Development of a High-sensitivity Inductively Coupled Plasma Mass Spectrometer for Actinide Measurement in the Femtogram Range

I Journal of 1 Analytical 1 Atomic Spectrometry

REMO CHIAPPINI , JEAN-MICHEL TAILLADE AND SOPHIE BREBION

CEA-DIRCENISMSRB: Service Mixte de Surveillance Radiologique et Biologique de 1 'homme et de 1 'environnement, BP No 208, 91 31 1 Montlhby Cedex, France

A quadrupole inductively coupled plasma mass spectrometer was modified and developed to achieve very low detection limits, down to a few femtograms in the actinide range. This type of performance was achieved through improvements in the interface pumping system of a conventional ICP-MS instrument, resulting approximately in a 3-fold gain in sensitivity, a 3-fold reduction in background noise, a 2-fold reduction in protenated uranium abundance and a 4-fold reduction in detection limits. A further gain in sensitivity and detection limits, by a factor of 7, was obtained using a high- efficiency desolvating nebulizer which further reduces protenated uranium abundance. The detection limit achieved with good reproducibility, using the high-efficiency desolvating nebulizer on standard solutions, is close to 1.2 fg. The performance of the new instrument is discussed and compared with that of radiometric techniques, with particular reference to plutonium, using standard solutions and real samples. The 'half-life cut off' between ICP-MS and alpha-spectrometry is close to 650 years. Intercomparison of plutonium measurements from 300 pg down to 35 fg in 16 environmental samples shows good agreement. The plutonium detection limits in real matrices are greater than those in clean samples and vary from 1.5 to 3 fg. The high-sensitivity ICP-MS system gives access to environmental contamination of other long-lived radionuclides, e.g., 99Tc, 237Np and uranium isotopes, which are sometimes difficult to measure at very low levels using conventional techniques.

Keywords: Inductively coupled plasma mass spectrometry; actinide measurement; environmental contamination; high- sensitivity; detection limits

Once considered as a research instrument, ICP-MS is now routinely used in many laboratories. Using a classification defined by Laitinen,' who discerns seven distinct ages in the lifetime of an analytical method, Horlick2 suggests that ICP-MS is now near the end of the fifth age of maturity and about to enter the age of reflection. In a recent overview, Gray3 conducts such a reflection and discusses the major current limitations of the technique. These limitations are inherent in the technological concept and identifying them makes it possible to take full advantage of instrumentation performance. One of the more remarkable aspects of that performance is sensitivity, together with speed and multi- element capability. However, instruments available in a con- ventional configuration generally provide detection limits that are too high to measure certain radionuclides present in the environment at very low levels. Detection limits achievable by quadrupole ICP-MS in the actinide range have been deter- mined by many ~orkers.~- ' ' They vary from one instrument to another. They are reported to be of the order of 1 pg ml-I and can be lowered by associating the ICP-MS with introduc- tion modules superior in performance to conventional liquid

nebulizers, namely the ultrasonic nebulizer or the ETV graph- ite furnace.

The use of a graphite furnace makes it possible to lower detection limits to a few f e m t ~ g r a m s . ~ q ~ ~ * ' ~ However, the ETV is difficult to operate since it presents problems in the areas of reproducibility, stability and memory effects. For that reason, it is rarely used on a routine basis. An ultrasonic nebulizer can increase sensitivity by a factor of 10 and provides lower detection limits.' A recent paper14 reports a detection limit of 20 fg ml-'. For radionuclides with a half-life close to 2 x lo4 years, this limit is comparable to that achieved through alpha- spectrometry with a 4000 min counting time, namely Bq. Thus, in a conventional configuration, even coupled to an ultrasonic nebulizer, ICP-MS is supposed to outperform alpha- spectrometry only for those radionuclides with a longer half- life than 239Pu.

In order to lower ICP-MS detection limits even further, a new approach has been explored by the SMSRB laboratory (part of the CEA: Atomic Energy Commission) in charge of radioactivity monitoring on the French nuclear test sites. A conventional instrument configuration was modified by using an oversized pumping system between the sampler and skim- mer. These changes have resulted in much improved perform- ance; sensitivity has been increased, background noise and protenated uranium abundance reduced. The work presented in this paper explores the possibilities offered by the new instrument and the capabilities obtained by coupling this instrument to a high-efficiency desolvating nebulizer.

EXPERIMENTAL

Instrumentation Description

A PlasmaQuad 2 + (Fisons Instruments, Winsford, Cheshire, UK) was modified by the manufacturer. A 40m3 per hour primary pump was added to the 25m3 per hour unit which controlled the vacuum in the sampler-skimmer interface. The interface pressure was upgraded from 1.8 x lop3 to 0.85 x bar. In consideration of the new pressure equilib- rium, cone spacing was adjusted. The conventional configur- ation can be restored simply by operating a switching valve. This new interface is commercially available as the Fisons 'S' option.

These are the only changes made to the basic instrumen- tation. The sample is delivered using a peristaltic pump (Minipuls 3, Gilson, Villiers-Le-Bel, France) and a glass concen- tric nebulizer (supplied by Fisons Instruments). The sample then enters a water-cooled Scott spray chamber.

A high-efficiency desolvating nebulizer system was also used. Designated 'Mistral' by its manufacturer (Fisons Instruments), it induces desolvation of the sample by raising its temperature with a radiative heat source, then cooling it by Peltier effect

Journal of Analytical Atomic Spectrometry, July 1996, Vol. 11 (497-503) 497

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online / Journal Homepage / Table of Contents for this issue

condensers. The glass concentric nebulizer itself is of the low- capacity type, less than 0.4 ml min- ', and it self-aspirates.

Performance Comparison of the Three Configurations Using Standard Solutions

ICP-MS operating conditions in the three configurations investigated

The performance of three distinct configurations was studied using standard solutions: the conventional configuration found in most of the recent systems currently used, a high- performance (HP) configuration resulting from the changes described above and a configuration (HP/Mistral) associating high performance with the Mistral desolvating nebulizer. Configuration operating conditions are summarized in Table 1.

In order to compare the performance, the nebulizer argon flow rate was allowed to vary. Ion optics and torch position were optimized each time the nebulizer flow rate was changed. The other conditions reported in Table 1 were held constant.

Standard solutions used for the measurements

Commercially available standard solutions of uranium, 99Tc, 236U, 237Np, 241Am (LMRI, Gif-sur-Yvette, France), 230Th, 236Pu, 239Pu, 241Pu, 241Pu, 242Pu, 243Am (Amersham QSA, Les Ulis, France) and a multi-element solution (Spex, Longjumeau, France) were used. For the measurements described herein, standards were diluted in 2% m/m nitric acid from 18 Mi2 cm de-ionized water (Mini-Q water, Millipore, S' Quentin- Yvelines, France) and concentrated, high-purity nitric acid (Ultra-pure Grade, Prolabo, Fontenay-sous-Bois, France).

Measurements on standard solutions

Tuning and data acquisition. Before studying the performance of each configuration, the instrument was allowed to equilibrate for at least 1 h, then instrument sensitivity was optimized on uranium. Mass calibration, short-term stability over 10 min, then long-term stability over 4 h were checked, using the multi- element solution of Li, Co, In, Cs, Pb, Bi and U (2 pg 1-l of each for conventional and HP configurations, 0.2 pg 1-' of each for HP/Mistral configuration).

The measurements were acquired with a resolution set at a value of 0.85 at 10% of peak height. Each measurement involved three acquisitions in the peak-jumping mode, with three points per peak and a dwell time of 10 ms. The integrating time was fixed at 5 s per peak in all instances.

Performance Jigures investigated. The performance figures investigated were selected based on criteria associated with trace actinide measurements:

Sensitivity determined at m/z 238. This is representative of most actinides since they show ionization potentials close to

that of uranium (about 7 eV). Sensitivity is expressed in counts s-' per ppm (parts-per-million).

Background noise in counts s-', measured at m/z 240. Protenated uranium abundance, 238UH+: 238Uf. This is a

net protenated uranium abundance since background noise measured at m/z 240 was subtracted from the signal measured at mlz239. The protenated uranium is a 239Pu isobar and is likely to interfere with plutonium measurements when uranium is in excess in the solution.

Memory effect at m/z 238. The memory effect represents the time required for signal attenuation by a factor of 1000.

Mass bias per m/z unit, as measured at m/z 235 and 238 on the uranium standard solution.

Short-term stability, which is the relative standard deviation for ten acquisitions, at 1 min intervals for a 10 min period, from the already specified major isotopes of the multi- element solution.

Long-term stability, which is the relative standard deviation for acquisitions at 10 min intervals, over a 4 h period, from the same isotopes.

Before the measurements, we checked that the 238U signal was linear with respect to concentration until it reached 1.6 x lo6 counts s-'. As such, we verified that a 20 ns dead time was adequate to compensate for counting losses. Variations in 238U sensitivity, background noise and protenated uranium abundance were studied by varying the nebulizer flow rate. For each nebulizer flow rate value, instrument lenses and torch position were tuned in order to optimize sensitivity on uranium. The other parameters shown in Table 1 were held constant. In these studies, the uranium standard solution (12 pg 1-l in the conventional configuration, 6 yg 1-l in the HP configuration and 1 pgl-' in the HP/Mistral configur- ation) was continuously introduced. After each study was complete, a measurement at the maximum uranium sensitivity was repeated to assess the influence of instrument drift. These data are reported as unconnected points in Fig. 1. A radio- chemical analysis, performed on the uranium standard solution, indicated that the 239Pu:238U ratio was less than

Short- and long-term stability, mass bias as well as memory effects were measured close to the maximum uranium sensitivity.

Measurements on Environmental Samples

In order to assess the analytical capabilities of the ICP-MS technique in the new configurations when applied to environ- mental samples, three intercomparisons were conducted in the SMSRB laboratory, for three distinct categories of activity.

In the first intercomparison, several radionuclides were measured by ICP-MS in the HP configuration and by one of the radiometric techniques, namely alpha-, gamma- and liquid scintillation spectrometries, in 1 g of a sediment from the

Table 1 ICP-MS operating conditions in the three configurations investigated

Incident power/W Nebulizer argon flow rate/l min-' Intermediate argon flow rate/l min-' Outer argon flow rate/l min-' Sample flow rate/ml min-l Condenser temperaturePC Heater temperature/"C Interface pressurelbar Quadrupole pressure/bar Multiplier dead time/ns

Conventional configuration

1350 0.7-0.85 1.1

13.5 0.85 5

1.8 x 1.8 x lo-''

-

20

HP configuration

1350 0.74-0.90 1.1

13.5 0.85 5

-

0.85 x 10-3 1.2 x 10-9

20

HP/mistral configuration

1350 0.79-0.93 1.1

13.5 0.27 & 0.02* 1

135 0.85 10-3 1.2 x 10-9

20

* This value varies because the nebulizer self-aspirates.

498 Journal of Analytical Atomic Spectrometry, July 1996, Vol. 11

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online

0.70 0.72 0.74 0.76 0.78 0.80 0.82 0.84

0.74 0.76 0.78 0.80 0.82 0.84 0.86 0.88 0.90- - 14

- 12

0.79 0.81 0.83 0.85 0.87 0.89 0.91 0.93 Nebulizer argon flow rate/ I min-'

Fig. 1 Variation in 238U signal (A), background noise at m/z 240 (B) and 238UH+: 238Uf (C) as a function of nebulizer argon flow rate for: (a) conventional configuration, (b) HP configuration and (c) HP/Mistral configuration

Sellafield region. The following isotopes were selected: 234U,

These measurements related to activity levels up to 1000 Bq kg-' of 239+240Pu. In order to validate ICP- MS/HP/Mistral plutonium measurements at lower levels, a specific intercomparison with alpha-spectrometry covering Mururoa environmental samples was carried out. These were lagoon sediment samples in which radioactivity originated from atmospheric testing fallout.

The third intercomparison related to very low plutonium activity measurements. Four celestial grouper fish samples from the Mururoa atoll lagoon were measured by ICP- MS/HP/Mistral and alpha-spectrometry.

238u, 239+240pu 241pu, 2 4 1 ~ ~ . Y

Environmental sample preparation procedure and measurement conditions

For Sellafield and Mururoa sediment samples, the same prep- aration procedure was performed on plutonium. The same chemical separations were also applied before MS or alpha- spectrometry. Known amounts of 236U, 236Pu, 242Pu and 243Am were added to the sample before treatment. The aliquots were dry-ashed and complete dissolution was carried out using nitric, hydrochloric and hydrofluoric acids. Finally, the samples were dissolved in an 8 mol I- ' nitric acid medium. Sodium nitrite was added to perform a redox cycle. Then, successive

anion-exchange chromatographic separations were performed to eliminate the matrix, concentrate the analyte and separate uranium from plutonium so as to minimize potential inter- ferences as protenated uranium. Dowex AGl-X8 resins (BioRad, Ivry-sur-Seine, France) in 8 moll-' nitric and 10 moll-' hydrochloric acid solutions were used to separate most of the matrix and most of the uranium from plutonium. Dowex AGl-X4 resins (BioRad) in hydrochloric and nitric acid media were then used to optimize the separation of plutonium from americium and uranium.

For the mass spectrometric measurements, the concentration of each actinide was calculated by isotope dilution using the following net isotope intensities from the samples diluted in 5 ml of 2% m/m ultra-pure nitric acid: 236U:234U, 236U.238U - 9

242Pu and 243Am were used as reference isotopes so there was no need to correct for sample dilution, matrix effect and chemical yield. The isotope equilibration between the spike and the sample was ensured through the dissolution process and the redox cycle. Spikes of 243Am and 242Pu were added simultaneously to the samples when plutonium and americium were measured in the Sellafield samples. 243Am was determined when plutonium isotopes were measured and 242Pu was deter- mined when americium isotopes were measured in order to assess the separation of 241Pu and 241Am. The protenated uranium abundance was determined using a uranium standard solution (6 pgl-' for HP configuration and 1 pgl-' for HP/Mistral configuration) before and after the measurements. Mass bias determined on uranium at rn/z235 and 238, on plutonium at m/z 239 and 242 and on standard solutions after the ICP-MS measurements were used to correct uranium, plutonium and americium raw data, assuming that the bias per m/z unit on americium was the same as that of plutonium. A blank procedure was performed in advance of the analysis of the samples and 2% m/m ultra-pure nitric acid was measured as a blank between each sample.

Mass spectrometric measurements were carried out at the optimum condition setting shown in Table 2 (see under Results and Discussion for the determination of the optimum condition setting), in the peak-jumping mode, using the acquisition parameters defined in the 'tuning and data acquisition' para- graph. The alpha-spectrometry counting time was fixed at 4000 min for the Sellafield and Mururoa sediment samples.

For the very low plutonium activity determinations, the chemical separation procedure was similar to that used for the Sellafield and Mururoa sediments but some differences were introduced. In order to minimize uranium concentration, anion-exchange chromatographic separation using Dowex AGl-X4 resins in 8 moll-' nitric acid solution was performed twice and ultra-pure reagents were used to rinse the columns and elute plutonium. Protenated uranium abundance was measured before, during and after the measurements using a 1 pg1-I uranium standard solution in order to correct the 239Pu concentration precisely. As the Mistral was used, total dissolved solids had to be minimized because they generate a decrease in instrument sensitivity. Therefore, after elution, the dried residues were weighed with a highly precise balance in order to assess the matrix purification. Residues were then dissolved in 1 ml of 2% m/m ultra-pure nitric acid containing 10 pg ml-' of 236U as an internal standard. This isotope was added in order to check the variation in sensitivity during the measurements. A 2% m/m ultra-pure nitric acid blank contain- ing 10pgml-' of 236U was acquired between each sample. Background noise was acquired at rnlz237, 241 and 243 in order to estimate the background carefully. Mass bias meas- ured at m/z235 and 238 on the 1 pgl-' uranium standard solution was used to correct plutonium measurements assuming that its bias was the same as that of uranium. In order to avoid any contamination, plutonium standard solu-

242pu.239pu, 242pu:240pu, 242pu:241pu and 2 4 3 ~ ~ ~ 2 4 1 ~ ~ . 236u 7

Journal of Analytical Atomic Spectrometry, July 1996, VoE. 1 1 499

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online

Table 2 ICP-MS average performance and plutonium detection limits at the optimum condition setting for the three configurations investigated

Optimum condition setting (nebulizer flow rate in 1 min-’) Sensitivity (counts s-’ per ppm) at m/z 238 Background noise (counts s-l) at m/z 240 Protenated uranium abundance Detection limits* for plutonium/fg ml-I

Conventional HP HP/mistral configuration configuration configuration

0.76-0.78 0.80-0.82 0.82-0.84 8 x lo7 2.1 x lo8 1.5 x 109

2.8 x 10-5 1.4 x 10-5 1.1 x 10-5 7.0f 1.1 2.0 f 0.6 2.0 & 0.6

40 9 1.2

* Detection limits based on 3a of the blank equivalent concentration at m/z 240.

tions were not used and a 6% m/m ultra-pure nitric acid solution was introduced for 3 min after each sample. Moreover, new tubing and completely decontaminated cones, nebulizer, spray chamber and torch were used for the measurements. Mass spectrometric measurements were carried out at the optimum condition setting shown in Table2 and the acqui- sition parameters were the same as those described for the Sellafield samples. For alpha-spectrometry, the acquisition time was 15 d.

RESULTS AND DISCUSSION

Performance Data for the Three Configurations Investigated Using Standard Solutions

Fig. 1 shows the variations in 238U sensitivity, protenated uranium abundance and background noise as a function of nebulizer flow rate for conventional, HP and HP/Mistral configurations. The error bars represent one standard deviation in the measurement. The standard deviation of the net proten- ated uranium abundance was calculated by error propagation using the variances in the background at m/z 240 and the gross signal at mlz239. Mass bias, memory effect at m/z238, and short- and long-term stability are reported in Table 3.

A number of observations can be made: In the conventional and HP configurations, sensitivity passes

through a maximum and background through a minimum. These optima do not correspond exactly to the same nebulizer flow rate. The H P configuration increases sensitivity while lowering background noise, which lowers the detection limit. An additional benefit is a reduced protenated uranium abun- dance at m/z 239. Mass bias, memory effect and stability (short- and long-term) are comparable in both configurations.

The gain in sensitivity results from the higher pressure difference between the plasma and the interface (see Table 1) . However, the background noise reduction process is less clear- cut. A study15 reports that noise generally originates from the photons which reach the detector, from a weak discharge in the lens system at the elevated ion-lens voltage or from the presence of charged species. Another recent study16 appears to indicate that it is the combination of plasma noise frequencies close to the analyser scanning rates that is at the root of the problem and that the greatest contribution to the noise is the random spatial distribution of the analyte in the central channel of the plasma. How the new configuration and the increased pressure differential at the interface affect these processes is not clear and cannot be explained with the data of this study. A connection between quadrupole pressure and background noise could be considered because the pressure is

lowered from 1.8 x lo-’ to 1.2 x lo-’ bar (Table 1); however, we could not find any published data that could confirm such a connection.

Different papers7*10*17 describe the interference of 238UH +

on 239Pu+ but do not quantitatively measure the 238UHf abundance. Only one studyx8 describes completely the para- metric behaviour of the protenated uranium. In our work, the protenated uranium abundance data obtained in the conven- tional configuration are lower but consistent with data men- tioned in all these reference^.^,^^,^^,^^

Coupling the ICP-MS in its HP configuration with the Mistral enhances sensitivity which leads to a further reduction in the detection limit. Furthermore, protenated uranium abun- dance is slightly reduced. All these improvements in perform- ance are due to the introduction of a dry aerosol into the plasma. Removal of water vapour (and consequently removal of hydrogen), which reduces plasma water loading, increases the signal and decreases the protenated uranium abundance. These observations are consistent with data already pub- lished.18 Background noise is close to that measured in the H P configuration. The only adverse effect of this new, more complex introduction system is a slight lowering of short- and long-term stability. The memory effect is longer because the sample flow rate is reduced by a factor of 3 and the rinse-out volume only reduced by a factor of about 2.

Plutonium detection limits in standard solutions at the optimum condition setting for the three conjigurations investigated

For each configuration, an optimum condition setting can be distinguished. Under these conditions, the best detection limits can be achieved in the actinide range. If average values of sensitivity and background noise under the optimum con- ditions are taken into account, a detection limit defined as three times the uncertainty in the blank equivalent concen- tration at m/z240 can be calculated for plutonium as its sensitivity was verified to be the same as that of uranium. The optimum condition setting, average values of sensitivity, back- ground noise, protenated uranium abundance and the detec- tion limits are reported in Table 2.

At the optimum condition setting, the HP configuration increases sensitivity by a factor of 2.5, and decreases back- ground noise by a factor of 3.5, which lowers 240Pu detection limits by a factor of 4-5 with respect to the conventional configuration. Protenated uranium abundance is lowered by a factor of 2. HP/Mistral further reduces protenated uranium abundance and the detection limit by a factor of 7 to reach 1.2 fg ml-l. These tendencies could differ slightly from one day

Table 3 ICP-MS performance data using standard solutions for the three configurations investigated

Conventional HP HP/mistral configuration configuration configuration

Memory effect at mlz 238 2 min 2 min 3.5 min Mass bias per m/z unit (measured on 235U and 238U) < 1% < 1% <1% Short-term stability (10 min) <2% < 2% <3% Long-term stability (4 h) <4% < 4% < 5%

500 Journal of Analytical Atomic Spectrometry, July 1996, Vol. 11

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online

to another but performance is reproducible for each configur- ation at the optimum condition setting.

As the HP configuration reduces protenated uranium abun- dance by a factor of 2, the 239Pu detection limit is lowered by more than a factor of 5 with respect to the conventional configuration, when the measurements are carried out on samples including uranium. Indeed, the 239Pu detection limit is based on three times the uncertainty in the blank equivalent concentration at m/z 239 and this uncertainty decreases when protenated uranium abundance is reduced. However, in the context of interference management, the separation of uranium from plutonium before the measurements remains an essential step because the U : Pu ratio in environmental samples gener- ally exceeds lo5 and can reach more than lo9 for sea-water.

None of the studies on quadrupole ICP-MS instruments reported in the literature show detection limits in the actinide range as low as those obtained by coupling an ICP-MS instrument in the HP configuration with the Mistral system. This confirms that, as stated by HoukIg in a recent overview, the ICP-MS technique is still making progress after 10 years of continuous improvement. Fig. 2 illustrates a comparison of the three configurations investigated. Mass spectra were acquired at the optimum condition setting for each configur- ation, from the same 2% m/m ultra-pure nitric acid solution containing 2 ng ml-1 of natural uranium, 10 pg ml-1 of 237Np, 239Pu and 242Pu, as well as 90fgml-1 of 240Pu. Acquisitions were carried out over 1 min, in the scan mode with a dwell time of 320ps, 25 channels per u and a resolution set at a value of 0.70 at 10% of peak height.

Enhanced sensitivity and reduction in background noise are clearly shown. 240Pu and 234U gradually emerge from the background as system performance improves. These two radio- nuclides are present in the solution at respective concentrations of 90 and 110 fg ml-', which attests to the very low detection limits provided by the ICP-MS/HP/Mistral system.

Comparison of Instrumental Detection Limits Achieved on Standard Solutions With Alpha-Spectrometry, Beta-Counting and With the ICP-MS/HP Associated With the Mistral Desolvating Nebulizer

The detection limits achieved with the ICP-MS/HP/Mistral combination, defined as three times the uncertainty in the blank equivalent concentration, were compared with those of competing techniques, namely alpha-spectrometry and beta- counting, for the radionuclides listed in Table 4. The limits were determined on standard solutions without interferences. Thus, data reported in Table 4 are the best limits achievable. For mass spectrometric determinations and for each isotope, a blank and two solutions with different concentrations dis- solved in 2% m/m ultra-pure nitric acid were successively measured under the peak-jumping mode conditions already described. As a 1 ml solution volume was used, it was possible to convert masses per unit volume into masses. Conversion of masses into activities was achieved using half-lives data.20 Alpha-spectrometry detection limits are given for a 4000 min measurement period, while the beta-counting detection limit relates to an acquisition period of 100 min.

The comparison shows that the ICP-MS technique could, theoretically, be used to advantage for measurements of radio- nuclides with a radioactive half-life in excess of 650 years. It should also be noted that, for plutonium, performance figures are not directly comparable since alpha-spectrometry measures the sum of 239Pu and 240Pu activities whereas the ICP-MS separates these isotopes. The only 'radioactivity versus ICP-MS cut off' of a few hundred years that has been reportedl3 was estimated from a fundamental standpoint assuming 1000 min counting times for the radiation measure-

I

loa

la2 10

1

lob 104

v)

3 0 3 los

g 1% 2

10

1

lo6

5 10

104

los

lo2

10

1

Fig. 2 Mass spectra acquired in three distinct ICP-MS configurations [(a) conventional configuration, (b) HP configuration, (c) HP/Mistral configuration] from the same 2% m/m ultra-pure nitric acid solution containing 2 ng ml-' of natural uranium, 10 pg ml-' of 237Np, 239Pu and 242Pu, as well as 90 fg ml-l of 240Pu

ments, complete ETV sample introduction efficiency for ICP-MS and a single ion monitoring detection mode.

Measurements on Environmental Samples

Intercomparison of ICP-MS/HP with radiometric techniques

This comparison was also the subject of an intercomparison at the European level with the participation of ten international laboratories under NPL sponsorship. In this context, 230Th, 232Th, 235U, 237Np and "Tc were also measured by ICP- MS/HP but the data are not reported here because they have to be published under National Physical Laboratory (Teddington, Middlesex, UK) sponsorship.

Protenated uranium interference was taken into account but the correction of 239Pu for 238UH was negligible because uranium was mostly separated from plutonium prior to the measurements. The spike of 236U was adjusted so that the

Journal of Analytical Atomic Spectrometry, July 1996, Vol. 11 501

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online

Table 4 ICP-MS/HP/Mistral instrumental detection limits as compared with those of alpha-spectrometry and beta-counting on standard solutions

ICP-MS/HP/Mistral*

Isotope

99Tc 230Th 232Tht 2 3 4 ~

235Ut 238Ut 237Np 238PUJ 239Pu 240Pu 241Am

Half-life/years

2.135 x lo5 7.541 x lo4 1.406 x 10" 2.455 x lo5 7.037 x lo8 4.468 x lo9 2.14 x lo6

2.411 x lo4 6.564 x lo3 4.322 x lo2

87.74

Mass/g

4.8 x 10-15 1.2 x 10-15 4.0 x 10 - 14

1.2 x 10-15 4.0 x 10-15 4.0 x 10-14 1.2 x 10-15

1.2 x 10-15 1.2 x 10-15 1.2 x 10-15

-

Activit y/Bq

3.0 x

1.6 x lo-''

3.2 x 10-l' 5.0 x 3.1 x lo-'

2.8 x

9.2 x 10-7

2.8 x 10-7

-

1.0 x 10-5 1.5 x 10-4

a-Spectrometry and P-counting/Bq

( P ) 10-4 10-4 (a) 10-4 (a) 10-4 (a) 10-4 (a) 10-4 (a) 10-4 (a) 10-4 (a) 10-4

(a)

* Detection limits are defined as three times the uncertainty in the blank equivalent concentration. t The relatively high detection limits for 232Th, 235U and 238U are attributable to the high blank values of the nuclides present in the reagents

and water used. As no 232Th standard solution was available, the 232Th detection limits was determined by assuming that the sensitivity of this isotope was the same as that of 230Th.

$ 238Pu cannot be measured at low levels because of the high blank value of 238U in the reagents and water used.

correction of 236U for 235UH on the uranium isotope determi- nation was negligible.

242Pu and 243Am measurements indicated that Pu-Am separations were complete so that isobaric interferences between 241Pu and 241Am were neglected.

Individual results are the arithmetical mean of five measure- ments on separate aliquots. Mean values and standard devi- ations of the determinations are shown in Table 5. Mass spectrometric standard deviations, derived from propagation of experimental uncertainty, include all random and systematic errors involved in the chemical treatment and measuring process.

Mean values are very close and do not differ by more than 5%. Paired, two-tailed t-tests indicated no significant difference at the 95% confidence interval between ICP-MS and radio- metric determinations. The enhanced performance of the new ICP-MS configuration made it possible to improve measure- ment conditions with respect to the conventional configuration and to determine 241Pu with a good precision despite its short half-life.

Intercomparison of ICP-MS/HP/Mistral with alpha- spectrometry covering low-level plutonium measurements

Mean values and standard deviations of the determinations are reported in Table 6.

The radioactivity of the samples analysed during the course of this investigation ranged from 8.1 to 148 Bq kg-' of 239+240Pu. Thus, considering the sample size, 1, 2.5 or 5 g, and the chemical yields, compared data from MS and alpha- spectrometry ranged from 0.025 and 0.14 Bq, i.e., from 9.5 and 55 pg of plutonium.

There is a small difference in the MS and alpha-spectrometric measurements of plutonium. Averaged over all the samples, the relative difference in the plutonium activities was approxi- mately 5%. This is supposed to be caused by the use of different isotope dilution spikes, i.e., 236Pu for alpha-

spectrometry uersus 242Pu for MS. Despite the small differences, paired, two-tailed t-tests indicated no significant difference at the 95% confidence interval between ICP-MS and alpha- spectrometry determinations,

Intercomparison of ICP-MS/HP/Mistral with alpha- spectrometry covering very low-level plutonium measurements

The comparison was also the subject of an intercomparison at the global level with the participation of eight laboratories. Results have been published under IAEA sponsorship.21

After the purification steps, the dried residues represented less than 0.2 mg. Less than 150 pg ml-' of uranium were also measured in the samples. Considering the HP/Mistral sensi- tivity, protenated uranium abundance represented not more than 2 counts at m/z 239. A loss of sensitivity varying from 10 to 30% was observed on the samples with respect to the blank.

Mean results in Bqkg-' of fresh sample and standard deviations are reported in Table 7.

Sample activities range from 4.3 x lo-' to 3.5 x lop4 Bq, corresponding to plutonium masses of 17-143 fg.

Paired, two-tailed t-tests, performed on 'West' and 'North' grouper samples indicated no significant difference at the 95% confidence interval between ICP-MS and alpha-spectrometry determinations.

This intercomparison confirms the very low detection limits of ICP-MS/HP/Mistral when applied to environmental samples, lower than those obtained with alpha-spectrometry. Plutonium detection limits in environmental samples are higher than in standard solutions. This is due to the lower sensitivity but also to the protenated uranium abundance that increases background noise at m/z 239 because uranium was not com- pletely separated from plutonium. A concentration of 150 pg ml-' increases the signal at m/z 239 by approximately 2 counts s-' and the background noise by a factor of 2. Consequently, plutonium detection limits in real samples vary from 1.5 to 3 fg.

Table 5 Intercomparison of measurements in a Sellafield sediment using ICP-MS/HP, and alpha-, gamma- and liquid scintillation spectrometries

Isotope 2 3 4 ~

238u

239 + 240pu

239Pu 240Pu 241Pu 241Am

ICP-MS/HP/Bq kg-' a-SpectrometrylBq kg- y-SpectrometrylBq kg - Liquid scintillation/Bq kg -' 34.9 f 1.7 34.5 & 1.4 999 & 35 560 & 27 439 f 22

15 650 +_ 890 1571 & 63

34.2 & 3.5 34.5 k 2.9 947 f 53

1560 & 55 14 900 & 1100

1552+84

502 Journal of Analytical Atomic Spectrometry, July 1996, Vol. 11

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online

Table 6 Intercomparison of 239+240Pu measurements in Mururoa lagoon sediment samples using ICP-MS/HP-Mistral and alpha-spectrometry

ICP-MS/HP/Mistral/Bq kg - ' Sediment number

1 2 3 4 5 6 7 8 9

10 11

Mass/g

2.5 2.5 2.5 2.5 2.5 2.5 2.5 5.0 1 .o 1 .o 1 .o

a-SpectrometrylBq kg - ' 10.3 0.6 12.1 k0.7 16.2 f 0.9 21.6+ 1.1 23.6 k 1.2 27.6 & 1.5 25.8 & 1.5 8.1 f0.4

73.3 f4.5 121 f 7 136 k 7

239 + 240pu

10.4 f 0.4 12.5 f0.5 16.1 rt_ 0.6 23.7 k 0.9 26.2+ 1.1 29.3+ 1.1 27.0 f 1 .O

8.4 f 0.4 68.9 f 2.3 128 f 5 148 k 5

239Pu

8.86 f 0.37 10.9 & 0.4 14.3 k0.5 21.1 k0.8 23.4 f 1.0 26.2 k 1.0 22.7 f 0.9 7.52 k 0.30 58.3 2.2

108.0k4.5 125.0 f 4.5

240Pu

1.57 f 0.14 1.63f0.12 1.79 f 0.1 1 2.60k0.13 2.76 f 0.17 3.13 f0.21 4.31 f0.40 0.86 k 0.09 10.6 k 0.5 19.7 f 1.0 23.1 k 1.1

Table 7 Intercomparison of 239f240P~ measurements in grouper fish samples using ICP-MS/HP/Mistral and alpha-spectrometry

ICP-MS/HP/Mistral (Bq kg-') Grouper sampling zone Mass/g a-SpectrometrylBq kg - 239 + 240pu 239Pu 240Pu

West South North East

10.0 (3.8 f 0.7) x ( 4 . 0 k 0 . 4 ) ~ lop4 (3 .64f0 .35)~ low4 (0 .36k0.12)~

15.1 (2.7f0.6) x lop4 (3.2f0.4) x lop4 (2.88k0.34) x (0 .32k0 .16)~ 7.6 ( 1 . 7 k 0 . 6 ) ~ (1.4k0.4) x ~ 2 . 1 x 10-5

5.1 < 1.7 x 10-4 (1.1 k0.3) x G 3.3 x 10-5

Routine Radionuclide Trace Measurements by ICP-MS in Environmental Samples

The ICP-MS technique, having been validated for many radionuclides in an extensive range of radioactivity levels, is now routinely used by the SMSRB laboratory for very low- level measurements on environmental samples. Soil and water samples containing 239+240Pu, 9 9 T ~ , 237Np and uranium iso- topes are routinely analysed after chemical separation.

CONCLUSION

ICP-MS is a technique in which great improvements are still being made. The technological development presented in this paper opens new prospects in the field of long-lived radio- nuclide measurements, at trace levels in environmental samples. The 'half-life cut off' between ICP-MS and competing tech- niques such as radiometry has been lowered by a factor of 4-5 in the actinide range. It is now below 1000 years. These new capabilities not only give access to very low activities of plutonium, down to a few femtograms in environmental samples, but also to radionuclides such as 99Tc, 237Np and uranium isotopes, which were formerly difficult to measure at very low levels by conventional techniques. Further develop- ments can be envisaged to reach even lower detection limits. This could be achieved through a more efficient ion introduc- tion into the interface by lowering the interface pressure even further. This is currently being investigated in our laboratory.

REFERENCES

1 Laitinen, H. A., Anal. Chem., 1973, 45, 2305. 2 Horlick, G., J . Anal. At. Spectrom., 1994, 9, 593. 3 Gray, A. L., Anal. Proc., 1994, 31, 371.

4

5 6

7

8

9

10

11 12

13

14

15

16

17 18 19 20

21

Hursthouse, A. S., Baxter, M. S., McKay, K., and Livens, F. R., J. Radioanal. Nucl. Chem., 1992, 157, 281. Crain, J. S., and Mikesell, B. L., Appl. Spectrosc., 1992, 46, 1498. Kim, C. K., Morita, S., Seki, R., Takaku, Y., Ikeda, N., and Assinder, D. J., J. Radioanal. Nucl. Chem., 1992, 156, 201. Scott, R. D., Baxter, M. S., Hursthouse, A. S., McKay, K., Sampson, K., and Toole, J., Anal. Proc., 1991, 28, 382. Shaw, T. J., and Francois, R., Geochim Cosmochim. Acta, 1991, 55, 2075. Kim, C. K., Takaku, Y., Yamamoto, M., Kawamura, H., Shiraishi, Y., Igarashi, Y., Igarashi, S., Takayama, H., and Ikeda, N., Radioisotopes., 1989, 38, 153. Igarashi, Y., Kim, C. K., Takaku, Y., Shiraishi, Y., Yamamoto, M., and Ikeda, N., Anal. Sci., 1990, 6, 157. Blair, P. D., Trends Anal. Chem., 1986, 5, 220. Hall, R. J. B., James, M. R., Wayman, T., and Hulmston, P., Plasma Source Mass Spectrosc., 1989, Special Publ. No. 85, 145. Smith, M. R., Wyse E. J., and Koppenaal, D. W., J. Radioanal. Nucl. Chem., 1992, 160, 341. Crain, J. S., Smith, L. L., Yaeger, J. S., and Alvarado, J., J. Radioanal. Nucl. Chem., Articles, 1995, 194, 133. Kim, C.-K., Seki, R., Morita, S., Yamasaki, S. I., Tsumura, A,, Takaku, Y., Igarashi, Y., and Yamamoto, M., J. Anal. At. Spectrom., 1991, 6, 205. Gray, A. L., Williams, J. G., Ince, A. T., and Liezers, M., J. Anal. At. Spectrom., 1994, 9, 1179. Russ, G., and Bazan, J., Spectrochim. Acta, Part B, 1987, 42, 49. Crain, J. S., and Alvarado, J., J. Anal. At. Spectrom., 1994,9, 1223. Houk, R. S., Acc. Chem. Res., 1994, 27, 333. Browne, E., and Firestone, R. B., in Table of Radioactive Isotopes, ed. Shirley, V. S., Wiley-Interscience, 1986. Ballestra, S., Baxter, M. S., Gastaud, J., Petterson, H., Povinec, P. P., and Vas, D., IAEA/AL/085 report, International Atomic Energy Agency, Monaco, 1995.

Paper 6/000161 Received January 2, 1996

Accepted April 11, 1996

Journal of Analytical Atomic Spectrometry, July 1996, Vol. 11 503

Publ

ishe

d on

01

Janu

ary

1996

. Dow

nloa

ded

by U

nive

rsity

of

Pitts

burg

h on

30/

10/2

014

18:5

3:16

. View Article Online