Measurement of the isotopic composition of uranium micrometer-size particles by femtosecond laser ablation-inductively coupled plasma mass spectrometry

Spectrochimica Acta Part B 93 (2014) 52–60

Contents lists available at ScienceDirect

Spectrochimica Acta Part B journal homepage: www.elsevier.com/locate/sab

Measurement of the isotopic composition of uranium micrometer-size particles by femtosecond laser ablation-inductively coupled plasma mass spectrometry Amélie Hubert a,⁎, Fanny Claverie b,c, Christophe Pécheyran b, Fabien Pointurier a,1 a b c

CEA, DAM, DIF, F-91297 Arpajon, France Laboratoire de Chimie Analytique Bio-Inorganique et Environnement, UMR 5254, Hélioparc Pau-Pyrénées, 2 Avenue du Président Angot, 64053 Pau, France Novalase SA, ZI de la Briqueterie, 6 Impasse du Bois de la Grange, 33610 Canejan, France

a r t i c l e

i n f o

Article history: Received 31 August 2012 Accepted 16 December 2013 Available online 24 December 2013 Keywords: Femtosecond laser ablation Uranium particles ICP-MS Fission tracks

a b s t r a c t In this paper, we will describe and indicate the performance of a new method based on the use of femtosecond laser ablation (fs-LA) coupled to a quadrupole-based inductively coupled plasma mass spectrometer (ICP-QMS) for analyzing the isotopic composition of micrometer-size uranium particles. The fs-LA device was equipped with a high frequency source (till 10 kHz). We applied this method to 1–2 μm diameter-uranium particles of known isotopic composition and we compared this technique with the two techniques currently used for uranium particle analysis: Secondary Ionization Mass Spectrometry (SIMS) and Fission Track Thermal Ionization Mass Spectrometry (FT-TIMS). By optimizing the experimental conditions, we achieved typical accuracy and reproducibility below 4% on 235U/238U for short transient signals of only 15 s related to 10 to 200 pg of uranium. The detection limit (at the 3 sigma level) was ~350 ag for the 235U isotope, meaning that 235U/238U isotope ratios in natural uranium particles of ~220 nm diameter can be measured. We also showed that the local contamination resulting from the side deposition of ablation debris at ~100 μm from the ablation crater represented only a small percentage of the initial uranium signal of the ablated particle. Despite the use of single collector ICP-MS, we were able to demonstrate that fs-LA-ICP-MS is a promising alternative technique for determining uranium isotopic composition in particle analysis. © 2014 Elsevier B.V. All rights reserved.

1. Introduction It has become increasingly important for safeguards and environmental monitoring purposes to develop and apply analytical techniques to particulate matter collected at established nuclear facilities in order to check the consistency of the declarations by facility operators [1]. Sampling consists in collecting dust material by wiping surfaces with cotton cloths inside nuclear facilities. This dust contains U particles that have been produced in the processes implemented in the facility. These highly mobile particles are found in many locations in the facility and their isotopic, elemental and structural compositions provide specific information about the activities carried out in the installation [2]. Uranium has three naturally occurring isotopes (234U, 235U and 238 U). 235U is the only U isotope that is fissile with thermal neutrons. The 235U/238U uranium isotopic ratio shows only minor variations in nature (around 7.25 × 10−3), but due to anthropogenic enrichment, it fluctuates from slightly enriched for civilian applications (in nuclear power plants) to highly enriched for research or military applications. Depleted uranium (lower 235U content than NU—natural uranium) is ⁎ Corresponding author. Tel.: +33 169 26 70 65; fax: +33 169 26 40 00. E-mail address: [email protected] (A. Hubert). 1 Tel.: +33 169 26 70 65; fax: +33 169 26 40 00. 0584-8547/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.sab.2013.12.007

also produced as a by-product of uranium enrichment processing. Consequently, uranium isotopic measurements allow a discrimination of the different origins (natural, enriched, etc.) and purposes of nuclear material detected in the samples (civilian, military). However, as the amount of material contained in each of the micrometer-size particles usually encountered in the sample is very low (in the picogram range), a very sensitive analytical technique is required. So far, the analysis of such environmental samples has been investigated using two different methodologies [2]: bulk analysis [3] which consists in measuring the uranium and plutonium isotopic signatures of the entire sample, and particle analysis which consists in separating particles in order to determine their individual isotopic signature. Individual particle analysis has the advantage of being selective and detecting different isotopic compositions (i.e. origins) in one location, whereas bulk analysis gives only the average uranium isotopic composition. Therefore, single particle analysis is a very powerful approach for determining the presence of undeclared nuclear activities. Currently, two analytical techniques are used for particle analysis, each with their advantages and drawbacks: Fission Track-Thermal Ionization Mass Spectrometry (FT-TIMS) [4,5] and Secondary Ionization Mass Spectrometry (SIMS) [6–9]. Considering the limitations of these techniques (TIMS has a low sample throughput and lacks sensitivity, measurement of minor isotopes

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

by SIMS can be biased by molecular interferences [8,9]), the feasibility of using Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) for uranium particle analysis was demonstrated by a few teams [10]. On the one hand, LA sampling allows bulk analysis as well as in-depth profiling, surface mapping and micro-analysis with good spatial resolution (at best a few μm). On the other hand, ICP-MS is considered to be a highly sensitive measurement technique and when coupled to LA it allows the direct analysis of solid samples. Particles can therefore be directly ablated individually for determining their corresponding isotopic composition. However, without a uranium particle localization technique, only relatively “large” uranium particles (at least 10 μm) can be analyzed [11–14]. Another group [15] presented direct analysis of U containing particle using ICP-MS after chemical purification of the particles. This method requires a perfect control of the blank level in all the reagents used. In a recent paper [16], we described the use of LA-ICP-MS in combination with precise uranium particle localization techniques (fission track and scanning electron microscope). After location, LA-ICP-MS particle analysis is fast (a few minutes per particle) and therefore a great number of particles can be analyzed per day. The sample preparation procedure allows particles to be fixed on the sample holder before localization and LA-ICP-MS measurements. Therefore, particles were not blown out by laser shots. Particles smaller than 1 μm were successfully analyzed using a commercially-available nanosecond-UV laser (Cetac LSX 213 nm) coupled to a quadrupole based-ICP-MS (ICP-QMS Thermo “X-Series II”) on two particle-containing samples already analyzed in the same laboratory by FT-TIMS and SIMS. Their 235U content was in the femtogram range. 235U/238U ratios were measured for all located particles. LA-ICP-MS results, although less precise and accurate (typically 10%) than the results obtained by FT-TIMS and SIMS due to short (20–40 s), transient and noisy signals, were in good agreement with the certified values or with the results obtained with other techniques. Thanks to a good measurement efficiency (~ 6 × 10−4) and a high signal/noise ratio during the analysis, LA-ICP-MS can be considered a very promising technique for fast particle analysis. The aim of this study is to test the use of a newly developed infra-red femtosecond laser ablation (fs-LA) system coupled with a quadrupole based ICP-MS in order to analyze NU micrometer size particles, and evaluate the reproducibility, accuracy and sensitivity of this method. A thorough study was conducted to optimize analytical conditions (wet/dry plasma, laser spot size, shot frequency, etc.) for uranium particle analysis. Statistical treatments were also applied on raw results to optimize the precision of 235U/238U ratio measurements. The isotope fractionation effect and the extent of deposition of ablation debris in the vicinity of the ablated area were also investigated. In addition to micrometer-size particles of known isotopic composition (NU), the fs-LA-ICP-QMS method was also applied to a real cotton cloth sample. Results were discussed and compared to those obtained with the FT-TIMS and SIMS techniques. 2. Experimental 2.1. Instrumentation The ablation system used in this study is a femtosecond LA device (Alfamet, Novalase SA, Amplitude Systèmes France) [17,18] fitted with a diode-pumped KGW-Yb crystal. The laser source delivers 360 fs pulses at an IR wavelength of 1030 nm and operates at high repetition rates (up to 10 kHz) and low energy (from 225 μJ at 100 Hz to 100 μJ at 10 kHz). The low energy level delivered by the laser source imposed the use of a narrow laser beam focused on the sample in order to keep the fluence above the ablation threshold. In the same way, the use of a high repetition rate laser source largely overcomes the lack of sensitivity which results from the low quantity of material ablated by low energy laser shots. A 50 mm focal length objective was fitted in the laser machine providing a 17 μm laser spot with maximum fluence varying from 39 J/cm2 at 100 Hz to 15 J/cm2 at 10 kHz). A fast 2D galvanometric

53

scanner allows the laser beam to be rapidly moved with high repositioning precision (better than 5 μm at 280 mm·s−1) in order to design complex trajectories in 2 dimensions. Then, when needed, the laser beam can be virtually enlarged by combining the high repetition rate to the fast movement of the laser beam as described below. Details of the LA device are described elsewhere [17,19–22]. These features are unusual compared to lower repetition rate and higher energy lasers commonly used for chemical analysis by commercially available nanosecond LA. The ablation cell used was cylindrical cell (15 cm3, 4,5 cm diameter). The ICP-MS used in this study was an “X Series II” (Thermo Fisher Scientific, Winsford, UK). Wet and dry plasma conditions were evaluated for this study. While dry plasma conditions were simply obtained introducing the laser produced aerosol using a conventional torch, wet plasma conditions were achieved by using a two inlet torch that mixes the dry aerosol together with a nebulized aerosol (HNO3 2%). The optimization of the ICP-MS was performed with a 1 μg·l−1 uranium solution in wet plasma conditions, and by ablating a 100 μm wide line of the NIST612 glass standard material in dry plasma conditions. Each isotope was acquired with a dwell time of 20 ms. In the case of the FT-TIMS technique, particles are extracted from the cotton and fixed on polycarbonate Lexan® disks with a polymer. These disks are then covered with a thinner Lexan® disk and irradiated in a nuclear reactor [16]. Fissions of fissile 235U atoms induced by thermal neutrons result in fission fragments which impact on the polymer. After chemical etching, characteristic figures called fission tracks can be observed on the thinnest disk under an optical microscope. The position of the center of a cluster of fission tracks indicates the position of the corresponding uranium particle. Once located, the particles are removed and deposited on rhenium filaments (one particle per filament). These filaments are then introduced in the TIMS source, and each particle is analyzed individually. Particles are chosen according to the number of their fission tracks which depend on the quantity of 235 U present in the particle. A large number of fission tracks reveals either a large NU particle or a smaller particle made of highly enriched uranium (HEU—higher abundance of 235U). This technique ensures the detection of a few HEU particles even if they are mixed with thousands of NU particles. However, the analysis time for this technique is quite long (at least several weeks). Decay of short-life activation radionuclides after irradiation of the sample prior to handling lasts one week. Sampling of individual particles is tricky and only 5 to 6 particles can be analyzed by TIMS per day. The analysis of a batch of 3 to 4 samples requires at least three weeks, and more often 2 months, provided the proper instruments and an irradiation facility are available. When using the SIMS technique, particles are extracted from the cotton and deposited on a carbon disk. They can either be located using a Scanning Electron Microscope (SEM) or directly located with SIMS. Each single particle is then analyzed individually. This technique allows fast measurements as sample preparation and particle locations can be carried out in 2 to 3 days. About 30 particles can be analyzed per day. However, this technique is not selective for highly enriched uranium particles, and a large number of particles have to be analyzed in order to detect one particular minor isotopic composition among a large quantity of particles coming from the major isotopic compositions (for instance NU particles). Moreover, measurements of minor isotopes are hindered by polyatomic interferences [8,9]. Modern large geometry SIMS equipped with automated particle search algorithms has a much higher throughput than 30 particles per day and are currently under evaluation [23]. 2.2. Samples The evaluation of the LA-ICP-MS technique for particle analysis was based on the analysis of three different samples: IRMM541, IRMM184 and particles of NU. The IRMM541 (from Institute for Reference Materials and Measurements, Geel, Belgium) is a certified glass material with a

54

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

nominal U mass fraction of 50 ppm and a certified amount ratio of 235 U/238U of 0.007277 ± 0.000007 (at the 2σ level) corresponding to the natural abundance. The IRMM184 (also from Institute for Reference Materials and Measurements) standard solution contains uranium in a natural composition with a certified 235U/238U isotopic ratio of 0.0072623 ± 0.0000022 (at the 2σ level). These samples were used in order to correct results from mass bias. Particles of NU (not certified) were used in order to evaluate the accuracy and the precision of the technique. These particles were produced from NU specifically for particle analysis. For the laser ablation analysis, the particles were first fixed on a transparent polycarbonate disk (25 mm diameter) using a polymer and then located using two techniques: (i) observation of Fission Track (FT) clusters obtained after irradiation in a nuclear reactor (7 × 1014 cm−2 thermal neutrons, Orphée, Saclay, France) or (ii) SEM equipped with an energy dispersive X-ray analysis (EDX) system and software called GSR (Gun Shot Residue) that allows the automated detection of U particles on the disk (Philips FEI, XL30 ESEM, Eindhoven, The Netherlands). Correspondence between the positions of each FT cluster and the positions of the corresponding particles was established using landmarks and a relatively straightforward two-point algorithm. Particles below 1 μm could be easily detected, especially those which contain U enriched in 235U. Conversely, particles below 1 μm in diameter were difficult to locate using the SEM/EDX technique. In both cases, coordinates of the particles were obtained according to the FT or SEM “reference system”. Therefore, a calculation had to be implemented to transpose the particles' coordinates into the reference system of the LA cell. For this purpose, 3 distinct marks were engraved on the polycarbonate disk and located on each reference system. Precision of the relocation was typically ±15 μm.

2.3. Ablation strategy The use of a relatively narrow laser beam (17 μm diameter at the sample surface) is particularly useful for high resolution microanalysis. Two ablation modes that resulted in 40 μm diameter craters were chosen. They consisted in ablating 3 concentric circles at either a) 1 mm·s−1 scanner speed and 1 kHz or b) 2 mm·s−1 scanner speed and 2 kHz (Fig. 1). The “a” ablation mode resulted in a 106 count·s−1 signal which lasts about 40 s at the mass to charge ratio (m/z) corresponding to 238U for the IRMM 541 sample. The “b” ablation mode resulted in a 2 × 106 count·s− 1 signal which lasts about 50 s at the m/z ratio corresponding to 238U for the IRMM 541 sample. The final choice is made according to analytical performance (accuracy and precision). Whatever the case may be, this ablation scheme appears to be a good compromise between the spatial resolution required for particle analysis and the sensitivity required for the analysis of the certified glass material. Additionally, this intermediate crater size will ensure the

Final crater Laser shot

Concentric circles

Fig. 1. Schematic representation of the ablation figure with concentric circles.

localization of the particles, given the error on the particles' relocation procedure (±15 μm). However, as it will be explained in detail in the following section, the repetition rate used for the ablation of particles was below 1 kHz. Therefore, the scan speed of the laser beam was decreased in order to keep the same overlapping of the pulses (92.5% between pulses on a same circle and 39.1% between pulses of concentric circles). 3. Results and discussion 3.1. Impact of the repetition rate on the degradation of the polycarbonate disk The use of a high repetition rate induces fast overlapping of pulses which could cause heating of the surface [24,25]. Consequently, helium was chosen as the carrier gas due to its high thermal conductivity which minimizes the thermal effect induced by the laser on the sample surface. Moreover, the use of He improves the detection efficiency of uranium [26]. Recent study shows that the mixing of He and Ar increases the plasma temperature [27], it results in a better diffusion and atomization of analyte ions [28] and then increases the detection efficiency. Tests were performed to evaluate the effect of shot frequency on ablation quality in terms of crater shape and surrounding damage. A blank deposit (a transparent polycarbonate disk coated with a thin polymer layer without U particles) was ablated at repetition rates ranging from 0.1 to 5 kHz. The number of shots was kept constant. Three replicates of each repetition rate but at different locations on the disk were performed and observed with an SEM (Fig. 2). The final size of the crater ranged from 45 to 65 μm depending on the repetition rate. In fact, results showed that the higher the repetition rates, the deeper the ablation. This led to an increase in crater diameters because of the Gaussian shape of the laser beam. At repetition rates above 1 kHz, the polycarbonate disks were pierced, their surfaces were distorted due to shock waves and craters were not circular. In contrast, craters obtained at repetition rates below 0.5 kHz showed a well defined geometry without melted rims. Even with the use of a femtosecond laser (which is known to reduce thermal effect [29,30]) in a helium atmosphere (which is also known to limit surface heating [31]) melted rims were observed. The polymer used to settle the particles was found to be very sensitive to high repetition rates (up to 1 kHz) which led to fusion effects that disappear below 0.5 kHz. Furthermore, it was observed that the shock wave can move the particles away from their initial positions thereby preventing their analysis. As a result and based on these qualitative observations, the other tests were carried out with repetition rates of 0.5, 0.3, and 0.1 kHz. 3.2. Contamination of the disk surface due to deposition of ablation debris Despite the use of helium which is supposed to reduce this surface deposition [32,33], ablation debris of unknown nature can be observed near the ablated area (see Fig. 2), whatever the repetition rate. Consequently, this surface contamination might induce a bias in the isotopic composition determination of particles located at the periphery of a previously ablated particle. To evaluate the extent of this phenomenon, two isolated particles were chosen on the polycarbonate disk. For each particle of interest, five spots evenly distributed along a 200 μm (diameter) circle centered on the particle were ablated at 300 Hz before and after the ablation of the located particles. A schematic representation of the experiment is presented in Fig. 3. The first 5 ablations, carried out before shooting at the uranium particle, allow the blank composition of the polycarbonate disk and its thin layer to be assessed. The other five ablations then allow the impact of the ablation debris deposition on the blank polycarbonate disk to be determined and thus the contamination caused by a neighboring ablation. The 238U signal measurements for all these ablations are shown in Fig. 3. For a given spot, deposits accounted for less than 4% of the initial particle signal for particle A and for less

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

55

Fig. 2. Effect of the repetition rate on the size and shape of the crater. SEM pictures of craters obtained with different repetition rates. A: 5000 Hz, B: 2000 Hz, C: 1000 Hz, D: 500 Hz, E: 300 Hz, F: 100 Hz.

than 2% in the case of the largest particle B. Considering the two particles studied, it should be noted that we were unable to find any systematic fall-out locations related to the direction of the helium gas flow. This

1.0 29.0

0.6 1.7

0.6 1.7

0.9 36.0

4

9 3

10

Before the ablation of the particle

6 100 µm 8

5 11 7.8 16.8

0.6 0.8

Part A: 194 Part B: 1871

1

7

0.6 1.0

1.0 13.9

7.8 14.8

Ablation of the particle After the ablation of the particle

2 0.7 0.6

contrasts to a certain extent with previous observations for a cylindrical shape ablation cell [34]. Although the reason this phenomenon was not observed here remains unclear, it may be related to the dynamic conditions governed by the cell geometry, helium flow rate and eventually back pressure. As expected, larger U content (particle B), induced a higher surface contamination. Therefore, it appears crucial to sufficiently separate the analyzed particles in order to avoid the re-ablation of aerosol deposits and consequently potential mixing of different isotopic signatures.

ion beam mesured (x103 counts.s-1) for particle A (up) and B (down)

3.3. Calculation of the 235U/238U ratios

0.7 238 U 0.6

Helium flux Fig. 3. Schematic representation of the ablation figure carried out to evaluate potential contamination due to redeposition of particle debris for two particles (part A and part B). The numbers inside circles are related to the sequencing of ablations. The ablation numbers: 1 to 5 (dark gray circles) correspond to blank ablations performed before the particle's ablation, 6 (black circle) corresponds to the ablation of the particle itself, 7 to 11 (light gray circles) correspond to blank ablations performed after the particle's ablation. The numbers outside the circles indicate the 238U ion beam measured for each ablation point.

Several NU particles were ablated with different operating conditions (LA frequency, wet or dry aerosol introduction in the plasma, dwell time of the ICP-MS detector). In a real environmental sample, each particle is unique and most particles are very small, and consequently are completely destroyed by the LA analysis. It is therefore not possible to evaluate the reproducibility of each measurement by repeating the analysis several times on the same particle. Internal and external precision are determined using NU particles with a known atomic ratio 235 U/238U of 7.252 × 10−3. The relevance of different data treatments has been evaluated. The mono-collection of the ion beam combined

56

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

with the transient signal obtained during particle ablation makes it difficult to evaluate isotopic ratios accurately. Two methods were used for the calculation of the 235U/238U ratios: i) overall 235U and 238U peak integration (no estimation of internal error) and ii) floating average (Eq. (1)) of point-to-point 235U/238U (estimation of internal error). 0 1 j¼iþ5 n X 235 U j U 1X 1 @ A; ¼ 238 U average n i¼1 10 j¼i−4 238 U j 235

ð1Þ

with n being the number of individual measurements acquired during particle ablation (depending on the size of the particle). The repetition of these data treatments on various NU particles of the same disk allows the calculation of an external error (reproducibility) of the measurement. A comparison of these calculations is presented in Table 1. Results from the second method (point to point) were more reproducible (3%) than those treated by peak integration (8%). Moreover, they showed better accuracy: the bias from the natural value is shorter with the second method. Accordingly, the results presented in this study were calculated using this point-to-point method. 3.4. Introduction mode Two introduction modes were implemented: wet and dry nebulization. The wet plasma condition is known to be more stable and more robust with respect to variations in quantity and the nature of materials injected in the plasma [21,34] than the dry plasma condition. However, unlike a dry aerosol, a wet aerosol can induce interferences due to oxide formation, and increases the risk of contamination by impurities contained in the solution. In order to study the impact of the plasma conditions on the accuracy and reproducibility of the results, the IRMM 541 certified glass material as well as particles of NU composition were analyzed with both introduction modes. Each data point from Fig. 4 represents the 235U/238U isotopic ratios measured for the LA-ICPMS analysis of single U particles and the IRMM 541 standard (with the same analytical conditions) and for 5 min of continuous measurement of the IRMM 184 U isotopic standard. The ablations of IRMM 541 glass material led to a reproducible signal with 238U peak intensities of 1 × 106 counts and 2 × 106 counts for 1 kHz and 2 kHz respectively, with no significant differences in terms of accuracy and uncertainty of the 235U/238U atomic ratios. In contrast, particle ablations produce a 238U ion beam of 10 to 20 s, with poor reproducibility of the intensity due to the wide particle-size (from 1 to 3 μm) and U-content (from 10 to 200 pg) distributions. Results obtained for particles were more dispersed, but the 235U/238U reference value remained within the uncertainties of the measured isotopic ratios. Finally, the IRMM 541 glass material analysis carried out with the various introduction modes showed no significant differences in the measured 235 U/238U ratios (uncertainties between 2.9 and 4.0% in wet conditions and between 3.3 and 3.8% in dry conditions). However, the 235 U/238U ratios obtained for particles were more dispersed in dry than in wet plasma conditions (standard deviations between 6 and 38% in wet conditions and between 7 and 60% in dry conditions). The analyzed particles showed comparable signal intensities, so that the variability observed in the results is most likely not due to variability in the size of the ablated particles. Furthermore, in wet plasma

Fig. 4. Plot of the 235U/238U isotopic atomic ratio in various materials containing NU (triangles: solution, squares: glass, and circles: particles) with different introduction modes in the ICP-MS (plain: wet, open: dry) at different repetition rates. Ablation of glass is performed at a higher repetition rate (1000–2000 Hz) than the particles (100–300 Hz).

conditions IRMM 184 isotopic standard solutions can also be measured continuously. Therefore, the wet introduction mode was chosen for the following steps of this study.

3.5. Accuracy, reproducibility and detection limit The temporal evolution of the 235U/238U ratio was investigated with NU particles and with the IRMM 541 certified material. Two typical examples of raw profiles are shown in Fig. 5. No significant deviation (reference values are within the uncertainties) can be observed either for the IRMM 541 or for the NU particles during the course of the ablation. However, a strong dispersion of the 235U/238U ratios can be observed for particle ablation. It can be explained by the poor counting statistics obtained for particle analysis (800,000 integrated counts for 235 U for the glass reference material versus only 21,000 integrated counts for the particle because of the short signal duration of 15 s). Moreover, 235U/238U ratios may be biased due to saturation of the detector for the highest 238U signals, when relatively large ablation scraps are injected into the plasma (see Fig. 5B). Lastly, as can be seen from Fig. 4 (IRMM184 standard solution) and from Fig. 5A (IRMM 541 glass standard) mass bias was not detectable. As mentioned above, ablations of NU particles were carried out with three different shot frequencies (500, 300 and 100 Hz). Results were compared in terms of accuracy, reproducibility and detection limits. These ablations were carried out the same day in order to keep similar operating conditions of the ICP-MS. Several particles were ablated with each shot frequency in order to take into account the variability of the size and of the U quantity of the particles (Fig. 6). External (from particle-to-particle) RSDs (relative standard deviations) on the 235 U/238U ratios were of 24.5%, 2.5% and 7.0% for shot frequencies of 500, 300, and 100 Hz respectively. Moreover, the internal precision of the data follows the same trend. Actually, internal RSDs (between

Table 1 Comparison of the two calculation methods for 6 particles with a dwell time of 50 ms on each isotope and a shot frequency of 300 Hz. Calculation methods

Peak integration Point to point

Average 235U/238U on 6 particles

7.92 × 10−3 7.49 × 10−3

Internal error

External error

Reproducibility (%)

Average standard deviation

Average standard error

Standard deviation on 6 particles

0.84 × 10−3

0.05 × 10−3

0.64 × 10−3 0.26 × 10−3

8.1 3.4

Bias from natural value (%)

9.2 3.3

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

57

A

B

Fig. 5. Temporal evolution of the 235U/238U during the ablation of: A) IRMM541 glass standard reference material at 1 kHz, scanner speed 1 mm·s−1, 28.6 J cm−2 and B) NU particles at 0.3 kHz, scanner speed 0.3 mm·s−1, 32.95 J cm−2 The horizontal dashed line represents the 235U/238U value for NU. The triangles represent the 235U/238U isotopic ratio during the ablation. The solid line represents the 235U ion beam measured during the ablation.

Fig. 6. Plot of the 235U/238U isotopic atomic ratio in NU particles at different repetition rates (plain squares: 500 Hz, plain circles: 300 Hz, open circles: 100 Hz). The horizontal line represents the 235U/238U value for NU.

different measurement points for a single particle) ranged from 4.9% to 89% at 500 Hz. In contrast, lower shot frequencies of 300 Hz and 100 Hz led to RSDs of ~ 10% for all particles. It should be noted that a quadrupole-based ICP-MS is not the ideal instrument for such isotopic measurements due to the sequential measurement (peak jumping mode) of very transient signals which relate to the different isotopes. Two methods were used to calculate detection limits. The first method involved blank measurements carried out during the contamination experiments described above. The sensitivity of the method was first calculated with the IRMM 541 glass. Given a crater with a diameter of 40 μm (Φcrater) for a 50 μm (h) depth and a2 U density of 2.55 g·cm−3, Φ −5 mm3 . the ablated glass volume is: V crater ¼ π  crater 2  h ¼ 6:3  10 This corresponds to a U ablated quantity of 8 pg. Knowing that the intensity of the 238U ion beam for this ablation was 2.1 × 106 counts·s−1, the sensitivity was 262,500 counts·s−1 per pg of U. In the course of the contamination experiments, we measured a mean background signal for the polycarbonate disk and its thin polymer layer of 930 ± 460 counts·s−1 for 238U and of 35 ± 20 counts·s− 1 for 235U (calculated from Fig. 3: average of the 5 blank measurements before LA analysis of particles A and B). The detection limit was then calculated using this background + 3 times the standard deviation over background giving minimum

58

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

detectable quantities of 9 fg and 350 attograms respectively for the 238U and 235U isotopes. This represents a minimum diameter for a perfectly spherical NU UO2 particle (density of 10.96 g cm−3) of ~220 nm, if we assume that the atomization and the diffusion processes for both measurements (particles from LA and one particle) are comparable. However, these tiny particles must be detected and localized before isotopic measurement by LA-ICP-MS. Now, the SEM used in the course of this study allowed only U particles larger than 0.8–0.9 μm to be detected. As for localization by FT, the theoretical size of the natural U particles can be derived from the number of observed FTs [16]. In the current irradiation conditions (integrated thermal neutron flux of ~1015 cm−2), a U particle is supposed to be detected when the corresponding cluster has at least 4 FTs, which represents an equivalent diameter of ~0.6 μm. To improve smaller U particle detection capability, up to ~ 0.2 μm, irradiation flux must be significantly increased, for instance by increasing irradiation time with the same reactor, typically from 1 min to 2 h. Moreover, significant improvements in SEM technology have been made recently resulting in the decrease of the minimum size of U particle that can be detected by SEM.

3.6. Performance comparison with other particle analysis techniques Analyses of the same NU particles were carried out using the TIMS (single collector Sector 54, VG) and SIMS (IMS7f, Cameca) techniques. Results are summarized in Table 2. The mean 235U/238U ratios obtained for the 3 techniques are in good agreement, within the stated expanded uncertainties (95% confidence level), with the NU value and with each other. All measurements were made with a single detector. The main differences lie in the ionization process, and the duration of the measurement. LA leads to the shortest signals (15 s), whereas the analysis carried out using SIMS and TIMS are longer (5 min and 15 min respectively) in this particular case. The duration of the ionization process was only controlled by the ablation process in the case of LA-ICP-MS technique. In the case of TIMS and SIMS, two parameters affect the duration of the ionization: the way the particle heats (temperature slope and ultimate value for filament heating in the case of TIMS and intensity of primary beam bombarding the sample in the case of SIMS) and the size of the particle (i.e. the mass of U contained in the particle). Thus, by controlling the intensity of filament heating (TIMS) or the intensity of the primary ion beam (SIMS), longer analysis time can be achieved. Analysis time with TIMS and SIMS cannot practically be shortened below a few minutes as it is necessary to adjust analytical parameters (temperature, primary ion beam intensity) and continuously focus the ion beams so as to optimize signal sensitivity and stability. Therefore, the LA-ICP-MS technique presents results comparable to the FT-TIMS technique with a shorter analysis time. However, 235U/238U ratios measured by SIMS showed better accuracy and reproducibility. With new modern TIMS, the standard deviation and reproducibility should be improved. It is also true for the performances of the SIMS, new SIMS instruments (LG-SIMS) are currently under evaluation [23] and shall provide better results. The three techniques were also compared on a real sample and the results confirm what has been stated above. All results are presented in Fig. 7. Several HEU particles were detected by FT-TIMS (235 U abundance N 75%, 19 particles were analyzed in 8 days). No HEU particle was detected by SIMS but many NU particles (29 particles

Fig. 7. Plot of 235U abundance (atomic %) measured in particles of one sample using different techniques: SIMS (plain squares), LA-ICP-MS (plain circles) and TIMS (open circles). The duration of the analysis is indicated for each technique (1.5 days to analyze 29 particles with SIMS, 2 days to analyze 51 particles with FT-LA-ICP-MS and 8 days to analyze 19 particles with FT-TIMS). Only FT-LA-ICP-MS and FT-TIMS succeed in detecting highly enriched uranium particles in this sample.

were analyzed in 1.5 days) were detected. 51 particles were analyzed in 2 days with the FT-LA-ICP-MS technique. The U isotopic compositions of these particles varied from NU to HEU. Different isotopic compositions were detected between these two extreme isotopic compositions. These results show that the vast majority of the particles were small NU particles, detected by SIMS. However, most of those particles were probably too small to induce a significant number of FTs. As a result, none of these particles were selected for TIMS analysis whereas the biggest NU particles were nevertheless analyzed by LA-ICP-MS. It should be mentioned that only the particles that induce the larger number of FTs were selected for TIMS measurements whereas all the particles that generated a significant number of FTs were analyzed by LA-ICP-MS (typically only 4 to 10 FTs per cluster). Intermediate enrichment (~ 1% to 75% 235U) can be due to the presence of both NU and HEU particles in the ablated area (for instance one or more very small NU particles in addition to an HEU particle), resulting in a mixing line between the different isotopic compositions. Regarding the analytical results, LA-ICP-MS in combination with FT (FT-LA-ICP-MS) has the advantage of covering all the U isotopic compositions of the sample, whereas FT-TIMS focuses essentially on the most enriched U particles, and SIMS focuses on the major isotopic composition (here NU). The fact that HEU was not detected by the SIMS technique is probably due to the scarcity of the HEU particles in comparison to NU particles. Regarding sample throughput, FT-LA-ICP-MS analysis can be carried out more quickly than FT-TIMS but remains more time consuming than SIMS analysis, as shown in Table 3.

4. Conclusions The results obtained in this study show that femtosecond LA-ICP-MS is an extremely sensitive technique for isotopic analysis of micrometersize and even sub-micrometer size U-bearing particles, provided U particles can be detected and located beforehand, either using fission tracks after irradiation in an appropriate nuclear reactor or using a SEM equipped with the appropriate software for detecting particles of a high average atomic number. An appropriate selection of experimental

Table 2 Performances of SEM-LA-ICP-MS, FT-TIMS and SIMS techniques obtained for the analysis of natural uranium particles. U/238U (×10−3)

Technique

Number of particles analyzed

235

Reproducibility (%)

Bias from natural uranium value (%)

LA-ICP-MS FT-TIMS SIMS

9 5 6

7.36 ± 0.27 7.40 ± 0.38 7.246 ± 0.058

3.7 5.2 0.80

1.5 2.0 0.092

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

59

Table 3 Comparison of analysis time of one sample with the different techniques.

Preparation time (deposition disk) Irradiation timea (transport + cooling) Deposition on filament Localization of particles (SEM for SIMS, optical microscope for FT) Isotopic analysis (30 particles) TOTAL a b

FT-TIMS

FT-LA-ICP-MS

SEM-SIMS

SEM-LA-ICP-MS

2 days 10 days 3 days 2 days 8 days 25 days

2 days 10 days – 2 days 1 day 15 days

2 days – – 2 daysb 2 days 6 days

2 – – 2 1 5

days

days day days

Provided that the irradiation facility is immediately available. This initial localization is not necessary for SIMS analysis which can be carried out directly on the carbon disk.

conditions was found to improve accuracy and reproducibility of the 235 U/238U ratio. The laser shot repetition rate seems to be one of the major factors affecting the isotopic measurements, whereas the plasma condition (wet or dry) has no significant effect. Although the use of a quadrupole-based ICP-MS limits the precision of the isotopic analysis, the optimization of the dwell time of the detector yielded better precision. Good accuracy and reproducibility below 4% could be achieved for transient signals of only 15 s related to a small U quantity of 10 to 200 pg (particle diameters close to 1 μm). fs-LA-ICP-MS can thus be considered an alternative technique to the two techniques currently used by the laboratories involved in particle analysis (fission track TIMS and SIMS) due to its accuracy, reproducibility and time effective analysis with a potential for analyzing about 100 particles per day. It should also be underscored that no isotopic fractionation could be detected with the quadrupole-based ICP-MS used in this study. Moreover, detection limits, based on sensitivity measurements with a glass CRM and background measurements without U particles, stand at ~ 350 ag for 235 U, which roughly corresponds to particles ~ 220 nm in diameter (particles assumed to be spherical, made of NU and whose density is bulk UO2). Local contamination due to deposition of ablation debris around the crater very likely accounts for only a few percent of the initial integrated signals. Further investigation is required for this technique in order to control the variability of the measurements. In particular, work must be done to find a way to increase signal duration and reduce the size of the ablation debris which enters the plasma and causes strong shorttime signal variations, whatever the size of the particles. Moreover, further work is necessary to measure U minor isotopes (234U and 236U). In this sense, coupling with a multi-collector ICP-MS may be necessary to further improve the precision of measurements and detection limits, especially for measuring minor isotopes. Acknowledgments The authors wish to thank Anne-Laure Fauré and Olivier Marie for the sample preparations and particle location with SEM. They also thank Laure Sangely for the data set and useful talks on SIMS measurements. References [1] D.L. Donohue, Strengthening IAEA safeguards through environmental sampling and analysis, J. Alloy Comp. 271–273 (1998) 11–18. [2] D.L. Donohue, Strengthened nuclear safeguards, Anal. Chem. 74 (2002) 28A–35A. [3] F. Pointurier, P. Hémet, A. Hubert, Assessment of plutonium measurement in the femtogram range by ICP-MS; correction from interferring polyatomic species, J. Anal. At. Spectrom. 23 (2008) 94–102. [4] C.G. Lee, K. Iguchi, J. Inagawa, D. Suzuki, F. Esaka, M. Magara, S. Sakurai, K. Watanabe, S. Usuda, Development in fission track-thermal ionization mass spectrometry for particle analysis of safeguards environmental samples, J. Radioanal. Nucl. Chem. 272 (2007) 299–302. [5] Y. Shen, Y. Zhao, S.L. Guo, J. Cui, Y. Liu, J. Li, J. Xu, H. Zhang, Study on analysis of isotopic ratio of uranium-bearing particle in swipe samples by FT-TIMS, Radiat. Meas. 43 (2008) S299–S302. [6] G. Tamborini, M. Betti, V. Forcina, T. Hiernaut, B. Giovannone, L. Koch, Application of secondary ion mass spectrometry to the identification of single particles of uranium and their isotopic measurement, Spectrochim. Acta B 53 (1998) 1289–1302.

[7] G. Tamborini, M. Betti, Characterisation of radioactive particles by SIMS, Mikrochim. Acta 132 (2000) 411–417. [8] F. Esaka, K.T. Esaka, C.G. Lee, M. Magara, S. Sakurai, S. Usuda, K. Watanabe, Particle isolation for analysis of uranium minor isotopes in individual particles by secondary ion mass spectrometry, Talanta 71 (2007) 1011–1015. [9] Y. Ranebo, P.M.L. Hedberg, M.J. Whitehouse, K. Ingeneri, S. Littman, Improved isotopic SIMS measurements of uranium particles for nuclear safeguards purposes, J. Anal. At. Spectrom. 24 (2009) 277–287. [10] S.F. Boulyga, T. Prohaska, Determining the isotopic compositions of uranium and fission products in radioactive environmental microsamples using laser ablation ICP-MS with multiple ion counters, Anal. Bioanal. Chem. 390 (2008) 531–539. [11] S.J. Becker, H. Sela, J. Dobrowolska, M. Zoriy, S.J. Becker, Recent applications on isotope ratio measurements by ICP-MS and LA-ICP-MS on biological samples and single particles, Int. J. Mass Spectrom. 270 (2008) 1–7. [12] Z. Varga, Application of laser ablation inductively coupled plasma mass spectrometry for the isotopic analysis of single uranium particles, Anal. Chim. Acta. 625 (2008) 1–7. [13] N.S. Lloyd, R.R. Parrish, M.S.A. Horstwood, S.R.N. Chenery, Precise and accurate isotopic analysis of microscopic uranium-oxide grains using LA-MC-ICP-MS, J. Anal. At. Spectrom. 24 (2009) 752–758. [14] S. Kappel, S.F. Boulyga, T. Prohaska, Direct uranium isotope ratio analysis of single micrometer-sized glass particles, J. Environ. Radioact. 113 (2012) 8–15. [15] F. Esaka, M. Magara, D. Suzuki, Y. Miyamoto, C.-G. Lee, T. Kimura, Faesibility study of isotope ratio analysis of individual uranium–plutonium mixed oxide particles with SIMS and ICP-MS, Mass Spectrom. Lett. 2 (2011) 80–83. [16] F. Pointurier, A.C. Pottin, A. Hubert, Application of nanosecond-UV laser ablationinductively coupled plasma mass spectrometry for the isotopic analysis of single submicrometer-size uranium particles, Anal. Chem. 83 (2011) 7841–7848. [17] B. Fernández, F. Claverie, C. Pécheyran, O.F.X. Donard, Solid-spiking isotope dilution laser ablation ICP-MS for the direct and simultaneaous determination of trace elements in soils and sediments, J. Anal. At. Spectrom. 23 (2008) 367–377. [18] C. Pécheyran, S. Cany, O.F.X. Donard, Ablation laser femtoseconde à haute cadence de tir et basse énergie couplée à une détection ICP-MS: résultats prélliminaires, Can. J. Anal. Sci. Spectrosc. 50 (2005) 272–283. [19] B. Fernandez, F. Claverie, C. Pécheyran, O.F.X. Donard, F. Claverie, Direct analysis of solid samples by fs-LA-ICP-MS, TrAC Trends Anal. Chem. 26 (2007) 951–966. [20] B. Fernández, F. Claverie, C. Pécheyran, J. Alexis, O.F.X. Donard, Direct determination of trace elements in powdered samples by in-cell isotope dilution femtosecond laser ablation ICP-MS, Anal. Chem. 80 (2008) 6981–6994. [21] F. Claverie, B. Fernandez, C. Pecheyran, J. Alexis, O.F.X. Donard, Elemental fractionation effects in high repetition rate IR femtosecond laser ablation ICP-MS analysis of glasses, J. Anal. At. Spectrom. 24 (2009) 891–902. [22] C. Pécheyran, S. Cany, P. Chabassier, E. Mottay, O.F.X. Donard, High repetition rate and Row energy femtosecond laser ablation coupled to ICP-MS detection: a new analytical approach for trace element determination in solid samples, J. Phys. Conf. Ser. 59 (2007) 112–117. [23] P. Peres, P.M.L. Hedberg, S. Walton, N. Montgomery, J.B. Cliff, F. Rabemananjara, M. Schuhmacher, Nuclear safeguards applications using LG-SIMS with automated screening capabilities, Surf. Interface Anal. 45 (2013) 561–565. [24] J. Koch, M. Wälle, J. Pisonero, D. Günther, Performance characteristics of ultraviolet femtosecond laser ablation inductively coupled plasma mass spectrometry at 265 and 200 nm, J. Anal. At. Spectrom. 21 (2006) 932–940. [25] J.J. Gonzalez, A. Fernandez, D. Oropeza, X. Mao, R.E. Russo, Femtosecond laser ablation: experimental study of the repetition rate influence on inductively coupled plasma mass spectrometry performance, Spectrochim. Acta B 63 (2008) 277–286. [26] M. Wälle, J. Koch, L. Flamigni, S. Heiroth, T. Lippert, W. Hartung, D. Günther, Detection efficiencies in nano- and femtosecond laser ablation inductively coupled plasma mass spectrometry, Spectrochim. Acta B 64 (2009) 109–112. [27] H. Lindner, A. Bogaerts, Multi-element model for the simulation of inductively coupled plasmas: effects of helium addition to the central gas stream, Spectrochim. Acta B 66 (2011) 421–431. [28] K. Niemax, Considerations about the detection efficiency in inductively coupled plasma mass spectrometry, Spectrochim. Acta B 76 (2012) 65–69. [29] O. Samek, V. Margetic, R. Hergenröder, Sampling of material using femtosecond pulses, Anal. Bioanal. Chem. 381 (2005) 54–56. [30] R. Le Harzic, N. Huot, E. Audouard, C. Jonin, P. Laporte, S. Valette, A. Fraczkiewicz, R. Fortunier, Comparison of heat-affected zones due to nanosecond and femtosecond laser pulses using transmission electronic microscopy, Appl. Phys. Lett. 80 (2002) 3886.

60

A. Hubert et al. / Spectrochimica Acta Part B 93 (2014) 52–60

[31] I. Horn, D. Günther, The influence of ablation carrier gasses Ar, He and Ne on the particle size distribution and transport efficiencies of laser ablation-induced aerosols: implications for LA-ICP-MS, Appl. Surf. Sci. 207 (2003) 144–157. [32] S.M. Eggins, L.P.J. Kinsley, J.M.G. Shelley, Deposition and element fractionation processes during atmospheric pressure laser sampling for analysis by ICP-MS, Appl. Surf. Sci. 127–129 (1998) 278–286.

[33] D. Günther, C.A. Heinrich, Enhanced sensitivity in laser ablation-ICP mass spectrometry using helium–argon mixtures as aerosol carrier, J. Anal. At. Spectrom. 14 (1999) 1363–1368. [34] O'Connor, B.L. Sharp, P. Evans, On-line additions of aqueous standards for calibration of laser ablation inductively coupled plasma mass spectrometry. Theory and comparison of wet and dry plasma conditions, J. Anal. At. Spectrom. 21 (2006) 556–565.