Dust Budget Crisis in Little Red Dots

TL;DR

LRDs had been interpreted as heavily dust-obscured AGNs at high redshift, risking a dust-budget crisis if their hosts are extremely low-mass or absent. The authors develop an energy-balance UV-to-IR modeling framework across multiple extinction laws and dust distributions, applying it to four LRDs (two individual, two stacks) with JWST/MIRI, Herschel, and ALMA data to constrain $A_V$. They find modest extinction ($A_V\lesssim 1$ mag, up to ~2 mag with some laws), implying dust reddening is not as extreme as previously thought and yielding radiative efficiencies compatible with the So\l tan--Paczy\ński argument; this shifts the paradigm toward moderate extinction or dust-free inner regions. The study discusses dust production constraints, alternative dense-gas reddening scenarios, and future tests with PRIMA to robustly determine the dust content and nature of LRDs in the early universe.

Abstract

Little red dots (LRDs), a population of active galactic nuclei (AGNs) recently identified by JWST, are characterized by their compact morphology and red optical continuum emission, which is often interpreted as evidence for significant dust extinction of $A_V \gtrsim 3$ mag. However, the dust-reddened AGN scenario is increasingly challenged by their faint near-to-far infrared emission and a potential "dust budget crisis" in cases when the host galaxy is either undetectably low-mass or absent. In this study, we re-evaluate the dust extinction level in LRDs by modeling the UV-to-infrared spectra for various extinction laws and a broad range of dusty distribution parameters. Comparing the predicted infrared fluxes with observational data from the JWST MIRI, Herschel, and ALMA, our analysis finds that the visual extinction is tightly and consistently constrained to $A_V \lesssim 1.0-1.5$ mag for A2744-45924, RUBIES-BLAGN-1, and stacked SEDs from a large sample of LRDs under the SMC extinction law, with slightly weaker constraints for those with gray extinction in the UV range. The revised $A_V$ values yield radiative efficiencies of $\sim 10\%$ for the LRD population, easing the tension with the Sołtan argument for the bulk AGN population at lower redshifts. Moreover, this moderate extinction (or dust-free) scenario, with reprocessed emission spectra testable by future far-infrared observatories, provides a paradigm shift in understanding their natures, environments, and evolutionary pathways of massive black holes in the early universe.

Figures (7)

• Figure 1: Extinction curves as a function of wavelength used in this work, covering from UV to infrared wavelengths: the SMC, the interstellar dust in the Milky Way with $R_V = 3.1$WeingartnerDrain_2001_extinction, and the Orion Nebula Baldwin1991. We also show the modified Calzetti's attenuation law proposed in Noll_2009_flexible_dust_law with $\delta = -1.8$ (see main text). The extinction $A_{\lambda}$ is normalized at the V band 5500 Å (dotted vertical line). The UV to optical part is used to compute dust attenuation of the incident radiation flux (blue shaded region), and the IR part is used to calculate the re-emission from heated dust (red shaded region).
• Figure 2: Multi-wavelength SED of the brightest LRD, A2744-45924 (orange curve) with JWST NIRCam/MIRI detection (circles) and non-detection in Herschel and ALMA bands (triangles). Left: Incident (corrected for extinction) and reprocessed IR re-emission spectra for fixed dust distribution parameters $\gamma = 0.5$ and $n_0 = 10^3~ {\rm cm}^{-3}$ with varying visual extinction $A_V$. Right: Re-emission SEDs for four different dust density configurations: $(\gamma, n_0)=(0.5,10^3~{\rm cm}^{-3})$, $(0.5,10~{\rm cm}^{-3})$, $(0,10~{\rm cm}^{-3})$, and $(2.0, 10^4~{\rm cm}^{-3})$. For each case, the maximum allowed $A_V$ is applied such that the resulting IR flux does not violate the observational constraints.
• Figure 3: Parameter survey showing the upper limit on $A_V$ as a function of the density distribution parameters $(\gamma, n_0)$, assuming the SMC (left) and Orion Nebula (right) extinction curves for LRD A2744-45924. Black solid curves divide the regions where the maximum allowed $A_V$ value is constrained by MIRI F1000W, F2100W, or ALMA data (from the top to the bottom). White dashed contours label the total dust mass $M_{\rm dust}$, computed using Equation (\ref{['eq:Mdust']}) based on the corresponding $A_V$ upper limit. Orange stars mark the density configurations shown in the right panel of Figure \ref{['fig:SED with Different Av']}. The right-bottom region (i.e., steeper profiles with lower density normalization) requires extended dust distribution with $r_{\rm out}\geq 10$ kpc to achieve sufficient column density, and is therefore ruled out as unphysical configurations.
• Figure 4: Left: Infrared re-emission SEDs with the maximum allowed $A_V$ for different dust density configurations, constrained with the stacked LRD SED from Delvecchio_agn-heated_2025 (filled circles and triangles). Stacked SED from Akins_COSMOS-Web_2025 is also shown for comparison (open circles and triangles). The parameter combinations are chosen as $(\gamma, n_0)=(0.5,10^3~{\rm cm}^{-3})$, $(0.25,10~{\rm cm}^{-3})$, $(0,10~{\rm cm}^{-3})$, and $(0, 10^5~{\rm cm}^{-3})$. Right: Parameter survey showing the upper limit on $A_V$ as a function of the density distribution parameters $(\gamma, n_0)$, assuming the stacked LRD SED and the SMC opacity curve. Black solid curves divide the regions where the maximum allowed $A_V$ value is constrained by MIRI F1280W, F1500W, F2100W, or ALMA data (from the top to the bottom).
• Figure 5: Summary of the upper limits on $A_V$ derived for the four different SEDs analyzed in this work: the two individual bright LRDs (A2744-45924 and RUBIES-BLAGN-1) and the two stacked LRD samples from Delvecchio_agn-heated_2025 and Akins_COSMOS-Web_2025. The bar chart compares the results across the four dust opacity laws (from the left to the right: a modified Calzetti law with $\delta = -1.8$, SMC, Milky Way, and Orion Nebula). Each solid rectangle shows the least restrictive $A_V$ upper limit ($A_V^{\rm lim}$) found from the entire $(\gamma, n_0)$ parameter space with an error bar representing $A_V^{\rm lim}$ under different dust mass limits: $M_{\rm dust} \leq 10^4$, $10^5$, and $10^6~M_\odot$ from the bottom to the top. The dashed lines indicate the more stringent limit obtained when restricting the analysis to physically-motivated, concentrated density profiles ($\gamma \geq 1$). The gray shaded region indicates the extinction range typically assumed in obscured AGN scenarios ($A_V \simeq 3.48\pm 0.70~{\rm mag}$, Akins_COSMOS-Web_2025).