1 Investigation into properties of neutron and electron irradiated CCD By Nick Sinev University of Oregon Jim Brau, Jan Strube, Olya Igonkina, Nick Sinev LCWS 06, Bangalore, India, March 9-13, 2006
2 Topics Measurement technique Puzzle to study Hypothesizes Problem Explanation What we could measure Conclusions
3 Measurement Technique ( it was used to study details of radiation damage caused by neutron irradiation – reported on IEEE NSS in 1999) Measurement of the number of charge traps in every pixel in CCD : Inject small equal amount of charge in each pixel Read out charge signal from the CCD Compare output from different pixels Accuracy The amount of charge traps in each pixel is expected to be small – may be just one or few traps, each trap can capture only 1 electron To achieve charge measurement accuracy of the order of 1 electron repeat the process tens of thousands times and average readings. What can be measured The distribution of number of traps per pixel – depends on the spectrum of momentum transfer from irradiating particles to silicon atoms The charge retention time by traps and it’s temperature dependence – this depends on traps energy level and so defined by traps nature The temperature and speed of damage annealing – also defined by traps nature. Mechanism of Charge Transfer Inefficiency (CTI) – timing of charge capture by traps Space distribution of damaged pixels – allows to estimate space distribution of particles, causing damage What cannot be measured Traps evenly distributed over all pixels
4 Data processing details Signal in each pixel is compared with average of 24 neighboring pixels If signal in given pixel is smaller than average and difference significantly larger measurement noise presence of charge traps is assumed. List of all pixels with traps is made, containing pixel position and number of traps (derived from signal deficiency). This list is used to select pixels in another measurements for comparison of number of traps in the same pixels ubder different conditions. Number of pixels with traps constitute very small fraction of all pixels. That means that each damaged pixel contains in most cases just one cluster of traps.
5 Puzzling effect Time required to capture electron from charge packet by charge traps was believed to be in nanoseconds range. However, from the picture at right we see, that milliseconds are required to fill all traps in charge traps cluster, created by neutron irradiation. Time of charge packet and charge trap cluster contact (in milliseconds)
6 Hypothesizes to explain puzzle Potential pockets between 2 I clock electrodes could lead to areas with very low charge density – Chris Damerrel, Konstantin Stefanov Distribution of the charge in the charge packet has very fuzzy borders with charge density varying by 5-6 orders of magnitude from center of packet.
7 Problem We decided to check these hypothesizes by doing new set of measurements on the same neutron irradiated CCD used in previous experiments (in year 1999 and 2003). However, contrary to the fact, that between year 1999 and 2003 amount of charge traps in the clusters created by neutron irradiation did not change, they almost all disappeared by year 2005
8 Problem - annealing Ratio of number of charge traps per damaged pixel observed at different times to the same number observed after completion of series of neutron irradiations in The CCD was additionally irradiated with /cm 2 of 60 MeV electrons at NLCTA
9 Explanation Charge traps which we observe are created when vacancy in crystalline structure of silicon (which is mobile by itself) meets another vacancy, or doping atom (P) or impurity atom (O). Such complex (V2 or VP or VO) is immobile and has ability to capture and retain for long time electron from passing by charge packet. These complexes are stable at room temperature, and annealing occurs only at high temperatures ( C). However, if there are large amount of interstitial Si atoms, wandering in the bulk of silicon, they can fill vacancies, destroying above complexes. After neutron irradiation amount of interstitial silicon atoms was not large enough to cause observable annealing. Irradiation with electrons has created large number of such atoms. The fact that similar to neutrons in the observable damage effect caused by electrons has created much larger number of interstitial Si may indicate that most of observable charge traps are, in fact, V2 complexes. In that case probability of generating such complex is proportional to squire of vacancy density, and because of high initial local density of vacancies created by mechanism of energy loss by neutrons, neutrons can create larger number of V2 complexes with the same overall number of V and interstitials.
10 What we could measure Because we are left with small amount of observable electron traps, it is difficult to do any quantitative measurements, as it requires large statistics. However, first of the hypothesizes about cause of slow trapping gives as qualitative criteria for verification – it predicts that effect should disappear if we keep only 1 phase of I clocks high during charge collection.
11 Experiment with I clocks On this picture you see distribution for fraction of filled traps (TFF) for 3 ms and 0.3 ms contact time between charge packet and traps cluster when 2 I clock phases are kept high during it On this picture you see distribution for fraction of filled traps (TFF) for 3 ms and 0.3 ms contact time between charge packet and traps cluster when only 1 I clock phase is kept high during it
12 What it shows It shows, that in the case of only 1 I-clock phase kept at high potential during charge collection, the effect of slow trapping did not disappear. Rather it is more pronounced in that case (trapping even slower), which can be explained by larger fraction of low charge density area.
13 Conclusions Effect of “induced annealing” of number of charge traps in the charge trap clusters created by neutrons after electron irradiation is observed. The fact that similar in the number of created charged traps dose of electron irradiation has created much larger number of interstitial silicon atoms may indicate that nature of such traps is V2 complex. Hypothesis about “potential pockets” as a reason for slow capture of electrons by traps is not working. Hypothesis about fuzzy distribution of electrons in charge packet does not contradict experiment