Anal. Chem. 2005, 77, 3390-3393
Measurement of Cosmogenic Rainwater and Lake Water
35
S Activity in
Young-Lim Hong and Guebuem Kim*
School of Earth & Environmental Sciences/RIO, Seoul National University, Seoul 151-742, Korea
Although cosmogenic 35S (t1/2 ) 87 d) has been found to be a unique and excellent radioactive tracer for stable S in atmospheric and aqueous environments, its application has been very limited because an analytical method for its detection has not been well-documented. Here, we report a rapid and robust method for analyzing extremely low levels of 35S in rainwater and lake water samples. About 20-L water samples were preconcentrated using an anion exchange column. The purified 35S was precipitated as BaSO4, and the precipitates were collected using a GF/B filter. The 35S in precipitates on the filter was directly counted using a super-low-background liquid scintillation counter with cocktail. We successfully measured 35S in precipitation and lake water samples using this method, which promises future diverse applications of the 35S tracer to S cycling in the environment and to age determination of lake water and shallow groundwater.
cosmogenic 35S has been used to determine the removal rates of atmospheric SO2.3,6,7 It has also been applied to estimating the traveling time of atmospherically derived SO4 thorough an alpinesubalpine watershed, to tracing SO4 cycling in a lake,8 to estimating and comparing the ages of water existing in small alpinesubalpine basins,9 and to estimating the residence times of shallow groundwater.10 Although, as shown, there have been a few excellent applications of cosmogenic 35S for tracing stable S, which is one of the most critical elements in the environment, its applications have been limited due to lack of information about analytical methods. In this study, therefore, we have attempted to document a rapid and accurate method for analyzing 35S in rainwater and lake water.
The cycling of S plays a critical role in three environmental processes: local air pollution and smog, acid rain and dry deposition, and global climate change.1,2 For example, sulfate aerosols derived from industrial emissions are a major component of cloud condensation nuclei3 and reduce the greenhouse effect by reflecting radiation. Due to industrialization, the presence of large quantities of anthropogenic sulfur has led to the development of a more acidic rainfall, so-called “acid rain”, which often results in ecological damage over large areas of the industrialized world. The deposition of such a high load of anthropogenic S into the watersheds has also caused a disturbance of weathering processes and ecosystem structure.4 Thus, to understand such important processes as the removal and cycling of S in the atmosphere and watersheds, the application of natural radioactive tracers, such as 35S, is very effective. The radionuclide 35S is produced naturally in the atmosphere by the spallation of argon atoms by cosmic rays.5 It rapidly oxidizes to SO4 and then is deposited on the Earth’s surface either through precipitation or dry fallout.3 Since 35S follows exactly the pathway of stable S in the environment, it is useful for tracing the fate of S in the atmospheric and aqueous environments. For example,
EXPERIMENTAL SECTION The overall approach to the preconcentration and measurement of 35S is shown in Figure 1. Preconcentration and Separation of S. For the analyses of 35S, rainwater (6-33 L) and lake water samples (20 L) were collected in plastic buckets (32-cm diameter) and collapsible containers, respectively. The water sample was passed through a 0.45-µm capsule filter in order to remove any particulates. The sample was acidified immediately to pH 3-4 using concentrated HCl, and then 0.02 g of a stable S carrier (Na2SO4, which was confirmed to contain no 35S) was added since the natural concentration of stable S is very low in rainwater and lake water. Then, the sample was immediately passed through an anion exchange resin (Amberlite IRA 400) column (15-50 mesh) using a peristaltic pump at a flow rate of 4 L/h. About 50 g of resin was packed in a column (i.d. 2.5 cm, length 25 cm) in order to achieve full recovery of the 35S in the sample. We tested the effect of salt on the adsorption of S onto the resin (Figure 2a). A full adsorption of S was confirmed in salinity lower than 2‰. This method seems not to be applicable to waters of salinity higher than 10‰. Then, the S in the column was eluted with 300 mL of 3 M NaCl.11 We achieved the full recovery of 35S using this volume with an elution test (Figure 2b). The recovery of each sample was checked on the basis of the weight of the recovered BaSO4, which will be explained in the next section.
* Corresponding author. Phone: +82-2-880-7508. Fax: +82-2-876-6508. E-mail:
[email protected]. (1) Charlson, R.; Langner, J.; Rodhe, H. Nature 1990, 348, 22. (2) Stern, D. I. Chemosphere 2005, 58, 163-175. (3) Tanaka, N.; Turekian K. K. J. Geophys Res. 1995, 100, 2841-2848. (4) Michel, R. L.; Turk, J. T.; Campbell, D. H.; Mast, M. A. Water, Air, Soil Pollut. 2002, (Focus 2), 5-18. (5) Lal, D.; Peters, B. In Handbuch der physic; Springer-Verlag: New York, 1996; pp 550-612.
(6) Tanaka, N.; Turekian, K. K. Nature 1991, 352, 226-228. (7) Turekian, K. K.; Tanaka, N. Geophys. Res. Lett. 1992, 19, 1767-1770. (8) Michel, R. L.; Campbell, D.; Clow, D.; Turk, J. T. Water Resour. Res. 2000, 36, 27-36. (9) Suker, J. K.; Turk, J. T.; Michel, R. L. Geomorphology 1999, 27, 61-74. (10) Plummer, L. N.; Busenberg, E.; Bohlke, J. K.; Nelms, D. L.; Michel, R. L. Chem. Geol. 2001, 179, 93-111. (11) Novak, M.; Michel, R. L.; Prechova, E.; Stepanova, M. Water, Air, Soil Pollut. 2004, 4, 517-529.
3390 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005
10.1021/ac048128c CCC: $30.25
© 2005 American Chemical Society Published on Web 04/19/2005
Figure 3. Spectrum of TR/SL.
Figure 1. Schematic illustration of the analytical procedure.
Figure 2. Dependency of S recovery on the salinity of samples (a) and elution volume of 3 M NaCl (b) using an anion exchange column (Amberlite 400 resin).
Source Preparation for LSC Counting. The entire S in the eluted solution from the IRA 400 resin column was precipitated as BaSO4 by adjusting the pH to 3-4 following the addition of excess Ba solution (5 mL of 10% BaCl2‚3H2O solution for a 5-L sample). The samples were stirred vigorously to achieve good
35S
using the Packard Tri-Carb 3170
precipitation and good 35S recovery. After ∼5 h, the precipitates were collected onto a preweighed GF/B filter. The samples were dried at 60 °C for 2 h in a dry oven and then stored in a desiccator until the final weight of the sample had been measured. The chemical recovery through all procedures, including preconcentration, BaSO4 precipitation, and filtration, was determined using the weight of BaSO4 (against the added S) on the filter. In general, we obtained a consistently high recovery of 35S (>94%) using this method for all samples that we measured. The selection of the filter is important since the filter should be translucent for efficient liquid scintillation counting without the chemical dissolution of the filter. We chose the Whatman GF/B 1.0-µm filter since it is translucent in Ultima Gold LLT cocktail. After the filter with BaSO4 was transferred into a 20-mL plastic vial, 10 mL of distilled water and 10 mL of LLT cocktail were added. The material in this vial was homogenized in an ultrasonic bath. Then, 35S in the precipitates was directly counted using a super-low-background LSC. The details of LSC counting methods are described in the next section. INSTRUMENTATION Background. Liquid scintillation counting has advantages over β counting using a gas-flow proportional counter because it allows for higher efficiencies in counting and easy source preparation. We used a Packard Tri-Carb 3170 TR/SL (Packard Instrument Co.), which is especially designed for counting low-level radioactivity in samples by utilizing a detector guard made of bismuth germanate, Bi4Ge3O12 (BGO). The BGO discriminates γ and muon components of cosmic background using the duration of pulses. Cosmic rays interacting with the BGO guard produce scintillations that primarily consist of a single burst followed by a number of afterpulses which can last up to 5 µs.12 In general, a larger number of afterpulses occur after an unquenchable event, caused by cosmic ray interactions with the vial wall and photomultiplier tubes, than after a true scintillation event. The programmable TR-LSC has a variable delay-before-burst feature, which enables a further reduction in background by modifying the instrument setting. Counting Optimization. We used a NIST-traceable 35S standard, purchased from American Radiolabeled Chemicals Inc., for (12) Benitez-Nelson, C. R.; Buesseler, K. O. Anal. Chem. 1998, 70, 64-72.
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005
3391
Table 1. Summary of LSC Optimization Conditions Depending on Window Setting, Cocktail Type, Vial Type, and with/without Filter cocktail type Ultima Gold LLT
Ultima Gold AB
vial type
GF/B filter
window setting (keV)
bkg (cpm)
efficiency (%)
E2/B
plastic plastic plastic plastic glass glass plastic plastic glass glass
x x x O x O x O x O
0-167 4-167 10-115 4-167 4-167 4-167 4-167 4-167 4-167 4-167
3.72 1.87 1.50 2.42 2.18 2.43 1.46 2.20 1.85 2.26
31.4 ( 1.5 25.2 ( 1.2 15.3 ( 1.0 20.7 ( 1.1 21.9 ( 1.2 21.4 ( 1.2 22.4 ( 1.1 20.6 ( 1.1 21.4 ( 1.1 20.2 ( 1.1
264 ( 18 340 ( 24 156 ( 15 177 ( 14 221 ( 17 189 ( 14 344 ( 24 192 ( 15 248 ( 18 181 ( 14
Table 2. Variation of FOM and Background Dependent on DBB Settings count mode
background (cpm) (4-160 keV)
DBB @ 75 ns DBB @ 100 ns DBB @ 150 ns DBB @ 200 ns DBB @ 300 ns DBB @ 400 ns DBB @ 500 ns DBB @ 600 ns
2.08 2.73 3.17 3.60 5.42 5.42 7.63 8.70
35S
efficiency (%) (4-160 keV) 26.6 ( 1.3 30.2 ( 1.5 36.6 ( 1.8 42.7 ( 2.1 46.7 ( 2.3 50.3 ( 2.4 53.0 ( 2.6 57.5 ( 2.9
E2/B 339 ( 24 341 ( 24 423 ( 30 507 ( 36 403 ( 28 466 ( 32 367 ( 26 381 ( 27
optimization of the instrument and for method checks. The liquid scintillation spectrum of 35S standard is shown in Figure 3. Although the maximum energy of 35S is theoretically 167 keV, the 35S spectrum shows only up to 60 keV due to a quenching effect (Figure 3). To optimize 35S counting, the figure of merit (FOM), which is the signal-to-background ratio (the square of the detection efficiency divided by the background, E2/B), was checked for various conditions such as variation in the types of vials and cocktails, window settings, and burst delay settings (Tables 1 and 2).13 We found that the best counting region in window setting is between 4 and 167 keV using a plastic vial. The FOM is similar for Ultima Gold LLT and Ultima Gold AB cocktails (Packard Instrument Co.). The highest FOM was obtained with 200 ns of DBB. We used these optimum settings using Ultima Gold LLT cocktail and plastic vials for all rainwater and lake water samples. Quenching and Detection Efficiency. The variation of detection efficiency against quenching effect was determined for the low-level count mode (Figure 4). The detection efficiency increased by ∼5% linearly as lake water transformed spectral index of the external standard (t-SIE) increased from 180 to 260 (as the quenching effect decreased). The quenching was produced by adding either acetone or BaSO4 (sample matrix). The t-SIE is referred to as a quenching indicating parameter. The efficiency of detector was corrected for the measured t-SIE. Limit of Detection. The lower limit of detection (LLD)14 may be calculated using the following eq 1, where B is the background (13) Nour, S.; Burnett, W. C.; Horwitz, E. P. Appl. Radiat. Isot. 2002, 57, 235241. (14) Biggin, C. D.; Cook, G. T.; Mackenzie, A. B.; Pates, J. M. Anal. Chem. 2002, 74, 671-677.
3392
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005
Figure 4. Dependency of detection efficiency of 35S on quenching effects. (The t-SIE value increases as quenching effects decrease.)
LLD (Bq L-1) ) 4.66xB/EVT
(1)
expressed in counts, E is the fractional counting efficiency, V is the sample volume (L), and T is the count time (s). The background and efficiency values in the S counting window for a DBB of 200 ns are presented in Table 2. We counted the sample for 360 min for this calculation. The LLD for 35S is calculated to be 18.2 mBq L-1 for 1 L but may be lowered to 0.91 mBq L-1 for a 20-L sample size. RESULTS AND DISCUSSION We have analyzed cosmogenic 35S in precipitation and lake water samples. The details of all preconcentration and measurement procedures were presented above. The extraction efficiency of 35S using the Amberlite resin column, including unknown natural S, was checked using a two-column method (tandem fashion). The absolute adsorption efficiency of the water sample can be calculated by eq 2, where A and B represent the activity
E ) 1 - [B]/[A]
(2)
of 35S in the first and second columns (tandem fashion), respectively. About five sets of tandem samples were used for this experiment. For example, the sample S1 showed A and B values of 40.8 ( 2.2 and 2.3 ( 1.0 dpm, for a counting time of 360 min, respectively. The calculated extraction efficiency of S for all
Figure 5. Decay curve of 35S for a rainwater sample source measured by a series of LSC counts. Table 3. 35S Activity in Rainwater Samples from Seoul, Korea, from May to July, 2004 sampling period
precipitation (mm)
t-SIE
May 2-May 3 May 8-May 9 May 12-May 13 May 19-May 21 May 27-May 28 June 18 June 29-June 21 July 1-July 5 July 6-July 8 July 11-July 13
18 41 15 20 73 23 95 102 95 113
164 218 224 162 200 211 217 204 170 184
35S
(mBq/L)
15.2 ( 0.9 6.0 ( 0.4 37.0 ( 2.0 24.5 ( 1.4 14.5 ( 0.8 136 ( 8 5.8 ( 0.3 20.1 ( 1.1 51.5 ( 2.8 26.1 ( 1.4
measured natural samples was higher than 94% at a flow rate of 5 L/h. This result also gives independent confirmation that the recovery efficiency of 35S by BaSO4 is higher than 94% as determined from the weight loss of BaSO4. If the natural S concentration in a sample is measured separately, the tandem column method is not necessary since the overall chemical recovery can be obtained from the recovered BaSO4 weight. The purity of 35S separated using the Amberlite column was checked by counting samples several times (Figure 5). The slope of this exponential decay curve fits well to the decay constant of 0.007 96 d-1, indicating a good separation of 35S using this method in the actual sample matrix. In addition, duplicate runs for the actual samples (3 duplicates) showed a good agreement, within 8%, indicating good reproducibility for this method. Rain. Measurement results of 35S in rainwater samples collected from the roof of Seoul National University, Korea (37.5°N129°E), between May and July 2004 are shown in Table 3. The specific activities of 35S ranged from 6 to 137 mBq L-1, which is similar to those (20-67 mBq L-1) observed in New Haven, CT.3 The seasonal variation of our 35S, together with 7Be, which is also a cosmogenic nuclide, will be used to estimate the relative residence time of SO2 in the atmosphere since 7Be is directly adsorbed onto aerosols following the production of 7Be and 35S by the spallation of cosmic rays.
Figure 6. Vertical profile of Lake in August 2004.
35S
activity at a station in Daechung
Lake. The vertical profile of 35S in Daechung Lake (36.2°N127°E), Korea, on 24 August 2004 is shown in Figure 6. The specific activities of 35S decrease from the surface (3.06 ( 0.13 mBq L-1) to the bottom (1.16 ( 0.06 mBq L-1) in the lake water column. This may be a typical level of 35S in lake water, since 35S is introduced only from the atmosphere (there is no artificial 35S source nearby). Therefore, the average residence time of water in the lake following precipitation, which may be a result of the mixing of various ages of water, could be calculated by constructing a box model including the measured atmospheric input flux and biogeochemical removal of S from the water column. CONCLUSIONS A simple and accurate method for determining 35S in rainwater and lake water samples has been developed. This method is applicable to any aqueous samples of salinity lower than 10‰. Thus, with this rapid and accurate technique for the measurement of 35S, we can effectively explore various environmental problems, such as the residence time of SO2 in the atmosphere, the microbial cycling of S in sedimentary and aqueous environments, the biological enrichment of S, and the age of young groundwater and lake water. If both radioactive 35S and stable 34S isotopes are used together as tracers, it will be even more powerful. ACKNOWLEDGMENT This work was supported by a research grant from the KOSEF/KRF (R08-2003-000-10328-0). We thank EMBL (SNU) members who helped in the sampling and preconcentration of rainwater and lake water samples. Received for review December 20, 2004. Accepted March 21, 2005. AC048128C
Analytical Chemistry, Vol. 77, No. 10, May 15, 2005
3393
