ATMO2020 Evolutionary Models

• ATMO2020 evolutionary models are comprehensive grids that couple self-consistent atmospheric boundary conditions with interior cooling tracks to simulate brown dwarfs and giant exoplanets.
• They incorporate revised input physics, including updated H-He equation of state and enhanced molecular opacities, which yield cooler, denser atmospheres and refined spectral predictions.
• The models enable practical inference of effective temperature, radius, mass, age, and surface gravity by integrating observational data from JWST, SPHEREx, and other platforms.

Searching arXiv for ATMO2020 and closely related papers to support the article. ATMO 2020 is a set of solar metallicity atmosphere and evolutionary models for very cool brown dwarfs and self-luminous giant exoplanets, generated with the 1D radiative-convective equilibrium code ATMO and used as surface boundary conditions to calculate the interior structure and evolution of $0.001-0.075\,\mathrm{M_\odot}$ objects (Phillips et al., 2020). In practice, the term denotes both the atmosphere grids and the coupled cooling tracks, so the “ATMO2020 evolutionary models” are best understood as non-gray evolutionary calculations anchored to self-consistent atmospheric boundary conditions. Subsequent work has used ATMO2020-based tracks, and the modified ATMO2020++ atmosphere grid, to infer $T_{\mathrm{eff}}$, radius, mass, age, and evolutionary $\log g$ for cold brown dwarfs and giant exoplanets observed with JWST and SPHEREx (Tu et al., 2024).

1. Scope, parameter space, and model construction

ATMO 2020 was introduced as a new set of solar metallicity atmosphere and evolutionary models for cool T-Y brown dwarfs and giant exoplanets (Phillips et al., 2020). The atmosphere models are generated with a state-of-the-art 1D radiative-convective equilibrium code, and the atmospheric pressure-temperature profiles at deep optical depths $(\tau=1000)$ are coupled to the interior structure/evolution code as a non-gray boundary condition. This replaces older outer-boundary prescriptions and is central to the model family’s evolutionary behavior.

The published grids span the following ranges:

Quantity	ATMO 2020 coverage
Mass	$0.001-0.075\,\mathrm{M_\odot}$
Age	$0.001-10\,\mathrm{Gyr}$
Atmosphere grid $T_{\mathrm{eff}}$	$200-3000\,\mathrm{K}$
Recommended $T_{\mathrm{eff}}$ range	up to $2000\,\mathrm{K}$
$T_{\mathrm{eff}}$0	$T_{\mathrm{eff}}$1
Chemistry grids	one equilibrium and two non-equilibrium grids

Radiative transfer is solved in discrete ordinates with 16 rays and Gauss-Legendre quadrature, including isotropic scattering, while opacities are combined with the correlated-$T_{\mathrm{eff}}$2 method for computational efficiency (Phillips et al., 2020). Convection is modeled using mixing length theory with a mixing length of 2 pressure scale heights. In later observational studies, the ATMO2020 and ATMO2020++ atmosphere grids are described as self-consistent, radiative-convective equilibrium, cloudless atmospheric models with disequilibrium chemistry (Tu et al., 2024).

2. Input-physics revisions relative to earlier evolutionary grids

A defining feature of ATMO2020 is the revision of the input physics relative to earlier AMES-Cond and Saumon & Marley 2008 evolutionary calculations (Phillips et al., 2020). The most prominent change is the adoption of the H-He equation of state from Chabrier et al. (2019), which incorporates ab initio quantum molecular dynamics calculations. In the stated comparison with the older SCVH equation of state, this produces cooler, denser, more degenerate objects, especially near the stellar/substellar boundary.

The quantitative consequences are explicit. The hydrogen-burning and deuterium-burning minimum masses are raised by about $T_{\mathrm{eff}}$3 in mass, and a $T_{\mathrm{eff}}$4, $T_{\mathrm{eff}}$5 object is predicted to be about $T_{\mathrm{eff}}$6 cooler and $T_{\mathrm{eff}}$7 less luminous than with the old equation of state (Phillips et al., 2020). The location of the deuterium burning minimum mass is also shifted, and the cooling tracks crossing the theoretical stellar/substellar boundary are altered, particularly for old, high-mass brown dwarfs.

A second major revision is the update to molecular opacities. The atmosphere code incorporates significantly more line transitions, including extensive ExoMol line data for CH$T_{\mathrm{eff}}$8 and NH$T_{\mathrm{eff}}$9, together with improved H$\log g$0O lists (Phillips et al., 2020). The stated effect is warmer atmospheric temperature structures at low $\log g$1, with corresponding changes to cooling curves, predicted emission spectra, and color-magnitude trajectories.

The treatment of the collisionally broadened potassium resonance doublet is also substantially improved. ATMO 2020 employs the Allard et al. (2016) profiles, which update interaction potentials, extend the validity of the profiles to higher densities, and include spin-orbit coupling (Phillips et al., 2020). These lines are highlighted as important in shaping the red-optical and near-infrared spectrum of brown dwarfs, and the revised broadening can induce up to several hundred kelvin differences in deep atmospheric temperatures together with a redistribution of near-infrared flux.

3. Chemistry, vertical mixing, and spectral diagnostics

The ATMO 2020 release contains three model grids: one computed in equilibrium chemistry and two computed self-consistently with non-equilibrium chemistry due to vertical mixing (Phillips et al., 2020). Chemical equilibrium is calculated via Gibbs free energy minimization, including 76 gas and 92 condensate species, and elemental rainout is implemented so that once condensates form, their elements are removed from higher atmospheric layers.

Vertical mixing is treated through a chemical relaxation scheme, with quench levels determined by the competition between chemical and vertical mixing timescales (Phillips et al., 2020). Two non-equilibrium grids are provided, corresponding to “strong” and “weak” vertical mixing, with $\log g$2 and $\log g$3 at $\log g$4. The physical effect is the familiar quenching of slow-reacting species: CO and N$\log g$5 are enhanced in the upper atmosphere, while CH$\log g$6 and NH$\log g$7 are depleted.

The reported spectral consequences are specific. Vertical mixing increases flux in the H and L$\log g$8 bands, lowers flux at $\log g$9 through increased CO and CO$(\tau=1000)$0, and produces redder J-H and H-L$(\tau=1000)$1 colors together with bluer H-W2 colors (Phillips et al., 2020). The paper explicitly emphasizes that the $(\tau=1000)$2 flux window can be used to calibrate vertical mixing in cool T-Y spectral type objects. This made the ATMO2020 family especially relevant for JWST and Spitzer/WISE analyses in which the $(\tau=1000)$3 region carries strong leverage on disequilibrium chemistry.

4. Cooling behavior, spectra, and color-magnitude predictions

The evolutionary consequences of the revised equation of state and atmosphere boundary conditions are not limited to small parameter shifts. ATMO 2020 predicts modified cooling tracks that are slower or cooler near the hydrogen-burning limit and around the deuterium-burning minimum mass (Phillips et al., 2020). At low masses, the warmer atmospheric boundary conditions make brown dwarfs fainter and cooler at ages below $(\tau=1000)$4, but warmer and brighter at older ages.

In synthetic spectra and color-magnitude diagrams, the updated CH$(\tau=1000)$5, NH$(\tau=1000)$6, H$(\tau=1000)$7O, and alkali opacities improve the reproduction of JHK spectra and of the red-optical through near-infrared features shaped by potassium resonance lines (Phillips et al., 2020). The models are reported to match the observed locus of cool T-Y dwarfs more faithfully than AMES-Cond, SM08, S12, and BT-Cond in both spectra and photometry. In particular, H-W2 and H-L$(\tau=1000)$8 colors are more accurately reproduced when non-equilibrium chemistry is included.

At the same time, the published discussion is explicit that agreement is not complete. J-H color discrepancies remain, and the “4 micron problem” is described as improved but not fully resolved (Phillips et al., 2020). A plausible implication is that ATMO2020 established a stronger baseline for cloud-free, disequilibrium calculations in the late-T/Y regime while still leaving room for additional physics in some wavelength regions and temperature ranges.

5. Use of ATMO2020-based tracks in brown-dwarf and exoplanet inference

Later observational studies use evolutionary models from Chabrier et al. (2023), described as adopting the ATMO2020 atmospheric grid, to convert spectroscopically inferred atmospheric quantities into mass, radius, age, and evolutionary $(\tau=1000)$9 (Tu et al., 2024). In these workflows, the key inputs are typically $0.001-0.075\,\mathrm{M_\odot}$0 and radius, where the radius comes from the fitted flux scaling $0.001-0.075\,\mathrm{M_\odot}$1 together with parallax. In SPHEREx SUDA, the pair $0.001-0.075\,\mathrm{M_\odot}$2 is used for interpolation in the C23 tracks because tests confirm that this pairing yields weaker covariance and less degeneracy than using $0.001-0.075\,\mathrm{M_\odot}$3 (Tu et al., 29 Apr 2026). The corresponding surface gravity is then

$0.001-0.075\,\mathrm{M_\odot}$4

In the JWST study of 20 T and Y dwarfs, fitted effective temperatures and absolute parallaxes were combined with evolutionary tracks to derive radii from approximately $0.001-0.075\,\mathrm{M_\odot}$5 to $0.001-0.075\,\mathrm{M_\odot}$6 (Tu et al., 2024). Using interpolated evolutionary tracks with fitted $0.001-0.075\,\mathrm{M_\odot}$7 and radius as input, most brown dwarfs in that sample were inferred to have masses below $0.001-0.075\,\mathrm{M_\odot}$8 and ages younger than $0.001-0.075\,\mathrm{M_\odot}$9, while Y dwarfs were placed in the ranges $0.001-10\,\mathrm{Gyr}$0 and $0.001-10\,\mathrm{Gyr}$1.

For COCONUTS-2 b, ATMO2020 evolutionary cooling tracks were combined with a direct luminosity estimate and a system age of $0.001-10\,\mathrm{Gyr}$2 to derive a mass of $0.001-10\,\mathrm{Gyr}$3, an evolutionary radius of $0.001-10\,\mathrm{Gyr}$4, an evolutionary $0.001-10\,\mathrm{Gyr}$5 of $0.001-10\,\mathrm{Gyr}$6, and an evolutionary $0.001-10\,\mathrm{Gyr}$7 of $0.001-10\,\mathrm{Gyr}$8 (Ravet et al., 8 Apr 2026). The luminosity estimate itself used the Stefan-Boltzmann relation,

$0.001-10\,\mathrm{Gyr}$9

and the MIRI wavelength range was reported to account for $T_{\mathrm{eff}}$0 of the bolometric flux (Ravet et al., 8 Apr 2026). These examples show the role of ATMO2020-based tracks as a bridge between atmospheric fitting and bulk-parameter inference.

6. ATMO2020++, comparative behavior, and recurring interpretation issues

ATMO2020++ is described as a modification of the ATMO2020 models and differs primarily in its treatment of convection (Tu et al., 2024). Whereas ATMO2020 uses the standard thermochemical-equilibrium adiabatic gradient with $T_{\mathrm{eff}}$1-1.4, ATMO2020++ incorporates a reduced temperature gradient to account for convective instabilities triggered by CO/CH$T_{\mathrm{eff}}$2 and N$T_{\mathrm{eff}}$3/NH$T_{\mathrm{eff}}$4 transitions, setting the effective adiabatic index to $T_{\mathrm{eff}}$5. In the JWST 20-object analysis, the ATMO2020++ grid covered $T_{\mathrm{eff}}$6, $T_{\mathrm{eff}}$7, and $T_{\mathrm{eff}}$8, with a fixed relation for vertical mixing,

$T_{\mathrm{eff}}$9

This altered convective prescription has a direct effect on inferred surface gravities. The same study reports that ATMO2020++ produces systematically higher $200-3000\,\mathrm{K}$0 values, with a median about $200-3000\,\mathrm{K}$1 higher than Sonora Elf Owl and higher than ATMO2020 for the same objects (Tu et al., 2024). The conclusion drawn there is that surface gravity from spectroscopy is highly model-dependent, especially sensitive to convection and effective adiabatic index implementations, whereas effective temperature and parallax-based radius are more robust.

A related issue appears in the SUDA analysis of 1675 ultracool dwarfs. There, ATMO2020++ is characterized as optimized for late T/Y regimes and used preferentially for sources colder than $200-3000\,\mathrm{K}$2 (Tu et al., 29 Apr 2026). However, in the $200-3000\,\mathrm{K}$3 range, atmospheric $200-3000\,\mathrm{K}$4 values from spectral fitting are systematically lower than evolutionary $200-3000\,\mathrm{K}$5 by a median offset of about $200-3000\,\mathrm{K}$6, which the paper attributes to residual degeneracy between $200-3000\,\mathrm{K}$7 and metallicity in low-resolution SPHEREx spectra. Outside this range, the offsets are smaller and less systematic, and for late T/Y dwarfs atmospheric and evolutionary gravities agree better.

The COCONUTS-2 b study adds a model-comparison perspective. Classical model comparison found the ATMO2020++ synthetic spectra, with and without PH$200-3000\,\mathrm{K}$8, to be statistically preferred for the combined $200-3000\,\mathrm{K}$9 dataset (Ravet et al., 8 Apr 2026). When Gaussian Processes were used to model correlated noise, Sonora Elf Owl attained the highest log-Bayes factor, but yielded bulk parameters inconsistent with cooling models and poorer MIRI fits; ATMO2020++ remained the most physically consistent model. This suggests that the practical value of the ATMO2020 family lies not only in fit quality, but also in consistency between spectral inference and evolutionary cooling constraints.

A recurrent misconception is that atmospheric $T_{\mathrm{eff}}$0, evolutionary $T_{\mathrm{eff}}$1, and masses inferred from different grids are directly interchangeable. The later brown-dwarf studies argue against that reading: surface gravity, mass, and age are highly sensitive to the atmospheric model grid choice and especially to the convection prescription, whereas reliable physical inference requires pairing evolutionary tracks with consistent atmospheric models (Tu et al., 2024). In that restricted but important sense, “ATMO2020 evolutionary models” denotes not just a cooling-track library, but a coupled atmosphere-evolution framework whose interpretive power depends on internal physical consistency.