We use the SPace Infrared telescope for Cosmology and Astrophysics (SPICA) project as a template to demonstrate how deep spectrophotometric surveys covering large cosmological volumes over extended fields (1– $15\, \rm{deg^2}$ ) with a mid-IR imaging spectrometer (17– $36\, \rm{\rm{\upmu m}}$ ) in conjunction with deep $70\, \rm{\rm{\upmu m}}$ photometry with a far-IR camera, at wavelengths which are not affected by dust extinction can answer the most crucial questions in current galaxy evolution studies. A SPICA-like mission will be able for the first time to provide an unobscured three-dimensional (3D, i.e. x, y, and redshift z) view of galaxy evolution back to an age of the universe of less than $\sim$ 2 Gyrs, in the mid-IR rest frame. This survey strategy will produce a full census of the Star Formation Rate (SFR) in the universe, using polycyclic aromatic hydrocarbons (PAH) bands and fine-structure ionic lines, reaching the characteristic knee of the galaxy luminosity function, where the bulk of the population is distributed, at any redshift up to $z \sim 3.5$ . Deep follow-up pointed spectroscopic observations with grating spectrometers onboard the satellite, across the full IR spectral range (17– $210\, \rm{\rm{\upmu m}}$ ), would simultaneously measure Black Hole Accretion Rate (BHAR), from high-ionisation fine-structure lines, and SFR, from PAH and low- to mid-ionisation lines in thousands of galaxies from solar to low metallicities, down to the knee of their luminosity functions. The analysis of the resulting atlas of IR spectra will reveal the physical processes at play in evolving galaxies across cosmic time, especially its heavily dust-embedded phase during the activity peak at the cosmic noon ( $z \sim 1$ –3), through IR emission lines and features that are insensitive to the dust obscuration.

We report fourteen and twenty-eight protocluster candidates at z = 5.7 and 6.6 over 14 and 19 deg2 areas, respectively, selected from 2,230 Lyα emitters (LAEs) photometrically identified with Subaru/Hyper Suprime-Cam (HSC) deep images. Six out of the 42 protocluster candidates include at least 1 spectroscopically confirmed LAEs at redshifts up to z = 6.574. By the comparisons with the cosmological Lyα radiative transfer (RT) model reproducing LAEs with the reionization effects, we find that more than a half of these protocluster candidates might be progenitors of the present-day clusters with a mass of ≳ 1014M⊙. We also investigate the correlation between LAE overdensity and Lya rest-frame equivalent width (EW), because the cosmological Lyα RT model suggests that a slope of EW-overdensity relation is steepened towards the epoch of cosmic reionization (EoR), due to the existence of the ionized bubbles around galaxy overdensities easing the escape of Lyα emission from the partly neutral intergalactic medium. The available HSC data suggest that the slope of the EW-overdensity correlation does not evolve from the post-reionization epoch z = 5.7 to the EoR z = 6.6 beyond the moderately large statistical errors.

Recent CO surveys of star-forming galaxies (SFGs) at z ~ 2 have revolutionized our picture of massive galaxies. It is time to expand these studies toward the more common z ~ 2 SFGs with SFR < 40 M⊙ yr−1 and M* < 2.5 × 1010 M⊙. We have derived molecular gas, stars, and dust in 8 such lensed SFGs. They extend the LIR–L'CO(1-0) distribution of massive z>1 SFGs and increase the spread of the SFG star formation efficiency (SFE). A single star formation relation is found when combining all existing CO-detected galaxies. Our low-M* SFGs also reveal a SFE decrease with M* as found locally. A rise of the molecular gas fraction (fgas) with redshift is observed up to z ~ 1.6, but it severely flattens toward higher redshifts. We provide the first insight into the fgas upturn at the low-M* end 109.4 < M*/M⊙ < 1010 reaching fgas ~ 0.7, it is followed by a fgas decrease toward higher M*. Finally, we find a non-universal dust-to-gas ratio among local and high-redshift SFGs and starbursts with near-solar metallicities.

We have conducted a LABOCA 870 μm follow-up of ten massive lensing clusters of the Herschel Lensing Survey (HLS, Egami et al. 2010) aiming at unveiling the yet hidden part of dusty star formation in the distant Universe. Among these clusters, A2744 and AS1063 are part of the Frontier Fields HST program. We also obtained 2 mm bolometer observations of A2744 and A370 with the GISMO array at the IRAM 30 m. We detected sources that are undetected with Herschel (PACS and SPIRE) implying either a very high redshift (z > 4) or a very low dust temperature (T<25 K). Their flux also imply a low intrinsic luminosity, LFIR < 1012L⊙. Some of them are extended and could correspond to multiple sources or to multiple images of a lensed source. Substructures in the Sunyaev-Zeldovich effect could also contribute this very red submm emission. An ALMA program is ongoing to unveil the nature of these sources.

Current star-forming galaxies (SFGs) with CO measurements at z ~ 2 suffer from a bias toward high star formation rates (SFR) and high stellar masses (M*). It is yet essential to extend the CO measurements to the more numerous z ~ 2 SFGs with LIR < L⋆ = 4× 1011 L⊙ and M* < 2.5× 1010 M⊙. We have achieved CO, stars, and dust measurements in 8 such sub-L⋆ SFGs with the help of gravitational lensing. Combined with CO-detected galaxies from the literature, we find that the LIR, L′CO(1−0) data are best-fitted with a single relation that favours a universal star formation. This picture emerges because of the enlarged star formation efficiency spread of the current z>1 SFGs sample. We show that this spread is mostly triggered by the combination of redshift, specific SFR, and M*. Finally, we find evidence for a non-universal dust-to-gas ratio (DGR) with a clear trend for a lower DGR mean in z>1 SFGs by a factor of 2 with respect to local galaxies and high-redshift sub-mm galaxies at fixed about solar metallicity.

To extend the molecular gas measurements to typical L* star-forming galaxies (SFGs) at z ∼ 1.5 − 3, we have observed CO emission for five strongly-lensed galaxies selected from the Herschel Lensing Survey. The combined sample of our L* SFGs with CO-detected SFGs at z >1 from the literature shows a large spread in star formation efficiency (SFE). We find that this spread in SFE is due to variations of several physical parameters, primarily the specific star formation rate, but also stellar mass and redshift. An increase of the molecular gas fraction (fgas) is observed from z ∼ 0.2 to z ∼ 1.2, followed by a quasi non-evolution toward higher redshifts, as found in earlier studies. We provide the first measure of fgas of z >1 SFGs at the low-stellar mass end between 109.4 < M∗/M⊙ < 109.9, which shows a clear fgas upturn.

We present the results obtained from our deep survey of lensing clusters aimed at constraining the abundance of star-forming galaxies at z∼6–11.

Recent progress in measuring the optical depth of neutral hydrogen in distant quasars and that of electron scattering of microwave background photons suggests that most of the sources responsible for cosmic re-ionisation probably lie in the redshift interval 6 to 10. We present two new observational results which, together, provide valuable constraints on the contribution from star-forming sources in this redshift interval. First, using a large sample of v-band dropouts with unconfused Spitzer-IRAC detections, we determine the integrated stellar mass density at z = 5. This provides a valuable ‘integral constraint’ on past star formation. It seems difficult to reconcile the observed stellar mass at z = 5 with the low abundance of luminous i-z- and J-band dropouts in deep Hubble Space Telescope data. Accordingly, we explore whether less luminous star-forming sources in the redshift interval 6 to 10 might be the dominant cause of cosmic re-ionization. In the second component of our research, we report on the results of two surveys for weak Lymanα emitters and z- and J-band dropouts highly-magnified by foreground lensing clusters. Although some promising z = 8–9 candidates are found, it seems unlikely that low luminosity sources in this redshift interval can dominate cosmic reionization. If our work is substantiated by more extensive and precise surveys, the bulk of the re-ionizing photons may come from yet earlier sources lying at redshifts z>10.