Post on 24-Dec-2015
Near Infrared Spectrometers and
Integral field spectroscopy
Andy Longmore (UK ATC)
(contact via supa@roe.ac.uk)
• Recap of principles of IR instrument design• Some spectroscopy related specifics• Introducing the dispersion element• Spectrometer design - slit width, resolution and instrument size• Novel dispersers• Integral field and multi-object spectroscopy
Outline of lecture
Principles of IR instrument design• As for imaging, control of the infrared background is key
• cryogenic temperatures for NIR (130K-77K) •use of cold pupil stop• conservation of etendue/throughput from telescope to detector
• Challenge remains to detect a faint source against a bright background
• 5sigma in 1 hour for magnitude 17 – 18.5• broadband backgrounds 13-16.5
• Nature of the background very different…..
NIR spectroscopy basics
The OH emission spectrum:
OH spectrum from Rousselot et al. (2000, A&A, 354,1134).
• Lines are bright– 2500 photons per
second per arcsec with an 8-m telescope, 1000 times target source flux
• Variable• BUT, the continuum
between them is very dark– 0.1photons per second
Resolving out the OH background
From Martini & DePoy 2000, Proc SPIE 4008, 695
Spectral resolving powers of >2000, preferably around4000 are a good choice.
Slit alignedon object
Collimator DispersingElement
Camera
X
Y
Y
Detectorarray
entrance aperture/slit + collimator + dispersing component (if possible at a pupil)
Basic Spectrometer layout
Introducing the dispersion element
Grating equation Angular dispersion
Spectral resolving power, R
Slit alignedon object
Collimator DispersingElement
Camera
X
Y
Y
Detectorarray
Angular separation of two wavelengths at the detector
From conservation of étendue:
Wavelengths are resolved when these are equal, so:
NB: max R depends on Pupil size, telescope D and angular size of the object.
Spectral resolving power, R =
Camera focal length, f
Pixel size, w
fw /)tan(
Relating R to instrument design parameters:
)tan(2
rw
fR
NB: often w = two pixels for Nyquist sampling of the spectral line width
KMOS - A practical example
The K-band multi-field spectrograph for the European Southern Observatory Very Large Telescopes
KMOS - A practical example
• Conservation of A and the detector pixel size determines the camera f-number :
• FOV = field of view per pixel in arcsec
f_number
hpixel_widt
206265
FOVD
detectortel
KMOS - a practical example
KMOS - Design choices: pixel FoV
• pixel FoV depends on– Size of images delivered to the
telescope
– the science case
• For maximum sensitivity – Match slit to seeing
• KMOS has designed
for the best seeing at VLT• Pixel scale selected is
0.2arcsec for Nyquist sampling of the seeing
KMOS - Camera f-ratio
• For VLT, D = 8m, and for KMOS pixel width = 18m which we want to match to FOV=0.2arcsec.
• The camera required is f/2.32• NB: a fast camera and difficult to design• Combination of large telescopes and sharp
image through AO provides design changes in future e.g. 40m telescope would require f/0.5 for 0.2arcsec pixels!
• Key requirements for KMOS on spectral resolving power:– high enough to resolve OH lines – Ideally, coverage of a complete transmission
window simultaneously– Allowing velocity resolution of 10km/s for
experiments on the Tully-Fischer relation in galaxies
KMOS - Design choices: gratings
KMOS - Resolving the OH background
From Martini & DePoy 2000, Proc SPIE 4008, 695
• Key requirements for KMOS on spectral resolving power:– high enough to resolve OH lines – Ideally, coverage of a complete transmission
window simultaneously– Allowing velocity resolution of 10km/s for
experiments on the Tully-Fischer relation in galaxies
KMOS - Design choices: gratings
• Atmospheric windows
KMOS - Design choices: gratings
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
wavelength (um)
tran
smis
sio
n
• Key requirements for KMOS on spectral resolving power:– high enough to resolve OH lines – Ideally, coverage of a complete transmission
window simultaneously– Allowing velocity resolution of 10km/s for
experiments on the Tully-Fischer relation in galaxies
KMOS - Design choices: gratings
• Key requirements for KMOS on spectral resolving power:
– high enough to resolve OH lines
– Allowing velocity resolution of 10km/s for experiments on the Tully-Fischer relation in galaxies
KMOS - galaxy dynamics
• Key requirements for KMOS on spectral resolving power:– high enough to resolve OH lines – Ideally, coverage of a complete transmission
window simultaneously– Allowing velocity resolution of 10km/s for
experiments on the Tully-Fischer relation in galaxies
– Resulting choice: / > 3000, maximum 4000 at J band.
KMOS - Design choices: gratings
• For a grating spectrometer :
– f is the focal length of the camera, – w is the size of the slit at the detector (two 18m
pixels for KMOS) – b the blaze angle for the grating (45º).– f ~ 70mm and beam size ~31mm for R=4000– In practice, KMOS beam size is 33mm
KMOS - gratings again
w
rfR
)tan(2
• Advantages– instrument layout– compact– simplified
Novel Dispersers 1: Grisms
Cam
era Module
Slit Module
Grism
Module
TSIU
Novel Dispersers 1: Grisms
Refraction in the prism:
Diffraction from the grating:
Resulting ‘grism’ equation:
NB: dependence on the refractive index of the glass
Novel Dispersers 1: Grisms
• Advantages– Compact– simplified instrument layout– Potential for higher R – for a given beam size
• Disadvantages: – difficult materials– Relatively low transmission
ND2: Volume Phase Holographic gratings
IR Laboratory prototypes
Lab testing and instrumentdevelopment underway for these gratings.
Real spectrometer throughput
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
wavelength (um)
tran
smis
sio
n
kmos
atmosphere
optimal
essential
Data courtesy UK Infrared Telescope, Tom Ray and Chris Davis
• With typical plate scales of 0.2arsec/pixel and 2048 x 2048 pixel detectors, modern spectrometers have slit lengths of 120arcmin.
The long slit: strengths/weaknesses
An image of a field of stars and galaxies is used to set positions for slitlets in a mask…
the multi-object slit mask is inserted at the telescope focal plane…..
and spectra of the selected sources measured at the detector.
Images from the Gemini Multi-object Spectrograph
The long slit versus MOS spectroscopy
•The distribution of 200,000 galaxies in the 2dF galaxy redshift survey•2dF is an optical multi-object spectrograph with a two-degree field of view and the ability to observe 400 objects simultaneously
A fibre-fed MOS spectrograph: 2DF
Techniques for 3D spectroscopy
Image: Stephen Todd, ROE and Douglas Pierce-Price, JAC
Integral Field Spectroscopy
stray light control
scattering from diamond-machined surfaces
scattering and diffractionfrom edges of the slicing mirror
triple-fold mirror
field andpupil baffle
input baffle
pupil baffle
Dekker mask
mask over unusedmirror surface
IFU
folded optical layout
compact - to fit on slit wheel
input focal plane and outputslit plane are equivalent
f / 10 input and output
deployable
IFU
externalappearance
reality
partiallydismantled
triple-foldmirror
f/conversionmirrorassembly
slicing mirrorassembly
main body
• 0.95mm thick Aluminium slices
• Individually manufactured• Diamond machined• Common radius of
curvature• Surface roughness
~10nm rms• ~700 spatial elements
Metal slicing mirrors
Integral field spectroscopy: data
IFU data
Dispersion
Spa
tial p
ositi
on
yx
Emission line images in 0.4 arcsec seeing
N
E
350 pc
unresolvedSpatially resolved
Wilman, Edge and Johnstone (2005)
IFU data: NGC1275
H2 located in a disk. Resolved for the first time.
ΔV=345 km/s with r=50 pc → MBH = 3.4 x 108 M
A BLACK HOLE MASS ESTIMATE
Techniques for 3D spectroscopy
IFUs 1: SAURON pupil slicing
• SAURON optical IFU on the William Herschel Telescope
• 33’’ x 41’’ total field of view with 0.94” per pixel• Fixed spectral format 4810-5400A (science case…..)
The pupillenslet array
• CIRPASS – Cambridge University
IFUs 2: Fibre bundles for image slicing
Now and in the future: Multi-IFUs
In the future: Multi-IFU systems
MOMSI for a 42-m E-ELT
Precision Radial Velocity Spectrometer: Optical Layout
• Input slit– Output from fibre slicer– 0.9 x 0.08mm effective size, f/5
• Focal reducer– 2x spherical lenses– Convert from f/5 to f/13
• Single pupil mirror– Parabola, f=2000mm, 500x446mm– Used in triple pass
• 140mm collimated beam diameter
• Spectrum mirror– Flat, 386 x 30mm
– Echelle• 31.6 lines/mm, R4 (75° blaze angle)• Small gamma angle (0.4°)• 610 x 200mm
– Cross disperser• Reflective grating, 100 lines/mm, m=1• 210mm diameter
– Camera• Refractive, no aspherics, 6 elements• f=437mm, f/2.78• 0.148 arcsec/pixel
– Detector• 2 x 2K2 HAWAII-2RG arrays
Spectral Format
Detector array footprint2 x 2K2 HAWAII-2RG arrays73.728 x 36.864mm
-30
-20
-10
0
10
20
30
-60 -40 -20 0 20 40 60
Detector X Position (mm)
Det
ecto
r Y
Po
siti
on
(m
m)
Order 35
Order 611.0m
1.75m
Realistic PRVS Simulations
M6VTeff = 2800 KLog g = 5v sin i = 0 km/s
ModelTelluricOH
Wavelength calibration
(Emission line density used in baseline simulations)
Combined Ar, Kr, Ne and Xe lamps
With Dyanamic range 100-300
Functions of Calibration Unit
• Track displacements at the sub pixel level and changes in Spectral Response Function (SRF)
• Allow accurate wavelength calibration for each order
• Provide Flat Field signal to the spectrograph using a Quartz-Halogen lamp
• Provide a gas cell reference to provide for tracking emission line lamps– Final calibration accuracy must be maintained over
timescales from hours to 5 years
Need large # of lines throughout spectrum
Combine Multiple
emission line lamps in fibers
And: gas absorption cell for additional long-term referenceOR: Laser frequency comb (new technology)
M6V: v vs v sini
John Rayner (University of Hawaii) + PRVS team
Standard Spectroscopy Data Reduction
• Preparation of Single Frames – Manipulation of Raw Data – Preliminaries – Bad pixels
• Creating a bad-pixel mask – Readnoise Variance
• CGS4 Readnoise • Michelle Readnoise • UIST Readnoise • IRIS2 Readnoise • ISAAC Readnoise
– Bias Subtraction – Poisson Variance – Chopping – Flat Fielding – Interleave and Coadd – Orient Image Normally – Wavelength Calibrate
• Group Formation – Sky Subtraction – Group Coaddition
• Spectrum Extraction – Counting Beams – Finding Beams – Beam Extraction – Derippling – Beam Cross Correlation – Extracted Beam Coaddition
• Signal-To-Noise Calculation • Division By Standard Star • Flux Calibration • Science Target Recipe Details • POINT_SOURCE • EXTENDED_SOURCE
– EXTENDED_SOURCE_WITH_SEPARATE_SKY
• FAINT_POINT_SOURCE • POINT_SOURCE_POL • Calibration Recipe Details
http://www.oracdr.org/docs/sun236.htx/sun236.html
3D spectroscopy data processing
3D spectroscopy DR – more examples
• Reduction steps– Wavelength calibrations– Spectral flat-fielding– Fringing– Spatial flat-fielding– Spatial calibration– Sky subtraction– Telluric absorption– Flux calibrations– Cosmic rays
Wavelength Calibration
• What wavelengths fall on which pixels?• How can we re-sample the data to have
linear wavelength axis?– Find dispersion function: relationship between
your pixels and absolute wavelength
• Re-sampling– A necessary evil, at least in spectral direction– Propagates ‘artefacts’ from a sinlge spectrum
to all final spectrum, but diluted so harder to identify.
Spatial Flat-Fielding
• Like an imager, IFUs need 2D spatial flat field• Optical train is much more complex than an
imager.• Vignetting typically due to optics, pupil masks,
etc.• Typically spatial resolution is too low to be very
sensitive to (e.g.) dust particles• Still need global illumination function• Best to use integrated light from the sky rather
than internal lamp
Spatial Flat-Fielding
Lamp
OASIS illumination
Sky
Spatial Calibration
• IFUs have complex optics that may introduce non-negligibledistortions in the spatial plan.
• Important for astrometric applications - in practice, compromisebetween FoV and spatial sampling reduces the significance
• Lenslet-based instruments have spatial stage fixed by lens array -usually well known
• Slicer-based instruments more prone to flexure, esp. along the slit -needs calibrated
• Very important for large-format slicers, like MUSE (ESO)
Reference image(NIRI, Gemini North) NIFS Reconstructed
Sky subtraction
Atmospheric dispersion => ADC(not so important for infrared unless working at diffraction limit)
Atmospheric refraction => image shifts as function of wavelength• Shifts largest at blue wavelengths• Can be corrected during reduction by shifting back each image plane
Co-Adding Data Cubes
• Two approaches:• 1. Dithering by non-integer number of spaxels:
– Allows over-sampling, via ‘drizzling’– Resampling introduces correlated noise– Good for fairly bright sources
• 2. Dither by integer number of spaxels– Allows direct ‘shift and add’ approach– No resampling:- better error characterisation– Assumes accurate (sub-pixel) offsetting– Suitable for ‘deep-field’ applications
3D spectroscopy workshops
• 3D Spectroscopy in Astronomy• Series: Canary Islands Winter School of
Astrophysics• Edited by Evencio Mediavilla • Science perspectives for 3D spectroscopy:
proceedings of the ESO ... Markus Kissler-Patig, Jeremy R. Walsh, … 2007
EAGLE at the
E-ELT Gravity-Invariant Focal Station
Assembly Sequence
The future is HUGE…
Instrument De-Rotator and Services Wrap
Instrument Core
space envelope
Electronics Racks
these can be placedat any point in theassembly sequence
Focal Plate
focal plate is 1226 mmbelow the top of theinstrument envelope
on-axis focus is 1200 mmbelow the top of theinstrument envelope
Receiving Mirrors
mounting plate notyet detailed
TRAMSs and ISSs
ISSs can be attached anddetached individually,following warm-up andback-filling of the cryostat
TRAMS will be mountedwithin an enclosure heldat constant temperature
Robotic Positioner Mounting Ring
Robotic Positioners
Entrance Window
for 5 arcmin science field
LGS Extraction
six LGSs in a ring at7.5 arcmin diameter
Overall Envelope
5m diameter5m high
SideView