Applied Radiation and Isotopes 56 (2002) 117–123 A national campaign for coincidence-summing correction in g-ray spectrometry Pierino De Felice*, Paola Angelini, Aldo Fazio, Marco Capogni ENEA, Istituto Nazionale di Metrologia delle Radiazioni Ionizzanti, C.R. Casaccia, P.O. Box 2400, I-00100 Rome A.D., Italy Abstract A national campaign was carried out in Italy for the application of coincidence-summing correction in g-ray spectrometry. Twelve laboratories, belonging to the National Environmental Radioactivity Surveillance Network, took part in the campaign. They are equipped with g-ray spectrometry systems based on p- and n-type HPGe detectors with different relative efficiencies. A simplified procedure was used for coincidence-summing correction. This procedure requires a single-photon, single-nuclide source to measure the peak-to-total efficiency ratio at just one photon energy value. All the laboratories were given a 137Cs standard source for total efficiency determination, and a 134Cs source in the same geometry whose activity they had to determine. The results show the usefulness of the procedure. The average deviation of all the laboratory results from the ENEA-INMRI reference value was reduced from�12% before correction to +1% after correction. The paper gives a description of the different organisational aspects of the campaign, reports the results obtained and draws conclusions in which the gain in measurement accuracy is evaluated in the light of the effort required for each participant to perform the correction. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Coincidence-summing correction; g-ray spectrometry; QA programmes; National Radioactivity Surveillance Network 1. Introduction A national network for radioactivity surveillance in foodstuffs and environment has been operating in Italy for more than 30 years. The network is formed, at present, by about 50 laboratories located in different regions in the country. A national programme for the quality assurance (QA) of the Italian network was started in 1983 and it was particularly intensified after the Chernobyl accident. The main purpose of this programme, that is still in effect, is to establish uniform levels of accuracy and reproduci- bility in measurement procedures routinely used by the laboratories belonging to the network. The relevant measurements include g-ray spectrometry and beta counting of different samples. The QA programme is based on periodical calibration and intercomparison campaigns. The organisers of the QA programme are the National Institute for Metrology of Ionising Radia- tion of ENEA (ENEA-INMRI) and the Italian Agency for Environmental Protection (ANPA). The role of ENEA-INMRI in the QA programme is to provide standard radioactive sources for calibration and techni- cal support for organisation and data analysis. The results of the first (1980–1990) comparisons showed a poor degree of uniformity among the laboratories and made it possible to identify the inadequacy of the calibration procedure as one of the major sources of error. Repeated calibration campaigns, in different experimental conditions, were then introduced in the QA programme. The calibration campaigns were very effective. In g-ray spectrometry measurements, for example, the deviation of the individual laboratory results from the ENEA-INMRI reference values were confined to 10–15% in more than 90% of the cases. This quality level was maintained even with increasing complexity of experimental conditions, e.g., introducing *Corresponding author. Fax: +39-06-304-83558. E-mail address:
[email protected] (P. De Felice). 0969-8043/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 9 - 8 0 4 3 ( 0 1 ) 0 0 1 7 6 - 2 serious spectral deconvolution problems and/or need of self-absorbtion correction due to changes in sample density and chemical composition. A deeper look shows a difference in the laboratory results obtained for single-g emitters and for multiple-g emitters. An example is given in Fig. 1. The reason is clearly due to the modest capability of the network laboratories to correct their measurements for coin- cidence-summing effects. This was then identified as one of the main sources of systematic errors. This was one of the main conclusions in several other recent national environmental radioactivity intercomparison exercises conducted by national standards laboratories in other countries (Dean et al., 1997). The errors made in neglecting coincidence-summing effects are, in frequent experimental conditions, greater than the accuracy required by the measurement procedure. As an example, for a 15% relative efficiency detector, this error can be 20–50% for close source–detector geometries and 5–10% for volume sources. It can increase notably (to 100–500%) when working with well- or n-type detectors or 100% relative efficiency detectors. With the increasing use of high efficiency detectors, coincidence-summing effects can easily make unachievable the 10% accuracy level even if good calibrations and other important corrections, such as self-absorption, are performed. It was after a general network meeting in 1998, when such a conclusion was again stressed, that a national working group was created to study the problem and find practical solutions. Members of the working group were the ENEA-INMRI, as co-ordinator, and 12 laboratories, about one-fifth of the network labora- tories. The list of participating laboratories is reported in Table 1. They were selected in consideration of the competence shown in this particular field and of the characteristics of their measurement systems. 2. Simplified procedure for coincidence-summing correction Accurate procedures, developed to correct for this effect, require considerable effort, in particular for the determination of the total efficiency, that makes them difficult to implement in routine g-ray spectrometry. It was for this reason that a preliminary study was performed by the ENEA-INMRI to identify possible simplified procedures. Three different procedures were developed and compared. These procedures are based on the same theoretical expressions of the correction factors used in the basic approach, but differ in that 0.00 0.05 0.10 0.15 0.20 -2 0 -1 8 -1 6 -1 4 -1 2 -1 0 -8 -6 -4 -2 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 Relative deviation (%) R e la ti v e f re q u e n c y Am-241 Cs-137 Co-60 Ba-133 Mn-54 Na-22 Cs-134 Fig. 1. Distribution of relative deviations from the ENEA reference value in the 1996 intercomparison campaign orga- nised for the laboratories of the national radioactivity network. The arrows indicate the sub-distribution relative to each one of the nuclides taken into account in the intercomparison. A fine structure is clearly emerging in the distribution resulting from the presence of systematic errors due to coincidence-summing effects on 134Cs, 22Na and 133Ba. The slight overestimation of the 54Mn activity is due to coincidence-summing errors in calibration (use of 60Co and 88Y). Deviations for nuclides not affected by coincidence-summing are confined within a few per cent. Table 1 Participants in the national campaign for coincidence-summing corrections Laboratory Scientist Agenzia Regionale per la Protezione dell’AmbienteFAOSTA G. Agnesod, C. Operti Presidio Multizonale di PrevenzioneFBARI V. Martucci, L. Vitucci Agenzia Nazionale di Protezione AmbientaleFROMA M. Belli, R. Ocone Agenzia Regionale per la Protezione dell’AmbienteFFIRENZE C. Giannardi, S. Nava Centro di Riferimento regionaleFCAMPOBASSO C. Cristofaro Universit"a di Napoli ‘‘Federico II’’FNAPOLI V. Roca Centro Comune di RicercaFISPRA V. Vocino Agenzia Regionale per la Protezione dell’AmbienteFIVREA M. Magnoni, S. Bertino Agenzia Provinciale per la Protezione dell’AmbienteFBOLZANO L. Verdi Agenzia Regionale per la Protezione dell’AmbienteFPIACENZA A. Gazzola, R. Sogni Agenzia Regionale per la Protezione dell’AmbienteFPADOVA A. Bertolo Seconda Universit"a di NapoliFCASERTA F. Terrasi, C. Sabbarese P. De Felice et al. / Applied Radiation and Isotopes 56 (2002) 117–123118 approximations are used in the determination of the total efficiency curve (De Felice et al., 2000). These procedures make it possible to reduce to acceptable levels the experimental effort required to correct g-ray spectrometry measurements for coincidence-summing effects. After discussion of the results obtained in the ENEA-INMRI study the participating laboratories agreed to perform a campaign for coincidence-summing corrections. The third of the ENEA-INMRI procedures was selected as it optimised the balance between the accuracy improvement and the effort needed. A summary of the selected procedure is reported hereafter. Further details are given in De Felice et al. (2000). The full-energy peak count rate, ni; for a generic photon, gi; emitted by a radionuclide is given by ni ¼ AIgi eiCi ; ð1Þ where A is the radionuclide activity, Igi the emission probability, ei the full-energy peak efficiency, and Ci the correction factor for coincidence summing. The equation for calculating Ci given by Semkow et al. (1990) is used: Ci ¼ 1þ P k;m PtkmPkPmekem Igi ei � � 1� P j Ptij PiPjetj Igi � � ; ð2Þ where Ptkm is the probability per decay that the coincidence transitions k and m occur, Pk and Pm are the probabilities that in each transition the respective photons gk and gm will be emitted (Ek þ Em ¼ Ei), ek and em are the full-energy peak efficiencies for the photons gk and gm and etj is the total efficiency for a generic photon gj : The first factor in Eq. (2) takes into account the ‘‘summing-in’’ effect. This occurs when two photons (gk and gm) are simultaneously detected at their full energies and a single count is recorded in the full-energy peak of gi: The second factor takes into account the ‘‘summing- out’’ effect that produces a loss of counts from the full- energy peak of gi due to the detection (not necessarily at full energy) of any other coincident photon gj : A simple relation is then assumed, above 160 keV, between the peak-to-total efficiency ratios, Re ¼ ej=etj ; and the photoelectric-to-total cross-section (in germa- nium) ratios, Rs ¼ sj=stj : R ¼ Re=Rs ¼ KEg; ð3Þ where Eg is the photon energy. In a crude approximation the parameter K is assumed to be constant for each counting geometry. The slope K ¼ R=Eg is then calculated only at one photon energy, i.e. 137Cs 661 keV photon, for each specific counting geometry. This value is then used, together with the full-energy peak efficiency and cross-section curves, to derive all the etj values for that counting geometry, by using Eq. (3), that can be re-written as etj ¼ ej KRsEg for Eg > 160 keV: ð4aÞ The etj below 160 keV is approximated with ej : etj ¼ ej for Ego160 keV: ð4bÞ This procedure requires one standard source of a single line emitter (e.g. 137Cs) and the collection of one spectrum for each counting geometry considered. In the first study, on a 15% relative efficiency p-type coaxial HPGe detector, the calculated Ci factors agreed within few per cent with those based on more accurate experimental determinations of the total efficiency curve from measurements of standard sources of eight different single-g-ray emitters. A further check of the described procedure was carried out in the measurement of radionuclides of the natural decay series with high efficiency detectors (Nuccetelli et al., 2000). 3. Organisation of the campaign Technical procedures and organisational aspects were agreed between all the participants and the co-ordinator in a starting meeting. Each laboratory had to measure the activity of a standard 134Cs source in a Marinelli beaker. This nuclide was chosen for the remarkable magnitude of the coincidence-summing effect arising from its cascade decay scheme and photon energies. In order to put all the laboratories in the same starting conditions, the participating laboratories agreed to avoid the application (in calibration and in analysis) of any coincidence-summing correction. Activity measure- ments were restricted to the analysis of the 563.2, 569.3, 604.7, 795.9 and 801.9 photons emitted in the 134Cs decay, as these are the main lines affected by summing- out events. Each laboratory had to use the most recent full-energy peak efficiency curve. It was also agreed that the total efficiency values had to be obtained by each laboratory by integration of the overall g spectrum with linear extrapolations to zero photon energy and with correction for 137Cs X-ray emission. The standard source needed for determination of the single total efficiency value and the parameter K was a 137Cs solution in the same Marinelli beaker used for the 134Cs source. In order to investigate the dependence of K on the counting geometry it was agreed that also a standard 137Cs point source had to be measured by each laboratory at 10 cm from the detector window. It was agreed that all the results had to be analysed by ENEA- INMRI and reported using closed codes. All of the standard sources needed for the campaign were prepared and calibrated by ENEA-INMRI which also provided the calibration certificates. The Marinelli beakers were filled with aqueous solution containing HCl 4M and stable caesium carrier. Activity values were 1.02 and 1.17 kBq respectively for the Marinelli beaker and point sources of 137Cs, and 1.31 kBq for the 134Cs source in Marinelli beaker, with relative combined P. De Felice et al. / Applied Radiation and Isotopes 56 (2002) 117–123 119 standard uncertainties of 1%. All the activity values were referred to the reference time of 15 March 1999. Participants were given a questionnaire to be filled in and to be sent back to the ENEA-INMRI. The most relevant information requested in the questionnaire were: live- and real-time for each spectrum acquired, values of ej and etj at 661.7 keV (with corresponding combined standard uncertainties) obtained for each counting geometry, full-energy peak efficiency curve used, the five values of the 134Cs activity for each photon emission considered. The main characteristics of the photon detectors used by the participants are reported in Table 2. Nominal relative efficiencies varied from 25% to 71%. n-Type HPGe crystals were used by all the laboratories except one, which used a 40% p-type crystal (detector code No. 10). 4. Results The five 134Cs activity values, as reported by each participant, without corrections for coincident summing were first considered. The arithmetic mean and the experimental standard deviation of the five values were then calculated (Fig. 2). The systematic effect of the common source of error due to coincident summing was clearly shown. As expected, the laboratories using higher efficiency detectors reported lower activity values by more than 10%. Most of the laboratories with 25–30% detectors reported relative deviations of about 10%. If experimental means and measurement procedures are correctly applied by each laboratory, the five activity values should coincide within the quoted uncertainties. As the magnitude of the coincident-summing effect is not the same for the five considered photons, the activity values are also spread out. The experimental standard deviation of the five activity values gives a measure of this disagreement. It varied from 3% to 10% (Fig. 2), with increasing detector efficiency. A number of preliminary checks were performed on the original participant data before calculation of the correction factors. The full-energy peak efficiency values at 661 keV for the two counting geometries considered were reported as a function of the detector relative efficiency, as given by the manufacturer of each detector (Fig. 3a). All the measured efficiency values were found in agreement with the corresponding nominal efficiency values. A similar check was carried out for the total efficiencies measured by the participants (Fig. 3b). In both cases, deviations from the interpolating curve can be attributed to uncertainties in the measured efficiency values and, more probably, to inaccuracy of the detector relative efficiency as given by the manufacturer. A further preliminary check considered the ratios of the full-energy peak efficiency values for point source and Marinelli beaker counting geometry (Fig. 4a). In a first approximation this ratio is not affected by the detector efficiency, as confirmed by the results obtained. Table 2 Main characteristics of the photon detectors used by the participants in the national campaign for coincidence-summing correction Detector code HPGe crystal type Nominal relative efficiency (%) 1 p 31 2 p 28 3 p 28 4 p 30 5 p 27 6 p 30 7 p 43 8 p 30 9 p 28 10 n 40 11 p 25 12 p 25 13 p 71 -30 -20 -10 0 10 20 30 1 2 3 4 5 6 7 8 9 10 11 12 13 Detector code R el a ti v e d ev ia ti o n ( % ) Not corrected Corrected (a) -30 -20 -10 0 10 20 30 2 0 3 0 4 0 5 0 6 0 7 0 8 0 Detector relative efficiency (%) R el a ti v e d ev ia ti o n ( % ) Not corrected Corrected (b) Fig. 2. Average deviations and experimental standard devia- tions of the 134Cs activity values reported by each laboratory. Original values (dark), not corrected for coincidence-summing effect, and corrected values (clear) are reported versus the detector code (a) and detector relative efficiency (b). P. De Felice et al. / Applied Radiation and Isotopes 56 (2002) 117–123120 Differently, a dependence on the detector efficiency has to be expected for the ratio of the full-energy peak efficiency at 661 keV (from 137Cs) and the uncorrected efficiency at 1333 keV (from 60Co). The data, reported in Fig. 4b, show this dependence and the magnitude of the coincidence-summing effect on the detector relative efficiency. The ratio of the full-energy peak counting efficiency at 661 keV as obtained from the measurement of the source circulated during the campaign was finally compared with the value obtained from the efficiency curve that each laboratory had previously determined. Agreement within 4% was obtained for all laboratories. An investigation on the origin of the efficiency calibra- tion curves used by the laboratories showed different traceability and different time periods span since the date of calibration. These factors can account for the observed deviations. Finally, the values of K were calculated for each detector and for the two counting geometries. As shown in Fig. 5, a small correlation with the detector relative efficiency and counting geometry was obtained. This may lead to further simplifications in the correction procedure as for example the use of tabulated K values eliminating any need for the measurement of the total efficiency. By means of Eqs. (4a) and (4b), and by using the individual K value, the total efficiency curve was generated for each detector. Typical results are reported in Fig. 6. 0.00 0.01 0.02 0.03 0.04 0.05 0 20 40 60 Relative efficiency (%) F u ll -e n e rg y p e a k e ff ic ie n cy (a) 80 0.00 0.05 0.10 0.15 0.20 0 20 40 60 Relative efficiency (%) T o ta l e ff ic ie n cy (b) 80 Fig. 3. Full-energy peak (a) and total (b) efficiency values at 661.7 keV 137Cs photon energy for the two counting geometries considered in the work: point source at 10 cm from the detector window (higher values) and 1 l Marinelli beaker (lower values) filled with aqueous solution. The correlation with relative efficiency is evident. 0.00 0.10 0.20 0.30 0.40 0 10 20 30 40 50 60 70 80 Relative efficiency (%) E ff ic ie n cy r a ti o (a) 1.70 1.75 1.80 1.85 1.90 1.95 2.00 0 10 20 30 40 50 60 70 80 Relative efficiency (%) E ff ic ie n cy r a ti o (b) Fig. 4. The full-energy peak efficiency ratios at 661.7 keV (137Cs) photon energy for point source and Marinelli beaker counting geometry are reported in graph (a) as a function of the relative efficiency. The 661 keV (137Cs) to the 1333keV (60Co) full-energy efficiency ratios for Marinelli beaker geometry are reported in graph (b). 0.008 0.010 0.012 0.014 0.016 0.018 0.020 0 20 40 60 Relative efficiency (%) K ( 1 /k e V ) (a) 80 0.008 0.010 0.012 0.014 0.016 0.018 0.020 0 20 40 60 80 Relative efficiency (%) K ( 1 /k e V ) (b) Fig. 5. Values of K for the two counting geometries considered: 1 l Marinelli beaker (a) and point source at 10 cm from the detector window (b), reported as a function of the detector relative efficiency. P. De Felice et al. / Applied Radiation and Isotopes 56 (2002) 117–123 121 Correction factors for summing-out effect were then determined for each detector and for each of the considered 134Cs photon emissions. Nuclear data and decay schemes were taken from IAEA (1991) and Lagoutine et al. (1983). Efficiency values were needed only at photon energy values higher than 500 keV. The procedure was then also applied to the n-type detector No. 10, assuming that in this energy range the energy response is not substantially different from that of p- type detectors for which the procedure was previously tested (De Felice et al., 2000). The correction factors obtained for each detector and each energy line are reported in Fig. 7. These are all greater than one, as is expected for summing-out effects. Furthermore, a remarkable dependence on the detector relative efficiency was obtained. It is also shown that the correction factor depends on the energy line considered. Numerical values resulted in the range from 7% to 17% for the 604.7 and 795.9 keV peaks and from 12% to 32% for the 563.2, 569.3 and 801.9 keV peaks, showing good reproducibility for similar experimental conditions. Application of the correction factors to the individual activity values given by the laboratories resulted in a new average activity value with associated experimental standard deviations. The results are reported in Fig. 2. The relative deviations from the reference value, that were in the interval [�18%, �9%] for the uncorrected values, became restricted to [�2.5%, +2.8%] after correction. Similarly, the experimental standard devia- tion values reduced from [3%, 11%] to [0.7%, 8%] after correction, with the average value reducing from 6% to 3%. In most of the cases (10 of 13) the reference value was within one standard deviation of the laboratory result. In the remaining three cases, the agreement was within two standard deviations. Furthermore, the correlation between the observed deviations and the detector efficiency, that was evident before the correc- tion, was removed after the correction, showing that the procedure can be equally applied to low or high efficiency detectors. The average activity obtained from all the laboratory results changes from �12% to +1%. This low residual average deviation is well within the combination of the other sources of uncertainty not explicitly considered in this work, namely those on: 134Cs and 137Cs standard sources (1%), full-energy efficiency curves (3–5%), extrapolation to zero energy for determination of the 661 keV total efficiency, counting statistic. Nevertheless, this little overestimation of the 134Cs activity can be explained with the argument that most of the laboratories used, for calibration of their detectors mixed nuclide solutions containing 60Co and 88Y for which summing-out events occur. With the same procedure used for 134Cs, correction factors were then calculated for the full-energy peaks of these two nuclides, for each experimental condition. The results are reported in Fig. 7. Correction factors resulted almost equal for the four photon emission considered. As for 134Cs, a dependence on the detector efficiency was obtained but, in this case, the correction factors were a factor of 3 lower. They were indeed between 5% and 8% for all the detectors except for detector 10 for which the correction factors are 12–13%. A change of such a magnitude in the 60Co and 88Y full-energy peak efficiency values produces a much lower corresponding variation of the values in the 500–700 keV region, important for 134Cs correction. This change, evaluated for a 30% relative efficiency detector, resulted from 1% Detector code 4 0.00 0.02 0.04 0.06 0.08 0.10 0 500 1000 1500 Energy (keV) T o ta l e ff ic ie n cy (a) Detector code 13 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0 500 1000 1500 Energy (keV) T o ta l e ff ic ie n cy (b) Fig. 6. Examples of calculated total efficiency curves. Graphs (a) and (b) refer, respectively, to a 30% and a 71% relative efficiency detector. 1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 20 40 60 80 Relative efficiency (%) C o rr e ct io n f a ct o r 563.2 keV 569.3 keV 604.7 keV 795.9 keV 801.9 keV 1173 keV 1332 keV 898 keV 1836 keV Fig. 7. Coincidence-summing correction factors calculated for each participant’s detector, for 1 l Marinelli beaker counting geometry and for the 134Cs, 60Co and 88Y photon emissions (photon energies reported). P. De Felice et al. / Applied Radiation and Isotopes 56 (2002) 117–123122 to 3% depending on the energy line. This is in agreement with the results of the campaign. This further correction was not applied to all the laboratory results as it was considered negligible for the type of measurements considered. 5. Conclusions The application of coincident-summing corrections made it possible to reduce to a few per cent the deviations for the average activity value obtained by each laboratory with notable improvement of the internal agreement of the activity values obtained from different energy lines. The experimental effort required for each laboratory to perform the correction was acceptable if compared with the accuracy improvement. The applied procedure is suitable for applications where extreme accuracy is not required and where time and costs are important factors to consider. These include many environmental and health physics measurements. The tested procedure will be proposed in the near future to all the network laboratories in the frame of the ongoing national QA programme and it will be extended to other nuclides. Comparison with the results of other correction procedures available (Sundgren, 1993; Blaauw, 1993), also implemented in computer software packages, will be of interest. References Blaauw, M., 1993. The use of sources emitting coincident g-rays for determination of absolute efficiency curves of highly efficient Ge detectors. Nucl. Instrum. Methods A 332, 493–500. Dean, J.C.J., Bowles, N.E., Brosnan, W.E., Carter, M., De Lavison, P.A.G., Jerome, S.M., Keightley, J.D., Woods, M.J., Woods, S.A., 1997. Environmental radioactivity intercomparison exercise, 1996. NPL Report CIRM 3, National Physical Laboratory, UK. De Felice, P., Angelini, P., Fazio, A., Biagini, R., 2000. Fast procedures for coincidence-summing correction in g-ray spectrometry. Appl. Radiat. Isot. 52, 745–752. IAEA International Atomic Energy Agency, 1991. X-ray and g- ray standards for detector calibration. Technical Document 619. Lagoutine, F., Coursol, N., Legrand, J., 1983. Table de radionucleides. CEA, Commissariat a l’Energie Atomique, France. Nuccetelli, C., Angelini, P., Bolzan, C., De Felice, P., Risica, S., 2000. Application of coincidence-summing correction to natural radionuclide activity measurements by gamma-ray spectrometry. Proceedings of the Fifth International Con- ference on High Levels of Natural Radiation and Radon Areas: Radiation Dose and Health Effects, Munich, September 4–7. Semkow, T.M., Mehmood, G., Parekh, P.P., Virgil, M., 1990. Coincidence summing in gamma-ray spectroscopy. Nucl. Instrum. Methods A 290, 437–444. Sundgren, O., 1993. Coincidence-summing correction in gam- ma spectrometry. Sci. Total Environ. 130, 131. P. De Felice et al. / Applied Radiation and Isotopes 56 (2002) 117–123 123