What is the nature of Little Red Dots and what is not, MIRI SMILES edition

TL;DR

This study uses JWST/JADES and SMILES MIRI data to investigate the nature of Little Red Dots (LRDs), a population of compact, high-redshift galaxies with blue rest-UV and red rest-IR SEDs. By combining NIRCam and deep MIRI photometry with four distinct SED-modeling codes that span star-formation- and AGN-dominated scenarios, the authors show that stellar emission typically dominates the rest-frame UV–optical, while rest-frame 1–2 μm can require additional dust or modest AGN contribution. The results yield typical stellar masses around 10^9.4 M⊙, large dust attenuation (A_V ≈ 3–4 mag), very young mass-weighted ages (5–10 Myr in some fits, up to ~150 Myr in others), and a number density of ~10^-4 Mpc^-3, representing about 14% of galaxies at similar redshifts and masses. Overall, LRDs are best viewed as extreme, compact starburst systems with efficient dust production, though a minority may host obscured AGN that contribute to their mid-IR luminosity. The work demonstrates the critical role of extending SED analyses into the MIRI regime to disentangle stellar and AGN signatures in the early Universe.

Abstract

We study little red dots (LRD) detected by JADES and covered by the SMILES MIRI survey. Our sample contains 31 sources, $\sim70$% detected in the two bluest MIRI bands, 40% in redder filters. The median/quartiles redshifts are $z=6.9_{5.9}^{7.7}$ (55% spectroscopic). We analyze the rest-frame ultraviolet through near/mid-infrared spectral energy distributions of LRDs combining NIRCam and MIRI observations, using a variety of modeling techniques that include emission from stars, dust, and (un)obscured active galactic nuclei (AGN). The NIRCam$-$MIRI colors, for $\geq10$ $μ$m, are bluer than direct pure emission from AGN tori; the spectral slope flattens in the rest-frame near-infrared, consistent with a 1.6 $μ$m stellar bump. Both observations imply that stellar emission makes the dominant contribution at these wavelengths, expediting a stellar mass estimation: the median/quartiles are $\log \mathrm{M_\star/M_\odot}=9.4_{9.1}^{9.7}$. The number density of LRDs is $10^{-4.0\pm0.1}$ Mpc$^{-3}$, accounting for $14\pm3$% of the global population of galaxies with similar redshifts and masses. The flat ultraviolet spectral range is dominated by young stars. The rest-frame near/mid-infrared (2-4 $μ$m) spectral slope reveals significant amounts of dust (bolometric stellar attenuation $\sim3-4$ mag) heated by strong radiation fields arising from highly embedded compact sources. Our models imply $<0.4$ kpc heating knots, containing dust-enshrouded OB stars or an AGN producing a similar radiation field, obscured by $\mathrm{A(V)}>10$ mag. We conclude that LRDs are extremely intense and compact starburst galaxies with mass-weighted ages 5-10 Myr, very efficient in producing dust, their global energy output dominated by the direct and dust-recycled emission from OB stars, with some contribution from obscured AGN in the mid-infrared.

Figures (14)

• Figure 1: The left and central panels show the $F277W-F444W$ vs$.$$F444W$ color-magnitude and $F277W-F444W$ vs. $F150W-F200W$ color-color diagrams, as well as histograms of NIRCam colors, indicating the selection thresholds for LRDs ($F277W-F444W>1$ mag, $F150W-F200W<0.5$ mag and $F444W<28$ mag; dashed lines) relative to the bulk of the JADES DR2 galaxy catalog. The different colors indicate the subsets of LRDs detected in different MIRI bands: up to F1280W and beyond (Golden Five galaxies), up to F1000W, up to F770W, or not detected in MIRI (at the SMILES depth). Comparison LRD samples from 2023arXiv230514418B and 2023arXiv230607320L in the CEERS and UNCOVER fields are shown with squares and triangles, respectively. The right panel shows the color $vs.$ redshift diagram for the LRDs and JADES galaxies and the redshift distribution of LRDs (including median and quartiles). The 55$\%$ of the LRDs in our paper that have secure NIRSpec- and NIRCam-based spectroscopic redshifts are marked with a black dot.
• Figure 2: NIRCam F150W+F277W+F444W color composite postages of the sample of 31 LRDs in this paper. North is up, East is left, sizes are 2.5$\times$2.5 arcsec$^2$. Galaxies detected by MIRI at wavelengths longer than F1280W (Golden Five galaxies), F1000W, and F770W are marked in gold, red, and purple (and the sources are ordered according to this MIRI detections). Those not detected in MIRI are marked in magenta. We display redshifts, with spectroscopic values written with 4 decimals.
• Figure 3: Stacked LRD SEDs for sources detected by MIRI (left panel) and not (right panel), normalized at rest-frame wavelength 0.4 $\mu$m. Black points show NIRCam fluxes for individual sources, while rainbow color points show MIRI fluxes. Arrows depict 5$\sigma$ upper limits. The average SED is shown with a magenta line (10 point averages) and its rms with a magenta shaded region. The observed average SED for LRDs is compared to 5 different templates. The orange lines show an average QSO spectrum (see text for details), and the same template extincted by $\mathrm{A(V)}=2$ mag using a 2000ApJ...533..682C attenuation law. The red lines stand for the torus template in 2007ApJ...663...81P, normalized to the same wavelength as the observations (continuous line) as well as normalized at the 2 $\mu$m average SED level (dashed line) in the case of the MIRI detections plot. The blue line shows the model for an intense starburst presented in 2007AA...461..445S, more specifically, the sub-LIRG model with a size of 350 pc, 90% of the total luminosity coming from OB stars, optical attenuation of $\mathrm{A(V)}=36$ mag, and $10^3$ cm$^{3}$ density of dust in hot spots (gas clouds surrounding and directly heated by OB stars).
• Figure 4: NIRCam and MIRI colors $vs.$ redshift for all LRDs detected at least in F770W. The lines illustrate the color-redshift tracks for the same templates shown in Figure \ref{['fig:average_sed']}. Galaxies from different samples are marked with different colors. The first two panels in the top row show the F444W-F560W(F770W) colors which are available for the majority of the sources. The last panel at the top and the bottom panels show the MIRI-MIRI colors relative to F770W. The median and scatter of the colors are indicated in the top left corner.
• Figure 5: SED fitting results for source JADES$-$57356, the LRD in our sample detected up to F2100W. The four upper panels show: (1) the fits for synthesizer-AGN$+$ on the top left panel, including the individual components of the model, i.e., young, old, and all stars in blue, cyan, and gray, (un)obscured AGN in orange, and regular dust emission (in principle, linked to star formation) in red; (2) results for Prospector-AGN$+$ on the top right panel, showing the three components, stars in gray, AGN in orange, and star-formation heated dust in red; (3) Prospector-SF on the bottom left, showing stars in gray and star-formation heated dust in red; and (4) Prospector-AGN on the bottom right, including young stars in blue, total AGN emission in orange, with the torus emission shown with a dashed line. Number of bands fitted and direct (i.e., not reduced) $\chi^2$ values are provided, as well a stellar masses, stellar mass-weighted ages, $V$-band stellar attenuation, ratio between the FUV and optical stellar attenuation, and fraction of bolometric luminosity coming from the AGN. The fits include NIRCam bands, shown in black, and MIRI fluxes, in color; upper limits, depicted with triangles, are also used. Below the SEDs, we give 10$^{\prime\prime}$$\times$10$^{\prime\prime}$ postage stamps in NIRCam (upper row), MIRI (middle row), and MIRI convolved with a 5-pixels wide tophat filter, the LRD being marked with a 0.3$^{\prime\prime}$ radius circle, and the $S/N$ provided (when it is above 5).