The Science
The Motivation
The decade prior to the turn of the millennium saw the first confirmed detections of extrasolar planets. Shortly thereafter, the first evidence for evaporating exoplanets was seen (Vidal-Madjar et al. 2003). With the thousands of detections thanks primarily to the Kepler misson in recent years, it has become possible to perform robust (though skewed, due to the detection bias toward large, close-in planets) statistical analyses of the exoplanetary population. These have shown a marked gap in the radius distribution of planets which has been termed the "photoevaporation valley" (Fulton et al. 2017, see Owen & Yanqin 2017 for a theoretical treatment).
This suggests that photoevaporation is an important factor in the evolution of some types of planets. In particular, it appears to result in two distinguishable families of terrestrial planets divided by insolation, with low-insolation (less than 10 Earth insolation) terrestrial planets having been born rocky and high-insolation (greater than 10 Earth insolation) terrestrial planets having been evaporated from gaseous sub-Neptunes (Swain et al. 2018). With a large part of the exoplanet search focused on finding Earth-analogues, it is therefore essential that we understand the process of photoevaporation.
We all know the real reason people work on exoplanets... 
Because there are only a handful of planets with observed evaporation signatures (five Jupiters and two Neptunes ), they provide the testbed for our theoretical understanding of photoevaporation. Observations in the Lyman-α line of hydrogen show absorption in the evaporated envelope of these planets out to Doppler shifts of as much as ±150 km/s, with varying degrees of asymmetry. A variety of physical mechanisms have been proposed whose effects might explain these observations, including radiation pressure, charge exchange, and confinement by the stellar wind or by magnetic fields. It is our purpose to determine the processes necessary to replicate these observations in simulations.
The Work
Working with AstroBEAR, we have implemented a planet consisting of a hydrostatic ball of hydrogen that represents large gaseous planets. We use plane-parallel radiative transfer to propagate a stellar flux of ionizing photons across the simulation, which, when absorbed by the planet, heat the atmosphere and cause a wind to blow off of the planet.
We have so far explored a small, highly-inflated region of the mass-radius-insolation parameter space, as well as modelling HD 209458b. In the lowest-mass region, we found that the evaporating wind causes a tail of neutral material to be pulled from the night side of the planet and extend a significant distance from the edge of the planet. On the other hand, the higher-mass planet shows no such tail. The most realistic planet, modelled after HD 209458b (shown above), has a very different flow pattern that also results in a stream of neutral material from the night side of the planet.
We are able to make synthetic observations in the Lyman-α line to view the effects of the physical processes implemented. We have begun by investigating the effects of radiation pressure due to the Lyman-α radiation from a host star, finding that the Lyman-α flux measured by Wood et al. (2005) (the intermediate-flux case in the following figure) is too low to create the observed absorption patterns.
Upcoming Projects
We have begun investigating the effects of charge exchange on the observations of HD 209458b. Charge exchange occurs when hot (~106 K) ionized material from the stellar wind interacts with cold (~104 K) neutral material from the planetary wind and an exchange of electrons occurs, creating a population of ~106 K neutral hydrogen. The thermal broadening of the Lyman-α line could then potentially create sufficient column density at the observed high velocities. This is implemented as multi-species advection, governed by the following equations:
It is also simple, though potentially computationally expensive, to attempt to simulate observations of hot Neptunes. The resolution per atmospheric scale height depends only on the size of the box relative to the planetary radius. Therefore, if we want to keep a simulation domain of the same size with the same resolution per scale height, we have to change the resolution by a factor of Rold/Rnew. To change from HD 209458b to GJ 436b, this is a factor of 3.6 (or, if the simulations are equivalent in computational cost per cell, a week's execution time becomes approximately a month, ignoring any savings from AMR).
Development Projects
New capabilities are continuously being added to AstroBEAR as they are required to tackle a problem. A short list of projects related to the problem of photoevaporation that we would like to implement:
- Spherical (from point source) line transfer
- Metastable He 23S triplet excited state, along with emission and absorption lines
In addition, there are significant portions of AstroBEAR ripe for refactoring and improved documentation.