Detailed study of a rare hyperluminous rotating disk in an Einstein ring 10 billion years ago

Wide-area infrared surveys (IRAS, WISE, Planck) have enabled the discovery of rare HyLIRGs, often strongly lensed with apparent µLIR > 1014 L⊙. High-resolution imaging (HST, ALMA, NOEMA, VLA) typically reveals HyLIRGs as gas-rich major mergers, frequently hosting AGN. The prevailing paradigm links HyLIRGs to the high-luminosity extension of local ULIRGs, where extreme starbursts are merger-driven. However, hydrodynamical simulations also predict an alternative: very young, massive, gas-rich disks at z≈2–3 can reach SFRs ≥1,000 M⊙ yr−1 via secular processes, with cold gas inspiralling and fueling intense, centrally concentrated star formation without recent major mergers. Observational tests of this secular scenario have been limited by the rarity of HyLIRGs with high-quality, multi-tracer, high-resolution data. To address this, the study targets PJ0116-24—the brightest southern Einstein-ring HyLIRG from PASSAGES at z = 2.125—for a comprehensive analysis of dust attenuation, metallicity and multi-phase gas kinematics via ERIS (Hα, Hβ, [N II], [S II]) and ALMA CO(3–2), exploiting strong lensing to probe 100–300 pc scales.

Previous work has established that most HyLIRGs are merger-driven starbursts, with strong evidence from high-resolution optical, near-IR, submm and radio imaging and spectroscopy revealing disturbed morphologies and kinematics, and frequent AGN activity. The merger-driven ULIRG–quasar evolutionary sequence is a widely accepted framework. In contrast, simulations of massive, turbulent, gas-rich high-z disks predict that secular evolution can also drive maximal SFRs via rapid inward transport of cold gas, producing ring- and arm-like structures, clump formation, and compact bulge growth without recent major mergers. Only a handful of lensed HyLIRGs have detailed kinematic constraints; notably, the lensed ring 9i09 (PJ020941.3+001559, z = 2.55) showed circular rotation with vmax/σ0,mol ≈ 4.9 ± 0.7 at ∼360 pc scales. Global dust obscuration diagnostics (e.g., Balmer decrement) in HyLIRGs are scarce; extreme Hα/Hβ values have been reported in some local ULIRGs and in a few high-z sources (e.g., G165 Arc 1a), but comprehensive, spatially integrated rest-frame optical spectroscopy of heavily obscured HyLIRGs is challenging. This study leverages strong lensing and multi-tracer datasets to fill this gap.

Target selection and facilities: PJ0116-24 (PJ011646.8-243702) was selected as the brightest Einstein-ring HyLIRG from the PASSAGES survey (z = 2.125). Multi-tracer observations include ALMA CO(3–2) and dust continuum imaging and VLT/ERIS near-IR IFU spectroscopy; VLT/MUSE IFU data constrain the foreground lens redshift; HST WFC3 F160W and Gemini r/z imaging provide high-resolution stellar light maps.

Spectroscopy (VLT/ERIS): ERIS Science Verification (ID 110.258 S) observed H-band (1.46–1.67 µm; covers Hβ, [O III] λλ4959,5007) and K-band (1.93–2.22 µm; covers Hα, [N II] λλ6548,6584, [S II] λλ6717,6731) with R≈11,200, 250 mas pixel scale, seeing-limited. Eight object and four sky exposures per band (DIT 300 s) with dithers; data reduced using ERIS pipeline (flat-fielding, wavelength calibration, OH subtraction, sky subtraction, DAR correction). Flux calibration from telluric stars; individual cubes combined with sigma clipping and weighted averaging; final PSF FWHM ≈0.7″ (K) and 1.2″ (H). Integrated spectra extracted over the Einstein ring; simultaneous line+continuum fitting (astropy LevMarLSQFitter). Stellar Balmer absorption corrections based on BC03 SED modeling.

Interferometry (ALMA): Multiple programs combined (2017.1.01214.S, 2019.1.01197.S, 2021.1.00353.S). Visibilities regridded (30 km s−1 channels), continuum-subtracted (uvcontsub). Imaging with tclean using multiscale and hogbom down to 1σ; natural weighting gives synthesized beam ≈0.19″×0.16″; Briggs cubes also produced. JvM residual correction applied to correct clean residual flux scaling. CO(3–2) cube and dust continuum used for lens modeling and kinematics; maximum recoverable scale 7.5″ adequate for the extended ring.

Lens redshift and imaging: VLT/MUSE (5–6 Oct 2022) confirms the foreground lens as a massive ETG at z = 0.554. HST WFC3 F160W (GO-14653) and Gemini r/z imaging provide stellar distribution constraints.

Lens modeling and delensing: HST astrometry corrected to Gaia DR3. Foreground lens subtracted with GALFIT (Sérsic models). Emission ‘knots’ identified in HST and ALMA CO channel maps to constrain the lens. Lens model built with glafic including a Navarro–Frenk–White halo, singular isothermal ellipsoid for the lens galaxy, and external shear; minimum χ2 optimization followed by MCMC to assess parameter uncertainties. Source-plane reconstructions derived through mesh-grid-based delensing with linear interpolation. Spatially varying ‘delensed beam’ characterized; best source-plane resolution reaches ≤100 pc near critical curves.

Kinematic modeling: State-of-the-art 3D forward modeling with DysmalPy+Lensing directly in the image plane, accounting for lensing, PSF/beam smearing, instrumental line spread, and projection effects. Source-plane mass distribution parameterized as geometrically thick Sérsic disk + bulge plus NFW dark halo. Free parameters include Mdisk,dyn, B/T, Rbulge/Rdisk, σ0 (intrinsic dispersion), MDM,dyn, inclination i, and PA, with Gaussian priors on log Mdisk,dyn, sin i, and PA. MCMC sampling (250 walkers, 300 steps post burn-in). Models compared to observed velocity and dispersion fields and P–V diagrams for CO and Hα.

Line diagnostics and SED fitting: Global Balmer decrement Hα/Hβ measured (after stellar absorption correction) to infer nebular attenuation (Calzetti RV=3.1). Strong-line metallicity diagnostics (Curti et al. calibrations) from Hα, Hβ, [N II], [S II], [O III] ratios. Ionization source assessed via BPT diagrams. Multi-wavelength SED fitted with MiChi2 including BC03 stellar component (constant SFH, solar metallicity, Chabrier IMF), empirical AGN mid-IR template, Draine & Li dust model, and synchrotron radio; Monte Carlo sampling yields apparent and intrinsic properties with lensing corrections.

• Strong lensing enables sub-kpc (∼100–300 pc) source-plane resolution along parts of the Einstein ring with total magnifications µ ≈ 16 (HST) and µ ≈ 17 (CO). • Extreme dust attenuation: global Balmer decrement Hα/Hβ = 8.73 ± 1.14 (corrected for stellar absorption), implying A_V,neb = 2.95 ± 0.13 mag and E(B−V)_neb,line = 0.95 ± 0.13. • Star formation: intrinsic SFRUV+IR = 1,490 ± 400 M⊙ yr−1; SFRHα,corr = 470 ± 60 M⊙ yr−1 (≈31% ± 9% of total), while apparent uncorrected SFRHα,not corr = 58 ± 2 M⊙ yr−1; SFRCO = 470 ± 60 M⊙ yr−1. The unobscured fraction is small, indicating heavy dust obscuration. • Stellar and dynamical masses (intrinsic): log(M*) = 11.16 ± 0.35; log(Mdyn) ≈ 11.26 ± 0.32. From kinematic modeling, log(Mdisk,bulge,dyn) = 11.29 (+0.02/−0.07), B/T = 0.07 (+0.09/−0.01), Reff = 3.62 (+0.02/−0.03) kpc, inclination i = 53.1 (+8.5/−8.2) deg, PA = 80.4 ± 0.8 deg. • Rotational support: Vrot,mol(R_eff) = 401 (+15/−12) km s−1, σmol = 42.8 (+3.8/−2.8) km s−1, yielding vrot/σ0,mol ≈ 9.4 ± 0.9. For ionized gas, Vrot,ion(R_eff) = 389 ± 31 km s−1, σion = 68.6 ± 7.7 km s−1, vrot/σ0,ion ≈ 5.7 ± 0.6. Velocity fields and P–V diagrams of CO and Hα are consistent and indicative of ordered rotation rather than a major merger. • Gas stability and substructure: Toomre Q ≈ 0.3 within ∼1 Reff, well below Qcrit ≈ 0.67, explaining fragmentation into rich (sub)kpc substructures. Cold gas exhibits central deficit (∼<0.5 kpc) and ring-/arm-like enhancements at 1–2 kpc, extending to ∼3–4 kpc. • Giant clump: A ∼kpc-scale clump ∼4 kpc from the centre is detected in HST, Hα, and CO. Its CO(3–2) emission accounts for ∼1.6% of the total; estimated stellar mass fraction ≤1–3% of the galaxy. The clump co-rotates with the disk and has similar dispersion, consistent with in-situ formation via disk instability. • Metallicity and ionization: Strong-line diagnostics indicate solar to ∼0.2 dex supersolar metallicity. BPT analysis places the source in the star-forming/composite regime, with star formation dominating and only low AGN contribution. • Secular evolution: The combination of strong rotational support, ordered kinematics, central gas depletion with high Σgas and ΣSFR at 1–2 kpc, and high metallicity supports a scenario of secular evolution (wet compaction and inside-out quenching) in a massive, gas-rich disk achieving maximal SFRs without recent major mergers.

The coherent, rapidly rotating disk kinematics (high vrot/σ0 for both molecular and ionized gas), centrally peaked apparent dispersion, and lack of disturbed morphology or massive companions argue against a recent major merger, differentiating PJ0116-24 from the majority of HyLIRGs. Instead, the observed gas ring/arms, low Toomre Q, and rich substructure align with a turbulent, gas-rich disk undergoing secular evolution and violent disk instability. The central deficit of cold gas and elevated Σgas and ΣSFR at 1–2 kpc fit the ‘wet compaction’ framework, wherein gas inflow builds a compact central mass concentration followed by inside-out quenching. The extreme dust attenuation (Hα/Hβ ≈ 8.7) and high SFRUV+IR highlight that most star formation is obscured, with lensing enabling detection of less obscured regions. The metallicity estimates (solar to supersolar) are consistent with an evolved, enriched system and with high central stellar surface densities that presage bulge growth and eventual quenching. Compared with the only other lensed HyLIRG with confirmed rotation (9i09), PJ0116-24 is even more rotationally supported and exhibits more complex substructure, strengthening the case that secular evolution can reach maximal SFRs predicted by simulations. This study thus provides critical observational evidence complementing the merger-driven paradigm, demonstrating that secular processes can power the most extreme starbursts at cosmic noon.

PJ0116-24 is a rare, hyperluminous, strongly lensed rotating disk at z = 2.125 with intrinsic SFR ≈ 1,490 M⊙ yr−1 and baryonic mass ≈ 1011.3 M⊙. Multi-tracer, high-resolution observations and lensing-aided kinematic modeling reveal ordered rotation with high vrot/σ0, rich (sub)kpc gaseous substructure, solar–supersolar metallicity, and heavy dust attenuation with an extreme Balmer decrement. The system shows central gas depletion and elevated Σgas and ΣSFR at 1–2 kpc, consistent with a secular ‘wet compaction’ phase leading to inside-out quenching and bulge growth, without evidence for a recent major merger. These results demonstrate that maximal star formation can occur in secularly evolving, gas-rich disks, providing a decisive complement to the established merger-driven view of HyLIRGs. Future work should expand the sample of HyLIRGs with comparable lensing-enabled, multi-tracer IFU and interferometric datasets to test the generality of these findings, refine metallicity calibrations at high obscuration, and pursue higher angular resolution AO IFU observations to search for resolved inflow/outflow signatures and noncircular motions.

• Spatial resolution varies across the source-plane reconstructions due to lensing; delensed Hα features near critical curves can be affected by beam-smearing, and some blobs may be delensing artefacts. • Kinematic modeling assumes idealized disk+bulge Sérsic profiles and circular motions; noncircular motions are not explicitly modeled and could contribute to residuals, especially near the centre. • ISM mass modeling relies on a simplified monotonic temperature–density relation; alternative multi-component models change inferred masses and introduce degeneracies among parameters. • SED fitting and mass estimates are subject to assumptions about the IMF, AGN contributions, and star formation histories; combined systematic uncertainties may reach ≲0.5 dex. • Some emission lines have modest S/N; metallicity estimates (especially supersolar indications) depend on calibration choice and require confirmation in a larger HyLIRG sample. • Global attenuation estimates from the Balmer decrement likely represent lower limits to total obscuration since the most obscured regions may be undetected in Hα.