In situ radiological characterisation can give decommissioning and waste-management teams information about radionuclides in components, structures and packaged materials without relying solely on destructive sampling. For this approach to produce useful activity estimates, the detector response must be understood for the measurement geometry, and the uncertainty in the result must be assessed. A methodology developed for a shut-down research reactor demonstrates how calibrated gamma spectrometry and package scanning can support these tasks.
The method combines high-purity germanium (HPGe) detectors, which provide strong energy resolution, with sodium iodide (NaI) scintillation detectors selected for robust, stable operation. Monte Carlo modelling is used to establish detector efficiencies for measurement conditions that may be difficult to reproduce with physical calibration sources.
Why detector calibration matters
Gamma-ray spectrometry identifies radionuclides through their characteristic photon energies and estimates activity from measured peak counts. Converting those counts into activity requires an efficiency calibration: an estimate of how likely the detector is to record a photon emitted by the material. That efficiency depends on photon energy, detector characteristics and the geometry between the detector and the measured object.
For historical radioactive waste, material may be distributed unevenly, and packages can have complex shapes or compositions. A calibration source with a simple, uniform geometry may not represent the item being measured. Monte Carlo simulations can model the detector and measurement arrangement to calculate response for more complex configurations. Those calculations, however, are only as reliable as the geometry and material assumptions used in the model.
Building and validating detector models
The methodology developed geometry models for two coaxial HPGe detectors and two NaI detectors using the MCNP-5 particle-transport code. Detector dimensions supplied by manufacturers provided a starting point, but physical verification and comparison with experimental measurements were needed to refine the models.
Accounting for inactive detector material
In an HPGe detector, the crystal’s inactive surface layer, commonly called the dead layer, does not contribute to signal collection in the same way as the active germanium volume. Its thickness is especially important for estimating efficiency at lower photon energies, where photons are more readily absorbed before reaching the active region. At higher energies, uncertainty in the active crystal volume can have a greater influence.
Radiographs were used to check crystal dimensions and its position inside the detector casing. The dead layer was investigated using a collimated americium-241 source, with measurements at two angles of incidence. The relative change in counts for 59.54 keV photons helped estimate the absorbing layer. The model was then adjusted in MCNP-5 so calculated efficiencies better matched measured efficiencies.
For the NaI detectors, radiography helped establish crystal dimensions. The model’s reflector thickness was adjusted to improve agreement between simulated and measured detector efficiencies. These steps illustrate why nominal specifications alone may not adequately describe a detector for efficiency calculations.
Comparing simulation with measured efficiency
Calibration measurements used certified gamma-ray standards spanning energies from 59.54 to 1408.0 keV. The standards included several radionuclides and source geometries. Experimental efficiencies were compared with MCNP-5 results, with measurement uncertainty considered as part of the comparison. After model adjustment, the discrepancy between calculated and measured efficiencies was below 5% for the tested sources, energies and angles.
This validation is essential before applying modelled efficiencies to unknown waste. The work also found that a separate efficiency-calibration software package could give results that differed from measurements for some detector configurations. The practical lesson is to verify calibration performance for the detector and geometry in use rather than assume that a software characterisation will transfer directly.
From detector response to package activity
Once detector models have been validated, a collimated detector can measure a package from multiple vertical or angular positions. Each position provides information about the response from different parts of the item. The measured peak counts are compared with simulated detector responses for package segments, and the activity in those segments is estimated by solving a system of equations. A non-negativity constraint prevents physically impossible negative activity estimates.
Scanning detail is a practical trade-off: more measurement positions can better reveal non-uniform activity, but require more time. The methodology therefore proposes selecting the number of scan segments based on representative packages, measurement duration and the accepted uncertainty. In an example involving a 200-litre barrel, increasing the number of vertical segments improved the estimate of caesium-137 activity; combining vertical and angular segments provided a reference for the converged result. The appropriate scan plan depends on the package and the required confidence in the result.
Application to contaminated HEPA filters
The approach was applied to three types of contaminated high-efficiency particulate air (HEPA) filter. Because filter construction affects photon attenuation, transmission measurements were made at different filter orientations and for selected gamma-ray energies. These measurements supported estimates of effective density and the proportions of glass-fibre paper and aluminium in each filter type.
Detector readings at ten positions showed that caesium-137 was sufficiently uniform across the examined filters for an open measurement geometry to be used. Activity estimates were then produced using different detector configurations, including collimated and bare detectors. For the first filter, for example, total activity estimates ranged from approximately 53,800 to 56,000 Bq across the reported methods. Across all three filter types, the results from the different spectrometers differed by less than 5%, within the reported total measurement uncertainties.
Practical implications for decommissioning teams
For plant and project teams, the methodology highlights several steps that make in situ gamma spectrometry more defensible:
- Characterise the detector: verify physical dimensions and account for factors such as HPGe dead-layer thickness or NaI reflector thickness.
- Validate the model: compare simulated efficiency with measurements across relevant energies and source geometries.
- Represent the item: account for package composition, attenuation and possible non-uniform activity.
- Plan the scan: choose measurement positions and segment count to balance survey time with the required uncertainty.
- Report uncertainty: include contributions from detector efficiency and measured counts, as well as the effects of spatial distribution.
For decommissioning planning and radioactive waste management, the value of in situ characterisation depends on more than taking a spectrum. Reliable results require detector models that have been checked against physical measurements and a measurement strategy suited to the package. With those elements in place, calibrated HPGe and NaI spectrometry can support radionuclide identification, activity estimates and more informed planning for contaminated materials.