Little Red Dot $-$ Host Galaxy $=$ Black Hole Star: A Gas-Enshrouded Heart at the Center of Every Little Red Dot

TL;DR

This work tests the BH* hypothesis for Little Red Dots (LRDs) by developing an empirical host–BH* decomposition using host galaxies drawn from the DAWN JWST Archive. By subtracting host templates matched in redshift and [OIII] luminosity, the authors recover a BH*-like spectrum with a strong Balmer break and a blackbody-like optical continuum, implying a dense gas cocoon surrounding accreting black holes. The results show LRD hosts are faint dwarf galaxies undergoing recent starbursts, while the BH* component increasingly dominates the SED from the rest-UV to the rest-optical, with a characteristic transition around 3600–4000 Å and a BH* fraction near 1 μm of >90%. They derive a BH* duty cycle of ~1% and lifetimes of ~10 Myr, suggesting BH*s are a common, transient phase in massive black hole growth, potentially linked to NSCs or supermassive star formation, and implying a broader prevalence of BH*s than inferred from LRDs alone.

Abstract

The central engines of Little Red Dots (LRDs) may be ``black hole stars" (BH*s), early stages of black hole growth characterized by dense gas envelopes. So far, the most direct evidence for BH*s comes from a handful of sources where the host galaxy is completely outshone as suggested by their remarkably steep Balmer breaks. Here we present a novel scheme to disentangle BH*s from their host galaxies assuming that the [OIII]5008Å line arises exclusively from the host. Using a sample of 98 LRDs ($z$~$2-9$) with high quality NIRSpec/PRISM spectra, we demonstrate that the host-subtracted median stack displays a Balmer break $>2\times$ stronger than massive quiescent galaxies, with the rest-optical continuum resembling a blackbody-like SED ($T_{\rm{eff}}$~$4050$ K, $\log(L_{\rm{bol}})$~$43.9$ erg s$^{-1}$, $R_{\rm{eff}}$~$1300$ au). We measure a steep Balmer decrement (H$α$/H$β>10$) and numerous density-sensitive features (e.g., FeII, HeI, OI). These are hallmark signatures of dense gas envelopes, providing population-level evidence that BH*s indeed power LRDs. In the median LRD, BH*s account for $\sim20\%$ of the UV emission, $\sim50\%$ at the Balmer break, and $\sim90\%$ at wavelengths longer than H$α$ with the remainder arising from the host. BH*s preferentially reside in low-mass galaxies ($M_{\rm{\star}}$~$10^{8}\,{\rm M}_{\rm{\odot}}$) undergoing recent starbursts, as evidenced by extreme emission line EWs (e.g., [OIII]5008Å~$1100$Å, CIII]~$12$Å), thereby favoring BH* origins linked to star-formation. We show V-shaped LRD selections are biased to high BH*/host fractions ($\gtrsim60\%$ at 5500Å) -- less dominant BH*s may be powering JWST's blue broad-line AGN. We find BH*s are so commonplace and transient (duty cycle $\sim1\%$, lifetime $\sim10$ Myrs) that every massive black hole may have once shone as a BH*.

Figures (32)

• Figure 1: Distributions of redshift (left) and [O$\;$] luminosity (right) of the LRD sample and the candidate host galaxy sample. Our empirical host subtraction method relies on matching LRDs with peer galaxies that lie at similar redshift and that display similar [O$\;$] luminosity (§\ref{['sec:methods']}). The vastness of the DAWN JWST archive ensures that NIRSpec/PRISM spectra of peer galaxies are available for every LRD in our sample.
• Figure 2: Illustration of our empirical LRD decomposition strategy. Every LRD (red; left) is matched with candidate host galaxies (navy; center) at similar redshift ($|\Delta z|<1.5$) and of comparable [O$\;$] luminosity ($\pm0.5$ dex) from the DAWN JWST Archive (DJA). By subtracting each candidate host from the LRD after scaling to match the [O$\;$] luminosity we derive a set of possible $\text{LRD}-\text{Host}$ SEDs (black; right). We show five candidate hosts here for simplicity, whereas in practice each LRD is matched to a median of 78 possible hosts -- e.g., the LRD shown above has 95 matches. The $\text{LRD}-\text{Host}$ SEDs display over/under-subtraction around the [O$\;$] wavelength due to subtle differences in the [O$\;$] line width, but note that every $\text{LRD}-\text{Host}$ spectrum has zero [O$\;$] luminosity by construction.
• Figure 3: Examples of LRDs spanning the full range of ($\text{LRD}- \text{Host}$)/$\text{LRD}$ ratios inferred for our sample. Median stacks for the inferred hosts (center) and $\text{LRD}-\text{Host}$ spectra (i.e., the central engine; right) are shown, sorted by the ($\text{LRD} - \text{Host}$ )/LRD ratio at 5500Å. Errors on the LRD are observed uncertainties, while on the Host and $\text{LRD}-\text{Host}$ spectra are posteriors based on all the possible host candidates. While the central engine produces almost all the optical light in the LRD in the top row (The Cliff from degraaff25), the host galaxy is dominant in the LRD in the bottom row (MoM-149501). Note the diversity in the $\text{LRD}-\text{Host}$ SEDs -- they are not carbon copies of the Cliff (top row), with the peak wavelength of the optical SED occurring at comparable or redder wavelengths. In our framework, this diversity of central engines (see Appendix \ref{['fig:appendix_gallery']} for decomposition of the full sample), along with the diversity of hosts and their relative fractions accounts for the diversity in LRD SED shapes.
• Figure 4: Mock test results to validate our host/BH* decomposition strategy.Top: We produce a sample of synthetic LRDs by combining spectra of observed galaxies (top-center) and CLOUDY models (top-right). Every LRD in our sample has a corresponding synthetic LRD with matched BH*/LRD fraction and noise properties (see §\ref{['sec:mocks']} for details). The median stack of the synthetic LRDs is shown in the top-left (red). While we show median stacks here (top-row) as the injected input we are attempting to reconstruct, our decomposition procedure is run on individual synthetic LRDs (exactly as in Fig. \ref{['fig:method']}). Center, Bottom: The reconstructed host and BH* stacks are generally in excellent agreement with the input stacks as evidenced by the residuals that agree within $1\sigma$ (bottom panels). The uncertainty in the recovered BH* stack is most pronounced in the rest-UV below $\lesssim2000$Å where spurious emission features appear as the subtracted hosts are not exactly the same as the true hosts -- in what follows, we do not interpret this portion of the reconstructed BH* spectrum.
• Figure 5: Stacks of $\text{LRD} - \text{Host}$ difference spectra (top) and host galaxies (bottom-right) constructed by applying our decomposition procedure to each of the 98 LRDs in our sample (stacked in the bottom-left).Top: The central engine of the median LRD shows a steep Balmer break ($f^{\nu}(4050\rm{\AA})$/$f^{\nu}(3670\rm{\AA})\approx7$), extremely strong H$\alpha$ emission ($\approx850$Å), a steep Balmer decrement (H$\alpha$/H$\beta\approx11$), and strong O$\;$ and Fe emission. These are hallmark signatures of radiative transfer in a dense gas envelope, strongly supporting the picture that the central engine of the typical LRD is indeed a BH*. Note that the [O$\;$] line by construction has zero luminosity, but appears over/under-subtracted because the kinematic width of the line is never exactly the same across the LRDs and the candidate hosts. Bottom-Left: In LRDs, these hallmark BH* features are mixed with galaxy light and are challenging to discern (e.g., there is only a weak Balmer break), underscoring the need for our procedure. Bottom-Right: The median host galaxy is highly star-forming, as evidenced by strong rest-optical emission lines (EW(O$\;$+H$\beta$)$\approx1700$Å) that are $\approx2-3\times$ stronger than the typical galaxy at these redshifts Endsley24. The detection of strong C$\;$] ($\approx12$Å, $>2\times$ the typical EW at $z\approx5$) points to the hosts having undergone a recent burst Roberts-Borsani25.