Atmospheric Features of Transiting Exoplanets

1
Atmospheric Features of Transiting Exoplanets

• Nassim Bozorgnia
• Thesis Advisor Dr. Debra Fischer

2
Outline

• Introduction
• Brown Dwarfs
• 2.1 Brown Dwarf Detections
• 2.2 Brown Dwarf Atmospheres
• 3. Searches for Exoplanet Atmospheres
• 3.1 Space-based Observations
• 3.2 Ground-based Observations
• HD 149026 A Star with a Transiting Planet
• 4.1 HD 149026b An L/T Dwarf
• 4.2 Observations
• 4.3 Data Analysis
• 5. Detectability Simulations
• 6. Summary and Conclusion

3
1. Introduction

• More than 170 extrasolar planets detected by
  Doppler surveys over the past ten years (Marcy et
  al. 2005)
• Exoplanets found to date are very diverse and
  have orbits and masses different from planets in
  our Solar System.

4
(No Transcript)
5

• Short-period massive planets that orbit within
  0.1 AU of their host stars are called hot
  Jupiters
• About 10 of all hot Jupiters transit their host
  star.
• Transiting exoplanets offer a unique opportunity
  to search for constituents in the planet
  atmosphere.

6
Artists concept of HD 149026b in transit
(artwork by Lynette Cook)
7

• N2K Consortium Carry out Doppler surveys of the
  Next 2000 FGK stars using Keck, Subaru, Magellan
  (Fischer et al. 2005)
• To date, six exoplanets with periods between 1.2
  to 12 days have been discovered by the N2K
  consortium.
• HD 149026b is a short-period, Saturn-mass planet
  that transits its host star, and was recently
  discovered by the N2K Consortium.

8
(No Transcript)
9

• Transiting planets are likely to reside close to
  their host stars. These strongly irradiated
  planets have atmospheres similar to brown dwarfs.
  
• the strongest absorption features in brown dwarf
  atmospheres are expected to be from alkali metals
  such as Na I and K I, and Li I resonance
  doublets.

10
2. Brown Dwarfs

• Brown dwarfs are low-mass substellar objects that
  do not fuse hydrogen into helium as do stars on
  the main-sequence and cannot stabilize their
  surface and central temperatures.
• Because of their low masses, the core contraction
  of brown dwarfs is not halted by gas pressure and
  hydrogen fusion, but by electron degeneracy.

11
Hydrostatic equilibrium in normal stars
12
2.1. Brown Dwarf Detections

• 1988 GD 165B was discovered to orbit the 32-pc
  distant DA4 white dwarf GD 165 (Becklin and
  Zuckerman 1988).
• 1995 Gliese 229B which was orbiting the 5.7-pc
  distant M1 dwarf Gl 229 was discovered (Nakajima
  et al. 1995).
• These object were not classifiable as normal M
  dwarfs and were thought to be a link between
  stars and planets.

13

• Follow-up discoveries made it clear that new
  spectral types were needed to link the previously
  known low-mass stars to cool, planet-like
  objects.
• The L and T spectral classes were suggested by
  Kirkpatrick et al. in 1999, and to date, a total
  of 403 L dwarfs and 62 T dwarfs have been
  discovered.

14
2.2. Brown Dwarf Atmospheres

• The spectra of L dwarfs, T dwarfs and irradiated
  giant planets have dominant absorption lines from
  neutral alkali metals like Na, K, Cs, Rb, and Li.
• The more refractory metals like aluminum,
  magnesium, iron, silicon, and calcium, condense
  at low temperatures, and form droplets and rain
  out of the atmosphere, leaving the upper
  atmosphere depleted of heavy elements.

15

• Various qualitative similarities exist between
  the atmosphere of giant exoplanets and those of
  late type L dwarfs.
• Based on the effective temperature range of the
  atmosphere of the planet, one can estimate the
  spectral type of the exoplanet, classify it as a
  particular extrasolar giant planet class, and
  predict its atmospheric composition.

16
Adopted from Burrows (2001). For an atmospheric
temperature of 1500 K and pressure of 1 bar,
neutral alkali lines are dominant.
17
3. Searches for Exoplanet Atmospheres

• HD 209458 a G0 dwarf with V7.65
• The transiting planet has an anomalously large
  radius of 1.3 RJup , making it an excellent
  candidate to search for atmospheric features.
• Searching for atmospheric signatures in other
  exoplanets is difficult
• Lower intrinsic brightness of the host star
• Smaller planet to star radius

18
3.1. Space-Based Observations

• HST Observations
• Absorption from sodium at 5893 Å in the
  atmosphere of HD 209458b (Charbonneau et al.
  2002)
• Detection of an escaping extended exosphere of
  hydrogen in HD 209458b (Vidal-Madjar et al. 2004)
• Spitzer Observations
• Detection of infrared radiation of 209458 and
  TrES-1 (Charbonneau et al. Deming et al. 2005)

19
3.2. Ground-Based Observations

• Lick Keck Weak upper limits on Na I and K I in
  the atmosphere of HD 209458b and 51 Peg (Bundy
  Marcy 2000).
• No carbon monoxide absorption features detected
  in the transmission spectrum of HD 209458b at
  2.3 µm using NIRSPEC on Keck II (Deming et al.
  2005).

20
4. HD 149026 A Star with a Transiting Planet

• HD 149026 is a metal-rich G0 IV subgiant with
  V8.15.
• HD 149026b was initially detected by the N2K
  consortium, using Subaru and Keck telescopes.
• Orbital characteristics
• a 0.045 AU
• P 2.8759650.000085-0.000135 days
• Msini 0.360.03 MJup

21
Keplerian fit overplotted on the phased RV data
obtained at Subaru and Keck (adopted from Sato et
al. 2005).
22
Photometric transits of HD 149026 observed at
Fairborn Observatory. Solid curves represent the
best-fit models (Adapted from Sato et al. 2005).
23
4.1. HD 149026b An L/T dwarf

• The equilibrium temperature of an exoplanet
  atmosphere at the substellar point

where f 1 for an isotropic emission, and the
Bond albedo, AB is 0.3. ? We calculate Teq 1593
K.
24

• Fortney et al. (2006) report different Teff for
  the planet based on different atmospheric models
• For a cloud free model with M/H0.5 and
  assuming isotropic radiation Teff 1734 K
• If radiation is not isotropic, but radiates from
  the day side only (2p steradians), this same
  atmospheric model yields Teff 2148 K
• ?Therefore, the temperature of HD 149026b is
  likely to be similar to the temperature of L/T
  dwarfs.

25
4.2. Observations

• A total of 30 spectra obtained at Keck using the
  HIRES echelle spectrometer.
• Eighteen of these spectra were obtained while the
  planet was in transit.
• Wavelength range 3500-8000 Å
• Resolution, R55000 S/N250

26
(No Transcript)
27
Two Sources of Contamination

• Iodine contamination 4800-6200 Å
• (Due to the iodine absorption cell on the
  telescope.)
• Telluric contamination wavelengths greater than
  6000 Å

28
(No Transcript)
29
4.3. Data Analysis

• Goal To Search for constituents of the planet
  atmosphere during transit.
• Compare spectra obtained during transit with
  those obtained out of transit
• First Step
• Identify the in and out of transit observations
  by using the ephemeris of HD 149026b and
  phase-folding the RV data.

30
(No Transcript)
31
The Rossiter-McLaughlin Effect
32

• Next Step Compare the in and out of transit
  spectra
• Constructed a high signal-to-noise continuum-
  normalized template for HD 149026 by averaging
  all 12 nontransit spectra.
• Normalized each individual spectrum (both in and
  out of transit), and cross-correlated with the
  template spectrum.

33
Number of Photons
Pixels
34

• We then constructed an average transit spectrum
  from 18 in-transit spectra
• Calculated the difference between the template
  nontransit spectrum and the average in-transit
  spectrum.
• We searched the full wavelength range of
  6500-8000 Å by eye for any residuals.

35

• First portion of 2nd order

36

• Nine spectral orders were covered by this method.
• Because of contamination from telluric and iodine
  lines, we were not able to cover other
  wavelengths by this technique.
• ?No significant variations between the in-transit
  and out-of-transit spectra greater than a few
  percent were found.

37

• A more robust analysis Looking at individual
  lines
• Searching for potassium
• Overplotted an in-transit spectrum on the
  template spectrum at the place of the K line at
  7698 Å.
• Subtract the two spectra and look for significant
  differences
• Considered a wavelength bin centered on the K
  line at 7698 Å.

38
(No Transcript)
39

• Calculated the percent differences at each pixel,
  by taking the difference between each spectrum
  and the template spectrum, and dividing by the
  template spectrum.
• Choosing bins with widths comparable to sizes of
  absorption lines (8 pixels or 10.4 km s-1)
• In order to average over photon noise, we binned
  the percent differences in wavelengths.
• Search in the percent deviations as a function of
  orbital phase phase0 corresponds to mid-point
  of transit.

40
(No Transcript)
41

• Uncertainties
• To assess uncertainties, we computed the standard
  deviation of the percent differences for 12
  out-of-transit spectra. This represents the level
  of intrinsic fluctuations.
• A Kolmogorov-Smirnov (K-S) test was applied to
  the in-transit and out-of-transit residuals.
• The K-S test suggests that the probability that
  the two samples were drawn from different
  populations is only 0.36.

42

• ?The K-S test implies that there is no evidence
  for an excess of potassium in the irradiated
  planet atmosphere.

43

• Searching for lithium
• The same analysis was applied to the Li I
  resonance doublet at 6707.82 Å.
• The template spectrum was overplotted on one
  in-transit spectrum, and the difference of the
  two was evaluated.
• We also considered a 10.4 km s-1 bin at the
  place of the lithium line.

44
(No Transcript)
45
(No Transcript)
46

• The percent deviations for Li do not exceed the
  limit of 3 times the intrinsic fluctuations
  measured in the out-of-transit spectra.
• The K-S test shows that the in-transit spectra
  are statistically identical to the out-of-transit
  spectra.
• No excess lithium absorption is seen during
  transit.

47
Possibility of a leading or trailing atmosphere

• Nontransit observations that are close to the
  transit window, correspond to the ingress and
  egress times.
• A leading atmosphere could cause contamination in
  the spectra near ingress
• A trailing atmosphere may be problematic near
  egress
• A leading/trailing atmosphere could enhance
  absorption when the planet is not transiting

48

• To investigate the possibility of a
  leading/trailing atmosphere, we modified our
  original analysis of the K and Li lines.
• We constructed a template spectrum from only 5
  out-of-transit spectra which were clearly not
  near the ingress and egress points.
• Then, we computed the percent deviations for
  individual observations of the K and Li lines
  relative to the new template, and plotted against
  orbital phase.

49
(No Transcript)
50
(No Transcript)
51

• The dashed horizontal lines show boundaries that
  represent three times the RMS scatter in the 5
  non-transit spectra.
• The K-S test shows that the line depths do not
  vary significantly whether the observations were
  obtained in-transit, near ingress or egress, or
  out of transit.
• We see no line enhancement that suggests a
  leading or trailing exosphere with a large
  optical depth

52
5. Detectability Simulations

• What is the required increase in optical depth in
  the planet atmosphere to produce a detectable
  signal?
• We scaled the K line and the Li line in 1 steps
  for each of our in transit spectra.
• For each of these pseudo optical depth scaling
  factors, we applied the two-sided K-S test to
  find the probability that the simulated transit
  data were drawn from the same distribution as our
  out-of-transit template.

53
Synthetic data. A 2enhancement in the signal at
the place of the K line.
54
The 2 fake signal would be detected at the 84
confidence level.
55
A 5 fake signal is introduce at the place the K
line.
56
The 5 fake signal would be detected at the
98.83 confidence level
57
An 8 fake signal is introduced at the place of
the K line.
58
The 8 fake signal would be detected at the
99.47 confidence level.
59
K-S Test for Potassium
60
K-S Test for Lithium
61
(No Transcript)
62
6. Summary and Conclusion

• Theoretical models of planetary atmospheres
• close-in giant planets similar to L/T dwarfs
• Alkali metals such as Na, K ,and Li have dominant
  absorption features in their spectra.
• We investigated the spectra of HD 149026b
  obtained at Keck, and searched the wavelength
  range of 6500-8000Å.
• ?Any spectral changes during transit must occur
  below a few percent.

63

• We specifically studied the K line at 7698 Å, and
  the Li line at 6707.8 Å.
• In this method the percent deviation between each
  spectrum, and a template comprised of co-added,
  out-of-transit spectra was calculated.
• ?We applied the K-S test and found no evidence
  for potassium or lithium in the atmosphere of the
  planet.

64

• Monte Carlo simulations show that deepening the
  spectral lines by just 3 results in a clearly
  detectable signal at 95.16 for both potassium
  and lithium.
• The lack of variations in the core of K and Li
  might be due to
• The small ratio of the planet to star radius
• Thin atmosphere of the planet may make detection
  difficult
• Having only 5 nontransit observations at points
  far from the ingress and egress