NGTS-8b and NGTS-9b: two non-inflated hot Jupiters Free

ABSTRACT

1 INTRODUCTION

Hot Jupiters are giant gas exoplanets similar to Jupiter, but with a shorter orbital period, inferior to 10 d. While rare, these planets are the easiest to detect from ground-based surveys due to their relatively deep transits (⁠|$\sim 1{{\ \rm per\ cent}}$|⁠), their large radial velocity (RV) signals, and their short periods, which make hot Jupiters important targets in order to understand the structure, composition, and evolution of planetary systems.

From the currently observed population of exoplanets with known radii, masses, and orbital distances, the evolution of planetary radii has been modelled (e.g. Fortney, Marley & Barnes 2007; Baraffe, Chabrier & Barman 2008). These models, where the effects of stellar irradiation and heavy element cores are included, agree with observations at low stellar irradiation. However, the observed radii of highly irradiated gas giants are discrepant with theoretical expectations. For instance, at fluxes greater than 2 × 10⁵ W m⁻² (Demory & Seager 2011; Miller & Fortney 2011) the gas giants are increasingly found with anomalously large radii. This is the case for the hot Jupiters WASP-17 b, WASP-121 b, and Kepler-435 b, which all have measured radii R > 1.8 R_J (Anderson et al. 2011; Almenara et al. 2015; Delrez et al. 2016). A number of possible mechanisms have been postulated to explain these inflated planetary radii including kinetic heating (Guillot & Showman 2002), enhanced atmospheric opacities (Burrows et al. 2007), double diffusive convection (Chabrier & Baraffe 2007), Ohmic heating through magnetohydrodynamic effects (Batygin & Stevenson 2010; Perna, Menou & Rauscher 2010; Wu & Lithwick 2012; Ginzburg & Sari 2016), tidal dissipation (Bodenheimer, Lin & Mardling 2001; Bodenheimer, Laughlin & Lin 2003; Arras & Socrates 2010; Jermyn, Tout & Ogilvie 2017), and vertical advection of potential temperature (Youdin & Mitchell 2010; Tremblin et al. 2017). However, the exact mechanisms responsible are still, as yet, unidentified and the problem remains unsolved. In order to perform robust statistical studies of hot-Jupiter radii and constrain the dominant ‘inflation’ mechanisms at work (e.g. as done by Sestovic, Demory & Queloz 2018), we need to increase the sample of planets spanning a range of planetary masses, radii, stellar irradiation levels, as well as planetary system ages.

In this paper, we present two hot Jupiters that appear to be non-inflated, despite being highly irradiated with an incident flux greater than 2 × 10⁵ W m⁻² (like many inflated planets). In Section 2, the NGTS discovery data is described. Section 3 explains the photometric follow-up campaigns and Section 4 reports the mass determination via RV monitoring from spectroscopy. Section 5 details the analysis of the stellar parameters, presents the stellar activity, and its relation with the stellar rotation and shows the global modelling process to characterize the planets. Section 6 presents an investigation regarding the incident flux, the planetary mass, and the radius. Finally, we finish with our conclusions in Section 7.

2 DISCOVERY PHOTOMETRY FROM NGTS

The next generation transit survey (NGTS), operating since early 2016, is a widefield transit survey located at ESO’s Paranal Observatory in Chile, whose primary goal is to discover Neptune-sized or bigger exoplanets. NGTS has a fully robotized array of 12 20 cm Newtonian telescopes, and each telescope is equipped with 2K × 2K e2V deep-depleted Andor IKon-L CCD cameras with 13.5 |$\mu$|m pixels and an instantaneous field of view of 8 deg². For a description of this facility and its capabilities, optimized for detecting planets, we refer the reader to Wheatley et al. (2018). NGTS has already detected four hot Jupiters: NGTS-1b (Bayliss et al. 2018), NGTS-2b (Raynard et al. 2018), NGTS-3Ab (Günther et al. 2018), and NGTS-6b (Vines et al. 2019). Here, we report the latest hot-Jupiter discoveries from NGTS: NGTS-8b and NGTS-9b.

NGTS-8 was observed using a single NGTS camera (#811) over a 227 night baseline between the 2016 April 21 and the 2016 December 3. NGTS-9 was also observed using a single camera (#806) over a 234 night baseline between the 2016 October  8 and the 2017 May 29. A total of 177 799 and 167 933 images were obtained, respectively, each with an exposure time of 10 s. These data were taken using the custom NGTS filter (520–890 nm; Wheatley et al. 2018) and the telescope was autoguided using an improved version of the donuts autoguiding algorithm (McCormac et al. 2013). The data were reduced and aperture photometry was extracted using the casutools¹ photometry package. A total of 177 120 and 166 043 valid data points were extracted from the raw images and then detrended for nightly trends, such as atmospheric extinction, using our implementation of the sysrem algorithm (Tamuz, Mazeh & Zucker 2005).

Both data sets were searched for transit-like signals using orion, an optimized implementation of the box-least-squares fitting algorithm (Collier Cameron et al. 2006). A 1.6 per cent deep transit signal was detected at a period of 2.49970 d for the K0V star, NGTS-8, and a 0.6 per cent deep transit signal at 4.43527 d for the F8V star, NGTS-9. These periods were distinguished from other aliases using the photometry and spectroscopy follow-up – see Sections 3 and  4 for details. The detrended NGTS data for the two stars, phase-folded on the planetary orbital periods, are shown in Figs 1 and 2. A sample of the NGTS reduced photometric measurements is presented in Tables 1 and  2, with the full data available electronically from the journal.

Figure 1.

From top to bottom: Fig. 1(a) represents the NGTS discovery light curve of NGTS-8b with residuals, phase-folded on the orbital period and zoomed on the transit. Fig. 1(b) represnts the ingress and mid-transit of NGTS-8b observed with SAAO, with residuals. Fig. 1(c) represents the ingress and mid-transit of NGTS-8b observed with Euler, with residuals. For all the plots, the blue data points are binned every 7 min to aid visualization. The red lines show 20 light-curve models generated from randomly drawn posterior samples of the allesfitter fit.

Open in new tabDownload slide

Figure 2.

From top to bottom: Fig. 2(a) represents the NGTS discovery light curve of NGTS-9b with residuals, phase-folded on the orbital period, and zoomed on the transit. Fig. 2(b) represents the full transit of NGTS-9b observed with SAAO, with residuals. This image uses night 20181221, where ingress and mid-transit can be seen, and night 20190103, with mid-transit and egress. For both figures, the blue data points are binned every 10 min to aid visualization. Fig. 2(c) represents the full transit of NGTS-9b observed with Euler, with residuals. The data were taken on a 100 s cadence, shown in blue. For all the plots, the red lines show 20 light-curve models generated from randomly drawn posterior samples of the allesfitter fit.

Open in new tabDownload slide

The NGTS data were searched for signs that would indicate that the planetary candidates were false positives. No evidence for a secondary eclipse or out-of-transit variations indicating an eclipsing binary system were identified in the NGTS light curves of the two stars. However, for both sources, some stars were found to be in close proximity to our targets. Using Gaia, we confirmed that these nearby stars did not appreciably dilute the light from NGTS-8 or NGTS-9, and also confirmed that NGTS-8 and NGTS-9 were not giants stars – see Section 5.1.1 for details. Based on the NGTS detection, NGTS-8 and NGTS-9 were followed up with further photometry and spectroscopy to confirm the planetary nature of the system and to measure the planetary parameters, which we report on in the next section.

3 FOLLOW-UP PHOTOMETRY

3.1 South African Astronomical Observatory photometry

Follow-up photometry of NGTS-8 was obtained with the 1.0 m Elizabeth telescope at the South African Astronomical Observatory (SAAO) on 2017 July 17 and 2017 July 18, utilizing the frame-transfer CCD Sutherland High-speed Optical Camera ‘SHOC’n’awe’ (SHOC; Coppejans et al. 2013).

With a pixel scale of 0.167 arcsec pixel⁻¹, the SHOC cameras on the 1 m telescope were binned 4 × 4 pixels in the X- and Y-directions. The field of view of its instruments is 2.85 arcmin × 2.85 arcmin. This allows to observe simultaneously the target and a comparison star of similar brightness for differential photometry. The data, obtained using a z′ filter with an exposure of 30 s, were bias and flat-field corrected. This was performed in python using the standard procedure with the ccdproc package (Craig et al. 2015). Then, using the ‘sep’ package (Barbary 2016), the aperture photometry of both the target and the comparison star were extracted. Finally, the sky background was measured and substracted using the SEP background map.

We also obtained two follow-up light curves of NGTS-9 on 2018 December 21 and 2019 January 30, with the same telescope and instrument set-up as described above. This time the data were obtained with an I filter and an exposure time of 20 s. The data were reduced with the local SAAO SHOC pipeline, which is driven by python scripts running iraf tasks (PYFITS and PYRAF), and incorporating the usual bias and flat-field calibrations. Aperture photometry was performed using the Starlink package autophotom. For the first observation of NGTS-9, we used a 4 pixel radius aperture that maximized the signal-to-noise ratio (SNR), and the background was measured in an annulus surrounding this aperture with inner and outer radii of 7 and 9 pixels, respectively. Two comparison stars were then used to perform differential photometry on the target. The 2019 January 30 observation was obtained in slightly poorer seeing conditions, and we therefore utilized a 6 pixel aperture, a correspondingly larger background annulus, and only one comparison star for differential photometry.

The transits of these two exoplanets observed from SAAO are shown in Figs 1(b) and 2(b). Regarding NGTS-8, only a partial transit was observed with SAAO. For NGTS-9, two nights were combined: night 20181221, where ingress and mid-transit were seen, and night 20190103, where mid-transit and egress were seen. While only partial transits were observed, these SAAO data were able to confirm the transits and the consistency of the transit depths and were used to revise the orbital ephemerides for subsequent follow-up observations. In particular, the 2017 July observations of NGTS-8 were helpful in confirming the orbital period and ruling out aliases of similar power in the original NGTS data.

3.2 Eulercam

We also observed both objects with Eulercam (Lendl et al. 2012) on the 1.2 m Euler Telescope at La Silla Observatory. NGTS-8 was observed on the 2017 August 21, 502 exposures were acquired using the Cousins-I filter, an exposure time of 12 s and a defocus of 0.05 mm. NGTS-9 was observed on the 2019 January 12. We acquired 134 images using the Gunn-R filter, a 100 s exposure time and no defocus. For both target, their data were reduced using the standard procedure of bias subtraction and flat-field correction. The aperture photometry as well as x- and y-position, full width at half-maximum (FWHM), airmass, and sky background of the target star were extracted using the PYRAF implementation of the phot routine. The comparison stars and the photometric aperture radius were carefully chosen in order to reduce the RMS in the scatter out of transit.

Using both follow-up photometry for the 2 stars, we can conclude that the nearby stars did not blend with the two targets. The Euler data for the two stars are shown in Figs 1(c) and 2(c). Regarding NGTS-8, only a partial transit was observed with Euler. Concerning NGTS-9, the full transit was observed, with some systematics that were removed using a Gaussian process. Again, these data confirm the transits as we see the ingress or the full transit around the predicted times. Table   3 presents the summary of the follow-up photometry of NGTS-8 and NGTS-9.

4 SPECTROSCOPY

4.1 NGTS-8

NGTS-8 was observed with the HARPS spectrograph (Mayor et al. 2003) on the ESO 3.6 m telescope at La Silla Observatory, Chile, between the 2017 August 5 and the 2017 October 28 under programmes 099.C-0303 and 0100.C-0474. We used the high-efficiency mode, EGGS, due to the faintness of the host star and large expected RV amplitude. The exposure times for each spectrum varied between 1800 and 1200 s resulting in a SNR, measured around 550 nm, of 10–15 per exposure. The standard HARPS data reduction software was used to measure the RVs of NGTS-8 at each epoch. This was done via cross-correlation with a K0 binary mask.

Three additional spectra were obtained with FEROS (Kaufer & Pasquini 1998), mounted on the MPG 2.2 m telescope at La Silla Observatory, Chile, on the 2017 August 20 and 21. All spectra were obtained with an exposure time of 1800 s, and the data were reduced using the FEROS routine of the CERES pipeline (Brahm, Jordán & Espinoza 2017). CERES also performed an RV extraction, by cross-correlating the spectra with a G2 binary mask. The resulting SNR of the spectra, taken around 550 nm per resolution element, was around 45.

The RVs from both HARPS and FEROS are listed in Table 4, along with their associated error, FWHM, and bisector span. While not presented in this table, the error on the BIS and on the FWHM were calculated using the same standard treatment as done previously (West et al. 2018). The errors for the BIS and for the FWHM are set to equal twice the error and 2.3548 times the error on the RV, respectively. The RV measurements of NGTS-8, shown phase folded in Fig. 3, are in phase with the period detected by orion, with a semi-amplitude of K = 149.95 ± 3.56 m s⁻¹. This indicates a transiting planet with the mass of a hot Jupiter. No evidence of a correlation between the RVs and the measured bisector spans or FWHMs were found, with a Spearman correlation of −0.05 and −0.21, respectively. Thus, the RV signal does not originate from cool stellar spots or a blended eclipsing binary (Queloz et al. 2001b).

Figure 3.

Phase-folded radial velocity data in km s⁻¹ and residuals from HARPS, in blue, and FEROS, in orange, for NGTS-8. The red lines show 50 light-curve models generated from randomly drawn posterior samples of the allesfitter fit.

Open in new tabDownload slide

4.2 NGTS-9

NGTS-9 was observed with the CORALIE spectrograph (Queloz et al. 2001a) on the 1.2 m Euler telescope at La Silla Observatory, Chile, between the 2017 December 24 and the 2018 April 5. Exposure times of either 1800 or 2700 s were used depending on seeing and general observing conditions at the time. RVs were calculated with a G2 binary mask using the standard data reduction pipelines. Initial analysis, shown phase folded in Fig. 4, confirm the planetary nature of the candidate with a mass of a hot Jupiter. The RV variations are in phase with the period detected by orion with a semi-amplitude of K = 293.44 ± 15.08 m s⁻¹. Correlations between RVs and bisector span and FWHM were also checked but no evidence for any such correlations was found, with a Spearman correlation of −0.05 and −0.15, respectively. The CORALIE RVs for NGTS-9 are listed in Table 5, along with their associated error, FWHM, and bisector span.

Figure 4.

Phase-folded radial velocity data in km s⁻¹ and residuals from CORALIE for NGTS-9. The red lines show 50 light-curve models generated from randomly drawn posterior samples of the allesfitter fit.

Open in new tabDownload slide

5 ANALYSIS

5.1 Stellar properties

5.1.1 Gaia

To obtain astrometric information for NGTS-8 and NGTS-9, we cross-matched both sources with Gaia DR2. To check the quality of the astrometric solutions, we calculated the unit weight error (UWE) and then renormalized UWE (RUWE). We find that both sources pass the filters recommended by the Gaia team (RUWE < 1.4, see Lindegren et al. 2018, for a discussion on the recommended UWE filters). Along with this, the two targets also have zero astrometric noise, giving us confidence that they are both single sources without evidence of unresolved binarity. They also pass the photometric filters specified by Arenou et al. (2018) to identify blended stars. With the Gaia information for each source, we calculate the absolute magnitude and plot their positions on the Hertzsprung–Russell diagram in Fig. 5, where we can see that both NGTS-8 and NGTS-9 lie in the region expected for single main-sequence stars.

Figure 5.

HR diagram using Gaia DR2 absolute magnitude. The K0V star, NGTS-8, is shown in red and the F8V, NGTS-9 in light blue.

Open in new tabDownload slide

5.1.2 SPECIES

The stacked spectra for both targets were also analysed using SPECIES (Soto & Jenkins 2018), a python tool to derive stellar parameters in an automated fashion from high-resolution echelle spectra. By measuring the equivalent widths for a list of irons lines, and using the ATLAS9 model atmospheres (Castelli & Kurucz 2004), SPECIES first solves the radiative transfer equation using MOOG (Sneden 1973). From an iterative process, SPECIES derives then the atmospheric parameters (T_eff, log g, [Fe/H]) of our target. By interpolation through a grid of MIST isochrones (Dotter 2016), the mass and radius are estimated, using a Bayesian approach. This method delivers an estimate of the age of the system as well. However, due to the fact that solar-type stars spend most of their lives in the evolutionary stage and because the dependence of their effective temperature and luminosity with the age of the system is weak, the estimation of the age of the system is very unconstrained for main-sequence stars. Finally, SPECIES derives the rotational and macroturbulent velocities from the stellar temperature and by line fitting to a set of five absorption lines. Parameters found by SPECIES for both targets are displayed in Tables 6 and  7.

5.2 Stellar activity and rotation on NGTS-8

In addition to modelling the stellar parameters, we also attempted to search for stellar activity and rotation signals. As mentioned earlier, the out-of-eclipse light curves of both targets show no appreciable variability, and there are no correlations with the measured RVs and the bisector or the FWHM for the two targets – see Section 4. None the less, determining the stellar rotation period, along with knowledge of the stellar radius and vsin i_*, can enable the inclination angle of the stellar rotation axis to be constrained. This can enable misaligned star–planet systems to be identified (Watson et al. 2010).

In order to put constraints on the stellar rotation period, two methods were used. The first one consists of using the activity of the star. Using the formulae described in Lovis et al. (2011), the |$\log R^{\prime }_{\mathrm{ HK}}$| was measured for each individual HARPS spectrum. Since the |$\log R^{\prime }_{\mathrm{ HK}}$| of NGTS-9, an F8V star, was not measurable, we will only focus on the K0V star, NGTS-8, in this section. The calculated value of the |$\log R^{\prime }_{\mathrm{ HK}}$| varied between −4.643 and −5.053 in the individual spectra due to a low SNR in the blue band, with a mean value of −4.783. In order to increase the precision of the measured data, a stacked spectrum of all the spectra was created. Fig. 6 shows this spectrum, zoomed on the H (3933.664 Å) and K (3968.470 Å) bands, represented with the dashed lines. Using this spectrum, we found a value of −4.817 ± 0.110 for the |$\log R^{\prime }_{\mathrm{ HK}}$|⁠. Finally, using the relation from Noyes et al. (1984), the stellar rotation period of NGTS-8 was derived to be 37.7 ± 4.1 d.

Figure 6.

Stacked spectrum of NGTS-8 from HARPS, zoomed on the H (3933.664 Å) and K (3968.470 Å) bands, represented with the dashed lines.

Open in new tabDownload slide

Our second approach used the vsin i_* of the star measured from SPECIES and assumed that the stellar inclination angle, i_*, was 90° [i.e. sin (i_*) = 1]. This then enables an upper limit to be placed on the stellar rotation period when the stellar radius is known (e.g. Watson et al. 2010). An upper limit of 13.92 ± 2.64 d was found, which is discrepant with our previous result by almost 5σ and would rule out the long rotation period inferred from the |$\log R^{\prime }_{\mathrm{ HK}}$| measurement. Even adopting the most extreme individual, |$\log R^{\prime }_{\mathrm{ HK}}$| measurement implies a stellar rotation period greater than 26 d.

Given this discrepancy, we decided to verify the vsin i_* value measured by SPECIES with another technique. We did this by taking a stellar spectrum of a slowly rotating star of the same spectral type as NGTS-8 and artificially broadening it by different vsin i_* amounts (using a Gray rotational broadening profile). The projected rotational broadening of NGTS-8 was then measured using an optimal-subtraction technique in which the broadened template spectra were multiplied by a constant and then subtracted from the NGTS-8 spectrum. This is done after correcting for RV shifts and reinterpolating to a constant velocity scale. The value of the rotational broadening is then the one that minimizes the scatter in the residual spectrum after performing the optimal subtraction. For our template spectrum, we used α Cen B, which has a spectral type very close to NGTS-8 and a low rotation rate. We constructed the template spectrum by stacking archival HARPS spectra taken over one night when α Cen B was known to be inactive. The result of this analysis yielded a value consistent with that found by SPECIES.

Adopting the firm upper limit on the stellar rotation period from the vsin i_* measurement would ordinarily lead to a much higher |$\log R^{\prime }_{\mathrm{ HK}}$| level than the one observed. While no definitive answer can explain this difference, some systems hosting hot Jupiters are known to have suppressed Ca ii H & K re-emission (e.g. WASP-12; Fossati et al. 2013), leading to a lower measured value of the |$\log R^{\prime }_{\mathrm{ HK}}$|⁠. However, these systems generally contain hot Jupiters very close to filling their Roche lobes, which is not the case for our planet, NGTS-8b. We therefore conclude that the most likely explanation of the discrepancy is that we have caught NGTS-8 in an extended low-activity state.

5.3 Global modelling

Analysis of the different photometric and spectroscopic data was performed on NGTS-8 and NGTS-9 data using allesfitter (Günther & Daylan 2019, and in prep.). allesfitter is a user-friendly and publicly available software package for modelling data from photometric and RV instruments. Its generative model can account for multistar systems, stellar flares, star-spots, and multiple exoplanets. For this, it constructs an inference framework that unites the versatile packages ellc (light-curve and RV models; Maxted 2016), aflare (flare model; Davenport et al. 2014), dynesty (nested samplingl; Speagle 2019), emcee [Markov chain Monte Carlo (MCMC) sampling; Foreman-Mackey et al. 2013], and celerite (GP models; Foreman-Mackey et al. 2017). allesfitter is accesible at https://github.com/MNGuenther/allesfitter.

For NGTS-8 and NGTS-9, the nested sampling approach (see Skilling 2004) was used, which enables simultaneous fitting of the transit light curves and RV data. In particular, we fit for the following astrophysical parameters: a planet’s orbital period P, the transit epoch T_C, the radius ratio R_p/R_⋆, the sum of radii over the semimajor axis (R_p + R_⋆)/a, the cosine of the inclination cos i, the eccentricity and argument of periastron parametrized as |$\sqrt{e} \sin {\omega }$| and |$\sqrt{e} \cos {\omega }$|⁠, and the RV semi-amplitude K. For the transit light-curve modelling, a quadratic limb-darkening law was adopted parametrized after Kipping (2013) as u₁ and u₂. Systematic trends in the transit light curves were modelled by a Gaussian process with Matern 3/2 kernels parametrized by the GP’s amplitude ln ρ and time-scale ln σ. For both planets, all photometric data were used for the fits as well as all spectroscopic data, with instrumental offsets taken into account, relevant for NGTS-8 where HARPS and FEROS data were combined for the modelling.

We find that NGTS-8b has a mass of |$0.93\, ^{+0.04}_{-0.03}$| M_J and a radius 1.09 ± 0.03 R_J, while NGTS-9b has a mass of 2.90 ± 0.17 M_J and a radius 1.07 ± 0.06 R_J. The results of the fits for the two planets are summarized in Tables 8 and 9 and shown in earlier plots. Fig. 1 shows (in red) 20 light-curve models generated from randomly drawn posterior samples of the allesfitter fit to the NGTS, SAAO, and Euler light curves, respectively, for NGTS-8. In the same way, Fig. 2 shows the photometric data of NGTS, SAAO, and Euler, respectively, for NGTS-9, with (in red) 20 light-curve models generated from randomly drawn posterior samples of the allesfitter fit. For the RV data, Fig. 3 shows the modelling of HARPS, the blue points, and FEROS, the orange points, for NGTS-8 and Fig. 4 shows the modelling of the CORALIE data for NGTS-9.

In order to check our results, we also performed another analysis of the photometric data from NGTS and available spectroscopic data on NGTS-8 and NGTS-9 using the EXOplanet traNsits and rAdIal veLocity fittER (EXO-NAILER; Espinoza et al. 2016). Using the MCMC with a total of 250 walkers for 20 000 jumps and 5000 burn-in steps, the modelling was done assuming pure white noise for the inputted light curves. A logarithmic limb-darkening law was adopted with limb-darkening coefficients taken from Claret, Hauschildt & Witte (2013), and sampled according to Espinoza & Jordán (2015). The results from this second analysis all agreed, within the error bars, with those found from allesfitter.

5.4 TESS

During the preparation of this manuscript, TESS photometry was released for NGTS-9, which was observed in Sector 8. In response to this data release, we re-analysed, using all available data, NGTS-9 with allesfitter. The TESS photometric data area presented in Fig. 7 with the models generated from the fit. We confirmed that using TESS data in our modelling of NGTS-9 did not change or improve the values obtained, and thus the TESS data were not taken into account in the analysis presented in this work. The fact that TESS does not improve the results can be explained by the magnitude of NGTS-9, V = 12.80 ± 0.02. At these magnitudes, we have found that NGTS and TESS perform similarly (Wheatley et al. 2018). No TESS data are available for NGTS-8.

Figure 7.

TESS light curve of NGTS-9b with residuals. The blue data points are binned every 7 min to aid visualization. The red lines show 20 light-curve models generated from randomly drawn posterior samples of the allesfitter fit.

Open in new tabDownload slide

6 DISCUSSION

As outlined in the introduction, at incident fluxes greater than 2 × 10⁵ W m⁻² (Demory & Seager 2011; Miller & Fortney 2011), hot Jupiters are increasingly found with radii that are significantly larger than theoretically predicted (Anderson et al. 2011; Almenara et al. 2015; Delrez et al. 2016). Using Gaia DR2 measurements for the stellar luminosity and the orbital parameters listed in Tables 6 and 8 for NGTS-8b and listed in Tables 7 and 9 for NGTS-9b, we calculated the flux received by both planets to be greater than this limit (6.85 ± 0.45 × 10⁵ W m⁻² and 9.92 ± 1.09 × 10⁵ W m⁻² for NGTS-8b and NGTS-9b, respectively). Thus, the stellar irradiation levels received by both of these planets puts them firmly in the regime where we might expect them to exhibit larger than predicted planetary radii.

Sestovic et al. (2018) conducted a statistical investigation on hot-Jupiter radii and found that above a threshold in incident flux (2 × 10⁵ W m⁻²; Miller & Fortney 2011; Demory & Seager 2011), the observed radius follow the thermal evolution models (Miller & Fortney 2011; Thorngren et al. 2016) with the addition of an inflation parameter, ΔR. This observed radius ‘inflation’ is dependent on both the incident stellar flux and the mass of the planet. They proposed a flux-mass–radius relationship that has distinct forms for four different planetary mass regimes: below 0.37 M_J, between 0.37 and 0.98 M_J, between 0.98 and 2.50 M_J, and over 2.50 M_J. Using these relationships, we calculated what would be the expected radius inflation (ΔR) values for our two planets. For NGTS-8b, its mass lies on the edge of two regimes in Sestovic et al. (2018) (M_p < 0.98 M_J and M_p > 0.98 M_J), we thus determined predicted ΔRs from both relationships of 0.24 ± 0.02 and 0.02 ± 0.01 R_J, respectively. While the first value would suggest a highly inflated radius, the second value, however, suggests almost no inflation. Concerning NGTS-9b, the predicted radius inflation, ΔR, is 0.18 ± 0.01 R_J. Thus, from the work of Sestovic et al. (2018), these two planets would be expected to exhibit planetary radii larger than predicted.

We finally compared the observed planetary radii of NGTS-8b and NGTS-9b to the mass–radius models of Baraffe et al. (2008) and Fortney et al. (2007) who present, assuming a solar-type star, tables of planetary radii as a function of core mass, mass of the planet, orbital separation, and age of the system. Since neither of the host stars of the planets presented here are solar-like, we had to renormalize the orbital separation in order to keep the same incident flux. In this scenario, the distance from their host star would be equal to 0.044 and 0.038 au for NGTS-8b and NGTS-9b, respectively. As one can see in Tables 10 and 11, both models seem consistent and correctly predict the measured radius of NGTS-8b, 1.09 ± 0.03 R_J, and NGTS-9b, 1.07 ± 0.06 R_J, using the described parameters.

To conclude, even if both planets are in a regime where we expect planets to exhibit larger than predicted radii, our two planets are non-inflated hot Jupiters. This could be due to the planets being enriched with heavy elements, yielding a more compact structure and thus a smaller radius, like HD 149026b (Sato et al. 2005).

7 CONCLUSIONS

We have presented the latest discovery by the NGTS of two non-inflated hot Jupiters: NGTS-8b and NGTS-9b. NGTS, SAAO, and Euler photometric data and spectroscopic data from HARPS, FEROS, and CORALIE were used to confirm the detection of these two planets. By combining some of these data, an analysis of the transiting planets was performed using allesfitter and confirmed with EXO-NAILER. From this model, both planets have orbits consistent with being circular, as expected for such short period hot Jupiters. The characteristic of the planets were calculated such as NGTS-8b with a mass of |$0.93\, ^{+0.04}_{-0.03}$| M_J and a radius 1.09 ± 0.03 R_J, and NGTS-9b with a mass of 2.90 ± 0.17 M_J and a radius of 1.07 ± 0.06 R_J. Fig. 8 shows these discoveries in comparison to known planets with radius higher than 0.4 R_J.

Figure 8.

Shows NGTS-8b and NGTS-9b in the planetary radius versus planetary mass plot, in regards with all exoplanets from the NASA Exoplanet Archive with radius uncertainty below 10 per cent or mass uncertainty below 50 per cent and with an incident flux received by the planets greater than 2 × 10⁵ W m⁻². The background and the dotted black lines represents the point density per grid element.

Open in new tabDownload slide

A study of the rotational period of the K0V star, NGTS-8, was performed using different models and despite a significant discrepancy that we assume is due to an extended low activity of the star, we measured an upper limit of 13.92 ± 2.64 d. While its host is considerably fainter, our analysis also suggests that its planet, NGTS-8b, could have similar properties to HD 189733b, one of the best-studied hot Jupiters. Further observations of NGTS-8b will allow direct comparisons to be drawn between these two hot Jupiters. The upcoming launch of JWST will enable high-precision observations of NGTS-8b’s full-phase curve, which would be of particular interest due to the efficient dayside to nightside heat recirculation that HD 189733b exhibits relative to other hot Jupiters (Knutson et al. 2007; Schwartz et al. 2017).

Concerning NGTS-9b, the planet is highly irradiated, with an incident flux around 9.59 ± 0.74 × 10⁵ W m⁻², yet non-inflated. This radius could be due to the planet being extremely enriched with heavy elements, explaining its density, |$2.93^{+0.53}_{-0.49}$| g cm⁻³, one of the highest compared to planets with similar masses, as shown in Fig. 8.

SUPPORTING INFORMATION

Table 1. A sample of the photometric data of NGTS-8 from NGTS.

Table 2. A sample of the photometric data of NGTS-9 from NGTS, SAAO, and Eulercam.

Please note: Oxford University Press is not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

ACKNOWLEDGEMENTS

Based on data collected under the NGTS project at the ESO La Silla Paranal Observatory. The NGTS facility is operated by the consortium institutes with support from the UK Science and Technology Facilities Council (STFC) project ST/M001962/1. This paper uses observations made at the SAAO. We thank Marissa Kotze (SAAO) for developing the SHOC camera data reduction pipeline. The contributions at the University of Warwick by PJW and RGW have been supported by STFC through consolidated grants ST/L000733/1 and ST/P000495/1. Contributions at the University of Geneva by DB, FB, BC, LM, and SU were carried out within the framework of the National Centre for Competence in Research ‘PlanetS’ supported by the Swiss National Science Foundation. The contributions at the University of Leicester by MRG and MRB have been supported by STFC through consolidated grant ST/N000757/1. CAW acknowledges support from the STFC grant ST/P000312/1. EG gratefully acknowledges support from Winton Philanthropies in the form of a Winton Exoplanet Fellowship. JSJ acknowledges support by Fondecyt grant 1161218 and partial support by CATA-Basal (PB06, CONICYT). DRA gratefully acknowledges support from the STFC via an Ernest Rutherford Fellowship (ST/R00384X/1). PE and HR acknowledge the support of the DFG priority program SPP 1992 ‘Exploring the Diversity of Extrasolar Planets’ (RA 714/13-1). LDN acknowledges support from the Gruber Foundation Fellowship. MNG acknowledges support from MIT’s Kavli Institute as a Torres postdoctoral fellow. The research leading to these results has received funding from the European Research Council under the FP/2007-2013 ERC grant 336480 and from the ARC grant for Concerted Research Actions, financed by the Wallonia-Brussels Federation. This work was also partially supported by a grant from the Simons Foundation (PI Queloz, ID 327127). This work has used 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. pyraf is a product of the Space Telescope Science Institute, which is operated by AURA for NASA. This research has used 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.

Footnotes

REFERENCES