6.5 UVIS Spectral Elements

6.5.1 Filter and Grism Summary

An overview of the UVIS spectral elements was given in Section 2.3. This section gives further details of the UVIS filters and grism. Table 6.2 contains a complete listing of the available spectral elements in the UVIS channel. Figures 6.3 through 6.10 show the effective throughput curves, including the filter transmission convolved with the OTA, WFC3 optics, and detector response. All of the UVIS filters are contained in a multi-wheel mechanism—identical to the mechanism on WFPC2—called the Selectable Optical Filter Assembly (SOFA). Values in Table 6.2 have been calculated for UVIS chip 1.

More detailed information on the throughput curves of all of the filters is given in Appendix A; in particular, Section A.2.1 describes how to generate tabular versions of the throughput curves using stsynphot. All measurements of the UVIS filters which involve wavelengths, as tabulated in Table 6.2 and plotted in Figures 6.3 through 6.10 and in Appendix A, were done in air. The data have been converted to vacuum wavelengths using the formula given by D. C. Morton (1991, ApJS 77, 119). It should also be noted that the laboratory measurements were done at a temperature of 20°C, whereas the UVIS filters are operated on orbit at 0°C. The temperature difference may lead to wavelength shifts that are no more than 1.4 Å in the worst cases, according to the filter manufacturing specifications.

The UVIS filters have been chosen to cover a wide variety of scientific applications, ranging from color selection of distant galaxies to accurate photometry of stellar sources and narrow-band imaging of nebular gas. The set includes several very wide-band filters for extremely deep imaging, filters that match the most commonly used filters on WFPC2 and ACS (to provide continuity with previous observations), the SDSS filters, and filters that are optimized to provide maximum sensitivity to various stellar parameters (e.g., the Strömgren and Washington systems, and the F300X filter for high sensitivity to the stellar Balmer jump). There are a variety of narrow-band filters, which allow investigations of a range of physical conditions in the interstellar medium, nebulae, and solar system. A few of the narrow-band filters are also provided with slightly redshifted wavelengths, for use in extragalactic applications. Finally, there is a UV grism that provides slitless spectra with useful dispersion covering 2000–8000 Å (although the grism transmission spans the full wavelength range of the CCD).

Table 6.2: WFC3/UVIS Filters and Grism. The pivot wavelength, width and peak system throughput listed values are for UVIS chip1, except for the quad filters (see text for more details).

1. The spectral-element naming convention is as follows for both the UVIS and IR channels. All filter names begin with F, and grisms with G; if the filter is part of a four-element quad mosaic, a Q follows F. Then there is a three-digit number giving the nominal effective wavelength of the bandpass, in nm (UVIS channel) or nm/10 (IR channel). (For long-pass filters, the number is instead the nominal blue cut-off wavelength in nm.) Finally, for the filters, one or two letters indicate the bandpass width: X (extremely wide), LP (long pass), W (wide), M (medium), or N (narrow).
2. Filters intended for imaging in a red-shifted bandpass are given descriptions similar to the following: “z (Hα + [N II])”.
3. “Pivot wavelength” is a measure of the effective wavelength of a filter (see Section 9.3 and Tokunaga & Vacca 2005, PASP, 117, 421). It is calculated here based on the integrated system throughput. Filter transmissions were measured in air, but the equivalent vacuum wavelengths are reported in this table.
4. Widths listed are passband rectangular width, defined as the equivalent width divided by the maximum throughput within the filter bandpass. Equivalent width is the integral with respect to wavelength of the throughput across the filter passband.
5. km-agt (km-amagat) is a unit of vertical column density, equal to 2.69×10²⁴ molecules/cm².
6. See Chapter 8 for UVIS Grism details.

Most of the UVIS filters, as well as the UVIS grism, are full-sized elements that cover the entire UVIS field of view. However, in order to provide a larger number of bandpasses, there are five sets of “quad” filters, identified with “FQ” in the filter name, where each bandpass covers ~1/6 of the WFC3 UVIS field of view (i.e., each bandpass covers less than half of a single CCD chip). The quad filters are discussed in more detail below.

The UVIS channel is designed to be used with a single filter or grism in the light path. Unfiltered imaging, or imaging through a combination of two filters (from two different SOFA wheels), although possible in principle, would lead to significantly degraded image quality and has not been tested; thus these options are not permitted. The F200LP filter provides a clear fused silica element that approximates unfiltered imaging.

While the red blocking in the WFC3 UV filters is generally very good, resulting in negligible red leaks for hot objects (typically far less than 1% for targets with effective temperature T_eff > 10,000 K), the red leak can become significant in some filters for cooler targets (e.g., ~10% in F225W for a star with T_eff = 5000 K). More details are available in Section 6.5.2; Table 6.5 in that section tabulates red-leak values as a function of stellar effective temperature.

UV Filters

As mentioned earlier, the WFC3 UVIS optics and CCDs have been optimized for UV imaging. As such, the UV filters play a key role and considerable effort has been made to procure filters with the best possible characteristics, including maximum throughput, maximum out-of-band blocking, and minimal ghosts.

The UV filters include the shortest-wavelength F218W, intended for studies of the ISM absorption feature; the wide F225W and F275W for broad-band UV imaging; the Strömgren u (F336W) and Washington C (F390W) for stellar astrophysics; the extremely wide F300X for very deep imaging; and narrow bands such as F280N (Mg II) and the quad filters FQ232N and FQ243N ([C II] and [Ne IV]).

Ultra-Wide Filters

The selection of extremely wide (X) and long-pass (LP) filters is suited for deep imaging of faint sources. The ultra-wide F200LP filter is simply a fused-silica element with a UV-optimized anti-reflection coating which covers the UVIS channel’s entire spectral range (2000 - 10000 Å). The F200LP filter is analogous to the clear filter on STIS.

WFC3’s maximum sensitivity to hot sources can be obtained by subtracting an F350LP image from an F200LP, thereby creating a very broad ultraviolet bandpass. In Figure 6.11, the blue curve shows the filter transmission for the F200LP filter for UVIS2 at the reference epoch of June 26, 2009, and the black curve shows the effective transmission for a F200LP minus F350LP difference image. For redder targets, some additional calibration may be necessary to account for differences in the transmission of the two filters longward of ~4500 Å.

The F850LP filter is part of the Sloan Digital Sky Survey (SDSS) griz filter set, and is the reddest of the ultra-wide filters.

Wide-Band Filters

The most commonly used WFPC2 and ACS wide filters are also included in the WFC3 filter set. In addition to a wide V-band filter (F606W), there is the Johnson-Cousins BVI set (F438W, F555W, F814W).
The Sloan Digital Sky Survey (SDSS) griz filter set (F475W, F625W, F775W, F850LP) is designed to provide high throughput for the wavelengths of interest and excellent rejection of out-of-band wavelengths. These filters provide wide, non-overlapping filter bands that cover the entire range of CCD sensitivity from blue to near-IR wavelengths.

Medium-Band Filters

The medium-band filters include the Strömgren set (u, ν, b, and y), as well as some continuum bands needed for use with narrow-band imaging (F390M, FQ422M). The four 11% passband filters were added to the WFC3 UVIS set in order to cover the ~ 6000 - 9000 Å wavelength region with equal-energy filters. The “11%” refers to the filter bandwidths, which are ~11% of the central wavelength.

Narrow-Band Filters

The WFC3 UVIS channel contains 36 different narrow-band filters, covering a variety of species and most of the astrophysically interesting transitions, including Hα, Hβ, Hγ, He II, C II], [N II], [O I], [O II], [O III], [Ne IV], [Ne V], [S II], and Ca II. The methane absorption bands seen in planets, cool stars, and brown dwarfs are also covered.

Cosmological emission lines can be detected across a range of redshifts within the bandpasses of the narrow-band filters. Table 6.3 lists the redshifts that can be probed using the specified narrow emission lines (hence, no entries for broad absorption bands or continuum or “off” bands). These redshift ranges are offered as a guide; exact values depend on the wavelengths of the filter cutoffs. Filter cutoffs used in Table 6.3 were defined using the passband rectangular widths (defined in Footnote ⁴ of Table 6.2). However, passband cutoffs were not centered on the filter pivot wavelengths λ_p (defined in Section 9.3), because red leaks shift the pivot wavelengths to longer wavelengths by 1-9% in some of the ultraviolet filters. Instead, the central wavelength for each filter was determined by maximizing the wavelength-integrated product of a rectangular passband of the specified width with the actual system throughput for the filter. In the most extreme case (FQ232N), the pivot wavelength of 2413 Å is more than two bandpass widths to the red of the rectangular passband equivalent central wavelength (2326 Å). 

Table 6.3: Nominal redshift ranges for WFC3/UVIS narrow-band filters.

Quad Filters

The WFC3 UVIS channel contains five quad filters: each is a 2 × 2 mosaic of filter elements occupying a single filter slot, with each quadrant providing a different bandpass (typically narrow-band, although there are also several bandpasses intended for continuum measurements). The five quad filter sets on WFC3 significantly increase the number of available narrow-band filters. The WFC3 quad filters are listed in Table 6.4 with their readout amplifiers. The pivot wavelength, equivalent width and peak system throughput for these filters (see Table 6.2) were calculated by using the filter curves updated in October 2020: these are based on the original UVIS filter curves, aperture corrections and in-flight corrections (for more details see WFC3 ISR 2021-04).

A quadrant nominally covers only 1/4 of the WFC3 total field of view or about 80"× 80", although edge effects (Figure 6.12) result in a useable field of about 1/6 of the field of view. The filter edges are out of focus on the focal plane, so light from multiple passbands reaches the detector in those areas.

In programs where targets are placed in different quadrants during a single orbit, spacecraft maneuvers may be large enough to force a new guide star acquisition. Guide star acquisition overheads are described in Section 10.2.

Table 6.4: Quad filter names and positions (identified by readout amplifier).

Grism

The UVIS channel has a UV grism (G280), a spare element from WF/PC-1. It provides slitless spectra with a dispersion of about 14 Å/pix and a spectral resolution of about 70 over the 2000 - 8000 Å wavelength range, but with transmission in the zeroth order over the entire response of the CCD (see Figure 8.3). At wavelengths longer than ~ 4000 Å, however, reduced sensitivity and overlapping orders may complicate analysis. Typically, a grism observation is accompanied by a direct image, for source identification and wavelength zero-point calibration; an ideal filter for the identification image is the F300X discussed above. Chapter 8 discusses WFC3 slitless spectroscopy in detail.

6.5.2 Filter Red Leaks

The design and manufacture of the UV filters was based on a careful balance of the in- and out-of-band transmissions: in general, higher in-band transmission results in poorer suppression of out-of-band transmission, and vice versa. The WFC3 filters represent an attempt to achieve an optimum result, maximizing the in-band transmission while keeping the out-of-band transmission as low as possible in order to minimize red leaks.

Table 6.5 below summarizes the red-leak levels for the WFC3 UV filters. The table lists the fraction of the total signal that is due to flux longward of 4000 Å, as a function of effective temperature. This was calculated by convolving a blackbody of the given T_eff with the system throughput in the listed filter. As can be seen from the table, red leaks should not be an issue for observations of any objects taken with F275W or F336W. The other UV filters have some red leaks, whose importance depends on stellar temperature. The red leaks in F218W and F300X, for example, exceed ~1% for objects cooler than ~6000 K, while in F225W the red leak reaches ~1% for objects with even cooler temperatures. The most extreme red leaks arise from F218W and F225W observations of objects with T_eff of ~4000 K or cooler, necessitating appropriate corrections.

Table 6.5: Fraction of flux longward of 4000 Å as a function of effective temperature.

6.5.3 Ghosts

The WFC3 UVIS channel exhibits three different types of optical ghosts: a) those due to reflections between the CCD front surface and the two detector package windows; b) those due to reflections between the window surfaces; and c) those due to reflections within the particular filter in use.

Window Ghosts

Window ghosts were predicted from early models of the UVIS detector (WFC3 ISR 2001-17) and were seen in ground testing using an optical stimulus (WFC3 ISR 2004-04).

When a point source is positioned in the lower right quadrant of the UVIS detector, out-of-focus reflections between the CCD and windows appear along a diagonal from the source towards the upper left, well removed from the source. These figure-eight shaped ghosts gradually move outside the field of view as the target moves out of the lower right corner. They contain a few percent of the flux of the target. Ghosts from a bright, saturated star are visible in the F606W exposure shown in Figure 6.13. Note that smaller window ghosts appear closer to the target; they are due to reflections between the window surfaces. In this case, the source from which the ghosts originate is so saturated that excessive charge has bloomed column-wise above and below the PSF (see Section 5.4.5 for a discussion of charge overflow due to saturation). 

To prevent the worst effects of window ghosts, avoid placing very bright targets on the D quadrant. Also, pay attention to the location of key science targets if bright sources are in the lower right area of the field of view. The production of window ghosts has been modeled and an aid to observers has been produced to enable them to estimate the position of ghosts for a given field of view and ORIENT (WFC3 ISR 2011-16); this map is available as an XML overlay in APT (when viewing in Aladin, select "Window Ghosts" from the "Instrument Fields of Views (FoV)" menu). If ORIENTs are necessary, they must be requested in the Phase I proposal; the ORIENT special requirements can then be imposed within APT at the Phase II proposal preparation stage to control the positioning of bright sources on the detector. See Section 7.2.2 of the Phase II Proposal Instructions, which gives detailed information on the relationship between detector coordinates, spacecraft coordinates, and ORIENT.

Filter Ghosts

Filter ghosts for the WFC3 filters were specified to be less than 0.2%, and in most cases were measured during ground testing to be less than 0.1%. A few filters however, were found during testing to have ghosts that exceeded the specification. Some of these, the ones deemed highest priority, were remanufactured and installed in the SOFA (WFC3 ISR 2007-01). Consequently, there are a relatively small number of filters that exhibit filter ghosts at a level >0.2%. These are listed in Table 6.6. They have been retained in the flight instrument either because they were of lower scientific priority, or because the ghost level was deemed acceptable in light of the otherwise excellent performance characteristics of the filters (e.g., in- and out-of-band transmission, sharpness of bandpass edges). While some scientific programs (e.g., stellar photometry) may be unaffected by filter ghosts, others (e.g., observations of extended targets or faint objects adjacent to bright ones) could be adversely affected. In such cases, extra planning and/or data-analysis efforts may be needed, e.g., combining images taken at different dither positions and/or roll angles, or applying a deconvolution algorithm.

Table 6.6: Filters exceeding the filter ghost requirement, measured during ground testing of the integrated instrument (see WFC3 ISR 2007-09).

¹ Laboratory measurement of stand-alone filter.