1: \documentstyle[11pt,twoside,colloqOHP,epsfig]{article}

2: \markboth{Deming \etal}{Infrared Radiation from Hot Jupiters}

3: \pagestyle{myheadings}

4: %

5: %

6: \begin{document}

7: %

8: \title{Infrared Radiation from Hot Jupiters}

9: \author{Drake Deming$^1$, L. Jeremy Richardson$^2$, Sara Seager$^3$, 

10: Joseph Harrington$^4$}

11: \affil{$^1$Planetary Systems Laboratory, NASA's Goddard Space Flight Center,

12:  Code 693, Greenbelt, MD 20771 USA [ddeming@pop600.gsfc.nasa.gov]} 

13: \affil{$^2$NRC Research Associate in the Exoplanet and Stellar Astrophysics 

14: Laboratory, NASA's Goddard Space Flight Center, Code 667, 

15: Greenbelt MD 20771 USA 

16: [richardsonlj@milkyway.gsfc.nasa.gov]} 

17: \affil{$^3$Department of Terrestrial Magnetism, Carnegie Institution 

18: of Washington, 5241 Broad Branch Road NW, Washington, DC 20015 USA 

19: [seager@dtm.ciw.edu]} 

20: \affil{$^4$Center for Radiophysics and Space Research, Cornell University,

21: 326 Space Sciences Bldg., Ithaca, NY 14853-6801 USA 

22: [jh@oobleck.astro.cornell.edu]} 

23: 

24: \begin{abstract}

25: Recent Spitzer infrared (IR) observations of two transiting hot

26: Jupiters during their secondary eclipses have provided the first direct

27: detection of planets orbiting other stars (Charbonneau et al. 2005;

28: Deming et al. 2005).  We here elaborate on some aspects of our

29: detection of HD\,209458b at 24~$\mu$m, and we compare to the detection

30: of TrES-1 by Charbonneau et al.  Spitzer will eventually determine the

31: IR spectral energy distribution of these and similar hot Jupiters,

32: opening the new field of comparative exoplanetology.  For now, we have

33: only three Spitzer data points, augmented by upper limits from the

34: ground.  We here interpret the available measurements from a purely

35: observational perspective, and we point out that a blackbody spectrum

36: having $T \sim 1100$K can account for all current IR measurements,

37: within the errors.  This will surely not remain true for long, since

38: ongoing Spitzer observations will be very sensitive to the IR

39: characteristics of hot Jupiters.

40: \end{abstract}

41: %

42: \section{Direct Detection of Extrasolar Planets}

43: 

44: There are now over 160 extrasolar planets known from Doppler surveys,

45: and about 15\% of these are hot Jupiters.  Strong stellar irradiation

46: heats these planets to $T > 1000$K (Seager and Sasselov 1998), so they

47: should emit most of their radiation in the IR.  A number of pioneering

48: attempts were made to detect hot Jupiters from the ground in the

49: combined IR light of star+planet (Wiedemann et al. 2001, Lucas and

50: Roche 2002, Richardson et al. 2003a,b). But success was only recently

51: achieved, and it required using the Spitzer Space Telescope

52: (Charbonneau et al. 2005; Deming et al. 2005).

53: 

54: The Spitzer observations of two hot Jupiters are the first direct

55: detections of extrasolar planets.  To be clear about terminology, by

56: direct detection we mean that photons emitted by the planet are

57: detected, and are separated from stellar photons by some method.  One

58: method to separate the stellar and planetary photons would be to

59: spatially resolve the planet and the star, e.g., by imaging.  Imaging

60: of extrasolar planets is being widely pursued, but it requires a very

61: high order of technology. The technique which is successful for

62: Spitzer is to observe the {\it secondary eclipse} of transiting hot

63: Jupiters.  We measure the star+planet when the planet is out of

64: eclipse, and the star alone when the planet is eclipsed, and the

65: diference tells us how many photons are due to the planet.

66: 

67: Note that the secondary eclipse technique does not specify {\it which}

68: of the detected photons are due to the planet, and which are from the

69: star, only {\it how many} photons are from the planet.  But for

70: scientific purposes, `how many' is the primary quantity of interest!

71: So secondary eclipses are a very sensitive and useful method for

72: direct detection and characterization of transiting extrasolar

73: planets.  There are {\it nine} ongoing Spitzer programs to detect and

74: characterize close-in extrasolar planets in combined IR light, and we

75: are now truly in the era of comparative exoplanetology.

76: 

77: In Sec 2, we describe our detection of the secondary eclipse of

78: HD\,209458b, elaborating on some points not mentioned in our detection

79: paper.  Sec. 3 interprets our results in combination with the TrES-1

80: measurements by Charbonneau et al. (2005), and factors in the

81: ground-based upper limits.  We point out the importance of measuring

82: these planets in the 2- to 5-$\mu$m wavelength region, where there is

83: currently a considerable question about their level of IR emission.

84: 

85: \section{The Secondary Eclipse of HD\,209458b}

86: 

87: Figure 1 shows the secondary eclipse of HD\,209458b at 24 $\mu$m

88: wavelength, observed using the Multiband Imaging Photometer for

89: Spitzer (MIPS, Rieke et al. 2004).  The scatter in the individual

90: points (upper panel) is about 0.008 magnitudes, poorer than is

91: possible from the ground in the visible.  However, the MIPS precision

92: is limited by statistical fluctuations in the thermal emission from

93: dust in our own solar system, the IR zodiacal light.  The Spitzer data

94: analysis pipeline provides an `error image' which quantifies these

95: statistical fluctuations.  We extract the brightness of the star using

96: an optimal weighting technique (Horne 1986), and we propagate the

97: error image thru the same procedure to derive the formal errors in the

98: photometry.  We find the scatter in our photometry to be in close

99: agreement with the formal errors.  We are background-limited by the

100: thermal zodiacal emission.

101: 

102: \begin{figure}

103: \epsfig{file=fig1.ps, width=8cm}

104: \caption{Photometric detection of the secondary eclipse of 

105: HD\,209458b using Spitzer/MIPS.  The upper panel shows all 1696

106: individual photometric points, which are binned in the lower panel to

107: show the eclipse.  The fitted eclipse curve is overplotted.  The

108: points above the eclipse are a control sequence, showing the stability

109: of the technique.  Note that the eclipse is accurately centered at

110: phase 0.5, as marked by the vertical line.}

111: \end{figure}

112: 

113: 

114: In addition to noise from the zodiacal background, there are some

115: instrument quirks in MIPS.  For example, there is a `first frame'

116: effect, where the signal from the MIPS detector is different if it has

117: just been reset.  Fortunately, we were able to use a simple trick to

118: remove the instrument quirks.  Our trick is to ratio the intensity of

119: the star to the total intensity of the zodiacal background in the

120: image.  Since the background is a large signal, its relative precision

121: is much better than the star's precision, and the ratio does not add

122: significant additional noise. Moreover, because of Spitzer's modest

123: (0.85-meter) aperture, diffraction spreads the stellar image over

124: multiple pixels, like the background.  Also, the per-pixel intensity

125: of the star is only modestly greater than the background.  To MIPS,

126: the star looks like a small patch of bright background, so the

127: instrument treats the background and stellar photons exactly the same

128: - it cannot `tell the difference'.  After normalizing the stellar

129: intensity to the background, we find the instrument quirks are gone,

130: and the noise is accurately characterized as Gaussian white noise,

131: which averages down as the square-root of the number of observations -

132: just as the text books predict.

133: 

134: The lower panel of Figure 1 averages our secondary eclipse photometry

135: into bins of 0.001 in phase, and adds error bars to the binned values.

136: Now the secondary eclipse is quite evident, and is seen to be centered

137: at phase 0.5.  We fit an eclipse curve to the data, and find a

138: best-fit depth of $0.26\pm0.046\%$, a $5.6\sigma$ detection. The

139: planet's brightness temperature is $1130\pm150$K.  Simultaneously with

140: our MIPS detection of HD\,209458b, Charbonneau et al. (2005) detected

141: TrES-1 at 8- and 4.5-$\mu$m using Spitzer's InfraRed Array Camera

142: (IRAC).  The TrES-1 measurements imply a very similar temperature

143: ($1060$K), and their 8 $\mu$m measurment has over $6\sigma$ statistical

144: significance.  Since these three data points are (so far) our only direct

145: measurements of hot Jupiters, they have elicited considerable

146: interest.  Four modeling papers have interpreted the measurements

147: (Barman et al. 2005, Burrows et al. 2005, Fortney et al. 2005, Seager

148: et al. 2005), but in Sec. 3 we will give a more simple-minded

149: interpretation, highlighting what we call the `short IR wavelength

150: question'.

151: 

152: The Spitzer detections have been described as surprising, in the sense

153: of being unexpected.  Certainly this application for Spitzer was

154: unanticipated in the early years when the observatory was being

155: developed.  But recently, the planet-to-star contrast ratio was

156: robustly predicted to be a significant fraction of one percent at long

157: IR wavelengths (Charbonneau 2003, Burrows et al. 2004).  Prior to

158: launch, we reasoned that if we couldn't detect a fraction of one

159: percent using a cryogenic telescope in space, then we should surely

160: give up!  Fortunately, the first public data from MIPS showed us that

161: Spitzer's sensitivity and stability were up to the task.  So we knew

162: in advance that we would detect HD\,209458b with MIPS, providing that

163: the planet was at least as hot as 700K - which seemed unavoidable.

164: 

165: Curiously, the Spitzer detections did not start with the best cases.

166: Since the flux from both the planet and star decrease with wavelength

167: (decline of the Planck function), 24~$\mu$m is not the best wavelength

168: for secondary eclipse detection.  The IRAC 8~$\mu$m channel is more

169: suitable, being source photon-limited, not background-limited.  But

170: for historical reasons, IRAC was first used by Charbonneau et al. for

171: TrES-1, not for HD\,209458.  So the brighter and closer system was not

172: observed at the best wavelength.  The new observing programs now

173: underway will soon fill this gap, and we will see secondary eclipses

174: of hot Jupiters even more clearly.  The new very hot Jupiter recently

175: announced by Bouchy et al. (2005) will be an especially important

176: target for Spitzer.

177: 

178: \section{The Short IR Wavelength Question}

179: 

180: \subsection{New 3.8 $\mu$m Photometry}

181: 

182: In order to discuss what we regard as the major unresolved

183: observational question concerning hot Jupiters, we need to include the

184: ground-based data. A flux peak is predicted to occur in hot Jupiter

185: spectra at 3.8 $\mu$m (Sudarsky et al. 2003).  Ground-based

186: observations can be very useful at this important wavelength, because

187: we can use a filter centered exactly on the predicted peak (the IRAC

188: bandpasses are offset).  Unfortunately, the high thermal background

189: from the ground has made broadband photometry impossible near 4~$\mu$m

190: - most instrument detector arrays saturate in quite short integration

191: times.

192: 

193: 

194: \begin{figure}

195: \epsfig{file=fig2.ps, width=8cm}

196: \caption{Window in the terrestrial atmosphere for photometry of hot 

197: Jupiters at 3.8 $\mu$m.  The solid line shows the telluric

198: transmittance at moderate spectral resolution.  The dashed curves are

199: the transmittance of the conventional L$^\prime$ filter (broad), and our

200: CVF (narrow).}

201: \end{figure}

202: 

203: 

204: In September 2003 we obtained 3.8 $\mu$m photometry of HD\,209458

205: during two secondary eclipses, using NSFCAM (Shure et al. 1994) on the

206: NASA Infrared Telescope Facility (IRTF).  We were able to avoid

207: detector saturation by using a narrow optical bandwidth, a $1.5\%$

208: circular-variable-filter (CVF).  Figure 2 shows the atmospheric

209: transmittance, and the profile of the conventional L$^\prime$ filter, and

210: our CVF at 3.8 $\mu$m.  To monitor changes in atmospheric absorption,

211: we observed a comparison star of similar brightness to HD\,209458. 

212: We have completed the analysis of one of the two eclipses.  The

213: eclipse amplitude in that case, from 329 individual 10-second

214: exposures, was $-0.0007\pm0.0014$, i.e. the system nominally becomes

215: brighter when the planet is hidden, consistent with seeing no eclipse.

216: Not surprisingly, the errors are larger in the ground-based

217: observations than with Spitzer, but nevertheless the results are

218: sufficiently precise to be somewhat puzzling, as we explain below.

219: 

220: \subsection{An Observational Perspective}

221: 

222: The two extrasolar planets currently observed by Spitzer - HD\,209458b

223: and TrES-1 - are different worlds, and they surely have their own

224: unique characteristics.  But, our ignorance of hot Jupiter spectra is

225: arguably much greater than the real differences between them. So here 

226: we compare observations of both planets to a single model.  To

227: quantify the similarity between the two planets, we assume thermal

228: equilibrium and the same Bond albedos and efficiency of heat

229: redistribution.  In the long wavelength (Rayleigh-Jeans) limit, it is

230: easy to show that the planet-to-star contrast ${c_\lambda}$ depends on

231: the stellar and orbital parameters as:

232: \begin{equation}

233: {c_\lambda} \sim  {R_p^2}{R_*^{-{\frac{3}{2}}}}{a^{-{\frac{1}{2}}}}

234: \end{equation} 

235: where $R_p$ and $R_*$ are the planetary and stellar radius, 

236: respectively, $a$ is the orbit semi-major axis, and we supress the

237: constant containing the Bond albedo.  Note that the temperature of the

238: star does not appear since it cancels when computing the

239: contrast. Using the parameters for HD\,209458 (Brown et al. 2001;

240: Wittenmyer 2005) and TrES-1 (Alonso et al. 2004), we find that the contrasts

241: for these two systems at a given wavelength would differ by only 8\%

242: under our simple assumptions.  So comparing them to a single model is

243: reasonable.

244: 

245: Figure 3 shows contrast versus wavelength for HD\,209458b and TrES-1,

246: and compares the observations to a model from Sudarsky et al. (2003),

247: as did Charbonneau et al. (2005, their Figure 3).  The dashed line is

248: the contrast from a blackody having $T=1100$K.  The Spitzer

249: observations cannot in themselves discriminate between the model curve

250: and a simple blackbody.  Considering the errors, all three Spitzer

251: observations could be consistent with either the Sudarsky et al. model

252: or the blackbody. 

253: 

254: 

255: \begin{figure}

256: %  \vspace{-0.2in}

257: \epsfig{file=fig3.ps, width=8cm}

258: \caption{Spitzer and ground-based observations of TrES-1 and

259: HD\,209458b in contrast units, compared to a single model (solid line)

260: and 1100K blackbody (dotted line).  The Spitzer IRAC and MIPS

261: bandpass funtions are plotted below.  Filled symbols are the Spitzer

262: points, and the open diamonds show the expectation value of the

263: contrast averaged over the Spitzer bands. The open squares show the

264: ground-based photometry of Snellen (2005) at 2.3~$\mu$m, and from this

265: paper at 3.8~$\mu$m.  The two horizontal lines at 2.2~$\mu$m represent

266: the limit from Richardson et al. (2003b).  Note log scale for

267: wavelength.}

268: \end{figure}

269: 

270: Now we consider the ground-based data.  The error bar on our CVF

271: photometry at 3.8 $\mu$m misses the Sudarsky et al. peak at this

272: wavelength, but it overlaps the blackbody.  The Snellen (2005)

273: photometry at 2.3 $\mu$m agrees with the model, but the error bar also

274: includes the blackbody curve.  The most stringent ground-based point

275: is the Richarson et al. (2003b) upper limit at 2.2 $\mu$m.  This is a

276: limit, derived from differential spectroscopy, on the shape of the

277: spectrum, and it is given by the two horizontal lines.  Unlike the

278: other error limits on the figure, this is a 3$\sigma$, not 1$\sigma$

279: limit.  These lines are relative limits, i.e. only the intensity

280: interval between them is significant (see Seager et al. 2005). To

281: conform to this limit, the contrast spectrum has to fit within the

282: lines. The Sudarsky et al. model doesn't fit, but the blackbody does.

283: This is the essence of the short-IR wavelength question - we don't see

284: the peaks predicted in the spectrum where planet flux should escape

285: between absorption bands, and we wonder how strong these peaks really

286: are.  Now, we do not seriously suggest that the planet is actually a

287: blackbody.  But to date, we do not have a definitive measurement of

288: departures from a blackbody spectrum in the IR.  The IRAC band at 3.5

289: $\mu$m should provide this, as evident from Figure 3.  Observations of

290: TrES-1 at 3.5 $\mu$m were made by Spitzer in September 2005, and are

291: now being analyzed by Dave Charbonneau.  We eagerly await the results.

292: 

293: %  \acknowledgments{   }

294: 

295: 

296: \begin{references}

297: \reference Alonso,~R., \& 11 co-authors 2004, ApJ 613, L153.

298: \reference Barman,~T.~S., Hauschildt~P.~H., \& Allard,~F. 2005, ApJ 632, 1132.

299: \reference Bouchy,~F., \& 11 co-authors 2005, Astr. Ap., in press.

300: \reference Brown,~T.~M., Charbonneau,~D., Gilliland,~R.~L., Noyes,~R.~W.,

301: 	\& Burrows,~A. 2001, ApJ 552, 699.	

302: \reference Burrows,~A., Hubeny,~I., \& Sudarsky,~D. 2005, ApJ 625, L135.

303: \reference Burrows,~A., Sudarsky,~D. \& Hubeny,~I. 2004, in {\it The 

304: 	Search for Other Worlds}, AIP Conf. Proceedings, 713, 143.

305: \reference Charbonneau, D. 2003, in {\it Scientific Frontiers in Research

306: 	on Extrasolar Planets}, ASP Conf. Series, 294, 449.

307: \reference Charbonneau,~D., \& 10 co-authors 2005, ApJ 626, 523.

308: \reference Deming,~D., Seager,~S., Richardson,~L.~J., \& 

309: 	Harrington,~J. 2005, Nature 434, 740.

310: \reference Fortney,~J.~J., Marley,~M.~S., Lodders,~K., Saumon,~D., \&

311: 	Freedman,~R. 2005, ApJ 627, L69.

312: \reference Horne, K. 1986, PASP 98, 609.

313: \reference Lucas,~P.~W. \& Roche,~P.~E. 2002, MNRAS 336, 637.

314: \reference Richardson,~L.~J., and 6 co-authors 2003a, ApJ, 154, 583.

315: \reference Richardsom,~L.~J., Deming,~D., \& Seager,~S. 2003b, ApJ 597, 581.

316: \reference Rieke,~G.~H. \& 42 co-authors 2004, ApJ(Suppl) 154, 25.

317: \reference Seager,~S. \& Sasselov,~D.~D. 1998, 502, L157.

318: \reference Seager,~S., Richardson,~L.~J., Hansen,~B.~M.~S., Menou,~K.,

319: 	Cho,~J.~Y.-K., \& Deming,~D. 2005, ApJ 632,1122.

320: \reference Shure,~M.~A., Toomey,~D.~W., Rayner,~J.~T., Onaka,~P.~M.,

321: 	\& Denault,~A.J. 1994, SPIE 2198, 614.  

322: \reference Snellen, I.~A.~G. 2005, MNRAS 363, 211.

323: \reference Sudarsky,~D., Burrows,~A., \& Hubeny,~I. 2003, ApJ 588, 1121.

324: \reference Wiedemann,~G., Deming,~D., \& Bjoraker,~G. 2001, ApJ 546, 1068.

325: \reference Wittenmyer,~R.~A., \& 12 co-authors 2005, ApJ 632, 1157. 

326: \end{references}   

327: 

328: 

329: 

330: \end{document}

331: 

332: