KELT-23Ab: A Hot Jupiter Transiting a Near-solar Twin Close to the TESS and JWST Continuous Viewing Zones

KELT-23A is located in KELT-North survey field 22, a $26^\circ \times 26^\circ$ region centered at (${\alpha }_{{\rm{J}}2000}={16}^{{\rm{h}}}{03}^{{\rm{m}}}12\buildrel{\rm{s}}\over{.} 0$, ${\delta }_{{\rm{J}}2000}\,=+57^\circ 00^{\prime} 00\buildrel{\prime\prime}\over{.} 0$). We observed the field a total of 3308 times from 2012 February to 2017 August. TYC 4187-996-1 was identified as a transit candidate within the field at $\alpha ={15}^{{\rm{h}}}{28}^{{\rm{m}}}35\buildrel{\rm{s}}\over{.} 1926,\,\delta =+66^\circ 21^{\prime} 31\buildrel{\prime\prime}\over{.} 544$. The initial discovery light curve, shown in Figure 1, contains a significant box-least-squares (Kovács et al. 2002) signal with a period of 2.2552375 days, a transit duration of 1.80 hr, and a transit depth of 12.8 mmag. The KELT image analysis and light-curve production pipeline is described in detail in Siverd et al. (2012). Table 1 shows photometric and astrometric properties of KELT-23A.

Zoom In Zoom Out Reset image size

Table 1.  Literature Properties for KELT-23A

Other Names
	BD+66 911
	TYC 4187-996-1
	TIC 458478250
Parameter	Description	Value	References
αJ2000	Right ascension (R.A.)	${15}^{{\rm{h}}}{28}^{{\rm{m}}}35\buildrel{\rm{s}}\over{.} 1926$	1
δJ2000	Declination (decl.)	$+66^\circ 21^{\prime} 31\buildrel{\prime\prime}\over{.} 544$	1
βJ2000	Ecliptic latitude	+75°6'43''.21	1
λJ2000	Ecliptic longitude	163°17'10''.38	1
BT	Tycho BT mag.	11.029 ± 0.049	2
VT	Tycho VT mag.	10.376 ± 0.039	2
B	APASS Johnson B mag.	10.973 ± 0.021	3
V	APASS Johnson V mag.	10.253 ± 0.024	3
g'	APASS Sloan g' mag.	10.647 ± 0.026	3
r'	APASS Sloan r' mag.	10.114 ± 0.080	3
i'	APASS Sloan i' mag.	9.980 ± 0.070	3
J	2MASS J mag.	9.208 ± 0.032	4
H	2MASS H mag.	≥8.951*	4
K	2MASS K mag.	≥8.904*	4
WISE1	WISE1 mag.	8.746 ± 0.022	5
WISE2	WISE2 mag.	8.778 ± 0.020	5
WISE3	WISE3 mag.	8.778 ± 0.019	5
WISE4	WISE4 mag.	8.672 ± 0.189	5
μα	Gaia DR2 proper motion	0.434 ± 0.039	1
	in R.A. (mas yr−1)
μδ	Gaia DR2 proper motion	−12.217 ± 0.041	1
	in decl. (mas yr−1)
Π	Gaia DR2 parallax (mas)	7.973 ± 0.021	1†
RV	Systemic radial	$-15.224\pm 0.10$	Section 2.3
	velocity (km s−1)
$v\sin {i}_{\star }$	Stellar rotational	$2.42\pm 0.50$	Section 3.1
	velocity (km s−1)
Sp. Type	Spectral type	G2V	Section 3
Age	Age (Gyr)	${6.3}_{-3.2}^{+3.5}$	Section 3.3
d⋆	Distance (pc)	$125.42\pm 0.35$	1†
AV	Visual extinction (mag)	0.075 ± 0.015	Section 3.2
${U}^{\dagger \dagger }$	Space motion (km s−1)	17.32 ± 0.03	Section 4.2
V	Space motion (km s−1)	0.95 ± 0.07	Section 4.2
W	Space motion (km s−1)	−0.90 ± 0.07	Section 4.2

Note. All photometric apertures are blended with the partner within 45. References are: (1) Gaia Collaboration et al. (2018) Gaia DR2 http://gea.esac.esa.int/archive/; (2) Høg et al. (2000); (3) Henden et al. (2015); (4) Cutri et al. (2003); (5) Cutri et al. (2012). *Two Micron All Sky Survey (2MASS) magnitude contains the "U" quality flag. † Gaia DR2 parallax and distance after correcting for the systematic offset of −0.082–mas as described in Stassun & Torres (2018). ††–U is taken to be positive in the direction of the galactic center.

Download table as:  ASCIITypeset image

The KELT-Follow-Up Network (FUN) is a collaboration of around 60 observatories spread throughout the northern and southern hemispheres.⁴²  The KELT-FUN follow-up photometry is used to better constrain the transit parameters, eliminate false positives, and determine accurate ephemerides. KELT-FUN members plan photometric follow-up observations through the use of TAPIR (Jensen 2013), a web-based transit prediction calculator. KELT-FUN members generally use the AstroImageJ (AIJ) package (Collins & Kielkopf 2013; Collins et al. 2017)⁴³ to reduce and perform a preliminary analysis of their light curves. We obtained 11 follow-up transits between 2018 January 26 and July 3 from the observatories listed in Table 2. The specifications listed in Table 2 are taken from Collins et al. (2018), which is the most current listing of KELT-FUN instrument specifications. The follow-up light curves span the Johnson BVRI and Sloan $g^{\prime} r^{\prime} i^{\prime} z^{\prime}$ filter sets. As part of the follow-up confirmation process the transits in different bandpasses were checked for chromatic variations. No chromatic variations were found. The individual light curves are shown in Figure 2.

Zoom In Zoom Out Reset image size

Table 2.  Photometric Follow-up Observations of KELT-23Ab

Observatory	Location	Aperture	Plate Scale	Date	Filter	Exposure	rmsa	Detrending Parametersb	Transit
		(m)	($^{\prime\prime} {\mathrm{pix}}^{-1}$)	(UT 2018)		Time (s)	(10−3)		Coverage
WCO	PA, USA	0.35	0.45	26 January	I	90	2.42	Airmass, total counts, Y	Full transit
PvdK	PA, USA	0.61	0.38	26 January	r'	60	2.67	Airmass, FWHM	Full transit
DEMONEXT	AZ, USA	0.50	0.90	03 February	g'	30	4.95	Airmass, X	Full transit
KeplerCam	AZ, USA	1.20	0.37	04 February	i'	15	3.58	Airmass	Mid-transit, egress
astroLAB	Zillebeke, Belgium	0.684	1.86	17 February	I	5	6.01	Airmass, sky brightness	Partial ingress, egress
PvdK	PA, USA	0.61	0.38	10 March	z'	60	2.86	Airmass, Y	Full transit
ASP	MA, USA	0.355	0.69	19 March	B	30	5.80	Airmass, total counts, X	Full transit
ASP	MA, USA	0.355	0.69	19 March	R	10	4.78	Airmass, total counts	Full transit
WCO	PA, USA	0.35	0.45	19 March	z'	120	4.60	Airmass	Full transit
SOTES	Suwalki, Poland	0.10	1.65	31 May	V	120	4.05	Airmass, total counts	Full transit
KUO	PA, USA	0.61	0.72	03 July	V	90	2.55	Airmass, total counts	Ingress, mid-transit, post-egress

Notes.

^aThe root-mean-square scatter of the best-fit transit model residuals. ^bPhotometric parameters are allowed to vary in the global fit and are described in the text. Abbreviations: Westminster College Observatory, WCO; Peter van de Kamp Observatory, PvdK; DEdicated MONitor of EXotransits and Transients, DEMONEXT; public observatory astroLAB IRIS,AstroLAB; Acton Sky Portal, ASP; Gabriel Murawski Private Observatory, SOTES; Kutztown University Observatory, KUO.

Download table as:  ASCIITypeset image

The KELT-FUN light curves were detrended by various parameters such as airmass, FWHM, X position, Y position, sky brightness, and the total comparison of star counts. To determine the best set of detrending parameters we used the Bayesian information criterion (BIC) of a preliminary fit to the detrended data. Detrending parameters that produced a drop in BIC of 10 or greater were selected for the global fit (Section 4.1). These free parameters can be found in Table 2. In all cases, airmass was selected as a detrending parameter due to the pervasive airmass-dependent systematic effects seen in differential photometry. Other parameters such as the X and Y target pixel position were evaluated as they can induce trends in the light curve due to flat fielding. Observatories that perform a meridian flip during observations additionally have the option of detrending by meridian flip in AIJ as a meridian flip can cause a shift in the position of the target. Meridian flip detrending was not found to be necessary for any KELT-FUN light curves in the current analysis.

2.3.1. Tillinghast Reflector Échelle Spectrograph at Fred Lawrence Whipple Observatory

We obtained 21 spectra from the Tillinghast Reflector Échelle Spectrograph (TRES⁴⁴ ; Szentgyorgyi & Fűrész 2007; Fűrész et al. 2008) on the 1.5 m telescope at the Fred Lawrence Whipple Observatory (FLWO) on Mount Hopkins, Arizona, USA. We observed KELT-23A with TRES between UT 2018 February 18 and UT 2018 June 7. TRES is a 23 (as projected on the sky) fiber-fed échelle spectrograph with a wavelength dispersion between 3900 and 9100 Å and a resolving power of R ∼ 44,000. Radial velocities were obtained through the procedure described below. These spectra were also used to calculate bisector spans (BSs) for use in the false-positive analysis in Section 5. The procedure for calculating BSs can be found in detail in Buchhave et al. (2010). The TRES radial velocities and BSs can be seen in Table 3.

Table 3.  Absolute Radial Velocities and Bisector Spans for KELT-23Ab

${\mathrm{BJD}}_{\mathrm{TDB}}$	RV	σRV	Bisector	σBisector	Source
	( s−1)	( s−1)	( s−1)	( s−1)
2458168.018618	−162.7	27.3	20.0	21.8	TRES
2458184.972170	116.6	24.9	19.4	23.1	TRES
2458200.902902	75.4	19.1	−20.8	14.2	TRES
2458201.932662	−222.8	21.0	−9.1	16.1	TRES
2458210.887269	−200.7	20.1	48.1	13.9	TRES
2458211.973315	127.7	20.1	26.2	11.7	TRES
2458218.822284	105.3	23.6	6.7	16.3	TRES
2458227.783676	101.9	21.1	−11.1	14.9	TRES
2458241.834480	−56.7	20.2	−22.3	14.2	TRES
2458243.807620	61.0	18.6	10.9	7.8	TRES
2458244.791342	−174.6	20.1	−6.9	11.0	TRES
2458245.744880	89.4	23.1	−44.5	25.2	TRES
2458255.919545	−217.4	16.3	−19.1	12.1	TRES
2458263.868286	72.6	13.4	−2.3	6.5	TRES
2458268.747169	16.4	14.6	4.0	12.5	TRES
2458269.736791	−194.2	13.7	−6.9	10.1	TRES
2458271.788423	−204.9	14.9	8.7	8.9	TRES
2458273.749890	−200.5	13.1	−7.0	14.3	TRES
2458274.733876	−19.2	12.3	−5.0	8.6	TRES
2458275.708487	−53.3	13.9	−10.2	10.9	TRES
2458276.745735	−113.8	13.0	21.2	17.7	TRES
2458167.067925	138.308	8.055	14.9	12.7	APF
2458207.851774	108.658	7.488	21.3	12.5	APF
2458210.942618	−130.985	10.568	42.4	12.2	APF
2458222.808525	36.404	6.853	18.2	13.1	APF
2458230.805805	−79.151	6.772	11.3	12.5	APF
2458233.858489	−49.625	7.187	21.0	12.6	APF
2458275.762324	−19.782	6.392	8.4	12.4	APF
2458283.751511	54.966	7.085	23.5	12.1	APF
2458285.739298	−59.795	6.269	23.4	12.0	APF
2458288.721088	147.346	6.433	25.7	12.5	APF
2458289.708981	−144.553	6.269	18.4	12.9	APF

Download table as:  ASCIITypeset image

Relative radial velocities from TRES are found by first designating the spectrum with the highest signal-to-noise ratio (S/N) as the reference spectrum. Each individual spectrum is then cross-correlated with the reference. This cross-correlation utilizes the portion of the spectrum between 4300 and 5660 Å. To obtain absolute radial velocities from TRES the reference spectrum's Mg b region (5190 Å) is cross-correlated with the appropriate Kurucz (1992) stellar atmosphere spectrum. From there, absolute radial velocities are adjusted to the International Astronomical Union (IAU) Radial Velocity Standard Star system (Stefanik et al. 1999). This correction is of the order of −0.61 km s⁻¹ and serves to correct for the noninclusion of gravitational redshift in the model spectrum. The absolute radial velocity of the KELT-23A system was found to be $-15.224\pm 0.10$ km s⁻¹, where the uncertainty is a result of the residual systematic error in the transformation to the IAU system. The TRES radial velocities are plotted in Figure 3 in green.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

2.3.2. Automated Planet Finder at Lick Observatory

As part of our standard process for validating transiting exoplanets, we observed KELT-23A with infrared high-resolution AO imaging at Keck Observatory. The high-resolution AO imaging has demonstrated the ability to resolve faint stellar companions, as in the case of the hierarchical triple KELT-21 system (Johnson et al. 2018b). The Keck Observatory observations were made with the NIRC2 instrument on Keck II behind the natural guide star AO system. The observations were made on 2018 June 6 in the standard three-point dither pattern that is used with NIRC2 to avoid the left lower quadrant of the detector, which is typically noisier than the other three quadrants. The dither pattern step size was $3^{\prime\prime}$ and was repeated twice, with each dither offset from the previous dither by 05.

The observations were made in the narrow-band Br − γ filter (λ_o = 2.1686; Δλ = 0.0326 μm) with an integration time of 10 seconds with one co-add per frame for a total of 90 s on target. The camera was in the narrow-angle mode with a full field of view of ∼10'' and a pixel scale of approximately 0099442 per pixel. The Keck AO observations clearly detected a companion star 45 to the southwest of the primary star, which is the known star Two Micron All Sky Survey (2MASS) J15283577+6621288 (Figure 5) and must be accounted for in the transit depth determinations for observations that do not resolve the stars (Ciardi et al. 2015). No additional stellar companions were detected to within a resolution ∼005 FWHM (Figure 6).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The sensitivities of the final combined AO image were determined by injecting simulated sources azimuthally around the primary target every 45° at separations of integer multiples of the central source's FWHM (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 the target set the contrast limits at that injection location. The final 5σ limit at each separation was determined from the average of all of the determined limits at that separation and the uncertainty on the limit was set by the rms dispersion of the azimuthal slices at a given radial distance. The sensitivity curve is shown in Figure 6 along with an inset image zoomed to the primary target showing no other companion stars.

Spectral analysis allows us to obtain initial estimates of several characteristics of KELT-23A. The TRES spectra were analyzed using the spectral parameter classification (SPC) procedure outlined in Buchhave et al. (2012). SPC cross-correlates the spectra against a grid of stellar atmospheres made by Kurucz (1979, 1992). Allowing SPC to vary all parameters, we obtain a preliminary ${T}_{\mathrm{eff}}\,=5915\pm 50$ K, $\mathrm{log}{g}_{\star }=4.45\pm 0.10$ (cgs), $\left[\mathrm{Fe}/{\rm{H}}\right]=-0.09\pm 0.08$, and $v\sin {I}_{* }=2.42\pm 0.502$ km s⁻¹. The errors reported represent mean errors. The low $v\sin {I}_{* }$ value indicates that the spectra are not significantly broadened by stellar rotation. The APF spectra were also analyzed and yielded a stellar effective temperature of $5775\pm 110\,{\rm{K}}$ and a [Fe/H] of $-0.03\pm 0.06$. The larger quantity and S/N of the TRES spectra led us to accept the values obtained by the SPC over those obtained on the APF spectra.

Using KELT-23A's spectral energy distribution (SED) together with the Gaia parallax, we were able to determine an empirical constraint on its radius. We use broadband photometry, shown in Table 1, spanning the wavelength range 0.2–22 μm from the Galaxy Evolution Explorer near-UV to the WISE4 passbands. We excluded the 2MASS H and K_S bands because they contained the "U" quality flag in the 2MASS catalog, meaning that the magnitude listed is an upper limit. We adopted the T_eff and [Fe/H] from the spectrosopic analysis (see Section 3.1), such that the one remaining free parameter of the fit is extinction, A_V. We use the stellar atmosphere models of Kurucz (1979, 1992) to fit the SED, and the bolometric flux F_bol is obtained by numerically integrating the (unreddened) fitted SED model. The existence of a known companion within 45 (Section 2.4) indicates that the fluxes in the above bandpasses are blended, so we simultaneously fitted the companion star's SED by enforcing the same distance and extinction (see Section 4.4) but in this case fitting for T_eff, and then subtracting the companion star's F_bol in order to obtain the true F_bol for KELT-23A alone.

We obtain a good fit with reduced χ² = 4.4, resulting in a best fit of A_V = 0.075 ± 0.015 and a (companion flux-corrected) F_bol for KELT-23A of (2.056 ± 0.071) × 10⁻⁹ erg s⁻¹ cm⁻². With the Gaia DR2 parallax (corrected for the systematic offset from Stassun & Torres 2018), we obtain a preliminary stellar radius for KELT-23A of 0.959 ± 0.016 $\,{R}_{\odot }$.

We estimate the age of KELT-23A using the stellar parameters determined from the circular Modules for Experiments in Stellar Astrophysics (MESA) Isochrones and Stellar Tracks (MIST; Paxton et al. 2011, 2013, 2015; Choi et al. 2016; Dotter 2016) model fit in Section 4.1 and Table 4. We also determine an evolutionary track for KELT-23A, using a similar process to that in Siverd et al. (2012). In short, we find the intersection of the evolutionary track corresponding to the M_⋆ and $\left[\mathrm{Fe}/{\rm{H}}\right]$ with the best fitting isochrone for our ${T}_{\mathrm{eff}}\,$ and $\mathrm{log}{g}_{\star }$, as displayed in Figure 8. We find an age of ${6.3}_{-3.2}^{+3.5}$ Gyr, where the uncertainty only accounts for observational uncertainty in the parameters used in the fit. This does not account for systematic uncertainties or calibration uncertainties in the MIST model. We find that KELT-23A is a main-sequence G2V-type star in the second half of its lifetime on the main sequence, making KELT-23A a slightly older solar-type star.

Table 4.  Median Values and 68% Confidence Intervals for the Physical and Orbital Parameters of the KELT-23A System

Parameter	Units	Value	Adopted Value
		(MIST eccentric)	(MIST circular; e = 0 fixed)
Stellar Parameters:
M*	Mass (M⊙)	${0.942}_{-0.055}^{+0.062}$	${0.944}_{-0.054}^{+0.060}$
R*	Radius (R⊙)	0.996 ± 0.015	0.996 ± 0.015
L*	Luminosity (L⊙)	${1.082}_{-0.049}^{+0.051}$	${1.081}_{-0.048}^{+0.051}$
${\rho }_{* }$	Density (cgs)	${1.345}_{-0.087}^{+0.098}$	${1.349}_{-0.083}^{+0.090}$
$\mathrm{log}g$	Surface gravity (cgs)	${4.416}_{-0.027}^{+0.028}$	${4.417}_{-0.025}^{+0.026}$
Teff	Effective temperature (K)	5899 ± 49	5899 ± 49
$[\mathrm{Fe}/{\rm{H}}]$	Metallicity (dex)	$-{0.106}_{-0.077}^{+0.078}$	$-{0.105}_{-0.077}^{+0.078}$
${[\mathrm{Fe}/{\rm{H}}]}_{0}$	Initial metallicity	$-{0.072}_{-0.071}^{+0.070}$	−0.071 ± 0.071
Age	Age (Gyr)	${6.5}_{-3.3}^{+3.7}$	${6.4}_{-3.2}^{+3.5}$
EEP	Equal evolutionary point	${374}_{-32}^{+27}$	${374}_{-30}^{+27}$
$\dot{\gamma }$	Radial velocity slope (m s−1 day−1)	$-{0.196}_{-0.090}^{+0.079}$	$-{0.276}_{-0.11}^{+0.098}$
Planetary Parameters:		b	b
P	Period (days)	2.255249 ± 0.000012	${2.255251}_{-0.000012}^{+0.000011}$
RP	Radius ($\,{R}_{{\rm{J}}}$)	1.323 ± 0.025	1.323 ± 0.025
TC	Time of conjunction (${\mathrm{BJD}}_{\mathrm{TDB}}$)	${2458007.3184}_{-0.0029}^{+0.0027}$	${2458007.3194}_{-0.0028}^{+0.0027}$
T0	Optimal conjunction time (${\mathrm{BJD}}_{\mathrm{TDB}}$)	${2458140.3781}_{-0.0030}^{+0.0028}$	${2458140.3792}_{-0.0029}^{+0.0027}$
a	Semimajor axis (au)	${0.03300}_{-0.00065}^{+0.00071}$	${0.03302}_{-0.00064}^{+0.00068}$
i	Inclination (Degrees)	${85.37}_{-0.30}^{+0.31}$	${85.37}_{-0.30}^{+0.31}$
e	Eccentricity	${0.036}_{-0.013}^{+0.014}$	⋯
ω*	Argument of periastron (Degrees)	${3}_{-40}^{+36}$	⋯
Teq	Equilibrium temperature (K)	1562 ± 21	1561 ± 20
MP	Mass (MJ)	${0.936}_{-0.043}^{+0.047}$	${0.938}_{-0.044}^{+0.048}$
K	RV semi-amplitude (m s−1)	${150.5}_{-3.7}^{+3.8}$	${150.6}_{-4.3}^{+4.4}$
logK	Log of RV semi-amplitude	2.178 ± 0.011	${2.178}_{-0.013}^{+0.012}$
${R}_{P}/{R}_{* }$	Radius of planet in stellar radii	0.1365 ± 0.0011	0.1365 ± 0.0010
$a/{R}_{* }$	Semimajor axis in stellar radii	${7.13}_{-0.16}^{+0.17}$	${7.13}_{-0.15}^{+0.16}$
δ	Transit depth (fraction)	${0.01863}_{-0.00030}^{+0.00031}$	0.01864 ± 0.00028
Depth	Flux decrement at mid-transit	${0.01863}_{-0.00030}^{+0.00031}$	0.01864 ± 0.00028
τ	Ingress/egress transit duration (days)	${0.01706}_{-0.00097}^{+0.0010}$	${0.01710}_{-0.00085}^{+0.00089}$
T14	Total transit duration (days)	0.0992 ± 0.0010	0.09921 ± 0.00092
TFWHM	FWHM transit duration (days)	${0.08211}_{-0.00061}^{+0.00060}$	${0.08208}_{-0.00060}^{+0.00061}$
b	Transit impact parameter	${0.575}_{-0.031}^{+0.028}$	${0.576}_{-0.027}^{+0.024}$
bS	Eclipse impact parameter	${0.575}_{-0.028}^{+0.027}$	⋯
${\tau }_{S}$	Ingress/egress eclipse duration (days)	${0.0171}_{-0.0010}^{+0.0011}$	⋯
${T}_{S,14}$	Total eclipse duration (days)	${0.0994}_{-0.0025}^{+0.0024}$	⋯
${T}_{S,\mathrm{FWHM}}$	FWHM eclipse duration (days)	${0.0822}_{-0.0017}^{+0.0018}$	⋯
${\delta }_{S,3.6\mu m}$	Blackbody eclipse depth at 3.6 μm (ppm)	1484 ± 63	${1482}_{-62}^{+64}$
${\delta }_{S,4.5\mu m}$	Blackbody eclipse depth at 4.5 μm (ppm)	1981 ± 74	${1980}_{-73}^{+76}$
ρP	Density (cgs)	${0.502}_{-0.035}^{+0.038}$	${0.503}_{-0.036}^{+0.039}$
loggP	Surface gravity	${3.123}_{-0.026}^{+0.027}$	3.124 ± 0.027
Θ	Safronov number	${0.0495}_{-0.0015}^{+0.0016}$	0.0496 ± 0.0017
$\langle F\rangle$	Incident flux (109 erg s−1 cm−2)	${1.349}_{-0.070}^{+0.073}$	${1.349}_{-0.067}^{+0.070}$
TP	Time of periastron (${\mathrm{BJD}}_{\mathrm{TDB}}$)	${2458006.80}_{-0.25}^{+0.23}$	${2458007.3194}_{-0.0028}^{+0.0027}$
TS	Time of eclipse (${\mathrm{BJD}}_{\mathrm{TDB}}$)	${2458006.234}_{-0.017}^{+0.015}$	${2458008.4470}_{-0.0028}^{+0.0027}$
TA	Time of ascending node (${\mathrm{BJD}}_{\mathrm{TDB}}$)	2458006.776 ± 0.018	${2458006.7556}_{-0.0028}^{+0.0027}$
TD	Time of descending node (${\mathrm{BJD}}_{\mathrm{TDB}}$)	${2458007.901}_{-0.017}^{+0.019}$	${2458007.8832}_{-0.0028}^{+0.0027}$
$e\cos {\omega }_{* }$		${0.030}_{-0.012}^{+0.011}$	⋯
$e\sin {\omega }_{* }$		0.001 ± 0.021	⋯
${M}_{P}\sin i$	Minimum mass (MJ)	${0.933}_{-0.043}^{+0.047}$	${0.935}_{-0.044}^{+0.048}$
${M}_{P}/{M}_{* }$	Mass ratio	0.000948 ± 0.000030	0.000949 ± 0.000033
$d/{R}_{* }$	Separation at mid-transit	${7.11}_{-0.25}^{+0.27}$	${7.13}_{-0.15}^{+0.16}$
PT	A priori nongrazing transit prob.	${0.1215}_{-0.0044}^{+0.0045}$	0.1210 ± 0.0025
${P}_{T,G}$	A priori transit prob.	${0.1599}_{-0.0058}^{+0.0059}$	0.1593 ± 0.0035
PS	A priori nongrazing eclipse prob.	0.1211 ± 0.0029	⋯
${P}_{S,G}$	A priori eclipse prob.	${0.1594}_{-0.0041}^{+0.0042}$	⋯

Download table as:  ASCIITypeset image

The star's metallicity suggests that it could also be slightly subsolar. A more detailed analysis of individual elemental abundances of the host star is encouraged in order to search for deviations from the solar abundance ratios.

In order to gain a more complete understanding of the parameters of the KELT-23A system, we simultaneously fit the ground-based photometric observations from the KELT-FUN (See Figure 2) and the radial velocities (See Figure 3) from TRES using EXOFASTv2 (Eastman et al. 2013; Eastman 2017). Within the global fit, the MIST stellar evolution models are simultaneously fit to constrain the stellar parameters. We place a Gaussian prior on the $\left[\mathrm{Fe}/{\rm{H}}\right]$ of −0.09 ± 0.08 and ${T}_{\mathrm{eff}}\,$ of 5915 ± 50 K from the SPC analysis of the TRES and APF spectra (See Section 3.1). We also place a Gaussian prior on R_⋆ from the value given by the SED analysis (see Section 3.2, Figure 7). We perform two separate fits of KELT-23Ab, one where eccentricity is a free parameter and the other where eccentricity is enforced to be zero. The results of our two fits are shown in Tables 4 and 5.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 5.  Median Values and 68% Confidence Intervals for the Additional Parameters of KELT-23A from EXOFASTv2

Wavelength Parameters:
	Linear Limb-darkening	Quadratic Limb-darkening	Dilution
	u1	u2	AD
B	0.612 ± 0.050	0.202 ± 0.050	0.0013336 ± 0.0000067
I	0.268 ± 0.034	0.292 ± 0.034	0.002565 ± 0.000013
R	0.342 ± 0.047	0.289 ± 0.048	0.003504 ± 0.000018
${g}^{{\prime} }$	${0.512}_{-0.046}^{+0.047}$	0.227 ± 0.047	0.0014614 ± 0.0000073
${i}^{{\prime} }$	${0.315}_{-0.049}^{+0.050}$	0.312 ± 0.049	0.009836 ± 0.000049
r'	0.374 ± 0.045	0.300 ± 0.047	0.003357 ± 0.000017
z'	0.200 ± 0.034	0.258 ± 0.034	0.02120 ± 0.00011
V	0.448 ± 0.036	0.300 ± 0.035	0.002566 ± 0.000013
Telescope Parameters:
		Mist Eccentric		Mist Circular
		APF	TRES	APF	TRES
γrel	Relative radial velocity offset (m s−1)	${7.2}_{-3.8}^{+4.0}$	$-{59.7}_{-4.5}^{+4.8}$	${7.6}_{-5.2}^{+5.5}$	$-{56.1}_{-4.8}^{+4.9}$
σJ	RV jitter (m s−1)	${9.1}_{-4.3}^{+5.5}$	${5.4}_{-5.4}^{+7.4}$	${14.4}_{-4.3}^{+6.3}$	${8.4}_{-8.4}^{+6.5}$
${\sigma }_{J}^{2}$	RV jitter variance	${82}_{-60}^{+130}$	${28}_{-78}^{+130}$	${210}_{-110}^{+220}$	${70}_{-89}^{+150}$
Transit Parameters (MIST Circular Fit):
Observation	Added variance	TTV (days)	Baseline flux	Detrending coeff.	Detrending coeff.	Detrending coeff.
	σ2	(days)	F0	C0	C1	C2
WCO UT 2018-01-26 (I)	$-{0.00000126}_{-0.00000062}^{+0.00000071}$	${0.0088}_{-0.0028}^{+0.0029}$	0.99953 ± 0.00022	$-{0.00120}_{-0.00061}^{+0.00062}$	$-{0.00034}_{-0.00046}^{+0.00047}$	$-{0.00117}_{-0.00063}^{+0.00062}$
PvdK UT 2018-01-26 (${r}^{{\prime} }$)	${0.00000673}_{-0.00000068}^{+0.00000077}$	${0.0076}_{-0.0028}^{+0.0029}$	1.00025 ± 0.00021	−0.00067 ± 0.00060	−0.00089 ± 0.00068	−0.00061 ± 0.00061
DEMONEXT UT 2018-02-03 (g')	${0.0000231}_{-0.0000018}^{+0.0000019}$	${0.0073}_{-0.0028}^{+0.0029}$	1.00001 ± 0.00028	−0.00017 ± 0.00064	${0.00079}_{-0.00064}^{+0.00063}$	⋯
KeplerCam UT 2018-02-04 (i')	${0.0000132}_{-0.0000013}^{+0.0000015}$	${0.0062}_{-0.0028}^{+0.0029}$	0.99974 ± 0.00031	0.00022 ± 0.00049	⋯	⋯
astroLAB UT 2018-02-17 (I)	${0.0000002}_{-0.0000051}^{+0.0000061}$	${0.0059}_{-0.0028}^{+0.0029}$	${1.00163}_{-0.00063}^{+0.00064}$	−0.0014 ± 0.0016	−0.0012 ± 0.0016	⋯
PvdK UT 2018-03-10 (z')	${0.00000648}_{-0.00000099}^{+0.0000011}$	${0.0062}_{-0.0029}^{+0.0030}$	0.99924 ± 0.00025	0.00081 ± 0.00072	0.00061 ± 0.00073	⋯
ASP UT 2018-03-19 (B)	${0.0000218}_{-0.0000039}^{+0.0000043}$	${0.0099}_{-0.0030}^{+0.0032}$	${0.99946}_{-0.00052}^{+0.00051}$	−0.0011 ± 0.0014	−0.0025 ± 0.0016	−0.0018 ± 0.0014
ASP UT 2018-03-19 (R)	${0.0000126}_{-0.0000017}^{+0.0000019}$	${0.0106}_{-0.0029}^{+0.0030}$	1.00072 ± 0.00027	$-{0.00143}_{-0.00072}^{+0.00073}$	−0.00108 ± 0.00073	⋯
WCO UT 2018-03-19 (z')	${0.0000008}_{-0.0000016}^{+0.0000019}$	${0.0138}_{-0.0032}^{+0.0034}$	${1.00005}_{-0.00062}^{+0.00061}$	${0.00071}_{-0.0010}^{+0.00099}$	⋯	⋯
SOTES UT 2018-05-31 (V)	${0.0000055}_{-0.0000020}^{+0.0000025}$	${0.0131}_{-0.0033}^{+0.0034}$	1.00059 ± 0.00041	$-{0.00183}_{-0.00096}^{+0.00095}$	0.00035 ± 0.00097	⋯
KUO UT 2018-07-03 (V)	${0.00000294}_{-0.00000074}^{+0.00000084}$	${0.0121}_{-0.0032}^{+0.0034}$	0.99865 ± 0.00032	$-{0.00298}_{-0.00095}^{+0.00094}$	$-{0.00147}_{-0.00097}^{+0.00096}$	⋯

Download table as:  ASCIITypeset image

We computed the three-dimensional space motion of KELT-23A in Galactic coordinates (UVW) to characterize it kinematically in a Galactic context. The systemic velocity was measured to be $-15.224\pm 0.10$ km s⁻¹ (Section 2.3), which is within 1σ of the value reported by Gaia DR2 (RV_Gaia = −14.891 ± 0.379; Katz et al. 2019), and the proper motion of the system was found to be μ_α = 0.434 ± 0.039 mas yr⁻¹ and μ_δ = −12.217 ±0.041 mas yr⁻¹ (Table 1). We adopt the distance value derived from the Gaia parallax (Table 1), and the standard of rest values from Coşkunoǧlu et al. (2011). We calculate the space motion to be (U, V, W) = (17.32 ± 0.03, 0.95 ± 0.07, −0.90 ±0.07) km s⁻¹. When compared with the distributions in Bensby et al. (2003), we find a 99.4% probability that KELT-23A is a member of the thin-disk population in the Galaxy. KELT-23A's near-solar metallicity and relatively low velocities are consistent with the age derived in Section 3.3.

Transit timing variation (TTVs) can be evidence of an unobserved planetary companion gravitationally perturbing the observed planet (Agol et al. 2005; Holman & Murray 2005). Hot Jupiters seldom show significant TTVs because they rarely have companions close to their mean motion resonances (Steffen et al. 2012). However, there is at least one example of a hot Jupiter that exhibits TTVs due to a nearby companion, specifically WASP-47b, which was discovered by Hellier et al. (2012) and whose TTVs due to a nearby Neptune–mass companion were discovered in K2 data by Becker et al. (2015). Therefore, it is important that we continue to search for hot Jupiters with detectable TTVs, as these rare systems can provide rich and precious information about the emplacement mechanisms of hot Jupiters.

To analyze any possible TTVs we first converted each observatory's photometric time stamps to ${\mathrm{BJD}}_{\mathrm{TDB}}$ (Eastman et al. 2010) before the global fit in Section 4.1. The accuracy of the time stamps is assured by the synchronization to a standard clock by each observatory during observations. This synchronization assures that any uncertainty in T₀ is a result of the global fit. We then plotted the O–C residuals between the observed transit times and the calculated transit times using the ephemeris determined in Section 4.1 (Table 4). The O–C residuals are in Table 6 and plotted in Figure 9. Similar to the KELT-14 TTV analysis (Rodriguez et al. 2016), it would appear that there are significant TTVs. However, observatory-dependent factors such as seeing conditions and astrophysical red noise (Carter & Winn 2009) can induce discrepancies in the observed transit mid-times. Due to these factors, we do not claim that the observed TTVs are astrophysical in nature and conclude that they are likely spurious and a result of systematics. Continued monitoring, especially at the precision that will be reached by TESS, will unveil any true TTVs.

Zoom In Zoom Out Reset image size

Table 6.  Transit Times for KELT-23Ab

Epoch	T0	${\sigma }_{{T}_{0}}$	O–C	O–C	Telescope
	(${\mathrm{BJD}}_{\mathrm{TDB}}$)	(s)	(s)	(${\sigma }_{{T}_{0}})$
−3	2458144.8984	40	116	2.9	WCO
−3	2458144.89724	38	16	0.4	PvdK
1	2458153.91793	51	−43	−0.8	DEMONEXT
0	2458153.91681	82	−138	−1.7	KeplerCam
6	2458167.4484	95	−190	−2.0	astroLAB
15	2458187.74583	55	−250	−4.5	PvdK
20	2458196.77035	95	17	0.2	ASP
20	2458196.77106	54	77	−1.4	ASP
20	2458196.7735	103	285	2.7	WCO
54	2458273.45249	85	25	0.3	SOTES
66	2458302.76997	70	−155	−2.2	KUO

Download table as:  ASCIITypeset image

As with some other recently discovered hot-Jupiter hosts, such as KELT-18 (McLeod et al. 2017), there is evidence that KELT-23A is a member of a wide binary star system. The nearby star described in Section 2.4 is most likely gravitationally bound to KELT-23A. The agreement of the parallaxes of both components within their sub-one-percent uncertainties, and the consistency of their relative proper motion with their expected orbital motion, lead us to conclude that the nearby star is bound to KELT-23A.

The nearby star's Gaia DR2 parallax (also corrected for the systematic offset from Stassun & Torres 2018) is 7.977 ± 0.052 mas and places the star at a distance of 125.36 ± 0.82 pc, equal to KELT-23A's Gaia-derived distance to within the small uncertainty. At that distance, with an apparent angular separation of 45, the projected (minimum) distance between the two stellar components would be 570 au.

The nearby star's Gaia DR2 proper motions of μ_α = 1.567 ± 0.092 mas yr⁻¹ and μ_δ = −11.902 ± 0.107 mas yr⁻¹ are close to KELT-23A's (Table 1) and give the two stars a relative proper motion between them of just 1.2 mas yr⁻¹. This corresponds to a relative tangential velocity of 0.72 km s⁻¹, which could be produced by orbital motion. For illustration, a 0.25 M_⊙ companion at 570 au from KELT-23A would be in a 12,000 yr orbit. KELT-23A would have a (circular) orbital speed of 0.28 km s⁻¹ and the companion's orbital speed would be 1.12 km s⁻¹. A 0.5 M_⊙ companion would orbit with a period of 11,000 yr giving KELT-23A and its companion (circular) orbital speeds of 0.52 km s⁻¹ and 1.0 km s⁻¹, respectively. The SED analysis (Section 3.2) produces relative fluxes for the companion (Table 7) that are consistent with a ∼0.3 M_⊙ star.

Table 7.  Relative Fluxes for the Companion Star, Determined by the SED Analysis (Section 3.2)

Filter	${F}_{B}/{F}_{A}$	Filter	${F}_{B}/{F}_{A}$
B	0.00134	g'	0.00146
V	0.00257	r'	0.00337
R	0.00352	i'	0.00993
I	0.01282	z'	0.02166

Download table as:  ASCIITypeset image

The presence of KELT-23A's distant companion adds it to a family of systems that are informing the developing theories about the formation of hot Jupiters. One way to drive a gas giant planet inward toward its host star is through eccentric orbital interactions with a distant companion star via Kozai–Lidov oscillations (Fabrycky & Tremaine 2007). The planet's orbit would later circularize due to tidal friction as the planet approaches the host star.

We simulate the past and future evolution of KELT-23Ab's orbit using the parameters derived in Section 4.1 as boundary conditions. We use the POET code for this simulation (Penev et al. 2014) under several assumptions: a circular orbit, no perturbing companion planets, and a constant phase lag (constant tidal quality factor). The simulation analyzes the change in the semimajor axis over time due to the tidal forces exchanged by the star and planet. The simulation also analyzes the change in incident flux received by the planet over time. This change in incident flux is a result of both the changing semimajor axis and host-star luminosity as it evolves. Tidal force strengths are parameterized through the tidal dissipation parameter, ${Q}_{* }^{{\prime} }$. This parameter is defined as the quotient of the tidal quality factor (Q_*) and the Love number (k₂). We test values of ${Q}_{* }^{{\prime} }$ ranging from 10⁵ to 10⁸ stepping up only the order of magnitude. This large range in tidal dissipation factor allows us to probe a large range of timescales for energy dissipation. These timescales correspond to the large range of proposed mechanisms for tidal dissipation.

If tides have been the primary cause of changing orbital properties, then KELT-23Ab has been above the inflation irradiation threshold, as defined in Demory & Seager (2011) (2 × 10⁸ erg s⁻¹ cm⁻²), for its entire history, regardless of the ${Q}_{* }^{{\prime} }$ chosen. Currently, KELT-23Ab receives an incident flux about six times larger than the Demory & Seagar inflation threshold (Figure 10, upper panel). KELT-23Ab's large radius, ${R}_{{\rm{P}}}=1.322\pm 0.025{R}_{{\rm{J}}}$, is almost certainly due to its long history of high stellar irradiation.

Zoom In Zoom Out Reset image size

The future of KELT-23Ab is highly dependent on the value of ${Q}_{* }^{{\prime} }$. A large value (10⁸) will keep the planet in orbit for several gigayears. A small value (10⁵) will quickly cause the planet to spiral into its host star, as seen in the bottom panel of Figure 10.

If, in addition to the orbital properties, the spin of the star is known, then it is possible to constrain ${Q}_{* }^{{\prime} }$, following the procedure of Penev et al. (2018). Briefly, the procedure relies on comparing the observed spin of the star to the one predicted by gyrochronology and assuming that any excess spin is due to tidal dissipation transferring orbital angular momentum to the star. Not having a measurement of the stellar spin, we can instead simply evaluate the frequency-dependent tidal dissipation found in Penev et al. (2018) at the tidal period of this system, resulting in an expected value of log(${Q}_{* }^{{\prime} }$) = 5.8. If this truly is the case, then KELT-23Ab is expected to spiral into its host within the next gigayear or so.

While KELT-23Ab is not an extreme case of star–planet tidal interactions like WASP-18b (Hellier et al. 2009), the differences in host-star properties, mainly the prevalence of a larger surface convection zone on KELT-23A, could potentially lead to observable TTVs in the system over a long time period. Nonetheless, tidal interactions within the KELT-23Ab system are almost certainly strong enough to conspire with potential Kozai–Lidov oscillations to produce a hot Jupiter, if the nearby star is indeed bound to KELT-23A.

The KELT-23A system presents a unique opportunity for follow up from TESS and the JWST. KELT-23Ab has the second-highest ecliptic latitude (β = 75°.112) of any transiting exoplanet, second only to Qatar-1b (Alsubai et al. 2011). The system's high ecliptic latitude makes it visible to the JWST 61% of the year. JWST will be able to characterize planetary atmospheres through the transmission spectroscopy method (Beichman et al. 2014) and KELT-23Ab will be a well-suited candidate for extensive atmospheric studies due to its high equilibrium temperature (${T}_{\mathrm{eq}}=1561$ K) and relatively low surface gravity ($\mathrm{log}{g}_{{\rm{P}}}=3.124$). It is also possible to probe the dayside thermal emission of the planet by obtaining infrared photometry near the secondary eclipse, as described by Rodriguez et al. (2016) in reference to KELT-14b.

The system will also receive multi-sector coverage from TESS, which can be used to determine the transit ephemeris with extremely high precision and detect any TTVs too small to observe from the ground. The photometric precision of TESS would allow for the detection of an Earth-sized companion at an orbital period of several days. TESS will observe many consecutive transits that can be used to study details of the system's transit morphology.

TESS's relatively long observing baseline for KELT-23A will also allow detailed characterization of the planet using the TESS light curve alone. KELT-23A should be observed by TESS in three sectors during year two of its mission. Given KELT-23A's TESS bandpass magnitude of T = 9.739, we estimate that TESS should achieve a per point photometric precision of 180 ppm hr^−1/2 on KELT-23A. Over the three 26-day sectors, TESS will observe ∼34.5 orbital periods of KELT-23Ab. Phase folding the TESS data on the period implies a binned light-curve precision of ∼30 ppm hr^−1/2. Additionally, the high photometric precision of TESS could detect microvariability of the host star due to stellar oscillations (Campante et al. 2016).

We use the Price & Rogers (2014) expressions for the analytic precision estimates on the transit observables given finite exposure times to estimate the precision with which TESS can measure the time of transit center T_C, the ingress/egress durations τ, and the transit and eclipse depths. With the aforementioned per point precision and estimated number of observable transits, TESS should measure T_C to 10⁻⁴ days, approximately 9 s, at both the two-minute or thirty-minute cadences (a factor of 10 improvement over the current precision), the ingress/egress duration τ to fractional uncertainties of 2% and 6% at each cadence, and the transit depth δ to 0.5% at both cadences. Following Stevens et al. (2018), the improvement in the precision on τ at the two-minute cadence would improve the stellar density precision to ≈3% (from the current 6%) in the limit that the ingress/egress duration is the dominant source of uncertainty.

We calculate the anticipated secondary eclipse depth due to reflected light via Equation (39) of Winn et al. (2010). The depth is 4 × 10⁻⁴A_λ, where A_λ is the geometric albedo. For even a very large albedo, A_λ = 0.3, the fractional uncertainties on the reflected light eclipse depths are 71% and 79% at the two-minute and thirty-minute cadence.

Assuming that the temperature of KELT-23Ab is equal to the equilibrium temperature, T_eq = 1534K, as well as perfect heat redistribution in its atmosphere, we calculate the expected secondary eclipse depth from KELT-23Ab's thermal emission to be 5 mmag in the TESS passband; the fractional uncertainty is calculated to be 2% at both TESS cadences, so thermal emission should be readily detectable.

Using this temperature, the results from our global fit, and Equation (35) of Winn et al. (2010), we calculate a scale height of 3 × 10⁻³$\,{R}_{{\rm{J}}}$ and an expected transmission signal of 10⁻⁴, as is expected for a typical hot Jupiter.

We estimated the amplitudes of three components of the phase curve of KELT-23Ab (Doppler beaming, ellipsoidal variations, and reflection) using the equations presented in Faigler & Mazeh (2011), Shporer et al. (2011), and Shporer et al. (2019), and our measured parameters of KELT-23Ab. The amplitude of each phase-curve component, however, depends upon a coefficient α, which is of order unity (see Equations (1)–3 of Shporer et al. 2011). We assumed that the beaming coefficient α_beam = 1, and calculated the ellipsoidal and reflection coefficients per Shporer et al. (2019) and Faigler & Mazeh (2011), respectively, using the linear limb-darkening coefficient from the i' band (as an approximation to the TESS bandpass) and estimated albedo, respectively. This resulted in α_ellip = 1.08, and α_refl = 0.3 (3.0) if the planetary geometric albedo is 0.03 (0.3). These imply phase-curve amplitudes of A_beam ∼ 2 ppm, A_ellip ∼ 3 ppm, and A_refl ∼ 11 (110) ppm. Comparing these to the estimated phase-curve precision, this suggests that the beaming and ellipsoidal components will be undetectable, while the reflected light may be detectable, depending upon the albedo. To the first order, however, the amplitude of the thermal phase curve should be similar to the secondary eclipse depth (assuming a large day–night temperature contrast). Our estimated TESS secondary eclipse depth of 4.9 mmag is much larger than the reflected light component, suggesting that the TESS phase curve will be dominated by thermal emission from the planet.