1. High Contrast Spectral Imaging: the Case of GQ Lup
0.73”
GQ Lup B
Keck AO + OSIRIS
OSIRIS: imaging + spectroscopy - 3-D cube
GQ Lup B - claimed to be a planet
Michael McElwain (UCLA)
James Larkin (UCLA)
Stanimir Metchev (UCLA)
OSIRIS commissioning team

2. GQ Lup B – An Exoplanet or a Brown Dwarf?
10–40 MJup brown dwarf?
Keck AO + OSIRIS spectroscopy
McElwain, Metchev, Larkin et al., ApJ, accepted
1–2 MJup planet?
VLT AO slit spectroscopy
Neuhaüser et al. (2005)

3. Discovery images of GQ Lup A/B
M8 V L2
2000 K, log g = 2.0
GQ Lup B
2.2
2.4
2.3
2.0
2.1
Wavelength [µm]
Spectral type: M9–L4
2.5
∆K = 6 mag
GQ Lup A K7 cTT star, -35 degrees declination
Classical T Tauri stars have extensive disks that result in strong emission lines. Weak-lined T Tauri stars are surrounded by a disk that is very weak or no longer in existence.
T Tauri stars are pre-main sequence stars - the youngest visible F,G,K,M spectral type stars (<2 Solar mass). Their surface temperatures are similar to those of main sequence stars of the same mass, but they are significantly more luminous because their radii are larger. Their central temperatures are too low for hydrogen burning. Instead, they are powered by gravitational energy released as the stars contract towards the main sequence, which they reach after about 100 million years. They typically rotate with a period between one and twelve days, compared to a month for the Sun, and are very active and variable.
Assign a spectral type of M9-L4 based on K I spectral index (Reid 2001), H20-D index (McLean 2003), Na I doublet (Cushing 2005, Cameron 2000, Gorlova 2003), or the CO index (Gorlova 2003). Average of these estimates is M9-L4.
log(L/Lsun)=-2.37+/-0.41
Na I
H2O
12CO
cTTS in Lupus 1; age 0.1–2 Myr
(Hughes et al. 1994)
(Neuhaüser et al. 2005)

4. OSIRIS (OH-Suppressing InfraRed Imaging Spectograph)
Integral Field Spectrograph
Spectra over a contiguous rectangular field.
Spatial resolution at the Keck Diffraction Limit (<0.050”)
Spectral resolution (l/Dl) ~ 3800
Full z, J, H, or K spectra with single exposure (16x64 lenslets)
Integrated Data Reduction Pipeline
H2 emission (2.122 microns) from NGC 4151, an AGN x y l

5. OSIRIS - A Lenslet Based Integral Field Spectrograph (IFS)
Focus Image onto
a Lenslet Array
1. Image on Lenslets
2. Pupil images
3. Pupil images dispersed
4. Extracted Data Cube l y
Talk about pipeline creating Extracted Data Cube
16x64x~1500 l x

6. Pre-observing planning checklist
Natural Guide Star – GQ Lup A
R magnitude of 11.0
Choose scale
0.020”
Choose integration time for desired sensitivity
From instrument zero points
Determine dither pattern
Make an execution file
Telluric HD , OSIRIS Observation Planning GUI

7. Keck/OSIRIS Spectra of GQ Lup B
0.73”
GQ Lup B
J-band J M6 M9 L2 L5 H
L0, 2 Gyr
L0, 10 Myr
K I
H2O
H2O
FeH
H2O
GQ Lup B
integral field spectrograph behind Keck II AO system
(PI: J. Larkin, UCLA)
OSIRIS commissioning data (June 2005)
GQ Lup B
OSIRIS - IFS, 3-D data cube
Before getting to what the spectra reveal
GQ Lup B delta J = 6.21+/-0.12 mag
GQ Lup B delta H = 7.5+/-0.5 mag
GQ Lup B delta K = ~6 mag
(McElwain, Metchev et al., ApJ, in press)

8. AO Integral Field Spectroscopy Is More Reliable Than AO Slit Spectroscopy
elevation,
differential refraction
H-band
53 mas-wide slit
GQ Lup A/B aligned on slit
AO slit spectroscopy:
slit width (40–100 mas), PSF (40–80 mas) comparable to pointing precision (~20–40 mas)
differential refraction (atmosphere, AO transmission optics)
especially important in high-contrast regime
IFS AO spectroscopy :
no slit losses due to centering on slit
no slit losses due to differential refraction
trace PSF centroid as a function of 
variable extraction aperture as PSF changes?
Important in high-contrast regime b/c: 1) need to sample radial dependence in halo; 2) need to find companion
With IFS: adjust spectroscopic aperture as a function of wavelength
However, wide slit in Neuhauser et al: 150 mas.
J band image moves by 70 mas due to atmospheric dispersion.
H band image moves by 44 mas due to atmospheric dispersion.
Dispersion due to the dichroic is about 20 mas, at the instrument angle.
5 images across the OSIRIS H-band

9. IFS is Good for Target Extraction and Primary Background Subtraction
Correct cube for differential dispersion.
Extract the companion spectrum.
Fit host star PSF to estimate the background contribution at the location of the secondary.
Subtract host background from the companion spectrum.
Important in high-contrast regime b/c: 1) need to sample radial dependence in halo; 2) need to find companion
With IFS: adjust spectroscopic aperture as a function of wavelength
However, wide slit in Neuhauser et al: 150 mas.
5 images across the OSIRIS H-band

10. Keck/OSIRIS Spectra of GQ Lup B
0.73”
GQ Lup B
J-band J M6 M9 L2 L5 H
L0, 2 Gyr
L0, 10 Myr
H2O
H2O
FeH
H2O
K I
GQ Lup B
commissioning OSIRIS data (Aug 2005)
J- and H-band
spectral type: M8 ± 2
Neuhaüser et al.: M9–L4
GQ Lup B
Spectra of GQ Lup B on a different vertical scale
H2O - good spectral type indicator, independent of gravity
K I absent, H-band triangular: low gravity
(McElwain, Metchev et al., ApJ, in press)

11. GQ Lup A/B Astrometry & Photometry
Similar to imaging
Photometry
Curve of growth for the telluric and GQ Lup A – find flux ratio and magnitude for GQ Lup A
Compare the flux ratios of the same aperture for GQ Lup A/B
Determine GQ Lup B magnitude
J-band

12. High Contrast Imaging: Speckle Suppression
Typical speckle pattern for
Keck II + OSIRIS Imager in the Kn3 filter
At moderate Strehl ratios (< 0.95) and small separations (< 1”), speckle noise produced by atmospheric wavefront distortion and imperfect optics are the dominate noise source.
Innovative techniques for enhancing contrast
Simultaneous Differential Imaging
Spectral Suppression
NOTES:
Diffraction limit at J (30 mas), H band (37 mas), K band is (50 mas).
Cube – HR 7672
Mention that OSIRIS does not have an occulting spot,
and that the black pixels on the center of the image are from saturation of the detector.
Keck II + OSIRIS Spec in the Kbb filter
Speckles are wavelength dependent and can be modelled for each wavelength.

13. Summary
AO integral field spectroscopy is more reliable than AO slit spectroscopy
An IFS is efficient for halo subtraction.
Astrometry and photometry procedures are the similar to those for direct imaging.
An IFS can perform speckle suppression.
GQ Lup B is probably a brown dwarf and not an exoplanet.

14. Steps in Characterizing Sub-Stellar Companions
Determine age and distance
from parent stellar association (best) or primary star
Determine spectral type, effective temperature
direct near-IR spectroscopy (with AO)
Determine mass, surface gravity
from evolutionary models

15. GQ Lup B is Hotter and Older Than Inferred by Neuhaüser et al.
McElwain, Metchev et al.:
spectral type: M6–L0 (~2600 K)
age: 1–10 Myr
Neuhaüser et al. (2005):
spectral type: M9–L4 (~2000 K)
AO slit losses affecting K-band continuum?
weakening H2 CIA absorption at 1.5–2.5 µm
age: 0.1–2 Myr
marginal difference
We use the PMS models of Baraffe et al in order to assign a revised age of the Lupus 1 molecular cloud complex.

16. Testing Evolutionary Models: “Hot-Start” Models Better at ≤3 Myr
2MASS 0535 A/B (0–3 Myr)
GQ Lup B (1–10 Myr)
A (0.054 MSun)
B (0.034 MSun)
N05
Back to which models to use. Could not determine at time of Neuhauser’s paper. Now, use the empirical evidence provided by the eclipsing sub-stellar spectroscopic binary in the Orion Nebula Cluster.
Dynamical masses!
How do we get the Bolometric Luminosity? J magnitude, our distance estimate, and the K-band bolometric corrections for M6-L0 dwarfs from Golimowski et al. (2004).
2MASS 0535 Characteristics
A Mass Radius Temperature Luminosity B
M06
3.0
(Stassun et al. 2006,
Chabrier et al models)
(Neuhaüser et al. 2005,
Wuchterl & Tscharnuter 2003 models)

17. GQ Lup B is Probably a Brown Dwarf
McElwain, Metchev et al.:
spectral type: M6–L0 (~2600 K)
age: 1–10 Myr
“hot-start” models (Burrows et al. 1997; Chabrier et al. 2000)
 mass: 10–40 MJup
Neuhaüser et al. (2005):
spectral type: M9–L4 (~2000 K)
AO slit losses affecting K-band continuum?
weakening H2 CIA absorption at 1.5–2.5 µm
age 0.1–2 Myr
“cold-start” models (Wuchterl & Tscharnuter 2003)
 mass: 1–2 MJup
Not a 1-2 Mjup planet! Our paper in press and posted to astro-ph. Very recently, Christian Marois + Bruce Macintosh independently came up with their own paper on GQ Lup.
Marois et al. (accepted), 0.6–3.5 µm SED analysis: 9–20 MJup

18. Which theoretical models are more accurate?
The Mass of GQ Lup B
“hot-start” models predict 3–42 MJup
Burrows et al. (1997), Baraffe et al. (2002)
uncertain at ≤3 Myr ages
nucleated instability and collapse models predict 1–2 MJup
Wuchterl et al. (2000), Wuchterl & Tscharnuter (2003)
better at young ages?
Before getting into model discussion, I will first show you our data on GQ Lup
“hot start” models assume 1. fully convective 2. pick radius 3. adiabatic at all stages of evolution.
Planet Formation scenarios: Core accretion/gas capture (. or
Which theoretical models are more accurate?
Is GQ Lup B an exoplanet?
(Neuhaüser et al. 2005)

19. Thanks to the OSIRIS team
ACADEMIC
Principal Investigator - James Larkin (UCLA)
Project Scientist - Andreas Quirrenbach (University of Heidelberg)
Co-Investigator – Alfred Krabbe (Cologne)
Research Astronomer – Inseok Song, Christof Iserlohe (Cologne)
Graduate Students - Matthew Barczys, David LaFreniere*, Michael McElwain, Tommer Wizansky, Shelley Wright
Close collaboration – Ian McLean, Eric Becklin
ENGINEERING
Project Engineer - George Brims
Mechanical – Ted Aliado, John Canfield, Nick Magnone, Evan Kress
Software – Tom Gasaway (UCSD), Chris Johnson, John Milburn, Jason Weiss
Electrical – Ken Magnone, Michael Spencer, Gunnar Skulason,
CARA - Paola Amico, Allan Honey, Junichi Meguro, Grant Tolleth, & others
ADMINISTRATIVE
CARA Project Manager – Sean Adkins, David Sprayberry*
Management – Juleen Moon, Jim Kolonko
Secretarial – Melinda Laraneta
(lead engineer in each area for OSIRIS in bold, * denotes non-active team members)
NOTES: - not sure if you need to go through every team member ? (saw)

20. Spectral Classification of Ultra-Cool Objects is Age-Dependent
K I
H2O
H2O
spectral type
proxy for Teff
determined by continuum shape in brown dwarfs
but: young (<100 Myr) brown dwarfs
larger radius
lower surface gravity
(g = GM/R2)
weaker K I, Na I absorption
weaker H2 CIA over 1.5–2.5 µm
spectral classification most reliable from H2O dip at 1.3 µm (Slesnick et al. 2004) J H K
H2 CIA
Na I
logg and Teff degenerate effects on spectra.
H and K bands: H2 CIA - decreases with decreasing logg and wavelength, peaks at K.
BUT: J-band can distinguish: little H2 CIA, H2O, K I
Other features increased VO absorption at 1.06 and 1.18 microns.
Why is the H band triangular? Not sure. H2O steam bands gravity sensitive, or H2 CIA wavelength dependence.
1-50 Myr
(Kirkpatrick et al. 2006)

21. Independent Confirmation of the Mass of GQ Lup B Is Necessary
AO slit spectroscopy near bright objects is challenging
spectroscopic classification is gravity (i.e., age) dependent
theoretical models for sub-stellar objects are unreliable at <3 Myr

22. GQ Lup B is Hotter Than Inferred by Neuhaüser et al.
McElwain, Metchev et al.:
spectral type: M6–L0 (~2600 K)
Neuhaüser et al. (2005):
spectral type: M9–L4 (~2000 K)
AO slit losses affecting K-band continuum?
weakening H2 CIA absorption at 1.5–2.5 µm (Borysow et al. 1997, Kirkpatrick et al. 2006)
marginal difference

23. An Updated Age for Lupus 1: 1–10 Myr
(McElwain, Metchev, et al., in press,
Baraffe et al tracks)
1–10 Myr
0.1 Myr
1 Myr
Also independently re-visit age of GQ Lup A
Benefit from cluster of stars (not single star like HD ) - better constrained age
Evolutionary tracks - point out direction of evolution, isochrones.
Baraffe+98 are best for dynamical masses of stars.
Ages > 10 Myr can be excluded because GQ Lup A is sometimes seen as a classical T Tauri star.
We use I vs. R-I because these bands are not strongly affected by excess UV and IR emission, and
are thus suitable proxies for the bolometric luminosities and effective temperatures of pre-main-sequence stars.
I vs. J-I is another suitable pair. Also, R and I band photometry is often taken simultaneously, thus obviating the problem of variability.
0.1–2 Myr
10 Myr
100 Myr
(Hughes et al. 1994,
D’Antona & Mazzitelli 1994 tracks)

24. GQ Lup: A Sub-Stellar Companion and a Disk
projected binary separation: 110 AU
disk mass: ~0.01 M*
GQ Lup B may be accreting:
J – KS = 1.8 ± 0.1 mag
~1.0 mag for late-M dwarfs
KS – L´ = 1.4 ± 0.3 mag
~0.8 mag for late-M dwarfs
relevant for theories of brown-dwarf formation
similar objects:
2MASS 1207–3932 B (Chauvin et al. 2004)
IRAS B (Apai et al. 2005)
J-Ks =1.8 may be due to the decreased CIA H2 absorption.

25. Summary Teff at L/T transition is a function of age at >0.1 Gyr
or: sub-stellar radius is ~independent of age at >0.1 Gyr
no observational data at <0.1 Gyr: planetary realm
Characterization of young brown dwarfs is challenging
Teff and g have degenerate spectroscopic signatures (H2 CIA)
mass estimates are strongly model dependent
GQ Lup B: 10–40 MJup brown dwarf rather than 1–2 MJup planet
Very low mass ratio (M2/M1 ≤ 0.03) resolved young systems
5 known: HD B, GQ Lup B, HR 7329 B, AB Pic B, HN Peg B
important for inferring the photospheric properties of extrasolar giant planets to be imaged in the future

26. AO Imaging Contrast is Limited by Variable Speckle Noise
Keck AO speckles at 2 µm
r = 1
HD 49197B
Sco PMS 214B
HII 1348B
exo-planets
Conventional AO imaging… ultimately limited …
Exo-planet locus dependent on multitude of assumptions
Variable shading to denote clustering near small separations
speckles limit the detectability of exo-planets
(Kalas et al. 2002)

27. Pushing the Contrast Limit: Speckle Suppression
L1V
G8V
(L1V/ G8V)-1
speckles are images of the primary:
same spectrum
sub-stellar companion is much cooler
optimal weighting:
adjust to any companion spectral type
caveats:
need Nyquist sampling of PSF
need broad spectral range
Optimal filter: each lenslet containing only flux from the primary becomes 0. Lenslets with companions are positive.
optimal weighting
(McElwain et al., in prep.)

28. Closing the Gap to R.V. Exo-Planets
companions discovered in direct imaging, ≥5–15 MJup
a ≥ 15–50 AU
limited by contrast of conventional AO
r.v. planets, 0.2–15 MJup
a ≤ 6 AU
limited by survey length
intermediate regime:
6–50 AU
to be probed by speckle suppression techniques
2M 1207 B (Chauvin et al. 2004)
GQ Lup B (Neühauser et al. 2005)
exo-planets
SDI, OSIRIS
Ultimately, since interested in the lowest-mass BDs… how this complements planet searches
Now you see, that I’ve conveniently chosen the shading so that it includes the 2 candidate exo-planets

29. OSIRIS: An Integral Field Spectrograph for Keck AO
1.0–2.4 µm
R = 3700
spectroscopy over a 2-D field of view:
3-D data cube
FOV:
from 0.32"1.20" (20 mas/lenslet)
to 4.8"6.4" (100 mas/lenslet)
Keck diffraction limit: mas at 2.2 µm
commissioned: 2005 x y l
(Larkin et al. 2006)

30. AO Slit Spectroscopy is Challenging
HD A
Why use an IFS (OSIRIS)? - slit spectroscopy is challenging: continuum slope
Example: HD
Challenges: object centering (width comparable to pointing accuracy; worse with dithering); contamination from primary; AO correction; ADR (if oriented along binary)
Gemini AO; K-band
HD B/C: L2 ± 2
(Potter et al. 2002)

31. AO Slit Spectroscopy is Challenging
Apply a slope correction, s.t. to maximize correlation btw observed and template spectra
-> improved precision.
HD B/C are well separated (2.7”) from primary. Closer-in: AO halo, non-linear effects.
HD C: L2 ± 2  L4 ± 1
(Goto et al. 2003)