Download Conference AS11 - Thirty Meter Telescope

A high-resolution optical spectrograph for the Thirty
Meter Telescope: design and performance
Steve Osterman, Cynthia Froning, Matthew Beasley, James Green,
Stephane Beland
Center for Astrophysics and Space Astronomy, University of
Colorado, Campus box 389, Boulder CO 80309
ABSTRACT
We have completed a conceptual design study of the High Resolution Optical
Spectrograph for the Thirty Meter Telescope project. We propose the use of a
fiber fed integral field unit and a dichroic tree to achieve R=100,000 spectroscopy
from 310 to 1100 nm independent of AO performance. The system relies on the
dichroic tree to provide coarse wavelength selection, and 32 first order spectrograph benches. This approach allows for simultaneous optimization of grating
and detector performance for all wavelengths, resulting in high efficiency, near
uniform dispersion, and reduced program risk and cost due to the high degree of
component commonality. We present projected performance and design details.
Keywords: Optical, spectroscopy, high resolution, dichroic, Thirty Meter
Telescope, HROS,CU-HROS
1. INTRODUCTION
This paper presents the results of a feasibility study performed for the Thirty Meter Telescope
project1, and is an extension of Froning, et al. (2006), “Conceptual design for the High Resolution
Optical Spectrograph on the Thirty-Meter Telescope: a new concept for a ground-based highresolution optical spectrograph”.2 Froning, et al. discuss the scientific motivation, top level design
and advantages of the CU-HROS design. This paper, in turn, focuses on the design details as presented in the feasibility study. As laid out in Froning, et al., it is clear that the large physical scales
of extremely large telescopes (ELTs) and the need for ever higher resolution and sensitivity
necessitate a fundamentally new approach to instrument design to break the cost-growth paradigm
inherent to these ambitious projects. In breaking with the simple scalings of existing instruments,
we developed an alternative to the conventional cross-dispersed echelle design and achieve high
resolution, high efficiency broad band optical spectroscopy by using a dichroic tree to sort the
science spectrum into 32 individual narrow band channels which then feed simple first-order
spectrographs. In this paper we discuss the design at the component level, and provide system level
performance estimate. Our organization is as follows: Section 2 presents an overview of the
concept and its principal components. In Section 3, we discuss the design in detail, and in Section 4
we present performance estimates and design constrains. Science drivers, signal to noise estimates,
system level advantages and upgrade paths are addressed in Froning, et al. The instrument
described in this paper is referred to as CU-HROS to distinguish it from the generic HROS
instrument intended for TMT.
2. INSTRUMENT OVERVIEW
We present a design concept for a high resolution optical spectrograph which is suitable for large
aperture, long focal length ELT projects – a design which combines new and emerging technologies to limit the growth in size, cost and risk as telescope apertures continue to grow. This design
features:
___________________________
* [email protected] voice 1.303.492.3656 fax 1.303.492.5941





A fiber-optic integral field unit (FIFU) to decouple throughput and spectral resolution
from telescope seeing or adaptive optics performance.
An array of high performance dichroic mirrors to perform primary wavelength selection
and reduce the spectral grasp needed by the individual first order spectrograph benches to
roughly 3% of the total band pass.
High efficiency coatings, detectors and diffraction gratings, made possible by the narrow
band over which these components must be optimized.
Small, low cost, low risk components with a high degree of replication, reducing risk and
cost.
Full spectral coverage in a single observation with no mechanisms after the foreoptics.
2.1. Instrument requirements
Our view of HROS is that of a large, multi-purpose high-resolution optical spectrograph. In order
to meet these requirements (Table 1), we have several issues to confront. First, the large size of
TMT requires either very large optics or small spatial sampling. Second, the telescope must
provide an R > 50,000 spectrum from 310 nm to 1100 nm (ideally in one exposure to maintain efficiency). Third, the quality of the spectrum has to be high across the band pass to allow science at
arbitrary wavelengths. Ideally, the spectrograph must maintain high efficiency over as a large a
band pass as possible. Finally, the instrument must maintain the effective area advantage provided
by the thirty meter primary mirror.
Instrument design and performance requirements are either dictated by the project or flow down
directly from the science goals as described in Froning, et al. In addition, the nominal
Specification
Requirement
Goal
Telescope F/#
f/15
f/15
Location
Nasmyth platform
Nasmyth platform
Wavelength
Coverage
310 – 1100 nm in a single, or at most
two, observations.
310 – 1100 nm in a single observation
Spatial Sampling
<0.2"/lenslet or image slice at the
telescope focal plane
0.1" /lenslet or image slice at the
telescope focal plane
Spatial Resolution
<0.2" FWHM at detector
<0.2" FWHM at detector
Field of View
10"
70"
Effective Slit
Length
1"x5"
At least 5 discrete 1 square arc second
IFUs
Sensitivity
Must maintain 30m advantage
Efficiency > than 15% over 80% of
band pass
Support Calibrated
Spectroscopy?
Yes
Yes
Spectral Resolution
R=50,000
R=100,000
Spectral Stability
Line center stable to 3 m/s for the
duration of an observation
Line center stable to 3 m/s for the
duration of an observation
Spectral
Repeatability
Wave cal. must provide centroid
knowledge to <3 m/s between
observations
Wave cal. must provide centroid
knowledge to <1 m/s between
observations
Table 1. CU-HROS Performance Requirements
requirement for a 1"x5" slit has been modified in favor of a requirement for areal coverage, and
instead of a requirement for long term stability, we require wavelength knowledge based on cotemporal or near cotemporal wavelength calibration. Finally, the field of view of the foreoptics has
been increased to 70" to ensure guide star acquisition and to accommodate a single laser ground
layer adaptive optics system (SL-GLAO).
2.2. Design Overview
When we conceived of CU-HROS, we started with the conventional echelle-based design, which
has long been the standard for high-resolution optical spectrographs for ground-based telescopes.
The echelle has many advantages, in particular making use of the 2D format of the CCD for the
high spectral band pass at high resolution. However, as the telescope apertures and focal lengths
grow and the instruments are seeing limited or near seeing limited (especially in the optical) the
size of the instrument grows dramatically. For an instrument such as the one specified in the original HROS requirements — a slit-limited spectral resolution of 50,000 matched to a 1" slit — the
beam size at the echelle is 0.9 m, requiring a very large echelle mosaic, with equally large optics at
other points in the system. In addition, since an echelle works over a large band pass, the variation
in grating efficiency dominates the system throughput. Prism cross dispersers becomes impractical
due to the very large size of the diffracted beam, further reducing efficiency. With these constraints, we chose to move in a new direction to address these challenges.
Our solution is to use a multiplexed first order spectrograph, as illustrated in Figure 1. Spectrograph components that have significant efficiency variations with wavelength (detectors and
gratings) may be individually optimized over a narrow wavelength range to maximize performance.
Broadband components can be either reflective (collimator) or very carefully chosen to minimize
loss of efficiency or spectral resolution (atmospheric dispersion compensator, or ADC, and the
fiber-optic IFU, or FIFU). The nominal design operates in a single mode with a minimum number
Figure 1. Schematic diagram of the CU-HROS Concept. Light from TMT enters the
enclosure and is reimaged onto a 0.1" pseudo-slit via a fiber optic IFU. The spectrum is
sorted into 32 spectral bins via the dichroic mirror array, feeding 32 narrow band, first
order spectrographs with near constant resolution and efficiency.
of mechanisms (all in the foreoptics), covering the wavelength range of 310 – 1100 nm at a spectral
resolution or R=100,000 in a single integration.
Light from the telescope is directed to the Nasmyth platform and into the CU-HROS foreoptics. These
include the ADC, derotator, stops, and baffles. A fast steering mirror will be used to increase image
stability, and the light will then pass to the focal plane reimaging optics to reform and possibly magnify
the focal plane onto the fiber-optic integral field unit (FIFU) array. Foreoptics also include a camera
for acquisition, an absorption cell for wavelength calibration, and reserve space for AO pickoff optics.
After being directed into the FIFU, the image is dissected in a coherent manner, and the 1"x5"
(2x10mm) entrance slit is reformatted to multiple 0.066 x 13 mm pseudo-slits. The FIFU includes
additional fibers dedicated to wavelength calibration sources.
Light exiting the FIFU is collimated and then passes through 5 banks of dichroic mirrors, subdividing
the spectrum into 32 channels of roughly equal spectral grasp. Each channel then feeds a single spectrograph bench operating in first order and optimized for its own narrow wavelength band. A possible
physical layout for the dichroic tree and spectrograph feed points is shown in Figure 2. This arrangement uses a single additional fold mirror to place the blue channels directly below the red channels to
increase packing efficiency.
3. DESIGN DETAILS
3.1. Instrument location and enclosure
The high resolution optical spectrograph for TMT will reside on the Nasmyth platform. This simplifies mechanical design since the gravity vector will remain constant. Image roll will be
accommodated by a derotator, as discussed in section 3.2.
The stability requirements imposed in Table 1 demand that the instrument be placed in a thermally
stable (better than ±1°C) and acoustically isolated environment. Our baseline design accomplishes
this by placing the instrument in a vacuum chamber, as indicated in Figure 1. We recognize that
the vacuum chamber represents a significant cost, weight and operational impact. If subsequent
analysis shows that this chamber is not required, then we will modify our design accordingly. Our
design philosophy throughout this project has been to baseline an instrument that we are certain
will achieve the design requirements, and to look for cost savings as the program evolves and
resources are available for more detailed analysis.
The arrangement shown in Figure 2 will compress the instrument footprint to 5x3x3 meters. With
vacuum chamber, outer enclosure and clean tent, the footprint on the Nasmyth platform should not
exceed 9x10x4.5 meters.
Figure 2: Possible layout of the dichroic array. Light enters from the upper left, and
either passes through to the upper red arm, or is reflected back and folded into the lower
blue arm. The feed points for the individual spectrograph benches are indicated by
spheres. Overall size is roughly 5x3x3 meters.
3.2. Foreoptics and focal plane
CU-HROS will use a series of foreoptics to derotate the image, correct for atmospheric dispersion,
allow target guiding and acquisition, change beam speed to optimize for the fiber optic feed, and
allow for an adaptive optics feed to correct the seeing.
Light from the telescope enters the foreoptics (ADC and derotator), and passes any pickoff mirrors
or beam splitters required for guidance or AO wave front sensor (WFS) feed (see Figure 2, Froning,
et al.). The focal plane is reimaged onto the fiber optic integral field unit which reformats the 1"
point source field of regard into a long, narrow pseudo-slit: in terms of the 2mm/arc second plate
scale, the pseudo-slit is approximately 0.1" wide by 10" long.
The fiber-optic integral field unit
CU-HROS relies on reformatting the 1" entrance slit to a pseudo-slit with an effective width of less
than 0.1". The current design implements this image dissection through the use of a fiber-optic integral field unit (FIFU).3,4 The design could be achieved with a conventional image slicer, but at
the cost of reduced performance. The narrow pseudoslit allows us to reduce the width of the
entrance slit to no more than 0.10" equivalent plate scale in order to reduce the size of our optics.
Also, since the image width is now fixed at output width of the FIFU, spectrograph performance is
not limited by telescope seeing or AO performance. If the TMT delivers a 1.0" PSF to the focal
plane rather than a 0.4" PSF, then the light will be spread across more fibers, diminishing signal to
noise per pixel, but the resolution will not be impacted. Similarly, if a tight PSF is not stable on the
focal plane, this will appear as a variable distribution of light in the cross dispersion direction at the
spectrograph focal plane, but will not change the line width.
The FIFU dissects the image plane and remaps a 2x2mm (1"x1") entrance aperture to roughly 13.7
by 0.066 mm. We accomplish this by placing an array of 91 0.066 mm core fibers behind a
microlens array at the focal plane on 0.2mm centers, and remapping that to a vertical array one
fiber wide (Figure 3). We are opting to use a hexagonal entrance aperture geometry for two
reasons: This allows us to maximize the packing efficiency while minimizing geometrical focal
ratio degradation4, and this geometry makes better use of focal plane real estate given the likely
PSF from the telescope.
Figure 3. (a) Illustration of one possible remapping of the fiber optic IFU input to the
pseudo-slit output. Tessellated hexagonal microlenses (used to minimize geometrical
FRD) are used to concentrate light onto the individual fibers, which are then remapped
into a long, narrow pseudo-slit. Various remapping schemes, such as the spiral mapping shown here, and various raster mappings will be considered prior to implementation. (b) Distribution of light at the pseudo-slit for a 0.5 arc second Gaussian PSF.
Spiral image dissection concentrates light towards the center of the pseudo-slit.
While the default design for HROS called for a rectangular entrance aperture for a point source
spectrograph, we will build an array of several one square arc second hexagonal fiber bundles. In
order to take advantage of improved performance when the seeing disk decreases, we are considering using a methodology similar to that described in Kenworthy et al. (2001), where the
microlens to pseudo-slit mapping spirals outwards from the center of the microlens array. 5 This
would concentrate light towards the center of the output array, would reduce the number of
illuminated pixels at the detector, and would allow on-chip binning at the detector to reduce detector read noise. This and other mapping schemes will be evaluated prior to implementation. Note
that the individual fibers are resolved in cross dispersion at the detector.
Rather than a fixed contiguous 1"x5" slit, we would provide an array of at least 5 one square arc
second IFUs in a nominally fixed grid. This would allow for sampling multiple points on an
extended object, and for a well separated sky channel. Alternatively, the central IFU could be
larger, perhaps 2 or 3 square arc seconds, with fewer sky channels in a fixed array around that.
In addition to decoupling CU-HROS performance from telescope seeing, the FIFU allows us to
insert dedicated calibration fibers into the IFU and inject the calibration signal directly into the
spectrograph while science spectra are being recorded. For example if two fibers at the top and
bottom of the pseudo-slit were dedicated to calibration, then the signal from these would be
spatially separated from the science signal, but could be correlated to the science signal with very
high repeatability since they are being emitted from the same physical structure. Finally, the output
geometry of the FIFU can be tailored to the needs of the collimating optics.
It is important to note that since the instrument will be designed with the FIFU in mind. As such
the collimator and camera optics will be optimized to accommodate the fast beam speed and the
additional vertical extent of the pseudo-slit. This will allow us to minimize or eliminate vignetting
and resolution degradation that might otherwise occur if such a pseudo-slit were introduced to an
optical system originally designed to accommodate a compact, slow source.
Wavelength calibration systems
We intend to implement at least two, and preferably three, calibration systems: These include an
iodine absorption cell, a thorium/argon lamp, and a laser frequency comb. As mentioned above,
one of the advantages of the FIFU is that it allows us to inject light from an external source such as
a discharge lamp or a frequency comb directly into the FIFU, making it possible to record calibration spectra during an observation. This provides real-time wavelength knowledge and a direct
measure of any drift or blurring of the spectra since any motion in the science spectra would also be
present in the calibration spectra.
Today, iodine absorption cells are standard wavelength references for high precision velocity
observations (for planet searches), and have provided reliable and cross-system replicatable
absorption spectra, but at the expense of depressing throughput of the target continuum by 50%.
Furthermore, an iodine cell will not provide absorption features across the full instrument band
pass, leaving some channels uncalibrated. However, a deployable iodine absorption cell is included
in the CU-HROS concept to maintain continuity with current observing programs and to act as a
calibration source for certain observations.
CU-HROS will also include either a Th/Ar lamp or a suite of emission line lamps to provide emission line calibration spectra across the instrument band pass. The light from these lamps can be
injected directly into dedicated fibers in the FIFU to provide simultaneous science and wavelength
calibration, or fed into the science beam via a beam splitter to provide knowledge of line center
across the pseudo-slit. The first option would allow continuous injection of the wavecal spectrum
without contaminating the science spectrum, while the second would provide direct measure of any
wavelength offset as a function of position across the IFU output (this could also be obtained from
the absorption cell). Th/Ar and He/Ne/Th lamps are dim but well characterized, and provide
Figure 4: Consecutive pulses emitted by a mode-locked laser generate a uniformly
spaced series of emission lines (“teeth”). Figure based on Udem, et al (2002).6
rich spectra in the region of interest. For example, Th/Ar lamps provide over 4000 stable, identified lines between 310 and 1100 nm, and have provided excellent reference spectra for many
instruments.
In addition to emission line spectra, recent work in at the National Institute for Standard Technology (NIST) in Boulder, Colorado, has produced laser frequency combs with the potential for use as
extremely stable, well characterized line sources.6,7 A frequency comb produces emission lines
(‘teeth’) at a regular frequency spacing: ω n = nωγ + ωo, where ωγ is the pulse frequency of the
stimulation laser, and the offset frequency ωo can be phase locked to a highly stable atomic source
(Figure 4).
Current work has generated line spacing as wide as 2GHz (approximately 250,000 λ/Δλ). This line
spacing can be increased by using an etalon to filter the unwanted lines, reducing the number of
lines by an order of magnitude.8 The advantage of this light source is that it produces regularly
spaced, extremely stable emission lines across the instrument band pass: the driving frequency and
offset frequency can be phase locked to an off the shelf atomic frequency source, yielding wavelength accuracies as great as 1 part in 1010 when coupled to a rubidium microwave transition
source. Even though the output intensity varies over several orders of magnitude, the output is very
intense relative to our requirements, so selective filtering could be used to normalize the beam.
3.3. Collimator and Dichroic Tree
Collimating optics
We will use an all-reflective collimator based on a three-mirror anastigmat to provide an achromatic collimated beam without residual astigmatism or coma (Figure 5).9 A three-mirror anastigmat provides control over coma, spherical aberration, and astigmatism with an arbitrary field
curvature. In addition, to preserve the post-grating camera performance, the collimator cannot leave
FIFU output and
collimator optics
Camera optics
and spectrograph
focal plane
Figure 5. The Collimator and all-up system. View from top demonstrating the location of the
exit pupil of the collimator, chosen to minimize the size of the dichroics. The large size of the
camera is an artifact of the rendering which includes the full size. See Figure 7 for side view.
aberration in the beam. The design uses off-axis elements and is designed to accommodate the
extended field and fast beam introduced by the fiber bundle in order to eliminate vignetting in the
collimator or dichroic array. With the freedom to place the fiber ends at arbitrary positions, a
curved input plane is easily created, reducing the constraints of the collimator.
We expect the optics to be ~1 meter class. With advances in mirror polishing technology this does
not represent a significant increase in cost, and reflective optics remain a cost savings relative to
refractive optics. In addition, the use of reflective surfaces instead of refractive optics will
minimize chromatic aberrations throughout the system.
Dichroic Tree
Dichroic mirrors provide the primary wavelength selection for the CU-HROS instrument, and the
performance of the instrument is dependent on these mirrors operating at high efficiency. Dichroic
mirror technology has undergone substantial advances in the last several years, yielding mirrors
with sharp transitions (>5nm from 90% rejection to 90% acceptance) and high efficiency (typically
>95% transmission/reflectance). Measured performance of dichroic mirrors already produced by
Barr Associates support this contention (Figure 6-a) and exceed our specification of transmission/reflection 95%, out of band transmission/reflection 5%, and 90% to 10% transition edge in
less than 5nm.
The ability of Barr Associates to deliver optics with the performance shown in Figure 6-a reinforces
our confidence in their ability to meet our performance requirements. Note that the fluctuations in
transmission/reflection shown in the curve also appear in the theoretical data for the individual
HROS dichroic mirrors; however, the locations of the dips in performance are uncorrelated, and
tend to average out after 5 filters (Figure 6-b). The dichroic mirrors will have a 200 mm clear
aperture, although this may grow if doing so relaxes design constraints on the collimator and
camera optics, and be optimized 15° angle of incidence in order to minimize polarization effects.
3.4. Spectrograph Benches
Diffraction Gratings
Holographic gratings are the preferred option for CU-HROS. Holographic gratings are produced
by the interference pattern of two stigmatic (laser) source point sources on a photo-sensitive
Figure 6. (a) Measured performance for an existing 125x125 mm dichroic mirror; data provided by
Barr Associates. While this optic is designed for longer wavelengths than we will be operating at, the
manufacturer considers optics designed for the visible to be less challenging to fabricate. Note the
high performance and low ripple show in this figure. Similar performance had been achieved with the
filters for the HST Wide Field 3 instrument. (b) Performance of dichroic tree for each channel based on
theoretical data provided by Barr Associates with modeled performance reduced to match specification.10 The dip in performance at short wavelengths is due to structure in the first dichroic – since this
optic must provide high efficiency across the greatest band width, it is the most difficult to optimize.
Overall performance for CU-HROS is discussed in section 4.2.
material. The photo- sensitive material is developed and a series of smooth grooves are produced on the
surface of the substrate. Due to the smooth nature of the grooves, holographic gratings have very low
scatter (measured at better than 10-5 – 10-6/Å at 122nm for the Cosmic Origins Spectrograph gratings).11
Our preferred option is to request a pseudo-sinusoidal groove profile by adjusting the exposure time to
produce deeper grooves. This produces high efficiency devices (~65%), although over a relatively narrow
bandpass (~5%). With the design for CU-HROS, each channel represents a 3% bandpass and very high
groove efficiency can be achieved at all wavelengths.
Camera Optics
CU-HROS requires fast (f/1.8) cameras with wide fields (14 degrees) and excellent aberration control (spot
FWHM = 15 μm). Preliminary investigations imply that several extant solutions meet our needs. In
discussions with Zygo Corp and Goodrich Space Optical Branch, both vendors felt that appropriate, costeffective solutions exist and could be produced. A significant advantage to the CU-HROS spectrograph
cameras is the narrow wavelength range required for each camera, simplifying design by removing the
need for achromatism. Our camera design is aided by the significant anamorphism in the beam from the
grating to the camera. By compressing the beam in the spectral direction, the final beam is closer to f/5,
increasing performance.
The trades that must be examined closely in the future are the relative advantages of a single all-reflective
system (simplifying overall system engineering) versus the slightly more complicated refractive. Refractive systems do not have the broadband performance to allow one design to be used in all channels;
however, perhaps as few as three different configurations will be needed to accommodate the broadband
performance of CU-HROS. Figure 7 is an example of an all-reflective camera suitable for our design. The
camera is based on a three-mirror anastigmat system that produces <15 μm images without considering
tolerancing or manufacturing error. The camera itself uses three aspheric elements, which will add to the
cost of manufacture, although recent advances in figuring technology will mitigate this cost. We are
considering options to permit the diamond turning of the mirrors, which presents a significant cost savings.
The optics are in the 200 mm size range, increasing availability and reducing cost and risk to the program.
We also take advantage of the anamorphism introduced by the grating to help control aberrations in the
spectral direction to smaller than the spatial on-chip binning size (5 pixels). The current design has less
than 4 pixels of distortion from the middle of the detector to the edge, so binning is limited by cosmic ray
rates rather than our ability to bin in a straight line.
Light from
dichroics
Diffraction
Grating
Spectrograph
focal plane
Figure 7. Example camera design employing a three mirror anastigmat that
performs to specification. Note that while the camera optics appear quite large,
we will be using small off axis portions of the two largest optics.
Detectors
The detectors for CU-HROS represent conventional technology in a large format application. Commercial
CCD vendors (e2v, Fairchild) have expressed interest in working with us to produce a series of custom
CCDs for our application. Our current design would use 12.5 micron pixel, 9K by 3K CCDs, which will fit
on a conventional six-inch silicon wafer. The well depth of such CCDs would be on the order of 100,000
electrons for the 12.5 microns pixel pitch.
While there is a trade between larger pixels and reduced read noise, the resulting cosmic ray rate must also
be considered. At the faint limit for a seeing limited telescope read noise dominates all other noise sources.
This is a function of the aperture and will affect all instruments equally: assuming a best-case F/1.0 final
beam speed, the spatial scale of a 30-meter telescope is roughly 150 microns per arc second, which for 15
m pixels translates to >10 pixels per resolution element (total area, including cross dispersion), independent of instrument design.
4. PERFORMANCE ESTIMATES
4.1. Spectral resolution
Each spectrograph is designed to provide similar resolution, with the band pass for each of the 32 spectrograph benches increasing with wavelength. Since the spectral coverage increases linearly with wavelength,
resolution is virtually unchanged from one channel to the next. Each spectrograph channel will have the
same band pass in terms of individual resolution elements so that the resolution will be close to constant
across the entire wavelength range (ranging from R = 100,000 to 104,000). One resolution element
encompasses 2x2 pixels (30x30μm) on the detectors, and the design 50% encircled energy width in the
dispersion direction is >20μm for the full optical train (Figure 8).
4.2. Efficiency
HROS efficiency is expected to be uniformly high with limited variation between the red and blue ends of
the spectrum. Component level efficiency estimates are summarized in Table 2, and system level performance is illustrated in Figure 9.
Figure 8. Ray trace from FIFU output to spectrograph focal plane. FFT diffraction
enclosed energy in the dispersion direction (half width) is shown for three fiber locations
near the central wavelength for channel 16 (547.5-571.4nm). At the red and blue ends of
the channel the enclosed half width increases to 10 μm. Inset shows spot diagrams for
two lines separated by λ/Δλ=100,000 (585.000nm (top) and 585.006nm (bottom)).
Component Level Efficiency Breakdown
Refl/Trans per
Qty
surface
Item Description
Derotator
ADC
Reimaging optics
Chamber window
FIFU
Collimator
Vignetting
Dichroic tree
Grating
Camera
CCD
.
3 mirror
4 surfaces
4 mirror
wide band AR
Durham best effort
3 mirror
Due to finite source size
(HROS 100)
Optimized for narrow band
3 surfaces, tuned AR/Refl
Optimized for narrow band
0.98
0.97
0.98
0.97
0.65
0.98
0.95
0.95
0.65
0.98
0.90
Net
Efficiency
3
4
4
2
1
3
1
5
1
3
1
Net HROS Performance:
0.941
0.885
0.922
0.941
0.650
0.941
0.950
0.774
0.650
0.941
0.900
0.18
Table 2. Projected CU-HROS efficiency breakdown. High reflectivities, transmissions and
grating and detector efficiencies are achievable due to narrow band over which many of the
components must be optimized. Wherever possible, vendor estimates have been used.
Since the spectrographs can be optimized for a narrow band, channel to channel variation will be dominated by dichroic mirror performance (Figure 6-b), and that is principally controlled by the first dichroic in
the array. High efficiencies have been assigned to the detectors, diffraction gratings and camera optics
since these can be optimized for the narrow bands of the individual spectrograph benches – ranging from
13nm at the blue end of the instrument band pass to 46 nm at the red end. Somewhat more conservative
estimates are made for optics which must perform across the entire 800nm instrument band pass. The
dichroic mirror efficiencies do not reflect the vendor estimates, but rather the more conservative specification, so that higher overall performance may be realized.
The overall 18% efficiency is modulated by a sharp drop to 10-14% efficiency at the channel boundaries
due to the finite width of the dichroic mirror transitions (Figure 6-b). This effect is mitigated by allowing
overlap in the spectral coverage of adjacent channels and then coadding the dispersed data.
Figure 9. CU-HROS end-to-end efficiency for the entire instrument band pass
including coaddition of overlapping spectral regions.
5. CONCLUSION
CU-HROS represents a novel alternative design for high resolution spectroscopy on the next generation of
extremely large telescopes. The design presented in this paper decouples spectrograph resolution from
telescope seeing and AO performance, provides near uniform resolution and efficiency across the instrument band pass. The design requires modestly sized optics and leverages a high degree of replication to
reduce cost, complexity and risk, breaking the cost-growth paradigm for the construction and instrumentation of extremely large observatories.
The authors gratefully acknowledge the support of the TMT partner institutions. They are the Association
of Canadian Universities for Research in Astronomy (ACURA), the Association of Universities for
Research in Astronomy (AURA), the California Institute of Technology and the University of California.
This work was supported, as well, by the Canada Foundation for Innovation, the Gordon and Betty Moore
Foundation, the National Optical Astronomy Observatory, which is operated by AURA under cooperative
agreement with the National Science Foundation, the Ontario Ministry of Research and Innovation, and the
National Research Council of Canada.
_________________________________________
REFERENCES
1.
The Thirty Meter Telescope project (http://www.tmt.org) is a public-private partnership to construct a
giant segmented mirror telescope in response to National Academy of Sciences report “Astronomy and
Astrophysics in the New Millennium”.
2. C. Froning, S. Osterman, M. Beasley, J. Green and Stephane Beland, “Conceptual design for the High
Resolution Optical Spectrograph on the Thirty-Meter Telescope: A new concept for a ground-based
high-resolution optical spectrograph”, Proc SPIE, paper 6269-69 (in press).
3. J. Allington-Smith, G. Murray, R. Content, G. Dodsworth, R. Davies, B. Miller, I. Jorensen, I. Hook,
D. Crampton and R. Murowindsk, “Integral Field Spectroscopy with the Gemini Multiobject
Spectrograph. I. Design, Construction, and Testing”, Pub. Astro Soc. Pac. 114, 892-912 (2002).
4. M.A. Kenworthy, The Development of New Techniques for Integral Field Spectroscopy in Astronomy,
Thesis prepared for the Institute of Astronomy and Clare College, Cambridge (1998).
5. M.A. Kenworthy, I. Parry and K. Taylor , “Spiral Phase A: A Prototype Integral Field Spectrograph for
the Anglo-Australian Telescope”, Pub. Astro Soc. Pac. 113, 215-226 (2001).
6. S.A. Diddams, D.J. Jones, J.Ye, S.T. Cindif, J.L. Hall, J.K. Ranka, R.S. Windeler, R. Holwarth, T.
Udem and T.W. Hänsch, “Direct Link between Microwave and Optical Frequencies with a 300THz
femptosecond Laser Comb”, Phys Rev Letters, 84 (22), 5102-5105 (2000).
7. Thomas Udem, Ronald Holwarth, and T.W. Hänsch, “Optical Frequency Metrology”, Nature, 416,
233-237 (2002)
8. S. A Diddams, Personal communication, , 29 December, 2005
9. D. Korsch, Reflective Optics, Academic Press, Inc. San Diego, CA (1981)
10. Joanne Choi, Barr Associates, Personal communication, 11 January, 2006.
11. S. Osterman, E. Wilkinson, J. Green and K. Redman, “FUV grating performance for the Cosmic
Origins Spectrograph”, Proc SPIE, 4013, 360-366 (2000).