Wide Field Camera 3 Instrument Mini-Handbook for Cycle 16

4. The IR Channel

---

The IR Channel is optimized over the wavelength range 900-1700 nm. The IR optics, with the exception of the WFC3 pick-off mirror (which intercepts the central bundle of light from the OTA and is shared with the UVIS channel), are coated with a protected silver layer for maximum IR throughput. The IR channel consists of the following components: (a) the Channel Select Mechanism (CSM), which diverts light from the UVIS channel optical train; (b) a two-mirror mechanism providing focus and alignment adjustments; (c) a lens (the Refractive Corrector Plate or RCP) for spherical-aberration correction; (d) the Filter Selection Mechanism (FSM); and finally (e) the HgCdTe IR detector assembly inside a vacuum enclosure sealed by a transparent window. The IR optical chain is supported by a thermal control subsystem and also by control and data-handling electronics subsystems. Because the detector is electronically shuttered, a mechanical shutter is not needed; bright-object protection when the IR channel is not in use is obtained by blocking the optical path with an opaque slot in the FSM.

4.1 Field of View and Pixel Size

The pixel size for the IR detector, as projected onto the sky, was selected by the SOC and Science IPT after careful analysis; it achieves an optimal balance between size of the field of view and PSF sampling. Simulations showed that dithering and image-reconstruction techniques would be able to reconstruct a good-quality PSF satisfactorily, as long as the pixel size was below 0.15 arcsec. The final choice of ~0.13×0.13 arcsec satisfies this requirement, while at the same time conforming to optical packaging considerations. Because the focal plane is tilted by about 22 degrees with respect to the beam, the field of view is rectangular in shape, with an aspect ratio of ~0.90. The scale of the rectangular pixels is 0.121×0.135 arcsec, yielding a useful field of view of 123×137 arcsec.

4.2 IR Detector Array

The WFC3 IR detector is a Rockwell HgCdTe 1024×1024 array with 18 µm pixels, bonded onto a silicon multiplexer (MUX). The inner 1014×1014 pixels are light-sensitive and are used for scientific imaging, while the bordering 5 pixels around the array edges provide a constant-voltage reference and are used for improved subtraction of the electronic bias level. This device is a direct descendant of the NICMOS 256×256 and Hawaii 1024×1024 arrays, widely used in space- and ground-based astronomy. A major effort has been made to eliminate the amplifier glow and bias jumps that have affected the NICMOS detectors.

To avoid the complexity and limited lifetime of a stored-cryogen system, while at the same time provide the low operating temperatures required for dark-current and thermal-background reduction, the WFC3 IR detector is refrigerated with a six-stage thermoelectric cooler to a nominal operating temperature of 145 K. The long-wavelength cutoff at 1700 nm (provided by tailoring the composition of the HgCdTe array) has been chosen to limit the "effective" dark current (i.e., the sum of the intrinsic detector dark current and the instrument's internal thermal background), given the restrictions imposed by the operating temperature. A longer cutoff wavelength would have increased the effective dark current at a given detector operating temperature, because of the thermal background from the telescope and optical bench.

The currently selected candidate flight detector (FPA 129) has a read-out noise of about 16 electrons rms per pixel after multiple reads. The sum of detector dark current and thermal background contributions is predicted to be lower than the requirement of 0.4 electrons per pixel per second.

The MTF of the IR detector is not expected to broaden the PSF by more than 10%. The intrinsic photometric accuracy is expected to be better than 5%.

4.3 Internal and OTA Thermal Background

Given the overall design of the IR channel, particular care has been put into minimizing the thermal background contribution to the detector effective dark current and into reducing the thermal load on the detector. This has been achieved by cooling the optical bench, by designing a cooled, baffled camera head for the IR detector, and by including a cold-stop mask in the design. This latter component greatly decreases both the thermal load and the effective background coming from the instrument itself and from the HST OTA, meeting the goal of 0.2 electrons/pixel/s in the "H" (F160W) band, at the cost of a modest throughput loss of ~10%. Because the contribution to the background by zodiacal light is ~0.2 electrons/pixel/s in the F160W band, the instrument will meet a goal of being near-background-limited in this passband. Similar performance is achieved in the F125W band. The IR channel filter wheel is also cooled to reduce its contribution to the background, due to the off-band emissivity of the filters.

4.4 Spectral Elements

The IR channel's FSM contains a single filter wheel with 18 slots, containing 15 passband filters, two grisms, and an opaque blank position. The filter complement, as recommended by the SOC with community input, samples the range from 900 to 1700 nm. It includes broad- and medium-band filters covering the entire range, as well as several filters centered on molecular bands and nearby continua. There are also several narrow-band filters probing interstellar and nebular diagnostic lines. Table 5 lists the names and characteristics of the IR spectral elements.

The IR channel's versions of the ground-based J and H filters are F125W and F160W, respectively. The F125W filter has a width somewhat wider than that of a typical J passband used in ground-based cameras. The F160W filter's bandpass has been blue-shifted relative to ground-based H in order to give a better fit to the quantum-efficiency curve of the IR detector. Specifically, the H bandpass has been narrowed to approximately 1400-1700 nm, in order to limit thermal background, and to have the filter transmission define the bandpass at the red end rather than the detector sensitivity cutoff. By contrast, the NICMOS H filter (NICMOS F160W) covers approximately 1400-1800 nm. This narrowing for WFC3 minimizes photometric errors due to spatial variations in the detector's sensitivity function.

The wide F140W filter covers the gap between the J and H bands that is inaccessible from the ground. F105W has a central wavelength similar to ground-based Y, but is considerably wider. The IR channel also includes a very wide filter, F110W, spanning the ground-based Y and J bands. This filter can be used for deep imaging, with a bandpass fairly similar to that of the corresponding broad-band filter (also called F110W) in NICMOS. .

The two grisms will provide the WFC3 IR channel with the ability to obtain slitless spectra. The "blue" G102 grism provides a dispersion of 2.5 nm/pix (or a 2-pixel resolving power of ~200) over the 900-1150 nm wavelength range. The "red" G141 grism has a dispersion of 4.9 nm/pix (resolving power of ~140) over the 1080-1700 nm range. These dispersions and resolving powers are still only predicted values at this writing, pending laboratory measurements.

In most cases, a grism observation will be accompanied by a direct image, for source identification and wavelength calibration. Recommended filters for this purpose are F098M for the G102 grism, and F140W for G141.

4.5 Operating Modes

The standard operating mode for the IR channel, "MULTIACCUM," begins with an array reset followed by one or more non-destructive readouts as the exposure charge accumulates. The number of readouts (up to 15 after the initial reset) can be requested by the observer, and all readouts are recorded and transmitted to the ground for analysis. The accumulated signal is converted to DN with a gain of 2.5 e^-/DN. The minimum exposure time is 4.3 s.

On-orbit experience with NICMOS shows that non-destructive readout is very effective, because it reduces the effective readout noise and corrects for most cosmic-ray hits. Many IR observations are expected to make use of sub-pixel dithering to improve the PSF sampling. Suitable dithering patterns will be included in the proposal software tools (APT), and can be requested by the observer. The IR channel also supports four subarray readout modes to allow for the short integration times required by bright targets (e.g., standard stars or bright solar-system targets), as well as reduction of data volume. The subarrays are always centered on the detector center, and the available sizes are 64×64, 128×128, 256×256, and 512×512 pixels.