Characterisation of atmospheres using direct exoplanet spectra at nearIR wavelengths

1
Characterisation of atmospheres using direct
exoplanet spectra at near-IR wavelengths John
Barnes University of Hertfordshire
Travis Barman Lisa Prato (Lowell Obs.) Hugh
Jones David Pinfield (Herts) Chris Leigh
(Liverpool JMU) Andrew Collier Cameron (St
Andrews) Brad Hansen (UCLA) James Jenkins
(Santiago)
2
Hot Jupiter models - A stratosphere?

• Stratospheric absorber included in models by
  Burrows, Budaj Hubeny (2008, ApJ, 678, 1436)
• constant opacity at l 0.42 - 1.0 mm below a
  given pressure (0.03 bars).

• TiO and VO give similar (Fortney et al., 2008,
  ApJ, 678, 1419) effects which result in a
  temperature inversion

Fortney et al. (2008)
3
Two classes of irradiated atmospheres?

• pL Cooler Ti, V as solid condensates
• absorb radiation deeper in atmosphere
• atmospheric dynamics will more readily
  redistribute energy leading to cooler day sides,
  warmer night sides and phase shifts in thermal
  emission lightcurves
• e.g. HD 189733b TrES-1
• pM Hot - TiO, VO opacities absorb incident
  flux hot stratospheres, molecular bands in
  emission
• Peaks/troughs evened out
• Contrast ratio increased telluric window regions
  where weak absorption features instead appear in
  emission - search for emission features?
• e.g. HD 209458b, Ups And b and probably HD
  179949b

4
Characterisation of atmospheres
Plot taken from Barman 2008, ApJ, 676, 61
Burrows et al. 2008, ApJ, 668, L171
5
Direct near infrared spectroscopic planetary
signal

• Fp/F 1/1000 of the combined starplanet
  spectrum in the near infrared 2.2 mm K
  band
• Extract the signal from a high resolution
  spectral timeseries planetary signature is
  modeled as a phase dependent spectrum
  superimposed on an unvarying stellar spectrum

• Does not require transiting system
• Contrast ratio determined
• Kp, hence orbital inclination and planet mass
  determined
• Test of model atomic/molecular linelists at high
  resolution
• Split data into wavebands to obtain a local SED
• Optimise the phase function fit to better
  constrain the energy distribution models

6
Phase dependency of model

• Planetary time dependent variations
• Doppler shift of the spectrum due to relative
  orbital position of planet
• Phase dependent flux ratio fp/f which is
  dependent on the atmospheric physics and heating
  due to the parent star

a phase angle e0(l) maximum
planet/star flux ratio at ?0.5
g phase function
7
Phase dependent model spectra

• Barman, Hauschildt Allard, 2005,
  ApJ, 632, 1132
• pL (Type 1) H2O and CO absorption
• pM (type 2) H2O and CO emission

Barman, Hauschildt, Allard (2005, ApJ, 632, 1132)
8
Deconvolution

• S/N in a single spectrum is typically a few
  hundred
• Several hundred to several thousand lines in a
  typical spectrum
• Use model spectrum to deconvolve a mean line
  profile from the observed spectra
• A weighted least squares profile can be derived
  which boosts the S/N ratio by a factor depending
  on the number of lines Typically factor of
  several to a few 10s gain
• S/N ratios of order 1000 can thus be achieved
  for a single spectrum, enabling search for planet
  signatures of similar magnitude (i.e. Fp/F
  1/1000)

9
Modeling the planetary motion

• High S/N average spectrum is scaled, shifted and
  subtracted from each spectrum in turn in order to
  remove stellar spectrum tellurics
• Residuals contain only a planetary absorption
  spectrum (not removed by mean spectrum
  subtraction due to radial velocity changing from
  spectrum to spectrum during motion of the planet
  in its orbit)
• Remaining trends moved using principal components
  analysis

10
Modeling/detecting a planet

• Use a travelling Gaussian scaled by the phase
  function which mimics the sinusoidal RV motion of
  the planet
• Since inclination is generally unknown run model
  for pairs of velocity amplitude, Kp, and maximum
  planet/star brightness, e0, and measure
  improvement in c2 for combination of e0 vs Kp
• Can test significance of the result by
  randomising the order of spectra within each
  night and re-performing the search as above. By
  using several thousand randomised data sets, we
  can plot confidence levels for detected
  enhancements in c2.
• Perform 3000 trials which implicitly define
  empirical probability distributions of e0 and c2
  that include both the photon statistics and the
  effects of correlated systematic errors at each
  trial value of Kp.

11
Simulation

• Wavelength setup equivalent to that afforded by
  IRCS/Subaru or NIRSPEC/Keck - i.e.1.90 2.45 mm

R 20,000

• 50 spectra per night (tellurics star RV
  varying planet at 1 part in 1000 of stellar flux)
• Spectra S/N 300 (S/N after deconvolution
  10,000)
• Top R 20,000
• Bottom R 40,000

R 40,000
Simulated planet is easily recovered at both
intermediate and high resolution
(68.3, 95.5, 99 99.9 confidence levels shown)
12
HD 189733bBarnes, Barman, Prato, Segransan,
Jones, Leigh, Collier Cameron, Pinfield, 2007,
MNRAS, 382, 473

• SpTK1-2V, P2.21 d, a0.031 AU, Kp 153 km/s
• NIRSPEC at Keck II on 22nd Jul 2006
• Aladdin III 10242 InSb array 225 spectra taken
  in highly variable cloud conditions.
• Spectral coverage 2.0 mm 2.4 mm at R 15,000
• S/N ratios of 136 ? 85 (27 min 382 max)
• Deconvolution with pL template - gain of 20
  yields profiles of with mean S/N 2751
  ? 1772 over both nights
• Because of variability only the 131 frames with
  S/N gt 97 were used in the final analysis

13
HD 189733 NIRSPEC coverage
pL
14
Timeseries analysis HD 189733b
All 225 spectra
131 spectra
68.3, 95.4, 99 and 99.9 confidence levels
plotted
15
HD 189733b SED
99.9, 99, 95.4, 68.3 levels (top bar to
bottom arrow)
Deming, Seager, Richardson Harrington (DHSR06),
2006, ApJ, 644, 560 Grillmair et al. (G07), 2007,
ApJ, 658, L115 Knutson et al. (K07), 2007, 447,
183 Charbonneau et al. (C08), 2008, astro-ph
(arXiv0802.0845v2)
16
2008 Results HD 189733b

• Planet is not detected at a contrast which is 2.2
  times (at 2s - 95.4) deeper than the model
  predicts

17
HD 179949bBarnes, Barman, Jones, Leigh, Collier
Cameron, Barber, Pinfield, 2008, MNRAS, In Press
(arXiv0806.0298)

• SpT F8V, P 3.09 d, a 0.045 AU
• CRIRES at VLT on 26th Jul and 2nd Aug 2007
• Single order (not cross-dispersed) spectrum on
  for Aladdin III 512 x 1024 InSb arrays
• 46 27 spectra (combined groups of 4) in
  excellent conditions lt 10 humidity and 0.5
  seeing
• Spectral coverage 2.122 mm 2.175 mm at R
  50,000

18
HD 179949 Spectral coverage
pL
pM
H2O opacities
19
HD 179949b

Phased dynamic spectrum Upper limits
Note relative significance between models
Type 1 (no Ti/V)
Orbital phase
N.B. no planet at Kp 115 kms-1.
Mean S/N after deconvolution 8300.
?
Velocity kms-1
Kp kms-1
Type 2 (with TiO/VO)
Orbital phase
Mean S/N after deconvolution 2600.
Velocity kms-1
Kp kms-1
20
Results HD 179949b
pL / Type 1
pM / Type 2

• We are able to rule out the presence of a pL /
  type 1 planetary atmosphere at a level of log10?0
  -3.53 (i.e. Fp/F 1/3388) with 99 confidence
  (i 30o model)
• We are not able to rule out the presence of a
  planet with a pM / type 2 atmosphere - require
  greater S/N

21
So where are the planets?

• Devil is in the systematics tellurics are the
  major concern and have to be dealt with carefully
• BUT Principal components analysis can move
  residuals, but at some level the sensitivity is
  compromised
• HOWEVER With the HD 189733b and HD 179949b data,
  we have
  reached sensitivities at which a planet should be
  visible
• For HD 189733b no detection at 2s level if
• - Line depths modified by 70
• - Wavelength postns. uncertain by 20
• Are model opacities reliable?
• - 90 opacities with 0.3 cm-1 (factor 3)
• - 49 within 0.1 cm-1 (factor 1.4)

Swain et al., Nature, 2008
22
SummaryFor Hot Jupiter atmospheres at high
resolution

• HD 189733b We can rule out the type 1 atmosphere
  where atomic species such as Ti and V have
  "rained out" resulting in an atmosphere dominated
  by H2O and CO absorption
• HD 179949b The unknown orbital inclination
  introduces a further degree of freedom into the
  interpretation of the results
• reject pL atmosphere (type 1 model)
• more observations needed to enable detection or
  rejection of the pM (type 2) atmosphere scenario
  where a high altitude absorbing species results
  in formation of a stratosphere, pushing many H2O
  transitions into emission
• H2O linelists not accurate at high resolution ?
  use observed
• Relative line depths uncertain ? investigate
  metallicity effects

23
(No Transcript)
24
Future

• Return to optical data sets to search for TiO/VO
  opacities
• Fortney et al. (2008) predict contrast ratios of
  order 10-4 in the red part of the optical for pM
  class (type 2) resulting from absorption and
  heating of TiO/VO
• High resolution/multiorder near infrared
  instruments would enable limits to be pushed into
  regime where sensitivities are of order 10-4 In
  the presence of a clear detection could
• Split data into wavebands to obtain a local SED
• Optimise the phase function fit to better
  constrain the energy distribution models

25
Planetary atmospheres

• Unlike isolated brown dwarfs, atmospheres are
  irradiated, strongly in the case of CEGPs
• e.g. Atmospheric T-P structure will vary
  substantially as a function of latitude and
  longitude
• Condensation of species sequesters most of the
  heavier elements such as Si, Mg, Ca, Al Fe in
  compounds that settle or rain out
• Upper atmospheres depleted of species that would
  otherwise contribute to the molecular chemistry

• The condensates may then provide substantial
  absorption and scattering opacities e.g. MgSiO3
  (enstatite) and Mg2SiO4 (forsterite) may exist
  high in atmosphere -gt Clouds?

26
Direct planet detection Ground based observations

• Snellen (MNRAS, 2005) UKIRT 2.3 mm photometric
  observations of secondary eclipse of HD 209458
  no eclipse detected but flux decrement of 0.10 ?
  0.10 during the in-eclipse event
• Richardson et al. (ApJ, 2003) Occultation
  spectroscopy to search for disappearance of the
  planet signal during eclipse
• Search for 2.2 mm bump led to strong rejection
  of models which re-radiate strongly on the
  dayside hemisphere no peak at a level of 3x10-4
  relies on accurate model shape and opacities

Cloudless heat reradiated on dayside
Cloudy complete heat redistn.
Cloudless complete heat redistn.
27
Direct planet detection Space based observations

• Deming (Nature, 2005) - first secondary eclipse
  detection of HD 209458b
• 0.0026 depth Tp 1130K
• Charbonneau et al. (2005, ApJ) - first secondary
  eclipse detection of TrES-1
• 0.00066/-0.00013 at 4.5 µm
• 0.00225/-0.00036 at 8.0 µm
• Tp 1026 K

Spitzer (HD 20945b and TrES-1)
Snellen (2005)
Deming (2005)

• Contrast ratio developments for a number of
  systems at mid-infrared wavelengths
• A clear need for near infrared measurements which
  reliably measure contrast ratio

28
Reflected light

• Searches for reflected light in the optical (400
  - 600nm) conducted by Charbonneau et al. (1999),
  Cameron et al. (1999-2002) and Leigh et al.
  (2003) can a faint copy of the stellar spectrum
  be detected at optical wavelengths?
• Cloud base forms deep in Class IV planet
  atmospheres spectrum dominated by very strong
  and deep pressure broadened Na K lines which
  absorb much of scattered light deep in the
  atmosphere. Weak Rayleigh scattered reflection
  signature at blue wavelengths remains
• Cloud base forms high in Class V roasters
  leading to less absorption by Na K giving
  atmospheres which possess albedos of typically
  60 that of Jupiter throughout most of the
  optical
• Results
• Upper albedo limits of plt0.12 with 99.9
  confidence for HD 75289b Leigh et al. 2003
  (revised to 0.5 by Rodler, Kürster Henning,
  2008) Planet/Star flux ratios of 1/70,000
  reached.
• p lt 0.39 for Tau Boo (Leight et al. 2003)
• MOST photometry of HD 209458 - albedo, p 0.08
  (1s) - Planet/Star flux ratios of 1/160,000
  probed

29
HD 189733(b) system parameters
Star Spectral type K1-2V mv / mK 7.67 /
5.54 Distance (pc) 19.3 ? 0.32 Teff (K) 4954 ?
50 M (M?) 0.8 ? 0.4 R (R?) 0.753 ?
0.025 Fe/H -0.03 ? 0.04 Prot (d) 10.9 K
(ms-1) 205 ? 6 vsini (kms-1) 3.5 Age (Gyr) gt
0.6 Gyr Planet Transit (HJD) 2453988.80336
? 0.00023 Period (d) 2.2185733 ?
0.0000019 Orbital axis (AU) 0.0312 ?
0.0004 Orbital incln. (deg) 85.79 ? 0.29 Mp
(Mjup) 1.13 ? 0.03 Kp (kms-1) 152.58 ? 1.96
30

• Star F8V with mv 6.25, mK 4.94, Teff 6260,
  Fe/H 0.22, M 1.28 M?, R 1.19R?
• Planet Msini 0.95 M?, a 0.045 AU, P 3.0925
  d
• - most probable inclination 62.5o
• -
  most probable velocity amplitude 141 kms-1

31
2006 Results HD 189733b

• No candidate feature is found in the data at the
  known velocity amplitude of the system,
• Systematic feature at 52 kms-1 is a systematic
• We do not detect the planet with 1-s and 2-s
  limits of log e0 -3.40 and
  -2.88 respectively. The equivalent planet/star
  contrast ratios are 1/2512 and 1/759 respectively