WASP-19b: THE SHORTEST PERIOD TRANSITING EXOPLANET YET DISCOVERED

Since the unexpected discovery of the first "hot Jupiter," 51 Peg b (Mayor & Queloz 1995), exoplanets with an exceptionally wide variety of properties have been detected which have dramatically changed our understanding of planetary physics. In particular, through the discovery of various transiting planets, we have learned that extrasolar planets can have radii much larger (e.g., Hebb et al. 2009) or densities much higher (Sato et al. 2005) than Jupiter. Many, but not all, "hot Jupiters" have temperature inversions in their atmospheres (e.g., Knutson et al. 2008), and they can have very low optical albedos (Rowe et al. 2008). Despite their short periods, not all transiting exoplanets have been tidally circularized (Gillon et al. 2009a), and both rocky (e.g., CoRoT-Exo-7, P ∼ 0.85 days) and gas giant (e.g., WASP-12b, P ∼ 1.09 days) planets can exist in extremely short period orbits. Here, we report on the discovery of a new extreme transiting extrasolar planet with the shortest orbital period yet detected which is on the verge of spiraling into its host star. This transiting planet not only can inform us about the properties and evolution of close-in planets, but also has the potential to provide information about the characteristics of its host star.

In this paper, we first describe all the observations that were obtained to detect and analyze the transiting star–planet system (Section 2). We describe the data analysis in Section 3 where we present the planet and its host star. Finally, in Section 4, we discuss the implications of the planet's short period and its future evolution.

2MASS J09534008-4539330 (hereafter WASP-19) is an apparently unremarkable 12th magnitude (V = 12.59), G8V star in the southern hemisphere located at α = 09:53:40.08, δ = −45:39:33.0 (J2000). The target was observed with the WASP-South telescope and instrumentation (Pollacco et al. 2006; Wilson et al. 2008) in the winter and spring observing seasons from 2006 to 2008. Between 2006 May 4–June 20, 1496 photometric data points were obtained, 6695 measurements were made between 2006 December 18–2007 May 18, and 8968 observations were taken from 2007 December 18–2008 May 22. All data sets were processed independently with the standard WASP data reduction pipeline and photometry package (Collier Cameron et al. 2006). The individual data points have typical uncertainties of ∼0.02 mag including Poisson noise and systematic noise. The resulting light curves were then run through our implementation of the box least squares algorithm (Kovács et al. 2002) designed to detect periodic transit-shaped dips in brightness.

The target was initially flagged as a transiting planet candidate because a strong periodic signal was detected in the 2007 data. The phase-folded light curve showed a square-shaped dip in brightness with a depth δ ∼ 25 mmag and duration τ ∼ 1.2 hr, consistent with a planet-sized object around a main-sequence star. Further, a periodic transit was also apparent in the 2006 data when phase folded with the 2007 ephemeris, and a transit was subsequently detected in the 2008 season of data. Therefore, we classified the object as needing follow-up photometry and spectroscopy to assess the planetary nature of the system. The phase-folded light curve containing all WASP-South data is shown in Figure 1.

Zoom In Zoom Out Reset image size

WASP-19 was observed photometrically with the 2 m Faulkes Telescope South (FTS) on 2008 December 17 during transit. Over 3.3 hr, 139 Pan-STARRS z-band¹¹ observations were made. The images were observed in 2 × 2 binning mode such that one binned pixel corresponds to 0279. They were processed in the standard way with IRAF¹² using a stacked bias image, dark frame, and sky flat. Minimal fringing was present in the z-band images due to the deep depletion CCD in the camera, so no fringe correction was applied. The DAOPHOT photometry package (Stetson 1987) was used to perform object detection and aperture photometry with an aperture size of eight binned pixels in radius. The 5' × 5' field of view of the instrument contained 53 comparison stars that were used in deriving the differential magnitudes with a photometric precision of 1.3 mmag. We measured the red noise (Pont et al. 2006) in the light curve on a 30 minute timescale to be 449 ppm and added this value in quadrature to the formal uncertainties on each data point. The resulting light curve is shown in Figure 2.

Zoom In Zoom Out Reset image size

Thirty-four radial velocity (RV) measurements were obtained with the CORALIE spectrograph on the 1.2 m Euler telescope (Baranne et al. 1996; Wilson et al. 2008). The stable, temperature-controlled, high-resolution echelle spectrograph has a resolution of R ∼ 55, 000 over the spectral region from 3800to6800 Å. The WASP-19 spectra, obtained between 2008 May 29 and 2009 April 23, were processed through a slightly updated version of the CORALIE data reduction pipeline. In addition to the standard pipeline described in Baranne et al. (1996), we corrected for the blaze function and scaled the fitted cross-correlation region to match the full-width half maximum of the object. The final RV values were obtained by cross-correlating the spectra with a G2 template mask. Table 1 presents the RV measurements of WASP-19 at each Barycentric Julian date, the 1σ Poisson errors on the velocities, and the line bisector span measurements (Gray 1988; Queloz et al. 2001). Based on our experience, we adopt uncertainties on the line bisector measurements of twice the measured RV errors.

The RV of the star varies sinusoidally with the same period measured from the photometry (see Figure 3). In addition, the line bisector spans, which are used to discriminate spot induced velocity variations and the effects of line-of-sight binarity, show no correlation with RV within the uncertainties. The slope of the bisector versus RV (Figure 4) is −0.006 ± 0.037, and the bisector measurements have an rms scatter of ∼50 m s⁻¹.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Although all existing photometric and spectroscopic data suggest WASP-19 is orbited by a short period transiting extrasolar planet, we explore the possibility of a false positive detection. In general, it is difficult to mimic the photometric and spectroscopic observations of a transiting planet without showing a visible second star in the spectrum, a significant bisector trend, and/or inconsistencies between the transit duration and host star spectral type. Starspots can cause periodic low-amplitude RV variations (e.g., Huélamo et al. 2008; Desort et al. 2007) but not photometric transits as well; thus, a single star blended with a fainter stellar eclipsing binary (EB) is the preferred false positive scenario for transiting planets. Unfortunately, no comprehensive simulations have been performed that model the expected RV variations, eclipse shapes, and bisector slopes for different blended EB scenarios, and performing such simulations is beyond the scope of this discovery paper. Instead, we explore the blended EB scenario through qualitative reasoning.

In order to produce a flat-bottomed, 2% transit, as seen for WASP-19, the flux ratio of the EB compared to WASP-19 would have to be small enough that the EB was undetectable as a peak in the cross-correlation function. However, it could not be so small that the necessary unblended eclipse depth would require nearly equal-sized EB components and therefore, a V-shaped eclipse (i.e., 0.05 < F_EB/F_W19 < 0.2). Although the eclipsing star would have to be small (and presumably less massive) compared to its primary to create the flat-bottomed eclipse, the RV amplitude of the visible EB primary would still be 40–90 km s⁻¹ for all reasonable mass ratios due to the high inclination angle needed to eclipse, the short orbital period of the observed transits, and the relatively deep transit (i.e., brown dwarf mass eclipsing objects would produce much shallower transit depths). Therefore, the RV variations of the EB could not be hidden within the WASP-19 spectral features given the resolution of the CORALIE data.

Furthermore, a short period stellar EB would almost certainly be tidally synchronized with vsin i ∼ 40–80 km s⁻¹. In the analysis of HD 41004 A (Santos et al. 2002), which exhibits planet-like RV variations due to a blended M dwarf + brown dwarf spectroscopic binary (SB), the bisector correlation is most dependent on the width of the visible SB component. This system, in which the SB is only 3% as bright as the single star and the width of the SB cross-correlation function is ∼8 km s⁻¹, shows a significant bisector correlation (slope of 0.67). According to their simulations, higher rotational broadening would produce an even greater bisector slope. Therefore, we would expect WASP-19 to show a significant bisector trend if the RV signal was caused by a rapidly rotating, blended EB, rather than a transiting planet.

Finally, we see no difference in the eclipse depth of the z-band FTS transit compared with the SuperWASP transit taken in a bluer, V+R filter. Although this is not a strong constraint given the scatter in the SuperWASP photometry, an eclipse by a stellar object could show eclipse depth variations in different filters which we do not see. In summary, we conclude that the existing photometric and RV variations of WASP-19 are most likely due to the presence of a transiting extrasolar planet.

A combined analysis of the RV curve and light curve of a transiting planet host star will provide direct measurements of the mass and radius of its orbiting planet with one additional constraint, e.g., the stellar mass. Below, we describe the determination of the stellar mass and other host star properties.

3.1. Spectroscopic Parameters

3.2. Host Star Mass

By comparing the effective temperature, metallicity, and mean stellar density (ρ_*) of WASP-19 to theoretical stellar models, we determine its mass. The stellar density is dependent on the shape of the transit and largely independent of any assumptions or models (Seager & Mallén-Ornelas 2003). Thus, after deriving the spectroscopic parameters, we model the WASP-South and FTS transit light curves of WASP-19 using the Markov-chain Monte Carlo (MCMC) routined described in Collier Cameron et al. (2007) to derive the stellar density and its uncertainty. The code uses the MCMC approach to simultaneously solve for the orbital and physical properties of the star-planet system. We apply the limb darkening coefficients of Claret (2000, 2004) for the appropriate temperature of the star and wavelength of the light curves. We note that the eccentricity value has an effect on the stellar density determination, thus we first allow the eccentricity to be a free parameter. Since the resulting eccentricity value gives a non-significant 1.5σ detection, we solve for the stellar density a second time while fixing the eccentricity to be zero.

Interpolating among the Girardi et al. (2000) stellar evolution tracks as described in Hebb et al. (2009), we derive a mass for the host star of M_* = 0.95^+0.09_−0.10 M_☉ using the circular orbit solution. The non-circular orbit value for the stellar density results in a mass of 0.96 M_☉, well within the existing error bars (adopted from the non-circular solution). Figure 5 shows a plot of the position of WASP-19 in a modified Hertzprung–Russell diagram as compared to the theoretical tracks. The errors on the stellar mass are dominated by the uncertainty on the metallicity. Here, we take into account uncertainties on the stellar temperature, metallicity, and density. We do not include systematic uncertainties on the chosen stellar model which are difficult to determine, but we might expect this additional uncertainty to be at least ∼4%. Southworth (2009) finds that with stellar parameters measured at the limit of our technological and theoretical ability for HD 209458b, the variation in the mass determination using four different stellar models is ∼4%.

Zoom In Zoom Out Reset image size

3.3. Host Star Age

3.4. Host Star Rotation Period

We search for variability in the WASP-South light curves of WASP-19 caused by asymmetric starspots on the photosphere which modulate the flux. Any starspots will rotate in and out of view with the stellar surface, such that the period of the variability gives the rotation period of the star. To exhibit rotational variability, the star must have a sufficient coverage of starspots to create a variable brightness signal which is detectable given the scatter in the photometric data. Furthermore, star spots evolve on timescales of weeks or months causing the amplitude and phase of the variability to change; therefore, each season of WASP-South data was examined independently.

To detect the rotational variability, we determine the improvement in χ² over a flat, non-variable model when a sine wave of the form y = a₀ + a₁sin(ωt + a₃) is fitted to each season of the WASP-South data phase folded at a set of trial periods, P_rot = 2π/ω. We subtract all transits from the light curves using the model derived from the parameters in Table 3 before fitting the sine curve model. We test periods between 0.2and50 days and find a strong periodic signal in the 2007 data with P_rot = 10.5 days and amplitude, a₁ = 7.6 mmag. The phase-folded light curve and periodogram of normalized Δχ² values are shown in Figures 6 and 7 (top), respectively. In the 2008 data, we also detect a weaker signal with a similar period, P_rot = 10.6 days, and with an amplitude of 3.6 mmag.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Table 3. WASP-19 System Parameters and 1σ Error Limits Derived from the MCMC Analysis

Parameter	Symbol	Value (e fixed)	Value (e free)	Units
Transit epoch (BJD)	T0	2454775.3372+0.0001−0.0002	2454775.3372+0.0002−0.0002	days
Orbital period	P	0.7888399+0.0000008−0.0000008	0.7888399+0.0000008−0.0000008	days
Planet/star area ratio	(Rp/Rs)2	0.0203+0.0004−0.0004	0.0203+0.0004−0.0004
Transit duration	tT	0.0642+0.0006−0.0006	0.0643+0.0006−0.0007	days
Impact parameter	b	0.62+0.03−0.03	0.62+0.03−0.03	R*
Stellar reflex velocity	K1	0.256+0.005−0.005	0.256+0.005−0.005	km s−1
Center-of-mass velocity	γ	20.78534+0.0002−0.0002	20.78535+0.0003−0.0003	km s−1
Orbital semimajor axis	a	0.0165+0.0005−0.0006	0.0164+0.0005−0.0006	AU
Orbital inclination	I	80.5+0.7−0.7	80.8+0.8−0.8	degrees
Orbital eccentricity	e	0 (fixed)	0.02+0.02−0.01
Longitude of periastron	ω	0 (fixed)	−76+112−23	deg
Eccentricity × cos(ω)	ecos ω	0 (fixed)	0.004+0.009−0.009
Eccentricity × sin(ω)	esin ω	0 (fixed)	−0.02+0.02−0.02
Stellar mass	M*	0.96+0.09−0.10	0.95+0.10−0.10	M☉
Stellar radius	R*	0.94+0.04−0.04	0.93+0.05−0.04	R☉
Stellar surface gravity	log g*	4.47+0.03−0.03	4.48+0.03−0.03	[cgs]
Stellar density	ρ*	1.13+0.09−0.09	1.19+0.12−0.11	ρ☉
Planet radius	Rp	1.31+0.06−0.06	1.28+0.07−0.07	RJ
Planet mass	Mp	1.15+0.08−0.08	1.14+0.07−0.07	MJ
Planetary surface gravity	log gp	3.19+0.03−0.03	3.20+0.03−0.03	[cgs]
Planet density	ρp	0.51+0.06−0.05	0.54+0.07−0.06	ρJ
Planet temperature (A = 0, F = 1)	Teq	2009+26−26	1993+32−33	K

Download table as:  ASCIITypeset image

To assess the veracity of the sinusoidal signal detected in the 2007 data, we determine the significance and the false alarm probability (FAP) following Zechmeister & Kürster (2009, employing the residual variance normalization). The equations include, N, the number of independent data points, and M, the number of independent frequencies, and the measured peak value in the periodogram. According to Cumming (2004), the number of independent frequencies can be approximated by the duration of the time-series data times the difference between the highest and lowest frequencies tested (here M = 753). The number of independent data points depends on the level of red noise in the light curve. We empirically determine N by generating a new light curve in which the red noise is preserved, but the periodic signal is destroyed. We randomly re-order the individual nights of data, so that the integer part of the light curve time values are shuffled, but the fractional parts (containing the red noise) remain the same. We then run the sine fitting program on the shuffled light curve and find N = 664 by assuming the highest peak in the resulting periodogram (Figure 7 (bottom)) has a 95% probability of being false (FAP = 0.95). Using our calculated N and M, we then apply the equations in Zechmeister & Kürster (2009) to the results of the sine fitting on the original 2007 light curve and find a highly significant periodic signal with a Prob(p>p_best) = 1.4 × 10⁻¹⁰ and FAP = 1 × 10⁻⁷. Thus, we adopt a rotation period for WASP-19 of P_rot = 10.5 days.

We calculate the error on this measurement using the formula in Horne & Baliunas (1986) and find σ_P = 0.08, but given the variation in period values we measure using two other techniques (Lomb–Scargle and auto-correlation), we find this to be underestimated and suggest σ_P = 0.2 days is a more realistic uncertainty. Finally, we note that the 10.5 day rotation period of the star measured via the photometry is consistent with the less precise v sin i value of 4 ± 2 km s⁻¹, which corresponds to a rotational period of between 8 and 24 days (assuming the spin axis of the star is perfectly aligned with the orbital axis).

The rotation period of a main-sequence star can also be used to estimate its age. According to the Barnes (2007) gyrochronology relationship, a rotation period of P_rot = 10.5 days corresponds to an age of 500–600 Myr for a G8V star like WASP-19 with B − V = 0.74. This value is not consistent with the older age inferred from the isochrones, lithium, and space motion. In Section 4, we discuss the implications of the possible discrepancy between the different age indicators.

3.5. Planet Properties

3.6. Transit Timing

The most striking aspect of WASP-19b is its extremely short orbital period. With a period, P = 0.7888399 ± 0.0000008 days, WASP-19b is the shortest period planet yet discovered. What physical processes in the evolution of the planet have led to such a close separation? Did the initial migration process leave the planet in its extremely short period orbit or has there been subsequent evolution of the orbital separation? The observations presented here suggest the possibility that WASP-19b has been spiraling into its host star throughout its lifetime and has spun up its host star in the process.

Most transiting extrasolar planets will ultimately spiral into their host stars because there is insufficient total angular momentum in the systems to reach a state of tidal equilibrium (Hut 1980; Rasio et al. 1996; Jackson et al. 2009). However, the timescale for this evolution is not well known because the stellar tidal quality factor, Q'_s, is uncertain to at least three orders of magnitude. Values between 10⁶ < Q'_s < 10⁹ are reasonable (Jackson et al. 2008, and references therein). Furthermore, the planetary tidal quality factor, Q'_p is equally uncertain, but it affects the evolution of the stellar spin and planetary orbital separation to a much lesser degree.

For WASP-19b, the ratio of total to critical angular momentum is well below the limiting value of one, thus ensuring the planet will ultimately collide with its host star, but the lifetime of this evolution ranges from 4 Myr to 4 Gyr depending on the value used for Q'_s. However, unlike most other transiting planets, we have measured the rotation period of WASP-19 which contributes additional information that can be used to place tighter constraints on the overall lifetime of the planet and on the tidal quality factor of the star.

By integrating the coupled equations of orbital eccentricity and separation (Dobbs-Dixon et al. 2004) while conserving the total angular momentum of the system, we can model the future evolution of the planet as it spirals inward and further spins up the host star. In the analysis, we include magnetic braking which takes the form $\dot{\Omega } = -\kappa \Omega ^3$ with κ = 3.88 × 10⁻⁷ s as evaluated using Equation (12) of Collier Cameron & Jianke (1994). We use the current orbital separation derived from the MCMC analysis (Table 3) and the measured rotation period (P_rot = 10.5 days) as initial conditions. Figure 8 shows the future evolution of the planet in orbital separation, and in Figure 9, we plot the corresponding rotational evolution of the star.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The results show that in the Q'_s = 10⁹ case, the tidal interaction is so weak that the star's future evolution is dominated by magnetic braking. In this situation, the relatively short rotation period could not have been caused by prior tidal interactions with the planet and would instead suggest a young stellar age similar to that of the Hyades (∼ 600 Myr). Although it is possible WASP-19 is as young as the Hyades, the isochrones analysis, the non-detection of lithium, and the stellar velocity all favor an older age for the star. Therefore, a lower value for Q'_s is preferred.

For all models with Q'_s < 10⁸, the star's spin period has already passed through a maximum and is decreasing as the planet spirals in. However, in the Q'_s = 10⁶ case, the remaining lifetime of the planet is extremely short (4 Myr). Therefore, the existing data appears to favor values of Q'_s = 10⁷–10⁸ which are high enough to give significant life expectancy, but low enough to have reversed magnetic braking and initiated spiral-in. This scenario would allow for the extremely short period of the planet to be a consequence of further evolution in orbital separation after the initial migration process early in its formation and evolution. However, the Q'_s would have to be 1–2 orders of magnitude greater than the nominal value of 10⁶ that is typically adopted (e.g., Bodenheimer et al. 2003; Jackson et al. 2008).

It is important to note that Q'_s is a simple parameterization of the complex physics involving the interaction between the tides raised by the planet on its star and the turbulent viscosity in the stellar convection zone and inertial wave modes excited in the stellar interior (Rasio et al. 1996; Sasselov 2003; Ogilvie & Lin 2007). A more detailed analysis of this system and others like it (e.g., WASP-18, OGLE-TR-56b, WASP-12) will hopefully lead to a better understanding of the physics modulating tidal dissipation in stars, since currently, there is no comprehensive theory that is able to explain observations of both main-sequence binary stars and close-in extrasolar planets with regard to their tidal evolution (Rasio et al. 1996; Sasselov 2003; Terquem 1998; Ogilvie & Lin 2007).

Finally, we note that the stellar flux incident on WASP-19b at the substellar point is 3.64 × 10⁹ erg cm⁻² s⁻¹ placing it in the class of highly irradiated planets like OGLE-TR-56b (Konacki et al. 2003; Torres et al. 2004) and WASP-1b (Cameron et al. 2007). Therefore, we expect the planet to have large secondary eclipse depths in the mid-IR, evidence of an atmospheric temperature inversion, molecular emission features, and a large day/night contrast (Fortney et al. 2008). The existence of a hot stratosphere can be tested using secondary eclipse measurements that are currently being obtained with Spitzer. Eclipse measurements in the near-IR and optical z-band are also possible given current technology (e.g., Gillon et al. 2009b). Furthermore, the density of WASP-19b is half that of Jupiter's (ρ_p = 0.51ρ_J), so it is slightly bloated for its mass, but not extremely so. The high irradiation, increased metallicity of the host star, and dissipation of tidal energy are possible factors causing the enhanced radius (Bodenheimer et al. 2003; Fortney et al. 2007; Burrows et al. 2007).

In summary, WASP-19b is the shortest period transiting planet yet detected. It has a mass M_pl = 1.15 M_J and radius R_pl = 1.31 R_J. The planet orbits a main-sequence G-dwarf with a slightly super-solar metallicity and a rotation period of P_rot = 10.5 ±  0.2 days. It is likely WASP-19b has been spiraling into its host star over its lifetime and has spun up the star in the process. A more precise age determination of WASP-19 will allow us to confirm this and to place stronger constraints on the star's tidal quality factor, Q'_s.

The SuperWASP Consortium consists of astronomers primarily from the Queen's University Belfast, St. Andrews, Keele, Leicester, The Open University, Isaac Newton Group La Palma and Instituto de Astrofísica de Canarias. The SuperWASP Cameras were constructed and operated with funds made available from Consortium Universities and the UK's Science and Technology Facilities Council.