NICMOS provides grism imaging spectroscopy in the spectral range
between 0.8 and 2.5 μ
m with Camera 3 (Storrs, NICMOS ISR-97-027
). NICMOS is used in this mode of operation without any slit or aperture at the input focus, so all objects in the field of view are dispersed for true multi-object spectroscopy. The grisms reside in the NIC3 filter wheel, therefore the spatial resolution of the spectroscopy is that of this Camera. The filter wheel contains three grisms (G096, G141, G206), of infrared grade fused silica, which cover the entire NICMOS wavelength range with a spectral resolving power of ~200 per pixel.
A grism is a combination of a prism and grating arranged to keep light at
a chosen central wavelength undeviated as it passes through the grism. The resolution of a grism is proportional to the tangent of the wedge angle of the prism in much the same way as the resolution of gratings are proportional to the angle between the input and the normal to the grating.
Grisms are normally inserted into a collimated camera beam. The grism
then creates a dispersed spectrum centered on the location of the object in the camera field of view. Figure 5.9
shows an example of grism spectra of point sources using G096, G141, and G206. The target is the brightest source in the FOV, although many other sources yield useful spectra as well. The band along the bottom of the images, about ~15–20 rows wide, is due to vignetting by the FDA mask, while the faint dispersed light on the right edge of the G206 grating image is due to the warm edge of the aperture mask.
The two shorter wavelength grisms exploit the low natural background
of HST while the longest wavelength grism is subject to the thermal background emission from HST.
Grism observations are carried out in a similar manner as other
NICMOS imaging. For accurate wavelength calibration, it is essential to pair each grism observation with a direct image of the field in NIC3, through an appropriate filter, at the same pointing. This provides the location of each object in the field and aids in the identification of their individual spectra. Because of this natural pairing, most spectroscopy observations will be a two image set, direct and grism images. For estimating the S/N for both the filter and grism observations a NICMOS spectroscopic ETC is available:
There are two separate considerations for the strategy to place the
spectra on the NICMOS detector. One is the choice of orientation, the other one is the placement relative to the edges of the detector. The orientation of the spectra relative to the detector is important because the spectra of random objects in the field might overlap the spectra of the target objects. The orientation is also important for slightly extended objects. In such cases, it may be possible to alleviate the superposition of spectra by requesting a specific orientation (roll-angle) of the telescope during the Phase II Proposal submission. For complex fields or extended targets, observations of the same field at three or more different spacecraft orientations are advisable to allow deconvolution of overlapping spectra. A review of the target field to determine if spectral contamination caused by field objects can be avoided, should be the first step in choosing a pointing strategy. If infrared images of the target field are not available, then DSS images should be examined. It is important to find a range of possible roll-angles, since restricting the roll-angle limits the schedulability of the program.
The next step of the pointing strategy is the placement of the target
spectra onto the detector. The four different quadrants of the NICMOS detector are read separately. Therefore, differences in the bias level, sensitivity and/or noise between the quadrants are often unavoidable. For that reason, spectra should be placed so that they do not cross the boundary between the quadrants. The best positions are close to the center of any of the four quadrants of the NICMOS detector, i.e. as close as possible to any of the points (64,64), (192,64), (64,192) or (192,192). The more important part is the x-coordinate of the detector. The pointing of NICMOS is specified for the undispersed image. To place the spectra at the desired position, one has to take into account that the spectra taken through any of the grisms are displaced relative to the position of the undispersed image taken through one of the filters. The offsets in pixels for each grism are listed in Table 5.4. These offsets have to be taken into account when placing the spectra on the detector. In order to place the center of the spectrum at a desired position, the pixel values listed in Table 5.4 have to be added to the specified pixel position on the detector. For example, in order to place a G141 spectrum at the center of the first quadrant, pixel coordinates x=54+6.7, y=64+13 have to be specified. If only a single target is observed, we recommend to split up the integration time into four exposures and place the spectrum at the center of all four quadrants. This improves the flux calibration of extracted spectra.
We encourage all grism observers to dither their observations. Dithering
the target on the detector will minimize image anomalies such as grot affected pixels, cosmic ray hits, pixel sensitivities, and residual persistence images. The sequence of images should always be: direct and grism images at the first dither point, move to next dither position, direct and grism images at the second point, etc. This can be achieved using the pattern syntax (see Appendix D). Because of intrapixel sensitivity variations (See Section 5.3.5), dither spacing should be a non-integer number of pixels, e.g. 2.1 arcsec (10 and a half pixels).
The best dither pattern is to move in both directions (pattern
NIC-SPIRAL-DITH). This will improve line fluxes, wavelength measurement of lines, and help to verify broader spectra features. Dithering parallel to the dispersion may result in loss of data off the edge of the detector, or move spectra at a position where they cross the boundary of a quadrant. For a single target, this can easily be avoided by choosing a dither pattern which places the spectrum close to the center of all four quadrants. In a more complex situation with several targets, the design of a dither strategy needs to include the consideration of roll angles.
The NICMOS spectroscopic grism mode calibrations were determined
from on-orbit observations. Wavelength calibration was carried out by observing planetary nebulae, Vy 2-2 and HB12. The inverse sensitivity curve is derived from observations of the white dwarf G191-B2B and G-dwarf P330E. Grism calibration data reductions were performed at the Space Telescope European Coordinating Facility (ST-ECF). An IDL software package (NICMOSlook), as well as a newer C/Python package (aXe) are available from the ST-ECF web site to extract and calibrate spectra from pairs of direct and grism images. NICMOSlook
is available from the ST-ECF NICMOS Web page:
gives the dispersion relationship in the form:
is the x coordinate relative to the center of the object on the direct image, and λ
is the wavelength in μ
m. The relationship is plotted in Figure 5.10
. The grisms were aligned as accurately as possible along a row or column of the array. However, there is a slight tilt to the spectra; 3.1 degrees for G096, 0.7 degrees for G141, and about 1 degree for G206. Distortion and curvature in the spectrum are negligible.
Background radiation is a greater concern for grisms than for imaging
observations. Every pixel on the array receives background radiation over the spectral bandpass of the particular grism, while the source spectrum is dispersed over many pixels. Therefore, the ratio of the source to background flux is much lower for the grisms than for the regular imaging mode filters. The background rate per pixel (sky + telescope) expected with NCS operations is presented in Table 5.6
below for the three grisms. Observing a source with flux at all wavelengths equal to the peak response for each grism will result in a peak count rate equal to the background. The increase in the background flux for the G206 grism is dramatic. Grisms G096 and G141 should therefore be used whenever possible. Despite its broad wavelength coverage, the G206 grism should be used for the longest wavelengths only. Dithered observations, especially when the field is uncrowded, can often be used to remove the background quite well.
gives the sensitivity of each grism as a function of wavelength, as measured for standard stars post-NCS. The signal was measured in an aperture of 10 pixels (2 arcsec) in the spatial direction. Tables 5.7
, and 5.9
present the basic information for the three NICMOS grisms, as well as the best direct imaging filter to associate with each.
The same intrapixel sensitivity problem which affects NIC3 images (see Chapter 4
) will affect the grism spectra, since the dispersion direction is not exactly aligned with the detector rows: as the heart of the spectrum crosses from one row to the next, the flux will dip by 10-20%. The size of the effect depends on the size of the object. This effect is not obvious in emission line spectra, but can be very clear in continuous spectra. The number and placement of the sensitivity minima within the spectrum will depend on exactly where the spectrum falls on the detector, and the angle between the dispersion direction and the detector X axis. Note that the former changes with the dithering position, and the latter changes between observations. A correction procedure for this effect is available in NICMOSlook
The decision chart given in Figure 5.12
helps summarize the recommendations of Section 5.3.