Journal of Radioanalytical and Nuclear Chemistry

, Volume 281, Issue 1, pp 113–117

234U and 238U in mineral water: reference value and uncertainty evaluation in the frame of an interlaboratory comparison


    • European Commission, Joint Research Centre, Institute for Reference Materials and Measurements
  • L. Benedik
    • European Commission, Joint Research Centre, Institute for Reference Materials and Measurements
  • M. Vasile
    • European Commission, Joint Research Centre, Institute for Reference Materials and Measurements
  • M. Beyermann
    • Bundesamt für Strahlenschutz
  • U. Wätjen
    • European Commission, Joint Research Centre, Institute for Reference Materials and Measurements
  • S. Pommé
    • European Commission, Joint Research Centre, Institute for Reference Materials and Measurements

DOI: 10.1007/s10967-009-0069-8

Cite this article as:
Spasova, Y., Benedik, L., Vasile, M. et al. J Radioanal Nucl Chem (2009) 281: 113. doi:10.1007/s10967-009-0069-8


In 2007/2008 the Institute for Reference Materials and Measurements (IRMM) organised an interlaboratory comparison (ILC) on the determination of 226Ra, 228Ra, 234U and 238U activity concentrations in mineral water. This paper describes the determination of the reference values for the activity concentrations of 234U and 238U by radiochemical separation and α-particle spectrometry performed at two independent laboratories. The experimental uncertainty of the reference values is discussed in detail.


Natural radioactivityUraniumMineral waterReference valueUncertainty


Radioactivity levels in environmental samples and foodstuff are of concern for international organisations, national authorities and the general public with respect to health aspects of the population. As a support to the European Union policies the Institute for Reference Materials and Measurements (IRMM) organises on regular basis interlaboratory comparisons (ILCs) among those national laboratories monitoring the radioactivity levels in the environment of their countries.

Determination of the reference value is a particularly important phase in conducting and evaluating an interlaboratory comparison. IRMM provides a reference value and corresponding uncertainty for the measurand, independently from the results of the participants in the ILC round, according to paragraph 5.5 of ISO 13528 [1].

The recent intercomparison exercise organised by the Institute for Reference Materials and Measurements was designed to evaluate the performance of the participating laboratories in measuring the 226Ra, 228Ra, 234U and 238U activity concentrations in mineral water. The intercomparison material was bought from three different companies in Hungary. All water samples were bottled in 1.5 L PET (polyethylene terephthalate) bottles.

In this paper the procedure to assess the reference values of 234U and 238U is presented. The most influential parameters contributing to the uncertainty of the reference value are investigated. In particular, the uncertainty contributions due to the alpha spectrometry measurements, the variation between different samples of one batch (homogeneity study) as well as the stability of the samples with respect to adsorption on the walls of the bottles is discussed.

In order to establish the reference values of the activity concentrations of the radionuclides studied measurements were done by two independent laboratories—Institute for Reference Materials and Measurements (IRMM, Belgium) and Bundesamt für Strahlenschutz (BfS, Germany). Whereas three batches were analysed, only one of them (Water 3—W3) will be discussed in detail.


Measurements at BfS

At the Bundesamt für Strahlenschutz (BfS, Germany) the analysis of uranium isotopes was performed with proven methods of radiochemical separation and activity measurement [2]. After addition of 232U tracer the 1 L water sample was evaporated to dryness and the residue was wet-ashed with HNO3 and H2SO4. Subsequently, the remaining residue was dissolved in diluted HNO3. The radiochemical separation of uranium ions was performed by extraction chromatography using a UTEVA column [3], and the measurement source was prepared from the uranium containing eluate by electrodeposition. The radionuclides were deposited within 4 h in cathodic electrochemical reactions at 300 mA on a stainless steel disc. The sources were measured by α-particle spectrometry applying Ortec alpha spectrometer chambers (SOLOIST) and Passivated Implanted Planar Silicon (PIPS) detectors with an active area of 450 mm2. Measuring times between 100,000 and 300,000 s and Ortec MAESTRO (Vers. 6.04) Alpha Analysis Software for the data evaluation were used.

Measurements at IRMM

Also at IRMM the activity concentrations of 234U and 238U in the water were determined by α-particle spectrometry after extensive chemical separation in order to remove interferences from other alpha emitters. After adding 232U as tracer, uranium was preconcentrated from water samples by coprecipitation with iron (III) hydroxide at pH 9–10 using an ammonia solution [4]. The radiochemical separation was performed using a UTEVA column and the source for alpha counting was prepared by micro-coprecipitation with CeF3 [5]. 232U as a tracer for the determination of the chemical yield of the radiochemical procedure was measured together with the two other isotopes by α-particle spectrometry. Special care was taken to ensure traceability to the SI Units by using a calibrated balance and 232U tracer standardised at IRMM within 0.6% by defined solid angle counting [6].

The experimental set-up used is a common alpha spectrometry system for environmental samples which consists of Canberra alpha spectrometer chambers (Model 7401 VR) and PIPS detectors with 450 mm2 sensitive area. The measured source is deposited on a flat substrate that is placed in a parallel plane, centred at the symmetry axis of the detector. The data acquisition and analysis was done using the Genie-2000 (Vers. 3.1) Alpha Analysis Software.

The activity concentration AU of 234U and 238U, respectively, was determined by relative measurements based on the comparison of the measured peak areas of these uranium isotopes (NU) and the 232U tracer (Ntracer):
$${\text{A}}_{{\text{U}}} = {\frac{{{\text{N}}_{{\text{U}}} \cdot {\text{m}}_{{{\text{tracer}}}} \cdot {\text{A}}^\prime}{}_{{{\text{tracer}}}}}{{{\text{N}}_{{{\text{tracer}}}} \cdot {\text{V}}_{{{\text{sample}}}}}}} \cdot {\text{f}}_{\text{decay}},$$
where NU and Ntracer are the net peak areas of the corresponding uranium isotopes; Atracer is the activity per mass of the tracer solution; mtracer is the mass of the added 232U solution in the sample; Vsample is the sample volume; fdecay is a correction factor for decay of the uranium isotopes (incl. decay correction to reference date and decay during measurement).
The net area (N) was calculated by taking the numerical integral of the number of counts in the peak region (Ng) and correcting for background (B):
$$ {\text{N}} = {\text{N}}_{\text{g}} - {\text{B}} $$
The samples were typically measured for 4 days, yielding net peak areas between 660 and 11000. Twelve samples were prepared of 1.5 L and ten samples of 3 L sample volume. In total 22 samples were measured two times each.

Results and discussion

Uncertainty of uranium determination

Table 1 shows a typical uncertainty budget of the 234U and 238U activity concentration for a single measurement of one sample at the 1 s level.
Table 1

Uncertainty budget for 234U and 238U in mineral water, showing the typical propagated uncertainty contributions for a single measurement at the 1 s level


234U (%)

238U (%)

Counting statistics measurand (incl. background)



Counting statistics tracer (incl. background)



Choice of region of interest (ROI)



Volume of sample



Weighing of tracer



Activity of tracer






Combined uncertainty (quadratic sum)



The combined uncertainty is the quadratic sum of all components

The first uncertainty contributions concern counting statistics in the 234,238U peaks and the tracer peak calculated from the number of gross counts Ng and the background counts B:
$$ {\text{u(N)}} = \sqrt {{\text {{N}}}_{\text{g}} + {\text {B}}}$$
This uncertainty component reduces by a factor of √44 when taking into account all the measured samples for the uncertainty of the reference value. However, also the choice of the regions of interest around the analysed peaks plays a role. The corresponding uncertainty was estimated at 2%, and taken as systematic, which does not reduce with repeated measurements.

With respect to sample preparation, the estimated uncertainty on the volume of the sample is 0.1%. The uncertainty on the weight and the activity per mass of the tracer solution is 0.5% and 0.6%, respectively. The propagated uncertainty due to the decay correction factor of the different isotopes (i.e. half-life) is negligible.

The uncertainty on the chemical yield cannot be easily determined. The uranium isotopes and 232U tracer should have the same chemical behaviour and therefore the same yield. Nevertheless, one cannot completely exclude that the uranium in the mineral water could behave differently from the tracer because of matrix effects, for example if it were encapsulated in silica grains which are not easily dissolved. At this stage no uncertainty component is added for such possible effects.


The uncertainty components in Table 1 can be verified by repeatability (swb) tests, as each sample was measured twice. A one-way analysis of variance (ANOVA) showed a repeatability of about 2.5% for both 234U and 238U (see Table 2), which is in agreement with the random component of the uncertainty budget.
Table 2

ANOVA test results for 234U and 238U in mineral water (%)

Sample volume (L)


















In spite of the good repeatability, the ANOVA test also reveals higher variability from one source to another (sbb) (Table 2). This can in principle be due to inhomogeneity from bottle to bottle, instability of the material or an unforeseen change in the chemical yield ratio.


The measured 234U and 238U activity concentrations in 1.5 L and 3 L mineral water are plotted in Fig. 1. The error bars indicate combined standard uncertainties of the individual measurements (cf. Table 1). The mean activity concentrations and standard deviations of 234U were (45.0 ± 1.8) mBq L−1 for 1.5 L and (44.7 ± 1.6) mBq L−1 for 3 L, while the mean 238U activity values were (22.3 ± 1.0) mBq L−1 for 1.5 L and (22.0 ± 1.0) mBq L−1 for 3 L of volume. Clearly, the 238U activities are lower than those of 234U, hence the uranium isotopes are not in equilibrium in the mineral water. The within bottle inhomogeneity [1] of the samples could not be determined due to the large sample volumes needed to measure the low activity concentrations.
Fig. 1

Measured activity concentrations of 234U in mineral water for sources prepared from 1.5 L (a) and 3 L (b) of sample volume. Same for 238U (c, d). Mean activity concentration (continuous line) and standard deviation (dashed line) are indicated

The determination of the uranium isotopes in the mineral water at IRMM was carried out in a period of about 300 days. During this period an increase of the measured activity concentrations with time was observed (Fig. 2). One could speculate that this time-dependency may be due to unknown changes in conditions that influence the radiochemical separation procedure or the solid source deposition. Another possible explanation is inhomogeneity of the material, which might show up as a trend because the bottles were taken in a particular order from the batch. The observed variation was finally interpreted as inhomogeneity and taken into account in the uncertainty budget by including a standard deviation sbb of 3% and 4% for 234U and 238U, respectively, determined by the ANOVA test.
Fig. 2

Time distribution of the measured activity concentrations of 234U and 238U in mineral water

Adsorption test

As stability check, the adsorption of uranium on the walls of the bottles was tested, considering that no special treatment of the water was carried out during bottling. The check was done by filling some of the emptied bottles with distilled water and a few millilitre of concentrated nitric acid. The bottles were stored for a period of at least 1 month. Then solid sources were prepared following the same radiochemical procedure as the one above. The sources were counted for 4 days. The measured activity concentrations were below the detection limit (<0.2 mBq L−1). Therefore, no uncertainty due to adsorption contributes to the combined uncertainty of the reference value.

Reference value and uncertainty calculation

In the field of ILCs different procedures on how a reference value and uncertainty should be calculated are suggested [1, 7, 8]. In this work the reference value is calculated from data obtained from two independent expert laboratories according to clause 5.5 of ISO 13528 [1]. As the typical measurement uncertainty in both expert laboratories is comparable, the unweighted mean of their results is taken as reference value. Table 3 shows the 234U and 238U activity values obtained in BfS and IRMM for Water 3 discussed in this paper as well as the other waters of the ILC round. The uncertainty values of the laboratory results refer to standard deviation of the data, whereas for the reference value the combined standard uncertainty is given (i.e. including all uncertainty components). The main uncertainty components are 2% on the uranium determination and 3% (4%) on homogeneity. The reference date is 1 May 2006, 0:00 UTC.
Table 3

Activity concentration results for 234U and 238U in mineral water (mBq L−1)





Reference value ± combined standard uncertainty uC

Mean ± s

Mean ± s

Water 1


14.4 ± 1.3

15.8 ± 1.1

15.1 ± 1.1


10.3 ± 0.9

12.2 ± 1.1

11.3 ± 0.8

Water 2


3.9 ± 0.3

4.2 ± 0.7

4.0 ± 0.3


0.91 ± 0.14

0.92 ± 0.15

0.91 ± 0.15

Water 3


44.7 ± 1.5

42.9 ± 2.3

43.8 ± 1.6


22.1 ± 0.9

21.3 ± 0.9

21.7 ± 1.0

Reference date, 1 May 2006, 0:00 UTC


The reference values of 234U and 238U in mineral water determined in the frame of an interlaboratory comparison (ILC) were obtained independently from the participants results of the ILC round. They were calculated as a mean of the results obtained at two independent expert laboratories, using partially different methods of determination. The mean values of both laboratories were in very good agreement. The main uncertainty contributions to the combined uncertainty of the reference value came from the uranium determination with α-particle spectrometry (2%) and the homogeneity (3% and 4% for 234U and 238U, respectively). The estimated 3.6% and 4.6% combined uncertainty was considered to be adequate for the purpose of the interlaboratory comparison.

Copyright information

© Akadémiai Kiadó, Budapest, Hungary 2009