Method to Determine 226Ra in Small Sediment Samples by Ultralow

Jul 16, 2010 - To whom correspondence should be addressed. Phone: +34 93 581 1915. E-mail: [email protected]., †. Autonomous University of ...
0 downloads 0 Views 409KB Size
Anal. Chem. 2010, 82, 6847–6853

Method to Determine 226Ra in Small Sediment Samples by Ultralow Background Liquid Scintillation Joan-Albert Sanchez-Cabeza,*,†,‡ Laval Liong Wee Kwong,§ and Maria Betti§ Institute of Environmental Science and Technology, and Physics Department, Autonomous University of Barcelona, ES-08193 Bellaterra, Spain, Departamento de Medio Ambiente, CIEMAT, 28040 Madrid, Spain, and Environment Laboratories, International Atomic Energy Agency, MC-98000 Monaco 210

Pb dating of sediment cores is a widely used tool to reconstruct ecosystem evolution and historical pollution during the last century. Although 226Ra can be determined by γ spectrometry, this method shows severe limitations which are, among others, sample size requirements and counting times. In this work, we propose a new strategy based on the analysis of 210 Pb through 210Po in equilibrium by r spectrometry, followed by the determination of 226Ra (base or supported 210Pb) without any further chemical purification by liquid scintillation and with a higher sample throughput. Although γ spectrometry might still be required to determine 137Cs as an independent tracer, the effort can then be focused only on those sections dated around 1963, when maximum activities are expected. In this work, we optimized the counting conditions, calibrated the system for changing quenching, and described the new method to determine 226Ra in small sediment samples, after 210Po determination, allowing a more precise determination of excess 210Pb (210Pbex). The method was validated with reference materials IAEA-384, IAEA-385, and IAEA-313. The natural radionuclide 210Pb is used as an environmental tracer of numerous biogeochemical processes in the aquatic, soil, and atmospheric sciences. Among these, possibly the most common application is its use in the reconstruction of recent environmental changes through 210Pb dating, as described in some seminal papers.1-5 210Pb (half-life ) 22.23 ± 0.12 years)6 * To whom correspondence should be addressed. Phone: +34 93 581 1915. E-mail: [email protected]. † Autonomous University of Barcelona. ‡ CIEMAT. § International Atomic Energy Agency. (1) Goldberg, E. D. In Radioactive Dating; International Atomic Energy Agency: Vienna, Austria, 1963; pp 121-131. (2) Crozaz, G.; Picciotto, E.; de Breuck, W. J. Geophys. Res. 1964, 69, 2597– 2604. (3) Krishnaswamy, S.; Lal, D.; Martin, J.; Meybeck, M. Earth Planet. Sci. Lett. 1971, 11, 407–414. (4) Robbins, J. A. In Biochemistry of Lead; Nriagu, J. O., Ed.; Elsevier: Amsterdam, The Netherlands, 1998; pp 285-393. (5) Appleby, P. G.; Oldfield, F. Catena 1978, 5, 1–8. (6) All half-lives used in this work are from Decay Data Evaluation Project, http://www.nucleide.org/DDEP.htm, updated by the “Laboratoire National Henri Becquerel” on January 22, 2010. 10.1021/ac1008332  2010 American Chemical Society Published on Web 07/16/2010

is a natural radionuclide of the 238U radioactive chain. In closed systems, 226Ra (T1/2 ) 1600 ± 7 years) decays, through various daughter radionuclides, to 210Pb, named base or supported 210 Pb, which should be in equilibrium with 226Ra (base 210Pb ) 226 Ra). On the other hand, some 210Pb can reach bottom aquatic sediments from either the atmosphere (after 222Rn exhalation from soils) or the water column (in situ production). This is called the excess or unsupported fraction (210Pbex) and is the basis of all 210Pb dating models.7 Therefore, bottom sediments contain a mixture of base and excess 210Pb, and 210Pbex is determined by the difference between the total 210Pb and 226Ra concentration for each sediment section (210Pbex ) 210Pb 226 Ra). This is commonly determined in sections (typically 1 cm width) of undisturbed sediment cores collected from areas of interest. High-resolution γ spectrometry with HPGe (high-purity Ge) detectors is commonly used to simultaneously determine both 210 Pb and 226Ra in sediment samples.8 Some of the reasons for its success are (i) 137Cs is also determined, which may be used to validate the 210Pb chronology,9 and (ii) it is nondestructive, as it does not require chemical separation. However, it has important drawbacks for 210Pb dating: (i) calibration in the low γ emission energy region is difficult and requires tedious selfabsorption corrections, thus affecting method accuracy, (ii) precision is limited by the unavoidable presence of a relevant background due to Compton scattered electrons in the detector, (iii) counting times are long, thus sample throughput is small, and (iv) most importantly for 210Pb dating applications, sample size requirement is usually large, typically more than 5 g dry weight (dw), although some laboratories with HPGe well-type detectors and/or special low-level counting conditions might use a sample size as low as 1-2 g dw. Sample size is in many cases a limitation for 210Pb dating as commonly a large number of other analyses need to be carried out. Although all samples could be measured by γ spectrometry and saved for further analysis, special care is needed during sample manipulation (e.g., if the sample cannot be ground) and in order to avoid (7) Appleby, P. G. Chronostratigraphic techniques in recent sediments. In Tracking Environmental Change Using Lake Sediments: Basin Analysis, Coring, and Chronological Techniques, Last, W. M., Smol, J. P., Eds.; Kluwer Academic, 2001; Vol. 1, pp 171-203. (8) Schelske, C. L.; Peplow, A.; Brenner, M.; Spencer, C. N. J. Paleolimnol. 1994, 10, 115–128. (9) Smith, J. N. J. Environ. Radioact. 2001, 55, 121–123.

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6847

contamination of the substances of interest (e.g., trace metals). When this strategy is used, the time needed to complete γ spectrometry before further analyses can be performed also needs to be taken into consideration. In many laboratories, the total 210Pb concentration is determined by R spectrometry, through its daughter radionuclide 210 Po in equilibrium. The technique is simple, rapid, reliable (recoveries >95%), and shows good precision.10,11 Furthermore, although the technique is destructive, only ca. 250 mg of dry sediment is needed, and because laboratories usually have several R detectors (they are much cheaper than Ge detectors), sample throughput is much larger than with γ spectrometry. Typical counting times range from 1 to 7 days, depending on sample activity and desired uncertainty. An experimented analyst might produce a full total 210Pb profile within a month after sample receipt (ca. 40 samples). When R spectrometry is used, the base 210Pb can only be indirectly determined by assuming that the 226Ra concentration is constant along the profile and therefore estimated as the average of concentrations in the profile bottom sections, as equilibrium should have been reached.12 However, as 226Ra may vary along the profile, this may lead to inaccurate 210Pbex values. It is not uncommon that, once the total 210Pb profile is known, γ spectrometry is carried out in selected sections, thus optimizing the use of this limiting resource, although because of sample size requirements this might be impracticable for many laboratories. As γ spectrometry is carried out on the untreated sample and 210Po on a digestate, these techniques sometimes yield results which are not fully consistent. Liquid scintillation counting (LSC), mostly used for the determination of β emitters,13,14 has also been extensively used to quantify R emitters, such as 226Ra, in environmental samples (mainly waters).15,16 Some of the reported methods include direct R LSC after water sample concentration,17 226Ra extraction from water samples with specific scintillation cocktails such as Radaex,18 or generation of a radium-barium sulfate coprecipitate that is transformed into a soluble chloride or nitrate.19,20 Villa et al.21 opted for this approach for sediment: the sediment is digested, and after the elimination of actinides as hydroxides, radium is recovered as Ra-Ba-SO4, dissolved in EDTA 0.2 M ammonia solution, and counted. However, most of these methods cannot be directly used with sediment digestates and/ or are excessively resource-consuming. (10) Sanchez-Cabeza, J. A.; Masque´, P.; Ani-Ragolta, I. J. Radioanal. Nucl. Chem. 1998, 227, 19–22. (11) Vesterbacka, P.; Ikaheimonen, T. K. Anal. Chim. Acta 2005, 545, 252– 261. (12) Binford, M. W. J. Paleolimnol. 1990, 3, 253–268. (13) Pujol, L.; Sanchez-Cabeza, J. A. J. Radioanal. Nucl. Chem. 1999, 2, 391– 398. (14) Liong Wee Kwong, L.; LaRosa, J. J.; Lee, S. H.; Povinec, P. P. J. Radioanal. Nucl. Chem. 2000, 248, 751–755. (15) Salonen, L. Sci. Total Environ. 1993, 130-131, 23–35. (16) Salonen, L.; Hukkanen, H. J. Radioanal. Nucl. Chem. 1997, 226, 67–74. (17) Sanchez-Cabeza, J. A.; Pujol, L. Analyst 1998, 123, 399–403. (18) Aupiais, J. Anal. Chim. Acta 2005, 532, 199–207. (19) Repinc, U.; Benedik, L. J. Radioanal. Nucl. Chem. 2002, 254, 181–185. (20) Galan-Lopez, M.; Martin-Sanchez, A.; Tosheva, Z.; Kies, A. In LSC 2005 Advances in Liquid Scintillation Spectrometry; Chalupnik, S., Schoenhofer, F., Noakes, J., Eds.; Radiocarbon: Tucson, AZ, 2006; pp 165-170. (21) Villa, M.; Moreno, H. P.; Manjo´n, G. Radiat. Meas. 2005, 39, 543–550.

6848

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

After 210Po analysis for R spectrometry, 226Ra remains in solution. In this work, we propose, develop, and validate a new method to determine 226Ra by 222Rn emanation to a scintillation cocktail, which eliminates the need to perform any further purification, and counting with an ultralow background liquid scintillation system. EXPERIMENTAL SECTION Equipment. An ultralow background liquid scintillation system, Quantulus 1220TM (Wallac, Turku, Finland), was used to carry out this work. In this system, background is reduced by an optimized combination of active and passive shields. A pulseshape analysis (PSA) circuit permits the discrimination of pulses produced by R and β radiation by comparing the area of the pulse tail after 50 ns from the start with its total area. Pulse-shape discrimination is accomplished using a software adjustable parameter (PSA parameter) which can vary between 1 and 256.16,22 Quenching (sample extinction) was quantified with the standard quenching parameter (SQP(E)) which is used to determine the counting efficiency for each sample through calibration curves.23,24 Counting was performed with Wallac OptiScint HiSafe III, a diisopropyl naphthalene based aqueous immiscible cocktail, and low-diffusion PE counting vials (Packard BioScience). Counting Solutions. The tracer solutions were prepared by gravimetrically spiking 226Ra/2 M HNO3 (NIST, SRM4967, U.S.A.) into known amounts of deionized water contained in 20 mL low-diffusion PE counting vials. OptiScint HiSafe was then added to reach a total admixture volume of 20 mL. These were stored for 3 weeks in a dark temperature-controlled area to allow in-growth and equilibrium of the radioactive progenies. The background solutions, used for calibration purposes, were prepared with 10 mL of deionized water, acidified to match the standard solutions, to which 10 mL of the scintillation cocktail was added. In all cases, quenching was changed by adding different amounts of CCl4, ranging from 0 to 200 µL. All reagents used in the experiments were of analytical grade (Fisher Scientific). RESULTS When counting a 226Ra aqueous solution with an immiscible scintillant (such as OptiScint Hisafe), the R emitter 226Ra decays to the R emitter 222Rn (T1/2 ) 3.8332 ± 0.0008 days). Radon is highly soluble in oil-based scintillators and is selectively extracted in the cocktail, suffering some decay while this process takes place. Once solubilized in the organic phase, 222 Rn decays to the R emitter 218Po (T1/2 ) 3.094 ± 0.006 min), this mainly decays to the β emitter 214Pb (T1/2 ) 26.8 ± 9 min), which decays to the β emitter 214Bi (T1/2 ) 19.9 ± 0.4 min), which mainly decays to the R emitter 214Po (T1/2 ) 162.3 ± 1.2 µs), and this one to 210Pb (T1/2 ) 22.23 ± 0.12 years). Therefore, in the scintillant, and after an appropriate equilibration time (usually set to about 3 weeks), the R emitters 222Rn, 218Pb, and 214 Pb are in secular equilibrium (Figure 1). As the probability of these R decays is in all cases close to one, the maximum (22) Kaihola, L. J. Radioanal. Nucl. Chem. 2000, 243, 313–317. (23) Villa, M.; Manjon, G.; Garcia-Leon, M. Nucl. Instrum. Methods Phys. Res., Sect. A 2003, 496 (2-3), 413–424. (24) Sanchez-Cabeza, J. A.; Pujol, L. Health Phys. 1995, 68 (5), 674–82.

Figure 1. 226Ra spectrum by liquid scintillation (only 6 × 10-4 Bq). Vertical dashed lines indicate the approximate counting windows.

efficiency reached when counting all R events should be 300%. This is another advantage of R versus γ spectrometry, where final counting efficiencies for 226Ra, depending on the counting configuration, normally do not usually exceed a few percent. The magnitudes used for optimization were those related to counting precision. • Counting efficiency E: this is the ratio between the observed number of counts and the 226Ra decays. The maximum observed efficiency is close to 300% (Figure 1) because with this method we simultaneously count R particles from 222Rn, 218 Pb, and 214Pb in equilibrium. • Minimum detectable activity (MDA):25 we used the following expression

Figure 2. Determination of the optimal admixture composition.

Figure 3. Change of background and 226Ra counting efficiency with PSA.

where B is the background count rate, t is the counting time, and V is the volume of solution used (in the case of sediment analysis, V was substituted by m, the mass of the aliquot analyzed). • Figure of merit: this is a magnitude commonly used to compare methods, which emphasizes counting efficiency (and therefore sample throughput). FM was calculated as FM ) E2/B. In this section, we describe the results of the optimization process for the relevant counting parameters. Optimal Admixture Composition. Once the scintillation vial and total volume were fixed, the optimal “scintillator-to-water” ratio was optimized. We prepared and counted several composition mixtures with tracer solutions and background solutions, by using 1, 5, 10, 15, and 19 mL of scintillant (Figure 2). Not unexpectedly, the highest efficiency was observed for the maximum volume of scintillant, as this maximizes radiation interaction with the detector (the scintillant). However, as the scintillant itself is an important source of background, this also increases the MDA value. This behavior is well-captured with the FM, which shows a maximum value for a 10:10 mL admixture. Therefore, we used this proportion in all experiments. Optimal PSA Parameter. The PSA circuit of Quantulus 1220 sends the signals to one of the two multichannel analyzers

depending on the result of the PSA analysis. However, this method is not error-free and shows interference as (i) R and β events can be wrongly assigned and (ii) the efficiency of this method depends on the energy of the incident radiation particles.26 In order to minimize the interference, tracer solutions of 226Ra activity 5.34 Bq and background solutions, with a 10:10 mL admixture composition, were counted by LSC during 12 h and with the PSA parameter ranging from PSA ) 0 (no R-β discrimination) to PSA ) 256 (maximum R-β discrimination) with a step increment of 5 (Figure 3). The background signals assigned to the R spectrum have a varied nature and may include γ interactions with the scintillant, β events wrongly assigned to the R spectrum, R emission from impurities in the vial, and a variety of cosmic-ray originated signals. Therefore, the background spectrum does not show prominent peaks. The background count rate versus PSA plot (Figure 3) shows an almost constant value until PSA ∼ 90, as all events are recorded in the R spectrum, and then its value decays smoothly to a value close to 0 cpm, when all background events are recorded in the β spectrum. On the other hand, the 226Ra spectrum shows well-resolved peaks within a quite narrow energy window (Figure 1), and the effect of the R particle energy on the interference is clearly shown in the efficiency versus PSA curve (Figure 3). Until PSA ∼ 110, the total efficiency exceeds the maximum value of 300% because of the misclassification of background and β events in the R spectrum. In the PSA range of 120-160, most background

(25) Currie, L. A. Anal. Chem. 1968, 40, 586–593.

(26) Pujol, L.; Sanchez-Cabeza, J. A. Analyst 1997, 122, 383–385.

MDA)

2.71 + 4.65√Bt tVE

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6849

Figure 4. Variation of MDA and FM with respect to PSA for standard solutions.

226

Ra

and β events are correctly discriminated and a counting plateau, of almost constant efficiency (ca. 265%), is observed. In order to choose the optimum value within the efficiency plateau, both MDA and FM were plotted (Figure 4). In broad terms, MDA followed the efficiency curve, suggesting that, from a limit of detection perspective and within the plateau, best results would be achieved at higher PSA, as the background slowly decreases with the PSA parameter and stabilizes for the PSA ranging from 145 to 190. It is worth noticing that, as from PSA ) 135 the background rate is very low, the MDA is correspondingly low. Higher PSA values are discarded from the discussion as the efficiency there tends rapidly to zero. On the other hand, the FM shows a clear peak at PSA ) 145, and shows the optimum combination of efficiency and background, taking advantage of a high efficiency (and therefore sample throughput) and providing the lowest possible MDA. Counting Window. Due to the presence of many substances in the digestate, which may be partially soluble in the scintillator, the sample may show different levels of quenching, which also affect the position of the peaks. In order to compensate for the peak shift, we opted to define counting windows with a fixed width but a variable position in the spectrum, manually set to comprise the peaks of interest. The first strategy was to define a wide counting window of 150 channels. This window allowed to include, irrespective of the quenching level (quantified through the SQP(E) parameter), the three R peaks (Figure 1). The main advantage of this strategy is that counting efficiency is large, up to the maximum theoretical level of 300%, and above 200% for the real samples commonly analyzed. However, as the spectrum resolution does not allow us to distinguish other R impurities present in the sample and some β interferences, this method might lead to slight activity overestimation. In order to minimize the effect of background from impurities and interferences, the second strategy was to define a 50 channel wide counting window to only comprise the higher energy R peak (214Po), which stands in an area of much smaller background and interference,15 and is easily identified. However, in this case the maximum theoretical efficiency is only 100%. Calibration. Quenching is the only parameter affecting the counting configuration that is sample-dependent and cannot be easily controlled. Therefore, several 226Ra standard solutions were quenched with CCl4 to values covering the expected SQP(E) 6850

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

Figure 5. Effect of quenching (SQP(E)) on R counting efficiency.

Figure 6. Time stability test for sample quenching (SQP(E)).

value in environmental samples (Figure 5). For both counting windows, efficiency is close to zero for SQP(E) < 575. Therefore, the calibration curves, obtained by regression analysis between the counting efficiency (y) and the SQP(E) parameter (x), were 150 channels: y ) (0.74 ( 0.06)x - (435 ( 51) and r ) 0.95 50 channels: y ) (0.21 ( 0.02)x - (119 ( 19) and r ) 0.91 Usually, LSC techniques involve the measurement of a batch of samples in cycles. As the typical activities expected in this type of work are low (of the order of 10 mBq), long-term counting, of the order of days for a single batch, is required. We successfully checked the stability of the scintillator through the monitoring of SQP(E) for a quenched tracer with an activity of 10 mBq during 4 days (Figure 6). Detection Limits. Each counting configuration (including counting window and quenching) shows different detection limits. In Table 1, we show the counting properties for a standard solution quenched to SQP(E) ) 850, which is an average value of sediment samples analyzed in our laboratory. The combination of the indirect detection of 226Ra by 222Rn emanation, the high resulting efficiency, and the ultralow background of Quantulus 1220 result in an extremely low MDA (0.29 Bq kg-1) for samples as low as 250 mg, a typical amount used when analyzing 210Pb by R spectrometry.10 Although both MDA are very low, and therefore any of the windows can be chosen, the larger efficiency (and therefore precision of measurements for the same counting time) favors the use of the largest window, which is the recommended value in this work. From a

Table 1. Counting Properties and Detection Limits for a

Ra Standard Solutiona

226

window (channels)

efficiency (%)

background (cpm)b

Ld (cpm)

MDA (Bq kg-1)c

710-860 810-860

173 ± 12 54 ± 4

0.0070 ± 0.0020 0.0024 ± 0.0012

0.0076 0.26

0.29 0.59

a Activity ) 0.01 Bq. Quenched to SQP(E) ) 850. b Counting time of the background was 55 h. c MDA was calculated by assuming a typical sediment mass of 250 mg.

spectrometric point of view, a smaller window and higher R energy minimize the possibility of interferences with other emissions and should be favored when expected sample activity or desired analytical throughput are not compromised. DISCUSSION Although this technique is well-suited to easily determine 226Ra in small samples from most environmental matrixes, the system has been designed and optimized for small sediment samples, subsequent to 210Pb determination through 210Po in equilibrium. Proposed Method. We recommend operating in batches of about 12 samples, where both a reference material and a blank are processed and measured at the same time. The solution from which the Po source was deposited (about 15 mL) is recovered and, in order to destroy ascorbic acid and its degradation compounds, heated under reflux with 5 mL of concentrated HNO3 for half an hour or until a clear solution is obtained. This solution is then evaporated to incipient dryness, and small amounts of 0.5 M HCl are added to dissolve the residue. This operation is carried out two more times to ensure the elimination of HNO3, and the total volume is then adjusted to 10 mL in a 20 mL low-diffusion PE counting vial. Then 10 mL of OptiScint HiSafe cocktail is carefully added by avoiding disturbing the interface formed by the two immiscible liquids in the vial (Figure 7). The mixture is kept for 3 weeks in a dark temperaturecontrolled area, in order to wait for radioactive equilibrium and to minimize chemiluminescence, and is counted by liquid scintillation by using the following conditions: • Counting time: all samples of the batch are sequentially counted for 1 h each, during a minimum of 30 cycles. • Counting windows are approximately set to channels 750-900 and 850-900. However, best results are obtained by summing all spectra and visually adjusting the windows manually to their optimum value. • Counting mode is set to discriminate R and other pulses, with PSA ) 145. Counting conditions and calibration are equipment-sensitive, and therefore, the quenching calibration must be performed for each instrument. We also recommend carrying out some tests with 226Ra tracer solutions to confirm that the chosen PSA value is correct for each instrument. The counting information needed to calculate the activity are sample count rate, background count rate, and quenching parameter (such as SQP(E) in Quantulus 1220). From the quenching value the efficiency can be obtained (Figure 5), and the sample activity finally calculated. Low-Activity Test. We tested the linearity and the low-activity response of the detector in the chosen configuration by measuring a set of 226Ra unquenched tracer solutions (0.5 M HCl) with

Figure 7. Proposed procedure to measure sediment samples.

210

Po and

226

Ra in

Table 2. Count Rates When Measuring 226Ra Unquenched Standard Solutions by LSC sample activity (mBq)

net count rate (cpm)

0.017 0.12 0.29 0.41 0.58 0.83 1.7 10

5 g dry weight in well detectors, >20 g in coaxial detectors) might be impossible to obtain when sediments are used for the analysis of multiple magnitudes (such as grain size, elemental composition, trace metals, organic substances, biomarkers, ...) and counting times (typically >2 days per sample) might be a limiting factor for many laboratories. The proposed strategy in this work is the determination of 210Pb through 210Po in equilibrium by R spectrometry10 followed by 226Ra by LSC without any further radiochemical processing. We optimized the counting parameters for an ultralow background scintillation system with R-β separation capabilities (Quantulus 1220, Wallac) and propose the use of a PSA parameter of 145. The system was calibrated with a series of quenched 226Ra standard solutions. For a typical sediment sample quenching (SQP(E) ) 850), the efficiency was (173 ± 12)% in the wide energy counting window (150 channels) due to the simultaneous counting of three radionuclides in equilibrium (222Rn, 218 Po, and 214Po). When analyzing 250 mg dw sediment samples, the MDA was as low as 0.29 Bq kg-1, which is about 2 orders of magnitude lower than typical sediment concentrations, showing the usefulness of the technique for many other environmental applications. The method was validated with three reference materials spanning 3 orders of magnitude of concentration. The proposed method can greatly improve the reliability of 210Pb chronologies of sediment cores, and can also be tested for 226Ra/210Pb dating of carbonates such as corals and speleothems. ACKNOWLEDGMENT The authors thank Mr. Jacobo Martı´n (IAEA) for providing samples of a sediment core from the DYFAMED site and helpful comments. The IAEA is grateful for the support provided to its Marine Environment Laboratories by the Government of the Principality of Monaco.

Received for review March 31, 2010. Accepted June 30, 2010. AC1008332

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6853