8.2 Slitless Spectroscopy with the UVIS G280 Grism

The G280 grism is a WF/PC1 spare. Figure 8.1 shows a spectrum of the wavelength calibration star WR14 observed as part of the Cycle 17 calibration program 11935. The circled spot shows the location of a direct image of the source obtained with a separate (undispersed) F300X filter exposure, but superposed on the grism image for illustrative purposes only. The prominent star-like feature near the center of the picture is the zeroth-order grism image, and the positive and negative higher orders extend towards the left and right of the zeroth order, respectively. The +1st order is defined to be the order with the higher throughput (due to the grating blaze), even though it falls at lower x-pixels than the position of the zeroth order. The +1st order extends to the left of the zeroth order a distance of about 1/4 of the image size. Further left there is heavy overlap with higher orders. Some prominent emission lines can be seen along the spectral trace.

There are several features of this grism that differ, for example, from the G800L grism on ACS. There is an offset of about 175 pixels in the y-direction between the direct image and the spectra, the zeroth-order is relatively bright due to a lower grating efficiency and clear substrate, and there is curvature of the spectra at the blue ends of the first orders (nearest the zeroth order). The amplitude of the curvature is about 30 pixels in the detector y-direction. Figure 8.2 shows a close up view of the first few positive orders of the WR14 spectrum, which illustrates the curvature at the short-wavelength end of each order.The spectral trace and dispersion have been measured both during ground calibration and on-orbit using the wavelength calibration star WR14. The mean dispersion in the 1st order at 2200 Å is 12.2 Å/pixel, varying from 10.8 Å/pixel at 1850 Å to 14.4 Å/pixel at 4000 Å. The dispersion per pixel in the higher orders is higher by the approximate ratio of the orders; for example, in +3rd order it is 48 Å/pixel.

The flux calibration of G280 spectra is best at the center of chip 2, and next best at the center of chip 1. Observers having a single target of interest are encouraged to place the target near the center of chip 2, due to the higher QE of chip 2 at very short wavelengths.

Field-dependent calibrations of the trace and dispersion for the -1 and +1 orders over the entire UVIS field of view are presented in WFC3 ISR 2017-20. The best fit to the shape of the G280 curved spectra on multiple locations on the detector was achieved using a 6th order trace polynomial with a linear 2D field dependence. The model reproduces the location of the curved traces of the +1 and -1 order spectra to within a fraction of a pixel. It fits a longer portion of the red part of each spectrum than the earlier ground calibration, and overall is a significant improvement over that calibration. The accuracy of the wavelength calibration is estimated to be ±7 Å, or about half of a UVIS resolution element.

A full calibration of the WFC3 UVIS G280 slitless spectroscopic mode was performed in 2020 (see WFC3 ISR 2020-09). We have combined all of the G280 calibration data (nearly 600 datasets) obtained over the years to determine the traces, wavelength calibration, and flux calibration across the entire field of view of both UVIS detectors, modeling the large amount of field dependence of the G280 grism. This was done for the ±1, ±2, ±3, and ± 4 orders, and we have calibrated the position of the 0th order as well.

Figure 8.3 shows the total 1st-order transmission for the WFC3 G280 mode (including the instrument and telescope) for observations on chip 2, as determined from observations of the HST standard star GD-71 during Cycle 17. The +1st order is more sensitive than the zeroth order at wavelengths less than 320 nm, but longward of this wavelength the zeroth order dominates. On deep exposures, orders out to at least -8 and +8 have been detected. The current absolute flux calibration of the G280 is estimated to be accurate to better than 10%. Due to the relatively high throughput of the 2nd order, 1st order spectra longward of 400 nm are likely to be overlapped by 2nd order light longward of 200 nm, at least for sufficiently hot sources.

In 2023, the first background sky images for G280 were released (WFC3 ISR 2023-06). One set (two files, one for each CCD) were derived from individual flat-fielded (FLT) science exposures, while a second set were derived from their corresponding CTE-corrected (FLC) frames. Multiple parameters, such as the Earth limb angle, higher sun altitude, smaller sun angle, and the presence of zodiacal light, can increase the sky background for any given observation. However, it was found that all stray light scatters onto the CCD in a similar manner, so the background is well-modeled by a single component. While He I emission was detected in the WFC3/IR grisms (WFC3 ISR 2014-03), no emission line features have been detected in G280.

Grism exposures of a given target field should always be accompanied by a direct image. The direct image provides source sizes and identifications, from which the corresponding locations of spectra in the grism images are determined. Knowledge of the direct-to-grism exposure source offsets is necessary to set the wavelength zero-point for each extracted spectrum and the source size measurements enable the software extraction slit to be tuned to each object (see Section 8.5). The 0th-order trace in G280 images is slightly dispersed (making it difficult to centroid) and often saturated (making it impossible to centroid), therefore it cannot be used in place of a direct image.

The natural choice of a direct-imaging filter to provide the reference image for G280 exposures is the F300X, because its response matches most closely the +1st-order grism response. The broader F200LP filter may be preferable for fainter objects. The shape of the spectra in G280 exposures varies significantly across the field of view but has been calibrated. The sensitivity of the +1 order also varies, but its field variation could not be handled by the aXe package and is only calibrated at the center of chip 2. The separate software package hstaxe, a Python follow-up to aXe, can now be used to extract and calibrate one-dimensional spectra from WFC3 grism exposures. Jupyter cookbooks illustrating the procedures for full- and subarray-frames of G280 data are available in the WFC3 section of the hstaxe Github repository. Small dithers of G280 exposures and accompanying direct image exposures are recommended to remove the effects of bad pixels. In addition, it is recommended that CR_SPLIT be used in both the direct and dispersed exposures to allow for cosmic-ray identification and removal.

G280 exposures can only be obtained using the “UVIS” aperture selection, which places the target at the reference point of the UVIS field of view, about 10" above the inter-chip gap, i.e. on chip 1. Accompanying direct images should use the G280-REF aperture selection, which places the target at the same location as in the dispersed exposures. If your observations are of a single primary target, it is best to specify a POS TARG Y of -50", for both the dispersed and direct exposures, to place the target at the center of chip 2, where the spectral calibrations are best determined and the near-UV sensitivity is somewhat higher than on chip 1.

For sufficiently bright objects, the multiple spectral orders may extend across the full field of view of the camera. This leads to overlapping of fainter spectra in the vicinity of the bright-object spectrum. Therefore, a careful determination of the optimum telescope roll angle is required to obtain non-overlapped spectra of faint objects in the vicinity of much brighter objects. i.e., the observer needs to set the orientation of the detector on the sky by using the Visit Orientation Requirements parameter “ORIENT” in the phase II proposal; e.g. ORIENT = 135.17 degrees aligns the Y axis of the UVIS detector with North for the G280 and G280-REF apertures. See Section 6.2. of the Phase II Proposal Instructions, which gives detailed information on the relationship between detector coordinates, spacecraft coordinates, and ORIENT. Note: if ORIENTs are necessary, they must be requested in the Phase I proposal.

CTE losses can be large for faint spectra on faint backgrounds. They should be taken into account when planning UVIS exposures. See Sections 5.4.11 and 6.9 for information on CTE and observing strategies.