A remarkable Ruby: Absorption in dense gas, rather than evolved stars, drives the extreme Balmer break of a Little Red Dot at $z=3.5$

TL;DR

This paper presents The Cliff, a luminous z≈3.55 LRD with an exceptionally strong Balmer break observed with JWST, and critically tests whether the rest-optical emission is stellar or AGN-driven. Through JWST/NIRSpec spectroscopy (PRISM and G395M), NIRCam/MIRI imaging, and X-ray data, the authors demonstrate that massive, ultra-dense stellar populations cannot reproduce the spectrum, even with extreme dust laws or IMF variations. They show that conventional galaxy+AGN decompositions fail to explain the Balmer-break magnitude, and argue that a black hole star (BH*) scenario—an AGN embedded in dense gas—best matches the rest-optical to near-IR continuum, albeit with ongoing modelling challenges in the infrared. The Cliff thus provides the strongest case to date that some LRDs are powered by central ionising sources in absorbing gas rather than by stars alone, informing our understanding of high-z compact red galaxies and BH* models.

Abstract

The origin of the rest-optical emission of compact, red, high-redshift sources known as `little red dots' (LRDs) poses a major puzzle. If interpreted as starlight, it would imply that LRDs would constitute the densest stellar systems in the Universe. However, alternative models suggest active galactic nuclei (AGN) may instead power the rest-optical continuum. Here, we present JWST/NIRSpec, NIRCam and MIRI observations from the RUBIES and PRIMER programs of The Cliff: a bright LRD at $z=3.55$ with an exceptional Balmer break, twice as strong as that of any high-redshift source previously observed. The spectra also reveal broad Hydrogen (H$α \rm FWHM\sim1500$km/s) and He I emission, but no significant metal lines. We demonstrate that massive evolved stellar populations cannot explain the observed spectrum, even when considering unusually steep and strong dust attenuation, or reasonable variations in the initial mass function. Moreover, the formally best-fit stellar mass and compact size ($M_*\sim10^{10.5}\,M_\odot,\ r_{e}\sim40\,$pc) would imply densities at which near-monthly stellar collisions might lead to significant X-ray emission. We argue that the Balmer break, emission lines, and H$α$ absorption line are instead most plausibly explained by a `black hole star' (BH*) scenario, in which dense gas surrounds a powerful ionising source. In contrast to recently proposed BH* models of dust-reddened AGN, we show that spectral fits in the rest UV to near-infrared favour an intrinsically redder continuum over strong dust reddening. This may point to a super-Eddington accreting massive black hole or, possibly, the presence of (super)massive stars in a nuclear star cluster. The Cliff is the clearest evidence to date that at least some LRDs are not ultra-dense, massive galaxies, and are instead powered by a central ionising source embedded in dense, absorbing gas.

Figures (13)

• Figure 1: NIRSpec/PRISM spectrum of The Cliff (RUBIES-UDS-154183) at $z_{\rm spec}=3.548$. Orange points show NIRCam and MIRI photometry spanning an observed wavelength range of $0.9-18\,\rm \mu m$; the inset colour image was constructed using the NIRCam F115W, F277W and F444W filters and shows the location of the NIRSpec microshutters. Coloured lines are the NIRSpec/PRISM spectra (rescaled using the median flux at rest-frame $[3200-3700]\,\AA$) of four sources with strong Balmer breaks: two of the most luminous LRDs in the literature Labbe2024Wang2024a and their MIRI detections Setton2025, the triple-imaged LRD at $z_{\rm spec}=7.04$ of Furtak2024, and a massive post-starburst galaxy at $z_{\rm spec}=4.62$ (a medium-resolution NIRSpec spectrum of which was presented by Carnall2024). Dotted lines indicate the locations of strong emission line features in the different spectra as well as the Balmer limit. The Cliff shows an exceptionally strong Balmer break, a declining SED in the rest near-infrared ($\sim1-4\,\rm \mu m$), strong H and He i emission lines, but (in comparison to the two luminous LRDs) very weak metal lines.
• Figure 2: Balmer break strength, measured as the ratio of the mean flux density in the rest-frame wavelength ranges $[3620,3720]\,\AA$ and $[4000,4100]\,\AA$ from public JWST/NIRSpec data. Grey squares show the median values of stacks of star-forming galaxies from RobertsBorsani2024; red markers represent a compilation of LRDs Furtak2024Labbe2024Setton2024Wang2024b; blue markers show a large sample of massive quiescent galaxies Barrufet2025Carnall2024deGraaff2024cGlazebrook2024Nanayakkara2024Weibel2024b. The recently discovered LRD of Naidu2025, MoM-BH*-1, is a higher-redshift analogue of The Cliff and discussed in Section \ref{['sec:cloudy']}.
• Figure 3: Kinematic emission line decomposition from simultaneous fitting to PRISM (black) and G395M (grey) spectra. Top: The H$\alpha$ emission and absorption complex is well-described by a broad Lorentzian profile (blue; $\rm FWHM\sim1400\,\,\rm km\,s^{-1}$), with weaker narrow emission (cyan) as well as redshifted absorption (purple). The total G395M model and residuals are shown in red. Bottom: Zoom-ins of other strong H and He features, also revealing a non-detection of the [O iii] doublet (see Table \ref{['tab:elines']}). Although (where possible) both dispersers were used in the fitting, in the bottom panels we show only the PRISM model components and total model (orange).
• Figure 4: Left: NIRCam/F200W image ($0.02\,\arcsec\,{\rm pix}^{-1}$) of The Cliff, the nearby foreground source at a lower redshift of $z_{\rm phot}\approx2.8$, and a star of approximately equal magnitude (at a distance of $\sim1\arcsec$). All sources are modelled as single Sérsic profiles and fit simultaneously to account for blending. Right: The posterior distribution of the effective radius of the star is clustered at the edge of the prior, as may be expected for an unresolved point source, while that of the LRD converges to a small value of $r_{\rm e,maj}\approx0.26\,\rm pix\approx 40\,$pc.
• Figure 5: Best-fit SED models and residuals (with respect to the PRISM spectrum) for four model variations, constructed as different mixtures of (dust-reddened) stellar population and power-law AGN model components. These AGN models do not consider the possible reddening by absorbing dense gas, which we explore in Section \ref{['sec:cloudy']}. Right-hand panels show zoom-ins of the region around the Balmer break. From top to bottom: the fiducial 'galaxy-only' model from Prospector; a galaxy + AGN model that maximises the stellar contribution, fit with Prospector (the 'maximal $M_*$' model of Wang2024b); the galaxy + AGN model following Labbe2024, but fitting only Hydrogen emission lines instead of a forest of metal lines; the galaxy + AGN model of Labbe2024, but with an even steeper dust law (see Section \ref{['sec:dust']}). All four models favour a massive post-starburst solution, with a very steep dust attenuation law and high optical depth. However, even with such extreme dust, none of these models can produce the strong Balmer break and shape of the rest-frame optical SED, as is evident from the systematic (and significant) features in the residuals blue- and redward of the Balmer break.