B. C.
Russell
*a,
P. E.
Warwick
bc,
H.
Mohamud
a,
O.
Pearson
a,
Y.
Yu
d,
H.
Thompkins
a,
S. L.
Goddard
e,
I. W.
Croudace
c and
Z.
Zacharauskas
ac
aNuclear Metrology, National Physical Laboratory, Hampton Road, Teddington, TW11 0LW, UK. E-mail: ben.russell@npl.co.uk
bGAU Radioanalytical, UK
cUniversity of Southampton, NOCS, Southampton, SO14 3ZH, UK
dRoyal Holloway University of London, Egham Hill, Egham, TW20 0EX, UK
eAir Quality and Aerosol Metrology, National Physical Laboratory, Hampton Road, Teddington, TW11 0LW, UK
First published on 22nd November 2022
Tandem inductively coupled plasma mass spectrometry (ICP-MS/MS) has demonstrated effective online removal of interferences for a number of medium and long-lived radionuclides including 90Sr, 129I, 135/137Cs and 237Np. This reduces or removes the dependence on relatively time consuming offline chemical separation, improving sample throughput and offering economic benefits for end users through reduced procedural time, analyst time, secondary waste arisings and quantity of resins and reagents required. This work demonstrates the development and application of a method for measuring multiple radionuclides (including 63Ni, 90Sr, 93Zr, 99Tc, 129I, 237Np, 239Pu) and stable elements in various sample matrices in a single measurement. Separate instrument modes customised for individual radionuclides were combined into a single run, each using a different instrument setup depending on the extent of interference removal required. The measurement time was approximately seven minutes and consumed 3.5 mL for each sample. Instrument conditions were optimised for each radionuclide using stable and radioactive standards, and then validated using samples including aqueous wastes from a nuclear site, air filters following microwave dissolution, groundwater, and soil, sediment and concrete samples following borate fusion dissolution. Matrix-matched calibration standards were prepared, and a multi-element standard solution was used as an internal standard. The limits of detection were compared to regulatory limits, which showed that longer-lived radionuclides such as 93Zr, 99Tc, 129I and 237Np can be directly measured, whilst shorter-lived radionuclides such as 90Sr and 63Ni require separation and pre-concentration prior to sample introduction. The method provides a significant amount of information on stable and radioactive composition in a single measurement, with the potential to further expand the number of radionuclides measurable as part of this single procedure.
Of the available instrumentation, tandem ICP-MS/MS has demonstrated an improvement in instrument-based interference removal, reducing the reliance on time-consuming offline chemical separation. ICP-MS/MS consists of two quadrupole mass filters, separated by a collision-reaction cell. The use of a collision or reaction gas can selectively remove polyatomic and isobaric interferences, respectively, whilst operating with two quadrupole mass filters means the peak tailing removal capability is greater than that of single quadrupole instruments, with values on the order of <10−10 achievable. Additionally, positioning a quadrupole before the entrance to the cell enables the sample to be mass filtered prior to entering the cell, simplifying the cell chemistry. This technique has been applied to the measurement of multiple radionuclides including 90Sr, 129I, 135Cs/137Cs, 226Ra and 236U/238U4–11 and its capabilities for nuclear waste characterisation has been recently reviewed.12 The combination of MS/MS mode with a collision-reaction cell increases user confidence in the direct measurement of samples without relying on offline separation prior to measurement.
Whilst difficult-to-measure radionuclides have been effectively measured using ICP-MS/MS, the combination of multiple instrument modes into a single procedure has not been utilised. Using the instrument software, it is possible to combine different modes to enable the measurement of multiple radionuclides in a single run. For example, measurement of a radionuclide that benefits from the use of O2 as a reaction gas (such as 90Sr)13 can be followed by measurement of a radionuclide that benefits from NH3 (such as 93Zr),14,15 and then no gas (e.g.237Np)12 and so on as part of a single run. This approach provides more information from a single measurement.
This study shows the development of optimised methods for multiple radionuclides, initially using single radionuclide standards and then spiking with increasing concentrations of interferences. These individual methods are then combined into a single procedure along with the stable element composition and tested for direct measurement of various samples. The limit of detection achieved is compared to regulatory limits to assess which radionuclides can be directly measured, and which require further sample preparation steps. Consideration is also given for additional radionuclides that can be measured as part of this single procedure.
The Agilent 8800 has 4 cell gas lines (Table 1). In this study, NH3 was introduced via the corrosive gas line, and O2via the non-corrosive gas line. To prevent NH3 from damaging the collision/reaction cell, the gas mixture is 10% NH3 in 90% He. Additionally, the He gas line automatically runs at a flow rate of 1 mL min−1 when the corrosive gas line is operating to protect the cell. Helium, O2 and NH3 gases were provided by BOC, and hydrogen was provided from a generator (Linde NM Plus), all with a purity of N 6.0 (99.9999%).
Cell gas line | Cell gas | Flow rate (mL min−1) |
---|---|---|
a When operating the corrosive gas line, the He line automatically runs at a flow rate of 1 mL min−1 to protect the collision/reaction cell | ||
Hydrogen | H2 | 0.5–10 |
Helium | He | 0.5–12 |
Cell line 3 | Corrosive gasae.g. NH3, CH3F | 0.5–10 |
Cell line 4 | Non-corrosive gas e.g. O2, N2O, NO2 | 0.05–1.0 |
The instrument was tuned daily in all modes using a mixed 1 ng g−1 standard solution. The sensitivity across the mass range was assessed at 9Be, 89Y and 205Tl, with the oxide and doubly charged ratio assessed by 140Ce16O/140Ce and 140Ce2+/140Ce+ ratios, respectively. The quality control of the instrument was performed by generating a tune report each time the instrument was run, and the results logged in a control chart to ensure the sensitivity and uncertainty thresholds were reached, and that oxide and doubly charged ion formation was <2%. The peak axes at m/z = 9, 89 and 205 were also assessed to ensure the peak was within 0.1 atomic mass unit e.g. 8.9–9.1 for m/z = 9.
For radionuclide-specific measurement, the instrument was custom tuned by manually adjusting instrument parameters, focusing on the properties of the collision/reaction cell to find the best balance between sensitivity and interference removal. Firstly, the cell gas flow rate for the one or more gases used for each radionuclide; the octopole bias voltage, which impacts the acceleration and kinetic energy of ions in the cell; and the energy discrimination voltage, which provides an energy barrier at the exit of the cell for filtering out interferences whilst retaining analyte ions. When using NH3, a range of cell product ions can be formed (NH+, NH2+etc.). This was initially assessed by performing a product ion scan using single element standards. Q1 is set to a single m/z value (e.g. 63 for Ni), and Q2 then measures all m/z values up to 260, with each peak representing a cell product that can then be investigated further. For samples with an unknown composition, a precursor ion scan can also be applied where Q2 is set to a single m/z value and the Q1 mass range is scanned to determine if interfering elements are present in the sample.
A Spex Katanax K2 Prime was used for borate fusion dissolution of soil, sediment and concrete samples. An Anton Paar Multiwave 3000 microwave oven was used for dissolution of filter samples.
Radionuclide | Half life (a) | Production route | Principal decay mode | Mass equivalent to an activity of 1 Bq (pg) | Decay counting measurement technique |
---|---|---|---|---|---|
63Ni | 98.7(24) | Activation | β | 0.5 | Liquid scintillation counting (LSC) |
90Sr | 28.8(7) | Fission | β | 0.2 | LSC |
93Zr | 1.61(6) × 106 | Fission and activation | β | 1.1 × 104 | LSC |
99Tc | 2.1(11) × 105 | Fission | β | 1.6 × 103 | LSC |
129I | 16.1(7) × 106 | Fission | β | 1.6 × 105 | LSC |
237Np | 2.1(7) × 106 | Neutron capture | α | 3.8 × 104 | Alpha spectrometry |
239Pu | 24.1(11) × 103 | Neutron capture | α | 4.4 × 102 | Alpha spectrometry |
Radionuclide | Isobar (% abundance) | Major polyatomics | Tailing (% abundance) |
---|---|---|---|
63Ni | 63Cu (69.15) | 62Ni1H, 49Ti14N, 47Ti16O, 31P32O, 35Cl28Si, 27Al36Ar, 25Mg38Ar, 23Na40Ar | 62Ni (3.63) |
64Zn (48.27) | |||
90Sr | 90Zr (51.45) | 89Y1H, 74Ge16O, 59Ni16O2,50Ti40Ar, 50V40Ar, 54Fe36Ar | 88Sr (82.58) |
93Zr | 93Nb (100) | 92Zr1H, 77Se16O, 61Ni16O2,53Cr40Ar | 92Zr (17.15) |
99Tc | 99Ru (12.76) | 98Mo1H, 83Kr16O, 64Zn35Cl, 59Co40Ar, 51V16O3 | 98Mo (24.19) |
100Mo (9.67) | |||
100Ru (12.60) | |||
129I | 129Xe (26.40) | 127IH2,97MoO2,113Cd16O, 115In14N, 89Y40Ar | 127I (100) |
237Np | — | 205Tl16O2,197Au40Ar | 238U (99.27) |
239Pu | — | 238U1H | 238U (99.27) |
For each radionuclide, the instrument was run in different modes to determine the impact on sensitivity and interference removal capability. This was initially assessed using interference-free solutions, followed by spiking with various concentrations of known interferences. Each collision and reaction gas was tested, along with the difference between Single Quad and MS/MS mode. For higher matrix samples, the instrument can be operated with argon gas aerosol dilution (termed High Matrix Introduction or HMI) throughout, which allows tolerance of high matrix samples (up to 2.5% total dissolved solids) and is a faster alternative to offline dilution.
Aqueous waste samples were provided from a UK nuclear site. Some contained radionuclides of interest for this study, others were spiked with radionuclide standards of interest over a range of activity concentrations. These were measured without prior sample preparation. Twelve groundwater samples from sites around the UK were received from the British Geological Survey, and after filtering were also tested by taking a 1 mL aliquot and diluting to 10 mL in 0.2 M HNO3.
Air filter samples were provided by the NPL Air Quality and Aerosol Metrology Group, who are responsible for operating the UK Metals Monitoring Network on behalf of the Environment Agency and the UK governmental Department for Environment, Food and Rural Affairs (Defra). The Network consists of monitoring sites all around the UK that sample airborne particulate matter with an aerodynamic diameter less than 10 μm (PM10) onto cellulose ester filters. The sampled filters from the monitoring sites were prepared for analysis by microwave digestion. The filters were digested in a mixture of hydrogen peroxide and nitric acid. After digestion, the solutions were diluted to a total volume of 50 mL in deionised water, then analysed by ICP-MS for a suite of twelve metals to be reported for the Network analysis.16 In this study, four of these samples originally prepared for Network analysis were measured to assess the potential of the combined procedure for measurement of radionuclides on air filter samples.
Several solid samples (soil, sediment, concrete) were measured following lithium borate fusion, a dissolution technique first used in the radioanalytical field by Croudace et al. (1998).17 The method is described in detail elsewhere.18 Briefly, a 1 g sample was added to a crucible (95% Pt/5% Au) along with 1 g of 49.75/49.75/0.5 lithium metaborate/lithium tetraborate/lithium bromide, which was mixed manually. Each sample was loaded onto an automated fusion furnace (Spex Katanax K2 Prime) for dissolution. The fused samples were automatically dispensed by the instrument into PTFE beakers containing 50 mL 4 M HNO3 with a PTFE magnetic stirrer to aid dissolution of the flux. The total time to prepare and digest a batch of five samples was approximately 40 minutes. A 1 mL aliquot of the dissolved flux was diluted to 50 mL with 0.2 M HNO3 and measured without further treatment.
Measurement of both stable and radioactive Ni isotopes by ICP-MS/MS has been demonstrated elsewhere, using a combination of NH3 and H2 to achieve separation from isobaric 63Cu through selective formation of Ni(NH3)3.25 In the case of 63Ni, operating in MS/MS mode improves tailing removal from 62Ni, although the low isotopic abundance (3.6%) means this is unlikely to be an issue. The main advantage of operating in MS/MS mode is filtering the ion beam to m/z = 63 prior to using NH3 cell gas, simplifying the cell chemistry where a number of potential products can be formed.
The relatively short half life of 63Ni with regards to ICP-MS measurement (1 Bq g−1 is equivalent to 0.5 pg g−1) means operating in MS/MS mode and using cell gas must be more carefully considered for its impact on sensitivity compared to longer-lived radionuclides.
When He cell gas was introduced, the background at m/z = 63 increased up to a flow rate of 3 mL min−1, followed by a decrease at higher gas flow rates (Fig. 1). The increase is likely a result of collisional focusing, which is evidence of an isobaric interference, in this case 63Cu. The background at m/z = 59 was also measured because of interest in potential 59Ni measurement. The background increased slightly from 0.5–1.5 mL min−1 before decreasing at higher gas flow rates. This slight increase is associated with stable 59Co, which overlaps with 59Ni, however, the decrease in signal at higher He flow rates is indicative of a polyatomic interference (most likely 58Ni1H) that is not affected by collisional focusing.
![]() | ||
Fig. 1 Background at m/z = 63 and m/z = 59 with increasing He flow rate to determine isobaric and polyatomic background interferences. |
Oxygen cell gas was investigated, however, a limited amount of NiO was formed (maximum 6% compared to the CPS measured for Ni on-mass), increasing to 12% when the energy discrimination and octopole bias were custom tuned. By comparison, 0.5% of the total Cu signal was measured as CuO under the same conditions, suggesting a decontamination factor of 10 was achievable. Using NH3 cell gas alone, a Ni/Cu formation rate of 35 (calculated as the CPS for stable 58Ni(NH3)3/63Cu(NH3)3, with isotopic abundance scaled to 100% for each element) was achieved at a flow rate of 3 mL min−1, further improved by decreasing the energy discrimination from a starting value of −7 V to −13 V. The starting octopole bias value of −4.5 V was determined to be optimal.
Compared to NH3 alone, the addition of H2 gas increased Ni(NH3)3 formation up to a flow rate of 1.5 mL min−1. There was also an increase in Cu(NH3)3 formation (Table 4) but the Ni/Cu separation factor remained constant. The Ni/Cu separation factor was improved to 100 at 1 mL min−1 NH3 and 3 mL min−1 H2, with a sensitivity of 6100 CPS for a 1 ng g−1 (2.1 × 103 Bq g−1) Ni solution. At 1 mL min−1 NH3 only, the Ni/Cu separation factor was 90–95, but the Ni sensitivity was significantly lower (15 CPS). Under optimised conditions (Fig. 2), the 63Ni detection limit was calculated as 0.52 Bq g−1 (0.25 pg g−1), with the background at m/z = 63 < 10 CPS, compared to 500 CPS in no-gas mode.
NH3 (mL min−1) | H2 (mL min−1) | Ni(NH3)3 (CPS) | Cu(NH3)3 (CPS) |
---|---|---|---|
2.0 | 0 | 3900 | 70 |
0.5 | 6500 | 110 | |
1.0 | 6800 | 130 | |
1.5 | 7200 | 140 | |
2.0 | 7100 | 125 | |
2.5 | 7100 | 120 | |
3.0 | 6600 | 130 |
Whilst ICP-MS/MS has been applied to 99Tc,39 this was following chemical separation, and the possibility of cell-based interference removal has not been investigated. Operating in MS/MS mode could improve tailing removal from 98Mo, whilst the use of cell gases may be used to remove isobaric 99Ru and polyatomic 98Mo1H. The instrument blank at m/z = 99 was 0–5 CPS in 0.2 M HNO3 solutions. Operating in MS/MS mode with no cell gas reduced the sensitivity by 50% compared to Single Quad mode (245000 and 450
000 CPS for 1 Bq g−1 (1.6 × 103 pg g−1), respectively). However, the peak tailing removal was improved, with MS/MS mode able to tolerate up to 1 μg g−1 98Mo before the signal increased at m/z = 99. The use of He as a collision gas was very effective at reducing the signal from polyatomic 98Mo1H, with the signal from a 10 μg g−1 Mo solution reducing from 2200 CPS at 0.5 mL min−1 to <5 CPS at 8 mL min−1. However, the use of collision gas did not offer separation from isobaric 99Ru and was not considered further.
A product ion scan using NH3 (3 mL min−1) showed the majority of Tc remained on mass (90%). Whilst multiple cell products were formed, in all cases the sensitivity was low, with a maximum of 2.4% of the total counts for a mass shift of 34 (Tc(NH3)2). The majority of Mo and Ru also remained on mass under the same conditions (85% and 80%, respectively). The shift from m/z = 99 to 185 (TcH(NH3)5+) was the most promising for Ru/Mo separation. Custom tuning of the cell parameters improved Tc sensitivity to 3 × 103 CPS, whilst the signal from 10 ng g−1 Ru was reduced by an order of magnitude from 5 × 102 CPS to 3.6 × 101 CPS. However, accompanied with this was the increase of the Mo signal by an order of magnitude, from 2.4 × 101 CPS to 1.2 × 102 CPS. This mode offers the optimal Ru decontamination, however, the Tc sensitivity remained low and contamination from Mo remained.
In O2 mode, Tc remained predominantly on-mass over an O2 flow rate of 0.1–0.7 mL min−1, with collisional focusing improving sensitivity by 65% from 0.10 mL min−1 up to 0.35 mL min−1 O2 before decreasing at higher flow rates. The shift from Tc to TcO ranged from 1% to 15% of the total Tc signal, peaking at 0.45 mL min−1 O2. The TcO2 shift also ranged from 1% to 15%, with an optimal O2 flow rate of 0.45 mL min−1. Additionally, the background at m/z = 131 (TcO2) was 0 CPS. A mixture of 10 ng g−1 Ru and 1 μg g−1 Mo was run in the optimised modes for measurement of 99Tc, 99Tc16O and 99Tc16O2 to determine the decontamination achievable. When measuring Tc as 99Tc16O2, the 99Ru and 98Mo1H interferences were effectively suppressed compared to other modes. There was limited formation of 99Ru16O2, whilst 98Mo1H readily formed 98Mo16O2, but limited 98Mo16O21H. Measurement of 99Tc as 99Tc16O2 therefore effectively changed an interfering polyatomic into another, non-interfering polyatomic in the case of 98Mo1H. Generally, custom tuning of the instrument was effective at suppressing 98Mo1H, but behaviour of Ru and Tc was similar. The Tc sensitivity was improved from 2.5 × 104 CPS per Bq g−1 to 3.8 × 104 CPS per Bq g−1 when optimised, however, the background contribution from Ru increased by a factor of 5, to approximately 500 CPS at a concentration of 10 ng g−1. It was therefore decided that the best compromise between sensitivity and interference removal was to use the ‘standard’ O2 auto-tune settings. A summary of the performance of different cell gases is shown in Table 5. In the optimised setup (Fig. 3), the instrument LOD for 99Tc was 0.3 mBq g−1, equivalent to 0.5 pg g−1.
Mode | Q1 | Q2 | CPS per Bq g−1 Tc | 10 ng g−1 Ru signal (CPS) | 1 μg g−1 Mo signal (CPS) |
---|---|---|---|---|---|
MS/MS no gas | 99 | 99 | 245![]() |
200![]() |
100 |
MS/MS O2 (on mass) | 99 | 99 | 126![]() |
111![]() |
200 |
MS/MS O2 (single oxide) | 99 | 115 | 24![]() |
15![]() |
35 |
MS/MS O 2 (double oxide) | 99 | 131 | 25![]() |
160 | 5 |
MS/MS NH3 | 99 | 185 | 3000 | 40 | 120 |
ICP-MS/MS has been demonstrated as a viable method for measurement of 93Zr.14,15 A combination of NH3 and H2 gases was shown to offer the most efficient separation from stable isobaric 93Nb and radioactive 93Mo, through the selective formation of 93Zr(NH3)6, measured at m/z = 195. After demonstrating this using stable and radioactive standard solutions,14 the method was modified for dissolved steel and aqueous waste samples,15 with detection limits of 1.1–8.3 pg g−1 (0.1–0.8 mBq g−1).
Iodine-129 has been measured by ICP-MS/MS in samples contaminated by the Fukushima Nuclear Power Plant accident.45,46 In all cases, O2 gas was used to minimise the interference from isobaric 129Xe present as an impurity in the plasma gas via a charge transfer reaction that has limited impact on 129I. The additional mass filter improved stable 127I tailing removal, as well as in cell oxide polyatomic formation, most notably 97Mo16O2. Recent work has also shown that carbon-based matrix modification considerably improved instrument sensitivity, helping to overcome the high first ionisation energy of iodine.47
Neptunium-237 is a long-lived radionuclide that has been measured using a range of ICP-MS designs.2,3,48,49 The most significant interference is tailing from 238U, which can be removed by offline chemical separation.50,51 Operating in MS/MS mode, 237Np detection is possible even at 238U concentrations of 10 μg g−1 (Fig. 4), whilst there was an increase in background at m/z = 237 at 238U concentrations of 5 ng g−1 and above in Single Quad mode. This instrument setup has been effectively used for measuring 237Np in spiked uranium samples52 and sediment samples following chemical separation,53 with the latter achieving a detection limit of 25.0 pg g−1 (6.5 × 10−4 Bq g−1).
![]() | ||
Fig. 4 Signal at m/z = 237 with increasing concentration of 238U in Single Quad mode. 0 CPS were measured in MS/MS mode. |
Plutonium-239 has been extensively measured by ICP-MS, often in combination with 240Pu to determine the source of nuclear contamination through the difference in 239Pu/240Pu isotopic ratio values.54–56 The ICP-MS/MS setup has been shown to reduce the reliance on offline chemical separation from 238U, which otherwise interferes via tailing and formation of polyatomic 238U1H.12,13 The Agilent 8800 used in this study has a maximum m/z value of 260, limiting the number of cell products that can be formed in NH3 and O2 mode. Recently developed ICP-MS/MS instruments can measure at higher m/z values, and this has been shown to be beneficial for actinide measurement and separation through double oxide formation.57
In NH3 mode at a flow rate of 3 mL min−1, the majority (96%) of the Pu signal remained on mass. By comparison, only 18% of the total 238U signal remained on mass, with the remainder measured as U(NH) and U(NH2). When 239Pu was measured on mass in NH3 mode, a 238U1H interference was still present, but was reduced by an order of magnitude compared to MS/MS no gas mode (Fig. 5). At an O2 flow rate of 0.3 mL min−1, >95% of the U and Pu was shifted to an oxide. Compared to MS/MS no-gas and NH3 mode, the interference from 238U1H is significantly reduced when measuring Pu as 239Pu16O. This is a result of 238U1H being converted to 238U16O in the reaction cell, with limited formation of 238U16O1H (Fig. 5). At a 238U concentration of 10 μg g−1, the signal at m/z = 239 was 45600 CPS in MS/MS no gas, 5100 CPS in MS/MS NH3, and 165 CPS in MS/MS O2 mode when measuring Pu as 239Pu16O.
![]() | ||
Fig. 5 Impact of 238U concentration on tailing and hydride formation at m/z = 239 in MS/MS mode, NH3 mode and O2 mode (Pu measured as 239Pu16O). |
Increasing the O2 gas flow rate reduced the 238U16O1H formation, at the expense of 239Pu16O sensitivity (Fig. 6). The optimal flow rate was determined to be 0.4 mL min−1, with the background at m/z = 255 (i.e.239Pu16O) from a 10 μg g−1 238U solution of 45 CPS, compared to 165 at 0.3 mL min−1. Under the optimised conditions, the instrument LOD for 239Pu in this study was 0.27 mBq g−1 (0.12 pg g−1).
Parameter | Stable | 63Ni | 90Sr | 93Zr | 99Tc | 129I | 237Np | 239Pu | |||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Mode | MS/MS | ||||||||||||
Q1–Q2 | Various | 63–114 | 90–90 | 93–195 | 99–131 | 129–129 | 237–237 | 239–255 | |||||
RF (W) | 1550 | ||||||||||||
Carrier gas (L min−1) | 0.65 | ||||||||||||
Dilution gas (L min−1) | 0.35 | ||||||||||||
Extract 1 (V) | 0 | ||||||||||||
Extract 2 (V) | −175 | ||||||||||||
H2 (mL min−1) | 3.00 | 3.00 | |||||||||||
He (mL min−1) | 4.00 | 1.00 | 2.00 | ||||||||||
O2 (mL min−1) | 0.40 | 0.30 | 0.55 | 0.40 | |||||||||
NH3 (mL min−1) | 1.00 | 0.15 | |||||||||||
Energy discrimination (V) | 5.0 | −13.0 | −18.0 | −9.0 | −7.0 | −6.0 | −7.0 | ||||||
Octopole bias (V) | −18.0 | −4.5 | 0.0 | −2.8 | −5.0 | −2.0 | −5.0 | ||||||
Internal standard | 115In/209Bi | 115In | 115In | 115In | 115In | 130Te | 209Bi | 209Bi |
Iodine-129 is included in Table 6, however, because it must be prepared in alkaline solution due to its volatility, it could not be included in the same run. Instead, following the measurement of other radionuclides, multiple washes were added to the procedure, which proceeded automatically after the end of the previous run: (1) 10 minutes in 0.3 M HNO3, (2) 20 minutes in deionised water to wash acid out of the instrument and (3) 20 minutes conditioning in 0.001 NaOH. The 129I procedure then started automatically. A 10 ng g−1 Te standard was monitored at m/z = 130, which was manually spiked into the solutions rather than using the internal standard line.47
The ‘stable’ mode in Table 6 operates in MS/MS mode with He collision gas at a flow rate of 4 mL min−1. It is therefore suitable for several radionuclides that do not suffer from isobaric interferences. Polyatomic interferences should be suppressed by the collision gas, whilst tailing removal is improved by operating in MS/MS mode. This mode is considered suitable for 226Ra, 231Pa, 232Th, 234U, 235U, 238U, 241Am and 243Am. Neptunium-237 could also be run in this mode, as the long half life means low detection limits can still be achieved even following signal suppression from the use of He. This approach also enables an additional mode to be used, with procedures published elsewhere for ICP-MS/MS measurement of 41Ca,58151Sm59 and 236U.9 It should also be noted that the procedure developed for 239Pu is also applicable to 240Pu.
For each sample type, matrix-matched calibration standards using multi-element stable standard solutions and radionuclide standard solutions were produced. The calibration curves were used to determine the method LOD, calculated as the equivalent concentration of three times the standard deviation of the blank in unspiked samples. Each sample run consisted of 10 replicates and 100 sweeps per replicate.
The concentration of a range of stable elements spanned several orders of magnitude (Table 7), indicating the range of matrix compositions of the samples measured. The values shown are the concentrations in the original samples, factoring in corrections for sample dilution where this was applied. It must be noted that a higher dilution factor was used for solid samples digested using borate fusion prior to measurement, resulting in a much lower concentration in the samples measured compared to in the starting material. Amongst the stable elements tested were those that represent isobaric interferences for radionuclides of interest to this study, including 90Zr (90Sr), 63Cu (63Ni), 93Nb (93Zr) and 99Ru (99Tc), as well as tailing interferences including 88Sr (90Sr), 98Mo (99Tc) and 238U (237Np and 239Pu).
Element | Concentration rangea (ng g−1) |
---|---|
a Corrected to concentration per unit mass of the starting sample. | |
Co | <1 to 3.7 × 103 |
Ni | <1 to 7.1 × 103 |
Cu | 2 to 4.4 × 103 |
Zn | 5 to 1.7 × 104 |
Sr | 2 to 1.9 × 104 |
Y | <1 to 1.6 × 103 |
Zr | 1 to 4.4 × 104 |
Nb | 1–470 |
Mo | 2–340 |
Ru | <1–35 |
Cs | <1–330 |
Ba | 20 to 5.7 × 104 |
Th | <1–700 |
U | 2–1000 |
A mixed 50 ng g−1 Be/In/Bi internal standard was used to assess instrument drift and any suppression of the signal and reduced ion transmission due to solid sample deposition on the interface cones over time.
Method detection limits for each of the radionuclides of interest were determined (calculated from the equivalent concentration of three times the standard deviation of the blanks in unspiked samples) in their optimised mode, compared to the instrument LOD for interference-free samples (Fig. 7).
The decontamination factor achieved by the collision/reaction cell was a key consideration. This was calculated for each sample matrix by dividing the CPS at the interfering m/z value compared by the CPS at the analyte m/z value in each optimal instrument mode. For example, for 90Sr, the 90Zr decontamination factor was calculated as the CPS at m/z = 106 (where 90Zr preferentially forms 90Zr16O) to the CPS at m/z = 90 (where 90Sr is expected). To ensure that the interference was the result of an isobaric interference, where possible an isotope of the same element as the isobar was also measured (e.g.91Zr for 90Zr) and the ratio between the counts compared to natural ratios. This was not possible for monoisotopic 93Nb, however, the maximum Zr concentration measured (880 ng g−1 in sediment) was not high enough to result in 92Zr tailing in MS/MS mode, whilst there was no evidence of polyatomic 92Zr1H formation following a mass shift of m/z +195. Therefore, the signal at m/z = 93 was expected to be the result of isobaric 93Nb.
Across all samples tested for 63Ni, the average cell decontamination factor of isobaric 63Cu using NH3 + H2 was 6.9 × 102. For 63Ni standards in the optimum instrument mode (measured as 63Ni(NH3)3), the target was a background of <5 CPS. This compared to a range of 7–182 CPS (average 38 CPS) in groundwater samples, 123–1560 CPS (average 940 CPS) for air filters, and 108–701 CPS (average 360 CPS) across the solid samples tested. The instrument LOD was calculated as 0.50 Bq g−1 (0.24 pg g−1), increasing to an average of 25.6 Bq g−1 (12.0 pg g−1) in aqueous waste samples, 30.7 Bq g−1 (14.4 pg g−1) in groundwaters, 128.5 (60.4 pg g−1) Bq g−1 across the solid samples tested and 331.2 Bq g−1 (155.7 pg g−1) in air filter samples.
The relatively short half life of 63Ni (1 Bq g−1 is equivalent of 0.5 pg g−1) results in both a lower sensitivity and higher impact of interferences relative to the same activity concentration of longer-lived radionuclides. The highest isobaric 63Cu concentration measured (4.4 ng g−1) in final samples was measured in air filters, which produced the highest calculated detection limit of the samples measured. Whilst the reaction cell does achieve some decontamination of interferences, even relatively low 63Cu concentrations will contribute to the 63Ni signal. For air filters and groundwater samples in particular, there was a significant range in instrument backgrounds measured. One groundwater sample had a background in the optimal setup of 182 CPS, compared to the next highest value of 49 CPS. If this higher background sample was not included, the method detection limit across the samples measured will have been 9.0 Bq g−1 (4.2 pg g−1), compared to 30.7 Bq g−1 (14.4 pg g−1). It was a similar case for air filter samples, with three samples having a background of > 900 CPS, compared to 123 CPS for the fourth air filter measured. In both cases, this shows the importance of measuring a range of sample matrices when determining the application of this method.
The results for 90Sr followed a similar trend to those for 63Ni and was the shortest-lived radionuclide in this study. In agreement with previous studies, O2 was an effective reaction gas in shifting 90Zr to 90Zr16O, whilst the majority of the 90Sr signal remained on mass. The average cell decontamination factor of 90Zr across the samples tested was 6.0 × 102. The instrument background in the optimised setup was <5 CPS. A similar value was achieved for groundwater samples (maximum 8 CPS), increasing to 9–56 CPS (average 40 CPS) in air filter samples and a significantly higher average value of 1880 CPS (168–2960 CPS) across the solid samples tested.
The relatively short half life for 90Sr meant it had the highest instrument LOD of any radionuclide in this study, with a calculated value of 1.0 Bq g−1 (0.2 pg g−1). This also meant that any increase in background at m/z = 90 had a more significant impact on the detection limit compared to the other radionuclides investigated. The low background across groundwater samples resulted in a method LOD that matched the instrument LOD of 1.0 Bq g−1 (0.2 pg g−1), increasing to 3.2 Bq g−1 (4.2 pg g−1) in air filters, 90.0 Bq g−1 (17.6 pg g−1) in aqueous waste samples and 173.9 Bq g−1 (34.1 pg g−1) across the solid samples tested. The increase in background is primarily from isobaric 90Zr. Despite the decontamination achieved by the reaction cell, concentrations of up to 880 ng g−1 were measured that will not be completely removed by reaction cell separation alone and will contribute significantly to the background. The highest stable 88Sr concentration of 380 ng g−1 is not expected to result in a tailing interference when measured in MS/MS mode.
For 93Zr, in instrument blank solutions, the background was 0 CPS, with a background of <10 CPS across all samples tested, suggesting the significant mass shift of 102 from m/z = 93 to m/z = 195 (i.e. (93Zr(NH3)6) effectively reduces interferences, primarily from 93Nb). The method LOD was on the order of 1 × 10−4 Bq g−1 (1.1 pg g−1), in agreement with those achieved in previous studies.14,15
Technetium-99 used O2 cell gas for interference removal. The background for samples tested was not significantly different to the instrument background (<10 CPS), with the instrument and method LOD values of 3.0 × 10−4 Bq g−1 (0.5 pg g−1) and 3.1 × 10−4 Bq g−1 (0.6 pg g−1). Direct measurement as 99Tc16O2 was possible in the samples measured due to the relatively low Ru (isobaric) and Mo (tailing and polyatomic 98Mo1H) concentrations. A Ru concentration of 10 ng g−1 increased the background at m/z = 131 by approximately 160 CPS, compared to 25000 CPS for a 1 Bq g−1 99Tc solution, suggesting that samples containing higher Ru concentrations will require offline chemical separation in combination with reaction cell separation. The maximum Mo concentration of 110 ng g−1 did not significantly contribute to the instrument background (Table 8).
Radionuclide | Instrument LOD (Bq g−1, pg g−1) | Method LOD (Bq g−1, pg g−1) |
---|---|---|
a Iodine-129 only measured in aqueous samples | ||
63Ni | 0.5 (0.3) | 25.6 (12.1)–331.2 (155.7) |
90Sr | 1.0 (0.2) | 1.0 (0.2)–173.9 (34.1) |
93Zr | 1.3 × 10−5 (0.1) | 1.7 × 10−4 (2.2) |
99Tc | 3.0 × 10−4 (0.5) | 3.1 × 10−4 (0.6) |
129Ia | 8.1 × 10−5 (5.2) | 8.6 × 10−5 (5.1) |
237Np | 1.0 × 10−5 (0.4) | 1.1 × 10−4 (4.2) |
239Pu | 1.6 × 10−4 (0.07) | 1.6 × 10−3 (0.7) |
Iodine-129 was measured on-mass, using O2 reaction gas to suppress the background from isobaric 129Xe present as an impurity in the Ar plasma gas, in agreement with previous studies.44–46 Operating in MS/MS mode effectively removed any potential interference from 129I tailing. MS/MS mode will not remove 127I1H2 formed in the plasma, but this can be suppressed by operating with O2 reaction gas. Polyatomic 97Mo16O2 has been recorded as interfering with the background at m/z = 129 at Mo concentrations of 500 ng g−1 and above,47 which is higher than the maximum value from samples in this study of 110 ng g−1. Iodine-129 was only measured in aqueous samples, as it’s volatile nature means borate fusion dissolution is not an appropriate sample preparation technique. As with 99Tc, the instrument and method LOD calculated were very similar (8.1 × 10−5 Bq g−1 (5.2 pg g−1) and 8.6 × 10−5 Bq g−1 (5.1 pg g−1), respectively).
Neptunium-237 was measured on mass in MS/MS mode, with a background of 0 CPS in the instrument and all method blank samples. This showed the MS/MS setup to effectively remove 238U tailing in the samples tested (with 238U concentrations up to 1000 ng g−1), removing the need for offline chemical separation. The method LOD was on the order of 1.1 × 10−4 Bq g−1 (4.2 pg g−1). The 238U tailing removal was also beneficial for 239Pu measurement, whilst measurement of 239Pu as 239Pu16O minimised the impact of 238U1H through effective conversion to 238U16O. The background was <10 CPS across the samples measured, with a method LOD calculated as 1.6 × 10−3 Bq g−1 (0.7 pg g−1).
Regulatory limits vary between organisations and countries depending on the nature of the sample being measured. The range of samples measured are a good indicator of whether the method developed for simultaneous radionuclide measurement is applicable for end users as a rapid approach for assessing sample composition. The detection limits calculated were compared to several regulatory limits related to drinking water, environmental permitting, radiation protection and waste exemption.60–63 For 93Zr, 99Tc, 129I and 237Np, the detection limits are lower than the regulatory limits (Fig. 7), whilst for 239Pu, the detection limit (1.6 × 10−3 Bq g−1) was lower than all limits other than the WHO drinking water limit (1 mBq mL−1).
As expected, radionuclides with a shorter half life are less likely to be candidates for direct low-level measurement compared to long-lived radionuclides. Despite the relatively high detection limits for 63Ni, (25.6–331.2 Bq g−1 (12.1–155.7 pg g−1)), this value is lower than some the regulatory limits tested, meaning that direct measurement is feasible for some applications. For 90Sr, a highly important radionuclide with regards to decommissioning, waste characterisation and public safety, the detection limit calculated is generally higher than the regulatory limits. It is therefore necessary that measurement is preceded by chemical separation and pre-concentration prior to measurement. This is the approach that has been taken in previous studies, with reduced procedural times achieved using online separation, for example using sequential injection.5,43,44 Chemical separation would also enable decay counting measurement of both 63Ni and 90Sr by LSC to check the agreement between decay counting and mass spectrometric techniques. Additionally, other sample dissolution techniques that can handle higher starting sample masses compared to borate fusion (such as open vessel acid leaching) have not been considered, which may be beneficial for relatively short-lived radionuclides.13
Preliminary results from stable Ni and Sr standards using an Agilent 8900 suggests the instrument sensitivity and cell-based interference removal is considerably improved compared to the previous generation Agilent 8800 used in this study, which could be particularly beneficial for shorter-lived radionuclides. In the case of Sr, the sensitivity was approximately five times higher using the newer instrument, with more efficient formation of 90Zr16O and no change in 90Sr16O formation. Compared to the Agilent 8800 used in this study, the updated Agilent 8900 has a new interface vacuum stage for improved ion transmission and an updated collision reaction cell with axial acceleration, which further improves sensitivity and reaction cell efficiency, respectively.
The results show the ability to directly measure multiple radionuclides without prior separation, which minimises the analyst time, secondary waste produced, loss of analyte during the procedure, and the need for tracer addition, which all offer economic benefits to the end user, and improves sample throughput. It should be noted that the radionuclides in Table 8 are an example of the combination that can be tested, and the user can select up to eight modes depending on the target radionuclides of interest. Initial data suggests that additional radionuclides can be included in the combined procedure, specifically 36Cl and 93Mo.25
In this study, several O2 modes have been used that have been customised for individual radionuclides, but it may be possible to develop a single mode that would be applicable to 90Sr, 99Tc, 151Sm and 239Pu. Such an approach is likely to reduce sensitivity for all radionuclides compared to the individually tailored method used in this study, which would have a more significant effect on shorter-lived radionuclides. Again, the approach used ultimately depends on the number of radionuclides that must be measured, and the limits of detection required.
The instrument used in this study has four gas lines-dedicated H2 and He, one non-corrosive line (O2) and one corrosive line (NH3). The use of other gases was not considered, for example 135Cs/137Cs has been effectively measured by ICP-MS/MS using N2O as a reaction gas to shift isobaric 135Ba and 137Ba to oxides. Alternative gases could potentially enable detection of radionuclides that to date have not been measured by ICP-MS, such as 59Ni. Additionally, the instrument used in this study requires the NH3 to be balanced in at least 90% He to protect the cell. Other ICP-MS/MS instrument designs can tolerate higher NH3 concentrations, which may offer more efficient cell-based interference separation and expand the number of radionuclides measurable.
The procedure developed demonstrates the capabilities of ICP-MS/MS for direct measurement of radionuclides without prior treatment, as well as showing the amount of information that can be generated about a sample in a single measurement. There are additional radionuclides that could potentially benefit from the high sample throughput and improved interference separation of ICP-MS/MS, for example 59Ni. Additionally, there are reaction gases not tested in this study that may further improve interference removal or expand the number of radionuclides measurable, such as N2O for measurement of 135Cs/137Cs. There is also the possibility of measuring isotopic ratios to determine the source of contamination, for example 127I/129I, 135Cs/137Cs, and 239Pu/240Pu.
This journal is © The Royal Society of Chemistry 2023 |