2.3.1.5 Flickering Pixels

The feature known as `flickering pixels' was first reported in 1992, but not widely known until the flight of ASCA. With the benefit of hindsight, other flight programs were also able to report on their appearance in CCDs. However the effects of flickering pixels were most dramatic in ASCA; thus an examination of the phenomenon is important for ACIS.

The manifestation is that a number of pixels show unstable states of dark current signature: sometimes the mean signal is comparable with the background level of dark current and sometimes it is at a much greater level. In the former state they do not affect the science data adversely; in the latter they appear as isolated pixel signals resembling X-ray events. If the bright state were stable, the pixels affected could be masked from the data stream through the use of a look-up table, but the instability of the defect prevents this (unless a dynamic monitoring system can be implemented).

The physical cause is poorly understood, but Hopkins and Hopkinson (IEEE Trans Nucl Sci 40, p 1567 1993) provide the most comprehensive investigation to date. They propose that defect states can exist in different configurations, probably depending on whether the defect is charged or not. The dark current generation rate depends on which state the defect is in. It is hard to extrapolate their data to lower temperatures appropriate for ASCA and ACIS, but the general model is still probably valid.

A bulk displacement damage state may be created by a proton or ion. Depending on the precise nature of the defect type, the energy level within the bandgap, and the energy level's susceptibility to occurring in different charged states, may vary. As a result, over a period of time, any pixel may accumulate a number of dark current generation sites as the damage increases. In addition to an overall increase in background dark current, some pixels may be sites with one or more of the multi-level states. The dark charge generated in those pixels in one frame time therefore will depend on the particular configuration of states in that frame time, so that the accumulated signal may correspond to one of two (or more) generation rates.

There are a number of factors which suggest the problem may be less important for ACIS than ASCA:

1.

The operating temperature will be over 50 C lower than on ASCA, and the temperature dependence is steep. Assuming the excess dark current is due to only a handful of states per pixel, the expected reduction in charge generation rate will typically be $\sim 10^{-3}$ times the rate seen with ASCA.

2.

The time scale of changes between high and low states will be temperature dependent, so that for any class of defect, the defect will stay in one state longer (pixels will flicker less often).

3.

The total time between successive CCD readouts on ACIS will generally be shorter, so that the charge accumulated per frame is lower; also there is less chance for the defect states to change during a frame (a more stable dark signal condition arises).

4.

ACIS will be able to employ a bad pixel look-up table to veto bad pixels; also the increased data rate capabilities and event processing power will allow more local diagnostics of the data to be included for ground-based rejection of bad pixel data.

On the deficit side, it is not clear if the lower temperatures might cause a different set of states to be activated on the time scales critical to ACIS. In addition, the different orbit environment will cause a greater proportion of heavy ions to be the species creating displacement damage. It is known that these produce greater damage per particle than low-earth orbit protons, but not whether they will produce a greater proportion of the multi-energy generation states. It is expected that further investigation may reveal a more quantitative explanation of all these effects.