Research

My research focuses on exoplanet atmospheres. Using a combination of observation and modelling, I try to understand the structure, composition, and dynamics of these distant worlds. I am particularly interested in observing small planets with JWST, but I have also been observing and modelling photoevaporation from young mini Neptunes. For more details, click one of the following links:

1. Atmospheric characterization with JWST
2. Photoevaporation from small planets
3. Atmosphere Retrievals
4. Other work (see Publications)

Atmospheric characterization with JWST

JWST is opening a new era for exoplanet atmosphere characterization, and I am fortunate to have been involved from the very beginning. I was heavily involved in writing the proposal for JWST GO 1952 (PI: Renyu Hu) and in reducing the data with my independent pipeline (SPARTA). This program led to the first convincing evidence for an atmosphere on a rocky planet (Hu et al 2024). I performed the fiducial data reduction for the first JWST phase curve (of the mini-Neptune GJ 1214b; Kempton 2023), which revealed a high-metallicity atmosphere with reflective aerosols. I have contributed reductions and retrievals to the Early Release Science program (WASP-39b, WASP-43b), and I am leading an effort to comprehensively characterize the atmosphere of the hot Jupiter HD 189733b using both transmission and emission spectra (Zhang et al, submitted).

My main focus is finding out which rocky planets around M dwarfs host atmospheres. This question holds enormous implications for habitability and our ability to detect biosignatures. To help answer it, I am using JWST to observe the thermal emission from these planets. I am the PI of JWST GO 2508, which resulted in the first JWST phase curve of the sub-Earth GJ 367b (Zhang et al 2024), as well as GO 3784, which observed a phase curve of a hot super-Earth (TOI 2445Ab). I am also playing an important role in the JWST programs of my collaborators, which have observed GJ 486b (Mansfield et al 2024), GJ 1132b (Xue et al 2024), and LTT 1445Ab (Wachiraphan et al 2024). None of these planets have thick atmospheres, as evidenced by their hot daysides and (for GJ 367b and TOI 2445Ab) cold nightsides; and neither does another planet I am closely involved in characterizing. Taken together, the emerging picture is that M dwarf rocky planets with equilibrium temperatures >~430 K do not have atmospheres.

To push toward lower temperatures, STScI has implemented a 500 hour Director's Discretionary Time program. I plan to work with graduate student Qiao Xue to write an independent pipeline for this data, and explore the use of AI in reducing systematics.

Photoevaporation from mini-Neptunes

Photoevaporation dramatically sculpts exoplanet demographics and atmospheric properties. For example, the observed radius distribution of close-in sub-Neptune-sized planets is bimodal, with peaks at < 1.5 R_⊕ and 2-3 R_⊕ (Fulton et al 2017, Fulton et al 2018). This bimodality can be explained by scenarios in which the observed population of sub-Neptune-sized planets formed with a few M_⊕ rocky cores and hydrogen-rich atmospheres, which were then stripped away from the most highly irradiated planets. To reveal the fundamental nature of mini-Neptunes and test whether and how they evolve into super-Earths, I designed programs with Hubble, Keck, and Magellan to observe the escaping atmospheres of young mini-Neptunes in both hydrogen (Ly alpha) and helium (the 1083 nm triplet). These programs achieved the first detection of atmospheric escape from a mini-Neptune, and account for 1/3 of the detections of escaping helium currently in the literature. Out of 7 young (~few hundred Myr) mini-Neptunes observed in the 1083 nm helium triplet, 6 are strongly detected and the seventh shows evidence of episodic outflows. These results show that mass loss from young mini-Neptunes is ubiquitous, that the mass loss rate is enough to turn many of them into super-Earths, and that the width of the absorption is consistent with photoevaporation, not with core-powered mass loss.

Recently, I have been exploring the use of the helium triplet as an alternative metallicity probe of mini-Neptune atmospheres. In the immediate future, I will focus on modelling helium outflows from multi-planet systems because I recently detected, for the first time, escaping helium from two planets (both mini-Neptunes) orbiting the same star. In a year or two, after accumulating more observations from the southern hemisphere, I will publish the final results from my all-sky survey of young mini Neptunes.

Atmosphere Retrievals

I am the main developer of Planetary Atmospheric Tool for Observer Noobs (PLATON). PLATON is a GPU-accelerated Python code that computes the eclipse and transit spectra of exoplanets, given user-specified parameters. It also solves the inverse problem of retrieving atmospheric parameters given observational data.

PLATON is fast, open source, easy to use, and easy to understand. Supported features include equilibrium chemistry with condensation, Mie scattering, opacities generated from up-to-late line lists, surface emission modelling (courtesy of Kim Paragas), and PSIS leave-one-out cross-validation. PLATON has enabled a range of research by multiple groups (e.g. retrievals on HAT-P-11b by Chachan et al 2019, on HD 97658b by Guo et al 2020, on WASP-76b by Fu et al 2020; correlated noise studies by Ih et al 2021).