MIRI Detector Non-Ideal Behavior
A brief description about 3 sentences.
Don't forget! Doesn't work at low light levels. Or subarrays (most extreme). ???
Last Frame Correction
No frame left behind. Initial algorithm seems to work.
Reference Pixel Correction
Put it into MRS simulator. Ressler doesn't believe he sees it in the real data.
Stacey and Maca's SPIE papers. MOR RESOURCES.
First frame affect goes away with Background subtraction!
Where does the 2% slope difference come from?
After jumping from Full Fast/Full Slow, there is a relation.
5 minutes. ???
Short integrations. Background and dithers are your friends.
Background Anomaly Feature
With the MIRI arrays, the first few samples starting an integration after a reset do not fall on the expected linear accumulation of signal. This behavior is seen in virtually all types of infrared arrays, both of Si:X IBC type (e.g., Gordon et al. 2004; Hora et al. 2004), and more generally (e.g., Rauscher et al. 2007; Rieke 2007). The reset anomaly can be removed largely (or perhaps entirely) by subtracting from each ramp under signal, a correction generated from the behavior of that ramp in the dark. The approach to doing this has to be robust to changing “dark” slopes due to slow detector settling (e.g., after a change in background level or following events such as a thermal anneal or just turning the detector on). It has been found that subtracting the median slope of a series of dark integration ramps before subtracting from a data ramp can provide a good correction that is largely immune to these effects.
The Rockwell/Boeing North America Si:X IBC arrays used in Spitzer IRS and MIPS had outputs for all the individual pixels that included a component proportional to the total signal received by the array (with a proportionality constant of 0.3-0.4). This phenomenon was termed “droop” (van Cleve et al. 1995). Similar behavior was exhibited by the WISE Si:As IBC arrays, but the effect is greatly reduced in the IRAC and MIRI arrays. A plausible explanation is that the latter two arrays use the Raytheon cryogenic readout manufacturing process that heavily dopes the multiplexer substrate to within a few microns of the circuitry to improve the performance of the ground plane in limiting long-term drifts and other undesirable types of behavior. It appears that it may not be necessary to make corrections for droop in the MIRI pipeline.
The array is reset sequentially by row pairs. The last frame of an integration ramp on a given pixel is influenced by signal coupled through the reset of the adjacent row pair. The result is that the odd and even rows both show anomalous off- sets in the last read on an integration ramp.
The MIRI arrays are subject to slow output drifts with amplitudes roughly proportional to the signal level (e.g., the drifts are greatly reduced when the arrays are in the dark). A simple description of this behavior, along with a prescription for removing it, is that an observing strategy that includes dithers so that images can be subtracted while retaining the source signals appears to be able to remove the effects of the drifts, so long as the dithers are performed at least every 5 minutes. There are some other aspects not captured in this simple picture; e.g., the amplifier zero points also drift in a way that can affect the linearity corrections. This behavior is similar to that for the Spitzer Si:X IBC arrays, both from Raytheon and Boeing. For example, it was necessary with IRAC to dither every 7 minutes to generate optimum self-calibrated flat fields, while the IRS allowed integrations only up to 8 minutes before moving the source on the slit (from the IRAC and IRS instrument handbooks, Spitzer Science Center, 2011, 2013).
Although dithering is an acceptable strategy for many MIRI operational modes, for coronagraphy, planetary transit observa- tions, and some others it is not a viable option. The MIRI array reference pixels do not follow the drifts accurately enough to remove them noiselessly. The decay of latent sources is likely one component of the cause of the drifts. Therefore, learning how to deal with latent images as well as other causes of the drifts is an open issue in the development of the MIRI pipeline.
The MIRI multiplexer can contribute significant levels of glow to the low-level signals from, e.g., the MIRI spectrometers. A primary source of this emission is both the column and row shift registers, where it is associated with the forward biasing of the n-FET substrates of the shift register MOSFETs. Adjustment of the bias potentials of the p-wells in which these MOSFETs reside can minimize their glow. Further reduction can be achieved by control of the potentials for the rails of each of the shift register clocks. The upgraded electronics boards al- low greater ability to optimize these settings than with the original MIRI FPE.
Low levels of trapped charge could result from impurities in the blocking layer, which might cause weak latent images at all levels of illumination. Another possibility is dielectric relaxation—the slow adjustment of the electrical equilibrium of some portion of the detector, basically because of its long RC time constant. M. G. Stapelbroek (private communication) has cal- culated that the time constant for this process in the undepleted region of the detectors is roughly appropriate to lead to the more rapidly decaying latent images. Smith et al. (2008) propose that latent images in photodiodes can arise because the depleted re- gion at the junction shrinks as the detector is debiased due to accumulation of signal, allowing trapping of photoelectrons. Once the bias is re-established, the traps lie within the depletion region; as the trapped electrons are slowly released, they appear as a signal. Although this argument cannot be applied directly to Si:As IBC detectors, a similar situation may exist if sufficient signal is gathered to produce a significant undepleted region (from the above section, a signal of order 200,000 electrons would suffice). A number of these processes could operate to- gether. Thus, why the latent image behavior might be complex is not difficult to understand, but a systematic set of measure- ments as a function of detector temperature, photon flux, and perhaps, wavelength is needed to obtain sufficient information to constrain the possibilities.
Bright sources leave latents on the MIRI arrays, typically at a level of about 1% immediately after the source has been removed. The decay of these images shows multiple time constants, suggesting that there are a number of mechanisms that contribute to the effect (Paper VII). Further characterization of this complex behavior is needed to determine ways to correct it in the MIRI pipeline.
Hollowed out PACS detectors.