A novel calibration technique for silicon semiconductor Rn detectors by using radioactive decay method

An innovative calibration method was developed for radon detectors by multipoint regression in this study investigated with two semiconductor radon detectors (the RAD-7 and MR107+), an external 226Ra standard source and a sealed stainless drum of radon chamber. Initial theoretical concentration of the standard source corrected by leakage rate was used to build up a decay curve over a 789-h period of measurement. Correlations between the data collected each hour from the detectors and those from the standard source were obtained by regression curves. The results indicated that binary quadratic equations well fitted the regression curves for the RAD-7 [coefficient of determination (R2) = 0.9995] in the range of 59–22856 Bq/m3 and MR107+ (R2 = 0.9949) in the range of 60–15663 Bq/m3. Regression analysis found a good linearity in 61.8–14700 Bq/m3 for the RAD-7 (R2 = 0.9996) and 62.3–9567.2 Bq/m3 for the MR107+ (R2 = 0.9933). Besides, the RAD-7 has wider linear range than the MR107+, while the MR107+ has higher sensitivity in this range. In low-concentration <400 Bq/m3, the linearity for RAD-7 in the range 59–398 Bq/m3 is still applicable (R2 = 0.9494). However, the correlation between the data from MR107+ and the standard in 60–398 Bq/m3 is better predicated by a binary quadratic regression. Further investigation found that the linearity for RAD-7 in 101–398 Bq/m3 changes little (R2 = 0.9434), while that in 59–100 Bq/m3 is medium. On the other hand, the linearity for MR107+ in concentration <100 Bq/m3 is good (R2 = 0.9494) and better than that for the RAD-7. The sensitivity of detectors was found dependent on radon concentration, revealing the imperative for calibration by multipoint concentrations. By this method, the linear range is easily discovered and the correlation between standard concentrations and measured values can be directly applied to calibrate detectors under different radon levels. This is more convenient than the traditional ones requiring calibration factors for different concentration and enhances the reliability for those using single standard concentration for calibrating samples with a wide range of concentration.