The Transit Light Source Effect

Abstract

Transmission spectroscopy provides a powerful probe of exoplanet atmospheres, enabling constraints on their compositions and structures. Recent advances in instrumentation and observational techniques have enabled detections of molecules in the atmospheres of exoplanets as small as Neptune as well as provided constraints on cloud properties for Earth-sized and super-Earth exoplanets. However, these precise observations have also revealed that the heterogeneous nature of stellar photospheres presents a significant challenge to high-precision transit depth determinations. This owes to a fundamental limitation of the transmission spectroscopy technique, which is that transiting exoplanet atmospheres are illuminated by a spatially resolved region of the stellar photosphere, the spectrum of which we cannot directly measure. Any difference between the out-of-transit disk-averaged emergent spectrum of the star—our necessary reference by which we measure transit depths—and the average emergent spectrum of the transit chord—the true light source for the transmission measurement—will imprint on the observed transmission spectrum. This phenomenon is what I term the transit light source effect. In this thesis, I present my work to understand the transit light source effect in F to M dwarf systems and constrain stellar contamination signals in transmission spectra from two M dwarf systems. I first describe a modeling effort to constrain spot and faculae covering fractions and the concomitant stellar contamination spectra on M dwarfs. I find that large covering fractions of active regions are possible for typically active M dwarfs, and therefore stellar contamination signals can be likewise large and even overwhelm planetary atmospheric features produced by small transiting planets. This is indeed what I find in two observational studies of transiting M dwarf systems: the M4.5V GJ 1214 system and the M8V TRAPPIST-1 system. I then expand the analysis to F5V to K9V spectral types, investigating stellar contamination signals with a model similar to that presented for M dwarfs. I find that stellar contamination signals are much weaker for typical F to K dwarfs than for M dwarfs, though signals are detectable in high-precision transmission spectra, and active G and K dwarfs, in particular, can impart relatively large transit depth changes. Finally, I summarize the findings of this thesis and conclude with a look toward future prospects for disentangling stellar and planetary signals in exoplanet transmission spectra.