Ground-based characterization of transiting exoplanets

  1. López Fernández-Nespral, David
Supervised by:
  1. Roi Alonso Sobrino Director
  2. Hans-Jörg Deeg Co-director

Defence university: Universidad de La Laguna

Fecha de defensa: 07 June 2019

  1. Victor Sánchez Béjar Chair
  2. Sergio David Hoyer Miranda Secretary
  3. Raine Karjalainen Committee member

Type: Thesis

Teseo: 590877 DIALNET lock_openRIULL editor


In the last decade the detection of new planets around stars outside our solar system has grown dramatically. These detections, made both from ground-based telescopes and from the different space missions dedicated to this type of discovery, have resulted in a large range of planets with different masses and radii. These values of masses and radii provide mean values of planetary density, which result in an infinite number of solutions relative to the chemical composition of the planet. A current research field of exoplanets is the study of their atmospheres. This type of study allows to study the chemical composition of the upper layers of the planetary atmosphere. This work is focused on the study of exoplanetary atmospheres from a sample of hot Jupiter. Thanks to large sizes, large masses and large temperatures and to the proximity of its central star, make them excellent targets for the study of their atmospheres through the technique of spectroscopy transmission. This technique detects the light of the central star that passes through the thin layers of the planet's atmosphere, showing the chemical composition of its upper layers. In this thesis we presented the spectro-photometry study of six Hot Jupiter and one Brown Dwarf, at optical wavelengths using ground based observations with the Grantecan telescope (GTC). These analyses are summarized below. In chapter 3, we analyzed the original data of a transit of TrES-3b taken by Parviainen et al. (2016) in the optical range with OSIRIS/GTC. The same data have been re-analyzed by different authors (Parviainen et al. 2016, Mackebrandt et al. 2017) whose transmission spectra disagree. The motivation of the chapter was to try to find the cause of such a discrepancy and obtain our transmission spectrum of TrES-3b. Our analysis took into account the effect of the close contaminating companion to the target as well as testing how Limb Darkening values (LDCs) affect the final spectrum of a grazing planet such as TrES-3b. Our results are consistent with those obtained by Mackebrandt et al. (2017), who performed a similar analysis, and disagree with those obtained by Parviainen et al. (2016) who reported a different analysis from ours. Note that Parviainen et al. (2016) also show an analysis similar to ours in their work and which is totally compatible with ours. In addition, we detected a larger planet radius at the bluest wavelength bins, associated with the Rayleigh scattering phenomenon in the TrES-3b atmosphere. This increment in radius cannot be explained with the theoretical models that explain this phenomenon, so other phenomena must be occurring in the atmosphere of TrES-3b and which we are unable to explain presently. In chapter 4, we present the optical transmission spectrum in the optical range of Qatar-1b This study allowed to verify the low instrumental systematics affecting GTC/OSIRIS data and showed that GTC/OSIRIS is a powerful instrument for determining transmission spectra. In the analysis of the white light curve it was necessary to introduce a periodical term in the systematic model to fit part of the curve correctly. We note that although we did not know the real origin of the systematic effect on the light curve, we could correct this effect. The results of the physical parameters obtained from the white light curve analysis are consistent with most previously performed studies. However, there are other older studies in the literature that present smaller values of transit depth. These discrepancies are compatible with the fact that Qatar-1 is a moderately active star, which causes the star's flux to vary during these phases of activity. In the analysis of the spectroscopic curves we verified the impact of the limb darkening coefficients (LDCs) on the shape of transit. In this case, Qatar-1b is planet with an intermediate impact parameter value. The results obtained conclude that the effect in modeling of the LDCs color light curves and the impact parameter value was minimum, i.e, we obtained similar transmission spectra when we fixed the LDCs or not. These results support the idea that the theoretical LDCs derived for each spectroscopic curve are correct. The analysis carried out in Qatar-1b does not give preference to any atmospheric model over any other. A small slope is also observed in the bluest part of the spectrum that may be due to Rayleigh. However, any conclusion about its atmosphere must be taken with caution since they are results obtained with a single transit. Our result is compatible with the result obtained by von Essen et al. (2017). In chapter 5, we present a transmission spectrum of WASP-36b as a result of the observation of two consecutive transits. WASP-36b is interesting because it is orbiting around a star of very low metallicity, and there are very few cases of giant gaseous planets with this feature. Both white light curves presented a same systematic effect at the end of the transit egress. As both observations were configured the same way, it was easy to detect where the systematic effect came from. In this case it was due to the variation in the rotator angle which exceeded 60 degrees. This caused small variations in the stars flux on detector which depends on the distance to the center of rotation in the detector. The results obtained from the transit depth are not consistent with previous studies, but we note that our results are the only ones obtained with complete light curves. Each night's spectroscopic analysis agrees with each other, and also with an analysis that includes both nights. The shape of the transmission spectrum obtained cannot be explained by exoplanetary theoretical models. Our transmission spectrum presents a sudden change in the size of the planet redwards / bluewards from 700 nm, this can only be explained with theoretical models that present strong absorptions without including Na or K. However, an alternative more likely would be to assume that there are sources of systematics in this spectral range that have not been characterized during the analysis and that cause this abrupt change. A previous study by Mancini et al. (2016) in broadband show a strong Rayleigh in the WASP-36b atmosphere, which cannot be explained with theoretical models. Mancini et al. (2016)'s result is nor compatible with ours. Another of the most relevant conclusions we can extract from the chapter is the advantage of observing a transit more than once to derive more reliable results of its transmission spectrum. In chapter 6, we present a study of the atmospheres of two pairs of twin planets: with similar physical and orbital parameters, except in some of the parameters, with the aim of looking for differences. However, due to instrumental problems on one of the nights of observation, the study of one of the pairs could not be carried out completely. The first pair was HAT-P-41b and HAT-P-33b, planets with a mass of about 0.8 Jupiter masses and a radius of 1.7 Jupiter radii, with a slight difference in orbital distances and orbiting around stars of about 6400K. The HAT-P-33 data were contaminated by internal reflections of the moonlight on the detector. This caused the spectrum of the reference star to be contaminated. After a careful reduction of the data we were able to minimize the effect of light on the spectrum and obtain a good white light curve. The results obtained of the physical parameters of the planet from the corrected light curve are compatible with previously published studies. However, the quality of the data was not sufficient to distinguish between various atmosphere models. The HAT-P-41 data lack sufficient points before the transit ingress, which caused the error in determining the physical parameters of the planet to be greater than if those points were present. The transit depth result is consistent with previously published studies. However, in the result of the transmission spectrum it was not possible to distinguish between the different atmosphere models studied due to the quality of the data. These results need to be confirmed with new observations since there are no previous spectroscopic studies. Kepler-7b and Kepler-12b is the other pair of twin planets. They have a mass of approximately 0.4 Jupiter masses and a radius of 1.5 Jupiter radii. Their host stars have a temperature around 5950 K. The main difference in this couple is the albedo of their atmospheres. In this case, we have explored the use of a new technique in GTC, a scan mode where the star is moved parallel to the slit while exposing. For Kepler-7b, the transit was not detected. The causes of the absence of detection, after ruling out problems with ephemeris and meteorological causes, we suspect that they are due to the variations of the rotator angle, that cause flux variations in the detector. This effect was already detected in transit observations carried out with OSIRIS/GTC (Nortmann et al. (2016), Chen et al. (2017)). As an extension to Nortmann et al. (2016)'s study, we carry out the characterization of the detector to find possible flux variations along the CCD as a function of the rotator angle, with flat field images and covering all the positions of the rotator angle. The study presented in Sect 6.7.1 shows how large variations of the rotator angle produce significant changes in the flux received in the detector, reaching maximum values of 1%. From our study together with Nortmann et al. (2016), it is concluded as technical preventive steps during transit observations with OSIRIS/GTC: to avoid great variations of the angle of the rotator during the time series, and to keep the target and the reference star as close as possible to the Nasmyth rotation center. In the case of Kepler-12b observations, two observations were carried out, one with the 'scan' mode and the other with the standard mode. In this last observation there was a system interruption just in the middle of the transit, causing the instrument's configuration not to be the same on both parts of the transit. After analysis of the data could conclude that no atmospheric model is significantly favored. The transmission spectrum of one of the two nights (standard mode) shows absorption around the sodium and potassium lines. These results need to be confirmed with new observations since the detections are not very significant. In chapter 7, we presented a spectro-photometric study of CoRoT-15b, a transiting brown dwarf. The motivation of the chapter was to observe the transmission spectrum of a brown dwarf to check that nothing should be detected in its atmosphere due to its high gravity value, which causes a very small-scale height (about a few kilometers). The result shows that none of the atmosphere models tested is more compatible than another, due to lack of data precision. Thanks to the transit obtained with GTC we have improved a 28% the density value of CoRoT-15b, from 59±37 g/cm3 (Bouchy et al. 2011) to 57±23 g/cm3. In addition, an update of the ephemeris of the CoRoT-3b BD could be carried out while its data were not suitable for a further analysis. In chapter 8 we discussed how our sample compares with the global sample of planets found, and also with the sample of planets characterized by the GTC in which alkaline metal signals have been detected in their atmospheres. A comparative summary of the results of those planets that have already been characterized by other authors is also carried out. It served to test the validity of the analysis technique used. In addition, we define and measure two broadband parameters in two different spectral ranges. These parameters compare the relative strength of scattering in the transmission spectrum. They were later compared with different theoretical models. Both parameters are independent of scale height and show trends between cloudy and clear atmospheres. No significant feature was found, except for cases where we suspect that systematic noises are predominant in our spectra. In chapter 9, we report the main conclusions of the thesis.