Introduction

The nuclear forensics methods aim at providing hints on the intended use, origin, production time and history of nuclear and other radioactive materials [1,2,3]. The scientific results, obtained in a timely manner, support the nuclear forensic findings and may serve law enforcement as investigative leads or as evidence. Several characteristic parameters (signatures) such as physical dimensions of the material, isotopic composition of U and Pu, impurities or production date can be used to re-establish the material history, hence link the material in question to a production process or even a facility. Moreover, traditional forensic evidence associated with the material may help to identify individuals who handled the material [1, 4,5,6,7,8,9]. Determination of the production date (age dating) is based on the radioactive decay of the material and the measurement of the formed daughter products relative to the parent nuclides. The (model) age of the material is a prominent signature as it is a so-called predictive signature and does not require any reference information [10, 11]. Uranium age dating is typically carried out using the decay of 234U to 230Th, achieving the quantification of both nuclides by isotope dilution mass spectrometry. Age dating measurements, however, require typically milligram amounts of material and a tedious separation due to the low abundance of the decay products [3, 12]. Using lower sample amount can result in high uncertainties due to the small quantity of the daughter nuclides and lower measurement precision, e.g. due to the higher contribution from the background. Age dating of U particles by secondary ion mass spectrometry (SIMS) would require relatively large particles (micrometer-sized) of an old material [13].

In June 2011 the Moldavian Police arrested six suspects alleged to sell kilograms of highly enriched uranium (HEU) in Chisinau, Moldova. During the arrest a glass vial, containing 4.4 g of uranium, was found (Fig. 1a). The sample was analyzed by portable gamma spectrometry (ORTEC, Micro-Detective) in the National Agency for Regulation of Nuclear and Radiological Activities (NARNRA) in Moldova. The result indicated that the sample had a 235U enrichment of more than 72%. JRC-Karlsruhe obtained a sample (a piece of paper that had been in contact with the HEU powder; referred later as a “swipe”) from the confiscated material for nuclear forensic analysis in 2017 (Fig. 1b). The HEU powder was put on a paper swipe without further alteration.

Fig. 1
figure 1

The confiscated HEU sample in Moldova in 2011 (a), a swipe sample received in JRC-Karlsruhe in 2017 (b)

Full size image

As the sample amount was very low (the material was not even visible on the swipe) and the particles were small (based on the scanning electron microscopy their size was between 100 and 200 nm), only a limited number of analysis could be performed. It was decided to look into the production date (age) of the material besides the determination of the isotopic composition of U, which was performed by Large-Geometry SIMS (LG-SIMS) as a primary technique for particle analysis [14]. Due to the small size of the particles production date measurement could not be performed by LG-SIMS. Thus, a new method had to be developed, which combined the measurement of “bulk” (i.e. the average of all the particles) U isotopic composition and the production date determination of nanogram amount of a uranium sample by multi-collector inductively coupled plasma mass spectrometry (MC-ICP-MS).

Theory of direct production date measurement

The 230Th and 234U signals are clearly visible in the MC-ICP-MS spectrum for highly enriched uranium (Fig. 2) without chemical separation, even though the lower abundant 230Th is at μg g−1 level relative to the total U and no Th and U separation was done. This is due to the fact that the “minor” 234U abundance is high in HEU in contrast to natural or low-enriched U samples. However, the quantification of the 230Th/234U ratio is difficult due to the different ionisation efficiencies of Th and U, the matrix effect and without using isotope dilution (i.e. use of 229Th or 233U as a spike).

Fig. 2
figure 2

Simultaneously detected signals of 230Th (ion counter) and 234U (Faraday detector) by MC-ICP-MS using 300 ng g−1 of CRM U630

Full size image

The relation between the 234U decaying to 230Th can be expressed with the radioactive decay rules (Bateman equation):

$$\frac{{N_{{ 2 3 0_{\text{Th}} }} }}{{N_{{ 2 3 4_{\text{U}} }} }} = \frac{{\lambda_{{ 2 3 4_{\text{U}} }} }}{{\lambda_{{ 2 3 0_{\text{Th}} }} - \lambda_{{ 2 3 4_{\text{U}} }} }}\left( {e^{{ - \lambda_{{ 2 3 4_{\text{U}} }} t}} - e^{{ - \lambda_{{ 2 3 0_{\text{Th}} }} t}} } \right) + \frac{{N_{{ 2 3 0_{\text{Th}} }}^{0} }}{{N_{{ 2 3 4_{\text{U}} }} }}e^{{ - \lambda_{{ 2 3 0_{\text{Th}} }} t}}$$
(1)

where the N refers to the amount of the nuclides, \({\lambda}\) is the decay constant. Assuming that the U purification resulted in a complete separation of the 230Th decay product, in consequence, there is no 230Th in the sample at time zero:

$$\frac{{N_{{ 2 3 0_{\text{Th}} }} }}{{N_{{ 2 3 4_{\text{U}} }} }} = \frac{{\lambda_{{ 2 3 4_{\text{U}} }} }}{{\lambda_{{ 2 3 0_{\text{Th}} }} - \lambda_{{ 2 3 4_{\text{U}} }} }}\left( {e^{{ - \lambda_{{ 2 3 4_{\text{U}} }} t}} - e^{{ - \lambda_{{ 2 3 0_{\text{Th}} }} t}} } \right)$$
(2)

As the decay constants are small, using \(e^{x} \approx 1 + x\) approximation if x ≪ 1 (Taylor-series), one can re-write the equation:

$$\frac{{N_{{ 2 3 0_{\text{Th}} }} }}{{N_{{ 2 3 4_{\text{U}} }} }} = \frac{{\lambda_{{ 2 3 4_{\text{U}} }} }}{{\lambda_{{ 2 3 0_{\text{Th}} }} - \lambda_{{ 2 3 4_{\text{U}} }} }}\left( {\lambda_{{ 2 3 0_{\text{Th}} }} - \lambda_{{ 2 3 4_{\text{U}} }} } \right)t \approx \lambda_{{ 2 3 4_{\text{U}} }} t$$
(3)

For a 20-year-old sample, the simplification results in − 0.13% bias. The equation implies that the ingrowth of 230Th is approximately linear and depends on the decay constant of 234U. Thus, measuring the 230Th/234U ratio the production date can be calculated easily. The directly obtained 230Th/234U intensity ratio measured by MC-ICP-MS, however, is not equal to the “true” 230Th and 234U ratio in the sample due to U/Th ionisation difference, the mass bias effects and relative detector efficiencies. Within one sequence these values can be considered constant, the measured 230Th/234U intensity ratio thus being proportional to the 230Th/234U amount ratio in the sample:

$$\frac{{N_{{ 2 3 0_{\text{Th}} }} }}{{N_{{ 2 3 4_{\text{U}} }} }} \approx \lambda_{{ 2 3 4_{\text{U}} }} t = f\frac{{I_{{ 2 3 0_{\text{Th}} }} }}{{I_{{ 2 3 4_{\text{U}} }} }}$$
(4)

where ITh-230 and IU-234 are the measured intensities by MC-ICP-MS for 230Th and 234U, respectively. Merging all the constants together, the measured 230Th/234U intensity ratio is proportional to the production date (time elapsed since the last chemical purification):

$$\frac{{I_{{ 2 3 0_{\text{Th}} }} }}{{I_{{ 2 3 4_{\text{U}} }} }} = f^{\prime } t$$
(5)

where f′ constant includes Th/U ionisation difference, detector relative efficiencies, mass bias of 230Th and 234U and the 234U decay constant. Thus, the f′ constant can be calculated using a reference material with certified production date (in our case CRM U630 was used) after the measurement of the 230Th/234U intensity ratio. This f′ value can be then applied to the unknown sample to calculate the production date after the measurement of its 230Th/234U intensity ratio.

The developed method is applicable for highly enriched uranium (i.e. containing relatively high amount of 234U), which is reasonably pure from other elements. For instance, “dirty” swipe samples can contain high amounts of Pb, e.g. from lead shielding, and due to the possible molecular interferences (e.g. 204Pb12C14N+) erroneous results can be obtained. This analysed sample contained HEU particles on a piece of paper, which had been in touch only with the bulk HEU powder. Therefore the blank level was expected to be low.

In addition to that, the method is obviously applicable only for HEU with single U composition and not for mixtures containing e.g. natural uranium. This sample was analysed prior to the age determination by SIMS, which showed that the HEU particles had all the same isotopic composition and no other uranium particles, with different composition, were detected.

Experimental

Reagents and materials

All labware was thoroughly cleaned before use. Suprapur grade nitric acid (Merck, Darmstadt, Germany) was used for the sample preparation, which was further purified by sub-boiling distillation (AHF analysentechnik AG, Tübingen, Germany). For dilutions ultrapure water was used (Elga LabWater, Celle, Germany). Perfluoralkoxy (PFA) vials with a diameter of 22 mm (volume: 3 mL) were purchased from AHF analysentechnik AG (Tübingen, Germany). Polished pyrolytic graphite planchets were obtained from ANAME Instrumentación Científica (Madrid, Spain).

Instrumentation

A NuPlasma™ (NU Instruments, Wrexham, United Kingdom) double-focusing multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS), equipped with 11 Faraday detectors and 3 discrete dynode electrode multipliers was used for the U isotope abundance and age dating measurements. The resistors fitted to the Faraday detectors were 1011 Ω. The instrument was operated in low mass resolution mode (R = 300). The samples were introduced using an Aridus II desolvation unit (CETAC Technologies Inc., Omaha, NE, USA) as a solution. Detailed optimised instrumental settings and data acquisition parameters are given in Table 1.

Table 1 MC-ICP-MS operating parameters
Full size table

The ICP-MS was optimised daily (torch position, gas flows, voltages) using a 50 ng g−1 multi-elemental solution (Inorganic Ventures, Christiansburg, USA). The optimisation aimed at achieving highest sensitivity and stability of the acquired U signal. The 234U signal was aimed to about 1 V intensity for the age dating measurement and only the 230Th and 234U signals were measured.

For the U isotopic composition determination the isotopes 234U, 235U, 236U and 238U were measured on the Faraday detectors, while for the age dating 230Th was measured on the ion counter equipped with a retardation filter and 234U was measured simultaneously on a Faraday detector. The retardation filter improves the abundance sensitivity on m/z = 230 by a factor of ~ 10. The measurement of the age dating was done in one sequence to make sure that Th and U sensitivities, detector efficiencies and mass bias (i.e. f′ in Eq. 5) were constant. Calibrant to calculate the f′ was measured before each sample. The MC-ICP-MS measurement was performed in the following order: Calibrant—Quality control—Calibrant—Sample—Calibrant—Quality control. Dilute HNO3/HF mixture between the samples was used to remove the Th and U traces.

Measured U samples

The used certified reference materials (CRMs), U500, U850 and CRM U630, are U standard reference materials in the form of U3O8 (New Brunswick Laboratory Argonne, IL, USA) and they have 50, 85 and 63% nominal enrichment of 235U, respectively. All CRMs are certified for their U isotopic composition, while the CRM U630 is certified for model purification date as well (certified model date: 6 June, 1989 with an uncertainty of 190 days). For the U isotopic measurements the mass bias was determined using the U500 CRM whereas for the quality control (to check the U isotopic abundances) U850 CRM was used. For the production date measurement the 230Th/234U amount ratio and the age was calculated after calibrating with the CRM U630 radiochronometric standard. For quality control purpose a 70% highly enriched uranium (HEU-70) was used (known production date: 19 July, 2011), which has been prepared in the JRC-Karlsruhe, afterwards analysed and validated by several international laboratories [10]. The concentration of the samples for age measurement was approximately 250 ng g−1 U.

Sample preparation

As the U amount in the sample was limited, chemical separation or spiking was not possible. After careful considerations and performing tests using a blank and standard, the following method was chosen: the particles on the swipe sample were collected on a pyrolytic graphite planchet using the vacuum impactor method in a clean laboratory as done for SIMS analysis routinely [14,15,16]. The sample was vacuumed along the whole swipe to collect as many particles as possible. For fixing the particles on the graphite planchet, polyisobutylene (PIB) was added. The planchet was baked after collection to remove the PIB to avoid any interference during the MC-ICP-MS measurement. The U particles were then leached out from the planchet and dissolved in ccHNO3 using a specially developed procedure to minimize the background and consumption of chemicals (Fig. 3). The graphite planchet was placed and fixed on the top of a PFA vial containing 3 mL of ccHNO3 facing the liquid phase. The planchet was not immersed in ccHNO3 to minimize the background and avoid graphite residues. Covering the PFA vial with the graphite planchet separated the leachate from the atmospheric environment. The diameter of the PFA vial was chosen in such a way that its diameter was slightly larger than the center (i.e. the U particles containing) part of the planchet. The vial was placed on a hot plate for 24 h to make sure that all U was leached out from the planchet and was got in the liquid phase. The temperature of the hot plate was slightly below the boiling point of the HNO3, thus it resulted in reflux, which dissolved the particles on the planchet and the nitric acid dropped back in the vial. As U is not volatile, the U solution remained in the PFA container. A blank was prepared together with the sample. The dissolution was tested beforehand with particles of known U isotopic composition (i.e. U010 CRM) to optimize the leaching time, to measure the background and check the leaching efficiency. Tests showed that more than 90% of the U could be recovered with the above described method, while the U background was approximately 17 pg. Lower recoveries were obtained if the amount of ccHNO3 was lower, reflux time was insufficient or the temperature was low. During testing the isotopic composition of the leached U010 CRM was measured by MC-ICP-MS and compared to SIMS results on individual particles. The results agreed well, thus no cross-contamination took place during the sample preparation procedure.

Fig. 3
figure 3

U leaching from the pyrolytic graphite planchet

Full size image

The same procedure was applied to the “real” sample. It was evaporated and taken up in 1.5 mL 4% HNO3. About 390 ng U could be recovered from the sample. The sample was split in two fractions in order to measure the “bulk” (i.e. the average of the particles) U isotopic composition (about 10% of the aliquot) and the production date (about 90% of the aliquot) by MC-ICP-MS.

Results and discussion

Minimum U amount for age dating

Using U630 one can estimate the minimum amount needed for the age dating using the direct measurement by MC-ICP-MS (Fig. 4). Measuring increasing concentration of U, the required U amount for an accurate and precise measurement can be determined.

Fig. 4
figure 4

Age dating results with expanded uncertainties (k = 2) using a CRM U630 by MC-ICP-MS with increasing U concentration. The solid line is the certified model purification age (29.01 ± 0.52 years). Reference date: 28 June, 2018

Full size image

Using about 300 ng of U (with about 0.15 pg 230Th) one can get an age result, which agrees with the model purification date and which is precise enough. Above this U amount the precision of the measurement, which is about 6% relative to the measured value, will not increase significantly.

Results of the swipe sample

“Bulk” U isotopic composition

The U isotopic composition of the swipe sample is given in Table 2. The MC-ICP-MS results for the U “bulk” isotopic composition agreed well with the LG-SIMS results (average 235U = 72.6%) performed on 41 individual particles. This indicated that no cross-contamination had taken place during the sample preparation. The blank was negligible.

Table 2 “Bulk” U isotopic composition of the swipe sample by MC-ICP-MS
Full size table

Age dating result

The measured production date for the sample together with the HEU-70 results is summarized in Table 3. The reported uncertainties are expressed as expanded uncertainties with a coverage factor of k = 2. The half-lives of 234U and 230Th applied are 245,250 ± 490 years and 75,690 ± 230 years, respectively. The uncertainties took into account the calibrant model age uncertainty as well as the measured calibrant and sample signal uncertainties. The f′ values calculated from the CRM U630, which were measured always right before the samples (HEU-70 or the swipe sample), were stable throughout the measurement sequence, and showed no change during the analysis. The results obtained for the HEU-70, before and after the swipe sample measurement, (September 2011 and June 2011, respectively) agreed well with the known production date of 19 July, 2011 [10]. The individual results are shown in the Supplementary Information.

Table 3 Age dating measurement sequence and obtained results
Full size table

The age dating result for the swipe sample is December 1992 with an uncertainty of 1.0 years. The uncertainty is mainly determined by the 230Th/234U ratio measurements and CRM U630 model purification date uncertainty making up > 95% of the overall uncertainty, other contributions (e.g. from the half-life uncertainties) are minimal.

Conclusions

A novel sample preparation and measurement procedure was developed and validated for the “bulk” U isotopic composition and age determination from a swipe sample containing only nanogram amounts of U by MC-ICP-MS. The U isotopic information could be used to complement the LG-SIMS results performed on individual U particles. Using these two main characteristics, i.e. isotopic composition and age of uranium, one could link the HEU seized in Moldova to two other similar illicit trafficking cases, which had taken place around 10 years earlier in Bulgaria [17] and in France [18]. First of all, the packaging of the sample in all cases was very similar: cylindrical lead container lined with yellow paraffin wax and the HEU powder was inside a flamed-sealed glass ampoule. The U isotopic composition in the Bulgarian case was 234U: 1.2%, 235U: 72.7%, 236U: 12.1% and 238U: 14.0% (all values are atom- %, uncertainties are less than 0.5%) [17], while in the French seizure the U isotopic composition was 234U: 1.17 ± 0.02, 235U: 72.57 ± 0.86, 236U: 12.15 ± 0.14 and 238U: 14.11 ± 0.08 [18]. Therefore, both the Bulgarian and French HEU cases agree well with the HEU seized in Moldova. The age dating results differed about one year between the Bulgarian and French cases being 30 October, 1993 ± 50 days, and November 1994 ± 100 days, respectively [19]. Thus, the age dating result for the HEU found in Moldova is in agreement with the Bulgarian seizure. In conclusion, the HEU material found in three different European countries within the time span of 12 years is very likely coming from the same source material. Whether the HEU in different cases originates from the same batch, i.e. was purified at the same time, can be questioned. However, the link between the cases is evident.