0803.1054/ms.tex
1: 
2: 
3: %% preprint produces a one-column, single-spaced document:
4:   %% manuscript produces a one-column, double-spaced document:
5:   %% preprint2 produces a double-column, single-spaced document:
6: % \documentclass[12pt,preprint]{aastex}
7: % \documentclass[manuscript]{aastex}
8: % \documentclass[preprint2]{aastex}
9: \documentclass{emulateapj}
10: 
11: \begin{document}
12: 
13: \title{Determining atmospheric conditions at the terminator of the hot-Jupiter HD209458\MakeLowercase{b}}
14: 
15: \author{David K. Sing\altaffilmark{1}}
16: \author{A. Vidal-Madjar\altaffilmark{1}} 
17: \author{A. Lecavelier des Etangs\altaffilmark{1}}
18: \author{J.-M. D\'{e}sert\altaffilmark{1}}
19: \author{G. Ballester\altaffilmark{2}}
20: \author{D. Ehrenreich\altaffilmark{3}}
21: 
22: \altaffiltext{1}{Institut d'Astrophysique de Paris, CNRS; Universit\'{e} 
23: Pierre et Marie Curie, 98 bis bv Arago, F- 75014 Paris, France; sing@iap.fr}
24: 
25: \altaffiltext{2}{Lunar and Planetary Laboratory, University of Arizona, 
26: Sonett Space Science Building, Tucson, AZ 85721-0063, USA}
27: 
28: \altaffiltext{3}{Laboratoire d'astrophysique de Grenoble, Universit\'{e} 
29: Joseph-Fourier, CNRS (UMR 5571), BP 53 F-38041 Grenoble cedex, France}
30: 
31: \submitted{Accepted to Ap.J May 6, 2008}
32: 
33: \begin{abstract}
34: 
35:  We present a theoretical model fit to the {\it HST}/STIS optical transit transmission spectrum
36: of HD209458b.  In our fit, we use the sodium absorption line profile along with
37: the Rayleigh scattering by H$_{2}$ to help determine the average temperature-pressure profile at
38: the planetary terminator, and infer the abundances of atomic and molecular species.
39: The observed sodium line profile spans an altitude range of $\sim$3,500 km, corresponding to
40: pressures between $\sim$0.001 and 50 mbar in our atmospheric models.
41: We find that the sodium line profile requires either condensation into sodium sulfide or ionization, 
42: necessary to deplete atomic sodium only at high altitudes below pressures of $\sim$3 mbar.
43: The depletion of sodium is supported by an observed sudden abundance change, from 2 times solar 
44: abundance in the lower atmosphere to 0.2 solar or lower in the upper atmosphere.
45: Our findings also indicate the
46: presence of a hot atmosphere near stratospheric altitudes corresponding to pressures of 33 mbar, consistent
47: with that of the observed dayside temperature inversion. In addition, we find a separate 
48: higher altitude temperature rise is necessary within both the condensation and ionization models, 
49: corresponding to pressures below $\sim$0.01 mbar.  
50: This hot higher altitude temperature indicates that absorption by atomic sodium can potentially 
51: probe the bottom of the thermosphere, and is possibly sensitive to
52: the temperature rise linked to atmospheric escape.
53: 
54: 
55: \end{abstract}
56: 
57: \keywords{planetary systems - radiative transfer - stars: individual(HD209458)}
58: 
59: \section{Introduction}
60: 
61: Early theoretical transit models of hot-Jupiter exoplanetary
62: atmospheres \citep{SS00,Brown01,Hubbard01} contained strong sodium absorption
63: signatures, predicted to be as high as 0.2\% for clear cloudless
64: atmospheres.  The original measured Na absorption signature in
65: HD209458b from medium resolution {\it HST} data \cite{Charby} was,
66: however, $\sim$3$\times$ smaller than initial predictions.
67: Additionally, \cite{Bar07} has also identified 
68: a potential broad band Na signature from photometry of low resolution STIS data in the same planet,
69: taken several years later \citep{Knut}.  
70: These observations have lead to multiple
71: theories to explain the small Na signature.  The most widely cited
72: explanations of reduced Na absorption include: high clouds or hazes,
73: ionization, low global Na abundance, rainout, as well as non-local
74: thermal equilibrium effects \citep{Charby,Bar02,Bar07,Fortney03}.  
75: High opaque clouds could potentially mask absorption
76: features by effectively cutting transit information for altitudes lower
77: than the cloud deck \citep{SS00,Fortney03}.  The small atmospheric signature has
78: made detection from ground-based telescopes difficult, with no Na
79: detection currently reported for HD209458b after multiple
80: attempts \citep{Narita,Arribas}.
81: 
82: In the grazing line of sight of transit geometry, the measured effective altitude 
83: depends upon the wavelength dependant opacities of the atmospheric
84: constituents giving a transit transmission spectrum its altitude sensitivity
85: (see relevant equations in \citealt{Lecavelier08a}).
86: The effective altitude is also dependant upon the abundance of
87: absorbing atmospheric particles, as well as the atmospheric pressure.
88: As the abundances are not in general a priori known, this leads to atmospheric models which have degenerate
89: solutions between modeled abundance and pressure, where an atmospheric transit model with increased
90: abundance is identical to models with lower abundances and higher pressures.
91: 
92: Here we interpret the optical {\it HST} STIS transmission spectra of HD209458b of
93: \cite{Sing08a}.  The spectra consists of STIS data obtained during planetary 
94: transit at low and medium spectral resolutions.
95: The two datasets are combined to extend 
96: the measurements over the entire optical regime, providing a
97: way to simultaneously measure, in the case of atomic Na, 
98: both the narrow line core and wide line wing 
99: absorption as well as other absorbing or scattering species.  Since the 
100: Na line core and wings are optically thick at different
101: atmospheric heights, a detailed analysis of the Na line profile in conjunction with
102: other observed species can potentially break the model degeneracies and reveal
103: atmospheric structure such as the altitude-temperature-pressure (z-T-P) relation,
104: abundance, and the possible presence and height of planetary clouds.
105: In this paper, we focus mainly on the resulting temperature pressure profile, as well
106: as the Na line itself.  We briefly describe the observations and resulting spectra in \S2, present our
107: transit model fits in \S3, report on an observed temperature inversion in \S3.3, and discuss our results
108: in \S4 \& 5. In addition to this work, \cite{Lecavelier08b} provides a detailed 
109: description of H$_{2}$ Rayleigh scattering while \cite{desert08} goes into modeling details of TiO, 
110: VO, and other molecular species.
111: 
112: \section{Observations}
113: 
114: The data reduction, procedures, and various tests resulting in the transmission spectra
115: used in this analysis are covered in detail by \cite{Sing08a}, though we 
116: briefly describe them here for completeness.
117: The {\it HST}/STIS optical transit data for HD209458b consist of data at low
118: resolution covering $\sim$3,000-10,000{\AA} with the G430L and G750L gratings
119: and medium resolution data from the G750M grating covering 5,813-6,382{\AA}.
120: 
121: With the medium resolution data, \cite{Sing08a} measured the Na absorption using
122: the double differential photometric method of \cite{Charby}, as well as a
123: double differential spectroscopic evaluation.  For the differential
124: photometric measurement, \cite{Sing08a} selected 23 wavelength bands from 4 to 100
125: \AA\ centered on the Na~D lines, \cite{Charby} used three different
126: wavelength bands, and compared the absorption to that of regions
127: red-ward and blue-ward of the Na~D line region.  The resulting
128: photometric Na absorption profile (see Sing et al. 2008, Fig. 3) matches
129: the original results of \cite{Charby} and, in addition, reveals the strong
130: absorption from the Na~D line centers. 
131: 
132: To analyze the low and medium resolution {\it HST} spectra together, \cite{Sing08a} used a
133: limb-darkened corrected transit spectral ratio to probe the full
134: wavelength dependence of atmospheric absorption features (see Fig. 1 \& Fig. 2).  From this
135: analysis, the Na~D wavelength region shows an absorption plateau between
136: $\sim$5600-6700 {\AA}, which is $\sim$0.045\% above the minimum absorption
137: level observed (see Fig. 1).  Note that the wavelength range of the medium
138: resolution data, originally used to detect atmospheric Na \citep{Charby}, 
139: is entirely contained within this broad plateau
140: absorption region.  The total absorption seen in
141: the Na~D line cores, therefore, is the result from the medium
142: resolution data, which is the absorption above the plateau region
143: measured to be $\sim$0.065\%, {\it plus} the absorption level of the plateau
144: region itself.  This total Na core absorption level is $\sim$0.11\%, within a 4.4{\AA} band,
145: and is much closer to the original predictions \citep{SS00,Brown01},
146: indicating that a large extent of the atmosphere is being probed
147: during transit.  
148: Additionally, the Na~D doublet is resolved
149: in our resulting spectra which shows two strong narrow
150: absorption lines (see Fig. 2).
151: 
152: 
153: \subsection{Observed Na line profile}
154: 
155: The observed Na~D transit line profile (Fig. 1), with it's narrow
156: absorption core and broader plateau-like shape, is in stark contrast
157: to cloudless theoretical models of
158: HD209458b \citep{SS00,Brown01,Fortney03,Bar07} which have smoothly decaying
159: Lorentzian ``Eiffel Tower'' shaped line wings.  This more complicated
160: observed line profile suggests a very low quantity of Na is present in
161: the middle atmosphere and much larger quantity in the lower
162: atmosphere, pointing to either high altitude Na condensation into Na$_2$S or
163: alternatively, depletion via ionization.
164: 
165: Previous studies have predicted models of HD209458b which
166: included Na$_2$S clouds \citep{Iro05}, as high altitude temperatures on
167: the night side reach levels cool enough for condensation of atomic
168: Na \citep{Lodd99,Show02}.  Sodium sulfide would effectively
169: decrease the amount of atomic sodium at altitudes above the
170: condensation temperature, creating a stratified-like atmospheric profile
171: depleted of atomic Na only at those high altitudes.  The transit 
172: signature with such a layered Na profile
173: produces a broad plateau shaped line wing along with narrow absorption
174: cores rising above the plateau to higher levels \citep{Iro05}.  
175: Na$_2$S would have to be transparent, since they are not observed in absorption during
176: transit at the limb.  If Na$_2$S or other clouds/hazes were opaque in absorption during transit
177: at the condensation altitude, significant absorption would extend over a very broad 
178: wavelength range \citep{Fortney05}, masking the low resolution absorption features we observe.
179: Such a flat, largely featureless transmission spectra has recently been observed
180: in HD189733b, the result of high altitude haze \citep{Pont08, Lecavelier08a}.
181: 
182: An alternative method to deplete neutral atomic sodium is by ionization \citep{Fortney03}.
183: In this scenario, UV photons between wavelengths of 1,107 and 2,412 {\AA} (corresponding to 
184: 11.2 and 5.14 eV respectively) photoionize sodium along the transit line of sight until recombination
185: matches ionization.  \cite{Fortney03} found a sharp boundary at the limb between higher atmospheric regions of complete
186: ionization and lower regions with no ionization.  Such a model could explain the observed Na line shape, with a narrow Na absorption 
187: core rising above broad line wings, if the atmosphere suddenly becomes ionized above a certain altitude.
188: 
189: 
190: \section{Model Fits to the transit transmission spectra of HD209458b}
191: 
192: We modeled the atmospheric transmission spectra calculating the
193: transit signature based on a theoretical model \citep{dav} containing absorption by
194: Na, H$_2$ Rayleigh scattering, TiO and VO.  The model
195: output absorption spectrum can then be directly compared to our
196: limb-darkened corrected transit absorption spectrum. 
197: We calculated the optical depth along a grazing line of sight through the planet's atmosphere, 
198: and calculated the absorption depth at a given wavelength using 
199: equations given in \cite{Lecavelier08a}. For the Na D 
200: lines, we used collision-broadened line-shapes calculated using 
201: Equations of \cite{Iro05}. Concerning atomic sodium, 
202: ultimately the free parameters to fit in the transit calculation 
203: include the atmospheric temperature profile as a function of altitude, and the mixing ratios of 
204: Na below and above the condensation or ionization level (see Fig. 3). 
205: 
206: We found two families of solutions, with each giving similarly good fits to the data.  In one model, Na is
207: allowed to condense into sodium sulfide at the chemical equilibrium temperature, while in the other model neutral Na is depleted at high altitudes,
208: presumably by ionization, below an arbitrary pressure.
209: 
210: 
211: \subsection{Condensation model}
212: \subsubsection{Temperature-Pressure profile}
213: 
214: For the calculation,
215: we fit various simplified T-P profiles in which
216: the temperature is taken as a linear function of the altitude and
217: pressure is calculated using hydrostatic equilibrium.  
218: Note that it is not specifically the T-P profile shape itself which is important in our fitting of
219: the data, but rather the different temperature and pressure regimes imposed by the observations.  At
220: our signal-to-noise levels, the general variation of the T-P profile in between the different atmospheric regimes
221: we find (hot at high pressure, cool at middle pressures, and hot at low pressures) is only loosely constrained.
222: A more complicated T-P profile would add undesired additional parameters into the fit, leading us to choose
223: as simple a T-P profile as possible.
224: 
225: We found it necessary to have a minimum of three variable points in the
226: T-P profile. The temperature as a function of the altitude, $T(z)$, is
227: thus composed of two linear functions below and above the altitude
228: $z_{m}$, where $z_{m}$ is the altitude of the minimum temperature,
229: $T_{m}$.  We took the lower reference altitude, $z_{s}$, for the
230: altitude corresponding to the mean planetary radius \citep{Knut}, and the
231: altitude $z_{th}$ is the fixed position, at high altitude, needed to
232: define the second part of the $T(z)$ linear function. We used $z_{th}$
233: corresponding to the altitude of the absorption depth of 1.55\%, which
234: corresponds to the mean absorption depth in the core of the sodium
235: doublet.  The temperature and pressure at the altitude $z_{s}$, the
236: altitude $z_{m}$ of minimum temperature $T_{m}$, and $T_{th}$ are
237: the five free parameters defining the T-P profile.  The sodium
238: abundance in the lower and the upper atmosphere are the two other free
239: parameters in the fit. 
240: 
241: In our condensation fit we found:
242: (1) a high temperature of 2,200$\pm$260 K at 33$\pm$5 mbar  ($\sim$500 km)
243: mainly constrained by Rayleigh scattering \citep{Lecavelier08b} seen at 
244: wavelengths shorter than $\sim$5,000 {\AA}, 
245: (2) a low temperature of 420$\pm$190 K at 0.63$\pm$0.27 mbar needed to explain the plateau at 0.045\% ($\sim$1,500 km)
246: above the minimum absorption level as a Na abundance change due to condensation, and finally 
247: (3) a high altitude ($\sim$3,500 km) high temperature of 1,770$^{+500}_{-570}$ K at 0.0063$\pm$0.003 mbar needed to 
248: fit the strong peak and narrow width of the cores of the Na~I doublet.
249: 
250: The pressure-temperature gradient we find when imposing condensation is close to and consistent (within 1$\sigma$)
251: of an adiabatic gradient.  A temperature gradient which is adiabatic or less is needed to be stable against 
252: convection, and solutions imposing adiabatic variations give acceptable fits to the data (an adiabatic fit is shown in Fig. 3).
253: We ultimately, however, choose to quote values found from the more general solution, as the linear temperature-altitude relation 
254: we assume includes adiabatic gradients and the general fit is located at a well behaved minimum $\chi^2$.  Although the
255: best fit solution is slightly superadiabatic, the relatively low quality of the data still allows for a wide range of T-P gradients and
256: the adiabatic-like profile found from our general fit indicates that an adiabatic behavior is likely at those altitudes.
257: 
258: 
259: \subsubsection{Atmospheric Na Condensation, TiO \& VO}
260: 
261: Our model contained two different amounts of atomic Na having separate variable
262: parameters for the mixing ratios of the middle and lower
263: atmospheres.  The pressure boundaries (or condensation pressure) where
264: the quantity changes occur is taken to be at an altitude where the
265: temperature profile crosses the Na$_2$S condensation curve.  In our
266: fits, we found a Na mixing ratio of 3.5$^{+2.9}_ {-1.9}\times10^{-7}$
267: in the middle atmosphere and 4.3$^{+2.1}_{-1.9}\times10^{-6}$ in the
268: lower atmosphere, corresponding to 0.2 and 2 times solar abundance.
269: 
270: \cite{Fortney05} estimated that
271: the optical depth of Na$_2$S condensates in transit geometry could be optically thick for HD209458b ($\tau\sim$0.73) for an atmosphere
272: of solar composition and cloud base at 30 mbar.  Given the fit T-P profile, we find a Na$_2$S cloud base of $\sim$3 mbar.
273: Lower pressures or larger particle sizes would provide a natural explanation for transparent clouds, as the condensate optical
274: depth is proportional to the pressure and inversely proportional to the particle size \citep{Marley2000}.  Given our determined Na abundances, the mixing ratio
275: of Na$_2$S we find is (2$\pm1$)$\times10^{-6}$, which matches the assumed value of \cite{Fortney05}.
276: Scaling the estimated optical depth of \cite{Fortney05}, a cloud base at 3 mbar gives $\tau\sim$0.073, indicating that transparent Na$_2$S 
277: condensates are plausible.  
278: 
279: The continuum blue-ward of the Na~D lines, between $\sim$4,000 and 5,500
280: \AA, shows a slightly lower absorption than the red-ward region,
281: 7,000-8,000 \AA, and a significant absorption `bump' around $\sim$6,250\AA\ which 
282: appears in both resolutions.  
283: Our fits indicate that the red-ward absorption levels (7,000-8,000 {\AA})
284: are likely due to TiO and VO at stratospheric heights, being largely confined to altitudes
285: below our Na condensation level (also see \citealt{desert08}).  The TiO and VO features mask part of the long wavelength
286: Na line wing, though not the short wavelength side, providing strong constraints on
287: the abundance and altitude of TiO and VO.  The hot low altitude temperatures we find from Rayleigh scattering
288: are consistent with the presence of TiO and VO, as both species begin to condense out
289: of the atmosphere at temperatures below $\sim$1,800 K. 
290: 
291: The low-altitude pressure assumed here is dependent on the detection of
292: Rayleigh scattering \citep{Lecavelier08b}, though 
293: multiple interpretations of the absorption rise at those wavelengths
294: exist \citep{Lecavelier08b,Ballester07,Bar07}.  Ultimately, it may be difficult
295: without Rayleigh scattering to assign an absolute pressure scale to
296: these measurements which are inherently differential. 
297: The effective planetary radius seen during transit is sensitive to the quantity of
298: the atmospheric pressure times the abundance of absorbing species (see 
299: equations of \citealt{Lecavelier08a}), making 
300: an atmospheric fit at higher altitude possible by simply increasing the absorbers abundance.
301: The advantage of detecting Rayleigh scattering by H$_2$ when interpreting these models
302: is that the abundance of H$_2$ is known, constituting the bulk of the atmosphere, which thereby fixes
303: the pressure, and defines the pressure scale for the rest of the transmission spectrum. 
304: Our adoption of Rayleigh scattering here is based on the goodness of
305: fit to the data and the necessity to probe a large pressure range, 
306: given the total Na profile observed.
307: 
308: Since the low-altitude pressure is an upper limit
309: constrained by Rayleigh scattering, a lower pressure
310: would have an effect on the lower temperature profile, increase the
311: lower Na mixing ratio, and decrease the condensate pressure determined
312: in our transit fit.  Whatever the true pressure $P_{s}$ or additional
313: absorption sources, however, the basic atmospheric characteristics of Na
314: determined here would remain the same.  Namely, the middle
315: atmosphere has a greatly reduced amount of Na with a temperature below
316: Na$_2$S condensation, while the lower and higher atmospheres are
317: warmer than the condensation temperature with the lower atmosphere
318: containing a solar-like Na abundance and higher one a
319: significantly reduced Na abundance.
320: 
321: 
322: \subsection{Photoionization model}
323: 
324: Given the large uncertainties modeling hot-Jupiter Na photoionization, we choose not to employ a complex 
325: ionization model, but rather follow the results of \cite{Fortney03} and assume in our model that 
326: neutral Na can be suddenly ionized above a given altitude leading to the observed Na depletion.  
327: The ionization depth found by \cite{Fortney03} is around 1.7 mbar, close to the pressure we find for a Na
328: abundance change in the condensation model, indicating the effects could be important at those pressures
329: and such a model is viable.
330: 
331: A model with Na ionization effectively lifts the constraint of the temperature profile dropping below the Na
332: condensation curve, which the condensation model uses to deplete atomic Na.  Effectively, a model with Na 
333: ionization leaves a wide range of temperatures possible
334: in the middle atmosphere (corresponding in our model to $T_{m}$) as the collisional broadened Na profile at those
335: altitudes (1,000 - 2,000 km) contains no temperature information.  The corresponding H$_2$ Rayleigh scattering slope 
336: (below $\sim$3,400 {\AA}) is at the edge of the spectrum and also provides no useful middle atmospheric temperature constraints.  Within the framework of 
337: equilibrium chemistry, the temperature $T_{m}$ is likely below
338: that of the TiO/VO condensation curve and above that of the Na$_2$S condensation curve.
339: However, Rayleigh scattering still implies a high temperature of 2,200$\pm$260 K at 33$\pm$5 mbar and
340: the deep narrow Na cores suggest high altitude temperatures of around 1,770$^{+500}_{-570}$ K.
341: Without a strong constraint on the middle atmosphere temperature, T-P profile fits are dominated
342: by these upper and lower atmospheric constrains, leading to anomalously high temperature T-P profiles fit to be everywhere above 2,000 K.  
343: Though technically good fits to the data, such profiles can be ruled out as the temperatures would
344: be everywhere higher than the dayside temperature inversion seen with Spitzer \citep{Knut08,Burrows07}.
345: Additionally, the lack of observed Ca I at $\lambda$4,227 {\AA} in the low resolution data further suggests 
346: temperatures reach values lower than that required for Ca-Ti and Ca-Al-bearing condensates \citep{Lodd06}.
347: 
348: Though the current data lacks enough constraints for a unique T-P profile solution in the photoionization model, 
349: the ionization pressure, $P_{ion}$, can be estimated given the likely range of temperatures.  We fit various isothermal models fitting $P_{ion}$ 
350: as free parameter along with the Na abundances (taken to be constant) above and below that pressure.  The 
351: resulting ionization pressure varies from 0.5 mbar to 5 mbar for temperatures of 1,100 and 2,000 K respectively.
352: Higher temperature fits result in smaller upper-atmospheric Na abundances and larger Na abundance contrasts.
353: The abundances found in the lower temperature fits are in general agreement with those found in the Na condensation model, while
354: the upper temperature fits find much smaller upper-abundances, with the 2,000 K model giving an upper Na abundance 180 times less than solar. 
355: Thus, a Na abundance drop by a factor of ten can be considered as a lower limit, and may be much larger depending on the depletion
356: mechanism.
357: 
358: \subsection{High altitude temperature inversion}
359: 
360: A $\sim$0.11\% absorption of Na indicates that our transmission spectra probes an altitude range of
361: $\sim$3,500 km.  Such a large altitude range with our slant transit geometry indicates we are sensitive to pressures 
362: between 10's of mbar and $\sim$0.001 mbar.  
363: These limb pressures are in agreement with those found from \cite{Fortney03}, who analyzed the original \cite{Charby} 
364: medium resolution Na measurements.
365: At pressures higher than 30-50 mbar, our transmission spectra is optically thick to Rayleigh scattering by the 
366: abundant molecule H$_{2}$, providing a firm upper pressure limit.
367: %while at pressures
368: %below $\sim$0.001 mbar, Na is likely {\bf completely} ionized by stellar UV radiation \citep{Fortney03}.  
369: Such pressure restrictions, in addition to the strong narrow Na line cores, facilitate the necessity of a hot high-altitude 
370: temperature (above $\sim$1,500), which then naturally provides the observed altitude range due to an increased scale height.  
371: Our high altitude high limb temperature may point toward the atmospheric escape mechanism being felt, which necessitates large temperature
372: rises to perhaps 10,000 or 15,000 K in the exosphere \citep{Yelle,Vidal02,Vidal03}.
373: 
374: The highest altitudes probed here are dependent on the Na core, however, and thus have 
375: a correspondingly larger associated error.  The large error coupled with the possible Na depletion mechanisms
376: leaves open a wide range of possible upper atmospheric temperature profiles.
377: However, a high altitude temperature rise can be seen when analyzing the narrowest Na wavelength bands alone, sensitive
378: to high altitudes well above the condensation/ionization altitude, making the high-altitude temperature measurement
379: independent of the depletion mechanism.  The hot high-altitude temperature is illustrated in Fig. 4, which shows photometric Na measurements
380: from \cite{Charby} and \cite{Sing08a} overplotted with various constant abundance isothermal models.  
381: A single isothermal model fits through the 12 and 17 {\AA} bands, indicating that
382: the temperature and Na abundance do not change significantly between those altitudes ($\sim$2,000 to 2,200 km).
383: However, the absorptions measured in the 4.4 and 8.8 {\AA} bands, which probe higher in altitude ($\sim$2,700 to 3,700 km), 
384: rise much faster than the isothermal model fitting the lower altitudes of the 12 and 17 {\AA} bands, indicating that warmer temperatures
385: are needed.  
386: 
387: In the ionization Na depletion scenario, there are less temperature constraints for the middle atmosphere compared to the Na condensation model. 
388: However, limb temperatures everywhere above $\sim$1,500 K are unlikely, and in contradiction with Spitzer measurements of infrared brightness temperature.  
389: T-P profiles applicable to Na ionization models have less severe temperature rises compared to the condensation model, 
390: though a hot high-altitude temperature is still necessary and at least a modest temperature rise probable.
391: %However, the temperature change can be determined and is estimated to increase by 900$\pm$250 K over 1,400$\pm$200 km from the 12 and 4.4 {\AA} measurements alone.
392: 
393: A similar high altitude temperature rise will also be likely necessary
394: to explain recent measurements of another hot-Jupiter, HD189733b.  The Na core in that hot-Jupiter
395: reaches $\sim$4,950 km above the average optical transit altitude \citep{Redfield, Pont07}.
396: As the atmospheric scale height of HD189733b is smaller than HD209458b at the same temperature, 
397: a more extreme high altitude temperature rise is likely necessary, however, to fully explain the 
398: existing measurements.
399: 
400: \subsection{Comparison with Spitzer anti-transit data}
401: 
402: A stratospheric temperature inversion has been used to interpret Spitzer infrared secondary
403: eclipse measurements of HD209458b \citep{Knut08}, 
404: thought to be caused by an absorber at optical wavelengths 
405: on the day-side of the planet \citep{Burrows07, Fortney07}.  
406: This optical absorber has been postulated to be clouds or molecules TiO 
407: and VO \citep{Hubeny03, Burrows07, Fortney07}, with a low albedo 
408: ruling out some species, though not TiO or VO \citep{Rowe,Burrows07}. 
409: 
410: Comparing our limb T-P profile to the day-side profile from \cite{Burrows07}
411: (see Fig. 3), the atmospheric pressures probed are similar, between 0.01 and 30 mbar,
412: despite our slant transit viewing geometry \citep{Fortney05}.
413: The atmosphere in the infrared has a higher opacity,
414: due to strong molecular transitions from species such as water \citep{Burrows07},
415: placing the infrared surface of the planet viewed at normal angles at similar
416: altitudes as those probed in slant geometry in the optical at the limb. 
417: 
418: The high altitude temperature inversion we observe at the limb below pressures of $\sim$1 mbar 
419: is separate from that of the dayside and occurs at a much higher altitude. 
420: However, the altitude of TiO and VO we find from both models, largely confined to pressures above $\sim$10 mbar, 
421: closely matches the model presented by \cite{Burrows07}, 
422: who placed the stratospheric absorber at pressures of $\sim$25 mbar and imposed heat 
423: redistribution between pressures of 10 and 100 mbar.  The probable presence of TiO and VO in our spectrum along
424: with the altitude similarity implies TiO and VO 
425: are indeed likely to be the stratospheric day-side temperature inversion instigators \citep{Fortney07}.  
426: As our fit of TiO and VO is based largely on low resolution data, a more robust detection and interpretation of TiO/VO will 
427: likely require higher resolution {\it HST} spectra than those used here,
428: such that multiple specific molecular bandheads could be observed.
429: 
430: Some form of heat redistribution at or near our low-altitude pressure ($\sim$30 mbar) would likely be needed to 
431: explain the high temperature we find there, as the temperatures there are significantly above that predicted by
432: radiative equilibrium \citep{Showman08}.  
433: Initial Spitzer mid-infrared measurements 
434: of the brightness temperature as a function of orbital
435: phase have revealed that at least 30\% of the energy from the stellar flux must be 
436: redistributed to the night side to explain their null results, implying efficient heat redistribution \citep{Cowan}. 
437: Comparing with models by \cite{Burrows07,Burrows07b} with
438: higher P$_{n}$s ($>$0.35), night
439: side temperatures in the region of imposed heat distribution (corresponding to our pressure at $z_{s}$) 
440: can have larger temperatures than the day side, perhaps explaining our high temperature at $z_{s}$.  
441: Alternatively, systematic pressure scale differences between the
442: two studies may indicate that our transmission spectra instead probes the top of the stratospheric temperature inversion,
443: giving a potential explanation of the high temperature at $z_{s}$ as those
444: altitudes could already be warm.  
445: The minimum temperature we find, however, is in general agreement with radiative equilibrium, perhaps implying the
446: approximation is valid at those pressures around 1 mbar.
447: Such scenarios can be further distinguished in future studies,
448: with tighter constraints on the average infrared and optical opacities, additional infrared observations, along 
449: with improved global circulation models.  
450: The presence of a temperature inversion, TiO and VO along with heat redistribution 
451: would seem to currently place HD209458b in a 
452: intermediate hot-Jupiter class between the pM and pL classes as proposed by \cite{Fortney07}.
453: 
454: \section{Discussion}
455: 
456: The atmospheric temperatures we find, within the condensation scenario, and presence of TiO and VO can also explain other HD209458b
457: observational features.  For instance, the low middle atmospheric
458: temperature and the presence of TiO and VO would also explain the
459: lack of observed potassium (K), originally predicted to be a major contributor to the optical
460: opacity.   At the lower temperatures of the middle atmosphere, K condenses into potassium chloride
461: allowing KCl to become a dominant K-bearing gas.  Condensed K would then naturally deplete the atmosphere
462: of atomic K, as seen here with Na, leading to reduced signatures.
463: In the lower warmer atmosphere, where wide atomic
464: K line wings would be observable even at low resolution, TiO and VO likely make up the
465: surrounding continuum further masking the signature.  
466: If Na is instead ionized, the minimum limb temperature is higher and ionization of K, which is easier than Na to
467: ionize, likely depletes neutral K.  
468: If present in the atmosphere, a positive detection of atomic K 
469: in HD209458b will probably require high resolution ground-based or medium resolution space-based 
470: data to detect the K line core.  
471: Further detections or upper-limits of other atomic species which have different ionization potentials and condensation
472: temperatures can offer clues in future studies when distinguishing between the two depletion scenarios.
473: 
474: The Na mixing ratio we find for the lower atmosphere is about two times
475: solar while the abundance in the middle and upper atmosphere is found
476: to be 0.2 times solar or less, thus our models have decreased Na
477: abundance without the need for a global deficiency of Na.
478: In chemical equilibrium, Na$_2$S condensation effectively removes all the Na below the condensation
479: temperature, since the protosolar Na abundance is $\sim$13\% that of S, limiting 
480: the amount of Na$_2$S which can condense directly to the amount of Na.  
481: Our condensation model T-P profile crosses the condensation curve again at higher altitude, making
482: the altitudes of our minimum temperature a ``cold trap'' for Na.  It is currently unclear
483: how the 2$^{nd}$ crossing of the condensation curve could affect the Na abundance, as the
484: S/N limits us to fit only a single value for the whole upper Na abundance.
485: 
486: Assuming equilibrium chemistry, our temperature profile
487: would also result in less CO in the middle atmosphere, as our
488: temperature profile at that level is closer to the CO = CH$_4$ curve
489: which would result in an enhancement of CH$_4$ and H$_2$O over that of
490: CO and H$_2$ (see Fig. 2).  This is consistent with the lack of CO
491: detection in HD209458b during transit \citep{Deming}, although deviations from
492: equilibrium have been shown to push the CH$_4$/CO ratio to small
493: values for the pressures probed here of less than 1
494: bar \cite{Cooper06}.
495: 
496: Atmospheric data resulting from transit measurements represent the
497: average atmospheric properties across the observed limb of the planet,
498: which would include the morning, evening, and pole terminators.  As
499: previously recognized by \cite{Iro05}, temperature and structure differences could result
500: in differing transit signatures between the morning and evening terminators.
501: Longitudinal differences might also play a role, though the transit
502: geometry generally limits significant opacity contributions to regions of the terminator itself, as
503: the density drops rapidly from the terminator along the transit line of sight \citep{Hansen08}.
504: A T-P profile decreasing in temperature and pressure with increasing altitude would further 
505: mitigate longitudinal differences, corresponding to pressures above 1 mbar in our condensation model fit.
506: Efficient heat redistribution \citep{Cowan} implies all these differences 
507: could be minimal, though future Spitzer observations and the next
508: generation of global climate models would be able to better quantify the differences.
509: Given the size of our temperature uncertainties, limb differences of even several hundred
510: degrees would have a negligible impact on our resulting model fit.
511: 
512: Primary transit data alone is currently not able to give detailed knowledge on atmospheric
513: differences between the morning and evening terminators.  Such transit signatures
514: would appear during ingress and egress, where the transit light curve is less
515: sensitive to changes in planetary radii and more sensitive to errors in limb-darkening.
516: Nevertheless, as our Na~D line cores remain relatively narrow in the medium
517: resolution data, sensitive to the middle and upper atmospheres, a solar abundance
518: of Na on the potentially warmer evening terminator in the middle and upper atmospheres
519: is ruled out.  For the condensation model, this would indicate that Na settles from the middle
520: atmosphere and remains at lower altitudes, presumably at the Na$_2$S
521: condensation altitude.  Condensed Na$_2$S would have little impact on anti-transit
522: observations and the observational constraint of a low
523: albedo \citep{Rowe}.  On the day side, the hotter temperatures would
524: sublimate Na$_2$S back into their atomic constituents, and are
525: likely transparent in normal viewing geometry anyway \citep{Fortney05}.  
526: 
527: \section{Conclusions}
528: 
529: As seen here, fitting primary transit transmission spectra can give detailed atmospheric information
530: over large altitude ranges, and thus large pressure regimes, where different physical processes
531: may be important.  In our case, strong sodium absorption coupled with a relatively 
532: clear atmosphere provides this large altitude range and reveals the pressure where Na depletion via condensation or ionization takes place. 
533: The depletion level naturally provides
534: a contrast level between the warmer deeper atmosphere, likely sensitive to global circulation,
535: and a cooler middle atmosphere where many atomic and molecular species have condensed out and/or 
536: are depleted via ionization.  
537: As strong opacity in the core of the Na line permits measurements up to very low pressures, allowing Na
538: to be sensitive to altitudes perhaps as high as the lower thermosphere.  The slant transit geometry and high 
539: associated column density restricts the lowest altitudes probed, however, limiting this study to 
540: altitudes at and above the stratosphere.
541: 
542: Interpreting transit spectra becomes easier with increased wavelength coverage and increased
543: resolution.  For HD209458b, low resolution data spanning the optical reveals a planetary continuum 
544: comprised of H$_2$ Rayleigh scattering, sodium absorption, and TiO/VO giving a self consistent picture
545: of a hot atmosphere at pressures around $\sim$10 mbar.  This result is consistent with infrared
546: Spitzer emission features and a temperature inversion at those altitudes.  The medium resolution data places
547: the detection of Na absorption on firm ground, as there is sufficient S/N and resolution to both resolve the Na
548: doublet and observe the depletion level.  This overall atmospheric picture is not possible without data
549: from all three spectroscopic gratings and highlights the importance of future observations at other wavelengths
550: and resolutions.  
551: 
552: Transit transmission spectra have become a powerful method to probe the abundances of atomic and molecular species
553: for hot-Jupiters, as seen here for HD209458b and in \cite{Swain08} for HD189733b.  Both Jupiter and Saturn are enhanced in metals
554: relative to the sun, as would seem to be the case here for HD209458b having a sodium abundance about twice that of it's parent star.
555: Such studies can begin to answer the question of whether or not all hot-Jupiters are enriched in heavy elements,
556: shedding light on giant planet formation. 
557: 
558: After the next servicing mission, {\it HST} will be capable of extending the transmission 
559: spectra of HD209458b from the UV to NIR, potentially linking important physical atmospheric 
560: processes such as the altitude dependence of UV/NUV energy deposition as well as abundance and altitudes of other 
561: atomic and molecular species.
562: As currently a patchwork of such observations exist for the two ideal targets HD209458b and HD189733b, future observations
563: should focus on covering wavelength and resolution gaps, which can lead to both theoretical and observational 
564: links between the separate studies.
565: 
566: \acknowledgments
567: D.K.S. is supported by CNES.  This work is based on observations
568: with the NASA/ESA Hubble Space Telescope, obtained at the
569: Space Telescope Science Institute (STScI) operated by AURA, Inc.  
570: Support for this work was provided by NASA through a grant from the STScI.
571: We thank G. Tinetti, F. Pont, G. H\'{e}brard, R. Ferlet, F. Bouchy and N. Allard for discussions and insight.
572: We also like to especially thank our referee Dr. Jonathan Fortney for his insightful comments and valuable suggestions,
573: especially concerning sodium ionization. 
574: 
575: \begin{thebibliography}{}
576: %\bibitem[Allende Prieto(1999)]{Allende99}Allende Prieto, C. \& Lambert, D. L. 1999, \aap, 352, 555-562 
577: \bibitem[Arribas et al.(2006)]{Arribas}Arribas, S., Gilliland, R. L., Sparks, W. B., Lopez-Martin, L. , Mediavilla, E. \& Gomez-Alvarez, P. 2006, \pasp, 118, 21-36
578: %\bibitem[Aufdenberg et al.(2005)]{Aufdenberg05}Aufdenberg, J. P., Ludwig, H.-G., \& Kervella, P. 2005, \apj, 633, 424-439
579: \bibitem[Ballester, Sing, \& Herbert(2007)]{Ballester07}Ballester, G. E., Sing, D. K. \& Herbert, F. 2007, Nature, 445, 511-514
580: \bibitem[Barman et al.(2002)]{Bar02} Barman, T. S., Hauschildt, P. H., Schweitzer, A., Stancil, P. C., Baron, E. \& Allard, F. 2002, \apj, 569, L51-L54 
581: \bibitem[Barman(2007)]{Bar07} Barman, T. S. 2007, \apj, 661, L191-L194 
582: %\bibitem[Bertone et al.(2004)]{Bertone04}Bertone, E., Buzzoni, A., Chavez, M., \& Rodriguez-Merino, L. H. 2004, \aj, 128, 829-841 
583: %\bibitem[Bigot et al.(2006)]{Bigot06}Bigot, L., Kervella, P., Thevenin, F., Segransan, D. 2006, \aap, 446, 635-641 
584: \bibitem[Brown(2001)]{Brown01}Brown, T. M., 2001 \apj, 553, 1006-1026 
585: %\bibitem[Burrows, Marley \& Sharp(2000)]{Burr}Burrows, A., Marley, M., \& Sharp. C. M., 2000, \apj, 531, 438-446 
586: %\bibitem[Burrows et al.(2006)]{Burrows06}Burrows, A., Sudarsky, D. \& Hubeny, I. 2006, \apj, 650, 1140-1149
587: \bibitem[Burrows et al.(2007a)]{Burrows07}Burrows, A., Hubeny, I., Budaj, J., Knutson, H. A., Charbonneau, D. 2007, \apjl, 668, 171-174
588: \bibitem[Burrows et al.(2007b)]{Burrows07b}Burrows, A., Budaj, J., Hubeny, I. 2007, \apj, arXiv:0709.4080
589: %\bibitem[Castelli et al.(1997)]{Castelli97}Castelli, F., Gratton, R. G., \& Kurucz, R. L. 1997, \aap , 318, 841-869 
590: \bibitem[Charbonneau et al.(2002)]{Charby}Charbonneau, D., Brown, T. M., Noyes, R. W. \& Gilliland, R. L. 2002, \apj, 568, 377-384 
591: %\bibitem[Claret(2000)]{Claret00}Claret, A. 2000, \aap , 363, 1081-1190 
592: \bibitem[Cowan, Agol, \& Charbonneau(2007)]{Cowan}Cowan, N. B, Agol, E. \& Charbonneau, D. 2007, MNRAS, 379, 641-646
593: \bibitem[D\'{e}sert et al.(2008)]{desert08}D\'{e}sert, J.-M., Vidal-Madjar, A., Lecavelier des Etangs, A., Sing, D. K., H\'{e}brard, G., Ferlet, R. 2008 \aap, in prep
594: \bibitem[Cooper \& Showman(2006)]{Cooper06}Cooper, C. S. \& Showman A. P. 2006, \apj, 649, 1048-1063 
595: \bibitem[Deming et al.(2005)]{Deming}Deming, D., Brown, T. M., Charbonneau, D., Harrington, J \& Richardson, L. J. 2005, \apj, 622, 1149-1159 
596: \bibitem[Ehrenreich et al.(2006)]{dav}Ehrenreich, D., Tinetti, G., Lecavelier Des Etangs, A., Vidal-Madjar, A. \& Selsis, F. 2006, A\&A, 448, 379-393
597: %\bibitem[Fisher \& Valenti(2005)]{Fisher05}Fisher, D. A. \& Valenti, J. 2005, \apj, 622, 1102-1117 
598: \bibitem[Fortney et al.(2003)]{Fortney03}Fortney, J. J., Sudarsky, D., Hubeny, I., Cooper, C. S., Hubbard, W. B., Burrows, A. \& Lunine, J. I. 2003, \apj, 589, 615-622 
599: \bibitem[Fortney(2005)]{Fortney05}Fortney, J. J. 2005, MNRAS, 364, 649-653
600: %\bibitem[Fortney et al.(2005)]{Fortney05}Fortney, J. J., Marley, M. S., Lodders, K., Saumon, D. \& Freedman, R. 2005, \apj, 627, L69-L72
601: \bibitem[Fortney et al.(2007)]{Fortney07}Fortney, J. J., Lodders, K., Marley, M. S., Freedman, R. S. 2007, \apj, arXiv:0710.2558
602: %\bibitem[Horne(1986)]{Horne86}Horne, K. 1986, \pasp, 98, 609-617 
603: \bibitem[Hansen(2008)]{Hansen08}Hansen, B. M. S., 2008, arXiv:0801.2972
604: \bibitem[Hubbard et al.(2001)]{Hubbard01}Hubbard, W. B., Fortney, J. J., Lunine, J. I., Burrows, A., Sudarsky, D., Pinto, P. 2001, \apj, 560, 413-419
605: \bibitem[Hubeny, Burrows \& Sudarsky(2003)]{Hubeny03}Hubeny, I., Burrows, A., \& Sudarsky, D. 2003, \apj, 594, 1011
606: \bibitem[Iro, B\'{e}zard, \& Guillot(2005) ]{Iro05}Iro, N., B\'{e}zard, B. \& Guillot, T. 2005, A\&A, 436, 719-727
607: \bibitem[Knutson et al.(2007)]{Knut}Knutson, H. A., Charbonneau, D., Noyes, R. W., Brown, T. M. \& Gilliland, R. L. 2007 \apj, 655, 564-575
608: \bibitem[Knutson et al.(2008)]{Knut08}Knutson, H. A., Charbonneau, D., Allen, L. E., Burrows, A. \& Megeath S. T 2008, \apj, 673, 526-531 
609: \bibitem[Lecavelier et al.(2008a)]{Lecavelier08a}Lecavelier des Etangs, Pont, F., A., Vidal-Madjar, A., Sing, D. K. 2008a, \aap, 481, 83
610: \bibitem[Lecavelier et al.(2008b)]{Lecavelier08b}Lecavelier des Etangs, A., Vidal-Madjar, A., \& D\'{e}sert, J. M., Sing, D. K. 2008b, \aap, accepted
611: \bibitem[Lodders(1999)]{Lodd99}Lodders, K. 1999, \apj, 519, 793-801
612: \bibitem[Loders \& Fegley(2006)]{Lodd06}Lodders, K., \& Fegley, Jr., B. 2006, in Astrophysics Update, Vol. 2, ed. J. W. Mason (Praxis Publishing, Chichester, UK)
613: %\bibitem[Mandel \& Agol(2002)]{Mandel}Mandel, K. \& Agol, E. 2002, \apj, 580, L171-175 
614: \bibitem[Marley(2000)]{Marley2000}Marley, M. S. 2000, in Griffith C. A., Marley, M. S., eds, ASP Conf. Ser. Vol. 212, From Giant Planets to Cool Stars. Astron. Soc. Pac., San Francisco, p. 152
615: \bibitem[Mazeh et al.(2000)]{Mazeh00}Mazeh, T. et al. 2000, \apj, 532, L55-L58 
616: \bibitem[Narita et al.(2005)]{Narita}Narita, N. et al. 2005, \pasj, 57, 471-480
617: %\bibitem[Neckel \& Labs(1994)]{Neckel94}Neckel, H. \& Labs, D. 1994, \solphys, 153, 91-114 
618: %\bibitem[Pont et al.(2006)]{Pont06}Pont, F., Zucker, S., \& Queloz, D. 2006, MNRAS, 373, 231-242 
619: \bibitem[Pont et al.(2007)]{Pont07}Pont, F. et al. 2007, \aap, 476, 1347-1355
620: \bibitem[Pont et al.(2008)]{Pont08}Pont, F., Knuston, H., Gilliland, R. L., Moutou, C., Charbonneau, D. 2008, MNRAS, 385, 109
621: \bibitem[Redfield et al.(2008)]{Redfield}Redfield, S., Endl, M., Cochran, W., Koesterke, L. 2007, \apjl, 673, 87-90
622: %\bibitem[Richardson et al.(2007)]{Rich07}Richardson, L. J., Deming, D., Horning, K., Seager, S. \& Harrington, J. 2007 Nature, 445, 892-895 
623: \bibitem[Rowe et al.(2006)]{Rowe}Rowe, J. F., et al. 2006, \apj, 646, 1241-1251 
624: %\bibitem[Santos, Israelian, \& Mayor(2004)]{Santos04}Santos, N. C., Israelian, G., \& Mayor, M. 2004, \aap, 415, 1153-1166
625: \bibitem[Seager \& Sasselov(2000)]{SS00}Seager, S. \& Sasselov, D. D. 2000, \apj, 537, 916-921 
626: \bibitem[Showman \& Guillot(2002)]{Show02}Showman, A. P. \& Guillot, T. 2002, A\&A, 385, 166-180
627: \bibitem[Showman et al.(2008)]{Showman08}Showman, A. P., Cooper, C. S., Fortney, J. J, Marley, M. S. 2008, ApJ, arXiv:0802.0327
628: \bibitem[Sing et al.(2008)]{Sing08a}Sing, D. K., Vidal-Madjar, A., Lecavelier des Etangs, A., D\'{e}sert, Ballester, G. 2008, ApJ, submitted, arXiv:0802.3864
629: \bibitem[Swain et al.(2008)]{Swain08}Swain, M. R., Vasisht, G., Tinetti, G. 2008, Nature, 452, 329
630: \bibitem[Vidal-Madjar et al.(2003)]{Vidal02}Vidal-Madjar, A., Lecavelier des Etangs, A., D\'{e}sert, J.-M., Ballester, G. E., Ferlet, R., H\'{e}brard, G. \& Mayor, M. 2003, Nature, 422, 143-146
631: \bibitem[Vidal-Madjar et al.(2004)]{Vidal03}Vidal-Madjar, A. et al. 2004, \apj, 604, L69-L72 
632: \bibitem[Yelle(2004)]{Yelle}Yelle, R. 2004,Icarus, 170, 167-179
633: \end{thebibliography}
634: 
635: \clearpage
636: %========================
637: %==========
638: %Figure 11.
639: \begin{figure*}
640: %  \includegraphics[width=12.5cm,angle=90]{Absorptionalt_rgb.ps}
641:   \vspace{1cm}
642:   \plotone{f1.eps}
643:   \figcaption{Transit Na~D absorption from \cite{Sing08a}. Plotted 
644: is the planetary absorption percentage compared to the minimum
645: absorption level referenced at 5,000{\AA} (1.440\%).  In black is the low resolution
646: data binned over 16 pixels and overplotted in red is the medium
647: resolution data binned over 18 pixels (10 \AA).  Representative error bars for the
648: binned data are also plotted (also see Fig. 10 \& 11 of \citealt{Sing08a}) as well as  
649: the mean planetary radius from \cite{Knut} (dotted line).
650: Plotted in blue is our transit absorption model
651: which contains opacity from atomic Na, Rayleigh scattering, and
652: contains upper-atmospheric Na depletion.  The model in purple includes Na, TiO, and VO illustrating
653: the wavelengths where those features are prominent.  
654: Higher absorption values
655: indicate higher atmospheric altitudes, with the right y-axis
656: labeled as the altitude above our minimum absorption depth, assuming
657: the system parameters of \cite{Knut}.  
658: The Na~D line cores can easily
659: be seen in the zoomed plot of Fig. 2.  The narrow Na~D1 and D2 cores
660: extend above a wide plateau of Na absorption indicating a low amount of
661: Na is present in the middle atmosphere compared to the lower.  The
662: drastic abundance change, along with the pressure and temperature
663: where the change occurs, indicates that Na is condensing into sodium
664: sulfide on the night side of the planet. 
665: }
666: \end{figure*}
667: %========================
668: %Figure 11.
669: \begin{figure}
670:  \vspace{1cm}
671:   \plotone{f2.eps}
672: %  \includegraphics[width=13cm,angle=90]{Absorptionalt2.ps}
673:   \figcaption{Same as in Fig. 1, but zoomed on the Na region with the medium resolution data
674: binned over 3 pixels (1.7 \AA).  The Na~D doublet is easily resolved and reaches up to an
675: average altitude of $\sim$3,500 km.}
676: 
677: \end{figure}
678: %========================
679: %========================
680: \begin{figure}
681:  \vspace{0.2cm}
682:   \plotone{f3.eps}
683: %  \includegraphics[width=13cm,angle=90]{fig2.ps}
684:   \figcaption{Limb temperature-pressure atmospheric profile of HD209458b for the sodium condensation model.
685: Plotted (thick line, three points with error bars) is our best fit terminator T-P profile, 
686: while (thin) is a day-side inverted profile
687: from \cite{Burrows07}.  Also shown are the
688: chemical equilibrium condensation curves of iron, TiO, and sodium (dashed
689: lines). The horizontal line represents the pressure as which the T-P
690: profile crosses the sodium condensation curve.  Above the condensation
691: curve, at high altitudes and low pressures, Na is depleted compared to
692: lower altitudes. The CH$_4$/CO line represents the profile where equal
693: amounts of CO and CH$_4$ are expected assuming chemical equilibrium.
694: The whole data set with low and medium
695: resolution observations are fitted with a total of 7 free
696: parameters, resulting in the plotted T-P profile and sodium abundance
697: below and above the sodium condensation temperature.  
698: A fit imposing an adiabatic lower T-P profile is also shown.
699: The sodium
700: abundance above the Na$_2$S condensation level is found to be
701: significantly lower than the sodium abundance below the condensation
702: level (the ratio of both abundances is significantly above 1 at
703: 3-$\sigma$ level); this reveals the presence of Na$_2$S condensation.}
704: 
705: \end{figure}
706: 
707: %========================
708: \begin{figure}
709:  \vspace{0.2cm}
710:   \plotone{f4.eps}
711: %  \includegraphics[width=13cm,angle=90]{fig4.ps}
712:   \figcaption{Differential spectrophotometric measurements of atmospheric sodium 
713: absorption from (diamonds) Charbonneau et al. (2002) and (circles) Sing et al. (2008) using the HST/STIS medium
714: resolution data.  Also plotted are the Na absorption predictions from isothermal models with temperatures (bottom to top)
715: of 500, 750, 1000, 1250, 1500, and 2000 K and constant sodium abundance.  The isothermal models illustrate the necessity
716: for a high-altitude temperature increase in the narrowest two photometric bands (4.4 and 8.9 {\AA}) which probe
717: the highest altitudes.}
718: 
719: \end{figure}
720: %========================
721: 
722: 
723: \end{document}
724: 
725: