Technical aspects of the new AAO/UKST H interference filter

Basic filter options considered

The chosen filter design and quality must preserve the improved resolution benefits of Tech Pan film for the best possible wide-field imaging while also satisfying the scientific criteria of the new survey; namely to be able to properly sample the full range of likely velocities of gaseous H emission in our galaxy of -400 to +600 km/s. We also wish to sample H emission from our nearest major external galaxies, groups and clusters such as the SMC, LMC, South Polar group and the rich Virgo and Fornax clusters at km/s.

Satisfying these requirements necessitates careful selection of the optical and physical filter specifications including choice of bandpass (which has to incorporate the effects of use in a converging beam), central wavelength and stringent manufacturing tolerances of thickness, flatness, rigidity etc.

A full-aperture objective filter

The ideal H filter is one where the incoming beam is normal to the filter surface. In this case we would need to place the filter across the entrance aperture of the UKST in front of the corrector where the 3.3 degree deviations from the optical axis (which define the 6.6 degree field of view of the UKST) have a negligible effect on the filter bandpass ( Å). We could also choose the minimum filter bandpass necessary to satisfy our scientific requirements for sampling galactic H emission. Due to the size of the entrance aperture (1.2 m), the filter would have to be a mosaic. In principle we could also tilt the individual filter elements to tune the filter blueward within a limited wavelength range (equation 1). Unfortunately this solution proved impractical, being technically difficult and very expensive, so after initial experiments with a mosaic mask was not considered further.

The focal surface filter

The adopted solution was to obtain a single element `monolithic' interference filter of the largest size possible. This would have no edge defects, unlike a mosaiced filter where stripes of slightly lower background density would be apparent on any exposures (Elliot & Meaburn, 1976) due shadowing by the thin opaque joins between the filter components. Matching each element of a mosaic to the same optical flatness is also difficult.

After initial groundwork by David Malin, one of us (QAP), identified Barr Associates as the only company able to accept our order for a full-field single element filter and come close to satisfying the stringent specifications set by the authors. A suitably large RG610 glass substrate was obtained with some difficulty and cut to the full mm size of standard UKST glass filters. Their largest available thin-film coating plant was used to coat the multi-layer di-electric stack to form a circular `clear aperture' of about 305 mm diameter. Since it is this circular aperture coating which forms the interference filter on the RG610 glass substrate the corners of the square substrate do not form part of the H filter. This does not quite permit standard UKST survey field overlap on 5 degree field centres. We thus adopted a conservative 4 degree field centre separation for the survey to ensure proper H coverage in each field. A filter bandpass of Å was necessary to ensure proper sampling of not only the extreme velocities of the H emitting gas in our galaxy but also the requirement to cover HII regions in nearby galaxy groups such as Virgo and Fornax (at km/s) while also accounting for the blueward shifts in central filter wavelength caused principally by the UKST's f/2.48 cone-angle variations (see section 4.1).

Interference filters placed at the focal surface of a fast telescope must be carefully specified if optical aberrations and other imaging problems are to be kept within acceptable limits. Given Tech Pan's fine grain, such effects would be more noticeable than with coarser grained emulsions previously used for H work such as 098-04 and 103aE. Elliot & Meaburn (1976), detail the aberrations expected and attempt to quantify the effects on resultant image size. Many of these aberrations can be minimised by achieving tight tolerances on thickness and flatness. The final filter choice was dictated mainly by the desire to provide a filter that yields the best possible imaging so that the resolution advantages of Tech Pan are properly exploited for maximum scientific gain.

© Copyright Astronomical Society of Australia 1997