Planets from the WTS/RoPACS projects

Transcription

1 Planets from WTS/RoPACS projects Roberto Saglia MPE/LMU Cappetta, Koppenhöfer, Steele, Zendejas (MPE, Germany), Pinfield (PI), Barnes, Campbell, Catalan, Goulding, Jones, Marocco, Napiwotzki, Sipocz (CAR, Herts, UK), Hodgkin (PI), Kovacs, Mislis (IoA, Cambridge, UK), Barrado, Bajo, Cruz, Galvez Ortiz, Sarro Baro, Solano, Stoev (CAB, Madrid, Spain), Martin, Lodieu, Murgas, Palle, Tata (IAC, Tenerife, Spain), Pavlenko, Ivanyuk, Kuznetsov (MAO, Ukraine), Birkby, Nefs, Snellen, (Leiden Observatory, The Nerlands), del Burgo (INAOE, Mexico), Fossati, Haswell (Open University, UK), Pollacco (Queens University, UK), WTS consortium, and members of EC funded RoPACS INT Planet formation and evolution 2012, Munich

2 Outline Introduction: The WFCAM Transit Survey and RoPACS Discoveries: giant planets WTS-1 b and WTS-2 b Discussion: upper limits on fraction of M- dwarfs with giant planets Conclusions

3 Hot Jupiters around M-Dwarfs? In core accretion paradigm, gas giants are formed by accumulating gas on solid core from disk. This does not work easily around low mass stars because of time scale differences (Laughlin et al. 2004, Ida & Lin 2005). The protoplanetary disk dissipates before runaway gas accretion phase is reached. The predicted frequency of giant planets should decrease towards lower primary masses. Neptunes and rocky planets should be common around low mass stars. Gravitational instability models can produce gas giants quickly (if protoplanetary disk is massive enough to become unstable, Boss 2006) Not many planets around M-dwarfs known: Johnson et al. (2007) find 3 Jupiters around 169 M dwarfs using RVs (1.8%). Two transiting planets found around M dwarfs from ground (Gillon e tal. 2007, Charbonneau et al. 2009). The Kepler Mission identifies 9 high-quality candidates around 1086 cool stars (3600K<Teff<4100), Howard et al search for transits in NIR sensitive to rocky planets in habitable zone of M-dwarfs

4 The WFCAM Transit Survey (WTS) Kovacs et al. (2012, in prep.) M-dwarf ongoing variability survey (started in August 2007) designed primarily to: i) tightly constrain M-dwarf hot Jupiter fraction ii) calibrate low-mass stellar evolution models with M-dwarf eclipsing binaries UKIRT+WFCAM wide-field view of four fields (4x1.5 = 6 sq. deg) 1% Infrared J-band (1.25μm) near peak of M-dwarf SED Poor seeing/sky back-up program - efficient but nonstandard observing pattern, one field always visible 3 mmag M-dwarfs WTS 17h FGK stars 1% 3R / 0.3R super-earth + late M-dwarf 10% 2RJ / 0.6R hot Jupiter + early M-dwarf WTS 19h 1% 1RJ / 1R Jupiter + Sun U/CMP/2

5 Light curves and analysis CASU Data reduction pipeline Aperture photometry (and difference imaging) light curves Box-fitting search for transits More than 1000 epochs in 19h field Multi-band (ugrizyjhk) spectral typing i-band light curve for interesting candidates low-resolution spectroscopy for better typing and detection of high amplitude radial velocity variations (binaries) high-resolution (HET) spectroscopy for precise radial velocity measurements

6 Rocky Planets Around Cool Stars (RoPACS) RoPACS is a Marie Curie Initial Training Network funded by European Community. It involves 6 nodes:opa University of Hertfordshire (UH), Hertfordshire - The coordinating node (PI D. Pinfield) Institute of Astronomy (IoA), Cambridge Institute de Astrofísica de Canarias (IAC), Tenerife Max-Planck Institut für Extraterrestriche Physik (MPE), Munich Laboratory of Stellar Astronomy and Exoplanets (LAEX-CAB), Madrid Main Astronomical Observatory (MAO), Kiev + Astrium RoPACS supports 12 PhD students and 4 PostDocs. The project will end in November 2012 with a Conference on: Hot planets and Cool Stars MPE,(Garching) November 2012 Nefs et al. 2012, Four ultra-short period eclipsing M-dwarf binaries in WFCAM Transit Survey, MNRAS, 425, 950 Birkby et al. 2012, Discovery and characterisation of detached M-dwarf eclipsing binaries in WFCAM Transit Survey, MNRAS, in press, astro-ph/ Cappetta et al. 2012, The first planet detected in WTS, an inflated hot-jupiter in a 3.35 days orbit around a late F-star MNRAS, in press, Kovacs et al. 2012, A sensitivity analysis of WFCAM Transit Survey for short-period giant planets around M dwarfs MNRAS, submitted Birkby et al. 2012, WTS-2b: an inflated hot Jupiter in a tight orbit around a K3V star, in prep.

7 Planetary parameters Transit fit 20 The light curves in J- and i! -band were fitted with ana21 lytic models presented by Mandel & Agol (2002). We used 22 quadratic limb-darkening coefficients for a star with ef23 fective temperature Teff =6250 K, surface gravity log g= and metallicity [Fe/H]=-0.5 dex, calculated as linear inter25 polations in Teff, log g and [Fe/H] of values tabulated 26 in Claret & Bloemen (2011). We use ir table derived 27 from ATLAS atmospheric models using flux conserva28 tion method (FCM) which gave a slightly better fit than 29 ones derived using least-squares method (LSM). The val30 ues of limb-darkening coefficients we used in our fitting 31 are given in Table 6. Scaling factors were applied to error 32 values of J- and i! -band light curves (0.94 and 0.9 re33 spectively) in order to achieve a reduced χ2 of constant 34 out-of-transit part equal to Using a simultaneous fit to both light curves, we fitted period P, time of central transit t0, radius 36 ratio Rp /Rs, mean stellar density, ρs = Ms /Rs3 in solar 37 units and impact parameter βimpact in units of Rs. The 38 light curves and model fit are shown in Figures 6 and 7, 39 while resulting parameters are listed in Table The errors were calculated using a multi-dimensional 41 grid on which we search for extreme grid points with χ2 =1 42 when varying one parameter and simultaneously minimizing 43 over ors. Figure 8 shows correlations between 44 parameters of WTS-1 system derived from simulta45 neous fit of J- and i! -band light curves. Considering 46 solar metallicity scenario, with different limb-darkening co47 efficients, change in final fitting parameters is smaller 48 than 1% of ir uncertainties. Note that combined fit as49 sumes a fixed radius ratio although in a hydrogen-rich atmo50 sphere, molecular absorption and scattering processes could 51 result in different radius ratios in each band (an attempt to detect such variations has recently been undertaken by de 52 Mooij et al. 2012). In our case, uncertainties are too large 53 to see this effect in light curves and assumption of 54 a fixed radius ratio is a good approximation. The estimated 55 stellar density of host star ( ρsun ) is consistent 56 with expected value based on its spectral type (Seager 57 & Mallen-Ornelas 2003). The noise in data did not allow 58 a secondary transit detection. 59 we searched furr signals in Figure WFCAMSubsequently, J-band light curve for data of periodic WTS-1. U pper Detections in WTS: WTS-1b Page 7 of M. Cappetta et al. 4 5 iraf.echelle package (Willmarth & Barnes 1994). 6 After standard calibration of raw science frame, 7 bias-subtraction and flat-fielding, stellar spectra were 8 extracted order by order and related sky spectrum 9 Figure 6. WFCAM light curve data ofwere WTS-1. U pper was subtracted.j-band The extracted ThAr spectra used to 10 panel: whole set of folded WFCAM J-band data points. compute dispersion functions, which are characterized 11iddle panel: M folded photometric data centred in between transit by an RMS of order of A. Consistency and 12 two (red solutions from ThArwith taken before best model line)(computed fitted in combination INTand i! -band 13 after science exposure) was checked for all nights data. Lower panel: residuals of best fit. 14 in order to detect possible drifts or any or technical 15 hitch that could take place during each run. Successively, 16 spectra were wavelength calibrated using a linear 17 interpolation of two dispersion functions. 18 In subsequent data analysis, custom Matlab programs were used. After continuum estimation and 19 following normalization, spectra were filtered to remove 20 residual cosmic rays peaks left after previous filtering 21 performed on raw science frame with iraf task cos22 micrays. Comparing spectra of all nights for each 23 single order, pixels with higher flux, due to a cosmic hit Figure 2. Broad band photometric data of WTS-1: SDSS (filled 24 dots), Johnson (empty dots), WFCAM (squares), 2MASS (trianon detector, were detected and masked. The telluric ab25 eff(stars). The best fitting template (black line) is gles) and WISE sorption lines present in redder orders were n removed 26 ATLAS9 Kurucz Teff =6500 K, log g=4.5, [Fe/H]=-0.5 model in order to avoid contaminations using a high SNR observed 27 with AV =0.44. Vertical errorbars correspond to flux uncerspectrum of a white dwarf as template for telluric lines. tainties while those along X-axis represent each band s EW 28 Finally, spectra related to two split science expo(see Table 3). 29 sures were averaged to obtain final set of spectra used in 30 following analysis. A total of 40 orders (18 from red 31 CCD and 22 from blue one) cover wavelength range NextGen models (Baraffe et al. 1998), more suitable at A. The spectra have a SNR 8. lower temperatures (< 4500 K). " 33 In order tosspeed up fitting procedure, we restricted 34 range of Teff and log g to K and , re spectively. These constrains in parameter space did not s " affect final results as it was checked a posteriori that 3 ANALYSIS AND RESULTS same results were obtained considering full avail Stellar parameters able range for 38 Vboth parameters. The resulting best fitting syntic template, plotted in Figure 2 with photo Spectral energy distribution metric data, corresponds to Teff =6500 K, log g=4.5 and 40 V A first characterization of parent star can be performed [Fe/H]=-0.5 Kurucz model and A =0.44. Uncertainties on V comparing shape of Spectral Energy Distribution parameters were estimated both using χ2ν statistical! 42 Figure 7. INTconstructed i -band light WTS-1. Uobservapper panel: analysis and a bayesian a 1 (SED), fromcurve broaddata bandof photometric (flat prior) approach. The related 43 photometric model (red line) fitted inthe combination tions,data with and a gridbest of syntic oretical spectra. data errors result to be of order of 250 K, 0.2, 0.5 and WFCAM J-band data. Lower panel: residuals of with relative to photometric bands, collected in Table 3, were for Teff, log g, [Fe/H] and AV, respectively As it can be W 2 data point is best fit. analysed with application VOSA (Virtual Observatory s seen in Figure 2, WISE sun SED Analyser, Bayo et al. 2008, 2012). VOSA offers a valunot consistent with best fitting model. Firstly, it is worth 47 able set of tools for SED analysis, allowing estimanoting that observed value of W 2 = (±0.157) is 48 tion of stellar parameters. It can be accessed through its below 5 sigma point source sensitivity expected in and do not live long enough to produce a stable signal W 2-band (> 15.5). The number of single source detections 49 y web-page interface and accepts as input file an ASCII table over of several a timescale with following data: years. source identifier, coordinates of used for W 2-band measurement is also considerably less 51 source, distance to source in parsecs, visual extinction than that of W 1-band (4 and 19 respectively) 0.4 increasing (AV ), filter label, observed flux or magnitude and related uncertainty in measurement. Finally, poor an52 uncertainty. For our purpose, we tried to estimate only gular resolution c of WISE in W 2-band (6.4!! ) could 1 add Radial velocity main intrinsic parameters of star: effective temperature, furr imprecisions to final measured flux, especially in 54 surface gravity and metallicity. The extinction A was asa field as crowded as WTS 19hrs field. For se reasons, V 55 of most pernicious transit mimics in WTS are One c 1 sumed as a furr free parameter. WISE W 2 data point was not considered in fitting 56 eclipsing binaries. On one hand, a transit can be mimicked procedure. The syntic photometry is calculated by convolving 57 Once magnitude values were corrected for 1 in response curve of used filter set with oretic 58 terstellar absorption according to best fitting value cal syntic spectra. Then a2012 statistical test is performed, c! RAS, MNRAS 000, WTS-1 b (J=15.4, i=16.2, g=16.8), F7V Inflated HJ in close orbit around a late F-st J band i Table band 5. Properties of WTS-1 host star. Parameter T log g [Fe/H] Value 6250±200 K 4.4±0.1 [-0.5, 0] dex M 1.2±0.1 M R 1.15 m R 16.13±0.04 M 3.55 Simultaneous MCMC fitting of J and i-band with Mandel & Agol v sin(i) 7±2 km s models ρ 0.79 ρ Evidence against a blended background eclipsing binary: [0.6, 4.5] Gyr Age - consistent transit depths between J and i band Distance 3.2 kpc - consistent stellar density from light curve and spectroscopy after removal of data points to panel: whole set oflight curve folded WFCAM J-band datarelated points. transit. No significant signals were detected in Lombe 5. M High-resolution HET stacked (black full iddle panel: folded photometric data centred in days. transit and line) Scargle periodogram up to aspectrum period of 400 Since WTS! best model (red at line) in withspots with INT -band syn1 is 6707 afitted late F-star, recomparison are not many on ithree surface WTS-1 Li i line A combination in data. Lower panel: residuals of best fit. True space motion U motion V Cappetta et al. 2012, MNRAS,Trueinspace press spectra calculated with final adopted stellar parameters True space motion W 13±28 km s 20±38 km s -12±26 km s

8 Δ ΔRpl 24 and 0.9 respectively) were applied in order to achieve a redu 2 25 χ2 =1 in constant out-of-transit part of light curves Parameter Value multaneous fit of transit in J and i -band light curves. The χ 28level (darker to lighter orrespond to 68%, 95% and 99% confidence cant correlations. HET Spectroscopy Porb = days 30 l. t0 = HJD R /R = p s Table 9. Properties of new extrasolar planet WTS-1b. bisector spans measurements for 33 ρs = ρsun ctra. The phases were computed and ions expressed in Julian date β = 0.69 impact Parameter Value 0.09 d with transit fit. 35 Mp 4.01±0.35 MJ 36 RV BS 37 Rp RJ 5 km s 1 ] [km s 1 ] Prot d by an eclipsing binary that is blended with foreground a 0.047±0.001 AU.46 ± ± background star. On or hand, grazing eclipsing.80 ± ± e < 0.1 (C.L.= 95%) naries with near-equal radius stars also have shallow, n ± ± inc deg equal depth eclipses that can+0.05 phase-fold into transit-like.06 ± ± βimpact nals at half binary orbital period. In order to rule out.09 ± ± t HJD 0 eclipsing binaries scenarios, RV variation of WT.70 ± ± ±0.56 cm 3, 1.21±0.42interm ρj p system were firstρmeasured using g ISIS/WHT.97 ± ± phase within errors No correlation between bisectors and Teq a 1500±100 Kbinaries systems typic ate resolution spectra. Eclipsing Figure 9. T op panel: RVs values measured with highfurr evidence 0.47 against a± blended.11 ± background 46 eclipsing binary 1 Radial Velocities ΔRstar [Rsun] Δt0 [min]! 2

9 Detections in WTS: WTS-2b WTS J-band WTS-2b: J=13.87, i=15.15, K3V INT i-band Birkby et al. 2012, in prep.

10 RV (km/s) Radial Velocities HET Spectroscopy Phase No correlation between bisectors and phase within errors - furr evidence against a blended background eclipsing binary M. Cappetta, RoPACS Me

11 Properties WTS-2b WTS-1b: one of largest radius anomalies among known hot Jupiters in mass range 3-5MJ WTS-2b: a short-period hot Jupiter around a sub-solar mass star that favours a substantially higher Q (efficiency of tidal dissipation) value than predicted by ory.

12 Planets around M-Dwarfs 19h field: 2844 M0-2, 1679 M2-4 to J=17 No planet candidates around M-dwarf sample in 19h field more stringent upper limit on hot Jupiters fraction (with periods less than 10 days) than Kepler. Still not possible to 0.05 discriminate between Planetary fraction M M M K5 formation ories. Potentially up to 31 hot Neptunes present in 4 fields, but with high false alarm rate Effective temperature Kovacs et al. 2012, submitted to MNRAS

13 Conclusions Ground-based infrared transit surveys with irregular observing patterns are capable of finding exoplanets. WTS-1b and 2b are hot Jupiters around an F7V and a K3V star. The fraction of M-dwarfs with hot Jupiters with periods less than 10 days is less than 2-3%. Larger M-dwarf samples are needed to put stronger constraints PanPlanets