IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 50, NO. 6, DECEMBER 2003 2011 Single Particle Dark Current Spikes Induced in CCDs by High Energy Neutrons A. M. Chugg, Member, IEEE, R. Jones, M. J. Moutrie, J. R. Armstrong, D. B. S. King, and N. Moreau Abstract—This paper presents an analysis of dark signal nonuniformity induced in a charge coupled device (CCD) by 90 MeV neutrons. Random telegraph signal switching between multiple levels was seen for some dark current spikes. The probability distribution of the dark current spikes is shown to be pseudo-exponential and the distribution remains exponential during annealing, but with an increasing decay constant. Similar dark current spikes were also observed to be generated in an APS device exposed to high energy neutrons at the WNR facility. Index Terms—Annealing, APS, CCD, dark current spike, dark signal nonuniformity (DSNU), neutron NIEL, random telegraph signal (RTS), single particle displacement damage effect (SPDDE). I. INTRODUCTION P ROTON induced dark signal nonuniformity (DSNU) in CCDs has been extensively reported and analyzed in the past decade [1]. Of special note are the single pixel dark current spikes, which exhibit random telegraph signal (RTS) behaviors [2]. A variety of modeling approaches and analysis techniques have tended to conclude that these spikes may be attributed to the displacement damage associated with the nonionizing energy loss (NIEL) [3], [4]. However, it has been observed by some commentators (e.g., Robbins [3]), that the models connecting the NIEL and the spikes are best validated with reference to neutron damage to CCDs, since proton data is invariably contaminated by incidental ionization effects. Nevertheless, there seems to be a relative dearth of suitable neutron data in the literature, and this is especially true of the tens to hundreds of MeV energy range, which is most pertinent to space applications. This paper contributes to the rectification of this deficiency in presenting for the first time an analysis of high energy neutron induced damage in an E2v Technologies CCD02–06 device at a peak energy of 90 MeV and in an OmniVision OV502AB Active Pixel Sensor (APS) array irradiated by high energy neutrons at the WNR facility at Los Alamos, NM. Furthermore, it presents a new empirical model of the probability distribution of the dark current spikes, which incorporates the annealing of the distribution over a wide range of timescales post-irradiation. Manuscript received July 22, 2003; revised August 26, 2003. A. M. Chugg, R. Jones, and M. J. Moutrie are with the Radiation Ef- fects Group, MBDA U.K. Ltd., Filton, Bristol BS34 7QW, U.K. (e-mail: andrew.chugg@mbda.co.uk). J. R. Armstrong and D. B. S. King are with BAE Systems, Warton, Lancashire PR4 1AX, U.K. N. Moreau is with Staffordshire University, Beaconside, Stafford ST18 0AB, U.K. Digital Object Identifier 10.1109/TNS.2003.821813 Fig. 1. The TSL neutron spectrum. The modest neutron flux combined with the relatively low number of ionization events has made it possible to trace the instigation and the early-time evolution of dark current spikes during the irradiation for the first time. Interesting new early-time behaviors have been observed, including multilevel RTS which seems to be due to single damage complexes. II. EXPERIMENTAL DETAILS The CCD testing was performed on a CCD02–06 with 22 square pixels using the neutron beam at the Theodor Svedberg Laboratory (TSL), Uppsala, Sweden. The neutrons were pro- duced via the reaction, when a 92.1 MeV proton beam was incident upon a 4 mm thick lithium target. Remaining pro- tons are magnetically deflected out of the beam. The neutron spectrum of the resultant beam is shown in Fig. 1. There was a relatively narrow peak at a mean energyof 89.6 MeV, which contained around 39% of all the neutrons. The balance of 61% of the neutrons arrived in a broad low-energy tail on the dis- tribution [5]. (All the neutron fluxes and fluences subsequently quoted are specific to the 39% of neutrons in the high energy peak.) The charged particle contamination of the beam is about 1 in 100 000 particles [6]. This is insignificant, since we know from previous work with this CCD [7] that protons are no more potent than neutrons at producing dark current spikes. The testing was conducted at an ambient temperature of 20.7 over a period of 31 hours at a neutron flux (in the 90 MeV peak) in the range 1.4 – 2.1 . The flux was continuous except for two quarter hour breaks, such that a total fluence of 1.83 of neutrons was accumulated. Nearly 200 CCD frames were captured at 14-bit resolution at intervals during irradiation and control frames were taken prior to irradiation, during the breaks and immediately post-irradiation. Several further sets of control frames were captured at approximately 21 at late 0018-9499/03$17.00 © 2003 IEEE