CFHT Information Bulletin, number 38, First Semester 1998

  

 This short note reports some comments about the problems encountered when one attempts to flux calibrate the spectra observed over a large wavelength range. It concerns all the modes of the MOS/SIS/ARGUS instrument in the reddest part of the optical range (from 7500 Å to 9000 Å ), now available thanks to the high efficiency of the CCDs from 4000 Å to 9000 Å. Similar problems were mentionned by Crawford and Vanderriest (1997, MNRAS 285, 580) who observed quasar environments with MOS/ARGUS. In order to correctly reproduce the line ratios of the quasar already published, they had to include a "correction factor" increasing up than a factor of 2 at 9000 Å.

 Contamination by the second order is responsible for it as no order separation filter is included in the instrument setup of MOS. An illustration of the effect is shown in Figure 32: two spectra are displayed, the top one being obtained with an Helium-Argon lamp and the bottom one with the same lamp and a B filter. The presence of the contaminating lines from the second order is obvious. In addition, the most UV lines (the strong He line at 3888 Å for example) are stronger in the second order than in the first one! This is due to the blaze function optimised in the first order, which is distorted in the second order towards much bluer wavelengths.

 The consequences of this effect which was not taken into account in the MOS instrument are multiple and we present only a few of them:

• Difficulties in flux calibration are essentially due to the fact the the spectrophotometric standard stars are generally much bluer than the faint objects observed in extragalactic astronomy, such as redshifted galaxies or distant quasars. This means that the contamination by the second order will be stronger for the standard star, with its red flux being overestimated by a significant factor redward of 7000 Å, because of its bright UV flux. The sensitivity function obtained by comparing this observed flux with absolute tabulated flux will be underestimated in the red, leading to a necessary correction curve which increases above 7000 Å. 
• Emission line objects can be polluted by the same phenomenon and some unidentified lines may arise by this kind of contamination. In particular, we suspect such an effect in the spectrum of the radiosource MG2016+112 obtained last summer at CFHT with OSIS/V and the R150 grism. The line seen at  = 8760 Å, although marginally detected, does not correspond to any known emission line at the quasar redshift. But it could correspond the the Ly line at the quasar redshift (z = 3.26), seen in the second order. Another example was found in the spectrum of a galaxy at a redshfit of 2.8 obtained last September with MOS and the O300 grism, in which the Ly emission line is seen both at 4624 Å and at 9248 Å. This is a serious warning about the detection of unidentified emission lines in the red!!!

• To overcome this difficulty and after discussions with people in charge of MOS/SIS at CFHT, a filter GG455 was ordered recently and added to the existing ones at CFHT. Its transmission covers the range 4800 to 9000 Å and will allow a correct flux calibration of the red spectra. In conclusion, no spectroscopy with accurate flux calibration from 4000 Å to 10000Å is possible with the existing spectrographs in low resolution mode in only one single exposure, and wide band filters to separate the orders are mandatory for a good restitution of the flux distribution, especially in the red.

[Editor's note: CFHT is grateful to the authors for bringing this problem to our attention and will ensure that future observers are fully appraised of the situation and appropriately advised as to its solution.]