Transcript Diffraction-Limited Imaging in the Visible On Large Ground-Based Telescopes Craig Mackay, Institute of Astronomy, University of Cambridge. 22 March, 2012: Open University.

Diffraction-Limited Imaging in the Visible
On Large Ground-Based Telescopes
Craig Mackay,
Institute of Astronomy, University of Cambridge.
22 March, 2012: Open University
Introduction and Outline
• Will talk about the extraordinary importance of CCDs to
scientific progress.
• Going fainter than ever before.
• Higher resolution imaging than ever before.
• Diffraction limited imaging on very large telescopes.
• Electron multiplying CCDs: their care and feeding.
• Applications of CCDs in physical and life sciences.
22 March, 2012: Open University
Going Faint with CCDs
• The first image forming CCDs were built around 1973, and
first used for astronomy soon after that.
• In Cambridge we started using GEC (now E2V) CCDs in the
late 1970s and built cameras that were used for astronomy in
1979.
• They were small (385 x 578 pixels), front illuminated, noisy
(20 electrons) and cosmetically dreadful by modern
standards (35% peak to peak non-uniformity).
• But they did have high quantum efficiency so good science
could be done with some care and effort.
• Results from 1981/1982 pushed the sensitivity limits for
faint galaxy counts deeper by a factor of ~300.
22 March, 2012: Open University
Drift-Scan Correction
• The Hubble Space Telescope (HST) will
not last forever.
• Mechanically
track the CCD
in synchronism
with the
progressive
parallel transfer
of charge.
• Each pixel of
the sky detected
with the mean
gain of all the
pixels in the
column.
• Gives detector
nonuniformities
<0.2% of sky.
Ultra Faint Galaxy Counts
• Observations were
done in March, 1982,
30 years ago.
• These were the
deepest galaxy counts
from 1984 to 1999.
• Down to R ~25.7.
• Bettered only by
Hubble Deep Field,
342 exposures over
10 days.
• Nearly x100 times
fainter.
Improving Resolution Using
Adaptive Optics
• For many years, instrumentalists
have tried to overcome
atmospheric degradation in image
quality.
• Adaptive optics techniques tries
to measure wavefront distortions
on scales of ~r0 (perhaps 20 cm)
and correct them on timescales of
~t0 (perhaps 10 ms).
• Usually Shack-Hartmann
wavefront sensors are used.
22 March, 2012: Open University
Why is AO so hard?
• Shack-Hartmann wavefront sensor is often used.
• Breaks up the pupil into many small cells, ~20-50 cm diameter.
• Each forms an image of a bright star tracked to deduce the
wavefront errors.
• Starlight divided amongst many cells so the reference star must be
bright.
• Need fast read-out as atmosphere
changes rapidly (~wind crossing
time for one cell, so <10 ms).
• Need to determine errors and
correct them before they all
change.
22 March, 2012: Open University
Angular Resolution Using Adaptive Optics
• Despite the ~$1B5 spent worldwide on AO, it only works
adequately in the near infrared.
• Bright (<13 mag) reference stars are essential for
conventional AO so sky coverage is poor.
• No AO system has yet delivered Hubble resolution images
(0.12 arcsec resolution) on a Hubble size (2.5 m) telescope.
• Our goals are near-diffraction limited resolution with very
faint reference stars for ~full sky coverage in the visible.
22 March, 2012: Open University
Why is AO so hard? Seeing is terribly variable.
• Image resolution
over a run of 85
seconds ~good
seeing.
• Changes by factor
of two in a few
frames (~ wind
crossing time of
telescope).
• AO systems would
struggle to follow
many of these
steps.
22 March, 2012: Open University
Why is AO so hard?
• This requires a bright reference star, typically 12-13 magnitude
(very scarce, ~0.1% sky coverage).
• Below this threshold, as star gets fainter, all cells loose lock
together, so cannot do any correction below a specific
threshold.
• Even at small angles away from the reference star, the
turbulence correction becomes uncorrelated.
• This gives a tiny isoplanatic patch size (~few arcsec in the
visible on the good site).
• Much easier in the near infrared because cells can be larger,
read rate slower and reference stars are brighter.
22 March, 2012: Open University
Lucky Imaging in the Visible.
• Technique originally suggested by Hufnagel (1966) and
developed by Fried (1978).
• Images taken fast enough to freeze the motion due to turbulence.
• On a 2.5 m telescope (the NOT) in I band, on a good site (LPO),
under typical conditions 10-30% of images are ~ diffraction
limited at 20 frames per sec.
• The best images are selected and combined to give a neardiffraction limited image.
• The isoplanatic patch size is much larger than with AO, typically
~ 60 arcsec rather than ~3-5 arcsec diameter.
22 March, 2012: Open University
Results with Lucky Astronomy
• 100Her is a double
star with 14 arc sec
separation.
• Here the two
components are
shown side by side.
• The scale is about 4
arc sec vertically
• Images were taken
with 10 millisec
frame time, and stars
are each 6.0
magnitude.
22 March, 2012: Open University
The Einstein Cross
• The image on the left is from the Hubble Space Telescope Advanced Camera for Surveys
(ACS) while the image on the right is the lucky image taken on the NOT in July 2009 through
significant amounts of dust.
• The central slightly fuzzy object is the core of the nearby Zwicky galaxy, ZW 2237+030
that gives four gravitationally lensed images of a distant quasar at redshift of 1.7
22 March, 2012: Open University
New Results with Lucky Astronomy
(Images courtesy Wah!, Hong-Kong)
• Techniques are
also very popular
with amateur
astronomers.
• This shows a short
movie of the moon
taken under poor
conditions (roof of
skyscraper in
Hong Kong!).
• Wah! used
Registax Lucky
software.
22 March, 2012: Open University
New Results with Lucky Astronomy
(Images courtesy Wah!, Hong-Kong)
• Techniques are
also very popular
with amateur
astronomers.
• This shows a short
movie of the moon
taken under poor
conditions (roof of
skyscraper in
Hong Kong!).
• Wah! used
Registax Lucky
software.
22 March, 2012: Open University
New Results with Lucky Imaging
• Image of the
International Space
Station, with Space
Shuttle Atlantis &
Soyuz, June 2007.
• Resolution ~20 cm at
an altitude of 330 km
altitude, or ~ 0.12
arcsec.
• Downward looking
resolution is much
better, ~20 milliarcsecs
or ~ 2 cm.
22 March, 2012: Open University
Large Telescope Lucky.
• Lucky imaging techniques
on larger telescopes will not
work.
• How to improve our luck?
• Remove much of the
turbulent power with a low
order AO system.
• We used the Palomar 5 m
telescope low-order
adaptive optics system plus
our Lucky Imaging camera.
22 March, 2012: Open University
Large Telescope Lucky Imaging.
• Globular cluster M13
on the Palomar 5m.
• Seeing ~650 mas.
• PALMAO system and
our EMCCD Camera.
• Achieved 17% Strehl
ratio in I-band, giving
~35 mas resolution.
• This is the highest
resolution image ever
taken in the visible.
22 March, 2012: Open University
Large
Telescope
Lucky
Imaging.
• Compare Lucky/AO and Hubble
Advanced Camera (ACS) is quite
dramatic.
• The Lucky/AO images have a
resolution ~35 milliarcseconds or
nearly 3 times that of Hubble.
22 March, 2012: Open University
Large Telescope Lucky.
•
•
•
•
•
AO usually needs a bright reference star.
Building a new kind of wavefront curvature sensor.
Much more sensitive than S-H sensors for low-order AO.
We use 4 out-of-pupil images, and fit a wavefront curvature.
Can work with reference objects x100-1000 fainter. Is
substantially achromatic.
(From Olivier Guyon, Subaru
telescope, Hawaii).
22 March, 2012: Open University
Lucky/AO Imager for the WHT.
• Building visitor instrument for WHT 4.2m and the GTC 10.5m.
• Will allow a wide range of problems to be tackled that require
>HST resolution in visible.
• Examples include globular cluster physics, quasar host galaxies,
AGN studies, compact gravitational lenses, MACHO surveys in
crowded regions and many others.
• Also works as high-time resolution instrument.
• Photon-counting CCDs allow limited fields at 1000Hz.
• May also be used with Integral Field Unit (IFU) based
spectrographs.
• First light in 18 months, complete in 3 years.
22 March, 2012: Open University
Lucky/AO Imager for the WHT.
Key technologies are:
• Electron Multiplying CCDs. They run
up to 30 MHz pixel rate, are thinned and
~zero read noise so count photons.
• Used both for wavefront detectors and
science detectors.
• Re-imager gives 2000 x 2000 px field of
view of 12 x 12 to 30 x 30 arcsec.
• Deformable mirror wavefront corrector.
• Large RAID arrays for data storage (200
Mb/sec continuous), plus NVIDIA Tesla
parallel processors for real-time
processing (512 64-bit FP processors,
3x109 transistors).
22 March, 2012: Open University
Introduction to EMCCDs: General Characteristics
• EMCCDs are
standard CCDs plus
an electron
multiplication stage.
• One serial electrode
runs at high voltage
(~45 volts).
• An electron has a
low probability
(~1%) of a 1
electron avalanche.
• Gives 604 1.01 or 1.02
~few x100 gain.
22 March, 2012: Open University
Photon Counting with EMCCDs
• All these tests are
with our own camera
design (available from
www.pixcellent.com).
• A big overscan
separates parallel
register effects from
serial register effects.
• A cut across an
EMCCD image at
very low signal level
and high gain shows a
wide range of events
sizes.
22 March, 2012: Open University
Photon Counting with EMCCDs
• The additional variance introduced by the multiplication stage
increases the amount of noise.
• Normally expect the SNR=√ (N).
• With the EMCCD the SNR=√ (2N).
• Equivalent to halving the DQE of the device.
• However, if we threshold the image and replace each event by a
single value then added noise is eliminated and DQE restored.
• In photon counting mode, long-term absolute gain stability is
much less important. An event is an event is an event.
22 March, 2012: Open University
Photon Counting with EMCCDs
• We can also tolerate some
loss in performance when
we run at pixel rates well
above those recommended.
• With the CCD201, E2V
give a max. speed of 13-20
MHz,
• All our results here use 30
MHz pixel rate.
• The gain may be
determined by looking at
the statistics of the photon
events sizes detected.
22 March, 2012: Open University
Solid-State Detectors: Applications
•
Application areas include
 life sciences (automated DNA sequencing,
protein electrophoresis, gene probe work,
monoclonal antibody searches).
 physical sciences (x-ray imaging for
materials inspection, electron beam imaging).
 medical sciences (dental, internal
examinations, x-ray fluoroscopy).
 surveillance (security, battlefield, crime
prevention and detection).
Automated DNA Sequencing
• The use of cooled, slowscan
CCD cameras for DNA
sequencing work patented in
Cambridge.
• Variant on drifts scan/TDI
technology.
• Every machine used for the
Human Genome Project used
one of these patented cameras.
• Many thousands of these
machines have been
manufactured, each costing
between £250K and £800K.
Two-Dimensional Gel Electrophoresis
• Proteins are uniquely
defined by their mass and
electric charge.
• Separate in two
dimensions allows analysis of
protein samples.
• Proteins fluorescently
labelled are much easier to
image.
• Specific protein patterns
associated with specific
diseases.
• Key for the development
of pharmaceuticals.
Other Life Science Applications
•
In optical microscopy :
• Reduces the amount of
fluorescent dye and UV
illumination.
• Bioluminescence shows
diseased leaves long before
they are visible to the eye
(far right).
• Gene probe
technologies allow
screening in high-volume
DNA sequencing research.
Material Science Applications
• Cooled CCDs greatly
improves x-ray imaging systems
for very low contrast images.
• Composite plastics with
hollow glass microspheres are
used in stealth aircraft.
• Ceramic ball bearings used in
next-generation jet engines.
• Micro cracks are very faint
(bottom right) and can cause
damaging disintegration.
22 March, 2012: Open University
Medical Science Applications
• Fluorescein angiography
has revolutionised treatment
for atherosclerosis.
• Many systems use cooled
CCD cameras along the lines
patented, also in Cambridge,
over 20 years ago.
• Systems are increasingly
wide field and are beginning
to use amorphous silicon
technology which is a direct
spin out from the
CCD/CMOS world.
22 March, 2012: Open University
Surveillance with Ground-Based Lucky Imaging
• Target is 33mm text
spacing at 400m range.
• Images show selections
of 100%, 10%, 1% and 1%
post-processing.
• The resolution
improvement is dramatic.
•We typically are able to
reduce the effective distance
of a target by a factor of 812.
22 March, 2012: Open University
Conclusions
• The development of CCDs has impacted almost every scientific
discipline.
• Now very widely used and of extraordinary quality.
• EMCCDs offer very high performance at the lowest signal
levels ever with two-dimensional imaging systems.
• In photon counting mode we achieve maximum DQE.
• Even in analog mode the excellent read noise achievable can
allow operation at extremely low signal levels indeed.
• These cameras can offer astronomers and scientists in other
areas the opportunity to carry out entirely new kinds of research
at the very faintest signal levels.
22 March, 2012: Open University
Instrumentation Group
Institute of Astronomy
University of Cambridge, UK
[email protected]
22 March, 2012: Open University