Abstract
We present the discovery of HD 221416 b, the first transiting planet identified by the Transiting Exoplanet Survey Satellite (TESS) for which asteroseismology of the host star is possible. HD 221416 b (HIP 116158, TOI-197) is a bright (V = 8.2 mag), spectroscopically classified subgiant that oscillates with an average frequency of about 430 μHz and displays a clear signature of mixed modes. The oscillation amplitude confirms that the redder TESS bandpass compared to Kepler has a small effect on the oscillations, supporting the expected yield of thousands of solar-like oscillators with TESS 2 minute cadence observations. Asteroseismic modeling yields a robust determination of the host star radius (R⋆ = 2.943 ± 0.064 R⊙), mass (M⋆ = 1.212 ± 0.074 M⊙), and age (4.9 ± 1.1 Gyr), and demonstrates that it has just started ascending the red-giant branch. Combining asteroseismology with transit modeling and radial-velocity observations, we show that the planet is a "hot Saturn" (Rp = 9.17 ± 0.33 R⊕) with an orbital period of ∼14.3 days, irradiance of F = 343 ± 24 F⊕, and moderate mass (Mp = 60.5 ± 5.7 M⊕) and density (ρp = 0.431 ± 0.062 g cm−3). The properties of HD 221416 b show that the host-star metallicity–planet mass correlation found in sub-Saturns (4–8 R⊕) does not extend to larger radii, indicating that planets in the transition between sub-Saturns and Jupiters follow a relatively narrow range of densities. With a density measured to ∼15%, HD 221416 b is one of the best characterized Saturn-size planets to date, augmenting the small number of known transiting planets around evolved stars and demonstrating the power of TESS to characterize exoplanets and their host stars using asteroseismology.
Export citation and abstract BibTeX RIS
1. Introduction
Asteroseismology is one of the major success stories of the space photometry revolution initiated by CoRoT (Baglin et al. 2006) and Kepler (Borucki et al. 2010). The detection of oscillations in thousands of stars has led to breakthroughs such as the discovery of rapidly rotating cores in subgiants and red giants, as well as the systematic measurement of stellar masses, radii, and ages (see Chaplin & Miglio 2013 for a review). Asteroseismology has also become the "gold standard" for calibrating more indirect methods to determine stellar parameters such as surface gravity (log g) from spectroscopy (Petigura et al. 2017a) and stellar granulation (Mathur et al. 2011; Bastien et al. 2013; Kallinger et al. 2016; Corsaro et al. 2017; Bugnet et al.2018; Pande et al. 2018), and age from rotation periods (gyrochronology; e.g., García et al. 2014; van Saders et al. 2016).
A remarkable synergy that emerged from space-based photometry is the systematic characterization of exoplanet host stars using asteroseismology. Following the first asteroseismic studies of exoplanet host stars using radial velocities (Bazot et al. 2005; Bouchy et al. 2005), the Hubble Space Telescope (Gilliland et al. 2011), and CoRoT (Ballot et al. 2011b; Lebreton & Goupil 2014), Kepler enabled the systematic characterization of exoplanets with over 100 detections of oscillations in host stars to date (Huber et al. 2013b; Lundkvist et al. 2016). In addition to the more precise characterization of exoplanet radii and masses (Ballard et al. 2014), the synergy also enabled systematic constraints on stellar spin–orbit alignments (Benomar et al. 2014; Chaplin et al. 2014a; Lund et al. 2014; Campante et al. 2016a) and statistical inferences on orbital eccentricities through constraints on the mean stellar density (Sliski & Kipping 2014; Van Eylen & Albrecht 2015; Van Eylen et al. 2019).
The recently launched NASA Transiting Exoplanet Survey Satellite (TESS) Mission (Ricker et al. 2014) is poised to continue the synergy between asteroseismology and exoplanet science. Using dedicated 2 minute cadence observations, TESS is expected to detect oscillations in thousands of main-sequence, subgiant, and early red-giant stars (Schofield et al. 2019), and simulations predict that at least 100 of these will host transiting or nontransiting exoplanets (Campante et al. 2016b). TESS host stars are on average significantly brighter than typical Kepler hosts, facilitating ground-based measurements of planet masses with precisely characterized exoplanet hosts from asteroseismology. While some of the first exoplanets discovered with TESS orbit stars that have evolved off the main sequence (Brahm et al. 2018; Nielsen et al. 2019; Wang et al. 2019), none of them were amenable to asteroseismology using TESS photometry. Here, we present the characterization of the HD 221416 (TESS Object of Interest 197, HIP 116158) system, the first discovery by TESS of a transiting exoplanet around a host star in which oscillations can be measured.
2. Observations
2.1. TESS Photometry
TESS observed HD 221416 in 2 minute cadence during Sector 2 of Cycle 1 for 27 days. We used the target pixel files produced by the TESS Science Processing Operations Center (Jenkins et al. 2016) as part of the TESS alerts on 2018 November 11.78 We produced a light curve using the photometry pipeline79 (R. Handberg et al. 2019, in preparation) maintained by the TESS Asteroseismic Science Operations Center (TASOC; Lund et al. 2017), which is based on software originally developed to generate light curves for data collected by the K2 Mission (Lund et al. 2015).
Figure 1(a) shows the raw light curve obtained from the TASOC pipeline. The coverage is nearly continuous (duty cycle ∼93%), with a ∼2 day gap separating the two spacecraft orbits in the observing sector. Two ∼0.1% brightness dips, which triggered the identification of TOI-197.01 as a planet candidate, are evident near the beginning of each TESS orbit (see triangles in Figure 1(a)). The structure with a period of ∼2.5 days corresponds to instrumental variations due to the angular momentum dumping cycle of the spacecraft.
Figure 1. Panel (a): raw TESS 2 minute cadence light curve of HD 221416 produced by the TESS Asteroseismic Science Operations Center (TASOC). The red line is the light curve smoothed with a 10 minute boxcar filter (shown for illustration purposes only). Triangles mark the two transit events. Panel (b): light curve after applying corrections by the TASOC pipeline. Panel (c): power spectrum of panel (b), showing a granulation background and power excess due to oscillations near ∼430 μHz. The solid red line is a global fit, consisting of granulation plus white noise and a Gaussian describing the power excess due to oscillations. Dashed red lines show the two granulation components and the white noise level, respectively.
Download figure:
Standard image High-resolution imageTo prepare the raw light curve for an asteroseismic analysis, the current TASOC pipeline implements a series of corrections as described by Handberg & Lund (2014), which includes the removal of instrumental artifacts and of the transit events using a combination of filters utilizing the estimated planetary period. Future TASOC-prepared light curves from full TESS data releases will use information from the ensemble of stars to remove common instrumental systematics (M. N. Lund et al. 2019, in preparation). Alternative light-curve corrections using transit removal and gap interpolation (García et al. 2011; Pires et al. 2015) yielded consistent results. The corrected TASOC light curve is shown in Figure 1(b). Figure 1(c) shows a power spectrum of this light curve, revealing the clear presence of a granulation background and a power excess from solar-like oscillations near ∼430 μHz, both characteristic of an evolved star near the base of the red-giant branch.
2.2. High-resolution Spectroscopy
We obtained high-resolution spectra of HD 221416 using several facilities within the TESS Follow-up Observation Program (TFOP), including HIRES (Vogt et al. 1994) on the 10 m telescope at Keck Observatory (Maunakea, Hawai'i); the Hertzsprung SONG Telescope at Teide Observatory (Tenerife; Grundahl et al. 2017); HARPS (Mayor et al. 2003), FEROS (Kaufer et al. 1999), Coralie (Queloz et al. 2001), and FIDEOS (Vanzi et al. 2018) on the MPG/ESO 3.6 m, 2.2 m, 1.2 m, and 1 m telescopes at La Silla Observatory (Chile); Veloce (Gilbert et al. 2018) on the 3.9 m Anglo-Australian Telescope at Siding Spring Observatory (Australia); TRES (Fürész 2008) on the 1.5 m Tillinghast reflector at the F. L. Whipple Observatory (Mt. Hopkins, Arizona); and iSHELL (Rayner et al. 2012) on the NASA IRTF Telescope (Maunakea, Hawai'i). All spectra used in this paper were obtained between 2018 November 11 and December 30 and have a minimum spectral resolution of R ≈ 44,000. FEROS, Coralie, and HARPS data were processed and analyzed with the CERES package (Brahm et al. 2017a), which performs the optimal extraction and wavelength calibration of each spectrum, along with the measurement of precision radial velocities and bisector spans via the cross-correlation technique. Most instruments have been previously used to obtain precise radial velocities to confirm exoplanets, and we refer to the publications listed above for details on the reduction methods.
To obtain stellar parameters, we analyzed a HIRES spectrum using Specmatch (Petigura 2015), which has been extensively applied for the classification of Kepler exoplanet host stars (Johnson et al. 2017; Petigura et al. 2017a). The resulting parameters were Teff = 5080 ± 70 K, log g = 3.60 ± 0.08 dex, [Fe/H] = −0.08 ± 0.05 dex, and v sin i = 2.8 ± 1.6 km s−1, consistent with an evolved star as identified from the power spectrum in Figure 1(c). To account for systematic differences between spectroscopic methods (Torres et al. 2012), we added 59 K in Teff and 0.062 dex in [Fe/H] in quadrature to the formal uncertainties, yielding final values of Teff = 5080 ± 90 K and [Fe/H] = −0.08 ± 0.08 dex. Independent spectroscopic analyses yielded consistent results, including an analysis of a HIRES spectrum using ARES+MOOG (Sousa 2014; Sousa et al. 2018), FEROS spectra using ZASPE (Brahm et al. 2017b), TRES spectra using SPC (Buchhave et al. 2012) and iSHELL spectra using BT-Settl models (Allard et al. 2012).
2.3. Broadband Photometry and Gaia Parallax
We fitted the spectral energy distribution (SED) of HD 221416 using broadband photometry following the method described by Stassun & Torres (2016). We used NUV photometry from GALEX, BTVT from Tycho-2 (Høg et al. 2000), BVgri from APASS, JHKS from 2MASS (Skrutskie et al. 2006), W1–W4 from WISE (Wright et al. 2010), and the G magnitude from Gaia (Evans et al. 2018). The data were fit using Kurucz atmosphere models, with Teff, [Fe/H], and extinction (AV) as free parameters. We restricted AV to the maximum line-of-sight value from the dust maps of Schlegel et al. (1998). The resulting fit yielded Teff = 5090 ± 85 K, [Fe/H] = −0.3 ± 0.3 dex, and AV =0.09 ± 0.02 mag with a reduced χ2 of 1.9, in good agreement with spectroscopy. Integrating the (dereddened) model SED gives the bolometric flux at Earth of Fbol = (1.88 ± 0.04) × 10−8 erg s cm−2. An independent SED fit using 2MASS, APASS9, USNO-B1, and WISE photometry and Kurucz models yielded excellent agreement, with Fbol = (1.83 ± 0.09) × 10−8 erg s cm−2 and Teff = 5150 ± 130 K. Additional independent analyses using the method by Mann et al. (2016) and PARAM (Rodrigues et al. 2014, 2017) yielded bolometric fluxes and extinction values that are consistent within 1σ with the values quoted above.
Combining the bolometric flux with the Gaia DR2 distance allows us to derive a nearly model-independent luminosity, which is a valuable constraint for asteroseismic modeling (see Section 3.3). Using a Gaia parallax of 10.518 ± 0.080 mas (adjusted for the 0.082 ± 0.033 mas zero-point offset for nearby stars reported by Stassun & Torres 2018) with the two methods described above yielded L⋆ = 5.30 ± 0.14 L⊙ (using Fbol = (1.88 ± 0.04) × 10−8 erg s cm−2) and L⋆ = 5.13 ± 0.13 L⊙ (using Fbol = (1.83 ± 0.09) × 10−8 erg s cm−2). We also derived a luminosity using isoclassify (Huber et al. 2017),80 adopting 2MASS K-band photometry, bolometric corrections from MIST isochrones (Choi et al. 2016), and the composite reddening map mwdust (Bovy et al. 2016), yielding L⋆ = 5.03 ± 0.13 L⊙. Our adopted luminosity was the mean of these methods with an uncertainty calculated by adding the mean uncertainty and scatter over all methods in quadrature, yielding L⋆ = 5.15 ± 0.17 L⊙.
2.4. High-resolution Imaging
HD 221416 was observed with the NIRC2 camera and Altair adaptive optics system on Keck II (Wizinowich et al. 2000) on UT 2018 November 25. Conditions were clear but seeing was poor (0
8–2''). We used the science target as the natural guide star, and images were obtained through a K-continuum plus KP501.5 filter using the narrow camera (10 mas pixel scale). We obtained eight images (four each at two dither positions), each consisting of 50 coadds of 0.2 s each, with correlated double-sampling mode and four reads. Frames were coadded, and we subtracted an average dark image, constructed from a set of darks with the same integration time and sampling mode. Flat-fielding was performed using a dome flat obtained in the K' filter. "Hot" pixels were identified in the dark image and corrected by median filtering with a 5 × 5 box centered on each affected pixel in the science image. Only a single star appears in the images. We performed tests in which "clones" of the stellar image reduced by a specified contrast ratio were added to the original image. These show that we would have been able to detect companions as faint as ΔK = 5.8 mag within 04 of HD 221416, 3.8 mag within 02, and 1.8 mag within 01.
Additional NIRC2 observations were obtained in the narrowband Br − γ filter (λo = 2.1686; Δλ = 0.0326 μm) on UT 2018 November 22. A standard three-point dither pattern with a step size of 3'' was repeated twice with each dither offset from the previous dither by 05. An integration time of 0.25 s was used with one coadd per frame for a total of 2.25 s on target, and the camera was used in the narrow-angle mode. No additional stellar companions were detected to within a resolution of ∼005 FWHM. The sensitivities of the final combined AO image were determined following Ciardi et al. (2015) and Furlan et al. (2017), with detection limits as faint as ΔBr − γ = 7.4 mag within 04, 6.1 mag within 02, and 3.2 mag within 01.
The results from NIRC2 are consistent with Speckle observations using HRCam (Tokovinin et al. 2010) on the 4.1 m SOAR telescope.81 Because the companion is unlikely to be bluer than HD 221416, these constraints exclude any significant dilution (both for oscillation amplitudes and the depth of transit events).
3. Asteroseismology
3.1. Global Oscillation Parameters
To extract oscillation parameters characterizing the average properties of the power spectrum, we used several automated analysis methods (e.g., Huber et al. 2009; Mathur et al. 2010; Benomar et al. 2012; Kallinger et al. 2012; Mosser et al. 2012a; Corsaro & De Ridder 2014; Lundkvist 2015; Stello et al. 2017; Campante 2018; Bell et al. 2019), many of which have been extensively tested on Kepler data (e.g., Hekker et al. 2011; Verner et al. 2011). In most of these analyses, the power contributions due to granulation noise and stellar activity were modeled by a combination of power laws and a flat contribution due to shot noise, and then corrected by dividing the power spectrum by the background model. The individual contributions and background model using the method by Huber et al. (2009) are shown as dashed and solid red lines in Figure 1(c), and a close-up of the power excess is shown in Figure 2(a).
Figure 2. Panel (a): power spectrum of HD 221416 centered on the frequency region showing oscillations. Vertical dashed lines mark identified individual frequencies. Panel (b): grayscale échelle diagram (see footnote 83) of the background-corrected and smoothed power spectrum in panel (a). Identified individual mode frequencies are marked with blue circles (l = 0, radial modes), green squares (l = 2, quadrupole modes), and red diamonds (l = 1, dipole modes). Note that the diagram is replicated for clarity (Bedding 2012).
Download figure:
Standard image High-resolution imageNext, the frequency of maximum power (νmax) was measured either by heavily smoothing the power spectrum or by fitting a Gaussian function to the power excess. Our analysis yielded νmax = 430 ± 18 μHz, with uncertainties calculated from the scatter between all fitting techniques. Finally, the mean oscillation amplitude per radial mode was determined by taking the peak of the smoothed, background-corrected oscillation envelope and correcting for the contribution of nonradial modes (Kjeldsen et al. 2008a), yielding A = 18.7 ± 3.5 ppm. We caution that the νmax and amplitude estimates could be significantly biased by the stochastic nature of the oscillations. The modes are not well resolved, as demonstrated by the non-Gaussian appearance of the power spectrum and the particularly strong peak at 420 μHz.
Global seismic parameters such as νmax and amplitude follow well-known scaling relations (Huber et al. 2011; Mosser et al. 2012b; Corsaro et al. 2013), allowing us to test whether the detected oscillations are consistent with expectations. Figure 3 compares our measured νmax and amplitude with results for ∼1500 stars observed by Kepler (Huber et al. 2011). We observe excellent agreement, confirming that the detected signal is consistent with solar-like oscillations. We note that the oscillations in the TESS bandpass are expected to be ∼15% smaller than in the bluer Kepler bandpass, which is well within the spread of amplitudes at a given νmax observed in the Kepler sample. The result confirms that the redder bandpass of TESS only has a small effect on the oscillation amplitude, supporting the expected rich yield of solar-like oscillators with TESS 2 minute cadence observations (Schofield et al. 2019).
Figure 3. Amplitude per radial mode vs. frequency of maximum power for a sample of ∼1500 stars spanning from the main sequence to the red-giant branch observed by Kepler (Huber et al. 2011). The red star shows the measured position of HD 221416 (TOI-197). The uncertainties are approximately equal to the symbol size.
Download figure:
Standard image High-resolution image3.2. Individual Mode Frequencies
The power spectrum in Figure 2(a) shows several clear peaks corresponding to individual oscillation modes. Given that TESS instrument artifacts are not yet well understood, we restricted our analysis to the frequency range 400–500 μHz where we observe peaks well above the background level.
To extract these individual mode frequencies, we used several independent methods ranging from traditional iterative sine-wave fitting, i.e., prewhitening (e.g., Kjeldsen et al. 2005; Lenz & Breger 2005; Bedding et al. 2007), to fitting of Lorentzian mode profiles (e.g., Handberg & Campante 2011; Appourchaux et al. 2012; Mosser et al. 2012b; Corsaro & De Ridder 2014; Corsaro et al. 2015; Vrard et al. 2015; Davies & Miglio 2016; Handberg et al. 2017; Roxburgh 2017; Kallinger et al. 2018), including publicly available code such as DIAMONDS.82 We required at least two independent methods to return the same frequency within uncertainties and that the posterior probability of each peak being a mode was ≥90% (Basu & Chaplin 2017). A comparison of the frequencies returned by different fitters showed very good agreement, at a level smaller than the uncertainties for all reported modes. For the final list of frequencies, we adopted values from one fitter who applied prewhitening (HK), with uncertainties derived from Monte Carlo simulations of the data, as listed in Table 1.
Table 1. Extracted Oscillation Frequencies and Mode Identifications for HD 221416
| f(μHz) | σf (μHz) | l |
|---|---|---|
| 413.12 | 0.29 | 1 |
| 420.06 | 0.11 | 0 |
| 429.26 | 0.14 | 1 |
| 436.77 | 0.24 | 1 |
| 445.85 | 0.21 | 2 |
| 448.89 | 0.21 | 0 |
| 460.16 | 0.33 | 1 |
| 463.81 | 0.43 | 1 |
| 477.08 | 0.31 | 1 |
| 478.07 | 0.35 | 0 |
Note. The large frequency separation derived from radial modes is Δν = 28.94 ± 0.15 μHz. Note that the l = 1 modes at ∼460 and ∼463 μHz are listed for completeness, but it is unlikely that both of them are genuine (see the text).
Download table as: ASCIITypeset image
To measure the large frequency separation Δν, we performed a linear fit to all identified radial modes, yielding Δν = 28.94 ± 0.15 μHz. Figure 2(b) shows a grayscale échelle diagram83
using this Δν measurement, including the extracted mode frequencies. The l = 1 modes are strongly affected by mode bumping, as expected for the mixed-mode coupling factors for evolved stars in this evolutionary stage. The offset 
of the l = 0 ridge is ∼1.5, consistent with the expected value from Kepler measurements for stars with similar Δν and Teff (White et al. 2011).
3.3. Frequency Modeling
We used a number of independent approaches to model the observed oscillation frequencies, including different stellar evolution codes (ASTEC, Cesam2K, GARSTEC, Iben, MESA, and YREC; Iben 1965; Christensen-Dalsgaard 2008; Demarque et al. 2008; Morel & Lebreton 2008; Scuflaire et al. 2008; Weiss et al. 2008; Paxton et al. 2011, 2013, 2015; Choi et al. 2016), oscillation codes (ADIPLS, GYRE, and Pesnell; Pesnell 1990; Christensen-Dalsgaard 2008; Townsend & Teitler 2013), and modeling methods (including AMP, ASTFIT, BeSSP, BASTA, and PARAM; Deheuvels & Michel 2011; Lebreton & Goupil 2014; Rodrigues et al. 2014, 2017; Silva Aguirre et al. 2015; Yıldız et al. 2016; Ball & Gizon 2017; Creevey et al. 2017; Serenelli et al. 2017; Mosumgaard et al. 2018; Tayar & Pinsonneault 2018; Ong & Basu 2019). Most of the adopted methods applied corrections for the surface effect (Kjeldsen et al. 2008b; Ball & Gizon 2017). Model inputs included the spectroscopic temperature and metallicity, individual frequencies, Δν, and the luminosity (Section 2.3). To investigate the effects of different input parameters, modelers were asked to provide solutions using both individual frequencies and only using Δν, with and without taking into account the luminosity constraint. The constraint on νmax was not used in the modeling because it may be affected by finite mode lifetimes (see Section 3.1).
Overall, the modeling efforts yielded consistent results, and most modeling codes were able to provide adequate fits to the observed oscillation frequencies. The modeling confirmed that only one of the two closely spaced mixed modes near ∼460 μHz is likely real, but we have retained both frequencies in Table 1 for consistency. An échelle diagram with observed frequencies and a representative best-fitting model is shown in Figure 4.
Figure 4. Échelle diagram showing observed oscillation frequencies (filled gray symbols) and a representative best-fitting model (open colored symbols) using GARSTEC, ADIPLS, and BeSSP (Serenelli et al. 2017). Model symbol sizes for nonradial modes are scaled using the mode inertia (a proxy for mode amplitude) as described in Cunha et al. (2015). Thick model symbols correspond to modes that were matched to observations. Uncertainties on the observed frequencies are smaller than or comparable to the symbol sizes. Note that the l = 1 mode at 460 μHz has been omitted from this plot (see the text).
Download figure:
Standard image High-resolution imageIndependent analyses confirmed a bimodality splitting into lower mass, older models (∼1.15 M⊙, ∼6 Gyr), and higher mass, younger models (∼1.3 M⊙, ∼4 Gyr). Surface rotation would provide an independent mass diagnostic (e.g., van Saders & Pinsonneault 2013), but the insufficiently constrained v sin i and the unknown stellar inclination mean that we cannot decisively break this degeneracy. Combining an independent constraint of log g = 3.603 ± 0.026 dex from an autocorrelation analysis of the light curve (Kallinger et al. 2016) with a radius from L and Teff favors a higher mass solution (M⋆ =1.27 ± 0.13 M⊙), but may be prone to small systematics in the νmax scaling relation (which was used for the calibration). To make use of the most observational constraints available, we used the set of nine modeling solutions, which used Teff, [Fe/H], frequencies, and the luminosity as input parameters. From this set of solutions, we adopted the self-consistent set of stellar parameters with the mass closest to the median mass over all results. A more detailed study of the individual modeling results will be presented in a follow-up paper (T. Li et al. 2019, in preparation).
For ease of propagating stellar parameters to exoplanet modeling (see the next section), uncertainties were calculated by adding the median uncertainty for a given stellar parameter in quadrature to the standard deviation of the parameter for all methods. This method has been commonly adopted for Kepler (e.g., Chaplin et al. 2014b) and captures both random and systematic errors estimated from the spread among different methods. For completeness, the individual random and systematic error estimates are R⋆ = 2.943 ± 0.041(ran) ± 0.049(sys) R⊙, M⋆ = 1.212 ± 0.052(ran) ± 0.055(sys) M⊙, ρ⋆ = 0.06702 ± 0.00019(ran) ± 0.00047(sys)gcc, and t = 4.9 ± 0.6(ran) ± 0.9(sys) Gyr. This demonstrates that systematic errors constitute a significant fraction of the error budget for all stellar properties (in particular stellar age), and emphasizes the need for using multiple model grids to derive realistic uncertainties for stars and exoplanets. The final estimates of the stellar parameters are summarized in Table 2, constraining the radius, mass, density, and age of HD 221416 to ∼2%, ∼6%, ∼1% and ∼22%, respectively.
Table 2. Host Star Parameters
| Basic Properties | |
|---|---|
| HD ID | 221416 |
| Hipparcos ID | 116158 |
| TIC ID | 441462736 |
| V magnitude | 8.15 |
| TESS magnitude | 7.30 |
| K magnitude | 6.04 |
| SED and Gaia Parallax | |
| Parallax, π (mas) | 10.518 ± 0.080 |
| Luminosity, L (L⊙) | 5.15 ± 0.17 |
| Spectroscopy | |
| Effective temperature, Teff (K) | 5080 ± 90 |
| Metallicity, [Fe/H] (dex) | −0.08 ± 0.08 |
| Projected rotation speed, v sin i (km s−1) | 2.8 ± 1.6 |
| Asteroseismology | |
| Stellar mass, M⋆ (M⊙) | 1.212 ± 0.074 |
| Stellar radius, R⋆ (R⊙) | 2.943 ± 0.064 |
| Stellar density, ρ⋆ (gcc) | 0.06702 ± 0.00067 |
| Surface gravity, log g (cgs) | 3.584 ± 0.010 |
| Age, t (Gyr) | 4.9 ± 1.1 |
Note. The TESS magnitude is adopted from the TESS Input Catalog (Stassun et al. 2018).
Download table as: ASCIITypeset image
4. Planet Characterization
To fit the transits observed in the TESS data, we used the PDC-MAP light curve provided by the TESS Science Processing and Operations Center, which has been optimized to remove instrumental variability and preserve transits (Smith et al. 2012; Stumpe et al. 2014). To optimize computation time, we discarded all data more than 2.5 days before and after each of the two observed transits. We have repeated the fit and data preparation procedure using the TASOC light curve and found consistent results.
A total of 107 radial-velocity measurements from five different instruments (see Section 2.2 and Table 3) were used to constrain the mass of the planet. No spectroscopic observations were taken during transits, and hence the measurements are unaffected by the Rossiter–McLaughlin effect (∼2.3 m s−1 based on the measured v sin i and Rp/R⋆). To remove variations from stellar oscillations, we calculated weighted nightly means for all instruments that obtained multiple observations per night. We performed a joint transit and radial-velocity fit using a Markov Chain Monte Carlo algorithm based on the exoplanet modeling code ktransit (Barclay 2018), as described in Chontos et al. (2019). We placed a strong Gaussian prior on the mean stellar density using the value derived from asteroseismology (Table 2) and weak priors on the linear and quadratic limb-darkening coefficients, derived from the closest I-band grid points in Claret & Bloemen (2011), with a width of 0.6 for both coefficients. We also adopted a prior for the radial-velocity jitter from granulation and oscillations of 2.5 ± 1.5 m s−1, following Yu et al. (2018; see also Tayar et al. 2018), and a 1/e prior on the eccentricity to account for the linear bias introduced by sampling in e cos ω and e sin ω (Eastman et al. 2013). Independent zero-point offsets and stellar jitter values for each of the five instruments that provided radial velocities. Independent joint fits using EXOFASTv2 (Eastman et al. 2013) yielded consistent results.
Table 3. High-precision Radial Velocities for HD 221416
| Time (BJD) | RV (m s−1) | σRV (m s−1) | Instrument |
|---|---|---|---|
| 2458426.334584 | 4.258 | 11.260 | SONG |
| 2458426.503655 | 6.328 | 11.270 | SONG |
| 2458427.575230 | −12.667 | 3.000 | FEROS |
| 2458428.547576 | 17.328 | 18.540 | SONG |
| ⋯ | ⋯ | ⋯ | ⋯ |
| 2458443.535340 | −14.667 | 3.600 | CORALIE |
| 2458443.541210 | −3.067 | 3.800 | CORALIE |
| 2458443.714865 | −6.815 | 0.780 | HIRES |
| 2458443.825283 | −4.375 | 0.720 | HIRES |
| ⋯ | ⋯ | ⋯ | ⋯ |
| 2458482.562290 | 19.433 | 2.000 | HARPS |
| 2458483.541710 | 16.133 | 2.000 | HARPS |
| 2458483.553240 | 19.233 | 2.000 | HARPS |
| 2458483.564690 | 16.233 | 2.000 | HARPS |
Note. Error bars do not include contributions from stellar jitter, and measurements have not been corrected for zero-point offsets.
Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.
Download table as: DataTypeset image
Figures 5 and 6 show the radial-velocity time series, phase-folded transit and RV data, and the corresponding best-fitting model. Table 4 lists the summary statistics for all planet and model parameters. The system is well described by a planet in a 14.3 day orbit, which is nearly equal in size but ∼35% less massive than Saturn (Rp = 0.836 ± 0.031 RJ, Mp = 0.190 ± 0.018 MJ), with tentative evidence for a mild eccentricity (e = 0.11 ± 0.03). The long transit duration (∼0.5 days) is consistent with a nongrazing (b ≈ 0.7) transit given the asteroseismic mean stellar density, providing further confirmation for a gas-giant planet orbiting an evolved star. The radial-velocity data do not show evidence for any other short-period companions. Continued monitoring past the ∼4 orbital periods covered here will further reveal details about the orbital architecture of this system.
Figure 5. Radial-velocity time series (panel a) and residuals after subtracting the best-fitting model (panel b) for HD 221416 b. Data points are corrected for zero-point offsets of individual instruments, and error bars include contributions from stellar jitter.
Download figure:
Standard image High-resolution image
Figure 6. TESS light curve (panel a) and radial-velocity measurements (panel b) folded with the best-fitting orbital period. Gray points in panel (a) show the original sampling, and black points are binned means over 10 minutes. Red lines in both panels show the best-fitting model from the joint fit using stellar parameters, transit, and radial velocities. Gray lines show random draws from the joint MCMC model. Error bars in panel (b) include contributions from stellar jitter.
Download figure:
Standard image High-resolution imageTable 4. Planet Parameters
| Parameter | Best Fit | Median | 84% | 16% |
|---|---|---|---|---|
| Model Parameters | ||||
| γHIRES | 4.8 | 5.4 | +1.6 | −1.6 |
| γSONG | 1.1 | 0.2 | +1.5 | −1.5 |
| γFEROS | −15.4 | −15.7 | +1.2 | −1.2 |
| γCORALIE | −5.4 | −5.0 | +1.2 | −1.2 |
| γHARPS | 8.1 | 8.8 | +1.5 | −1.5 |
| σHIRES | 2.71 | 2.68 | +0.85 | −0.80 |
| σSONG | 2.06 | 2.11 | +0.91 | −0.89 |
| σFEROS | 3.49 | 3.47 | +0.75 | −0.71 |
| σCORALIE | 1.88 | 2.50 | +0.75 | −0.64 |
| σHARPS | 2.41 | 2.69 | +0.75 | −0.63 |
| z (ppm) | 199.4 | 199.1 | +10.6 | −10.7 |
| P (days) | 14.2762 | 14.2767 | +0.0037 | −0.0037 |
| T0 (BTJD) | 1357.0135 | 1357.0149 | +0.0025 | −0.0026 |
| b | 0.744 | 0.728 | +0.040 | −0.049 |
| Rp/R⋆ | 0.02846 | 0.02854 | +0.00084 | −0.00071 |
| e cos ω | −0.054 | −0.028 | +0.063 | −0.061 |
| e sin ω | −0.099 | −0.096 | +0.029 | −0.030 |
| K (m s−1) | 14.6 | 14.1 | +1.2 | −1.2 |
| ρ⋆ (gcc) | 0.06674 | 0.06702 | +0.00052 | −0.00052 |
| u1 | 0.12 | 0.35 | +0.36 | −0.24 |
| u2 | 0.71 | 0.44 | +0.30 | −0.44 |
| Derived Properties | ||||
| e | 0.113 | 0.115 | +0.034 | −0.030 |
| ω | −118.7 | −106.0 | +34.7 | −31.1 |
| a (AU) | 0.1233 | 0.1228 | +0.0025 | −0.0026 |
| a/R⋆ | 9.00 | 8.97 | +0.27 | −0.27 |
| i (o) | 85.67 | 85.75 | +0.36 | −0.35 |
| Rp(R⊕) | 9.16 | 9.17 | +0.34 | −0.31 |
| Rp(RJ) | 0.835 | 0.836 | +0.031 | −0.028 |
| Mp(M⊕) | 63.4 | 60.5 | +5.7 | −5.7 |
| Mp(MJ) | 0.200 | 0.190 | +0.018 | −0.018 |
| ρp(gcc) | 0.455 | 0.431 | +0.064 | −0.060 |
Note. Parameters denote velocity zero points γ, radial-velocity jitter σ, photometric zero-point z, orbital period P, time of transit T0, impact parameter b, star-to-planet radius ratio Rp/R⋆, eccentricity e, argument of periastron ω, radial-velocity semi-amplitude K, mean stellar density ρ⋆, linear and quadratic limb-darkening coefficients u1 and u2, semimajor axis a, orbital inclination i, as well as planet radius (Rp), mass (Mp) and density (ρp).
Download table as: ASCIITypeset image
5. Discussion
HD 221416 b joins an enigmatic but growing class of transiting planets orbiting stars that have significantly evolved off the main sequence. Figure 7 compares the position of HD 221416 within the expected population of solar-like oscillators to be detected with TESS (panel a) and within the known population of exoplanet host stars. Evolutionary states in Figure 7(b) have been assigned using solar-metallicity PARSEC evolutionary tracks (Bressan et al. 2012) as described in Berger et al. (2018).84 HD 221416 sits at the boundary between subgiants and red giants, with the measured Δν value indicating that the star has just started its ascent on the red-giant branch (Mosser et al. 2014). HD 221416 is a typical target for which we expect to detect solar-like oscillations with TESS, predominantly due to the increased oscillation amplitude, which are well known to scale with luminosity (Kjeldsen & Bedding 1995). On the contrary, HD 221416 is rare among known exoplanet hosts: while radial-velocity searches have uncovered a large number of planets orbiting red giants on long orbital periods (e.g., Wittenmyer et al. 2011), fewer than 15 transiting planets are known around red-giant stars (as defined in Figure 7(b)). HD 221416 b is the sixth example of a transiting planet orbiting a late subgiant/early red giant with detected oscillations, following Kepler-91 (Barclay et al. 2013; Lillo-Box et al. 2014a, 2014b), Kepler-56 (Steffen et al. 2012; Huber et al. 2013a), Kepler-432 (Quinn et al. 2015; Ciceri et al.2015), K2-97 (Grunblatt et al. 2016), and K2-132 (Grunblatt et al. 2017; Jones et al. 2018).
Figure 7. Stellar radius vs. effective temperature for the expected TESS Cycle 1 yield of solar-like oscillators (panel a; Schofield et al. 2019) and for all stars with confirmed transiting planets (panel b). The blue dashed line in panel (a) marks the approximate limit below which 2 minute cadence data are required to sample the oscillations. Symbols in panel (b) are color coded according to the evolutionary state of the star using solar-metallicity PARSEC evolutionary tracks. HD 221416 falls on the border between subgiants and red giants, and is highlighted with an orange/red/blue star symbol. HD 221416 is a typical target for which we expect to detect solar-like oscillations with TESS, but occupies a rare parameter space for an exoplanet host.
Download figure:
Standard image High-resolution imageTransiting planets orbiting evolved stars are excellent systems to advance our understanding of the effects of stellar evolution on the structure and evolution of planets (see, e.g., Veras 2016 for a review). For example, such systems provide the possibility of testing the effects of stellar mass, evolution, and binarity on planet occurrence (e.g., Johnson et al. 2010; Schlaufman & Winn 2013; Stephan et al. 2018), which are still poorly understood. Furthermore, the increased irradiance on the planet caused by the evolution of the host star has been proposed as a means to distinguish between proposed mechanisms to explain the inflation of gas-giant planets beyond the limits expected from gravitational contraction and cooling (Hubbard et al. 2002; Lopez & Fortney 2016). Recent discoveries by the K2 mission have indeed yielded evidence that planets orbiting low-luminosity RGB stars are consistent with being inflated by the evolution of the host star (Grunblatt et al. 2016, 2017), favoring scenarios in which the energy from the star is deposited into the deep planetary interior (Bodenheimer et al. 2001).
Based on its radius and orbital period, HD 221416 b would nominally be classified as a warm Saturn, sitting between the well-known population of hot Jupiters and the ubiquitous population of sub-Neptunes uncovered by Kepler (Figure 8(a)). Taking into account the evolutionary state of the host star, however, HD 221416 b falls at the beginning of the "inflation sequence" in the radius–incident flux diagram (Figure 8(b)), with planet radius strongly increasing with stellar incident flux (Kovács et al. 2010; Demory & Seager 2011; Miller & Fortney 2011; Thorngren & Fortney 2018). Because HD 221416 b is currently not anomalously large compared to the observed trend and scatter for similar planets (Figure 8(b)) and low-mass planets are expected to be more susceptible to planet reinflation (Lopez & Fortney 2016), HD 221416 b may be a progenitor of a class of reinflated gas-giant planets orbiting RGB stars.
Figure 8. Planet radius vs. orbital period (panel a) and incident flux (panel b) for confirmed exoplanets. Symbols are color coded according to the evolutionary state of the host star (see Figure 7). HD 221416 b is highlighted in both panels with an orange/red/blue star symbol.
Download figure:
Standard image High-resolution imageIf confirmed, the mild eccentricity of HD 221416 b would be consistent with predictions of a population of planets around evolved stars for which orbital decay occurs faster than tidal circularization (Villaver et al. 2014; Grunblatt et al. 2018). Moreover, combining the asteroseismic age of the system with the possible nonzero eccentricity would allow constraints on the tidal dissipation in the planet, which drives the circularization of the orbit. Using the formalism by Mardling (2011; see also Gizon et al. 2013; Ceillier et al. 2016; Davies et al. 2016), the current constraints would imply a minimum value of the planetary tidal quality factor Qp;min ≈ 3.2 × 104, below which the system would have been already circularized in ∼5 Gyr. Compared to the value measured in Saturn (Q ≈ 1800; Lainey et al. 2017), this would demonstrate the broad diversity of dissipation observed in giant planets. Because tidal dissipation mechanisms vary strongly with internal structure (see, e.g., Guenel et al. 2014; Ogilvie 2014; André et al. 2017), this may also contribute to understanding the internal composition of such planets. We caution, however, that further RV measurements will be needed to confirm a possible nonzero eccentricity for HD 221416 b.
The precise characterization of planets orbiting evolved, oscillating stars also provides valuable insights into the diversity of compositions of planets through their mean densities. HD 221416 b falls in the transition region between Neptune and sub-Saturn-size planets for which radii increase as RP ≈ 
, and Jovian planets for which radius is nearly constant with mass (Weiss et al. 2013; Chen & Kipping 2017; Figure 9). Recent studies of a population of sub-Saturns in the range ∼4–8 R⊕ also found a wide variety of masses, approximately 6–60 M⊕, regardless of size (Petigura et al. 2017b; Van Eylen et al. 2018). Furthermore, masses of sub-Saturns correlate strongly with host star metallicity, suggesting that metal-rich disks form more massive planet cores. HD 221416 b demonstrates that this trend does not appear to extend to planets with sizes >8 R⊕, given its mass of ∼60 M⊕ and a roughly subsolar metallicity host star ([Fe/H] ≈ −0.08 dex). This suggests that Saturn-size planets may follow a relatively narrow range of densities, a possible signature of the transition in the interior structure (such as the increased importance of electron degeneracy pressure; Zapolsky & Salpeter 1969) leading to different mass–radius relations between sub-Saturns and Jupiters. We note that HD 221416 b is one of the most precisely characterized Saturn-size planets to date, with a density uncertainty of ∼15%.
Figure 9. Mass–radius diagram for confirmed planets with densities measured to better than 50%. Symbols are color coded according to the evolutionary state of the host star (see Figure 7). HD 221416 b is highlighted with a orange/red/blue star symbol. Magenta letters show the position of solar system planets.
Download figure:
Standard image High-resolution image
6. Conclusions
We have presented the discovery of HD 221416 b, the first transiting planet orbiting an oscillating host star identified by TESS. Our main conclusions are as follows:
- 1.HD 221416 is a late subgiant/early red giant with a clear presence of mixed modes. Combined spectroscopy and asteroseismic modeling revealed that the star has just started its ascent on the red-giant branch, with R⋆ = 2.943 ± 0.064 R⊙, M⋆ = 1.212 ± 0.074 M⊙, and near-solar age (4.9 ± 1.1 Gyr). HD 221416 is a typical oscillating star expected to be detected with TESS, and it demonstrates the power of asteroseismology even with only 27 days of data.
- 2.The oscillation amplitude of HD 221416 is consistent with ensemble measurements from Kepler. This confirms that the redder bandpass of TESS compared to Kepler only has a small effect on the oscillation amplitude (as expected from scaling relations; Kjeldsen & Bedding 1995; Ballot et al. 2011a), supporting the expected yield of thousands of solar-like oscillators with 2 minute cadence observations in the nominal TESS mission (Schofield et al. 2019). A detailed study of the asteroseismic performance of TESS will have to await ensemble measurements of noise levels and amplitudes.
- 3.HD 221416 b is a "hot Saturn" (F = 343 ± 24 F⊕, Rp = 0.836 ± 0.031 RJ, Mp = 0.190 ± 0.018 MJ) and joins a small but growing population of close-in, transiting planets orbiting evolved stars. Based on its incident flux, radius, and mass, HD 221416 b may be a precursor to the population of gas giants that undergo radius reinflation, due to the increased irradiance as their host star evolves up the red-giant branch.
- 4.HD 221416 b is one the most precisely characterized Saturn-size planets to date, with a density measured to ∼15%. HD 221416 b does not follow the trend of increasing planet mass with host star metallicity discovered in sub-Saturns with sizes between 4 and 8 R⊕, which has been linked to metal-rich disks preferentially forming more massive planet cores (Petigura et al. 2017b). The moderate density (ρp = 0.431 ± 0.062 g cm−3) suggests that Saturn-size planets may follow a relatively narrow range of densities, a possible signature of the transition in the interior structure leading to different mass–radius relations for sub-Saturns and Jupiters.
HD 221416 provides a first glimpse at the strong potential of TESS to characterize exoplanets using asteroseismology. HD 221416 b has one the most precisely characterized densities of known Saturn-size planets to date, with an uncertainty of ∼15%. Thanks to asteroseismology, the planet density uncertainty is dominated by measurements of the transit depth and the radial-velocity amplitude, and thus can be expected to further decrease with continued transit observations and radial-velocity follow-up, which is readily performed given the brightness (V = 8) of the star. Ensemble studies of such precisely characterized planets orbiting oscillating subgiants can be expected to yield significant new insights into the effects of stellar evolution on exoplanets, complementing current intensive efforts to characterize planets orbiting dwarfs.
The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawai'ian community. We are most fortunate to have the opportunity to conduct observations from this mountain. We thank Andrei Tokovinin for helpful information on the Speckle observations obtained with SOAR. D.H. acknowledges support by the National Aeronautics and Space Administration through the TESS Guest Investigator Program (80NSSC18K1585) and by the National Science Foundation (AST-1717000). A.C. acknowledges support by the National Science Foundation under the Graduate Research Fellowship Program. W.J.C., W.H.B., A.M., O.J.H., and G.R.D. acknowledge support from the Science and Technology Facilities Council and UK Space Agency. H.K. and F.G. acknowledge support from the European Social Fund via the Lithuanian Science Council grant No. 09.3.3-LMT-K-712-01-0103. Funding for the Stellar Astrophysics Centre is provided by The Danish National Research Foundation (grant DNRF106). A.J. acknowledges support from FONDECYT project 1171208, CONICYT project BASAL AFB-170002, and by the Ministry for the Economy, Development, and Tourism's Programa Iniciativa Científica Milenio through grant IC 120009, awarded to the Millennium Institute of Astrophysics (MAS). R.B. acknowledges support from FONDECYT Post-doctoral Fellowship Project 3180246, and from the Millennium Institute of Astrophysics (MAS). A.M.S. is supported by grants ESP2017-82674-R (MINECO) and SGR2017-1131 (AGAUR). R.A.G. and L.B. acknowledge the support of the PLATO grant from the CNES. The research leading to the presented results has received funding from the European Research Council under the European Community's Seventh Framework Programme (FP72007-2013)ERC grant agreement No. 338251 (StellarAges). S.M. acknowledges support from the European Research Council through the SPIRE grant 647383. This work was also supported by FCT (Portugal) through national funds and by FEDER through COMPETE2020 by these grants: UID/FIS/04434/2013 and POCI-01-0145-FEDER-007672, PTDC/FIS-AST/30389/2017, and POCI-01-0145-FEDER-030389. T.L.C. acknowledges support from the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 792848 (PULSATION). E.C. is funded by the European Union's Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement No. 664931. V.S.A. acknowledges support from the Independent Research Fund Denmark (Research grant 7027-00096B). D.S. acknowledges support from the Australian Research Council. S.B. acknowledges NASA grant NNX16AI09G and NSF grant AST-1514676. T.R.W. acknowledges support from the Australian Research Council through grant DP150100250. A.M. acknowledges support from the ERC Consolidator Grant funding scheme (project ASTEROCHRONOMETRY, G.A. n. 772293). S.M. acknowledges support from the Ramon y Cajal fellowship number RYC-2015-17697. M.S.L. is supported by the Carlsberg Foundation (grant agreement No. CF17-0760). A.M. and P.R. acknowledge support from the HBCSE-NIUS programme. J.K.T. and J.T. acknowledge that support for this work was provided by NASA through Hubble Fellowship grants HST-HF2-51399.001 and HST-HF2-51424.001 awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS5-26555. T.S.R. acknowledges financial support from Premiale 2015 MITiC (PI B. Garilli). This project has been supported by the NKFIH K-115709 grant and the Lendület Program of the Hungarian Academy of Sciences, project No. LP2018-7/2018.
Based on observations made with the Hertzsprung SONG telescope operated on the Spanish Observatorio del Teide on the island of Tenerife by the Aarhus and Copenhagen Universities and by the Instituto de Astrofísica de Canarias. Funding for the TESS mission is provided by NASA's Science Mission directorate. We acknowledge the use of public TESS Alert data from pipelines at the TESS Science Office and at the TESS Science Processing Operations Center. This research has made use of the Exoplanet Follow-up Observation Program website, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program. This paper includes data collected by the TESS mission, which are publicly available from the Mikulski Archive for Space Telescopes (MAST).
Software: Astropy (Astropy Collaboration et al. 2018), Matplotlib (Hunter 2007), DIAMONDS (Corsaro & De Ridder 2014), isoclassify (Huber et al. 2017), EXOFASTv2 (Eastman 2017), ktransit (Barclay 2018).
Footnotes
- 78
- 79
- 80
- 81
- 82
- 83
Échelle diagrams are constructed by dividing a power spectrum into equal segments with length Δν and stacking one above the other, so that modes with a given spherical degree align vertically in ridges (Grec et al. 1983). Departures from regularity arise from sound speed discontinuities and from mixed modes, and thus probe the interior structure of a star.
- 84
See also https://github.com/danxhuber/evolstate.