Thermal Emission of Exoplanet XO-1b
Pavel Machalek, Peter R. McCullough, Christopher J. Burke, Jeff A. Valenti, Adam Burrows, Joseph L. Hora
aa r X i v : . [ a s t r o - ph ] M a y Accepted for publication in The Astrophysical Journal
Preprint typeset using L A TEX style emulateapj v. 08/22/09
THERMAL EMISSION OF EXOPLANET XO-1B
Pavel Machalek , Peter R. McCullough , Christopher J. Burke , Jeff A. Valenti , Adam Burrows ,Joseph L. Hora Accepted for publication in The Astrophysical Journal
ABSTRACTWe estimate flux ratios of the extrasolar planet XO-1b to its host star XO-1 at 3.6, 4.5, 5.8 and 8.0microns with the IRAC on the
Spitzer Space Telescope to be 0.00086 ± ± ± ± ∼ × erg cm − s − sets a new lower limit for the occurrence of a thermal inversion in a planetaryatmosphere. Subject headings: stars:individual(XO-1) — binaries:eclipsing — infrared:stars — planetary systems INTRODUCTION
Over 270 extrasolar planets have been reported,more than 30 of which transit their primary star .In addition to the mass, radius and inclination ofthe planet evident from transits, atmospheric com-position can also be studied through transmissionspectroscopy, leading to detections of sodium, waterand methane (Charbonneau et al. 2002; Tinetti et al.2007; Swain et al. 2008). In addition, secondaryeclipse observations provide broadband emission spectra(Knutson et al. 2008; Charbonneau et al. 2008), plane-tary brightness temperatures (Charbonneau et al. 2005;Deming et al. 2005; Harrington et al. 2007) and evenday-night temperature contrast (Knutson et al. 2007).Torres et al. (2008) provide a reanalysis of light curvesand RV measurements of all then known transitingplanets.For “hot Jupiters”, planets with P orb .
10 days, afavorable planet-star ratio in the IR allows for directdetection of the planetary atmosphere by comparing thecombined flux from the star and the planet during andout of secondary eclipse at the superior conjunction.The contrast ratio in the mid-IR (1-10 microns) canbe higher than 10 − (cf. observations of HD 189733bby Charbonneau et al. (2008)) and theoretical predic-tions by Burrows et al. (2006), Fortney et al. (2006)and Burrows et al. (2007a), allowing for detection offluxes from planets at secondary eclipse using theIRAC, IRS, and MIPS cameras of the Spitzer SpaceTelescope (Werner et al. (2004)). Five planets havehad their secondary eclipse fluxes measured in one
Electronic address: [email protected] Department of Physics and Astronomy, Johns Hopkins Univer-sity, 3400 North Charles St., Baltimore MD 21218 Space Telescope Science Institute, 3700 San Martin Dr., Balti-more MD 21218 Department of Astrophysical Sciences, Princeton University,Princeton, NJ 08544 Harvard-Smithsonian Center for Astrophysics, 60 Garden St.,MS-65, Cambridge, MA 02138 or more IRAC bands: TrES-1 (Charbonneau et al.2005), HD 209458b (Deming et al. 2005; Knutson et al.2008), HD 189733b (Knutson et al. 2007), HD 149026b(Harrington et al. 2007) and GJ 436b (Deming et al.2007). In addition low resolution spectra of 2 transit-ing planets were obtained with the IRS spectrometerbetween ∼ µm : HD 189733b (Grillmair et al.2007) and HD 209458b (Richardson et al. 2007).Recently a detection of an atmospheric feature at-tributed to water has been claimed by Tinetti et al.(2007) and Barman (2007) by studying the transit fluxratios of HD 189733b and HD 209458b, respectively.Burrows et al. (2007a) analyzed the secondary trans-mission spectra of HD 209458b at all 4 infrared IRACSpitzer channels observed by Knutson et al. (2008)and suggested the observations are consistent with anatmospheric thermal inversion layer and yet unknownstratospheric absorber. A detailed study of the IRsecondary eclipse planetary spectra of HD 209458b, HD189733b, TrES-1, HD 149026b and non-eclipsing HD179949b, and υ And b by Burrows et al. (2007b) sug-gests that the presence of such a stratospheric absorbermight be dependent on the flux from the star at thesub-stellar point on the planet as well as second ordereffects like metallicity and planetary surface gravity. Inthe Burrows et al. (2007b) interpretation planets withhigh sub-stellar point flux (e.g., HD 209458b, OGLE-Tr-56b, OGLE-Tr-132b, TrES-2b and XO-3b) would havea stratospheric layer and a water feature in emissionwhile planets with lower fluxes (XO-1b, TrES-1, XO-2band HD 189733b) would have no such layer and a waterfeature in absorption. Fortney et al. (2007) also suggesta similar division of planetary spectra based on incidentstellar flux. Based on the planetary sub-stellar pointflux from the star, both Burrows et al. (2007b) andFortney et al. (2007) predict that XO-1b should notexhibit a thermal inversion in its atmosphere.We present observations of the infrared spec-tral energy distribution (SED) of the planet XO- Machalek et al.1b (McCullough et al. 2006) in all 4 IRAC channelsobtained during secondary eclipses with the IRAC of
Spitzer Space Telescope . By comparing our ∼ µm SED with atmospheric models, we test for the presenceof a thermal inversion layer in XO-1b. OBSERVATIONS
The InfraRed Array Camera (IRAC; Fazio et al. 2004)has a field of view of 5.2 ′ × ′ in each of its four bands.Two adjacent fields are imaged in pairs (3.6 and 5.8 mi-crons; 4.5 and 8.0 microns). The detector arrays eachmeasure 256 ×
256 pixels, with a pixel size of approx-imately 1.22 ′′ × ′′ . We have observed XO-1 in all4 channels in two separate Astronomical Observing Re-quests (AORs) in two different sessions: the 4.5 and 8.0micron channels for 5.9 hours on UT 2007 April 02 (AOR21374464) and the 3.6 and 5.8 micron channels for 5.9hours on UT 2007 Aug 10 (AOR 21374208). We usedthe 12 s frame time, obtaining 1620 full-array images ineach bandpass with a cadence of 13.2 s and an effectiveintegration time of 10.4 s . The pointing was not ditheredand was selected such that for the Apr 2007 observationsin the 4.5 and 8.0 micron channels, two bright calibra-tors for XO-1 (2MASS J16021184+2810105 J=9.939 J -K =0.412) were used: 2MASS J16021795+2813328: J =9.913 J - K = 0.564 and 2MASS J16020133+2809268:J = 12.542 J - K = 0.795. The Aug 2007 observationsin the 3.6 and 5.8 micron channels used 2 bright cali-brators: 2MASS J16021311+2809004: J=11.045 J - K =0.652 and 2MASS J16020133+2809268: J = 12.542 J -K = 0.795.We used the standard IRAC Basic Calibrated Dataproducts (version 16.1) described in the Spitzer DataHandbook , which includes dark frame subtraction, mul-tiplexer bleed correction, detector linearization, andflat-fielding of the images. We converted the timesrecorded by the spacecraft in the FITS file header key-word DATE-OBS to heliocentric Julian dates using theorbital ephemeris of the spacecraft provided by the Hori-zons Ephemeris System .Prior to performing aperture photometry, we resam-pled the images in all four channels to a 10 times finergrid in each spatial direction using flux-conserving bi-linear interpolation (similar to Harrington et al. (2007)).With the implementation for aperture photometry thatwe used, resampling makes a marginal improvement inthe photometry, i.e. a slightly lower r.m.s. of out-of-eclipse points, presumably related to how the routinehandles fractional pixels at the edge of the aperture.The zodiacal background was subtracted in eachchannel by constructing a histogram of all pixels in eachimage and fitting a Gaussian to the distribution of thezodiacal background brightness. A constant mean valueof the Gaussian was then subtracted from each pixel inthe image to construct a background-subtracted image.The centroids of XO-1 and the 2 calibrators were evalu-ated by fitting a Gaussian to the stellar flux distribution.The pointing varied by 0.3 pixel, and the shifting of thestellar centroid within a pixel, which have sub-pixel sen-sitivity variations, resulted in a modulation of the stellarflux in the 3.6 and 4.5 micron channels (described below). http://ssc.spitzer.caltech.edu/irac/dh/ http://ssd.jpl.nasa.gov/ Aperture photometry was then performed on the im-ages with an aperture radius of 4 pixels, which was foundto be the optimum value for all 4 channels. The sizeof the aperture was determined by minimizing the rmsscatter in the light curve for observations outside of theeclipse. Apertures smaller than 4 pixels contained insuf-ficient stellar flux and larger apertures were more con-taminated by the sky (especially in the high backgroundsignal of the 5.8 and 8.0 micron channels). An appro-priate aperture correction for each channel was appliedto the stellar flux value according to the Spitzer DataHandbook of [1.112, 1.113, 1.125, 1.218] for the [3.6; 4.5;5.8 and 8.0] micron channels, respectively. The σ of theout-of-eclipse points was calculated iteratively using 3- σ outlier rejection at each step until no more points wererejected. To remove cosmic rays the resultant robust σ was used to reject entire images which contain the 3- σ outliers above and below the mean of the light curve.1.4%; 1.9%; 3.0 %, 2.4 % of images from the 3.6, 4.5,5.8 and 8.0 micron channels, respectively, were removedin this fashion. The higher rejection rate in the two red-der channels is consistent with a higher number of cosmicray-affected pixels in these channels (Patten et al. 2004).Throughout the analysis we have preserved flux units. The 3.6- µ m time series exhibited a sharp increase dur-ing the first ∼
30 minutes of exposure for XO-1 and the2 calibrators, presumably as a result of the instrumentreaching a new equilibrium after previous observations.Such relaxation effects can reach several percent and usu-ally stabilize within the first hour of observations of a newtarget. We have ignored the first 125 points ( ∼
30 min)in the 3.6- µ m time series in addition to the high-sigmaoutlier rejection as described in the previous section.A strong correlation between the sub-pixel centroidand stellar brightness was observed in both the 3.6- µ m and 4.5- µ m channels, with flux magnitudes of ∼ ∼ ∼ . The uncorrectedsub-pixel intensity variations are clearly visible in thetime series of XO-1 in the 3.6- µ m and 4.5- µ m channelsin Fig. 1. We have corrected for this sub-pixel intensityvariations after Charbonneau et al. (2008) by fitting aquadratic function to the photometric flux points of XO-1b observed out-of-eclipse as a function of the x and ysub-pixel centroids: I subpixel = b + b × x + b × x + b × y + b × y , (1)where x and y are the subpixel centroids of center oflight of the star and b n are the fit parameters. The rmsresidual of the XO-1 time series for points outside of theeclipse after correction for the sub-pixel intensity vari-ation was 0.0020, which is 18% higher than a theoret-ical estimate of XO-1 Poisson noise based on detector http://ssc.spitzer.caltech.edu/documents/exoplanetmemo.txt O-1b Thermal emission 3 -0.15 -0.10 -0.05 -0.00 0.05 0.10 0.15Time from predicted eclipse center (days)0.9200.9300.9400.9500.9600.9700.9800.9901.0001.010 R e l a t i v e f l u x + c on s t an t Fig. 1.—
Secondary eclipse observations of XO-1b withIRAC on
Spitzer Space Telescope obtained on UT 2007 April02 and UT 2007 Aug 10 in 3.6, 4.5, 5.8 and 8.0 micron chan-nels (from top to bottom) binned in 6-minute interval andnormalized to 1 and offset by 0.02 for clarity. The overplot-ted solid lines do not represent a fit to the data, but rathershow the correction for the detector effects. The 3.6 and 4.6micron time series are decorelated using XO-1 out-of-eclipsepoints and the 5.8 and 8.0 micron time series is detrendedusing a fit to a calibrator star in the field (see § § read noise and background noise. The time series wasthen normalized by taking the robust average of out-of-eclipse points and binned in 6-minute intervals contain-ing approximately 30 individual measurements each (seeFig. 2).The 4.5- µ m time series also exhibited an initialrelaxation-induced brightness increase and consequently139 points corresponding to the first ∼
30 minutes of ob-servations were rejected, which is more points than inthe 3.6- µ m time series. The analysis of the time serieswas identical to that of the 3.6- µ m time series. The rmsof out-of-eclipse points was 0.0024, which is 19% higher TABLE 1XO-1 absolute fluxes
IRAC channel XO-1 flux XO-1 instrumentaleffective λ magnitude(microns) (mJy) (mag)3.6 45.1 ± ± ± ± than the theoretical estimate and is similar to that ofTrES-1 (rms=0.0027 Charbonneau et al. (2005)).We have tested for the linearity of the detector re-sponse in the 3.6 and 4.5 micron channels in whichXO-1 is close to the onset of detector non-linear re-sponse. Using a subset of data from the SAGE survey(Meixner et al. 2006) obtained in the high dynamic range(HDR) mode of IRAC camera with both 0.6s and 12.0sintegration times we are able to determine that both the3.6 micron and 4.5 micron XO-1 fluxes are unsaturatedand in the detector linear regime response. Table 1 showsthe absolute XO-1 fluxes and the instrumental magni-tudes in the four IRAC channels. The 5.8 and 8.0 micron time series were recorded withSi:As detectors and do not thus exhibit the prominentsub-pixel intensity variations evident in the 3.6 micronand 4.5 micron channels. The first ∼
30 minutes of ob-servations (139 data points) were rejected as the instru-ment settled into a new equilibrium state. Fig. 1 showsintensity variation with time, which is caused by changesin the effective gain of individual pixels over time. Thiseffect has also been been observed by Harrington et al.(2007) at 8 micron with IRAC and by Deming et al.(2006) at 16 micron with IRS. The intensity variationsare dependent on the illumination level of the individ-ual pixel (Knutson et al. 2007, 2008), pixels with highillumination will reach their equilibrium within ∼ I model = a + a × ∆ t + a × ln ∆ t + a × ( ln ∆ t ) , (2)where I model is the model flux, ∆ t is the time since thebeginning of observations and a i are the free parameters.The detector ramp intensity decreased in flux during the5 hours of observation by ∼ The nor-malized and binned 5.8 micron time series is depicted inFig. 2.Fitting the detector ramp using a calibrator star, whichis non-variable at the 0.012 level after detector ramp re-moval, allows us to bypass using XO-1 itself to removethe detector ramp by making a fit to its 5.8 micron out-of-eclipse points. The choice whether to correct for thedetector ramp of XO-1 using the photometry of a calibra-tor star or the target star itself could be a limiting factorin our analysis. Deming (2008, p.c.) has indicated thatin a long ( >
10 hours) series of full-frame 12s-exposureIRAC photometry, fitting the detector ramp to the tar-get star while masking out the points in the eclipse couldbe more appropriate. We implemented each of the twoalternate methods of correction, and for these observa-tions of XO-1b, the measured depths at 5.8 micron arewithin 1- σ of each other.Charbonneau et al. (2005), Knutson et al. (2007),Harrington et al. (2007) and Deming et al. (2007) havereported a nonlinear flux increase over time in the 8.0 mi-cron IRAC channel. We also detect a nonlinear increasein the brightness of XO-1 and in the 2 calibrators, inaddition to the sharp increase during the first ∼
30 min-utes of observations (137 points). The initial ramp-updata were discarded and the ∼ ANALYSIS
We fit the secondary eclipse light curves using the for-malism of Mandel & Agol (2002) with no stellar limbdarkening and adopt stellar and orbital parameters(Holman et al. 2006) : R ⋆ = 0.928 +0 . − . R ⊙ , R p =1.184 +0 . − . R Jup , i = 89.31 +0 . − . degrees, and a =0.0488 ± T c ( E ) = 2 , , . HJD ) + E (3 . days ) . (3)We fit the depth of the eclipse ∆ F and the timing ofthe centroid ∆ T independently in all 4 channels in theunbinned light series using Levenberg-Marquardt mini-mization (Press et al. 1992) with an equal error assignedto all points, which is equal to the rms of out-of-eclipsepoints in each time series. Best-fit eclipse curves areplotted in Fig. 2 and the eclipse parameters are listed inTable 2. They are the channel wavelength, eclipse depth http://ssc.spitzer.caltech.edu/documents/irac_memo.txt We have also reduced the data using the stellar and orbital pa-rameters from Torres et al. (2008) as a test, but the eclipse depthschanged negligibly and eclipse mid-center timings were all within1- σ . Jup = 71,492 km. -0.15 -0.10 -0.05 -0.00 0.05 0.10 0.15Time from predicted eclipse center (days)0.9550.9600.9650.9700.9750.9800.9850.9900.9951.0001.005 R e l a t i v e f l u x + c on s t an t Fig. 2.—
Secondary eclipse of XO-1b observed with IRAC on
Spitzer Space Telescope in 3.6, 4.5, 5.8, and 8.0 micron chan-nels (top to bottom) corrected for detector effects, normalized andbinned in 6-minute intervals and offset for clarity. The best-fiteclipse curves are overplotted. ∆ F , eclipse mid-center time in HJD, and the timing off-set ∆ t in minutes from the expected secondary eclipsemid-center time for an assumed eccentricity of zero, andthe reduced χ . The reduced χ is close to 1.0 in all 4channels, indicating a good fit to the data.To estimate the errors on the depth and mid-eclipsetiming we performed the error analysis using the boot-strap method from Press et al. (1992). The bootstrapmethod makes no prior assumptions about the distribu-tion of the noise in the data and the data points are notaltered as in the Monte-Carlo analysis. For 10,000 trialruns we have randomly drawn with replacement pointsfrom the normalized- and detector-effect-decorrelated,but otherwise unaltered, light curve, until we had thesame number of data points in the light curve thatwe started with. During each iteration we performedthe full eclipse fitting for eclipse depth ∆ F and eclipsemid-center ∆ T . The 1 − σ errors for ∆ F and ∆ T werecomputed by fitting a Gaussian to the respective 1-D dis-tribution of bootstrap points and are reported in Table 2.O-1b Thermal emission 5 TABLE 2Secondary eclipse best fit parameters λ Eclipse Depth ∆ F Eclipse Center Time Time offset ∆ T Reduced χ (microns) (HJD) (min)3.6 0.00086 ± ± ± ± ± ± ± ± ± ± ± ± The eclipse depth errors ∆ F = [0.00007, 0.00009,0.00031, 0.00029] for the J = 9.939 XO-1 compare fa-vorably with the eclipse depth errors [0.00009, 0.00015,0.00043, 0.00026] of Knutson et al. (2008) for the J =6.591 HD 209458, despite the fact that XO-1 is dimmer.XO-1 observations were made in the full-array modewith 10.4-s integration time and readout time 2.8 s for atotal of 1,620 images, while the HD 209458 observationswere made in sub-array mode with exposure time of 0.1s in sets of 64 in each channel totaling 26,880 usableimages. The S/N scales as ∝ √ n exp * f e * ∆ t / σ total ,where n exp is the number of exposures during theduration of the eclipse, f e is the stellar flux in signalelectrons, ∆ t is the integration time and σ total is thecombined Poisson, readout and background noise. Thepredicted eclipse depth errors for XO-1 in the four IRACchannels are then [0.4; 0.5; 0.9; 2.5] times the respectiveeclipse depth errors for the 21 times brighter HD 209458.To test the robustness of our data reduction andanalysis technique and consistency with other observa-tions in the IRAC full-array mode we have re-reducedthe 4.5 and 8.0 micron IRAC secondary eclipse dataof TrES-1 by Charbonneau et al. (2005) with ourpipeline, rejecting the first 30 minutes in both channels.The procedure described in § § F . µm = 0.00043 and∆ F . µm = 0.00194, which are -1.1 σ and -0.9 σ awayfrom the Charbonneau et al. (2005) secondary eclipsedepths of ∆ F . µm = 0.00057 ± F . µm = 0.00225 ± σ ; -2.0 σ ] away fromthe Charbonneau et al. (2005) [4.5; 8.0]micron channelvalues +19.6 ± ± ∼ σ level, the eclipse mid-center timing offset inthe 8.0 micron channel is only mildly consistent at the2.0 σ level, probably because Charbonneau et al. (2005)observations were made using multiple AstronomicalObservation Requests (AORs) which resulted in arbi-trary flux shifts in the time series. Charbonneau et al.(2005) do not mention how they have corrected for theseflux shifts. Despite this fact our pipeline is capableof reproducing their results to within ∼ σ in eclipsedepth and ∼ σ in mid-eclipse timing. We have thusdemonstrated that our reduction pipeline is robust andthe secondary eclipse depth estimates are consistentwith other major full-array pipelines. DISCUSSION
The eclipse mid-center timings for XO-1b in Table2 are individually consistent with zero timing resid-uals for a circular orbit based on the ephemeris byMcCullough et al. (2006), but the April and August 2007combined timings show a time shift. The UT Apr 2 2007observations of 4.5 micron and 8.0 micron channels havea combined eclipse mid-center timing of 2454193.21366 ± ± ± ± e × cos ( ω ) ≃ π ∆ t P , (4)where e is the eccentricity, ω is the longitude of perias-tron, P is the orbital period, and ∆ t is the centroid timeshift from expected time of secondary eclipse, allows usto set a 2 σ upper limit on e × cos( ω ) < µm channel, with a decrease in the 3.6 micronand 4.5 micron channel and a slight decrease towardsthe 8.0 micron channel. Furthermore the flux in the 4.5 µm channel is higher than in the 3.6 micron channel,which does not match the general character of the cloud-less models of Burrows et al. (2006) for redistributionparameter P n =0.3 (dot-dashed line and open circlesfor band-averaged ratios in Fig. 3), which predict alower flux at 4.5 microns than at 3.6 microns; the modelwith thermal inversion predicts the opposite. Since theobservations manifest a higher flux in the 4.5 micronchannel than at 3.5 micron, a thermal inversion in theatmosphere might be indicated. P n =0 corresponds tono heat redistribution from the planetary day-side tothe night-side and P n = 0.5 stands for full redistribution(see Burrows et al. (2007b) for details). The possibilityof thermal inversion in a planetary atmosphere hasbeen suggested by Hubeny et al. (2003), Burrows et al.(2006, 2007a) and Fortney et al. (2007). Recently,Burrows et al. (2007a,b) suggested that model spectracould match the observations of HD 209458b andHD 149026b if a stratospheric absorber of unknowncomposition (possibly tholins, polyacetylenes, TiO or Machalek et al.VO) were present in the atmosphere of the planet.The presence of a stratospheric absorber would yield athermal inversion in the planetary atmosphere and thepresence of the water features in emission for a varietyof heat redistribution parameters P n . µ m0.00000.00100.00200.00300.0040 P l ane t / s t a r f l u x r a t i o Fig. 3.—
Spitzer Space Telescope
IRAC secondary eclipse depthsfor XO-1b with bootstrap error bars (filled squares). The pre-dicted emission spectrum of the planet with an upper atmosphericabsorber of κ e = 0.1 cm /g and a redistribution parameter ofP n =0.3 is plotted as a solid line. A model with no atmosphericabsorber and a redistribution parameter of P n =0.3 is over plottedwith dot-dashed line (see § Spitzer Space Telescope
IRAC response curves for the 3.6-, 4.5-, 5.8-, and 8.0 micron channels are plotted at the bottom of the fig-ure (doted lines). The theoretical flux ratios obtained from a XO-1stellar spectrum (from http://kurucz.harvard.edu/stars/XO1/ )and an assumed black-body spectrum for the planet at [1200, 1600]K are plotted as dashed lines.
Our observations suggest the presence of a thermalinversion layer and a possible stratospheric absorberin the atmosphere of the XO-1b planet. The solid lineand open squares in Fig. 3 depict an atmospheric modelof XO-1b, following the methodology of Burrows et al.(2006, 2007a) with a thermal inversion and a strato-spheric absorber of opacity of κ e = 0.1 cm /g andredistribution parameter of P n = 0.3. The latter modelfits the data better than the canonical cloudless modelwith P n = 0.3 (dot-dashed curve and open circles foraveraged band ratios). The band-averaged flux ratiosfor the model with a stratospheric absorber (opensquares) are within the error bars for the 3.6, 4.5, and8.0 micron channels, but are inconsistent by 2.7 σ withthe band-averaged flux ratios for the 5.8 micron channel.This is similar to the situation for the IRAC fit to theobservations by HD 209458b by Knutson et al. (2008).The absorber-free canonical model (dot-dashed line)is clearly inconsistent with our observations (Fig. 3)of XO-1b in all 4 channels by [3.4 σ , 7.1 σ , 6.3 σ , 3.7 σ ],respectively.Burrows et al. (2007b) and Fortney et al. (2007) sug-gested that the presence of the stratospheric absorbermight be correlated with the incident flux from thestar at the sub-stellar point on the planet, the preciselevel of which is yet to be refined. The presence ofan irradiation-induced stratospheric absorber has been suggested by Burrows et al. (2007a) for HD 209458b(see our Fig. 4) with a sub-stellar flux of ∼ × erg cm − s − at a distance a = 0.045 AU. Interestingly,XO-1b has a lower sub-stellar flux of ∼ × ergcm − s − and a semi-major axis of a = 0.0488 AU,but still manifests evidence for a thermal inversion. Arecent study of the broadband infrared spectrum ofHD 189733b (see our Fig. 4) by Charbonneau et al.(2008) finds no evidence for an atmospheric thermalinversion, despite a similar sub-stellar point flux of ∼ × erg cm − s − (Burrows et al. 2007b) with asmaller semi-major axis a = 0.0313 AU. Further studyof planetary atmospheres should refine the concept ofthis sub-stellar flux boundary with respect to the pres-ence/absence of a stratospheric absorber and thermalinversion. µ m0.00000.00100.00200.00300.0040 P l ane t / s t a r f l u x r a t i o XO-1bHD 209458bHD 189733bGJ 436bTrES-1HD 149026b
Fig. 4.—
Comparison of
Spitzer Space Telescope
IRAC secondaryeclipse depths: XO-1b (filled square) from this paper; HD 209458b(filled circle) from Knutson et al. (2008); HD 189733b (filled tri-angle) from Charbonneau et al. (2008); GJ 436b (filled star) fromDeming et al. (2007); TrES-1 (filled upside down triangle) fromCharbonneau et al. (2005); and HD 149026b (open circle) fromHarrington et al. (2007). The central wavelengths have been offsetby [+0.1;-0.1] microns for clarity. The normalized
Spitzer SpaceTelescope
IRAC response curves for the 3.6-, 4.5-, 5.8-, and 8.0 mi-cron channels are plotted at the bottom of the figure (doted lines).
Atmospheric water detection has been claimed inthe transit broadband spectra of HD 189733b byTinetti et al. (2007) and in its secondary eclipse spec-tra by Fortney & Marley (2007). Burrows et al. (2007a)also found evidence for water vapor emission in the at-mosphere of HD 209458b. Our data can be interpreted asevidence for rovibrational band of water emission long-ward of ∼ n (Burrows et al. (2007b), especially their Fig. 4). Fur-ther modeling would allow tighter constraints on the P n ,not just for XO-1b, but for a variety of planets. Thefortuitous importance of the 3.6 micron IRAC channelto the study of planetary atmospheres is likely to be en-hanced as the Spitzer Space Telescope runs out of cryo-coolant in 2009, when only the 3.6 micron and 4.5 micronchannels will be available.O-1b Thermal emission 7 CONCLUSION
We report the estimated flux ratios of the planet XO-1b in the
Spitzer Space Telescope
IRAC 3.6, 4.5, 5.8 and8.0- µm channels. We find that the estimated fluxes arenot consistent with a canonical cloudless model for ther-mal emission from the planet and instead may indicatean atmosphere with as yet unknown stratospheric ab-sorber and a likely thermal inversion, which would causethe water band longward of 4.0 microns to switch fromabsorption to emission. The atmospheric model with athermal inversion produces a tight match to the data at3.6, 4.5, and 8.0 microns, but is inconsistent by 2.7 σ withobservations at 5.8 microns. This is similar to observa-tions of HD 209458b (Knutson et al. 2008).The presence or absence of the stratospheric absorberand thermal inversion layer has been linked to the fluxfrom the parent star at the sub-stellar point on theplanet. The XO-1b sub-stellar point flux of ∼ × erg cm − s − is the lowest so far reported for aplanetary atmosphere with a thermal inversion. Obser-vations of atmospheres of other planets may permit abetter understanding of the thermal inversion layer andparametrization of the characteristics that create such athermal inversion. The authors would like to thank J. E. Stys,R. Gilliland, C. M. Johns-Krull, and K. A. Janesfor helpful discussions. The authors would also liketo acknowledge the use of publicly available routinesby Eric Agol and Levenberg-Marquardt least-squaresminimization routine MPFITFUN by Craig Markwardt.P.M. and P.R.M. were supported by the Spitzer ScienceCenter Grant C4030 to the Space Telescope ScienceInstitute. A.B. was supported in part by NASA undergrants NAG5-10760 and NNG04GL22G. This work isbased on observations made with the Spitzer SpaceTelescope, which is operated by the Jet PropulsionLaboratory, California Institute of Technology undera contract with NASA. This publication also makesuse of data products from the Two Micron All SkySurvey, which is a joint project of the University ofMassachusetts and the Infrared Processing and AnalysisCenter/California Institute of Technology, funded bythe National Aeronautics and Space Administration andthe National Science Foundation. The authors wouldlike to thank the reviewer Dr. Drake Deming for hishelpful comments which have substantially improvedthe manuscript.is the lowest so far reported for aplanetary atmosphere with a thermal inversion. Obser-vations of atmospheres of other planets may permit abetter understanding of the thermal inversion layer andparametrization of the characteristics that create such athermal inversion. The authors would like to thank J. E. Stys,R. Gilliland, C. M. Johns-Krull, and K. A. Janesfor helpful discussions. The authors would also liketo acknowledge the use of publicly available routinesby Eric Agol and Levenberg-Marquardt least-squaresminimization routine MPFITFUN by Craig Markwardt.P.M. and P.R.M. were supported by the Spitzer ScienceCenter Grant C4030 to the Space Telescope ScienceInstitute. A.B. was supported in part by NASA undergrants NAG5-10760 and NNG04GL22G. This work isbased on observations made with the Spitzer SpaceTelescope, which is operated by the Jet PropulsionLaboratory, California Institute of Technology undera contract with NASA. This publication also makesuse of data products from the Two Micron All SkySurvey, which is a joint project of the University ofMassachusetts and the Infrared Processing and AnalysisCenter/California Institute of Technology, funded bythe National Aeronautics and Space Administration andthe National Science Foundation. The authors wouldlike to thank the reviewer Dr. Drake Deming for hishelpful comments which have substantially improvedthe manuscript.