Direct Imaging of Exoplanets Closer to Stars

Citation

Agrawal, Shubh (2022) Direct Imaging of Exoplanets Closer to Stars. Senior thesis (Major), California Institute of Technology. doi:10.7907/17sv-vf40. https://resolver.caltech.edu/CaltechTHESIS:06192022-010740124

Abstract

Detecting exoplanets through direct imaging at lower angular separations, where more planets are expected to be, is limited by the variability of the stellar point spread function. Integral field spectrographs like OSIRIS at the Keck Observatory can leverage high spectral resolution to search for new planets at smaller separations (< 0.3 arcseconds) by detecting their distinct spectral signature compared to the diffracted starlight. In this thesis, we present the mid-survey results of a search for planets around 23 targets in the Ophiuchus and Taurus star-forming regions.

We use this pathfinder survey with Keck/OSIRIS to demonstrate our technique and compare the final sensitivities to other classical imaging techniques, particularly at separations of 0.05-0.3 arcseconds. We detect an M dwarf companion around HD 148352 at a ≈ 34σ significance level. We measure this binary star companion to be at an angular separation of roughly $0.11$ milliarcseconds, with a contrast of $0.38\%$, effective temperature T_eff ≈ 3200 K, and radial velocity RV ≈ 12 km/s. We also present other low-significance objects, along with detection maps and sensitivity limits around these 23 targets.

We use our open-source data analysis pipeline, called the Broad Repository for Exoplanet Analysis, Detection, and Spectroscopy (breads), as the framework for this planet search. breads operates on high spectral resolution data from existing and in-development instruments. Our code is based on a forward-modeling framework, which is statistically more accurate than classical cross-correlation techniques. It includes a built-in optimization and analytical marginalization of linear parameters in the forward model, therefore limiting the number of parameters to be explored by the posterior sampling method. We allow users to select forward models, parameters to detect and analyze, and fitting methods like Markov Chain Monte Carlo sampling, grid optimization, and gradient descent. breads provides a flexible framework to retrieve radial velocity, spin, and atmospheric parameters of high-contrast companions. We also describe wavelength and resolution calibration, transmission and spectra calculation, and bad pixel identification techniques.

Our work will be applicable to future integral field spectrographs like NIRSpec on the James Webb Space Telescope and other first light instruments on the future Extremely Large Telescopes, which are poised to become the next generation of exoplanet detection facilities.

Thesis Files