Dominic
Larivière
a,
Karima
Benkhedda
*b,
Stephen
Kiser
b,
Sonia
Johnson
b and
R. Jack
Cornett
b
aLaboratoire de Radioécologie, Département de chimie, Université Laval, 1045 Avenue de la Médécine, Québec, QC, Canada G1V 0A6
bKarima Benkhedda, Radiation Protection Bureau, 775 Brookfield Road, Address Locator 6302D1, Ottawa, ON, Canada K1A 1C1. E-mail: karima_benkhedda@hc-sc.gc.ca; Fax: (613) 957-1089; Tel: (613) 957-9039
First published on 15th January 2010
A novel, rapid and completely automated method for the determination of long-lived actinides in air particulate samples was developed using ion chromatography separation prior to ICP-MS analyses. U, Pu and Am were pre-concentrated and separated from radioactive and stable interferences through a sequential arrangement of three columns (TEVA, U/TEVA, DGA) arranged in order to allow selective retention of the radioisotope on the designated resin. Detection and quantification of the various radioisotopes were performed on-line using inductively coupled plasma mass spectrometry (ICP-MS). Detection limits of 0.0006 (238U), 0.0063 (239Pu), 0.0041 (240Pu), 0.062 (241Am) μBq/m3 were obtained for air filters (based on 3000 m3 of air filtered). The automated procedure has been successfully tested on SRMs and spiked samples and showed high recovery yields (>90%). A total analysis time of 18 min is required for the separation, detection, and column rinsing/regeneration for subsequent analysis.
Alpha-spectrometry has been frequently applied to the determination of long-lived α-emitters. However, the fact that many of the isotopes of Pu, Am, and U have low-specific activity indicate that either large amount of samples will be required or that longer counting time will be needed for adequate detection.11–13 In addition, separation protocols used for α-spectrometry samples are generally labor intensive, time-consuming, and focused on the separation of only one radioelement to produce thin chemically pure sources free of energy interferences.1,4,5,7,14 For all these reasons, α-spectrometry is not perfectly suited for emergency response. In comparison, accelerator mass spectrometry (AMS)1,15–17 and thermal ionization mass spectrometry (TIMS),17–19 are exceptionally sensitive ion-counting methods that can detect a variety of long-lived actinides down to the sub-fg range. However, the scarcity of those instruments, especially AMS, and their low sample throughput make them less attractive instrumentation for emergency response scenarios.20,17 Inductively coupled plasma mass spectrometry (ICP-MS), while less sensitive than the above-mentioned techniques17 but far more common in analytical laboratories, has demonstrated its effectiveness for the measurement of many long-lived radionuclides at higher environmental concentrations.20–23 The simplicity of the interface of the ICP, which allows coupling with a variety of analytical techniques, including a multi-solvent delivery system (MDS) such as high-performance liquid chromatography on ion-exchange pumps, and its multi-elemental ability are attractive features to develop methods that provide significant sample throughput without sacrificing instrumental sensitivity.20
Many protocols have demonstrated the analytical capability of ICP-MS for the on-line determination of specific long-lived radioelements in a variety of matrices such urine24–27 water28–30 soil, sediment,3,31–34 biological3,33 and vegetation35 samples and food.36,33 However, most of these protocols lack the multi-elemental capacity and rapidity required for post-event assessment. The ones that have embraced the multi-elemental capacity of the ICP-MS have focused primarily on nuclear fuels and wastes quantification2,37–40 While these methods provide selectivity, rapidity, and a certain degree of automation, they lack the sensitivity required for environmental radioactivity monitoring, which can be achieved only through sample pre-concentration5,7
In this work and based on the knowledge available for the automation of nuclear waste analysis, we have investigated and demonstrated the automation of a sequential injection system for the identification and quantification of airborne actinides. The automated system is composed of a multi-solvent delivery system for pre-concentration/separation coupled to a sector-field ICP-MS (ICP-SFMS) for on-line detection. The determination of 238U, 239,240,242Pu, and 241,243Am in certified materials (airborne particulate matter and loaded air filters) and spiked filters was performed using the method developed. Parameters such as interferences from uranium hydride, abundance sensitivity from 238U peaks, matrix effect, and instrumental sensitivity have been investigated and method performance criteria such as detection limits, sample throughput, accuracy and precision are also reported.
Instrumental parameters | Element2 ICP-SFMS |
Torch position | Optimized daily |
Gas flow (L/min) | |
Cooling | 16.08 |
Auxiliary | 0.81 |
Sample | 0.99 |
RF power (W) | 1200 |
Lenses (V) | |
Extraction | −2000 |
Focus | −902 |
X-deflection | 0.00 |
Shape | 120.00 |
Y-deflection | −3.8 |
Detector voltage (V) | 1600 |
Sampling cone | 1.1 mm Nickel |
Skimmer cone | 0.8 mm Nickel |
m/z monitored | 235, 238, 239, 240, 241, 242, 243 |
Number of passes | 1 |
Number of replicates | 350 |
Acquisition time (min) | 7 |
Apex-Q system parameters | |
---|---|
Nebulizer | 1 mL/min PFA microflow |
Spray chamber temperature (°C) | 140 |
Peltier-cooled multipass condenser temperature (°C) | −5 |
A multi-solvent delivery system (ICS-3000, Dionex, Sunnyvale, CA) was used to deliver the various solvents in a gradient mode. An autosampler, (AS-HV, Dionex, Sunnyvale, CA) was used for the automated sample loading through the columns. The direction of the solvent flow into the column was controlled using four biocompatible analytical-scale two-positions, six-port switching modules (MX9900-000, Upchurch Scientific, Oak Habor, WA). Biocompatible modules were used instead of stainless steel ones to reduce the risk of contamination as well as deterioration due to corrosion when using acid solutions. The ICS-3000 unit, the switching modules and the autosampler were all controlled by the CHROMELEON® chromatography data system (Dionex, Sunnyvale, CA). Fig. 1 is a schematic diagram of the setup of the system coupled to the ICP-SFMS. Isotemp® programmable muffle furnace (Fisher Scientific, Ottawa, ON) and a MARS 5 microwave system (CEM Corporation, Matthews, NC, USA) were used for the thermal destruction and the acid digestion of the samples, respectively.
![]() | ||
Fig. 1 Schematic of the HPLC-ICP-MS system for the sequential separation and detection of actinides. |
Three stainless steel analytical grade columns (Alltech, Columbia, MD), two 4.6 mm i.d. × 50 mm long and one 2.1 mm i.d. × 50 mm long, coated inside with polyether ether ketone (PEEK) were filled, respectively, with TEVA,41 U/TEVA,42 and n-DGA43 resins (50–100 μm particle size, Eichrom Technologies Inc., Darien, IL). All transport and reagent lines used to design the flow injection (FI) unit, were made of 0.762 mm i.d. PEEK tubing (Alltech, Columbia, MD) with the exception of the transfer line between the switching module four and the PFA nebulizer which was assembled using 1.1 mm i.d. polytetrafluoroethylene (PTFE) tubing (Alltech, Columbia, IL). 10–32 PEEK high pressure fittings with PEEK ferrules (Upchurch Scientific, Oakhabor, WA) for coned ports were used to connect the switching modules, multisolvent delivery system, and the analytical-grade columns.
In addition, in order to assess the efficiency of the sample preparation and analyte detection developed in this work on actual air filters currently used by the Canadian Radiological Monitoring Network, polypropylene (PP) 8′′x 10′′ (20.5 × 25.4 cm) filters, (3M, London, ON) were spiked with known amounts of the radioisotopes (U, 239,240,242Pu and 241,243Am) and their recovery, following sample preparation and analysis, was evaluated.
Step | Time (s) | Medium | Flow rate (mL/min) | Switching modules (SM) positiona | ||||
---|---|---|---|---|---|---|---|---|
SV-1 | SV-2 | SV-3 | SV-4 | SLPb | ||||
a SV-1 to SV-4: switching valves. b Sample loop position, elute signifies that the medium is passing through the sample loop while load is the opposite. | ||||||||
1 | 240 | 3M HNO3 | 2.5 | On | On | On | Off | elute |
2 | 270 | 0.1M (NH4)2C2O4 | 1 | Off | On | Off | On | load |
3 | 120 | 0.1M (NH4)2C2O4 | 1 | Off | Off | On | On | load |
4 | 180 | 0.01M (NH4)2C2O4 | 1 | On | Off | Off | On | load |
5 | 90 | Milli-Q water | 2.5 | On | On | On | Off | load |
6 | 60 | 3M HNO3 | 3 | On | On | On | Off | load |
Step | Step description |
---|---|
1 | 3 M HNO3 is pumped through the sample loop to load the sample and rinse the residual elements from the three resins |
2 | 0.1 M (NH4)2C2O4 is pumped through U/TEVA to elute U |
3 | 0.1 M (NH4)2C2O4 is pumped through n-DGA to elute Am |
4 | 0.01 M (NH4)2C2O4 is pumped through TEVA to elute Pu |
5 | Milli-Q water pumped to rinse all three resins from any residual elements |
6 | 3 M HNO3 is pumped through all three resins to pre-condition the resins for the next analysis |
In order to evaluate the analytical performance of the protocol developed with respect to a nuclear emergency scenario, the maximum exposure by airborne particulate (MEAP) that would result in a annual dose commitment of 0.1 mSv y−1 dose for each radioisotope was calculated. The following equation was used:
Variable | Optimal | Effect of variable on the system |
---|---|---|
Filter ashing | 550 °C for 4 h or 400 °C for 24 h | critical |
Media for MW digestion | HNO3 | critical |
Sample acidity | 3 mol L−1 HNO3 | can vary between 2–5 mol L−1 |
Diameter of TEVA column | 4.2 mm i.d. | critical |
Diameter of U/TEVA column | 4.2 mm i.d. | critical |
Diameter of DGA column | 2.1 mm i.d. | critical |
Sample flow rate | 2.5 mL min−1 | can vary between 1–5 mL min−1 |
Eluent (U, Am) | 0.1 mol L−1 (NH4)2C2O4 | critical |
Elution flow rate | 1 mL min−1 | critical for ICPMS performance |
Eluent (Pu) | 0.01 mol L−1 (NH4)2C2O4 | critical |
Effect of uranium on 239Pu determination | <10 μg L−1 uranium | critical |
Effect of uranium on 241Am determination | <100 μg L−1 uranium | critical |
The adsorption efficiency using different column diameters was tested using a matrix-matched standard at an uptake rate of 1 mL min−1. Two diameter sizes of columns (i.d. 2.1 and 4.2 mm) were tested for each resin. The results showed that the smaller diameter column did not contain sufficient amount of TEVA and U/TEVA resin to completely retain the analyte. For this reason, larger i.d. columns were used, which showed a complete retention of the analytes of interest. In the case of the n-DGA resin, the extraction of Am was complete with the smaller diameter column. The smaller amount of packing material required for the Am retention is probably associated with the better capacity factor and the higher selectivity achieved with the n-DGA resin.43 Therefore, for the rest of this work, 4.2 mm i.d. columns were used for TEVA and U/TEVA and a 2.1 mm i.d. column was used for DGA resin.
The effect of the sample uptake flow rate on the retention capacity of the resin was studied in order to determine the optimal flow rate for analyte retention, which also determines the time required for sample loading. This factor is critical for minimizing the overall analysis time. The maximum sample flow rate for Pu, Am, and U was tested by varying the flow rate from 1 to 5 mL min−1. No significant effect of the uptake flow rate on the adsorption of the analytes was found even for flow rates as high as 5 mL min−1. This observation is consistent with the findings from other groups using different extraction chromatography resins.25,49 While 5 mL min−1 flow rate could theoretically be used in this system, an uptake flow rate of 2.5 mL min−1 was chosen in order to minimize the time required for sample loading and rinsing steps while avoiding working at a pressure near the maximum allowable pressure of the fittings (1500 psi).
It was found that 0.1 mol L−1 (NH4)2C2O4 is the most suitable solvent for the elution of Am from the n-DGA resin. The use of the cumulative recovery at maximum peak signal clearly indicates that this solvent produces a peak much narrower than H2O or HCl as shown in Fig. 2. In addition, the elution occurred much faster (∼40 s) than for the other eluents tested. The choice of the eluent for Pu from TEVA resin was not as clear as in the case of Am. The use of 0.1 mol L−1 (NH4)2C2O4 as eluent did provide the fastest elution, however, the signal intensity was not maximal, due to incomplete elution. Similarly, elution with H2O did produce a reasonable peak, but with some apparent tailing. Elution of Pu with HCl leads to a peak significantly later than the other eluents tested with no significant improvement over 0.01 mol L−1 (NH4)2C2O4 which seems to be an ideal compromise as it did provide equivalent signal and elution time than H2O, but helped minimize the effect of tailing. In the case of U adsorption onto the U/TEVA resin, 0.1 mol L−1 (NH4)2C2O4 was chosen because it provides a narrower peak with an earlier elution time. In addition, the tailing was significantly reduced using this eluent when compared to H2O. The final elution parameters used for the rest of this work are as described in Table 2.
![]() | ||
Fig. 2 Cumulative recovery profiles for Pu, Am, and U using various elution solvents (○: H2O; ◇: 0.01 M (NH4)2C2O4; □: 0.1 M (NH4)2C2O4; ▽: 0.05 M HCl; △: 0.1 M HCl). Cumulative recovery of the radioelement at the peak maximum intensity is represented by the + symbol. |
Many methods have been published on the preparation of atmospheric particulate samples for chemical analysis based on digestion and extraction. However, airborne particulate matter collected on membrane filters is a difficult to digest sample because it contains a variety of matrix constituents, such as organics, oxides and silicates and therefore acidic decomposition is a critical step and in some cases total digestion is required particularly when a small amount of particles is collected on the filters.
In this work, for the in-house filters (PP), we have investigated whether it was necessary to completely dissolve the filters for effective extraction of the radioisotopes and whether it was possible to omit the dry ashing step to speed-up the process of sample preparation. The direct digestion of the whole filters in the microwave, without the dry ashing step, created a high pressure and could cause minor explosions inside the MW cavity. The leaching of PP filters in concentrated HNO3 caused the filter to aggregate and float on the surface of the acid, leading us to believe that efficiency of the extraction would be limited. Therefore, the ashing of the filters followed by digestion with HNO3 in a MW was found the most suitable procedure and produced nearly complete dissolution of the filters. Since time is a crucial issue in emergency situations, the effect of temperature and time of ashing step was also investigated. We have compared dry ashing the filters at 400 °C for 24 h to 550 °C for 4 h. It was found that recoveries obtained for all the radioisotopes investigated were comparable. Therefore, in case of emergencies, the short procedure is recommended.
The digestion of urban particulate matter (SRM-1648) resulted in a slight precipitate probably due to the incomplete dissolution of silica in HNO3 only and this will be probably the case of real loaded filters. Therefore the addition of HF during the digestion was investigated. It was found that the use of HF produced clear digest solutions and the recoveries for uranium and americium were comparable to the procedure using HNO3 only. However Pu isotopes suffered from low recoveries, 10 to 20%, in the spiked samples and therefore for the rest of this work, the use of HF was omitted. The samples were filtered before column separation to avoid clogging of the columns. Similar low recoveries for 242Pu were observed by Komosa et al.50 in their work when they compared different acid combinations, including HNO3/HF/HClO4 for leaching of Pu from air filters made of chlorinated polyvinyl chloride.
It is important to note that the efficiency of the sample preparation optimized in this work was evaluated only for our analytes of interest. An additional advantage of this sample preparation protocol is that the resulting acidity of the digested samples is optimal for loading onto the separation columns, with no additional pre-treatment.
Sample volume, mL | 5 | |||
Time of analysis, min | 16 | |||
Sample frequency, sample d−1 | 80 | |||
Sensitivity (105 counts pg−1) | 1.69 (242Pu), 1.81 (243Am), 1.64 (238U) | |||
Precision, % RSD (5 replicates) | 50 pg L−1 | 150 pg L−1 | 1 μg L−1 | 2 μg L−1 |
242Pu | 3.97 | 3.50 | — | — |
243Am | 5.66 | 4.53 | — | — |
238U | — | — | 0.89 | 2.86 |
![]() | ||
Fig. 3 Chromatogram of the sequential separation for PP filters spiked with 5 μg L−1 (0.06 Bq L−1) U, 0.14 Bq L−1242Pu and 1 Bq L−1243Am. |
The analytical performances of the present work were assessed against exposure risk to radionuclides from airborne particulate as shown in Table 5. As can be seen from column 5 and 6, the limits of quantifications (evaluated as 5 times the DL) for all radioisotopes investigated are several orders of magnitude below the maximum exposure limit by airborne particulate which makes this technique particularly suitable for both emergency purposes and for routine radiological monitoring of air.
Isotope | Half-life (y) | Effective dose conversion factor (Sv Bq−1)46,47 | Detection limits (mBq L−1) | Detection limit equivalence in air (DLEA, mBq m−3) | Maximum exposure by airborne particulate (MEAP, mBq m−3) |
---|---|---|---|---|---|
238U | 4.46 × 109 | 3.2 × 10−5 | 0.052 | 5.60 × 10−7 | 0.386 |
239Pu | 2.411 × 104 | 8.1 × 10−5 | 0.595 | 6.34 × 10−6 | 0.152 |
240Pu | 6.537 × 103 | 8.1 × 10−5 | 0.392 | 4.18 × 10−6 | 0.152 |
242Pu | 3.76 × 105 | 7.92 × 10−5 | 0.0022 | 2.34 × 10−8 | 0.156 |
241Am | 432.2 | 1.2 × 10−4 | 5.80 | 6.1 × 10−5 | 0.103 |
243Am | 7.37 × 103 | 1.19 × 10−4 | 0.088 | 9.38 × 10−7 | 0.104 |
A significant challenge when quantifying Pu and Am by ICP-MS is the presence of U at high concentrations. In this work we have investigated the effect of increasing levels of uranium in the sample on m/z= 239, 240, 241, 242 and 243 to determine the maximum concentration of uranium at which the separation protocol became obsolete for Pu and Am determination. As shown in Fig. 4, the presence of U will affect the background at masses from 239 and above, depending on the concentration of uranium present in the sample. In this work we determined that the ratio of the background signal at masses 239, 240 and 241 to the signal of uranium represent only 3 × 10−7, 4 × 10−8 and 3 × 10−8 for 239Pu, 240Pu and 241Am, respectively. However, the presence of uranium at a level of 10 μg L−1 in the loading sample is shown to be sufficient to prevent proper identification of 10 mBq L−1239Pu, which represents 2 times the quantification limit achieved with this method. The same level of uranium (10 μg L−1) was found not to affect the quantification limits for 240Pu and 241Am. However, the quantification limits for both 240Pu and 241Am decreases by a factor of 10 in the presence of 100 μg L−1 of uranium in the measured solution. As a result, it appears that the present method is suitable for the simultaneous quantification of 238U, 239Pu, 240Pu and 241Am in air for environmental monitoring purposes since uranium levels in air are unlikely to be higher than the maximum tolerated levels of uranium to allow the simultaneous quantification of radionuclides. However, the application of this method to a radiological event monitoring should be taken with caution, considering that presumably the uranium concentration will be significantly higher than the tolerable maximum level identified for the present method.
![]() | ||
Fig. 4 Effect of increasing levels of uranium in loading sample on the determination of (A) 239Pu, (B) 240Pu, (C) 242Pu and (D) 241Am. |
Contrary to the other automated systems that were developed in the past for the measurement of actinides,25,29,35,37 this system is designed in such a way that the resins can be reused multiple times. This helps improving the sample throughput and decreases the analytical cost per sample by reducing the amount of resins used.27 However, this situation could lead to possible sample cross-contamination resulting from sample carry-over. Therefore, in order to assess the level of cross-contamination, a sequence composed of 1 spiked sample (1 Bq L−1 Pu and Am; 2 μg L−1 U) followed by 3 blank solutions (3 mol L−1 HNO3) was used. The amount of each actinide was reported as a percentage of the signal compared to the one obtained by the sample. Carry-over percentages of 0.32, 0.73, and 0.02% for U, Pu, and Am, respectively, were measured in the first blank solution. Undetectable carry-over levels were found for the second rinse. These levels of carry-over are comparable to those found by others using similar extraction chromatography resins.29,34,51 Considering that the maximum carry-over rate is below 1% for all analytes, this indicates that a range of 100-fold in the analyte concentration will not significantly affect the quantification process. Therefore, in case of emergency, only a consecutive sample that exhibits a concentration difference of 100-fold should be retreated after either re-rinsing the resins or changing the columns, depending on the time and resources available.
The levels of uranium in the reference materials and the recoveries of spiked radioisotopes are compiled in Table 6. The concentration of uranium found in the SRM-2783 loaded filter with particulate matter agrees well with the observed values reported by Kulkarni52 and NIST. These results also show the capability of this method for accurate determinations of uranium at low ng L−1 levels which makes this method particularly attractive for monitoring of uranium levels in air. The concentration of uranium reported for the urban particulate matter (SRM-1648) was calculated with a 2.7% (RSD) precision and a bias of 6% from the certified value which demonstrates the efficiency of sample preparation methodology optimized in this work. This value, however, lay well outside the uncertainty interval of the certified value which could be explained by the fact that this reference material was purchased many years ago and the content of some elements is possibly altered. Note that the urban particulate matter SRM-1648 has been replaced by SRM-1648a which is the same particulate matter than SRM-1648 that has been re-blended and reanalyzed. Curiously, no value for uranium is provided for SRM-1648a. Another possible explanation to the lower recoveries of uranium in the urban dust is the acid resistant residual material, for example, refractory oxide that requires the use of more drastic sample preparation conditions such as fusion.
Sample | U, mg Kg−1 | U, μg L−1 | 239Pu, Bq L−1 | 240Pu, Bq L−1 | 241Am, Bq L−1 | |||||
---|---|---|---|---|---|---|---|---|---|---|
Found | Certified | Added | Found | Added | Found | Added | Found | Added | Found | |
a Only reference value is provided by NIST. b Reference value from Ref. 52. | ||||||||||
SRM-2783 Air particulate matter on filter | 1.216 ± 0.054 | 1.234 ± 0.024a 1.2 ± 0.1b | — | — | 2.52 | 2.44 ± 0.15 | 2.52 | 2.42 ± 0.02 | 2.52 | 2.36 ± 0.18 |
SRM-1648 Urban dust | 5.17 ± 0.14 | 5.5 ± 0.1 | — | — | 2.52 | 2.34 ± 0.13 | 2.52 | 2.29 ± 0.20 | 2.52 | 2.32 ± 0.20 |
Spiked PP filter | — | — | 5 | 4.66 ± 0.23 | 2.52 | 2.31 ± 0.21 | 2.52 | 2.30 ± 0.15 | 2.52 | 2.44 ± 0.15 |
PP filters are being used by Health Canada in various stations throughout Canada for air sampling for the monitoring of radioactivity in air. Therefore it was important for us to investigate the efficiency of this technique for the multi-isotopic determination of U, Pu and Am in air samples in a single filter (or a portion of a filter). As shown in Table 5, the recoveries for U, Pu and Am in the spiked PP filters were in the range 90 to 95%.
This journal is © The Royal Society of Chemistry 2010 |