Abstract
We report the discovery of two transiting Neptunes by the HATSouth survey. The planet HATS-37Ab has a mass of (31.5 ± 13.4 M⊕) and a radius of , and is on a day orbit around a mag, star with a radius of . We also present evidence that the star HATS-37A has an unresolved stellar companion HATS-37B, with a photometrically estimated mass of . The planet HATS-38b has a mass of (23.5 ± 3.5 M⊕) and a radius of , and is on a day orbit around a mag, star with a radius of . Both systems appear to be old, with isochrone-based ages of Gyr, and Gyr, respectively. Both HATS-37Ab and HATS-38b lie in the Neptune desert and are thus examples of a population with a low occurrence rate. They are also among the lowest-mass planets found from ground-based wide-field surveys to date.
Export citation and abstract BibTeX RIS
1. Introduction
Over the past two decades the population of known transiting exoplanets has grown at an accelerating pace, with the Kepler satellite (Borucki et al. 2010) dominating the overall number of discoveries. The distribution of the discoveries is far from homogeneous in terms of the planetary parameters, both due to observational biases and variations in the intrinsic occurrence of planets as a function of their physical parameters and those of their host stars. An example of an observational bias is the paucity of known transiting exoplanets with periods P ≳ 10 days, a region of parameter space that the ground-based survey HATSouth (Bakos et al. 2013) was designed to target, and that is currently being explored efficiently by the Transiting Exoplanet Survey Satellite mission (TESS, Ricker et al. 2015). An example of intrinsically low occurrence rates is the so-called Neptune desert, a term coined by Mazeh et al. (2016) to describe a wedge in the period-mass or period–radius diagram where close-in (P ≲ 5 days) planets with radii similar to Neptune are very rare, and essentially nonexistent for P ≲ 3 days (see also Szabó & Kiss 2011; Beaugé & Nesvorný 2013).
In order to uncover more planetary systems in sparsely populated regions such as the Neptune desert it pays to survey to fainter magnitudes than what TESS is optimized for. Ground-based wide-field surveys that are currently in operation, such as HATSouth or NGTS (Wheatley et al. 2018), can complement TESS by uncovering an additional number of intrinsically rare systems. Indeed, one of the most extreme systems in the Neptune desert was recently uncovered by the NGTS (NGTS4-b, West et al. 2019). The reason for the existence of the desert is under investigation. The physical processes thought to be relevant are photoevaporation and the tidal disruption barrier for gas giants after high-eccentricity migration (see Owen & Lai 2018, and references therein).
In this paper we report the discovery by the HATSouth survey of two transiting Neptunes in the desert. They both have similar radii and period values, and fairly similar masses. We thus contribute two more systems to the sparsely populated Neptune desert. The paper is structured as follows. In Section 2 we describe the observational data that were used to perform the modeling of the system as described in Section 3. The results are discussed in Section 4.
2. Observations
Figures 1 and 2 show the observations collected for HATS-37 and HATS-38, respectively. Each figure shows the HATSouth light curve used to detect the transits, the ground-based follow-up transit light curves, the high-precision radial velocities (RVs) and spectral line bisector spans (BSs), and the catalog broadband photometry, including parallax corrections from Gaia DR2, used for characterizing the host stars. Below we describe the observations of these objects that were collected by our team.
2.1. Photometric Detection
Both of the systems presented here were initially detected as transiting planet candidates based on observations by the HATSouth network. The operations of the network are described in Bakos et al. (2013), while our methods for reducing the data to trend-filtered light curves (filtered using the method of Kovács et al. 2005) and identifying transiting planet signals (using the Box-fitting Least Squares or BLS method; Kovács et al. 2002) are described in Penev et al. (2013). The HATSouth observations of each system are summarized in Table 1, while the light curve data are made available in Table 3.
Table 1. Summary of Photometric Observations
Instrument/Fielda | Date(s) | # Images | Cadenceb | Filter | Precisionc |
---|---|---|---|---|---|
(s) | (mmag) | ||||
HATS-37 | |||||
HS-1/G567.1 | 2011 Mar–2011 Aug | 4975 | 294 | r | 5.2 |
HS-3/G567.1 | 2011 Jul–2011 Aug | 735 | 297 | r | 5.7 |
HS-5/G567.1 | 2011 Mar–2011 Aug | 3217 | 291 | r | 5.0 |
PEST 0.3 m | 2016 Feb 16 | 113 | 132 | RC | 2.8 |
Swope 1 m/e2v | 2017 Apr 4 | 161 | 104 | i | 1.6 |
LCO 1 m/sinistro | 2016 Apr 16 | 108 | 159 | i' | 1.0 |
LCO 1 m/sinistro | 2018 Mar 19 | 82 | 163 | i' | 0.8 |
CHAT 0.7 m | 2018 Apr 5 | 217 | 113 | i | 1.4 |
HATS-38 | |||||
HS-1/G561.1 | 2014 Dec–2015 Jul | 4892 | 319 | r | 6.7 |
HS-2/G561.1 | 2014 Dec–2015 Jul | 5718 | 349 | r | 4.7 |
HS-3/G561.1 | 2014 Dec–2015 Jul | 3691 | 353 | r | 5.1 |
HS-4/G561.1 | 2014 Dec–2015 Jul | 2862 | 352 | r | 6.9 |
HS-5/G561.1 | 2014 Dec–2015 Jul | 2959 | 356 | r | 5.7 |
HS-6/G561.1 | 2014 Dec–2015 Jul | 3058 | 342 | r | 6.9 |
HS-1/G561.1.focus | 2014 Dec–2015 Jul | 2026 | 1122 | r | 14.5 |
HS-2/G561.1.focus | 2014 Dec–2015 Jul | 2134 | 1204 | r | 13.4 |
HS-3/G561.1.focus | 2014 Dec–2015 Jul | 1217 | 1227 | r | 14.1 |
HS-4/G561.1.focus | 2014 Dec–2015 Jul | 977 | 1221 | r | 15.5 |
HS-5/G561.1.focus | 2014 Dec–2015 Jul | 1190 | 1232 | r | 15.0 |
HS-6/G561.1.focus | 2014 Dec–2015 Jul | 1174 | 1206 | r | 15.7 |
CHAT 0.7 m | 2017 Feb 5 | 146 | 112 | r | 1.1 |
LCO 1 m/sinistro | 2017 Mar 30 | 83 | 161 | i' | 0.9 |
LCO 1 m/sinistro | 2017 Apr 3 | 118 | 160 | i' | 1.0 |
Notes.
aFor HATSouth data we list the HATSouth unit, CCD, and field name from which the observations are taken. HS-1 and -2 are located at Las Campanas Observatory in Chile, HS-3 and -4 are located at the High Energy Spectroscopic Survey site in Namibia, and HS-5 and -6 are located at Siding Spring Observatory in Australia. Each unit has four CCDs. Each field corresponds to 1 of 838 fixed pointings used to cover the full 4π celestial sphere. All data from a given HATSouth field and CCD number are reduced together, while detrending through external parameter decorrelation (EPD) is done independently for each unique unit+CCD+field combination. For HATS-38 we also derived light curves from short (30 s) focus frames that were taken by the HATSouth instruments every ∼20 minutes. The Swope 1 m light curve for HATS-37 covered a predicted secondary eclipse event. bThe median time between consecutive images rounded to the nearest second. Due to factors such as weather, the day–night cycle, and guiding and focus corrections the cadence is only approximately uniform over short timescales. cThe rms of the residuals from the best-fit model.Download table as: ASCIITypeset image
We also searched the light curves for other periodic signals using the generalized Lomb–Scargle method (Zechmeister & Kürster 2009), and for additional transit signals by applying a second iteration of BLS. Both of these searches were performed on the residual light curves after subtracting the best-fit primary transit models. No additional periodic signals are detected for HATS-37. For HATS-38 we detect a periodic signal at a period of P = 21.52 days, semiamplitude of 0.43 mmag, and a false alarm probability, determined via bootstrap simulations, of 10−6.3. We do not detect any additional transit signals in its light curve. The periodic signal detected for HATS-38 may correspond to the photometric rotation period of this K star. The star has , which gives an upper limit of 18.7 ± 1.7 days on the equatorial rotation period. The photometric period of 21.52 days is 1.7σ larger than this upper limit, but a larger value is possible if the rotation axis has and the spots are at a latitude that is rotating more slowly than the equator.
2.2. Spectroscopic Observations
The spectroscopic observations carried out to confirm and characterize both of the transiting planet systems are summarized in Table 2. The facilities used include FEROS on the MPG 2.2 m (Kaufer & Pasquini 1998), Coralie on the Euler 1.2 m (Queloz et al. 2001), HARPS on the ESO 3.6 m (Mayor et al. 2003), WiFeS on the ANU 2.3 m (Dopita et al. 2007), and PFS on the Magellan 6.5 m (Crane et al. 2006, 2008, 2010).
Table 2. Summary of Spectroscopy Observations
Instrument | UT Date(s) | # Spec. | Res. | S/N Rangea | b | RV Precisionc |
---|---|---|---|---|---|---|
Δλ/λ/1000 | () | () | ||||
HATS-37 | ||||||
ANU 2.3 m/WiFeS | 2014 Feb 20 | 1 | 3 | 35 | ⋯ | ⋯ |
ANU 2.3 m/WiFeS | 2014 Feb 20–23 | 3 | 7 | 38–72 | 8.2 | 4000 |
Euler 1.2 m/Coralie | 2014 Mar–2016 Jun | 6 | 60 | 20–29 | 7.05 | 149 |
ESO 3.6 m/HARPS | 2016 Feb 27–29 | 2 | 115 | 19–22 | 6.417 | 38 |
Magellan 6.5 m/PFS+I2 | 2016 Jun–2017 Apr | 11 | 76 | ⋯ | ⋯ | 8.7 |
Magellan 6.5 m/PFS | 2016 Jun 20 | 1 | 76 | ⋯ | ⋯ | ⋯ |
HATS-38 | ||||||
Euler 1.2 m/Coralie | 2016 Nov 16–18 | 3 | 60 | 17–20 | 4.143 | 54 |
ESO 3.6 m/HARPS | 2016 Nov–2017 May | 18 | 115 | 17–47 | 4.144 | 9.2 |
MPG 2.2 m/FEROS | 2016 Dec–2017 Mar | 10 | 48 | 36–67 | 4.130 | 19 |
Magellan 6.5 m/PFS+I2 | 2017 Apr 5–8 | 4 | 76 | ⋯ | ⋯ | 5.7 |
Magellan 6.5 m/PFS | 2017 Apr 19 | 1 | 76 | ⋯ | ⋯ | ⋯ |
Notes.
aSignal-to-noise ratio (S/N) per resolution element near 5180 Å. This was not measured for all of the instruments. bFor high-precision RV observations included in the orbit determination this is the zeropoint RV from the best-fit orbit. For other instruments it is the mean value. We only provide this quantity when applicable. cFor high-precision RV observations included in the orbit determination this is the scatter in the RV residuals from the best-fit orbit (which may include astrophysical jitter), for other instruments this is either an estimate of the precision (not including jitter), or the measured standard deviation. We only provide this quantity when applicable.Download table as: ASCIITypeset image
The FEROS, Coralie, and HARPS observations were reduced to wavelength-calibrated spectra and high-precision RV and bisector span (BS) measurements using the CERES pipeline (Brahm et al. 2017a).
The WiFeS observations of HATS-37, which were used for reconnaissance, were reduced following Bayliss et al. (2013). We obtained a single spectrum at resolution R ≡ Δλ/λ ≈ 3000 from which we estimated the effective temperature, and of the star. Three observations at R ≈ 7000 were also obtained to search for any large amplitude RV variations at the ∼4 level, which would indicate a stellar mass companion.
The PFS observations of both HATS-37 and HATS-38 include observations through an I2 cell, and observations without the cell used to construct a spectral template. The observations were reduced to spectra and used to determine high-precision relative RV measurements following Butler et al. (1996). Spectral line bisector spans and their uncertainties were measured as described by Jordán et al. (2014) and Brahm et al. (2017a).
We also used the HARPS and I2-free PFS observations to determine high-precision stellar atmospheric parameters, including the effective temperature , surface gravity , metallicity , and via the ZASPE package (Brahm et al. 2017b). For HATS-37 we used the PFS observations to perform this analysis, while for HATS-38 this analysis was performed on the HARPS observations.
The high-precision RV and BS measurements are given in Table 4 for both systems.
2.3. Photometric Follow-up Observations
Follow-up higher-precision ground-based photometric transit observations were obtained for both systems, as summarized in Table 1. The facilities used for this purpose include: the Chilean-Hungarian Automated Telescope (CHAT) 0.7 m telescope at Las Campanas Observatory, Chile (A. Jordán et al. 2018, in preparation); 1 m telescopes from the Las Cumbres Observatory (LCO) network, (Brown et al. 2013); the 0.3 m Perth Exoplanet Survey Telescope in Australia (PEST);20 and the Swope 1 m telescope at Las Campanas Observatory in Chile.
Our methods for carrying out the observations with these facilities and reducing the data to light curves are described in our previous papers (Bayliss et al. 2013; Mohler-Fischer et al. 2013; Penev et al. 2013; Jordán et al. 2014; Hartman et al. 2015, 2019; Rabus et al. 2016).
The time-series photometry data are available in Table 3, and are plotted for each object in Figures 1 and 2.
Download figure:
Standard image High-resolution imageTable 3. Light Curve Data for HATS-37 and HATS-38
Objecta | BJDb | Magc | Mag(orig)d | Filter | Instrument | |
---|---|---|---|---|---|---|
HATS-37 | 2455765.23265 | 0.00383 | 0.00267 | ⋯ | r | HS |
HATS-37 | 2455691.59689 | −0.00538 | 0.00294 | ⋯ | r | HS |
HATS-37 | 2455678.60275 | 0.00353 | 0.00337 | ⋯ | r | HS |
HATS-37 | 2455747.90786 | 0.00160 | 0.00295 | ⋯ | r | HS |
HATS-37 | 2455682.93494 | −0.00208 | 0.00250 | ⋯ | r | HS |
HATS-37 | 2455665.61003 | −0.00583 | 0.00305 | ⋯ | r | HS |
HATS-37 | 2455708.92562 | 0.00583 | 0.00249 | ⋯ | r | HS |
HATS-37 | 2455691.60029 | 0.00317 | 0.00300 | ⋯ | r | HS |
HATS-37 | 2455652.61745 | −0.00718 | 0.00291 | ⋯ | r | HS |
HATS-37 | 2455747.91127 | 0.00013 | 0.00262 | ⋯ | r | HS |
Notes.
aEither HATS-37 or HATS-38. bBarycentric Julian date is computed directly from the UTC time without correction for leap seconds. cThe out-of-transit level has been subtracted. For observations made with the HATSouth instruments (identified by "HS" in the "Instrument" column) these magnitudes have been corrected for trends using the EPD and TFA procedures applied prior to fitting the transit model. This procedure may lead to an artificial dilution in the transit depths. The blend factors for the HATSouth light curves are listed in Table 7. For observations made with follow-up instruments (anything other than "HS" in the "Instrument" column), the magnitudes have been corrected for a quadratic trend in time, and for variations correlated with up to three PSF shape parameters, fit simultaneously with the transit. dRaw magnitude values without correction for the quadratic trend in time, or for trends correlated with the seeing. These are only reported for the follow-up observations.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
Table 4. Relative Radial Velocities and Bisector Spans for HATS-37 and HATS-38
System | BJD | RVa | b | BS | Phase | Instrument | |
---|---|---|---|---|---|---|---|
(2,450,000+) | () | () | () | () | |||
HATS-37 | |||||||
HATS-37 | 7505.50363 | −43.39 | 15.00 | 83.0 | 19.0 | 0.268 | HARPS |
HATS-37 | 7507.79700 | 32.57 | 12.00 | 108.0 | 16.0 | 0.797 | HARPS |
HATS-37 | 7557.51991 | 0.29 | 2.71 | 31.5 | 10.7 | 0.277 | PFS |
HATS-37 | 7559.58237 | ⋯ | ⋯ | 15.2 | 8.4 | 0.753 | PFS |
HATS-37 | 7559.61286 | 12.06 | 3.25 | 0.0 | 11.2 | 0.760 | PFS |
HATS-37 | 7614.48266 | 0.50 | 4.95 | 32.2 | 17.6 | 0.427 | PFS |
HATS-37 | 7615.48351 | 4.63 | 4.11 | 3.8 | 10.5 | 0.658 | PFS |
HATS-37 | 7617.48123 | −5.21 | 3.30 | −36.9 | 10.0 | 0.120 | PFS |
HATS-37 | 7619.48263 | 18.92 | 7.62 | −107.5 | 33.0 | 0.582 | PFS |
HATS-37 | 7622.48860 | −15.30 | 3.16 | 27.1 | 11.4 | 0.276 | PFS |
HATS-37 | 7623.47962 | −7.39 | 3.12 | 10.3 | 11.8 | 0.504 | PFS |
HATS-37 | 7624.49150 | 6.33 | 4.59 | 36.5 | 13.4 | 0.738 | PFS |
HATS-37 | 7849.68766 | 25.22 | 4.06 | −52.0 | 10.3 | 0.728 | PFS |
HATS-37 | 7858.79784 | −0.39 | 4.54 | −130.3 | 16.1 | 0.831 | PFS |
HATS-38 | |||||||
HATS-38 | 7708.80458 | −23.72 | 7.70 | 18.0 | 10.0 | 0.262 | HARPS |
HATS-38 | 7736.74573 | 9.76 | 8.60 | −43.0 | 13.0 | 0.648 | FEROS |
HATS-38 | 7737.84703 | −0.54 | 8.20 | −17.0 | 12.0 | 0.900 | FEROS |
HATS-38 | 7740.84458 | −8.94 | 9.70 | 38.0 | 14.0 | 0.585 | FEROS |
HATS-38 | 7741.83263 | 20.76 | 7.90 | −7.0 | 12.0 | 0.811 | FEROS |
HATS-38 | 7759.79504 | 50.26 | 9.00 | −29.0 | 13.0 | 0.917 | FEROS |
HATS-38 | 7804.60303 | −6.22 | 8.40 | −7.0 | 11.0 | 0.158 | HARPS |
HATS-38 | 7805.54338 | −8.04 | 10.10 | 35.0 | 14.0 | 0.373 | FEROS |
HATS-38 | 7805.60189 | −20.22 | 6.60 | −11.0 | 9.0 | 0.387 | HARPS |
HATS-38 | 7806.61869 | −1.12 | 8.40 | 15.0 | 11.0 | 0.619 | HARPS |
HATS-38 | 7806.78686 | 5.26 | 12.10 | 8.0 | 17.0 | 0.658 | FEROS |
HATS-38 | 7807.54628 | −21.74 | 13.50 | 125.0 | 18.0 | 0.831 | FEROS |
HATS-38 | 7829.55448 | 7.86 | 8.80 | −27.0 | 13.0 | 0.862 | FEROS |
HATS-38 | 7830.69385 | −10.24 | 8.50 | 12.0 | 13.0 | 0.122 | FEROS |
HATS-38 | 7848.60266 | −18.23 | 3.02 | −0.5 | 14.9 | 0.215 | PFS |
HATS-38 | 7849.55287 | −1.57 | 3.11 | 12.9 | 19.6 | 0.433 | PFS |
HATS-38 | 7850.56508 | 5.85 | 3.02 | −3.7 | 11.4 | 0.664 | PFS |
HATS-38 | 7851.62685 | 11.30 | 3.89 | −25.4 | 40.4 | 0.907 | PFS |
HATS-38 | 7862.63029 | ⋯ | ⋯ | 9.0 | 10.3 | 0.422 | PFS |
HATS-38 | 7866.48140 | −6.92 | 3.90 | 10.0 | 5.0 | 0.302 | HARPS |
HATS-38 | 7866.52106 | −14.82 | 4.40 | 0.0 | 6.0 | 0.311 | HARPS |
HATS-38 | 7867.48250 | 1.88 | 5.80 | 1.0 | 7.0 | 0.531 | HARPS |
HATS-38 | 7867.50397 | −1.52 | 6.70 | −5.0 | 9.0 | 0.536 | HARPS |
HATS-38 | 7868.49777 | −6.32 | 7.70 | 0.0 | 10.0 | 0.763 | HARPS |
HATS-38 | 7868.52092 | 2.58 | 7.80 | −9.0 | 10.0 | 0.768 | HARPS |
HATS-38 | 7870.52143 | −1.52 | 4.40 | −3.0 | 6.0 | 0.225 | HARPS |
HATS-38 | 7870.54312 | −9.22 | 4.40 | −1.0 | 6.0 | 0.230 | HARPS |
HATS-38 | 7871.52764 | 0.18 | 3.90 | 1.0 | 5.0 | 0.455 | HARPS |
HATS-38 | 7871.54726 | −0.32 | 4.40 | −3.0 | 6.0 | 0.460 | HARPS |
HATS-38 | 7887.50781 | −35.12 | 13.40 | 5.0 | 18.0 | 0.108 | HARPS |
HATS-38 | 7887.53008 | −14.52 | 15.00 | 5.0 | 20.0 | 0.113 | HARPS |
HATS-38 | 7890.47840 | 10.18 | 7.80 | −4.0 | 10.0 | 0.787 | HARPS |
HATS-38 | 7890.50175 | 9.78 | 8.40 | −1.0 | 11.0 | 0.792 | HARPS |
Notes.
aThe zeropoint of these velocities is arbitrary. An overall offset fitted independently to the velocities from each instrument has been subtracted. bInternal errors excluding the component of astrophysical jitter are listed in Table 7.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
2.4. TESS Light Curves
During its primary mission, TESS observed both of our targets. HATS-37 (TIC6036597) was observed on Sector 10, CCD 3 of camera 1, but the source lies within the bleed of a nearby bright star, making the photometry unusable. HATS-38 (TIC168281028) was observed by the TESS primary mission during its first year of operations. The target star fell on Camera 2, CCD 4 of the Sector 9 observations. Photometry was extracted from the Science Processing Operations Center (SPOC Jenkins et al. 2016) calibrated Full Frame Images (FFIs), retrieved via the MAST tesscut tool. Aperture photometry was performed using selected pixels of a 7 × 7 pixel cutout of the FFIs with the lightkurve package (Barentsen et al. 2019). The background flux was estimated from the remainder pixels that excluded nearby stars. We corrected for the flux contribution from nearby stars within our photometric aperture. A list of nearby stars was queried from the TICv8 catalog (Stassun et al. 2019), and their flux contributions to the photometric aperture were computed assuming each star has a Gaussian profile with FWHM of 1.63 pixels, as measured from the TESS pixel response function at the location of the target star. The TESS light curve for HATS-38 is shown in Figure 3.
Download figure:
Standard image High-resolution image2.5. Search for Resolved Stellar Companions
The Gaia DR2 catalog provides the highest spatial resolution optical imaging for both of these targets. Gaia DR2 is sensitive to neighbors with G ≲ 20 mag down to a limiting resolution of ∼1'' (e.g., Ziegler et al. 2018). We find that neither object has a resolved neighbor in the Gaia DR2 catalog within 10''.
For HATS-38 we also obtained J and KS-band images using the WIYN High-Resolution Infrared Camera (WHIRC) on the WIYN 3.5 m telescope at Kitt Peak National Observatory (KPNO) in Arizona. The observations were carried out on the night of 2018 March 18, and have an effective FWHM of 043 in J and 035 in KS. The images were collected at four different node positions in each filter. These were calibrated, background-subtracted, registered, and median-combined using the fitsh software package (Pál 2012).
We find a faint source separated from HATS-38 by 6''. The source is detected at about ∼3σ confidence in both bands, and has a magnitude contrast of ΔJ = 8.05 ± 0.09 mag and ΔKs = 7.18 ± 0.08 mag compared to HATS-38. The object is too faint, and too distant from HATS-38 to be responsible for the transit signal. The J and Ks magnitudes are consistent with it being a 0.09 star that is physically bound to HATS-38, at a current projected separation of ∼2100 au. In that case the source would have G ∼ 23 mag, consistent with the object not being included in Gaia DR2. It could also be an extragalactic source, an earlier M dwarf star that is in the background of HATS-38, or a foreground brown dwarf.
No other sources are detected closer to HATS-38 in the WIYN/WHIRC images. Figure 4 shows the resulting 5σ contrast curves for HATS-38. These curves were generated using the tools described by Espinoza et al. (2016). We can rule out neighbors with ΔJ < 3 mag and ΔKs < 3 mag at a separation of 05, and ΔJ < 7 mag and ΔKs < 6 mag at a separation of 15.
Download figure:
Standard image High-resolution image3. Analysis
We analyzed the photometric and spectroscopic observations of each system to determine the stellar and planetary parameters following the methods described in Hartman et al. (2019), with modifications as summarized most recently in Bakos et al. (2020). Briefly, the modeling involves performing a global fit of all the light curves and RV curves described in Section 2, spectroscopically measured stellar atmospheric parameters, catalog broadband photometry, and stellar parallax using a differential evolution Markov Chain Monte Carlo (DEMCMC) method. We fit the observations in two modes: (1) using an empirical method to determine the stellar mass given the direct observational constraint on the stellar radius and bulk density; and (2) constraining the stellar physical parameters using the PARSEC stellar evolution models (Marigo et al. 2017). We use the MWDUST Galactic extinction model (Bovy et al. 2016) to place a prior constraint on the line-of-sight extinction, but we allow the value to vary in the fit.
We also performed a blend model of each system following Hartman et al. (2019), where we attempt to fit all of the observations, except the RV data, using various combinations of stars, with parameters constrained by the PARSEC models. This is done both to rule out blended stellar eclipsing binary scenarios, and to identify systems that may have an unresolved stellar companion. For the blend modeling we consider five scenarios: (1) a single star with a transiting planet (the H-p scenario); (2) an unresolved binary star system with a transiting planet around the brighter stellar component (the H-p,s scenario); (3) an unresolved binary star system with a transiting planet around the fainter stellar component (the H,s-p scenario); (4) a hierarchical triple star system consisting of a bright star and a fainter eclipsing binary system (the H,s-s scenario); and (5) a blend between a bright foreground star, and a background stellar eclipsing binary system (the H,s-sBGEB scenario). For each case we perform an initial grid search over the most difficult to optimize parameters to find the global maximum likelihood (ML) fit, and then perform a DEMCMC analysis, initializing the chain near the ML location. As part of this analysis we also predict spectral line bisector span (BS) measurements, and RV measurements from the composite system. These are compared to the observed RV and BS measurements to rule out any blend scenarios that, while consistent with the photometric observations, predict much larger RV and BS variations than observed. For blend scenarios containing a transiting planet, we use these simulated RV observations to determine a scaling factor by which we expect the RV semiamplitude K to be reduced by dilution from the stellar companion. We then use this factor to scale the value of K determined from our H-p model of the RV observations to obtain corrected values for the H-p,s and H-s,p models. We assume a 20% uncertainty on the scaling factor.
For HATS-37 we find that the H-p,s scenario provides the best fit to the photometric data, with and , and even greater improvements relative to the H,s-s and H,s-p scenarios. Based on this we conclude that HATS-37 is not a blended stellar eclipsing binary object, but rather is best interpreted as a star with a transiting planet and a fainter, unresolved stellar companion. Note that here the use of the MWDUST Galactic extinction model is critical for coming to this conclusion. When the extinction is allowed to vary without the constraint, we find that the H,s-sBGEB scenario provides a slightly better fit to the data than the H-p,s model, while the improvement of the H-p,s model compared to the H-p model is less significant. These models, however, require much greater extinction (AV > 3 mag in the case of the H,s-sBGEB model, and AV ∼ 1 mag in the case of the H-p model) that is at odds with the total line-of-sight extinction of 0.274 mag based on dust maps. The best-fit H-p,s model, however, yields mag, which is in good agreement with the dust maps.
In addition to the photometric evidence for an unresolved stellar companion to HATS-37A, we also find evidence for such a companion in the RV observations. The PFS RVs of this system show a strong linear trend of day−1 (Figure 5). We included this trend, together with a Keplerian orbit for the transiting system, in our modeling of the RV observations. If the trend corresponds to the line-of-sight acceleration of HATS-37A due to HATS-37B, then given the estimated mass of from our H-p,s model, we can place an upper limit on the current physical separation between the two stars of aAB < 27.2 au. This upper limit corresponds to the case where there is no projected separation between the two stars. The maximum projected separation consistent with this acceleration is , corresponding to a maximum current angular separation between the stars of θAB < 008.
Download figure:
Standard image High-resolution imageFor HATS-38 we find that the H-p, H-p,s, and H,s-sBGEB models provide comparable fits to the photometric data, with , and . These differences are comparable to the 1σ scatter in χ2 for a given model as measured from the Markov Chains, and consistent with the slight improvement in the fit for the H,s-sBGEB and H-p,s models being solely due to the increased complexity of these models. In this case we make use of the RV and BS observations to rule out the H,s-sBGEB model. The simulated HARPS RV and BS observations for the H,s-sBGEB model show significantly larger variations than observed, with the simulated RV rms in excess of 200 , and the simulated BS rms in excess of 300 . The actual HARPS RV and BS observations have rms scatters of only 12 and 8 , respectively, with the RV observations following a Keplerian orbit as expected for the case of a transiting planet system. We can also rule out the H,s-s and H,s-p models based on the photometry, as these both provide significantly worse fits to the data than the H-p model. Since the H-p,s model does not provide a significant improvement over the H-p model, we choose to adopt the parameters for the system assuming it is a single star with a transiting planet. We place a 95% confidence upper limit on the mass of any unresolved companion star of MB < 0.62 . If we adopted the H-p,s model instead, the estimated planetary radius would be smaller by 4%, with a 1σ uncertainty of 5% in the difference. Note that the planet would be smaller due to its host star being smaller, even though the transits would be somewhat diluted.
Figures 1 and 2 compare the best-fit models to the observations for both HATS-37 and HATS-38. The astrometric, spectroscopic, and photometric parameters for both stars are listed in Table 5. Our final set of adopted stellar parameters derived from this analysis is listed in Table 6, while the adopted planetary parameters are listed in Table 7.
Table 5. Astrometric, Spectroscopic, and Photometric Parameters for HATS-37 and HATS-38
HATS-37 | HATS-38 | ||
---|---|---|---|
Parameter | Value | Value | Source |
Astrometric properties and cross-identifications | |||
2MASS-ID | 13191246–2259127 | 10170509–2516345 | |
GAIA DR2-ID | 6194574671813047424 | 5472386851683941376 | |
TIC-ID | 6036597 | 168281028 | |
R.A. (J2000) | 13h19m124637 | 10h17m050796 | GAIA DR2 |
Decl. (J2000) | −22°59'127306 | −25°16'345568 | GAIA DR2 |
() | GAIA DR2 | ||
() | GAIA DR2 | ||
parallax (mas) | GAIA DR2 | ||
Spectroscopic properties | |||
(K) | ZASPEa | ||
ZASPE | |||
() | ZASPE | ||
() | 3.175 ± 0.076 | Assumed | |
() | 0.818 ± 0.023 | Assumed | |
() | HARPSb | ||
Photometric properties | |||
G (mag)c | GAIA DR2 | ||
BP (mag)c | GAIA DR2 | ||
RP (mag)c | GAIA DR2 | ||
B (mag) | APASSd | ||
V (mag) | APASSd | ||
g (mag) | APASSd | ||
r (mag) | APASSd | ||
i (mag) | APASSd | ||
J (mag) | 2MASS | ||
H (mag) | 2MASS | ||
Ks (mag) | 2MASS | ||
W1 (mag) | WISE | ||
W2 (mag) | WISE | ||
W3 (mag) | WISE |
Notes.
aZASPE = zonal atmospherical stellar parameter estimator routine for the analysis of high-resolution spectra (Brahm et al. 2017b), applied to the FEROS spectra of each system. These parameters rely primarily on ZASPE, but have a small dependence also on the iterative analysis incorporating the isochrone search and global modeling of the data. bThe error on is determined from the orbital fit to the RV measurements, and does not include the systematic uncertainty in transforming the velocities to the IAU standard system. The velocities have not been corrected for gravitational redshifts. cThe listed uncertainties for the Gaia DR2 photometry are taken from the catalog. For the analysis we assume additional systematic uncertainties of 0.002 mag, 0.005 mag and 0.003 mag for the G, BP, and RP bands, respectively. dFrom APASS DR6, as listed in the UCAC 4 catalog (Zacharias et al. 2013).Download table as: ASCIITypeset image
Table 6. Adopted Derived Stellar Parameters for HATS-37 and HATS-38
HATS-37 | HATS-38 | |
---|---|---|
Parameter | Value | Value |
Planet Hosting Star HATS-37A and HATS-38 | ||
() | ||
() | ||
(cgs) | ||
() | ||
(K) | ||
Age (Gyr) | ||
AV (mag) | ||
Distance (pc) | ||
Binary Star Companion HATS-37B | ||
() | ⋯ | |
() | ⋯ | |
(cgs) | ⋯ | |
() | ⋯ | |
(K) | ⋯ |
Note. The listed parameters are those determined through the joint differential evolution Markov Chain analysis described in Section 3. For both systems the RV observations are consistent with a circular orbit, and we assume a fixed circular orbit in generating the parameters listed here. Systematic errors in the bolometric correction tables or stellar evolution models are not included, and likely dominate the error budget.
Download table as: ASCIITypeset image
Table 7. Adopted Orbital and Planetary Parameters for HATS-37Ab and HATS-38b
HATS-37Ab | HATS-38b | |
---|---|---|
Parameter | Value | Value |
Light curve parameters | ||
P (days) | ||
Tc ()a | ||
T14 (days)a | ||
(days)a | ||
b | ||
/ | ||
b2 | ||
i (deg) | 89.33 ± 0.45 | 87.21 ± 0.18 |
HATSouth dilution factorsc | ||
Dilution factor 1 | ||
Dilution factor 2 | ⋯ | |
Limb-darkening coefficientsd | ||
⋯ | ||
⋯ | ||
RV parameters | ||
K () | ||
() | ||
( day−1) | ⋯ | |
ee | <0.345 | <0.122 |
RV jitter FEROS ()f | ⋯ | |
RV jitter HARPS () | <72.8 | <2.4 |
RV jitter PFS () | <5.4 | |
Planetary parameters | ||
() | ||
() | ||
g | ||
() | ||
(cgs) | ||
a (au) | ||
(K) | ||
Θ h | ||
(cgs)i |
Notes. For all systems we adopt a model in which the orbit is assumed to be circular. See the discussion in Section 3.
aTimes are in Barycentric Julian date calculated directly from UTC without correction for leap seconds. : reference epoch of midtransit that minimizes the correlation with the orbital period. : total transit duration, time between first to last contact; : ingress/egress time, time between first and second, or third and fourth contact. bReciprocal of the half duration of the transit used as a jump parameter in our MCMC analysis in place of . It is related to by the expression (Bakos et al. 2010). cScaling factor applied to the model transit that is fit to the HATSouth light curves. This factor accounts for dilution of the transit due to blending from neighboring stars and overfiltering of the light curve. These factors are varied in the fit, with independent values adopted for each HATSouth light curve. The factor listed for HATS-37 is for the G567.1 light curve, while for HATS-38 we list the factors for the G561.1, and G561.1.focus light curves in order. dValues for a quadratic law. For HATS-37 the values were determined from the tabulations of Claret (2004) for values of the stellar atmospheric parameters, which varied in the modeling. We list here the values for the spectroscopically determined atmospheric parameters. For HATS-38, the limb-darkening parameters were directly varied in the fit, using the tabulations from Claret et al. (2012, 2013) and Claret (2018) to place prior constraints on their values. The difference in treatment between the two systems stems from differences in the software used to model the blended system HATS-37 and the unblended system HATS-38. eThe 95% confidence upper limit on the eccentricity determined when and are allowed to vary in the fit. fTerm added in quadrature to the formal RV uncertainties for each instrument. This is treated as a free parameter in the fitting routine. In cases where the jitter is consistent with zero, we list its 95% confidence upper limit. gCorrelation coefficient between the planetary mass and radius estimated from the posterior parameter distribution. hThe Safronov number is given by (see Hansen & Barman 2007). iIncoming flux per unit surface area, averaged over the orbit.Download table as: ASCIITypeset image
4. Discussion
We put HATS-37Ab and HATS-38b in the context of the population of known, well-characterized21 transiting exoplanets in Figure 6, where we show a scatter plot of planetary mass versus planetary radius, coding with color the equilibrium temperature. Both planets have a relatively low density close to 0.3 g cm−3, which among with their other properties translates into a transmission spectroscopy metric (TSM, Kempton et al. 2018) of ≈120 for HATS-37Ab and ≈165 for HATS-38b. The latter figure makes HATS-38b an attractive target among the currently known set of transiting Neptunes for transmission spectroscopy. Both targets populate a region in the mass–radius plane that is sparsely populated and where the transition between gas giants and the population of smaller planets occurs. We note that both HATS-37Ab and HATS-38b are among the lowest-mass planets found from ground-based wide-field surveys to date, joining a select group of systems uncovered by such surveys with masses Mp ≲ 0.1MJ: HAT-P-26 b (0.059 ± 0.007 MJ, Hartman et al. 2011), NGTS-4 b (0.0648 ± 0.0094 MJ, West et al. 2019), HAT-P-11 b (0.0736 ± 0.0047 MJ, Bakos et al. 2010; Yee et al. 2018), and WASP-166 b (0.101 ± 0.005 MJ, Hellier et al. 2019).
Download figure:
Standard image High-resolution imageIn Figure 7 we show the population of well-characterized planets in the period–radius plane, where HATS-37Ab and HATS-38b are extremely similar. In this figure we show the region defined as the Neptune desert by Mazeh et al. (2016). While they do not lie in the region with P ≲ 3 days and 0.4 ≲ (Rp/RJ) ≲ 0.8 that is essentially devoid of planets, both HATS-37Ab and HATS-38b lie within the region defined as the Neptune desert, which has an intrinsically low occurrence rate of planets.
Download figure:
Standard image High-resolution imageWhen we consider parameters other than planetary mass and radius and consider the properties of the host stars, the properties of HATS-37Ab and HATS-38b emerge as being particularly rare. Dong et al. (2018) used a large sample of stellar parameters obtained with the Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) to further characterize the Neptune desert region. Their study revealed a dearth of planets in the radius range , which they termed the Saturn valley, and a population of hot Neptunes with radii 2 ≲ (Rp/R⊕) ≲ 6, which are rare (occurrence rate of ≈1% for FGK stars) and whose occurrence is correlated with metallicity in the sense that hot Neptunes appear preferentially around metal-rich stars. In fact, Dong et al. (2018) found the great majority of the hot Neptunes in their sample to be hosted by stars with [Fe/H] ≥ 0.1. Both HATS-37Ab and HATS-38b have radii ≈6.7R⊕, making them large specimens for hot Neptunes and veering into the Saturn valley as defined by Dong et al. (2018). More strikingly, HATS-38 has an estimated metallicity of ≈−0.1, making it a very metal-poor star to host a hot Neptune given the expected occurrence rate at that metallicity of order ∼10−3 (Dong et al. 2018; see their Figure 4). Even if the metallicity was as high as ≈0.05, as allowed at the ≈3.5σ level, the expected occurrence rate is ≲5 × 10−3. Thus, we can see that HATS-37Ab and HATS-38b contribute a new pair of exoplanetary systems with uncommon properties and showcase the continuing contributions of wide-field ground-based surveys to better map the variety of landscapes present in the exoplanetary realm.
Development of the HATSouth project was funded by NSF MRI grant NSF/AST-0723074, operations have been supported by NASA grants NNX09AB29G, NNX12AH91H, and NNX17AB61G, and follow-up observations have received partial support from grant NSF/AST-1108686. A.J., R.B., and V.S. acknowledge support from project IC120009 "Millennium Institute of Astrophysics (MAS)" of the Millenium Science Initiative, Chilean Ministry of Economy. A.J. acknowledges additional support from FONDECYT project 1171208. R.B. acknowledges additional support from FONDECYT Post-doctoral Fellowship Project No. 3180246. L.M. acknowledges support from the Italian Minister of Instruction, University and Research (MIUR) through FFABR 2017 fund. T.H. acknowledges support from the European Research Council under the Horizon 2020 Framework Program via the ERC Advanced Grant Origins 83 24 28. K.P. acknowledges support from NASA grant 80NSSC18K1009.2 This work is based on observations made with ESO Telescopes at the La Silla Observatory. This paper also makes use of observations from the LCOGT network. Some of this time was awarded by NOAO. We acknowledge the use of the AAVSO Photometric All-Sky Survey (APASS), funded by the Robert Martin Ayers Sciences Fund, and the SIMBAD database, operated at CDS, Strasbourg, France. Operations at the MPG 2.2 m Telescope are jointly performed by the Max Planck Gesellschaft and the European Southern Observatory. We thank the MPG 2.2 m telescope support team for their technical assistance during observations. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. This research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program.
Facilities: HATSouth - , LCOGT - , FTS - , CTIO:0.9 m - , Danish 1.54 m Telescope (DFOSC) - , Swope - , Max Planck:2.2 m (FEROS) - , ESO:3.6 m (HARPS) - , Euler1.2 m (Coralie) - , ATT (WiFeS) - , AAT (CYCLOPS) - , Magellan:Clay (PFS) - , VLT:Kueyen (UVES) - , NTT (Astralux Sur) - , Gaia. -
Footnotes
- *
The HATSouth network is operated by a collaboration consisting of Princeton University (PU), the Max Planck Institute für Astronomie (MPIA), the Australian National University (ANU), and Universidad Adolfo Ibáñez (UAI). The station at Las Campanas Observatory of the Carnegie Institution for Science is operated by PU in conjunction with UAI, the station at the High Energy Spectroscopic Survey site is operated in conjunction with MPIA, and the station at Siding Spring Observatory (SSO) is operated jointly with ANU. This work is based in part on observations made with the MPG 2.2 m Telescope at the ESO Observatory in La Silla and based on observations collected at the European Southern Observatory under ESO programmes 094.C-0428(A), 095.C-0367(A), 097.C-0571(A), 098.C-0292(A), 099.C-0374(A), 0100.C-0406(A), 0100.C-0406(B). This paper includes data gathered with the 6.5 m Magellan Telescopes at Las Campanas Observatory, Chile, and is based in part on observations made with the Anglo-Australian Telescope operated by the Australian Astronomical Observatory.
- 20
- 21
We use the catalog of well-characterized planets of Southworth (2011). The catalog is kept updated online at https://www.astro.keele.ac.uk/jkt/tepcat/ and the data we used were retrieved in 2019 November. We restrict the sample to systems whose fractional error on their planetary masses are <50%, and planetary radii are <25%.