Exoplanets orbiting close to their parent stars could lose some fraction of their atmospheres because of the extreme irradiation 1-6 . Atmospheric mass loss primarily affects low-mass exoplanets, leading to suggest that hot rocky planets 7-9 might have begun as Neptune-like 10-16 , but subsequently lost all of their atmospheres; however, no confident measurements have hitherto been available. The signature of this loss could be observed in the ultraviolet spectrum, when the planet and its escaping atmosphere transit the star, giving rise to deeper and longer transit signatures than in the optical spectrum 17 . Here we report that in the ultraviolet the Neptune-mass exoplanet GJ 436b (also known as Gliese 436b) has transit depths of 56.3 ± 3.5% (1σ), far beyond the 0.69% optical transit depth. The ultraviolet transits repeatedly start ~2 h before, and end >3 h after the ~1 h optical transit, which is substantially different from one previous claim 6 (based on an inaccurate ephemeris). We infer from this that the planet is surrounded and trailed by a large exospheric cloud composed mainly of hydrogen atoms. We estimate a mass-loss rate in the range of ~10 8 -10 9 g s −1 , which today is far too small to deplete the atmosphere of a Neptune-like planet in the lifetime of the parent star, but would have been much greater in the past.Three transits of GJ 436b, which occur every 2.64 days, were observed on 7 A stellar spectrum acquired using similar settings in January 2010 (ref. 17) (visit 0) was retrieved from the archive for comparison purposes. HST data in visits 2 and 3 were complemented with simultaneous Chandra X-ray observations. The HST data consist of timetagged, far-ultraviolet spectra obtained with a grating dispersing light over the 1,195-1,248 Å domain, with a spectral resolution of ~20 km s −1 at 1,215.6 Å (the Lyman-α transition of 2 atomic hydrogen, H I). Exposure times of 1,500 s to 2,900 s were used to observe the star for four successive HST orbits during each visit. Each HST orbit lasts for 96 min, during which GJ 436 is visible for 56 min before being occulted by the Earth, yielding 40 min gaps in the data.The most prominent spectral feature is the H I Lyman-α emission (Fig. 1). Absorption in the blue wing of this line has been reported in other systems, during transits of hot Jupiters. This is interpreted by the presence of escaping hydrogen exospheres surrounding giant planets 1,5,[18][19][20] . Tentative evidence that the Neptune-mass planet GJ 436b possesses such an extended atmosphere was drawn from visit 1 data despite the signal being observed after one optical transit 6 , raising questions on its planetary origin. Visits 2 and 3 were carried out to determine the signal nature.We performed a careful analysis to check for the existence of instrumental systematics in the data and correct for them (see Methods). Large variations are detected over a part of the Lyman-α line at times corresponding to the optical transit, which cannot be explained by any known instrumental effects. The most notab...

Context. KELT-9 b exemplifies a newly emerging class of short-period gaseous exoplanets that tend to orbit hot, early type stars – termed ultra-hot Jupiters. The severe stellar irradiation heats their atmospheres to temperatures of ~4000 K, similar to temperatures of photospheres of dwarf stars. Due to the absence of aerosols and complex molecular chemistry at such temperatures, these planets offer the potential of detailed chemical characterization through transit and day-side spectroscopy. Detailed studies of their chemical inventories may provide crucial constraints on their formation process(es) and evolution history. Aims. We aim to search the optical transmission spectrum of KELT-9 b for absorption lines by metals using the cross-correlation technique. Methods. We analysed two transit observations obtained with the HARPS-N spectrograph. We used an isothermal equilibrium chemistry model to predict the transmission spectrum for each of the neutral and singly ionized atoms with atomic numbers between three and 78. Of these, we identified the elements that are expected to have spectral lines in the visible wavelength range and used those as cross-correlation templates. Results. We detect (>5σ) absorption by Na I, Cr II, Sc II and Y II, and confirm previous detections of Mg I, Fe I, Fe II, and Ti II. In addition, we find evidence of Ca I, Cr I, Co I, and Sr II that will require further observations to verify. The detected absorption lines are significantly deeper than predicted by our model, suggesting that the material is transported to higher altitudes where the density is enhanced compared to a hydrostatic profile, and that the material is part of an extended or outflowing envelope. There appears to be no significant blue-shift of the absorption spectrum due to a net day-to-night side wind. In particular, the strong Fe II feature is shifted by 0.18 ± 0.27 km s−1, consistent with zero. Using the orbital velocity of the planet we derive revised masses and radii of the star and the planet: M* = 1.978 ± 0.023 M⊙, R* = 2.178 ± 0.011 R⊙, mp = 2.44 ± 0.70 MJ and Rp = 1.783 ± 0.009 RJ.

Helium is the second-most abundant element in the Universe after hydrogen and is one of the main constituents of gas-giant planets in our Solar System. Early theoretical models predicted helium to be among the most readily detectable species in the atmospheres of exoplanets, especially in extended and escaping atmospheres . Searches for helium, however, have hitherto been unsuccessful . Here we report observations of helium on an exoplanet, at a confidence level of 4.5 standard deviations. We measured the near-infrared transmission spectrum of the warm gas giant WASP-107b and identified the narrow absorption feature of excited metastable helium at 10,833 angstroms. The amplitude of the feature, in transit depth, is 0.049 ± 0.011 per cent in a bandpass of 98 angstroms, which is more than five times greater than what could be caused by nominal stellar chromospheric activity. This large absorption signal suggests that WASP-107b has an extended atmosphere that is eroding at a total rate of 10 to 3 × 10 grams per second (0.1-4 per cent of its total mass per billion years), and may have a comet-like tail of gas shaped by radiation pressure.

Infrared radiation emitted from a planet contains information about the chemical composition and vertical temperature profile of its atmosphere. If upper layers are cooler than lower layers, molecular gases will produce absorption features in the planetary thermal spectrum. Conversely, if there is a stratosphere-where temperature increases with altitude-these molecular features will be observed in emission. It has been suggested that stratospheres could form in highly irradiated exoplanets, but the extent to which this occurs is unresolved both theoretically and observationally. A previous claim for the presence of a stratosphere remains open to question, owing to the challenges posed by the highly variable host star and the low spectral resolution of the measurements. Here we report a near-infrared thermal spectrum for the ultrahot gas giant WASP-121b, which has an equilibrium temperature of approximately 2,500 kelvin. Water is resolved in emission, providing a detection of an exoplanet stratosphere at 5σ confidence. These observations imply that a substantial fraction of incident stellar radiation is retained at high altitudes in the atmosphere, possibly by absorbing chemical species such as gaseous vanadium oxide and titanium oxide.

Atmospheric escape has been detected from the exoplanet HD 209458b through transit observations of the hydrogen Lyman-α line. Here we present spectrally resolved Lyman-α transit observations of the exoplanet HD 189733b at two different epochs. These HST/STIS observations show for the first time that there are significant temporal variations in the physical conditions of an evaporating planetary atmosphere. While atmospheric hydrogen is not detected in the first epoch observations, it is observed at the second epoch, producing a transit absorption depth of 14.4 ± 3.6% between velocities of −230 to −140 km s −1 . Contrary to HD 209458b, these high velocities cannot arise from radiation pressure alone and require an additional acceleration mechanism, such as interactions with stellar wind protons. The observed absorption can be explained by an atmospheric escape rate of neutral hydrogen atoms of about 10 9 g s −1 , a stellar wind with a velocity of 190 km s −1 and a temperature of ∼10 5 K. An X-ray flare from the active star seen with Swift/XRT 8 h before the second-epoch observation supports the idea that the observed changes within the upper atmosphere of the planet can be caused by variations in the stellar wind properties, or by variations in the stellar energy input to the planetary escaping gas (or a mix of the two effects). These observations provide the first indication of interaction between the exoplanet's atmosphere and stellar variations.