HYPERION: Early Hyperluminous Quasars

• HYPERION is a multiwavelength experiment studying z>6 hyperluminous quasars with extreme supermassive black holes to test early accretion and feedback mechanisms.
• It integrates X-ray spectroscopy, UV–NIR photometry, and ALMA submillimeter imaging to reconstruct detailed spectral energy distributions and host galaxy properties.
• The experiment refines growth models by deriving revised bolometric corrections and gas-to-dust ratios, thereby challenging standard black hole evolution scenarios.

HYPERION, expanded as HYPerluminous quasars at the Epoch of ReionizatION, is a multiwavelength experiment centered on the brightest quasars at $z \gtrsim 6$, with the aim of characterizing their X-ray-to-near-infrared spectral energy distributions, X-ray nuclear properties, cold interstellar medium, and host-galaxy growth within the first billion years of cosmic history. The experiment is motivated by the existence of quasars powered by supermassive black holes with masses $M_{\rm BH}\gtrsim 10^9\,M_\odot$ at the Epoch of Reionization, a regime that challenges standard black-hole growth scenarios and makes these sources a natural laboratory for testing the early establishment of accretion physics, coronal emission, dust reprocessing, and black-hole–galaxy coevolution (Saccheo et al., 2024).

1. Scientific rationale and sample definition

The central selection principle of HYPERION is to isolate the quasars whose black holes must have undergone the most extreme growth during the Universe’s first gigayear. In the experiment’s growth-based framing, black-hole mass evolves under continuous Eddington-limited accretion as

$$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$

so reaching $10^9\,M_\odot$ by $z\sim 6$ requires either heavy seeds or prolonged super-Eddington episodes from light seeds. HYPERION therefore selects objects for which, under continuous Eddington-limited accretion since $z=20$, the seed masses required to reach the observed black-hole masses exceed $10^3\,M_\odot$ (Zappacosta et al., 2023).

From 46 unlensed, radio-quiet hyperluminous quasars with published single-epoch Mg II virial masses, 18 satisfy the HYPERION criterion. These objects span $z\simeq 6.0$–7.5, with mean $z\approx 6.7$, $\log(L_{\rm bol}/{\rm erg\,s^{-1}})\approx 47.3$, and black-hole masses of approximately $M_{\rm BH}\gtrsim 10^9\,M_\odot$0–$M_{\rm BH}\gtrsim 10^9\,M_\odot$1, corresponding to Eddington ratios $M_{\rm BH}\gtrsim 10^9\,M_\odot$2–2.6. For the broad-band SED analysis, 36 additional quasars at $M_{\rm BH}\gtrsim 10^9\,M_\odot$3 from the E-XQR-30 XSHOOTER legacy survey were included to strengthen rest-UV–NIR statistics (Saccheo et al., 2024).

This selection defines HYPERION not simply as a high-redshift quasar survey, but as an experiment on the extremal tail of early SMBH growth. A plausible implication is that any regularities found across this sample pertain to the most rapidly assembled and most luminous accretion systems at the Epoch of Reionization, rather than to the full underlying quasar population.

2. Observational architecture across wavebands

The observational backbone of HYPERION is an XMM–Newton Multi-Year Heritage Key Project. One description of the program emphasizes a 2.4 Ms XMM–Newton campaign designed to deliver uniform, high-quality spectra with more than 100 net counts per source in the observed 0.5–10 keV band for all 18 quasars, while the broad-band SED study uses XMM–Newton EPIC (PN + MOS) data totaling $M_{\rm BH}\gtrsim 10^9\,M_\odot$4 ks to provide 0.3–7 keV spectra for the HYPERION objects (Zappacosta et al., 2023).

The experiment is explicitly multiwavelength. Rest-frame UV-to-NIR coverage combines archival and new imaging in the $M_{\rm BH}\gtrsim 10^9\,M_\odot$5, $M_{\rm BH}\gtrsim 10^9\,M_\odot$6, $M_{\rm BH}\gtrsim 10^9\,M_\odot$7, $M_{\rm BH}\gtrsim 10^9\,M_\odot$8, and $M_{\rm BH}\gtrsim 10^9\,M_\odot$9 bands from DES, VHS, UKIDSS, and Pan-STARRS, together with proprietary programs at TNG/NICS and NTT/SOFI. Mid-infrared photometry comes from WISE W1–W4 and, when available, Spitzer IRAC and MIPS; for two bright targets, JWST/MIRI and Bosman et al. (2023) provided 1 $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$0m spectral luminosities (Saccheo et al., 2024).

The host-galaxy arm of HYPERION uses low- to high-frequency ALMA observations of ten quasars at $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$1, selected to have robust black-hole masses, $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$2erg s$$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$3, continuum coverage from $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$4 GHz up to $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$5 GHz, and observations of at least one mid-$$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$6 CO line or [C II]. The ALMA data set spans Bands 3, 6, 8, and 9, with a supplementary NOEMA 3 mm measurement for one source, thereby sampling both the Rayleigh–Jeans tail and the peak of the cold-dust SED (Tripodi et al., 2024).

A particularly illustrative extension is the deep ALMA study of SDSS J0100+2802, the most luminous known $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$7 quasar, observed in Band 6 around the [C II] 158 $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$8m transition and in Band 3 continuum. That study was optimized for faint extended emission rather than only the compact quasar core, an observational design choice that proved consequential for identifying both a close companion and an outflow (Tripodi et al., 2023).

3. Spectral modeling, SED reconstruction, and derived quantities

The broad-band analysis combines photometric reduction, X-ray spectral fitting, and EUV interpolation into a single SED framework. UV–NIR data reduction uses standard sky subtraction, flat fielding, coaddition, and astrometric alignment, followed by aperture photometry with calibration against 2MASS or local standard stars. Each quasar’s UV–NIR points at $$M_{\rm BH}(t)=M_{\rm seed}\,\exp\Bigl(\frac{t}{t_s}\Bigr), \qquad t_s = 0.45\,\frac{\epsilon}{(1-\epsilon)}\,f_{\rm duty}^{-1}\;\mathrm{Gyr},$$9 Å are corrected for emission-line contamination by subtracting $10^9\,M_\odot$0 computed from the Vanden Berk et al. (2001) composite (Saccheo et al., 2024).

The UV–NIR continuum is fit with two high-luminosity empirical templates, lum-K13 and WISSH-S23, plus a dust-reddening screen $10^9\,M_\odot$1 following the Prevot et al. (1984) law with color excess $10^9\,M_\odot$2. The X-ray spectra are modeled with Galactic-absorbed power laws; in the first HYPERION X-ray analysis, spectra were grouped with the Kaastra–Bleeker optimal scheme and fit in XSPEC with

$10^9\,M_\odot$3

using Cash statistics with background subtraction, and no additional intrinsic column was required (Zappacosta et al., 2023).

The unobserved EUV interval is reconstructed with a double power law anchored at 500 Å and 1 keV, using the dereddened UV template at long wavelengths and the measured X-ray luminosity to determine the connecting slope. This yields uniform monochromatic luminosities, bolometric luminosities, and Eddington ratios across the sample. The central definitions are

$10^9\,M_\odot$4

Hot-dust emission strength is parameterized at rest-frame 2 $10^9\,M_\odot$5m as

$10^9\,M_\odot$6

with the NIR modeled as a single-temperature blackbody at $10^9\,M_\odot$7 K plus a power-law continuum (Saccheo et al., 2024).

One of the experiment’s methodological consequences is a revised optical bolometric correction. Comparing integrated $10^9\,M_\odot$8 to $10^9\,M_\odot$9 gives

$z\sim 6$0

rather than the canonical 5.15. In the HYPERION analysis, the older correction overestimates $z\sim 6$1 by $z\sim 6$2 dex because it includes reprocessed infrared emission (Saccheo et al., 2024).

4. Cold interstellar medium, star formation, and environmental structure

The ALMA branch of HYPERION addresses the host-galaxy side of the experiment. All ALMA visibilities were calibrated by the ALMA pipeline and imaged in CASA v5.1.1 with tclean in multifrequency-synthesis mode, natural weighting, and a $z\sim 6$3 clean threshold. Spectral cubes were continuum-subtracted with uvcontsub, rebinned to 20–50 km s$z\sim 6$4, and fit with single Gaussians to obtain line centroids, FWHM values, integrated CO fluxes, and peak fluxes. Continuum photometry from $z\sim 6$5 GHz to $z\sim 6$6 GHz was fit with an optically thick modified black body plus CMB correction using MCMC (EMCEE) over $z\sim 6$7 (Tripodi et al., 2024).

The derived cold-gas and dust quantities are based on

$z\sim 6$8

with $z\sim 6$9 and $z=20$0 at $z=20$1 GHz. Star-formation rates and depletion times are then expressed as

$z=20$2

Across the ten quasars, the mean molecular-gas mass is $z=20$3, dust temperatures are 50–85 K, the mean emissivity index is $z=20$4, gas-to-dust ratios span 30 to 250 with an average of about 120, and the HYPERION objects tend toward higher gas-to-dust ratios of about 150 because of lower dust masses. After correcting for a 50% AGN heating contribution, the inferred star-formation rates are 200–2500 $z=20$5 and the gas-depletion timescales are $z=20$6–0.05 Gyr (Tripodi et al., 2024).

The J0100+2802 study adds a resolved environmental example. Deep ALMA observations reveal a clumpy, tidally stretched companion extending $z=20$7 ($z=20$8 kpc) west of the quasar, with [C II] peaking at $z=20$9. The companion has a derived dust mass of $10^3\,M_\odot$0–$10^3\,M_\odot$1 and a star-formation rate of 35–344 $10^3\,M_\odot$2, while the quasar itself shows a broad blueshifted [C II] component interpreted as an outflow with $10^3\,M_\odot$3–269 $10^3\,M_\odot$4. The host and companion are both gas rich, and the reported morphology and kinematics are consistent with a major merger and a jet-aligned cold outflow (Tripodi et al., 2023).

5. Principal empirical results

In X-rays, the first-year joint analysis of ten HYPERION quasars reports a steep mean photon index of $10^3\,M_\odot$5, inconsistent at $10^3\,M_\odot$6 with the canonical $10^3\,M_\odot$7–2.0 measured in lower-redshift quasars. An alternative fit fixes $10^3\,M_\odot$8 and yields a high-energy cutoff of $10^3\,M_\odot$9 keV in the rest frame. The same analysis finds that the optical-to-X-ray spectral index lies systematically above the low-$z\simeq 6.0$0 $z\simeq 6.0$1–$z\simeq 6.0$2 relation by $z\simeq 6.0$3, implying relative X-ray brightness at fixed UV luminosity (Zappacosta et al., 2023).

In the UV–optical, the broad-band HYPERION study finds that the quasars’ continua can be modeled with templates of luminous $z\simeq 6.0$4 type-1 quasars. When 54 quasars are normalized at rest-frame 3500 Å and combined into a mean SED, the result matches the lum-K13 and WISSH-S23 templates within $z\simeq 6.0$5 up to 1 $z\simeq 6.0$6m. The fitted broken-power-law continuum has $z\simeq 6.0$7 for $z\simeq 6.0$8 and $z\simeq 6.0$9 for $z\approx 6.7$0. The UV slope is close to the values reported by Lusso et al. (2015) and Telfer et al. (2002), while the optical slope is significantly steeper than the Vanden Berk et al. (2001) composite for lower-luminosity quasars (Saccheo et al., 2024).

In the near infrared, the mean Epoch-of-Reionization SED shows a mildly enhanced hot-dust bump relative to lower-redshift templates, although the analysis explicitly notes large scatter and possible selection bias toward dust-rich objects. For the subsample with rest-frame NIR photometry, hot-dust strengths span approximately 1–6 with median $z\approx 6.7$1, consistent with lower-$z\approx 6.7$2 hyperluminous quasars, but two objects—J0100+2802 and SDSS J0836+0054—lie near the dust-poor regime. No clear correlation with C IV blueshift is visible (Saccheo et al., 2024).

On the host-galaxy side, the experiment finds that the HYPERION quasars have molecular-gas reservoirs consistent with other quasars at the same redshift, but lower dust masses and higher gas-to-dust ratios on average. Their high star-formation rates imply high star-formation efficiencies and extremely short depletion times. In the growth-efficiency plane,

$z\approx 6.7$3

most $z\approx 6.7$4 quasars are reported to lie in a galaxy-dominated or symbiotic regime with $z\approx 6.7$5, unlike lower-$z\approx 6.7$6 luminous quasars that were BH-dominated (Tripodi et al., 2024).

6. Interpretation, caveats, and position within early-universe quasar studies

Taken together, the HYPERION results present a structured picture of early hyperluminous quasars. On one side, the UV–optical continua are statistically indistinguishable from those of lower-$z\approx 6.7$7, luminous quasars when matched with appropriate templates; on the other, the X-ray spectra indicate either a genuinely steeper coronal continuum or an unusually low high-energy cutoff. This suggests that any redshift dependence may be concentrated more strongly in the corona-disc coupling than in the UV–optical accretion-disc continuum itself (Saccheo et al., 2024).

The host-galaxy results add a second layer to that picture. The combination of $z\approx 6.7$8 molecular-gas reservoirs, star-formation rates of 200–2500 $z\approx 6.7$9, and depletion timescales as short as $\log(L_{\rm bol}/{\rm erg\,s^{-1}})\approx 47.3$0 Gyr is interpreted as rapid galaxy growth, plausibly regulated by strong outflows. The experiment’s proposed evolutionary path is an early phase of BH-dominated growth, followed by intense galaxy growth with strong feedback, and then convergence toward the massive end of the local black-hole–bulge relation (Tripodi et al., 2024).

Several caveats are explicit in the HYPERION literature. The mean X-ray-to-NIR SED is affected by non-uniform photometric coverage and possible selection bias toward dust-rich quasars. The mildly enhanced NIR hot-dust bump is therefore not presented as definitive evidence for evolution. In X-rays, the steep mean photon index has an explicit alternative interpretation in terms of $\log(L_{\rm bol}/{\rm erg\,s^{-1}})\approx 47.3$1 keV on a standard $\log(L_{\rm bol}/{\rm erg\,s^{-1}})\approx 47.3$2 power law. In ALMA imaging, the J0100+2802 study emphasizes that very high angular resolution can filter out diffuse extended emission, whereas deep medium-resolution data are better suited to recovering tidal features and outflows on 1–20 kpc scales (Tripodi et al., 2023).

Within the study of quasars in the first gigayear, HYPERION thus occupies a distinct niche: it is an experiment on the fastest-growing, most luminous SMBH engines, combining X-ray spectroscopy, broad-band SED reconstruction, and (sub)millimeter host-galaxy measurements in a uniform framework. The aggregate result is that, despite the extreme youth of their host halos, the earliest known hyperluminous quasars already exhibit disk continuum shapes, dust-reprocessing behavior, and host-galaxy gas reservoirs that closely resemble those of luminous quasars at later epochs, while their X-ray coronae and immediate feedback environment may retain stronger signatures of early-universe conditions (Zappacosta et al., 2023).