Understanding the evolution of radiation damage on the Gaia CCDs after 72 months at L2

Saad Ahmed; David J. Hall; Cian M. Crowley; Jesper M. Skottfelt; Ben Dryer; George Seabroke; José Hernández; Andrew D. Holland

doi:10.1117/1.JATIS.8.1.016003

1 February 2022 Understanding the evolution of radiation damage on the Gaia CCDs after 72 months at L2

Saad Ahmed, David J. Hall, Cian M. Crowley, Jesper M. Skottfelt, Ben Dryer, George Seabroke, José Hernández, Andrew D. Holland

Author Affiliations +

Journal of Astronomical Telescopes, Instruments, and Systems, Vol. 8, Issue 1, 016003 (February 2022). https://doi.org/10.1117/1.JATIS.8.1.016003

Abstract

The European Space Agency’s Gaia spacecraft has been operating in L2 ever since its launch in December 2013 with a payload that includes 106 scientific charge-coupled devices (CCDs). Due to the predicted radiation environment at the pre-flight testing stage in addition to the high level of accuracy demanded by the science objectives, the non-ionizing energy loss (NIEL) damage on the detectors was identified as a major factor that could affect the science goals of the mission. Here, we present the analysis of an extended set of charge calibration data, taken up to almost six years after launch. It is found that the rate of radiation damage accumulation by the CCDs has not differed significantly from previous results. While the parallel and serial CTI measure an increase in time, the trap defect landscape is still dominated by the pre-flight defects rather than the radiation-induced traps. CCD devices that were predicted to have a lower NIEL dose measure comparatively larger rates of CTI increase. In addition to this, thicker devices have been measured to have lower serial CTI values compared to thinner devices. The initial parallel CTI values have also been found to be dependent on manufacture year.

1. Introduction

The Gaia spacecraft is an European Space Agency (ESA) medium-class mission from the Horizon 2000 Plus program, successfully launched in December 2013. Over its mission, it will observe and take measurements of more than one thousand million stars with the goal of making the largest catalog of astronomical objects in the Milky Way. Rotating with a period of $\sim 6 h$ , Gaia scans the sky with its two optical telescopes, making repeated observations of stars, galaxies, quasars, and other objects that are autonomously detected onboard. In addition to astrometric parameters, Gaia data releases (DRs) contain (or will contain) photometric results, low-resolution spectra, colors, and astrophysical parameters of most stars in addition to radial velocities and medium-resolution spectra of a subset of objects.¹^,² As of the time of this publication, two complete DRs and an Early third DR have already been released based on the large quantities of data collected by the spacecraft²^,³ with the full third DR and future DRs to be produced in the future (for the latest updates on the DRs and the mission in general see Ref. 4). Gaia has a focal plane of 106 charge-coupled devices (CCDs), which continually make measurements of objects in its fields of view. Figure 1 shows a schematic of the focal plane where the CCDs are color-coded by their instrument. Due to the continually scanning nature of the mission, the CCDs operate in time delay and integration (TDI) mode, transferring charge at the same rate as the objects move across the detectors.²

Fig. 1

The Gaia focal plane with 106 large-format CCDs. The CCDs are split into seven rows and 17 columns, each with a set of readout electronics known as proximity electronic modules. The devices all run in synchronisation with the aid of an on-board computer. The CCDs are also split into 17 different strips. The CCDs in this figure are color-coded according to the functional group they belong to. The color designations are as follows: black for the wavefront sensors (WFS), purple for the BAM, maroon for the SM, turquoise for the AF, blue for the BP, red for the RP, and orange for the RVS. Figure reproduced from previous work.⁵

Gaia orbits Lagrange Point 2 (L2), the same destination of several other future space missions, such as ESA’s PLAnetary Transits and Oscillations of stars, Atmospheric Remote-sensing Infrared Exoplanet Large-survey and Euclid missions, NASA’s Roman Space Telescope as well as NASA and ESA’s James Webb Space Telescope. Due to the predicted radiation environment in this region at the time of Gaia’s pre-launch phase, as well as the high-precision measurement requirements of the mission, the increase of charge transfer inefficiency (CTI) in the CCDs from non-ionizing radiation damage was identified in the pre-launch studies as a major factor that would impact the useful scientific output of the mission.⁵ As a result, many tests and radiation campaigns were conducted during the pre-flight stage to quantify the impact of radiation damage, and several CTI mitigation strategies were devised.⁶^,⁷ However, a comparatively more recent analysis of in-flight data $\sim 30$ months after launch has revealed that Gaia is predicted to accumulate roughly an order of magnitude less damage than was initially expected during its nominal lifetime.⁸^,⁹ In fact, the overall spacecraft performance has been so good that Gaia’s mission time has been extended past its original five-year plan. Although this is good news for the science goals of the mission, it is important to understand, as much as possible, the reasons behind the differences between the predicted and measured results to better understand the physics and mechanisms for more accurate testing and predictions for other missions.

This paper analyzes an additional set of in-flight data, now taken up to six years after its launch date, to examine the on-board radiation damage (using specially acquired calibration datasets) and check whether there have been any major deviations from the behavior measured in 2016. A previous analysis only looked at the CTI between two different points in the mission. This paper is an extension of the previous work and analyses the data across multiple points in the mission and quantifies the evolution and increase of the CTI in time.⁵

2. Radiation Damage in CCDs

CCDs have a rich history of use in many space applications and missions in the past. CCDs consist of a metal-oxide-semiconductor structure where charge in the form of photo-electrons is generated in the semiconductor from incident photons. In space, CCDs aboard spacecraft are exposed to radiation, which consists of energetic particles, such as protons, electrons, and neutrons from galactic cosmic rays (GCRs) and solar events. These particles can cause displacement damage in the CCDs.

When displacement damage occurs, atoms are knocked out of the silicon lattice by the high energy particles. The exact spectrum and energy of particles varies with respect to orbit, location in space, and the behavior of the Sun. The vacancy, or absence of the silicon atom, diffuses through the silicon and can form a stable defect with either another vacancy or an impurity atoms, such as oxygen or phosphorus. These defects, or traps, have their own characteristic energy levels and the traps capture and emit photo-electrons from passing charge packets on probabilistic timescales. Depending on the charge transfer period, trails of charge can be formed behind the signal packets, which reduces the quality of the data and distorts the signal.¹⁰

3. Gaia CCD Details

Gaia’s CCDs, which have been custom-designed and manufactured by Teledyne e2v, have a number of special features to reduce the impact of radiation damage and charge trapping. The CCDs were designed with a supplementary buried channel (SBC), which is a notch in the buried channel region that can confine small charge packets to a smaller volume so that they encounter fewer traps. In addition to this, the CCDs have a charge injection (CI) gate structure where rows of charge can be periodically injected and transferred through the CCDs to keep traps filled so signals can pass by undistorted.¹¹ While all of Gaia’s CCDs were manufactured with these features and with identical chip architectures, they were also manufactured in three different variants, each optimized to improve their quantum efficiency in different wavelength regimes to fulfill different objectives.

3.1.

Device Variants and Year of Manufacture

About 78 of the focal plane CCDs were of the astrometric field (AF) variant device, which were built on $25 - μ m$ standard silicon, thinned to $16 μ m$ with a resistivity of $100 Ω cm$ , and have an anti-reflection (AR) coating centered at 650 nm. These devices were implemented on the AF and Sky Mapper (SM) instruments as well as the wave-front sensors (WFS). The seven CCDs of the Blue Photometer (BP) instrument used the BP variant CCDs, which were made with $16 μ m$ , $100 Ω cm$ resistivity standard silicon but had an AR-coating that was centered on 360 nm, making them more sensitive to bluer wavelengths. The third CCD variant was the red photometer (RP) variant that was used on instruments that required greater sensitivity to redder wavelengths. They were manufactured with a thickness of $40 μ m$ using deep-depleted silicon that had a resistivity of $1500 Ω cm$ and had an AR-coating centered at 750 nm. These variants were used in the seven CCDs of the RP Instrument, the 12 CCDs of the radial velocity spectrometer (RVS), and the two basic-angle monitor (BAM) CCDs.⁵^,⁸^,¹²

Many different devices were manufactured in the time before launch, to be used in performing radiation testing and sector analysis or to be implemented as in-flight devices. Due to the comparatively larger number of devices required on the focal plane as compared to other space missions, the in-flight devices were all manufactured at different times and came from different batches of silicon. In a previous study, it was found that the CCDs that were manufactured after 2004 were more likely to contain a reduced SBC full well capacity as compared to earlier devices.¹³ To check for any correlations between device performance in the results of this paper and the time of production, Fig. 2 was produced, which illustrates the corresponding year of manufacture of the in-flight devices.⁵

Fig. 2

The year of manufacture of all the in-flight devices on the Gaia focal plane; reproduced from previous work.⁵

3.2.

Pre-Flight Tests

As has been mentioned previously, a number of pre-flight tests were conducted on the Gaia CCDs to work out how to optimize the CCD performance as well as to investigate the effects of radiation damage. Surrey Satellite Technology (formerly Sira) and Airbus Defence and Space (formerly EADS Astrium) were the main industrial partners of the mission who performed these tests, the latter of whom performed five radiation campaigns to test device performance. While hardware optimization structures were implemented into the design, as outlined previously, operating techniques were also investigated, which would optimize the performance as much as possible. Some of the main conclusions derived from these campaigns were selecting a 163 K operating temperature to optimize CTI between the parallel and serial directions, the use of periodic CIs to reset the illumination history in the CCDs, and the rejection of artificial light to produce flat-fields for calibrations.⁶^,¹²

A number of computer models and simulations were also used in conjunction with these experimental tests. To measure the CCD irradiation levels to be used in the radiation campaigns, Airbus Defence and Space simulated a prediction for the 10-MeV equivalent NIEL dose of protons across the focal plane; Fig. 3 reproduces this predicted NIEL dose.¹⁴ This prediction was made in 2006, using the predicted level of solar activity at the time. This NIEL dose prediction only considered particles from solar events, neglecting the effects of GCRs, as well as only considered the primary emission of particles. In Fig. 3, it can be seen that the AF devices receive the largest fluence of 10-MeV equivalent protons in this prediction, with the devices in the center receiving the largest dose due to these devices being exposed the most to radiation. The BP, RP, and RVS instruments have optics and photometers setup in front of the devices which causes a reduced impact from the 10-MeV equivalent dose, as is noted in Fig. 3. This predicted pattern formed the basis for the irradiations used in all the pre-flight tests; because the AF CCDs were predicted to receive a dose of $3.1 \times 10^{9} p^{+} / {cm}^{2}$ 10-MeV equivalent protons, irradiations were performed at a slightly higher level of $4 \times 10^{9} p^{+} / {cm}^{2}$ 10-MeV equivalence to overestimate the damage.¹⁴^,¹⁵

Fig. 3

A graphical representation of the 10-MeV equivalent proton NIEL dose, measured in ${protons / cm}^{2}$ , predicted by Airbus Defence and Space a number of years before launch in 2006. The EOM NIEL prediction for the CCDs ended up being lower than this at the time of launch but irradiation levels for the on-ground tests were based on this set of predictions.

A semi-empirical model was also developed during the pre-flight testing called the charge distortion model (CDM);¹⁶ it was developed for forward modelling to reproduce the image distortion from the effects of radiation damage. CDM uses an electron confinement volume model where the signal level and the volume confined by the charge cloud are related by a power-law, defined by a $β$ parameter.¹⁶ Previous work has been successfully conducted to model observations with CTI to obtain the true image parameters from the damaged observations.⁶^,⁷

4. Charge Data Monitoring

Calibration activities are run periodically on-board Gaia, initially with a cadence of every 3 to 4 months, to monitor the performance of the devices. Unique strategies are adopted to make measurements of different detector characteristics. Previous analyses of these measurements have revealed that the detectors are functioning as expected with no major issues.⁸^,¹⁷

The CTI is measured regularly by examining the charge loss from CI lines. Previous analysis of both the in-flight parallel and serial [along-scan (AL) and across-scan (AC) respectively, using the Gaia terminology] CTI data in 2016 revealed that the radiation damage experienced by the in-flight CCDs has been well below the predicted end-of-mission (EOM) 10-MeV equivalent fluence values and the degradation in performance has been less than expected.⁸^,⁹ While this is beneficial for the science objectives of the mission, it is important to understand, as well as possible, why this discrepancy has occurred between the pre-flight tests and predictions and the in-flight data. This is to verify the reliability in the pre-flight testing as well as to obtain a better overall understanding of what is going on which can positively feedback into other missions and studies. The previous analysis of charge calibration data concluded that the lower than expected CTI levels are most likely the result of a lower solar activity level during solar cycle 24 as well as due to mitigation of charge loss from background straylight.⁹ There is potential work to be done to accurately quantify all the factors causing the differences between the in-flight and on-ground results and the results of this paper provide a starting point from which to go forward.

This paper analyzes more in-flight data, using data taken all the way up to the end of 2019, beyond the original EOM date, to see how the CTI level and radiation damage compares to the previous results. In previous studies, AL and AC CTI were measured from alternate CI schemes. Regular CIs are conducted periodically at fixed levels in the AF, BP, and RP devices; these are carried out every $\sim 2 s$ (corresponding to a 2000 TDI period) for the AF devices and every 5 s (5000 TDI) for the BP and RP devices. While the RVS devices were built with a CI structure, one that is used in some of the calibration activities, regular CIs are not implemented on these devices as the signal would interfere too much with the finer measurements and long spectral windows and the leading edge of the long spectrum would act to fill the traps.⁸

Compared to the CI period of 2000 TDIs in the AF CCDs and 5000 TDIs in the photometers, the CIs investigated in this study had a smaller period of 455 TDIs. This means that more traps would have been filled between the charge-injection generated virtual objects (VOs) and so the measured parallel CTI values will be lower as compared to the values measured with the regular CIs. The trends of the CTI behavior with time should still be clearly visible. In addition, the use of these datasets to measure parallel CTI means that parallel CTI can also be investigated in the RVS devices. As previous studies only looked at parallel CTI measurements from CIs, new insights are obtained about the behavior of the RVS devices. The CTI measurements can also be more easily compared between different devices because the CIs of the calibration activity all have similar charge levels and periods across all the devices as opposed to the regular CIs, which are not only non-uniform between CCDs, but, as mentioned previously, had different periods depending on the instrument.⁸^,¹⁵

The technique that is used to analyze the CTI in the serial direction involves generating charge in the image pixels and monitoring any trailing in the post-scan pixels after transferring through the serial register. Five different levels of CI are generated in this activity with VOs placed over the end of the image area. This calibration activity was initially proposed to be run monthly or bi-monthly but due to the lower than expected amount of damage, it is now run every three to four months; it is likely to be less frequent in the future.⁸^,¹⁵ Figure 4 details a set of serial CTI results with respective signal levels for a single CCD at all the different times the calibration activity was run. It can be seen that there is a small but finite increase of CTI with time. As the charge is generated at one end of the CCD and is transferred across the device to the serial register at the other end, it is possible to investigate the parallel CTI using this set of data as well. In this study, both the serial and parallel CTI were measured using just the calibration data; data were only available for science CCDs, except for the SM CCDs.

Fig. 4

The five CI values with corresponding CTI values for each calibration run for a device near the center of the focal plane. A slow but finite degradation in CTI is visible with respect to time. For the sake of completeness, data from January 2014 is produced here; the data from this time are not used in the rest of this study as the focal plane was known to be at a higher temperature.

An important complexity should be noted about the CTI results. The parallel transfer speed is at the TDI scanning speed of $982.8 μ s$ so the parallel CTI should be reflective of any trap species emitting around those timescales. In comparison, transfer in the serial direction takes place at two different speeds; there is a 10-MHz flush to transfer data to the readout node and a slower kHz speed for readout.⁸ This means the charge tail will be formed from trap emission at two different timescales. For these results, only emission at MHz timescales is considered as it dominates the serial CTI but it is noted that it would be useful to conduct a study in the future to look at the differences between different readout speeds.

To obtain the next set of measurements and compare the CTI of all the devices against each other, a number of processing steps were conducted. A power-law relationship was fit between the parallel and serial CTI values and their corresponding CI levels.⁵ The reason a power law was chosen was due to a similar model used in Gaia’s CDM ¹⁶ and the verification of such a relationship from other experimental studies.¹⁸ Once this relationship was extrapolated, a CTI value was calculated for the same signal level (10,000 electrons was used for all the results) for all the devices. By using the same virtual signal, the CTI between all the devices can be more easily compared against each other.⁵

5. Serial CTI (AC Direction)

5.1.

2014 and 2019 Data

Using the data analysis steps outlined in the previous section, the diagrams in Fig. 5 were generated, which highlights the serial CTI of all the CCDs at two different times, initially computed and analyzed in a previous work.⁵ Figure 5(a) shows the CTI in April 2014, just a few months after launch, while Fig. 5(b) shows the CTI in December 2019, more than five years later. The mean CTI for each instrument was also calculated at the two times and these values are given in Table 1. The calibration data from January 2014 was not used as the cool-down of the payload was still not complete. Figure 5(a) is similar to the plot generated in the previous set of results in 2016, verifying the reliability of these results. It is assumed that the oxygen defect, called the A-center, is likely to be responsible for the bulk of the serial CTI as it has a comparable release time with the 10-MHz flush that dominates the serial transfer. A previous pre-flight test revealed a trap species that emitted faster than the A-center so there is a chance for the AC CTI to be caused from this trap species as well.⁸^,¹⁹

Fig. 5

The distribution of the derived serial CTI values across the focal plane for a signal of 10,000 electrons as measured in (a) April 2014 and (b) December 2019; reproduced from previous work.⁵

Table 1

Mean serial CTI (at 10,000 electrons) for each instrument as measured in April 2014 and December 2019.

CCD instruments	2014 mean CTI (×10−5)	2019 mean CTI (×10−5)
AF	$3.5 \pm 0.3$	$3.8 \pm 0.3$
BP	$3.6 \pm 0.2$	$3.8 \pm 0.2$
RP	$1.5 \pm 0.1$	$1.9 \pm 0.1$
RVS	$1.5 \pm 0.1$	$1.9 \pm 0.1$

The 2014 results in Fig. 5(a) and Table 1 should be reflective of the initial CTI state of the CCDs as the accumulated radiation damage within the first few months is known to be minimal. The results show that the thicker RP devices, found in both the RVS and the RP, all measure a lower serial CTI as compared to the other devices. Even though thicker devices are known to have different properties to thinner devices, such as being more sensitive to cosmic rays and having larger amounts of dark current, it is currently unknown why these devices would be manufactured measuring a lower serial CTI compared to thinner devices with the same architecture.²⁰ If the serial CTI is indeed primarily caused by A-centers, then this could be due to a smaller amount of oxygen and A-center defects being present in the buried channel region. This could have been the result of impurity oxygen having a larger amount of silicon to diffuse through during manufacture. It would be useful to perform trap-pumping on unirradiated thick and thin devices to understand the trap landscape and narrow down the reasons behind this behavior. Trap-pumping is an analysis technique that makes it possible to observe individual trap species and their properties with a high level of accuracy and quantify their densities. It involves the clocking of a flat-field signal between phases which leads to the formation of dipole maps that can be studied in more detail.²¹^–²³

It can be seen that the pattern of serial CTI distribution across the focal plane is the same in 2014 and 2019. This indicates that the serial CTI is still dominated by the initial CTI state and the amount of radiation damage has not been significant enough to alter this distribution pattern. The 2019 CTI values in Table 1 are within the same order of magnitude as the 2014 CTI values; this demonstrates a very low rate of increase. The AF and BP CCDs measure a smaller percentage increase in their serial CTI ( $\sim 6 %$ to 9%) as compared to the red variant devices ( $\sim 27 %$ ).⁵ With Gaia having been in orbit for just over its original mission lifetime, no major deviations are seen from the previous set of CTI results, as measured in 2016.

5.2.

Rate of Increase

Using the serial CTI values for each CCD at the time of each calibration activity, a linear relationship was fitted between serial CTI and time. The slopes derived from these relationships were taken as the rate of increase of CTI with time and were computed for each CCD; these values are shown for each CCD in Fig. 6. If outliers are ignored, the column-wise CTI distribution would seem loosely similar to the NIEL-dose prediction from Fig. 3, however, the pattern is not very clearly defined so a correlation cannot be established.

Fig. 6

The extrapolated rate of increase of serial CTI (at 10,000 electrons) for each CCD during the six-year period of the charge calibration.

Combining the data from the CCDs for each instrument, the values in Table 2 are produced which gives the mean rate of increase for each instrument, in units of both Gaia revolutions and in years. Figure 7 shows the mean CTI for each instrument and how it increases with time; the values were calculated for each calibration activity run. Three major sets of solar flares are also marked out; the events in 2014 and 2015 had obvious effects in the previous set of data analysis of in-flight parallel CTI.⁹ Interestingly, these events do not seem to have a significant impact on the serial CTI. In contrast, the solar flare events in 2017, which were more larger and powerful, make a slightly more noticeable impact on this set of charge calibration data. This could be due to the relatively high level of CTI from the beginning of the mission which means the relative contribution from a solar event is much less evident on serial CTI. There is a loosely, linear increase of CTI in time, justifying the use of a linear extrapolation for the results of Fig. 6 and Table 2. The fluctuations and uncertainties seen in the data are likely due to photon noise from straylight and noise from bias and background subtraction during data processing.

Table 2

Mean increase in serial CTI per time for each instrument on the focal plane for a signal of 10,000 electrons.

CCD instruments	CTI increase/revs (×10−10)	CTI increase/year (×10−7)
AF	$5.5 \pm 0.9$	$8.0 \pm 1.0$
BP	$4.0 \pm 1.0$	$6.0 \pm 1.0$
RP	$6.0 \pm 0.3$	$8.8 \pm 0.4$
RVS	$5.7 \pm 0.8$	$8.0 \pm 1.0$

Fig. 7

Evolution of the mean serial CTI for each instrument, calculated from the CTI measured from all the devices at each different time. There is a steady, linear increase in time with fluctuations likely due to noise. Three main sets of solar flare events are all marked out. The $x$ axis spans a little under six years.

From the results of Figs. 6, 7, and Table 2, it can be seen that the RP devices in the RP and RVS measure comparatively higher rates of increases as compared to the other devices. It is known from Fig. 3 that the Blue and RPs receive an equivalent amount of fluence, likely due to similar levels of shielding. More tests will need to be conducted to more precisely determine the physical reasons behind the calculated higher increase. Current hypotheses include the initial CTI state or the geometry of the devices affecting the rate of radiation damage as well as interactions between the shielding and the radiation altering the energy spectrum of incoming particles.

5.3.

NIEL-Scaled Rate of Increase

It is useful to track the evolution of CTI of each individual device to work out the correct data processing procedure for each CCD. However, by scaling the rate of increase results by the NIEL values from Fig. 3, it becomes possible to remove the variation in dose and shielding effects and examine the CTI increase between each CCD variant against each other more reliably. For example, in Fig. 6 and Table 2, it can be seen that there is a large variation in the CTI increase of the various AF devices due to the NIEL dose. Figure 8 reproduces Fig. 6 but with the rate of increase values of each CCD scaled by their corresponding NIEL dose value from Fig. 3. Table 3 also shows the new rate of CTI increase values for each different instrument after the CCD values are scaled.

Fig. 8

The extrapolated rate of increase of serial CTI (at 10,000 electrons) for each CCD during the six-year period of the charge calibration data with each CCD scaled by its corresponding NIEL dose value from Fig. 3.

Table 3

NIEL-scaled mean increase of serial CTI per time for each instrument on the focal plane for a signal of 10,000 electrons.

CCD instruments	CTI increase (NIEL scaled)/revs (×10−10)	CTI increase (NIEL scaled)/Year (×10−7)
AF	$1.8 \pm 0.3$	$2.6 \pm 0.4$
BP	$2.6 \pm 0.5$	$3.8 \pm 0.8$
RP	$3.8 \pm 0.3$	$5.6 \pm 0.4$
RVS	$3.1 \pm 0.5$	$4.5 \pm 0.7$

In Fig. 8, the AF devices now all highlight rates of CTI increase that are more comparable to each other as the shielding pattern from Fig. 3 is removed. This would indicate that all the AF CCDs experience radiation induced CTI increase on equivalent levels to each other. An interesting observation from Fig. 8 and Table 3 is that even after accounting for the NIEL dose, the RP devices still measure larger rates of CTI increase which are not measured by any of the other device variants. This gives more credence to the idea that it is a unique property of the thick RP devices that is causing this larger increase in CTI. It remains to be seen whether this feature is due to the lower initial CTI, an additional unknown effect of shielding not accounted for in the NIEL dose calculation, the manufacture of the devices, or due to device geometry.

It is noted from Fig. 8 that some of the BP devices seem to have rates of CTI increase that are more comparable to the RP devices than with the AF and BP devices. This would suggest that despite the reduced impact from the 10-MeV equivalent NIEL-dose, the shielding from the optics and photometers could somehow be affecting the radiation damage in a counterintuitively more destructive manner. This could be due to radiation interfering with the optics and photometers in some way, either causing secondary emissions or slowing down higher energy particles that may not have caused damage otherwise. In previous on-ground studies, it has been reported that aluminium shields in front of CCDs have caused higher energy particles to lose some of their kinetic energy so a similar phenomenon could be occurring here.¹⁹ Given only some BP devices measure this relatively higher increase though, this effect appears to be minimal, at least for the serial CTI.

From Tables 2 and 3, it can be seen that the rate of increase of CTI per year is around the order of magnitude of $10^{- 7}$ . This is at least two orders of magnitude smaller than the CTI values at the beginning of the mission, as noted in Table 1. This illustrates that the degradation of the CTI from the current status is very low.

6. Parallel Charge Transfer Inefficiency (Along-Scan Direction)