Anal. Chem. 2009, 81, 1901–1906
Effective Determination of the Long-lived Nuclide 41 Ca in Nuclear Reactor Bioshield Concretes: Comparison of Liquid Scintillation Counting and Accelerator Mass Spectrometry P. E. Warwick,*,† I. W. Croudace,† and D. J. Hillegonds‡ GAU-Radioanalytical, National Oceanography Centre, Southampton, European Way, Southampton SO14 3ZH, U.K., and Chemical Sciences Division, Center for Accelerator Mass Spectrometry, Lawrence Livermore National Laboratory, 7000 East Avenue, L-231, Livermore, California 94550 The routine application of liquid scintillation counting to 41 Ca determination has been hindered by the absence of traceable calibration standards of known 41Ca activity concentrations. The introduction of the new IRMM 41 Ca mass-spectrometric standards with sufficiently high 41Ca activities for radiometric detection has partly overcome this although accurate measurement of stable Ca concentrations coupled with precise half-life data are still required to correct the certified 41Ca:40Ca ratios to 41Ca activity concentrations. In this study, 41 Ca efficiency versus quench curves have been produced using the IRMM standard, and their accuracy validated by comparison with theoretical calculations of 41Ca efficiencies. Further verification of the technique was achieved through the analysis of 41Ca in a reactor bioshield core that had been previously investigated for other radionuclide variations. Calcium-41 activity concentrations of up to 25 Bq/g were detected. Accelerator mass spectrometry (AMS) measurements of the same suite of samples showed a very good agreement, providing validation of the procedure. Calcium-41 activity concentrations declined exponentially with distance from the core of the nuclear reactor and correlated well with the predicted neutron flux. The expanding program of decommissioning the first and second generation nuclear reactors has resulted in an increasing demand for the quantification of low abundance, long-lived radionuclides in diverse waste forms. A volumetrically significant waste is reactor bioshield concrete that acts as an important structural component and radiological shield. During routine reactor operations these radionuclides are much less significant either radiologically or environmentally than the shorter lived, more abundant activation and fission products. Until recently, the development of novel analytical and measurement procedures for these radionuclides has historically been a low priority. However, the quantification of these long-lived radionuclides is now becom* To whom correspondence should be addressed. noc.soton.ac.uk. † GAU-Radioanalytical, National Oceanography Centre. ‡ Lawrence Livermore National Laboratory.
E-mail:
10.1021/ac802225a CCC: $40.75 2009 American Chemical Society Published on Web 01/29/2009
pew@
ing important in accurate waste characterization and for assessing the long-term impact of nuclear waste repository inventories. One long-lived radionuclide generated in concretes is 41Ca (t1/2 ) 1.04 ± 0.05 × 105 a1 which, along with the shorter lived 45Ca (t1/2 ) 162.7 ± 0.4 d 2) is produced via neutron capture by stable Ca. The reactor bioshield concretes also contain other activation products, including 54Mn, 55Fe, 60Co, 133Ba, (particularly in Barite-enriched concrete), 134Cs, 152Eu, and 154Eu. It is notable that after 100 years of decay 41Ca is the predominant activation product remaining, hence its significance in waste sentencing. Calcium-45 is readily quantifiable via measurement of its 256.1 keV beta emission (100% yield) with beta-counting techniques such as liquid scintillation counting. Measurement of 41Ca is more demanding as the radionuclide decays via electron capture with the emission of weak X-rays (3.3 keV). Low energy X-ray spectrometry has been used to quantify 41Ca in concretes following purification of Ca although reported detection limits are rather poor at 8 Bq/g for a 1 g sample size.3 Standard mass spectrometric techniques such as thermal ionization mass spectrometry (TIMS) and inductively coupled plasma mass spectrometry do not have sufficient mass resolution to separate the very low 41Ca signal from the very large natural 40Ca signal. Resonance ionization mass spectrometry (RIMS) is capable of measuring such challenging ratios and can achieve significantly better limits of detection (∼0.1 Bq/g) compared with X-ray spectrometry.4 Detection limits have been reduced further to ∼0.02 Bq/g through the use of triple-resonance measurements.5 Detection of 41Ca in a magneto-optical trap configuration has also been demonstrated but not widely applied.6 Currently the most sensitive technique for measuring 41Ca is accelerator mass spectrometry (AMS) which is capable of (1) Nishiizumi, K.; Caffee, M. W.; DePaolo, D. J. Nucl. Instrum. Methods Phys. Res., Sect. B 2000, 172, 399–403. (2) Radiochemical Manual, AEA Technology, Harwell, U.K., 1998. (3) Itoh, M.; Watanabe, K.; Hatakeyama, M.; Tachibana, M. Anal. Bioanal. Chem. 2002, 372, 532–536. (4) Muller, P.; Blaum, K.; Bushaw, B. A.; Diel, S.; Geppert, Ch.; Nahler, A.; Nortershauser, W.; Trautmann, N.; Wendt, K. Radiochim. Acta 2000, 88, 487–493. (5) Muller, P.; Bushaw, B. A.; Blaum, K.; Diel, S.; Geppert, Ch.; Nahler, A.; Trautmann, N.; Nortershauser, W.; Wendt, K. Fresenius’ J. Anal. Chem. 2001, 370, 508–512. (6) Moore, I. D.; Bailey, K.; Lu, Z.-T.; Mu ¨ ller, P.; O’Connor, T. P.; Young, L. Nucl. Instrum. Methods Phys. Res., Sect. B 2003, 204, 701–704.
Analytical Chemistry, Vol. 81, No. 5, March 1, 2009
1901
routinely detecting 41Ca:total Ca ratios as low as 10-14 (equivalent to 2 × 10-6 Bq/g for a concrete containing 7 wt % Ca). However, suitably configured AMS facilities are not widely accessible and not necessarily ideal for routine determination of 41Ca to 0.1 Bq/g levels as would be typically required for waste characterization. Liquid scintillation counting offers a potential intermediate, providing sufficient sensitivity for most waste characterization requirements while permitting the simultaneous determination of both 41Ca and 45Ca following radiochemical purification of the Ca.7 Liquid scintillation counting has been routinely used for 45Ca quantification in the biomedical field 8-15 and has also been employed for the standardization of 41Ca sources and for the determination of 41Ca in bioshield concretes.7,16,17 Liquid scintillation counting efficiencies for 41Ca are low at 1-11% but are still practical, and limits of detection of ∼0.3 Bq/g have been reported.7 However, calibration of liquid scintillation counters for 41Ca measurement is problematic as the original sources of standardized 41Ca prepared by the Oak Ridge National Laboratory, U.S.A., and previously used by researchers for calibration are no longer available. The lack of suitable calibration standards was considered a significant barrier to the routine use of liquid scintillation counting for 41Ca measurements.3 More recently sources designed for AMS calibration have been prepared.1 In general, however, these sources do not contain sufficient 41 Ca to permit calibration of liquid scintillation counters although one source (IRMM3701) produced by the Institute for Reference Materials and Measurements (Geel Belgium) contains ∼6 Bq/g 41Ca and has been used for LSC calibration.17 In this study, the routine calibration of liquid scintillation counters is evaluated under various conditions using certified 41Ca standard solutions. Experimentally derived calibrations are compared with theoretical predictions of counting efficiency obtained using the software program, CN2003. The program can be used to predict the counting efficiencies for a wide range of beta, beta-gamma, electron capture (EC), and EC-gamma emitting radionuclides and is based on the principles of the CIEMAT programs EFFY,18 EMI,19 Cega2, and KB.20 An overview of the program is given by Gunther.21 Although the model has been used for the standardization of pure radionuclide solutions, including 41Ca,22 it has not found widespread use in routine radioanalytical determinations. The approach offers an attractive alternative for determining quench curves (7) Sua´rez, J. A.; Rodriquez, M.; Espartero, A. G.; Pina, G. Appl. Radiat. Isot. 2000, 52, 407–413. (8) Lutwak, L. Anal. Chem. 1959, 31, 340–343. (9) Sarnat, M.; Jeffay, H. Anal. Chem. 1962, 34, 643–646. (10) Humphreys, E. R. Int. J. Appl. Radiat. Isot. 1965, 16, 345–348. (11) Hardcastle, J. E.; Hannapel, R. J.; Fuller, W. H. Int. J. Appl. Radiat. Isot. 1967, 18, 193–199. (12) Turpin, R. A.; Bethune, J. E. Anal. Chem. 1967, 39, 362–364. (13) Blanusa, M.; Kastelan, M. Int. J. Appl. Radiat. Isot. 1971, 22, 723–728. (14) Gibbons, R. A.; Sellwood, R. Int. J. Appl. Radiat. Isot. 1968, 19, 129–134. (15) Waller, S. S.; Dodd, J. D. Health Phys. 1977, 32, 185–187. (16) Krasznai, J. P. Waste Management 1993, 13, 131–140. (17) Hou, X. Radiochim. Acta 2005, 93, 1–7. (18) Garcia-Torano, E.; Grau, A. Comput. Phys. Commun. 1985, 36, 307–321. (19) Grau, C. A.; Malonda, A.; Grau, C. P. Comput. Phys. Commun. 1994, 79, 115–123. (20) Los Arcos, J. M.; Ortiz, F. Comput. Phys. Commun. 1997, 103, 83–94. (21) Gunther, E. Appl. Radiat. Isot. 2002, 56, 357–360. (22) Rodriguez Barquero, L.; Los Arcos, J. M. Nucl. Instrum. Methods Phys. Res., Sect. A 1996, 369, 353–358.
1902
Analytical Chemistry, Vol. 81, No. 5, March 1, 2009
of radionuclides that are either not readily available in a certified form or have short half-lives making repeated purchases of the standards uneconomic. The calibration approach was combined with an optimized chemical separation procedure for the efficient purification of Ca from concretes and low level liquid scintillation counting to obtain the optimum limits of detection for this measurement approach. The procedure was applied to the determination of 41Ca activity concentrations in a concrete bioshield core taken during decommissioning from the Steam Generating Heavy Water Reactor (SGHWR), a 100 MWe research reactor that operated on the UKAEA Winfrith site in Dorset, U.K., between 1967 and 1990 (average load factor 57.3%). Data obtained using this approach are compared with values determined for the same sample set by AMS to further validate the measurements, giving the first direct comparison between liquid scintillation and AMS-based measurements in such sample types. EXPERIMENTAL SECTION Equipment and Reagents. Sr-resin was supplied by Eichrom Europe, Paris, France. Gold Star (di isopropyl naphthalene based) cocktail and 22 mL polythene scintillation vials were supplied by Meridian, Epsom, U.K. All other reagents were supplied by Fisher Scientific, Loughborough, U.K. Unless otherwise stated all reagents were of analytical grade or better. Milli-Q water was used throughout. The 41Ca standard solution (IRMM 3701) was supplied by the Institute of Reference Materials and Measurements (IRMM), Geel, Belgium. All other radionuclide solutions were supplied by Amersham QSA, U.K. Standards for AMS measurements were prepared and supplied by K. Nishiizumi.1 Liquid scintillation measurements were performed using a Wallac 1220 Quantulus ultralow level LSC. Gamma spectrometric analysis was performed using a Canberra 30% HPGe well-detector calibrated for the energy range 40-2000 keV against a matrixmatched multi-radionuclide gamma standard which included 210Pb. Radionuclide quantification was performed using Fitzpeaks spectral deconvolution software. Where appropriate, nuclide activities were corrected for cascade summing effects. Elemental composition of the concrete was determined by X-ray fluorescence analysis of fusion beads using a Philips MagiPro XRF. Accelerator Mass Spectrometry measurements were performed using the tandem-AMS facility at the Lawrence Livermore National Laboratories, U.S.A. Separation and Purification of Radiocalcium. Samples were fused with lithium borate to ensure complete dissolution of Ca using a technique originally developed by Croudace and coworkers23 for the dissolution of Pu and U in soils and sediments. Total sample dissolution is desirable to ensure that all activated Ca is recovered from the bulk sample matrix. The fusion mixture was dissolved in nitric acid. Fusion with NaOH + Na2CO3,17 Na2CO3 + K2CO3,4,5 or acid digestion with HCl/HF/HClO43 have also previously been used to achieve complete sample dissolution. Subsequent isolation of Ca was achieved using the separation scheme shown in Figure 1. The purified CaCO3 was dissolved in 4 mL of 3 M HCl in a 22 mL polythene scintillation (23) Croudace, I. W.; Warwick, P. E.; Taylor, R. N.; Dee, S. J. Anal. Chim. Acta 1998, 371, 217–225.
Figure 1. Flow diagram of analytical separation scheme for the purification of
vial, and the resulting solution was mixed with 16 mL Gold Star scintillation cocktails prior to dark adaptation and LSC measurement. Liquid Scintillation Analysis. Calcium-41 quench curves were determined experimentally using the 41Ca mass spectrometry standard (IRMM 3701) supplied by IRMM. Conversion of the certified 41Ca:40Ca atomic ratio to an activity concentration requires knowledge of the Ca isotopic composition of the standard, the total Ca concentration, the half life and the atomic mass of 41Ca. The Ca concentration of the standard is not certified although a nominal concentration of 2000 ppm is
41
Ca.
reported. ICPAES measurement confirmed Ca concentrations in the original standard of 2050 ± 70 ppm. Ca concentrations were also measured in subsamples of the standards and used in all subsequent calculations of 41Ca activity concentrations. IRMM-3701 was originally prepared via dilution of IRMM3703 with excess natural Ca, and the Ca isotopic ratio is effectively (24) Hennessy, C.; Berglund, M.; Ostermann, M.; Walczyk, T.; Synal, H.-A.; Geppert, C.; Wendt, K.; Taylor, P. D. P. Nucl. Instrum. Methods Phys. Res., Sect. B 2005, 229, 281–292. (25) De Laeter, J. R.; Bo ¨hlke, J. K.; De Bievre, P.; Hidaka, H.; Peiser, H. S.; Rosman, K. J. R.; Taylor, P. D. B. Pure Appl. Chem. 2003, 75, 683–800.
Analytical Chemistry, Vol. 81, No. 5, March 1, 2009
1903
natural.24 Isotopic abundances for natural Ca were taken from ref 25, and the half-life of 41Ca was taken as 1.04 × 105 years.1 On the basis of the above data and taking [Ca] ) 2050 ppm, the activity concentration of IRMM 3701 was calculated as 6.4 ± 0.2 Bq/mL. A known mass of standard was evaporated to dryness, and the residue redissolved in 3 M HCl to produce a higher activity concentration (35 Bq/g), matrix matched standard for subsequent LSC calibrations. The stable Ca concentration of this solution was measured to calculate the activity concentration. Calibration of the liquid scintillation counter for 41Ca was achieved by mixing known quantities of the radionuclide solution (typically 100 µL, containing ∼3.5 Bq of 41Ca) with 3 M HCl. Sets of standards containing 41Ca and varying quantities of Ca2+ from 0-250 mg were prepared and counted on the Quantulus liquid scintillation counter. In all cases, the aqueous/ scintillant ratio was 4:16. An additional set of standards containing no additional Ca and varying quantities of 3 M HCl was prepared. Sufficient Gold Star scintillation cocktail was then added to produce a total volume of 20 mL in each vial. The vials were dark-adapted for 2 h prior to measurement. Sources were counted for sufficient time to record >30000 counts giving a measurement uncertainty of 8000 700 >7000 >9000 >700 >6000 >700
element Mn Ni Ra Sr Y Zn
decontamination factor 1300 >1000 >250 >9000 >880 >2000
tion and single particle counting of 41Ca8+ in a multianode gas ionization detector after additional mass and velocity filters. Raw ratios of 41Ca8+ counts (collected for 100-200 s) to total 40 Ca8+ charge (collected in 50 microsecond pulses at ∼3 Hz) were calculated by the data collection system. Samples and standards were run at least four times and final 41Ca/Ca calculated via normalization to the known standard ratios and assuming natural 40Ca abundance. The measured 41Ca/Ca ratios along with the measured Ca concentrations of the prepared 100 mL stock solutions were used to calculate 41Ca activity concentrations for the concrete. RESULTS AND DISCUSSION The most notable differences in this separation scheme compared with previously reported separations are the initial preconcentration step involving calcium oxalate precipitation and the use of Sr-resin to isolate Sr. The oxalate precipitation ensures effective separation of 54Mn, 60Co, and 65Zn and also removes most 133Ba and 226Ra which remain in the supernatant. Any residual Ba and Ra are effectively removed using a BaCrO4 precipitation. Sr will follow Ca but is later quantitatively removed on the Sr-resin column from 8 M HNO3. An adaptation of this procedure, but using MnO2 in place of BaCrO4 for the removal of Ra, has been reported elsewhere.30 Separation of Sr/Ba from Ca has also been reported by selective precipitation of Sr(Ba)Cl2 from >10 M HCl solutions.17 However, although the approach is suitable for the majority of samples approximately 1% of Sr was observed in the Ca fraction following three precipitation stages, and the technique may not be appropriate where high Sr decontamination factors are required. Overall decontamination factors for key interfering radionuclides were high (Table 1). Source Preparation and Experimental Determination of 41 Ca and 45Ca Efficiency Curves. The purified Ca was dissolved in 3 M HCl and mixed with an appropriate emulsifierbased cocktail (similar to that described elsewhere7,15). The 41Ca counting efficiency was highly dependent on both the acid loading and the Ca concentration decreasing from 4.6% to ∼1% for Ca loadings from 6 to 200 mg. For subsequent measurements, sources were prepared by mixing 4 mL of 3 M HCl (containing the sample) with 16 mL of Gold Star scintillation cocktail in a 22 mL polythene vial. The experimentally determined efficiencies for 41Ca were within 10% of the values predicted using the CN2003 program (30) Rowlands, F.; Warwick, P. E.; Croudace, I. W. Environmental Radiochemical Analysis III. Proceedings of the 10th International Conference of Environmental Radiochemical Analysis, 2006; Warwick, P., Ed.; Royal Society of Chemistry, Cambridge, U.K., 2007.
Figure 2. Comparison of measured and predicted counting efficiencies for 41Ca (note full LSC energy window used for measurements).
(Figure 2). The theoretical determination of liquid scintillation counting efficiencies for electron capture radionuclides have been previously highlighted as problematic because of the uncertainties in atomic data (in particular the intensity of the KLL Auger electron, the intensity of the KL X-ray, and the K-shell capture probability) and ionization quenching in the low energy region.21,22 However, for routine radioanalytical applications, these uncertainties are acceptable and have been incorporated into the overall method uncertainty budget. Predicted counting efficiencies ranged from 3-11% over the quench range of 700 to 820 giving typical efficiencies of 4.8 ± 1.4% for the quench values observed for concrete samples. These values are comparable with previously published ranges of 1-6%7 and 1-11%.22 Gamma spectrometric analysis identified 58Co, 60Co, 152Eu, and 154 Eu being produced as activation products and arising from neutron capture reactions. Europium-152 was the dominant activation product with activities ranging from 4.2 to 728 Bq/g whereas 58Co, 60Co, and 154Eu existed at significantly lower activity concentrations. In all instances, the radionuclide activity concentrations declined exponentially with distance from the reactor core, and given the relatively constant composition of the concrete this clearly indicates the effect of falling neutron flux with distance. This gradient in activity concentration was similarly found for tritium which was surprising given the expected mobility of the nuclide with time. The resistance to migration is explained by the tritium being locked in mineral lattices since it was derived from lithium activation (see Kim et al.26 for further details). Calcium-41, in line with the other activation products, also showed an exponential decline with distance from the reactor and ranged from 25 to 0.1 Bq/g. The profile (Figure 3) shows a high correlation with 152Eu (r2 ) 0.98), and the mean measured 41Ca/152Eu ratio of 0.038 is in excellent agreement with the predicted value of 0.036 (based on predicted neutron fluxes and measured Ca and Eu concentrations). Inspection of the liquid scintillation spectra showed only low energy components consistent with only 41Ca being present following the radiochemical separation (i.e., no higher energy radionuclides were present). Gamma spectrometric analysis of the purified Ca fraction from the concrete sample exposed to the highest integrated neutron flux did not detect any 152Eu Analytical Chemistry, Vol. 81, No. 5, March 1, 2009
1905
applied in calibrating the liquid scintillation counter, the observed agreement provides strong reassurance of using the relatively straightforward and more accessible liquid scintillation approach for quantifying 41Ca activities. For comparison purposes the 41Ca activity concentrations in the SGHWR bioshield are considerably higher than values reported previously in research reactor bioshields of
