Coronography at La Silla: High Resolution Imaging of Faint Features Near Bright Objects F. PARESCE and c. BURROWS, Space Telescope Science Institute, Baltimore, USA, and Astrophysics Division, Space Science Dept. of ESA Introduction of this article and available from the sponds to 5.10-3 square arcseconds in authors. the sky with a total rectangular field of Some of the more interesting as­ view of 22.5 by 35.9 arcseconds. tronomical sources happen to lie very The optical and detector system just close to bright objects. If the source is Experimental Setup described was calibrated absolutely us­ very faint, it becomes extremely difficult ing bright, stable stars as calibration to discern it against the glare of the The optical configuration of our ob­ sources. Using standard Johnson S, V, bright object. This situation arises, for servational setup at La Silla is shown and Rand Cousins I filters available at example, when attempting to image the schematically in Figure 1. Light from the La Silla (ESO filter numbers 445, 446, ionized tori around satellites such as 10, object under investigation is gathered 447 and 465), we obtained overall peak protoplanetary or circumstellar systems and focused by the f/8 MPI 2.2-metre counting efficiencies between 4,000 and of nearby stars, accretion disks within Cassegrain telescope onto a blackened 8,000 Aof 0.2 to 0.4 CCO counts per close binaries, faint emission photoetched mask located in the focal photon depending on bandpass. The nebulosities in the vicinity of old novae plane. This occulting mask is shaped in transmission of the achromatic doublet and possible fuzz around bright the form of a long thin wedge which can decreases rapidly below 4000 Aand is quasars. In all these cases, it is impera­ be moved longitudinally by a microme­ essentially opaque below 3500 A. Thus tive to reduce dramatically the scattered ter in order to vary its projected width in for U band measurements the lens light in the wings of the bright object the sky from 2 to 10 arcseconds de­ should be replaced by a singlet of fused seeing disk and/or to prevent the bright pending on seeing conditions, source quartz or silica and the detector object from saturating the detector and brightness, etc. It can also be moved changed to the UV sensitive GEC CCO thereby disabling it in the adjacent transversely to occult any desired por­ available at the 2.2 metre. areas. The latter effect can be important tion of the field of view. Following the even for relatively faint objects such as mask, an achromatic doublet reimages 15th magnitude quasars if long integra­ the telescope focal plane with a magnifi­ tion times are required. cation of 5 onto the detector so that the Observing Methodology Accomplishing this objective would effective focal ratio of the system be­ First, the coronograph is focused by open up a very fertile field of inquiry; one comes f/40. adjusting the position of the lens mount that is ideally suited to the performance An apodizing mask especially de­ until the image of the occulting wedge characteristics of modern high resolu­ signed to reduce diffracted stellar light under flat field illumination comes into tion, large aperture telescopes and de­ from the 2.2-metre telescope pupil is sharp focus. The coronograph comes tectors. Success requires a well-de­ located in the exit pupil and is rigidly equipped with internat LEOs to provide signed coronograph coupled to a high attached to the lens mount to ensure such illumination so that the focusing quality photon collecting device located that it remains there as the coronograph can be carried out even when the tele­ on a site with excellent seeing condi­ is focused. The mask obscures 30 per scope is not in operation. Second, a tions. This note briefly describes the cent of the exit pupil close to the images suitable star is made to fall somewhere techniques we have devised in our first of the entrance pupil edges. After this on the unocculted area of the detector attempts with a simple coronograph mask, the light passes through optical through a broad band filter and the re­ mated to the 2.2-metre MPI Cassegrain filters in a rotating commandable two­ sultant image is used to focus the tele­ telescope at La Silla and some of the wheel assembly mounted just in front of scope secondary. The coronograph results obtained in the first year of oper­ the detector. In our two runs of Sep­ need not be refocused if a filter is ation. A more complete description of tember and November 1986 at La Silla, changed since the filters are in an f/40 the experimental and data analysis we employed the 512 x 320 30 micron beam. Third, a suitably opaque neutral techniques and the scientific results can square pixels RCA CCO. In this particu­ density filter (N03 or 4) is moved into be found in the references at the end lar configuration, each pixel corre- the beam and the bright primary source acquired. Using the telescope TV autoguiding system described by Ouchateau and Tele~cope Coronagraph ~---- 1\ Ziebell in the Messenger No. 45, 1986, \ ( the source is placed behind the occult­ ing wedge in aseries of short acquisi­ tion exposures. With aseries of increas­ ingly smaller offsets, the source is pre­ ---- cisely centred behind the occulting wedge. This is accomplished by making the spillover light distribution above and below the wedge obtained with progres­ -- - sively less ND attenuation as symmetri­ cal as possible. Finally, any ND filter remaining is removed and the long ex­ 1: Optical-------configuration. Figure posure in the specified bandpass be- 43 set mainly by the time necessary to read ing. Gosmic ray events and local GGO out, process, and display the relevant defects are also located by comparison GGO acquisition images. A detector with neighbouring pixels using standard with real time display capabilities or a statistical tests and iterating to avoid faster data handling system for the GGO including bad pixels in the averaging would obviously shorten considerably process. Finally, the edges of the this set up period and allow a faster turn occulting mask are defined interactively around. This is a critical factor for a and stored in a file associated with the lengthy survey programme. image. No further user interaction is re­ quired in the subsequent processing which can proceed automatically. Data Analysis and Results The results of this process are shown All the raw images are processed in a in Figures 2 and 3 that represent images standard way to prepare them for scien­ of Beta and Alpha Pictoris, respectively, tific analysis. The GGO bias level and the taken with the system we have just de­ variations in response across the detec­ scribed with the R band filter on the tor area are removed by subtracting a night of November 27, 1986. Beta Pic­ constant bias level and dividing by an toris is a star suspected of having a appropriate flat field. Overclocked, satu­ giant protoplanetary system seen edge­ Figure 2: Image of Alpha Pictoris. The image rated pixels, and bad columns are easily on in emission surrounding the central has been flat fielded, and minor blemishes located and masked out so that they will star while Alpha Pictoris is used as the have been removed. The horizontal bar is the not contribute to the rest of the process- reference star. Both images are mainly occulting mask and has been masked black. The masked circular area in the centre with triangular appendages is due to saturated and bleeding pixels. The deviations from cir­ cular symmetry are due to instrumental scat­ tering and diffraction effects. The image is truncated at 5,000 counts, and has been displayed with the image greyscale changed to give equal numbers of pixels at any given intensity. (Histogram equalization.) gun. If a faint feature is expected to be lost in the light from the wings of the bright object seeing profile, the tele­ scope is moved to a control star nearby and the procedure just described re­ peated. After some practice, the over­ head time used to acquire and precisely register the source can be kept within approximately 20 minutes. This time is Figure 3: An image of Beta Pictoris with trun­ Figure 4: R band image of Beta Pictoris circumstellar disk .obtained as a difference image of cation at 3,000 counts. It is otherwise treated Figures 2 and 1. North is up and East is to the left. The image displayed is about 23 arcsec in identically to that of Figure 1. All instrumental width. The disk extr:mds diagonally from NE to Sw. The dWk portions of the image are due to effects are similar but there is an additional saturated pixels or the focal plane mask. The bright line rUlining vertically through the frame bulge in the image at bottom left (NE) through centre, and light close to the mask particularly to the right of centre are due to residual top right, due to the circumstellar disko diffractive scattering. 44 composed of the stellar seeing disk with small contributions due to scattering within the instrument and, of course, for Beta Pictoris, the circumstellar disko This feature, however, is almost lost in the image shown in Figure 2 without benefit of a direct comparison with the reference image shown in Figure 3. Although a feature running in a rough­ Iy NE to SW direction through the Beta Pic image, without a corresponding one to be discerned in the control image, stands out rather clearly in the image shown in Figure 2, considerably more analysis effort has to be expended in order to generate a photometrically true or correct image of the immediate surroundings of Beta Pictoris.