Ghost images generated in the Gemini Near-infrared Integral Field Spectrograph (NIFS) are required to be at a level below 10^-5 of the primary image (NIFS Functional and Performance Requirements Document, SDN0003.02). Minimizing ghost image intensities is necessary in order to faithfully record source spectra and to successfully reject OH airglow emission. NIFS will operate at spectral resolving powers sufficient to significantly separate terrestrial OH airglow emission-lines. This will allow sensitive observations to be performed in the dark regions between these emission-lines if background light levels can be maintained below the detector dark current. Ghost images of strong, variable, OH airglow emission-lines could compromise observations by lowering sensitivity and producing a time-variable background.

Ghost images generated in the NIFS optics have been identified using the Zemax ray trace program and the OptiCAD non-sequential scattering analysis program. Intensities of the major ghost images are quantified below. Further work is planned using OptiCAD to model the end-to-end ghost image performance.

4 Field Flattener Final Surface

4.1 Ghost Intensity

A relatively strong ghost is formed by light reflected off the detector and then off the rear surface of the camera field flattener (Figure 1). An image at radius r from the optical axis forms a ghost centered at 0.8r which is nearly uniformly spread over a circular region ~ 256 pixels in diameter. The reflectivity of a HAWAII-2 detector is estimated to be ~ 0.3, the reflectivity of the field flattener surface is ~ 0.02, and the quantum efficiency of the detector is ~ 0.6. The mean intensity of this ghost from a single pixel is therefore ~ 1.2´10^-7 per pixel. However, a single emission-line will form ghost images that overlap in the spatial direction. This increases the mean central intensity of the integrated ghost to ~ 3.0´10^-5 per pixel. Adjacent OH airglow emission-lines will also produce overlapping ghost images. Typically, there will be of order ~ 6 airglow emission-lines per 256 pixel spectral band with the H grating. The integrated ghost image of these six emission-lines may reach ~ 1.9´10^-4 of the mean per pixel emission-line intensity. This exceeds specification by a factor of ~ 19.

Figure 1: NIFS spectrograph camera optical design. The grating is at right. The detector is at left. Camera surfaces are numbered 1 to 10 from right to left.

The strongest OH airglow emission-lines are expected to produce ~ 25000 e in 3600 s. Airglow emission-lines in the H band typically have intensities approximately half this value. The integrated ghost image of ten such emission-lines will therefore have a mean intensity of ~ 2.4 e in 3600 s. This corresponds to a signal rate of ~ 7´10^-4 e/s/pix, which is a factor of ~ 14 below the expected detector dark current. Thus, while the ghost image formally exceeds specification, its effect due to airglow emission-lines is expected to be unmeasurable.

A bright star measured with NIFS/ALTAIR will illuminate ~ 2 full detector rows in one slitlet image (ignoring the AO seeing halo). Overlapping ghost images will then increase the per pixel ghost intensity by a factor of ~ 2 in the spatial direction and a further factor of ~ 256 in the spectral direction. The integrated ghost intensity will therefore be ~ 6.1´10^-5.

A stellar spectrum reaching the detector full-well of ~ 50,000 e will produce a ghost with ~ 3.1 e/pixel.

4.2 Mitigation Strategy

Increasing the detector/field flattener separation from the present 30 mm to 50 mm allows the offending surface to change from its present concave form to a flat surface. However, this does little to degrade the ghost intensity and significantly degrades image quality. The ghost intensity is essentially a function of the detector and field flattener reflectivities; moving the field flattener further from the detector dilutes the per pixel ghost intensity, but more pixels along each spectral feature (airglow emission-line or stellar spectrum) and more spectral features (airglow emission-lines) contribute to the integrated ghost. The integrated ghost intensity remains largely constant until the ghost images begin to over-fill the detector.

Eliminating the ghost requires either a refractive camera with no lens near the detector or a reflective camera design. Refractive designs of this sort will not be found easily. All-reflective three mirror anastigmats have been used in NIRSPEC on Keck and ISAAC on the VLT. These are expensive, have tight alignment tolerances, and may not be capable of operating over the large field angles required for NIFS.

NIFS is also required to have an upgrade path for a 5 mm cutoff detector operating to 2.5 mm. It is proposed to use the silica field flattener in the present design (cooled to the detector temperature of ~ 65 K) to block background radiation longward of 4 mm. No such blocking would be available with a three mirror anastigmat, and the whole camera of an alternate refractive design would need to be cooled to 65 K.

In view of these complexities, it is proposed that the current refractive camera design not be changed and that the ghost be tolerated.

5 Camera Shell Ghost

5.1 Ghost Intensity

A second significant ghost is generated by double reflection in the ZnSe shell (C6/C5) in the camera. This ghost is manifest as an out-of-focus halo centered at ~ 1.1r with an extent of ~ 300 pixels. The per pixel ghost intensity is ~ 5.7´10^-9 based on two reflections with reflectivities of ~ 0.02. The integrated airglow emission-line ghost intensity (as above) should then be at a level of ~ 1.3´10^-5.

5.2 Mitigation Strategy

In view of the minor contribution of the camera shell ghost to the total ghost image intensity, modifications to the camera optical design are not considered to be worthwhile.

6 C9/C7 Ghost

6.1 Ghost Intensity

A third significant ghost is generated by reflection off the front face of the camera field flattener (C9) followed by reflection off the front face of the immediately upstream camera lens (C7). We designate this the C9/C7 ghost. This ghost is manifest as a compact out-of-focus image on the detector with diameter of ~ 100 pixels for the on-axis image. The per pixel ghost intensity is estimated to be ~ 4.9´10^-8. The integrated airglow emission-line ghost intensity (as above) should then be at a level of ~ 1.2´10^-5.

All ghost images are likely to add, so in general this should be added to the field flattener final surface ghost and the camera shell ghost for a total ghost image intensity of ~ 2.2´10^-4.

6.2 Mitigation Strategy

In view of the minor contribution of the C9/C7 ghost to the total ghost image intensity, modifications to the camera optical design are not considered to be worthwhile.

7 Other Ghosts

The NIFS camera design has been checked for additional ghosts due to reflections from the detector surface. Table 1 shows the ratio of ghost intensity to image intensity for a one-pixel image in the center of the field for each camera surface. This assumes a detector quantum efficiency of 0.6 and surface reflectivities for each camera lens of 0.02. Surface 10 is the Field Flattener Final Surface ghost discussed in §4. All other ghosts listed in Table 1 are insignificant relative to that ghost. However, it should be noted that extended ghost images may add spatially.

Table 1: Single-Surface Ghost Image Intensities

Camera Surface #	Camera Component	Interface	Ghost Intensity Ratio

1	Camera Lens 1	Vac/CaF₂	3.5´10^-10
2	Camera Lens 1	CaF₂/Vac	1.3´10^-10
3	Camera Lens 2	Vac/Silica	1.3´10^-10
4	Camera Lens 2	Silica/Vac	8.0´10^-10
5	Camera Lens 3	Vac/ZnSe	6.3´10^-10
6	Camera Lens 3	ZnSe/Vac	4.8´10^-10
7	Camera Lens 4	Vac/CaF₂	1.4´10^-8
8	Camera Lens 4	CaF₂/Vac	2.0´10^-9
9	Field flattener	Vac/Silica	2.3´10^-8
10	Field flattener	Silica/Vac	1.2´10^-7

Appendix A: List of Figures

Figure 1	camera layout.wmf