Transcript script (powerpoint)

Direct Imaging of Exoplanets
I. Techniques
a) Adaptive Optics
b) Coronographs
c) Differential Imaging
II. Results
Challenge 1: Large ratio between star and planet flux (Star/Planet)
Reflected light from Jupiter ≈ 10–9
Challenge 2: Close proximity of planet to host star
Planet
Mass
a
(MJup) (AU)
d
(LYr)
D
(arcsecs)
e Eri b
1.5
3.39
10.4
1.0
GJ 832 b
0.6
3.4
16.1
0.69
GJ 317 b
1.2
0.95
29.3
0.10
GJ 849
0.8
2.35
28.7
0.27
HD 62509 b
2.9
1.7
33.7
0.16
55 Cnc d
3.8
5.8
42.4
0.44
u And c
2.0
0.8
44.0
0.61
u And d
4.0
2.5
44.0
0.19
g Cep b
1.6
2.0
45.0
0.15
47 UMa b
2.6
2.1
45.6
0.15
47 UMa c
0.5
3.4
45.6
0.24
HD 160691 c
3.1
4.2
49.9
0.27
a= semi-major axis, d = distance (light
years), D = planet-star separation
Direct Detections need
contrast ratios of 10–9 to
10–10
At separations of 0.01 to
1 arcseconds
Earth : ~10–10 separation = 0.1
arcseconds for a star at 10
parsecs
Jupiter: ~10–9 separation = 0.5
arcseconds for a star at 10 parsecs
1 AU = 1 arcsec separation
at 1 parsec
Younger planets are hotter and they emit more radiated light. These are
easier to detect.
A Little Background:
Fourier Transforms
f(x) =  F(s) e2pixs ds
F(s) =  f(x) e−2pixs dx
The Fourier transform of a function (frequency spectrum)
tells you the amplitude (contribution) of each sin (cos)
function at the frequency that is in the function under
consideration.
The square of the Fourier transform is the power spectra
and is related to the intensity when dealing with light.
Fourier Transforms
Two important features of Fourier transforms:
1) The “spatial or time coordinate” x maps into a “frequency”
coordinate 1/x (= s or n)
A function that is narrow in x is wide in s
A Pictoral Catalog of Fourier Transforms
Time/Space Domain
Time
Fourier/Frequency Domain
0
Frequency (1/time)
Period = 1/frequency
Comb of Shah function
(sampling function)
x
1/x
Time/Space Domain
Fourier/Frequency Domain
Negative
frequencies
Cosine is an even function:
cos(–x) = cos(x)
Positive
frequencies
Time/Space Domain
Sine is an odd function: sin(–x)
= –sin(x)
Fourier/Frequency Domain
Time/Space Domain
e–px
w
Fourier/Frequency Domain
2
e–ps
2
1/w
The Fourier Transform of a Gausssian is another Gaussian. If the
Gaussian is wide (narrow) in the temporal/spatial domain, it is
narrow(wide) in the Fourier/frequency domain. In the limit of an infinitely
narrow Gaussian (d-function) the Fourier transform is infinitely wide
(constant)
Time/Space Domain
All functions are interchangeable. If
it is a sinc function in time, it is a slit
function in frequency space
Fourier/Frequency Domain
Note: these are the diffraction
patterns of a slit, triangular and
circular apertures
Fourier Transforms : Convolution
Convolution
 f(u)f(x–u)du = f * f
f(x):
f(x):
Cross Correlation
f(x-u)
a2
a1
a3
g(x)
CCF
a3
a2
a1
Background: Fourier Transforms
In Fourier space the convolution (smoothing of a
function) is just the product of the two transforms:
Normal Space
f*g
Fourier Space
F×G
Suppose you wanted to smooth
your data by n points.
x
You can either:
1. Move your box to a place in your data, average all the points in that box
for value 1, then slide the box to point two, average all points in box and
continue.
2. Compute FT of data, the FT of box function, multiply the two and inverse
Fourier transform
Fourier Transforms
The second important features of Fourier transforms:
2) In Fourier space the convolution is just the product of the
two transforms:
Normal Space
f*g
Fourier Space
F G
f ×g
F*G
sinc
sinc2
Adaptive Optics : An important component
for any coronagraph instrument
Seeing →
0.25“
0.5“
1“
2“
Atmospheric turbulence distorts stellar images making them much larger than
point sources. This seeing image makes it impossible to detect nearby faint
companions.
Adaptive Optics
The scientific and engineering discipline whereby the performance of an optical
signal is improved by using information about the environment through which it
passes
AO Deals with the control of light in a real time closed loop and is a subset of active
optics.
Adaptive Optics: Systems operating below 1/10 Hz
Active Optics: Systems operating above 1/10 Hz
Example of an Adaptive Optics
System: The Eye-Brain
The brain interprets an image, determines its correction, and
applies the correction either voluntarily of involuntarily
Lens compression: Focus corrected mode
Tracking an Object: Tilt mode optics system
Iris opening and closing to intensity levels: Intensity control
mode
Eyes squinting: An aperture stop, spatial filter, and phase
controlling mechanism
The Ideal Telescope
This is the Fourier transform
of the telescope aperture
where:
• P(a) is the light intensity in the focal plane, as a function of angular coordinates a ;
• l is the wavelength of light;
• D is the diameter of the telescope aperture;
• J1 is the so-called Bessel function.
The first dark ring is at an angular distance Dl of from the center.
This is often taken as a measure of resolution (diffraction limit) in an ideal telescope.
Dl = 1.22 l/D = 251643 l/D (arcsecs)
Diffraction Limit
Telescope
5500 Å
2 mm
10 mm
Seeing
TLS 2m
0.06“
0.2“
1.0“
2“
VLT 8m
0.017“
0.06“
0.3“
0.2“
Keck 10m
0.014“
0.05“
0.25“
0.2“
ELT 42m
0.003“
0.01“
0.1“
0.2“
Even at the best sites AO is needed to improve image quality and reach
the diffraction limit of the telescope. This is easier to do in the infrared
Atmospheric Turbulence
A Turbulent atmosphere is characterized by eddy (cells) that decay
from larger to smaller elements.
The largest elements define the upper scale turbulence Lu which is the
scale at which the original turbulence is generated.
The lower scale of turbulence Ll is the size below which viscous effects
are important and the energy is dissipated into heat.
Lu: 10–100 m
Ll: mm–cm (can be ignored)
Atmospheric Turbulence
Original wavefront
• Turbulence causes temperature
fluctuations
• Temperature fluctuations cause
refractive index variations
- Turbulent eddies are like
lenses
• Plane wavefronts are wrinkled
and star images are blurred
Distorted wavefront
Atmospheric Turbulence
ro: the coherence length or „Fried parameter“ is
r0 = 0.185 l6/5 cos3/5z(∫Cn² dh)–3/5
ro is the maximum diameter of a collector before atmospheric distortions
limit performance (l is in meters and z is the zenith distance)
r0 is 10-20 cm at zero zenith distance at good sites
To compensate adequately the wavefront the AO should have at least
D/r0 elements
Definitions
to: the timescale over which changes in the atmospheric turbulence
becomes important. This is approximately r0 divided by the wind
velocity.
t0 ≈ r0/Vwind
For r0 = 10 cm and Vwind = 5 m/s, t0 = 20 milliseconds
t0 tells you the time scale for AO corrections
Definitions
Strehl ratio (SR): This is the ratio of the peak intensity observed at
the detector of the telescope compared to the peak intensity of the
telescope working at the diffraction limit.
If D is the residual amplitude of phase variations then
D = 1 – SR
The Strehl ratio is a figure of merit as to how well your AO
system is working. SR = 1 means you are at the diffraction
limit. Good AO systems can get SR as high as 0.8. SR=0.3-0.4
is more typical.
Definitions
Isoplanetic Angle: Maximum angular separation (q0) between two
wavefronts that have the same wavefront errors. Two wavefronts
separated by less than q0 should have good adaptive optics
compensation
q0 ≈ 0.6 r0/L
Where L is the propagation distance. q0 is typically about
20 arcseconds.
If you are
observing an
object here
You do not want to correct using a
reference star in this direction
Basic Components for an AO System
1. You need to have a mathematical model representation of the
wavefront
2. You need to measure the incoming wavefront with a point
source (real or artifical).
3. You need to correct the wavefront using a deformable mirror
Describing the Wavefronts
The aberrated wavefront is compared to an ideal spherical
wavefront called a the reference wavefront. The optical path
difference (OPD) is measured between the spherical reference
surface (SRS) and aberated wavefront (AWF)
The OPD function can be described by a polynomial where each
term describes a specific aberation and how much it is present.
Describing the Wavefronts
Zernike Polynomials:
Z= SKn,m,1rn cosmq + Kn,m,2rn sinm q
Measuring the Wavefront
A wavefront sensor is used to measure the aberration function W(x,y)
Types of Wavefront Sensors:
1. Foucault Knife Edge Sensor (Babcock 1953)
2. Shearing Interferometer
3. Shack-Hartmann Wavefront Sensor
4. Curvature Wavefront Sensor
Shack-Hartmann Wavefront Sensor
Shack-Hartmann Wavefront Sensor
Lenslet array
Image Pattern
Focal Plane
detector
reference
af
disturbed
a
f
Shack-Hartmann Wavefront Sensor
Correcting the Wavefront Distortion
Adaptive Optical Components:
1. Segmented mirrors
Corrects the wavefront tilt by an array of mirrors. Currently up to
512 segements are available, but 10000 elements appear
feasible.
2. Continuous faceplate mirrors
Uses pistons or actuators to distort a thin mirror (liquid mirror)
Unperturbed wavefront
Wavefront at telescope
Liquid
Mirror
wavefront
sensor
corrected
wavefront to
camera
Reference Stars
You need a reference point source (star) for the wavefront
measurement. The reference star must be within the isoplanatic angle,
of about 10-30 arcseconds
If there is no bright (mag ~ 14-15) nearby star then you must use an
artificial star or „laser guide star“.
All laser guide AO systems use a sodium laser tuned to Na 5890 Å
pointed to the 11.5 km thick layer of enhanced sodium at an altitude of
90 km.
Much of this research was done by the U.S. Air Force and was
declassified in the early 1990s.
dhNa
Mesospheric
Sodium Layer
a
dhNa ≈ 11.5 km
hNa ≈ 90 km
H ≈ 4 km
hNa
Average seeing layer
H
LaserL Lab
Telescope
d
Applications of Adaptive Optics
1. Imaging
Sun, planets, stellar envelopes and dusty disks, young
stellar objects, etc. Can get 1/20 arcsecond resolution in
the K band, 1/100 in the visible (eventually)
Applications of Adaptive Optics
2. Resolution of complex configurations
Globular clusters, the galactic center, stars in the spiral arms
of other galaxies
Applications of Adaptive Optics
3. Detection of faint point sources
Going from seeing to diffraction limited observations
improves the contrast of sources by SR D2/r02. One will
see many more Quasars and other unknown objects
Applications of Adaptive Optics
4. Faint companions
The seeing disk will normally destroy the image of faint
companion. Is needed to detect substellar companions
(e.g. GQ Lupi)
Applications of Adaptive Optics
5. Coronography
With a smaller image you can better block the light. Needed
for planet detection
Coronagraphs
Basic Coronagraph
Dl = D/l = number of wavelengths across the telescope aperture
b)
The telescope optics then forms the incoming wave into an image. The
electric field in the image plane is the Fourier transform of the electric
field in the aperture plane – a sinc function (in 2 dimensions this is of
course the Bessel function)
Eb ∝ sinc(Dl, q)
Normally this is where we place the detector
c) d)
In the image plane the star is occulted by an image stop. This stop has
a shape function w(Dlq/s). It has unity where the stop is opaque and
zero where the stop is absent. If w(q) has width of order unity, the stop
will be of order s resolution elements. The transfer function in the
image planet is 1 – w(Dlq/s).
W(q) = exp(–q2/2)
e)
The occulted image is then relayed to a detector through a second
pupil plane e)
This is the convolution of the step function of the original pupil
with a Gaussian
e)
f) g)
One then places a Lyot stop in the pupil plane
At h) the detector observes the Fourier transform of the second pupil
Difference Imaging : Subtracting the
Point Spread Function (PSF)
To detect close companions one has to subtract the PSF of the central star
(even with coronagraphs) which is complicated by atmospheric speckles.
One solution: Differential Imaging
Planet Bright
Planet Faint
Since the star has no methane, the PSF in all
filters will look (almost) the same.
Spectral Differential Imaging (SDI)
1.58 mm
1.68 mm
1.625 mm
Split the image with a beam splitter. In one beam place a filter where the planet is
faint (Methane) and in the other beam a filter where it is bright (continuum). The
atmospheric speckles and PSF of the star (with no methane) should be the same in
both images. By taking the difference one gets a very good subtraction of the PSF
Results!
Coronography of Debris Disks
Structure in the disks give hints to the presence of sub-stellar
companions
Coronographic Detection of a Brown Dwarf
Cs
Spectral Features show Methane and Water
The Planet Candidate around GQ Lupi
But there is large
uncertainty in the surface
gravity and mass can be as
low as 4 and as high as 155
MJup.
Another brown dwarf detected with the NACO adaptive optics system on the VLT
M = 4 MJup
Estimated mass from
evolutionary tracks: 13-14 MJup
Coronographic observations with HST
a ~ 115 AU
P ~ 870 years
Mass < 3 MJup,
any more and
the gravitation
of the planet
would disrupt
the dust ring
Photometry of Fomalhaut b
Planet model with
T = 400 K and R =
1.2 RJup.
Reflected light from
circumplanetary disk
with R = 20 RJup
Detection of the
planet in the optical
may be due to a disk
around the planet.
Possible since the
star is only 30 Million
years old.
SPITZER Observations
of Fomalhaut at 4.5 mm
Marengo et al. 2009
Not detected in the
Infrared. Limits of 3 MJup
and age of 200 Million
years
2010
2012
Galaxies
Planet
Kalas et al. 2012
Recent observations by Kalas using HST confirm presence of planet.
Imaged using
Angular
Differential
Imaging (i.e.
Spectral
Differential
Imaging)
Image of the planetary system around HR 8799 taken with a
„Vortex Phase“ coronagraph at the 5m Palomar Telescope
A fourth planet has also been detected around HR 8799
The 2009-2010 orbital motions of the four planets are shown in the larger plot. A square
symbol denotes the first 2009 epoch. The upper-right small panel shows a zoomed version
of e's astrometry including the expected motion (curved line) if it is an unrelated background
object. Planet e is confirmed as bound to HR 8799 and it is moving 46 ± 10 mas/year
counter-clockwise. The orbits of the solar system's giant planets (Jupiter, Saturn, Uranus
and Neptune) are drawn to scale (light gray circles). With a period of ~50 years, the orbit of
HR 8799e will be rapidly constrained by future observations; at our current measurement
accuracy it will be possible to measure orbital curvature after only 2 years.
HR 8799 Compared to Our Solar System
asteroid belt
The Planet around b Pic
Mass ~ 8 MJup
2003
2009
UScoCTIO 108
Some Imaging Planets
Planet
Mass
(MJ)
Period
(yrs)
a
(AU)
e
Sp.T.
Mass
Star
2M1207b
4
-
46
-
M8 V
AB Pic
13.5
-
275
-
K2 V
GQ Lupi
4-21
-
103
-
K7 V
0.7
b Pic
8
12
~5
-
A6 V
1.8
HR 8799 b
7
465
68
-
F2 V1
HR 8799 c
10
190
38
´-
HR 8799 d
10
10
24
-
HR 8799 e
9
50
14.5
Fomalhaut b
<3
88
115
-
A3 V
0.025
2.06
lists this as an A5 V star, but it is a g Dor variable which have spectral types
F0-F2. Tautenburg spectra confirm that it is F-type
1SIMBAD
Darwin
Beam combiner
Data storage and transfer
station
Darwin will use Nulling
Interferometry to image
terrestrial planets
Earth
Venus
Mars
Advantages:
Summary
1. Finds planets at large orbital radii. This fills an
important region of the parameter space
inaccessible with other methods.
2. Can get spectroscopy of the planet directly
3. Can Planets around hot stars as well
4. Seeing is believing!
Disadvantages:
1.Only works for nearby stars
2.Planet mass relies on evolutionary tracks that are
model dependent – mass uncertain!
3.Orbital parameters poorly known (wait a long time!)
4.Only massive and young planets detected so far
5.Only planets far from the star have been detected
 Smaller, close in planets will require space
missions or extemely large telescopes (30m)