Binary Host Star Spectroscopy

In the last year, I have expanded my study of host star compositions to include a wider suite of elements (spanning condensation temperatures of 40-1800 K) and am approaching the star-planet connection from a di erent angle. Many investigations have focused on the "planet-centric" perspective -- how does host star composition influence the type of planets that form? -- but a growing body of work aims to address the "star-centric" perspective -- how does the formation and evolution of planets change the original, pre-planet composition of stars? -- by examining di erences between very similar stars that do/do not host known planets, or that are known to host di erent types of planets. Binary stars provide an ideal laboratory for investigating the eff ects of planet formation on stellar composition because they likely formed in the same environment/from the same material, so are not a ected di erently by, e.g., chemical evolution, motion through the Galaxy, or dust cleansing by luminous stars. Furthermore,
studying "twin" stars (Delta-Teff < 100 K; Delta-log g < 0.1 dex) allows a very precise relative abundance measurement, as systematic uncertainties that usually dominate abundance analyses are so similar that they essentially cancel out, leaving only observational noise that can be pushed down with very high S/N, high-resolution spectra.

Above: From Teske et al. (2015). The (XO-2N - XO-2S) relative abundances versus Tc (Lodders 2003). Black asterisks, blue closed/open triangles, and red circles show results from using the “original” parameters, “alternative params 1”, “alternative params 2”, and “alternative params 3”, respectively. A dotted line shows zero difference. The important take-away is that all models show a Si, Fe, and possibly Ni enhancement in XO-2N. Below: From Ramirez et al. (2015), who confirmed and extended my results on the same binary system. Chemical composition di erence between XO-2N and XO-2S as a function of the elements' condensation temperature. The dashed line corresponds to identical chemical composition. The solid line is a double linear t to the data, broken at T_C = 841 K. Red vertical lines connect two species of the same chemical element (e.g., CH and C I, Ti I and Ti II, etc.).

While previous work suggested abundance di erences of 0.015-0.03 dex are anomalous for binary stars (Gratton et al. 2001; Desidera et al. 2004; 2006), in these studies the binaries were not always "twins" , and the data were not always high S/N (> 400). More recent work on the true "twin" system XO-2, in which both stars host planets, suggests that diff erences in stellar abundances as small as 0.015 dex could be due to a di erent type of planet formation around one star versus the other (Teske et al. 2015a; Ramrez et al. 2015; Biazzo et al. 2015). Even this small di erence amounts to 0.57 MJ of volatile-rich material potentially "missing" from XO-2S, instead locked up in its two gas giant planets. Studying "twin" star abundance di erences may prove to be the most useful for constraining both the eff ects of planet formation on stellar abundances, and the bulk compositions of giant planets, but we need more data to understand their signi cance.

Above: From Teske et al. (2016a). The Δ(A-B) relative abundances versus T_c (Lodders 2003), calculated using the (A-B) stellar parameters derived in this work. Shown in orange is the best fit to the abundances, not including K, with a smooth T_c break point. The grey line shows the solar twin trend from Melendez et al. (2009) with a constant 0.015 dex added to account for the [Fe/H] difference between A and B.

I recently had another first-author paper accepted (with Ivan and a fantastic undergrad student at UT Austin) that reports a detection of volatile depletion ( -0.02 dex) and refractory enhancement (~0.01 dex) in one component of the one of the only remaining unstudied stellar "twin" system in which both stars host planets, WASP-94. The 1.45 M_Jupiter, 1.62 R_Jupiter WASP-94Ab transits its host star, orbits at 0.055 AU, and has an orbit consistent with zero eccentricity, but displays strong evidence of the Rossiter-McLaughlin effect (it is misaligned with the host star spin axis and likely orbits in retrograde). WASP-94Bb does not transit, orbits at 0.034 AU, has a minimum mass of 0.62 M_Jupiter, and also has an orbit consistent with circular. The two stars are very similar in spectral type (F8 and F9), but their wide separation (~2700 AU) suggests their protoplanetary disks were likely not influenced by stellar interactions. And yet WASP-94Ab's orbit -- misaligned with the host star spin axis and likely orbits in retrograde -- points towards a dynamically active formation mechanism, perhaps different than that of WASP-94Bb.

Our results for WASP-94A&B are di fferent than every other published case of binary host star abundances, in which either no signi cant abundance differences are reported, or there is some degree of enhancement in all elements, including volatiles. There is not a single obvious explanation for the measured abundance differences, but they may actually be related to the small difference in convection zone sizes of WASP-94A (6194+/-5 K) and B (6112+/-6 K). The effective temperature of both stars put them near the border, where the convection zone becomes negligible, that Winn et al. (2010) suggest separates systems in which tidal dissipation damps planetary obliquities within a
few Gyr versus those in which dissipation is ineffective.  Perhaps the 2x larger convection zone of WASP-94B versus A (based on Pinsonneault et al. 2001) was enough to better facilitate tidal circularization of its giant planet; this may have affected how material was accreted into the stellar photosphere during giant planet migration.

Studying ``twin'' star abundance differences may prove to be the most useful for constraining both the effects of planet formation on stellar abundances, and the bulk compositions of giant planets, but we need more data to understand their significance; there are only eight such detailed studies of planet-hosting binary ``twin'' systems, of which I have led three. The most recent of these is another first-author paper that announces the discovery of planets around stars in a known "twin" binary system and an accompanying precision stellar abundance analysis of both stars. This discovery relies almost entirely on data from the Magellan Planet Finder Spectrograph, and is exciting because of the somewhat-unique discovered planet properties. Check it out here!