instrumentation concepts ground-based optical telescopes
DESCRIPTION
Instrumentation Concepts Ground-based Optical Telescopes. Keith Taylor (IAG/USP) Aug-Nov, 2008. Imaging considerations. Trading field of view vs. angular resolution A large field at coarse spatial resolution or smaller field of view at high fidelity? - PowerPoint PPT PresentationTRANSCRIPT
Instrumentation Concepts
Ground-based Optical Telescopes
Keith Taylor(IAG/USP)
Aug-Nov, 2008
Aug-Sep, 2008 IAG-USP (Keith Taylor)
Imaging considerations
Trading field of view vs. angular resolution A large field at coarse spatial resolution or
smaller field of view at high fidelity? Generally detector defines fixed pixel format
Constraints also driven by seeing and pixel scale (or camera f-ratio)
Photometric accuracy Simple morphological discrimination or accurate
flux measures as well? Definition of passbands and central
wavelengths Multiple simultaneous passbands?
Astronomical CCD Imaging
Simplest astronomical instrument (in principle)
Undispersed 2D images of field of view
Generally use filters to limit spectral band-pass Detector itself may
supply band-width Polarimetric capabilities?
Rapid reads can give limited time resolution information
Standardized Filter Systems
Variety of different filter systems prevalent in optical/IR domain. eg: UBVRI / JHKLM – Johnson/Cousins
(UV/optical) (NIR) u g r i z - Sloan Digital Sky Survey (SDSS)
optical bands Extras and modifications
y - UKIDSS IR band K’ and K* - modifications to K to avoid thermal
radiation HST, Spitzer etc. defined by wavelength
rather than name (revolutionary!)
Examples of CCD Imaging
CCD = Charge-coupled deviceMore sensitive than photographic plates by factors ~50
Data can be read directly into computer memory, allowing digital enhancement and manipulation
Negative image can enhance contrasts
False-color image to visualize brightness contours
cf: Photographic Imaging(eg: AAT c1990)
14inch (=350mm)• ~5 Giga Pixels
cf: Largest CCD currently:• 8 Mega Pixels
(4k-by-2k)• CCD arrays up to
~0.1 Giga Pixels
Generally seeing-limited
• How do we obtain higher spatial definition?
Adaptive OpticsComputer-controlled “adaptive” mirror adjusts the mirror surface (many times per second) to compensate for distortions induced by atmospheric turbulence
And yet furtherspatial resolution?
InterferometryRecall: Resolving power of a telescope depends on diameter D:
min = 1.22 /D.
This holds true even if not the entire surface is filled out.
Sparsely filled aperture:
•Combine the signals from several smaller telescopes to simulate one big mirror
Interferometry
Spectroscopic considerations
What kind of spectral feature are of interest? Emission or absorption lines; continuum
shapes Broad, narrow or spectrally unresolved Low or high contrast with continuum Spectral Energy Distributions (SEDs)
Line centres ; equivalent widths ; line shapes ; kinematic mapping? and/or precise spectrophotometry?
One or many targets simultaneously?
The simplest spectrographUsing a prism (or a grating), light can
be split up into different wavelengths (colors!) to produce a spectrum.
Spectral linesSpectral lines in a spectrum in a spectrum tell us about the chemical tell us about the chemical composition and other composition and other properties of the observed properties of the observed object object
Typical grating spectrograph
Simple grating spectrograph
Spectrum extracted along a slit so ‘imaging’ in one dimension
Off source light along slit used to measure and subtract sky background
What you get Optical long slit
spectrum of a galaxy Minimal data
reduction in displayed spectral image
Can see galaxy, bad pixels, cosmic ray hits and sky lines Need off source
signal to measure and extract target (sky subtraction)
Sky lines
Target
Considerations for Spectroscopy
Basic parameters - resolution and central wavelength for spectrum
Slit width (if selectable) affects resolution Wavelength range
Set by combination of detector geometry and spectral resolution
Some spectrographs provide large -range at low-R; others provide only a few 1,000kms-1 range, so centering on a critical line of interest (eg: H)
But, what if you need both high-R and large -range?
High Resolution and lots ofSpectrum
• X-dispersed echelle grating spectrometers allow high resolution and lots of spectral coverage• Achieve this by having two
orthogonal gratings• One gives the high resolution (in y-axis) the other spreads the spectrum across the detector(in x-axis)• However, the slit is consequently much shorter
STELES echelle spectrograph(for SOAR)
Primary disperser (echelle grating)
Secondary (orthogonal) disperser (VPHG)
Redchannel
Bluechannel
Multiobject Spectroscopy
To get spectra for lots of objects at once. Can use two approaches Multislit - have several slits in the
image plane and get spectra for all of them
Use fibres to pipe light from different parts of the focal plane while reformatting them along the spectrograph slit
Both techniques were developed in the 80s and perfected in the 90s
Fibre Fed Systems
AAT 2dF (now AAT 2dF (now replaced by replaced by AAOmega)AAOmega) Pickoff fibres Pickoff fibres
positioned by robotpositioned by robot Include sky fibres Include sky fibres
for each objectfor each object
Multi-slit spectroscopy
Example of multislit spectral image
Easier to achieve at telescope (can use holes in a mask) but preparation and reduction can be more complex
Need to ensure spectra don’t overlap
LDSS-2mask
superimposed on sky image
Great care has to be taken in selecting objects to study so that they don’t overlap in wavelength direction.
Also need objects of similar brightness so the SNRs are similar.
Mask optimization is NOT trivial!
Field acquisition is NOT trivial
But what if you want images and spectroscopy simultaneuously?
Integral Field Spectroscopy Extended (diffuse) object with lots of spectra
Use “contiguous 2D array of fibres or ‘mirror slicer’ to obtain a spectrum at each point in an image
SIFS
Tiger
MPI’s 3D
Large-field imagingspectrographs
Narrow band filters Image a field in a single narrow band Use enough narrow bands and you
have very low res. spectroscopy Fabry-Perot
Effectively acts as a narrow tunable filter
Can thus image a field in emission lines of choice (eg. TTF)
Fabry-Perot Light enters etalon and is
subjected to multiple reflections
Transmission spectrum has numerous narrow peaks at wavelengths where path difference results in constructive interference need ‘blocking filters’ to use as
narrow band filter Width and depth of peaks
depends on reflectivity of etalon surfaces: finesse
Fourier Transform Spectrometer
• As translation mirror scans an interference pattern is produced that is the FT of the source spectrum• Scan distance defines the resolution of the spectrum• Advantage - get spectrum of whole field• Disadvantage - get broad band noise
IFTS for NGST
Detectors for Opical/near-IR
(current) Photon Counters:
Image tube + TV camera + real-time discrimination (not solid state)
eg: IPCS - c70s to c80s CCDs now dominate - Hi QE but …
Integrate signal on detector – no time resolution Finite read-noise Finite read-time
EMCCDs – new generation of Photon Counters CCD-like QEs V. high frame-rates
DQE - the key to gooddetectors
Detector quantum efficiency - the fraction of incident photons detected - is the key measure for the effectiveness of a detector;
Traditional photographic plates, while large in size, have DQE of only about 10%
CCDs and similar semiconductor devices can have DQE as high as 90% (though wavelength dependent) Like having a telescope with 9 times the
collecting area
CCDs
CCDs combine photon detection with integration and multiplexing
Incident photons excite charge carriers which are stored and integrated in a capacitor
CCDs are also uniquely effective in transferring charge from 2D to 1D charge ‘clocked’ from pixel to pixel and
read out at fixed point ideal for multiplexing
CCD Array Camera
Semiconductor fabrication limits the size of a CCD detector
To get a large area need to mosaic detectors together
Subaru Mosaic CCD Camera
Near-IR Detectors
CCDs use Silicon as their substrate Valance to conduction bandgap in silicon is 1.1eV so
restricted to detecting photons with wavelength < 1 micron
Need different materials for infrared InSb for 1 to 5 micron, HgCdTe for 1 to 2.5 micron Detector elements bonded to Si CCD system to provide
multiplexing readout
IR Arrays vs. Optical
IR arrays are smaller, more expensive (by factor of ~10/pixel)
Readout has to be faster because of higher backgrounds
Use of different materials can push to longer wavelengths More difficult to work with, less helpful
characteristics, more expensive At longest wavelengths have to stress the
detector to produce lower energy band gaps
Lecture 1 (2 sessions) - Synopsis
Fundamental Principles Introduction Information Theory Seeing-limited observations Diffraction-limited observations Signal-to-noise estimates
Lecture 2
Fundamentals
Aug-Nov, 2008 IAG/USP (Keith Taylor)
UVOIR Astronomy
Definition: UVOIR = the "UV, Optical, Near-Infrared"
region of EM spectrum Shortest wavelength: 912 Å (or 91.2 nm) --- Lyman
edge of H I; interstellar medium is opaque for hundreds of Å below here
Longest wavelength: ~3µm (or 3000 nm) --- serious H2O absorption in Earth's atmosphere above here
Ground-based UVOIR: 0.3µm (or 300nm) < < 2.5µm (or 2,500nm)
UVOIR Astronomy
Uniqueness: Best developed instrumentation; Best understood astrophysically; Highest density of astrophysical information; Provides prime diagnostics on the two most important
physical tracers.
===> UVOIR observations/identifications are almost always prerequisites to a
thorough understanding of cosmic sources in other EM bands.
Proof
Stars Plasma (to 105K)
Observational Priorities
Astronomy driven by discoveries rather than theoretical insights Direction of field shaped by observations in about
3/4 of instances. Few important astronomical discoveries were
predicted; many were actually accidental
Accidental Discoveries
Uranus Expanding universe Pulsars Supermassive black holes/AGN‘s Large scale structure Dark matter in spiral galaxies X-ray emitting gas in clusters of galaxies Gamma ray bursts Extra-solar planets High redshift evolution of galaxies
HST contributions were actually hindered by theoretical prejudice. A deep pencil-beam survey was delayed by 5 years.
Counterexamples: theory-driven discoveries Neptune General relativistic distortion of space-
time near Sun 21 cm line of HI Helioseismology Cosmic microwave background
ConclusionsIs Observational Astronomy a Science?
(Build, don’t Think)
Technology drives Discovery
Key technology development for UVOIR astronomy: 17th century: telescopes 19th century: spectroscopy, photography, quality
lens making, large structure engineering 20th century: large mirror fabrication,
electronic detectors, digital computers, space astronomy
Since 1980: array detectors
Telescope size: determines ultimate
sensitivity Diameter doubling time ~45 years Largest telescopes now 8-10m diameter
Collecting area of 10-m is 4*106 that of the dark-adapted eye
In planning: 15-m to 40-m For a given technology, cost D2.6
Cost is roughly proportional to mass Even using new technologies, next
generation of large ground-based telescopes will cross the $1 billion threshold.
The Future?
NB: Number of ground-based telescopes is NOT inversely proportional to their size Almost as many 8m
telescopes as there are 4m telescopes
How many 30m telescopes are there going to be in the next 50 years? (at US$1B a pop)
Review of some Basics• c = ν = 3.1010 cm/s• E = hν (ergs)• F = L/4d2
• G = 6.67.10-8 (c.g.s)• c = 3.1010 cm.s-1
• k = 1.38.10-16
• h = 6.626.10-27
• mH ~ mproton = 1.67.10-24 grams
• me = 0.91.10-27 grams• eV = 1.602.10-12 ergs• Luminosity of Sun = 4.1033 ergs/sec• Mass of the Sun = 2.1033 grams
Flux measurements Signal-to-Noise Ratio
"Sensitivity"---i.e. the faintest source measurable---is not simply a matter of the size of the photon collector.
It is instead a signal-to-noise (SNR) issue: SNR = measured value / uncertainty and is
dependant on many things, including: Structure of source (point vs. extended) Nature of luminous background & surroundings Foreground absorption Telescope & instrument throughput Characteristics of detectors (quantum efficiency,
noise)
SNRs in Astronomy
Fundamental limit set by photon statistics: SNR < √ N, where N = no. of detected source photons
Typical SNR's in Astronomy: Measures of astronomical EM fluxes:
Best precision: SNR ~ 1000 (0.1% error) Low by lab standards! Problems: difficulty of
calibration; faintness of interesting sources. Typical "good" measures: SNR ~ 20-30 Threshold detections: SNR ~ 5-10
Noise Sources Detector Noise (CCDs)
Read-noise (rms ~3-10e-1/read) Dark noise (3.10-4e-1/s/pixel)
Determined by Temperature of detector
Background Noise (Diffuse) Artificial light pollution Earth's atmosphere Ecliptic scattered sunlight Scattered Galactic light
Background Noise (Discrete) Exclusion zone around bright stars caused by scattered
light within instrument Source "confusion" caused by diffractive blending of
multiple faint sources
Magnitude System
An ancient and arcane, but compact and by now unchangeable, way of expressing brightnesses of astronomical sources. Magnitudes are a logarithmic measure of
spectral flux density (not flux!) Monochromatic Apparent Magnitudes
m = −2.5 log10 f − 21.1
where f is in units of erg.s−1.cm−2.Å−1
A system of “monochromatic magnitudes per unit wavelength”
Magnitude Normalization
Normalization is chosen to coincide with the zero point of the widely-used “visual” or standard “broad-band” V magnitude system: i.e. m(5500Å) = V
Zero Point: fluxes at 5500Å corresponding to m (5500Å) = 0, are (Bessell 1998)
f0 = 3.63.10−9 erg.s−1.cm−2.Å−1; or
fν0 = 3.63.10−20 erg.s−1.cm−2.Hz−1; or = 3630 Janskys
f0/hν = 1005 photons.s−1.cm−2.Å−1 is the
corresponding photon rate per unit wavelength
Surface Brightness
Surface Brightnesses (extended objects): μ = m + 2.5 log10
where m is the integrated magnitude of the source and is the angular area of the source in units of arcsec2.
1 arcsec2 = 2.35.10−11 steradians
μ is the magnitude corresponding to the mean flux in one arcsec2 of the source.
Absolute Magnitudes
M = m − 5 log10(D/10), where D is the distance to the source in Parsecs (pc) 1pc = 3.258 light-years or 3.086.1013
kilometers 1 pico-pc = a good day’s walk
M is the apparent magnitude the source would have if it were placed at a distance of 10 pc.
M is an intrinsic property of a source For the Sun, MV = 4.83
Source characterization Luminosity (L)
Power (energy/sec) radiated by source into 4 sterad Units: ergs.s−1
Flux (f) Power from source crossing normal to unit area at
specified location a distance D from source f = L/4D2 if source isotropic, no absorption Units: ergs.s−1.cm−2
Surface Brightness (I) Power per unit area per solid angle Units: ergs.s−1.cm−2.sterad−1 (f = I.)
I is independent of distance if source remains resolved
Point Source Sensitivity
Faintest UVOIR point sources detected: Naked eye: 5-6 mag Galileo telescope (1610): 8-9 mag Palomar 5m (1948): 21-22 mag (pg)
25-26 mag (CCD) Keck 10m (1992): 27-28 mag HST (2.4m in space, 1990): 29-30 mag
NB: current optical detectors approach 100% QE ie: We can't improve sensitivity via detector
development. Improvements require new instrumentation.
Spatial Resolution Fundamental limit governed by diffraction in
telescope/instruments Min. image dia. (min)= 2.5/D rads(diffr. limit)
where D is the dia. of the telescope At 5500Å … min= 28”/D(cm) Inside Earth's atmosphere, turbulence strongly
affects image diameter. Resulting image blur & motion is called "seeing", and
typically yields: atm~0.7-1.5” i.e. spatial resolution in most instances is governed by
the atmosphere, not the telescope. Good site + Good environmental control Good AO – approaches diffraction limit
Spectral Resolution
Theoretical maximum governed by diffraction limts set by optical components: Practical limit set by photon rates High resolution devices are typically photon-
starved (except for Sun). ID's, surveys, classification at low resolution
10-500Å or 10 <R< 500 Physical analysis at moderate-to-high
resolution 0.1-10Å or 500 <R< 50,000
Highest to date: ~ 0.01Å or R = 500,000
Basic Lens formulae
Basic Mirror formulae
Refraction
2
1 n1
n2
Snell’s Law: n1 sin(1) = n2 sin(2)• n1 = refractive index in region 1• n2 = refractive index in region 2
where: n = c/v = vacuum /medium
Refraction andTotal Internal Reflection
ConstructiveInterference
DestructiveInterference
N=2 interfering beams
N >> 2 interfering Beams(eg: Grating Spectrograph)
Diffraction grating(N ~ 100,000)
Interferenceorder
wavelength
groove spacing incidence angle
diffracted angle
Michelson Interfermeter(N = 2 interference)
Fabry-Perot(N ~500)
Optics and Focus
Optics below represents a doublet lens Parallel rays from the left are made to
converge
Location where the rays cross is the “focal point”
Distance from the fiducial point in the lens is the “focal length” (fl)
fl
Images Object Plane Image Plane
These are “conjugates” of each other
Conjugate distances are: s1 ; s2
Lens formula 1/s1 + 1/s2 = 1/f
Magnification (m) is given by m = s1/s2
s1 s2
Object
Image
Focal length and focal-ratio (f/#)
Effective focal length (EFL = fl) is the distance from the optics to the focal point
f/# is the ratio focal length to the optic diameter (f/# = d/fl) f/1 is fast (v.difficult to control
aberrations) f/30 is slow (simple optics)fl
d
Plate Scale For a given optic with EFL = fl, the image
plane scale is given by: P.S. = 1c/fl (radians/m) = 206265 /fl (arcsec/mm)
For instance, a telescope with an EFL = 10m (eg: 1.2m @ f/8), plate scale is: 206265/104 = 20.6 arcsec/mm
However, a telescope with an EFL = 170m (eg: 10m @ f/17), plate scale is: 206265/1.7.105 = 1.2 arcsec/mm
ie: If you want a wide-field you have to have a small telescope
Entendu
Entendu = A (area * solid angle) Entendu is conserved for any optical
system ie: Conservation of Energy However, entendu can be lost in fibre systems
High entendu is a figure of merit for an optical system Equivalent to more energy (or information)
transport Telescope have generally to trade:
High A with High Spectrographs try to maximize A.R