Autoguidance improves IR spectrograph resolution

Image stabilization with adaptive optics and autoguidance makes high-resolution near-IR imaging spectroscopy possible for astronomy.

Th Lfw28514 40

Autoguidance improves IR spectrograph resolution

Image stabilization with adaptive optics and autoguidance makes high-resolution near-IR imaging spectroscopy possible for astronomy.

David E. L. Freeman, Niranjan Thatte, and Harald Kroker

Near-infrared imaging spectroscopy at spatial resolutions of 0.5 arc sec may change our fundamental understanding of active galactic nuclei. Such resolutions are now achievable by combining the spatial and spectral resolution of an IR-imaging spectrometer with a first-order adaptive-optics system and accurate autoguider (see image at right).

An IR-imaging spectrometer, "3D," designed at the Max Planck Institut für Extraterrestrische Physik (Garching, Germany) is capable of simultaneous imaging and spectroscopy over an 8 ¥ 8-arc sec field in the H (1.4 to 1.8 µm) and K (1.95 to 2.45 µm) atmospheric windows. Although the imaging spectrometer itself can resolve down to 0.5 arc sec, its actual performance is limited by atmospheric seeing. This limitation is reduced by ROGUE (rapid off-axis guider experiment), a device that interfaces between the telescope and 3D and provides both first-order correction for atmospheric seeing and autoguidance. The resulting performance improvement of 3D is such that, for the first time, spectroscopy is possible down to 0.5-arc sec resolution over an extended object.

ROGUE design

The heart of ROGUE is a simple tip/tilt mirror that is moved over small angles (typically 1 mrad) to stabilize the image. The mirror is driven by signals from the guidance system, which detects movement in an auxiliary image viewed through a beamsplitter. To avoid loss of IR signal, the guidance system operates at visible wavelengths (see Fig. 1).

The performance of 3D is affected by variations in the background radiation picked up from the warm telescope optics, particularly in the K band; positioning the tip/tilt mirror at an image of the telescope pupil minimizes this problem. The mirror must also be positioned so both IR and visible (guidance) images are affected by tip/tilt corrections. Furthermore, only reflective surfaces can be used for optics prior to the tip/tilt mirror because chromatic correction is difficult over the dual waveband and because suitable antireflection coatings are expensive and inefficient. These requirements constrain the design of 3D.

In a situation in which an astronomical object under study is not bright or sharp enough to provide a sufficient visual signal for accurate guiding, a nearby star can be used for first-order stabilization, provided that it is within the isokinetic patch (the area of sky over which the first-order term is essentially constant), that is, within ~ 2 arc min of the IR target.

The guidance system forms a 4 ¥ 4-arc min image of the sky. A beamsplitter close to this image plane divides the system into two channels--a full-field acquisition channel for selecting a suitable guide star and a narrow-field detection system for tracking small movements of the guide-star image caused by atmospheric turbulence and telescope guidance errors.

The resolution of the acquisition system is not as demanding as for detection because acquisition only requires locating a suitable guide star and directing the detection system onto it. Hence, the acquisition channel images the 4 ¥ 4-arc min field onto a 512 ¥ 512 CCD array of 12-µm-square pixels.

The detection system is mounted on an x-y stage that enables it to pick up the image of any selected guide star within the 4 ¥ 4-arc min field centered on the IR target. The selected star is re-imaged onto the tip of a square reflecting pyramid that splits the image into four parts. Each part is re-imaged onto an avalanche photo diode detector (APD) with a 0.10-mm sensitive area. Movements of the guide-star image are sensed by changes in the four signals from these detectors.

The telescope

Image scale and numerical aperture at the focus and exit-pupil location all change significantly from one telescope to another. Thus, a single version of ROGUE cannot interface with every large telescope. It is possible, however, to design a modular version such that acquisition and detection systems can be common.

The first version of ROGUE was designed for the European Southern Observatory 2.2-m telescope (ESO, La Silla, Chile) and was commissioned for an observing run in August 1994. A second version will be used with either the 4.2-m William Herschel telescope (WHT, La Palma) or the 3.5-m Max Planck Institute for Astronomy telescope (MPIA, Calar Alto). The image parameters of these telescopes are close enough that a single instrument can be used with some slight rearrangement of the optical components (see table). This version of ROGUE is scheduled to be used first on the WHT in early 1996. We completed the optical design of ROGUE with Kidger`s Sigma 2000 optical design software. Its 3-D capabilities permit the system to be configured to meet the space/volume requirements. Additional features important to the design of ROGUE were operation over multiple wavebands and its use of off-the-shelf components.

The detection system

The size of the 4 ¥ 4-arc min intermediate image field is dictated by the limits of movement of the x-y translation stage carrying the detection system. Increasing the size of the field eases the optical design, but the bulk and weight of the system increase rapidly. The best compromise is a commercially available x-y stage with a 50-mm travel scanning a 45-mm-square intermediate image. For the 2.2-m ESO telescope, this gives an image scale of 5.33 arc sec/mm at a numerical aperture of 0.028.

The detection system moves on its x-y stage to view just a small fraction of the intermediate image. The system has two sections: a probe--which images the selected guide star onto the apex of the square pyramid--and the APD optics--which image each quadrant of the image onto a detector. The image at the pyramid must be of high quality--performance must be near the diffraction limit. In particular, the residual aberrations should not deform the image to a non-circular shape; that is, some spherical aberration, chromatic aberration, or field curvature are acceptable, but coma and astigmatism are not. The detection system is constructed entirely from simple off-the-shelf optics except for the square reflecting pyramid.

Magnification for the probe optics sets the image scale at 1 arc sec/mm, which is large enough to avoid performance loss caused by manufacturing defects on the pyramid but small enough to contain the image within the sensitive area of the detector under all usable states of atmospheric seeing. The probe optics are a short-focus, infinity corrected, cemented doublet. An oversized lens is used so it can accommodate the telecentric (chief rays must be parallel) input field and to allow for alignment.

The APD optics have to work at a high reduction ratio (and numerical aperture) so that each quadrant of the image is sufficiently small to be completely contained within the sensitive area of the detector under all seeing conditions. If the image overfills the sensitive area then the efficiency of the detection system is impaired. To minimize path length, the detector lens should have a short focal length, but there is not much choice in the component catalogs. A 10-mm infinity-corrected cemented doublet is used at 35X reduction and 386-mm track length. This length is folded to reduce the overall size of the detector, so care is taken to ensure there is no crosstalk between the four channels.

Although the detection field of view is small, a field lens is essential to re-image the telescope pupil at the APD lenses. This is a plano-convex lens in the probe, situated just above the pyramid and common to all four APD channels. Careful selection of the probe optics and field lens is required for all the imaging conditions to be met simultaneously.

The CCD acquisition system

The CCD views the entire 4 ¥ 4-arc min intermediate image field at 0.137X reduction and 0.205 image numerical aperture. Performance is less critical here; the only requirements are to identify a guide star within the field of view and to move the x-y table so the detection system can lock onto it.

The image reducer comprises two infinity focus systems back-to-back. A commercially available 22.5-mm, f/1.4 CCD camera lens is used for the short conjugate, and a combination of catalog doublets is used for the long conjugate. To avoid vignetting, the collimated air space is adjusted so that the telescope pupil is imaged into the camera lens aperture.

The performance from the catalog doublets improves with optimization using the air spaces between them as variables. Different combinations of doublets give widely different results. A thorough search of the lens catalogs was worthwhile, and the best combination had surprisingly good performance--much better than was predicted. The residual aberrations are principally field curvature and a little chromatic difference of magnification.

The relay system

The relay forms the intermediate image and incorporates the tip/tilt mirror and the visible/IR beamsplitter. The optics prior to the beamsplitter are common to all channels and are reflective. Any point in the image can be analyzed by the detection system. Consequently, the entire image must be flat, of high quality, and telecentric.

The relay system for the ESO version of ROGUE uses two concave mirrors with collimated light between them. The tip/tilt mirror is in the collimated beam at the image of the telescope pupil. The mirrors are used at large off-axis angles to give good clearance between incident and reflected beams. The separation of the mirrors is set to make the image telecentric.

The off-axis angles are large compared to the 4 ¥ 4-arc min field. The coma and astigmatism contributions are significant but vary only slowly over the field. The coma can be mostly corrected by setting the two mirrors off-axis in the opposite sense and balancing the off-axis angles approximately in the ratio of the magnification. The astigmatism is mostly corrected by using a shallow cylindrical mirror for the tip/tilt.

Although the basic layout of the optics is simple, it is difficult to fit them within the space available and to provide adequate access to the components for alignment. Thus, the system is folded in three dimensions.

The two-mirror relay has field curvature. It could be corrected by a field flattener near the intermediate image, but there is a shortage of space here and it would destroy the telecentric correction. As an alternative, the second mirror (visible only) is turned into a Mangin mirror using a balance between the reflective and refractive powers in the Mangin to correct field curvature. The Mangin has to be achromatized but this is possible using a pair of ordinary optical glasses in a cemented configuration.

The residuals of coma and astigmatism are small but troublesome, and they do not follow the normal field dependence of an axially symmetric system. These residuals can be corrected by decentering a fourth-power-figured aspheric plate positioned near the tip/tilt mirror and optimizing the aspheric power with the cylindrical figure on the tip/tilt mirror and the off-axis mirror angles. To save manufacturing time, the aspheric plate was replaced by a decentered and tilted, spherically surfaced, cemented doublet corrector. By optimization, all the aberrations are corrected to an exceptionally low level across the entire field.

The IR system

Although the IR channel is the raison d`être for ROGUE, it presents the least problem in optical design. The 8 ¥ 8-arc sec field must be re-imaged at a scale of 1.25 arc sec/mm for 3D, and the telescope pupil must be re-imaged out close to infinity. The first mirror of the IR relay system is common with the visible channels. A second off-axis concave mirror completes the system (see Fig. 2).

The used field of the IR system is so small that there is negligible variation of coma and astigmatism. The invariant levels can be balanced in a manner similar to the visible relay, but this would require a conflicting cylindrical figure to be applied to the tip/tilt mirror. Re-optimizing the IR and visible relays together avoids this conflict.

ROGUE version 2

The requirements for the WHT and MPIA telescopes can be met in a single instrument. Although the concept is similar, the next version of ROGUE will be quite different from the ESO version. In particular, the magnifications of the visible and IR relay systems will be different and the numerical aperture will be higher.

Version 2 will use a new relay layout derived from the Offner 1X system. The tip/tilt is a convex mirror now placed in a converging beam. The intermediate image will be formed before the image is split into the visible and IR channels. The acquisition and detection channels will be essentially the same. n

DAVID E. L. FREEMAN is an optical design consultant at Kidger Optics Ltd., 9A High Street, Crowborough, East Sussex TN6 2QA, England. NIRANJAN THATTE is a research fellow and HARALD KROKER is a PhD student at the Max Planck Institut für Extraterrestrische Physik, Postfach 1603, D-85740, Garching, Germany.

Click here to enlarge image

Image of the Seyfert 2 galaxy NGC 1068 in the [Si VI] line at 1.96 µm obtained with ROGUE and 3D at the 2.2-m ESO telescope on La Silla shows the [Si VI] emission is spatially resolved. This is the first time that extended [Si VI] emission has been observed in Seyfert nuclei. The [Si VI] line traces coronal gas, presumably ionized by the hot ionizing radiation from the central active galactic nucleus. The high spatial resolution in these data would not have been possible without ROGUE.

Click here to enlarge image

FIGURE 1. The rapid off-axis guider experiment (ROGUE) system uses a tip/tilt mirror with quadrant detector and guidance system (operating at visible wavelengths) to detect and correct for any movement of the guide star image.

Click here to enlarge image

Click here to enlarge image

FIGURE 2. The IR channel of the ROGUE system presents the least problem in optical design.

More in Optics