Early Results from GLASS-JWST. III. Galaxy Candidates at z ∼9–15^*

Since the discovery of the first object at a cosmological distance (Schmidt 1963), the quest for "the most distant source ever seen" has been both a motivation and a powerful means to expand our understanding of cosmology and of how astronomical objects formed and evolved. Substantial progress in the study of distant, star-forming galaxies has been enabled by the Lyman-break technique yielding to samples of so-called Lyman-break galaxies (LBGs) at progressively higher redshifts (e.g., Giavalisco 2002).

In recent years, the combination of Hubble Space Telescope (HST) and ground-based data enabled the selection of relatively large samples of LBGs at z ∼ 2–8 (e.g., Reddy & Steidel 2009; McLure et al. 2013; Finkelstein et al. 2015; Roberts-Borsani et al. 2016; Atek et al. 2018; Ono et al. 2018). The inferred UV luminosity functions (LF), together with assumptions on the ionizing output of LBGs, enabled the first investigations of the epoch of reionization (e.g., Dayal & Ferrara 2018). The data are consistent with a scenario in which the universe was still largely neutral at z ∼ 8 and was rapidly reionized by z ∼ 5.5–6 (e.g., Mitra et al. 2015; Greig & Mesinger 2017; Mason et al. 2018), likely by faint star-forming galaxies (e.g., Castellano et al. 2016b; Yue et al. 2018; Finkelstein et al. 2019; Dayal et al. 2020).

Very few constraints are available at higher redshifts. Small samples of increasingly bright objects have been selected up to z ∼ 9–11 (Zheng et al. 2012; Coe et al. 2013; Oesch et al. 2013; McLeod et al. 2016; Hashimoto et al. 2018; Ishigaki et al. 2018; Morishita et al. 2018; Rojas-Ruiz et al. 2020; Bouwens et al. 2021; Bagley et al. 2022; Roberts-Borsani et al. 2022a). The available estimates of the Schechter UV LF show significant discrepancies, with some results in agreement with a smooth evolution from lower redshifts, driven either by a decrease of ϕ^* or by a dimming of M^* (e.g., McLeod et al. 2016), and other analyses suggesting either a dearth (Oesch et al. 2018) or an excess (Bagley et al. 2022) of bright galaxies at z ∼ 10. As a result, the cosmic star formation rate (SFR) density is poorly constrained, and it is not clear whether the ratio between SFR and dark matter accretion rate remains approximately constant at z ∼ 9–11 as found at z ≲ 7 (Harikane et al. 2022a).

The discovery of the population of galaxies that ended the so-called dark ages at z ∼ 12 and beyond has been so far prevented by the limited IR spectral coverage of HST, and by the insufficient sensitivity and resolution of Spitzer and ground-based telescopes. In fact, at such high redshifts, only tentative candidates have been found (Harikane et al. 2022a). The highest spectroscopically confirmed redshift is that of LBG GN-z11 at z = 10.957 (Oesch et al. 2016; Jiang et al. 2021). Even with their limitations, current observations have challenged our understanding of galaxy buildup in the early universe. The number densities of bright z > 10 candidates and GN-z11 lie above the typical empirical Schechter function (e.g., Bowler et al. 2020) while theoretical models are in tension with the abundance of bright galaxies (Finkelstein et al. 2022). Thus, these remarkably bright galaxies may mark a significant transition in the mode of galaxy formation in the early universe: likely significantly less dust-obscured and with a higher star formation efficiency enabling more stars to form in high-mass halos, and/or the presence of young quasars. Understanding how these galaxies formed requires building a census of the galaxy population at z > 10 with deeper photometric data, especially at longer wavelengths than those covered by HST.

In this paper we present the search for LBG candidates at z ∼ 9–15 in the first NIRCam imaging data from the GLASS-James Webb Space Telescope (JWST) Early Release Science Program (JWST-ERS-1324; Treu et al. 2022, hereafter T22). GLASS-JWST targets the Hubble Frontier Fields (HFF) galaxy cluster A2744 (Lotz et al. 2014; Castellano et al. 2016a; Merlin et al. 2016a) with NIRISS and NIRSpec spectroscopy, while obtaining coordinated parallel observations with NIRCam on two flanking fields. The GLASS NIRCam parallels are the deepest images collected by the 13 ERS programs. Details on the design and strategy of the GLASS-JWST survey can be found in T22. Throughout the paper, we adopt AB magnitudes (Oke & Gunn 1983), and a flat Λ-cold-dark-matter concordance model (H₀ = 70.0 km s⁻¹ Mpc⁻¹, Ω_M = 0.30).

The GLASS-JWST NIRCam observations discussed in this paper were taken in parallel to NIRISS on 2022 June 28–29, They are centered at R.A. = 3.5017025 deg and decl. = −30.3375436 deg, and consist of images in seven bands: F090W (total exposure time 11,520 s), F115W (11,520 s), F150W (6120 s), F200W (5400 s), F277W (5400 s), F356W (6120 s), and F444W (23,400 s). The image reduction and calibration, and the methods used to detect sources and measure multiband photometry are described in a companion paper (Paper II in this series, Merlin et al. 2022, hereafter M22). The NIRISS observations are described in Paper I (Roberts-Borsani et al. 2022b).

For the convenience of the reader, we briefly summarize below the information relevant for the present paper. Data reduction and flux calibration were obtained using the official JWST pipeline ²³ and exploiting the calibration files made available by STScI on 2022 July 29. A customized step has been introduced to improve masking of image defects (snowballs and wisps) and to improve the subtraction of 1/f noise and background. Images have been aligned to a common reference grid (defined on Gaia Data Release 3 calibrated ground-based images) using a custom pipeline already adopted in similar projects (Fontana et al. 2014).

We warn the reader that flat-fielding, flux calibration, and other steps of the data processing are inevitably based on preliminary calibrations. Similarly, the cleaning of defects, especially in the bluest bands that are used here to identify the Lyman break, will also be improved as the knowledge of the telescope and instrument improves.

Objects were detected using a customized version of SExtractor (Bertin & Arnouts 1996; Guo et al. 2013, v2.8.6) on the F444W coadded image. Total F444W fluxes in elliptical apertures as defined by Kron (1980) were measured with a-phot (Merlin et al. 2019). Fluxes in the other bands were measured with a-phot in several apertures at the positions of the detected sources using images point-spread function-matched to the F444W one. For the final search of high-z candidates, we have used both an aperture of 2FWHM diameter (=0''.28) as well as a smaller, 0''.2 one. The latter has been adopted to optimize the search of potentially blended sources, which may be useful in our crowded field.

The final 5σ depths for point sources within a circular aperture of diameter of 0''.2 are 28.78, 29.03, 28.84, 28.89, 29.26, 29.33, 29.71 in the F090W − F444W bands, respectively (M22). We estimate a total area of ∼7 arcmin² available for the selection of high-redshift candidates by considering the regions observed in all bands, and excluding the pixels occupied by foreground sources. Modest lensing magnification is expected to be present in the NIRCam parallel fields. In this initial set of papers, we neglect the effect. The issue will be revisited after the completion of the campaign, which includes a second NIRCam field, in parallel to NIRSpec observations (T22).

3.1. Color–Color Selection

We defined the color selection criteria for galaxies at z ∼ 9–15 using a catalog of objects at z = 0–15 that mimics our observations. The catalog comprises objects at 0 < z ≤ 4.5 generated with the Empirical Galaxy Generator (EGG) code (Schreiber et al. 2017a), which is based on empirical relations to reproduce observed number counts and color distributions of galaxies at low and intermediate redshifts. Galaxies at higher redshifts have been generated according to observed UV LFs over z ∼ 5–10 (McLeod et al. 2015; Oesch et al. 2018; Bouwens et al. 2021) and randomly associating to each object a template from a library based on Bruzual & Charlot (2003, hereafter BC03) models with metallicity = 0.02,0.2 Z_⊙, 0 < E(B − V) < 0.2, constant SFH, Salpeter (1955) initial mass function, and Calzetti et al. (2000) extinction law to predict photometry in 7 bands.

In order to provide sufficient statistics to design appropriate selection criteria, we simulated the catalog over an area of ∼0.12 deg², which is ∼60 times larger than the NIRCam parallel field, and we assumed no evolution of the UV LF beyond z = 10. We also artificially boosted the number counts at z > 9.5 by a factor of 20. The overrepresentation of high-z sources in the mock catalogs is taken into account when considering purity and completeness. Late-type dwarf stars are another potential source of contamination for the color selection of high-redshift galaxies. We assessed their impact using synthetic JWST photometry for the models by Marley et al. (2021), which include brown dwarfs and self-luminous extrasolar planets with 200 ≤ T_eff ≤ 2400 and metallicity [M/H] from 0.5 to +0.5. The brown dwarf models were normalized at 26.0 ≤ F444W ≤ 28.0 in 0.5 mag steps.

Finally, the catalogs were perturbed by adding noise in order to reproduce the expected relation and scatter between magnitudes and errors in each band.

We find that the following color–color selection window, which is a modified version of the selection proposed by Hainline et al. (2020), identifies objects at z ∼ 9–11.5, as shown also in the top panels of Figure 1:

$$\begin{eqnarray}&&\begin{array}{c}({\rm{F}}115{\rm{W}}-{\rm{F}}150{\rm{W}})\gt 1.7,\\ ({\rm{F}}115{\rm{W}}-{\rm{F}}150{\rm{W}})\gt 2.17\times ({\rm{F}}200{\rm{W}}-{\rm{F}}277{\rm{W}})+1.7,\\ \,\,-\,0.8\lt ({\rm{F}}200{\rm{W}}-{\rm{F}}270{\rm{W}})\lt \mathrm{0.6.}\end{array}\end{eqnarray} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

In addition, we require a signal-to-noise ratio (S/N) > 3 in all bands redward of the Lyman break (F150W, F200W, F277W, F356W), and S/N < 1.5 in the F090W band.

Objects in the wider range z ∼ 9–15 are well identified by the following selection criteria, which leverages also on the redder bands (e.g., F444W):

$$\begin{eqnarray}&&\begin{array}{c}({\rm{F}}115{\rm{W}}-{\rm{F}}444{\rm{W}})\gt 2.0,\\ ({\rm{F}}115{\rm{W}}-{\rm{F}}444{\rm{W}})\gt 2.5\times ({\rm{F}}200{\rm{W}}-{\rm{F}}444{\rm{W}})+1.5,\\ \,\,-\,1.0\lt ({\rm{F}}200{\rm{W}}-{\rm{F}}444{\rm{W}})\lt \mathrm{1.0.}\end{array}\end{eqnarray} \tag{ 2 }$$

By requiring all the S/N requirements above, plus S/N(F150W) < 2 and S/N(F115W) < 2, the selection in Equation (2) proves to be complementary to the one in Equation (1), sampling the z ∼ 11–15 redshift range (Figure 1, bottom panels). We verified that most of the z ≲ 11 objects selected in the F115W − F150W versus F200W − F277W diagram occupy the expected color space in F115W − F444W versus the F200W − F444W one, although photometric scatter may move them outside the color selection window. For this reason, and given also the differences in depths among the various bands, the two diagrams are not expected to select coincident samples in the z ∼ 9–11 range.

For both color selections described above, we consider only objects with S/N(F444W) > 8, corresponding to F444W ∼ 28.2, that, given the relevant depths in the bluer bands, enable the measurement of the large color terms (∼1.7–2) required to separate the high-redshift galaxies from the bulk of contaminants. While the proposed diagrams efficiently select high-redshift targets, some contamination from low-redshift galaxies is evident from the redshift distribution of the selected objects (right panels in Figure 1). The impact of such contamination and the completeness of the selection criteria will be discussed in Section 3.3. Instead, contamination from brown dwarfs is not a concern for our z ∼ 9–15 selection. We show in Figure 1 late-type stars with 26.0 ≤ F444W ≤ 28.0 that meet the above-mentioned S/N criteria in the F150W to F444W bands: the brown dwarfs occupy a distinct region of the color diagrams.

3.2. Photometric Redshift Selection

3.3. Completeness and Purity

The resulting observed color–color diagrams are shown in Figure 2. We select five candidates at z > 9 by combining the samples from the two selections described in Section 3. One additional candidate is selected from the photometric redshift analysis described in Section 3.2. The positions and magnitudes of the candidates are listed in Table 1. All objects are selected with both the 2FWHM and the 0''.2 aperture colors, with the exception of GHZ5 whose F115W − F150W color is slightly outside the selection boundary when using the 2FWHM ones. The six selected objects have been visually inspected to ensure that they are not affected by defects such as spurious sources, stellar spikes, etc. We found that GHZ2 is very compact, and it is close (∼0''.5) to a bright, foreground galaxy that contaminates the total flux measured within the Kron ellipse. We thus estimated its total flux using the T-PHOT software (Merlin et al. 2015, 2016b), which has been shown to provide accurate photometry for blended sources. We used the GHZ2 light profile in the F277W band, which provides the best compromise between resolution, S/N, and contamination from the low-redshift neighbor, as a high-resolution prior to reextract the F444W photometry. We found a total magnitude F444W = 27.21 ± 0.04. Finally, we have used GALFIT (Peng et al. 2010) to fit its light profile in both the F200W and F444W bands in order to measure its spatial extent and to further assess the measurement of the total flux under different assumptions. We used a Sersic (1968) model and fixed the index n at different values from n = 0.5 to n = 4. We found that GHZ2 is compact but resolved, with effective radius 1.0 ± 0.5 pixels. The n = 1 fit provides the best residuals. We found from GALFIT a total F444W magnitude of F444W = 27.25 ± 0.20, in agreement with the T-PHOT measurement but with a larger error due to the systematic uncertainty on the profile parameters. We adopt this conservative error estimate in the forthcoming analysis.

Zoom In Zoom Out Reset image size

Table 1. Galaxy Candidates at z > 9 in GLASS-JWST-ERS ^a

ID	R.A.	Decl.	F444W	zEAzY	zzphot	MUV	β	SFR	Re	Selection
	(deg.)	(deg.)						(M⊙ yr−1)	(kpc)
GHZ1	3.511929	−30.371848	26.36 ± 0.05	10.53	10.63	−20.98 ± 0.06	−1.99 ± 0.10	25${}_{-16}^{+68}$	0.43 ± 0.02	(1), (2)
GHZ2 b	3.498985	−30.324771	27.21 ± 0.20	12.11	12.30	−21.19 ± 0.20	−3.00 ± 0.12	${20}_{-13}^{+14}$	0.12 ± 0.01	(2)
GHZ3	3.528937	−30.363811	26.73 ± 0.07	2.69	9.33	−20.69 ± 0.09	−1.85 ± 0.17	${31}_{-8}^{+10}$	0.88 ± 0.09	(1), (2)
GHZ4	3.513743	−30.351554	27.74 ± 0.12	10.08	9.93	−19.98 ± 0.27	−2.86 ± 0.55	${5}_{-3}^{+19}$	0.39 ± 0.09	(1), (2)
GHZ5	3.494437	−30.307620	27.25 ± 0.08	2.49	9.20	−20.17 ± 0.18	−1.82 ± 0.33	${18}_{-15}^{+21}$	0.21 ± 0.08	(1)
GHZ6	3.479054	−30.314925	27.43 ± 0.11	9.85	9.05	−19.66 ± 0.21	−1.67 ± 0.38	${10}_{-8}^{+30}$	0.45 ± 0.09	photo z

Notes.

^a Coordinates and fluxes are from the GLASS-JWST catalog by M22; the rest-frame M_UV has been obtained by converting the F200W magnitude at the average photometric redshift, except for GHZ3 and GHZ5 where the only solution at z > 9 was used. The UV slope β is obtained by fitting the F200W, F277W, and F356W bands; the uncertainties in the fit account for photometric errors (Castellano et al. 2012). The R_e has been estimated from the F200W image by Yang et al. (2022b). Physical parameters from zphot (Santini et al. 2022). The last column indicates selections (1) and (2) in Section 3, and additional photo-z candidates not meeting color selection criteria. ^b All errors are computed considering the systematic uncertainty from different photometric techniques (Section 4).

Download table as:  ASCIITypeset image

Table 1 also reports the UV rest-frame magnitudes estimated from the observed F200W at the median photometric redshift, the UV slope obtained by a fit on the observed F200W, F277W, and F356W bands, the effective radius (Yang et al. 2022b), and the SFR from the SED fitting performed with zphot (Santini et al. 2022). We show in Figures 3 and 4 the best-fit SEDs, the probability density function (z), and thumbnails for all candidates.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We highlight the detection of the two bright sources GHZ1 and GHZ2, at z ≃ 10.6 and z ≃ 12.2, respectively. These objects are located in the color–color diagram fairly away from color selection borders, and have no significant low-redshift solution. Their P(z) is single peaked, the 68% confidence level photometric redshift range being z_zphot = 9.68–11.15 (z_EAzY = 10.28–10.80) for GHZ1, and z_zphot = 12.03–12.58 (z_EAzY = 11.85–12.39) for GHZ2. They are among the most robust candidates at z > 9 ever detected, as demonstrated by the several recent works that have presented their selection with redshift and properties comparable to those discussed here (Donnan et al. 2022; Harikane et al. 2022b; Naidu et al. 2022). They are remarkably bright (M_UV ≲ −21) and star forming (SFR ≳ 20M_⊙ yr⁻¹). GHZ2 is compact with R_e = 120 ± 10 pc (Yang et al. 2022b), and it has a very steep UV slope β = −3.0. This extreme value is partly driven by the F200W band, as the slope value is β = −2.4 ± 0.2 when using the F277W and F356W bands only. These UV slope estimates are anyway indicative of very young ages and very-low-metallicity, possibly pristine, stellar populations, which will need to be confirmed through JWST spectroscopy. A third bright candidate (GHZ3, M_UV ∼ −20.7, SFR ∼ 30 M_⊙ yr⁻¹) is also selected by both of our color diagrams, but its photometric redshift is more uncertain: it is found to be at z ∼ 9.3 by zphot, while EAzY gives a best-fit solution at z ∼ 2.7. The remaining candidates are fainter (M_UV ≳ −20.2) galaxies likely at z ∼ 9–10.

The GHZ3-6 candidates exhibit in general alternative lower-z solutions with nonnegligible probability. These lower-z solutions are driven in part by our conservative error estimates, especially in the bluer bands, accounting for data reduction and calibration systematics (M22). However, definitive confirmation will require deep imaging at shorter wavelengths. Instead, the contamination from galactic stars can be excluded because all objects show resolved profiles as discussed by Yang et al. (2022b). ²⁵

The physical properties, morphology, and size of our candidates are discussed in detail in the companion papers by Santini et al. (2022), Treu et al. (2022), and Yang et al. (2022b) respectively.

We have presented a search for galaxy candidates at z ∼ 9–15 from the seven-band, mag ∼29 (5σ) NIRCam imaging observations, acquired by the GLASS-JWST Early Release Science Program in parallel with NIRISS primary observations of the cluster A2744. The NIRCam images enable the first multiband selection of Lyman-break candidates at z ∼ 9–15 through two independent color–color diagrams.

We have identified two bright (M_UV ≃ − 21) sources that, thanks to the depth of our images, are unambiguously placed at z = 10.6 and z ≃ 12.2, respectively. These are among the most reliable candidates ever selected at z > 9 via photometric techniques.

We have also identified five other fainter candidates at z > 9. Some of these objects have significant alternative solutions at lower redshift, that it is difficult to exclude given the limits that are currently achieved by our data, and residual systematic uncertainties especially in the bluest bands. Definitive confirmation will require deeper images at short wavelengths as well as NIRSpec spectroscopic follow-up.

Keeping in mind the uncertainties associated with our limited area and to the presence of a foreground cluster that may boost the luminosity of our candidates, it is intriguing to explore the potential implications of our results on the evolution of galaxies at large redshift.

Noting that the F444W ∼ 28.2 limit corresponds to M_UV ≲ −19. (−19.5) at z = 10(13), we estimate the average number of sources expected in our field under several assumptions about the evolution of the LF.

At z ∼ 9–11, we expect ${1.7}_{-0.6}^{+1.1}$ on the basis of the z ∼ 9 and z ∼ 10 UV LFs by Bouwens et al. (2021) and Oesch et al. (2018), respectively, where the uncertainties consider the relevant 68% c.l. confidence range in the LF parameters. The double power-law (DPL) model by Bowler et al. (2020) predicts ${2.4}_{-1.9}^{+1.6}$ galaxies. We expect as many as ${6.8}_{-1.8}^{+2.9}$ objects at z ∼ 9–11 on the basis of the model by Mason et al. (2015).

At z > 11, where no direct constraints exists, we can estimate the average number of sources with M_UV < −19.5 under four scenarios. In the first (optimistic) case, we assume that the z = 10 UV LF remains constant up to z = 15 and predict ${0.6}_{-0.3}^{+0.5}$ objects in one GLASS-JWST-ERS parallel field. A perhaps more realistic scenario is provided by the evolving UV LF by Bouwens et al. (2021) that extrapolates at z > 11 the observed evolution in the Schechter (1976) parameters of the UV LFs measured at lower redshifts. In this second scenario, on average, only ${0.1}_{-0.05}^{+0.1}$ objects fall within one NIRCAM parallel field. A similar number of ${0.08}_{-0.03}^{+0.06}$ objects is predicted based on Mason et al. (2015). Finally, the extrapolation at z = 11–15 of the evolving DPL model by Bowler et al. (2020) predicts ${0.5}_{-0.2}^{+0.3}$ objects.

We note that such predictions are affected by cosmic variance at the level of ∼40%. ²⁶ We also note that the adopted M_UV limits may be altered by photometric scatter. In combination with the steep increase in the number counts beyond the knee of the LF, a nonnegligible number of fainter sources can thus be scattered within the observable range. As a reference, by adopting 0.5 mag fainter M_UV cuts, the predicted number of objects at z ∼ 9–11 increases to 3.5 on the basis of the observed UV LFs, and to 7.5 in the Mason et al. (2015) model. The predicted number of sources at z > 11 increases to 1.5 (0.2) assuming a constant (evolving) UV LF as above, and to 0.26 on the basis of Mason et al. (2015).

Considering the incompleteness of our sample, as estimated above, we conclude that the total number of detected sources is roughly in line with the expectations of a nonevolving LF.

What is most remarkable of this first search is that we have found two bright (M_UV ≲ −21) sources, one at z ∼ 10 and one at z ∼ 12, well beyond the expectations based on the extrapolation of the LF at lower redshift. In fact, all observations and models predict a negligible number of sources brighter than this limit in the redshift range z = 9–15 on an area equal to the GLASS one: the predictions being of ≲0.2 objects, compared to an observed number of ${2}_{-1.4}^{+2.7}$ (where the uncertainty includes both cosmic variance and the Poisson uncertainty for a small number of events computed following Gehrels 1986). The aforementioned predictions are summarized in Table 2. Clearly it is premature to draw broad conclusion based on a single field, considering small number statistics, cosmic variance, clustering, and lensing. In particular, the impact of lensing on these results will need to be assessed. A preliminary investigation (based on the model by Bergamini et al. 2022) suggests that magnification may be 2–10 times higher than the 10%–15% value measured in the HFF Abell2744 Parallel (Castellano et al. 2016a). At the present stage, this indication is highly uncertain because it is extrapolated from the cluster core to the NIRCam field 3' away, and most importantly, it is not possible to obtain lensing constraints for single sources. We note that our candidates do not show any evidence of being sheared, so although magnification of up to 50% is possible, it is unlikely to be much higher. A thorough analysis will be performed upon availability of additional spectroscopic information in the GLASS-JWST-ERS field. Despite the aforementioned sources of uncertainty, these results show that JWST can obtain the scientific results it has been designed for.

Table 2. Predictions on the Number of z > 9 Objects in GLASS-JWST-ERS ^a

UV LF	z = 9–11		z > 11
	MUV < −21.0	MUV < −19.0	MUV < −21.0	MUV < −19.5
Oesch+18 b	<0.06	${1.7}_{-0.6}^{+1.1}$	<0.04	${0.6}_{-0.3}^{+0.5}$
LF(z) Bouwens+21 c	''	''	<0.01	${0.1}_{-0.05}^{+0.1}$
Mason+15 d	${0.16}_{-0.05}^{+0.07}$	${6.8}_{-1.8}^{+2.9}$	0.002 ± 0.001	${0.08}_{-0.03}^{+0.06}$
Bowler+20 e	${0.06}_{-0.03}^{+0.30}$	${2.4}_{-1.9}^{+1.6}$	${0.05}_{-0.04}^{+0.1}$	${0.5}_{-0.2}^{+0.3}$

Notes.

^a Uncertainties on the predicted numbers are based on the 68% c.l. errors on the Schechter parameters. ^b UV LF at z ∼ 10 by Oesch et al. (2018), assumed fixed at z > 11. ^c Extrapolation at z > 11 of the UV LF as a function of redshift by Bouwens et al. (2021). ^d The semiempirical UV LF model by Mason et al. (2015). ^e Extrapolation of the DPL model as a function of redshift by Bowler et al. (2020).

Download table as:  ASCIITypeset image

There is no doubt that the deeper and/or wider surveys already planned will eventually gather the long-awaited-for sample of galaxies in the reionization epoch and revise our understanding of how and when bright galaxies formed.

This first JWST-based search for candidate galaxies also provides us with compelling targets for spectroscopic follow-up in JWST Cycle-2. In addition to providing the necessary redshift confirmation, the relatively high luminosity of these galaxies means that NIRSpec spectroscopy should enable the study of their physical properties such as SFR, dust content, and gas metallicity.

This work is based on observations made with the NASA/ESA/Canadian Space Agency James Webb Space Telescope. The data were obtained from the Mikulski Archive for Space Telescopes at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-03127 for JWST. These observations are associated with program JWST-ERS-1324. The JWST data used in this paper can be found on MAST: 10.17909/fqaq-p393. We acknowledge financial support from NASA through grants JWST-ERS-1342. K.G. and T.N. acknowledge support from Australian Research Council Laureate Fellowship FL180100060. C.M. acknowledges support by the VILLUM FONDEN under grant 37459. The Cosmic Dawn Center (DAWN) is funded by the Danish National Research Foundation under grant DNRF140. We acknowledge support from INAF Minigrant "Reionization and fundamental cosmology with high-redshift galaxies." We thank Pascal Oesch and Rohan Naidu for the useful feedback and discussion.

Software: Astropy (Astropy Collaboration et al. 2013), EAzY (Brammer et al. 2008), Empirical Galaxy Generator (Schreiber et al. 2017b), Matplotlib (Hunter 2007), SExtractor (v2.8.6 Bertin & Arnouts 1996; Guo et al. 2013), T-PHOT (Merlin et al. 2015, 2016b), zphot (Fontana et al. 2000).