Calibration of buried NaI(Tl) scintillator detectors for 4π natural radionuclide measurement based on MCNP modelling.

NaI(Tl) scintillation detectors are widely used for field measurements of gamma rays due to their robustness and low cost; at least in case of high activity concentrations or large samples, they can provide accurate measurements. Measurements of naturally occurring activity concentrations of 40K, and of the decay series of 238U and 232Th, are of interest in the earth sciences in general, and in particular, NaI(Tl) scintillator-based gamma spectrometers can be used for the low cost determination of burial dose rates in natural geological samples [1]. We are currently developing a robust, portable, and wireless detector specifically intended for field measurement of natural radionuclide concentrations and dose rates. One of the challenges in developing such an instrument is reliable calibration. Currently most calibrations of field instruments depend on non-finite matrices of known K, U, Th activity concentrations, in either a 4π or 2π geometry [2]. There are only a limited number of these facilities available in the world, and for most laboratories repeated access for regular calibration is clearly difficult. We are investigating an alternative approach, based on the measurement of small samples (~300 g) containing well-known activity concentrations of only K or U or Th, and MCNP modelling to convert the observed spectra to those expected from specific activity concentrations in an non-finite 4π geometry. The determination of the non-finite matrix calibration spectra is based on three main steps: * MCNP simulations of NaI spectra for individual K, U and Th wax impregnated calibration cups of known activity and major element composition, validated against observed spectra. * MCNP simulations of individual K, U, Th spectra expected from field measurement (nonfinite matrix) for a chosen activity and major element composition [3]. * The resulting spectra ratios Infinite modelling/Cup modelling are used to multiply the observed calibration cup spectra to give predicted non-finite matrix calibration spectra. These modelled calibration spectra are validated by (i) combining in appropriate proportions, and comparing with measured spectra from non-finite matrices of known mixed K, U, Th composition, and (ii) by deriving these (known) K, U, Th concentrations using least squares fitting of the calibration spectra to the measured spectra (after subtraction of instrument background) [4]. This modelling approach to calibration also allows us to investigate the sensitivity of our analytical results to variations in measurement geometry, water content and major element composition. [ABSTRACT FROM AUTHOR]