A super-Eddington-accreting black hole ~1.5 Gyr after the Big Bang observed with JWST

The study addresses how massive black holes formed and grew rapidly in the early universe. Luminous quasars observed at z > 6–7 imply the existence of supermassive black holes (SMBHs) of 10⁹–10¹⁰ M⊙ within the first billion years, challenging standard growth scenarios and motivating investigations into black hole seeding and accretion histories. Seed black holes may originate as light seeds from Population III stellar remnants or as heavy seeds from direct gas collapse; each pathway implies distinct early growth requirements. With JWST’s sensitivity, faint, compact, dust-reddened AGNs at z > 4 have been uncovered, often hosting overmassive black holes relative to their galaxies and typically lacking X-ray detections. The paper presents LID-568, an extremely red, compact, X-ray-bright AGN at z_spec = 3.965 discovered among near-IR-dropout X-ray sources. The research aims to measure its black hole mass, accretion rate, outflow properties, spectral energy distribution, and environment to test whether super-Eddington accretion phases can drive rapid early black hole growth in low-mass hosts.

Prior surveys have identified hundreds of z > 6–7 quasars, implying rapid SMBH assembly and prompting theoretical models of black hole seeds divided into light (∼10²–10³ M⊙, stellar remnants) and heavy (∼10⁴ M⊙, direct collapse) classes, with intermediate channels via massive stars, stellar mergers, and black hole mergers. Theoretical expectations differ in number densities and required accretion histories, with heavy seeds easing early growth demands but forming in rarer environments. JWST has revealed an abundant population of faint, compact, dust-reddened AGNs at z ≳ 4–7 (often called little red dots), with overmassive BHs relative to their stellar hosts and typically moderate Eddington ratios (∼0.2). Most are X-ray undetected, with only a few reported X-ray detections. These findings suggest early AGN phases in relatively low-mass galaxies and indicate that black hole growth can precede galaxy growth. Theoretical work also predicts that super-Eddington accretion may be episodic and could be sustained for tens of Myr, potentially producing distinctive SEDs (either extremely red IR-dominated or, in alternative models, UV-excess).

Parent sample and selection: The study began from a catalog of near-IR-dropout X-ray sources in the Chandra COSMOS Legacy Survey (4016 X-ray sources over ~2.2 deg²). Using COSMOS2020 and HELP photometry spanning UV to far-IR, sources with no optical counterparts within 2″ of the Chandra positions were selected, excluding contaminated or diffuse X-ray cases, yielding 62 IR-dropout X-ray sources. ALMA observations: Band 7 (275–373 GHz) continuum observations (Cycle 7 program 2019.1.01275.S) were obtained for all 62 sources (42–46 antennas, ~5 min integration per source). Data were reduced with the ALMA pipeline (CASA v6.2.1.7). LID-568 has S_870µm = 545 ± 158 µJy, spatially consistent with Spitzer/IRAC positions. JWST spectroscopy and reduction: NIRSpec/IFU G395M/F290LP (3–5 µm, R ≈ 1000) observations in April 2023 (1.45 h total, four-point dither, NRSIRS2 readout) and MIRI/LRS slit spectroscopy in Jan 2023 (5–12 µm, R ≈ 100, 1.1 h total). NIRSpec data were reduced with JWST calibration pipeline v1.11.4 (CRDS jwst_1149.pmap) with additional steps: 1/f noise removal via per-column median subtraction with sigma clipping, flagging saturated/bad pixels as DO_NOT_USE, additional outlier voxel filtering, and channel-by-channel background subtraction to remove zodiacal/stray light gradients. MIRI data were processed via v1.12.5 (CRDS jwst_1135.pmap) from MAST. Spectral analysis and line fitting: NIRSpec spectra (aperture r = 0.2″) were continuum-subtracted with a power-law and fitted with combinations of narrow and broad Gaussians for Hα, [N II] 6548,6583 (fixed flux ratio 2.96), and [S II] 6716,6731, including blue-shifted broad components to capture outflows. A broad Hα component represents the BLR; additional blue-shifted broad components trace outflowing gas. Best-fit parameters yield the broad-line width and Hα luminosity used for single-epoch virial black hole mass estimates. IFU channel maps around Hα were constructed in velocity bins (165 km s⁻¹ steps) to map spatially extended components. X-ray spectral modeling: Chandra spectrum (0.5–7 keV) was fit in XSPEC v12.13.0 with a power law (Γ fixed to 1.9), Galactic absorption (N_H = 2.6×10²⁰ cm⁻²), and intrinsic absorption at source redshift. Column density and absorption-corrected L_2–10 keV were derived. MYtorus modeling was also performed (inclination 75°, Γ=1.9) to account for reflection and fluorescence; results were consistent with Compton-thin obscuration. Allowing Γ to vary produced softer spectra and higher inferred N_H and luminosities; adopting Γ=1.9 is conservative. SED fitting: Multiwavelength SEDs were fit with AGNfitter-like libraries and with CIGALE/X-CIGALE configurations. The source exhibits an unusually red IR continuum not reproducible by standard AGN/galaxy SED libraries. A modified IR SED model with a mid-IR power law and two greybodies (β=1.5, α free) was used to derive dust temperatures, total IR luminosity, and dust mass. Bolometric luminosity estimation: L_bol was inferred from L_2–10 keV via luminosity-dependent bolometric corrections and by integrating intrinsic X-ray (0.1–100 keV) and corrected torus IR luminosity accounting for covering factor and anisotropy. An Hα-based bolometric estimate was also computed using Hα-to-5100 Å conversion and a standard bolometric correction. Environment: Galaxy overdensity around LID-568 was evaluated using Voronoi tessellation Monte Carlo density maps (2<z<5) combining spectroscopic and photometric redshifts. Post-processing linked overdensity detections across slices to identify coherent structures and estimate their masses. Completeness was >50% only for massive protoclusters (z=0 mass >10¹⁴.⁵ M⊙) at z=4 in this region.

Redshift and nature: LID-568 is at z_spec = 3.965, extremely red and compact in the IR, invisible in deep optical/near-IR imaging, and X-ray bright relative to other JWST faint AGNs. X-ray properties: Observed 0.5–10 keV flux = 5.16×10⁻¹⁵ erg cm⁻² s⁻¹. Intrinsic obscuration log N_H = 23.44 (−0.34 + 0.47) cm⁻². Absorption-corrected luminosity log L_2–10 keV = 44.79 (−0.33 + 0.27) erg s⁻¹. Bolometric luminosity: log L_bol ≈ 46.6 (−0.44 + 0.36) erg s⁻¹ from X-ray, consistent with SED-based L_bol ≈ 46.68. Hα-based bolometric estimate is ∼1 dex lower (log L_bol ≈ 45.60), indicating heavy obscuration of Hα. Black hole mass and accretion: Single-epoch virial mass M_BH = 7.2 (−5.4 + 10.8) × 10⁶ M⊙ from broad Hα. Eddington ratio L_bol/L_Edd ≈ 41.5, indicating extreme super-Eddington accretion. IR SED and dust: Rest-frame IR continuum is a very steep power law with α ≈ 4.5 for λ_rest ≥ 1 µm, steeper than typical JWST faint AGNs (α ≈ 2). Two-temperature greybody fit yields hot and warm dust components at ~656 K and ~71.5 K. Total IR luminosity log L_8–1000µm ≈ 46.1 erg s⁻¹, comparable to L_bol. Dust mass M_dust ≈ 2.95×10⁸ M⊙; inferred stellar mass ≈ 2×10⁹ M⊙, implying a low-mass host galaxy. Little evidence for ongoing star formation. Outflows: IFU channel maps reveal spatially extended, blue-shifted Hα emission at −600 to −500 km s⁻¹ relative to the BLR, peaking at projected distances of ~0.4″ (~3 kpc, north; component B) and ~1″ (~7 kpc, south; component D) from the nucleus; a northeastern component (A) is near systemic. Nuclear ionized outflow velocity is ~−540 km s⁻¹, consistent across Hα and [S II]. Ionized outflow mass from the blue-shifted Hα component is ~1.4×10⁷ M⊙; adopting v_out ≈ 540 km s⁻¹ and r_out ≈ 7 kpc yields an outflow rate of ~3.1 M⊙ yr⁻¹. AGN lifetime estimate: If the extended Hα emission traces the outflow, the AGN lifetime is τ ≈ (7 kpc)/(540 km s⁻¹) ≈ 1.2×10⁷ yr. Environment: No evidence for a significant overdensity; local overdensity log(1+δ) ≈ 0.11 (~1σ above mean), and no coherent structure within Δz = 0.04 and R < 5 proper Mpc. Preburst mass estimate: Assuming super-Eddington accretion at the observed Eddington ratio for ~12 Myr with ε=0.1, the preburst BH mass could be as low as ~10² M⊙ (consistent with a light seed), though this is a lower limit and model-dependent.

The discovery of LID-568 demonstrates that low-mass black holes at z ≈ 4 can undergo phases of extreme, super-Eddington accretion, offering a viable pathway to rapid early SMBH growth. Its high L_bol and low M_BH place it in a previously underexplored accretion regime among JWST-discovered faint AGNs, with the key distinction of being X-ray luminous. The unusually red IR SED, hot dust temperatures, and lack of clear star-formation signatures indicate that the IR emission is dominated by an obscured accretion disk/torus rather than the host galaxy. Spatially resolved, blue-shifted Hα components at several kiloparsecs from the nucleus, together with the nuclear outflow signature, suggest AGN-driven feedback, potentially suppressing star formation in a low-mass host and contributing to overmassive BHs relative to their galaxies. The inferred AGN lifetime (~10⁷ yr) and the possibility of sustained super-Eddington phases over tens of Myr are consistent with theoretical models positing episodic supercritical accretion as a key ingredient in early BH growth. A back-extrapolation of the observed growth implies a light-seed-scale preburst mass; however, this remains a lower limit contingent on assumptions about radiative efficiency, duty cycle, and gas supply. The lack of an environmental overdensity suggests that such phases need not be restricted to massive protocluster environments. Overall, the results support scenarios in which intermittent super-Eddington episodes, regardless of seed origin, drive rapid BH mass assembly and contribute to the observed population of overmassive BHs in low-mass galaxies at high redshift.

LID-568 is a z ≈ 4, X-ray-bright, low-mass (∼7×10⁶ M⊙) black hole accreting at L_bol/L_Edd ≈ 41.5, with powerful ionized outflows extending to ~7 kpc and an IR SED dominated by hot dust from a heavily obscured nucleus. These observations capture an extreme, likely episodic super-Eddington phase, illuminating a key parameter space of early black hole growth and providing empirical support for rapid mass assembly in low-mass hosts. The source’s properties suggest that AGN feedback can regulate or quench star formation in such systems and that super-Eddington accretion may be an important, recurrent process in building SMBHs. Future work should assemble larger samples of similar objects to measure the duty cycle of super-Eddington phases, refine black hole mass estimates (e.g., with reverberation mapping or alternative lines), better constrain obscuration and orientation effects through multiwavelength SED and X-ray modeling, and map environments with deeper spectroscopy and additional outflow tracers such as [O III].

Key limitations include uncertainties in single-epoch virial black hole mass estimates (systematic scatter ~0.3 dex), potential overestimation of M_BH for high accretion rate AGNs due to smaller-than-expected BLR sizes, and heavy obscuration that likely causes the Hα-based bolometric luminosity to be underestimated by ~1 dex. The interpretation of extended Hα as outflow is plausible but not definitive; a merger origin cannot be excluded, and no continuum counterpart is detected for the extended components. The [O III] line, a standard outflow tracer, is not covered by the data. X-ray luminosity estimates depend on assumed spectral indices and torus geometry; adopting a fixed Γ=1.9 is conservative, but different Γ would change inferred N_H and L_X. SED modeling required custom components as standard templates could not reproduce the extremely red IR continuum, introducing model dependence in dust and luminosity estimates. The environment analysis is limited by sparse spectroscopy at z ≈ 4 and is complete only for very massive structures, leaving open the possibility of association with lower-mass overdensities. Results are based on a single object, limiting generalizability until larger samples are studied.