Cool Jupiter Host Star Abundances
The relationship between the compositions of giant planets and their host stars is of fundamental interest in understanding planet formation. The solar system giant planets are enhanced above solar composition in metals (anything heavier than H and He), both in their visible atmospheres and bulk compositions. (Check out the latest on Jupiter’s oxygen abundance!) A key question is whether the metal enrichment of giant exoplanets is correlated with that of their host stars. We know giant planets are more often found around stars with higher iron abundances ([Fe/H]) and/or higher bulk metal ([M/H]) abundances, so we were curious, do metal-rich stars also make metal-rich planets? I set out to investigate this in collaboration with Daniel Thorngren and Jonathan Fortney, along with Natalie Hinkel and John Brewer.
Daniel previously looked into this question by estimating the heavy element mass of transiting cool Jupiter exoplanets (which are not affected by anomalous radius inflation like hot Jupiters), based on their measured masses and radii. He found that as the planet mass increased, so did the heavy element mass, just like in our much more limited sample of solar system gas giants (see figure below).
From Thorngren et al. (2016). Heavy element masses of planets and their masses. The lines of constant Zplanet are shown at values of 1 (black), 0.5, 0.1, and .01 (Gray). Distributions for points near Zplanet=1 tend to be strongly correlated (have well-defined Zplanet values) but may have high-mass uncertainties. No models have a Zplanet larger than one. The distribution of fits (see Section 4 for discussion) is shown by a red median line with 1, 2, and 3σ contours. Note Kepler-75b at 10.1 MJ, which only has an upper limit.
Interestingly, Daniel also found a relationship between the planet metal-enrichment relative to the parent star (Zplanet/Zstar) and the planet total mass Mp, but no relation between just Zplanet and Mp. This made Daniel and his coauthors think that maybe Zstar was playing a role in setting the planet heavy element mass. In my paper, dug into this idea further, comparing the heavy element masses of a smaller sample of 24 cool Jupiter planets to a wider range of host star elemental abundances, focusing on elements important for forming planets (C, O, Mg, Si, Fe, and Ni). Obtaining the spectra and doing all of the abundance measurements for these stars (two of which came from John’s previous work) was not easy and not quick -- it required multiple semesters of time on the Keck and Magellan telescopes.
In the end, we were surprised to find no clear correlation between stellar metallicity and planet residual metallicity (the relative amount of metal versus that expected from the planet mass/radius alone; see left column in plot below), which is in conflict with common predictions from formation models. However, we also found a potential correlation between residual planet metals and stellar volatile-to-refractory (so, C and O versus the other elements) element ratios (see middle column in the plot below).
Main summary plot from Teske et al. (2019). Plots, fits, Kendall’s tau correlation tests, and slope histograms of the three major results of this paper. Shaded regions indicate the 1σ uncertainty in the fit, and dotted lines indicate the 1σ predictive interval. The abundances for the lighter points are from Brewer & Fischer (2018). On the left, we show the lack of a strong correlation between the total metallicity of a star and the residual metallicity of the planet. A slope of 1 is ruled out at the 2σ level. In the center, we show a correlation we observe between the residual metallicity of a planet and the ratio of stellar volatiles (C and O) to refractory elements (Fe, Si, Mg, and Ni). On the right, we show that our data reproduce the relationship between planetary metal enrichment relative to the parent star and planet mass. “Aggregate” indicates that we consider the total metallicity from all of the measured elements, rather than simply using one.
So, what does this mean? We find the same Zplanet/Zstar correlation as in Thorngren et al. (2016) (see right column above, across all of the elements we studied, but given the lack of correlation between the residual planet mass — the heavy-element mass beyond what is expected given its total mass—and all [Z/H] values, it seems more likely that the Zplanet/Zstar versus Mp relationship is driven by the planet metallicity without much input from stellar metallicity.
Overall, this work presents several new observational relationships between host star and planet composition that should be addressed in future theoretical studies of planet formation. Daniel’s previous work and that of others has shown that we should expect giant planet atmospheres to be highly metal enriched — given the heavy element mass versus the total planet mass, the heavy elements cannot just be in the “core”. So I’m excited to see whether atmospheric observations of these targets with JWST will reveal a different relationship with the host star composition.