Calibration of a NaI (Tl) detector for low level counting of naturally occurring radionuclides in soil
The Physics Department at the University of the Western Cape and the Environmental Physics group at iThemba labs have been conducting radiometric studies on both land and water. In this study a 7.5 cm X 7.5 cm NaI (Tl) detector was used to study activity concentrations of primordial radionuclides in soil and sand samples. The detector and the sample were placed inside a lead castle to reduce background in the laboratory from the surroundings such as the wall and the floor. The samples were placed inside a 1 L Marinelli beaker which surrounds the detector for better relative efficiency as almost the whole sample is exposed to the detector. Additional lead bricks were placed below the detector to further reduce the background by 20%. The NaI detector is known to be prone to spectral drift caused by temperature differences inside and around the detector. The spectral drift was investigated by using a ¹³⁷Cs source to monitor the movements in the 662 keV peak. The maximum centroid shift was about 4 keV (for a period of 24 hours) which is enough to cause disturbances in spectral fitting. There was no correlation between the centroid shift and small room temperature fluctuations of 1.56 ºC. A Full Spectrum Analysis (FSA) method was used to extract the activity concentrations of ²³⁸U, ²³²Th and ⁴⁰K from the measured data. The FSA method is different from the usual Windows Analysis (WA) as it uses the whole spectrum instead of only putting a ‘window’ around the region of interest to measure the counts around a certain energy peak. The FSA method uses standard spectra corresponding to the radionuclides being investigated, and is expected to have an advantage when low-activity samples are measured. The standard spectra are multiplied by the activity concentrations and then added to fit the measured spectrum. Accurate concentrations are then extracted using a chi-squared (χ²) minimization procedure. Eight samples were measured in the laboratory using the NaI detector and analyzed using the FSA method. The samples were measured for about 24 hours for good statistics. Microsoft Excel and MATLAB were used to calculate the activity concentrations. The ²³⁸U activity concentration values varied from 14 ± 1 Bq/kg (iThemba soil, HS6) to 256 ± 10 Bq/kg (Kloof sample). The ²³²Th activity concentration values varied from 7 ± 1 Bq/kg (Anstip beach sand) to 53 ± 3 Bq/kg (Rawsonville soil #B31). The ⁴⁰K activity concentration values varied from 60 ± 20 Bq/kg (iThemba soil, HS6) to 190 ± 20 Bq/kg (Kloof sample). The χ² values also varied from sample to sample with the lowest being 12 (Anstip beach sand) and the highest (for samples without contamination of anthropogenic nuclei) being 357 (Rawsonville soil #B28). A high χ² value usually represents incomplete gain drift corrections, improper set of fitting functions, proper inclusion of coincidence summing or the presence of anthropogenic (man made) radionuclei in the source [Hen03]. Activity concentrations of ⁴⁰K, ²³²Th and ²³⁸U were measured at four stationary points on the Kloof mine dump. The fifth stationary point was located on the Southdeep mine dump. These measurements were analysed using the FSA method and fitting by "eye" the standard spectra to the measured spectra using Microsoft Excel. These values were then compared to values obtained using an automated minimization procedure in MATLAB. There was a good correlation between these results except for ²³²Th which had higher concentrations when MATLAB was used, where 16 Bq/kg was the average value in Excel and 24 Bq/kg was the average value in MATLAB.