Accurate high-contrast imaging polarimetry of exoplanets with SPHERE/IRDIS

Abstract

The search for and characterization of exoplanets and other sub-stellar companions are hot topics in contemporary astronomy. Currently, the characterization of the atmospheres of exoplanets through direct imaging leverages on the analysis of only the intensity of their light as a function of wavelength and time. Additional information on the composition and structure of planetary atmospheres - that cannot be obtained with spectroscopy - can be deduced with polarimetry, i.e. measuring the direction of oscillation of the electric fields of light. Not only the starlight that an exoplanet reflects is expected to be polarized, but also the thermal emission of a planet, as this radiation from inside the atmosphere will be scattered by cloud and haze particles on its way up. Indeed, the polarized thermal radiation of several field brown dwarfs has already been measured and is attributed to the scattering of the radiation by patchy clouds in their atmospheres. The degree of linear polarization of hot exoplanets at near-infrared wavelengths is expected to generally be larger than 0.1% and could be up to several percent in some cases. Measurements of the polarized thermal emission of exoplanets can provide information on the presence and patchiness of atmospheric clouds and hazes, the cloud top pressure, spatial structure such as rotational flattening and cloud bands, the atmospheric rotation rate, and the surface gravity and mass of the companion. By determining the angle of linear polarization, the planet's projected spin axis could be constrained. The recently commissioned VLT instrument SPHERE is a high spatial resolution, high-contrast, direct imaging instrument that is specifically designed to detect and characterize giant exoplanets orbiting nearby stars. SPHERE's near-infrared instrument arm IRDIS has a dual-beam polarimetric mode that is primarily used for high-contrast imaging of circumstellar disks, as it is expected to be too insensitive to directly measure the polarized thermal emission of exoplanets. However, IRDIS has already detected exoplanets with huge signal-to-noise ratio using angular differential imaging (ADI) of thermal fluxes. The aim of this thesis is therefore to investigate the feasibility of combining ADI and accurately calibrated polarimetry with SPHERE/IRDIS to for the first time detect and possibly characterize exoplanetary atmospheres through direct imaging polarimetry at near-infrared wavelengths. To assess whether IRDIS can detect the expected polarization signal of exoplanets, and using HR 8799's planetary system as a study case, IRDIS' polarimetric sensitivity is estimated by extrapolating the results from VLT/NaCo observations to IRDIS, and by simulating single-beam polarimetric measurements with real IRDIS data. Based on these estimates, SPHERE/IRDIS is expected to reach sub-percent polarimetric sensitivity when combining extreme adaptive optics, coronagraphy, ADI with advanced data reduction techniques and dual-beam polarimetry. A model describing the modification of the polarization signal induced by the telescope and instrument is established and validated with available internal calibration measurements and observations of a standard star. It appears that for some filters and particular combinations of the parallactic and altitude angle, only a very small part (~10%) of the incident linearly polarized signal is actually measured by the instrument. This loss of signal is accompanied by large offsets in the angle of linear polarization. It is also found that the instrumental polarization of the telescope and SPHERE's first mirror varies with telescope altitude angle and can reach values of a few percent. To limit the instrumental polarization and the loss of signal, an observation strategy is presented for IRDIS' polarimetric mode. In addition, a data reduction method is developed that uses the instrument model to derive an exoplanet's true degree and angle of linear polarization from the measured polarization signal. It is estimated that, after correcting for the modification of the polarization signal induced by the complete optical system, a polarimetric accuracy of ≤0.1% is reached. Given that an exoplanet will generally be between a few tenths of a percent and a percent polarized, it is concluded that, if SPHERE/IRDIS can indeed attain the predicted sub-percent polarimetric sensitivity when combining angular differential imaging and polarimetry, exoplanetary atmospheres can be characterized for the first time through direct imaging polarimetry.