Shown above on the left is the complete history of measured CTI while the focal plane temperature was set to -110 C. CTI is plotted in units of 10-5. This is an average of the I-array CTI so the signal to noise is relatively high. Typical errors are around 2.3 x 10-7, so the scatter and structure seen are real.
We know that CTI is anticorrelated with the cosmic ray rate (cosmic rays work as sacrificial charge). Shown above on the right is the counting rate of events with pulseheights greater than 3750 ADU which are rejected by the flight software. Using S3, which is relatively unaffected by CTI, should remove any radiation damage induced structure in the amplitude rejection rate. The structure seen is real and should be proportional to the actual cosmic ray background rate seen by the CCDs.
Below on the left is shown the strong anticorrelation between the amplitude rejection rate on S3 and the average CTI of the I-array. The dotted line is the best linear fit with an intercept of (19.46 ± 0.04) x 10-5 and a slope of (-5.62 ± 0.06) x 10-7 CTI / (s3 counts/frame). The scatter around the linear fit is larger than would be expected from the errors on the points. The RMS deviation of the CTI data from the linear correlation is 9.2 x 10-7
The bottom right plot shows the CTI corrected by normalizing to the mean S3 reject rate. Much of the structure seen before is removed and no obvious increase in CTI is seen over this period. A linear fit to the corrected CTI values yields an intercept of (15.28 ± 0.02) x 10-5 and a slope of (1.17 ± 0.06) x 10-8 CTI increase per day. The RMS deviation of the data from the linear fit is 7.8 x 10-7.
Reference: Bev Lamarr, ACIS Memo #180 (PDF)
Above is shown a close up of the famous `day 286 event' in which the CTI appeared to jump up by a few x 10-6. The solid line is the mean CTI value during this period. After correcting the CTI for the changes in background rate, the jump seems to disappear. Below is a table showing some statistics of the period around day 286. The remaining scatter in the corrected CTI values is consistent with the scatter around the anticorrelation used in correcting the data except for one anomalous point.
| Without CTI Correction | With CTI Correction | ||||
|---|---|---|---|---|---|
| Interval | Mean CTI | RMS CTI | Mean CTI | RMS CTI | Typical CTI Error |
| (DOY 1999) | (x 10-5) | (x 10-5) | (x 10-5) | (x 10-5) | (x 10-5) |
| 262-310 | 15.40 | 0.13 | 15.65 | 0.10 | 0.02 |
| 262-285 | 15.29 | 0.06 | 15.67 | 0.11 | 0.02 |
| 286-310 | 15.53 | 0.05 | 15.62 | 0.06 | 0.02 |
| Change | +0.24 | -0.05 | |||
If there was any kind of small degradation of the ACIS CCDs associated with the day 286 event (or any other time for that matter) one would expect to see the correlation between CTI and the s3 background rate change since the effect of sacrificial charge is dependent on the density of traps. Plotted above is the correlation of I-array CTI with s3 high energy reject rate, using different symbols for data before and after day 286.4. Unfortunately there is no overlap in s3 rate between the two time periods. It is not clear whether any change occurred at all. Removing the two anomalous points ((67,15.3) and (81,15.3)), the correlation before and after day 286 is:
| Intercept | Slope | |
|---|---|---|
| (x 10-5) | (x 10-7/(s3 reject rate)) | |
| Before 284 | 18.61 ± 0.66 | -4.48 ± 0.90 |
| After 286 | 20.04 ± 0.07 | -6.48 ± 0.11 |