The JWST EXCELS survey: Too much, too young, too fast? Ultra-massive quiescent galaxies at 3 < z < 5

TL;DR

The JWST EXCELS survey presents deep, medium-resolution NIRSpec spectra of four ultra-massive quiescent galaxies at $3<z<5$, including a $z=4.62$ pair, a compact $z=3.99$ PSB, and a fossil at $z=3.19$. Through full spectral fitting with Bagpipes and supplementary ALF analysis, the study derives detailed star-formation histories, metallicities, sizes, and kinematics, revealing formation epochs $z_{form}\approx5.5-11$ and extreme compactness ($r_e\sim$ few hundred pc) with high central densities. Using extreme value statistics, the authors compare these galaxies to the $oldsymbol{ m \Lambda CDM}$ halo-mass function under two stellar-fraction models, finding that three systems are difficult to accommodate with fiducial $f_*$ values but can be reconciled if $f_*=1$, implying exceptionally efficient early star formation. The work concludes that there is no fundamental conflict with $oldsymbol{\Lambda CDM}$ cosmology, but the results point to extraordinary baryonic physics during the first billion years and highlight the need for further JWST follow-up, including IFU data, to fully map the assembly and merger histories of these extreme systems.

Abstract

We report ultra-deep, medium-resolution spectroscopic observations for 4 quiescent galaxies with log$_{10}(M_*/\mathrm{M_\odot})>11$ at $3 < z < 5$. These data were obtained with JWST NIRSpec as part of the Early eXtragalactic Continuum and Emission Line Science (EXCELS) survey, which we introduce in this work. The first two galaxies are newly selected from PRIMER UDS imaging, both at $z=4.62$ and separated by $860$ pkpc on the sky, within a larger structure for which we confirm several other members. Both formed at $z\simeq8-10$. These systems could plausibly merge by the present day to produce a local massive elliptical galaxy. The other two ultra-massive quiescent galaxies are previously known at $z=3.99$ and $3.19$, with the latter (ZF-UDS-7329) having been the subject of debate as potentially too old and too massive to be accommodated by the $Λ$-CDM halo-mass function. Both exhibit high stellar metallicities, and for ZF-UDS-7329 we are able to measure the $α-$enhancement, obtaining [Mg/Fe] = $0.42^{+0.19}_{-0.17}$. We finally evaluate whether these 4 galaxies are consistent with the $Λ$-CDM halo-mass function using an extreme value statistics approach. We find that the $z=4.62$ objects and the $z=3.19$ object are unlikely within our area under the assumption of standard stellar fractions ($f_*\simeq0.1-0.2$). However, these objects roughly align with the most massive galaxies expected under the assumption of 100 per cent conversion of baryons to stars ($f_*$=1). Our results suggest extreme galaxy formation physics during the first billion years, but no conflict with $Λ$-CDM cosmology.

Figures (10)

• Figure 1: Spectral energy distributions (SEDs) for the 4 most massive $z>3$ quiescent galaxy candidates selected from our PRIMER catalogue by the process described in Section \ref{['subsec:primer_passive']}. These are the only 4 candidates at $z>3$ that have inferred stellar masses, log$_{10}(M_*/\mathrm{M_\odot}) > 11$. Photometric data are shown with blue points and our posterior-median fitted models with red lines. All 4 were observed spectroscopically as part of EXCELS, and their spectra are shown in Fig. \ref{['fig:spec2']}. The top two are our newly confirmed $z=4.62$ galaxies. The bottom two were previously selected by Schreiber2018 and spectroscopically confirmed by Nanayakkara2024. The top two objects also have PRIMER MIRI coverage, and we include the F770W fluxes, which are consistent with our fitted models.
• Figure 2: Stellar mass vs redshift distribution for the 341 out of 401 EXCELS galaxies that could be assigned secure (flag 3, 4 or 9) spectroscopic redshifts (see Section \ref{['subsec:method_specz']}). Stellar masses were measured by repeating the photometric fitting analysis described in Section \ref{['subsec:primer_passive']} with redshifts fixed to the values derived from the EXCELS spectra. The EXCELS sample has been split into quiescent galaxies (red squares), star-forming galaxies (blue large circles) and filler galaxies (green small circles). The selection of these sub-samples is described in Section \ref{['sec:excels_sample']}. The bars at the top show the redshift ranges over which key spectral features are visible in the three medium-resolution gratings used. A full list of EXCELS galaxy IDs, coordinates, spectroscopic redshifts and their associated quality flags is given in Table \ref{['table:redshifts']}.
• Figure 3: JWST EXCELS NIRSpec observations of our 4 ultra-massive quiescent galaxies at $3 < z < 5$: zoom in on the rest-frame $3540-7350$Å region included in our Bagpipes full-spectral-fitting analysis (see Section \ref{['subsec:spec_fitting']}). The spectroscopic data are shown in blue, with PRIMER NIRCam photometry shown as red points. The posterior-median fitted Bagpipes models are shown with black lines. The vertical blue shaded regions were masked from the fits. The spectra and our full-spectral-fitting results are described in Section \ref{['sec:results']}.
• Figure 4: Star-formation histories for our 4 ultra-massive quiescent galaxies at $3 < z < 5$ from full spectral fitting. To the left the SFR as a function of time is shown, whereas to the right the total mass in stars as a function of time is shown. Results for GS-9209 Carnall2023c at $z=4.658$, which is $\simeq0.4-0.5$ dex less massive than the other galaxies, are also shown in grey. Three of the new galaxies are older than GS-9209, having formed at $z\gtrsim8$, whereas ZF-UDS-6946 is younger, having formed in a very rapid burst at $z\simeq5.5$. It is instructive to view the shaded areas as confidence intervals on SFR and stellar mass at fixed redshift (i.e., in the vertical direction). Taking PRIMER-EXCELS-109760 as an example, stellar masses of both log$_{10}(M_*/\mathrm{M_\odot})\simeq10.5$ and log$_{10}(M_*/\mathrm{M_\odot})<9$ are within the 1$\sigma$ contour at $z=12$. The right panel therefore indicates we have virtually no constraint on the stellar mass of this galaxy before $z\simeq10$.
• Figure 5: PRIMER F277W cutout images ($4^{\prime\prime}\times4^{\prime\prime}$) for our 4 ultra-massive quiescent galaxies. The positions of the open NIRSpec MSA shutters in the EXCELS G235M observations are shown in red. The positions shown are for the first of three nod positions. Objects were shifted to the top and then bottom shutters for equal thirds of the total exposure time. All four objects are extremely compact ($r_e$ < 1 kpc). PRIMER-EXCELS-117560 and ZF-UDS-7329 appear significantly elongated, whereas the other two objects appear almost round.