Little Red Dots host Black Hole Stars: A unified family of gas-reddened AGN revealed by JWST/NIRSpec spectroscopy

TL;DR

This study assembled 116 Little Red Dots (LRDs) from JWST/NIRSpec PRISM and NIRCam data, spanning 2.3<z<9.3, to investigate their nature. The rest-optical to near-IR continua are predominantly well-described by single-temperature modified blackbodies with typical temperatures around 5000 K, forming an HR-diagram-like Hayashi-track locus and suggesting photospheres in dense, optically thick gas envelopes around an AGN. Emission-line analysis reveals tight linear relations between Hα and the optical/blackbody continua, high Balmer decrements likely driven by gas physics (collisional excitation and resonant scattering), and a clear distinction between AGN-dominated line emission and host-galaxy [O III] emission. The authors interpret LRDs as BH*-type objects embedded in dense envelopes—consistent with black hole star/quasistar scenarios—and show that a simple host+BH* model can reproduce several observed line-continuum trends, providing empirical constraints for theoretical models and offering a public data release for community use.

Abstract

We use the DAWN JWST Archive to construct and characterise a sample of 116 little red dots (LRDs) across 2.3<z<9.3, selecting all sources with v-shaped UV-optical continua from NIRSpec/PRISM spectra and compact morphologies in NIRCam/F444W imaging. We show that LRD continuum spectra are ubiquitously well described by modified blackbodies across ~$0.4-1.0μ$m, with typical T~5000K or $λ_{peak}$~$0.65μ$m across 2 dex in luminosity, and a tail toward T~2000K. LRDs therefore trace a locus in the Hertzsprung-Russell diagram that is directly analogous to stars on the Hayashi track, strongly supporting the picture that LRDs are AGN embedded in thermalised dense gas envelopes in approximate hydrostatic equilibrium. Hotter LRDs with $λ_{peak}<0.65μ$m typically have strong Balmer breaks, redder UV slopes and high optical luminosities; other LRDs show weak or no Balmer breaks, and wide variety in $β_{UV}$ and $L_{5100}$. Crucially, we demonstrate that the UV-optical continuum shapes and luminosities are strongly linked to the $Hα,\ Hβ$, [OIII] and OI line properties. There is a tight linear relation between the H$α$ and optical continuum luminosities, as well as H$α$ and OI$_{8446}$, indicating that Balmer, OI and optical emission must primarily be powered by the same source. The Balmer decrement increases strongly toward higher $L_{Hα}$, $L_{5100}$ and Balmer break strength, providing key evidence for luminosity-dependent effects of collisional (de-)excitation and resonant scattering in the gaseous envelopes. In contrast, we show that [OIII] emission likely originates from star-forming host galaxies, and that its strong correlation with Balmer break strength arises naturally from variation in the AGN-to-host ratio. Our work presents an empirical description of the nature and structure of LRDs, defining a new benchmark for ongoing LRD model developments.

Figures (18)

• Figure 1: Top: Redshift distribution of all PRISM spectra selected from the public DJA that satisfy the spectroscopic v-shape criteria of Hviding2025 (black). The sample of v-shaped objects contains both LRDs and a large number of dust-reddened star-forming galaxies at cosmic noon. We define LRDs (116 unique sources; red) as the subset of sources satisfying strict compactness criteria in NIRCam/F444W imaging. Bottom: NIRCam/F444W magnitudes measured in circular apertures of radius $0.1\arcsec$ (half the MSA slit width) of all LRDs as a function of redshift. Of these sources, $47\%$ (55/116) also have medium or high resolution spectra from NIRSpec. Dashed lines indicate the subsample used for the modified blackbody fitting of Section \ref{['sec:MBB']}.
• Figure 2: Left: NIRSpec/PRISM spectra (normalised at $1.2\,\micron$) of four RUBIES LRDs showcasing their diverse spectral shapes. From light to dark: RUBIES-UDS-144195 Hviding2025, The CliffdeGraaff2025, RUBIES-BLAGN-1 Wang2024a, and RUBIES-UDS-175975. The colour coded vertical dashed lines indicate the peak wavelengths of the modified blackbody fits to the spectra. Right: Example of the 3-parameter modified blackbody fitting to the PRISM spectrum of The Cliff, with its error spectrum shown in blue. Regions around strong emission lines and the region blueward of $\lambda_{\rm v}$, marked in grey, were masked in the fit. Model lines (pink) represent 100 random draws from the parameter posteriors.
• Figure 3: Location of the v-shape turnover wavelength $\lambda_{\rm v}$ versus modified blackbody temperature and $\lambda_{\rm peak}$. For most LRDs $\lambda_{\rm v}$ is near the Balmer limit (there is a small systematic bias, see main text), but cooler LRDs with long peak wavelengths can have substantially larger $\lambda_{\rm v}$ than this.
• Figure 4: Distribution of the power law slopes and peak wavelengths from modified blackbody fits to the LRD sample restricted to $z<4.5$ (red). For reference, the results of fits to the complete RUBIES galaxy sample at $2<z<4.5$ and $\rm F444W<26$ are shown in grey. A slope of $\beta_{\rm MBB}=0$ corresponds to a perfect single-temperature blackbody. Whereas normal galaxies typically have broad SED shapes, as they consist of a mixture of stellar types (and hence temperatures), many LRDs are well approximated by single-temperature blackbodies. The peak wavelengths of galaxies span a broad range due to variation in recent star formation activity and dust attenuation, but LRDs have a characteristic continuum $\lambda_{\rm peak}\sim0.6-0.7\,\micron$.
• Figure 5: HR diagram of LRDs at $z<4.5$ for the modified blackbody temperature (left) and peak wavelength (middle). LRDs span a wide range in luminosity, with a lower luminosity limit that may be part physical and part driven by the imposed magnitude limit and depth of the PRISM spectra. The rest-optical continua of LRDs typically peak at a wavelength of $\sim 0.65\,\micron$, corresponding to a temperature of $\sim 5000\,$K for a perfect blackbody ($\beta=0$), although some systems are as cold as $\sim2000\,$K. The right panel shows examples of extremes in the HR diagram: a remarkably cold source ($T\sim2300\,$K, $\beta_{\rm MBB}\sim2$, $\lambda_{\rm peak}\sim0.83\,\mu$m; UNCOVER-20698), and a source that peaks in the near-IR ($\lambda_{\rm peak}\sim1.05\,\mu$m, $\beta_{\rm MBB}\sim-2$; CAPERS-COSMOS-30440).