Michael Krachler*,
Maria Wallenius,
Adrian Nicholl and
Klaus Mayer
European Commission, Joint Research Centre (JRC), Directorate for Nuclear Safety and Security, P. O. Box 2340, D-76125 Karlsruhe, Germany. E-mail: michael.krachler@ec.europa.eu
First published on 27th April 2020
Laser ablation multi-collector inductively coupled plasma mass spectrometry (LA-MC-ICP-MS) was applied to the detailed investigation of the uranium (U) isotopic composition (234U, 235U, 236U, and 238U) of five contaminated scrap metal samples found within the European Union. Pressed pellets of the two certified U isotopic reference materials CRM U-020 and CRM U-030 were included in the measurement protocol for mass bias correction, calculation of the ion counter gains and for quality assurance. Since the investigated samples had low U content (0.15–14.3 wt%) compared to typically analysed pure U compounds (>60 wt%), the applied experimental parameters had to be adjusted. Spatially-resolved U isotopic information was obtained by line scan analysis of each sample. While other analytical techniques used typically in nuclear forensic investigations, such as γ-spectrometry and thermal ionisation mass spectrometry (TIMS) yielded average U isotopic compositions of the entire sample, LA-MC-ICP-MS provided substantial added value, highlighting the inhomogeneous distribution of U isotopes within various scrap metal samples. Analysis of individual particles via secondary ion mass spectrometry (SIMS) confirmed the large range of 235U enrichment levels in heterogeneous scrap metal samples. Four out of five scrap metal samples contained 236U (∼0.05–∼0.11 wt%), indicating the presence of reprocessed U. Taken together, LA-MC-ICP-MS analysis provided fast and accurate spatially-resolved U isotopic information without consuming or altering the scrap metal samples, a key feature for nuclear forensics investigations.
Scrap metal is an important source material for the metal production industry, contributing a large fraction of the final product (in the case of steel, about 50%). Large scrapyards handle some ten million tonnes of scrap metal each year.4 The number of metal works and foundries worldwide that buy scrap to melt and refine or cast to shape is in the tens of thousands.5 Furthermore, there is substantial transboundary movement of scrap metal and other products of the metal recycling and production industries.6 As a consequence, radioactive material – being either carelessly or intentionally disposed of in an unlawful way and mixed with scrap metal – may inadvertently be transported across international borders. The number of such incidents involves up to 200 items per years in incoming loads of scrap in a single scrapyard.7 Given that radioactive sources in scrap metal have become a growing threat to the recycling industry, the International Atomic Energy Agency (IAEA) has issued guidelines for the scrap industry.4
Besides scrap metal, many nuclear forensic investigations frequently involve nuclear fuel cycle material such as uranium dioxide (UO2) pellets, U ores, U ore concentrates (UOCs or yellow cakes) or powders containing high enriched U (HEU).1,8–12 Important measurable material properties – referred to as “signatures” – include physical dimensions, chemical impurities, microstructure, and age of the material.1–3,11 In this context, the accurate and precise determination of the U isotopic composition of seized nuclear materials is a key signature. The abundance of 235U indicates whether the material contains depleted, low enriched U (LEU), HEU or natural U helping to identify its intended use. Because natural U does not comprise noteworthy amounts of 236U, the apparent presence of this U isotope indicates the exposure of the nuclear material to neutrons, i.e. its prior irradiation in a nuclear reactor.1
For the determination of the U isotopic composition of nuclear samples both non-destructive and destructive analytical procedures are routinely employed. Non-destructive γ-spectrometry analysis typically provides the first assessment of the U enrichment without altering or consuming the nuclear forensic evidence.13 The U isotopic information obtained using γ-spectrometry, however, is an average value and no information on the spatial distribution of the U isotopes within a solid sample can be acquired. At the other end of the analytical toolbox used in nuclear forensic analysis of solid samples is secondary ion mass spectrometry (SIMS), which can be used for single U particle measurements.14 To this end, items are swiped with dedicated cotton swipes removing thousands of particles having diameters typically ranging from <1 to 100 μm from their surface.
For more precise U isotopic analysis, destructive analytical techniques such as single or multiple collector (MC) inductively coupled plasma mass spectrometry (ICP-MS),9,10,15 thermal ionization mass spectrometry (TIMS),15 or even high resolution ICP-optical emission spectrometry (OES),16 are required. For the analysis of selected nuclear forensic samples having a very low abundance of 236U such as U ores or U ore concentrates (UOCs, yellow cakes), the use of accelerator mass spectrometry (AMS) can be advantageous.12 While the above mentioned solution-based, destructive methods allow the accurate determination of the U isotopic abundance, they also lack spatial information and only yield average values.
The knowledge of the spatial distribution of the abundance of the U isotopes of a seized solid sample provides added value for nuclear forensic investigations.17,18 Detailed knowledge of this characteristic gives insights into the U isotopic homogeneity of the material in question and allows drawing conclusions on the process of its production, e.g. powder blending or co-precipitation. Laser ablation (LA) combined with ICP-MS not only offers access to horizontal spatially-resolved U isotopic information of solid materials,17,18 but also to vertical spatially-resolved data, i.e. depth profiling.19 So far, however, only U bearing materials containing U as the matrix element such as U dioxide (UO2) pellets or UO2 single crystals were investigated for this characteristic.17–19
This study explored the potential of LA-MC-ICP-MS for the spatially-resolved U isotopic analysis (234U, 235U, 236U, and 238U) of contaminated scrap metal samples facing their low U concentrations and large elemental heterogeneity. Comparing LA-MC-ICP-MS data to corresponding results of bulk analysis using both γ-spectrometry and TIMS as well as to results of single U particle analysis using SIMS clearly demonstrated the added value of LA-MC-ICP-MS providing valuable information on such inhomogeneous samples.
Microscopic portions of solid samples were introduced into the plasma of the MC-ICP-MS via a ns-laser ablation (LA) system (ESI Lasers, Bozeman, MT, USA) operated at a wavelength of 213 nm. Essential operating parameters of the entire LA-MC-ICP-MS set-up were reported earlier.17,18,20
Prior to the measurement of the quality control sample (CRM U-030) or each actual sample, CRM U-020 was analysed, i.e. applying the standard-sample bracketing technique. This way, the mass bias correction factor as well as the efficiency of the ion counters were determined frequently in order to detect any possible drifting of the response in a timely manner.18
Measured | Certified | |
---|---|---|
n(234U)/n(238U) | 0.0001945 ± 0.0000012 | 0.0001960 ± 0.0000010 |
n(235U)/n(238U) | 0.031474 ± 0.000021 | 0.031430 ± 0.000031 |
n(236U)/n(238U) | 0.0002065 ± 0.0000060 | 0.0002105 ± 0.0000010 |
The uneven distribution of U within the contaminated scrap metal samples lead to higher signal variability in LA-MC-ICP-MS analysis as compared to homogeneous specimens, including the above mentioned certified reference material.17,18 The relatively small dynamic range of the linear response of the Faraday detectors (∼10 mV to 10 V) used to measure 235U and 238U abundances in the samples further contributed to the actual complexity because signal intensities of 235U and 238U needed to be kept within these limits. On top of that, the average U content of the investigated samples was low, ranging from 0.15 wt% in scrap metal 4 to 14.3 wt% in scrap metal 3 which is much lower than that of UO2 materials (>88 wt%), for example. Concurrently, the presence of large amounts of other elements such as Al, Be, Ca, Co, Cr, Fe, K, Mg, Ni, Ti in the scrap metal samples leads potentially to matrix effects.
The LA parameters (fluence, laser spot size) were optimised for every investigated sample to obtain high ion currents on the detectors of the MC-ICP-MS. Therefore, the laser fluence and/or the nominal diameter of the circular laser beam were increased to compensate for lower U content in a scrap metal sample, whenever necessary. This was, however, only possible within certain limits: if sample fragments were getting too small in size and too light in weight, application of laser powers higher than ∼10 J cm−2 caused movement of sample fragments within the LA chamber. Eventually, line scan analysis employing laser fluence exceeding 10 J cm−2 was impossible. The negative impact of such comparatively high laser energy on tiny samples would have required their fixation in the LA chamber. However, as the sample fragments were typically very porous, this feature did not favour fixing specimens with a glue or resin, for example. Additionally, fixation of specimens would have largely complicated their use for further nuclear forensic analysis.
Consequently, laser power was kept below 10 J cm−2 in such cases. Instead, the diameter of the circular laser beam was increased from 5 μm up to 20 μm in order to ablate sufficiently large amounts of material resulting in sufficiently high ion currents for statistically meaningful measurements. Simultaneously, the speed of sample translocation was increased from 20 μm s−1 to 80 μm s−1 to avoid overlaps between individual laser shots when scanning the sample surface with a frequency of 4 Hz.
LA-MC-ICP-MS | Data points | ||
---|---|---|---|
Average ± σ | Median | ||
Scrap metal 1 | |||
234U | 0.161 ± 0.109 | 0.132 | 3495 |
235U | 27.6 ± 15.0 | 24.4 | |
236U | 0.0535 ± 0.0380 | 0.0442 | |
238U | 72.3 ± 15.0 | 75.3 | |
Scrap metal 2 | |||
234U | 0.210 ± 0.184 | 0.161 | 4772 |
235U | 37.7 ± 20.8 | 34.7 | |
236U | 0.0723 ± 0.0782 | 0.0519 | |
238U | 62.0 ± 20.9 | 65.0 | |
Scrap metal 3 | |||
234U | 0.312 ± 0.111 | 0.312 | 4588 |
235U | 34.5 ± 9.73 | 35.2 | |
236U | 0.114 ± 0.044 | 0.112 | |
238U | 65.1 ± 9.80 | 64.3 | |
Scrap metal 4 | |||
234U | 0.128 ± 0.049 | 0.124 | 2126 |
235U | 16.5 ± 0.98 | 16.6 | |
236U | 0.114 ± 0.045 | 0.109 | |
238U | 83.3 ± 0.99 | 83.2 | |
Scrap metal 5 | |||
234U | 0.0053 ± 0.0015 | 0.0052 | 3789 |
235U | 0.714 ± 0.012 | 0.713 | |
236U | <0.001 | <0.001 | |
238U | 99.28 ± 0.01 | 99.28 |
The inhomogeneity in the U isotopic composition is illustrated in detail in Fig. 2. The point-by-point line scan analysis of scrap metal 3 reveals the spatially-resolved heterogeneity of its n(235U)/n(238U) amount ratio within a distance of ∼2 mm along the sample surface, where the n(235U)/n(238U) amount ratio varied by a factor of ∼40-times, ranging from ∼0.075 to ∼2.8. Besides the pronounced isotopic heterogeneity, there were, however, also regions in which the n(235U)/n(238U) amount ratio was quite constant. The first 140 data points in Fig. 2 (dotted horizontal line A), for example, were centred around a value of ∼0.55 before dropping to ∼0.20 for the next ∼70 measurements (dotted horizontal line B). These regions of constant n(235U)/n(238U) amount ratios each indicated a single component (U isotope ratio) involved. The last ∼40 data points in Fig. 2 (grey rectangle C) indicate mixing of these two components with another one having a much higher U enrichment.
Scrap metal samples 1 and 2 originated from the same find and both contained highly enriched U. Both scrap metal samples also showed a low U content of <1 wt%. According to bulk TIMS analysis, their U isotopic composition appeared very similar, with a 235U abundance of ∼24.5 wt% and ∼25.9 wt%, respectively. The LA-MC-ICP-MS data, however, revealed contrasting average 235U enrichments of ∼27.6 wt% and ∼37.7 wt%, together with high standard deviations (Table 2). The difference between median and average 235U enrichment calculated from the LA-MC-ICP-MS data indicated an asymmetry of the corresponding frequency distribution. In fact, a distinct tailing towards higher 235U enrichments was obvious on the right side of the maximum 235U enrichment of both frequency distributions (Fig. 3A and B). This feature was more prominent for scrap metal 2. The obvious discrepancy in the 235U enrichment of bulk and spatially-resolved analysis was due to the large U isotopic inhomogeneity of these two samples. Based on the LA-MC-ICP-MS analysis, both scrap metal samples covered 235U enrichments ranging from natural to >90 wt% (Fig. 3A and B).
Scrap metal 3 had the highest U concentration among the investigated specimens in this study, i.e. ∼14 wt%. Both the average (34.5 wt%) and median (35.2 wt%) 235U enrichment found with LA-MC-ICP-MS on the surface of scrap metal 3 were almost identical. The 235U enrichment was, however, ∼10 wt% higher than the corresponding value established using TIMS (24.7 wt%). Again, spatially resolved U isotopic analysis revealed the large variability of the 235U enrichment present on a μm scale in this sample (Fig. 3C). Simultaneously, the obvious discrepancy of U isotopic information obtained by bulk and spatially-resolved analysis demonstrates the limits of the former in case of inhomogeneous samples and stresses the added value of LA-MC-ICP-MS analysis in this context. Even though the LA-MC-ICP-MS results also included low and highly enriched U, the corresponding frequency distribution of the 235U enrichment was much narrower than that of the scrap metal 1 and 2.
Scrap metal 4 had the lowest U content (∼0.15 wt%) of all specimens included in this study. This time, the average and median values obtained by LA-MC-ICP-MS (16.5 wt% and 16.6 wt%, respectively) and TIMS (16.8 wt%) matched with each other. Line scan analysis by LA showed that the 235U enrichment of scrap metal 4 consisted of a relative narrow, mono-modal frequency distribution.
Scrap metal 5 consisted of natural U, having a very narrow distribution of the abundance of all U isotopes. This sample was the only specimen in this study that contained no 236U. As for scrap metal 4, also here LA-MC-ICP-MS and TIMS results (0.714 wt% and 0.711 wt%, respectively) matched well with each other.
Scrap metal | Type of Analysis | |||
---|---|---|---|---|
Bulk, NDAa | Bulk, DAb | Spatially-resolved | Particle | |
γ-spectrometry | TIMS | LA-MC-ICP-MS | SIMS | |
a Non-destructive analysis.b Destructive analysis.c Compare to Fig. 6. Dispersion of data is as low as for certified U isotopic reference materials18 and caused by the uncertainty of individual measurement points.d Not measured. | ||||
1 | 18.1 ± 0.4 | 24.50 ± 0.01 | 0.88–90.5 | 0.65–96.2 |
2 | 28.3 ± 2.5 | 25.92 ± 0.01 | 0.52–94.4 | 0.65–95.9 |
3 | 23.9 ± 0.5 | 24.65 ± 0.07 | 1.94–88.0 | 0.75–87.8 |
4 | 17.0 ± 3.5 | 16.76 ± 0.01 | 10.1–22.0 | 7.13–21.2 |
5 | 0.73 ± 0.03 | 0.711 ± 0.001 | 0.66–0.97c | —d |
Comparison of average 235U enrichments obtained by γ-spectrometry and TIMS revealed large differences within the stated uncertainties (Table 3). This holds especially true for scrap metal samples 1 to 3 pointing to substantial inhomogeneity of their U isotopic composition. In such cases high precision analysis of the U isotopic composition of a relatively small fraction of the sample using TIMS is not necessarily representative for the entire specimen and might lead to wrong conclusions about the 235U enrichment of the sample.
The large range of various 235U enrichments in scrap metal samples 1 to 3 was confirmed independently by SIMS (Table 3). One should note that LA-MC-ICP-MS and SIMS analysis utilise different sampling approaches that might lead to slightly different results. While particles were swiped off from the scrap metal samples and collected on a cotton swipe for SIMS analysis, LA-MC-ICP-MS analysis generated spatially resolved U isotopic data directly from the solid sample fragment. Despite the different sampling methods, the 235U enrichment range obtained by both techniques overlapped largely showing that the endpoints of the enrichments (i.e. lowest and highest) can be determined accurately also by LA-MC-ICP-MS (Table 3). The main difference between the LA-MC-ICP-MS and SIMS results are seen between the 235U enrichment endpoints. While LA-MC-ICP-MS results detected the entire range of 235U enrichments, i.e. also all enrichments between the endpoints (Fig. 3A), SIMS revealed two populations of 235U enrichments peaking at a few atom% and at >90 atom% (Fig. 4).
The difference of the results obtained via the two analytical approaches is not yet fully understood, however, the most reasonable explanation is that the LA-MC-ICP-MS data suffer from mixing of different 235U enrichments within a sample because of the constraints related to the temporal resolution of the measurement. As the laser was operated at 4 Hz and the signal of a single laser shot lasted for ∼1.5 s, the particle plume of several laser spots were mixed with each other. This mixing led to a smoothing of the results, and might be one of the reasons why the entire range of 235U enrichments was observable using LA-MC-ICP-MS. The use of single shot LA analysis might be advantageous in this context19 and will be tested in future studies. Depending on the particle size of the sample, also the diameter of the circular laser beam (5 μm to 20 μm in this study) may possibly influence the measured 235U enrichment obtained via LA-MC-ICP-MS. In other words, if the particle size of the uranium is smaller than the diameter of the laser beam, then the different U isotopic composition of individual particles would remain unresolved.
The difference in the information obtained by LA-MC-ICP-MS and SIMS is exemplified graphically for scrap metal 3 in Fig. 5. Corresponding three-isotope plots showed a reasonable correlation between both n(234U)/n(238U)/n(235U)/n(238U) amount ratios (Fig. 5A) and n(236U)/n(238U)/n(235U)/n(238U) amount ratios (Fig. 5B) obtained via LA-MC-ICP-MS. The relatively large scatter of the LA-MC-ICP-MS data comes from the fact that the data was obtained by a quick line scan, whereas the SIMS data was acquired via precise micro beam analysis. Again, this demonstrates that both techniques provide excellent information about the range of the different 235U enrichments in the samples. However, where SIMS shows mainly two clusters (low-enriched and highly enriched U), LA-MC-ICP-MS data also shows the mixtures.
Applying the statistical methodology developed earlier,18 only scrap metal 5 was considered homogeneous, while scrap metal samples 1–4 were inhomogeneous with respect to their U isotopic composition. The half width of the Gaussian distribution fitted from the frequency distribution of the 235U enrichment of scrap metal 5 revealed a value of 1.3 × 10−4 (Fig. 6) that was within the range of 0.8 × 10−4 to 1.8 × 10−4 obtained for homogenous reference materials.18 The corresponding half width of scrap metal sample 4, the next best candidate of what could potentially be considered homogeneous, was as high as 1.3 × 10−2, hence clearly above the upper limit of the reference range indicated above. Because scrap metal 5 showed to be isotopically homogeneous, the spatially-resolved and bulk analysis results of the 235U enrichment (Table 3) were statistically not different.
Fig. 6 Frequency distribution of the 235U enrichment determined in scrap metal 5. The black solid line indicates the Gauss function fitted from the frequency distribution. The half width of this Gauss function is well within the range of 0.8 × 10−4 to 1.8 × 10−4 obtained earlier for certified U isotopic reference materials18 indicating that scrap metal 5 is homogeneous with respect to its U isotopic composition. |
This journal is © The Royal Society of Chemistry 2020 |