KELT-14b and KELT-15b

KELT-14b and KELT-15b: An Independent Discovery of WASP-122b and a New Hot Jupiter

Abstract

We report the discovery of KELT-14b and KELT-15b, two hot Jupiters from the KELT-South survey. KELT-14b, an independent discovery of the recently announced WASP-122b, is an inflated Jupiter mass planet that orbits a Gyr, = 11.0, G2 star that is near the main sequence turnoff. The host star, KELT-14 (TYC 7638-981-1), has an inferred mass  =   and radius  =  , and has  =  K,  =  and  = . The planet orbits with a period of days (=2457091.028630.00047) and has a radius R =  R and mass M =  M, and the eccentricity is consistent with zero. KELT-15b is another inflated Jupiter mass planet that orbits a Gyr, = 11.2, G0 star (TYC 8146-86-1) that is near the “blue hook” stage of evolution prior to the Hertzsprung gap, and has an inferred mass  =   and radius  =  , and  =  K,  =  and  = . The planet orbits on a period of days ( = 2457029.16630.0073) and has a radius R =  R and mass M =  M and an eccentricity consistent with zero. KELT-14b has the second largest expected emission signal in the K-band for known transiting planets brighter than . Both KELT-14b and KELT-15b are predicted to have large enough emission signals that their secondary eclipses should be detectable using ground-based observatories.

Subject headings:
planetary systems, stars: individual: KELT-14, stars: individual: KELT-15, techniques: photometric, techniques: radial velocities, techniques: spectroscopic

1. Introduction

The confirmation of over 1000 transiting exoplanets to date is due to the success of ground-based photometric surveys such as HATNet (Bakos et al. 2004), SuperWASP (Pollacco et al. 2006), XO (McCullough et al. 2006), and TrES (Alonso et al. 2004), and the space-based missions CoRoT (Baglin et al. 2006) and Kepler (Borucki et al. 2010). The field has shifted from pure discovery to understanding the demographics of exoplanets and atmospheric characterization. However, many of the discovered planets are too faint or too small for performing atmospheric characterization with current facilities. To date, there are only 29 giant transiting planets orbiting stars with 11.5 in the southern hemisphere1.

It is believed that “hot Jupiters,” gas giant planets that orbit extremely close (orbital periods of a few days) to their host stars, must form beyond the “Snow Line.” Once formed, the giant planets can migrate inward through various methods (Tanaka et al. 2002; Masset & Papaloizou 2003; D’Angelo & Lubow 2008; Jackson et al. 2008; Cloutier & Lin 2013). It has been proposed that Jupiter experienced migration early in its lifetime, but did not migrate all the way inward due to the gravitational pull of Saturn (Walsh et al. 2011). These hot Jupiters, specifically ones orbiting solar-like stars, provide insight into alternate evolutionary scenarios.

The Kilodegree Extremely Little Telescope (KELT) exoplanet survey, operated and owned by Vanderbilt University, Ohio State University, and Lehigh University, has been observing 60 of the sky with a cadence of 10 to 20 minutes for many years. The project uses two telescopes, KELT-North at Winer Observatory in Sonoita, Arizona and KELT-South at the South African Astronomical Observatory (SAAO) in Sutherland, South Africa. The survey is optimized for high-precision (1% RMS) photometry for stars with 8 11 to enable transit discovery of giant planets. Each telescope has a 42 mm aperture, 26 field of view, and a pixel scale of 23/pixel (Pepper et al. 2007, 2012). The first telescope in the survey, KELT-North, has announced six planets orbiting stars brighter than = 11 (Siverd et al. 2012; Beatty et al. 2012; Pepper et al. 2013; Collins et al. 2014; Bieryla et al. 2015; Fulton et al. 2015). The younger counterpart in the survey, KELT-South, has already announced one planet, KELT-10b (Kuhn et al. 2015).

In this paper, we present the discovery of a new hot Jupiter by KELT-South, which we name KELT-15b. We also present another hot Jupiter, which we refer to in this paper as KELT-14b. Shortly before the completion of this paper, a draft manuscript was posted to the arXiv (Turner et al. 2015) describing the discovery of three new exoplanets by the SuperWASP survey. One of the planets they name WASP-122b, which is the same planet we designate as KELT-14b. Since the data we present in this paper were collected independently and the analysis performed before the announcement of WASP-122b, we have chosen to discuss our findings as an independent discovery of this planet, and we refer to it here as KELT-14b. However, we acknowledge the prior announcement of it as WASP-122b.

The paper is organized as follows with each section including both discovered systems, KELT-14b and KELT-15b. In §2 we present our discovery and follow-up observations (photometric and spectroscopic). We present our stellar characterization analysis and results in §3. The global modeling and resulting planetary parameters are discussed in §4 with our false positive analysis described in §5. In §6 we describe the evolutionary analysis, long-term follow-up to look for additional companions in each system, and the value each planetary system has for future atmospheric characterization observations. We summarize our results and conclusions in §7.

Figure 1.— Discovery light curve of KELT-14b (Top) and KELT-15b (Bottom) from the KELT-South telescope. The light curves are phase-folded to the discovery periods of P = 1.7100596 and 3.329442 days respectively; the red points show the light curve binned in phase using a bin size of 0.01.
Figure 2.— (Top) The follow-up photometry of KELT-14b from the KELT follow-up network. The red line is the best model for each follow-up lightcurve. (Bottom) The individual follow-up lightcurves combined and binned in 5 minute intervals. This combined and binned plot represents the true nature of the transit. The combined and binned light curve is for display and is not used in the analysis. The red line represents the combined and binned individual models (red) of each follow-up observation.

2. Discovery and Follow-Up Observations

2.1. KELT-South

KELT-14 and KELT-15 are located in the KELT-South field 34, which is centered at J2000 08 16 12 -54 00 00. Field 34 was monitored in two separate campaigns: First from UT 2010 January 03 to UT 2010 February 19 as part of the KELT-South commissioning campaign, and then again from UT 2012 September 16 to UT 2014 June 14, acquiring a total of 5780 images after post-processing and removal of bad images. Following the strategy described in Kuhn et al. (2015), we reduced the raw images, extracted the light curves, and searched for transit candidates. Two stars emerged as top candidates from this process: KS34C030815 (TYC 7638-981-1, GSC 07638-00981, 2MASS J07131235-4224350) located at 07 13 12347 -42 24 3517 J2000, hereafter as KELT-14, and KS34C034621 (TYC 8146-86-1,GSC 08146-00086, 2MASS J07493960-5207136) located at 07 49 39606 -52 07 1358 J2000, designated as KELT-15 (see Figure 1). The host star properties for both targets are listed in Table 1. We used the box-fitting least squares (BLS) algorithm (Kovács et al. 2002; Hartman 2012) to select these candidates, and the BLS selection criteria and values for both are shown in Table 2.

Parameter Description KELT-14 Value KELT-15 Value Source Reference(s)
TYC 7638-981-1 TYC 8146-86-1
GSC 07638-00981 GSC 08146-00086
2MASS J07131235-4224350 2MASS J07493960-5207136
Right Ascension (RA) 07:13:12.347 07:49:39.606 Tycho-2 Høg et al. (2000)
Declination (Dec) -42:24:35.17 -52:07:13.58 Tycho-2 Høg et al. (2000)
NUV 17.06 0.1 N/A GALEX
B Tycho B magnitude 11.963 11.889 Tycho-2 Høg et al. (2000)
V Tycho V magnitude 11.088 11.440 Tycho-2 Høg et al. (2000)
Johnson V APASS magnitude 10.948 0.05 11.189 0.05 APASS Henden et al. (2015)
Johnson B APASS magnitude 11.64 0.05 11.745 0.05 APASS Henden et al. (2015)
Sloan g’ APASS magnitude 11.247 0.051 11.438 0.03 APASS Henden et al. (2015)
Sloan r’ APASS magnitude 10.733 0.053 11.048 0.03 APASS Henden et al. (2015)
Sloan i’ APASS magnitude 10.631 0.05 10.935 0.05 APASS Henden et al. (2015)
J 2MASS magnitude 9.808 0.024 10.205 0.024 2MASS Cutri et al. (2003)
H 2MASS magnitude 9.487 0.024 9.919 0.023 2MASS Cutri et al. (2003)
K 2MASS magnitude 9.424 0.023 9.854 0.025 2MASS Cutri et al. (2003)
WISE1 WISE passband 9.369 0.023 9.775 0.023 WISE Cutri & et al. (2012)
WISE2 WISE passband 9.414 0.021 9.805 0.020 WISE Cutri & et al. (2012)
WISE3 WISE passband 9.339 0.026 9.919 0.048 WISE Cutri & et al. (2012)
WISE4 WISE passband 9.442 0.495 9.580 WISE Cutri & et al. (2012)
Proper Motion in RA (mas yr) -13.9 2.2 -3.4 2.3 NOMAD Zacharias et al. (2004)
Proper Motion in DEC (mas yr) -1.3 2.0 -2.0 2.9 NOMAD Zacharias et al. (2004)
U Space motion ( km s) -4.6 1.9 7.8 3.8 This work
V Space motion ( km s) -14.6 0.9 2.6 0.8 This work
W Space motion ( km s) -14.0 2.3 -1.5 3.3 This work
Distance Estimated Distance (pc) 20119 29130 This work
RV Absolute RV ( km s) 34.62 0.13 12.20 0.11 This work
Stellar Rotational Velocity ( km s) 7.70.4 7.60.4 This work

NOTES
Red value correspond to upper limits (S/N 2)

U is positive in the direction of the Galactic Center

Table 1Stellar Properties of KELT-14 and KELT-15 obtained from the literature.
BLS Selection KELT-14b KELT-15b
Statistic Criteria KS34C030815 KS34C034621
Signal detection SDE 7.0 7.75403 11.04677
efficiency
Signal to pink-noise SPN 7.0 8.26194 9.78164
Transit depth 0.05 0.01072 0.00841
ratio 1.5 2.16 2.56
Duty cycle q 0.1 0.03333 0.04667
Table 2KELT-South BLS selection criteria

2.2. Photometric Follow-up

Target Observatory Date (UT) Filter FOV Pixel Scale Exposure (s) FWHM Detrending parameters for global fit
KELT-14b PEST UT 2015 January 20 31 21 1.2 60 6.04 airmass, y coordinates
KELT-14b PEST UT 2015 January 25 31 21 1.2 120 7.48 airmass, y coordinates
KELT-14b PEST UT 2015 March 09 31 21 1.2 120 5.56 airmass
KELT-14b Adelaide UT 2015 March 09 16.6 12.3 0.62 60 10.48 airmass, total counts
KELT-14b Hazelwood UT 2015 March 09 18 12 0.73 120 6.10 airmass
KELT-14b Hazelwood UT 2015 March 21 18 12 0.73 120 6.31 airmass
KELT-14b LCOGT UT 2015 March 29 27 27 0.39 39 11.24 airmass, pixel width, total counts
KELT-14b Hazelwood UT 2015 April 02 18 12 0.73 120 7.19 airmass
KELT-15b Adelaide UT 2014 December 27 16.6 12.3 0.62 60 9.95 airmass, y coordinates, total counts
KELT-15b Adelaide UT 2015 January 06 16.6 12.3 0.62 120 13.8 airmass, y coordinates
KELT-15b PEST UT 2015 January 16 31 21 1.2 120 6.35 airmass, sky counts per pixel, total counts

NOTES

All the follow-up photometry presented in this paper is available in machine-readable form in the online journal.

Table 3Photometric follow-up observations and the detrending parameters found by AIJ for the global fit.

To precisely measure the transits of KELT-14b and KELT-15b, we obtained high-cadence, high-precision photometric follow-up using larger telescopes that cleanly resolve the hosts from their neighbors within a few arcseconds. These observations better constrain the period, depth, and duration of the transit and also rule out various false positive scenarios. To predict the transits, we use the web interface, TAPIR (Jensen 2013). For consistency, all follow-up observations were analyzed using AstroImageJ (AIJ) (Collins & Kielkopf 2013; Collins 2015). This software also provides the best detrending parameters that are included in the global fit (see §4.1). The follow-up photometry for KELT-14b and KELT-15b are shown in Figures 2 and 3 respectively.

Figure 3.— (Top) The follow-up photometry of KELT-15b from the KELT follow-up network. The red line is the best model for each follow-up lightcurve. (Bottom) All the follow-up lightcurves combined and binned in 5 minute intervals. This best represents the true nature of the transit. The combined and binned light curve is for display and is not used in the analysis. The red line represents the combined and binned individual models (red) of each follow-up observation.

Lcogt

We observed a nearly full transit of KELT-14b in the Sloan -band on UT 2015 March 29 from a 1-m telescope in the Las Cumbres Observatory Global Telescope (LCOGT) network2 located at Cerro Tololo Inter-American Observatory (CTIO) in Chile. The LCOGT telescopes at CTIO have a 4K 4K Sinistro detector with a 27 27 field of view and a pixel scale of 0.39 per pixel. The typical FWHM of the star in this data set was 11.24 pixels. The reduced data were downloaded from the LCOGT archive and analyzed using the AstroImageJ software. In a portion of the light curve surrounding the transit ingress the target was saturated, therefore we exclude this portion of the data from the global parameter analysis in §4.1.

PEST Observatory

PEST (Perth Exoplanet Survey Telescope) observatory is a home observatory owned and operated by Thiam-Guan (TG) Tan. It is equipped with a 12-inch Meade LX200 SCT f/10 telescope with focal reducer yielding f/5. The camera is an SBIG ST-8XME with a filter wheel having , , , and Clear filters. The focusing is computer controlled with an Optec TCF-Si focuser. The image scale obtained is 1.2 per pixel and a full frame image covers 31 21 . For images in focus the usual star FWHM achieved is about 2.5 to 3.5 pixels. The PEST observatory clock is synced on start up to the atomic clock in Bolder, CO and is resynced every 3 hours. PEST observed full transits of KELT-14b on UT 2015 January 20 () and UT 2015 January 25 (), and a nearly full transit on UT 2015 March 09 (). PEST observed a full transit of KELT-15b on UT 2015 January 16 ().

Hazelwood Observatory

The Hazelwood Observatory is a backyard observatory with 0.32 m Planewave CDK telescope working at f/8, a SBIG ST8XME 1.5K 1K CCD, giving a 18 12 field of view and 0.73 per pixel. The camera is equipped with Clear, B, V, Rc, and Ic filters (Astrodon Interference). Typical FWHM is 2.4 to 2.7. The Hazelwood Observatory, operated by Chris Stockdale in Victoria, Australia, obtained an ingress of KELT-14b in -band on UT 2015 March 09, a full transit in the -band on UT 2015 March 21 and a full transit in -band on UT 2015 April 02. The observatory computer clock is synchronised at the start of each observing session and then every 15 minutes using NTP protocol to time.nist.gov. ACP, ACP Scheduler and MaximDL are used to acquire the images. The camera shutter latency (0.5s) is allowed for within MaximDL and the adjusted exposure time is recorded within the FITS header. Experience with another project has shown that the exposure start time is recorded in the FITS header to within one second of the actual exposure start time.

Adelaide Observations

The Adelaide Observatory, owned and operated by Ivan Curtis is located in Adelaide, Australia (labeled “ICO” in the figures). The observatory is equipped with a 9.25-in Celestron SCT telescope with an Antares 0.63x focal reducer yielding an overall focal ratio of f/6.3. The camera is an Atik 320e, which uses a cooled Sony ICX274 CCD of pixels. The field of view is 16.6 12.3 with a pixel scale of 0.62 per pixel and a typical FWHM around 2.5 to 3.1 . The observatory’s computer clock is synced with an internet time server before each observation session and has an overall timing uncertainty of a few seconds. The Adelaide Observatory observed a full transit of KELT-14b on UT 2015 March 09 () and full transits of KELT-15b on UT 2014 December 27 () and UT 2015 January 06 ().

2.3. Spectroscopic Follow-up

Target Telescope/Instrument Date Range Type of Observation Resolution Wavelength Range Mean S/N Epochs
KELT-14 ANU 2.3/WiFes UT 2015 Feb 02 Reconnaissance R Å 75 1
KELT-14 ANU 2.3/WiFes UT 2015 Feb 02 – UT 2015 Feb 04 Reconnaissance R Å 85 3
KELT-15 ANU 2.3/WiFes UT 2014 Dec 29 Reconnaissance R Å 110 1
KELT-15 ANU 2.3/WiFes UT 2014 Dec 29 – UT 2015 Jan 02 Reconnaissance R Å 80 3
KELT-14 AAT/CYCLOPS2 UT 2015 Feb 26 – UT 2015 May 13 High Resolution R Å 41.6 15
KELT-15 AAT/CYCLOPS2 UT 2015 Feb 27 – UT 2015 May 15 High Resolution R Å 41.2 14
KELT-15 Euler/CORALIE UT 2015 Sep 04 – UT 2015 Sep 13 High Resolution R Å 28.25 5
Table 4Spectroscopic follow-up observations

Reconnaissance Spectroscopy

Since many astrophysical phenomena can create photometric signals that mimic planetary transits, it is important to follow up all candidates carefully to eliminate false positives. After identifying the targets as planet candidates from the KELT photometry, a first stage of spectroscopic reconnaissance was done using the WiFeS spectrograph mounted on the 2.3m ANU telescope at Siding Spring Observatory (Dopita et al. 2007). This instrument is an optical dual-beam, image-slicing integral-field spectrograph. The full WiFeS observing strategy and reduction procedure is described in Bayliss et al. (2013).

RV RV error Instrument
( m s) ( m s)
2457079.939623842 34621.30 16.70 CYCLOPS2
2457079.991772522 34658.60 8.20 CYCLOPS2
2457080.950428010 34456.10 5.20 CYCLOPS2
2457081.937382183 34725.20 5.50 CYCLOPS2
2457083.075531623 34431.30 6.00 CYCLOPS2
2457148.892669835 34776.70 9.80 CYCLOPS2
2457148.924568460 34792.00 12.60 CYCLOPS2
2457149.929456027 34505.50 19.10 CYCLOPS2
2457150.967262368 34528.10 13.10 CYCLOPS2
2457151.873724511 34733.40 14.30 CYCLOPS2
2457153.886260897 34794.20 13.20 CYCLOPS2
2457153.916879608 34729.70 14.90 CYCLOPS2
2457154.898109197 34468.40 8.30 CYCLOPS2
2457155.867229521 34533.50 112.90 CYCLOPS2
2457155.900058155 34560.60 111.50 CYCLOPS2

NOTES

This table is available in its entirety in a machine-readable form in the online journal.

Table 5KELT-14 radial velocity observations with CYCLOPS2.
RV RV error Instrument
( m s) ( m s)
2457081.094367965 12320.4 10.8 CYCLOPS2
2457083.091453823 12105.7 17.6 CYCLOPS2
2457148.910598987 12074.4 16.3 CYCLOPS2
2457148.942507928 12247.1 16.6 CYCLOPS2
2457149.947425124 12072.9 25.1 CYCLOPS2
2457150.985251112 12191.4 17.6 CYCLOPS2
2457151.891071429 12281.0 15.0 CYCLOPS2
2457151.953179348 12291.2 12.7 CYCLOPS2
2457153.903635089 12196.2 16.0 CYCLOPS2
2457153.934254059 12188.1 19.9 CYCLOPS2
2457154.912681718 12334.9 13.3 CYCLOPS2
2457154.921681414 12354.9 17.5 CYCLOPS2
2457155.886209486 12085.3 118.0 CYCLOPS2
2457155.918108410 12057.5 114.8 CYCLOPS2
2457269.908610 12096.39 53.81 CORALIE
2457272.903199 12125.96 81.47 CORALIE
2457273.907330 12221.40 57.97 CORALIE
2457276.897042 12216.76 44.60 CORALIE
2457278.894140 12161.43 24.16 CORALIE

NOTES

This table is available in its entirety in a machine-readable form in the online journal.

Table 6KELT-15 radial velocity observations with CYCLOPS2 and CORALIE.

First, observations of both stars were performed at low resolution (R) in the 3500-6000 Å range to determine their stellar type. Both KELT-14 and KELT-15 were identified with the following parameters: KELT-14 has , (cgs) and ; KELT-15b has , (cgs) and . The low resolution spectra provide poor precision on the and therefore, these values aren’t very reliable.

Additionally, three observations for each target were performed in medium-resolution (R) using the red camera arm of the WiFeS spectrograph (5500-9000 Å) across the expected orbital phase based on the photometrically detected period. These observations were aimed at performing multiple radial velocity (RV) measurements of each target to detect signals higher than 5 km/s amplitude, allowing us to identify grazing binary systems or blended eclipsing binaries. The typical RV precision achieved with this instrument is around 1.5 km/s, and both targets showed no significant variations among the three measurements.

Figure 4.— (Top) the AAT radial velocity measurements (the median absolute RV has been subtracted off) and residuals for KELT-14. The best-fitting orbit model is shown in red. The residuals of the RV measurements to the best fitting model are shown below. (Bottom) The KELT-14 AAT measurements phase-folded to the final global fit ephemeris.

High Precision Spectroscopic Follow-up

To confirm the planetary nature of the companion, we obtain multi-epoch high-resolution spectroscopy. These spectra allow us to very accurately measure the radial velocity of the host star providing us with a precise measurement of the companion’s mass. Also, these spectra provide a much better estimate of the stellar properties.

Cyclops2

Spectroscopic observations of KELT-14 and KELT-15 were carried out using the CYCLOPS2 fibre feed with the UCLES spectrograph instrument on the Anglo-Australian Telescope (AAT) over two observing runs: UT 2015 February 02 - UT 2015 March 01 and UT 2015 May 6 - UT 2015 May 13 (See Figure 4 and 5). The instrumental set-up and observing strategy for these observations closely follow that described in earlier CYCLOPS radial velocity papers (Addison et al. 2013, 2014).

Figure 5.— (Top) the AAT (black) and CORALIE (red) radial velocity measurements (the median absolute RV has been subtracted off) and residuals for KELT-15. The best-fitting orbit model is shown in red. The residuals of the RV measurements to the best model are shown below. (Bottom) The KELT-15 AAT (black) and CORALIE (red) measurements phase-folded to the final global fit ephemeris.

CYCLOPS2 is a Cassegrain fiber-based integral field unit which reformats a 2.5″ diameter on-sky aperture into a pseudo-slit of dimensions equivalent to 0.6” wide 14.5” long (Horton et al. 2012). CYCLOPS2 has 16 on-sky fibers, plus one fiber illuminated by a ThUXe lamp. Each fiber delivers a spectral resolution of over 19 echelle orders in the wavelength range of Å, when used with the UCLES spectrograph in its 79 line/mm grating configuration.

We use a ThAr calibration lamp to illuminate all of the on-sky fibers at the beginning of observations to create a reference ThAr wavelength solution. We then use simultaneous ThUXe data from each exposure to determine low-order distortions which differentially calibrate observations through the night onto the reference ThAr solution. These reductions are carried out using custom MATLAB routines (Wright and Tinney, in prep.). Calibration precision is estimated from the scatter of fits to the simultaneous ThUXe spectral features and these are tested against velocity standards taken each night. The typical calibration precision is < 10 m s. This calibration error is combined with the error from a fit to the cross-correlation profile to give a final uncertainty for each observation.

The cross-correlation profiles are obtained using a weighted cross-correlation (Baranne et al. 1996; Pepe et al. 2002) of a stellar template produced with synspec (Hubeny & Lanz 2011). The velocities are determined from the fit of a generalised normal distribution to the cross-correlation profiles and the errors are estimated from the Jacobian matrix for each fit. We find no correlation between the bisector spans and the measured radial velocities. This provides strong evidence against a blended eclipsing binary scenario.

Coralie

CORALIE is a fibre-fed echelle spectrograph (Queloz et al. 2001) attached to the Swiss 1.2 m Leonard Euler telescope at the ESO La Silla Observatory in Chile. It has a spectral resolution of R60000, a wavelength range of Å, and is able to measure radial velocities of bright stars to a precision of 3 m.s or better (Pepe et al. 2002). In June 2015, the CORALIE spectrograph was equipped with a new Fabry-Peŕot-based calibration system (Wildi et al. 2011). This system replaces the ThAr lamp for the simultaneous reference method that determines and corrects for instrumental drift occurring between the calibration and the science exposure (Baranne et al. 1996). The data-reduction software has been adapted to take into account the new operational mode and take benefit from the higher spectral content, and hence the lower photon noise, on the drift measurement, provided by the Fabry-Peŕot based calibration source. We obtained spectra at five epochs of KELT-15 from UT 2015 September 02 to UT 2015 September 14. All observations were reduced and radial velocities were computed in real time using the standard CORALIE pipeline. The observations from CORALIE are consistent with the CYCLOPS2 measured radial velocities. The results are shown in Figure 5. We find no correlation between the bisector spans and the measured radial velocities (see Figure 6).

Figure 6.— The AAT Bisector measurements for the (Top) KELT-14 and the combined AAT and CORALIE bisector measurements for (Bottom) the KELT-15 spectra used for radial velocity measurements. We find no significant correlation between RV and the bisector spans.

3. Analysis and Results

3.1. SME Stellar Analysis

In order to determine precise stellar parameters for KELT-14 and KELT-15, we use the available high-resolution, low S/N AAT CYCLOPS2 spectra acquired for radial velocity confirmation of the two planetary systems. For each CYCLOPS2 dataset, we took the flux weighted mean of the individual fibers, continuum normalized each spectral order, and stitched the orders into a single 1-D spectrum. We shifted each resulting spectrum to rest wavelength by accounting for barycentric motion, and median combined all observations into a single spectrum with a S/N 50, sufficient for detailed spectroscopic analysis.

Stellar parameters for KELT-14 and KELT-15 are determined using an implementation of Spectroscopy Made Easy (SME) (Valenti & Piskunov 1996). Our Monte Carlo approach to using SME for measuring stellar parameters is detailed in Kuhn et al. (2015). Briefly, we use a multi-trial minimization of 500 randomly selected initial parameter values, each solving for 5 free parameters: effective temperature (), surface gravity (), iron abundance (), metal abundance ([m/H]), and rotational velocity of the star (). We determine our final measured stellar properties by identifying the output parameters that give the optimal SME solution (i.e., the solution with the lowest ). The overall SME measurement uncertainties in the final parameters are calculated by adding in quadrature the internal error determined from the 68.3% confidence region in the map, the median absolute deviation of the parameters from the 500 output SME solutions to account for the correlation between the initial guess and the final fit, and an estimate for the systematic errors in our method when compared to other common stellar spectral analysis tools (see Gómez Maqueo Chew et al. 2013).

Due to the instrument setup used for measuring high-precision radial velocities, the AAT CYCLOPS2 spectra do not include the full MgB triplet wavelength region, a pressure-broadened set of lines commonly used in spectral synthesis modeling to constrain (Valenti & Fischer 2005). The available spectra only include one of the three strong Mg lines in this region. In order to investigate the effect of this constraint on our stellar parameters, we run two separate SME runs for both KELT-14 and KELT-15, one with as a free parameter and the other with fixed from our preliminary global fit of the photometric observations.

Our final SME spectroscopic parameters for KELT-14 are:  = 581790 K,  = 4.160.12, [m/H] = 0.390.03,  = 0.340.09 and a projected rotational velocity  =7.70.4  km s. Similarly, with a fixed =4.23;  = 583475 K, [m/H] = 0.390.03,  = 0.340.09 and  =7.60.4  km s. For KELT-15 we find:  = 602361 K,  = 3.800.08, [m/H] = 0.060.03,  = 0.050.03 and  =11.10.5  km s. With a fixed =4.17 we find;  = 610251 K, [m/H] = 0.020.03,  = 0.050.03 and  =11.10.5  km s. We constrain the macro- and microturbulent velocities to the empirically constrained relationship (Gómez Maqueo Chew et al. 2013). However, we do allow them to change during our modelling according to the other stellar parameters. Our best fitting stellar parameters result in v = 4.05  km s and v = 1.00  km sfor KELT-14, and for KELT-15 v = 4.37  km s and v = 1.19  km s.

3.2. SED Analysis

We construct empirical spectral energy distributions (SEDs) of KELT-14 and KELT-15 using all available broadband photometry in the literature, shown in Figure 7. We use the near-UV flux from GALEX (Martin et al. 2005), the and fluxes from the Tycho-2 catalogue, , , , , and fluxes from the AAVSO APASS catalogue, NIR fluxes in the , , and bands from the 2MASS Point Source Catalogue (Cutri et al. 2003; Skrutskie et al. 2006), and near-and mid-infrared fluxes in the WISE passbands (Wright et al. 2010).

We fit these fluxes using the Kurucz atmosphere models (Castelli & Kurucz 2004) by fixing the values of , and [Fe/H] inferred from the global fit to the lightcurve and RV data as described in §4.1 and listed in Table 5 and Table 6, and then finding the values of the visual extinction and distance that minimize , with a maximum permitted based on the full line-of-sight extinction from the dust maps of Schlegel et al. (1998) (maximum = 0.50 mag and 0.89 mag for KELT-14 and KELT-15, respectively). Note that while the final best SED fits below are in fact well fit with , we did include as a free fit parameter because of the a priori likelihood of as large as 0.50–0.89 mag.

For KELT-14 we find A mag and = 201 pc with the best fit model having a reduced . For KELT-15 we find A and = 291 pc with the best fit model having a reduced . This implies a very good quality of fit and further corroborates the final derived stellar parameters for the KELT-14 and KELT-15 host stars. We note that the quoted statistical uncertainties on and are likely to be underestimated because alternate model atmospheres would predict somewhat different SEDs and thus values of extinction and distance, but for stars of the masses and temperatures of KELT-14 and KELT-15 the systematic differences among various model atmospheres are not expected to be large.

Figure 7.— The SED fit for (top) KELT-14 and (bottom) KELT-15. The red points show the photometric values and errors given in Table 1. The blue points are the predicted integrated fluxes at the corresponding bandpass. The black line represents the best fit stellar atmospheric model.

3.3. Evolutionary State

To better place the KELT-14 and KELT-15 systems in context, we show in Figure 8 the H-R diagrams for the two systems in the versus plane. In each case, we use the Yonsei-Yale stellar evolution model track (Demarque et al. 2004) for a star with the mass and metallicity inferred from the final global fit. Specifically, we are using the global fit where the SME determined and , where was not fixed, as priors (See Section 4.1). The shaded region represents the mass and [Fe/H] fit uncertainties. The model isochrone ages are indicated as blue points, and the final best global fit and values are represented by the red error bars. For comparison, the and values determined from spectroscopy alone (without fixing ) are represented by the green error bars, while the blue error bars represent the case with fixed in the SME analysis (Figure 8).

KELT-14 is a G2 type star near the main-sequence turnoff but not yet in the Hertzsprung gap, with an age of Gyr. KELT-15 is a G0 type star with an age of Gyr, on or near the “blue hook” just prior to the Hertzsprung gap. These classifications are also consistent with those reported in the catalogs of Pickles & Depagne (2010) and Ammons et al. (2006). Note that the observed rotational velocities of the stars (7–11  km s; see Section 3.1) are consistent with the 2–15  km s range observed for solar-type stars with the masses and ages of KELT-14 and KELT-15 (e.g., Soderblom 1983).

Figure 8.— The theoretical H-R diagrams for (top) KELT-14 and (bottom) KELT-15 using the Yonsei-Yale stellar evolution models (Demarque et al. 2004). The values are in cgs units. The red cross represents the values from the final global fit. The blue cross is the position and errors of the SME analysis when was fixed at the initial global fit value and the green cross is when was not fixed. The dashed lines at the edge of the gray shaded region represent the 1 uncertainties on M and [Fe/H] from the global fit. The various ages along the tracks are represented by the blue points.

3.4. UVW Space motion

To better understand the place of KELT-14 and KELT-15 in the galaxy, we calculate the UVW space motion. This exercise can allow us to determine the membership and possibly the age of a star if it is associated with any known stellar groups. To calculate the UVW space motion, we combine the information presented in Table 1 with the determined distance to KELT-14 and KELT-15 from the SED analysis (20119 pc and 29130 pc respectively). We also estimated the absolute radial velocity and error by taking the average and standard deviation of all the measured radial velocities by AAT. This gave us an estimated absolute radial velocity of 34.62 0.13  km sand 12.20 0.11  m sfor KELT-14b and KELT-15b, respectively. We calculate the space motion to be U = -4.6 1.9  km s, V = -14.6 0.9  km s, W = -14.0 2.3  km s for KELT-14 and U = 7.8 3.8  km s, V = 2.6 0.8  km s, W = -1.5 3.3  km s for KELT-15 (positive U pointing toward the Galactic center). Using the peculiar velocity of the Sun with respect local standard rest (U = 8.5  km s, V = 13.38  km s, and W = 6.49  km s), we have corrected for this motion in our calculations of the UVW space motion of KELT-14 and KELT-15 (Coşkunoǧlu et al. 2011). These space motion values give a 99% chance that both KELT-14 and KELT-15 belong to the thin disk, according to the classification scheme of Bensby et al. (2003)

4. Planetary Properties

4.1. EXOFAST Global Fit

To perform a global fit of our photometric and spectroscopic data, we use a modified version of the IDL exoplanet fitting tool, EXOFAST (Eastman et al. 2013). More detailed explanation of the global modeling is provided in Siverd et al. (2012). To determine a system’s final parameters, simultaneous Markov Chain Monte Carlo (MCMC) analysis is performed on the AAT radial velocity measurements and the follow-up photometric observations. To constrain M and R EXOFAST uses either the Yonsei-Yale stellar evolution models (Demarque et al. 2004) or the empirical Torres relations (Torres et al. 2010). Each photometric observation’s raw light curve and the detrending parameters determined from the light curve are inputs for the final fit. We impose a prior on and [Fe/H] using the determined values and errors from the SME analysis of the AAT spectra. From analysis of the KELT-South and follow-up photometric observations, we set a prior on the period. For both KELT-14b and KELT-15b, we perform four global fits: 1) Using the Yonsei-Yale (YY) stellar models with eccentricity fixed at zero. 2) Using the YY stellar models with eccentricity as a free parameter. 3) Using the empirical Torres relations with eccentricity fixed at zero. 4) Using the empirical Torres relations with eccentricity as a free parameter. The results from these four global fits can be seen in Table 7 for the KELT-14 system and Table 8 for the KELT-15 system. For the parameters shown in solar or jovian units, the values for these constants are G = 1.3271244 10 m s, = 6.9566 10 m, = 0.000954638698 , and = 0.102792236   (Standish 1995; Torres et al. 2010; Eastman et al. 2013). All determined values for the four separate global fits are consistent with each other (within 1). We adopt the YY circular fit for all analysis and interpretation for KELT-14b and KELT-15b.

Figure 9.— Transit time residuals for KELT-14b using our final global fit ephemeris. The times are listed in Table 9.
Parameter Units Adopted Value Value Value Value
(YY circular) (YY eccentric) (Torres circular) (Torres eccentric)
Stellar Parameters
    Mass ()
    Radius ()
    Luminosity ()
    Density (cgs)
    Surface gravity (cgs)
    Effective temperature (K)
    Metallicity
Planet Parameters
    Eccentricity
    Argument of periastron (degrees)
    Period (days)
    Semi-major axis (AU)
    Mass ()
    Radius ()
    Density (cgs)
    Surface gravity
    Equilibrium temperature (K)
    Incident flux (10 erg s cm)
RV Parameters
    Time of inferior conjunction ()
    Time of periastron ()
    RV semi-amplitude (m/s)
    Minimum mass () &
    Mass ratio
    RM linear limb darkening
    m/s
    RV slope (m/s/day)
   
   
Primary Transit
    Radius of the planet in stellar radii
    Semi-major axis in stellar radii
    Inclination (degrees)
    Impact parameter
    Transit depth
    Ephemeris from transits () 2457091.0286320.00047
    Ephemeris period from transits (days) 1.71005880.0000025
    FWHM duration (days)
    Ingress/egress duration (days)
    Total duration (days)
    A priori non-grazing transit probability
    A priori transit probability
    Linear Limb-darkening
    Quadratic Limb-darkening
    Linear Limb-darkening
    Quadratic Limb-darkening
    Linear Limb-darkening
    Quadratic Limb-darkening
    Linear Limb-darkening
    Quadratic Limb-darkening
    Linear Limb-darkening
    Quadratic Limb-darkening
Secondary Eclipse
    Time of eclipse ()
    Impact parameter
    FWHM duration (days)
    Ingress/egress duration (days)
    Total duration (days)
    A priori non-grazing eclipse probability
    A priori eclipse probability
Table 7Median values and 68% confidence interval for the physical and orbital parameters of the KELT-14 system
Parameter Units Adopted Value Value Value Value
(YY circular) (YY eccentric) (Torres circular) (Torres eccentric)
Stellar Parameters
    Mass ()
    Radius ()
    Luminosity ()
    Density (cgs)
    Surface gravity (cgs)
    Effective temperature (K)
    Metallicity
Planet Parameters
    Eccentricity
    Argument of periastron (degrees)
    Period (days)
    Semi-major axis (AU)
    Mass ()
    Radius ()
    Density (cgs)
    Surface gravity
    Equilibrium temperature (K)
    Incident flux (10 erg s cm)
RV Parameters
    Time of inferior conjunction ()
    Time of periastron ()
    RV semi-amplitude (m/s)
    Minimum mass ()
    Mass ratio
    RM linear limb darkening
    m/s
    m/s
   
   
Primary Transit
    Radius of the planet in stellar radii
    Semi-major axis in stellar radii
    Inclination (degrees)
    Impact parameter
    Transit depth
    FWHM duration (days)
    Ingress/egress duration (days)
    Total duration (days)
    A priori non-grazing transit probability
    A priori transit probability
    Linear Limb-darkening
    Quadratic Limb-darkening