10  MeV  protons/cm 2   at  the   Tandem   van   de Graaff

accelerator at AEA Technology, Harwell, UK。 One device also received proton irradiations at 22。5°, 45° and 67。5° angles of incidence。 Also, two small area devices were irradiated with 100 McV protons by ESA at the SATURNE accelerator in France。 Apart from one device all the proton irradiations were carried out unbiased。 All irradiations  were at ro‹›m temperature and post-irradiation measurements were performed, aftcr periods of unbiased room temperature storage, to investigate annealing behaviour。

Dosimetry was carried out by staff at the radiation facilities and is believed to be accurate to better than 5% The data gathering system used an AT-compatible computer with an Imaging Technology VS 100 framegrabber board (1012x1012x12 bit, variable scan up to 10 MHz pixel rate)。 Also used was a purpose built programmable CCD timing generator employing fast access RAM and a specially developed software command interpreter with facilities for building automated test  sequences。   The CCD camera   head

IX11t-9499f92$03。0O  iO  1992 IEEE

had computer controlled bias voltages and a Peltier cooler which controlled the CCD temperature over the range 6°C to 40°C with an accuracy of better than +0。05°C。 The CCDs were operated with a line move rate of 2ps/line and a pixel ieadout rate of lps per pixel (though this could be increased during periods of fast dumping of unwanted charge)。 The clv:›ck waveforms had exponential edges and overlap times were approximately 40ns for the horizontal (serial) clocks and 200ns for the vertical (parallel) clocks。

Measurements of the dark voltage signal were converted to dark current densities using the charge to voltage conversion factor (CVF) of the output amplifier and knowing the integration time and pixel area。 The CVF value was determined for each dose step from measurements of output voltage and current in the reset drain bias line (corrections were also made for the change in CVF with  substrate voltage)。 For frame transfer operation the first line readout contains only dark current generated in the image region。 Later lines have progressively more dark charge resulting from the increased time spent in the storage region。 The difference between dark signals for the first and last line gives the storage region component。 CTE measurements were made

using 22 keV soft x-rays from a Cd 109 radioactive sourGe aRd

looking at the signals in single pixel events (as described later in the text)。 The particular CCDs studied here did not  have an input structure for injecting charge and so periodic pulse measurements of CTE [5] were not easily applicable。

III。RESULTS  FOR DISPLACEMENT  DAMAGE

A。 Changes in Charge Transfer Efficiency

During readout charge is transferred in parallel from pixel to pixel down each column and then serially along the output register to an output amplifier。 In this process, radiation- induced traps within the buried channel can trap charge and release it sometime later into a following charge packet - thus giving a loss in charge transfer efficiency (CTE)。

One method of measuring CTE is to examine the point- like signals from x-ray events as a function of their position in the CCD image (which determines how many pixel transfers have occurred prior to readout)。 Figure 1 shows the charge loss from single pixel x-ray events versus the number of parallel transfers for a CCD irradiated with 2 krad of 10 MeV protons。 CTE is found from the slope of the line。 Previous measurements using this technique on devices near room temperature [6, 7] did not have enough accuracy to look at CTE changes in detail (only upper limits on CTE damage could be given)。 In this study three methods have been used to reduce the CCD dark current so that CTE effects were not masked by dark current nonuniformities and the accuracy of signal digitization。 These were firstly to operate the CCD in inversion so that surface dark current is suppressed (cf。 section IV), secondly to slightly reduce the temperature (to 15°C) and thirdly to use a large area device but to reduce the integration  time  (and  hence  dark  charge build-up)  by only