- The paper maps Lyα escape in CRISTAL-02 by combining MUSE Lyα, JWST/NIRSpec Balmer lines, and NIRCam UV imaging at matched 4.5-kpc resolution.
- Lyα extends about 33 kpc along the [C II]-traced outflow, compared with roughly 14 kpc for Hα and UV emission, indicating resonant scattering through outflowing neutral gas.
- Effective Lyα escape reaches only 2–12% centrally but rises to approximately 27–87% near Clump A, showing that dust, H I column density, and gas geometry matter more than ionizing efficiency alone.
Context and motivation
Lyα emission is a resonant recombination line whose visibility in high-redshift galaxies is governed by H\,i radiative transfer, dust attenuation, and gas geometry, making it a powerful but difficult-to-interpret diagnostic of the interstellar and circumgalactic medium. The study presented here (2608.19439) targets CRISTAL-02 (also known as DC-848185, LBG-1, or HZ6), a main-sequence star-forming galaxy at z[CII]=5.293 located in the AzTEC-3 protocluster overdensity in COSMOS. Previous MUSE tomography revealed extended, diffuse Lyα emission connected to interactions within the protocluster core, while ALMA observations identified a prominent [C\,ii] outflow with a biconical geometry. Two compact regions—Clump A and Clump B—had been flagged as potential AGN candidates on the basis of broad Hα components detected at native NIRSpec resolution.
The central question addressed is how star formation, feedback-driven outflows, dust, and possible AGN activity jointly regulate the spatial distribution and escape of Lyα photons in a system where all of these ingredients coexist.
Data and methodology
The analysis combines three datasets, all matched to the VLT/MUSE spatial resolution (PSM FWHM ≈0.74′′, corresponding to ∼4.5 kpc):
- VLT/MUSE IFU: an inverse-variance-weighted narrowband Lyα map constructed over 7655–7670 Å, redward of systemic, capturing the dominant redshifted peak; the map reaches a 1σ sensitivity of ∼0.18×10−20 erg sz[CII]=5.2930 cmz[CII]=5.2931.
- JWST/NIRSpec G395M IFU (Cycle 2 GO Program 3045): spaxel-by-spaxel single-Gaussian fits to Hz[CII]=5.2932 (with tied [N\,ii]) and Hz[CII]=5.2933. The JWST cube was convolved to the MUSE PSF using a Gaussian kernel derived in Fourier space.
- JWST/NIRCam F115W imaging (COSMOS-Web): used as a proxy for rest-frame UV (z[CII]=5.2934 Å) continuum. A combined HST+NIRCam UV-slope (z[CII]=5.2935) map was attempted but rejected as unreliable owing to low HST S/N—a methodological concession that limits constraints on dust via UV slopes.
The authors emphasize that the MUSE-resolution data cannot robustly decompose broad Hz[CII]=5.2936 components reported at native NIRSpec resolution, so single-Gaussian fits are adopted throughout for total line fluxes.
Morphology: Lyz[CII]=5.2937 decoupled from non-resonant tracers
The resolved maps reveal a clear morphological dichotomy. Hz[CII]=5.2938 and UV emission trace the galactic disk (PA z[CII]=5.2939), extending α0 and α1 kpc toward the northwest respectively, while Lyα2 is substantially more extended, reaching α3 kpc toward the northeast along the direction of the [C\,ii] outflow (PA α4, bicone half-opening angle 15–30°). A faint southern Lyα5 extension at α6 from the center may trace a weaker counter-component of the outflow.
This alignment implies that a significant fraction of Lyα7 photons do not emerge from their production sites directly but are resonantly scattered or reprocessed by outflowing neutral gas, so that the extended Lyα8 morphology traces circumgalactic H\,i rather than in-situ star formation. Notably, the peaks of Hα9 and UV coincide spatially with the dust continuum maximum near Clump B; under standard conditions this would produce strong attenuation, yet both tracers remain bright there, which the authors attribute to multiphase mixing and outflow-carved low-opacity channels.
Slit-extracted velocity profiles reinforce this picture: Lyα0 linewidths exceed those of Hα1 in all three slit orientations, and the Lyα2 peaks are redshifted relative to systemic—the classic signature of back-scattering off receding far-side gas in an outflowing H\,i envelope.
Clump-scale contrasts in line ratios and extinction
Clumps A and B exhibit contrasting properties across every diagnostic:
| Quantity |
Clump A |
Clump B |
| Lyα3/Hα4 |
α5 |
α6 |
| Lyα7/UV |
α8 |
α9 |
| α0 (Balmer decrement, Calzetti law) |
α1 |
α2 |
| α3 |
α4 |
α5 |
Both Lyα6/Hα7 values fall far below the Case B intrinsic ratio of α8, indicating severe effective Lyα9 suppression even in the favored region. The contrast between the clumps is consistent with the dust continuum peak coinciding with Clump B. A check against the native-resolution H≈0.74′′0/H≈0.74′′1 map confirms that the smooth gradient between the clumps is primarily physical rather than an artifact of PSF convolution, although the convolution does smooth intrinsic variations.
Radial profiles along the disk PA show that Ly≈0.74′′2 surface brightness dips precisely where H≈0.74′′3 and UV peak (at Clump B) and peaks sharply at Clump A, where H≈0.74′′4 and UV decline—an offset indicating that Ly≈0.74′′5 there arises from scattering of photons produced elsewhere rather than local recombination.
Spatially resolved escape fraction and ionizing efficiency
Using the dust-corrected H≈0.74′′6 flux and the canonical intrinsic Ly≈0.74′′7/H≈0.74′′8 ratio, the authors derive spaxel-by-spaxel effective Ly≈0.74′′9 escape fractions. These include IGM transmission at ∼4.50 and should therefore be read as effective, not purely ISM, escape fractions. Across the disk strip, ∼4.51 reaches only ∼4.52–12% in the central regions but rises sharply toward Clump A, where zELDA II modeling yields a global escape fraction of ∼4.53–87% (with large uncertainties). This spatial variation demonstrates that local ISM conditions—column density and dust, not just ionizing output—dominate the observed Ly∼4.54 morphology.
For the ionizing photon production efficiency, assuming Case B recombination and ∼4.55 (so that any escaping LyC photons would make the estimate a lower limit), the galaxy-wide value is ∼4.56, within the typical range for ∼4.57–5 star-forming galaxies. Clump B shows a marginally higher ∼4.58 than Clump A (∼4.59 dex, consistent within uncertainties)—yet it is precisely this region that suppresses Lyα0. The juxtaposition constitutes one of the paper's clearest conclusions: high local ionizing photon production alone does not guarantee Lyα1 visibility; H\,i column density, geometry, and kinematics govern escape.
zELDA II fits to the slit spectra yield a low-velocity outflow (α2–190 km sα3) with a relatively high characteristic H\,i column density (α4 cmα5) along the outflow-direction slit, together with negligible dust optical depth in the Lyα6-emitting envelope. For the integrated disk-direction spectrum, the model gives α7 and IGM transmission α8, broadly consistent with the Hα9-based measurement given uncertainties.
Limitations and open questions
Several caveats bear directly on the interpretation:
- AGN status remains unresolved. Broad H1σ0 components (FWHM 1σ1 km s1σ2 for Clump A, 955 km s1σ3 for Clump B) place the candidates in composite regions of diagnostic diagrams without definitive classification, and auroral [O\,iii]1σ4 searches yielded only 1σ5 upper limits. Independent AGN diagnostics were impossible here because the MUSE-matched data preclude broad/narrow decomposition. If AGN contributions are significant, both 1σ6 and 1σ7 become biased effective quantities, since they assume stellar photoionization and Case B recombination.
- Feedback driver uncertain. Companion work reports outflow velocities of 550–740 km s1σ8, a mass-loading factor 1σ9, and outflow kinetic power (∼0.18×10−200 erg s∼0.18×10−201) comparable to the supernova energy injection rate (∼0.18×10−202 erg s∼0.18×10−203), favoring stellar feedback—but an AGN contribution cannot be fully excluded.
- Blending at matched resolution. At MUSE resolution, Clump B likely blends multiple physically distinct components, including the AGN candidate region and the outflow footpoint, complicating aperture-based measurements.
- Low S/N in key regions. Clump B's Ly∼0.18×10−204 profile lacks sufficient S/N for zELDA II modeling, and radial profile uncertainties prevent definitive conclusions about surface-brightness behavior.
The paper explicitly leaves open whether AGN feedback plays any role, identifying deeper spatially resolved multiwavelength IFU observations combining Ly∼0.18×10−205, H∼0.18×10−206, and [C\,ii], detailed forward-modeling of Ly∼0.18×10−207 profiles and spatial distributions, and additional non-resonant tracers such as [O\,iii]∼0.18×10−208 as requirements for resolving it.
Conclusion
By matching MUSE Ly∼0.18×10−209, JWST/NIRSpec Hz[CII]=5.29300/Hz[CII]=5.29301, and NIRCam UV data to a common spatial resolution, this work provides some of the first spatially resolved measurements of z[CII]=5.29302 and z[CII]=5.29303 in a z[CII]=5.29304 galaxy. The results show that Lyz[CII]=5.29305 escape in CRISTAL-02 is strongly anisotropic, preferentially channelled along the [C\,ii]-traced outflow, and modulated by local dust column density rather than by ionizing photon production alone. The observational evidence supports feedback-driven gas clearing as the dominant regulator of Lyz[CII]=5.29306 visibility, while conceding that the available data cannot uniquely distinguish stellar from AGN-driven feedback. The study establishes a template for resolved Lyz[CII]=5.29307–Hz[CII]=5.29308 comparisons that will require larger samples and higher-S/N spectroscopy to determine whether these mechanisms operate generally in high-redshift galaxies.
