Looking for reflected light from Boo b in high-cadence HARPS-N - PDF document

Twenty years of giant exoplanets - Proceedings of the Haute Provence Observatory Colloquium, 5-9 October 2015 Edited by I. Boisse, O. Demangeon, F. Bouchy & L. Arnold Looking for reflected light from τ Boo b in high-cadence HARPS-N observations F. Borsa 1 and the GAPS team Poster presented at OHP-2015 Colloquium 1 INAF – Osservatorio Astronomico di Brera, Via E. Bianchi 46, 23807 Merate (LC), Italy ( francesco.borsa@brera.inaf.it ) Abstract We observed the τ Boo system with the HARPS-N spectrograph to test a new observational strat- egy aimed at jointly studying asteroseismology, the planetary orbit, and star-planet magnetic interac- tion (Borsa et al. 2015, A&A, 578, A64). Given the very high signal-to-noise ratio of our spectra, we tested if it was possible to retrieve also the signal of the CCF given by the light reflected from the planet. We found that night-to-night variations of the stellar CCF are dominant, and add noise up to more of the expected planetary signal. We can give an upper limit to the albedo of the planet of 0.25. 1 Introduction With the intent of characterizing planetary systems with a spectro-photometric approach, in Borsa et al. 2015 (hereafter B15) we selected the well-known system τ Bootis A (HD 120136, F6V, V = 4.49) as a test case in the Global Architecture of Planetary Systems (GAPS, Covino et al. 2013) programme. τ Boo’s brightness allows for high-resolution spectroscopy and asteroseismology with a limited investment of telescope time. We collected with the HARPS-N spectrograph (Cosentino et al. 2012) high-cadence observations on 11 nearly consecutive nights, with which we made an asteroseismic analysis. We detected solar-like oscillations in the radial velocity (RV) time series, and estimated asteroseismic quantities that agree well with theoretical predictions. With this quantities we created a stellar model that could constrain the age of the system and the value of stellar (and thus planetary) mass. With the same spectra, applying a dedicated technique for averaging the raw FITS files of each night, we could analyse the variation of the CaII H&K lines with very high signal-to-noise ratio (S / N) spectra. We could thus study the star-planet interaction: the correlation between the chromospheric activity and the planetary orbital phase remained unclear. Using averaged spectra, we could also obtain RV values free from stellar oscillations: in this way we could update the planetary ephemeris and show the acceleration caused by the stellar binary companion. The very high S / N spectra available made us thinking about another possible analysis with the same dataset: the search for reflected light from the planet, as already done by Martins et al. 2015 (hereafter M15) for 51 Peg b. 2 Looking for reflected light 51 Peg b is the best candidate to look for reflected light spectroscopically. The host star is very bright ( M V = 5 . 49), its VsinI is low ( VsinI ≃ 2 . 8 km / s), the orbital period short (4.2 days) and the inclination of the system favourable ( i ≃ 80 degrees). The method utilized by M15 for the claimed possible detection of reflected light from the planet makes use of the cross-correlation function (CCF) of a binary mask with high-resolution spectra, to amplify the planetary signal that is present in the spectra by a factor proportional to the number of spectral lines. τ Boo, with respect to 51 Peg, presents two problems: the orbital inclination of the planet to the line of sight is less favourable to reflected light ( i ≃ 45 degrees), and the VsinI ( ≃ 15 km / s , B15) is very large. The inclination angle, however, makes reflected light possible to be observed at various orbital phases at the same percentage. In this way we can use our high-cadence HARPS-N spectra (see B15 for an observations log) even if they are taken 109

Twenty years of giant exoplanets - Proceedings of the Haute Provence Observatory Colloquium, 5-9 October 2015 Edited by I. Boisse, O. Demangeon, F. Bouchy & L. Arnold with a complete orbital phase coverage, and not only close to phase 0.5 (taking the inferior conjunction at phase 0.0). To understand the feasibility of the analysis, we first calculated which is the albedo that we could expect from the planet: given a mass of ≃ 6 . 13 M J , we hypotesized a radius varying from 1.2 R J to 1.7 R J . We then estimated the maximum reflection predicted as a function of the albedo of the planet (Figure 1). −4 1.2 Rj 1.2× 10 1.3 Rj 1.4 Rj −4 1.0× 10 1.5 Rj 1.6 Rj 1.7 Rj max reflection −5 8.0× 10 −5 6.0× 10 −5 4.0× 10 −5 2.0× 10 0.1 0.2 0.3 0.4 0.5 albedo Figure 1: Maximum reflection expected from τ Boo b as a function of its albedo. Di ff erent colors represent the results for various radii of the planet, from 1.2 R J to 1.7 R J Verified that the reflection could be expected even > 10 − 5 times the stellar signal, which is of the order of the noise for one night of our observations, we analysed our data (11 nights) looking for the planetary signal, using a method that is very similar to that of M15: in the CCF of the star, there should be hidden also the CCF of the planet at its RV. We analysed 281 spectra taken in 11 di ff erent nights, reducing the data with the DRS pipeline of HARPS-N with the YABI platform. This time we enlarged the CCF width up to 300 km / s, and used a CCF step of 0.1 km / s. As a reference mask, we used the ad hoc one we created in B15. We then made an average CCF (CCF re f ) by shifting all the CCFs for the respective RV, and normalized every CCF with this CCF re f . The resulting residuals should contain the CCF of the planet (CCF pl ). We shifted all the CCF pl for the theoric planetary RVs, estimated using the ephemerides of B15 and an inclination angle of 44.5 degrees (Brogi et al. 2012), and summed them toghether. We found that night-to-night variations of the CCF are dominant, resulting in a noise up to 5x10 − 4 (Figure 2), greater than the signal we are looking for. A possible cause of these variations could be contamination due to the Moon light: our data were not taken to look for a signal so small, and this source of noise was not taken into account given the brightness of the star. We tried then a slightly di ff erent approach, to reduce this e ff ect. We verified that normalizing the CCFs of each night for the average CCF of the night, allows us to reach a noise of ≃ 10 − 5 . In the time interval of our high- cadence observations of one night, about 1hr, however the planet is changing its RV too much slowly to be seen, and when averaging the CCFs we also average the planetary signal we are looking for. We then decided to make a CCF re f for each three consecutive nights, normalizing each CCF for the respective CCF re f depending on the night of observation. In this way most of the noise disappears, but anyway we are not able to find a significant planetary CCF signal. If we try to make the analysis excluding one night at a time, the final shape and level of noise obtained changes (e.g., Figure 3). 110

Twenty years of giant exoplanets - Proceedings of the Haute Provence Observatory Colloquium, 5-9 October 2015 Edited by I. Boisse, O. Demangeon, F. Bouchy & L. Arnold Figure 2: An example of night-to-night variations of the continuum of the stellar CCF. Di ff erent colors refer to di ff erent nights. 3 Conclusion Night-to-night variations of the CCF are dominant in the analysis, and do not allow us to fully exploit the very high S / N of our spectra for the search of the reflected light from τ Boo b using high-cadence HARPS-N data. When looking for the reflected light of a planet, care has to be taken to consider the e ff ects of night-to-night variations, that can be due to Moon contamination or atmospheric conditions. Since the planetary CCF should have a very low signal, at about 10 − 5 with respect to the stellar one, even small variations of the stellar CCF between di ff erent nights could result in adding noise and in creating false ”bumps” that could be misunderstood as the signal searched. With this analysis, we can only give an upper limit for the albedo of τ Boo b, which is 0.25. Acknowledgments: The GAPS project acknowledges support from INAF through the ”Progetti Premiali” funding scheme of the Italian Ministry of Education, University, and Research. We thank all the organizers of this conference at OHP. References Borsa, F., Scandariato, G., Rainer, M., et al. 2015, A&A, 578, A64 Brogi, M., Snellen, I. A. G., de Kok, R. J., et al. 2012, Nature, 486, 502 Cosentino, R., Lovis, C., Pepe, F., et al. 2012, in Proceeding of the SPIE conference, Vol. 8446, Ground-based and Airborne Instrumentation for Astronomy IV, 84461V Covino, E., Esposito, M., Barbieri, M., et al. 2013, A&A, 554, A28 Martins, J. H. C., Santos, N. C., Figueira, P., et al. 2015, A&A, 576, A134 111

Twenty years of giant exoplanets - Proceedings of the Haute Provence Observatory Colloquium, 5-9 October 2015 Edited by I. Boisse, O. Demangeon, F. Bouchy & L. Arnold Figure 3: Result for the planetary CCF of τ Boo b. The ”bump” seen at velocity 0, where the CCF is expected, is an artifact due to the non-homogeneities between the di ff erent nights of observation. 112