2 Measurements

The data were collected in May 2001 using NICS, the near infrared camera and spectrometer which is permanently mounted on the Italian TNG telescope in La Palma. This instrument is a FOSC-type cryogenic focal reducer equipped with two interchangeable cameras feeding a Rockwell Hawaii 1024² array. The camera used for the spectroscopic observations has a focal ratio of F/4.3 and yields a sky projected scale of $0.25\arcsec\!$/pixel. The spectroscopic modes are achieved by means of an Amici prism and a series of glass-resin grisms which can be inserted in the $\oslash$ 22 mm collimated beam (see e.g. Baffa et al. 2001).

All the grisms used for these measurements have their dedicated order sorter rigidly mounted inside the grism holder. Consequently, normal long-slit spectroscopic data are collected with the filter wheel in the "open'' position. Therefore, by inserting a filter in the filter wheel, one can measure its transmission at the wavelengths covered by the disperser.

The measurements were performed as follows. A cold entrance slit of 0.095 mm, with a projected size of 2 pixels onto the array, was illuminated by an halogen lamp through a diffuser. After the collimator, the light passed through a Lyot stop, the filter and grism wheels before reaching the camera and the detector. A first series of spectra of the halogen lamp were taken by inserting one of the dispersers available in the grism wheel followed, immediately after, by a series of spectra with a filter inserted in the filter wheel and, again, by measurements of the halogen lamp without any filter. Integration times were identical for all frames and long enough to obtain between 5000 and 30  000 ${\rm e}^-$ per pixels at all the wavelengths of interest, i.e. a good level of illumination remaining well below the limits of non-linearity. Each of these cycles lasted a few minutes during which the input flux from the lamp was found to be stable within better than 1%.

The dispersers used were the grisms IJ (0.89-1.45 $\mu$m, 5.5 Å/pix), JH (1.15-1.75 $\mu$m, 6.6 Å/pix), JK (1.15-2.20 $\mu$m, 11.6 Å/pix) and HK (1.40-2.50 $\mu$m, 11.2 Å/pix) which allowed determining the filters transmission with quite fine spectral details. Measurements with a factor of about 10 lower dispersion, i.e. at $\lambda/\Delta\lambda\!\simeq\!100$, were also taken with the Amici prism (0.8-2.6 $\mu$m), these data were particularly useful to estimate the out-of-band blocking factor which, in the red part of the spectrum, and up to array cutoff wavelength of 2.6 $\mu$m, could be checked to a level of about 0.01%.

The transmission of a given filter was simply determined by dividing frames taken with and without the filter in the collimated beam. Given the very high s/n ratio of the spectra, the internal errors of the measurements are negligibly small. The actual accuracy of the curves is practically limited by systematic effects due, for example, to the fact that the filters are tilted by 3-5 degrees relative to the optical axis and, when they are inserted in the beam, shift the pupil image by $\simeq$0.2 mm, i.e. $\simeq$1% of the pupil diameter. Therefore, slightly different parts of the dispersers were illuminated in the measurements with and without the filter. Nevertheless, each filter transmission could be independently determined using at least two different dispersers and, consequently, we could estimate that the systematic errors are <4% in the absolute values and $\le$1% in the shape of the curves.

Wavelength calibration was performed using exposures of Ar and Xe lamps which were taken before/after every change of disperser, the dispersion was always found to be stable within <0.1 pixels, i.e. a factor of >20 better than the $\lambda/\Delta\lambda \simeq 1000$ resolving power of the spectra.

The variation of the transmission curves with incidence angle was determined by extracting spectra at different distances for the array center. The data for narrow band filters spanned a range of incidence angle from $3^\circ$ to $6.5^\circ$ while those for broad band filters extended between $5^\circ$ and $7.6^\circ$.

 

 

Table 1: Summary of filters characteristics¹.

Name	cuton2	cutoff2	$<\!t\!>^{3}$
	($\mu$m)	($\mu$m)	(%)
$J_{\rm {MKO}}$	1.172	1.335	84
$H_{\rm {MKO}}$	1.481	1.779	81
$K\arcmin_{\rm {MKO}}$	1.949	2.295	88
$K_{\rm {MKO}}$	2.035	2.374	85
Hcont	1.5580	1.5807	67
FeII	1.6304	1.6576	72
H2	2.1061	2.1376	66
Br $\gamma^{4}$	2.1515	2.1868	62
Kcont4	2.2551	2.2941	60

¹  Measured at T=78 K and at an incidence angle of 5$^\circ$ (broad band filters) and $3^\circ$ (narrow band filters).
²  Half-power points.
³  Average transmission between 75% points.
⁴  The transmission curve is quite irregular, see Fig. 2.