Automated Radioanalytical System for the Determination of 90Sr in

Anal. Chem. 2009, 81, 1228–1237

Automated Radioanalytical System for the Determination of 90Sr in Environmental Water Samples by 90Y Cherenkov Radiation Counting Matthew J. O’Hara,*,† Scott R. Burge,‡ and Jay W. Grate† Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352, and Burge Environmental, Incorporated, 6100 South Maple Avenue, Suite 114, Tempe, Arizona 85283 Strontium-90 is an environmental contaminant at several U.S. Department of Energy sites, including the Hanford site, Washington. Due to its high biological toxicity and moderately long half-life of ∼29 years, groundwater and surface water contamination plumes containing 90Sr must be closely monitored. The highly energetic β radiation from the short-lived 90Y daughter of 90Sr generates Cherenkov photons in aqueous media that can be detected by photomultiplier tubes with good sensitivity, without the use of scintillation cocktails. A laboratorybased automated fluid handling system coupled to a Cherenkov radiation detector for measuring 90Sr via the high-energy β decay of its daughter, 90Y, has been assembled and tested using standards prepared in Hanford groundwater. A SuperLig 620 column in the system enables preconcentration and separation of 90Sr from matrix and radiological interferences and, by removing the 90Y present in the sample, creates a pure 90 Sr source from which subsequent 90Y ingrowth can be measured. This 90Y is fluidically transferred from the column to the Cherenkov detection flow cell for quantification and calculation of the original 90Sr concentration. Preconcentrating 0.35 L sample volumes by this approach, we have demonstrated a detection limit of 0.057 Bq/L using a 5 mL volume Cherenkov flow cell, which is below the drinking water limit of 0.30 Bq/L. This method does not require that the sample be at secular equilibrium prior to measurement. The system can also deliver water samples directly to the counting cell for analysis without preconcentration or separation, assuming that the sample is in secular equilibrium, with a detection limit of 7 Bq/L. The performance of the analysis method using a preconcentrating separation column is characterized in detail and compared with direct counting. This method is proposed as the basis for an automated fluidic monitor for 90Sr for unattended atsite operation. 1228

Analytical Chemistry, Vol. 81, No. 3, February 1, 2009

Strontium-90 (90Sr), a radioactive isotope with a half-life of 28.9 years, is produced from the fission of 235U and 239Pu (yield ∼6%). It is a pure β emitter (βmax ) 0.546 MeV) leading to its radioactive daughter product, 90Y (eq 1). This daughter product is also a pure β emitter; it has a much shorter half-life of 64.1 h and a much more energetic β particle emission (βmax ) 2.280 MeV). 90

Sr f 90Y f 90Zr(stable)

(1)

The relative half-lives of 90Sr and 90Y lead to secular equilibrium where the activities of the parent and daughter in a sample become equal. A significant number of β particles from 90Y are emitted with energy in excess of the threshold level of 0.263 MeV, where the particles may travel through condensed media at a velocity exceeding the speed of light in that medium. This leads to the production of Cherenkov photons with energies in the ultraviolet and visible wavelengths.1-4 Whereas the average β particle energy for 90Y is 0.934 MeV, that for 90Sr is only about 0.2 MeV.5 Hence, the number of detectable Cherenkov photons from 90Sr β particles is quite low. Previously, researchers have reported the Cherenkov radiation detection efficiency of 90Sr to be 99%) that was diluted into working solutions in a matrix of unacidified and acidified Hanford groundwater. The in-house standard and the subse(46) Measurement of Radionuclides in Food and the Environment; A Guidebook; Technical Reports Series No. 295; IAEA: Vienna, 1989. (47) Currie, L. A. Anal. Chem. 1968, 40, 586–591.


1231

quent solutions in groundwater had 90Y in secular equilibrium with 90Sr. The strontium-selective sorbent SuperLig 620 (230-420 mesh) was obtained from IBC Advanced Technologies, Inc. (American Fork, UT). Groundwater Sample Preparation and Characterization. Uncontaminated groundwater samples from the Hanford site (wells 699-49-100C and 699-19-88) were stored cool and in the dark without acidification. Prior to use, these samples were filtered using a 47 mm diameter, 0.45 µm type HA filter (Millipore, Billerica, MA) and then acidified to 0.1 M nitric acid. Radiotracers were always added postfiltration and postacidification. The activity concentration of the 85Sr- and 88Y-spiked samples was verified by direct analysis using a liquid scintillation analyzer Tri-Carb 3100TR (Perkin-Elmer, Boston, MA). The activity concentration of 90Y in groundwater samples was determined by Cherenkov radiation detection using the Tri-Carb 3100TR with no scintillation cocktail; the instrument’s low-level count mode was utilized. Determination of the Cherenkov radiation detection efficiency on this off-line liquid scintillation (LS) instrument was determined by dispensing 5 mL of a 90Sr stock solution in groundwater (90Sr and 90Y were in secular equilibrium) into two 20 mL LSC vials. Ultima Gold scintillation cocktail (15 mL) was added to one vial; dilute acid (15 mL) was added to the other. Blanks were prepared in similar fashion using unspiked groundwater. The net 90Y count rate was taken as 50% of the total count rate of 90Sr/90Y given by the LS spectrometer signal (in normal count mode) from the 90Sr standard in cocktail, assuming 100% detection efficiency for both isotopes in LS cocktail. The ratio of LS counts of the aqueous Cherenkov signal to the counts from the 90Sr standard in scintillation cocktail provided a detection efficiency of 0.531 for off-line Cherenkov-based measurements. Distribution Constant Measurements on SuperLig 620 in Groundwater. The 85Sr and 88Y distribution constants were measured on SuperLig 620 using uncontaminated Hanford groundwater (699-19-88) that was acidified to various concentrations of nitric acid. A small amount of resin was weighed into a tared 20 mL glass scintillation vial and mixed with a known volume of 88Y- or 85Sr-spiked groundwater. The samples were placed on an orbital shaker for a minimum of 4 h. After the completion of the contact period, an aliquot of each sample was filtered into a clean vial using a 13 mm diameter Puradisc syringe filter with 0.45 µm polysulfone membrane (Whatman, Inc., Florham Park, NJ) to remove all sorbent from the liquid. Finally, a precise volume of the filtrate was delivered to 15 mL of Ultima Gold scintillation cocktail and analyzed using the Tri-Carb 3100TR LS analyzer. The calculation of distribution constant, Kd, was performed using the following equation:

Kd )

[

Ai - Af Af

][

V M

]

(13)

where Ai and Af are the activities of the solution prior to and after contact with sorbent, respectively; M is the mass of the sorbent; V is the volume of contact solution. 1232


On-Line Cherenkov Detection Instrumentation. A Cherenkov detection flow cell (Figure 1A) was constructed in-house using a 2 dram glass vial and cap. The cell was designed to allow 5.0 mL of the 90Y-containing analyte to be delivered to and aspirated from the base of the vial via the inlet line. The second line, placed just within the cap, allowed the displaced air to vent from the cell. The resulting Cherenkov detection flow cell was positioned between the two PMTs of a flow scintillation analyzer (FSA) 610TR scintillation counter (Perkin-Elmer, Boston, MA). Counting data was acquired using Flow-One software (version 3.65) running on a laptop computer. The Cherenkov signal was acquired from the pulse height spectral region below 30 channels. The FSA’s low-level count mode was utilized. Automated Fluid Handling System. An automated fluid handling system was configured using off-the-shelf components and controlled by LabWindows/CVI version 5.0.1 software (National Instruments, Austin, TX) running on a laptop computer. The Kloehn 48 000-step digital syringe pump with a zero-dead-volume syringe (2.5 mL volume) (Kloehn Company, Las Vegas NV) featured a six-position distribution valve at its head. All reagents and samples were aspirated into the syringe at 12 mL/min and were delivered either to the Cherenkov detector or to the SuperLig 620 column at 0.5 mL/min. A four-port two-position diverter valve (Valco Instruments Co., Inc., Houston, TX) enabled the delivery of effluents either to a fraction collector or the FSA detector. Fluid lines were constructed of 1/16 in. o.d./0.03 in. i.d. Teflon FEP tubing and were held in place using 1/4-28 and 10-32 flangeless fittings made from PEEK (Upchurch Scientific, Oak Harbor, WA). To eliminate light piping into the detector flow cell, fluid lines leading to and from the FSA detector were constructed of black Teflon FEP tubing. The SuperLig 620 sorbent was slurry packed into an Omegachrom analytical column system (Upchurch Scientific) with internal dimensions of 4.6 mm × 50 mm (bed volume ) 0.83 cm3). Column effluent fractions were collected using a Gilson FC 203B (Gilson, Inc., Middleton, WI). When the fraction collector was not engaged, the effluents were routed directly to waste via its on-board solenoid valve. Caution! Radioactive solutions used in this work present radiological hazards. RESULTS AND DISCUSSION Analytical Approach and System Design. The automated fluidic system illustrated in Figure 1B uses a syringe pump to pull the sample into the system and deliver it, via the selection valve on the syringe pump head, to the separation column containing SuperLig 620. A diverter valve sends column effluents to a fraction collector where they can be saved for analysis in the characterization of system performance. After the 90Sr load and 90 Y ingrowth interval is complete, the diverter valve is toggled to connect the column with the Cherenkov detector, and the 90 Y is transferred from the column to the detection chamber using 5 mL of acidified groundwater. At this point, the Cherenkov detector (Figure 1A) is engaged to count for a predetermined time. The absolute detection efficiency, Ed, of the Cherenkov detection flow cell was measured in off-line experiments by injecting a 90Sr standard in secular equilibrium

Table 1. Details of a Single Analysis Cycle for the Determination of 90Sr from Acidified Hanford Groundwater (GW) Solutionsa step column condition 90 Sr sample load 90 Y ingrowth period transfer of 90Y to detection cell Cherenkov counting 90 Sr column strip total volume per cycle Figure 2. Entry of 90Y into the Cherenkov detection flow cell shown as a function of eluent solution volume. Flow rate is 0.5 mL/min; detector integration time is 5 s. 90Y transfer solution is uncontaminated Hanford groundwater acidified to 0.1 M HNO3.

with 90Y into the detection cell; the observed value was 0.426 (counts/s)/Bq, or 0.426 counts/decay. While the 90Y is being counted in the detector, the system regenerates the SuperLig 620 column by first delivering 15 mL of ammonium citrate to elute 90Sr and then delivering 20 mL of acidified groundwater in order to condition the column for the next analysis. These column effluents were diverted to the fraction collector or to waste. After the detection of 90Y within the Cherenkov detection chamber is complete, the syringe pump aspirates the analyte liquid from the chamber and purges it to waste. Next, it performs a triple rinse of the detection chamber (using acidified groundwater) to eliminate carryover of 90Y between samples. Alternatively, with the use of the selection valve on the pump head and the diverter valve, solutions can be delivered directly to the Cherenkov detector. This permits delivery of standards to characterize detector parameters or to directly analyze water samples without preconcentration and separation. Whereas the laboratory-based system in Figure 1B extracted water samples from a bottle, a field-deployable device would be capable of direct sampling from wells or aquifer tubes with inline filtration. Additionally, it would be set up to be capable of automated sample acidification, as required by the separation chemistry to be described next. Preconcentration and Separation. SuperLig 620, a strontiumselective solid-phase extraction material, was selected for use as the sorbent material to extract 90Sr from groundwater. Extraction chromatographic materials, typically utilized in modern laboratory radiochemical analysis as single-use columns, are known to suffer longevity issues in repeated use24 due to leaching of extractants. SuperLig solid-phase extraction materials have covalently bound ligands48-51 providing stability for repeated use. SuperLig 620, composed of a proprietary Sr-selective ligand covalently attached to a silica support, retains Sr from chemically (48) Izatt, R. M.; Bradshaw, J. S.; Bruening, R. L. Pure Appl. Chem. 1996, 68, 1237–1241. (49) Izatt, R. M. J. Inclusion Phenom. Mol. Recognit. Chem. 1997, 29, 197–220. (50) Izatt, R. M.; Bradshaw, J. S.; Bruening, R. L.; Bruening, M. L. Am. Lab. 1994, 26, 28C. (51) Izatt, S. R.; Bruening, R. L.; Krakowiak, K. E.; Izatt, R. M. The Selective Separation of Anions and Cations in Nuclear Waste Using Commercially Available Molecular Recognition Technology (MRT) Products. In Proceedings of Waste Management 2003 Symposium WM’03, Tuscon, AZ, Feb 2327, 2003; pp 1-11. http://www.wmsym.org/abstracts/2003/html/prof379.html (accessed Dec 2008).

a

reagent

volume, mL

uncontaminated GW 90 Sr spiked GW

20 350

uncontaminated GW

5

0.5 M ammonium citrate

15 390

All groundwater was acidified to 0.1 M HNO3.

untreated or acidified Hanford groundwater. However, in untreated groundwater, we have found that Y is also retained (Kd > 1000 mL/g). Therefore, groundwater is first made acidic, where Y is not retained, in order to obtain selective retention of 90Sr and isolation from 90Y. The distribution constants, Kd, for radiostrontium and radioyttrium on SuperLig 620 were measured using 85Sr and 88Y radiotracers in Hanford groundwater over a concentration range from 0.005 to 0.5 M HNO3. The results are shown in Figure S1 of the Supporting Information. The Sr Kd increases with increasing nitric acid concentration over this range, whereas Y has no affinity. At an acid concentration of 0.1 M nitric acid, which we selected for further experiments, the Kd value for Sr was 4700 mL/g. Experiments on a packed column confirmed that 90Y passed through the column unretained, while 90Sr was retained by the column during an entire 0.35 L sample load without detectable breakthrough. These results are described in the Supporting Information and shown in Figure S2. Ingrowth and Delivery of 90Y to the Cherenkov Detection Flow Cell. The SuperLig 620 column, immediately upon completion of the 90Sr preconcentration step, becomes the source of pure 90Sr as described above. Aside from an insignificant amount of 90Y in the column interstitial space, 90Y accumulation within the column can only be through the radioactive decay of its parent in a well-defined quantitative relationship (eq 3). The 90Y ingrowth time, t, begins at the completion of the delivery of the groundwater sample to the preconcentrating column. When the degree of ingrowth reaches the desired level for measurement to occur, the 90Y is fluidically transported to the analytical system’s Cherenkov detection flow cell. Figure 2 shows a detector trace from the delivery of 90Y to the Cherenkov detection flow cell using 5 mL of acidified groundwater. The detector trace clearly shows the entry of 90Y into the detection cell as a steep increase in count rate beginning at ca. 1 mL of eluent volume. At ∼1.5 mL, the count rate plateaus, showing that all available 90Y from the SuperLig 620 column was transported to the flow cell. This experiment was based on a 0.35 L sample of 111 Bq/L 90Sr solution, with 90 Y allowed to ingrow for 64.25 h (at which point 90Y was 50.1% of 90Sr activity on the column). The background-corrected Cherenkov count rate over time upon delivery of the 90Y sample from the column was determined to be 8.13 counts/s. The decay of 90Y removed from its parent isotope by delivery to the detector is shown in Figure 3, based on backgroundcorrected 200 min counts taken periodically over 10 days (9). Analytical Chemistry, Vol. 81, No. 3, February 1, 2009

1233

Figure 3. Inset graph shows a series of 200 min counts of 90Y acquired from the Cherenkov detection flow cell over 10 days. 90Y count rate at time zero is 8.13 counts/s. The average net count rate of each 200 min count cycle is plotted on the main graph (9). The observed count rate decrease is in excellent agreement with the theoretical decay curve of 90Y (- - -), which is expressed as y ) 8.13e-0.2595x.

Figure 4. Plot showing the recovery of 90Y (9) and 90Sr (2) from the SuperLig 620 column over six analysis cycles. Horizontal lines represent average recovery values: 99.9% ( 2.8% for 90Sr between cycles 1 and 6; 78.9% ( 1.5% for 90Y between cycles 3 and 6.

Detector traces are shown in the inset graph. The observed decrease in count rate over time fits the theoretical decay function of 90Y (dashed line), where λ2 is 0.2595 day-1 and 8.13 counts/s is the measured 90Y count rate at time zero. The strict adherence of the count data to the theoretical 90Y decay curve attests to the purity of 90Y in the detection cell (i.e., no 90Sr is present). While the 90Y was being counted in the Cherenkov detection flow cell, the automated fluidic system prepared the preconcentrating column for the next analysis, delivering 15 mL of 0.5 M ammonium citrate solution to the column to elute 90Sr. During the elution, the fraction collector was programmed to collect six 2.5 mL fractions. The average of three 90Sr elution profiles, each using the same SuperLig 620 column, resulted in an overall 90Sr recovery of 99.3% ± 0.8%. The elution of 90Sr was quite rapid, as over 96% of the 90Sr had been stripped from the column within the first 2.5 mL of each of the three 15 mL elution tests. Separation Reproducibility. Six consecutive analysis cycles using the same column were performed. A summary of the reagent and sample volumes required for a single 90Sr analysis cycle is outlined in Table 1. For an analysis requiring the loading of 0.35 L of 90Sr-containing groundwater sample, the total liquid volume delivered to the column per cycle was 0.39 L. The results are shown in Figure 4. The chemical recovery of 90 Sr (2) remained excellent throughout the duration of the 1234


Figure 5. Several MDAs expressed as a function of detector count time over a 24 h count interval. The horizontal bar represents the regulatory 90Sr DWL of 0.30 Bq/L (8 pCi/L). The four plots represent the MDAs for 90Sr as the volume of sampled groundwater volume increases from 0.1 to 0.5 L; other parameters are listed in the text.

testing (average ) 99.9% ± 2.8%). 90Sr recovery was calculated as the activity of 90Sr eluted from the column (in 0.5 M ammonium citrate) as compared to the theoretical activity of 90 Sr concentrated onto the column from the 0.35 L of 90Sr-spiked groundwater standard (111 Bq/L). The recovery of 90Y is plotted (9) as the percent of ingrown 90 Y that was fluidically delivered to the Cherenkov detection flow cell relative to the amount of 90Y that was theoretically present in the column following the 90Y ingrowth period. This is the recovery efficiency, Er, in eqs 7, 9, and 10. The first analysis cycle yielded nearly quantitative recovery of 90Y from the column (96.8%), whereas the next two analysis cycles showed a decline in 90Y recovery. The column performance was observed to reach a steady level of 90Y recovery beginning at the third analysis cycle. Cycles 3 through 6 averaged a recovery of 78.9% ± 1.5%. However, the 90Y that is not eluted does not accumulate on the column; experiments confirmed that the 90Y that did not elute from the aged column during the 90Y transfer to the detection cell was removed along with 90Sr during the ammonium citrate strip. A 90Y yield from the column that is less than 100% (e.g., due to column aging as just observed) is accounted for by the Er parameter. A decrease in detector sensitivity could adversely affect Ed. If either, or both, of the Er and Ed terms drift with increasing numbers of analytical cycles, a bias could result. The matrix spike addition technique of measuring a sample, then measuring the sample containing a known 90Sr spike, would allow the product of the Ed and Er terms to be determined in a single term, Espike ) EdEr. This new term would replace Ed and Er in eqs 7, 9, and 10. Matrix spike addition can be automated for unattended monitoring systems. Changes in Espike over time would indicate when the monitoring system required attention or maintenance. Minimum Detectable Activity for 90Sr Using the Preconcentrating Separation Column. The MDA (eq 12) depends the Cherenkov detector data integration time, t, detector background count rate, Cb, and the various terms comprising the measurement efficiency, Em, as defined in eq 10. Focusing on two experimentally adjustable parameters in system operation, MDAC,Sr values were calculated as a function of detector count time for four increasing sample volumes and plotted in Figure 5. The ingrowth term, I, of 0.541 (based on ingrowth time of 72 h), and experimentally determined Cb ) 0.153 counts/s, Ed ) 0.426

Table 2. Determination of

a

90

Sr Activities in Spiked Hanford Groundwater

standard

factor above/below the DWL

net detector count rate, counts/sa

calcd 90Sr activity, Bq/L

known 90Sr activity, Bq/L

1 2 3 4

0.25 0.5 1 2

0.0102 ± 0.0019 0.0162 ± 0.0019 0.0332 ± 0.0020 0.0614 ± 0.0021

0.095 ± 0.018 0.150 ± 0.018 0.307 ± 0.018 0.569 ± 0.019

0.074 ± 0.002 0.148 ± 0.004 0.295 ± 0.008 0.592 ± 0.016

analytical bias, % 28 1.3 4.2 -3.8

Detector background count rate, Cb ) 0.152 ± 0.001 counts/s.

(counts/s)/Bq, and Er ) 0.789, were used for these calculations. The Er value assumes the use of an aged SuperLig 620 column. As expected, increasing the data integration time, t, or the sample volume, V, can significantly lower the MDA for the 90Sr analytical system. Data integration time, t, greatly affects the MDA over relatively short count times, but the advantages are asymptotically decreased as count times increase. With regard to volumes, a 0.1 L sample volume with a 90Y ingrowth period of 72 h cannot measure at the DWL of 0.30 Bq/L within a 24 h detection cycle, whereas 0.2, 0.35, and 0.5 L groundwater samples can measure at the DWL within only ∼9, ∼3, and ∼1.5 h detection intervals, respectively. Note that sample concentration volume, V, can only be increased up to the point that quantitative uptake of 90Sr from the sample is maintained on the column (i.e., column breakthrough of 90Sr does not occur). The 90Y ingrowth, I, (not varied in these calculations) can only be increased up to the point that secular equilibrium of 90Y with 90Sr has been achieved (i.e., I ) 1). Analysis of Hanford Groundwater Spiked with Low Levels of 90Sr. With the use of these considerations as a guide, the analytical system using the preconcentrating separation column was tested at 90Sr concentrations between 0.25 and 2 times the regulatory DWL of 0.30 Bq/L. Four 90Sr standards in a matrix of acidified Hanford groundwater were prepared with the following activity concentrations: 0.074, 0.148, 0.295, and 0.592 Bq/L. A fresh SuperLig 620 column was prepared for each standard, so a 90Y Er of 0.968 was assumed based on prior experiments described above. A 0.35 L sample of each standard was delivered to each column, after which a 90Y ingrowth period of 175 h (90Y at 84.9% of 90Sr activity) was performed. After the transfer of the ingrown 90Y to the Cherenkov detection flow cell, a 24 h count was performed. The count rate data from these standards are listed in Table 2 and shown as Figure S3 in the Supporting Information. The plot is linear (y ) 0.1027x + 0.0014, R2 ) 0.9972), and the slope of 0.1027 is the experimentally determined measurement efficiency Em. The measurement efficiency can also be determined from the individual parameters in eq 10, which have been determined above or are known parameters from the experiment. By this method, the Em was calculated to be 0.108 (counts/s)/ (Bq/L), which is in good agreement with the slope of the calibration line; the following parameters were used: Vp ) 0.35 L, I ) 0.849, D ) 0.881, Ed ) 0.426 (counts/s)/Bq, and Er ) 0.968). Taking the calculated Em value as the system calibration, observed count rates for the standards were converted to 90Sr activity concentrations according to eq 7. The calculated results are in excellent agreement with the known values. This demon-

strates the viability of the reported system for the measurement of groundwater samples with 90Sr levels near the DWL. The MDA according to eq 12 (using same parameters listed previously in this subsection, a detector background count rate of 0.152 counts/s, and a 24 h count interval) was 0.057 Bq/L 90Sr. This MDA is a factor of ∼1.3 times below the lowest calibration standard concentration of 0.074 Bq/L and ∼5.2 times below the DWL. Ingrowth Time versus Counting Time. For a given total time devoted to ingrowth plus counting, we were curious to know how the time can best be divided. We devised a spreadsheet that calculatessfor a given total time equal to the sum of the ingrowth time and the counting timesthe 90Sr MDA for varying relative times spent on ingrowth and counting. The modeling results for several total times are shown in Figure 6, each plotted against the fraction of the total time spent counting (parameters: Vp ) 0.35 L, Cb ) 0.153 counts/s, Ed ) 0.426 (counts/s)/Bq, and Er ) 0.789). We found that the optimum split between 90Y ingrowth time and Cherenkov counting time was 2:1, which results in a count interval that is one-third of total time. This optimal count time fraction is essentially independent of total time for up to at least several days. However, the MDA curves’ minima are more sensitive to this fraction at shorter total times while being rather insensitive at long total times. For example, at a “long” total time of 96 h, there is only a few percent difference in the MDA if the fraction of time spent counting is one-quarter rather than one-third. At the optimal fraction of one-third, the improvement in MDA with increasing total time (Figure 7) approaches a point of diminishing returns at 1-2 90Y half-lives of ingrowth time (64-128 h) or 96-192 total hours including the counting time. Obtaining an MDA that is equivalent to the DWL of 0.3 Bq/L (horizontal line) in the shortest possible time necessitates 41.5 h of total time, of which 27 h (0.42 90Y half-lives) are spent on 90Y ingrowth and 13.5 h are spent counting. At one-half of a 90Y half-life of ingrowth (32 h) plus 16 h of counting, the MDA is calculated to be 0.24 Bq/L. Direct Determination of 90Sr in Groundwater without Preconcentration or Separation. At the Hanford site, 90Sr plumes can exceed concentrations of 185 Bq/L (5000 pCi/ L) in certain locations. At these levels, direct measurement of 90Y-induced Cherenkov radiation in small volumes (∼5 mL) of sample can be made using PMT-based detectors. However, the analyst must ensure that the 90Sr and 90Y exist in secular equilibrium or know the ratio of the parent/ daughter in the system. This can be determined either (1) through knowledge of the plume characteristics, (2) by performing a measurement immediately upon receipt folAnalytical Chemistry, Vol. 81, No. 3, February 1, 2009

1235

Table 3. Calculated Minimum Detectable Activities (MDAs) for the System’s Direct Counting Mode Using Count Intervals Ranging from 0.5 to 24 h count interval(h) 0.5 1 2 3 4 5a 8 12 16 24 a

Figure 6. Effects of 90Y ingrowth time vs counting time on the 90Sr MDA plotted for several specific total times given in hours in the legend. The horizontal line indicates the DWL.

Figure 7. Effect of time on the 90Sr MDA when the fraction of time spent counting is one-third of the 90Y ingrowth time plus counting time. The lower X-axis indicates the half-lives of 90Y ingrowth time while the upper x-axis gives the total time corresponding to ingrowth time plus counting time (counting time ) 1/2(ingrowth time) ) 1/3(ingrowth time plus counting time)). The intersection of the horizontal and vertical lines shows the minimum time required (41.5 h) to achieve an MDA equivalent to the DWL.

lowed by a second measurement at a later time, or (3) by aging the samples prior to measurement. A series of five groundwater standards (well 699-49-100C) were prepared with known activity concentrations of 90Sr (in secular equilibrium with 90Y): 20, 39, 79, 158, and 316 Bq/L (530-8530 pCi/L). Each solution was aspirated into the syringe pump, dispensed into the Cherenkov detection flow cell, and counted for 5 h. Plotting the background-corrected count rates (counts/ s) against the known activity concentrations of 90Sr (Bq/L) yielded the calibration line y ) 0.00193x - 0.00259. The calibration demonstrates excellent linearity, and using the 5 h count interval, the calculated MDA for 90Sr was 7 Bq/L (190 pCi/L). Several theoretical MDAs were calculated for the direct counting method, using values of 0.146 and 0.00193 (counts/s)/ (Bq/L) for Cb and Em, respectively. Table 3 provides the resulting calculated MDAs for count intervals ranging from 0.5 to 24 h. A 24 h count time results in an MDA that is 3.2 Bq/L, approximately 10 times the DWL. Discussion. The detection limit requirements for 90Sr in groundwater are extremely challenging. The 0.30 Bq/L DWL 1236


MDA (Bq/L) 22.5 15.8 11.1 9.0 7.8 7.0 5.5 4.5 3.9 3.2

Detection interval utilized for direct counting mode calibration.

for 90Sr compares with a 33 Bq/L DWL for 99Tc, for example. The mass-based concentration equivalent for the 90Sr radioactivity detection limit requirement is 5.8 × 10-11 ppm, a level that is normally extremely difficult to measure using fielddeployable instruments. Furthermore, the method must capture and determine this isotope in the presence of a 109-fold greater concentration of stable Sr. We have demonstrated a fully automated radioanalytical system capable of determination of 90 Sr activity concentrations in natural water samples to levels below the DWL, using a preconcentrating separation column and a small-volume Cherenkov detection flow cell. Cherenkov measurements can be subject to bias from other radionuclides that may contribute to Cherenkov radiation in the sample, such as naturally occurring 40K and potentially the progeny from natural U and Th (e.g., 228Ac, 214Bi). Cherenkov measurement does not distinguish between 90Sr and 89Sr, the latter of which can also lead to Cherenkov radiation. Potential interferences in direct Cherenkov counting of samples have been discussed.1,3,7,34,36,41,42 The separation approach can provide selectivity over other radionuclides and removal of matrix interferences. A radionuclide will only cause a positive bias in the preconcentrating separation approach if it meets the following four conditions: (1) it is retained on the column with 90Sr, (2) it produces a daughter product that is unretained and transfers to the detection flow cell with 90 Y, (3) the daughter decays by a mechanism that generates Cherenkov radiation, and (4) it decays at a rate that produces a significant signal in relation to 90Y in the sample. Potassium40 is retained on the column and thus will not appear in the fraction in the Cherenkov detection cell. It decays to 40Ca, which is stable and hence will not contribute to a signal. The detection approach using the preconcentrating method is specific for 90Sr; 89 Sr does not lead to a signal because it is retained on the column and its daughter product, 89Y, is stable. Retention of various other natural radionuclides, such as radium, and the radioactivity of their various progeny, remains to be fully investigated in detail. At the Hanford site, plume concentrations of 90Sr are well above the DWL in some locations. In such high level plumes, the contribution of Cherenkov radiation by natural 40K and natural U/Th progeny would be insignificant compared to the signal from 90Y. In addition, 89Sr is not an issue because it has a short half life and the waste is aged. The direct counting mode

designed into the fluidic system may be useful in locations where detection all the way down to the DWL is not required. The development of a field-deployable 90Sr analytical system could increase monitoring data frequency at a fraction of the cost of current manual sampling and analytical techniques, where groundwater wells are sampled once every 3 months at most. These data could allow researchers and engineers to better understand hydrogeologic processes within active 90Sr plumes, observe the effectiveness of plume remediation processes, provide process control feedback for pump and treat processes, and measure 90Sr at trace levels for regulatory or routine monitoring requirements. ACKNOWLEDGMENT The authors acknowledge the Department of Energy’s SBIR/ STTR program for funding to enable the creation of the laboratory prototype 90Sr monitoring system and Department of Energy’s Environmental Management Science Program (EMSP) and

Environmental Remediation Sciences Program (ERSP) for funding new science and technology for measuring radionuclides in water. J.W.G. acknowledges that a portion of the research was performed using EMSL, a national scientific user facility sponsored by the Department of Energy’s Office of Biological and Environmental Research located at Pacific Northwest National Laboratory. The Pacific Northwest National Laboratory is a multiprogram national laboratory operated for the U.S. Department of Energy by Battelle Memorial Institute. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review October 9, 2008. Accepted December 3, 2008. AC8021407


1237

Automated Radioanalytical System for the Determination of 90Sr in

Recommend Documents