Introduction
The result of a gamma spectrometry measurement is initially a histogram, a list of pulse values for consecutive ranges of the histogram. Each of these ranges is described by a channel number.
Typical frequency distribution of a gamma spectrum. It shows the frequency distribution of the measured pulses as a function of the channel number. The spectrum was measured with an HPGe detector for 8192 channels.
By calibrating the measurement system, it is achieved that
- the gamma spectrum is interpreted as a function of the energy of the gamma radiation instead of the channel number and
- the amount of a radionuclide is interpreted by its activity instead of a count of pulses.
This is particularly useful for automatic evaluation routines, as knowledge of the energy dependence of peak widths is also a very useful property.
This information is determined through calibrations. Specifically, these are
- Energy calibration,
- Peak width calibration, and
- Efficiency calibration.
In practice, good calibrations are always based on high-quality measured gamma spectra and nuclide data.
Note:
In this context, we want to point out GIGO: GIGO is the acronym for Garbage In – Garbage Out, which refers to the fundamental principle of gamma spectrometry that the accuracy of quantifying radionuclides depends entirely on the quality of the measurement data, calibration, and sample preparation.
In high-quality measured gamma spectra, the peaks of the nuclides used for calibration should be identifiable and evaluable with a very high degree of accuracy. This means that the associated measurement uncertainties should be as low as possible (ideally negligible) and that individual peaks should not be overlapped (or disturbed) by other lines.
Measurement uncertainty in determining the peak areas for an HPGe spectrum of 60Co. Left: Measurement with a very short measurement time; the peak at 1332 keV shows extreme fluctuations in the individual channel values (small image); a fit of a Gaussian function to determine the peak areas would be associated with large uncertainties. Right: Measurement with a long measurement time; the peak at 1332 keV shows an almost perfect Gaussian distribution, meaning the uncertainty in determining the peak area would be very small.
We will address this topic more closely in the section Peak Evaluation.
Another factor to consider in the quality of calibrations is the nuclide data used.
As we will see, during the determination of the various calibration factors, data concerning the specific nuclear states of the calibration nuclides are applied. Examples of such data include half-lives, gamma and X-ray energies of the characteristic lines, transition probabilities, etc.
Decay scheme for the radioactive decay of 60Co into the daughter nuclide 60Ni. It includes information about the half-life (5.2714 a), type and frequency of the transition (β1 with 99.925 %, β2 with 2E-4 %, β3 with 0.12 %) and the corresponding energy levels from which the energies of the characteristic lines can be determined (the red arrows mark the main transition lines at 1173.2 keV (= 2505.7 keV -1332.5 keV) and 1332.492 keV).
This data is usually already integrated as nuclide libraries in (commercial) gamma spectrometry programs. Other sources for nuclide data are publicly accessible databases provided online and free of charge by federal authorities (e.g., National Institute of Standards and Technology (NIST)) and international organizations (e.g., International Atomic Energy Agency (IAEA)). However, they are also available through commercial providers (e.g., Nucleonica GmbH, Institute for Spectrometry and Radiation Protection (ISUS)).
This immediately raises the question for the user as to why they should not exclusively use the data already included in the gamma spectrometry programs? This usually allows for automatic evaluation of their measured data, and they do not need to worry about which data to use and where to obtain it. This is fundamentally correct and will usually be effective. However, it has happened on multiple occasions that
- the integrated nuclide libraries contained erroneous data,
- new measurements of nuclide data resulted in corrected or new data (previously unavailable), or
- existing evaluations were revised, leading to corrected values.
Using incorrect nuclide data results in erroneous calibration values during calibration measurements, which – when used in the evaluations of gamma measurements – lead to erroneous results (in addition to the contribution that incorrect nuclide data generate when evaluating the measured characteristic peaks).
You see, there are good reasons to closely inspect the nuclide data you use. You should not rely on updates of new findings from, for example, manufacturers of gamma spectrometry software. Experience has shown that despite multiple notifications from manufacturers about erroneous values, no corrections were made in the delivered nuclide data libraries.
Here, it is advisable to compare the data from the nuclide libraries with data from other independent sources and to investigate the reason for any discrepancies.
At this point, just one more thing…
Tip from Practice:
All measurements and evaluations should be traceable for quality assurance, meaning that not only the measured spectrum with all measurement parameters but all calibration spectra or files and also the data used in the evaluation of the measurement data must be available again at any later time.
Here, a revision procedure for the various calibration measurements and the nuclide libraries is recommended. With every change, whether due to a new measurement or the correction of individual values, the original files should be saved. Your successors will inherit an increased revision index in future measurements and evaluations and will be documented there.
But now back to the actual topic, the calibration of a gamma spectrometry measurement system, and some general points for carrying out calibrations.
Every calibration…
- must be complete: the energy range of the calibration must cover the entire range intended to be captured by the (planned) measurements. It must also be covered by a sufficient number of measurement points (reference points) to allow for the interpolation of any calibration values with sufficient accuracy.
Schematic example of a complete calibration in the energy range from 100 keV to 2250 keV. The range is covered by a sufficient number of reference points that allow interpolation (Note: A simple exponential function was chosen for the schematic representation. Real calibration functions almost always look different!).
- must be unambiguous: the reference points should be formed from non-overlapping peaks, dead times of the detector over 1 %, and “true coincidence summing” must be avoided by a sufficient distance of the calibration sample from the detector.
- should consider the external circumstances: to avoid later recalculations of the calibration values, calibration measurements should be conducted in the same measurement and sample geometry as in the later measurements of the samples. If possible, the structure of the samples should also be considered (Example: if sand-like samples are planned to be measured in a Marinelli beaker, then an identical Marinelli beaker with homogeneous sand and activity mixture could be chosen for the calibration measurements).
