The high-energy environment and atmospheric escape of the mini-Neptune K2-18 b
Leonardo A. dos Santos, David Ehrenreich, Vincent Bourrier, Nicola Astudillo-Defru, Xavier Bonfils, François Forget, Christophe Lovis, Francesco Pepe, Stéphane Udry
AAstronomy & Astrophysics manuscript no. 37327 c (cid:13)
ESO 2020January 22, 2020 L etter to the E ditor The high-energy environment and atmospheric escape of themini-Neptune K2-18 b (cid:63)
Leonardo A. dos Santos , David Ehrenreich , Vincent Bourrier , Nicola Astudillo-Defru , Xavier Bonfils , FrançoisForget , Christophe Lovis , Francesco Pepe , and Stéphane Udry Observatoire astronomique de l’Université de Genève, 51 chemin des Maillettes, 1290 Versoix, Switzerlande-mail:
[email protected] Departamento de Matemática y Física Aplicadas, Universidad Católica de la Santísima Concepción, Alonso de Rivera, 2850 Con-cepción, Chile Université Grenoble Alpes, CNRS, IPAG, 38000 Grenoble, France Laboratoire de Météorologie Dynamique, Institut Pierre Simon Laplace, Université Paris 6 Boite Postale 99, 75252 Paris cedex 05,FranceReceived 6 December 2019; accepted 13 January 2020
ABSTRACT
K2-18 b is a transiting mini-Neptune that orbits a nearby (38 pc), cool M3 dwarf and is located inside its region of temperateirradiation. We report on the search for hydrogen escape from the atmosphere K2-18 b using Lyman- α transit spectroscopy with theSpace Telescope Imaging Spectrograph (STIS) instrument installed on the Hubble Space Telescope ( HST ). We analyzed the time-series of fluxes of the stellar Lyman- α emission of K2-18 in both its blue- and redshifted wings. We found that the average blueshiftedemission of K2-18 decreases by 67% ±
18% during the transit of the planet compared to the pre-transit emission, tentatively indicatingthe presence of H atoms escaping vigorously and being blown away by radiation pressure. This interpretation is not definitive becauseit relies on one partial transit. Based on the reconstructed Lyman- α emission of K2-18, we estimate an EUV irradiation in the range10 − erg s − cm − and a total escape rate on the order of 10 g s − . The inferred escape rate suggests that the planet will lose onlya small fraction ( < ff ects, confirm the in-transit absorption, and better assess the atmospheric escape and high-energy environmentof K2-18 b. Key words.
Stars: individual: K2-18 – stars: chromospheres – planets and satellites: atmospheres – ISM: kinematics and dynamics
1. Introduction
Short-period exoplanets orbiting nearby, cool M dwarfs areprime targets for the search and characterization of atmospheresof low-mass, sub-Neptune-sized worlds. One particular targetthat falls in this category is K2-18 b, which was first pointedout as a transiting planet candidate by Montet et al. (2015) andlater confirmed with
Spitzer photometry (Benneke et al. 2017)and Doppler velocity measurements (Cloutier et al. 2017). Thisplanet has a radius of R P = . ± .
065 R ⊕ , a mass of M P = . ± .
35 M ⊕ and an orbital period of T orb = . / He envelopeor a 100% H O composition (Sarkis et al. 2018; Cloutier et al.2019). The host star is a nearby M2.8-type dwarf located at 38pc (Gaia Collaboration et al. 2018), rendering K2-18 b one ofthe best mini-Neptunes suitable for atmospheric follow-up us-ing the
Hubble Space Telescope ( HST ), the
James Webb SpaceTelescope ( JWST ), and high-resolution infrared spectrographs. (cid:63)
The
HST
Lyman- α spectra are available in electronic form at theCDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or viahttp: // cdsweb.u-strasbg.fr / cgi-bin / qcat?J / A + A / The 0.4–5.0 µ m transmission spectrum of K2-18 b mea-sured with data from the Kepler satellite ( K2 mission), the Wide-Field Camera 3 (WFC3 / HST ), and the Infrared Array Camera(IRAC / Spitzer ) revealed the presence of water vapor in its loweratmosphere (Benneke et al. 2019; Tsiaras et al. 2019). By com-paring atmospheric models with the data, Benneke et al. con-cludes that the best match is a H -dominated atmosphere withwater vapor absorbing above the cloud deck below the 10-1000mbar pressure level. While these observations provided us withsome initial information regarding the composition of its loweratmosphere, they do not constrain the abundances of molecularspecies.Models predict that the deposition of high-energy photons(X-rays and far-ultraviolet) produced by the host star leads toan expansion of the planetary upper atmosphere, as well as theproduction of H atoms due to photodissociation of H O (e.g.,Ip 1983; Wu & Chen 1993). This expansion populates the outerlayers of the planetary atmosphere where the gas is collision-less, also known as the exosphere. It is therefore likely that theatmosphere of K2-18 b, which is rich in H and H O, possessesa H-rich exosphere. Previous
HST observations have shown evi-dence for the presence of large-scale, H-rich exospheres aroundthe warm Neptunes GJ 436 b (Ehrenreich et al. 2015; Lavieet al. 2017; dos Santos et al. 2019) and GJ 3470 b (Bourrieret al. 2018). However, to date, no evidence for extended atmo-
Article number, page 1 of 5 a r X i v : . [ a s t r o - ph . E P ] J a n & A proofs: manuscript no. 37327 spheres has been found for planets smaller than Neptune, as non-detections were reported for the super-Earths 55 Cnc e (Ehrenre-ich et al. 2012), HD 97658 b (Bourrier et al. 2017c), GJ 1132 b(Waalkes et al. 2019), and π Men c (García Muñoz et al. 2019),and marginal detections were reported for the small rocky-planetsystems in TRAPPIST-1 (Bourrier et al. 2017a) and Kepler-444(Bourrier et al. 2017b).In this letter, we report the results of a series of far-ultraviolet(FUV) observations of two transits of K2-18 b using the SpaceTelescope Imaging Spectrograph (STIS) installed on
HST . Theaim of these observations was to perform Lyman- α transmissionspectroscopy of K2-18 b in order to detect its H-rich exosphere,which should produce an excess absorption in the stellar Lyman- α emission during the transit of the planet.
2. Observations and data reduction
K2-18 was observed with
HST / STIS and the grating G140M (re-solving power R ≈ α emission and rendered this visit unsuitable foranalysis (see Fig. 1). The second visit (B) contained four suc-cessful exposures, of which two were performed before the tran-sit ingress and the other two in-transit.The data were reduced using the standard STIS pipeline, ex-cept for the spectral extraction. Since the star is faint, the au-tomated extraction is unable to accurately find the stellar spec-trum in the flat-fielded frames. Furthermore, the dark currentbackground of the FUV-MAMA detector of STIS reaches levelshigh enough to be comparable with the stellar spectrum. In or-der to correctly extract the spectrum and remove the dark currentbackground, we use the x1d method of stistools with user-defined values for the location of: i) the spectrum in the cross-dispersion direction, and ii) the regions where the dark currentbackground near the spectrum can be accurately estimated.Using visual inspection, we determined the location of thespectrum in the cross-dispersion direction to be y =
389 px(parameter a2center in the pipeline). Determining the best lo-cation of the background is not as straightforward; normally,the pipeline extracts the background from regions far from thespectrum, but these regions have discrepant levels of dark cur-rent compared to the region of the spectrum. Therefore, wechose to use regions immediately near the location of the spec-trum to determine the background (parameters bk1offst and bk2offst in the pipeline), namely at a distance of ∆ =
20 pxfrom a2center . We combine the two out-of-transit spectra andthe two in-transit spectra separately in order to isolate potentialsignals of an atmospheric signal of K2-18 b, and the resultingspectra are shown in Fig. 2.
3. Tentative detection of a H-rich exosphere inK2-18 b
Since the interstellar medium (ISM) absorbs the core of theLyman- α emission line, we can only observe the attenuatedfluxes in the blue and red wings of the stellar line. We integrated Software freely available at https://stistools.readthedocs.io/en/latest/ . the flux densities in wavelength space between Doppler veloci-ties [-160, -50] km s − and [ + + − , respectively, toproduce the Lyman- α light curves in the blue and red wings.During Visit B, we observe a steep decrease in theblueshifted Lyman- α fluxes during the transit of K2-18 b (leftpanel in Fig. 3), reaching almost zero emission near the plane-tary egress. This decrease in flux is also seen when we comparethe combined out-of-transit and in-transit spectra (see Fig. 2).The redshifted Lyman- α emission of the combined in-transit or-bits varies by 14% ± ff erent positions in the detector.The blue wing flux measured in the combined in-transit spec-tra decreases by 67% ±
18% in relation to the combined out-of-transit spectra; in particular, the last orbit displays an absorptionof 93% ±
18% in relation to the combined out-of-transit spectra.Although statistically significant, we conservatively deem thisresult tentative until it is repeated in future observations; for areference, the intrinsic stellar variability of the Lyman- α emis-sion of HD 97658 b is on the order of a few tens of percent at ∼ σ confidence (Bourrier et al. 2017c). If confirmed to be linkedto the transit of K2-18 b, the variation in Lyman- α flux can beinterpreted as the absorption caused by an extended, H-rich ex-osphere of the planet. Such a large absorption signal can be ex-plained by a combination of large atmospheric escape rate andlong photoionization lifetime of the H atoms in the exosphere.This result gives further support to the hypothesis that K2-18 bpossesses a H -dominated envelope (as in the conclusions ofBenneke et al. 2019), unlike the super-Earths 55 Cnc e andHD 219134 b. Previous results for the super-Earths HD 97658 band GJ 1132 b were inconclusive due to stellar variability for thefirst (Bourrier et al. 2017c) and lack of stellar blue wing emissionin the second (Waalkes et al. 2019).
4. The high-energy environment of K2-18 b
Determining the high-energy environment of K2-18 b provides acritical piece of information to interpret the evolution and currentstate of its atmosphere. To that end, we used the STIS observa-tions of K2-18 to reconstruct its intrinsic Lyman- α spectrum andestimate the high-energy irradiation received by planet b. The re-construction process follows the standard method used in, for ex-ample, Bourrier et al. (2017a) and Bourrier et al. (2018). In short,we fit the observed spectrum to a model of the intrinsic emissionline attenuated by ISM absorption, scaled for distance and con-volved with the instrumental response; the fit yields an estimateof the intrinsic emission and certain properties of the ISM in theline of sight. In this process, we assume that the intrinsic Lyman- α emission of K2-18 possesses a Gaussian profile (applicable forM dwarfs and for the quality of the available spectra; see Bour-rier et al. 2017b, 2018), and fix the temperature and turbulentvelocity of the ISM to 8000 K and 1.23 km s − (for the NGPcloud, as estimated by the LISM calculator Redfield & Linsky2008), respectively. We also set the deuterium-to-hydrogen ratio(D i / H i ) to 1 . × − and the systemic velocity to 0.6537 km s − (measured with high-resolution spectra by Cloutier et al. 2017).The result of the Lyman- α line reconstruction is shown inFig. 4. We simultaneously fit each exposure to a global Lyman- α line model within ±
300 km s − in the stellar rest frame, ex-cluding the band that corresponds to significant airglow con-tamination (approximately the range ±
50 km s − , shown in or-ange in Fig. 4). When fitting the in-transit orbits we excluded Article number, page 2 of 5eonardo A. dos Santos et al.: The high-energy environment and atmospheric escape of the mini-Neptune K2-18 b C o un t r a t e ( × c o un t s s )
400 300 200 100 0 100 200 300 400Doppler velocity (km s ) Orbit 1Orbit 2 C o un t r a t e ( × c o un t s s )
400 300 200 100 0 100 200 300 400Doppler velocity (km s ) Orbit 1Orbit 2Orbit 3Orbit 4
Fig. 1.
HST / STIS spectra of K2-18 in Visits A (left panel) and B (right panel) obtained after data reduction (lines). The geocoronal and backgroundcontamination is shown as vertical bars; in Visit A, the geocoronal emission overwhelms the stellar fluxes, preventing us from reliably measuringthe latter. The Doppler velocities are in the stellar rest frame. F l u x d e n s i t y ( × e r g s c m Å )
400 300 200 100 0 100 200 300 400Doppler velocity (km s ) Out of transitIn transitGeocoronal contamination
Fig. 2.
HST / STIS spectra of K2-18 in Visit B. The Doppler velocitiesare in the stellar rest frame. The shaded interval is the region with geo-coronal contamination. the pixels corresponding to the range potentially absorbed bythe planet. We explore the parameter space using the Markovchain Monte Carlo ensemble sampler implementation of emcee (Foreman-Mackey et al. 2013); in addition, we report the un-certainties based on the highest density interval (HDI), whichcontains 68.3% of the distribution mass such that no point out-side the interval has a higher density than any point within it. Wefit for four free parameters in total: the temperature (assuminga Gaussian thermal broadening) and amplitude of the intrinsicstellar Lyman- α line, and the radial velocity and H i density ofthe ISM. We estimated the EUV flux in K2-18 b using the rela-tion from Linsky et al. (2014).We determined that the heliocentric radial velocity of theISM is V R = . + . − . km s − , and the H i column density inthe line of sight is log η = . + . − . cm − . While η is consis-tent with the results from Wood et al. (2005), the value we de-termined for V R di ff ers by 2 σ with the value predicted by LISMcalculator (Redfield & Linsky 2008).We estimated the properties of the high-energy environmentof K2-18 b based on the reconstructed Lyman- α emission and theresults are shown in Table 1; the large uncertainties in these esti-mates are due to the low signal-to-noise ratio (S / N) of the spectra
Table 1.
Properties of the high-energy environment of K2-18 b.
Lyman- α flux (erg s − cm − ) 100 . + . − . EUV 10-91.2 nm flux (erg s − cm − ) 107 . + . − . Photoionization rate ( × − s − ) 3 . + . − . Photoionization lifetime (h) 3600 + − Escape rate at 100% e ffi ciency ( × g s − ) 3 . + . − . and the uncertainties in the semi-empirical relations used to esti-mate the EUV flux. The best fit model of the intrinsic Lyman- α emission of K2-18 results in a β (ratio between radiation pres-sure and stellar gravity; see right axis of Fig. 4) of about 2.2. Al-though more data would be necessary to confirm this line shape,and the corresponding high β , it nonetheless suggests that ra-diation pressure could blow away the escaping hydrogen atomsmore strongly in K2-18 b than in GJ 436 b (Bourrier et al. 2016)or the TRAPPIST-1 planets (Bourrier et al. 2017a). This value of β is similar to that inferred for GJ 3470 b (Bourrier et al. 2018),which could indicate that the exosphere of K2-18 b possesses asimilar shape. Using the energy-limited escape (e.g., Salz et al.2015) as an initial estimate, we predict that the total escape ratein K2-18 b is two orders of magnitude lower when comparedto the values inferred for GJ 436 b ( ∼ × g s − ; Bourrieret al. 2016) and GJ 3470 b ( ∼ × g s − ; Bourrier et al.2018). We estimated the photoionization rate and lifetime as inBourrier et al. (2017c), which yields a value likely above 1200h for the latter; this long lifetime means that the H atoms in theexosphere of K2-18 b could stay neutral for much longer thanin GJ 436 (12 h) and GJ 3470 b (3.5 h). These results are notsurprising because K2-18 b is subject to lower irradiation levelsthan the aforementioned warm Neptunes.The inferred escape rate of K2-18 b is likely underestimatedbecause we do not take into account the stellar X-ray flux, whichis currently unknown. For a similar star like GJ 436, the ratio be-tween X-ray and EUV emission is ∼ Article number, page 3 of 5 & A proofs: manuscript no. 37327 F l u x ( e r g s c m ) Visit BVisit A F l u x ( e r g s c m ) Visit BVisit A
Fig. 3.
Light curves of the blue (left panel) and red (right panel) wings of the Lyman- α emission of K2-18 during the transit of planet b. Time = ff ected by strong geocoronal contamination, and therefore the measured stellar fluxes are likely inaccurate. -300 -200 -100 0 100 200 300Doppler velocity in stellar rest frame (km s )0123 F l u x d e n s i t y ( e r g s c m Å ) Fig. 4.
Reconstructed intrinsic Lyman- α spectrum of K2-18 (green) andthe average Lyman- α profile as observed with HST / STIS (black bars).The red (blue) curve shows the inferred observable spectrum with (with-out) instrumental convolution and the shaded region represents the 1 σ uncertainty for the red curve. The dotted curve shows the inferred ISMabsorption profile.
5. Conclusions
K2-18 b is currently one of the best targets for transit spec-troscopy among sub-Neptune planets due to its large scaleheight, its short distance from the Sun, and the infrared bright-ness of the host star. Previous results have shown evidence thatthe atmosphere of the planet is dominated by H / He and con-tains water vapor. In these atmospheric conditions and under theexpected high levels of EUV irradiation, K2-18 b is prone to ef-ficiently losing its atmosphere and producing a detectable excessabsorption of H in Lyman- α caused by a H-rich exosphere dur-ing transit. In this study we analyzed four HST orbits before andduring the transit of K2-18 b with the STIS instrument to searchfor this feature.We analyzed the flux time series of both the blue- and red-shifted wings of the stellar Lyman- α emission. The blue wingdisplays a significant excess absorption during the transit; in par-ticular, near the egress of K2-18 b, the flux in the blue wing isconsistent with 100% absorption. The in-transit red wing fluxesvary by 14% ±
23% and are significantly more stable than theblue wing fluxes. A blueshifted absorption could indicate thepresence of a H-rich exosphere around K2-18 b being swept away by radiation pressure from its host star towards the direc-tion of the observer, similar to the exospheres of GJ 436 b andGJ 3470 b.Despite the low S / N of the observed spectra, we were ableto reconstruct the intrinsic stellar emission (without the ISM ab-sorption) to assess the high-energy environment of K2-18 b. Ourfirst estimate for the expected total escape rate of K2-18 b leadsto a value on the order of 10 g s − . The ratio between radiationpressure and gravity ( β ) suggests that the exosphere of K2-18 bis in a similar state to that observed for GJ 3470 b. We estimatethat the EUV (10 − . − erg s − cm − . At the estimated escape rate, it is likelythat the planet will lose only a small fraction (1% or less) of itsmass during its remaining lifetime, and therefore it is probablynot an archetypal planet crossing the radius valley to become abare rock (Fulton et al. 2017; Van Eylen et al. 2018; Fulton &Petigura 2018); as such, the planet will likely retain its volatile-rich atmosphere due to the more amenable EUV irradiation fluxthan for example GJ 3470 b, which is at least ten times moreEUV irradiated than K2-18 b.Since we observed only one partial transit of K2-18 b, weconclude that the H-rich exosphere detection is only tentativefor now, and more observations are needed to rule out stellaractivity e ff ects and confirm the reported feature. Furthermore,additional observations of the Lyman- α spectrum of K2-18 willhelp in better constraining the high-energy environment of theplanet and its atmospheric escape history. Acknowledgements.
LAdS thanks M. Turbet and M. López-Morales for the in-sightful discussions about K2-18 b. This project has received funding fromthe European Research Council (ERC) under the European Union’s Horizon2020 research and innovation programme (project F our A ces ; grant agreementNo 724427), and it has been carried out in the frame of the National Cen-tre for Competence in Research PlanetS supported by the Swiss National Sci-ence Foundation (SNSF). This research is based on observations made withthe NASA / ESA Hubble Space Telescope. NA-D acknowledges the support ofFONDECYT project 3180063. The authors are grateful to the anonymous refereefor the quick and helpful review. The data are openly available in the MikulskiArchive for Space Telescopes (MAST), which is maintained by the Space Tele-scope Science Institute (STScI). STScI is operated by the Association of Uni-versities for Research in Astronomy, Inc. under NASA contract NAS 5-26555.This research made use of the NASA Exoplanet Archive, which is operated bythe California Institute of Technology, under contract with the National Aero-nautics and Space Administration under the Exoplanet Exploration Program. Weused the open source software SciPy (Jones et al. 2001), Jupyter (Kluyver et al.2016), Astropy (Astropy Collaboration et al. 2013), Matplotlib (Hunter 2007)and emcee (Foreman-Mackey et al. 2013).
Article number, page 4 of 5eonardo A. dos Santos et al.: The high-energy environment and atmospheric escape of the mini-Neptune K2-18 b
References