Active galactic nuclei-heated dust revealed in "little red dots"

I. Delvecchio, E. Daddi, B. Magnelli, D. Elbaz, M. Giavalisco, A. Traina, G. Lanzuisi, H. B. Akins, S. Belli, C. M. Casey, F. Gentile, C. Gruppioni, F. Pozzi, G. Zamorani

First listed 2025-09-08 | Last updated 2025-12-20

Abstract

Little red dots (LRDs) are a puzzling population of extragalactic sources whose origin is highly debated. In this {work}, we performed a comprehensive stacking analysis of NIRCam, MIRI, and ALMA images of a large and homogeneously selected sample of LRDs from multiple JWST Legacy fields. We report clear evidence of hot-dust emission in the median stacked spectral energy distribution (SED) that features a rising near-infrared continuum up to rest-frame $λ_{\rm rest}$$\sim$ 3$μ$m, which is best explained by a standard dusty active galactic nucleus (AGN) structure. Although LRDs are likely to be a heterogeneous population, our findings suggest that most ($\gtrsim$50 %) LRDs show AGN-heated dust emission, regardless of whether the optical and ultraviolet (UV) continua are stellar or AGN-dominated. In either case, the best-fit dusty-AGN SED, combined with the lack of X-ray detection in the deep Chandra stacks, suggests that Compton-thick ($N_{\rm H}$$>$3$\times$10$^{24}$ cm$^{-2}$) gas obscuration is common, and likely confined within the dust sublimation radius ($R$$_{\rm sub}$$\sim$0.1 pc). Therefore, we argue that AGN-heated dust does not directly obscure either the optical-UV continuum or the broad-line region emission, in order to explain the observed blue UV slopes and prominent Balmer features. While a gas-dust displacement is in line with several models, the formation scenario (in-situ or ex-situ) of this pre-enriched hot dust remains unclear.

Short digest

This Letter median-stacks NIRCam, MIRI, and ALMA imaging for 302 photometrically selected little red dots across CEERS, JADES(+SMILES), and PRIMER (z≈2–11; median z≈6.2). The stacked SED rises to ≈3 µm in the rest frame, and CIGALE fits require a dusty AGN component in most (≳50%) LRDs; combined with ALMA non-detections and deep Chandra stack null results, this points to Compton-thick gas (NH>3×10^24 cm^-2) likely confined within Rsub≈0.1 pc. The inferred geometry leaves optical/UV and broad-line emission largely unobscured, matching blue UV slopes and strong Balmer features. While LRDs are heterogeneous, the prevalence of AGN-heated dust implies obscured black-hole growth is common in this population.

Key figures to inspect

• Fig. 1 stacked cutouts (NIRCam→MIRI→ALMA): verify the band-by-band detections (S/N>3 in all but MIRI/F2550W) and that emission is unresolved compared to the PSF, highlighting the emergence of MIR flux relative to the NIRCam bands.
• Fig. 2 SED fits (CIGALE): compare galaxy-only versus galaxy+AGN solutions to see why MIRI points demand hot dust; inspect the multi-temperature greybody decomposition and the mismatch with the normalized BH-star SED.
• Table 2 stacked photometry: check the measured fluxes and errors used in the SED fit, especially the strength of the 7.7–21 µm stacks and the ALMA/B6 upper limit constraining cool dust.
• Fig. 4 redshift distribution and filter-convolution: assess rest-frame coverage around 1–3 µm at the sample’s median z≈6.2, confirming the lever arm that isolates hot-dust emission.