Evaluation of the calibration factor uncertainty of an automatic TLD reader

The reading of a calibration set is a prerequisite for the reading of T.L. detectors used in dosimetry service. The detectors, used for extremity dosimetry (two for each dosemeter) are LiF:Mg, Ti (TLD-100) chips. For each batch of detectors ( ∼250), 30 are used for the calibration set. They are divided in six equal groups: five of them are irradiated with a Cs137 source, in the 1 mGy to 12 mGy range of air kerma; the sixth is left unirradiated, to provide for the correction for the natural and reader backgrounds. The dosemeters are read on a Harshaw Model 5500 automatic reader. This reader features non-contact hot Nitrogen gas heating with a programmable, linearly ramped time-temperature profile. Glow curves are recorded using 300 degrees Celsius maximum temperature and a heating rate of 14 degrees Celsius s-1. The detectors are individually coded and characterized by the sensitivity factor, evaluated before the first use and then at the conclusion of every six readouts. After each readout, the batch is annealed with the standard procedure: 1 h at 400 degrees Celsius and 2 h at 100 degrees Celsius. Before the detector reading, the reader has to be switched on, together with the Nitrogen flux, for at least 30 min. Detectors are inserted in the holes of a 50 holes disc and the disc is inserted in the reader. Before detector reading, a readout of the instrument noise and a readout of the reference light are made (10 sec each) and recorded. These readings are repeated after 20 dosemeters readouts and after the 50. readout. If the value of either noise or reference light are outside a fixed limit the reader stops automatically. The limits are set so to avoid frequent reading stops, but they are not to be too far apart to give a negligible contribution to the uncertainty. The value of the P.M. high voltage is automatically corrected, if necessary, after each reference light readout. The reader stability in eight-hours cycle has been checked, through readings of calibration sets, made up by detectors from the same batch. Two sets (A and C) were prepared irradiating five groups of five detectors and two (B and D) irradiating ten groups of five detectors. The sets were read in the A, B, C, D sequence. A group of dummy detectors was read between the calibration sets to simulate the continuos operation of the reader. The irradiations of the detectors were performed using a Cs137 source at the Calibration Centre of the Department, with 1.5% irradiation value uncertainty. The irradiation values for the four sets are reported in Table 1.The calibration factors (Fc) was determined for each set as the reciprocal of the slope of the line obtained from the best fit. Fig. 1 shows the best fit for set A. Each dot is the average detector readout, sensitivity factor corrected, minus the average of readouts, sensitivity factor corrected, of the unirradiated detectors. The sensitivity factor Si of each detector is defined as: Si 1/nΣ1nLi/Li The calibration factors of the four sets are listed in Table 2 together with the uncertainty.The results confirm the reader stability during a daily readout cycle. The final line in Table 2 shows the calibration factor of the set obtained as a result of readouts of all detector in the batch irradiated at the same irradiation value. Table 3 lists some values of the calibration factors for the same batch, by readout date, between the first and the last readout. The value of the calibration factor appears now nearly 25% higher, but the percentage uncertainty is about the same. This may be explained trough a change in the P.M.. The readout of the reference light was 173 nC in 10 sec for the first set (5-00) and 127 nC in 10 sec for the last one (40-05), with a variation ∼ 30%. The same trend has been found for other batches of detectors used in same years.The uncertainty of the calibration factor is evaluated taking into account the 1.5% uncertainty of the conventional true irradiation values and the readout uncertainty. The readout uncertainty is evaluated as the uncertainty of the sensitivity factor: u(Si) = u(L-bar) / Li = √1/n-1Σ1n(Li-L-bar)2/Li assuming that this uncertainty includes the contribution by the reader noise. The variation of the reader noise during the daily readout cycle amounts to a few mGy: it is rated as negligible, if compared with the irradiation values used for the calibration set. The effectiveness of the assumption was tested applying Fc of the A set to the readout of detectors used for B, C, and D set. The results for the B set are reported, as an example, in Table 4: nearly all results overestimate the corresponding conventional true values, while the uncertainty ranges from 13 to 16%, irrespective of the irradiation values.This result is satisfactory. Moreover, two detectors are inserted in the extremity dosemeter supplied by the service, so lowering the uncertainty of the evaluated value. (authors)