REIMEP-22 inter-laboratory comparison: U Age Dating determination of the production date of a uranium certified test sample

Transcription

1 Radiochim. Acta 2015; 103(12): Celia Venchiarutti*, Zsolt Varga, Stephan Richter, Rozle Jakopič, Klaus Mayer, and Yetunde Aregbe REIMEP-22 inter-laboratory comparison: U Age Dating determination of the production date of a uranium certified test sample DOI /ract Received May 3, 2015; accepted July 28, 2015; published online September 14, 2015 Abstract: The REIMEP-22 inter-laboratory comparison aimed at determining the production date of a uranium certified test sample (i.e. the last chemical separation date of the material). Participants in REIMEP-22 on U Age Dating Determination of the production date of a uranium certified test sample received one low-enriched 20 mg uranium sample for mass spectrometry measurements and/or one 50 mg uranium sample for α-spectrometry measurements, with an undisclosed value for the production date. They were asked to report the isotope amount ratios n( 230 Th)/n( 234 U) for the 20 mg uranium sample and/or the activity ratios A( 230 Th)/A( 234 U) for the 50 mg uranium sample in addition to the calculated production date of the certified test samples with its uncertainty. Reporting of the 231 Pa/ 235 U ratio and the respective calculated production date was optional. Eleven laboratories reported results in REIMEP-22. Two of them reported results for both the 20 mg and 50 mg uranium certified test samples. The measurement capability of the participants was assessed against the independent REIMEP-22 reference value by means of z-andzeta-scores in compliance with ISO 13528:2005. Furthermore a performance assessment criterion for acceptable uncertainty was applied to evaluate the participants results. In general, the REIMEP-22 participants results were satisfactory. This confirms the analytical capabilities of laboratories to *Corresponding author: Celia Venchiarutti, European Commission, Joint Research Centre (JRC), Institute for Reference Materials and Measurements (IRMM), Retieseweg 111, 2440 Geel, Belgium, cvenchiarutti@gmail.com Stephan Richter, Rozle Jakopič, Yetunde Aregbe: European Commission, Joint Research Centre (JRC), Institute for Reference Materials and Measurements (IRMM), Retieseweg 111, 2440 Geel, Belgium Zsolt Varga, Klaus Mayer: European Commission, Joint Research Centre (JRC), Institute for Transuranium Elements (ITU), Postfach 2340, Karlsruhe, Germany determineaccuratelytheageofuraniummaterialswith low amount of ingrown thorium (young certified test sample). The Joint Research Centre of the European Commission (EC-JRC) organised REIMEP-22 in parallel to the preparation and certification of a uranium reference material certified for the production date (IRMM-1000a and IRMM- 1000b). Keywords: Age dating, thorium, uranium, nuclear forensics, inter-laboratory comparison, quality control. 1 Introduction Nuclear forensics is a key element of nuclear security aiming at the identification and characterisation of seized nuclear material, such as uranium or plutonium, to reestablish the history of the nuclear material of unknown origin. By applying advanced analytical techniques, the isotopic composition, the chemical impurities and the macro- or microstructure of the nuclear material can be determined [1]. Thepotentialand advantagesofthe age determination of the material has been successfully demonstrated [1, 2]. The age of a nuclear material refers to its production date, i.e. the time elapsed since the last chemical separation of the daughter nuclides from the parent U or Pu radionuclides [3, 4]. During its production, the nuclear material is chemically purified from impurities including radioactive decay products. However, up to now, no certified age dating reference materials existed for the validation of mass spectrometric or radiometric methods, which in combination with the proper uncertainty evaluation [5], are required to characterise intercepted nuclear material, establishing its age and origin without ambiguity. This determined origin can be then verified against the declared origin of the seized material and therefore provide the necessary evidence for nuclear safeguards and nuclear forensic investigations in a court of law. The European Commission Joint Research Centre Institute for Reference Materials and Measurements (JRC- IRMM) is a renowned producer of certified reference ma-

2 826 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison terials (CRMs) and of quality control/conformity assessment tools such as inter-laboratory comparisons (ILCs) supporting nuclear safeguards and security. In this context, the two EC-JRC institutes, JRC-IRMM and JRC-ITU (Institute for Transuranium Elements) joined efforts to produce uranium reference materials certified for the production date suitable to serve as a reference material for method validation in age dating of uranium materials. These CRMs, called IRMM-1000a (20 mg uranium) and IRMM-1000b (50 mg uranium), were prepared from a lowenriched uranium solution by a complete chemical separation of thorium from uranium at a well-known time with subsequent monitoring of the ingrowth of the daughter nuclides in the purified material, which confirmed the very high Th separation efficiency. Before release, units from these CRMs were used as Proficiency Test (PT) items for the Regular European Inter-laboratory Measurement Evaluation Programme, REIMEP-22 U Age Dating Determination of the production date of a uranium certified test sample [6]. REIMEP-22 was organised, according to ISO/IEC 17043:2010 [7], in support to the Nuclear Forensics International Technical Working Group (ITWG). The ITWG is an international network of nuclear forensics experts, including nuclear scientists, law enforcement and regulators, contributing to advances in nuclear forensics through a variety of activities, such as comparative material analysis exercises, guidelines and best practices. Beyond the ITWG, network laboratories or institutions in the field of nuclear and environmental sciences also participated in REIMEP-22. Participants received one 20 mg and/or one 50 mg uranium certified test sample with an undisclosed value for the production date, depending on the applied measurement technique. They were asked to report the two parent/daughter pairs: 234 U/ 230 Th (compulsory) and 235 U/ 231 Pa (optional) to determine the production date of the uranium certified test sample and its associated uncertainty. Participants were requested to apply their routine measurement procedures and to complete a questionnaire on the measurement procedures applied in their laboratories. Fourteen laboratories registered for REIMEP-22; three laboratories could not report their results due to technical problems. Eleven laboratories reported results; among those, two laboratories submitted results for both 20 mg and50 mg uranium certified test samples. Thirteen results were reported for the 234 U/ 230 Th ratio. A specific lab code per participant was attributed to each of the thirteen results. Six out of the eleven participating laboratories were members of the ITWG and are involved in the measurements of nuclear forensics samples. This paper presents the results reported by REIMEP- 22 participants, the evaluation of the participant performances and discusses the questionnaire and participants feedback in order to gain insight in the current techniques applied in age dating and the expertise applicable in the field of nuclear forensics. 2 Materials and methods 2.1 Preparation of REIMEP-22 The REIMEP-22 certified test samples were prepared in the framework of the production and certification of the reference materials, IRMM-1000a and IRMM-1000b, in compliance with ISO Guide 34 [8]. They were produced from a low-enriched uranium solution (with a relative mass fraction m( 235 U)/m(U)of 3.6%) after chemical separation of thorium decay products from the material, at a wellknown time. The production date was then confirmed by measuring the ingrown 230 Th in the material. The methodology by Varga et al. [9, 10] was the analytical method used for the production of the certified test samples. The resulting purified uranium solution was dispensed into precleaned PFA (perfluoroalkoxy alkane) vials to produce 161 units in two sizes: 20 mg (IRMM-1000a) and 50 mg uranium (IRMM-1000b) in dried uranyl-nitrate form. 2.2 Assignment of the reference value The reference value is the carefully recorded date and time of the last chemical separation and corresponds to the complete removal of thorium from the original uranium material. The reference value is the production date expressed as dd/mm/yyyy with an expanded uncertainty in days and is based on the 230 Th/ 234 U radiochronometer. A complete uncertainty budget was established in accordance with the Guide to the Expression of Uncertainty in Measurement (GUM) [11]. The completeness of the separation of thorium from the uranium was assessed during the confirmation and homogeneity assessments carried out in compliance with ISO Guide 34, ISO Guide 35:2006 [12], andiso [13] aspart of the certification of the reference material. Detailed results of these assessments are described in Venchiarutti et al. [14]. The confirmation study demonstrated the successful purification of the uranium material (resulting in an expanded uncertainty of 0.17 d, k = 2), whereas the homogeneity study using one-way analysis variance (ANOVA) showed that the REIMEP-22 certified test samples were

3 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison 827 considered sufficiently homogeneous for the purpose of this PT with an expanded uncertainty for homogeneity of 7.8 d (k =2). As a result, the REIMEP-22 reference value, which corresponds to the production date of the uranium test sample based on the 230 Th/ 234 U radiochronometer is: 09/07/2012 (as 9 July, 2012) with an uncertainty, at the time of the REIMEP-22 ILC, of 7.8 d (k = 2) based on the confirmation and homogeneity study results during the certification of the candidate reference materials. 3 Discussion 3.1 Measurements results Nine results were reported for the n( 230 Th)/n( 234 U) amount ratios in the 20 mg uranium certified test samples and four for the A( 230 Th)/A( 234 U) activity ratios in the 50 mg uranium certified test sample. In addition, two laboratories reported the n( 231 Pa)/n( 235 U) amount ratios in the 20 mg uranium certified test sample, as well as the production dates and associated uncertainties. Participants were requested to report three replicates of the ratio and the average value normalised to a common reference date specified by the ILC organiser as 06/03/2013 (6 March, 2013). This enabled the evaluation of the measurement results without any data treatment by the ILC organiser. This approach leaves the responsibility for the reported result with the participant and enables to identify directly any differences in the reported ratio results. The participants results are presented with their respective lab codes in Figures 1 2 and Table 1. All the results are displayed as reported by the participants, i.e. with uncertainties with coverage factors of either k=1or k=2; however laboratories results that were reported as standard uncertainties with k=1are clearly identified in the figures and in the text. The average (amount and activity) ratios in the figures are reported for the reference date of 06/03/2013. For the results of the 20 mg uranium certified samples analysed by the mass spectrometry, four participants reported production date values that agreed well with the Fig. 1: Reported results for the 20 mg uranium certified sample a) for production dates (squares), as dd/mm/yyyy and uncertainty in day as stated by participants (i.e. k=1or k=2), with reference value X ref on 09/07/2012 (full line) and its expanded uncertainty (dotted lines) as described in Section 2.2. b) for the average n( 230 Th)/n( 234 U) amount ratios (diamonds) on 06/03/2013 with uncertainties as stated by participants (i.e. k=1or k=2). The asterisks in the lab codes legend indicate values reported by laboratories with uncertainties k=1 (standard uncertainty). The lab codes given by the PT organisers were 10249, 10246, etc and are now presented as L49, L46, etc.

4 828 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison Fig. 2: Reported results for the 50 mg uranium certified sample a) for production dates (squares), as dd/mm/yyyy and uncertainty in day (i.e. k=1or k=2), with reference value X ref on 09/07/2012 (full line) and its uncertainty (dotted lines) as described in Section 2.2. b) for the average A( 230 Th)/A( 234 U) amount ratios (diamonds) on 06/03/2013 with uncertainties as stated by participants (i.e. k=1or k=2). Same legend for lab codes as in Figure 1. The asterisks indicate values reported by laboratories with uncertainties k=1(standard uncertainty). reference value (Figure 1a), whereas all the other participants reported systematically production dates that corresponded to a younger age than the known age, i.e. the known time elapsed between the reference value (09/07/2012) and the reference date (06/03/2013). This shift towards younger age might result from an incomplete re- Table 1: Reported production date (as dd/mm/yyyy) and uncertainties in day based on the average measured n( 231 Pa)/n( 235 U) amount ratios on 06/03/2013 with uncertainties as stated by participants (i.e. k=1or k=2) for the measurements of the 20 mg uranium certified sample. X ref 1 Production dates (± day) n( 231 Pa)/n( 235 U) amount ratios 09/07/2012 ±7.8(k =2) L46 11/04/2012 ±22(k =2) (8.86 ± 0.59) (k=2) L52 23/07/2012 ±27(k =1) (6.10 ± 0.30) (k=1) 1 Note that REIMEP-22 is not certified for the production date based on this radiochronometer, but only on the 230 Th/ 234 U radiochronometer. Therefore, the reference value of 09/07/2012 is only given in this figure as indicative value. covery (due to loss) of thorium in the REIMEP-22 samples, prior to the addition of the Th spike in the samples [9]. All participants stated to report the n( 230 Th)/n( 234 U) amount ratios for the reference date of 06/03/2013. Therefore, the analysis of the reported average n( 230 Th)/n( 234 U) ratios with respect to the resulting production dates in Figure 1b should allow to depict any bias in the reported results. A lower n( 230 Th)/n( 234 U) amount ratio should correspond to a younger age (production date after 09/07/2012), whereas a higher n( 230 Th)/n( 234 U) amount ratio would result in an older age of the sample (production date before 09/07/2012). The latter can be observed from the participants results as shown in Figure 1a and 1b. Two participants with laboratory codes L42 and L43 had reported ratios contradicting the associated calculated production dates. When calculating the production date from the average n( 230 Th)/n( 234 U) amountratioasreported by laboratory L43 on the reference date, one would deriveaproductiondateofaboutoneyearbeforethecertified test sample was actually produced. On the other hand, for the participant L42, neither the reported average n( 230 Th)/n( 234 U) amount ratio nor the production date agree. The reported/measured average n( 230 Th)/n( 234 U)

5 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison 829 amount ratio is far too high compared to the amount ratio expected in the REIMEP-22 sample leading to a positive bias of ca. 238% in the resulting age. Whereas the reported production date leads to a negative bias of ca. 49% in the calculated age compared to the known age. It is very likely a calculation error in combination with an inadequate preparation of the sample prior to measurement (e.g. spiking) or insufficient noise or abundance sensitivity correction for 230 Th in the mass spectrometric measurements. However, the participant reported that the 230 Th noise correction was the major contribution to the final uncertainty budget. For the results of the 50 mg uranium certified samples measured by the α-spectrometry (Figure 2a), two participants L58 and L54 reported production dates that agreed within uncertainties with the REIMEP-22 reference value, though L54 reported a larger uncertainty than L58. Laboratory L57 reported a value close to the reference value and standard uncertainty while L59 reported a production date that significantly deviates from the reference value (Figure 2a). The good agreement of the reported average ratios for the A( 230 Th)/A( 234 U) amount ratios with the resulting production dates in Figure 2a and 2b confirmed that all the participants reported correctly their average activity ratios for 06/03/2013. In general, the reported uncertainties for the α-spectrometry are larger than those for the mass spectrometry measurements (Figs. 1 and 2). The relative uncertainties for the mass spectrometric measurements on the 20 mg certified test samples are in the range of 2% to 15%, whereas they are within 8 to 30% in the case of α-spectrometry measurements for the 50 mg certified test samples. The reported production dates for REIMEP-22 (Figure 2a) estimated by α-spectrometry do not appear to display a systematic shift towards younger age as observed for the mass spectrometry results (Figure 1a). In the past, a negative bias inthe α-spectrometry results was observed by Wallenius et al. [15], resulting in younger ages than the known ages. The absence of such bias in the REIMEP- 22 α-spectrometrymaybeduetotheuseofbothtechniques by some participants. Indeed, laboratories L54 and L58 (Figure 2) participated as well in the measurements of the 20 mg sample with mass-spectrometry (identified by L50 and L48 respectively in Figure 1), while laboratory L57 measured the sample with TIMS and α-spectrometry. This could have influenced the way how these participants treated the 50 mg sample prior to α-spectrometry measurements, but, as can be seen from the difference in the reported results of laboratories L54-L50 and L58-L48 in Figure 2-Figure 1, these participants reported independentlytheproductiondatesbasedontheirα-spectrometry or mass spectrometry measurements, respectively. Most REIMEP-22 participants reported uncertainties according to the Guide for Quantifying Measurement Uncertainty (GUM) [11] issued by the International Organization for Standardization as ISO 99:2005. Six laboratories reported expanded uncertainties with a coverage factor k=2and three others reported standard uncertainties (k =1) according to the GUM [11]. One laboratory reported uncertainties with k=2using another standard for the quantification of uncertainty (here the GOST R-ISO ), while another laboratory propagated the analytical uncertainties (with a coverage factor k=2and using a Student factor for the average ratio). As can be seen from Figures 1 and 2 some participants underestimated the uncertainties associated with their measurements and the production date. On the one hand, laboratories L45, L50, L47, and L42 reported results with a significant deviation from the reference value (Figure 1a). Their uncertainties did not reflect these biases and there is no overlap with the certified range. On the other hand, laboratory L52 (Figure 1a and b)reported only a standard uncertainty (k =1) for the measurement of the 20 mg uranium certified test sample. Therefore, although the reported production date did not deviate much from the reference value, the difference is significant. The same can be observed in Figure 2a for the measurement of the 50 mg uranium certified test sample for laboratory L57, which also underestimated the uncertainty by reporting the production date with a standard uncertainty (k =1), instead of reporting the expanded uncertainty with a coverage factor k=2, as it was done for the average A( 230 Th)/A( 234 U) amount ratio. The reporting of the n( 231 Pa)/n( 235 U) amount ratios or A( 231 Pa)/A( 235 U) activity ratios was optional since the REIMEP-22 reference value corresponds to the production date based on the 230 Th/ 234 U radiochronometer and not the 231 Pa/ 235 U radiochronometer. The verification of the completeness of the 231 Pa from its mother 235 U in the material was beyond the scope of the IRMM certification project [10, 14]. Nevertheless, two participants reported the n( 231 Pa)/n( 235 U)amount ratio measured by mass spectrometry (Table 1) and the derived production date [16]. Laboratory L46 reported only one value for one replicate due to technical problems. Laboratory L52, reported an average n( 231 Pa)/n( 235 U) amount ratio with an associated production date, which confirmed the production date of the certified test sample, which is for the 231 Pa/ 235 U radiochronometer only given as indicative value (Table 1). Most of the other participants stated that they had neither experience in protactinium measure-

6 830 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison ments, nor they had an appropriate and validated measurement procedure to measure the n( 231 Pa)/n( 235 U)ratio in the REIMEP-22 samples. As there are only two reported production dates based on the 231 Pa/ 235 U radiochronometer, it is difficult to conclude the REIMEP-22 reference value is also applicable for this radiochronometer. However the result of laboratory L52 is in good agreement with the REIMEP-22 reference value (Table 1), and may indicate that the separation of 231 Pa from its mother 235 U might have been complete. 3.2 Evaluation of laboratory performances The ITWG does not recommend quality goals or performance criteria to the network laboratories that could have been used in REIMEP-22 to assess the laboratories measurement capabilities. It was therefore agreed to evaluate the measurement performance in REIMEP-22 by means of z- and zeta-scores in compliance with ISO [13]. The zeta-scores were calculated for each laboratory results reported with an uncertainty and give an indication of whether the estimate of the uncertainty is consistent with the laboratory s deviation from the reference value as giveninsection2.2[17]. In this paper, the authors suggest 5% of the known age of d of the material on 16/10/2013 as relative standard deviation ( σ) for proficiency assessment [13]. σ was set to 23.2 d. The same criterion was used in REIMEP-22 for the homogeneity assessment [17]. The z-scoreprovides an indication whether a laboratory is able to perform the measurement in accordance with the σ. The scores are expressed as follows: zeta = x lab X ref u 2 ref +u2 lab z= x lab X ref (2) σ where x lab is the result reported by a participant (based on their measurement), X ref is the certified reference value (assigned value), u ref is the standard uncertainty of the reference value and u lab is the standard uncertainty reported by a participant. Both scores can be interpreted as: satisfactory performance for score 2, questionable performance for 2< score 3and unsatisfactory performance for score >3. An unsatisfactory laboratory performance may be caused by an underestimated uncertainty or by a large deviation from the reference value. Since all the laboratories participating in REIMEP-22 provided uncertainties with a coverage factor (k), the standard uncertainty of (1) the laboratory (u lab ) was calculated as the reported uncertainty divided by the coverage factor. Deviation from the reference value and tendencies of laboratories to underestimate their uncertainties are discussed in Figure 3 using the Naji plot as a straightforward graphical tool to evaluate participants results [18]. The two scores are combined as expressed in Eq. (3), which display the participants results/performances by means of z-scores (x-axis) with respect to the acceptable uncertainty (y-axis) in areas delimited by zeta 2and u min u lab u max with u min =u ref, and u max =2 σ. ( z C ) 2 ( u 2 ref σ ) =( u 2 lab σ ) Results fall in the Naji plot within areas defined in Figure 3 by two parabolas (C =2delimits the performance criteriadomainfor zeta =2andC =3the one for zeta =3) delimiting the different performance criteria domains. Results falling in the area delimited by u max and zeta 2 are satisfactory and results in the area delimited by u max and zeta 3are questionable. Figure 3 shows that five results (L43, L46, L48/L58 and L49) fall within the area corresponding to satisfactory performances, i.e. that their reported value and its uncertainty falls well within the range of the acceptable uncertainty based on a σ = 0.05X ref ;eventhoughlaboratory L46 reported an uncertainty that may be underestimated (smaller than u ref,figure3). Two laboratories L52 and L57 reported questionable results. Among the laboratories having reported satisfactory or questionable results, five are part of the ITWG (L48 and L58 represent the same laboratory having measured both REIMEP-22 samples). It is interesting to see that laboratories L49 and L43 considered themselves as not very experienced in Th-U mass spectrometry measurements, yet they reported satisfactory results. Moreover, laboratories L42, L47 and L48 reported that they did not have a routine measurement procedure in place to measure such low amount of Th samples and had to set-up completely new methods to analyse the REIMEP-22 sample(s) with mass spectrometry. The results for laboratory L48 are therefore very encouraging and confirm that the newly developed analytical procedures are suitable for this kind of measurements. For the two other laboratories, the determination of low amount of Th in the certified samples remains an analytical challenge. Concerningtheanalysisofthe50 mg samples, only laboratory L58 reported to be experienced in the radiometry measurements of Th-U samples; experience that is confirmed in Figure 3 by the satisfactory laboratory performance within reported uncertainty. (3)

7 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison 831 Fig. 3: Naji plots of all REIMEP-22 participants results for the measurements of the 20 mg uranium test samples (squares) and of the 50 mg uranium test samples (triangles). The ITWG laboratories are identified by open symbols. Two laboratory results (not ITWG) for the 50 mg uranium test samples fall outside the scale of this plot and are indicated by dashed arrows. While results using α-spectrometry have larger but realistic uncertainties, the majority of the laboratories using mass spectrometry underestimated the uncertainty ( zeta >2inFigure 3). This clearly shows the importance of providing reasonable and realistic measurement uncertainties, which do not have to be necessarily as small as possible. The questionable performance of laboratory L57 indicates that the calculation of realistic uncertainties might be an issue. This laboratory possibly underestimated the uncertainties on the reported ratios or/and production dates, as may have done laboratory L45. In addition, laboratories L46 and L47 reported an expanded combined uncertainty smaller than the expanded uncertainty of the reference value (Figure 3). Five out of the eleven REIMEP-22 participants reported results that met the laboratory performance criteria set by the PT organisers (based on the scores and acceptable uncertainty), by reporting values and expanded uncertainties within the range of the expanded uncertainty of the reference value. 4 Discussions on the participants answers to the questionnaire 4.1 Analytical methods used for REIMEP-22 samples Participants used from 0.1 mg up to almost 6mgof sample for one aliquot/replicate for the mass spectrometry measurements and up to 15 mg of sample for the α-spectrometry measurements. Three laboratories using mass spectrometry technique did not perform any chemical separation prior to measurements. Other participants applied a chemical treatment by dissolving the samples in nitric or hydrochloric acid followed by a separation using TEVA resin or anion exchange (e.g. AG 1-X8). Some participants used co-precipitation with lanthanides to separate Th from U before α-spectrometry measurements. All the participants applied isotope dilution for the determination of the amount or activity of Th and U in the samples. For the mass spectrometry measurements, seven out of the nine participants used Multi-Collector ICP-MS (MC- ICP-MS) for the U/Th measurements while two other laboratories used Sector-Field ICP-MS (SF-ICP-MS). For the measurement of the 50 mg uranium certified samples, all the participants used α-spectrometers. Laboratories L59 and L57 measured Th by α-spectrometry and U by Thermal Ionisation Mass Spectrometry (TIMS). Certified reference materials were used for instrument calibration, mass bias and abundance sensitivity correction. Furthermore, CRMs were used for the quantification of in-house spikes ( 229 Th). 4.2 The components of the uncertainty budgets For the determination of the production date based on the (amount or activity) 230 Th/ 234 U ratio, all participants reported that the major contribution to the final uncer-

8 832 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison tainty came from the thorium determination, whether using mass spectrometry or α-spectrometry. This is mainly due to the low amount of thorium present in the relatively young REIMEP-22 certified test samples. For the mass spectrometry analysis, the low 230 Th signal was often the problem, since it could then be hindered by background or high abundance sensitivity if 232 Th was used as spike for the isotope dilution. Participants reported as well uncertainty related to the calibration of the Th spikes against other reference materials, which is mostly due to the low availability of thorium certified reference materials. For the α-spectrometry measurements, the 230 Th counting statistics as well as the estimation of the detection efficiencies (if no isotope dilution was used) may be the reasons for the higher uncertainties observed for αspectrometry results. For the Pa-U analysis of the REIMEP-22 samples, though no detailed measurement procedure was given, both participants indicated that the separation yield of Pa and the 233 Pa spike calibration were the major contributors to the final uncertainty on the reported values. Since the thorium determination was identified as the major component of the final uncertainty on the production date, we may expect a reduction with ageing of the uncertainty on the 230 Th determination and 230 Th/ 234 U ratio of the REIMEP-22 certified test sample [5]. With increasing amount of 230 Th in the material with time, the activity ratio will tend to unity and the measurements of the 230 Th should be achievable by most of the laboratories. However, other components of uncertainties, such as the half-lives of 234 U or 230 Th, may then become more significant with time [5]. 4.3 Use of half-lives and molar masses REIMEP-22 participants were asked to report the half-lives (in years) and molar masses (gmol 1 ) with associated uncertainties (with a coverage factor k=2)usedintheircalculations. Pommé et al. [5] have shown recently that halflives and their uncertainties play a significant role in the final calculation of the age of the radioactive material and its uncertainty. This is of high importance in nuclear forensic science. Therefore, the REIMEP-22 organisers explicitly asked participants to report as well the bibliographic references for these values to get an overview of the commonly used half-lives within the nuclear forensic community. The values for the half-lives and the expanded uncertainties as reported by the participants [17] indicate that participants in REIMEP-22 used the 234 U half-lives provided by DDEP-BIPM [19] andbychengetal.[20], while more references were cited for the 230 Th half-life. There seems to be a possible misuse of the uncertainties associated with these values. Laboratories L48, L50 and L57 reported the same half-life values as given in DDEP-BIPM [19], but L48 reported associated uncertainties that were twice the uncertainty value reported by the two other laboratories [17]. Similarly, L46, L49, L42 and L47 reported similar half-life values as those given in Cheng et al. [20], while L42 reported uncertainties based on [21] that were twice the uncertainty value reported by the three other participants [17]. For the values published in [19], it appears that the expanded uncertainties as reported by L48 are correct, i.e. with k=2, so that the two other laboratories reported in reality standard uncertainties with k=1. On the other hand, L42 may have considered that the uncertainties for the 230 Th and 234 U half-lives provided in [20, 21] werestandard uncertainties instead of expanded uncertainties with k=2. Therefore, the correct half-lives and their respective expanded uncertainties are summarised in Table 2. More variations in the reported values were observed for molar masses than for half-lives, even though it has less influence on the final results [17]. Even if most of the reported values agreed well on the three first digits, the last significant digits were very different. Among the few reported uncertainties associated with molar masses, L48 reported an uncertainty for the 230 Th molar mass twice that of L50 and more than twice that of L47 [17]. Finally, the correct molar masses and uncertainties for M( 234 U) and M( 230 Th) in gmol 1 are those reported by laboratory L48, based on Audi et al. [22] and are respectively: ± and ± (k =2). This study clearly indicates that harmonisation of the half-life and molar mass values from bibliographic sources/nuclear data references within the nuclear forensic community is necessary for a more accurate and robust determination of the age (production date) of a uranium sample and of its associated uncertainty. The references provided in this paper shall be used as a starting point towards the harmonisation of these values. The nuclear forensics community shall participate in or even initiate Table 2: Half-lives (in years) and expanded uncertainties (with k=2) as from literature. References T 1/2 ( 234 U) T 1/2 ( 230 Th) DDEP-BIPM [19] ± ± 600 Cheng, H., et al. [20] ± ± 230

9 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison 833 the evaluation of the existing half-life values, especially when new values are published. For instance, Cheng et al. [23] recently published updated 234 U and 230 Th half-life values (as ±260 and ±110 a), to replace the previous ones ([20],see Table 2), improved due the use of new gravimetrically prepared reference materials for mass spectrometric measurements and a justified new choice of secular equilibrium materials. The new 234 U half-life value closely approaches the DDEP-BIPM value ([19], see Table 2) and has a much smaller uncertainty. However, the new 234 U and 230 Th half-life values could only be derived by using the old measurement of the 238 U half-life by Jaffey et al. [24], the uncertainty of which might be underestimated. The value by Jaffey et al. shall either be combined with a more realistic commonly agreed uncertainty, or new measurements for the 238 U half-life shall be performed (on-going at IRMM and elsewhere). Subsequently by using the new equilibrium 234 U/ 238 U and 232 Th/ 238 U values provided by [23], the 232 Th and 234 U half-lives could be re-calculated with improved uncertainties. As a conclusion, a critical evaluation and harmonisation of published half-life values used for nuclear forensics shall be performed at international level by suitable organisations like BIPM, the IAEA, or dedicated nuclear forensics organisations, such as the ITWG. 5 Conclusions and outlook REIMEP-22 offered auniqueopportunity to laboratories involved in nuclear forensic to demonstrate that their measurement results for the characterisation of the age of uranium materials are fit for the intended purpose and within the required measurement uncertainties for stateof-practice sample analysis. The challenge for REIMEP-22 participants was to measure 230 Th/ 234 U in a young uranium sample containing a low amount of thorium. Most of the participants in REIMEP-22 performed well using mass spectrometry or αspectrometry. The spread of results was larger for the measurements performed by α-spectrometry than those performed by mass spectrometry. REIMEP-22 confirmed the quality of the analytical capabilities of the participating laboratories to determine the production date of similar intercepted uranium materials. However there is room for improvement in the estimation and reporting of measurement uncertainties. Moreover, discrepancies were identified among participants using different half-lives and molar masses and the corresponding uncertainties from bibliographic references. Attention should be brought to the reporting of results and determination of the associated uncertainties. Proper harmonisation of nuclear reference data, such as half-lives, should be ensured within the nuclear forensics community. The additional results reported by two of the participants indicate that the separation of 231 Pa from its mother 235 U might be complete in the certified test sample and that the 231 Pa/ 235 U radiochronometer could be used to determine the production date. This needs to be further confirmed by additional measurements. Acknowledgement: The authors are very grateful to Adrian Nicholl and Judit Krajko (both at EC-JRC-ITU) and Monika Sturm (former EC-JRC-IRMM and now at IAEA- SGAS) for their technical help with the preparation and certification of the certified test samples. The authors would like to thank the EC-JRC-IRMM colleagues: Theo Droogmans, Frederik Van der Straat and Andreas Fessler and Marleen Peetermans who helped with the logistics of REIMEP-22. The authors are also very thankful to the internal reviewer at EC-JRC-IRMM, Piotr Robouch and an anonymous peer-reviewer for their helpful and constructive comments. References 1. Mayer, K., Wallenius, M., Varga, Z. (2013). Nuclear Forensic Science: Correlating Measurable Material Parameters to the History of Nuclear Material. Chem Rev 113: Varga, Z., Suranyi, G. (2007). Production date determination of uranium-oxide materials by inductively coupled plasma mass spectrometry. Anal Chim Acta 599: Wallenius, M., Mayer, K. (2000). Age determination of plutonium material in nuclear forensics by thermal ionisation mass spectrometry. Fresenius J Anal Chem 366: Morgenstern, A., Apostolidis, C., Mayer, K. (2002). Age Determination of Highly Enriched Uranium: Separation and Analysis of 231 Pa. Anal Chem 74: Pommé, S., Jerome, S. M., Venchiarutti, C. (2014). Uncertainty propagation in nuclear forensics. J Appl Radiat Isotopes 89C: EC-JRC-IRMM Webpage for REIMEP Inter-Laboratory Comparisons ISO/IEC 17043:2010 (2010). Conformity assessment General requirements for proficiency testing. 8. ISO Guide 34:2009 (2009). General requirements for the competence of reference material producers. 9. Varga, Z., Nicholl, A., Wallenius, M., Mayer, K. (2012). Development and validation of a methodology for uranium radiochronometry reference material preparation. Analytica Chimica Acta 718: Varga, Z., Venchiarutti, C., Nicholl, A., Krajkó, J., Jakopič, R., Mayer, K., Richter, S., Aregbe, Y. (2015). IRMM-1000a and IRMM-1000b uranium reference materials certified for the

10 834 C. Venchiarutti et al., REIMEP-22 inter-laboratory comparison production date Part I: Methodology, preparation and reference value. Journal of Radioanalytical and Nuclear Chemistry. DOI /s x. 11. ISO (2005). Guide to the Expression of Uncertainty in Measurement ISO Guide 35:2006 (2006). Reference materials General and statistical principles for certification. 13. ISO 13528:2005 (2005). Statistical methods for use in proficiency testing by inter-laboratory comparisons. 14. Venchiarutti, C., Varga, Z., Richter, S., Nicholl, A., Krajkó, J., Jakopič, R., Mayer, K., Aregbe, Y. (2015). IRMM-1000a and IRMM-1000b: uranium reference materials certified for the production date based on the 230 Th/ 234 U radiochronometer Part II: Certification. Journal of Radioanalytical and Nuclear Chemistry. DOI /s y. 15. Wallenius, M., Morgenstern, A., Apostolidis, C., Mayer, K. (2002). Determination of the age of highly enriched uranium. AnalBioanalChem374: Eppich, G., et al. (2013). 235 U- 231 Pa age dating of uranium materials for nuclear forensic investigations. J Anal At Spectrom 28: Venchiarutti, C., Varga, Z., Richter, S., Nicholl, A., Krajkó, J., Jakopič, R., Mayer, K., Aregbe, Y. (2015) REIMEP-22 U Age Dating Determination of the production date of a uranium certified test sample. JRC Science and Policy report. EUR DOI / interlaboratory-comparisons/reimep. 18. Robouch, P., Younes, N., Vermaercke, P. (2003). PTB IT-10, pp DDEP-BIPM ( ). Table of radionuclides. Monographie BIPM-5. Retrieved from DDEPdata.htm Cheng, H., Edwards, R. L., Hoff, J., Gallup, C. D., Richards, D. A., Asmerom, Y. (2000). The half-lives of uranium-234 and thorium-230. Chemical Geology 169: Nucleonica , developed under a License of the European Atomic Energy. (European Commission) nucleonica.net/. 22. Audi, G., Wapstra, A. H., Thibault, C. (2003). The AME2003 atomic mass evaluation (II). Tables, graphs and references. Nucl Phys A 729: Cheng, H., Edwards, L., Shen, C.-C., Polyak, V. (2013). Improvements in 230 Th dating, 230 Th and 234 U half-life values, and U-Th isotopic measurements by multi-collector inductively coupled plasma mass spectrometry. Earth Planet Sci Lett , Jaffey, A. H., Flynn, K., F Glendenin, L. E., Bentley, C., Essling, A. M. (1971). Precision Measurement of Half-Lives and Specific Activities of 235 U and 238 U. Phys Rev C 4(5),