Open Access
Add to BibSonomyAbstract
Under seeing limited conditions very high resolution spectroscopy becomes very difficult for extremely large telescopes (ELTs). Using adaptive optics (AO) the stellar image size decreases proportional with the telescope diameter. This makes the spectrograph optics and hence its resolution independent of the telescope diameter. However AO for use with ELTs at visible wavelengths require deformable mirrors with many elements. Those are not likely to be available for quite some time. We propose to use the pupil slicing technique to create a number of sub-pupils each of which having its own deformable mirror. The images from all sub-pupils are combined incoherently with a diameter corresponding to the diffraction limit of the sub-pupil. The technique is referred to as “Pupil Slicing Adaptive Optics” or PSAO.
©2007 Optical Society of America
1. Introduction
Coupling extremely large telescopes to high resolution spectrographs (SGs) is difficult. Assuming perfect gratings and optics their spectral resolution R (=λ)/Δλ) equals:
where tanβ is the tangent of the grating tilt angle β (tanβ = 4 for a state of the art R4 grating), Dgr is the diameter of the collimated light beam incident on the grating (which has a grove length close to Dgr), FRcoll the focal ratio of the collimator of the light coming from the entrance slit and SW is the width of the entrance slit. Sometimes the slit width is also given as θ, the angle it subtends on the sky in arcsec (or φ when expressed in radians). Generally the focal ratio FRtel of the telescope beam incident on the spectrograph slit equals that of the collimator FRcoll. In that case Eq. (1) converts into:
where D is the telescope diameter. Sometimes the Resolution-Slit-Width Product Rθ is used as a characteristic of the spectrograph [1].
Typical ELTs and hence some of their grating spectrographs have collimator focal ratios of about f/15 (FRcoll = 15) which sets together with Dgr the collimator focal length which in turn sets the spectrograph size. UVES (f/10) and HIRES (f/16) are examples of successful high resolution spectrographs on the largest telescopes in existence now, the 8.2-meter VLT and 10-meter Keck telescopes respectively [1]. Both achieve an R of 40000 for 1 arcsec slitwidth/seeing (or Rθ = 40000). R is proportionally better for better seeing. Both spectrographs are large instruments. To achieve spectral high resolution the UVES uses a so called image slicer allowing it to decrease SW and a high angle incident grating (tanβ=4 with Dgr=200 mm). HIRES used a three grating mosaic thus increasing Dgr to 305 mm (but consequently the SG size) and hence R. Table 1 summarizes the properties of these and other high-R spectrographs.
Table 1. Modern High-Resolution Spectrographsa
Here we will assume D to be 50 meters as is the case for the Euro50 ELT design [2, 3]. The results are however applicable to all ELTs, including the Thirty Meter Telescope (TMT), the 23-meter Giant Magellan Telescope (GMT) and the 42-meter European ELT (E-ELT). The difficulty of achieving high spectral resolution with ELTs under seeing limited conditions is best seen with Eq. (2): increasing D decreases R proportionally. The other instrument parameters in Eq. (2) are difficult to improve to compensate for this R decrease, and most have already been pushed to the present limits to achieve the desired resolution at the smaller 8 – 10 meter telescopes. Large grating mosaics might be envisioned, however their larger Dgr would increase the spectrograph sizes proportional to D if R is to be maintained, leading to spectrographs with very large dimensions (see table 1, last column).
In this paper we propose to combine two existing and tested concepts for achieving high R values (105 to 106) at ELTs with spectrographs small enough to be mountable in the Cassegrain/Gregorian focus. These concepts are: (i) pupil slicing [4, 5]. It effectively decreases D in Eq. (2), and (ii) decreasing the image size φ by adaptive optics [6, 7]. The two are synergistic in the sense that pupil slicing enables the use of adaptive optics at optical wavelengths for ELTs for high resolution spectroscopy purposes. We will refer to this combination of pupil slicing and adaptive optics as Pupil-Slicing Adaptive Optics or PSAO.
2. Science requirements for very high resolution spectrographs
A number of important scientific objectives for ELTs require very high spectral resolution. Table 2 summarizes many of them. They of course refer to relatively low temperature objects (like the interstellar medium and planetary atmospheres), to objects where high precision radial velocity observations require that spectral lines are well resolved (like planet searches using the RV method) and to objects where precision spectroscopy and polarimetry of Doppler and Zeeman effects are important (like stellar surface structure). ELTs are needed to achieve the required signal-to-noise ratio for the faint objects and for their high angular resolution (like resolving stars in clusters and galaxies and increasing their contrast with respect to a background).
The R values given in Table 2 are somewhat larger than those often listed for these science targets. The PSAO spectrograph proposed in this paper promises much higher spectral resolutions in a much smaller volume so that the past R restrictions are of no consequence. The values in Table 2 are those we consider needed to well resolve the spectral lines of concern rather than to just be able to detect them. The resulting increased strength of the lines (absorption or emission) tends to offset some or all of the signal-to-noise losses resulting from the smaller wavelength intervals seen by the detector pixels.
Table 2. Some science targets requiring ELTs with high resolution spectrographs
In addition to very high spectral resolution, there are other requirements for ELT spectrographs. These include: (i) broad wavelength coverage; this often requires a cross-dispersed Echelle configuration and large CCD detectors, (ii) mechanical and thermal stability to avoid spectrum drift resulting in R degradation during long exposures, (iii) a fiber coupling of the telescope image and of the calibration source to the spectrograph to assure stable wavelength referencing, (iv) a high quality atmospheric dispersion compensator (ADC) to avoid spreading out of the image in the focal plane across the fiber or spectrograph slit, (v) high throughput to optimize the limiting magnitude and signal-to-noise ratio, and (vi) the ability to do Stokes polarimetry for the study of magnetic fields and scattering effects. Some of these requirements (stability and wavelength referencing) are best met by placing the spectrograph on a gravitationally constant environment (Nasmyth or coudé foci), others (polarimetry, throughput) prefer location at a Cassegrain/Gregorian focus.
The Euro50 PSAO spectrograph we propose here will be much smaller than the dimensions indicated in table 1, probably small enough to allow it to be placed at the Cassegrain/Gregorian focus and meet all requirements. We estimate it to have 7 % efficiency in converting stellar photons incident on the telescope in producing detected photoelectrons. This conversion includes estimates of the effects of telescope throughput, instrument throughput, PSAO module (figure 3) throughput, fiber throughput losses, imperfect adaptive optics, grating spectrograph efficiency and CCD quantum efficiency. The state-of-the-art CCD array is assumed to have 1.5 electrons RMS read-out noise and 25 electrons dark current in 30 minutes. Using this 7% efficiency and including sky background of V = 21.5/arcsec2 the Euro50 should reach a limiting magnitude of V = 21.7 and 19.2 for R = 105 and 106 respectively for an exposure time of 30 minutes (S/N = 10 at 500 nm with critical sampling) using routinely used estimating techniques.
3. Pupil slicing spectroscopy
Fastie [4] in a short OSA meeting abstract described his proposal of using pupil slicing to match large aperture telescopes to spectrographs. To the best of our knowledge pupil slicing was first implemented in the Echelle spectrograph of the early, 6-mirror MMT [5, 8]. Figure 1 shows the MMT arrangement.
Fig. 1. Pupil slicing arrangement for the 6-mirror MMT A through F represents the 6 1.8-m mirrors. The column of 6 squares in the center represents the wedgelet array at the spectrograph slit. (Courtesy Chaffee and Latham).
In normal operation the 6 star images from MMT telescopes A, B, C, D, E and F would be made to coincide by means of the MMT beamcombiner (represented as the circle in the center of Fig. 1). In the pupil slicing mode the 6 images are not combined but instead placed next to each other in a line coinciding with the Echelle spectrograph slit. Because the chief rays of the individual six f/31.6 telescopes are not parallel at this point, each of the images are placed on a small optical wedge, shown as little squares in the beamcombining circle in Fig. 1.
Fig. 2. Slicing the 618 element Euro50 primary mirror. The 18 (+ vignetted central) hexagons shown are the hexagonal sub-pupils segments each containing 32 to 33 primary mirror hexagonal elements. The large dashed hexagon represents the primary mirror outline.
These wedgelets force the chief rays to be parallel so that the spectrograph collimator combines their pupil images on the spectrograph grating. The resulting 6 parallel spectra are then recorded by the CCD in the focal plane. As a result the MMT Echelle needed to accommodate only a D = 1.8-meter (f/31.6) telescope rather than the full edge-to-edge diameter D = 6.3-meter (f/9) telescope thus giving a spectral resolution for which otherwise a much larger spectrograph would have been needed. As is also the case for image slicers, pupil slicers of this kind require a substantial slit length to accommodate the images (15 arcsec for the MMT) which compromise the spectral coverage in cross-dispersed Echelles. As will be shown later that is not the case for PSAO spectrographs.
Pupil slicing at the MMT was a natural because the telescope pupils themselves were already separate. But the same principle will work for telescopes for which this is not the case. The Euro50 telescope uses 618 hexagonal elements 2 meter in size. Referring to the primary mirror as the telescope pupil, we propose to subdivide the pupil into sub-pupils in the pattern used for the hexagonal segments in the Keck telescope. The function of the pupil slicer is to “slice” the ELT pupil into these sub-pupils. The sub-pupil configuration in Fig. 2 has 2 rings around a central segment with 19 segments total. We considered three pupil slicing configurations for the Euro 50 having 1, 2 and 3 rings respectively resulting in 6, 18 and 36 sub-pupils (one more if the strongly vignetted central segment is included) with equivalent diameters of 18.9, 11.5 and 8.2 meters respectively. Figure 3 shows a schematic of the proposed Euro50 pupil slicing arrangement. Another pupil slicing option would be to differentially tilt the sets of the primary mirror elements corresponding to the sub-pupils. It would avoid the reimaging optics in Fig. 3. Because of the proposed inclusion of adaptive optics (section 5) in the pupil slicer we pursue the option shown in Fig. 3.
Fig. 3. Schematic of the proposed Euro50 pupil slicer. The collimator mirror images the pupil on a mirror consisting of an array of flat hexagonal segments (see Fig. 2 for the 18/19 segment array). The individual segments are slightly tilted to give the (seeing limited) stellar image pattern wanted on the spectrograph slit/wedgelet combination. In the Pupil Slicing Adaptive Optics configuration (section 5) the flat hexagonal segments are replaced by adaptive mirrors, which also allow for tip-tilt control of the wavefront error. In the PSAO mode the (diffraction limited) images are recombined incoherently on the spectrograph entrance.
The equivalent mirror diameter for the 18 and 36 sub-pupil configuration is close to those of the existing very large telescopes which according to table 1 achieve spectral resolutions around between 80,000 and 160,000. Pupil slicing using seeing limited images therefore can achieve quite high R values. However image sizes for ELTs are very large (for the 50 meter f/15 focus 1 arcsec equals 3.7 mm) resulting in a very large length on the slit to accommodate the 19 or 37 images, making the cross-dispersed Echelle problematic and causing as a result serious limitations in wavelength coverage.
4. Adaptive optics spectroscopy
In diffraction limited telescopes (using adaptive optics) image sizes are dramatically smaller than the seeing disk. That allows the use of much narrower spectrograph slits. This is one of the major gains resulting from the implementation of adaptive optics [6]. When the slit width is made equal to the image FWHM the product (φ∙D) equals λ, and Eq. (2) changes to:
independent of D. The attainable spectroscopic resolution is only dependent on β and on the size of the grating expressed in wavelengths. It is independent of the telescope diameter! For Dgr = 200 mm; tanβ = 4 (the grating used in UVES and HARPS) this theoretical resolution Rtheo would be 3.2 million also independent of the sub-pupil size! Such a resolution has never (to our knowledge) been achieved in any astronomical grating spectrograph. Ultimately the resolution may be limited to 2 million due to achievable grating performance.
To our knowledge, the first successful attempt to construct an Adaptive Optics spectrograph (ACES) is described in [7]. Using the Starfire Optical Range 1.5 meter AO telescope in New Mexico and a cross-dispersed Echelle spectrograph with Dgr = 116 mm and tanβ = 2 it achieved a R of 660,000 with a He-Ne light source, close to the value derived from Eq. (3). Astronomical observations only achieved R = 250,000 because the absence of an Atmospheric Dispersion Compensator (ADC) forced the use of a wider slit. High quality ADCs are essential when doing diffraction limited imaging and spectroscopy over a wide spectral range. A rotary ADC using Risley prisms consisting of 3 different glasses achieves the required performance [10].
Present limitations in adaptive optics technology do not allow diffraction limited imaging with full aperture ELTs at visible wavelengths. A 50-meter telescope on a good site (r0 = 15 cm at 500 nm) would require some 105 adaptive elements, well beyond the present state-of-the-art. But because in the diffraction limited case R is independent of D, one can subdivide the telescope pupil without spectral resolution loss into sub-pupil diameters which can be accommodated within the likely state-of-the-art AO technology at the time of the ELT implementation. At present MEMS mirrors with 1024 elements and 2 μm stroke exist (Boston Micromachine Corporation) and mirrors with 4096 (64 × 64) elements with ∼ 4 μm stroke are anticipated to be available in the near future in time to be incorporated in the Gemini-S ExAOC instrument scheduled for commissioning in 2009. In the following we will assume that at the time of the ELT commissioning: (i) 4K element MEMS deformable mirrors (DMs) with the desired properties (∼ 4 μm stroke, ∼ 450 μm pitch) will be available, (ii) these DMs can be produced in a hexagonal format (with 3997 elements and ∼ 33 mm corner-to-corner distances), and (iii) these DMs will be “buttable” to allow a close-packed MEMS array configuration with minimal “butting losses”. Some of these DM requirements could be relaxed, at the cost of higher complexity. It should be noted that the total number of DM elements is comparable to the number needed for a non-pupil sliced telescope, however the number is divided over 18 smaller DMs which can be produced whereas a ∼105 element, monolithic DM is well beyond the present state of the art. In this paper we assume that the 18 individual DMs are not phased so that the image combination described later is incoherent.
5. A pupil slicing adaptive optics spectrograph
We will assume here the 18/19 segment pupil slicing arrangement shown in Fig. 2 resulting in 11.5 meter sub-pupil diameters and a ∼ 145 mm corner-to corner pupil image in Fig. 3. Each of the segments will be a MEMS DM whose pitch corresponds to ∼ 14 cm on the Euro50 primary mirror, close to the median r0(500 nm) of 15 cm for a good telescope site. At 500 nm the FWHM diameter of the Airy disk will be 0.009 arcsec (diameter first dark ring is 0.011 arcsec) or ∼ 70 × less than the FWHM of the seeing disk. The resulting improvement in R is much more than needed to meet the science requirements. As a result there is plenty of opportunity to optimize spectrograph parameters like size, stability, slit width etc. We describe below the components of our PSAO spectrograph strawman for the Euro50.
5.1 Pupil slicer arrangement
Except for the imaging part, the PSAO configuration for the assumed hexagonal MEMS DMs is the same as that shown for the seeing limited pupil slicing arrangement shown in Fig. 3. Instead of flat tilted mirrors, the DMs are used. In case these mirrors are not hexagonalized one could replace the pattern with one with square mirrors. That may even be more desirable for a different primary mirror configuration like that used in the TMT. Each of the DMs are mounted on a tip-tilt device to align the 18/19 images in the desired pattern and to remove the tilt component of the atmospheric wavefront distortions. For r0 = 15 cm and Dsubpupil = 11.5 meter the atmospheric wavefront distortions are 3 and 1 μm RMS without and with the tilt removed (we ignore outer-scale-of-turbulence effects). The required PTV DM stroke is therefore ∼ 6 and 2 μm respectively so that the rapidly changing (∼ 1Hz) tip-tilt corrections to the DMs are necessary.
5.2 Wavefront control
Figure 4 shows the proposed wavefront control of the PSAO equipped telescope. It relies on the laser guide star (LGS) adaptive optics system proposed for the Euro50 [2, 3]. For bright object (mV < 8) spectroscopy a LGS system would not be needed since the object itself could be used for wavefront sensing, but the power of the Euro50 lies in the spectroscopy of faint objects in which case LGS AO is essential.
A number of Na LGSs will be needed to remove the so-called cone-effect in the single conjugate adaptive optics (SCAO) arrangement. Their wavefront measurements made in the Euro50 Gregorian focus are combined by the atmospheric tomography technique [11, 12] to give the full pupil wavefront distortion for an on-axis star. The low-order distortions will be corrected by the deformable secondary mirror since their amplitudes will exceed those of the stroke of the MEMS DMs when the effects of wind forces on the telescope are included [13]. The high order wavefront distortions are extracted and divided between the sub-pupil DMs for correction. To make this correction the so-called “virtual wavefront sensing” technique is used [2, 3]. It uses an “artificial star” in the Gregorian focus whose light is re-imaged through the PSAO optics and which then is used to measure the wavefront distortion in the PSAO DM array. That distortion is then combined with the atmospheric wavefront distortion and the actuators of the DMs are adjusted to make the combination of the two zero. The same artificial star is also used to adjust the tip-tilt of the DMs to give, the desired configuration of the object images on the entrance of the spectrograph. Since the LGS system does not sense the overall tilt aberration of the atmosphere the latter has to be measured as well in the SG slit focal plane on a natural star, probably on the stellar object itself [14].
5.3 Image combination
As in the case for the seeing limited pupil slicing case (section 3), one could put the PSAO images on a line on the spectrograph slit followed by a set of wedgelets. We decided not to do so for the PSAO spectrograph since the additional (unnecessary) gain in spectrograph design dimensions was offset by the advantage of having to deal only with one image. Instead the PSAO images will be combined incoherently into one at the entrance of a fiber. The exit of this fiber serves as the spectrograph entrance slit.
5.4 The spectrograph
It is often desirable to feed the spectrograph with (a) fiber(s). This is most clearly the case when radial velocity observations of highest precision are required, using Th-Ar wavelength references as in HARPS. The fibers (one for the star, one for the Th-Ar reference source) being fixed in position and illumination with respect to the spectrograph give spectra that do not move around in the focal plane. We show our proposed fiber-fed Echelle spectrograph schematically in Fig. 5.
In this spectrograph the star image formed by incoherent addition of the 18 AO corrected images would be placed at the entrance of the star fiber after an atmospheric dispersion compensator (ADC) and a focal reducer that changes the f-ratio of the combined images to f/7.5. This f-ratio was chosen to match a 50 μm core fiber accommodating the full pupil light cone formed by the individual segment bundles entering the fiber under different angles. The f-ratio of the segment bundles is f/37.5. The fiber will scramble the light of the combined sub-pupil image and change the f-ratio of the light exiting from the fiber and entering the spectrograph due to f-ratio degradation (FRD) by the fiber to about f/6. The spectral resolution is now set by the fiber diameter (50 μm). The use of a high grade ADC is essential. Over the 380 nm to 690 nm spectral range (chosen to be equal to that of HARPS) the atmospheric dispersion at a high site (3000 m) amounts to over 2 arcsec at a zenith distance of 60°. That is 200× the combined image size of the 18 element PSAO system! It is well beyond the capabilities of a Rotational ADC (RADC) with two 2-element Risley prisms to handle. The same is the case for a Linear ADC (LADC, [15]). A RADC with well chosen glasses in 3-element Risley prisms reduced the dispersion to less than 0.001 milliarcsec. [10]. That is small enough to not affect the combined image quality and the coupling of the fiber to the combined star image.
The 50 μm fiber diameter is significantly bigger than the theoretical diameter of the perfect diffraction limited stellar image of the sub-pupil (the Airy disk). It therefore collects a larger fraction of the starlight of the sub-pupil AO corrected star images than given by the AO Strehl ratio. This fraction of stellar photons captured was assumed to be 42%, the number used in the efficiency estimate at the end of section 2. On the other hand the fiber diameter is small enough to result in a compact R = 200000 spectrograph and in a low contribution of the sky background. The spectrograph resolution according to Eq. (1) is 200000 if the following spectrograph parameters are used: tanβ = 4; Dgr = 208 mm; FRcoll = f/6 and SW = 50 μm. The focal length of the spectrograph collimator is then only 125 cm making the spectrograph very compact and stable. The spectrograph camera has a similar focal length giving good sampling of a 15 μm pixel CCD camera.
Higher spectral resolutions can be achieved by (i) using a near single mode 10 μm fiber as done with the ACES spectrograph [16] using a fiber stretcher to minimize fiber modal noise, or (ii) imaging the star directly onto the spectrograph slit rather than through a fiber. By using the same configuration shown in Fig. 5 except for replacing the fiber by a slit with a width of 18.75 μm (the FWHM of the star image) R increases by a factor 2.67 to R = 533000. The removal of the focal ratio degradation by the fiber adds to R to give 667000 with a small increase in collimator focal length (to 156 cm). Further optimization of parameters (e.g. collimator focal length, grating mosaic, separating the images as in original pupil slicing) would allow one to approach R of about one million (and more).
Fig. 5. Schematic of the Euro50 High Resolution Optical Echelle Spectrograph using an 18 element PSAO configuration. Preceding the spectrograph entrance fiber is a focal reducer and an atmospheric dispersion compensator consisting of a pair of 3-element Risley prisms. The spectrograph is a cross-dispersed Echelle spectrograph similar to the ESO HARPS. The MEMS-DM tilts in the PSAO unit are all adjusted to combine the 18 images on the fiber entrance. Not shown is the cross-disperser.
5.5 Photometry and sky-subtraction
As already pointed out a long time ago [17] photometry in Adaptive Optics aided imaging is complicated by the variable point-spread-function (PSF) across the field-of-view resulting from isoplanatic patch limitations. The same holds for spectroscopy, except that (i) PSF variations with wavelength and (ii) inadequate atmospheric dispersion correction might add to the difficulties. This complication is not just a complication for the single conjugate adaptive optics proposed here, but applies as well to wide field adaptive optics systems like those using Multi-Conjugate Adaptive Optics. As the photometry requirements for imaging and spectroscopy using AO develop this issue has to be addressed.
Sky-subtraction on the other hand should become less of an issue for point source observations since the sky background decreases as the smaller sky areas associated with AO imaging decrease.
6. Other possible applications and refinements of the PSAO concept
Pupil Slicing AO has other possible applications some of which we mention here shortly.
6.1 Near infrared high resolution spectroscopy
The requirement for DM actuator numbers in the near infrared (1 – 2.2 μm) is less stringent because of the λ1.2 dependence of r0. Nonetheless PSAO (using only 6+1) slices and DMs) may find good applications for high resolution spectroscopy even at those wavelengths. At longer wavelengths thermal emissivity considerations and less demanding DM requirements make PSAO unnecessary.
Although we expect that AO developments going on now will lead to visible light PSAO at the time of commissioning of the ELTs, that may be too optimistic an assumption. As is the case now with AO use on very large telescopes it may therefore be that the near-IR PSAO application may be a more realistic option.
6.2 High resolution imaging
PSAO can also be used for imaging. By zeroing the tilts of the MEMS-DMs the 18 images can be combined into one with a PSF equal to the diffraction limit of the sub-pupil. The diffraction limited FOV of the combined image (∼10 milli-arcsec resolution at 500 nm) equals that of the size of the isoplanatic patch or about 5 arcsec diameter, resulting in ∼ 500 resolution elements across the FOV. With the f/75 sub-pupil f-ratio the image can be well sampled by a 15 µm pixel 1K × 1K CCD.
6.3 Limited multi-object adaptive optics using sub-pupils
One way to achieve diffraction limited imaging for a number of objects in Integral Field Spectroscopy (with fields larger than the isoplanatic patch) is by the use of “AO buttons” placed in front of each of the objects [18]. Each AO button contains a DM which corrects the wavefront for that object. The technique is commonly referred to as Multi-Object Adaptive Optics, or MOAO.
PSAO presents another option for MOAO which uses the sub-pupils rather than the full telescope pupil by pointing each sub-pupil DM (or a subset of DMs) to an object of interest. In contrast to full MOAO which uses the full telescope aperture, this limited MOAO sacrifices telescope collecting area in favor of increased simplicity and realizability at visible wavelengths. It effectively uses the ELT as a battery of smaller independently pointed AO telescopes.
6.4 The use of PSAO for ground-layer adaptive optics
Ground-Layer Adaptive Optics (GLAO) is an AO mode in which only the low-altitude seeing (“ground-layer”) is corrected by a DM placed at the conjugate of that layer. The DM could be an adaptive secondary mirror (preferably a Gregorian secondary) or a mirror in the conjugate of the ground layer. Some form of atmospheric tomography would determine the wavefront distortions by that layer. Because of the uncorrected seeing at high(er) altitudes the resulting images are still seeing limited but only by the seeing at those layers. Modeling shows a seeing improvement of a factor of 2 – 5 depending on the circumstances. Because of the proximity of the ground layer the GLAO image improvement occurs over a relatively wide field-of-view. That makes it attractive for observations, like integral field spectroscopy, where a large FOV is desired. The DM for the ground layer correction has to have a high enough actuator density to correct that layer. That number is somewhat less than that discussed for PSAO, but still quite high.
Even without using the full pupil AO mode (section 6.4) one can use the described PSAO for a GLAO use. Each sub-pupil will give the seeing improved image quality of about 0.2 arcsec over a wide FOV. Incoherent combination of these images should maintain this quality.
7. Conclusion
Pupil slicing offers interesting possibilities for high resolution spectroscopy for ELTs at visible wavelengths. In seeing limited applications it allows for high resolution spectroscopy with existing spectrographs sized for current 8 – 10 meter telescopes. When adaptive optics is applied to each of the sub-pupils, diffraction limited spectroscopy and imaging becomes within reach at visible wavelengths for 20 – 50 meter aperture ELTs using high actuator number deformable mirrors like the soon to be available MEMS deformable mirrors. The diffraction limit corresponds to that of the sub-pupil diameter. As larger deformable mirrors (more actuators) become available it will be possible to gradually approach the full diffraction limit of the ELTs. Pupil Slicing AO (PSAO) thus presents a way to gradually upgrade the ELTs to their full potential starting with an already powerful high resolution capability at their first light.
In the mean time pupil-slicing using PSAO will result in the capability to do high resolution spectroscopy with ELTs with spectrographs a couple of meters in size. The technology for the PSAO is likely to exist before the commissioning of the first ELTs. This technology consists of (i) the production of ∼ 4000 actuators DMs, presently already planned for the Gemini ExAOC, but buttable to produce a DM array, and (ii) 50W Na lasers capable of creating V = 8 laser beacons including ways of removing the perspective elongation effect [19, 20]. PSAO technology can be tried and applied without pupil slicing at existing 8 – 10 meter telescopes.
The prospect of doing diffraction limited imaging and spectroscopy with those telescopes at visible wavelengths is in itself exciting. Going to the much larger aperture ELT then makes fully use of the enhanced sensitivity (S/N) those telescopes offer. Finally it should be pointed out that achieving the spectroscopy advantages using PSAO diffraction limited imaging eliminates the problems with sky background during bright time observing, thus allowing deep sky spectroscopy at normally unfavorable lunations.
Acknowledgments
Many discussions with Arne Ardeberg contributed very much to the development of the concept of the high dispersion spectrograph concept described in this paper. Don York of the University of Chicago aided in the discussion of science applications. We value the referee comments.
References and links
1. C. Pilachowski, H. Dekker, K. Hinkle, R. Tull, S. Vogt, D.D. Walker, F. Diego, and R. Angel, “High-Resolution Spectrographs for Large Telescopes,” Publ. Astron. Soc. Pac. 107, 983–989 (1995). [CrossRef]
2. T. Andersen, M. Owner-Petersen, and A. Gontcharov, “The Swedish 50 m ELT,” in Beyond conventional adaptive optics : a conference devoted to the development of adaptive optics for extremely large telescopes, E. Vernet, R. Ragazzoni, S. Esposito, and N. Hubineds., ESO Conference and Workshop Proceedings 58, 7 (2001).
3. T. Andersen, A. Ardeberg, and M. Owner-Petersen, eds., Euro50 Design Study of a 50m Adaptive Optics Telescope, (2003).
4. W.G. Fastie, J. Opt. Soc. America51, 1472 (1961).
5. F.H. Chaffee and D.W. Latham, “An image stacker for high-resolution spectroscopy on the multiple mirror telescope,” Pub. Astron. Soc. Pac. 94, 386 (1982). [CrossRef]
6. J.M. Beckers, “Adaptive optics for astronomy - Principles, performance, and applications,” Annual Rev. Astron. Astrophys. 31, 13–62 (1993). [CrossRef]
7. J. Ge, J.R.P. Angel, B. Jacobsen, N. Woolf, R.G. Fugate, J.H. Black, and M. Lloyd-Hart, “An Optical Ultrahigh-Resolution Cross-dispersed Echelle Spectrograph with Adaptive Optics,” Publ. Astron. Soc. Pac. 114, 879–891 (2002). [CrossRef]
8. J.R.P. Angel, R.L. Hilliard, and R.J. Weymann, “The MMT and the Future of Ground-Based Astronomy”, SAO Special Report, T.C. Weekes ed., 385, 87 (1979).
9. F. Diego, A.C. Fish, M.J. Barlow, I.A. Crawford, J. Spyromilio, J.M. Dryburgh, D. Brooks, I.D. Howarth, and D.D. Walker, “The Ultra-High-Resolution Facility at the Anglo-Australian Telescope,” Monthly Notices RAS 272, 323–332 (1995).
10. T.G. Hawarden, E. Atad, C.R. Cunningham, D.M. Henry, C.J. Norrie, CD. Dainty, N. Devaney, and A.S. Goncharov, “Atmospheric dispersion correction for ELT instruments,” Final Report on Atmospheric Dispersion Correction for the EC Framework Programme 6 (2006), http://www.mpia-hd.mpg.de/ELT/Manuscripts/Atad.pdf
11. J.M. Beckers, “Increasing the Size of the Isoplanatic Patch with Multiconjugate Adaptive Optics,” in Proceedings of a ESO Conference on Very Large Telescopes and their Instrumentation, M.H. Ulrich, ed., ESO Conference and Workshop Proceedings No. 30, 693–700 (1988).
12. J.M. Beckers, “Detailed compensation of atmospheric seeing using multiconjugate adaptive optics,” in Active Telescope Systems, Francois J. Roddier, ed., Proc. SPIE 1114, 215–221 (1989).
13. A. Enmark, T. Andersen, M. Browne, and M. Owner-Petersen, “Status of the integrated model of the Euro50,” in Modeling, Systems Engineering, and Project Management for Astronomy II, Martin J. Cullum and George Z. Angeli eds.,. Proc. SPIE 6271, in press (2006). [CrossRef]
14. F. Rigaut and E. Gendron, “Laser guide star in adaptive optics - The tilt determination problem,” Astron. Astroph. 261, 677–684 (1992).
15. G. Avila, G. Rupprecht, and J.M. Beckers, “Atmospheric dispersion correction for the FORS focal reducers at the ESO VLT,” in Optical Telescopes of Today and Tomorrow, Arne L Ardeberg, ed., Proc. SPIE 2871, 1135–1143 (1997). [CrossRef]
16. R.O Reynolds and M. Lloyd-Hart, “First science observations with the ACES echelle spectrograph,” in Ground-based Instrumentation for Astronomy;Alan F. Moorwood and Masanori Iye, eds., Proc. SPIE 5492, 172–180 (2004). [CrossRef]
17. J.M. Beckers and L.E. Goad, “Image reconstruction using adaptive optics,” in Instrumentation for Ground-Based Optical AstronomyLB. Robinson, ed.(Springer-Verlag, 1987), pp. 315–336.
18. M. Puech and F. Sayéde, “FALCON: extending adaptive optics corrections to cosmological fields,” in Ground-based Instrumentation for Astronomy, Alan F.M. Moorwood and Masanori Iye, eds., Proc. SPIE 5492, 303–311 (2004) [CrossRef]
19. J.M. Beckers, “Overcoming perspective elongation effects in laser guide star adaptive optics,” Appl. Opt. 31, 6592–6594 (1992). [CrossRef] [PubMed]
20. J.M. Beckers, M. Owner-Petersen, and T. Andersen, ”A rapid refocusing system for the Euro50 telescope aimed at removing the perspective elongation of sodium beacons,” in Astronomical Adaptive Optics Systems and Applications. Robert K. Tyson and Michael Lloyd-Hart, eds., Proc. SPIE 5169, 123–136 (2004). [CrossRef]
