IntroductionObservationsData AnalysisDiscussionConclusionsarXiv:0812.1844v2 [astro-ph] 10 Dec 2008Accepted in Astrophysical Journal LettersMolecular Signatures in the Near Infrared Dayside Spectrum ofHD 189733bM. R. Swain1, G. Vasisht1, G. Tinetti2, J. Bouwman3, Pin Chen1, Y. Yung4, D. Deming5,& P. Deroo1ABSTRACTWe have measured the dayside spectrum of HD 189733b between 1.5 and2.5 µm using the NICMOS instrument on the Hubble Space Telescope. Theemergent spectrum contains significant modulation, which we attribute to thepresence of molecular bands seen in absorption. We find that water (H2O),carbon monoxide (CO), and carbon dioxide (CO2) are needed to explain theobservations, and we are able to estimate the mixing ratios for these molecules.We also find temperature decreases with altitude in the ∼ 0.01 < P < ∼ 1 barregion of the dayside near-infrared photosphere and set an upper limit to thedayside abundance of methane ( CH4) at these pressures.Subject headings: planetary systems — techniques: spectroscopic1. IntroductionHD 18973 3b is a transiting hot-Jupiter planet in a 2.2-day orbit around a K2V stellarprimary (Bouchy et al. 2005). Due to the relatively large depth of the eclipse (∼ 2.5% at Kband) and the bright stellar primary (K mag = 5.5), this system was immediately recognizedas an important target for atmospheric characterization observations, and its emission has1Jet Propulsion Laboratory California, Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA911092University College London, Gower Street, London WC1E 6BT, UK3Max-Planck Institute for Astronomy, Konigstuhl 17, D-69117 Heidelberg, Germany4Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA 911255Goddard Space Flight Center, Planetary Systems Branch, Code 693, Greenbelt, MD– 2 –been studied extensively durning secondary eclipse (Deming et al. 2005; Knutson et al.2007; Grillmair et al. 2007; Knutson et al. 2008; Charbonneau et al. 2008). Multicolorphotometric observations revealed the presence of H2O (Tinetti et al. 2007a) and the likelypresence of CO (Charbonneau et al. 2008), while optical transmission spectroscopy suggestsscattering by high a ltitude haze (Pont et al. 2008). In addition, extensive theoretical workhas also been done on the atmosphere of this planet by Fortney et al. (2006); Barman(2008); Burrows et al. (2008); Showman et al. (2008) and others.Recently, the NICMOS camera on the Hubble Space Telescope (HST) was used duringthe primary eclipse of HD 189733b to obtain a near-IR transmission spectrum of the planet’satmosphere; these results showed the presence of H2O and CH4(Swain et al. 2008; hereafterSVT08). The near-IR transmission spectrum of HD 18 9733b probes the upper (P ∼ 10−4bar estimated from our models) regions of the of the atmosphere at the terminator. In thispaper, we report the r esults o f HST observations of the dayside spectrum (derived fromsecondary eclipse measurements). At near infrar ed wavelengths, the dayside portion of theatmosphere of HD 189433b is probed in deeper (P ∼ 0.1 bar estimated from our models)regions.2. ObservationsWe observed HD 189733 with the Hubble Space Telescope for five contiguous spacecraftorbits, using NICMOS in imaging-spectroscopy mode with the G206 grism (wavelength cov-erage 1.4-2.5 µm). Observations began on April 29, 2007 at UT 23:47:58 and ended at UT06:51:52 on the next day. The first two orbits (O1& O2) observed the target pre-ingress, thethird orbit (O3) was phased to capture the occultation, while the fourth and fifth orbits (O4& O5) were post-egress (see figure 1). A total of 638 usable snapshot spectra were acquiredduring the five orbits. The third orbit (the occultation) contained 130 spectra. Using thebest available ephemerides (Winn et al. 2007; Pont et al. 2008) we determine that full-occultation starts ∼ 13 exposures into O3, and lasts for the remainder of this orbit. Theeffective exposure time for each spectrum was T = 1.624 s; the overall instrumental configu-ration, including the location of the spectrum on the f ocal plane array, was identical to thatin SVT08. A few spectral calibration exposures (in O1) were acquired using a narrow-bandfilter.Because the NIC-3 camera is severely undersampled a nd because of large gain drop-offat the edges in the NICMOS detector pixels, the instrument was configured in DEFOCUSmode (FWHM ≃ 5 pixels). This defocus reduces the level of photometric fluctuations due toboth pointing jitter (random) and beam wander (systematic errors). While the defocus also– 3 –helps to minimize the bright-source overheads for this extremely bright target, it neverthelesslimits the effective spectral resolution to R ≃ 40.2.1. Data AnalysisA complete description of the data-analysis methods is provided in the supplementaryinformation to SVT08. Herein, we discuss only departures and addenda to these analysismethods.First, each image is reduced to a one-dimensional spectrum covering 1.5-2.5 µm in spec-tral ra ng e. This includes combining data from the grating’s first a nd second orders, whichare simultaneously imaged onto the detector array. Inclusion of the second order improvesthe SNR in the blue because the second order, which is partially imaged onto t he detector,contains more detected photon-flux at the shortest wavelengths (shortwards of 1.7 µm, long-wards of which t he efficiency of the 2ndorder falls rapidly) as a consequence of the peculiarblaze of t he parent grating. Prior to combination, wavelength solutions for both orders wereobtained using the calibration exposures and the wavelength-scale coefficients in Nicmoslook(Freudling et al. 1997). Merger of the spectral orders takes advantage of the near integerratio, ≃ 2, of the spectral resolutions of the two orders. The orders are coadded around afiducial wavelength, and any smearing is captured by assigning the coadded channel a wave-length equivalent to the weighted average of wavelengths of component channels. Prior tocoadding, the 2ndorder spectra were multiplied by a broa d Tukey window so as to suppressthe effects of (1) the sharp band- edge at 1.5 µm and (2) the detector array edge at ∼ 1.85 µmon the final spectrophotometry. Lastly, coadded spectra are rebinned to wavelength chan-nels at t he binwidth of the DEFOCUS full-width-at- half -maximum, guaranteeing adjacentchannels have independent spectral content. As the state variables are channel independent,the two
View Full Document
Unlocking...