Abstract
A brief summary of JWST observations of the earliest galaxies is presented, in particular the evolution of the luminosity function at high redshift up to z=11 as compared to a variety of models. Surprising results on elemental abundances (e.g., high N/O) and ionization parameters are also found.
1. Introduction
The search for and characterization of the very first galaxies has been one of the main drivers for the development of the James Webb Space Telescope (JWST). Before I summarize the status of this field – that is evolving literally every day – and present few fresh results, it is wort re-assessing why this search is important.
Undeniably, part of its fascination arises from the exploratory nature of this search. By detecting galaxies at higher and higher redshift we are literally exploring backward in time, tracing the formation and evolution of galaxies as we approach the post-Big Bang epoch.
Beside this aspect, though, there are solid physical reasons to study these objects. The very early Universe, indeed, is remarkably different from the local Universe, as also were the physical processes at play. The early Universe is indeed much denser, leading to dynamical times that were certainly much smaller than nowadays. Average metallicity was certainly lower, and islands with (nearly) zero metal pollution were also present. Due to a largely neutral Inter Galactic Medium, the large-scale UV background was extremely suppressed, while the CMB radiation was 10x warmer than nowadays. There is a general expectation that newly born stars were distributed along a top-heavy IMF, which combined with a low metallicity, determined the conditions for a high ionization parameter and a hard UV ionizing flux within the interstellar medium of early galaxies. These conditions led to a formation of stars and first structures through paths that are not well understood and very poorly characterised and tested now.
In addition, the early Universe is an ideal lab to test the existence of ‘new physics’ – i.e. deviations from the current cosmological scenario. Among the various options, it is intriguing to speculate the existence of black holes of intermediate and large mass, originated not by the standard mechanism (supernovae) but formed through different channels either in the very first seconds after the Big Band and/or in the direct collapse of massive gaseous halos at z~100-20. If these objects existed, they triggered additional star formation and acted as the seeds of Super Massive Black Holes in galaxies. An example of the effect of a population of primordial black holes is reported (Figure 1 left), from Cappelluti et al. 2022, where it is shown that they may significantly increase the average star-formation rate only at z>10, when their effect is still significant with respect to the accretion on normal dark matter (DM) haloes.
Figure 1 Left: Evolution of the Cosmic Star Formation Density as a function of redshift, from Cappelluti et al. 2022. Dots and green area represent the observed data with relative uncertainty. Blue and yellow curve represent models with different types of primordial black holes. Right: Halo mass functions for the standard cosmological model Λ-CDM and for Extended Dark Energy models (EDE), at different redshifts, from Liu et al. 2024.
At high redshift it is also possible to identify deviations from the predictions of the standard cosmological model. For instance, the so-called Early Dark Energy (EDE) models predict more dark matter haloes at z~8 and beyond than the standard Λ-CDM model, resulting in larger numbers of galaxies at the same redshift. An example is shown in Figure 1, right, from Liu et al. 2024, where the distribution of DM haloes is shown at different redshifts in the standard and in the EDE model – the latter being significantly higher at z>7. Conversely, Warm Dark Matter models predict a significant deficit of small mass haloes, resulting in a lower number of small-mass galaxies (e.g. Menci et al. 2022). These effects are noticeable, and therefore can be directly tested directly by observation, only at high redshift.
2. The JWST surveys: context and early results
2.1 Photometric surveys
Previous surveys executed with HST and groundbased telescopes have been able to gather extensive statistics of galaxies up to z~8, and sparser samples at galaxies up to z~10. For most of the galaxies beyond z~7, the available information was mostly limited to HST photometry covering the UV range of the spectrum, together with a few spectroscopic redshifts obtained on the rare Lyman α emitters, and detections at very low S/N up to 5.6μm obtained from the warm Spitzer mission. Overall, these data sets were certainly able to reveal the existence of these galaxies and establish their number evolution, showing a progressive decline of the Luminosity Function from z~4 to z~9, assuming that the photometric selection is reasonably complete and free from major contamination.
Based on extrapolations from these data, a steady decline of the number density beyond z~10 was expected, which is not surprising as we approach the Big Bang. While the statistics were reasonably well assessed, the physical characterization of these objects remained elusive. Reliable measures of stellar masses and star-formation rates were hampered by the lack of proper sampling of the rest-frame optical bands, and metallicity and ionization status were very uncertain or impossible to measure given the lack of spectroscopic follow-up.
This situation has been completely and immediately revolutionised by the advent of JWST. Thanks to its thousand-fold increase in sensitivity with respect to Spitzer, and spatial resolution 3x HST, it has been immediately possible to search for galaxies well beyond z~10 and characterize the entire sample at z>7 with large spectroscopic follow up.
In only 18 months, a spectacular data set has indeed been collected by a wealth of large deep surveys, which released images and spectra over at least 12 different public fields, each with a coverage of at least 7-8 bands NIRCAM data, some of which with MIRI counterparts, and hundredths of redshifts at z>7 (Roberts-Borsani et al. 2024). As of today, more than 210arcmin2 have been observed in imaging at typical 5σ magnitude limits between 28 and 30.
These earliest results based on these data have not been short of surprises. As demonstrated by a number of different surveys, (e.g. Finkelstein et al. 2022, Finkelstein et al. 2023, Harikane et al. 2023, Bouwens et al. 2023, Pérez González et al. 2023, McLeod et al. 2024), the density of galaxies (and in particular of the brightest ones) at z>9 is significantly larger than previously estimated by extrapolation of lower redshift observations as well as by theoretical models.
Figure 2 Left: Luminosity Function at z=9, 11 and 14 as observed by the JWST survey CEERS (dots) compared with the predicted LF computed by extrapolation of lower-z LF. Right: the z=11 LF compared with a suite of theoretical models. Both panels from Finkelstein et al. 2024.
This is clearly shown in Figure 2, where the computation is reported of the Luminosity Function (LF) at z~9, 11 and 14, respectively (Finkelstein et al. 2024). These LFs are compared with the extrapolations based on the LF computed on HST survey (left) and the expectations of theoretical models (right). Both extrapolations of lower-z LF and theoretical pre-JWST models were predicting a significant evolution in the LF that is not observed. These results lead to a scenario where the formation of early galaxies is accelerated after the Big Bang, and to the expectation that the epoch of very first assembly is shifted to z>14-15.
Several scenarios have been proposed to explain these findings, ranging from a higher star-formation efficiency, the effect of stochastic star-formation histories, a lower dust extinction, an increased luminosity owing to the contribution of active galactic nuclei (AGNs) or of low-metallicity stars, or even nonstandard cosmological models (e.g. Ferrara et al. 2023, Mason et al. 2023, Melia et al. 2023, Padmanabhan et al. 2023, Trinca et al. 2024).
2.2 Spectroscopic surveys
Follow-up spectroscopy of the newly discovered high-redshift candidates is fundamental both to confirm the measured excess compared to theoretical predictions and to understand its physical origin. Early spectroscopic campaigns carried out with JWST NIRSpec have already provided support to the robustness of photometric selections and enabled the exploration of the physical conditions of galaxies at unprecedented redshifts (e.g. Curtis-Lake et al. 2023, Arrabal-Haro et al. 2023b, Boyett et al. 2023, Roberts-Borsani et al. 2023, Wang et al. 2023).
Early results have found a trend of decreasing metallicity and increasing excitation and ionization efficiency with increasing redshift (Trump et al. 2023, Tang et al. 2023, Nakajima et al. 2023, Curti et al. 2023), although most of the sources show physical conditions comparable to those of low-redshift analogues (Schaerer et al. 2022). A relatively small number of objects have shown features that are not usually found in low-redshift counterparts, and which may be due to physical properties unique to the first phases of star formation and galaxy assembly. A tantalizing example is the bright galaxy GNz11 at z=10.6 (first discovered by Oesch et al. 2016), whose NIRSpec spectrum shows evidence of a nitrogen abundance which is higher than expected for its metallicity (Bunker et al. 2023, Cameron et al. 2023). The discovery of other objects with a comparable nitrogen enrichment (Topping et al. 2024) has suggested that we may be witnessing the formation of globular-cluster progenitors (D’Antona et al. 2023, Watanabe et al. 2024). Instead, a high C/O ratio in galaxy GSz12 at z~12.5 has been interpreted as the imprint of ejecta from a previous generation of Population III stars (D’Eugenio et al. 2023).
A surprisingly large incidence of AGNs has also been suggested by NIRSpec follow-up observations of high-redshift objects (Larson et al. 2023, Maiolino et al. 2023b), with candidates reaching z~10 and beyond including GNz11 itself (Maiolino et al. 2023).
While these early results are tantalising, the emerging picture is far from being accessed. It is fundamental to push spectroscopic investigations to larger samples and higher redshifts to achieve a deeper understanding of the physical conditions of early star-forming regions and to assess the potential contribution of AGN accretion to the ultraviolet (UV) emission of distant galaxies.
3. GHZ2
The galaxy dubbed GHZ2 provided the first example of an unexpected population of high-redshift bright galaxies being discovered in the very first JWST data released on July 14, 2022. Two discovery papers appeared on arXiv the same day, and named the galaxy GHZ2 (Castellano et al. 2022b) and GLASS-z12 (Naidu et al. 2022) (we refer to it only as GHZ2 for conciseness). Initially discovered as a robust z~12.0-12.5 candidate, GHZ2 was targeted with JWST NIRSpec multi-object spectroscopy through Program GO-3073 (PI M. Castellano), which is aimed at extensive follow-up observations of the z≥ 10 candidates selected in the GLASS-JWST region. A full discussion of the results is given in Castellano et al. 2024.
Figure 3 Observed 2D (top) and 1D (bottom) NIRSpec PRISM spectra of GHZ2. In the bottom panel the gray line shows the RMS noise, and red dashed lines highlight the wavelength of the UV features discussed in the present paper. From Castellano et al. 2024.
The NIRSpec spectrum of GHZ2 (Figure 3) shows a sharp Lyman break and list of very prominent emission features, with a full suite of high ionization lines (CIII, CIV etc) all consistent with z~12.3. From a weighted average of the measurements of the best resolved, high-SNR lines (NIV, CIV, CIII, and NeIII), we obtain an accurate redshift z=12.342 ±0.009, which we adopt hereafter.
The spectroscopic redshift is in remarkable agreement with the estimates obtained from NIRCam photometry, lending support to the accuracy of JWST-based photometric selections of high-redshift galaxies, at least as far as bright objects are concerned. Together with similar results obtained from NIRSpec follow-up spectroscopy (e.g. Arrabal-Haro et al. 2023b), this provides a crucial confirmation that the relatively large density of bright galaxies at z>9 is real and deserves detailed investigation in order to understand the earliest phases of galaxy and structure formation.
The spectrum of GHZ2 shows strong NIV, CIV, HeII, OIII, CIII, OII, and NeIII emission lines. The prominent CIV line puts GHZ2 in the category of strong CIV emitters (Izotov et al. 2024). In fact, GHZ2 is the most distant, brightest, and most massive member of this recently discovered class of objects. Assessing the main source of ionizing photons from UV spectroscopy is known to be challenging, in particular for high-redshift objects likely dominated by young, low-metallicity stellar populations.
Unfortunately, the diagnostics more often used to discriminate between star-forming and AGN-dominated objects yield inconclusive results when applied to the observed lines in GHZ2.
Figure 4 Upper panels: line ratio for GHZ2 (orange square) and other high z objects, compared to star-forming (blue dots) and AGN-driven models (red dots). Lower panels. EW of high exhitation line for GHZ2 (orange square) and other high z objects, compared to star-forming (blue dots) and AGN-driven models (red dots).
This is shown in Figure 4, where we plot indexes obtained from line ratios (the two upper panels) and from line equivalent widths. The former are essentially inconclusive with GHZ2 being at the boundary between the two classes and eventually compatible with both.The latter seem to indicate the preference for AGN-driven values, but the AGN solution is at the same time disfavoured by the lack of high ionization lines with ionization potential > 60eV, such as NeIV and NeV. Clearly, high resolution spectra are needed to ascertain or not the presence of an AGN in this object.
Regardless of the nature of the dominant ionizing flux, we found that GHZ2 has a very low metallicity (below 10% solar) and a high ionization parameter (logU > -2). The N/O abundance is found to be 4-5 times the solar value, while the C/O is subsolar, similar to a number of recently discovered high-redshift objects (Topping et al. 2024). Given its small effective radius (Re ~ 100pc), GHZ2 has a high ΣSFR and a high stellar mass density similar to gravitationally bound stellar clusters; it is intriguing to speculate that GHZ2 is undergoing a phase of intense star formation in a dense configuration that may evolve into the nitrogen-enhanced stellar populations of globular clusters and other dense environments that are observed at low redshifts.
The origin of the copious amounts of ionizing photons in objects such as GHZ2 is currently unknown, but scenarios of dense star formation at very low metallicity including supermassive stars and high-mass X-ray binaries have the potential to also explain the atypical abundance patterns and the high-ionization spectra. A detailed investigation of the rare detection of the OIII-Bowen fluorescence line in GHZ2 can provide further insight into its sources of ionizing photons and their local environment. The high CIV/CIII ratio and the abundance patterns of GHZ2 are also suggestive of a high escape fraction of ionizing photons. A comprehensive search for sources in this evolutionary stage can reveal if they play a significant role in the reionization of the inter-galactic medium.
We caution, however, that in-depth studies will be needed to consolidate the proposed scenario. The N/O ratio needs to be assessed based on a robust detection of the NIII line in a higher-resolution spectrum considering the strong dependence of our measurement on the extrapolated continuum level at its position. Similarly, a high-resolution spectrum is needed to estimate the density and temperature of the ionized gas, to ascertain the presence of broad components due to AGNs or stellar winds, and to separate nebular and stellar contributions to the CIV emission.
The remarkable brightness of GHZ2 makes it accessible to a wealth of follow-up strategies, as showcased by the results described in this paper and by the MIRI detection discussed in a companion paper (Zavala et al., 2024). As such, GHZ2 has the potential to become a reference object for understanding galaxy formation at only 360~Myr after the Big Bang.
Bibliography
Arrabal-Haro et al. 2023, Nature, 622, 707.
Bouwens et al. 2023, MNRAS, 523, 1036.
Boyett et al. 2023, arXiv e-prints, arXiv:2303.00306.
Bunker et al. 2023, A&A, 677, A88.
Cameron et al. 2023, MNRAS, 523, 3516.
Cappelluti et al. 2022, ApJ, 926, 205.
Castellano et al. 2022, ApJL, 938, L15.
Castellano et al. 2022b, A&A, 662, A115.
Castellano et al. 2024, arXiv e-prints, arXiv:2403.10238.
Curti et al. 2023, arXiv e-prints, arXiv:2304.08516.
Curtis-Lake et al. 2023, Nature Astronomy, Volume 7, p. 622-632.
D’Antona et al. 2023, A&A, 680, L19.
D’Eugenio et al. 2023, arXiv e-prints, arXiv:2311.09908.
Ferrara et al. 2023, arXiv e-prints, arXiv:2310.12197.
Finkelstein et al. 2022, ApJL, 940, L55.
Finkelstein et al. 2023, ApJ, 946, L13.
Finkelstein et al. 2024, ApJ, 969, L2.
Harikane et al. 2023, ApJS, 265, 5.
Izotov et al. 2024, MNRAS, 528, L10.
Larson et al. 2023, ApJL, 953, L29.
Liu et al. 2024. arXiv:2406.03936.
Maiolino et al. 2023°, arXiv e-prints, arXiv:2308.01230.
Maiolino et al. 2023b, arXiv e-prints, arXiv:2305.12492.
Mason et al. 2023, MNRAS, 521, 497.
McLeod et al. 2024, MNRAS, 527, 5004.
Melia et al. 2023, MNRAS, 521, L85.
Menci et al. 2022, Apj 938, L5.
Naidu et al. 2022, ApJL, 940, L14.
Nakajima et al. 2023, ApJS, 269, 33.
Oesch et al. 2016, ApJ, 819, 129.
Padmanabhan et al. 2023, ApJL, 953, L4.
PerezGonzalez et al. 2023, arXiv e-prints, arXiv:2302.02429.
Roberts-Borsani et al. 2023, Nature, 618, 480.
Roberts-Borsani et al. 2024, arXiv e-prints, arXiv:2403.07103.
Schaerer et al. 2022, A&A, 665, L4.
Tang et al. 2023, MNRAS, 526, 1657.
Topping et al. 2024, arXiv e-prints, arXiv:2401.08764.
Trinca et al. 2024, MNRAS. 529, 3563.
Trump et al. 2023, ApJ, 945, 35.
Wang et al. 2023, ApJL, 957, L34.
Watanabe et al. 2024, ApJ, 962, 50.
Zavala et al. 2024, arXiv e-prints, arXiv:2403.10491.