Tracing planetary formation with chemistry

Planets form in disks made of gas, dust, and ice that surround young stars. These 'protoplanetary' disks are vast structures, covering regions of space many times larger than the entire Solar System. They are also chemically complex, with compositions that vary dramatically from the centre to the outer edge. A young planet grows by accreting material from the disk, and we can therefore expect the composition of a planet to be directly related to the region of the disk in which it formed. Drawing links between the composition of planets and disks is fundamental to understanding planetary formation processes.

Two elements of particular importance are carbon and oxygen, both of which are among the most abundant elements in the Universe. In a protoplanetary disk, most of the carbon and oxygen is locked into in simple molecular species such as CO and H₂O. In the warm inner part of the disk, these molecules can easily exist in the gas phase. In the outer parts of the disk, further from the central star, temperatures decrease and molecules 'freeze out' on to the surfaces of dust grains. Different molecules freeze out at different temperatures, so the overall effect is that the total amount of carbon or oxygen in the gas phase varies as a function of radius. By measuring the total amount of carbon and oxygen in an exoplanet's atmosphere, we can link it to the region of the disk in which it formed. This is typically done using the total carbon-to-oxygen ratio (C/O).

Determining the precise C/O in a disk is not easy, but we can generally distinguish between two different scenarios; oxygen-dominated chemistry (C/O < 1), or carbon-dominated chemistry (C/O > 1). One way to do this is by using observations of two sulfur-bearing molecules, CS and SO. Studies have shown that the CS/SO ratio varies dramatically for small changes in C/O, meaning we can use it as a robust C/O 'probe'. 

In our paper, we used observations of CS and SO to uncover an entirely unexpected chemical variation in a protoplanetary disk. We found that these molecules trace an azimuthal variation in the C/O ratio. Unlike the 'classical' radial variations described above, this variation affects only a small angular region of the disk. This is the first time such a variation has been observed.

Observations: CS and SO with ALMA

We obtained observations of both the CS and SO molecules using the Atacama Large Millimeter/sub-millimeter Array (ALMA), towards a protoplanetary disk called HD 100546. The observations show emission from molecular gas at millimeter wavelengths (see figure below). In each of these observations it's clear that the emission is asymmetric, ie. offset from the central star (denoted by the green 'x'). The SO emission is offset on one side of the disk, while the CS emission if offset on the other side.

The peculiar asymmetries in the emission of these molecules is also evident in their spectral line profiles. The SO spectral line exhibits a double-peaked structure, with the intensity of the blue-shifted peak approximately twice that of the red-shifted peak. Conversely, the CS spectral line exhibits a narrow peak that is slightly red-shifted (see figure below).

Using complex thermo-chemical models, we showed that these asymmetries can be explained by an azimuthal C/O variation in the disk chemistry. Throughout most of the disk, the chemistry is oxygen-dominated, with a 'Solar-like' C/O ratio of ~0.5. In this region, the production of the oxygen-bearing SO molecule is favoured. However, a small angular region on the western side of the disk (right side of the image) has an elevated C/O ratio of ~1.5. Here, the chemistry is carbon-dominated, favouring the production of CS. It is this chemical dichotomy that leads to the observed asymmetries.

What's causing the azimuthal C/O variation?

Asymmetries in molecular line emission have been observed in many protoplanetary disks. While a number of mechanisms can account for these, the cause is not always clear. In our paper we proposed a new mechanism, linked to a newly-forming planet in the inner part of the HD 100546 disk. 

Observations and modelling provide strong evidence for the presence of a protoplanet orbiting the central star at around 13 au. We hypothesised that an 'overdensity' of dust associated with the planet casts a shadow on the outer disk. This causes temperatures to drop, leading to the additional freeze-out of water, locking a considerable fraction of gas-phase oxygen into ices. The net effect is to increase the C/O ratio, which leads to more CS forming in the shadowed region, and less SO.

The interesting thing here is that, since the shadow is linked to a planet, the shadow 'orbits' the disk just like the planet does. We can use the orbital period of the planet, and the size of the shadow, to calculate how long a given part of the disk remains in shadow. This turns out to be about five years. 

Using this information, we tested the feasibility of shadowing mechanism using an analytical model. By considering the timescales of important processes, such as cooling, freeze-out, and chemical conversion, we showed that it is possible to reproduce the observed asymmetries within the five year window. Now that we have established a baseline, new observations with ALMA could potentially show if the shadow moves as expected.

Implications for planets

Observations also suggest that a second planet may exist in the outer parts of the HD 100546 disk. How might such a planet be affected by these azimuthal C/O variations?

Orbital timescales in the outer disk are much longer that the inner disk. This means that any planet situated at large distances from the star will move in and out of the shadowed region many times over the course of its lifetime. A planet which moves in and out of two chemically distinct regions as it evolves can be expected to have a chemically complex atmosphere, formed of material accreted from both regions. The degree to which the atmospheric composition mirrors that of either region in the disk will be governed by complex chemical and physical processes.

The results presented in our paper therefore add a new consideration to the way in which we interpret observations of both disks and planets. The classical view of a radially varying C/O ratio must be readdressed if we are to draw meaningful links between the composition of exoplanet atmospheres and the disks in which they form. Determining the C/O ratio at small spatial scales must be a major goal of future observations, if we are to build models that can meaningfully predict planetary formation pathways.