Non-LTE atmosphere models of very luminous sources and their applicability to Little Red Dots, quasi-stars, and similar objects

Fabrice Martins, Daniel Schaerer, Rui Marques-Chaves

First listed 2026-05-12 | Last updated 2026-05-12

Abstract

We investigate whether atmosphere models traditionally used for massive stars with strong winds can produce synthetic spectra morphologically similar to those of Little Red Dots (LRDs). We compute atmosphere models and synthetic spectra with the code CMFGEN. The models assume a thermalized radiation field at the inner boundary, parameterized by a temperature varying between 5000 and 12000~K. We adopt a typical luminosity of 1e10 Lsun. The models are spherical, assume an expanding atmosphere, and are computed under non-LTE conditions and for several metallicities. The spectral energy distribution (SED) is different from a blackbody, with a blue optical spectrum. Broad hydrogen emission lines are produced, their wings being formed by electron scattering. The SED near the Balmer and Paschen limit is rather continuous. A Balmer break is predicted for the coolest temperature models provided the wind density is reduced. The SED and Balmer decrement of most LRDs is reproduced by the models, provided they are dust-attenuated with Av~1.9-2.7. Assuming the absorbed luminosity is re-radiated in the infrared, the energy output at these wavelengths is consistent with observational constraints. The models predict FeII, oxygen and calcium lines. OI lines at 8446 A and 1.129 um are produced mostly by Lybeta fluorescence. The strength of emission lines from metals depends on input temperature, metallicity, and details of the radiative transfer models. CMFGEN atmosphere models predict a large number of spectral properties observed in many LRDs. They struggle to simultaneously produce a genuine Balmer break and strong emission lines. Whether they are more relevant or not to explain LRDs' spectra compared to alternative models is unclear, leaving open the question of the physical conditions in LRDs.

Short digest

Motivated in part by the spectral resemblance between the lensed LRD GLIMPSE 17775 and Eta Car-like wind spectra, Martins et al. test whether CMFGEN non-LTE, spherically expanding atmosphere models for extremely luminous sources can reproduce Little Red Dot phenomenology without imposing an external AGN continuum. The models naturally produce blue optical continua, broad hydrogen lines with electron-scattering wings, and metal features including Fe II, O I, and Ca II, with the O I 8446 Å and 1.129 μm lines arising mainly from Lyβ fluorescence. After dust attenuation of roughly A_V≈1.9-2.7, they recover the SED shape and Balmer decrement of many LRDs, and the absorbed power can be reradiated in the infrared without violating current constraints. The main limitation is that these same models have difficulty producing both a genuine Balmer break and strong emission lines at the same time, so they make dense-envelope or quasi-star-like atmospheres more plausible while not yet settling the physical origin of LRD spectra.

Key figures to inspect

• Figure 4. This density experiment is one of the most conclusion-driving figures in the paper because it shows how lowering the wind density in the cool T6 model changes the emergent continuum around the Balmer region. It is the clearest visual demonstration of the paper's main caveat: conditions that help produce a Balmer break also alter the line-emission behavior, making it hard to recover both a genuine break and strong broad lines simultaneously.
• Figure 5. This is the key SED synthesis figure, comparing attenuated CMFGEN models to the stacked spectra of the different LRD subclasses defined by Pérez-González et al. It shows where the models succeed across the observed continuum diversity, why A_V in the range of about 1.9-2.7 is central to the match, and how the intrinsic, transmitted, and absorbed luminosities feed the paper's infrared energy-budget argument.
• Figure 6. This normalized comparison against GN 9771 and GLIMPSE 17775 tests the model spectra at the level of observed line morphology rather than just broadband color. It is especially useful for judging the paper's headline claim that non-LTE expanding atmospheres can reproduce the mix of broad Balmer emission and multiple metallic features seen in real LRD spectra.
• Figure 7. This figure isolates the hydrogen-line profiles and explores how clumping and added turbulent velocity affect the match to observed LRD Balmer lines. It matters because the broad wings are a central observational puzzle, and this plot shows how far electron-scattering atmospheres can go toward reproducing those profiles and what extra line-shaping ingredients the models need.
• Figure 11. This is the cleanest process-identification figure in the paper: by suppressing Lyβ and showing the response of O I 8446 and O I 11287, it demonstrates that these lines are powered mainly by fluorescence. That directly supports one of the paper's most specific spectral predictions and gives readers a concrete physical handle on why the modeled metal-line spectrum resembles some of the best LRD data.