On the dayside of the rocky planet TOI-561 b, temperatures are lower than expected for a world so close to its star. Its nightside also appears to emit more heat than a bare, airless surface would be expected to retain. Observations from the James Webb Space Telescope point to a possible explanation: a reflective atmosphere that sends much of the incoming starlight back into space while allowing some heat to circulate around the planet.
TOI-561 b is an ultra-short-period rocky planet, a world so close to its star that its surface is expected to be intensely hot. Its mass is about 2.24 times Earth’s, and its radius is about 1.42 times Earth’s, according to measurements cited in the study. At such close range, the planet receives enough stellar energy to make its dayside a natural place to look for signs of molten rock and a possible atmosphere made from vaporized surface material.
But the temperature pattern measured by JWST does not fit the simplest picture of a dark, airless planet that absorbs nearly all the light falling on it.
In observations made over roughly 37 hours in May 2024, the telescope tracked the planet through several stages of its orbit, measuring changes in the combined infrared light from the planet and its star. The resulting phase curve, a record of how the system’s brightness changes as different parts of the planet face the telescope, allowed the researchers to estimate temperatures on the dayside and nightside.
The dayside’s estimated brightness temperature was about 2,050 kelvin. That is substantially below the roughly 3,200 K expected for a planet with no reflected starlight and no redistribution of heat to its nightside. The nightside, meanwhile, had an estimated brightness temperature of about 1,100 K.
Together, the measurements suggest that TOI-561 b reflects a considerable amount of incoming starlight and moves some heat away from its dayside. The researchers argue that a reflective atmosphere is the explanation most consistent with the observations. Their analysis does not, however, establish the atmosphere’s composition or prove that any particular cloud or gas is present.
Tracking a planet through its orbit
The observations used JWST’s Near-Infrared Spectrograph, or NIRSpec, with its G395H grating. The instrument recorded light between roughly 3 and 5 micrometers while the planet completed a little more than one orbit during the observing sequence.
That continuous view captured four secondary eclipses and three transits. A secondary eclipse occurs when a planet passes behind its star, temporarily removing the planet’s contribution from the light measured by the telescope. A transit happens when the planet crosses in front of the star. Between these events, the amount of the planet’s dayside and nightside visible to the telescope changes, producing the phase curve.
The technique gives astronomers a way to follow the planet’s changing infrared emission rather than relying only on brief snapshots of its dayside. As the planet moves around its star, the changing brightness can reveal how much energy is emitted from different regions and whether the hottest part of the dayside is displaced from the point directly facing the star.
But measuring a small planet’s changing infrared signal is difficult. The telescope records light from the star as well as the planet, and the star itself varies. Stellar surface activity can produce changes in brightness that resemble or obscure the planet’s much smaller signal. Instrument behavior can also introduce unwanted changes into the measurements.
The team therefore had to separate three contributions: the planet’s changing brightness, variations from the star, and instrumental effects.
Separating the planet’s signal from stellar variability
A central part of the analysis was accounting for changes caused by stellar granulation. Granulation is associated with the movement of hot and cooler gas near a star’s surface. It produces brightness fluctuations that can appear as noise in a time series, even when the telescope is measuring the star very precisely.
The researchers modeled this variability alongside the planet’s phase curve and the detector systematics, rather than treating every change in brightness as part of the planetary signal. They used a Gaussian process, a statistical method that can describe correlated variations over time, including a component designed to represent the characteristic behavior of stellar granulation.
The observations were recorded on two detectors, NRS1 and NRS2. The analysis treated their measurements separately, allowing the team to compare the planetary signals and the noise properties in the two detector channels. They also produced white-light curves, which combine the measured light across the full wavelength range, and seven spectroscopic light curves covering narrower wavelength intervals.
The team tested the results using two independent data-reduction approaches, Eureka! and ExoTiC JEDI. The resulting measurements were consistent to within one standard deviation in the key comparisons reported by the researchers.
The analysis also found granulation signals with amplitudes of about 61.9 parts per million in NRS1 and 54.6 parts per million in NRS2. Their characteristic timescales were about 21.8 and 25.4 minutes, respectively. The researchers checked these estimates against the frequency structure of the residual data and found them consistent with the expected behavior of stellar granulation.
That check matters because the planet’s phase curve is not the only source of changing light in the system. If stellar fluctuations were mistaken for planetary emission, the inferred temperatures and heat distribution could be misleading. Modeling the stellar variability gave the team a way to test whether the planet’s signal remained after those fluctuations were accounted for.
The white-light analysis detected the secondary eclipses at significance levels of at least about three standard deviations. The team then used the combined measurements to examine the planet’s changing brightness and its emission across the observed infrared wavelengths.
A cooler dayside and a warm nightside
For an airless planet with little reflected starlight, the dayside absorbs incoming energy and becomes extremely hot. If the planet cannot move that energy to its nightside, the contrast between the two hemispheres should be large.
TOI-561 b’s measured dayside brightness temperature is lower than that simple scenario predicts. One way to reduce the energy absorbed at the surface is to reflect more of the star’s light. Another is to transport some of the absorbed energy away from the dayside.
The nightside measurement provides a second part of the picture. At about 1,100 K, it is not as cold as a completely unheated hemisphere would be. The researchers interpret this emission as evidence that some energy is being carried from the dayside to the nightside.
These two effects are related but distinct. A high reflectivity reduces how much stellar energy the planet absorbs in the first place. Heat redistribution changes where the absorbed energy is released. The observed combination of a relatively cool dayside and a warm nightside points to both effects.
The temperature estimates are brightness temperatures derived from the infrared observations, not direct readings of the planet’s surface. They also cover only the 3–5 micrometer range. Because the observations do not capture the full thermal spectrum, the inferred energy balance depends on assumptions about how the planet emits at other wavelengths. Non-gray atmospheric behavior, in which different wavelengths are absorbed and emitted differently, can also affect the interpretation.
Even with those qualifications, the researchers found that a bare-rock model with no reflected starlight does not reproduce the observed phase curve. Their atmospheric modeling likewise indicated that heat redistribution alone was not enough to explain the measurements if the atmosphere absorbed essentially all the incoming energy.
Testing possible atmospheres
To explore what could produce the observed pattern, the team compared the phase curves with simulations from the GENESIS general circulation model. These simulations examine how an atmosphere moves heat around a planet and how its temperature and emitted radiation change across the globe.
The researchers tested several illustrative atmospheric compositions: a pure water-vapor atmosphere, a mixture of 50 percent carbon dioxide and 50 percent water vapor, and an atmosphere made mostly of oxygen with a small amount of water vapor. They considered surface pressures of 0.1, 1, 3, 5 and 10 bars, as well as different assumptions about the planet’s Bond albedo.
Bond albedo is the fraction of all incoming stellar energy that a planet reflects back into space, across the full range of wavelengths. A higher Bond albedo means the planet absorbs less of the energy reaching it.
In the simulations, the phase curves with a Bond albedo around 0.6 provided the best overall match to the observed day-night pattern. The authors also found that substantial atmospheric pressure, in the approximate range of 1 to 10 bars in the tested models, was needed to approach the observations.
The results varied between the two detectors. For NRS2, the model with equal parts carbon dioxide and water vapor gave the best matches among the tested cases at an albedo of 0.6. For NRS1, the oxygen-dominated model gave the best match in some of the comparisons. The reduced chi-square values reported for the models also varied across the tested pressures and detector channels.
Those differences do not identify the planet’s atmosphere. The simulations cover only a limited set of possible compositions and conditions, and several model curves are similar. The authors caution that the available data points and the differences between the detector channels do not support ranking the atmospheric compositions as though the best numerical fit established what the planet is made of.
The simulations also did not reproduce every part of the observations. In particular, they generally predicted more nightside emission in NRS1 than the telescope measured. The researchers therefore treat the models as examples of atmospheric conditions that could produce the broad observed pattern, not as a definitive reconstruction of TOI-561 b.
Clouds could help reflect the starlight
The high reflectivity suggested by the phase curve raises another question: what could make the atmosphere so bright?
The researchers considered clouds formed from material vaporized from the planet’s hot rocky surface. On a world this hot, minerals can evaporate and enter the atmosphere. As atmospheric gases move into cooler regions, some of that material could condense into mineral clouds.
The GENESIS simulations and the team’s cloud analysis point to possible magnesium silicate, or MgSiO3, and silicon dioxide, or SiO2, clouds near the morning terminator, the boundary between the dayside and nightside. These are model-based possibilities, not clouds directly detected by JWST.
The optical properties of the particles matter. The study notes that silicon dioxide clouds can scatter a large fraction of near-infrared light. In an illustrative calculation, the researchers considered SiO2 cloud particles with a single-scattering albedo of 0.9, meaning that scattering rather than absorption dominates their interaction with light. In that scenario, the estimated Bond albedo was about 0.224.
A reflective cloud layer could reduce the amount of stellar energy absorbed by the planet. The authors discuss the possibility that this cooling could allow additional cloud formation, creating a feedback between cloud reflectivity and atmospheric temperature. The calculation illustrates a possible mechanism; it does not establish that such a cloud layer exists on TOI-561 b.
The models also suggest that clouds could form on the nightside. Depending on the assumed chemistry of the planet’s mantle, the predicted cloud materials include silicon dioxide and sodium silicate. In the oxidized-mantle scenario, silicon dioxide clouds could occur higher in the atmosphere, with sodium silicate clouds forming deeper down.
Clouds may also help explain why the measured nightside spectrum is faint in part of the NIRSpec range. The authors discuss the possibility that nightside clouds suppress infrared emission, with the effect depending in part on particle size. This remains one possible explanation for the mismatch between the simulated and measured nightside brightness.
The two detectors do not tell exactly the same story
Although the two detector channels support the broad conclusion that a simple bare-rock model is inadequate, their measured phase curves are not identical.
NRS1 shows a larger day-night brightness variation and particularly low nightside emission. NRS2 shows a smaller variation and a westward displacement of the hottest region, known as a hotspot offset. The measured offset is within three standard deviations of zero, so the data do not establish a significant displacement. Still, the difference between the channels is notable.
The researchers investigated stellar granulation in both channels, but the constraints were weaker for NRS2. They also note that NRS1 is affected by stronger long-timescale instrumental systematics. Although their analysis accounts for detector behavior and stellar variability, residual effects could still contribute to the unusually low NRS1 nightside signal or to the disagreement between the channels.
This is an important qualification. The atmosphere interpretation is based on the overall pattern in the phase curve, but the differences between the detector measurements leave open the possibility that some details of the inferred day-night behavior are affected by remaining stellar or instrumental signals.
Hints in the dayside spectrum
The team also examined how the planet’s emitted light changes with wavelength. The dayside spectrum is consistent with an earlier analysis based on secondary-eclipse observations. Across the measured range, the spectrum trends downward from about 2.8 to 4 micrometers before rising again between roughly 4.0 and 4.5 micrometers.
That shape could contain information about the atmosphere’s chemical makeup and temperature structure. The researchers discuss a possible carbon dioxide emission feature, which could occur if the atmosphere has an inverted temperature profile, with some higher layers hotter than the layers below. Clouds could be involved in producing such a structure.
But the available spectrum does not establish that interpretation. The possible molecular features are tentative, and the observations cover only a restricted wavelength range. The team does not claim a confirmed detection of carbon dioxide from this shape alone.
The observations also provided the first nightside spectrum of TOI-561 b reported in the study. It shows particularly low emission in the NRS1 channel between about 2.67 and 3.72 micrometers. The model calculations tend to predict more nightside emission than observed, leaving open questions about the atmosphere, clouds and possible remaining measurement effects.
An atmosphere that is difficult to explain
If TOI-561 b has a substantial atmosphere, its presence raises a problem for understanding the planet’s history. The planet is small and close to its star, conditions under which an atmosphere could be vulnerable to escape into space over billions of years. Yet the observations are most naturally explained by a reflective atmosphere capable of redistributing heat.
The researchers outline several possible ways such an atmosphere could have survived or developed. The planet might have formed with a volatile-rich envelope, perhaps as a larger sub-Neptune, before losing much of its original atmosphere. A later atmosphere produced by processes within the rocky planet could also escape less efficiently if it is rich in heavier molecules. Another possibility is that the planet formed farther from its star and migrated inward later. Volatile material could also have been stored in the interior and released through later outgassing.
These are proposed scenarios, not conclusions drawn directly from the phase curve. The study does not determine how TOI-561 b formed, how its atmosphere originated or how it could have been retained. The authors note that understanding these possibilities will require models that connect the planet’s interior, atmosphere and long-term evolution.
Further observations could help distinguish among the explanations. The team points in particular to a phase curve measured with JWST’s Mid-Infrared Instrument, or MIRI. Observations at longer wavelengths, including the carbon dioxide band near 15 micrometers, could provide information that the 3–5 micrometer NIRSpec data cannot. The mid-infrared measurements would also be less affected by the stellar variability that complicates the current analysis.
For now, the JWST phase curve supports an atmosphere that is both reflective and capable of moving heat around TOI-561 b. The exact gases, cloud properties and atmospheric structure remain unresolved.
The study was published in The Astronomical Journal.