RV detection

Prior to the first observations of the new near-infrared high-resolution spectropolarimeter SPIRou at the Canada-France-Hawaii Telescope (CFHT), I conducted simulations of velocimetric follow-up of two representative targets of the SPIRou large observing program: the SPIRou Legacy Survey (SLS; 300 nights over 4 years). The first target, TRAPPIST-1, host seven transiting planets, three of which are in the so-called Habitable Zone of the star, where liquid water can theoretically form at the planetary surface. The second, K2-33, hosts the youngest transiting planet known to date, K2-33 b, a 8-11 Myr close-in Neptune-sized planet. Both stars are M dwarfs, significantly cooler than the Sun, which implies that they mostly emit in the near-infrared and are too faint to be observed with state-of-the-art optical velocimeters such as HARPS (at La Silla) or ESPRESSO (at the VLT). Moreover, both stars are magnetically active: they host strong magnetic fields, amplified and sustained by so-called dynamo processes, which produce several time-evolving phenomena such as surface brightness inhomogeneities (spot/plages) and intense flares (these phenomena are also observed for our Sun by the Solar Dynamics Observatory). These phenomena induce photometric and radial velocity (RV) signals that hamper the search for planet signatures. As a consequence, none of the planets in the aforementioned system have seen their mass directly measured with the RV technique.

AU Microscopii (AU Mic), the second closest pre-main-sequence star (~22 Myr) old, hosts at least 2 transiting Neptune-sized planets unveiled from TESS lightcurves in Plavchan et al. 2020 and Martioli et al. 2021. Due to its young age, AU Mic is significantly more active than its more evolved counterparts. Once again, the manifestations of this activity such as giant bright and dark surface features induce RV signals of several 100 m/s in the optical, terribly hampering RV mass measurements of the transiting planets. We collected 27 high-precision nIR observations of the star with SPIRou from September to November 2019. We find that, as expected, the stellar activity RV signal is 3 times lower in the nIR that in the optical. Using two independent methods to model this signal: Gaussian Process Regression and Doppler Imaging, we report a 3.8σ detection of the mass of the close-in planet AU Mic b in Klein et al. 2021b. From the mass and radius, we obtained the first estimate of the inner density of a planet younger than 100 Myr.

During my postdoc at the University of Oxford, I participated to the analysis of the observations collected as part of an intensive spectroscopic monitoring of AU Mic with HARPS. Using the high-performance framework of multidimensional Gaussian processes to jointly model the stellar activity signals and the planet RV signatures, we obtained, for the first time, a robust detection of the mass of AU Mic c (the framework is included in the open-source software PYANETI - check also the website of Oscar Barragan for detailed explanations). Surprisingly, planet c, which is smaller than planet b, seems to be significantly more massive. This result is in tension with standard planet formation paradigms where more massive planets should accrete more material. The origin of this tension is unclear, but could be due to a giant impact between two further out planets forming planets c. Further observations are needed to test different scenarios explaning the observations. Stay tune, the article should soon pop up on arXiv (Zicher, Barragan, Klein et al. 2022, in prep).