1. Observing the Atmospheres of Transiting Exoplanets
Seth Redfield
University of Texas at Austin

2. Outline Background Observations Results Future
Transmission spectroscopy and exoplanet atmospheres
Observations
Modeling subtle spectral effects (Rossiter-McLaughlin Effect and differential limb-darkening)
Development of ground-based observational program
Results
First ground-based detection of exoplanetary atmosphere
Control line comparison
Empirical Monte Carlo
Future
Short-term plans for transmission spectroscopy program
Rapidly growing sample and motivation to find and study Earth-like exoplanets indicates strong long-term research potential

3. Measuring Atmospheric Properties
Transmission Spectroscopy: Comparison of light that has passed through a medium (e.g., a planetary atmosphere) and light that has not. Atmospheric constituents will provide additional absorption beyond the opaque body and produce a deeper effective transit at the cores of their lines.
(Brown 2001)

4. Transmission Model (Ca I) (Li I) (Rb I) (Cs I)
(Barman 2007)
(Ca I)
(Li I)
(Rb I)
(Cs I)
Early models (Seager & Sasselov 2000, Brown 2001)
Models include: clouds, planetary abundances, rainout of condensates, and photoionization
Most gases in molecular form except He and alkali metals
Strongest features are narrow lines of NaI and KI
Volatile elements (e.g., Mg, Ca, Ti) have condensed into grains

5. Measuring Exoplanet Atmospheres
The presence of an exoplanetary atmosphere will cause a persistent absorption signature for both NaI lines.
With Atmosphere
NaI doublet

6. Measuring Exoplanet Atmospheres
In addition to possible atmospheric absorption, as the planet travels across the nonuniform stellar disk, it will be blocking different sections in velocity (Rossiter-McLaughlin Effect) and intensity (relative limb darkening)

7. Measuring Exoplanet Atmospheres
However, when averaged over a complete transit, the stellar obscuration signal is minimized, and the persistent exoplanetary atmospheric absorption dominates.

8. First Detection
Spectroscopic observations using the Hubble Space Telescope led to first detection of atmospheric absorption (Charbonneau et al. 2002)
Use “narrow” band and measure 23.2  5.7 x 10-5 abs.
Currently no operating high resolution spectrograph in space (SM4 in Aug 2008)
R = / ~ 5540, “narrow” bin: R ~ 490
Models had predicted ~3x stronger absorption
(Charbonneau et al. 2002)

9. Ground-Based Prospects?
Previous attempts were unsuccessful, because typically only one effective transit was observed and use of an iodine cell prevented sodium analysis (Bundy & Marcy 2000; Moutou et al. 2001; Winn et al. 2004; Narita et al. 2005; Bozorgnia et al. 2006; Arribas et al. 2006).
Requires high resolution spectrograph, large telescope, and ability to observe many sporadic transiting events.
Hobby-Eberly Telescope (HET)
At McDonald Obs. in west Texas
9.2-meter effective aperture
Fixed-elevation design necessitates queue-scheduled observing
Hi-Res Spectrograph (R ~ 60,000)

10. Observational Details
Many, many transit observations (and out-of-transits)
~11 1-hour in-transit (S/N~1800)
~25 1-hour out-of-transit (S/N~2700)
HD (to scale)
Top View
View from Earth
Period = 2.22 days
Inclination = 85.6 deg
Mass = 1.14 MJup
Radius = 1.14 RJup
Orbital distance = 0.03 AU
Stellar distance = 19.7 pc

11. Calibrations: Telluric (NaI)
Telluric standard
Telluric — ISM/stellar
HD189733
HD — telluric only
ISM
NaI

12. Calibrations: Telluric (CaI)
Telluric stan.
HD189733
HD — Telluric

13. NaI Detection
-67.2  7.2 x (9.3)
Clear absorption in both NaI lines
First ground-based detection of exoplanetary atmos.
Random error approximately that of Hubble (5.7 x 10-5)
RM and limb-darkening effects are minimal
Abs. extends an additional 6% above planetary surface
Abs. is blue-shifted
(Redfield et al. 2008)

14. Control Line (CaI) +15.4  6.8 x 10-5 (2.3)
Transmission models predict no absorption in volatile elements (e.g., Ca)
No significant signal in transmission spectrum at CaI
Random error approximately that of Hubble (5.7 x 10-5)
CaI obtained at slightly higher S/N
(Redfield et al. 2008)

15. Empirical Monte Carlo What about systematic errors?
Calibrating over many obs. runs
Telluric removal
Normalization of continuum
RM and limb-darkening correction
Stellar activity (e.g., flares)
Changing distributions of starspots
Use existing observations to create 1000s of alternative datasets
“out-out” test compares a subset of out-of-transit observations with the remaining out-of-transit sample
Run an “in-in” test to compare all in-transit observations
Top View

16. Empirical Monte Carlo (NaI)
“out-out” and “in-in” distributions are centered at zero
As expected, systematic errors dominate
Width of “out-out” distribution provides robust 1 value
Detection >3
Demonstrates consistency among in- and out-of-transit observations
-3.5  20.7 x 10-5
3.2
(Redfield et al. 2008)

17. Empirical Monte Carlo (CaI)
“out-out” and “in-in” distributions are centered at zero
As expected, systematic errors dominate
Width of “out-out” distribution provides robust 1 value
“Signal” at CaI consistent with zero
Higher S/N of CaI region and simplified telluric spectrum results in narrower distribution
-0.5  13.7 x 10-5
1.1
(Redfield et al. 2008)

18. Physical Properties
Tenuous atmosphere reaches to a height of 6% of planetary radius
~3x stronger than absorption detected toward HD209458b
Temperature inversion detected for HD209458b, requiring a high altitude absorber (e.g., clouds) which could explain stronger signal toward HD189733b (Knutson et al. 2007; Burrows et al. 2007)
Blue-shift may be explained by high-speed winds flowing from the hot dayside to the cold nightside (e.g., thermal tidal winds)
Models predict winds at velocities ~10 km/s (Cooper & Showman 2003)
(Burrows et al. 2008)

19. Physical Properties
Tenuous atmosphere reaches to a height of 6% of planetary radius
~3x stronger than absorption detected toward HD209458b
Temperature inversion detected for HD209458b, requiring a high altitude absorber (e.g., clouds) which could explain stronger signal toward HD189733b (Knutson et al. 2007; Burrows et al. 2007)
Blue-shift may be explained by high-speed winds flowing from the hot dayside to the cold nightside (e.g., thermal tidal winds)
Models predict winds at velocities ~10 km/s (Cooper & Showman 2003)
(Cooper & Showman 2003)

20. Summary
Compare in-transit observations to out-of-transit observations and detect additional NaI absorption when in-transit
First ground-based detection of atmospheric absorption in the transmission spectrum of a transiting extrasolar planet.
First resolved (i.e., high resolution) transmission spectrum.
Detection supported by analysis of a neighboring strong absorption line (CaI), that is not expected to, and does not show any exoplanetary atmospheric absorption
Contribution of systematic errors derived from empirical Monte Carlo analysis