Wide Field Camera 3 Instrument Mini-Handbook for Cycle 16

3. The UVIS Channel

---

The UVIS channel is optimized for highest performance in the 200-400 nm wavelength range. In concept and functionality, as well as in many design details, the UVIS channel is patterned after the ACS/WFC channel. The UVIS channel contains an optical train providing focus and alignment adjustments as well as correction for the OTA spherical aberration, a filter selection mechanism, a shutter mechanism, and a CCD detector assembly. These are supported by a thermal-control subsystem and also by control and data-handling electronics subsystems.

3.1 Field of View and Pixel Size

The field of view of WFC3 is limited by the size of the ~45 degree pick-off mirror (POM), which must be small enough to avoid vignetting of the other off-axis HST scientific instruments. The UVIS channel detector adopts the same ACS/WFC camera-head design, with 4096×4102 pixels in the focal plane. The projected pixel shape is not perfectly square due to the geometric distortions discussed in Section 2.2, and the average values of the x and y scales across the field of view are 0.0397 and 0.0395 arcsec/pixel, yielding a field of view of 163×162 arcsec.

3.2 CCD Detector

The UVIS channel detectors are two 4096×2051 pixel CCDs, butted together to yield a total 4096×4102 array with a gap between the two chips of approximately 50 pixels (2 arcsec). The gap can be filled in final images, when desired, by using appropriate dithering strategies. The CCDs are thinned, back-illuminated Marconi Corporation (now e2v technologies Ltd.) devices with UV-optimized anti-reflection coatings and 15 µm pixels. As shown in Figure 2, the CCD sensitivity extends down to 200 nm, while remaining relatively high at visible and near-IR wavelengths. The UVIS channel optics are UV-optimized, with high throughput in the 200-350 nm range achieved by employing aluminized mirrors with magnesium fluoride (MgF₂) coatings.

Although highly optimized for the UV, this design works reasonably well out into the red part of the spectrum too, yielding a reflectivity of ~88%. The overall system throughput of WFC3 in the R band is ~50% that of ACS/WFC.

Detector readout noise is expected to be 3.1 electrons rms, and the dark current less than 0.5 electrons per pixel per hour. Such low values allow background-limited observations in broad-band filters at visible and longer wavelengths in single one-orbit-long exposures.

Accurate photometric performance requires uniform response within each pixel and excellent charge-transfer efficiency (CTE), which must be stable over a relatively long lifetime in the high-radiation environment in which HST operates. The non-optimal CTE characteristics of the WFPC2 CCDs at launch, and their further degradation on orbit, have significantly limited high-accuracy photometry. The anticipated performance of the WFC3 CCD detectors has been improved by providing shielding to the CCDs (same as ACS/WFC and similar to WFPC2) and designing the CCDs with a mini-channel (improved over ACS/WFC), which will reduce the number of traps seen by small charge packets during transfers. In addition, WFC3 is the first HST instrument with the option of charge injection, which can mitigate the effects of CTE losses. In the later years, this option can be activated to fill in charge traps, without an excessive increase in the noise level of the images.

Another important detector parameter is the modulation transfer function (MTF), which defines how accurately a detector responds to a scene with high-spatial-frequency structure. MTF performance decreases with increasingly non-uniform response within a pixel (which determines the potential photometric accuracy), and with increasing cross-talk between pixels (which degrades PSF sharpness). Measured limits to the MTF of the UVIS CCDs imply less than 10% degradation of the PSF width over the whole spectral range, and a capability to do better than 2% photometry in the visible.

3.3 Spectral Elements

The UVIS channel makes use of the refurbished WF/PC1 Selectable Optical Filter Assembly (SOFA) unit. The SOFA contains 12 filter wheels, each one accommodating four spectral elements along with an open position, making a total of 48 slots available for filters or grisms. These slots have been populated with 42 narrow-, medium-, and broad-band filters, one UV grism, and 5 quad filters (each one containing 4 individual filters in a 2×2 mosaic configuration), bringing the total to 63 individual spectral elements.

Table 4 lists the available UVIS spectral elements. They include several very broad-band filters for extremely deep imaging; filters that match the most commonly used filters on WFPC2 (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 is also a wide array of narrow-band filters, which will 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.

The UV grism, G280, will provide slitless spectra with a dispersion of about 1.4 nm/pix, or a 2-pixel resolving power of about R=125. These spectra cover the approximate spectral range 200-500 nm.

In most cases, a grism observation will be accompanied by a direct image, for source identification and wavelength calibration. A suitable filter for this purpose is F200LP.

3.4 Operating Modes

The UVIS channel's only observing mode offered to General Observers (GOs) is called "ACCUM," a configuration in which photons are counted on the CCD detectors as accumulated charge after an initial reset. The minimum exposure time is 0.5 s. The charge is read out at the end of an exposure, converted to Data Numbers (DN) at a pre-defined gain (1.5 e^-/DN), stored in a data memory array, and, when the array is full, dumped to the science data recorder onboard HST.

In addition to a full-array image, the user can request subarray or on-chip binned images. A suite of fixed, pre-defined subarray apertures will be available to GOs, as well as a limited capability for user-defined subarrays. The subarray option may be used to minimize data volume and read-out time, in situations where it is scientifically acceptable to reduce the size of the field of view. Examples include observations of bright point sources or small fields, solar-system objects, or time-series monitoring of rapidly varying objects.

The on-chip binning option, with choices of 2×2 or 3×3, allows for better sensitivity in cases of low-background observations (e.g., narrow-band and/or UV imaging). Binning can reduce the contribution from CCD read-out noise, when maximum spatial resolution is of secondary importance. Binning will also reduce data volume. However, it will increase the fraction of pixels impacted by cosmic-ray hits.

GOs will be able to select from among a suite of standard dithering patterns, including both integer and fractional-pixel shifts, obtained by moving the telescope. (Note, however, that integer pixel shifts can effectively take place only over relatively small portions of the field of view because of geometric distortions.) The fractional-pixel patterns will help improve the effective resolution by allowing the user to improve the sampling of the PSF by suitable combination of the dithered images (drizzling), while at the same time minimizing the effects of detector cosmetic defects. Dither patterns will also be offered for the purpose of filling in the gap between the two CCD chips.