WASP-37b: A 1.8 M_J EXOPLANET TRANSITING A METAL-POOR STAR

WASP-37 is a V = 12.7 mag star in the constellation Virgo, located at α_J2000 = 14^h47^m4657, δ_J2000 = +01°03'539 (GSC 00326-00658; 2MASS 14474655+0103538). It was observed by SuperWASP (La Palma) between March and June in 2008 and 2009, and by WASP-S (South Africa) during 2008 June to July and 2009 March to July. A total of 22,593 photometric data points were obtained during these intervals. The pipeline-processed data were de-trended and searched for transits using the methods described in Cameron et al. (2006). A periodic, transit-like signature was detected independently at the same period (3.58 d) in multiple cameras and in successive seasons, strongly suggesting that the transits were real, despite the faint magnitude. The folded light curve is shown in Figure 1 with the best-fit model.

Zoom In Zoom Out Reset image size

The star underwent several consistency tests aimed at eliminating false positives. It is isolated within the WASP photometric aperture in the Digitized Sky Survey (DSS) image, ruling out resolved blends. We also checked for signatures of unresolved blends (see Cameron et al. 2007): no significant ellipsoidal variability was measurable in the WASP light curve folded on the transit period, nor any signs of a secondary eclipse. Moreover, the transit depth and duration yielded a planet-like radius for the companion, and a stellar density appropriate to a main-sequence host star of the effective temperature derived from the 2MASS colors. Having passed all these tests, the star was selected for follow-up observations.

We obtained two further transit light curves of WASP-37 in order to refine the photometric parameters, and they are shown in Figure 2. All photometric data presented in this paper are available from the NStED database.¹⁵

Zoom In Zoom Out Reset image size

A partial transit was observed with LT/RISE on 2010 May 21. The full transit was not obtainable as the star set below the observing limits of the LT prior to egress. RISE is a frame transfer CCD located on the robotic 2.0 m Liverpool Telescope (LT) on La Palma with a broadband V + R filter (Steele et al. 2008; Gibson et al. 2008). The telescope was defocused by −1.0 mm to give FWHM = 17 pixels = 9.2 arcsec, and the CCD was used in 2×2 binning mode with an exposure time of 65 s and effectively no dead time. This allowed 215 images to be taken over the 3.9 hr period, including 1 hr of observations before transit. The data were reduced using the ULTRACAM pipeline (Dhillon et al. 2007) and differential photometry was performed relative to five nearby bright stars using an 18 pixel radius aperture.

A further partial transit was obtained on 2010 June 30 with the Spectral¹⁶ camera (FS01) on the LCOGT 2.0 m Faulkes Telescope South (FTS, Siding Spring, Australia). The Pan-STARRS-Z filter was used with the instrument in binning 2×2 mode, giving 0.303 arcsec/pixel, and no defocusing. The exposure time was 45 s, and 169 images were taken in the 3.5 hr period. Data were reduced using a pipeline¹⁷ written at the Liverpool John Moores University, then differential photometry performed using the IRAF/DAOPHOT¹⁸ package. Eight comparison stars were used with an 8 pixel radius aperture. Points with error bars larger than 0.01 mag were removed from the second half of the data set. These were likely caused by passing cloud, and the observation was stopped prematurely due to bad weather.

In order to establish the planetary nature and determine the stellar parameters, we obtained follow-up spectroscopic observations. Five spectra were taken between 2010 April 17 and May 28 with the stabilized echelle spectrograph SOPHIE at the 1.93 m telescope of Observatoire de Haute-Provence (Perruchot et al. 2008; Bouchy et al. 2009). The observations were all made with a signal-to-noise ratio (S/N) of ∼20 in order to minimize the charge transfer inefficiency (CTI) effect (Bouchy et al. 2009). Two 3 arcsec diameter optical fibers were used: the first centered on the target and the second on the sky to simultaneously measure the background to remove contamination from scattered moonlight. A further eight spectra were obtained with the CORALIE Fiber-Fed Echelle Spectrograph on the Swiss 1.2 m telescope at ESO-La Silla, Chile, between 2010 April 19 and June 29 with S/N ∼ 10–20 in dark/gray time to minimize contamination from scattered moonlight. The data were processed using the standard pipeline (Baranne et al. 1996; Pepe et al. 2002).

The radial velocities (RVs) and line bisector spans (V_span; see Toner & Gray 1988) derived from cross correlation are shown in Table 1 and plotted with the best-fit model in Figure 3. No significant correlation is seen between the bisector span and RV, which supports the signal's origin as a planetary companion rather than a blended eclipsing binary system (Queloz et al. 2001). A Spearman rank–order correlation test also indicates that any correlation between the RVs and bisector is weak, with a probability of correlation of 0.12.

Zoom In Zoom Out Reset image size

We used several techniques to infer the fundamental stellar properties of WASP-37 and the results are shown in Table 2. We first performed a spectral analysis using the methods given in Gillon et al. (2009). The individual CORALIE and SOPHIE spectra from the standard reduction pipeline were co-added to produce a single spectrum with an average S/N of around 60:1. The Hα line was used to determine the effective temperature (T_eff), while the Na i D and Mg i b lines were used as surface gravity (log g) diagnostics. This yielded the values T_eff = 5800 ± 150 K and log g = 4.25 ± 0.15. The effective temperature indicates that the star is of spectral type G2 (Gray 2008).

The value for microturbulence (v_mic) was determined from Fe i using the Magain (1984) method. We assumed a value for macroturbulence (v_mac) of 3.6 ± 0.3 km s⁻¹  based on the tabulation by Gray (2008), and an instrumental FWHM of 0.11 ± 0.01 Å, determined from the telluric lines around 6300 Å. The projected stellar rotation velocity (vsin i) was determined by fitting the profiles of several unblended Fe i lines, giving a best-fitting value of vsin i = 2.4 ± 1.6 km s⁻¹.

Elemental abundances were determined from equivalent-width measurements of several clean and unblended lines. The quoted errors include the uncertainties in T_eff, log g, and v_mic, as well as the scatter due to measurement and atomic data uncertainties. We obtain [Fe/H] = −0.4 ± 0.12 and log A(Li) = 2.23 ± 0.13. Interstellar Na D lines are present in the spectra with equivalent widths of ∼100–150 mÅ, indicating an extinction of E(B − V) = 0.05 using the calibrations of Munari & Zwitter (1997).

Transit light curves provide a measure of the mean stellar density (ρ_⋆) which can be used as a luminosity indicator for stellar evolutionary models (Sozzetti et al. 2007). We used a Markov Chain Monte Carlo (MCMC) approach to globally model the photometric and RV data (see Section 3.2) and obtained ρ_⋆ = 0.931^+0.064_−0.099 ρ_☉. Values of T_eff, [Fe/H], and ρ_⋆ were compared with the theoretical stellar evolutionary models of Yi et al. (2001) to obtain the following stellar properties: M_⋆ = 0.849^+0.067_−0.040 M_☉, R_⋆ = 0.977^+0.045_−0.042 R_☉, L_* = 0.953^+0.184_−0.114 L_☉, log g_iso = 4.39^+0.02_−0.03, and an age of 11⁺³₋₄ Gyr. The model isochrones are shown in Figure 4.

Zoom In Zoom Out Reset image size

We find that the measurement of log g from evolutionary models, log g_iso = 4.39^+0.02_−0.03, is consistent with that found from the spectroscopic analysis, log g = 4.25 ± 0.15. We re-fitted the isochrones using the 1σ upper and lower limits of [Fe/H] and found that this affected the age and mass determinations somewhat, increasing the uncertainties in the age (8–13 Gyr) and stellar mass (0.806–0.914 M_☉).

We computed the stellar distance using the following two methods. First, the absolute bolometric magnitude (calculated from the stellar luminosity) was compared to the extinction-corrected V-band magnitude from TASS (m_v = 12.7; Droege et al. 2006). The bolometric correction was taken as the range of values for a G2-type star given in Cox (2000) and Gray (2008). This yielded a distance D = 338 ± 41 pc. Second, we compared the angular diameter Θ = 2R_⋆/D = 0.025 ± 0.005 mas derived from the flux fitting method (see Section 3.1.1) to the stellar radius, yielding a distance D = 360 ± 75 pc. It is reassuring that these values are similar, given that both methods rely on parameters from the stellar models and are therefore not independent. We take the distance to be the weighted mean of the two values, D = 343 ± 36 pc.

3.1.1. Independent Checks

To determine the properties of the planet, we simultaneously modeled the SuperWASP, LT, and FTN light curves and the CORALIE and SOPHIE RVs with a global MCMC fit. Details of this process are described in Cameron et al. (2007) and Pollacco et al. (2008). The free parameters in the fit are orbital period P, transit epoch T₀, transit duration T_dur, ratio of planet to star radius (R_p/R_*)², impact parameter b, RV semi-amplitude K, Lagrangian elements ecos ω and esin ω where e is the eccentricity and ω is the longitude of periastron, and the systematic offset velocity γ. In this particular case, two systematic velocities were fitted to allow for instrumental offsets between the SOPHIE and CORALIE data sets.

We propagated the uncertainty in the stellar mass through to the derived parameters (M_⋆ = 0.925 ± 0.120 M_☉; see Section 3.1.1). We used the four Claret (2000, 2004) nonlinear limb darkening coefficients for the appropriate stellar temperature and photometric passband appropriate for each light curve; see Table 5. The limb darkening coefficients are re-calculated at each step in the MCMC chain to take into account the uncertainties in the stellar temperature and radius.

The photometric error bars were scaled to account for any underestimations so that the best-fitting model results in χ²_red = χ²/dof = 1 (dof = number of points − number of fitted parameters). The RV error bars did not require rescaling with a jitter term in order to obtain χ²_red = 1. The rms of the residuals from the best-fit model are rms_WASP = 0.0292 mag, rms_LT = 0.0013 mag, rms_FTS = 0.0025 mag, rms_SOPHIE = 0.0194 km s⁻¹, and rms_CORALIE = 0.0380 km s⁻¹. By removing the outlying point at 8.064 km s⁻¹, we obtain rms_CORALIE = 0.0164 km s⁻¹.

An initial fit was performed in which the eccentricity was allowed to float, yielding e = 0.036 ± 0.021. From the Lucy–Sweeney test (Lucy & Sweeney 1971), we find it is only significant at the 1σ level, and hence a circular orbit was adopted (χ²_circ = 18.3, χ²_ecc = 14.0 with 13 RVs and 5 free parameters: K, γ_CORALIE, γ_SOPHIE, ecos ω, and esin ω). We determined an upper limit for the eccentricity as e_upper = e + 3σ = 0.078. We also fitted a linear trend in the RVs to search for a third body in the system, but found that the decrease in χ² did not warrant the extra free parameter. From the F-test, we found that the significance of the trend was less than 1σ (χ² = 18.3, χ²_trend = 15.4 with 13 RVs and 4 free parameters: K, γ_CORALIE, γ_SOPHIE, and linear trend $\dot{\gamma }$).

We performed the final MCMC fit with a chain length of 20,000 points and the resulting best-fit parameters and uncertainties are shown in Table 4. WASP-37b has P = 3.577469 ± 0.000011 d, M_p = 1.80 ± 0.17 M_J, R_p = 1.16^+0.07_−0.06 R_J, and ρ_p = 1.15^+0.12_−0.15 ρ_J.