A total of 7 silicon diodes were examined systematically in view of their radiological properties. The diodes varied in their cross sectional area, the thickness of their depletion layer (50 to 300 um) and their conductivity. The response of the diodes was investigated without and with a series of filters consisting of the materials aluminium, copper, tin, lead and a rare earth mixture. A pulse height spectrum was recorded for any given combination of radiation quality, diode type and filter. About 600 such spectra have been recorded. The pulse height spectra were analysed in different ways in order to develop an algorithm which, employed in an individual monitor, will provide the best possible dosimeter performance.
The simplest approach was to calculate the integral over the pulse height spectra as a function of the position of a lower level discriminator. By positioning this discriminator to a level corresponding to an energy of about 80 keV an acceptably flat energy dependence of response was obtained but at the obvious price that there was no response to radiation with energies below that threshold.
Another method of analysis was termed few channel analyser. For this algorithm the pulse height spectra were subdivided into up to 10 subregions. The sum of events in each subregion was multiplied with a reactor chosen in such a way that a linear combination resulted in a response which was sufficiently independent of energy. With this method quite acceptable results in view of an energy independent response were obtained. There were, however, certain disadvantages in a dose rate dependence. For higher dose rates the combination of diode and electronics produced a certain pile up, ie 2 pulses occurring in a time inverval shorter than the integration time are interpreted as one single event. A set of vectors can be determined, however, each set being used in a certain dose rate range. Alternatively the weighting factors could be a function of the dose rate. Another problem is that, depending on the number of channels, a negative or near zero response could result from some of the weighting factors having a negative sign or low positive value and the pulses fall predominantly into one of the bins. It must therefore be demonstrated for all possible spectral distributions of the radiation that the small or negative contributions are always compensated by larger positive contributions form the other bins.
The general proof that the system behaves satisfactorily for all spectral distributions of the radiation can be given by means of a more theoretical treatment only. As the number of realizable spectral distributions in fairly limited the system must be simulated and optimized by Monte Carlo techniques. A concept has been developed on how the optimization can be carried out allowing a reliable estimate of the worst case error.
CALIBRATION OF THE DOSIMETER WILL BE DEALT WITH PREDOMINANTLY IN THE SECOND HALF OF THE PROJECT. THE WIDE RANGE OF CALIBRATION FACILITIES OF THE PTB FOR PHOTON- AND NEUTRON FIELDS WILL ALLOW A COMPREHENSIVE ASSESSMENT OF THE DOSIMETER.
THE EXTENSION OF THE ENERGY RANGE FOR PHOTONS DOWN TO ENERGIES OF ABOUT 20 KEV IS IMPORTANT IN VIEW OF THE MANY WORKING PLACES IN RADIODIAGNOSTIC AND IN FUEL REPROCESSING PLANTS WHERE SUCH LOW ENERGY PHOTON RADIATION CONTRIBUTES SIGNIFICANTLY TO THE DOSE OF THE INDIVIDUAL. IT IS PLANNED TO USE THE TWO DETECTORS CONSISTING OF TWO DIFFERENT MATERIALS TO OBTAIN INFORMATION ON THE PHOTON SPECTRUM. THIS INFORMATION CAN BE USED TO DETERMINE ENERGY DEPENDENT CORRECTION FACTORS BY MEANS OF WHICH A CORRECT INDICATION CAN BE OBTAINED OVER A LARGE ENERGY RANGE. AS ONE OF THE DETECTORS CONSISTS OF SI A LOWER THRESHOLD WELL BELOW 50 KEV CAN BE OBTAINED. THE METHOD FOR OPTIMISING THE RESPONSE OF THE TWO DETECTORS WILL BE A COMBINATION OF MONTE CARLO SIMULATION AND EXPERIMENTAL VERIFICATIONS. THE PARAMETERS TO BE VARIED ARE THE KIND AND AMOUNT OF THE MATERIALS AROUND THE DETECTORS.