Determination of Radium-226 in Aqueous Solutions by α-Spectrometry

Anal. Chem. 2001, 73, 4218-4224

Determination of Radium-226 in Aqueous Solutions by r-Spectrometry Katell Morvan,*,† Yves Andres,‡,§ Bandombele Mokili,†,§ and Jean-Charles Abbe|,⊥

Laboratoire SUBATECH, UMR 6457, Ecole des Mines de Nantes, IN2P3/CNRS, Universite´ de Nantes, Nantes Cedex 3, France, Ecole des Mines de Nantes, GEPEA, 4, rue Alfred Kastler, La Chantrerie, BP 20722-44307, Nantes Cedex 3, France, and IRCCyN, 1 rue de la Noe¨, BP 92101, 44321 Nantes Cedex 3, France

The new European legislation imposes a lower threshold for radioactivity in drinking water. This requires the development of more sensitive and reliable analytical methods. This work presents an improved r-spectrometric technique to determine the radium-226 activity in aqueous solution relying on the radium adsorption onto a thin manganese oxide layer followed by r-measurement. The preparation of the MnO2 deposit has been optimized as well as the radium adsorption conditions. Detection threshold and limit of 5 and 10 mBq‚L-1, respectively, with a 10% (95% confidence) uncertainty are currently reached. This paper reports on the overall technique and on its application to assess the radium-226 activity in 28 French mineral waters. In addition, the gross r- and β-activities have been evaluated using proportional counting while the uranium concentrations were derived from ICPMS. During the past decades, governments all over the world, aware of the importance of the quality of drinking water, have introduced new regulations in line with the recommendations of the World Health Organization (WHO). The latter has published microbiological, chemical, and also radiological guideline values in drinking water1,2 which have been regularly updated in accordance with epidemiological studies. In 1996, WHO recommended that the gross R- and β-activities should not exceed 0.1 and 1 Bq‚L-1, respectively. Consequently, the activity of each radionuclide in water has to be determined to evaluate the committed effective dose for which the reference is established at 0.1 mSv‚y-1. Among the naturally occurring radionuclides, the R-emitters are usually considered as the most radiotoxic. Among them, radium-226 has been particularly studied over the recent years since its metabolic behavior in the human body is similar to that of calcium.3 Indeed, many measurements of radium-226 radioactivity have been published, giving a good * Corresponding author: (e-mail) [email protected]; (fax) (+33) 251858452. † Ecole des Mines de Nantes, IN2P3/CNRS. ‡ Ecole des Mines de Nantes, GEPEA. § E-mail: [email protected]; [email protected]. | IRCCyN. ⊥ E-mail: [email protected]. Fax: (+33) 251810577. (1) Guidelines for drinking-water quality, 2nd ed.; World Health Organization: Geneva, 1996; Vol. 2. (2) Guidelines for drinking-water quality, 2nd ed.; World Health Organization: Geneva, 1998; Addendum to Vol. 2.

4218 Analytical Chemistry, Vol. 73, No. 17, September 1, 2001

overview of its distribution in drinking waters around the world; the specific activity is usually low, but some values over 1 Bq‚L-1 have been reported.4-11 Although the latter are scarce, only a few countries have so far decided to take regulatory measures. Whereas the new EU directive12 for drinking water sets a dose limit of 0.1 mSv‚y-1 from radioactive elements, excluding 40K, radon (mainly 220Rn and 222Rn), and its daughter products, American and Australian regulations are somewhat different.13,14 All regulations generally aim at keeping radioactivity at relatively low levels. This is particularly true for radium-226 for which the analytical methods require sensitivity and reliability. Different analytical methods such as ICPMS, the emanation method, γ-ray spectrometry, liquid scintillation techniques, and R-spectrometry are currently applied. Although ICPMS is a promising technique, mass interferences require a pretreatment of the solution in order to isolate the radium isotopes and it is still rarely applied. All the other techniques are very effective and commonly used. The emanation method9,15 consists of measuring the γ-activity of the radium-226 daughter, radon-222, when the radionuclides are in secular equilibrium, which at least is reached after 20 days. This technique offers good sensitivity and is characterized by a detection limit of 30 mBq‚L-1 with a 250-mL sample. Radium-226 can be easily determined by γ-spectrometry16,17 and the detection limit can be decreased through the evaporation (3) Stather, J. W. The environmental behaviour of radium: The behaviour, effects and radiation dosimetry of radium in man; IAEA: Technical Reports Series 310; 1990; Vol. 2, Chapter 3.1. (4) Sanchez, M.; Montero, R. Appl. Radiat. Isot. 1999, 50, 1049-1055. (5) Marovic, G.; Sencar, J.; Franic, Z.; Lokobauer, N. Environ. Monit. Assess. 1997, 46, 233-239. (6) Bettencourt, A. O.; Teixeira, M. M. G. R.; Faisca, M. C.; Vieira, I. A.; Ferrador, G. C. Radiat. Prot. Dosim. 1988, 24, 139-142. (7) Salonen, L. Radiat. Prot. Dosim. 1988, 24, 163-166. (8) Gans, I. Sci. Total Environ. 1985, 45, 93-99. (9) Re´my, M. L.; Lemaitre, N. Hydroge´ ologie 1990, 4, 267-278. (10) Pires do Rio, M. A.; Godoy, J. M.; Amaral, E. C. S Radiat. Prot. Dosim. 1988, 24, 159-161. (11) Zelenski, A. V.; Buzinny, M. G.; Los, I. P. Liquid Scintillation Spectrometry 1992; Noakes; Scho ¨nhofer; Polach; Radiocarbon, 1993, 405-411. (12) European Council Directive 98/83/EC on the quality of household water. Official J. Eur. Communities 1998, L330, 32. (13) Pontius, F. W. J. Am. Water Works Assoc. 1999, 91, 46-58. (14) Australian Drinking Water Guidelines; National Health and Medical Research Council Agriculture and Ressource Management Council of Australia and New Zealand, 1996. (15) Lucas, H. F.; Markun, F.; Boulenger, R. The environmental behaviour of radium: methods for measuring radium isotopes: emanometry; IAEA: Technical Reports Series 310, 1990; Vol. 1, Chapter 3.2. 10.1021/ac0015220 CCC: $20.00

© 2001 American Chemical Society Published on Web 07/27/2001

of several liters of water. However, uranium-235, often present simultaneously in the solution, disintegrates with the emission of a γ-ray at 185.7 keV, interfering with that associated with radium226, 186.2 keV, which severely restricts the applicability of this technique. The R-liquid scintillation technique18 with rejection of β-/γemitters (PERALS), has a high efficiency, but the aliquot to be measured is small (6 mL) and the energy resolution is very poor. As a result, 226Ra measurements can only be performed after an extraction/concentration procedure using, for instance, an efficient and selective ligand. Recently, new selective extractants scintillating cocktails for different radionuclides have become commercially available, including Ra as Radaex. In comparison, R-spectrometry with solid-state detectors allows the measurement of the energy of the particles and consequently a safer assignment and, in principle, quantification. However, as a preliminary step, the source must be deposited as a very thin film in order to avoid self-absorption and energy resolution distortion. Some authors have used the coprecipitation of Ra by adding a Ba salt, as Ba(Ra)SO4.19 However, with this technique, autoabsorption of the R-particles may appear. The present work relies on a R-spectrometric method, as previously studied by Bland20 and Surbeck,21 who showed that MnO2 thin layer could selectively take off radium from the solution. Our aim was first to optimize the quality of the manganese oxide layer in order to avoid the autoabsorption of the R-particles and consequently to obtain a good energy resolution. We attempted to optimize the adsorption of radium-226 onto these thin layers. The characteristics of this analytical method are given in detail. As a test of the validity of this technique, 28 samples of French mineral waters have been analyzed and the data compared to the previously published ones. EXPERIMENTAL SECTION (A) Preparation of Thin Layers of Manganese Oxide. The capacity of hydrated manganese oxides to adsorb alkaline-earth cations, and particularly radium, was first reported by Moore.22 This property offers an alternative to other extraction techniques. The manganese oxide thin layers are prepared by chemical bath deposition from the reduction of MnO4- to MnO2. Four 1-mmthick 6,6-polyamide disks from Faber (Sedan, France) (22-mm diameter), degreased with ethanol, washed in deionized water, and then dried, are introduced into a Bakelite top followed by a rubber gasket. Such set was screwed on the reactor vessel (cf. Figure 1); the deposit is thus well limited on a single face of the disk. After filling of the reaction vessel with deionized water (350 mL) and stabilization of the temperature, potassium permanganate (5.53 g) is added to the solution up to a concentration of 0.1 M. The reaction is carried out under magnetic stirring, until the expected thickness of the manganese oxide layer is reached. The (16) Von Gunten, H. R.; Surbeck, H.; Ro¨ssler, E. Environ. Sci. Technol. 1996, 30, 1268-1274. (17) Kobal, I.; Williams, A. R. IAEA 1984, TECDOC-301, 23-44. (18) Aupiais, J.; Fayolle, C.; Gilbert, P.; Dacheux, N. Anal. Chem. 1998, 70, 23532359. (19) Baeza, A.; del Rio, L. M.; Jimenez, A. Radiochim. Acta 1998, 83, 53-60. (20) Bland, C. J. Int. J. Appl. Radiat. Isotopes 1979, 30, 557-561. (21) Surbeck, H. Sci. Total Environ. 1995, 173/174, 91-99. (22) Moore, W. S.; Reid, D. F. J. Geophy. Res. 1973, 78, 8880-8886.

Figure 1. Reactor for the preparation of the MnO2 disks: 1, Bakelite top; 2, 6,6PA disk; 3, rubber gasket.

disk is first washed with deionized water and then dried and is ready to proceed to the radium adsorption step. The growth rate of the MnO2 layer depends on various parameters such as the bath temperature, the deposition time, the chemical composition of the aqueous solution, and the stirring rate. As mentioned above, the KMnO4 concentration is fixed at 0.1 M.21 A first set of experiments was carried out at 80 °C for 15-420 min in order to derive a range of “MnO2” thicknesses corresponding to an optimal radium adsorption. Once established, another series of radium analyses was set up to determine the best experimental conditions to get a good energy resolution in the R-spectrum. The temperatures of the potassium permanganate bath were fixed between 50 and 80 °C with a constant contact time of 3 h. The deposited surfaces were observed by optical microscopy and thicknesses determined by Talystep at the Ecole Nationale Supe´rieure de Chimie de Paris. The full width at halfmaximum (fwhm) of the R-line was calculated using the Genie2000 software from Canberra. (B) Conditions of the Radium Adsorption. A device similar to that for the deposition, of a smaller volume and with a single hole instead of four, containing the aqueous sample (100 mL) under stirring is used for radium adsorption onto previously prepared manganese oxide layers. At the end of the procedure, the circular carrier is washed with deionized water and dried before the R-spectrometric analysis (planar implanted silicon detector, PIPS, from Canberra). All the kinetics and complexation experiments were carried out with a solution obtained from the leaching of an uranium mining ore. The samples were acidified to pH 1 with nitric acid and bubbled with nitrogen to decarbonate the solution. The pH was then set to the desired value with sodium hydroxide. The runs for kinetics evaluations were stirred between 60 and 2900 min before analyses. Some complexation tests were carried out to verify the inhibition of the thorium and uranium adsorption on the manganese oxide disks. Concerning the thorium adsorption, ethylenediaminetetracetic acid (EDTA) was added to samples at concentrations of 5 × 10-4 M for different pH values. As for the uranium adsorption, a 1 M solution of sodium carbonate was used to adjust the pH in addition to NaOH. To determine the radium adsorption yield, barium-133, having a behavior similar to Ra, was tested as a tracer preferably to spiked Analytical Chemistry, Vol. 73, No. 17, September 1, 2001

4219

226Ra

solutions. To that purpose, five 100-mL samples were probed from the solution containing ∼0.2 Bq‚L-1 226Ra. In one of them, 200 µL of 133Ba solution (75 000 Bq‚L-1) supplied by Damri-Lmri (CEA-Saclay) was injected. The analysis of radium-226 was then done with the 133Ba yield, calculated from the γ-line at 356 keV (HPGe detector, Canberra). The other samples were spiked with 0, 50, 100, or 150 mBq of 226Ra (maximal activity of 226Ra handling on the benchtop, 7 × 104 Bq). The radium-226 activities were then determined. (C) Studies of the Analytical Characteristics. The 226Ra activity (A) contained in a 100-mL sample (V) was calculated from eq 1, where N and N0 are the number of counts from the sample

A ) (N - N0)/TRηV

[ ] [ ] [ ]]

[

2

+

u(η) η

2

+

u(V) V

2 1/2

(2)

was estimated at 2%. The analytical characteristics were carried out with MnO2 disks prepared at 65 °C for 3 h. Concerning the repeatability and reproducibility studies, seven samples from the same solution were treated under the same conditions to determine the 226Ra activity using an R-chamber from Canberra. The reproducibility measurements were carried out in R-chambers supplied by Canberra and Eurisys Mesures. A 226Ra solution (442 Bq‚L-1) was diluted in deionized water to get the samples used during the linearity study. Their concentrations ranged from 0.005 to 10 Bq‚L-1. A Si detector from Canberra was used for the R-spectrometric determination. The detection limit was calculated from eq 3, which is derived from

DL )

226Ra

adsorption versus MnO2 layer thickness.

(1)

and from the background at corresponding counting times T and T0, respectively. We assumed T ) T0 in order to get a better comparison between background and signal. The recovery yield (R) was calculated from the 133Ba adsorption while the detector efficiency (η) was evaluated with a 226Ra source, which we considered as an internal standard. It was prepared in the laboratory from a 5 Bq‚L-1 solution of 226Ra. The mother solution of radium was provided by Amersham. The relative uncertainty, uA/A, of the 226Ra measurement was assessed by eq 2 with 95% confidence. The error on the volume

uA N + N0 u(R) )2 + 2 A R (N - N0)

Figure 2.

41 + x1 + (2N0) TVηR

(3)

the detection threshold given by Neuilly.23 (D) Mineral Water Analyses. The analyzed mineral waters, commercially available in French supermarkets, were bought in late 1999. The gross R- and β-activities were evaluated by proportional counting (IN20 Eurisys Mesures) on samples prepared by evaporating on a steel plate (55-mm diameter) a volume leading to a deposited mass of 20 mg, low enough to get rid of R-autoabsorption. The uranium concentration was determined by ICPMS (Pierre Sue laboratory/CEA-Saclay). The elements con(23) Neuilly, M. Statistique applique´ e a` l’exploitation des mesures; Masson: Paris, 1986; pp 113-117.

4220 Analytical Chemistry, Vol. 73, No. 17, September 1, 2001

centrations were picked from the bottle labels or measured by ICPOES. The radium-226 activity was calculated from the method described here. (E) Safety Considerations. Potassium permanganate is a powerful oxidizing agent. It should be handled accordingly, avoiding contact with skin. By their nature, all radioactive materials can be hazardous if they are not handled correctly. Precautions should be taken during the physical handling, by using disposable latex gloves. RESULTS AND DISCUSSION The recovery of the radium-226 adsorption is followed through a tracer, barium-133, that is γ-emitter. This choice will be discussed later. (A) Optimized Preparation of Thin Layers of Manganese Oxide. Before establishing the usual characteristics of an analytical method, it appeared interesting to optimize the preparation of the thin layers of manganese oxide, on which the radium-226 is adsorbed, on the basis of two criteria: (i) R-particle detection efficiency and (ii) energy resolution measured as the fwhm of the 226Ra peak. The variations of the 226Ra R-particle counts recorded as a function of the thickness of the MnO2 layer show three different regions (see Figure 2). Up to ∼13 µm, the radium adsorption grows with the thickness of the layers in relation to the number of adsorbing sites. For thicknesses between 14 and 20 µm, a plateau is reached, followed by a decrease for thicker deposits revealing a loss in the counting, attributed to the R-particles’ absorption in the MnO2 film. Indeed, deep cracks are observed on the MnO2 surface at these thicknesses (cf. Figure 3). For thicknesses between 14 and 23 µm, further investigations have been carried out to determine the best conditions with reference to the fwhm. Table 1 shows that fwhm decreases with MnO2 thickness. Although the best resolution was obtained when the film was prepared at 50 °C (see Figure 4), we have excluded this procedure as the recovery yield of the 133Ba tracer was incomplete. We have not selected the supports prepared at 70 and 80 °C because of the resulting poor energy resolution. At that stage, the layers prepared either at 60 or 65 °C appeared as the best compromises. We have opted for the latter which entailed fewer drawbacks, essentially a better reproducibility of the recovery yield. Consequently, all the following results were obtained with MnO2 films prepared at 65 °C for 3 h. (B) Parameters of the Radium-226 Adsorption. According to Surbeck,24 the adsorption of radium-226 onto manganese oxide

Figure 5. Radium adsorption kinetics at pH 7.6 and pH 8.5.

Figure 3. Pictures of MnO2 deposit on polyamide-6,6 with a 20× enlargement: (a) thickness of 15 µm and (b) thickness of 23 µm. Table 1. Data from the Preparation of the MnO2 Layer at Different Temperatures prepn temp (°C)

MnO2 layer thickness (µm)

fwhm (keV)

50 60 65 70 80

14 15 16 18 23

61.5 78.3 87.0 109.7 130.2

133Ba

rec (%)

88 100 100 100 100

226Ra counts in R-spectrum

2576 ( 52 2569 ( 51 2657 ( 50 2586 ( 52 2675 ( 54

Figure 4. R-Spectrum obtained from a MnO2 disk prepared at 50 °C. The emission percentages are 95 and 5% for 4784 and 4602 keV, respectively.

fluctuates with the pH. So, with the aim of shortening the adsorption time, the adsorption kinetics was established at two pH values: 7.6 and 8.5. The data illustrated on Figure 5 show that the system required at least 35 h to reach equilibrium at pH 7.6 and ∼10 h at pH 8.5: in this range of pH, the higher the pH, the faster the adsorption kinetics. These kinetics studies confirmed that the adsorption is pH dependent as expected considering the amphoteric behavior of manganese oxide, though the actual nature of the deposit has to be confirmed. (24) Surbeck, H. Appl. Radiat. Isot. 2000, 53, 97-100.

Figure 6. Adsorption of radium and thorium with EDTA (5 × 10-4 M).

Other R-emitters, with energies close to those of 226Ra (4784 keV, 94%), such as 230Th (4688 keV, 73%) and 234U (4775 keV, 71%), can lead to interferences. On one hand, the uranium complexation with carbonates, at basic pH, leads, in solution, to a stable anionic species, (UO2)(CO3)22-, unlikely to be absorbed on the manganese oxide, an amphoteric compound leading to a negatively charged surface at such pH. So, the addition of carbonate to the sample implies a decrease of the uranium adsorption.21 However, it should be noticed that a precipitate, (CaCO3, KS ) 10-8.4), might appear if the carbonate concentration is too high. On the other hand, thorium may be removed by adding strong complexing reagents, such as EDTA, at a concentration of ∼5 × 10-4 M. In such conditions, the thorium adsorption is fully inhibited whereas that of the radium is only slightly affected (Figure 6). Concerning the competing effects of chemical species on the 226Ra adsorption, Bland20 showed in 1979 that sodium did not have any effect, contrary to potassium and iron. As to the alkaline-earth species, the recovery of radium decreases by 25% when calcium and magnesium concentrations are, respectively, equal to 250 and 200 ppm. As far as we know, no data exist about a possible interference of barium and strontium. In conclusion, the manganese oxide film seems to be highly selective toward radium elements. However, Table 1 shows that while the radium counts do not vary significantly whatever the layer, this is no longer true for the barium recovery when the deposit was prepared at 50 °C. This might be due to a slight difference of the adsorption behavior between radium and barium; films prepared at 65 °C imply a complete recovery of both elements. To determine the radium adsorption yield on manganese oxide, the best alternative would be to use another radium isotope. Analytical Chemistry, Vol. 73, No. 17, September 1, 2001

4221

Figure 7. Linearity from 0 to 10 Bq‚L-1.

Unfortunately, 224Ra (R-emitter, T1/2 ) 3.6 days) has a too short half-life and 228Ra not only disintegrates through β-emission but may naturally be present in the solution. As previously mentioned, barium was used as radium tracer. Since the discovery of radium by Marie Curie,25 both elements being isoelectronic are considered as having similar chemical properties and behavior. So, barium-133, a γ-emitter, easy to measure, appeared as a good potential radium tracer. This assumption has been verified by comparing the results obtained using spiked 226Ra and 133Ba solutions. For a near 200 mBq‚L-1 226Ra solution, the 226Ra activity has been determined as 224 ( 33 and 190 ( 32 mBq‚L-1, respectively, through spiked Ra measurements or using the recovery yield as determined through 133Ba. The calculated deviation En normalized with respect to the stated uncertainty (see eq 4), has an absolute value less than 1, meaning that both results are statistically acceptable.

En ) (x1 - x2)/xU12 + U22

(4)

where xi is the measurement result and Ui is the uncertainty of xi (C) Analytical Characteristics of the Described Method. Repeatability and Reproducibility. The tests of repeatability included seven measurements in series by the same person, in the very same conditions. The mean, calculated from these data, is 0.74 Bq‚L-1 with a standard deviation (95% confidence) of 0.08 Bq‚L-1 indicating an error of ∼10% for the determination of radium-226. The reproducibility tests were carried out in order to ascertain that different measuring conditions, especially using several R-chambers, did not influence the results. The data listed in Table 2 are the means of seven measurements with four detection devices. The results show a good reproducibility. Linearity. The response linearity has been studied between 0 and 10 Bq‚L-1 226Ra. Figure 7 illustrates the curves obtained from several samples. Assuming a linear variation between the expected and the experimentally measured concentration, a good χ2 is effectively calculated, the deviation being quite reasonable for such (25) Curie, P.; Curie, M.; Be´mont, G. C. R. Acad. 1898, 127, 1215-1218.

4222

Analytical Chemistry, Vol. 73, No. 17, September 1, 2001

Table 2. Reproducibility Data no. of chamber

radium-226 activity (Bq‚L-1)

1 2 3 4

0.725 ( 0.068 0.743 ( 0.084 0.739 ( 0.104 0.710 ( 0.068

measurements. Thus, as the linearity is expected in the range of 0-10 Bq‚L-1, eq 3 can be applied to determine a detection threshold (DT) for each analysis. Fixing a detection threshold at 5 mBq‚L-1, a detection limit of 10 mBq‚L-1 can be derived (2σ). (D) Results from French Mineral Waters. To validate our method of 226Ra measurements, we analyzed 28 mineral waters and compared our results with those previously published in 1998 by Aupiais et al.18 and in 1990 by Remy and Lemaitre.9 The uncertainties are obtained from eq 2. In Table 3, wr, means the results are given with reservation because the Ba-133 recovery was under 50%. In addition to the radium-226 data, Table 3 provides the gross R- and gross β-activities as well as the uranium238 concentration and the potassium-40 activity, which is extrapolated from the potassium concentration. When looking at the relationship between the gross and individual R-activities, it can be noticed that our results are selfconsistent. Indeed, the 226Ra and the 238U activities are lower than the gross R. Now, if we focus more particularly on the 226Ra activities, the comparison between our data and those published by Re´my and Lemaitre9 obtained using the emanation technique shows that they are in fair agreement with the exception of the data referring to the Badoit sample. This case will be discussed later. 226Ra activities have also been determined in 10 of the water sources by R-liquid scintillation18 with rejection of β-/γ-emitters (Perals). Again, the data compare fairly well with our data shown in Table 3. R-Spectrometry then appears at least as powerful as the Perals with the advantage of a better energy resolution allowing an easier discrimination if some interfering elements appear on the spectrum and thus a safer assessment.

Table 3. r- and β-Activities of French Mineral Waters mineral water

226Ra (mBq‚L-1)

238U (µg‚L-1)

238U (mBq‚L-1)

gross R-activity (Bq‚L-1)

gross β-activity (Bq‚L-1)

40K (Bq‚L-1)

Que´zac Saint-Yorre Badoit Ste. Marguerite Chateauneuf Ogeu Vernet St. Jean de source Vals Arcens Vichy Ce´lestins Parot Saint-Alban Ce´sar Vernie`re Font Picant Arvie Plancoe¨t Chambon Volvic Thonon Celtic Wattwiller Aix-les-Bains Luchon St. Amand Sail-les-Bains Valvert La Franc¸ aise

699 ( 70 312 ( 37 16 ( 4 16 ( 4 706 ( 83 65 ( 10 67 ( 9 62 ( 9 114 ( 14 10 ( 3 457 ( 51 512 ( 56 1931 ( 201 1891 ( 200 829 ( 96 9(4 502 ( 54 (wr) 68 ( 11 102 ( 14 (wr)