Ultra-trace determination of ²²⁶Ra in thermal waters by high sensitivity quadrupole ICP-mass spectrometry following selective extraction and concentration using radium-specific membrane disks

Abstract

A method has been developed for the determination of ²²⁶Ra in highly saline (TDS ≈ 6 g l⁻¹) thermal waters, using ultrasonic nebulization and a quadrupole-based instrument equipped with a high sensitivity pumping interface. A very low instrumental limit of detection of 2 fg g⁻¹ was achieved when the pressure in the expansion chamber was decreased from ca. 2 to 0.85 mbar. Prior to ICP-MS measurements, Ra was preconcentrated and isolated from matrix elements using various separation schemes. The conventional, single-stage cation-exchange chromatographic techniques proved to be limited, for waters with high TDS, by the fact that all the barium and the tail of Sr and Ca peaks were found in the Ra fraction and induced severe ion signal suppression. Two-step procedures involving a clean-up column based either on cation exchange in perchloric acid medium (to remove Ca and reduce Sr) or on specific extraction chromatographic material (to remove Sr and Ba) gave satisfactory results but required time-consuming sample handling. Commercially available radium-specific solid phase extraction disks, originally designed for radioactive counting methods, provided a straightforward alternative, allowing a radium fraction suitable for ICP-MS measurements to be obtained in a single step, using a simple filtration device.

---

Introduction

²²⁶Ra is widespread at the ultra-trace level in natural samples, where it occurs as a relatively long-lived (half-life = 1600 years) intermediate isotope of the radioactive decay series of ²³⁸U. Disequilibria between parent and daughter isotopes of the U and Th series, caused by chemical fractionation processes, are extensively used as chronometers and tracers in geological, oceanographic and environmental sciences.¹ These allow one to calculate residence times and draw inferences on fluid movements, whatever their igneous or aqueous nature. These applications often require, among others, precise measurements of Ra concentrations.

In additon to this academic interest, ²²⁶Ra attracts much attention as one of the most toxic natural radioelements, owing to its high specific activity, 1 pCi pg ⁻¹, compared with that of ²³⁸U, 3 pCi g⁻¹. For these reasons, the determination of ²²⁶Ra in environmental samples is an important analytical task.²

Methods used to measure ²²⁶Ra largely rely on α-spectrometry, either directly after electrodeposition³ or indirectly using the ²²²Rn emanation technique.⁴ γ-Spectrometry can also be used for measuring both ²²⁶Ra and ²²⁸Ra, especially in rock samples.⁵ More recently, isotope dilution thermal ionization mass spectrometry (ID-TIMS)^6–8 and inductively coupled plasma mass spectrometry (ICP-MS)^9–11 have also been used. As shown for several other radioelements,¹² mass spectrometric methods offer better sensitivity and alleviate the need for long α-counting times and a ca. 1 month waiting period for ²²²Rn ingrowth in the indirect emanation technique.

In every case, it is necessary to preconcentrate and purify theradium fraction prior to spectrometric analyses. Time-consuming coprecipitation and cation-exchange methods have largely been used for that purpose. However, it is difficult to achieve a good separation of Ra from Ba owing to their closely related chemical properties, and several purification steps are required.^7,8,13 Therefore, there is still a need for more efficient and straightforward methods for the separation of Ra from water and other environmental samples.¹³ In this paper, we describe an ICP-MS method for the determination of ²²⁶Ra in highly saline thermal waters, using ultrasonic nebulization and a quadrupole-based instrument equipped with a high sensitivity pumping interface. Prior to ICP-MS measurements, the analyte was preconcentrated and isolated from matrix elements using various separation schemes. Among these, a simple and effective method was designed, based on commercially available Ra-specific solid phase extraction membrane disks, which were originally developed for nuclear spectrometry purposes.

Experimental

Instrumentation

A VG PlasmaQuad 2+ (VG Elemental, Winsford, UK) equipped with a high efficiency ion transmission interface (S-Option) was used. The enhanced sensitivity interface has already been described.^14–17 Briefly, this involves a large additional rotary pump (one stage, 80 m³ h⁻¹, Edwards, Crawley, Sussex, UK) connected in parallel with the standard pump, in order to enhance the vacuum in the sampler–skimmer volume (expansion chamber). With this switchable pump on, a pressure of ca. 0.8 mbar (instead of ca. 2 mbar) can be obtained in the expansion chamber. A U-5000 AT ultrasonic nebulizer (Cetac Technologies, Omaha, NE, USA) was used for sample introduction. In this device,¹⁸ the solution is continuously fed using a peristaltic pump (Perimax 12; Spetec, Erding, Germany). The dense aerosol produced by the ultrasonic transducer is swept by an Ar carrier flow into a desolvation unit consisting of an electrically heated furnace followed by a conventional, water-cooled condenser.

Reagents and separation materials

De-ionised water was further purified with a Milli-Q (Millipore) system to reach a 18.2 MΩ cm resistivity. All acids were of analytical reagent grade, purified by sub-boiling distillation in silica glass or PTFE stills. Cation-exchange resin (AG50W-X8 and AG50W-X4, both 200–400 mesh particle size; Bio-Rad, Richmond, CA, USA) and Sr. Spec (100–150 µm particle size) extraction chromatography material (Eichrom Industries, Darien, IL, USA) were used for column work. Empore Radium Rad disks of 47  mm diameter (3M Filtration Products, St. Paul, MN, USA) were used for direct concentration and isolation of Ra from water samples, using a polycarbonate vacuum filtering device (SM 165 10; Sartorius, Palaiseau, France) and a manual vacuum pump (Poly-Labo, Strasbourg, France).

Samples

Water samples were taken from three CO₂-rich thermal springs, located near Châtel-Guyon (France), that had already been studied for radium by γ-spectrometry and/or ID-TIMS.¹⁹ A detailed description of the hydrogeological context of these springs has been given elsewhere.^20,21 These waters are characterized by a high content of total dissolved solids (TDS ≈ 6 g l⁻¹) with Na, Ca and Mg as major cations (Table 1). After filtration on a 0.2 µm cellulose acetate membrane (Millipore), the samples were acidified to ca. 0.05 N with distilled 7 M HNO₃. Samplings were made over a period of several months during the development stage of this study. Aliquots of 500 ml were used for experimental work throughout this study.

Table 1 Typical concentration ranges of major cationic and anionic species contained in Châtel-Guyon thermal waters

Cation	Concentration/mg l^−1	Anion	Concentration/mg l^−1
K	80–95	SO^2−4	360
Na	750–950	Cl^−	1700–2200
Ca	570–680	HCO^−3	2500
Mg	350–400

---

Radium preconcentration

Taking into account the extreme salinity of the studied waters (TDS ≈ 6 g l⁻¹), their direct ICP-MS analysis for Ra was not possible. Therefore, several methods were investigated for eliminating most of the unwanted ions while preconcentrating Ra.

Method 1. First, a simple separation based on cation-exchange chromatography was used to remove all the anionic species and those elements forming cations which are not strongly retained from dilute mineral acids. A 10 ml amount of resin (AG50W-X8, 200–400 mesh particle size) suspended in H₂O was placed into a 15 mm id polypropylene column (Bio-Rad, ref. 732–1010), cleaned successively with 6 M HCl, H₂O and 5 M HNO₃ before preconditioning with 2 × 5 ml of 0.05 M HNO₃. After loading the 500 ml sample overnight with a peristaltic pump, Mg, Fe, Na, Mn and most of the Ca were washed off from the resin with 7 × 10 ml of 1.75 M HCl. Then, 2 ml of H₂O were added, before eluting Ba (+Ra) with 5 × 5 ml of 5 M HNO₃ and evaporation to dryness in a FEP Teflon beaker. Using this scheme, an Ra fraction also containing all the barium and most of the strontium of the sample was obtained. A variant of this procedure, involving 100 ml (instead of 70 ml) of 1.75 M HCl, was later used in order to improve the elimination of Ca.

Method 2. A two-step separation was used, combining the cation-exchange column described above and a second small, thermostated (50 °C) column (8 mm id) containing 1 ml of AG50W-X4 (200–400 mesh) resin. The dry residue left after the first column was dissolved with 0.3 ml of 9 M HClO₄ and loaded on to the column preconditioned with 2 × 0.5 ml of 9 M HClO₄. After rinsing the sample beaker with 0.3 ml of 9 M HClO₄, the radium (and most of the Ba) was eluted with 5 ml of 9 M HClO₄ while Ca and Sr remained on the column. These elements were subsequently washed off with 10 ml of 6 M HCl.

Method 3. In this case, a column containing 2 ml of the extraction chromatographic material Sr.Spec (100–150 µm particle size) was used for a clean-up step following the 10 ml cation-exchange resin (Method 1). The dry residue from the first step was taken up in 0.5 ml of 5 M HNO₃ and loaded on the Sr. Spec column preconditioned with 2 ml of 5 M HNO₃. After rinsing the FEP beaker with 0.5 ml of 5 M HNO₃, Ra was eluted with 6 ml of 5 M HNO₃. Under these conditions, breakthrough of Ba did not occur earlier than 8 ml and Sr was strongly retained on the column.

Method 4. Finally, a single-pass procedure based on the Empore Ra-specific disks was evaluated. In this scheme, the 500 ml water sample was brought to a nitric acid strength of ca. 2 N with 175 ml of 7.55 M HNO₃ and allowed to drain through the disk, previously wetted with 20 ml of 2 M HNO₃. The disk was held in a filtration device and a weak vacuum was applied at the bottom in order to keep the flow rate to slightly less than 50 ml min⁻¹. Then, the disk was rinsed with 20 ml of 2 M HNO₃ to displace any unwanted element. In order to recover Ra, the silica support of the Ra-specific organic extracting agent was dissolved with 20 ml of 5% HF, followed by 25 ml of aqua regia containing 10% H₂O₂. This fraction was evaporated to dryness. The residue was treated with 1 ml of 48% HF and 25 µl of 70% HClO₄ in order to bring to completion the removal of Si as volatile SiF₄ and destroy any remaining organic compounds.

ICP-mass spectrometry

The ICP-MS operating conditions and data acquisition parameters are given in Tables 2 and 3, along with those relevant to the ultrasonic nebulizer used for sample introduction. The Ra fraction was taken up in 5 ml of 0.5 M HNO₃ containing 1 ppb ²⁰⁹Bi as an internal standard. The ²²⁶Ra calibration standard solution, containing 550 ppq (fg g⁻¹), was prepared by diluting a stock standard solution kindly provided by Dr. M. Condomines. This primary solution was obtained by separation from a natural uranium ore (pitchblende from Joachimstal, Czech Republic) and calibrated by isotope dilution thermal ionization mass spectrometry using a ²²⁸Ra tracer.

Table 2 ICP-MS and ultrasonic nebulizer operating conditions

ICP and interface—
Forward power/W	1350
Reflected power/W	<10
Plasma gas flow rate/l min^−1	13.0
Auxiliary gas flow rate/l min^−1	2.11
Expansion chamber pressure/mbar	0.85
Ultrasonic nebulizer—
Sample solution delivery/ml min^−1	0.7
Carrier gas flow rate/l min^−1	0.75
USN furnace temperature/ °C	140 ± 1
Condenser temperature/ °C	3 ± 1

---

Table 3 Data acquisition parameters

Acquisition mode	Peak jumping
Selected masses	209, 226
Number of points per peak	1
Dwell time/ms	2.56
Settle time/ms	1.5
Acquisition time/s	30
Number of repeat measurements	6
Detector	Channeltron, pulse counting
Dead time/ns	20

---

Results and discussion

Precision and accuracy

The Ra concentration data obtained on seven different samples processed through the four different separation schemes evaluated in this work are listed in Table 4. Values measured by γ-spectrometry and/or ID-TIMS on aliquots of the same samples¹⁸ are also given for comparison. The Ra concentrations determined in the thermal waters range from 35 to 60 fg g⁻¹. Taking into account the 100-fold enrichment factor (i.e., 500 ml of sample processed, 5 ml of solution analysed by ICP-MS), the Ra content of the solutions analysed ranged from 3.5 to 6 pg g⁻¹. The corresponding relative standard deviations (RSDs) were between 1 and 3%, with only one higher value of 5.8%. The results obtained for duplicate or triplicate analyses of the same batch of sample do not differ by more than 2.5%, demonstrating the good overall reproducibility of the separation and mass spectrometric procedures. Likewise, a satisfactory accuracy is indicated by the fair agreement of our ICP-MS results with the concentrations independently determined, for some samples, by γ-spectrometry and/or ID-TIMS.

Table 4 Ra concentrations measured in different samples of thermal waters. These data were obtained after Ra separation (with a concentration factor of 100) using the four different methods investigated in this study. I, II, III stand for replicate analyses, throughout the whole procedure, of the same batch of sample

a Values determined on the same sample by ID-TIMS(*) or γ-spectrometry (**) with their 1σ errors are also given (in parentheses) for comparison.^31,34 b In this experiment, 100 ml (instead of 75 ml) of 1.75 M HCl were used to achieve a more complete elution of calcium.
Separation method		Ra/fg g^−1	RSD (%)^a
Method 1: single cation-exchange column—
Sample 1	I	59.3	2.5
	II	58.3	2.5
Sample 2	I	39.1	5.8 (41 ± 0.8*, 39.1 ± 0.7**)
	II	39.4	3.1
Sample 3^b	I	35.2	1.2
	II	35.6	1.0
Method 2: two-step cation-exchange separation—
Sample 1		59.7	2.7
Sample 2		39.7	1.3 (41 ± 0.8*, 39.1 ± 0.7**)
Sample 4	I	29.4	1.3 (29.8 ± 0.5**)
	II	28.7	1.5
Sample 5	I	25.7	1.3(22.0 ± 0.6**)
	II	26.3	2.0
Method 3: cation-exchange + extraction chromatography—
Sample 6		44.7	0.6
Method 4: single pass separation using Ra-specific solid phase extraction disks—
Sample 7	I	55.6	1.6
	II	56.6	1.4
	III	55.9	1.7

---

Separation–preconcentration procedures

Four distinct separation schemes were evaluated in the course of this study. Method 1 involved a conventional cation-exchange resin column, as used previously for Ra preconcentration from water samples prior to ICP-MS analyses.^9–11 In this procedure, based on distribution coefficients available for most elements in HCl²² and HNO₃²³ media, 70 ml of 1.75 M HCl were used to remove all the elements having distribution coefficients lower than that of Sr. Ra was eluted, together with all the barium and part of the strontium contained in the sample, with 5 M HNO₃. Nitric acid was preferred to HCl, based on lower distribution ratios (D) of alkaline earth elements for the same acid strength.^22,23 Although reasonably precise and accurate results were obtained (Table 4), a strong ion signal suppression was indicated by the poor (<10%) internal standard response (ISR, defined as 100 × ²⁰⁹Bi sample/²⁰⁹Bi blank). Semi-quantitative analyses performed on these Ra separates showed that, in addition to Ba, large amounts of Ca and Sr (representing 2% and ca. 10%, respectively, of the amounts initially present in the 500 ml sample) still occurred in the analysed solution (Table 5). Presumably, these account for the considerable signal attenuation which was observed. Based on these data, the 1.75 M HCl wash volume was increased to 100 ml. This resulted in a significant improvement of the internal standard response, which increased to 35–40% (Table 6).

Table 5 Amounts (µg) and proportion (%) of matrix elements found as residual impurities in Ra fractions (5 ml) separated using a single cation-exchange column. These data are based on semi-quantitative analyses with an estimated accuracy of ±10%

Sample No.	Mg	Ca	Fe	Sr	Ba
1	26	1100	35	250	52
	(0.06%)	(2%)	(0.8%)	(7%)	(95%)
2	15	1750	60	220	28
	(0.02%)	(2%)	(0.8%)	(13%)	(98%)

---

Table 6 Signal attenuation caused by residual matrix elements, as monitored by the internal standard response ISR (%) = 100 × ²⁰⁹Bi sample/²⁰⁹Bi blank. I, II, III stand for replicate analyses, throughout the whole procedure, of the same batch of sample

Separation method		ISR (%)
a In this experiment, 100 ml (instead of 75 ml) of 1.75 M HCl were used to achieve a more complete elution of calcium.
Method 1: single cation-exchange column—
Sample 1	I	7
	II	8
Sample 2	I	5
	II	8
Sample 3^a	I	40
	II	35
Method 2: two-step cation-exchange separation—
Sample 1		65
Sample 2		71
Sample 4	I	61
	II	64
Sample 5	I	49
	II	45
Method 3: cation-exchange + extraction chromatography—
Sample 6		87
Method 4: single pass separation using Ra-specific solid phase extraction disks—
Sample 7	I	75
	II	82
	III	78

---

In an attempt to improve further the purity of the Ra fraction, Method 2 was evaluated. This involved a clean-up step, based on the strong affinity of Ca (D = 140) and Sr (D = 29) relative to Ba (D = 4.9) and Ra (D = 2.0) for the AG50 cation-exchange resin from 9 M HClO₄ solutions.²⁴ In order to overcome the slow kinetics of ion exchange in concentrated perchloric acid, a resin with a lower degree of cross-linking (AG50W-X4) and higher temperature (50 °C) were used, as recommended by Nelson et al.²⁴ This second column acted as a filter capable of retaining the Ca and most of the Sr left as impurities in the Ba + Ra fraction of the first cation-exchange column. A distinct improvement of the internal standard response (which ranged from 45 to 71%, depending on the samples) was obtained in this way (Table 6).

Another experiment (Method 3), aimed at removing Ba and Sr from the Ra fraction obtained from the initial cation-exchange column, involved a commercially available column filled with an Sr-selective extraction chromatographic material, Sr. Spec. In addition to Sr (weight distribution ratio D ≈ 400 in 5 M HNO₃), the crown ether used in this material has the potential to extract Ba to a reasonable extent from nitric acid solutions (D ≈ 80 in 5 M HNO₃), whereas Ra is not appreciably extracted under these conditions (D ≈ 5 in 5 M HNO₃).²⁵ This property has already been used to isolate Ra from silicate rocks prior to TIMS measurements.⁸ The Ra fraction purified in this way gave the best internal standard response (87%) measured in this work (Table 6).

As an alternative to the time-consuming and relatively tedious two-step procedures, a fourth method was investigated, making use of commercially available Ra-specific solid phase extraction (SPE) disks (Empore Radium Rad disks). These disks are ca. 0.5 mm thick membranes consisting of silica chromatographic particles loaded with a radium-selective organic extractant and enmeshed in an inert network of PTFE fibrils.²⁶ These and other related SPE membranes have been described in some detail elsewhere,^27,28 but the precise nature of the molecule used for Ra extraction has not been specified. These membranes are conveniently used in the filtration mode, but perform as SPE columns. As yet, their application seems to have been restricted to radiochemical analyses of aqueous samples using counting methods, for which they are extremely well suited, since the disks can be assayed directly.^28–31 It has been shown that quantitative (95% or better) recoveries of radium can be achieved for 1 l samples (2 M in HNO₃) containing each of the potentially interfering cations Na⁺, Mg²⁺ and Ca²⁺ at the 10 000 ppm level, while Ba²⁺ and Sr²⁺ can be tolerated up to the 10 ppm level.^26,28 Sample volumes up to 3 l have been successfully processed²⁸ but Ba levels >500 µg may interfere.³² In the case of the highly saline thermal waters studied in this work, the sum of the cations Ca, Mg, Sr and Ba was decreased by a factor of about 5000 when Ra was extracted from 0.5 l samples (Table 7), thereby confirming the high degree of selectivity achievable with these SPE membranes.

Table 7 Amounts (µg) of matrix elements left as residual impurities in the Ra fractions (5 ml) separated using solid phase extraction disks, based on semi-quantitative analyses. The sum of these cations is about 0.02% of the amount initially present in the 500 ml water samples, corresponding to a decontamination factor of ca. 5000

	Mg	Ca	Sr	Ba	Total
I	4	6	7	8	25
II	4	5	5	6	20
III	8	10	4	2	24

---

Basic solutions of the Na or K salt of EDTA have been recommended by the manufacturer for eluting Ra from the disk.³² However, these stripping agents are not suitable for ICP-MS because they would add large amounts of easily ionizable elements responsible for severe matrix effects. An attempt to elute Ra using 40 ml of 6 M HCl was not successful. Therefore, we decided to dissolve the inert silica particles supporting the Ra-specific extracting agent. A 20 ml volume of dilute hydrofluoric acid was used for that purpose; 25 ml of aqua regia containing 10% H₂O₂ were further passed through the disk in order to rinse the PTFE membrane and digest the organic extractant. Albeit crude, this method provides an effective removal of Ra and gives an Ra fraction ready for ICP-MS analysis. Matrix induced signal suppression, as estimated from the internal standard response, was limited to about 20% (Table 6).

In order to estimate the overall blank associated with the Radium Rad disk preconcentration–separation method, 500 ml of Milli-Q water were acidified to 2 N HNO₃ and processed using Method 4. The blank measured under these conditions was 300 fg. Although not of great concern in this study (in which the amounts of radium processed ranged from 17 to 30 pg), this blank level is relatively high, and is interpreted to reflect a significant contribution of Ra-rich water samples processed previously. Indeed, the polycarbonate vacuum filtering device showed an obvious degradation (as indicated by a distinct yellowish colour and some microcracks), probably resulting from the aqua regia treatments. Altered polymer surfaces are likely to be prone to memory effects, as documented, for example, in a study dealing with neodymium.³³ Clearly, it would be advisable to use a filtering device made of more resistant material (e.g., Teflon PFA). Alternatively the Radium Rad disk could be removed from the filtering unit after the extraction and rinsing steps, and subsequently dissolved in a Teflon beaker.

This Ra extraction procedure is extremely fast, as it requires less than 30 min of analyst time for sample filtering and Ra stripping, and would be easily amenable to batch processing of a fairly large number of samples. A single evaporation step is required, making the whole sample preparation relatively straightforward.

Instrumental sensitivity, background level and limit of detection

In order to reduce as far as possible handling of the highly radioactive ²²⁶Ra, ²³⁸U was used as an analogue during the early stage of method set-up. Accordingly, the background intensity for m/z 226 and the ²³⁸U⁺ ion signal obtained for a 1 ppb U solution were measured as a function of the pressure in the expansion chamber (S-Mode, Table 8). A ca. five fold increase in sensitivity occurred when the pressure in the expansion chamber was reduced to 1 mbar or less. A concomitant three fold improvement of background noise at m/z 226 was observed. This behaviour compares very well with that reported by Chiappini et al.,¹⁶ who observed both a gain in sensitivity and a three fold decrease in background level for plutonium, compared with the same instrument operated in the conventional mode.

Table 8 Instrumental response (m/z = 238) and background intensity and standard deviation measured at m/z = 226 while nebulizing a 0.5 M HNO₃ solution containing 1 ppb U as a function of decreasing pressure in the expansion chamber (S-mode)

Pressure/mbar	Intensity m/z = 238/10^5 counts s^−1	Background m/z = 226/counts s^−1	s/counts s^−1
2.0	1.30	14.3	4.1
1.0	5.11	3.7	2.4
0.95	5.37	3.2	1.8
0.90	6.36	3.3	1.6
0.85	5.99	3.0	1.0
0.80	6.02	3.7	1.6

---

For each expansion chamber pressure, the approximate instrumental limit of detection (LOD) was calculated as the concentration equivalent to three times the standard deviation of 10 replicate measurements of the m/z 226 signal. The results depicted in Fig. 1 show that an optimum (LOD ≈ 5 fg g⁻¹) was reached for a pressure of 0.85 mbar. A similar optimum pressure was observed previously in studies on this enhanced sensitivity pumping interface.^14,15 The effective detection limit for ²²⁶Ra measured subsequently under the same conditions was found to be only 2 fg g⁻¹. This single figure ppq detection limit obtained for ²²⁶Ra is in excellent agreement with those measured previously for actinide elements, using quadrupole ICP mass spectrometers equipped with the same interface working at low pressure and high efficiency nebulization systems.^14–17

Fig. 1 Variation of the limit of detection for Ra as a function of decreasing pressure in the expansion chamber (S-mode).

The 2 fg g⁻¹ LOD achieved in this work corresponds to a 100-fold improvement for ²²⁶Ra determination compared with the performance allowed by quadrupole-based ICP-MS with conventional solution delivery using a Meinhard type pneumatic concentric nebulizer.⁹ In this study, the radium from 500 ml of water was concentrated into 5 ml for ICP-MS analysis. Taking into account this 100-fold enrichment factor, a method detection limit of 0.02 fg g⁻¹ can be achieved for the analysis of thermal waters having very high salt contents. This value is identical with that obtained by Park et al.¹⁰ for mineral waters (0.01 fg g⁻¹ using an enrichment factor of 200) with a high resolution ICP-MS instrument. Also, it compares well with the ²²⁶Ra LOD in groundwaters (0.19 mBq l⁻¹ = 5 fg l⁻¹ for an enrichment factor of 900) achieved by Kim et al.¹¹ using another model of high-resolution ICP-MS combined with an ultrasonic nebulization (USN). Therefore, it appears that the combination of USN with the high sensitivity interface used in this work has the potential to bring a conventional, quadrupole-based instrument to the same performance level as a double-focusing mass spectrometer operated in the low mass resolution mode, despite the higher background level (ca. 3 versusca. 0.1 counts s⁻¹) of the quadrupole ICP-MS.

Conclusion

An instrumental limit of detection in the very low ppq range (2 fg g⁻¹) was achieved for ²²⁶Ra using a standard quadrupole-based ICP-MS instrument equipped with a high efficiency ion transmission interface. Similar performances have been demonstrated for other high mass elements,^14–17 including several radionuclides.^14,16 Therefore, it appears that this improved quadrupole ICP-MS system has the potential to rival more sophisticated double-focusing instruments, when high mass resolution is not required for ultra-trace analyses. This can be achieved fairly easily with a simple, high ion transmission interface (S-mode) in which a ca. three fold decrease in the expansion chamber pressure strongly enhances the instrument sensitivity, while the background level and fluctuations are significantly reduced. In this way, extremely low detection limits can be obtained without resorting to more expensive pieces of equipment such as magnetic sector mass spectrometers.

Several methods for the preconcentration and separation of radium from saline thermal waters in preparation for ICP-MS measurements were evaluated in this work. The conventional, single-stage cation-exchange chromatographic techniques used in previous work are limited, in the case of waters having high TDS, by the fact that all the Ba and the tail of the Sr and Ca peaks are found in the Ra fraction. These easily ionizable elements induce severe ion signal suppression. Two-step procedures involving a clean-up column based either on cation exchange in perchloric acid medium (for Ca and Sr removal), or on specific extraction chromatographic material (to remove Sr and Ba ), give satisfactory results but require complicated, time-consuming sample handling. Radium-specific solid phase extraction disks, originally designed for counting methods largely used in radioanalytical chemistry, provide a straightforward alternative. Using these disks and a simple filtration device, a radium fraction suitable for ICP-MS measurement can be extracted in a single pass from large volumes of water, following a very simple separation scheme which requires little analyst time. As yet, a stripping agent for Ra which would be compatible with ICP-MS analysis has not been found. Therefore, the silica support of the Ra-selective extracting agent must be dissolved in order to achieve a quantitative recovery, thereby making the method relatively expensive (about US$ 30 per disk). Taking into account the unusually high salt load of the waters studied (6 g l⁻¹), it is expected that this method could be used for preconcentrating Ra from significantly larger volumes of more common mineral or tap water. Conversely, based on the good tolerance of the Radium Rad disks to high concentrations of rock-forming cations, it is suggested that they might be used for separating Ra from nitric acid solutions of even more complex matrices such as geological or environmental solid samples.

acknowledgement

We gratefully acknowledge 3M Filtrete, Breda, The Netherlands, for kindly donating some samples of Radium Rad disks, S. Rihs for discussions on Radium Rad disks and M. Condomines for discussions on measurements of ²²⁶Ra in thermal waters and providing the ²²⁶Ra stock standard solution. We also thank B. Le Fèvre for critical reading of the manuscript. Région Auvergne, and CNRS are thanked for financial support to SJ (`bourse de thèse de docteur-ingénieur', contract No. 4543).

References

1. Uranium-series Disequilibrium: Applications to Earth, Marine and Environmental Sciences, ed. M. Ivanovich and R. S. Harmon, Clarendon Press, Oxford, 2nd edn., 1992..
2. C. W. Sill, Nucl. Chem. Waste Manage., 1987, 7, 239.
3. M. Koide and K. W. Bruland, Anal. Chim. Acta, 1975, 75, 1.
4. G. G. Mathieu, P. E. Biscaye, R. A. Lupton and D. Hammond, Health Phys., 1988, 55, 989.
5. M. Condomines, R. Bouchez, J. L. Ma, J. C. Tanguy, J. Amosse and M. Piboule, Nature (London), 1987, 325, 607.
6. A. Volpe, J. A Olivares and M. T. Murrell, Anal. Chem., 1991, 63, 913.
7. A. S. Cohen and R. K. O'Nions, Anal. Chem., 1991, 63, 2705.
8. F. Chabeaux, D. Ben Othman and J. L. Birck, Chem. Geol., 1994, 119, 191.
9. V. F. Hodge and G. A. Laing, Radiochim. Acta, 1994, 64, 211.
10. C. J. Park, P. J. Oh, H. Y. Kim and D. S. Lee, J. Anal. At. Spectrom., 1999, 14, 223.
11. Y. J. Kim, C. K. Kim, C. S. Kim, J. Y. Yun and B. H. Rho, J. Radioanal. Nucl. Chem., 1999, 240, 613.
12. M. R. Smith, E. J. Wyse and D. W. Koppenaal, J. Radioanal. Nucl. Chem., 1992, 160, 341.
13. R. Chiarizia, M. L. Dietz, E. P. Horwitz, W. C. Burnett and P. H. Cable, Sep. Sci. Technol., 1999, 34, 931.
14. M. Liezers, C. T. Tye, D. Mennie and D. Koller, in Applications of Inductively Coupled Plasma-Mass Spectrometry to Radionuclide Determinations, ASTM STP 1291, ed. R. W. Morrow and J. S. Crain, American Society for Testing and Materials, West Conshohocken, 1995, pp. 61−72..
15. D. Günther, H. P. Longerich and S. E. Jackson, Can. J. Appl. Spectrosc., 1995, 40, 111.
16. R. Chiappini, J. M. Taillade and S. Brebion, J. Anal. At. Spectrom., 1996, 11, 497.
17. I. B. Brenner, M. Liezers, J. Godfret, S. Nelms and J. Cantle, Spectrochim. Acta, Part B, 1998, 53, 1087.
18. R. J. Thomas and C. Anderau, At. Spectrosc., 1989, 10, 71.
19. S. Rihs, PhD Thesis, University Blaise Pascal, Clermont-Ferrand, 1999..
20. A. Aubignat, L'Hydraulique Souterraine, La Houille Blanche, Grenoble, France, 1960, vol. 2, p. 486..
21. R. Thuizat, PhD Thesis, University Blaise Pascal, Clermont-Ferrand, 1973..
22. F. W. E. Strelow, Anal. Chem., 1960, 32, 1186.
23. F. W. E. Strelow, R. Rethmeyer and C. J. C. Bothma, Anal. Chem., 1965, 37, 107.
24. F. T. Nelson, T. Murase and K. A. Kraus, J. Chromatogr., 1964, 13, 503.
25. E. P. Horwitz, R. Chiarizia and L. Dietz, Solvent Extract. Ion Exch., 1992, 10, 313.
26. 3M Empore^TM Radium Rad Disks. Product Information Sheet, 3M Filtration Products, St. Paul, MN, 1997..
27. D. F. Hagen, C. G. Markell and G. A. Schmitt, Anal. Chim. Acta, 1990, 236, 157.
28. L. L. Smith, J. S. Alvarado, F. J. Markun, K. M. Hoffmann, D. C. Seely and R. T. Shannon, Radioact. Radiochem., 1997, 8, 30.
29. A. Durecova, J. Radioanal. Nucl. Chem., 1997, 223, 225.
30. D. C. Seely and J. A. Osterheim, J. Radioanal. Nucl. Chem., 1998, 236, 175.
31. S. Rihs and M. Condomines, Chem. Geol., submitted for publication..
32. 3M Empore^TM Radium Rad Disks. 3M Test Method RA-395, 3M New Products Department, St. Paul, MN, 1995..
33. C. Pin, D. Briot, C. Bassin and F. Poitrasson, Anal. Chim. Acta, 1994, 298, 209.
34. S. Rihs, M. Condomines and O. Sigmarsson, Geochim. Cosmochim. Acta, 2000, 64, 661.

---