Environ. Sci. Technol. 2004, 38, 581-586
Validated Measurements of the Uranium Isotopic Signature in Human Urine Samples Using Magnetic Sector-Field Inductively Coupled Plasma Mass Spectrometry I V A N T R E Sˇ L , † GU ¨ N T H E R D E W A N N E M A C K E R , ‡,§ C H R I S T O P H E R . Q U EÄ T E L , * , † IVAN PETROV,† FRANK VANHAECKE,‡ LUC MOENS,‡ AND PHILIP D. P. TAYLOR† European Commission Joint Research Centre, Institute for Reference Materials and Measurements, Retieseweg, B-2440 Geel, Belgium, and Laboratory of Analytical Chemistry, Ghent University, Institute for Nuclear Sciences, Proeftuinstraat 86, B-9000 Ghent, Belgium
Increased interest in measuring uranium isotope ratios in environmental samples (biological materials, soils, dust particles, water) has come from the necessity to assess the health impact of the use of depleted uranium (DU) based ammunitions during recent military conflicts (e.g., Gulf war, Kosovo) and from the need to identify nondeclared nuclear activities (nuclear safeguards). In this context, very important decisions can arise which have to be based on measurement data of nondisputable uncertainty. The present study describes the certification to 2.5% (k ) 2) relative combined uncertainty of n(235U)/n(238U) at ultralow uranium levels (∼5-20 pg g-1) in human urine samples. After sample decomposition and matrix separation, the isotope ratios were measured by means of a single-detector magnetic sector-field inductively coupled plasma mass spectrometry instrument fitted with an ultrasonic nebulizer. Correction for mass discrimination effects was obtained by means of the certified isotopic reference material IRMM184. The analytical procedure developed was validated in three complementary ways. First, all major sources of uncertainty were identified and propagated together following the ISO/GUM guidelines. Second, this quality was controlled with a matrix matching NUSIMEP-3 sample (∼0.060.7% difference from certified). Third, the instrumental part of the procedure was proven to be reproducible from the confirmation of the results obtained for three samples remeasured 7 months later (∼1.5% difference). The results obtained for 33 individuals indicated that none seemed to have been exposed to contamination by DU.
* Corresponding author phone: +32-14-571658; fax: +32-14571863; e-mail:
[email protected]. † European Commission Joint Research Centre, Institute for Reference Materials and Measurements. ‡ Laboratory of Analytical Chemistry, Ghent University, Institute for Nuclear Sciences. § Present address: Provinciaal Centrum voor Milieuonderzoek, Krijgslaan 281 S4 bis, B-9000 Ghent, Belgium. 10.1021/es0346025 CCC: $27.50 Published on Web 12/02/2003
2004 American Chemical Society
Introduction The discovery of uranium fission in 1938 induced a revolution in terms of usage of this element, and since the 1940s very large amounts of uranium have been mined and treated (1). The U isotopes that are normally measurable in nature are 234U, 235U, and 238U with abundances (%) of ∼0.0055, ∼0.72, and ∼99.27, respectively (2). For nuclear applications, the natural uranium ore is usually enriched industrially in 235U and, as a consequence, large amounts of “depleted uranium” (DU) are generated. This byproduct found use in gyrocompasses, as counterweights for aircraft, or as shielding material (3). As the abundance of 235U in DU is down to about 0.2%, it is possible to demonstrate the presence of DU from the measurement of the n(235U)/n(238U) isotope ratio. The specific activity of natural uranium is relatively low (owing to the relatively long half-life, 105-109 years, of its isotopes), and the specific activity of DU is even considerably lower. All natural uranium isotopes emit alpha particles unable to penetrate the human skin and, as a result, represent principally an internal radiation hazard (4). However, rather than its radioactivity, the major concern of health effects due to uranium is its chemical toxicity, and in that respect, natural and depleted uranium are indistinguishable (4-6). Quoting Bleise et al. (4), “uranium is about as abundant as molybdenum and arsenic and more plentiful than mercury, antimony, tungsten and cadmium. It is the heaviest naturally occurring element and is found at an average concentration of 3 µg g-1 in the earth crust. Due to its presence in soil, rocks, surface and underground water, air, plants and animals it occurs also in trace amounts in many foods and in drinking water”. In addition, release into the environment or illegal possession of uranium and other radioactive materials follow from accidents involving dispersed material, illegal dumping of nuclear scrap or waste, releases of traces from declared or clandestine sources, orphaned radioactive sources, diverted nuclear material, and illicit trafficking of nuclear or other radioactive material (7). Interest in U isotope ratio measurements in environmental samples increased in the past decade when it was reported that DU was also being used in military conflicts (e.g., Gulf war, Kosovo) as ballast for missiles/bullets (permitting these to penetrate armor plates; “antitank” weapons). When hitting hard targets, DU ammunition can generate DU dust. After deposition on the ground, resuspension can take place if the size of the particles containing DU is sufficiently small and lead to a transfer to the local human food chain and/or to inhalation by individuals. Biomonitoring (in human urine, feces, hair, and nails) is a common way to assess the exposure to DU under these conditions (4). Similar measurement capabilities (for environmental matrixes at large, that is, including soils, dust particles, etc.) are also increasingly needed for the identification of nondeclared nuclear activities from samples collected in the vicinity of possible production plants, an important field of the “nuclear safeguards” activities (8). Very important decisions of a political, economic, military, or medical nature can arise which have to be based on measurement data of nondisputable uncertainty. Since the uranium concentration is often ultralow (down to the pg g-1 level) and the sample size small (a few grams), the instrumental techniques employed for these measurements need to be sensitive. Successful application of alphaspectrometry has been reported for the determination of uranium concentration and uranium isotopic ratios in soil samples (9) or human urine samples (10, 11). However, this technique requires tedious sample treatment steps and VOL. 38, NO. 2, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
581
several days of acquisition time per sample may be necessary (9). Over the past years, the most quoted techniques for the determination of uranium at trace and ultratrace levels in environmental and biological samples are those based on mass spectrometry. Thermal ionization mass spectrometry is traditionally used for high-quality (small uncertainty) isotope ratio measurements (12, 13). Kelly and Fassett (13) reported less than 5% combined uncertainty on the determination of ∼100 pg mL-1 uranium by isotope dilution in urine samples. Inductively coupled plasma mass spectrometry (ICP-MS) is generally very sensitive and has now become the instrumentation of choice in many cases. It was used for the determination of the uranium content and/or the uranium isotope ratios in natural water samples (14), freshwater sediments (15), soil samples (16), radioactive waste material (17), and biological material such as human urine (18-21), serum (20), or hair (22). The pretreatments described in the literature for urine samples depend on the sample introduction systems involved, the detection limits required, and the combined uncertainties expected for the isotopes monitored. These sample preparation strategies vary from “no treatment” (10, 20) or simple 10-50-fold sample dilution (18, 19, 21, 23) to sample decomposition followed by chemical matrix separation such as coprecipitation (23) or extraction chromatography (9, 24). This brief literature review illustrates the main analytical difficulties at stake for the measurement of actinides in biological materials and the wide range of sample preparation methods and instrumental techniques available for this purpose. It is, by far, not comprehensive, but to our knowledge, none of the papers available on the subject describe robust validation schemes as we apply in our study. For instance, sound and realistic combined uncertainty estimation associated with the declared measurement result is nearly always lacking, which makes the comparison between the different published sets of results difficult. The aim of the present work was to certify the isotope amount ratio n(235U)/n(238U) in human urine samples containing ultralow levels of uranium (∼5-20 pg g-1) and collected in a urban area where there had been suspicion of contamination by DU. A procedure based on complete digestion of the samples and absolute uranium isotope ratio measurement by ICP-MS was developed with the objective of achieving a relative expanded uncertainty on the final results of 1.5-2.5% (k ) 2).
Experimental Section Instrumentation. The measurements were carried out on a single-detector magnetic sector-field double focusing inductively coupled plasma mass spectrometer Element2 (ThermoFinnigan MAT, Bremen, Germany). The instrument was operated at low mass resolution (m/∆m g 300), providing flat-top-shaped peaks. An ultrasonic nebulizer (U-6000AT+, CETAC Technologies, Omaha, NE), fitted with a peristaltic pump, was used to enhance sensitivity. The ASX-500, model 510 (CETAC Technologies), was used for automation of the sampling. Microwave-assisted digestion of the urine samples was performed using a Milestone Microwave System MLS 1200 (Milestone Laboratory Systems, Bergamo, Italy). Samples, Reagents, and Materials. The urine samples were available frozen (-18 °C) in polypropylene (PP) vials. Highly deionized water from a Milli-Q system (Millipore, Bedford, MA) was used throughout this work. Ultrapure concentrated nitric acid (70%) and hydrochloric acid (35%) were supplied from J.T. Baker (Ultrex, Phillipsburg, NJ). Concentrated H2O2 (30%) was purchased from Merck (Suprapur, Darmstadt, Germany). The certified isotopic reference material IRMM-184 (available as uranium solution in 5 mol L-1 HNO3) was used for the correction for mass discrimination effects. A ∼60 pg g-1 uranium solution was prepared for this 582
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 2, 2004
FIGURE 1. Analytical procedure for the n(235U)/n(238U) isotope ratio measurement in urine samples. purpose. “Sample B”, from the set of four test samples produced by IRMM for the third round of the Nuclear Signatures Measurement Evaluation Programme (NUSIMEP3), was incorporated as quality control (QC) for the analytical procedure developed. Sample B contains about 2 ng g-1 uranium, and besides the fact that it is certified for uranium isotope ratios (n(235U)/n(238U) ) 0.051 241 ( 0.000 024, k ) 2), it has a high salt content (∼35 g L-1) that mimics those of urine samples (25). Separation of uranium from the matrix was achieved using prepacked U-TEVA columns (Eichrom Technologies, Darien IL). U(VI) is extracted as a neutral nitrato complex, and the stationary phase is highly selective for U(VI) over all commonly occurring constituents such as Al, Fe, alkali and alkaline earth ions. In this project, columns with 100-150 µm mesh size working under gravitational flow (i.e., about 0.6 mL min-1) were employed. The separation was performed in an ultraclean chemical laboratory (26) to suppress the risk of airborne contamination. Only new labware (bottles, vessels, tips, syringes, etc.) was employed, and it was cleaned thoroughly following a cold cleaning procedure described in detail elsewhere (27). For the microwave digestion vessels, the cleaning strategy was different. In this case, 4 mL of HCl and 1 mL of HNO3 were added to the vessels and the same microwave program as for the urine samples was applied. Subsequently, the vessels were rinsed with Milli-Q water and dried the same way as the other labware (under less than class 100 laminar air-flow conditions). Sample Preparation. The complete analytical procedure developed is shown in Figure 1. Direct nebulization of urine would have led to a rapid degradation of the ICP-MS measurement conditions as the urine matrix is complex with up to several percent of total dissolved solids, organic
TABLE 1. Experimental Settings for the Element2 ICP-MS sample gas plasma power focus lens secondary electron multiplier voltage guard electrode total analysis time isotopes monitored no. of scans mass resolution mass window sample time samples per peak detection mode data acquisition mode nebulizer sensitivity (50 pg g-1 of natural U) dead time correction sample uptake rate
1.08 L min-1 1120 W -880 V 2500 V grounded 2:46 min 235U and 238U 4 × 800 low (m/∆m g 300) 10% 5 ms 50 pulse-counting E-scan ultrasonic U-6000AT+ ∼1 000 000 cps for 238U 14 ( 6 ns (k ) 2) ∼0.5 mL min-1
compounds included. The samples were therefore submitted to a microwave-assisted acid digestion treatment. Different mixtures of reagents (HNO3/HCl/H2O2), added to different amounts of urine, were tested, and eventually the most appropriate combination selected was 4 mL of concentrated HNO3 + 1 mL of concentrated H2O2 + 10 mL of urine. Urine samples were digested as batches of five plus one procedural blank sample. After decomposition, the samples were transferred to 50 mL PP tubes and ∼6 mL of Milli-Q water was added (i.e., estimated 2-3 mol L-1 HNO3 medium). Samples were then loaded on the U-TEVA columns (pretreating the columns with 50 mL of 0.1 mol L-1 HNO3 was found to be sufficient to remove uranium impurities). Subsequently, the matrix components were removed by rinsing the column with 30 mL of 6 mol L-1 HCl, and uranium was eluted with 0.1 mol L-1 HNO3. As sample B from the NUSIMEP-3 series is an inorganic solution, it only needed to be acidified prior to its loading on the column, without microwave-assisted acid digestion pretreatment. At that stage, the salt content in the NUSIMEP-3 samples was ∼28 g L-1 since the acidification had introduced a 1.25 dilution factor. Coupling the Ultrasonic Nebulizer to the Element2. The Element2 was coupled to the U-6000AT+ nebulizer by means of Tygon tubing (∼1 m long, 5 mm i.d.). Use of the membrane desolvator unit led to a slight drop of the overall sensitivity but helped to improve the ICP-MS signal stability. Remaining fluctuations could be successfully alleviated by carefully tuning the plasma and lens parameters. Under these conditions, the sensitivity was increased approximately by a factor of 10 (∼20 × 106 cps per ng g-1) compared to a more usual sample introduction setup (0.1 mL min-1 concentric nebulizer and Scott type spray chamber; Que´tel et al., ref 28). Optimization of the ICP-MS Settings and Isotope Ratio Measurements. The influence of the crucial tune parameters (sample gas flow rate, focus lens voltage) on the peak shape and the signal stability as well as the mass calibration were checked and optimized daily. The secondary electron multiplier was operated in the pulse-counting mode. The value of the detector dead time and its associated standard uncertainty were determined applying “method 2” to n(206Pb)/n(204Pb) ratio measurements (natural like Pb solution) as described by Nelms et al. (29). The Element2 was operated using the grounded guard electrode (GE) to eliminate the secondary discharge in the plasma and enhance the overall sensitivity. The main experimental settings are listed in Table 1 and Table 2. The instrument was allowed an hour of warm-up prior to its optimization, and another 30 min “cleanup” period running 2% HNO3 (until stabilization of the instrument
TABLE 2. Operating Conditions for the U-6000AT+ Sample Introduction System
heating (°C) cooling (°C) Ar sweep gas flow (L min-1)
ultrasonic nebulizer
membrane desolvator unit
140 3
160 2.2
FIGURE 2. Monitoring of the 238U signal intensity when rinsing the instrument with 2% HNO3 after aspiration of a ∼50 pg g-1 uranium solution (with ∼600 cps signal before aspiration). background signal around 300 cps on m/z 238) preceded the measurement session. Blank samples (from three types, corresponding to each category of samples described hereafter) were run first. Urine samples were run “semirandomly”; that is, samples from different digestion batches were alternatively measured. Isotope ratio measurement results were corrected for mass discrimination effects (external correction by bracketing with IRMM-184 every two urine samples). Quality control samples (NUSIMEP-3, sample B) were included every four or five urine samples. Three minutes of rinsing with 2% HNO3 between samples was found sufficient to prevent cross-contamination (Figure 2). Moreover, the instrument background was measured before and after the IRMM-184 samples, and whenever this background had increased significantly, an extra 1.5 min of rinsing was added. Evaluation of Combined Uncertainties. Uncertainties were estimated for all the isotope ratio values measured for this project. These were expressed as expanded uncertainties Uc ) kuc where uc is the combined standard uncertainty and k is a coverage factor equal to 2. Combined standard uncertainties were obtained by propagating individual uncertainty components according to the ISO/GUM guide (30). In practice, a dedicated software program (31) was used, based on the numerical method of differentiation described by Kragten (32). “Additive” corrections applied to the individual isotope signal intensities measured for this project include those for the instrumental background and, to some extent, the dead time effect. These factors cannot be neglected as is shown later. However, propagating these uncertainties directly with the repeatability of the measurements of the individual isotope signal intensities can lead to a gross overestimation of the resulting combined uncertainty. To avoid this risk, additive corrections on intensities were translated into multiplicative correction factors on ratios for the combined uncertainty calculations following a method described elsewhere (33).
Results and Discussion The uranium concentration in each urine sample was screened in the first place. The limit of detection and limit of quantification achievable with our instrumental setup, estimated from the results of the measurements of the blank samples, were about 0.14 pg g-1 (3σ) and 0.46 pg g-1 (10σ), VOL. 38, NO. 2, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
583
FIGURE 3. Uranium elution profile from a U-TEVA column of a spiked urine sample by means of 4 mL fractions: (1) 238U signal for a single 4 mL fraction; (2) U recovery in cumulated 4 mL fractions (expressed as percentage of total amount of U in a sample). respectively. This high sensitivity performance allowed us to certify the n(235U)/n(238U) ratios within acceptable combined uncertainty limits on urine samples down to ∼5 pg g-1 uranium (33 samples in total, including 11 containing not more than ∼10 pg g-1 U). Optimization of the Matrix Separation. Very low uranium level (
