HD 202772A b: A Transiting Hot Jupiter around a Bright, Mildly Evolved Star in a Visual Binary Discovered by TESS

HD 202772 (TIC 290131778, TOI 123) was observed by Camera 1 of the TESS spacecraft during the first sector of science operations, between 2018 July 25 and 2018 August 22 (BJD 2458325 to 2458353). The available data have two-minute time sampling ("short cadence"). Some basic parameters of the target are given in Table 1. Given its position in the sky, HD 202772 will not be re-observed during the TESS primary mission. The photometric data were analyzed by the Science Processing Operations Center (SPOC) pipeline, based on the NASA Kepler mission pipeline (J. Jenkins et al. 2018, in preparation). The light curve of HD 202772 presented in Figure 1 shows a clear transit signal. It was listed among the TESS Alerts published online on 2018 September 5, prompting us to download the photometric time series.³⁷

Zoom In Zoom Out Reset image size

Table 1.  HD 202772

Parameter	HD 202772A	HD 202772B	Source
R.A. (hh:mm:ss)	21:18:47.901	21:18:47.813	Gaia DR2
Decl. (dd:mm:ss)	−26:36:58.95	−26:36:58.42	Gaia DR2
μα (mas yr−1)	28.360 ± 0.269	23.236 ± 0.157	Gaia DR2
μδ (mas yr−1)	−56.533 ± 0.418	−57.557 ± 0.152	Gaia DR2
Parallax (mas)	6.166 ± 0.092	6.686 ± 0.109	Gaia DR2
B (mag)	8.81 ± 0.02	10.65 ± 0.02	Tycho
V (mag)	8.320 ± 0.05	10.15 ± 0.05	Tycho
TESS (mag)	7.92 ± 0.09	9.62 ± 0.09	TIC V7a
J (mag)	7.437 ± 0.027	9.142 ± 0.029	NIRC2; this paper
H (mag)	7.266 ± 0.021	8.897 ± 0.022	NIRC2; this paper
Ks (mag)	7.149 ± 0.017	8.858 ± 0.018	NIRC2; this paper
Spectroscopic and Derived Properties
${T}_{\mathrm{eff}}$ (K)	6330 ± 100	6156 ± 100	Keck/HIRES; this paper
$\mathrm{log}{g}_{* }$ (cgs)	4.03 ± 0.10	4.24 ± 0.10	Keck/HIRES; this paper
$\left[\mathrm{Fe}/{\rm{H}}\right]$ (dex)	0.29 ± 0.06	0.25 ± 0.06	Keck/HIRES; this paper
M* (${M}_{\odot }$)	${1.69}_{-0.04}^{+0.05}$	1.21 ± 0.04	Keck/HIRES; this paper
R* (${R}_{\odot }$)	${2.515}_{-0.127}^{+0.137}$	1.16 ± 0.06	Keck/HIRES; this paper
Age (age)	${1.52}_{-0.20}^{+0.19}$	${1.27}_{-0.80}^{+1.32}$	Keck/HIRES; this paper

Note.

^aStassun et al. (2018).

Download table as:  ASCIITypeset image

We detrended the raw light curves in the following way (see, e.g., Günther et al. 2017, 2018). After masking out all of the data obtained during transits, we fitted a Gaussian Process (GP) model to the data, using a Matern 3/2 kernel and a white noise kernel. For this task, we employed the celerite package, which uses a Taylor-series expansion of these kernel functions. Once the parameters of the GP were constrained based on the out-of-transit data, we used it to detrend the entire light curve.

The SPOC pipeline produces flags for poor-quality exposures. These include exposures taken during the 30-minute momentum dumps that occurred every 2.5 days (10 times in total). All flagged exposures were omitted from our analysis. The resulting light curve is plotted in Figure 1.

HD 202772 was reported to be a pair of stars separated by ≈1.5'' in several wide-field surveys (e.g., Tycho-2, Høg et al. 2000; PPMXL, Roeser et al. 2010;Gaia DR2, Gaia Collaboration et al. 2018; see also Holden 1978; Horch et al. 2001), although the Gaia DR2 catalog flags the brighter star as a "duplicate" entry. To check on these earlier findings, we performed high-resolution AO imaging at Keck Observatory.

The Keck observations were made with the NIRC2 instrument on Keck II behind the natural guide star AO system. The observations were made on 2018 September 18 on a night with partial cirrus conditions. We used the standard 3-point dither pattern that avoids the left lower quadrant of the detector (which is typically noisier than the other three quadrants). The dither pattern step size was 3'', and it was repeated twice, with the second dither offset from the first dither by 05. Observations were made with three different filters: narrow-band Br_γ (λ_o = 2.1686; Δλ = 0.0326 μm), H-continuum (λ_o = 1.5804; Δλ = 0.0232 μm), and J-continuum (λ_o = 1.2132; Δλ = 0.0198 μm), using integration times of 1.45, 5.0, and 1.5 s, respectively. The camera was in the narrow-angle mode with a full field of view of 10'' and a pixel scale of approximately 001 per pixel.

Two stars were clearly detected, with a separation of 13 (Figure 2). The resolution of the 2 μm image is approximately 005 FWHM. The sensitivity of the final combined AO image was determined by injecting simulated sources azimuthally around the primary target every 45° at separations of integer multiples of the FWHM of the central source (Furlan et al. 2017). The brightness of each injected source was scaled until standard aperture photometry detected it with 5σ significance. The resulting brightness of the injected sources relative to our target was taken to be the contrast limit for the injected location. The final 5σ limit at each separation was determined from the average of all of the determined limits at that separation, with an uncertainty given by the rms dispersion of the results for different azimuthal slices. Figure 2 shows the 2 μm sensitivity curve in black, with the 1σ(rms) dispersion marked in purple. The inset image shows the primary target in the center and the second source located 13 to the northwest.

Zoom In Zoom Out Reset image size

The two stars were detected in all three filters. The presence of the blended companion must be taken into account to obtain the correct transit depth and planetary radius (Ciardi et al. 2015). The stars have blended 2MASS magnitudes of J = 7.232 ± 0.026 mag, H = 7.048 ± 0.021 mag, and K_s = 6.945 ± 0.026 mag. The stars have measured magnitude differences of ΔJ = 1.705 ± 0.0015 mag, ΔH = 1.631 ± 0.008 mag, and ΔK_s = 1.709 ± 0.011 mag. The primary star has deblended apparent magnitudes of J₁ = 7.437 ± 0.027 mag, H₁ = 7.266 ± 0.021 mag, and Ks₁ = 7.149 ± 0.017 mag, corresponding to (J − H)₁ =0.171 ± 0.034 mag and ${(H-{K}_{s})}_{1}\,=\,0.117\pm 0.027$ mag. The secondary star has deblended apparent magnitudes of J₂ = 9.142 ± 0.029 mag, H₂ = 8.897 ± 0.022 mag, and Ks₂ = 8.858 ± 0.018 mag, corresponding to ${(J-H)}_{2}=0.245\,\pm 0.036$ mag and ${(H-{K}_{s})}_{2}=0.039\pm 0.029$ mag. The infrared colors of the primary star are consistent with an early-G or late-F main-sequence star, in agreement with the derived stellar parameters. The companion star has infrared colors that are consistent with a later G-type main-sequence star.

Based on the TESS magnitude and 2MASS color relationships established for the TESS Input Catalog (Stassun et al. 2018), we estimate the deblended TESS magnitudes for the two components to be T₁ = 7.92 ± 0.09 mag and T₂ = 9.62 ± 0.09 mag for a TESS magnitude difference of Δ_T = 1.70 ± 0.13 mag and a TESS flux ratio of ${F}_{2}/{F}_{1}=0.21\pm 0.02$. We used this value of the flux ratio to correct the apparent transit depth in the TESS light curve and derive the unblended transit depth.

We will refer to the brighter target (hosting the planet) as HD 202772A and the fainter companion as HD 202772B. Given the similarity between the two stars, and a projected separation of only ∼200 au, it seems very likely that the two stars are gravitationally bound. The chance alignment probability is negligible, as estimated from the Besanon Galactic Model (Robin et al. 2003). We have also checked all 131 stars in the Gaia DR2 catalog within 300'' of HD 202772A with measured parallax and proper motion, and the only nearby source with a similar projected velocity is HD 202772B.

To obtain independent estimates of the stellar parameters, we performed high-resolution optical spectroscopy with the LCO robotic network of telescopes (Brown et al. 2013). We obtained three 20-minute exposures with a total signal-to-noise ratio (S/N) ≈ 100 with the NRES (Siverd et al. 2016, 2018) mounted on a 1.0 m telescope at the South African Astronomical Observatory.

Since the NRES fiber diameter corresponds to 28, it captured the light from both stars in the visual binary system. Using TODCOR (Zucker & Mazeh 1994), we identified two RV components separated by 5.8 ± 1.0 km s⁻¹. This RV difference is compatible with the order of magnitude of the RV variation one would expect from the orbital motion of the stars, given their masses and sky-projected separation. The stellar parameters derived from collected spectra are listed in Table 2.

In order to obtain a spectrum of each of the two stars with minimal contamination from the other star, we observed both stars with Keck/HIRES (Vogt et al. 1994) on 2018 September 23. We obtained one spectrum of each star, with the HIRES slit oriented perpendicular to the separation between the two stars. Given the angular separation between the two stars, the slit width (086), and the astronomical seeing at Keck at the night of the observation (≈07), the level of cross-contamination is expected to be less than 10%. Both spectra were obtained without the iodine (I₂) cell, at a spectroscopic resolution of R ≈ 65000, and at an S/N per pixel of 150 at 5500 Å. A similar technique was successfully applied by Shporer et al. (2014) to a visual binary system in which both members have similar brightness and a smaller angular separation.

We obtained a total of 14 spectra of HD 202772A using CHIRON (Tokovinin et al. 2013), a fiber-fed high-resolution optical spectrograph mounted on the SMARTS 1.5 m telescope at Cerro Tololo in Chile. We collected the spectra using the image slicer, which delivers a resolution of ∼80000 and a higher throughput than the standard slit mode. Our 15 minutes exposures yielded an S/N per pixel of ∼60–80 at 5500 Å.

Table 2.  HD 202772A

Parameter	SMARTS 1.5 m/CHIRON	FLWO 1.5 m/TRES	LCO/NRES	KECK/HIRES	EXOFASTv2 FIT
${T}_{\mathrm{eff}}$ [K]	6470 ± 100	6270 ± 50	6255 ± 100	6330 ± 100	${6272}_{-71}^{+77}$
$\mathrm{log}{g}_{* }$ [cgs]	3.90 ± 0.15	3.91 ± 0.10	4.0 ± 0.1	4.03 ± 0.10	${3.848}_{-0.027}^{+0.030}$
$\left[\mathrm{Fe}/{\rm{H}}\right]$ [dex]	0.30 ± 0.10	0.16 ± 0.08	0.27 ± 0.06	0.29 ± 0.06	0.300 ± 0.060
M* $[{M}_{\odot }]$	1.73 ± 0.05	...	${1.78}_{-0.06}^{+0.02}$	${1.69}_{-0.04}^{+0.05}$	${1.720}_{-0.066}^{+0.064}$
R* $[{R}_{\odot }]$	2.65 ± 0.15	...	${2.87}_{-0.10}^{+0.11}$	${2.515}_{-0.127}^{+0.137}$	${2.594}_{-0.090}^{+0.076}$
Age [Gyr]	1.6 ± 0.1	...	${1.48}_{-0.16}^{+0.24}$	${1.52}_{-0.20}^{+0.19}$	${1.70}_{-0.26}^{+0.32}$
V sin i [km s−1]	...	8.1 ± 0.5	5.5 ± 1.4	7.0 ± 1.0	...

Note.¹The spectroscopic priors on ${T}_{\mathrm{eff}}$ and $\left[\mathrm{Fe}/{\rm{H}}\right]$ from Keck/HIRES are imposed in the EXOFASTv2 analysis.

Download table as:  ASCIITypeset image

Although CHIRON is equipped with an iodine cell to obtain a precise wavelength solution that permits long-term RV precision better than ∼2–3 m s⁻¹ (e.g., Jones et al. 2017), we did not use the iodine cell for these observations. This is because the cell absorbs ∼25% of the light at 5500 Å, significantly decreasing the S/N. Moreover, using the I₂ cell requires a time-consuming acquisition of a high-S/N template spectrum of the target star.

Instead, we derived the RVs using the Cross-Correlation-Function (CCF) method, in a manner similar to Jones et al. (2017). CHIRON is not equipped with a simultaneous calibration fiber. Instead, we acquired a Th-Ar lamp exposure before and after each target exposure. We computed the wavelength solution for the target spectra by interpolating line positions for the lamps to match the temporal midpoint of each observation. We thereby achieved an RV stability of ∼5–6 m s⁻¹, which was verified with two RV standard stars observed nightly. The resulting RVs of HD 202772A are listed in Table 3. RVs collected by CHIRON show a $\sim 95\,{\rm{m}}\,{{\rm{s}}}^{-1}$ sinusoidal variation in phase with the transit ephemeris.

Table 3.  Relative Radial Velocities for HD 202772A

$\mathrm{BJD}$	RV	${\sigma }_{\mathrm{RV}}$	BVS1	${\sigma }_{\mathrm{BVS}}$	FWHMa	${\sigma }_{\mathrm{FWHM}}$	Instrument
-2458300	(m s−1)	(m s−1)	(m s−1)	(m s−1)	(m s−1)	(m s−1)
69.5363	−87.5	7.5	15.4	20.1	16302.0	141.5	CHIRON
69.7550	−52.0	6.2	15.4	11.3	16258.7	128.1	CHIRON
70.5656	68.7	4.9	32.6	23.2	16279.1	130.0	CHIRON
71.5411	29.5	8.9	17.1	19.0	16133.5	132.2	CHIRON
71.6964	11.2	6.3	39.4	16.5	16171.8	130.5	CHIRON
72.5988	−76.2	5.4	25.7	18.1	16274.6	132.2	CHIRON
73.6179	50.1	5.5	22.3	15.0	16225.0	133.9	CHIRON
79.6314	−49.4	4.8	−15.4	15.5	16292.1	131.5	CHIRON
79.7038	−54.5	5.7	34.3	20.5	16320.8	133.3	CHIRON
79.7466	−49.3	9.8	15.4	19.0	16243.6	132.4	CHIRON
80.5925	80.0	6.7	−8.6	21.7	16264.7	140.5	CHIRON
80.6924	72.3	11.1	−30.9	25.3	16162.8	154.1	CHIRON
81.5400	−24.9	13.6	20.6	35.9	16166.2	136.7	CHIRON
83.6838	81.7	7.7	60.0	17.5	16229.4	127.2	CHIRON
83.5215	34.8	2.0	...	...	...	...	HARPS
83.5252	36.3	2.0	...	...	...	...	HARPS
84.5863	80.3	2.0	...	...	...	...	HARPS
84.5837	97.8	2.4	...	...	...	...	HARPS
85.5194	−90.3	2.8	...	...	...	...	HARPS
85.5128	−96.0	2.8	...	...	...	...	HARPS
85.7274	−62.7	2.0	...	...	...	...	HARPS
75.7239	−0.5	12.0	−4.6	12.8	18858	151	TRES
77.6977	174.7	19.8	−10.4	19.4	18997	231	TRES
79.7226	56.6	14.9	−25.4	8.3	18861	149	TRES
82.7170	0.0	12.0	−11.4	8.3	18869	144	TRES
83.7441	231.0	16.1	5.7	19.6	18673	222	TRES
84.7374	154.6	20.2	−20.6	23.7	18843	137	TRES
86.7178	124.7	17.0	5.5	11.6	18831	112	TRES
87.6938	163.5	21.3	38.3	14.7	18944	157	TRES
88.6862	20.2	17.4	22.0	24.4	18909	218	TRES
89.7004	72.1	16.2	−9.3	12.6	18796	109	TRES
90.7214	203.7	18.7	−4.1	8.3	18814	153	TRES
91.6617	30.3	20.1	14.4	11.7	18966	132	TRES

Note.

^aThe HARPS BVS and FWHM are not listed because those measurements were corrupted by moonlight (Section 2.7).

Download table as:  ASCIITypeset image

Finally, from the CCF, we measured the bisector velocity span (BVS) and FWHM variations, to check on the possibility that the observed RV variation results from stellar activity or a background eclipsing binary system (see, e.g., Santerne et al. 2015).

The CHIRON fiber has a 27 diameter on the sky, but HD 202772A and B are separated by only 13, which means that we must expect some of the light from the binary companion to be present in the spectra. Given that the two stars have a similar radial velocity, there is a risk that the stationary CCF of the binary companion causes the apparent amplitude of the RV variation to be lower than the true RV variation of the planet host. Such a "peak pulling" effect was observed in a study of the Kepler-14 system (Buchhave et al. 2011). We note, however, that HD 202772B is fainter than HD 202772A and only emits about 20% of the total light from the binary system. As described in Section 2.7, we did not find any evidence that the RVs from CHIRON were significantly affected by light contamination from HD 202772B.

We obtained 12 spectra of HD 202772A with the TRES (Fűrész et al. 2008) on the 1.5 m Tillinghast Reflector at Fred L. Whipple Observatory (FLWO) on Mt. Hopkins, AZ between UT 2018 September 14 and UT 2018 September 30. TRES is a fiber-fed, cross-dispersed echelle spectrograph with a resolving power of R ∼ 44000 and an instrumental precision of ∼10–15 m s⁻¹. The typical exposure time was ∼4 minutes, resulting in S/N per resolution element of ∼75 at 5200 Å. The spectra are calibrated using a Th-Ar lamp, exposed through the science fiber before and after each set of science exposures. We note that the TRES fiber is 23, and the exposures therefore include light from HD 202772B.

We reduced and analyzed the spectra according to the procedures outlined in Buchhave et al. (2010). Namely, the spectra were optimally extracted and then cross-correlated, order by order, against the strongest spectrum of HD 202772A. We exclude spectral orders far to the blue where the S/N is low, in the red where telluric lines contaminate the spectrum, and a few orders in between with little information content or affected by broad feature (e.g., Balmer lines) toward the edge of the order that affect continuum fitting. RVs were ultimately derived from a region spanning 4130–6280 Å. The peak of the summed CCF across all orders is fit to derive the final RV, and the scatter between orders within each spectrum is taken to be the internal error estimate. These relative RVs and their uncertainties are reported in Table 3.

We also derive the BVS and FWHM from the cross-correlation function of each spectrum against a non-rotating synthetic spectrum with appropriate ${T}_{\mathrm{eff}}$, $\mathrm{log}{g}_{* }$, and $\left[\mathrm{Fe}/{\rm{H}}\right]$. These values are also reported in Table 3 and shown in Figure 6, and show no correlation with the RVs.

To provide further confirmation of the planetary origin of the transit signal, we obtained nine spectra using the HARPS (Mayor et al. 2003).

These data were obtained simultaneously with Th-Ar reference spectra during three consecutive nights in good seeing conditions (≲10). Exposure time was 200–300 seconds, leading to an S/N of ∼70–80 at 6000 Å. We carefully centered the brighter star within the 1'' aperture fiber, to ensure that no light contamination from the companion was reaching the detector. We also carefully adjusted the size of the guiding box, to avoid guiding problems due to the secondary star, which was clearly visible in the acquisition camera. During the HARPS observations, the Moon was between 13° and 36° from our target, with an illuminated fraction between 92% and 99%. This led to some lunar contamination in the spectra. Moreover, the RV of the Moon was very close to the RV of the target star, which severely affected the shape of the CCF. For this reason, we discarded the two RV data points that were most affected and had very deviant values. Also, due to the contamination, the derived BVS and FWHM of the CCF are not reliable, and are not listed in Table 3. We processed the HARPS data using the CERES code (Brahm et al. 2017). The derived RVs from the CERES code also agree well with those from the SERVAL pipeline (Zechmeister et al. 2018). The resulting RVs are listed in Table 3 and also shown in Figure 4. As can be seen, the HARPS data agree with the CHIRON data, although the scatter around the fit is larger than expected. This is most likely caused by the lunar light contamination.

To derive the stellar atmospheric parameters of HD 202772A, we measured the equivalent widths (EWs) of about 150 relatively weak Fe i and Fe ii absorption lines (EW ≲ 120 mÅ ). The EWs were measured in the high-S/N template obtained by stacking the individual spectra (see Section 2.5), using the ARES v2 automatic tool (Sousa et al. 2015).

We then used the MOOG code (Sneden 1973) along with the Kurucz (1993) stellar atmosphere models to solve the radiative transfer equations under the assumptions of local excitation and ionization equilibrium via the Saha and Boltzmann equations. For each iron line, MOOG computes the corresponding iron abundance by matching the measured EW in the curve of growth computed from the input stellar model. This procedure is performed iteratively for models with different effective temperatures (${T}_{\mathrm{eff}}$), iron abundances ($\left[\mathrm{Fe}/{\rm{H}}\right]$), and microturbulent velocities (V_micro) until there is no correlation between the line excitation potential and wavelength with the model abundance. Finally, we obtained the surface gravity ($\mathrm{log}{g}_{* }$) using the constraint that the iron abundances derived from both the Fe i and Fe ii lines should be the same (for a more thorough description of the procedure, see Jones et al. 2011). Table 2 gives the resulting stellar parameters.

We computed the luminosity of HD 202772A based on the Gaia DR2 parallax (π = 6.166 ± 0.092, Gaia Collaboration et al. 2018), the apparent V magnitude (after correcting for interstellar absorption by A_V = 0.10 mag), and the bolometric correction of Alonso et al. (1999). Using this information and the stellar atmospheric parameters (${T}_{\mathrm{eff}}$ and $\left[\mathrm{Fe}/{\rm{H}}\right]$), we derived the stellar physical parameters using the PARSEC stellar-evolutionary models (Bressan et al. 2012). The results are also listed in Table 2.

We used the Spectral Parameter Classification (SPC) tool (Buchhave et al. 2012) to derive stellar parameters from the TRES spectra. We allowed ${T}_{\mathrm{eff}}$, $\mathrm{log}{g}_{* }$, $\left[\mathrm{Fe}/{\rm{H}}\right]$, and $V\sin i$ to be free parameters. SPC works by cross correlating an observed spectrum against a grid of synthetic spectra based on Kurucz atmospheric models (Kurucz 1993). The weighted average results are listed in Table 2.

We analyzed the LCO/NRES spectrum using the methodology of Fulton & Petigura (2018). We measured ${T}_{\mathrm{eff}}$, $\mathrm{log}{g}_{* }$, $\left[\mathrm{Fe}/{\rm{H}}\right]$, and $V\sin i$ using SpecMatch (Petigura 2015),³⁸ which compares the observed spectrum with a grid of model spectra (Coelho et al. 2005). The resulting parameters are listed in Table 2.

To calculate the star's physical parameters, we used isoclassify (Huber et al. 2017), which takes as input the effective temperature, metallicity, parallax, and apparent K_s magnitude. Using the isoclassify "direct" mode, we calculated the posterior probability distributions for R_⋆ and L_⋆ by applying the StefanBoltzmann law. Using the isoclassify "grid" mode, we calculated the range of MIST isochrone models (Choi et al. 2016; Dotter 2016) that are consistent with the spectroscopic parameters to estimate the stellar mass and age. The results of the SpecMatch+isoclassify analysis are listed in Table 2.

To compute stellar parameters from Keck/HIRES spectrum for each of the two stars, we used the same technique (SpecMatch+isoclassify) that was applied on NRES spectra.

The resulting atmospheric and physical stellar parameters obtained from SpecMatch+isoclassify analysis for both HD 202772A and B are listed in Tables 1 and 2.

The stellar parameters derived from resolved HIRES spectra show good agreement with those from the CHIRON, TRES, and NRES spectra, except that TRES finds modestly lower $\left[\mathrm{Fe}/{\rm{H}}\right]$. Evidently, the contaminating light from the secondary star in the CHIRON, TRES, and NRES spectra did not strongly affect the determination of the basic stellar parameters.