Under The Magnifying Glass: A Combined 3D Model Applied To Cloudy Warm Saturn Type Exoplanets Around M-Dwarfs

Astronomy & Astrophysics manuscript no.
current_version ©ESO 2024

October 24, 2024
Under the magnifying glass: A combined 3D model applied to

cloudy warm Saturn type exoplanets around M-dwarfs
S. Kiefer1, 2, 3 , N. Bach-Møller2, 3, 4 , D. Samra2 , D. A. Lewis2, 3 , A. D. Schneider1, 4 , F. Amadio4, 1 ,
H. Lecoq-Molinos2, 3, 1 , L. Carone2 , L. Decin1 , U. G. Jørgensen4 , and Ch. Helling2, 3
1
Institute of Astronomy, KU Leuven, Celestijnenlaan 200D, 3001 Leuven, Belgium
e-mail: [email protected]
2
Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria
3
Institute for Theoretical Physics and Computational Physics, Graz University of Technology, Petersgasse 16 8010 Graz
4
Centre for ExoLife Sciences, Niels Bohr Institute, Øster Voldgade 5, 1350 Copenhagen, Denmark
arXiv:2410.17716v1 [astro-ph.EP] 23 Oct 2024
Received ...; accepted ...
ABSTRACT
Context. Warm Saturn type exoplanets orbiting M-dwarfs are particularly suitable for in-depth cloud characterisation through trans-
mission spectroscopy due to their favourable stellar to planetary radius contrast. The global temperatures of warm Saturns suggest
efficient cloud formation in their atmospheres which in return affects the temperature, velocity, and chemical structure. However,
modelling the formation processes of cloud particles consistently within the 3D atmosphere remains computationally challenging.
Aims. The aim is to explore the combined atmospheric and micro-physical cloud structure, and the kinetic gas-phase chemistry of
warm Saturn-like exoplanets in the irradiation field of an M-dwarf. The combined modelling approach will support the interpretation
of observational data from present (e.g. JWST, CHEOPS) and future missions (PLATO, Ariel, HWO).
Methods. A combined 3D cloudy atmosphere model for HATS-6b is constructed by iteratively executing the 3D General Circulation
Model (GCM) expeRT/MITgcm and a detailed, kinetic cloud formation model, each in its full complexity. Resulting cloud particle
number densities, particles sizes, and material compositions are used to derive the local cloud opacity which is then utilised in the next
GCM iteration. The disequilibrium H/C/O/N gas-phase chemistry is calculated for each iteration to assess the resulting transmission
spectrum in post-processing.
Results. The first model atmosphere that iteratively combines cloud formation and 3D GCM simulation is presented and applied to
the warm Saturn HATS-6b. The cloud opacity feedback causes a temperature inversion at the sub-stellar point and at the evening
terminator at gas pressures higher than 10−2 bar. Furthermore, clouds cool the atmosphere between 10−2 bar and 10 bar, and narrow
the equatorial wind jet. The transmission spectrum shows muted gas-phase absorption and a cloud particle silicate feature at ∼ 10µm.
Conclusions. The combined atmosphere-cloud model retains the full physical complexity of each component and therefore enables
a detailed physical interpretation with JWST NIRSpec and MIRI LRS observational accuracy. The model shows that warm Saturn
type exoplanets around M-dwarfs are ideal candidates to search for limb asymmetries in clouds and chemistry, identify cloud particle
composition by observing their spectral features, and identify the cloud-induced strong thermal inversion that arises on these planets
specifically.
Key words. Planets and satellites: individual: HATS-6b – Planets and satellites: atmospheres – Methods: numerical – Planets and
satellites: gaseous planets – Techniques: spectroscopic
1. Introduction tunity for in-detail cloud and chemistry characterisation across

very different host stars. M-dwarfs are too faint for planetary at-
Warm Saturns are a class of Saturn-sized gas-giants with equilib- mosphere characterisation in the optical and around 1 µm with
rium temperatures T eq between 500 K and 1200 K (Fig. 1). The HST. Their luminosities peak at longer wavelengths, making
atmospheres of warm Saturns have so far been difficult to char- warm Saturns around M-dwarfs together with their favourable
acterise because observations with the Hubble Space Telescope stellar to planetary radius ratio ideal targets for infrared trans-
(HST) showed either absent or muted spectral features. Such mission spectroscopy with JWST.
observations can be interpreted either with a cloud-free high First studies with cloudless General Circulation Models
metallicity composition or a global extended cloud coverage (GCMs) predict more uniform temperatures of warm Saturns
with lower metallicity (Komacek et al. 2020; Carone et al. 2021; (Christie et al. 2022; Helling et al. 2023) compared to (ultra) hot
Wong et al. 2022). James Webb Space Telescope (JWST) obser- Jupiters. Moreover, due to their relatively low planetary global
vations together with better models can resolve the ambiguous temperature, a very efficient horizontal heat circulation has been
results for warm Saturns (or warm Neptunes like WASP 107b). inferred (e.g. Kataria et al. 2016; Komacek & Showman 2016).
The majority of warm Saturns have been observed around Thus, a global and highly mixed composition cloud coverage
metal rich F, G, or K stars (Buchhave et al. 2018). A selected few can be expected for warm Saturn type exoplanets (Christie et al.
have now been detected around M-dwarfs (Cañas et al. 2022; Lin 2022; Helling et al. 2023). Clouds scatter light at the top of
et al. 2023; Hartman et al. 2015). Atmosphere characterisation of the planet (Rowe et al. 2008) leading to less absorbed incom-
warm Saturns in the era of JWST provides an excellent oppor- ing stellar light, which tends to cool a planetary atmosphere. At
Article number, page 1 of 21
A&A proofs: manuscript no. current_version
solar planets. These models predict the hot spot offset within
2 HAT-P-7b
WASP-43b
HATS-6b
TOI-1231b
HATS-71 b
LP 714-47 b hot Jupiters due to equatorial superrotation (Showman & Guillot
HD 189733b WASP-96b TOI-1728 b 2002; Showman & Polvani 2011; Carone et al. 2020) which has
HD 209458b WASP-39b TOI-674 b
1 TOI-1634b GJ 1214b JWST targets been observed since Knutson et al. (2007). To model the interac-
tion of clouds and climate, cloud models and 3D GCMs need to
log10 (MPl [MJ])
be combined, which is computationally intensive. There are cur-

0 rently several approaches to tackle this problem. In a hierarchical
approach, the output of GCMs can be used to post-process cloud
structures (Helling et al. 2016; Kataria et al. 2016; Parmentier
-1 et al. 2016; Helling et al. 2019b, 2021; Robbins-Blanch et al.
2022; Savel et al. 2022; Helling et al. 2023). While these models
-2 can account for detailed micro-physical cloud formation, they
are missing the feedback of clouds on the GCM. So far, only
a few models couple micro-physical cloud models with GCMs
-3 (Lee et al. 2016, 2017; Lines et al. 2018b,a). To reduce the com-
500 1000 1500 2000 2500 putational cost, some models use simplifications for the atmo-
Teq [K] sphere models (Ormel & Min 2019; Min et al. 2020) or the cloud
modelling (Christie et al. 2021; Lee et al. 2023). Others use ra-
Fig. 1. Known gas-giants and JWST targets. The blue shaded area rep- diatively passive (Parmentier et al. 2013; Charnay et al. 2015b;
resents the approximate parameter space of warm Saturns. Several often Komacek et al. 2019) or active (Charnay et al. 2015a) cloud par-
studied exoplanets are shown. ticle tracers. Others parameterise the advection of clouds and
use condensation curves to prescribe the location of clouds in
the 3D GCM (Parmentier et al. 2016, 2018; Tan & Showman
the same time, clouds absorb and re-emit outgoing thermal ra- 2017, 2021; Roman & Rauscher 2017, 2019; Harada et al. 2021;
diation of the planet and thus impose an additional greenhouse Christie et al. 2021, 2022). In all cases, there is a trade off be-
effect that can warm the underlying atmosphere. Understanding tween computational cost and the detail by which the models
which effects determine the thermodynamic environment of the describe the interaction of clouds and climate.
planet requires 3D modelling of the cloud properties and their Warm Saturns around M-dwarf stars are interesting targets
horizontal and vertical distribution, as has been demonstrated for to study cloud formation and chemistry. The existence of such
hot Jupiters (Lines et al. 2018b; Powell et al. 2019; Parmentier planets question formation models that predict that only more
et al. 2021; Lee et al. 2023) and rocky exoplanets (Yang et al. massive host stars produce a protoplanetary disk with enough
2014; Turbet et al. 2021). Recent JWST observations, as well material to form a gas-giant (e.g. Pascucci et al. 2016). M-dwarfs
as recent theoretical work, showcase that the properties of cloud have a higher stellar activity than solar-type stars (Mignon et al.
particles within a planetary atmosphere are not necessarily uni- 2023). Specifically, the higher magnetic activity of M-dwarfs is
form, even on planets where global cloud cover is expected. In expected to expose planets around these stars to higher amounts
the case of the hot Jupiter WASP-39b, cloud asymmetries be- of stellar energetic particles (SEPs, e.g. Fraschetti et al. 2019;
tween the morning and evening terminator have been predicted Rodgers-Lee et al. 2023), which can affect the chemistry on these
(Carone et al. 2023; Arfaux & Lavvas 2024) and confirmed (Es- planets (Venot et al. 2016; Barth et al. 2021; Konings et al. 2022).
pinoza et al. 2024). HATS-6b is a rare case of a transiting warm Saturn that or-
For the in-detail characterisation of cloud properties, gas- bits an M-dwarf host star. HATS-6b was discovered by the HAT-
phase chemistry, and feedback with the 3D wind flow in ex- South survey in 2015 and has well-constrained planetary and
oplanet atmospheres, complex 3D cloud models are needed. stellar parameters. The planet has a mass of 0.319 ± 0.070 M J ,
Clouds and their formation in gas-giant exoplanets have been a radius of 0.998 ± 0.019 R J and an orbital period of 3.3253 d,
described with multiple levels of theory. Simplified descriptions where M J is the mass of Jupiter and R J the radius of Jupiter. The
assume phase equilibrium to determine where clouds can poten- zero-albedo equilibrium temperature is ≈ 700 K (Hartman et al.
tially be present (e.g. Demory et al. 2013; Webber et al. 2015; 2015). The host star, HATS-6, is an early M-dwarf (M1V) with
Crossfield 2015; Kempton et al. 2017; Roman & Rauscher 2017; a mass of 0.57 M⊙ , a radius of 0.57 R⊙ , a metallicity of [Fe/H] =
Roman et al. 2021). To be able to make predictions of the ma- 0.2 ± 0.09, and an effective temperature of T eff = 3724 K, where
terial composition, cloud particle sizes, location of formation, M⊙ is the mass of the Sun and R⊙ is the radius of the Sun. Con-
and effect on the gas-phase abundances by depletion, fully self- sequently, HATS-6b has one of the deepest transit depths known
consistent, micro-physical theories are needed (e.g. Woitke & with (RP /R⋆ )2 = 0.0323 ± 0.0003 around 600 nm, where RP is
Helling 2003; Helling & Woitke 2006; Helling & Fomins 2013; the observed planetary radius and R⋆ the host star’s radius. In
Powell et al. 2018; Woitke et al. 2020; Gao et al. 2020). Atmo- addition, the host star has a J band magnitude of 12.05, making
sphere models combining radiative transfer and micro-physical HATS-6b very well suitable for atmospheric characterisation in
cloud formation are often solved for one dimension (Helling the infrared. HATS-6b, together with eight other warm Saturns,
et al. 2008a; Witte et al. 2011; Juncher et al. 2017; Gao et al. have been selected as targets for two general observer programs
2020). However, atmospheric processes such as equatorial wind for the James Webb Space Telescope (JWST) Cycle 2 (GO 3171
jets (e.g. Showman & Guillot 2002), day-night cold traps (e.g. and 3731).
Parmentier et al. 2013; Pelletier et al. 2023), day-night asymme- This paper has two aims. Firstly, we aim to explore the at-
try (e.g. Perez-Becker & Showman 2013; Komacek et al. 2017; mospheric, micro-physical cloud and gas-phase structure of the
Helling et al. 2019a) and patchy clouds (e.g. Line & Parmentier warm Saturn HATS-6b orbiting an M-dwarf. We use a combined
2016; Tan & Showman 2021) cannot be captured by 1D models. model in the form of step-wise iterations between a detailed
3D GCMs are successfully used to understand the cloud- cloud formation description and expeRT/MITgcm, a 3D GCM
free wind flow and basic thermodynamic structures of extra- with full radiative transfer and deep atmosphere extension. The
S. Kiefer et al.: Under the magnifying glass
second goal is to demonstrate how the combined modelling ap- The thermal forcing part of expeRT/MITgcm solves the ra-
proach can help to support the interpretation of the data from diative transfer equation including isotropic scattering using the
space missions (e.g. JWST) for warm Saturn type planets. The Feautrier method (Feautrier 1964) with approximate lambda iter-
step-wise iterative approach between the GCM (Sect. 2.1) and ation (Olson et al. 1986). expeRT/MITgcm uses 11 correlated-k
cloud structure (Sect. 2.2) is described in Sect. 2. The evaluation wavelength bins for the radiative transfer calculations (S1 res-
of the combined model for the warm Saturn HATS-6b is pre- olution; Schneider et al. 2022c). We chose to use a radiative
sented in Sect. 3. The resulting atmospheric solution and trans- time-step of 100 s, which is four times the dynamical time-step2 .
mission spectra of HATS-6b are shown in Sect. 4. The discussion GGchem (Woitke et al. 2018) is used to pre-calculate a grid of
is in Sect. 5 and the conclusion in Sect. 6. chemical equilibrium abundances that are then used to gener-
ate a premixed opacity grid for expeRT/MITgcm. The elemen-
tal abundances are assumed solar (Asplund et al. 2009) with a
2. Approach solar C/O ratio of C/O = 0.54 and scaled for the metalicity of
HATS-6b of [Fe/H] = 0.2 ± 0.09 (Hartman et al. 2015). A gas
The step-wise iterative process applied in this study is initiated temperature T gas and gas pressure pgas grid for 100 K - 10000
by 3D GCM simulations of the temperature, gas, densities, and K, 10−5 bar - 650 bar is calculated. Taking the temperature and
wind velocities (Sect. 2.1). 1D profiles are extracted from this pressure averages from this premixed grid, the specific gas con-
3D atmosphere solution and used as input for a kinetic cloud stant R = 3925 J kg−1 K−1 and specific heat capacity at constant
formation model and the local cloud opacities are calculated pressure of c p = 14966 J K−1 are obtained.
(Sect. 2.2). The cloud opacity structure is then used as input for Most absorption cross sections are obtained from exomol3 .
the 3D GCM and a new atmosphere structure is calculated. Each H2 O, CO2 , CH4 , NH3 , CO, H2 S, HCN, PH3 , TiO, VO, FeH,
of the models is executed in their full complexity. Such an it- Na, and K opacities are used for the gas absorbers4 . Further-
eration is possible because the cloud formation time scales are more, we include Rayleigh scattering with H2 (Dalgarno &
considerably shorter than the hydrodynamic timescales (Helling Williams 1962) and He (Chan & Dalgarno 1965) and collision-
et al. 2001; Helling & Woitke 2006; Powell et al. 2018; Kiefer induced absorption (CIA) with H2 –H2 and H2 –He (Borysow
et al. 2024). The stopping criterion for the iteration is described et al. 1988, 1989; Borysow & Frommhold 1989; Borysow et al.
in Sect. 2.3. In order to explore how the iterative processes 2001; Richard et al. 2012; Borysow 2002) and H− (Gray 2008).
may affect the spectral information, transmission spectra are cre-
ated for each iteration. This is done by first modelling the non-
equilibrium gas-phase chemistry of H/C/N/O species based on 2.2. Cloud structure
the final output from each GCM run (Sect. 2.4). The resulting
relative concentrations of the gas species are then used to calcu- The results from expeRT/MITgcm are used as input to
late the transmission spectra for each iteration and the JWST’s a kinetic cloud formation model. 1D (T gas (z), pgas (z),
observability of the differences is assessed (Sect. 2.5) vz (z))-profiles are extracted for each point in a longitude-
latitude grid, with spacing of 45◦ in longitude (ϕ =
{−135◦ , −90◦ , −45◦ , 0◦ , 45◦ , 90◦ , 135◦ , 180◦ }) and ∼ 22.5◦ in lat-
2.1. 3D atmosphere itude (θ = {0◦ , 23◦ , 45◦ , 68◦ , 86◦ }), similar to previous works
(Helling et al. 2016, 2019b,a, 2020, 2021; Samra et al. 2022;
The 3D temperature and horizontal (zonal and meridional) wind Carone et al. 2023; Helling et al. 2023). T gas (z) [K] is the lo-
velocities in the atmosphere of HATS-6b are simulated with the cal gas temperature, pgas (z) [bar] is the local gas pressure, and
non-gray 3D GCM expeRT/MITgcm (Schneider et al. 2022c; vz (z) [cm s−1 ] is the local vertical velocity component of the 1D
Carone et al. 2020). This model solves the hydrostatic primi- profile extracted from the 3D GCM. Each 1D model was run
tive equations (HPE) in vertical pressure coordinates on a rotat- top-down in the atmosphere until it reached at least 0.1 bar. For
ing sphere (see, e.g. Showman et al. 2009). Hydrostatic equilib- the southern hemisphere (θ < 0◦ ), all grid points are mirrored
rium and the ideal gas law as equation of state are assumed. The across the equator (θ = 0◦ ). GGChem is used to determine the
model uses the dynamical core of MITgcm (Adcroft et al. 2004) local gas-phase composition in chemical equilibrium for which
which solves the HPE on an Arakawa C type cubed-sphere grid. solar elemental abundances from Asplund et al. (2009) are ini-
Here, we employ the nominal horizontal resolution C321 and a tially assumed which are scaled for the metallicity of HATS-6b.
vertical grid with 41 logarithmically spaced grid cells between
1 × 10−5 bar and 100 bar and six linearly spaced grid cells be-
tween 100 bar and 700 bar. expeRT/MITgcm couples the dy- 2.2.1. Cloud formation
namical core of the MITgcm (Adcroft et al. 2004) to the radiative
transfer solver of petitRADTRANS (Mollière et al. 2019). The Our kinetic cloud formation model treats the micro-physical pro-
model includes Rayleigh friction at the top (at 10−5 bar) and the cesses of nucleation, bulk growth, and evaporation in combina-
bottom (at 700 bar) of the computational domain in order to sta- tion with gravitational settling, element consumption, and re-
bilize the atmosphere against unphysical gravity waves and to plenishment using the moment method for the moments Lj (V)
mimic Ohmic dissipation (Carone et al. 2020). in volume space V (Gail & Sedlmayr 1986, 1988; Dominik
et al. 1993; Woitke & Helling 2003, 2004; Helling et al. 2004;
We use a stellar effective temperature and stellar radius of
Helling & Woitke 2006; Helling et al. 2008b). Cloud conden-
HATS-6 of 3724 K and 0.57 R⊙ , respectively, and assume HATS-
sation nucleii (CCNs) are considered to form from four nu-
6b to be tidally locked on a circular orbit with an orbital sepa-
cleating species: TiO2 , SiO, KCl, and NaCl. This choice was
ration and period of 0.03623 AU and 3.3252725 days, respec-
based on the ones included in previous studies (Lee et al. 2015;
tively. The surface gravity is assumed to be 7.94 ms−2 and the
internal temperature is set to T int = 50 K following Thorngren 2
This is a typical choice for hot Jupiters (see, e.g. Showman et al.
et al. (2019). 2009; Lee et al. 2021; Schneider et al. 2022c,b)
3
https://www.exomol.com/
1 4
C32 corresponds to a resolution of 128x64 in longitude and latitude References can be found in Table. B.1.

Bromley et al. 2016; Lee et al. 2018; Gao & Benneke 2018; timescales of cloud particles on the other hand can still be shorter
Köhn et al. 2021; Sindel et al. 2022; Kiefer et al. 2023). How- than the advection timescales (Helling & Woitke 2006; Lee et al.
ever, we note that there are current efforts to expand the list 2018; Powell & Zhang 2024; Kiefer et al. 2024). Studies which
of nucleating species (e.g. Gobrecht et al. 2022, 2023; Lecoq- considered the horizontal transport of cloud particles (e.g. Lee
Molinos et al. 2024). The growth of the mixed material cloud et al. 2016; Lines et al. 2018b; Komacek et al. 2022; Lee 2023;
particles is modelled through the surface growth of 16 bulk Powell & Zhang 2024) suggest that the number density and size
materials: TiO2 [s], Mg2 SiO4 [s], MgSiO3 [s], MgO[s], SiO[s], of cloud particles are more longitudinally uniform than studies
SiO2 [s], Fe[s], FeO[s], FeS[s], Fe2 O3 [s], Fe2 SiO4 [s], Al2 O3 [s], which only considered vertical mixing as cloud particle trans-
CaTiO3 [s], CaSiO3 [s], KCl[s], NaCl[s]. These materials grow port mechanism (e.g. Helling et al. 2021; Roman et al. 2021;
and evaporate through 132 surface reactions (Helling et al. Samra et al. 2022; Helling et al. 2023). Similarly, the replenish-
2019a) onto the surface of the CCN. The formation and evap- ment of gas-phase species can happen through both horizontal
oration of cloud particles depletes and replenishes the 11 par- and vertical mixing. Which of the two dominates depends on the
ticipating elements Mg, Si, Ti, O, Fe, Al, Ca, S, Na, K and Cl, atmospheric conditions. Studies comparing vertical and horizon-
which affects the gas-phase equilibrium abundances. The kinetic tal mixing timescales for gas-phase species found that at pres-
cloud model assumes a cloud free element reservoir deep in the sures below 0.1 bar, vertical mixing often dominates (Helling
planet. Through mixing, the upper layers are replenished with et al. 2019b; Baeyens 2021; Zamyatina et al. 2024). In general,
cloud forming elements. The exact strength of mixing within ex- considering the horizontal transport of cloud particles is compu-
oplanet atmospheres is difficult to determine (Parmentier et al. tationally expensive. Studies evaluating the effect of horizontal
2013; Steinrueck et al. 2019; Samra et al. 2022). For this study, transport are either limited to few simulation days (e.g. Lee et al.
we used a quasi-diffusive approach utilising the standard devi- 2016; Lines et al. 2018b), make simplifying assumptions on the
ation of neighbouring cells to compute a relaxation timescale cloud formation (e.g. Komacek et al. 2022; Roman et al. 2021;
(see Appendix B in Helling et al. (2023)). Previous work (e.g. Lee 2023), or make simplifying assumptions on the horizontal
Gao et al. 2020) considered sulfur, manganese and zinc-bearing transport (e.g. Powell & Zhang 2024).
species which typically condense below 1000 K for pressure
ranges between 10−4 bar to 102 bar. However, these elements are
2.2.3. Numerical aspects of cloud-GCM interface
much less abundant and therefore do not contribute significantly
to the cloud particle material composition. The 3D GCM employs a sponge layer to stabilize the upper
The resulting cloud particle material volume fractions boundary and basal Rayleigh drag to stabilize the lower bound-
Vs /Vtot , where Vs is the cloud particle volume of a given species ary. While both layers ensure numerical stability, they are also
s and Vtot the total cloud particle volume, average particle size physically justified.
⟨a⟩ [cm], and cloud particle number density nd [cm−3 ] are pro- The cloud model that interfaces with the GCM, similarly,
vided as input for the opacity calculation in the 3D GCM. All requires numerical stabilisation measures that are justified by
cloud particles are assumed to be spherical. physical mechanisms that limit the condensate cloud model. At
the upper boundary of the modelling domain, the growth of
2.2.2. Cloud formation mixing treatment the cloud particles is limited by decreasing collisional rates in
the upper atmosphere. At the lower boundary, cloud growth is
Since our cloud model calculates the cloud structure in the sta- limited by evaporation of cloud materials in the dense and hot
tionary case, the elemental replenishment has to balance the de- deeper atmosphere. To ensure numerical stability, the cloud par-
pletion through nucleation, bulk growth, and gravitational set- ticle opacities were decreased linearly at the upper and lower
tling (Woitke & Helling 2004). For each atmospheric layer, the pressure limit of the expeRT/MITgcm pressure grid until they
rate at which elements are replenished from the deep atmosphere reach 0. This decrease prevents a sudden drop in opacities that
is described by the mixing timescale τmix . This timescale is cal- would otherwise trigger instabilities in the radiative transfer. At
culated using the vertical velocities extracted from the GCM as the top of the atmosphere, the cloud structure is only interpolated
described in Appendix B in Helling et al. (2023). Consistent with in the upper most grid cell at p = 10−5 bar. At the high pres-
our cloud formation model, the mixing timescale is therefore de- sure limit of the cloud structures, the interpolation starts from
rived locally and is not constant through the atmosphere (Helling the the lowest pressure of the cloud structure which is at pres-
et al. 2019b, 2021). Replenishment is also sometimes described sures higher than 10−1 bar.
as a diffusive process which is common for chemical kinetics The expeRT/MITgcm uses a cubed-sphere C32 grid (Adcroft
models where the diffusion coefficient is called Kzz (e.g. Moses et al. 2004). The cells of this grid are more uniformly distributed
et al. 2011; Agundez et al. 2014; Tsai et al. 2017; Baeyens et al. than a longitude-latitude grid which prevents overcrowding at
2021; Konings et al. 2022). Other studies have used Kzz to derive the poles. The resolution of each cell in the C32 grid is approxi-
a replenishment timescale through τdif ∼ H 2p /Kzz where H p [cm] mately 2.8 by 2.8 degrees. The interpolation from the low resolu-
is the scale height (see e.g. Charnay et al. 2018; Helling et al. tion longitude-latitude cloud model grids to the expeRT/MITgcm
2019a). The difference between τdif and τmix is that τdif describes cubed-sphere grid was done using two interpolation steps to re-
the exchange between adjacent atmospheric layers whereas τmix duce interpolation artifacts while keeping the structure of the
describes the rate at which elements are replenished from the low resolution grid. First, a bilinear interpolation to a longitude-
deep atmosphere through convective processes to a given atmo- latitude grid with cell size ∆lon = ∆lat = 3 degrees was per-
spheric layer. formed. Afterwards, a bilinear interpolation to a cubed-sphere
A hallmark of the climate of close-in, tidally locked gas- grid is used. During runtime, clouds are added incrementally to
giants is a strong equatorial wind jet (Showman & Polvani 2011). prevent sharp changes in the opacity structure. Out of the 2000
The high hydrodynamic velocities lead to advection timescales total simulation days, the first 100 simulation days are run with-
that are typically orders of magnitude shorter than the gravita- out clouds. Then, cloud opacities are linearly increased over the
tional settling or diffusion of cloud particles (Woitke & Helling next 100 simulation days until they are fully added at simulation
2003; Powell & Zhang 2024). The nucleation and bulk growth day 200.
The step-wise iterative process applied here is a hands- cloud particles (Helling et al. 2008b) would require to assume a
on version of the iteration processes executed in every com- functional form. We therefore assume monodisperse cloud par-
plex model implemented as, for example, an operator splitting ticles with a local mean particle size derived from the kinetic
method in Helling et al. (2001). Such methods make use of the cloud model as used in Helling et al. (2020). A study on the ef-
time-scale difference that may occur, for example, between con- fect of the assumptions of monodisperse cloud particles can be
densation and hydrodynamical processes. In the case of cloud found in Samra et al. (2020). The absorption and scattering coef-
formation, the formation processes modelled here (nucleation, ficients are then added to the gas-phase opacities of the radiative
grow and evaporation) act on very short time scales since they transfer of expeRT/MITgcm:
predominantly occur in collisionally dominated gases in exo- gas
planet atmospheres. This may, however, change in the upper at- κabs
tot
(λ) = κabs (λ) + κabs
cloud
(λ) (3)
mosphere regions between 10−4 − 10−5 bar where the densities κsca
tot
(λ) =
gas
κsca (λ) + κsca
cloud
(λ) (4)
are so low that e.g. photochemistry affects the gas-phase. In this
work, photochemistry is taken into account in post-processing. gas gas
where κabs [m2 kg−1 ] and κsca [m2 kg−1 ] are the absorption and
The majority of the cloud formation occurs at deeper levels and,
generally, cloud haze models that extend higher up (Steinrueck scattering coefficients for the gas, respectively. κabs
tot
[m2 kg−1 ]
et al. 2021) still find that the hydrodynamic assumption for these and κsca [m kg ] are the total absorption and scattering coeffi-
tot 2 −1
layers is adequate. Further, we aim to resolve here the atmo- cients, respectively.
sphere regions accessible in the infrared by JWST which are
typically at pressures higher than 10−4 bar. 2.3. Stopping the step-wise iteration
Possible long-term changes of the thermodynamic atmo-
sphere structure may be linked to the deep atmosphere which One of the aims of this work is to demonstrate that a 3D atmo-
can take more than 80000 simulation days to converge (Wang & sphere solution including detailed cloud formation is computa-
Wordsworth 2020; Schneider et al. 2022b). Similarly, 1D time tionally feasible for warm Saturn type exoplanets. We do this
dependent cloud models have shown that convergence times can through a step-wise iteration between the two modelling com-
reach up to 8000 simulation days (Woitke et al. 2020). Fully cou- plexes. This enables the full complexity of both the 3D GCM
pled cloudy GCMs are computationally intensive and therefore and the cloud model, to achieve a conceivable accuracy for a
often limited to evaluation times below 5000 simulation days global exoplanet atmosphere structure.
(e.g. Lee et al. 2016, 2017; Roman & Rauscher 2017, 2019; Here, the conceivable accuracy is determined from the ob-
Lines et al. 2018a,b; Roman et al. 2021; Komacek et al. 2022). servational accuracy, hence, the stopping criterion depends on
However, recent run-time improvements of GCMs (Schneider the observational facilities used. For this project, the spectral
et al. 2022c) might allow for longer simulation times in the fu- precision of transmission spectra from the JWST instruments
ture. It should be noted, however, that this work showed only a NIRspec Prism and MIRI LRS after two and after ten transit
minor change in the upper atmosphere structure (p < 1 bar) after measurements are used as the primary stopping criterion. After
2000 days simulation time (see Appendix C). each iteration, a transmission spectrum is calculated (Sect. 2.5)
and compared to the previous iteration step. If the observational
differences between iterations fall below the spectral precision
2.2.4. Cloud opacities (dotted line in Fig. 7), the step-wise iteration is stopped. Since
To calculate the interaction between the photons and the cloud our model iterates between the cloud model and the GCM, the
particle, we use Mie-theory (Mie 1908). This theory is an ana- main goal is to have no observable impact of the changing cloud
lytical solution to the Maxwell equations under the assumption structure on the transmission spectra. The abundances in chemi-
of spherical particle with an effective refractive index ϵeff . cal disequilibrium are calculated in post-processing and the im-
To find the effective refractive index ϵeff of a given mixture pact on the transmission spectra is used as a secondary stopping
of bulk material, we follow the approach of Lee et al. (2016) and criteria.
start by using the Bruggeman mixing rule (Bruggeman 1935). In Additional further stopping criteria may be derived from the
case of non-convergence, we fall back to the Landau-Lifshitz- (T gas , pgas ) - structures and the cloud opacity. As will be demon-
Looyenga method (Looyenga 1965). The homogeneous opacity strated later, once changes in transmission spectra fall below the
values for all species s used in this paper can be found in the observable accuracy, the cloud properties between the two suc-
Appendix B. Using the effective refractive index, Mie-theory is cessive iterations remain similar. This suggests that the still ex-
used to calculate the absorption efficiency Qabs , the scattering ef- isting changes between the (T gas , pgas )-structures, which may be
ficiency Qsca and the anisotropy factor g. From these, the wave- substantial but locally confined, do not affect the gas-phase and
length dependent absorption coefficient κabscloud
(λ) [cm2 kg−1 ] and cloud opacity structure sufficiently enough to change the spec-
the scattering coefficient κsca
cloud
(λ) [cm2 kg−1 ] are calculated: trum beyond spectral precision. Hence, these changes will not
affect the interpretation of the data.
nd
κabs
cloud
(λ) = π⟨a⟩2 Qabs (⟨a⟩, λ, ϵeff ) (1)
ρgas
2.4. Disequilibrium gas-phase chemistry
nd
κsca
cloud
(λ) = π⟨a⟩2 Qsca (⟨a⟩, λ, ϵeff ) (1 − g) (2)
ρgas The disequilibrium chemistry for the H/C/N/O complex of
HATS-6b is modelled to assess how each iteration affects the
where nd [cm−3 ] is the number density of cloud particles, ⟨a⟩ atmospheric chemistry, and as a result the transmission spec-
[cm] the mean cloud particle radius, ρgas [g cm−3 ] the gas den- tra. This is done using an updated version of the chemical net-
sity, and λ [cm] the wavelength of the photon. Our cloud model work STAND2020 (Rimmer & Helling 2016, 2019; Rimmer &
uses the moment method which allows the efficient modelling Rugheimer 2019) in combination with the 1D photochemistry
of heterogeneous cloud particles without an explicit cloud parti- and diffusion code ARGO (Rimmer & Helling 2016). ARGO mod-
cle size distribution. Reconstructing the full size distribution of els chemical transport, wavelength-dependent photochemistry,
and cosmic ray transport by following a parcel of gas as eddy SEPs are not scaled based on the incident angle. The output of
diffusion leads it up and down through the atmosphere (further the STAND2020-ARGO run is the relative concentrations (ni /ngas )
description in e.g. Rimmer & Helling (2016) and Barth et al. of more than 511 gas-phase species. Based on these relative con-
(2021)). STAND2020 is a chemical H/C/N/O network contain- centrations as well as the cloud opacities mentioned previously,
ing the reaction rates for more than 6600 reactions in the temper- the transmission spectra can be produced.
ature range of 100 K to 30000 K. STAND2020 involves all reac-
tions for species containing up to six H, two C, two N, and three
2.5. Transmission radiative transfer
O atoms and also contains reactions with He, Na, Mg, Si, Cl, Ar,
K, Ti, and Fe bearing species. The network has been tested for The transmission spectrum of HATS-6b is produced by adding
the atmospheres of Earth and Jupiter (Rimmer & Helling 2016) the cloud opacities to petitRADTRANS (Mollière et al. 2019,
in addition to a number of hot-Jupiter models (Barth et al. 2021; 2020; Alei et al. 2022), where the cloud opacities from the
Hobbs et al. 2021; Tsai et al. 2023). The chemical network is run micro-pyhsical cloud model are included as opacity source
through the 1D photochemistry and diffusion code, ARGO, which (Sect. 2.2.4). Each transmission spectrum uses the input cloud
models chemical transport and the effect of photochemistry and structure for the GCM, the temperature structure of the GCM,
cosmic rays. and the post-processed gas-phase relative concentrations from
The inputs for ARGO and STAND2020 have been chosen as ARGO. The transmission spectrum of the cloudless GCM run
follows: (iteration 0) is calculated cloud free.
To calculate the transmission spectrum, the terminator region
1. (T gas , pgas ) profiles and vertical eddy diffusion profile: Eight is divided into ten equally spaced, and equally sized grid cells.
different 1D profiles are extracted from the output of the Each cell covers a latitudinal range of ∆lat = 22.5◦ . Choosing
expeRT/MITgcm by averaging over areas of the 3D grid. The a constant longitudinal range would lead to the neglect of data
eight profiles are six terminator regions with the longitudes points within the terminator region, especially close to the poles.
(ϕ = {90◦ , 270◦ }) and latitudes (θ = {0◦ , 23◦ , 68◦ }), and the Therefore, a latitude dependent longitudinal range was chosen
sub-stellar and anti-stellar points. to ensure equally sized grid cells. At the equator the longitu-
2. Atmospheric element abundances: Solar abundances adapted
dinal range is ±∆lon = 10◦ (Lacy & Burrows 2020). The total
for metallicity ([Fe/H] = 0.2) in accordance with the initial
transmission spectrum is then calculated as the average of the
abundances used for the GCM.
3. Stellar XUV spectrum driving photochemistry: Spectrum transmission spectra from all regions:
obtained from the MUSCLES survey where the M1.5V star, v
u
N
t
GJ667C, is chosen as a proxy to HATS6. The spectrum cov- 1 X 2
ers the XUV range and is composed of a combination of Rtot (λ) = R (λ) (5)
N i=1 i
modelled and observed spectra (France et al. 2016; Young-
blood et al. 2016, 2017; Loyd et al. 2016, 2018). The XUV Similar to previous studies (Lee et al. 2019; Baeyens et al. 2021;
spectrum used in this study can be seen in Appendix D.1. Nixon & Madhusudhan 2022), the following gas-phase species
4. Cosmic rays: Implemented based on the ionization rate of
were considered as line opacity species5 : H2 O, CO, CH4 , CO2 ,
low energy cosmic rays (LECR) as explained by Rimmer &
C2 H2 , OH, NH3 , and HCN. Furthermore, H2 and He were con-
Helling (2013) and Barth et al. (2021).
sidered as Rayleigh scattering opacities, and collision-induced
5. Stellar energetic particles: To account for the higher activ-
absorption (CIA) from H2 –H2 and H2 –He.
ity of M-dwarf host stars, SEPs have been included in the
model. The SEPs are introduced by scaling the spectrum of
a solar SEP event to fit the host star based on the X-ray flare 3. Evaluation of the combined model
intensity (method further described in Barth et al. (2021)). A
number of X-ray flare intensities has been reported for M- To study the atmosphere of the warm-Saturn HATS-6b, 6 itera-
dwarf stars (Hünsch et al. 2003; Namekata et al. 2020) rang- tions using expeRT/MITgcm and the kinetic cloud model have
ing from ∼ 0.001 to 0.2 W m−2 at 1 AU. For this study we been conducted. After each iteration of the GCM, the thermo-
chose to implement SEPs corresponding to X-ray flare inten- dynamic structure is used to produce a cloud-structure which is
sities of 0.1 W m−2 at 1AU. The effect of the SEPs will be included in the next iteration of the GCM. The cloud-free simu-
included continuously throughout the run, and not as a finite lations are shown in Sect. 3.1. The effect of clouds are presented
event. in Sect. 3.2. After each GCM run, the disequilibrium chemistry
(Sect. 3.3) is post-processed and the differences in the transmis-
Different locations on the planet will experience a different sion spectrum are determined (Sect. 3.4).
influx of stellar radiation and SEPs, and the model inputs are
therefore varied accordingly. For the sub-stellar point, both the
stellar spectrum and SEPs are included. For the anti-stellar point, 3.1. Cloud-free GCM and post-processed clouds
neither the stellar spectrum nor SEPs are included. For the ter- The first run, i.e. iteration 0, of the GCM was conducted
minator coordinates, only SEPs are included. The reason for in- without clouds. A total of 2000 days were simulated using
cluding SEPs but not the stellar spectrum for the terminator re- expeRT/MITgcm. After 2000 days, there was little change in
gion is that the shallow angle of incidence for radiation from the global average of the (T gas , pgas )-profiles for pressure lay-
the host star causes the radiation to pass through so much atmo- ers lower than 1 bar (see Appendix C). For pressures higher than
spheric layers before it reach the bulk of the 1D simulated atmo- 1 bar, the temperature did not change by more than 20 K from
sphere profile above the terminator that its influence is negligi- the initial conditions for the GCM. It is well known, that the con-
ble. Since XUV radiation is easily scattered by the atmosphere, vergence deeper in the atmosphere is computationally intensive
the stellar spectrum is not included for the terminator regions, and difficult to achieve (Wang & Wordsworth 2020; Skinner &
whereas SEPs have been shown to penetrate deeper into the at-
5
mosphere (Barth et al. 2021) and are therefore included. The References can be found in Table. B.1.

5
Iteration 0
Iteration 1
4 Iteration 2
Iteration 3
3 Iteration 4
Iteration 5
log10 (pgas [bar])
2
1
0
1
2 Substellar point Antistellar point Morning terminator Evening terminator
3 600 1000 1400 600 1000 1400 600 1000 1400 600 1000 1400
Tgas Tgas Tgas Tgas
Fig. 2. (T gas , pgas )-profiles of the sub-stellar point, anti-stellar point, evening terminator, and morning terminator.
Cho 2022; Schneider et al. 2022c,b). In the cloudy iterations (see condensation still occurs in these upper regions - hence the in-
Sect. 3.2), the clouds become opaque for pressures higher than crease in average particle size. Going to deeper layers, the cloud
10−2 bar. Therefore, the atmospheric layers of interest for this particle sizes steadily increase while cloud particle number den-
study can be assumed to be reasonably converged. sities steadily decrease. This inverse correlation between cloud
The (T gas , pgas ) profiles for the sub-stellar point, anti-stellar particle number density and cloud particle size is well known
point, morning terminator, and evening terminator are shown in (see Helling et al. 2021). If more cloud particles are present, the
Fig. 2. The iteration 0 profiles only show minor differences in same mass of condensation material is spread out over more par-
temperature for pressures greater than 10−1 bar. Higher up in ticles hence reducing the average size of the particles.
the atmosphere, the day-side is roughly 200 K hotter than the
night side. The isobaric temperature maps for p = 1.2 × 10−3
bar and p = 1.2 × 10−2 bar can be seen in the left panels of 3.2. Impact of clouds on the atmospheric structure
Fig. 3. The zonal mean wind velocities are shown in the left Iteration 1, 2, 3, 4, and 5 of the GCM are created by including
panel of Fig. 4 which describe the longitudinal average of winds the post-processed cloud opacities from the previous iteration
parallel to the equator and are known to redistribute heat lon- into expeRT/MITgcm as described in Sect. 2.2. The GCM was
gitudinally (Vuitton 2021). Overall, the global temperature and again run for 2000 simulation days. Similar to iteration 0, there
wind velocity structure of HATS-6b show the characteristics of was little change in the global average temperature above 1 bar
a highly irradiated gas-giant including a super-rotating equato- after 2000 simulation days (see Appendix C).
rial jet stream. This jet stream transports heat from the dayside
Going from iteration 0 to iteration 1 highlights the impact
to the nightside and causes a hot spot offset (e.g. Showman &
of clouds on the temperature and wind structure due to the in-
Polvani 2011; Cowan et al. 2012; Dang et al. 2018). In addition
creased local opacities. The most striking difference is a strong
to the equatorial jet stream, two weak polar jet streams can be
temperature inversion for pressures lower than 10−2 bar (higher
observed.
altitudes) on the sub-stellar point and the evening terminator
Using the output from the cloudless GCM (iteration 0) the
(Fig. 2). This increased temperature is also visible in the iso-
cloud structure was post-processed. This cloud structure is la-
baric plots (Fig. 3). At pressures between 10−2 bar and 10 bar,
belled iteration 1 because it will be later used as input for the
the (T gas , pgas ) profiles of iteration 1 show a considerably colder
first cloudy iteration of the GCM (iteration 1). The iteration 1
temperature. This drop can be explained with an anti-greenhouse
cloud profile of HATS-6b shows global cloud coverage. In Fig. 5,
effect (see Sect. 5.1). For pressures higher than 10 bar, no signif-
the nucleation rate J⋆ , cloud particle number density nd , average
icant differences in the (T gas , pgas ) profiles were found. Further-
cloud particle size ⟨a⟩, and cloud mass fraction6 ρd /ρ are shown
more, iteration 1 has a stronger and narrower equatorial wind
for the morning and evening terminator. The sub-stellar and anti-
jet than iteration 0 (see left panel of Fig. 4). This matches the
stellar point can be found in Fig. A.1. The cloud particle proper-
results of Baeyens et al. (2021), who showed that for the temper-
ties of these profiles are largely the same, although there is an in-
ature range of warm Saturns (500 K to 1200 K) an increase in
crease in the nucleation rate for the upper atmosphere (p < 10−3
equilibrium temperature results in a faster and narrower jet. The
bar) for the cooler profiles (morning terminator and anti-stellar
cloud-induced temperature inversion seems to have a similar ef-
point) with a corresponding increases in cloud particle number
fect. Furthermore, the weak polar jets of the iteration 0 are not
density and cloud mass fraction for these points. The formation
observed in iteration 1.
of cloud condensation nuclei (CCNs) happens predominantly in
the upper atmosphere (p < 10−2 ). Though it should be noted that Going from iteration 1 to iteration 2 shows less differences
in the thermal and wind structure of HATS-6b than from itera-
6
Also sometimes called dust-to-gas ratio. tion 0 to 1. However, the temperature profile still changes by up
iteration 0 iteration 1 iteration 2

45°N 45°N 45°N
10 3 bar
0° 0° 0°
45°S 45°S 45°S
90°W 0° 90°E 90°W 0° 90°E 90°W 0° 90°E
45°N 45°N 45°N

10 2 bar
0° 0° 0°
45°S 45°S 45°S
iteration 3 iteration 4 iteration 5

45°N 45°N 45°N
10 3 bar
0° 0° 0°
45°S 45°S 45°S
90°W 0° 90°E 90°W 0° 90°E 90°W 0° 90°E
45°N 45°N 45°N

10 2 bar
0° 0° 0°
45°S 45°S 45°S
90°W 0° 90°E 90°W 0° 90°E 90°W 0° 90°E
400 500 600 700 800 900 1000 1100

T [K]
Fig. 3. Isobaric slices of the expeRT/MITgcm runs at t = 2000 simulation days. The white lines indicate the horizontal wind velocity fields.
to 90 K. The exception to this is the morning terminator around Iteration 3, 4, and 5 all show the same general characteristics
10−4 bar where a temperature increase of up to 350 K can be as iteration 1 and 2: a temperature inversion around 10−3 bar, a
seen. This increase in temperature also reduces the nucleation cooling around 0.1 bar to 1 bar, a narrow equatorial wind jet,
rate which in turn leads to fewer but larger particles (Fig. 5). The and global cloud coverage. However, the temperature between
sudden change in the morning terminator is caused by a dynam- iterations after iteration 2 still vary by up to ∼130 K. Similarly,
ical instability caused by the iteration 2 cloud structure which differences in the temperature structure around 10−3 bar can be
leads to hot air being advected from the day-side into the morn- seen in the isobaric plots (Fig. 3) and in the zonal mean winds
ing terminator. The general thermal instabilities of the morning (Fig. 4). The cloud particle properties on the other hand vary
terminator are discussed in more detail in Sect. 5.2. While an in- little between iterations 3, 4, and 5. In particular the nucleation
stability is present in all cloudy iterations, it is more pronounced rate between iteration 4 and 5 is close to identical. Nevertheless,
in iteration 2. The persistence of this instability is a result of the there are still changes in the cloud particle number density and
cloud structures being static. Since no other iteration shows a average size between iteration 4 and 5. As mentioned previously,
similar behaviour, the temperature increase in the morning ter- the goal of this work is to find a solution within the observational
minator of iteration 2 is considered an artefact of the specific accuracy of JWST and therefore only observable changes within
configurations of the static clouds. the atmospheric structure matter. Whether the changes described
here have an observable effect is analysed in Sect. 3.4.

Iteration 0 Iteration 1 Iteration 2 Iteration 3 Iteration 4 Iteration 5

5
5000
4
3 4000
Mean zonal wind [m/s]

log10(pgas [bar])
2 3000
1
2000
0
1 1000
2 0
50 0 50 50 0 50 50 0 50 50 0 50 50 0 50 50 0 50
latitude latitude latitude latitude latitude latitude
Fig. 4. Zonally averaged zonal wind velocities at t = 2000 simulation days.
3.3. Disequilibrium chemistry spectrum of the cloudy iterations included the cloud structures
as well. All 6 spectra are shown in the top panel of Fig. 7. The
To assess the impact of the kinetic gas-phase chemistry and residuals between iterations 1, 2, 3, 4, and 5 are shown in the bot-
photo-chemistry on the observable atmosphere, and more specif- tom panel of the same figure. Additionally, the spectral precision
ically on transmission spectra of HATS-6b, the disequilibrium for the JWST instruments NIRSpec Prism (Birkmann et al. 2022;
chemistry was modelled for each iteration at six coordinates Ferruit et al. 2022) and MIRI LRS (Kendrew et al. 2015, 2016)
along the terminator region (longitudes (ϕ = {90◦ , 270◦ }) and lat- are shown. The spectral precision was calculated using PandExo
itudes (θ = {0◦ , 23◦ , 68◦ })). The resulting gas concentration pro- (Batalha et al. 2017) assuming a spectral resolution7 of R = 10
files were averaged over the six coordinates. The relative concen- for two and ten observed transits. If the residuals are significantly
trations ni /ngas of H, H2 , CO, CO2 , H2 O, CH4 , HCN, and NH3 larger than the spectral precision, the differences between the it-
can be seen in the top panel of Fig. 6. These eight molecules have erations are observable. If the residuals are of the same order or
been chosen due to their high concentrations in the atmospheres lower, observing the differences between iterations will be chal-
or because they are some of the more interesting atmospheric lenging.
species when looking at the effects of external radiation (e.g. The transmission spectra of iteration 0 and 1 show clear dif-
Barth et al. (2021); Baeyens et al. (2022)). The concentration ferences due to the presence of clouds in iteration 1. Compared
profiles show very little variation among the five iterations with to iteration 0, the molecular features of iteration 1 are muted
clouds, whereas the difference between the cloudless (iteration for wavelengths below 2 µm. Around 10 µm, an additional in-
0) and cloudy iterations (1 - 5) is somewhat larger. The largest crease of the relative transit depth can be seen which is caused
differences between the cloudless and cloudy iterations occurs by silicate species within the cloud particles. Both these effects
between pgas ∼ 10−2 −100 bar (top of Fig. 6). This pressure range are known and expected if clouds are present in exoplanet atmo-
corresponds to a cloud-induced cooling which is present in iter- spheres (Wakeford & Sing 2015; Powell et al. 2019; Grant et al.
ations 1 - 5. Some species (e.g. NH3 , CO2 , and CH4 ) also show 2023).
a difference between cloudy and cloudless at lower pressures
The residuals in transit depth from iterations 1 to 2 and 2 to
(Top panel of Fig. 6) corresponding to a cloud-induced heating
3 are around 25 ppm for wavelengths between 0.7 µm to 10 µm.
in iterations 1 - 5. The cloud-induced heating in the upper atmo-
This can be related to a shift in the cloud top, which is defined
sphere in combination with a cooling of the lower atmosphere
as the pressure level at which clouds become optically thick (see
comprises the temperature inversion (see Fig. 2).
e.g. Estrela et al. 2022). From iteration 3 to 4, the transit depth
The chemical variations along the equator are illustrated in differs by roughly 30 ppm for wavelengths between 0.7 µm to
the middle (sub-stellar and anti-stellar point) and bottom panel 10 µm. For the wavelength range of MIRI LRS, the differences
(morning and evening terminator). Comparing the top and mid- between iterations is always below the spectral accuracy even
dle panel we notice that the relative concentration of all species if 10 transits are observed. From iteration 4 to 5, the cloud top
show significantly larger differences between the day- and night- again changes but by less than 25 ppm. The residuals between
side (the solid and the dashed lines in the middle panel) than iteration 4 to 5 are now consistently below the spectral precision
the differences between the different iterations (all lines in the of NIRSpec Prism, as illustrated by the black data points. The
top panel). We will further describe the chemistry observed for only exception to this are the CH4 features around 2 µm to 4 µm,
HATS-6b in Sect. 4.1. that still have residuals around 30 ppm. Since the differences
between the transmission spectra of iteration 3, 4, and 5 are close
3.4. Observable differences between iterations to or below the spectral precision of JWST NIRSpec Prism, we
stop our iterative procedure after iteration 5.
The goal of the step-wise iteration approach is to reach observa- Furthermore, comparing the residuals to the differences be-
tional accuracy here chosen to be the JWST NIRspec and MIRI tween the morning and evening terminator (see Fig. 9) shows
LRS. For HATS-6b, this was reached after six iterations, i.e. af- that the differences between the limbs are generally larger then
ter iteration 5.
All transmission spectra were produced using the temper- 7
This resolution was chosen to maximise the spectral precision while
ature structure of the GCM and the chemistry of ARGO. The keeping enough data points to detect gas-phase features.

5.0 5.0
Iteration 1 Iteration 1
4.5 Iteration 2
Iteration 3
4.5 Iteration 2
Iteration 3
4.0 Iteration 4
Iteration 5
4.0 Iteration 4
Iteration 5
log10 (pgas [bar])
log10 (pgas [bar])

3.5 3.5
3.0 3.0
2.5 2.5
morning terminator evening terminator
2.0 cloud mass fraction 2.0 cloud mass fraction
1.5 1.5
1.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
ρd /ρ [×10 −3 ] ρd /ρ [×10 −3 ]
5.0 5.0
4.5 4.5
4.0 4.0
log10 (pgas [bar])
log10 (pgas [bar])

3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 cloud particle density 1.5 cloud particle density
1.00 1 2 3 4 5 6 7 8 1.00 1 2 3 4 5 6 7 8
log10 (nd [cm−3 ]) log10 (nd [cm−3 ])
5.0 5.0
4.5 morning terminator 4.5 evening terminator
average size average size
4.0 4.0
log10 (pgas [bar])
log10 (pgas [bar])
3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 1.5
1.03.0 2.5 2.0 1.5 1.0 0.5 0.0 1.03.0 2.5 2.0 1.5 1.0 0.5 0.0
® ®
log10 ( a [µm]) log10 ( a [µm])
5.0 5.0
4.5 4.5
4.0 4.0
log10 (pgas [bar])
log10 (pgas [bar])
3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 nucleation rate 1.5 nucleation rate
1.012 8 4 0 1.012 8 4 0
log10 (J [cm−3 s−1 ]) log10 (J [cm−3 s−1 ])
Fig. 5. Cloud particle properties of the step-wise iterated cloud structure for the warm Saturn example HATS-6b. Iteration 5 is shown as a solid line
to highlight the final result. Left: morning terminator. Right: evening terminator. Top: cloud mass fraction ρd /ρ. Upper middle: cloud particle
number density nd . Lower middle: average cloud particle size ⟨a⟩. Bottom: Nucleation rate J⋆ . The sub- and anti-stellar point can be found in
Fig A.1.

5 4. The combined 3D model for the warm Saturn

H HATS-6b that orbits an M-dwarf
4 H2
3 CO The final combined 3D model for the warm Saturn HATS-6b
CO2
log10 (pgas [bar])
H2 O
that orbits an M-dwarf has been demonstrated in Sect 3. The two
2 CH4 isobar maps at pressure ranges that are accessible to transmis-
HCN sion spectroscopy are shown in the bottom right panel of Fig. 3
1 NH3 (iteration 5). HATS-6b is affected by a strong equatorial jet that
0 Iteration 5
Iteration 4
reaches down to depth of pgas ≈ 1bar. Horizontally converg-
ing flows8 further contribute to the global mixing of the atmo-
1 Iteration 3
Iteration 2 sphere in the pole regions (e.g. Baeyens et al. 2021). The tem-
Iteration 1
2 Iteration 0 perature inversion is well represented in the upper atmosphere
with a temperature difference of ≈400K as compared to lower
14 12 10 8 6 4 2 0 regions where pgas ≈ 10−2 bar. Sect. 4.1 presents the final results
log10 (ni /ntot ) from the combined model and Sect. 4.2 shows the transmission
spectrum as observable with JWST NIRspec Prism and MIRI
5 LRS.
4
3 H 4.1. The atmospheric structure of HATS-6b
log10 (pgas [bar])
H2
2 CO HATS-6b is a warm Saturn with a hydrogen dominated atmo-
CO2 sphere that is assumed to be oxygen rich in this study. Figure 8
1 H2 O
presents the solution of the combined 3D model. The morning
CH4
0 HCN (left) and the evening (right) 1D terminator profiles were se-
NH3 lected to demonstrate the cloud and kinetic chemistry results in
1 Sub-stellar enough detail to gain insights into the physical reasons for the
Sub-stellar no SEPs
2 Anti-stellar
differences between the respective transmission spectra that are
shown in Sect. 4.2.
14 12 10 8 6 4 2 0 Figure 8 (top) shows that the local gas temperatures differ
log10 (ni /ntot ) between the morning and the evening terminator which has di-
rect consequences for the cloud formation efficiency. By com-
5 paring the nucleation rates (J∗ [cm−3 s−1 ]), we notice that the
4 cooler morning terminator (left) forms more cloud particles than
the hotter evening terminator. Consequently, the mean particles
3 H size, ⟨a⟩ [µm], is smaller in the respective low gas pressure re-
log10 (pgas [bar])
H2
2 CO
gion of the morning terminator. The lower panels in Fig. 8 show
CO2 which material groups dominate the cloud particle composition,
1 H2 O and hence, their opacities. Fig. A.2 contains the detailed break-
CH4 down of all 16 modelled bulk materials. Materials that are deter-
0 HCN
mined by low-abundance elements like Cl, Mn etc. can be ther-
NH3
1 mally stable but appear with ∼1% volume contribution (Helling
Evening terminator et al. 2017; Woitke et al. 2018).
2 Morning terminator
Figure 8 demonstrates that it is reasonable to expect the
14 12 10 8 6 4 2 0 cloud particles that form in the atmosphere of a warm Saturn
log10 (ni /ntot ) like HATS-6b to be made of a mix of materials and that their
mean particles sizes changes throughout the atmosphere. It fur-
Fig. 6. Concentrations of non-equilibrium gas species for the warm ther demonstrates that the bulk cloud mass (ρd /ρ, black line,
Saturn example HATS-6b. Top: Terminator region (both morning and middle panel) is not necessarily equally distributed through the
evening) averaged over all six coordinates for all iterations. Middle: atmospheres and differs between the two terminators, hence,
Sub- and anti-stellar point for the final iteration (Iteration 5). Bottom: emphasising the terminator asymmetry. Comparing to the ma-
Morning and evening terminators for the final iteration (Iteration 5). terial volume fractions, Vs /Vgas , in the lower panels demon-
strates that the bulk mass comes from silicates (MgSiO3 [s],
Mg2 SiO4 [s], CaSiO3 [s], Fe2 SiO4 [s]) and metal oxides (SiO[s],
SiO2 [s], MgO[s], FeO[s], Fe2 O3 [s]) because Mg/Si/Fe/O are the
most abundant elements amongst those considered here. It is,
hence, the result of mass conservation. Far less important for the
cloud material composition are the high temperature condensates
(TiO2 [s], Al2 O3 [s]) and salts (KCl[s], NaCl[s]) for the same rea-
the residuals between iterations 4 an 5. This holds true for son. Since most dominant condensation species are oxygen bear-
the cloud continuum (limb ≈ 250 ppm versus residuals < 30 ing species and oxygen is far more abundant then Ti, Mg, Si, and
ppm), the methane features (limbs ≈ 200 ppm versus residuals
< 40 ppm), and the water features (limb ≈ 100 ppm versus resid- 8
Convergence is here defined as ∇ · v < 0, where v is the horizontal
uals ≈ 30 ppm). velocity vector.

32500
32000
R2p/R2s [ppm]
31500
31000 Iteration 0 Iteration 2 Iteration 4
Iteration 1 Iteration 3 Iteration 5
30500
100 MIRI 10
1
residuals 1 2 residuals 3 4 NIRSpec Prism Ntransit = 10 LRS Ntransit = 10
75
| (R2p/R2s )| [ppm]
residuals 2 3 residuals 4 5 NIRSpec Prism Ntransit = 2 MIRI LRS Ntransit = 2

50
25
0
1 10
[ m]
Fig. 7. Comparison of the transmission spectra for the warm Saturn example HATS-6b for λ = 0.3 µm to 25 µm Top: Transmission spectrum for
each iteration. Bottom: Absolute residuals between subsequent iterations and the spectral precision for JWST observations with NIRspec Prism
and MIRI LRS for 2 and 10 transits.
0 Morning 0 Evening
Tgas 1400 Tgas 1400
2 Jtot 2 Jtot
1200 1200
log10 (J [cm s ])
log10 (J [cm−3 s−1 ])
JTiO2 JTiO2
−3 −1
JSiO JSiO
4 1000 4 1000
Tgas [K]
Tgas [K]
6 800 6 800
8 600 8 600
10 400 10 400
0 3 0 3
2 2 2 2
4 4
log10 (ni /ntot )
log10 (ni /ntot )

ρd /ρ [×10 −3 ]
ρd /ρ [×10 −3 ]
1 1
6 6
0 0
8 CO2
8 CO2
H HCN H HCN
10 H2 H2 O NH3 1 10 H2 H2 O NH3 1
CO CH4 ρd /ρ CO CH4 ρd /ρ
1.0 2 1.0 2
log10 (pgas [bar])
®
a HTC 4 log10 (pgas [bar])
®
a HTC 4
0.8 MO Salts 3 0.8 MO Salts 3
SC 2 SC 2
log10 ( a [µm])
log10 ( a [µm])
0.6 1 0.6 1
Vs /Vtot
Vs /Vtot
®
®
0 0
0.4 1 0.4 1
0.2 2 0.2 2
3 3
0.0 4 0.0 4
5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0
log10 (pgas [bar]) log10 (pgas [bar])
Fig. 8. Cloud structure of the morning (left) and evening (right) terminator of HATS-6b. Top: gas temperature T gas and total J⋆ and JTiO2 , and JSiO .
Middle: gas-phase relative concentrations ni /ngas and cloud mass fraction ρd /ρ. Bottom: cloud particle volume fractions Vs /Vtot and mean cloud
particle size ⟨a⟩ (MO = Metal oxides, SC = Silicates, HTC = High temperature condensates). Salts are a minor component and have a volume
fraction close to 0.

H2 O NH3 CH4 cloud Si HCN

33000
32500
R2p/R2s [ppm]
32000
Iteration 5 Evening
31500 clouds Ntransit = 2
Morning Ntransit = 10
31000
1 10
[ m]
Fig. 9. Transmission spectra for the warm Saturn HATS-6b for λ = 0.3 µm to 25 µm including the total spectrum of the final iteration (iteration 5)
as well as the morning and evening terminator separately. Also shown are the contributions of selected gas-phase specie, the contribution of the
Si-bearing cloud particle materials, and the spectral accuracy of the JWST instruments NIRSpec Prism and MIRI LRS for 2 and 10 transits.
Fe, these condensation species do not significantly impact the which for M-dwarf stars has been reported to range from 0.001
gas-phase chemistry of non Ti, Mg, Si, and Fe bearing species. to 0.2 Wm−2 at 1AU. In this study we scale our SEP spectrum
As introduced in Sect. 3.3, the disequilibrium chemistry of based on a X-ray flare intensity of 0.1 Wm−2 at 1AU, indicat-
HATS-6b is shown in Figure 6. The middle and bottom panels ing that the amount of SEPs could be significantly higher than
show the relative concentrations for a selection of molecular gas what we show, leading to a greater impact on the disequilibrium
species for the last iteration at four coordinates along the equa- gas-phase chemistry.
tor of the planet: sub-stellar point, anti-stellar point, and the two
terminators. The variations along the equator are caused both by
differences in the (T gas , pgas ) profiles for the four coordinates, 4.2. Transmission spectra
and by differences in the stellar XUV radiation and the SEPs.
The sub-stellar point is irradiated both by XUV radiation and Transmission spectra are used to study the chemical composi-
SEPs, the terminators are irradiated only by SEPs, and the anti- tion and atmospheric asymmetries of gaseous planets. To deter-
stellar point is irradiated by neither. Comparing the four cases mine the concentrations of known non-equilibrium species (e.g.
it can be seen that the sub-stellar point and the two terminators CH4 and HCN), the H/C/N/O complex of the gas-phase is de-
are relatively similar, whereas the anti-stellar point differs sig- termined using ARGO (Sect. 3.3). As the methods used in this
nificantly from the rest. By comparing the sub- and anti-stellar study may help to efficiently interpret the data of JWST and fu-
point, we notice a steep decrease in the concentration of many ture space missions, the focus was set on spectral features rel-
of the gas species at the sub-stellar point (incl. H2 , CH4 , and evant for the JWST instruments NIRspec and MIRI. In the fol-
NH3 ) in the upper atmosphere, indicating a break down of these lowing, the near-IR transmission spectra that also include non-
molecules through photolysis by the XUV radiation or ionisation equilibrium species like CH4 and HCN for a wavelength range
by the SEPs. Other species (such as H, HCN, and partly CO and of λ = 0.3 µm to 12 µm are presented.
CO2 ) show an increase in concentration for the sub-stellar point To create the transmission spectrum of HATS-6b, we used
compared to the anti-stellar point as we move further up into the (T gas , pgas ) profiles from the GCM, the cloud opacities from
the atmosphere, indicating that these are positively influenced the kinetic cloud model and the disequilibrium chemistry from
by photochemical reactions. Comparing model runs for the sub- ARGO. The full transmission spectra is shown in Fig. 9. Also
stellar point with and without SEPs (middle panel), we notice a shown are the morning and evening spectra separately as well
significant contribution by the SEPs on the gas-phase concentra- as the cloud and the gas-phase contributions. The transmission
tions and that this effect reaches far down into the atmosphere. spectrum shows a roughly flat spectra for wavelengths below
The middle and bottom panels show that the sub-stellar point 2 µm, owing to scattering which is typical for cloudy exoplanets
with SEPs bear a strong resemblance to the terminator regions, (Bean et al. 2010; Crossfield et al. 2013; Kreidberg et al. 2014;
which could indicate that SEPs can have a larger effect than the Knutson et al. 2014; Sing et al. 2015, 2016; Benneke et al. 2019).
XUV radiation on the concentrations of the gas species, includ- Observing HATS-6b at these wavelengths would allow to deter-
ing observationally interesting species such as CH4 and HCN. mine the height of the cloud deck (Mukherjee et al. 2021). Above
The bottom panel shows that the differences between the termi- 3 µm the gas-phase species start to contribute significantly to
nator regions are generally higher at lower pressures, with the the spectrum. Especially CH4 (around 3 µm and 8 µm), H2 O
exception of H that also show differences deeper into the atmo- (around 6 µm), and HCN (around 10.5 µm) can be seen. How-
sphere. ever, the clouds still heavily mute the spectral features, making
As mentioned in Sect. 1, M-dwarf stars such as HATS-6 are them harder to detect.
known to have higher magnetic activities compared to more mas- Around 10 µm, the cloud opacities show a broad silicate fea-
sive stars like the Sun. This increased activity leads to increased ture from the Si-bearing cloud particle materials. This feature
amounts of SEPs. As explained in Sect. 2.4, the amount of SEPs originates from Si-O vibrations (Wakeford & Sing 2015). De-
has been found to scale with the X-ray flare intensity of the star, tecting this feature could confirm the presence of silicon-bearing
cloud particle materials in HATS-6b as has been done previously absorption of stellar radiation in the optical wavelength range.
for WASP-17b (Grant et al. 2023) and WASP-107b (Dyrek et al. These absorption-driven temperature inversions have been ob-
2023). served in (ultra) hot Jupiters (Haynes et al. 2015; Yan et al. 2020,
HATS-6b itself may reveal differences in cloud coverage and 2022). HATS-6b orbits an M-dwarf that emits less flux in the op-
chemistry between the morning and evening terminator similar tical wavelength range than a hotter star (Lothringer & Barman
to WASP 39b (Carone et al. 2023; Espinoza et al. 2024). The 2019). Furthermore, Hu & Ding (2011) predicted that an exo-
cloud deck is slightly higher at the evening terminator compared planet with a CO2 dominated atmosphere orbiting an M-dwarf
to the morning terminator leading to a ≈ 250 ppm offset at the stars would also have an anti-greenhouse effect mainly due to
short wavelengths (0.3 µm < λ < 2 µm). The observable differ- highly efficient Rayleigh scattering of CO2 at the top of the at-
ence in the gas-phase is mainly caused by CH4 which has ≈ 200 mosphere. In this study, however, CO2 is a minor species and
ppm higher signal at the evening terminator than at the morning Rayleigh scattering is dominated by hydrogen and helium. In
terminator. H2 O has ≈ 100 ppm higher signal in the evening ter- our simulation of HATS-6b, no temperature inversion is present
minator than the morning terminator. At the morning terminator, in the GCM runs without clouds. Therefore, we are confident in
the contribution of chemical species to the transmission spec- our conclusion that the temperature inversion is indeed caused
trum can vary strongly with latitude (see Fig. A.3). Especially at mainly by scattering of stellar irradiation at the top of the ex-
±45◦ of the morning terminator, stronger transmission features tended cloud deck similar to the anti-greenhouse effect observed
of CH4 and HCN arise as a consequence of the Rossby gyres in Titan due to scattering at the top of the atmosphere due to
that represent particularly cold regions in the atmosphere (see hazes.
Fig. 3). Warm Saturns and Jupiters in the temperature range be- To observe a temperature inversion directly, emission from
tween 800 K and 1200 K typically exhibit extended Rossby wave the lower atmosphere needs to be observable (Gandhi & Mad-
gyres at the morning terminator (e.g. Baeyens et al. 2021, Fig. 2). husudhan 2019). The strong cloud coverage will likely block all
Thus, a study of morning terminator chemistry in warm Saturns emissions from the lower atmosphere and make direct detections
needs to take into account special locations like the dynamically of the temperature inversion unlikely. However, HATS-6b lies in
cool Rossby gyres to correctly infer global atmosphere proper- a temperature regime where quenching is expected to affect the
ties. chemistry (e.g. Baeyens et al. 2021). Fig. 7 and 9 shows a CH4
feature that is visible despite the cloud layer, which indicates that
5. Discussion the CH4 extends into the upper layers of the atmosphere where it
is exposed to photolysis and high degrees of SEPs. Since CH4 is
In this work, a 3D climate model in combination with a mi- very susceptible to photo-chemical reactions (Moses et al. 2011;
crophysical cloud model was implemented to capture the feed- Line et al. 2011; Baeyens et al. 2021, 2022; Konings et al. 2022),
back between the heating, the 3D climate, and cloud formation it will most likely break down in the upper atmosphere, and in
in its full complexity. The model was applied to the particu- order to maintain a visible CH4 feature, a constant up-welling
lar example of a warm Saturn around an M-dwarf, HATS-6b. from the lower protected part of the atmosphere would be neces-
Many exoplanet theories predict temperature inversions in ex- sary. A visible CH4 feature could therefore indicate that vertical
oplanet atmospheres due to various reasons. In Sect. 5.1, the mixing has connected the observable gas-phase chemistry above
strong cloud-induced temperature inversion of HATS-6b is dis- the clouds to deeper atmosphere layers, thereby forming a probe
cussed. Furthermore, we discuss the dynamics of the termina- through the temperature inversion and into the layers cooled by
tors in Sect. 5.2. While several studies have modelled the climate the anti-greenhouse effect (Agundez et al. 2014; Fortney et al.
and cloud structure of warm Saturns, there have been no detailed 2020). In addition, this work confirmed the influence of stellar
studies of warm Saturns around M-dwarf stars like HATS-6b so energetic particles (SEP) on the day side chemistry of a planet
far. Therefore in Sect. 5.3, the results for HATS-6b are compared around a relatively active planet as was already pointed out by
to grid studies including warm Saturns and detailed models of Barth et al. (2021) for HD 189733b.
other Saturn-mass planets similar to HATS-6b.
5.2. Dynamics of the terminator regions
5.1. Anti-greenhouse effect
In Sect. 3, we seen that the morning terminator of iteration 2 de-
The clouds in HATS-6b have considerable cloud particle sizes velops a temperature spike in the upper atmosphere (see Fig. 2).
and number densities for pressures lower than 10−3 bar. The Since no other iteration shows such a spike, we can conclude
high cloud deck leads to scattering and absorption of the incom- that it is caused by the impact of the specific configuration of the
ing short wavelength radiation on cloud particles. This reduces static cloud opacities of iteration 2 on the temperature and wind
the radiative heating of the lower atmospheric layers while si- structure of HATS-6b. A closer inspection of the potential tem-
multaneously heating up the upper layers. Both effects lead to perature profiles across the morning and evening terminators of
a temperature inversion in the layers where clouds are located all iterations (Fig. 10) reveals a potential temperature anomaly at
and a cooling of the layers below. This effect is called the anti- the equator for the morning terminator in iteration 2. The poten-
greenhouse effect and was first observed in Titan (McKay et al. tial temperature is linked to atmospheric stability. If the poten-
1991). It has also been theoretically predicted for exoplanet at- tial temperature increases monotonically with height, the atmo-
mospheres with either an extended cloud deck or photochemical sphere is dynamically stable (e.g. Holton & Hakim 2013).
hazes (Heng et al. 2012; Morley et al. 2012; Steinrueck et al. The equatorial morning terminator is a region, where the
2023). super-rotating wind jet advects colder air from the night side to
In hotter exoplanets, temperature inversions are not caused the day-side that is heated by the cloud feedback (see Fig. 3).
by scattering from high-altitude clouds but by gas-phase species The advection of cold air results in a minimum in potential tem-
like TiO and VO (Hubeny et al. 2003; Fortney et al. 2008) or perature between -40 degree and +40 degree latitude. In contrast
AlO, CaO, NaH, and MgH (Gandhi & Madhusudhan 2019). to that, the potential temperature across the evening terminator is
These species heat up the upper atmosphere due to very efficient more barotropic. Here, the wind jet broadens and encompasses
homogenous cloud coverage. Our results suggest that at least

5 for exoplanets around M-dwarf stars like HATS-6b an anti-
log10(pgas)
greenhouse effect is expected for these class (i) planets caused

4
by the clouds.
3 Christie et al. (2022) studied the impact of clouds on the cli-
5 50 0 50 50 0 50 mate of the warm Neptune GJ 1214b. They considered phase
log10(pgas)
4 equilibrium clouds within the EddySed model (Ackerman &

Marley 2001). Their cloud model parameterises the settling of
3 cloud particles with the parameter fsed , where a lower fsed re-
5 50 0 50 50 0 50 sults in vertically extended clouds. In contrast to our work, and
log10(pgas)
4 matching other studies of cloud composition of GJ 1214b (Gao

& Benneke 2018; Ormel & Min 2019), they consider only KCl
3 and ZnS as cloud particle material. Christie et al. (2022) found
5 50 0 50 50 0 50 that clouds can cause cooling in the lower atmosphere (10−2 bar
log10(pgas)
4 < p < 1 bar) which matches our findings. This temperature de-
crease is most pronounced for higher metallicities and lower fsed .
3 However, they did not find a significant temperature increase in
5 50 0 50 50 0 50 the upper atmosphere, as we find in our work.
log10(pgas)
4 Another planet similar to HATS-6b, is the Saturn-mass exo-

planet WASP-39b that is part of the JWST Early Release Science
3 program (Feinstein et al. 2023; Ahrer et al. 2023; Rustamkulov
5 50 0 50 50 0 50 et al. 2023; Alderson et al. 2023; JWST Transiting Exoplanet
log10(pgas)
4 Community Early Release Science-Team et al. 2023). WASP-

39b is at the upper limit of the class(i) cloud temperature regime
3 identified by Helling et al. (2023) with T eq ∼ 1100 K. Cloud-
50 0 50 50 0 50 less GCMs of WASP-39b show relatively small day-night tem-
Latitude Latitude perature differences of ∆T ∼ 500 K (Carone et al. 2023; Lee
et al. 2023), similar to the cloudless simulations of HATS-6b.
10000 20000 30000 40000 50000
Tpot Post-processed cloud modelling by Carone et al. (2023) pre-
dicts global cloud coverage of WASP-39b. Pre-JWST observa-
Fig. 10. Potential temperature cross section across the morning (left tions pointed towards a relatively cloud free atmosphere, with
panels) and evening terminators (right panels) for iterations 0 to 5 (from estimates of atmospheric metallicities ranging from 0.1 × −100×
top to bottom). solar (Sing et al. 2016; Nikolov et al. 2016; Fischer et al. 2016;
Wakeford et al. 2018). Observations with JWST revised these
earlier observational results and indicate the presence of clouds
almost the whole evening terminator region (Fig. 3), resulting in and a ≳ 10× solar metallicity, with some models favouring
a more uniform advection of warm air towards the night side. inhomogeneous cloud coverage (Feinstein et al. 2023). Thus,
The aim of this work is to identify a stable climate solution JWST observations of WASP-39b demonstrated that it is pos-
for a cloudy warm Saturn orbiting an M-dwarf star. However, it sible to break the high metallicity and cloudiness degeneracy
may be worthwhile investigating if the strong temperature gra- that also plagued warm Saturns pre-JWST (Carone et al. 2021).
dients of the morning terminator at the boundary of the super- JWST observations of WASP-39b could reveal cloud asymme-
rotating jet give rise to instabilities and thus variations at the tries between the morning and evening terminator as predicted
morning terminators as is evident in iteration 2. Observations by Carone et al. (2023). Also in this work, we find a tendency
that aim to disentangle the atmospheric properties of the morn- for cloud asymmetries between both terminators (see Sect. 4.2.)
ing and evening terminator of gas-giants with JWST could con-
firm the stronger dynamical variability of the morning termina-
tor compared to the evening terminator. However for HATS-6b, 6. Conclusion
we do not see observable differences due to variations across the In this paper, we explored the atmospheric, micro-physical cloud
terminators. and gas-phase structure of the warm Saturn HATS-6b, orbit-
ing an M-dwarf using a combined model in the form of step-
5.3. Comparison to other models wise iterations between a detailed cloud formation description
and expeRT/MITgcm, a 3D GCM with full radiative transfer
The cloud structure and climate of HATS-6b was simulated as a and deep atmosphere extension. We demonstrated how the com-
first example of a warm Saturn around an M-dwarf host star. To bined modelling approach can help to support the interpretation
extend our findings from HATS-6b to general warm Saturns, the of the data from space missions (e.g. JWST) for warm Saturn
results are compared to other studies focusing on including the type planets.
3D cloud and climate of warm Saturn type exoplanets. We find a significant cloud coverage on both the day and
Helling et al. (2023) conducted a grid study of post- night side of HATS-6b, which is to be expected for the gener-
processed cloud structures for temperatures between 400 K and ally low temperatures (500 K < T < 1200 K) of warm Saturn
2600 K for F, G, K and M-dwarf stars. For exoplanets with type exoplanets. On the day-side and evening terminator, we find
an equilibrium temperature of 700 K around M-dwarf stars, that clouds cause a significant temperature inversion in the up-
they predict a strong uniform cloud coverage. The equilibrium per atmosphere (p < 10−2 bar). This also results in a globally
temperature and cloud structure of HATS-6b falls with their reduced temperature of the deeper atmospheric layers (10−2 bar
"class (i)" planets which are characterised by global and mostly < p < 1 bar) which is consistent with an anti-greenhouse effect.
Including clouds in the GCM also leads to a stronger and nar- Baeyens, R., Decin, L., Carone, L., et al. 2021, MNRAS, 505, 5603
rower equatorial jet stream. The transmission spectra of HATS- Baeyens, R., Konings, T., Venot, O., Carone, L., & Decin, L. 2022, MNRAS,
512, 4877
6b shows a characteristically flat spectrum in the optical. In Barber, R. J., Strange, J. K., Hill, C., et al. 2014, MNRAS, 437, 1828
the infrared, however, major molecular absorption features like Barth, P., Helling, C., Stüeken, E. E., et al. 2021, MNRAS, 502, 6201
CO2 , CH4 and H2 O are visible despite the global cloud coverage. Batalha, N. E., Mandell, A., Pontoppidan, K., et al. 2017, PASP, 129, 064501
Around 10 µm, a silicate cloud feature is predicted. The small Bean, J. L., Miller-Ricci Kempton, E., & Homeier, D. 2010, Nature, 468, 669
Begemann, B., Dorschner, J., Henning, T., et al. 1997, ApJ, 476, 199
radii of M-dwarf stars compared to more massive stars leads the Benneke, B., Knutson, H. A., Lothringer, J., et al. 2019, Nat. Astron., 3, 813
to a ’magnifying effect’ of spectral features within transmission Birkmann, S. M., Ferruit, P., Giardino, G., et al. 2022, A&A, 661, A83
spectra. This makes planets like HATS-6b prime targets for deci- Borysow, A. 2002, A&A, 390, 779
phering gas chemistry and cloud compositions for warm Saturn Borysow, A. & Frommhold, L. 1989, ApJ, 341, 549
Borysow, A., Frommhold, L., & Moraldi, M. 1989, ApJ, 336, 495
type exoplanets. For wavelengths up to 8 µm, it may even be Borysow, A., Jorgensen, U. G., & Fu, Y. 2001, JQSRT, 68, 235
possible to identify morning and evening terminator differences Borysow, J., Frommhold, L., & Birnbaum, G. 1988, ApJ, 326, 509
with the CH4 feature around 3 µm. Bromley, S. T., Martín, J. C. G., & C. Plane, J. M. 2016, PCCP, 18, 26913
When iterating between the GCM and the cloud model, the Bruggeman, D. a. G. 1935, Annalen der Physik, 416, 636
Buchhave, L. A., Bitsch, B., Johansen, A., et al. 2018, ApJ, 856, 37
differences in the atmospheric structure of HATS-6b quickly Carone, L., Baeyens, R., Mollière, P., et al. 2020, MNRAS, 496, 3582
drop below the observational accuracy of the JWST NIRSpec Carone, L., Lewis, D. A., Samra, D., Schneider, A. D., & Helling, C. 2023,
Prism and MIRI LRS. For HATS-6b this was reached after 5 it- ArXiv, 2301.08492
erations. Even after 1 iteration (GCM → Clouds → GCM) the Carone, L., Mollière, P., Zhou, Y., et al. 2021, A&A, 646, A168
Cañas, C. I., Kanodia, S., Bender, C. F., et al. 2022, AJ, 164, 50
GCM predicts an atmospheric structure that differs by less than Chan, Y. M. & Dalgarno, A. 1965, Proc. Phys. Soc., 85, 227
50ppm to iteration 5. Furthermore, differences in the transmis- Charnay, B., Bézard, B., Baudino, J. L., et al. 2018, The Astrophysical Journal,
sion spectra between the iteration steps are smaller than the dif- 854, 172, publisher: IOP ADS Bibcode: 2018ApJ...854..172C
ferences between the morning and evening terminator. Therefore Charnay, B., Meadows, V., & Leconte, J. 2015a, ApJ, 813, 15
Charnay, B., Meadows, V., Misra, A., Leconte, J., & Arney, G. 2015b, ApJ, 813,
a combined model that enables the full complexity of all mod- L1
elling components (here 3D GCM and cloud formation) proves Christie, D. A., Mayne, N. J., Gillard, R. M., et al. 2022, MNRAS, 517, 1407
useful for the scientific interpretation of observational data for Christie, D. A., Mayne, N. J., Lines, S., et al. 2021, MNRAS, 506, 4500
larger sets of exoplanets to be studied with CHEOPS, JWST, Coles, P. A., Yurchenko, S. N., & Tennyson, J. 2019, MNRAS, 490, 4638
Cowan, N. B., Machalek, P., Croll, B., et al. 2012, ApJ, 747, 82
and also ELT, PLATO, and Ariel in the future. Crossfield, I. J. M. 2015, PASP, 127, 941
Warm Saturn type exoplanets around M-dwarfs are ideal Crossfield, I. J. M., Barman, T., Hansen, B. M. S., & Howard, A. W. 2013, A&A,
candidates to identify cloud particle composition by observing 559, A33
their spectral features and identify the cloud-induced thermal in- Dalgarno, A. & Williams, D. A. 1962, ApJ, 136, 690
Dang, L., Cowan, N. B., Schwartz, J. C., et al. 2018, Nat. Astron., 2, 220
version that arises on these planets. In particular limb asymme- Demory, B.-O., Wit, J. d., Lewis, N., et al. 2013, ApJL, 776, L25
try studies will allow us to study differences in the chemistry Dominik, C., Sedlmayr, E., & Gail, H.-P. 1993, A&A, 277, 578
and cloud top which are caused by the feedback of clouds on the Dorschner, J., Begemann, B., Henning, T., Jaeger, C., & Mutschke, H. 1995,
atmospheric structure. Using the combined model allows us to A&A, 300, 503
Dyrek, A., Min, M., Decin, L., et al. 2023, Nature, 625, 51
account for the full physical complexity of each model in a com- Espinoza, N., Steinrueck, M. E., Kirk, J., et al. 2024, Nature, 632, 1017
putationally feasible manner and enables a detailed interpreta- Estrela, R., Swain, M. R., & Roudier, G. M. 2022, ApJL, 941, L5
tion of observational data within the accuracy of JWST NIRSpec Feautrier, P. 1964, Comptes Rendus Academie des Sciences Paris, 258, 3189
and MIRI LRS. Feinstein, A. D., Radica, M., Welbanks, L., et al. 2023, Nature, 614, 670
Ferruit, P., Jakobsen, P., Giardino, G., et al. 2022, A&A, 661, A81
Acknowledgements. S.K. N.BM., A.D.S, F.A, H.LM., L.C., Ch.H., L.D., and Fischer, P. D., Knutson, H. A., Sing, D. K., et al. 2016, ApJ, 827, 19
U.G.J. acknowledge funding from the European Union H2020-MSCA-ITN-2019 Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678,
under grant agreement no. 860470 (CHAMELEON). D.S. and D.A.L. acknowl- 1419
edge financial support and use of the computational facilities of the Space Re- Fortney, J. J., Visscher, C., Marley, M. S., et al. 2020, AJ, 160, 288
search Institute of the Austrian Academy of Sciences. We acknowledged the France, K., Loyd, R. O. P., Youngblood, A., et al. 2016, ApJ, 820, 89
computation support at CPH and at the IWF Graz through the Vienna Science Fraschetti, F., Drake, J. J., Alvarado-Gómez, J. D., et al. 2019, ApJ, 874, 21
Cluster (VSC project 72245). LD acknowledges support from the KU Leuven Gail, H.-P. & Sedlmayr, E. 1986, A&A, 166, 225
IDN grant IDN/19/028 grant Escher. U.G.J. further acknowledges funding from Gail, H.-P. & Sedlmayr, E. 1988, A&A, 206, 153
the Novo Nordisk Foundation Interdisciplinary Synergy Programme grant no. Gandhi, S. & Madhusudhan, N. 2019, MNRAS, 485, 5817
NNF19OC0057374. To achieve the scientific results presented in this article we Gao, P. & Benneke, B. 2018, ApJ, 863, 165
made use of the Python, especially the NumPy (Harris et al. 2020) and Mat- Gao, P., Thorngren, D. P., Lee, E. K. H., et al. 2020, Nat. Astron., 4, 951
plotlib (Hunter 2007). The post-processing of GCM datawas performed with Gobrecht, D., Hashemi, S. R., Plane, J. M. C., et al. 2023, A&A, 680, A18
gcm-toolkit (Schneider et al. 2022a). Gobrecht, D., Plane, J. M. C., Bromley, S. T., et al. 2022, A&A, 658, A167
Grant, D., Lewis, N. K., Wakeford, H. R., et al. 2023, ApJL, 956, L29
Gray, D. F. 2008, The Observation and Analysis of Stellar Photospheres (Cam-
bridge University Press)
Harada, C. K., Kempton, E. M.-R., Rauscher, E., et al. 2021, ApJ, 909, 85
References Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357
Ackerman, A. S. & Marley, M. S. 2001, ApJ, 556, 872 Hartman, J. D., Bayliss, D., Brahm, R., et al. 2015, ApJ, 149, 166
Adcroft, A., Campin, J.-M., Hill, C., & Marshall, J. 2004, Mon. Weather Rev., Haynes, K., Mandell, A. M., Madhusudhan, N., Deming, D., & Knutson, H.
132, 2845 2015, ApJ, 806, 146
Agundez, M., Parmentier, V., Venot, O., Hersant, F., & Selsis, F. 2014, A&A, Helling, C., Dehn, M., Woitke, P., & Hauschildt, P. H. 2008a, ApJ, 675, L105
564, A73 Helling, C. & Fomins, A. 2013, Phil. Trans. R. Soc. A, 371, 20110581
Ahrer, E.-M., Stevenson, K. B., Mansfield, M., et al. 2023, Nature, 614, 653 Helling, C., Gourbin, P., Woitke, P., & Parmentier, V. 2019a, A&A, 626, A133
Alderson, L., Wakeford, H. R., Alam, M. K., et al. 2023, Nature, 614, 664 Helling, C., Iro, N., Corrales, L., et al. 2019b, A&A, 631, A79
Alei, E., Konrad, B. S., Angerhausen, D., et al. 2022, A&A, 665, A106 Helling, C., Kawashima, Y., Graham, V., et al. 2020, A&A, 641, A178
Arfaux, A. & Lavvas, P. 2024, MNRAS, 530, 482 Helling, C., Klein, R., Woitke, P., Nowak, U., & Sedlmayr, E. 2004, A&A, 423,
Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481 657
Azzam, A. A. A., Tennyson, J., Yurchenko, S. N., & Naumenko, O. V. 2016, Helling, C., Lee, E., Dobbs-Dixon, I., et al. 2016, MNRAS, 460, 855
MNRAS, 460, 4063 Helling, C., Lewis, D., Samra, D., et al. 2021, A&A, 649, A44
Baeyens, R. 2021, PhD thesis, book Title: On the Climate and Chemistry of Helling, C., Oevermann, M., Lüttke, M. J. H., Klein, R., & Sedlmayr, E. 2001,
Irradiated Exoplanet Atmospheres A&A, 376, 194

Helling, C., Samra, D., Lewis, D., et al. 2023, A&A, 671, A122 Palik, E. D. 1985, Academic Press Handbook Series, New York: Academic Press,
Helling, C., Tootill, D., Woitke, P., & Lee, E. K. H. 2017, A&A, 603, A123 1985, edited by Palik, Edward D.
Helling, C. & Woitke, P. 2006, A&A, 455, 325 Parmentier, V., Fortney, J. J., Showman, A. P., Morley, C., & Marley, M. S. 2016,
Helling, C., Woitke, P., & Thi, W.-F. 2008b, A&A, 485, 547 ApJ, 828, 22
Heng, K., Hayek, W., Pont, F., & Sing, D. K. 2012, MNRAS, 420, 20 Parmentier, V., Line, M. R., Bean, J. L., et al. 2018, A&A, 617, A110
Henning, T., Begemann, B., Mutschke, H., & Dorschner, J. 1995, A&A Supple- Parmentier, V., Showman, A. P., & Fortney, J. J. 2021, MNRAS, 501, 78
Parmentier, V., Showman, A. P., & Lian, Y. 2013, A&A, 558, A91
ment Series, 112, 143 Pascucci, I., Testi, L., Herczeg, G. J., et al. 2016, ApJ, 831, 125
Hobbs, R., Rimmer, P. B., Shorttle, O., & Madhusudhan, N. 2021, MNRAS, 506, Pelletier, S., Benneke, B., Ali-Dib, M., et al. 2023, Nature, 619, 491
3186 Perez-Becker, D. & Showman, A. P. 2013, ApJ, 776, 134
Holton, J. R. & Hakim, G. J. 2013, An introduction to dynamic meteorology, Piskunov, N. E., Kupka, F., Ryabchikova, T. A., Weiss, W. W., & Jeffery, C. S.
fifth edition edn. (Amsterdam: Academic Press) 1995, A&A Supplement Series, 112, 525
Hu, Y. & Ding, F. 2011, A&A, 526, A135 Polyansky, O. L., Kyuberis, A. A., Zobov, N. F., et al. 2018, MNRAS, 480, 2597
Posch, T., Kerschbaum, F., Fabian, D., et al. 2003, ApJ Supplement Series, 149,
Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011 437
Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90 Powell, D., Louden, T., Kreidberg, L., et al. 2019, ApJ, 887, 170
Hünsch, M., Weidner, C., & Schmitt, J. H. M. M. 2003, A&A, 402, 571 Powell, D. & Zhang, X. 2024, Two-Dimensional Models of Microphysical
Juncher, D., Jørgensen, U. G., & Helling, C. 2017, A&A, 608, A70 Clouds on Hot Jupiters I: Cloud Properties, arXiv:2404.08759 [astro-ph]
JWST Transiting Exoplanet Community Early Release Science-Team, Ahrer, E.- Powell, D., Zhang, X., Gao, P., & Parmentier, V. 2018, ApJ, 860, 18
M., Alderson, L., et al. 2023, Nature, 614, 649 Querry, M. R. 1998, Optical Constants of Minerals and Other Materials from the
Kataria, T., Sing, D. K., Lewis, N. K., et al. 2016, ApJ, 821, 9 Millimeter to the Ultraviolet (Chemical Research, Development & Engineer-
Kempton, E. M.-R., Bean, J. L., & Parmentier, V. 2017, ApJL, 845, L20 ing Center, U.S. Army Armament Munitions Chemical Command)
Kendrew, S., Scheithauer, S., Bouchet, P., et al. 2016, Proceedings of the SPIE, Richard, C., Gordon, I. E., Rothman, L. S., et al. 2012, JQSRT, 113, 1276
Rimmer, P. B. & Helling, C. 2013, ApJ, 774, 108
9904, 990443 Rimmer, P. B. & Helling, C. 2016, ApJS, 224, 9
Kendrew, S., Scheithauer, S., Bouchet, P., et al. 2015, PASP, 127, 623 Rimmer, P. B. & Helling, C. 2019, ApJS, 245, 20
Kiefer, S., Gobrecht, D., Decin, L., & Helling, C. 2023, A&A, 671, A169 Rimmer, P. B. & Rugheimer, S. 2019, Icarus, 329, 124
Kiefer, S., Lecoq-Molinos, H., Helling, C., Bangera, N., & Decin, L. 2024, A&A, Robbins-Blanch, N., Kataria, T., Batalha, N. E., & Adams, D. J. 2022, ApJ, 930,
682, A150 93
Knutson, H. A., Charbonneau, D., Allen, L. E., et al. 2007, Nature, 447, 183 Rodgers-Lee, D., Rimmer, P. B., Vidotto, A. A., et al. 2023, MNRAS, 521, 5880
Roman, M. & Rauscher, E. 2017, ApJ, 850, 17
Knutson, H. A., Dragomir, D., Kreidberg, L., et al. 2014, ApJ, 794, 155 Roman, M. & Rauscher, E. 2019, ApJ, 872, 1
Komacek, T. D., Fauchez, T. J., Wolf, E. T., & Abbot, D. S. 2020, ApJ, 888, L20 Roman, M. T., Kempton, E. M.-R., Rauscher, E., et al. 2021, ApJ, 908, 101
Komacek, T. D. & Showman, A. P. 2016, ApJ, 821, 16 Rowe, J. F., Matthews, J. M., Seager, S., et al. 2008, ApJ, 689, 1345
Komacek, T. D., Showman, A. P., & Parmentier, V. 2019, ApJ, 881, 152 Rustamkulov, Z., Sing, D. K., Mukherjee, S., et al. 2023, Nature, 614, 659
Komacek, T. D., Showman, A. P., & Tan, X. 2017, ApJ, 835, 198 Samra, D., Helling, C., Chubb, K. L., et al. 2022, A&A
Komacek, T. D., Tan, X., Gao, P., & Lee, E. K. H. 2022, ApJ, 934, 79 Samra, D., Helling, C., & Min, M. 2020, A&A, 639, A107
Savel, A. B., Kempton, E. M.-R., Malik, M., et al. 2022, ApJ, 926, 85
Konings, T., Baeyens, R., & Decin, L. 2022, A&A, 667, A15 Schneider, A. D., Baeyens, R., & Kiefer, S. 2022a, gcm_toolkit, DOI:
Kreidberg, L., Bean, J. L., Désert, J.-M., et al. 2014, Nature, 505, 69 10.5281/zenodo.7116787
Köhn, C., Helling, C., Enghoff, M. B., et al. 2021, A&A, 654, A120 Schneider, A. D., Carone, L., Decin, L., Jørgensen, U. G., & Helling, C. 2022b,
Lacy, B. I. & Burrows, A. 2020, ApJ, 905, 131 A&A, 666, L11
Lecoq-Molinos, H., Gobrecht, D., Sindel, J. P., Helling, C., & Decin, L. 2024, Schneider, A. D., Carone, L., Decin, L., et al. 2022c, A&A, 664, A56
A&A, 690, A34 Showman, A. P., Fortney, J. J., Lian, Y., et al. 2009, ApJ, 699, 564
Lee, E., Taylor, J., Grimm, S. L., et al. 2019, MNRAS, 487, 2082 Showman, A. P. & Guillot, T. 2002, A&A, 385, 166
Showman, A. P. & Polvani, L. M. 2011, ApJ, 738, 71
Lee, E. K. H. 2023, MNRAS, 2918 Sindel, J. P., Gobrecht, D., Helling, C., & Decin, L. 2022, A&A, 668, A35
Lee, E. K. H., Blecic, J., & Helling, C. 2018, A&A, 614, A126 Sing, D. K., Fortney, J. J., Nikolov, N., et al. 2016, Nature, 529, 59
Lee, E. K. H., Dobbs-Dixon, I., Helling, C., Bognar, K., & Woitke, P. 2016, Sing, D. K., Wakeford, H. R., Showman, A. P., et al. 2015, MNRAS, 446, 2428
A&A, 594, A48 Skinner, J. W. & Cho, J. Y.-K. 2022, MNRAS, 511, 3584
Lee, E. K. H., Helling, C., Giles, H., & Bromley, S. T. 2015, A&A, 575, A11 Sousa-Silva, C., Al-Refaie, A. F., Tennyson, J., & Yurchenko, S. N. 2015, MN-
Lee, E. K. H., Parmentier, V., Hammond, M., et al. 2021, MNRAS, 506, 2695 RAS, 446, 2337
Steinrueck, M. E., Koskinen, T., Lavvas, P., et al. 2023, ApJ, 951, 117
Lee, E. K. H., Tsai, S.-M., Hammond, M., & Tan, X. 2023, A&A, 672, A110 Steinrueck, M. E., Parmentier, V., Showman, A. P., Lothringer, J. D., & Lupu,
Lee, E. K. H., Wood, K., Dobbs-Dixon, I., Rice, A., & Helling, C. 2017, A&A, R. E. 2019, ApJ, 880, 14
601, A22 Steinrueck, M. E., Showman, A. P., Lavvas, P., et al. 2021, MNRAS, 504, 2783
Li, G., Gordon, I. E., Rothman, L. S., et al. 2015, ApJ Supplement Series, 216, Suto, H., Sogawa, H., Tachibana, S., et al. 2006, MNRAS, 370, 1599
15 Tan, X. & Showman, A. P. 2017, ApJ, 835, 186
Lin, A. S. J., Libby-Roberts, J. E., Alvarado-Montes, J. A., et al. 2023, AJ, 166, Tan, X. & Showman, A. P. 2021, MNRAS, 502, 2198
Thorngren, D., Gao, P., & Fortney, J. J. 2019, ApJ, 884, L6
90 Tsai, S.-M., Lee, E. K. H., Powell, D., et al. 2023, Nature, 617, 483
Line, M. R. & Parmentier, V. 2016, ApJ, 820, 78 Tsai, S.-M., Lyons, J. R., Grosheintz, L., et al. 2017, ApJS, 228, 20
Line, M. R., Vasisht, G., Chen, P., Angerhausen, D., & Yung, Y. L. 2011, ApJ, Turbet, M., Bolmont, E., Chaverot, G., et al. 2021, Nature, 598, 276
738, 32 Venot, O., Rocchetto, M., Carl, S., Roshni Hashim, A., & Decin, L. 2016, ApJ,
Lines, S., Manners, J., Mayne, N. J., et al. 2018a, MNRAS, 481, 194 830, 77
Lines, S., Mayne, N. J., Boutle, I. A., et al. 2018b, A&A, 615, A97 Vuitton, V. 2021, in Encyclopedia of Geology (Elsevier), 217–230
Wakeford, H. R. & Sing, D. K. 2015, A&A, 573, A122
Looyenga, H. 1965, Physica, 31, 401 Wakeford, H. R., Sing, D. K., Deming, D., et al. 2018, ApJ, 155, 29
Lothringer, J. D. & Barman, T. 2019, ApJ, 876, 69 Wang, H. & Wordsworth, R. 2020, ApJ, 891, 7
Loyd, R. O. P., France, K., Youngblood, A., et al. 2016, ApJ, 824, 102 Webber, M. W., Lewis, N. K., Marley, M., et al. 2015, ApJ, 804, 94
Loyd, R. O. P., France, K., Youngblood, A., et al. 2018, ApJ, 867, 71 Wende, S., Reiners, A., Seifahrt, A., & Bernath, P. F. 2010, A&A, 523, A58
McKay, C. P., Pollack, J. B., & Courtin, R. 1991, Science, 253, 1118 Witte, S., Helling, C., Barman, T., Heidrich, N., & Hauschildt, P. H. 2011, A&A,
McKemmish, L. K., Masseron, T., Hoeijmakers, H. J., et al. 2019, MNRAS, 488, 529, A44
Woitke, P. & Helling, C. 2003, A&A, 399, 297
2836 Woitke, P. & Helling, C. 2004, A&A, 414, 335
McKemmish, L. K., Yurchenko, S. N., & Tennyson, J. 2016, MNRAS, 463, 771 Woitke, P., Helling, C., & Gunn, O. 2020, A&A, 634, A23
Mie, G. 1908, Annalen der Physik, 330, 377 Woitke, P., Helling, C., Hunter, G. H., et al. 2018, A&A, 614, A1
Mignon, L., Meunier, N., Delfosse, X., et al. 2023, A&A, 675, A168 Wong, I., Chachan, Y., Knutson, H. A., et al. 2022, Astron. J., 164, 30
Min, M., Ormel, C. W., Chubb, K., Helling, C., & Kawashima, Y. 2020, A&A, Yan, F., Pallé, E., Reiners, A., et al. 2020, A&A, 640, L5
Yan, F., Reiners, A., Pallé, E., et al. 2022, A&A, 659, A7
642, A28 Yang, J., Boué, G., Fabrycky, D. C., & Abbot, D. S. 2014, ApJ, 787, L2
Mollière, P., Stolker, T., Lacour, S., et al. 2020, A&A, 640, A131 Youngblood, A., France, K., Loyd, R. O. P., et al. 2017, ApJ, 843, 31
Mollière, P., Wardenier, J. P., Boekel, R. v., et al. 2019, A&A, 627, A67 Youngblood, A., France, K., Loyd, R. O. P., et al. 2016, ApJ, 824, 101
Morley, C. V., Fortney, J. J., Marley, M. S., et al. 2012, ApJ, 756, 172 Yurchenko, S. N., Amundsen, D. S., Tennyson, J., & Waldmann, I. P. 2017,
Moses, J. I., Visscher, C., Fortney, J. J., et al. 2011, ApJ, 737, 15 A&A, 605, A95
Mukherjee, S., Batalha, N. E., & Marley, M. S. 2021, ApJ, 910, 158 Yurchenko, S. N., Mellor, T. M., Freedman, R. S., & Tennyson, J. 2020, MNRAS,
Namekata, K., Maehara, H., Sasaki, R., et al. 2020, PASJ, 72, 68 496, 5282
Nikolov, N., Sing, D. K., Gibson, N. P., et al. 2016, ApJ, 832, 191 Zamyatina, M., Christie, D. A., Hébrard, E., et al. 2024, MNRAS, 529, 1776
Zeidler, S., Posch, T., & Mutschke, H. 2013, A&A, 553, A81
Nixon, M. C. & Madhusudhan, N. 2022, ApJ, 935, 73 Zeidler, S., Posch, T., Mutschke, H., Richter, H., & Wehrhan, O. 2011, A&A,
Olson, G. L., Auer, L. H., & Buchler, J. R. 1986, JQSRT, 35, 431 526, A68
Ormel, C. W. & Min, M. 2019, A&A, 622, A121

Table B.1. References for cross sections of gas-phase species. Appendix D: Input SED for ARGO
Gas-phase species Reference The input SED used for the photo-chemistry of ARGO is shown
H2 O (Polyansky et al. 2018) in Fig. D.1. The SED is obtained from the MUSCLES survey
CO2 (Yurchenko et al. 2020) where the star GJ667C is chosen as the closest analogue to
CH4 (Yurchenko et al. 2017) HATS-6. The SED is combined from different spectra achieved
NH3 (Coles et al. 2019) through observations and models, as indicated in the sections
CO (Li et al. 2015) separated by the dashed lines (see e.g. France et al. (2016) for
H2 S (Azzam et al. 2016) details). The spectrum has been binned and adapted to the de-
HCN (Barber et al. 2014) sired wavelength range 1-10,000 nm.
PH3 (Sousa-Silva et al. 2015)
TiO (McKemmish et al. 2019)
VO (McKemmish et al. 2016)
FeH (Wende et al. 2010)
Na (Piskunov et al. 1995)
K (Piskunov et al. 1995)
Table B.2. References for the cloud material opacities.
Material species Reference

TiO2 [s] (Rutile) Zeidler et al. (2011)
SiO2 [s] (alpha-Quartz) Palik (1985), Zeidler et al. (2013)
SiO[s] (polycrystalline) Philipp in Palik (1985)
MgSiO3 [s] (grass) Dorschner et al. (1995)
Mg2 SiO4 [s] (crystalline) Suto et al. (2006)
MgO[s] (Cubic) Palik (1985)
Fe[s] (metallic) Palik (1985)
FeO[s] (amorphous) Henning et al. (1995)
Fe2 O3 [s] (amorphous) A.H.M.J. Triaud (unpublished)
Fe2 SiO4 [s] (amorphous) Dorschner et al. (1995)
FeS[s] (amorphous) Henning (unpublished)
CaTiO3 [s] (amorphous) Posch et al. (2003)
Al2 O3 [s] (grass) Begemann et al. (1997)
KCl [s] (cubic) Palik (1985); Querry (1998)
NaCl [s] (cubic) Palik (1985); Querry (1998)
Appendix A: Additional cloud structure plots

The cloud particle properties of the sub-stellar and anti-stellar
point for all iterations are shown in Fig. A.1. Additionally for
iteration 5, the cloud particle material composition of the sub-
stellar point, anti-stellar point, morning terminator and evening
terminator are shown in Fig. A.2. Lastly, the transmission spec-
trum of each terminator grid cell (lat = { -68, -23, 0, 23, 68}) of
iteration 5 before averaging is shown in Fig. A.3.
Appendix B: Opacity data

The references for the absorption cross sections used in the ra-
diative transfer of the GCM and transmission spectra calcula-
tions are listed in Table B.1. The references for the cloud par-
ticle material opacities are listed in Table B.2. Due to missing
data, CaSiO3 is treated as vacuum.
Appendix C: GCM convergence tests

To test the convergence of the expeRT/MITgcm runs, we anal-
yse the time dependent temperature evolution of the GCM. The
results can be seen in Fig. C.1. The results show that the temper-
ature structure of all three runs are reasonably converged in the
upper layers of the GCM.
5.0 5.0
Iteration 1 Iteration 1
4.5 Iteration 2
Iteration 3
4.5 Iteration 2
Iteration 3
4.0 Iteration 4
Iteration 5
4.0 Iteration 4
Iteration 5
log10 (pgas [bar])
log10 (pgas [bar])

3.5 3.5
3.0 3.0
2.5 2.5
Sub-stellar point Anti-stellar point
2.0 cloud mass fraction 2.0 cloud mass fraction
1.5 1.5
1.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 1.00.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
ρd /ρ [×10 −3 ] ρd /ρ [×10 −3 ]
5.0 5.0
4.5 4.5
4.0 4.0
log10 (pgas [bar])
log10 (pgas [bar])

3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 cloud particle density 1.5 cloud particle density
1.00 1 2 3 4 5 6 7 8 1.00 1 2 3 4 5 6 7 8
log10 (nd [cm−3 ]) log10 (nd [cm−3 ])
5.0 5.0
4.5 Sub-stellar point 4.5 Anti-stellar point
average size average size
4.0 4.0
log10 (pgas [bar])
log10 (pgas [bar])
3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 1.5
1.03.0 2.5 2.0 1.5 1.0 0.5 0.0 1.03.0 2.5 2.0 1.5 1.0 0.5 0.0
® ®
log10 ( a [µm]) log10 ( a [µm])
5.0 5.0
4.5 4.5
4.0 4.0
log10 (pgas [bar])
log10 (pgas [bar])
3.5 3.5
3.0 3.0
2.5 2.5
2.0 2.0
1.5 nucleation rate 1.5 nucleation rate
1.012 8 4 0 1.012 8 4 0
log10 (J [cm−3 s−1 ]) log10 (J [cm−3 s−1 ])
Fig. A.1. Cloud particle properties of the step-wise iterated cloud structure for the warm Saturn example HATS-6b. Iteration 5 is shown as a solid
line to highlight the final result. Left: sub-stellar point. Right: anti-stellar point. Top: cloud mass fraction ρd /ρ. Upper middle: cloud particle
number density nd . Lower middle: average cloud particle size ⟨a⟩. Bottom: Nucleation rate J⋆ . The morning and evening terminator can be found
in Fig 5.

TiO2[s] SiO2[s] CaSiO3[s] MgSiO3[s] Fe[s] FeS[s] Fe2SiO4[s] KCl[s]

SiO[s] CaTiO3[s] MgO[s] Mg2SiO4[s] FeO[s] Fe2O3[s] Al2O3[s] NaCl[s]
5 5
4 4
log10 (pgas [bar])
log10 (pgas [bar])

3 3
2 2
1 1
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Vs /Vtot Vs /Vtot
5 5
4 4
log10 (pgas [bar])
log10 (pgas [bar])

3 3
2 2
1 1
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Vs /Vtot Vs /Vtot
Fig. A.2. Cloud particle material composition of iteration 5. 1st row: Sub-stellar point. 2nd row: Anti-stellar point. 3rd row: Evening terminator.
4th row: Morning terminator. Here the cloud profiles only reach down to 0.3 bar.
33000 lat = 0 lat = 23 lat = 45 lat = 68

Morning
R2p/R2s [ppm]
32500
32000
31500
33000 100 101
Evening
R2p/R2s [ppm]
32500
32000
31500
1 10
[ m]
Fig. A.3. Transmission spectrum for each terminator grid cell. Top: Morning terminator. Bottom: Evening terminator.

Iteration 0 Iteration 1 Iteration 2 Iteration 3 Iteration 4 Iteration 5

2000
1750
-4
-3 1500
-2 1250
log10(pgas [bar])
Time [s]
-1 1000
0 750
1
500
2
250
600 850 1100 1400 600 850 1100 1400 600 850 1100 1400 600 850 1100 1400 600 850 1100 1400 600 850 1100 1400
T [K] T [K] T [K] T [K] T [K] T [K]
Fig. C.1. Evolution of the global average temperature for every 100 days of the expeRT/MITgcm runs.
16
Chandra EUV HST/Ly PHOENIX
log10(Actinic flux [cm 2 s 1 Å 1])
14
12
10
6
1.0 1.5 2.0 2.5 3.0 3.5 4.0
log10( [Å])
Fig. D.1. Input spectral energy distribution (SED) for host star expressed in actinic flux.

Under The Magnifying Glass: A Combined 3D Model Applied To Cloudy Warm Saturn Type Exoplanets Around M-Dwarfs

Uploaded by

Copyright:

Available Formats

Under The Magnifying Glass: A Combined 3D Model Applied To Cloudy Warm Saturn Type Exoplanets Around M-Dwarfs

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Under The Magnifying Glass: A Combined 3D Model Applied To Cloudy Warm Saturn Type Exoplanets Around M-Dwarfs

Uploaded by

Copyright:

Available Formats

Astronomy & Astrophysics manuscript no.

current_version ©ESO 2024

Under the magnifying glass: A combined 3D model applied to

Received ...; accepted ...

1. Introduction tunity for in-detail cloud and chemistry characterisation across

be combined, which is computationally intensive. There are cur-

Article number, page 3 of 21

Article number, page 6 of 21

iteration 0 iteration 1 iteration 2

45°S 45°S 45°S

90°W 0° 90°E 90°W 0° 90°E 90°W 0° 90°E

45°N 45°N 45°N

45°S 45°S 45°S

iteration 3 iteration 4 iteration 5

45°S 45°S 45°S

90°W 0° 90°E 90°W 0° 90°E 90°W 0° 90°E

45°N 45°N 45°N

45°S 45°S 45°S

90°W 0° 90°E 90°W 0° 90°E 90°W 0° 90°E

400 500 600 700 800 900 1000 1100

Article number, page 8 of 21

Iteration 0 Iteration 1 Iteration 2 Iteration 3 Iteration 4 Iteration 5

Mean zonal wind [m/s]

Article number, page 9 of 21

log10 (pgas [bar])

log10 (pgas [bar])

log10 (pgas [bar])

log10 (pgas [bar])

Article number, page 10 of 21

5 4. The combined 3D model for the warm Saturn

Article number, page 11 of 21

residuals 2 3 residuals 4 5 NIRSpec Prism Ntransit = 2 MIRI LRS Ntransit = 2

log10 (J [cm−3 s−1 ])

log10 (ni /ntot )

Article number, page 12 of 21

H2 O NH3 CH4 cloud Si HCN

homogenous cloud coverage. Our results suggest that at least

greenhouse effect is expected for these class (i) planets caused

4 equilibrium clouds within the EddySed model (Ackerman &

4 matching other studies of cloud composition of GJ 1214b (Gao

4 Another planet similar to HATS-6b, is the Saturn-mass exo-

4 Community Early Release Science-Team et al. 2023). WASP-

Article number, page 16 of 21

Article number, page 17 of 21

Table B.2. References for the cloud material opacities.

Material species Reference

Appendix A: Additional cloud structure plots

Appendix B: Opacity data

Appendix C: GCM convergence tests

log10 (pgas [bar])

log10 (pgas [bar])

log10 (pgas [bar])

log10 (pgas [bar])

Article number, page 19 of 21

TiO2[s] SiO2[s] CaSiO3[s] MgSiO3[s] Fe[s] FeS[s] Fe2SiO4[s] KCl[s]

log10 (pgas [bar])

log10 (pgas [bar])

33000 lat = 0 lat = 23 lat = 45 lat = 68

Article number, page 20 of 21