sometimes the [CII] line and even the [OIII] line at 88 µm are spatially offset from the Lyα or UV clumps (Carniani et al, 2017, Matthee et al, 2017). These offsets may be explained by obscuration, different excitation or metallicity of the different tracers. Alternatively, strong feedback could have removed a large fraction of gas and dust, or several parts of the systems are interacting while assembling, as suggested by theoretical models (Katz et al, 2017).
Some of the highest redshifts found in the EoR with ALMA are the z = 8.38 gravitationally lensed galaxy selected from deep HST imaging in the Frontier Field cluster Abell 2744 (Laporte et al, 2017), or the Lyman Break galaxy at z = 8.31 behind the Frontier Field cluster MACS J0416.1-2403 (Tamura et al, 2018). Dust emission and the [OIII] line have been detected, raising the problem of forming such dust amounts ∼ 600 Myr after the Big Bang. This would imply that each SN-II explosion has been able to produce 0.5 M⊙ of dust, during the SFR = 15-20 M⊙/yr star forming phase, since z = 10-12.
The highest redshift is MACS1149-JD1 at z = 9.11, a lensed galaxy dectected in the [OIII] line. No redshift was known from the optical before, and the [OIII] line was used to measure the redshift. The colors of its stellar population show that star formation began at z = 15 in this galaxy (Hashimoto et al, 2018).
Contrary to many ALMA surveys, finding a drop in their source number at high redshift, Strandet et al (2016) find a redshift distribution much more weighted towards the high-z, because of a low-frequency selection. Dusty sources were selected from the South Pole Telescope (SPT) survey, from their 1.4mm continuum flux; eliminating the synchrotron sources, by requiring 1.4mm flux being twice higher than the 2mm flux.
Although most of high-z star forming objects selected optically have low dust content, exceptional objects exist, like HFLS3, at z = 6.34, with SFR = 2900 M⊙/yr, a gas mass of 1011 M⊙, including 2 × 1011 M⊙ of atomic gas, and a depletion time of 36 Myr (Riechers et al, 2013). These must be located in proto-clusters, and are the progenitors of massive ellipticals in clusters today.