betaFeedback

WorksPhysics and Astronomy

EMBER: A Little Red Dot Candidate at zspec=10.36z_{\rm spec}=10.36 in the Seed Black Hole Era

Abstract

Little red dots (LRDs) are a newly discovered population that may trace rapidly growing black holes (BHs) and provide a link between seed BHs and luminous active galactic nuclei (AGNs). However, no LRD had previously been spectroscopically confirmed at z>10z>10, leaving this earliest phase observationally unexplored. Using JWST/NIRSpec spectroscopy from the EMBER program, we measure a spectroscopic redshift of zspec=10.361−0.002+0.003z_{\rm spec}=10.361^{+0.003}_{-0.002} for \target, the highest-redshift LRD candidate known to date. The redshift is constrained by a prominent Lyαα break and multiple tentative emission lines, including C,{\sc iv},λλ1549,1551λ\lambda1549,1551 and [O,{\sc iii}],λ4363\lambda4363 detected at S/N=3.9S/N=3.9 and 3.63.6, respectively. Its extreme NIRCam-MIRI color of mF444W−mF770W=1.82±0.1m_{\rm F444W}-m_{\rm F770W}=1.82\pm0.1, characteristic V-shaped continuum with a turnover near the Balmer limit, and continuum slopes of βUV=−2.26−0.11+0.11β_{\rm UV}=-2.26^{+0.11}_{-0.11} and βopt=1.17−0.70+0.74β_{\rm opt}=1.17^{+0.74}_{-0.70} all support its LRD nature. A possible iron complex at λrest∼0.24λ_{\rm rest}\sim0.24–0.25 μm0.25\,{\rm μm} and a possibly high log⁡([O III] λ4363/Hγ)=0.17−0.22+0.24\log\left([{\rm O\,III]\,\lambda4363/\rm Hγ}\right)=0.17^{+0.24}_{-0.22} further support an AGN-like interpretation, as seen in previously known LRDs. Tentative detection of Lyαα emission may indicate enhanced Lyαα transmission at z∼10z\sim10. Our findings extend the observational frontier of BH growth into the seed-BH formation era, and upcoming MIRI spectroscopy will further establish the physical nature of the source.

Introduction

Supermassive black holes (SMBHs) are ubiquitous in massive galaxies in the local Universe and are thought to co-evolve with their host galaxies (e.g., [60]). However, the origin of SMBHs remains one of the major unsolved problems in modern astrophysics (e.g., [113, 43, 89]).

As a possible key population for uncovering the origin of SMBHs, “little red dots” (LRDs) have recently attracted significant attention (e.g., [61, 25, 76, 55, 34, 6, 2, 33, 37, 40]). LRDs were first revealed by JWST thanks to its unprecedented sensitivity at infrared wavelengths ([90]) and are characterized by rest-optical compact morphologies and a unique V-shaped spectral energy distribution (SED), with a blue rest-UV component and a steep, red rest-optical component (e.g., [61, 25, 76, 55, 6, 2]). These components transition near the Balmer limit (e.g., [95]) and often show a Balmer break (e.g., [58, 62, 114]), and in some cases exhibit an extremely strong Balmer break that cannot be explained by stellar populations alone (e.g., [17, 80, 105]).

Spectroscopic observations have revealed broad Balmer emission lines with absorption features in many LRDs (e.g., [63, 25, 76, 29]). Despite having broad Balmer lines similar to those of type-I active galactic nuclei (AGN), LRDs show several unusual properties that distinguish them from typical AGNs. First, most LRDs are undetected in X-rays [120, 3, 2] (R. Maiolino et al. 2025), with rare exceptions including objects proposed to be in transition from LRDs to conventional AGNs [38] (S. Fu et al. 2026). LRDs are also generally undetected in radio [2, 88, 78, 28], although radio emission has been detected in some low-zz LRD analogs [92, 93]. Recent studies have revealed hot-dust emission in individual LRDs and stacked samples [20, 67, 87, 8] (X. Ji et al. 2026a). However, the detection of hot-dust emission does not necessarily imply that dust attenuation is responsible for the steep red rest-optical continua of LRDs, and the relative contributions of dust attenuation and other emission mechanisms remain uncertain [20, 96]. Many LRDs have not shown significant photometric variability [59, 121, 106, 123], although photometric variability has been reported in gravitationally lensed high-redshift LRDs in a century time-scale [122] and low-zz analogs (R. Lin et al. 2026). Multi-epoch observations have also revealed variations in broad Balmer-line equivalent widths and emission-line fluxes in individual LRDs [26, 64, 9, 80] (X. Ji et al. 2025), whereas recent systematic monitoring studies have found substantially weaker Hα\alpha variability in LRDs than expected for typical AGNs [69, 100]. In summary, these unusual properties leave their physical nature uncertain.

One proposed framework is the BH envelope (or BH-star) model [e.g. 80, 54, 42], in which a BH rapidly grows from a seed, possibly through super-Eddington accretion, while surrounded by dense, optically thick gas with an effective temperature of ∼4000\sim4000–50005000 K. The rest-optical-to-near-infrared SEDs of LRDs can be reproduced by blackbody-like emission from BH envelopes, whose Wien tails explain the steep rest-optical continua [e.g. 67, 54, 110] (A. de Graaff et al. 2026). Furthermore, dense, predominantly neutral gas (nH≳109n_{\mathrm{H}} \gtrsim10^{9} cm−3^{-3}) can produce strong Balmer absorption and a Balmer break through collisional excitation of hydrogen to the n=2n=2 state [41, 118]. However, the geometry and physical properties of the surrounding gas remain uncertain, and the gas may have a covering factor below unity, as in some clumpy, disk-like, or toroidal configurations [86, 104, 15, 124, 119] (X. Ji et al. 2026b). In addition, dense ionized gas with electron column densities of Ne∼1024N_{\mathrm{e}} \sim10^{24} cm−2^{-2} may contribute to broad Balmer-line profiles through electron scattering [94], suggesting that the emitting, absorbing, and scattering regions may not be identical. Other models invoke nebular emission from ionized gas cocoons [98 b], dense accretion flows around direct-collapse BHs [85], or dust-reddened AGN emission [70, 71] to explain the characteristic red rest-optical continua of LRDs. Therefore, we have not yet achieved a unified model of the nature of LRDs.

Regardless of the specific physical framework invoked to explain their unusual properties, LRDs may trace an early stage of BH growth, potentially close to the epoch of BH seeding. This picture is supported by K. Inayoshi (2025), who compared the number density evolution of LRDs with that of non-LRD AGNs and suggested that LRDs may represent the first episode of BH growth after seed BH formation. Therefore, LRDs provide a crucial population for observationally probing seed BHs, which have long been a theoretical target of study [113, 43, 89]. However, almost all previous LRD studies have been limited to z<9z < 9. Prior to this work, the most distant LRD with a spectroscopic redshift was CAPERS-LRD-z9 at z=9.288z = 9.288 with a broad Hβ\beta detection [105]. More recently, X. Cai et al. [11] reported a broad-line AGN at z=9.984z = 9.984 and discussed its possible classification as an LRD. Extending LRD studies to z>10z > 10 is crucial for constraining their number-density evolution during their emergence and for uncovering when and through which mechanisms seed BHs formed, a regime that has remained observationally inaccessible until now.

As shown in Figure 1 of T. S. Tanaka et al. [102], searching for z>10z > 10 LRDs requires the mid-infrared coverage provided by MIRI to sample their steep rest-optical slope and distinguish LRDs from galaxies. Motivated by this, T. S. Tanaka et al. [102] developed a joint NIRCam-MIRI color selection method and applied it to the COSMOS-Web imaging data (GO#1727, PIs J. Kartaltepe and C. M. Casey; C. M. Casey et al. 2023) and identified a promising candidate for the first z>10z > 10 LRD, CW-LRD-z10, with a photometric redshift

JWST/NIRSpec PRISM spectrum of CW-LRD-z10

Figure 1. JWST/NIRSpec PRISM spectrum of CW-LRD-z10. The top panel shows the nod-subtracted 2D spectrum produced by the JWST science calibration pipeline. The middle panel shows the optimally extracted 1D spectrum, and the bottom panel compares the spectrum, binned by taking the inverse-variance-weighted average over every four spectral channels, with the photometry reported in T. S. Tanaka et al. [102]. Uncertainties (1σ1\sigma) are shown as the orange shaded region. The inset in the middle panel shows the NIRCam 3′′×3′′3^{\prime\prime} \times3^{\prime\prime} RGB (R: F356W+F444W; G: F200W+F277W; B: F115W+F150W) image overlaid with the NIRSpec slitlet, while the upper-right inset shows a 2′′×2′′2^{\prime\prime} \times2^{\prime\prime} F770W image. In the bottom panel, we overplot a stacked spectrum of 274 LRDs compiled from the EMBER survey and archival observations with the dashed line ((H. B. Akins et al. in prep.)), scaled to match the continuum of CW-LRD-z10 over λrest=0.28−0.32 μm\lambda_{\mathrm{rest}} = 0.28 - 0.32\ \mu\mathrm{m}. The spectrum exhibits the Lyα\mathrm{Ly}\alpha break at ∼1.3−1.4 μm\sim1.3 - 1.4\ \mu\mathrm{m} and a slight upturn redward of ∼4.2 μm\sim4.2\ \mu\mathrm{m}, consistent with a VV-shaped SED of typical LRDs redshifted to z∼10z \sim10.

of zphot=10.5−0.6+0.7z_{\mathrm{phot}} = 10.5^{+0.7}_{-0.6}. They used this candidate to place the first constraint on the abundance of LRDs at z∼10z \sim10, ΦLRD=1.2−1.0+2.7×10−6 Mpc−3 mag−1\Phi_{\mathrm{LRD}} = 1.2^{+2.7}_{-1.0} \times10^{-6}\ \mathrm{Mpc}^{-3}\ \mathrm{mag}^{-1} for MUV=−20M_{\mathrm{UV}} = -20, suggesting the redshift evolution of their number density following the log-normal evolution hypothesized by K. Inayoshi (2025). However, CW-LRD-z10 had not yet been spectroscopically confirmed, leaving both its nature and the inferred z∼10z \sim10 LRD abundance dependent on the photometric interpretation.

In this paper, we report NIRSpec spectroscopic observations of CW-LRD-z10. In Section 2, we describe the data. In Sections 3 and 4, we present the spectroscopic analysis and the results, respectively. Then Section 5 discusses the nature of CW-LRD-z10, and Section 6 provides the summary. Throughout this work, we assume a flat Λ\LambdaCDM cosmology with H0=67.4 km s−1 Mpc−1H_0 = 67.4\ \mathrm{km\ s}^{-1}\ \mathrm{Mpc}^{-1}, Ωm=0.315\Omega_{\mathrm{m}} = 0.315, and ΩΛ=0.685\Omega_{\Lambda} = 0.685 (Planck Collaboration et al. [1]).

Data

CW-LRD-z10 at (R.A., Decl.) = (10h00m20s.183610^\mathrm{h}00^\mathrm{m}20^\mathrm{s}.1836, +02∘27′14′′.022+02^\circ27'14''.022) was observed with JWST/NIRSpec Multi-Object Spectroscopy as part of the Early Massive Black Hole Evolution & Reddening (EMBER) program (GO#7076, PI H. Akins). EMBER was designed to conduct a comprehensive spectroscopic census of the brightest LRDs selected across multiple extragalactic fields. While the highest priority EMBER targets were bright (mF444W<24m_{\mathrm{F444W}} < 24) z∼3z \sim3–7 LRDs, CW-LRD-z10 was selected for observation based on its robust photometric redshift of zphot=10.5−0.6+0.7z_{\mathrm{phot}} = 10.5^{+0.7}_{-0.6} and previous identification as an LRD candidate based on MIRI photometry [102]. CW-LRD-z10 was observed as part of Observation 1 of the EMBER program on November 27, 2025, with on-source exposure times of 1.48 hr using the PRISM/CLEAR disperser/filter pair (0.6–5.3 μ\mum, R∼100R \sim100) and 0.92 hr using G395M/F290LP (2.87–5.10 μ\mum, R∼1000R \sim1000).

We process the NIRSpec data starting from the uncalibrated uncal products using version 1.20.2 of the JWST Calibration Pipeline (H. Bushouse et al. [10]) and the Calibration Reference Data System (CRDS) mapping pmap-1481. The reduction follows the standard workflow implemented by the pipeline, with a few modifications to improve the data quality (described in more detail in H. B. Akins et al. in prep). First, following the detector-level corrections applied in stage 1, we employ a custom iterative procedure to subtract the thermal “picture frame” effect and 1/f1/f noise on the detector, masking spectral traces from all open micro-shutter assembly (MSA) shutters. The picture frame subtraction uses the recently released CRDS pictureframe reference file, and is particularly important for reliable background subtraction. We extract the 2D spectra for CW-LRD-z10 from each nod in stage 2 and apply flat-field, photometric calibration, and wavelength calibration. We inspect each nod to ensure that no shutters were stuck closed, and then perform the standard nodded background subtraction and combination of the individual exposures into rectified products. We extract a 1D spectrum from the combined 2D product using the optimal-extraction method of K. Horne (K. Horne [35]) with a profile constructed as the median of the rectified 2D spectrum along the cross-dispersion direction. We note that the G395M spectrum for CW-LRD-z10 is contaminated by the spectral trace from a nearby bright object, as G395M spectra were allowed to overlap in the MSA design. Given the contamination and the faint nature of this source, the analysis here focuses only on the deeper PRISM spectrum. We present the processed NIRSpec/PRISM spectrum in Figure 1. The spectrum is consistent with the observed photometry, supporting the robustness of the pipeline slit-loss correction.

Spectroscopic fitting

To derive the spectroscopic redshift and characterize CW-LRD-z10, we perform Markov Chain Monte Carlo spectrum fitting using emcee (D. Foreman-Mackey et al. [23]). For the intrinsic continuum, we adopt a broken power law (BPL) with independent rest-frame UV and optical slopes, βUV\beta_{\mathrm{UV}} and βopt\beta_{\mathrm{opt}} (fλ∝λβf_{\lambda} \propto\lambda^{\beta}). The two power laws are joined continuously at a turnover rest-frame wavelength, λturn\lambda_{\mathrm{turn}}, and the overall normalization is parameterized by the absolute magnitude in rest-frame 1450 Å (MUVM_{\mathrm{UV}}). To test whether the BPL model is required, we additionally fit a single power-law (SPL) continuum and compare the goodness of fit using the Bayesian information criterion (BIC).

We include Gaussian emission lines for Lyα\alpha, C III] λ1335\lambda1335, N IV] λλ1483,1487\lambda\lambda1483, 1487, C IV λλ1549,1551\lambda\lambda1549, 1551, [Ne III] λ3869\lambda3869, [Ne III] λ3967\lambda3967 + Hϵ\epsilon, Hδ\delta, Hγ\gamma, and [O III] λ4363\lambda4363. Because these lines are unresolved at the spectral resolution of NIRSpec/PRISM, their observed profiles are modeled using a Gaussian line-spread function (LSF) with a wavelength-dependent resolution for a point source provided by the unite package (R∼50R \sim50–500; R. E. Hviding [36]). This prescription assumes a point source centered in an MSA shutter, which is a reasonable approximation for the observation of CW-LRD-z10 (Figure 1), and accounts for the higher effective spectral resolution of compact sources compared to the nominal resolution derived assuming uniform illumination across the shutter (e.g., see A. de Graaff et al. 2024). Note that, at the fitted spectroscopic

Note— The quoted uncertainties correspond to the 16th–84th percentile ranges of the posterior distributions. All lines except the iron complex and Lyα\alpha are assumed to be unresolved, with their observed widths determined by the wavelength-dependent NIRSpec/PRISM line-spread function. The iron complex is treated as a resolved feature, with a fitted effective rest-frame central wavelength of 0.2432−0.0010+0.0011 μm0.2432^{+0.0011}_{-0.0010}\,\mu\mathrm{m} and an intrinsic FWHM of 1.22−0.32+0.58×104 km s−11.22^{+0.58}_{-0.32} \times10^{4}\ \mathrm{km\ s^{-1}}.

redshift of zspec=10.36z_{\mathrm{spec}} = 10.36 (§ 4.1), [O iii] λ4363\lambda4363 and Hγ\gamma are observed near λobs≃5 μm\lambda_{\mathrm{obs}} \simeq5\,\mu\mathrm{m}, where the PRISM resolving power reaches R≃450R \simeq450 based on the unit LSF, corresponding to a rest-frame spectral resolution of Δλrest∼10 A˚\Delta\lambda_{\mathrm{rest}} \sim10\,\text{\AA}. Therefore, [O iii] λ4363\lambda4363 and Hγ\gamma, separated by ∼23 A˚\sim23\,\text{\AA} in the rest frame, are spectrally distinguishable in principle.

Given the limited spectral resolution of PRISM, we fit the unresolved N iv] λλ1483\lambda\lambda1483, 1487 and C iv λλ1549\lambda\lambda1549, 1551 doublets with a single Gaussian profile each, and tie the centroids of all lines except Lyα\alpha and the iron complex to a common systemic redshift without allowing individual velocity offsets. Thus, the only free parameter for each of these unresolved lines is its line flux. Note that the iron complex, which likely consists of a blend of multiple iron emission lines, is treated separately, and both its intrinsic FWHM and effective rest-frame central wavelength are fitted as free parameters.

Lyα\alpha is also treated separately, and both its intrinsic FWHM and velocity offset are fitted as free parameters. We consider two model setups: (i) only intergalactic medium (IGM) absorption and (ii) both IGM and circumgalactic medium (CGM) Lyα\alpha absorption. IGM absorption is modeled following the formalism of T. Totani et al. (2006), which is based on J. Miralda-Escudé (1998), with two free parameters: the neutral hydrogen fraction xHIx_{\mathrm{HI}} and the ionized bubble radius RbR_{\mathrm{b}}. Note that, following H. Umeda et al. (2024), we fix the redshift corresponding to the end of reionization at zend=5z_{\mathrm{end}} = 5 and neglect the contribution of neutral IGM at z<zendz < z_{\mathrm{end}} to the damping-wing absorption. We confirm that varying zendz_{\mathrm{end}} between 5 and 6 has a negligible effect on the fitting results. For the CGM, we assume a Voigt absorption profile with three free parameters: the neutral hydrogen column density log⁡(NHI)\log(N_{\mathrm{HI}}), the velocity offset of the CGM from the galaxy Δv\Delta v, and the velocity dispersion of the CGM σCGM\sigma_{\mathrm{CGM}}.

We also fit an emission feature that possibly corresponds to an iron complex around λrest∼0.24\lambda_{\mathrm{rest}} \sim0.24–0.25 μm0.25\,\mu\mathrm{m} (see also § 4.5). Since the individual lines are blended and cannot be resolved, we model the entire feature with a single Gaussian profile, treating its FWHM and central wavelength as free parameters. We confirm that the overall feature can be adequately reproduced by a single broad Gaussian with FWHM ≃1.22−0.32+0.58×104 km s−1\simeq1.22^{+0.58}_{-0.32} \times10^{4}\ \mathrm{km\ s^{-1}} at the central wavelength of λrest=0.2432−0.0010+0.0011 μm\lambda_{\mathrm{rest}} = 0.2432^{+0.0011}_{-0.0010}\,\mu\mathrm{m}. We emphasize that the primary purpose of this fit is to assess the significance of the iron-complex detection by using the integrated flux derived from the fitted Gaussian component, and the fitted line width is not assigned any physical interpretation. The best-fit SEDs are shown in Figure 2, and the fitted parameters are shown in Table 1 and Table 2.

Zoomed-in versions of Figure 1 including emission lines

Figure 2. Zoomed-in versions of Figure 1 including emission lines. The middle rows show the unbinned original resolution spectrum and two fitted model SEDs: IGM+lines (dotted) and IGM+CGM+lines (dashed). The bottom panels show the residuals normalized by the 1σ1\sigma uncertainties. We identify multiple line features with S/N∼2−4S/N \sim2 - 4 at wavelengths consistent with zspec=10.36z_{\mathrm{spec}} = 10.36.

ModelContinuumAbsorptionLines†zspecz_{\mathrm{spec}}MUVM_{\mathrm{UV}}βUV\beta_{\mathrm{UV}}βopt\beta_{\mathrm{opt}}λturn\lambda_{\mathrm{turn}}log⁡Rb\log R_{\mathrm{b}}log⁡NHI\log N_{\mathrm{HI}}ΔBIC∗\Delta\mathrm{BIC}^{*}
BPLIGM+CGMIncluded10.361−0.002+0.00310.361^{+0.003}_{-0.002}−20.16−0.07+0.07-20.16^{+0.07}_{-0.07}−2.26−0.11+0.11-2.26^{+0.11}_{-0.11}1.17−0.70+0.741.17^{+0.74}_{-0.70}0.356−0.013+0.0100.356^{+0.010}_{-0.013}0.42−0.97+1.040.42^{+1.04}_{-0.97}22.81−0.25+0.2022.81^{+0.20}_{-0.25}—
BPLIGMIncluded10.361−0.002+0.00310.361^{+0.003}_{-0.002}−20.01−0.05+0.05-20.01^{+0.05}_{-0.05}−2.03−0.08+0.08-2.03^{+0.08}_{-0.08}1.01−0.69+0.711.01^{+0.71}_{-0.69}0.359−0.013+0.0110.359^{+0.011}_{-0.013}0.65−1.10+0.630.65^{+0.63}_{-1.10}—2.62.6
BPLIGM+CGMLyα\alpha only10.41−0.06+0.0610.41^{+0.06}_{-0.06}−20.13−0.08+0.07-20.13^{+0.07}_{-0.08}−2.20−0.12+0.12-2.20^{+0.12}_{-0.12}1.09−0.70+0.771.09^{+0.77}_{-0.70}0.355−0.014+0.0120.355^{+0.012}_{-0.014}0.56−1.06+1.100.56^{+1.10}_{-1.06}22.52−0.53+0.3122.52^{+0.31}_{-0.53}—
BPLIGMLyα\alpha only10.32−0.09+0.1610.32^{+0.16}_{-0.09}−20.00−0.06+0.05-20.00^{+0.05}_{-0.06}−2.03−0.09+0.08-2.03^{+0.08}_{-0.09}1.02−0.75+0.801.02^{+0.80}_{-0.75}0.362−0.016+0.0130.362^{+0.013}_{-0.016}0.38−0.95+1.080.38^{+1.08}_{-0.95}——
SPLIGM+CGMIncluded10.361−0.002+0.00210.361^{+0.002}_{-0.002}−20.05−0.07+0.07-20.05^{+0.07}_{-0.07}−2.01−0.10+0.11-2.01^{+0.11}_{-0.10}——0.46−0.97+0.960.46^{+0.96}_{-0.97}22.56−0.42+0.2822.56^{+0.28}_{-0.42}17.717.7

Table 1. Estimated properties of CW-LRD-z10 from the spectral fits.

LineFline/10−20F_{\mathrm{line}}/10^{-20}S/NlineS/N_{\mathrm{line}}EWrestEW_{\mathrm{rest}}
Lyα\alpha243−51+40243^{+40}_{-51}4.24.234.1−7.3+6.634.1^{+6.6}_{-7.3}
C ii λ1335\lambda133573.5−29.2+30.673.5^{+30.6}_{-29.2}2.52.512.4−5.1+5.212.4^{+5.2}_{-5.1}
N iv] λλ1483\lambda\lambda1483, 148746.9−25.2+26.846.9^{+26.8}_{-25.2}1.91.910.0−5.4+6.010.0^{+6.0}_{-5.4}
C iv λλ1549\lambda\lambda1549, 155194.3−24.4+23.794.3^{+23.7}_{-24.4}3.93.922.2−5.9+6.122.2^{+6.1}_{-5.9}
Iron complex121.4−27.2+27.5121.4^{+27.5}_{-27.2}4.44.479.2−18.7+18.879.2^{+18.8}_{-18.7}
[Ne iii] λ3869\lambda386914.8−6.3+6.914.8^{+6.9}_{-6.3}2.22.220.7−9.1+9.720.7^{+9.7}_{-9.1}
[Ne iii] λ3967\lambda3967 + Hϵ\epsilon14.9−6.1+6.314.9^{+6.3}_{-6.1}2.42.420.0−8.5+8.920.0^{+8.9}_{-8.5}
Hδ\delta16.6−6.8+6.216.6^{+6.2}_{-6.8}2.62.621.6−9.0+8.621.6^{+8.6}_{-9.0}
Hγ\gamma17.8−7.1+7.417.8^{+7.4}_{-7.1}2.52.521.4−8.6+10.221.4^{+10.2}_{-8.6}
[O iii] λ4363\lambda436326.5−7.8+7.026.5^{+7.0}_{-7.8}3.63.631.8−9.8+9.731.8^{+9.7}_{-9.8}

Table 2. Estimated line fluxes and equivalent widths of CW-LRD-z10 from the fitting with the BPL model including IGM+CGM absorption and emission lines.

Results

Spectroscopic redshift

The derived redshift is zspec=10.361−0.002+0.003z_{\mathrm{spec}} = 10.361^{+0.003}_{-0.002} from both the IGM+lines and the IGM+CGM+lines models, consistent with each other and with the previous photometric redshift, zphot=10.5−0.6+0.7z_{\mathrm{phot}} = 10.5^{+0.7}_{-0.6} (T. S. Tanaka et al. [102]). These tight constraints are largely driven by the tentative line features discussed in § 4.3.

Independently of these lines, the spectrum also exhibits a clear break feature around λobs∼1.4 μm\lambda_{\mathrm{obs}} \sim1.4\,\mu\mathrm{m}, which corresponds to the Lyα\alpha break. CW-LRD-z10 is undetected blueward of the break, while the continuum is detected at a signal-to-noise ratio of S/N∼4S/N \sim4–55 redward of the break. To assess the redshift constraint primarily from the continuum break, we repeat the fit after masking the wavelength regions of the tentative line detections. Note that Lyα\alpha is left unmasked and remains included in the model, because masking the Lyα\alpha region would also remove the Lyα\alpha break. As shown in Table 1, the derived redshifts from the IGM and the IGM+CGM models, 10.32−0.09+0.1610.32^{+0.16}_{-0.09} and zspec=10.41−0.06+0.06z_{\mathrm{spec}} = 10.41^{+0.06}_{-0.06}, respectively, are consistent with zspecz_{\mathrm{spec}} obtained including the lines.

Note that this break cannot be interpreted as a Balmer break at z∼3z \sim3. When combined with the non-detections in filters blueward of 1.4 μm1.4\,\mu\mathrm{m}, the implied break strength, defined as the fνf_\nu ratio across the break, is ≳10\gtrsim10. Such an extreme Balmer break is far stronger than expected for a conventional post-starburst galaxy, for which the Balmer break strength is typically ≲3\lesssim3. We also confirm this by fitting the photometry with z=3z = 3–44 galaxy models using Prospector (B. D. Johnson et al. [50]). The best-fit model SED predicts a continuum flux density of ∼0.02 μJy\sim0.02\,\mu\mathrm{Jy} at λobs<1.5 μm\lambda_{\mathrm{obs}} < 1.5\,\mu\mathrm{m}, which is inconsistent with both the observed spectrum and the photometric upper limits, e.g., F115W <3.3< 3.3 nJy (2σ2\sigma upper limit by T. S. Tanaka et al. [102]). Although similarly extreme Balmer breaks have been reported in some LRDs (R. P. Naidu et al. [80]; A. de Graaff et al. [17]; A. J. Taylor et al. [105]), an LRD-like solution at z∼3z \sim3 is also disfavored by the absence of strong rest-optical emission lines, such as Hα\alpha. Furthermore, the continuum slope at λobs>1.5 μm\lambda_{\mathrm{obs}} > 1.5\,\mu\mathrm{m}, β=−2.01−0.10+0.11\beta= -2.01^{+0.11}_{-0.10} from the SPL fit, is substantially bluer than the rest-optical slopes expected for LRDs, for example βopt>0\beta_{\mathrm{opt}} > 0 (D. D. Kocevski et al. [56]) or βopt>−0.52\beta_{\mathrm{opt}} > -0.52 (D. D. Kocevski et al. [57]).

Therefore, we conclude that CW-LRD-z10 is at zspec=10.36z_{\mathrm{spec}} = 10.36, which establishes CW-LRD-z10 as the first zspec>10z_{\mathrm{spec}} > 10 LRD candidate. The obtained UV absolute magnitude of MUV=−20.16−0.07+0.07M_{\mathrm{UV}} = -20.16^{+0.07}_{-0.07} places it among the most luminous LRDs (Figure 3). It is substantially brighter than the only previously published LRD at zspec>9z_{\mathrm{spec}} > 9, CAPERS-LRD-z9, at z=9.288z = 9.288 and MUV=−18.2±0.2M_{\mathrm{UV}} = -18.2 \pm0.2 (A. J. Taylor et al. [105]).

Figure 3. UV absolute magnitudes (MUVM_{\mathrm{UV}}) of previously spectroscopically confirmed LRDs (J. Matthee et al. [76]; D. D. Kocevski et al. [56]; X. Lin et al. [68]; A. de Graaff et al. 2026) as a function of redshift. Note that there can be an overlap between different studies. The red star indicates CW-LRD-z10 (this study), representing the first LRD candidate with zspec>10z_{\mathrm{spec}} > 10, corresponding to a cosmic age of <0.5< 0.5 Gyr.

Spectroscopic confirmation of a V-shaped SED

The fit with the BPL continuum with IGM+CGM absorption and emission lines results in βUV=−2.26−0.11+0.11\beta_{\mathrm{UV}} = -2.26^{+0.11}_{-0.11} and βopt=1.17−0.70+0.74\beta_{\mathrm{opt}} = 1.17^{+0.74}_{-0.70}. Both of the fitted spectral slopes, as well as their large difference across the turnover, are consistent with those typically observed in lower-zz LRDs (e.g., D. D. Kocevski et al. [56]; A. de Graaff et al. 2026; G. Barro et al. [7]; M. Ando et al. [4]). Furthermore, the BPL model is strongly favored over the SPL model with β=−2.01−0.10+0.11\beta= -2.01^{+0.11}_{-0.10} by the BIC improvement of ΔBIC=17.7>10\Delta\mathrm{BIC} = 17.7 > 10, a conventional threshold for strong evidence in model selection, confirming a significant change in the continuum slope. This continuum-slope change is more clearly visible in the 4-channel binned spectrum (Figure 1, lower). Furthermore, the fitted turnover wavelength is λturn=0.356−0.013+0.010 μm\lambda_{\mathrm{turn}} = 0.356^{+0.010}_{-0.013}\,\mu\mathrm{m}, which is consistent with the Balmer limit (λrest=0.3646 μm\lambda_{\mathrm{rest}} = 0.3646\,\mu\mathrm{m}) within the uncertainties, as commonly seen in LRDs (e.g., D. J. Setton et al. [95]; A. de Graaff et al. 2026).

The UV-to-optical continuum shape and characteristic V-shape turnover in CW-LRD-z10 are consistent with typical LRDs at lower redshift. In the bottom panel of Figure 1, we overplot a stack of PRISM spectra of 274 LRDs from z∼2.3z \sim2.3–9.3, combining the EMBER survey with PRISM spectra from the ZENITH program (GO#7417, PIs C. Casey, H. B. Akins, and M. Franco) and archival observations from CAPERS (GO#6368, A. J. Taylor et al. 2025), RUBIES (GO#4233, A. de Graaff et al. 2025b), and other archival programs. The details of the LRD selection and stacking procedure will be provided in H. B. Akins et al. (in prep). The stacked spectrum shows a clear turnover at the Balmer limit, similar to CW-LRD-z10, and continues to rise toward redder wavelengths, consistent with the observed F770W photometry of CW-LRD-z10.

Moreover, CW-LRD-z10’s Balmer-break strength, defined as the ratio of the continuum flux densities across the Balmer limit, is

⟨fν⟩4000–4100 A˚⟨fν⟩3620–3720 A˚=1.22−0.35+0.45.(1)\frac{\langle f_{\nu}\rangle_{4000\text{--}4100\,\mathrm{\mathring{A}}}}{\langle f_{\nu}\rangle_{3620\text{--}3720\,\mathrm{\mathring{A}}}} = 1.22^{+0.45}_{-0.35}. \tag*{(1)}

This value is comparable to those commonly observed in LRDs at z∼4z \sim4–7 (A. de Graaff et al. 2026) and the value from the stacked spectrum (∼1.6\sim1.6). However, the measured ratio is consistent with unity within a 1σ1\sigma uncertainty and does not provide significant evidence for a Balmer break. Nevertheless, spectral fitting over a broader wavelength range strongly favors a broken power-law continuum, with different UV and optical slopes joined near the Balmer limit. We therefore conclude that the characteristic V-shaped continuum is still supported by the spectrum.

Detection of rest-UV and optical lines

Based on the spectroscopic fitting, we detect multiple emission-line features at the S/N∼2S/N \sim2–4 level. Other than Lyα\alpha (§ 4.6), C IV λλ1549\lambda\lambda1549, 1551 and [O III] λ4363\lambda4363 show the highest values with S/N=3.9S/N = 3.9 and 3.6, respectively.

To test these tentative line detections, we perform spectral stacking of these lines. After subtracting the fitted BPL continuum, we stack the residual spectrum around five rest-optical lines, [Ne III] λ3869\lambda3869, [Ne III] λ3967\lambda3967 + Hϵ\epsilon, Hδ\delta, Hγ\gamma, and [O III] λ4363\lambda4363 (Figure 4 left), in the velocity plane within v∈[−5000,5000]v \in[-5000, 5000] km s−1^{-1}. In the stacking process, other lines within the extracted velocity range are masked within a ±1000\pm1000 km s−1^{-1} window, and we construct both an inverse-variance-weighted mean stack and a median stack. We repeat this stacking over a redshift range of z=8z = 8–11 and fit the stacked spectrum with a Gaussian centered at v=0v = 0 to evaluate its S/NS/N. Note that we use the stacked S/NS/N as a consistency check between each line rather than as an independent global detection significance. We do not include the rest-UV lines shown in Figure 2 left (Lyα\alpha, C III] λ1335\lambda1335, N IV] λλ1483\lambda\lambda1483, 1487, and C IV λλ1549\lambda\lambda1549, 1551) in the stacking analysis, because they have a much lower spectral resolution than optical lines.

Figure 4. (Left) Continuum-subtracted stacked spectrum of [Ne III] λ3869\lambda3869, [Ne III] λ3967\lambda3967 + Hϵ\epsilon, Hδ\delta, Hγ\gamma, and [O III] λ4363\lambda4363 assuming z=10.36z = 10.36. The red solid and black dashed lines show mean- and median-stacked spectra, respectively. 1σ1\sigma uncertainties are shown as the shaded regions. (Right) S/NS/N from the Gaussian fitting when redshift is varied. The gray thin lines indicate the results from the leave-one-out test. The stack S/NS/N peaks at z∼10.36z \sim10.36, providing additional support for zspec=10.36z_{\mathrm{spec}} = 10.36.

The stacked signals from the mean and median stacking methods reach their maximum significance at z=10.36z = 10.36, with S/N=6.9S/N = 6.9 and 5.5 for mean and median, respectively (Figure 4). As a robustness check, we perform a leave-one-out test by repeating the stacking analysis five times, excluding one of the five lines in each realization. The stacked excess with S/N=6.2S/N = 6.2–6.4 at z=10.36z = 10.36 remains present in all cases, indicating that the stacking result is not driven by any single feature. Therefore, we conclude that these tentative detections are unlikely to be coincidental and support the fitted spectroscopic redshift of zspec=10.36z_{\mathrm{spec}} = 10.36.

Line diagnostics

Among the line-flux ratios accessible from the detected lines, O3Hg≡log⁡([O III] λ4363/Hγ)\mathrm{O3Hg} \equiv\log([\mathrm{O\,III}]\,\lambda4363/\mathrm{H}\gamma) provides an important diagnostic of the nature of the ionizing source. [O III] λ4363[\mathrm{O\,III}]\,\lambda4363 is sensitive to the electron temperature and the O3Hg ratio can be enhanced by the hard ionizing radiation field produced by an AGN (e.g., J. Gutkin et al. [32]; A. Feltre et al. [22]; G. Mazzolari et al. [77]). Some studies (e.g., J. Gutkin et al. [32]; A. Feltre et al. [22]; G. Mazzolari et al. [77]; C. L. Martin et al. [75]) have also used line diagnostics involving O3Hg in combination with line ratios such as O33\mathrm{O33} ([O III] λ5007/[O III] λ4363[\mathrm{O\,III}]\,\lambda5007/[\mathrm{O\,III}]\,\lambda4363), O32\mathrm{O32} ([O III] λ5007/[O II] λλ3727,3729[\mathrm{O\,III}]\,\lambda5007/[\mathrm{O\,II}]\,\lambda\lambda3727,3729), and Ne3O2\mathrm{Ne3O2} ([Ne III] λ3869/[O II] λ3727[\mathrm{Ne\,III}]\,\lambda3869/[\mathrm{O\,II}]\,\lambda3727) for discussing the nature of the ionizing source. However, the limited depth and wavelength coverage of the available data result in a marginal detection of [Ne III] λ3869[\mathrm{Ne\,III}]\,\lambda3869 (S/N=2.2S/N = 2.2), a non-detection of [O II] λλ3727,3729[\mathrm{O\,II}]\,\lambda\lambda3727,3729, and no coverage of [O III] λ5007[\mathrm{O\,III}]\,\lambda5007.

The O3Hg ratio derived from the posterior samples is O3Hg=0.17−0.22+0.24\mathrm{O3Hg} = 0.17^{+0.24}_{-0.22}. This value is higher than those measured from stacked spectra of LRDs, LBDs (S. Geris et al. [27]), and broad-line AGNs (I. Juodžbalis et al. [52]) observed with JWST, and is relatively high even among measurements for individual LRDs (P. Rinaldi et al. [91]; G. C. Jones et al. [51]; M. Brazzini et al. [8]; X. Ji et al. 2026a). It is consistent with the values expected from AGN-only models over a wide range of O32≲2\mathrm{O32} \lesssim2 and Ne3O2≲1\mathrm{Ne3O2} \lesssim1 (Figure 5). We also estimate a 3σ3\sigma upper limit on the [O II] λλ3727,3729[\mathrm{O\,II}]\,\lambda\lambda3727,3729 flux using mock spectra and obtain F[O II]<1.73×10−19 erg s−1 cm−2F_{[\mathrm{O\,II}]} < 1.73 \times10^{-19}\ \mathrm{erg\ s^{-1}\ cm^{-2}}. Combining this upper limit with the 16th percentile of the [Ne III] λ3869[\mathrm{Ne\,III}]\,\lambda3869 flux posterior, we derive a conservative lower limit of Ne3O2>−0.31\mathrm{Ne3O2} > -0.31. This constraint is consistent with the AGN-only region of the diagnostic diagram over a wide range of plausible Ne3O2 values (Figure 5). However, neither of these line diagnostics rule out ionization by star formation.

Figure 5. (left) O3Hg–O32 and (right) O3Hg–Ne3O2 line diagnostic diagrams. The black dashed line shows the empirical threshold proposed by G. Mazzolari et al. [77], above which sources can be uniquely classified as AGNs (“AGN only”), while both AGNs and star-forming galaxies can occupy the region below (“AGN or SF”). In the left panel, the red solid and blue dot-dashed lines show the median O3Hg and the O3Hg after blending correction assuming the Case-B Balmer-line ratios between Hγ and Hδ. Note that no constraint on O32 is available because [O iii] λ5007 falls outside the observed wavelength range. In the right panel, the stars and arrows indicate the conservative lower limit of Ne3O2, derived from the 3σ upper limit on the [O ii] λλ3727, 3729 flux and the 16th percentile of the [Ne iii] λ3869 flux posterior. 1σ confidence levels are shown as the shaded regions. For comparison, we show the stacked spectra of JADES broad-line AGNs [52] and LRDs and LBDs [27] as gray filled markers, and individual sources from the literature as light-gray open markers: UNCOVER_20466 [51], GN-28074 and GS-3073 [8], the Lord of LRDs (X. Ji et al. 2026a), and the LRD sample from P. Rinaldi et al. [91]. Although the inferred line ratios are consistent with the AGN-only model and those observed in LRDs and AGNs, ionization by star formation cannot be ruled out given the uncertainties and possible line blending.

The observed Hγ/Hδ\mathrm{H}\gamma/\mathrm{H}\delta flux ratio Hγ/Hδ=1.07−0.47+0.95\mathrm{H}\gamma/\mathrm{H}\delta= 1.07^{+0.95}_{-0.47} is lower than the value expected for Case B recombination, which is ≃1.8\simeq1.8 for Te=1×104 KT_{\mathrm{e}} = 1 \times10^{4}\ \mathrm{K} and ne=1×102 cm−3n_{\mathrm{e}} = 1 \times10^{2}\ \mathrm{cm^{-3}}. This discrepancy cannot be explained by dust attenuation, since dust would preferentially suppress Hδ\mathrm{H}\delta relative to Hγ\mathrm{H}\gamma and therefore increase the Hγ/Hδ\mathrm{H}\gamma/\mathrm{H}\delta ratio. Previous studies have discussed that Balmer-line ratios in the dense gas associated with LRDs may deviate from the Case B recombination predictions owing to high-density radiative-transfer effects (e.g., G. P. Nikopoulos et al. [83]; S.-J. Chang et al. [14]). According to Table 2 of G. P. Nikopoulos et al. [83], JADES-GN-1181_73488 and CEERS-EGS-1345_1244 show Hγ/Hδ\mathrm{H}\gamma/\mathrm{H}\delta flux ratios of ≃1.4\simeq1.4 and ≃1.1\simeq1.1, respectively, comparable to the value inferred from our fit (1.07−0.47+0.951.07^{+0.95}_{-0.47}). Moreover, as mentioned in § 3, the wavelength separation between Hγ\mathrm{H}\gamma and [O III] λ4363[\mathrm{O\,III}]\,\lambda4363 is in principle resolvable at the PRISM spectral resolution in their observed wavelength range.

Nevertheless, given the limited S/NS/N, we cannot completely exclude the possibility that imperfect deblending of Hγ\mathrm{H}\gamma and [O III] λ4363[\mathrm{O\,III}]\,\lambda4363 substantially affects the inferred line fluxes. We therefore perform an additional test, assuming the Case B flux ratio between Hγ\mathrm{H}\gamma and Hδ\mathrm{H}\delta. If part of the flux attributed to [O III] λ4363[\mathrm{O\,III}]\,\lambda4363 in the fiducial fit is instead assigned to Hγ\mathrm{H}\gamma to satisfy Hγ/Hδ=1.8\mathrm{H}\gamma/\mathrm{H}\delta= 1.8, the inferred O3Hg ratio decreases to −0.21−0.53+0.43-0.21^{+0.43}_{-0.53}. This value could also be consistent with non-AGN ionization, depending on the O32 and Ne3O2 ratios, but is comparable to those observed in some LRDs and AGNs (B. E. Backhaus et al. [5]; G. C. Jones et al. [51]; M. Brazzini et al. [8]). Therefore, the limited wavelength coverage and depth of the available data prevent us from drawing a definitive conclusion about the nature of the ionizing source.

Possible iron feature

We find that there is also an emission-like feature at λobs∼2.7\lambda_{\mathrm{obs}} \sim2.7–2.9 μm2.9\ \mu\mathrm{m}, corresponding to λrest∼0.24\lambda_{\mathrm{rest}} \sim0.24–0.25 μm0.25\ \mu\mathrm{m} at z=10.36z = 10.36. Fitting this feature with a single Gaussian (§ 3) gives Fline=121.4−21.2+27.5×10−20 erg s−1 cm−2F_{\mathrm{line}} = 121.4^{+27.5}_{-21.2} \times10^{-20}\ \mathrm{erg\ s^{-1}\ cm^{-2}} (EWrest=79.2−18.7+18.8 A˚\mathrm{EW}_{\mathrm{rest}} = 79.2^{+18.8}_{-18.7}\ \text{\AA}, S/N=4.4S/N = 4.4) with a large FWHM of 1.22−0.32+0.58×104 km s−11.22^{+0.58}_{-0.32} \times10^{4}\ \mathrm{km\ s^{-1}}. This feature may correspond to the Fe II UV bump or the Fe III UV47 multiplet, with no other prominent emission lines expected at this wavelength. Such iron emission lines are typically observed in quasars and are thought to originate from the broad-line region (BLR, e.g., B. J. Wills et al. [115]; M. Vestergaard & B. J. Wilkes [112]; T. A. A. Sigut et al. [97]). The large fitted FWHM is also consistent with an iron-line blend composed of multiple transitions.

Similar iron emission features have been reported in several LRDs (e.g., I. Labbe et al. [62]; X. Ji et al. 2026a; R. Tripodi et al. [109]), including detections from stacked spectra (e.g., P. G. Pérez-González et al. [87]; W. Q. Sun et al. [101]; M. Ando et al. [4]). In contrast, such features have not been commonly reported in normal star-forming galaxies at high redshift. If the observed feature is indeed due to iron emission, it would further support an AGN-powered LRD interpretation for CW-LRD-z10 and may suggest partial UV leakage from the central engine through a clumpy or porous envelope

[4]. Strong UV iron emission may also provide information on the physical conditions of the nuclear gas, since its strength depends sensitively on gas density, ionization parameter, column density, and radiative-transfer effects [e.g. 109, 107, 4]. If confirmed as iron emission, this feature would suggest substantial chemical enrichment within the first ∼ 450 Myr of cosmic history. Both core-collapse supernovae and prompt Type Ia supernovae with short delay times of ∼ 30–100 Myr could have contributed to such early enrichment [e.g. 73, 74, 81, 82].

Possible Lyα line

As shown in Figure 2, the IGM+lines model overpredicts the continuum around λobs=1.4\lambda_{\mathrm{obs}} = 1.4–1.5 μm1.5\,\mu\mathrm{m}, and this remains the case even when fixing xHI=1.0x_{\mathrm{HI}} = 1.0, which maximizes the neutral IGM absorption. When using the IGM+lines model, Lyα emission is not detected (S/N=0.5S/N = 0.5). In contrast, the IGM+CGM+lines model can reproduce the flux decrement using a strong Lyα damping-wing absorption with log⁡(NHI/cm−2)=22.81−0.25+0.20\log(N_{\mathrm{HI}}/\mathrm{cm}^{-2}) = 22.81^{+0.20}_{-0.25} as seen in M. Ando et al. [4]. In this IGM+CGM+lines model, the excess flux around λobs∼1.4 μm\lambda_{\mathrm{obs}} \sim1.4\,\mu\mathrm{m} can be explained by a Lyα emission line with S/N=4.2S/N = 4.2. Lyα emission has also been reported in a number of LRDs in previous studies (e.g., Z. Ji et al. 2026; Y. Kageura et al. 2026; M. Ando et al. 2026).

A similar spectral feature of a Lyα line with a damping wing that cannot be explained by IGM-only models was reported even in a galaxy at z=13.0z = 13.0 [117]. At such a high redshift, Lyα photons are expected to be strongly attenuated by the substantially neutral IGM; therefore, the detection of a Lyα line requires conditions that enhance Lyα transmission [e.g. 21, 84]. One possibility is that the source resides in a large ionized bubble, within which Lyα photons can redshift away from resonance before encountering the neutral IGM [e.g. 103, 111]. Alternatively, radiative-transfer effects in the ISM and CGM, such as resonant scattering and outflows, can shift Lyα\alpha photons redward of the systemic velocity and thereby enhance their transmission even through a smaller ionized region. Thus, if the Lyα\alpha feature is real, it may indicate a locally ionized environment and/or favorable Lyα\alpha radiative-transfer conditions around CW-LRD-z10. However, the significance of the Lyα\alpha detection is model-dependent. Furthermore, the BIC improvement of the IGM+CGM+lines model relative to the IGM+lines model is only ΔBIC=2.6\Delta\mathrm{BIC} = 2.6, much less significant than the conventional ΔBIC≳10\Delta\mathrm{BIC} \gtrsim10 threshold for strong evidence. This indicates that the statistical preference for the CGM-included model is weak. We also note that the downturn around λobs∼1.4\lambda_{\mathrm{obs}} \sim1.4-1.5 μm1.5\,\mu\mathrm{m}, reproduced by the IGM+CGM model in our fitting, may alternatively be consistent with nebular continuum emission (M. Ando et al. [4]). For these reasons, we do not conclude that the CGM-included model is preferred. Instead, we regard the Lyα\alpha detection as model-dependent and refrain from interpreting in detail the fitted parameters sensitive to the Lyα\alpha modeling, such as RbR_b and NHIN_{\mathrm{HI}}.

Discussion: Is it truly an LRD?

In § 4.1, we established that the source lies at z=10.36z = 10.36 rather than being a low-redshift interloper. In this section, we discuss whether the observed SED at z=10.36z = 10.36 necessarily requires an LRD interpretation, or whether it could instead arise from a non-LRD component.

There is currently no community-wide consensus on a definitive definition of LRDs. Nevertheless, three key characteristics are commonly used to identify them: a compact morphology, a V-shaped continuum with a minimum around or redward of the Balmer limit, and broad Balmer emission lines (e.g., R. E. Hviding et al. [37]; A. de Graaff et al. 2026). For CW-LRD-z10, T. S. Tanaka et al. [102] confirmed its compact morphology, and, in this work, we spectroscopically confirm its redshift (zspec=10.36z_{\mathrm{spec}} = 10.36, § 4.1) and the presence of the characteristic V-shaped continuum, with its turnover at the Balmer limit (§ 4.2). However, broad Balmer lines have not yet been spectroscopically confirmed because Hβ\beta and Hα\alpha fall outside the wavelength coverage of the current NIRSpec/PRISM spectrum (Figure 1). Although the possibly high [O iii] λ4363\lambda4363/Hγ\gamma ratio (§ 4.4) and the possible iron emission (§ 4.5) support its LRD or AGN nature, an unambiguous spectroscopic confirmation has not yet been achieved.

We therefore consider whether the observed SED at z=10.36z = 10.36 could instead be explained by a stellar-origin Balmer break. Such a model could in principle reproduce the turnover, and the iron complex and the high O3Hg ratio could arise from an AGN component. However, this alternative solution is strongly disfavored by the extremely red NIRCam-MIRI color of CW-LRD-z10, mF444W−mF770W=1.82m_{\mathrm{F444W}} - m_{\mathrm{F770W}} = 1.82, based on its robust S/N=18S/N = 18 detection in F770W (T. S. Tanaka et al. [102]). If the rest-frame optical continuum was approximately flat in fνf_\nu from λrest∼0.4\lambda_{\mathrm{rest}} \sim0.4-0.9 μm0.9\,\mu\mathrm{m}, as expected for a typical Balmer-break galaxy, reproducing the F770W flux would require an Hα\alpha rest-frame equivalent width (EW) of ∼(6±1)×103\sim(6 \pm1) \times10^3 Å. Such an extreme EW is substantially larger than those expected theoretically (≲4000\lesssim4000 Å; A. K. Inoue [44]) or typically observed in extreme emission line galaxies (∼2000\sim2000 Å; e.g., S. Withers et al. [116]). The F770W detection therefore provides important support for the LRD nature of CW-LRD-z10. We further confirm that CW-LRD-z10 is independently detected in each of the four individual COSMOS-Web F770W exposures prior to mosaicing, indicating that the F770W signal is unlikely to be caused by an imaging artifact.

Another possible explanation for the red rest-optical continuum is dust-reddened AGN emission (e.g., D. D. Kocsevki et al. 2023; J. Matthee et al. [76]; J. E. Greene et al. [29]; P. Madau & R. Maiolino [70]; P. Madau et al. [71]). However, the observed turnover near the Balmer limit is characteristic of previously observed LRDs (D. J. Setton et al. [95]; A. de Graaff et al. 2026), whereas such a coincidence is not necessarily expected from simple dust reddening. Furthermore, the tentative C iv and UV iron emission features, commonly associated with AGN activity, suggest that at least part of the nuclear UV emission is visible without severe obscuration. This may be in tension with a uniformly dust-obscured AGN, although more complex geometries involving partial covering or scattering, as well as non-standard dust attenuation curves (Z. Li et al. 2025), remain possible. Moreover, despite the ALMA non-detection (T. S. Tanaka et al. [102]), obscuration by dust hotter than that effectively probed by the available ALMA observations, or by a compact dust distribution with a small global covering factor, cannot be excluded (C. M. Casey et al. [13]; P. Madau & R. Maiolino [70]; I. Delvecchio et al. [20]). Similarly, the X-ray non-detection (T. S. Tanaka et al. [102]) does not rule out the possibility of a Compton-thick AGN. Therefore, while the observed properties are consistent with an LRD-like interpretation, we cannot rule out dust-obscured AGN scenarios with the current data. On the other hand, previous studies have not ruled out the possibility that dust attenuation also contributes to the red rest-optical continua of LRDs themselves (e.g., P. Madau & R. Maiolino [70]). Thus, dust-reddened AGN and other proposed LRD scenarios are not necessarily mutually exclusive, and it remains unclear whether they can be observationally distinguished with the currently available data. Further observations at longer wavelengths are needed to distinguish between these scenarios and better constrain the physical origin of the red rest-optical continuum.

Conclusion

CW-LRD-z10 is a robust z>10z > 10 LRD candidate with zphot=10.5−0.6+0.7z_{\mathrm{phot}} = 10.5^{+0.7}_{-0.6} selected from the COSMOS-Web dataset using a joint NIRCam-MIRI color selection [102]. We present the NIRSpec/PRISM spectrum of CW-LRD-z10 (Figure 1) taken in the EMBER program. The key results are as follows:

  • Based on the clearly detected Lyα\alpha break and emission lines with S/N∼2S/N \sim2–44 (Figure 2), we spectroscopically measure the redshift as z=10.361−0.002+0.003z = 10.361^{+0.003}_{-0.002}, consistent with the photometric redshift (§ 4.1). The corresponding UV absolute magnitude is MUV=−20.16−0.07+0.07M_{\mathrm{UV}} = -20.16^{+0.07}_{-0.07}, comparable to those of the most UV-luminous LRDs ever found (Figure 3). The zspecz_{\mathrm{spec}} confirmation removes the principal redshift uncertainty underlying the number density estimates in T. S. Tanaka et al. [102].

  • We spectroscopically confirm a V-shaped SED with UV and optical continuum slopes of βUV=−2.26−0.11+0.11\beta_{\mathrm{UV}} = -2.26^{+0.11}_{-0.11} and βopt=1.17−0.70+0.74\beta_{\mathrm{opt}} = 1.17^{+0.74}_{-0.70}, respectively (§ 4.2). The turnover wavelength, λturn=0.356−0.013+0.010 μm\lambda_{\mathrm{turn}} = 0.356^{+0.010}_{-0.013}\,\mu\mathrm{m}, is consistent with the Balmer limit. Both of these findings satisfy key spectroscopic characteristics commonly associated with LRDs.

  • Line detections are tentative with S/N=2S/N = 2–44, and C iv λλ1549,1551\lambda\lambda1549, 1551 and [O iii] λ4363\lambda4363 have the highest significances of S/N=3.9S/N = 3.9 and 3.63.6, respectively. However, the stacking analysis provides additional support for the line interpretation (Figure 4, § 4.3).

  • The line fitting suggests a possibly high [O iii] λ4363\lambda4363/Hγ\gamma flux ratio of O3Hg =log⁡([O III] λ4363/Hγ)=0.17−0.22+0.24= \log([\mathrm{O\,III}]\,\lambda4363/\mathrm{H}\gamma) = 0.17^{+0.24}_{-0.22}, or −0.21−0.53+0.43-0.21^{+0.43}_{-0.53} after accounting for possible deblending between [O iii] λ4363\lambda4363 and Hγ\gamma by imposing the Case B Balmer-line ratio between Hγ\gamma and Hδ\delta (Figure 5, § 4.4). These high line ratios are consistent with those previously observed in LRDs and AGNs, but do not rule out ionization by star formation.

  • We detect a possible UV iron emission feature at λrest∼0.24\lambda_{\mathrm{rest}} \sim0.24–0.25 μm0.25\,\mu\mathrm{m} (§ 4.5). If confirmed as iron emission, this feature would be consistent with those observed in previously known AGNs and LRDs.

  • We also find a possible, absorption-model-dependent detection of Lyα\alpha emission (S/N=4.2S/N = 4.2 with the IGM+CGM model, § 4.6) with a strong absorption feature that cannot be reproduced by IGM absorption alone. This may suggest enhanced Lyα\alpha transmission around CW-LRD-z10 even at zspec=10.36z_{\mathrm{spec}} = 10.36.

  • Due to the lack of spectroscopic confirmation of broad Balmer lines, the LRD nature of CW-LRD-z10 has not yet been established definitively. Nevertheless, when combined with the previously reported large mF444W−mF770Wm_{\mathrm{F444W}} - m_{\mathrm{F770W}} color excess and the spectroscopic features presented here, the LRD interpretation provides the most plausible explanation for the observed properties of CW-LRD-z10 (§ 5). We therefore regard CW-LRD-z10 as the first z>10z > 10 LRD candidate with a spectroscopically confirmed redshift.

The planned MIRI/LRS spectroscopy in Cycle 5 (GO#9483, PI T. S. Tanaka) will provide a deep mid-infrared spectrum covering Hα\alpha, Hβ\beta, [O iii] λ5007\lambda5007 and the rest-optical continuum, which will enable further tests and a more detailed characterization of its nature. Complementing the MIRI spectroscopy, deeper NIRSpec observations at higher spectral resolution will enable more detailed spectroscopic characterization through more robust line detections, improved Lyα\alpha and IGM+CGM modeling, and more reliable photoionization modeling.

To constrain the evolution of the LRD population, we need a larger z>10z > 10 LRD sample, which requires more extensive MIRI observations. Beyond JWST-only searches, future MIRI follow-up of bright z>10z > 10 candidates identified by the recently launched Roman Space Telescope will be effective for searching for more luminous z>10z > 10 LRDs over much wider areas. On a longer timescale, the proposed GREX-PLUS mission will further extend such searches through its wide-field near- to mid-infrared imaging capabilities [30, 45, 31].

Acknowledgements

This work is based on observations made with the NASA/ESA/CSA 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 NAS5-03127 for JWST. These observations are associated with program GO-7076. Support for US investigators in program GO-7076 was provided by NASA through a grant from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-03127. The observation can be accessed via doi: 10.17909/7mg3-kh29. Numerical computations were in part carried out on the iDark cluster, Kavli IPMU. This work was made possible by utilizing the CANDIDE cluster at the Institut d’Astrophysique de Paris. The CANDIDE cluster was funded through grants from the PNCG, CNES, DIM-ACAV, the Euclid Consortium, and the Danish National Research Foundation Cosmic Dawn Center (DNRF140) and is maintained by Stephane Rouberol. The COSMOS team is partly supported by the Centre National d’Etudes Spatiales (CNES). Kavli IPMU is supported by World Premier International Research Center Initiative (WPI), MEXT, Japan. The Cosmic Dawn Center is funded by the Danish National Research Foundation (DNRF140). We thank Yoshihisa Asada, Ignas Juodžbalis, and Lucy R. Ivey for fruitful discussion. We thank the anonymous referee for helpful feedback. T. S. Tanaka is supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number JP25KJ0750 and the Forefront Physics and Mathematics Program to Drive Transformation (FoPM), a World-leading Innovative Graduate Study (WINGS) Program at the University of Tokyo. T. S. Tanaka gratefully acknowledges UNISON SQUARE GARDEN, whose full-color music motivated him throughout this research and inspired him to keep surprising the world. M. Aravena is supported by FONDECYT grant number 1252054, and gratefully acknowledges support from ANID Basal Project FB210003, ANID MILENIO NCN2024_112 and ANID + Vinculación Internacional + FOVI250261. B. Trakhtenbrot acknowledges support from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement No. 950533). K. Inayoshi acknowledges support from the National Natural Science Foundation of China (12573015, W2532003), the Beijing Natural Science Foundation (IS25003), and the China Manned Space Program (CMS-CSST-2025-A09). J.-T. Schindler acknowledges funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) - Project number 518006966 - and support by the DFG under Germany’s Excellence Strategy – EXC 2121 „Quantum Universe“ – 39083306. A. K. Inoue is supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number JP26H02069. O. Ilbert acknowledges the funding of the French Agence Nationale de la Recherche for the project iMAGE (grant ANR-22-CE31-0007). D. B. Sanders gratefully acknowledges support from NSF Grant 2407752. E. Treister acknowledges support from the ANID CATA-BASAL program FB210003, and FONDECYT Regular 1241005 and 1250821. OpenAI ChatGPT was used under author supervision to assist with language editing and refinement of the manuscript text. All scientific content, interpretations, references, and final wording were reviewed and approved by the authors, who take full responsibility for the content of the manuscript.

All Authors and Affiliations

Takumi S. Tanaka,1,2,3^{1,2,3} Hollis B. Akins,4,∗^{4,*} John D. Silverman,1,2,3,5^{1,2,3,5} Caitlin M. Casey,6,7^{6,7} Ezequiel Treister,8^{8} Manuel Aravena,9,10^{9,10} Yuichi Harikane,11^{11} Benny Trakhtenbrot,12^{12} Kohei Inayoshi,13^{13} Kazuhiro Shimasaku,2,14^{2,14} Maximilien Franco,15^{15} Seiji Fujimoto,16,17^{16,17} Mingyu Li,18,19,20^{18,19,20} Hiroya Umeda,21^{21} Jan-Torge Schindler,22^{22} Akio K. Inoue,23,24^{23,24} Jean-Baptiste Billand,15^{15} Darshan Kakkad,25^{25} Xuheng Ding,26^{26} Steven L. Finkelstein,27,28^{27,28} Michaela Hirschmann,29,30^{29,30} Z. Ghaffari,31,32^{31,32} Andrew J. Battisti,33,34^{33,34} Lauro Moscardini,35,36,37^{35,36,37} Nicole E. Drakos,38^{38} Rasha M. Samir,39^{39} Jeyhan S. Kartaltepe,40^{40} Qinyue Fei,16^{16} Ming-Yang Zhuang,41^{41} Olivier Ilbert,42^{42} Wuji Wang,43^{43} Rafael C. Arango-Toro,44^{44} Hiddo S. B. Algera,45^{45} Greta Toni,35,36,46^{35,36,46} Junya Arita,2^{2} Kei Ito,7,47^{7,47} Ghassem Gozaliasl,48,49^{48,49} Dale D. Kocevski,50^{50} Yunjing Wu,1,3^{1,3} Masafusa Onoue,51,1^{51,1} Marc Huertas-Company,52,53,54,55^{52,53,54,55} Yiyang Zhang,26^{26} Jason Rhodes,56^{56} Andreas L. Faisst,43^{43} Zihao Li,7,57^{7,57} Crystal L. Martin,6^{6} Makoto Ando,11^{11} Vasily Kokorev,58^{58} Sophie L. Newman,59^{59} Can Xu,60,61^{60,61} Suin Matsui,2^{2} Jialai Wang,62,63,1,3^{62,63,1,3} Carter Flayhart,40^{40} Hossein Hatamnia,64,65^{64,65} Ali Ahmad Khostovan,66,40^{66,40} Ronaldo Laishram,67^{67} Felix Martinez III,40^{40} David B. Sanders,68^{68} Sune Toft,7,57^{7,57} Francesco Valentino7,47^{7,47} And Jorge Zavala69^{69}

(Affiliations can be found after the references)

1{}^{1}Kavli Institute for the Physics and Mathematics of the Universe (WPI), The University of Tokyo Institutes for Advanced Study, The University of Tokyo, Kashiwa, Chiba 277-8583, Japan

2{}^{2}Department of Astronomy, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

3{}^{3}Center for Data-Driven Discovery, Kavli IPMU (WPI), UTIAS, The University of Tokyo, Kashiwa, Chiba 277-8583, Japan

4{}^{4}Department of Astrophysical Sciences, Princeton University, 4 Ivy Lane, Princeton, NJ 08544, USA

5{}^{5}Center for Astrophysical Sciences, Department of Physics & Astronomy, Johns Hopkins University, Baltimore, MD 21218, USA

6{}^{6}Department of Physics, University of California, Santa Barbara, CA 93106, USA

7{}^{7}Cosmic Dawn Center (DAWN), Denmark

8{}^{8}Instituto de Alta Investigación, Universidad de Tarapacá, Casilla 7D, Arica, Chile

9{}^{9}Instituto de Estudios Astrofísicos, Facultad de Ingeniería y Ciencias, Universidad Diego Portales, Av. Ejército 441, Santiago, Chile

10{}^{10}Millenium Nucleus for Galaxies (MINGAL)

11{}^{11}Institute for Cosmic Ray Research, The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8582, Japan

12{}^{12}School of Physics and Astronomy, Tel Aviv University, Tel Aviv 69978, Israel

13{}^{13}Kavli Institute for Astronomy and Astrophysics, Peking University, Beijing 100871, China

14{}^{14}Research Center for the Early Universe, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

15{}^{15}Université Paris-Saclay, Université Paris Cité, CEA, CNRS, AIM, 91191, Gif-sur-Yvette, France

16{}^{16}David A. Dunlap Department of Astronomy and Astrophysics, University of Toronto, 50 St. George Street, Toronto, Ontario, M5S 3H4, Canada

17{}^{17}Dunlap Institute for Astronomy and Astrophysics, 50 St. George Street, Toronto, Ontario, M5S 3H4, Canada

18{}^{18}Department of Astronomy, Tsinghua University, Beijing 100084, China

19{}^{19}Kavli Institute for Cosmology, University of Cambridge, Madingley Road, Cambridge CB3 0HA, UK

20{}^{20}Cavendish Laboratory, University of Cambridge, 19 JJ Thomson Avenue, Cambridge CB3 0HE, UK

21{}^{21}Center for Computational Sciences, University of Tsukuba, Ten-no-dai, 1-1-1 Tsukuba, Ibaraki 305-8577, Japan

22{}^{22}Hamburger Sternwarte, University of Hamburg, Gojenbergsweg 112, D-21029 Hamburg, Germany

23{}^{23}Department of Pure and Applied Physics, Graduate School of Advanced Science and Engineering, Faculty of Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku, Tokyo 169-8555, Japan

24{}^{24}Waseda Research Institute for Science and Engineering, Faculty of Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku, Tokyo 169-8555, Japan

25{}^{25}Centre for Astrophysics Research, University of Hertfordshire, Hatfield, AL10 9AB, UK

26{}^{26}School of Physics and Technology, Wuhan University, Wuhan 430072, China

27{}^{27}Department of Astronomy, The University of Texas at Austin, Austin, TX, USA

28{}^{28}Cosmic Frontier Center, The University of Texas at Austin, Austin, TX, USA

29{}^{29}Institute of Physics, GalSpec, Ecole Polytechnique Federale de Lausanne, Observatoire de Sauverny, Chemin Pegasi 51, 1290 Versoix, Switzerland

30{}^{30}INAF, Astronomical Observatory of Trieste, Via Tiepolo 11, 34131 Trieste, Italy

31{}^{31}INAF Osservatorio Astronomico di Trieste, Via G.B. Tiepolo 11 e via Bazzoni 2. 34100, Trieste, Italy

32{}^{32}IFPU: Institute for Fundamental Physics of the Universe, Via Beirut, 2, 34151, Trieste, Italy

33{}^{33}International Centre for Radio Astronomy Research, University of Western Australia, 7 Fairway, Crawley, WA 6009, Australia

34{}^{34}Research School of Astronomy and Astrophysics, Australian National University, Cotter Road, Weston Creek, ACT 2611, Australia

35{}^{35}University of Bologna - Department of Physics and Astronomy “Augusto Righi” (DIFA), Via Gobetti 93/2, I-40129 Bologna, Italy

36{}^{36}INAF- Osservatorio di Astrofisica e Scienza dello Spazio, Via Gobetti 93/3, I-40129 Bologna, Italy

37{}^{37}INFN- Sezione di Bologna, Viale Berti Pichat 6/2, I-40127 Bologna, Italy {sup}38`Department of Physics and Astronomy, University of Hawaii, Hilo, 200 W Kawili St, Hilo, HI 96720, USA

{sup}39`Department of Astronomy, National Research Institute of Astronomy and Geophysics (NRIAG), Cairo, 11421, Egypt

{sup}40`Laboratory for Multiwavelength Astrophysics, School of Physics and Astronomy, Rochester Institute of Technology, 84 Lomb Memorial Drive, Rochester, NY 14623, USA

{sup}41`Department of Astronomy, University of Illinois Urbana-Champaign, Urbana, IL 61801, USA

{sup}42`Aix Marseille Univ, CNRS, CNES, LAM, Marseille, France

{sup}43`IPAC, California Institute of Technology, 1200 E. California Blvd. Pasadena, CA 91125, USA

{sup}44`Institut d’Astrophysique de Paris, UMR 7095, CNRS, and Sorbonne Université, 98bis boulevard Arago, 75014 Paris, France

{sup}45`Institute of Astronomy and Astrophysics, Academia Sinica, 11F of Astronomy-Mathematics Building, No.1, Section 4, Roosevelt Rd, Taipei 106319, Taiwan, R.O.C

{sup}46`Zentrum für Astronomie, Universität Heidelberg, Philosophenweg 12, D-69120, Heidelberg, Germany

{sup}47`DTU Space, Technical University of Denmark, Elektrovej 327 DK2800 Kgs. Lyngby, Denmark

{sup}48`Department of Computer Science, Aalto University, P.O. Box 15400, FI-00076 Espoo, Finland

{sup}49`Department of Physics, University of Helsinki, P.O. Box 64, FI-00014 Helsinki, Finland

{sup}50`Department of Physics and Astronomy, Colby College, Waterville, ME 04901, USA

{sup}51`Waseda Institute for Advanced Study (WIAS), Waseda University, 1-21-1, Nishi-Waseda, Shinjuku, Tokyo 169-0051, Japan

{sup}52`Instituto de Astrofísica de Canarias (IAC), La Laguna, E-38205, Spain

{sup}53`Universidad de La Laguna. Avda. Astrofísico Fco. Sanchez, La Laguna, Tenerife, Spain

{sup}54`Observatoire de Paris, LUX, PSL University, 61 avenue de l’Observatoire, F-75014 Paris, France

{sup}55`Université de Paris, 5 Rue Thomas Mann, 75014 Paris, France

{sup}56`Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91001, USA

{sup}57`Niels Bohr Institute, University of Copenhagen, Jagtvej 128, DK2200 Copenhagen N, Denmark

{sup}58`Kapteyn Astronomical Institute, University of Groningen, PO Box 800, 9700 AV Groningen, The Netherlands

{sup}59`Institute of Cosmology and Gravitation, University of Portsmouth, Dennis Sciama Building, Burnaby Road, Portsmouth, PO1 3FX, UK

{sup}60`School of Astronomy and Space Science, Nanjing University, Nanjing, Jiangsu 210093, China

{sup}61`Key Laboratory of Modern Astronomy and Astrophysics, Nanjing University, Ministry of Education, Nanjing 210093, China

{sup}62`Department of Astronomy, University of Science and Technology of China, Hefei 230026, China

{sup}63`School of Astronomy and Space Science, University of Science and Technology of China, Hefei 230026, China

{sup}64`Department of Physics and Astronomy, University of California, Riverside, 900 University Avenue, Riverside, CA 92521, USA

{sup}65`Observatories of the Carnegie Institution for Science, 813 Santa Barbara St., Pasadena, CA 91101, USA

{sup}66`Department of Physics and Astronomy, University of Kentucky, 505 Rose Street, Lexington, KY 40506, USA

{sup}67`National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan

{sup}68`Institute for Astronomy, University of Hawaii, 2680 Woodlawn Drive, Honolulu, HI 96822, USA

{sup}69`Department of Astronomy, University of Massachusetts Amherst, 710 N Pleasant Street, Amherst, MA 01003, USA

References

  1. [1]Planck Collaboration, Aghanim, N., Akrami, Y., et al. 2020, A&A, 641, A6, doi: 10.1051/0004-6361/201833910
  2. [2]Akins, H. B., Casey, C. M., Lambrid es, E., et al. 2025, ApJ, 991, 37, doi: 10.3847/1538-4357/ade984DOI
  3. [3]Ananna, T. T., Bogdán, Á., Kovács, O. E., Natarajan, P., & Hickox, R. C. 2024, ApJL, 969, L18, doi: 10.3847/2041-8213/ad5669DOI
  4. [4]Ando, M., Harikane, Y., Katz, H., Inayoshi, K., & Tanaka, T. S. 2026, arXiv e-prints, arXiv:2606.03522, doi: 10.48550/arXiv.2606.03522DOI
  5. [5]Backhaus, B. E., Cleri, N. J., Trump, J. R., et al. 2025, ApJ, 994, 125, doi: 10.3847/1538-4357/ae1136DOI
  6. [6]Barro, G., Pérez-González, P. G., Kocevski, D. D., et al. 2024, ApJ, 963, 128, doi: 10.3847/1538-4357/ad167eDOI
  7. [7]Barro, G., Pérez-González, P. G., Kocevski, D., et al. 2026, ApJ, 1003, 96, doi: 10.3847/1538-4357/ae5d2cDOI
  8. [8]Brazzini, M., D’Eugenio, F., Maiolino, R., et al. 2026, arXiv e-prints, arXiv:2601.22214, doi: 10.48550/arXiv.2601.22214DOI
  9. [9]Brooks, M., Davis, K., Trump, J. R., et al. 2026, arXiv e-prints, arXiv:2608.12487, doi: 10.48550/arXiv.2608.12487arxiv.org/abs/2608.12487
  10. [10]Bushouse, H., Eisenhamer, J., Dencheva, N., et al. 2023, Zenodo, doi: 10.5281/zenodo.8380331
  11. [11]Cai, X., Inoue, A. K., Onoue, M., et al. 2026, arXiv e-prints, arXiv:2610.28735, doi: 10.48550/arXiv.2610.28735DOI
  12. [13]Casey, C. M., Akins, H. B., Finkelstein, S. L., et al. 2025, arXiv e-prints, arXiv:2505.18873, doi: 10.48550/arXiv.2505.18873DOI
  13. [14]Chang, S.-J., Gronke, M., Matthee, J., & Mason, C. 2026, MNRAS, 545, staf2131, doi: 10.1093/mnras/staf2131DOI
  14. [15]Chen, C.-H., Shangguan, J., Ho, L. C., et al. 2026, arXiv e-prints, arXiv:2606.04711, doi: 10.48550/arXiv.2606.04711DOI
  15. [17]de Graaff, A., Rix, H.-W., Naidu, R. P., et al. 2025a, A&A, 701, A168, doi: 10.1051/0004-6361/202554681DOI
  16. [20]Delvecchio, I., Daddi, E., Magnelli, B., et al. 2025, A&A, 704, A313, doi: 10.1051/0004-6361/202557164DOI
  17. [21]Dijkstra, M. 2014, PASA, 31, e040, doi: 10.1017/pasa.2014.33DOI
  18. [22]Feltre, A., Charlot, S., & Gutkin, J. 2016, MNRAS, 456, 3354, doi: 10.1093/mnras/stv2794DOI
  19. [23]Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306, doi: 10.1086/670067
  20. [25]Furtak, L. J., Labbé, I., Zitrin, A., et al. 2024, Nature, 628, 57, doi: 10.1038/s41586-024-07184-8DOI
  21. [26]Furtak, L. J., Secunda, A. R., Greene, J. E., et al. 2025, A&A, 698, A227, doi: 10.1051/0004-6361/202554110DOI
  22. [27]Geris, S., Maiolino, R., Ji, X., et al. 2026, arXiv e-prints, arXiv:2606.21614, doi: 10.48550/arXiv.2606.21614DOI
  23. [28]Gloudemans, A. J., Duncan, K. J., Eilers, A.-C., et al. 2025, ApJ, 986, 130, doi: 10.3847/1538-4357/adddb9DOI
  24. [29]Greene, J. E., Labbe, I., Goulding, A. D., et al. 2024, ApJ, 964, 39, doi: 10.3847/1538-4357/ad1e5fDOI
  25. [30]GREX-PLUS Science Team, Inoue, A. K., Harikane, Y., et al. 2023, arXiv e-prints, arXiv:2304.08104, doi: 10.48550/arXiv.2304.08104DOI
  26. [31]GREX-PLUS Science Team, Baba, S., Belli, S., et al. 2026, arXiv e-prints, arXiv:2606.05237, doi: 10.48550/arXiv.2606.05237DOI
  27. [32]Gutkin, J., Charlot, S., & Bruzual, G. 2016, MNRAS, 462, 1757, doi: 10.1093/mnras/stw1716DOI
  28. [33]Hainline, K. N., Maiolino, R., Juodžbalis, I., et al. 2025, ApJ, 979, 138, doi: 10.3847/1538-4357/ad9920DOI
  29. [34]Harikane, Y., Zhang, Y., Nakajima, K., et al. 2023, ApJ, 959, 39, doi: 10.3847/1538-4357/ad029eDOI
  30. [35]Horne, K. 1986, Publications of the Astronomical Society of the Pacific, 98, 609, doi: 10.1086/131801DOI
  31. [36]Hviding, R. E. 2026, unite: Unified liNe Integration Turbo Engine, v3.5.0 Zenodo, doi: 10.5281/zenodo.23366301
  32. [37]Hviding, R. E., de Graaff, A., Miller, T. B., et al. 2025, A&A, 702, A57, doi: 10.1051/0004-6361/202555816DOI
  33. [38]Hviding, R. E., de Graaff, A., Liu, H., et al. 2026, ApJL, 1000, L18, doi: 10.3847/2041-8213/ae4c88DOI
  34. [40]Inayoshi, K., & Ho, L. C. 2025, arXiv e-prints, arXiv:2512.03130, doi: 10.48550/arXiv.2512.03130arxiv.org/abs/2512.03130
  35. [41]Inayoshi, K., & Maiolino, R. 2025, ApJL, 980, L27, doi: 10.3847/2041-8213/adeabd
  36. [42]Inayoshi, K., Murase, K., & Kashiyama, K. 2026, ApJ, 1000, 90, doi: 10.3847/1538-4357/ae42ceDOI
  37. [43]Inayoshi, K., Visbal, E., & Haiman, Z. 2020, Annual Review of Astronomy and Astrophysics, 58, 27, doi: 10.1146/annurev-astro-120419-014455DOI
  38. [44]Inoue, A. K. 2011, MNRAS, 415, 2920, doi: 10.1111/j.1365-2966.2011.18906.xDOI
  39. [45]Inoue, A. K., Yamamura, I., Suzuki, T., et al. 2024, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 13092, Space Telescopes and Instrumentation 2024: Optical, Infrared, and Millimeter Wave, ed. L. E. Coyle, S. Matsuura, & M. D. Perrin, 130920Y, doi: 10.1117/12.3017780DOI
  40. [50]Johnson, B. D., Leja, J., Conroy, C., & Speagle, J. S. 2021, ApJS, 254, 22, doi: 10.3847/1538-4357/abef67
  41. [51]Jones, G. C., Übler, H., Maiolino, R., et al. 2026, MNRAS, 546, stag115, doi: 10.1093/mnras/stag115DOI
  42. [52]Juodžbalis, I., Maiolino, R., Baker, W. M., et al. 2025, arXiv e-prints, arXiv:2504.03551, doi: 10.48550/arXiv.2504.03551arxiv.org/abs/2504.03551
  43. [54]Kido, D., Ioka, K., Hotokezaka, K., Inayoshi, K., & Irwin, C. M. 2025, MNRAS, 544, 3407, doi: 10.1093/mnras/staf1898
  44. [55]Kocevski, D. D., Onoue, M., Inayoshi, K., et al. 2023, ApJL, 954, L4, doi: 10.3847/2041-8213/ace5a0DOI
  45. [56]Kocevski, D. D., Finkelstein, S. L., Barro, G., et al. 2025, ApJ, 986, 126, doi: 10.3847/1538-4357/adbc7dDOI
  46. [57]Kocevski, D. D., Taylor, A. J., Onoue, M., et al. 2026, arXiv e-prints, arXiv:2609.00112, https://arxiv.org/abs/2609.00112
  47. [58]Kokorev, V., Chisholm, J., Endsley, R., et al. 2024, ApJ, 975, 178, doi: 10.3847/1538-4357/ad7d03DOI
  48. [59]Kokubo, M., & Harikane, Y. 2025, ApJ, 995, 24, doi: 10.3847/1538-4357/ae119eDOI
  49. [60]Kormendy, J., & Ho, L. C. 2013, ARA&A, 51, 511, doi: 10.1146/annurev-astro-082708-101811DOI
  50. [61]Labbé, I., Dokkum, P. N., Nelson, E., et al. 2023, Nature, 616, 266, doi: 10.1038/s41586-023-05786-2DOI
  51. [62]Labbe, I., Greene, J. E., Matthee, J., et al. 2024, arXiv e-prints, arXiv:2412.04557, doi: 10.48550/arXiv.2412.04557DOI
  52. [63]Labbe, I., Greene, J. E., Bezanson, R., et al. 2025, ApJ, 978, 92, doi: 10.3847/1538-4357/ad3551DOI
  53. [64]Lambrides, E., Hutchison, T. A., Larson, R. L., et al. 2026, arXiv e-prints, arXiv:2604.25991, doi: 10.48550/arXiv.2604.25991DOI
  54. [67]Lin, X., Fan, X., Cai, Z., et al. 2026a, ApJ, 997, 364, doi: 10.3847/1538-4357/ae2bdfDOI
  55. [68]Lin, X., Fan, X., Wang, F., et al. 2026b, ApJ, 996, 93, doi: 10.3847/1538-4357/ae1b9bDOI
  56. [69]Liu, Z., Naidu, R. P., Secunda, A., et al. 2026, arXiv e-prints, arXiv:2604.13000, doi: 10.48550/arXiv.2604.13000DOI
  57. [70]Madau, P., & Maiolino, R. 2026, arXiv e-prints, arXiv:2602.22386, doi: 10.48550/arXiv.2602.22386arxiv.org/abs/2602.22386
  58. [71]Madau, P., Maiolino, R., & D’Eugenio, F. 2026, arXiv e-prints, arXiv:2609.06239, doi: 10.48550/arXiv.2609.06239DOI
  59. [73]Mannucci, F., Della Valle, M., & Panagia, N. 2006, MNRAS, 370, 773, doi: 10.1111/j.1365-2966.2006.10501.xDOI
  60. [74]Maoz, D., Mannucci, F., & Nelemans, G. 2014, ARA&A, 52, 107, doi: 10.1146/annurev-astro-082812-141031
  61. [75]Martin, C. L., Hu, W., Wold, I. G. B., et al. 2026, ApJ, 997, 102, doi: 10.3847/1538-4357/ae1d6bDOI
  62. [76]Matthee, J., Naidu, R. P., Brammer, G., et al. 2024, ApJ, 963, 129, doi: 10.3847/1538-4357/ad2345DOI
  63. [77]Mazzolari, G., Übler, H., Maiolino, R., et al. 2024, A&A, 691, A345, doi: 10.1051/0004-6361/202450407DOI
  64. [78]Mazzolari, G., Gilli, R., Maiolino, R., et al. 2026, A&A, 706, A372, doi: 10.1051/0004-6361/202453317DOI
  65. [80]Naidu, R. P., Matthee, J., Katz, H., et al. 2026, Nature, 656, 329, doi: 10.1038/s41586-026-10846-4DOI
  66. [81]Nakane, M., Ouchi, M., Nakajima, K., et al. 2024, ApJ, 976, 122, doi: 10.3847/1538-4357/ad84e8DOI
  67. [82]Nakane, M., Ouchi, M., Nakajima, K., et al. 2025, ApJ, 994, 65, doi: 10.3847/1538-4357/ae04e6DOI
  68. [83]Nikopoulos, G. P., Watson, D., Sneppen, A., et al. 2026, A&A, 710, A136, doi: 10.1051/0004-6361/202557604DOI
  69. [84]Ouchi, M., Ono, Y., & Shibuya, T. 2020, ARA&A, 58, 617, doi: 10.1146/annurev-astro-032620-021859DOI
  70. [85]Pacucci, F., Ferrara, A., & Kocevski, D. D. 2026, arXiv e-prints, arXiv:2601.14368, doi: 10.48550/arXiv.2601.14368DOI
  71. [86]Pang, Y., Wang, X., Cheng, C., et al. 2026, ApJ, 1004, 185, doi: 10.3847/1538-4357/ae6e3fDOI
  72. [87]Pérez-González, P. G., Barro, G., Carniani, S., et al. 2026, arXiv e-prints, arXiv:2602.20247, doi: 10.48550/arXiv.2602.20247DOI
  73. [88]Perger, K., Fogasy, J., Frey, S., & Gabányi, K. É. 2025, A&A, 693, L2, doi: 10.1051/0004-6361/202452422DOI
  74. [89]Regan, J., & Volonteri, M. 2024, The Open Journal of Astrophysics, 7, 72, doi: 10.33332/001c.123239
  75. [90]Rigby, J., Perrin, M., McElwain, M., et al. 2023, PASP, 135, 048001, doi: 10.1088/1538-3873/acb293DOI
  76. [91]Rinaldi, P., Bonaventura, N., Rieke, G. H., et al. 2024, arXiv e-prints, arXiv:2411.14383, doi: 10.48550/arXiv.2411.14383arxiv.org/abs/2411.14383
  77. [92]Rodriguez, L. F., & Mirabel, I. F. 2026, A&A, 707, L17, doi: 10.1051/0004-6361/202558326DOI
  78. [93]Rodríguez, L. F., Mirabel, I. F., González-Lópezlira, R. A., & Loinard, L. 2026, ApJ, 1009, 217, doi: 10.3847/1538-4357/ae9b01DOI
  79. [94]Rusakov, V., Watson, D., Nikopoulos, G. P., et al. 2025, arXiv e-prints, arXiv:2503.16595, doi: 10.48550/arXiv.2503.16595DOI
  80. [95]Setton, D. J., Greene, J. E., de Graaff, A., et al. 2025a, ApJ, 995, 118, doi: 10.3847/1538-4357/ae1500DOI
  81. [96]Setton, D. J., Greene, J. E., Spilker, J. S., et al. 2025b, ApJL, 991, L10, doi: 10.3847/2041-8213/ade78bDOI
  82. [97]Sigut, T. A. A., Pradhan, A. K., & Nahar, S. N. 2004, ApJ, 611, 81, doi: 10.1086/422027DOI
  83. [98]Sneppen, A., Watson, D., Matthews, J. H., et al. 2026a, arXiv e-prints, arXiv:2601.18864, doi: 10.48550/arXiv.2601.18864arxiv.org/abs/2601.18864
  84. [100]Stone, Z., Shen, Y., Zhuang, M.-Y., et al. 2026, arXiv e-prints, arXiv:2608.01647, doi: 10.48550/arXiv.2608.01647DOI
  85. [101]Sun, W. Q., Naidu, R. P., Matthee, J., et al. 2026, The Open Journal of Astrophysics, 9, 62505, doi: 10.33332/001c.162505
  86. [102]Tanaka, T. S., Akins, H. B., Harikane, Y., et al. 2025, ApJ, 995, 21, doi: 10.3847/1538-4357/ae145fDOI
  87. [103]Tang, M., Stark, D. P., Chen, Z., et al. 2023, MNRAS, 526, 1657, doi: 10.1093/mnras/stad2763DOI
  88. [104]Tang, M., Stark, D. P., Mason, C. A., et al. 2026, arXiv e-prints, arXiv:2604.03563, doi: 10.48550/arXiv.2604.03563DOI
  89. [105]Taylor, A. J., Kokorev, V., Kocevski, D. D., et al. 2025, arXiv e-prints, arXiv:2505.04609, doi: 10.48550/arXiv.2505.04609
  90. [106]Tee, W. L., Fan, X., Wang, F., & Yang, J. 2025, ApJL, 983, L26, doi: 10.3847/2041-8213/adc5e3DOI
  91. [107]Torralba, A., Matthee, J., Pezzulli, G., et al. 2026, A&A, 707, A75, doi: 10.1051/0004-6361/202557537DOI
  92. [109]Tripodi, R., Bradač, M., D’Eugenio, F., et al. 2025, ApJL, 994, L6, doi: 10.3847/2041-8213/ae13a9DOI
  93. [110]Umeda, H., Inayoshi, K., Harikane, Y., & Murase, K. 2026, ApJ, 999, 183, doi: 10.3847/1538-4357/ae4101DOI
  94. [111]Umeda, H., Ouchi, M., Nakajima, K., et al. 2024, ApJ, 971, 124, doi: 10.3847/1538-4357/ad554eDOI
  95. [112]Vestergaard, M., & Wilkes, B. J. 2001, ApJS, 134, 1, doi: 10.1086/320357DOI
  96. [113]Volonteri, M. 2010, A&A Rev, 18, 279, doi: 10.1007/s00159-010-0029-xDOI
  97. [114]Wang, B., de Graaff, A., Davies, R. L., et al. 2025, ApJ, 984, 121, doi: 10.3847/1538-4357/adc1caDOI
  98. [115]Wills, B. J., Netzer, H., & Wills, D. 1980, ApJL, 242, L1, doi: 10.1086/183388DOI
  99. [116]Withers, S., Muzzin, A., Ravindranath, S., et al. 2023, ApJL, 958, L14, doi: 10.3847/2041-8213/ad01c0DOI
  100. [117]Witstok, J., Jakobsen, P., Maiolino, R., et al. 2025, Nature, 639, 897, doi: 10.1038/s41586-025-08779-5DOI
  101. [118]Yan, Z., Inayoshi, K., Chen, K., & Guo, J. 2026, ApJ, 1002, 159, doi: 10.3847/1538-4357/ae5dbdDOI
  102. [119]Yanagisawa, H., Ouchi, M., Kiyota, T., et al. 2026, arXiv e-prints, arXiv:2606.09726, doi: 10.48550/arXiv.2606.09726DOI
  103. [120]Yue, M., Eilers, A.-C., Annan, T. T., et al. 2024, ApJL, 974, L26, doi: 10.3847/2041-8213/ad7ebaDOI
  104. [121]Zhang, Z., Jiang, L., Liu, W., & Ho, L. C. 2025a, ApJ, 985, 119, doi: 10.3847/1538-4357/adcb3eDOI
  105. [122]Zhang, Z., Li, M., Oguri, M., et al. 2025b, arXiv e-prints, arXiv:2512.05180, doi: 10.48550/arXiv.2512.05180DOI
  106. [123]Zhou, S., Sun, M., Zhang, Z., Chen, J., & Ho, L. C. 2025, ApJ, 991, 137, doi: 10.3847/1538-4357/adfd5fDOI
  107. [124]Zhou, S., Sun, M., Ji, X., et al. 2026, arXiv e-prints, arXiv:2606.21105, doi: 10.48550/arXiv.2606.21105DOI

Paper details

Contents