Applied Radiation and Isotopes 61 (2004) 319–323

ARTICLE IN PRESS

*Correspond

4554445.

E-mail addr

0969-8043/$ - se

doi:10.1016/j.ap

Low-level measurements of tritium in water

M. Villaa,*, G. Manj !onb

aDepartamento de F!ısica At !omica, Molecular y Nuclear, Facultad de F!ısica, Universidad de Sevilla,

Av. Reina Mercedes s/n, E41080 Sevilla, SpainbDepartamento de F!ısica Aplicada II, E.T.S. Arquitectura, Universidad de Sevilla, Av. Reina Mercedes 2, E41012 –Sevilla, Spain

Abstract

Using a liquid scintillation counter, an experimental procedure for measuring low-level activity concentrations of

tritium in environmental water has been developed by our laboratory, using the electrolytic tritium enrichment.

Additionally, some quality tests were applied in order to assure the goodness of the method. Well-known water samples

collected in the Tagus River (West of Spain) and the Danube River (Bulgaria), both affected by nuclear plant releases,

were analysed and results were compared to previous data. The analytical procedure was applied to drinking water

samples from the public water supply of Seville and mineral waters from different springs in Spain in order to

characterize its origin. Due to the very low levels of tritium in the analysed samples, some results were reported as lower

than the minimum detectable activity concentration (MDA). However, the count rate of these measurements was over

the background count rate of LS counter in all the cases. For that reason, an exhaustive discussion about the meaning

of the MDA, using an experimental essay, was made in order to establish a rigorous criterion that leads to a reliable

value in the case of low-level measurements.

r 2004 Elsevier Ltd. All rights reserved.

Keywords: Tritium levels; Water; Detection limit; Critical level

Introduction

Natural levels of 3H in environmental samples, mainly

due to interactions of the cosmic rays with the atmo-

sphere, were enhanced by the atmospheric nuclear test

conducted from 1945 until 1963. Since 1963, the stop of

atmospheric nuclear weapon tests the released activity of3H has decayed, with a half-life of 12.33 yr, in the

atmosphere (UNSCEAR, 1982). Nowadays, the levels of

tritium in the atmosphere are those of natural origin

before the nuclear tests.

Furthermore, nuclear power plants (NPP), for both

civilian and military uses of nuclear energy, have become

a significant source of tritium in the environment. These

facilities release tritium mostly into the close surface

waters and therefore the global distribution of tritium is

not uniform. Thus, a local radioactive impact can be

ing author. Tel.: +34-95-4550924; fax: +34-95-

ess: mvilla@us.es (M. Villa).

e front matter r 2004 Elsevier Ltd. All rights reserve

radiso.2004.03.027

observed in the vicinity of Nuclear Power Plants, where

the tritium activity concentration in waters might be

over the natural levels.

The purpose of this work was to establish a rigorous

method to measure low-levels of tritium in natural water

samples, as well as waters affected by nuclear power

plants. Analyses of river water samples collected around

NPP’s were done. Results were compared to the values

established by other laboratories. In addition, we have

developed a method for electrolytic enrichment of

tritium. Finally, the method was applied to drinking

water, commercial mineral waters and rainwater sam-

ples, where low activity concentrations of tritium were

expected.

All the results were obtained over the background,

however, the activity levels in some samples were close

to the Critical Level (Currie, 1968). For that reason, an

additional essay was made in order to establish a

threshold level in order to determine the activity in

samples with very low count rates just over the back-

ground count rate.

d.

ARTICLE IN PRESSM. Villa, G. Manj !on / Applied Radiation and Isotopes 61 (2004) 319–323320

2. Experimental procedure

2.1. Sampling

Different types of water samples were involved in this

work. Rainwater samples were collected using a 1m2

conic stainless steel funnel. The tap water was taken

directly from the public water supply of Seville

(EMASESATM). Mineral waters from different springs

in Spain are commercially available (Zambra, Fontvella,

Lanjaron, Fuente Primavera, Sol!an de Cabras). Finally,

river waters were provided by other laboratories. The

samples collected in the Tagus River (Spain), were

provided by Dr. Baeza (Universidad de Extremadura)

and the samples collected in the Danube River

(Bulgaria) were supplied by Mrs. Abramova (Kozloduy

NPP).

2.2. Sample treatment and measurement

The samples that were measured without enrichment

were previously distilled and then mixed with the

scintillation cocktail Optiphase Hisafe3.

The electrolytic concentration of tritium was made

following the procedure and the cells design described by

Baeza (Baeza et al., 1999) with slight variations. The

cells we have used have cylindrical geometry with

stainless electrodes. Each electrolytic run comprises 1

blank, 3 reference dilutions of tritiated water and 6

samples for measurement. The dead water (free of

tritium) used as blank is provided by Dr. Trilla from the

Servicio de Dataci !on de 3H y de 14C (Universidad

Aut !onoma de Barcelona).

An ultra-low-background Wallac Quantulus 1220

liquid scintillation spectrometer has been used for the

tritium measurements.

For an electrolyticly concentrated sample, Taylor

(1977) suggests the relationship

Ci

Cf

¼Wf

Wi

� �P

; ð1Þ

where Ci/Cf are the initial/final3H concentration in the

electrolyte. Wi and Wf are, respectively, the masses of

the electrolysed sample before and after the electrolytic

enrichment. P is the enrichment parameter, and is

assumed to be constant for all the cells in every

electrolytic run. It is determined from the cells contain-

ing traced samples.

Thus, activity concentration A (Bq/l) for an electro-

lyzed sample is deduced as

A ¼Rs � Rb

e � V � 60� eðln2�teÞ=T1=2

1

Wsi=Wsf

� �P; ð2Þ

where Rs is the net count rate (cpm) of the electrolysed

sample and Wsi and Wsf are, respectively, the masses of

the sample before and after the enrichment. V (l) is the

volume of the sample contained in the counting vial, te

(years) is the time elapsed between the collection and the

measurement of the sample, T1/2 is the tritium half-life

(12.3 years) and e is the counting efficiency. The tritiumcounting efficiency e is determined with the measure-ment of tritium standard sources.

According to Currie (1968), the Detection Limit LD is

defined as

LD ¼2:71þ 4:65 �

ffiffiffiffiffiffiffiffiffiffiffiRb � t

pt

; ð3Þ

where t is the measurement time of the background

sample that coincide with the measurement time of the

sample.

LD is expressed in cpm, although it is very common to

give units to this value to obtain the Minimum

Detectable Activity, which can be directly compared to

the activity concentration of a sample. Thus, for non

electrolysed and electrolysed samples it is defined as

MDA ¼2:71þ 4:65

ffiffiffiffiffiffiffiRbt

peV60t

; ð4Þ

MDA ¼2:71þ 4:65

ffiffiffiffiffiffiffiRbt

peV60t

�1

Wsi=Wsf

� �P: ð5Þ

The Critical Level defined by Currie (1968) is the one

used to establish whether or not a measurement is over

the background. The formula for the Critical Level, also

called Decision Limit, is.

LC ¼ 1:64 �

ffiffiffiffiffiffiffiffi2Rb

t

r: ð6Þ

The criterion is the following. If Rs�Rb > LC, with a

probability of error of 5%, the registered pulses include

a contribution by the sample. In that case we calculate

the tritium activity in the sample.

3. Results and discussion

3.1. Study of Currie’s critical level

It is usually found in the bibliography, that the

Minimum Detectable Activity is used to discriminate a

measurement from the background. According to

Currie, it must be the Critical Level LC, and not the

LD, or the equivalent MDA, the discrimination thresh-

old.

In the water samples analysed up to now, we have

observed that the counting rate in this samples is very

close to the background rate. In addition, the natural

activity present in the sample is so small that is close or

lower to the Minimum Detectable Activity of the

technique. For these reasons, the discrimination of the

tritium activity from background in these cases is not a

trivial problem.

ARTICLE IN PRESS

Table 1

Tritium activity concentration for traced samples measured after an electrolytic processes —Being Am the measured activity and At the

true activity of the tritiated water— Critical Level (LC) and Minimum Detectable Activity (MDA) of the samples and Currie’s

discrimination criterion

Sample Am (Bq/l) 7s At (Bq/l) 7s LC (cpm) (Rs�Rb) >LC MDA (Bq/l)

1 0,84 0,11 0,90 0,04 0,085 Yes 0,29

2 0,72 0,07 0,60 0,02 0,087 Yes 0,14

3 0,68 0,06 0,51 0,03 0,087 Yes 0,14

4 0,33 0,05 0,34 0,01 0,087 Yes 0,16

5 0,20 0,05 0,25 0,01 0,087 Yes 0,14

6 0,19 0,07 0,23 0,01 0,086 Yes 0,22

7 0,18 0,08 0,24 0,01 0,087 Yes 0,24

8 0,13 0,08 0,27 0,01 0,085 No 0,26

9 �0,08 �0,07 0,18 0,01 0,086 No 0,24

M. Villa, G. Manj !on / Applied Radiation and Isotopes 61 (2004) 319–323 321

Thus, it must be established if Currie’s discrimination

criterion can be applied to tritium measurements of very

low activity concentration. For that purpose, we have

made a set of electrolytic tritium concentrations using

diluted tritiated water samples of known activity; the

samples were traced with small amounts of tritium and

then concentrated electrolytically, we calculate their

activity in the usual way. Afterwards, we have compared

the true activity with the results obtained using the

electrolytic enrichment, the values are shown in Table 1.

The measurement of samples 1–3 gives counting rates

which are over the background and the activity

measured, although low, is appreciably higher than the

MDA. Furthermore, the activity measured is in coin-

cidence, within the error, with the true activity of the

tritiated water.

In two cases, samples 8 and 9, Rs�Rb is below LC.

Specifically, sample 8 gives a positive result for the

tritium activity, however, the measured activity is not in

agreement with the true activity of the sample. This

result stands out the importance of LC to discriminate a

true result from a background measurement.

For the rest of the samples, Rs�Rb is higher than LC.

In addition, in samples 6 and 7 the results of the activity

are below the MDA. However, within the error, in all the

samples where Rs�Rb > LC, the activities measured are

in agreement with the true activity of the traced samples.

This result confirm that the Minimum Detectable

Activity is an a priori limit that does not indicate whether

or not a measurement is a background result. Thus, it is

possible to present an accurate activity of a sample which

is below the MDA. On the other hand, we have verified

that if the counting rate does not exceed the Critical Limit

LC, the result of the activity concentration in this samples

is not reliable and do not has to be considered.

3.2. Riverwater samples around nuclear power plants

The radioactive impact of two NPP, both equipped

with a pressure water reactor (PWR), can be compared

by measuring the activity concentration of 3H in the

riverwater used as coolant, which is opened to the

close environment. For that reason, high 3H levels

are expected in these samples. In this work, the selected

NPP were Almaraz (West of Spain), which is located

in the bank of Tagus River, and Kozloduy (Bulgaria),

which is located in the bank of Danube River. Table 2

presents the activity concentration of 3H in the

Tagus River, obtained without electrolytic enrichment.

Table 3 presents the activity concentrations of 3H in

the Danube River. The levels observed in the Tagus

River were an order of magnitude higher than the ones

observed in the Danube River. Then, a higher radio-

active impact by Almaraz NPP in the Tagus River can

be inferred.

The direct measurement of tritiated water samples is

checked comparing our results in Tagus river to

previous data obtained by Dr. Baeza in the same

samples (Table 2). Both measurements coincide within

the error, confirming the goodness of the procedure.

Furthermore, our results have been confirmed in an

intercomparison procedure, concerning to the Tajo

River water samples. In this intercomparison were

included the measures reported by 12 laboratories from

Portugal and Spain. The average activity of all the

measures, excluding the results of two outliers, was

9579Bq/l. Our laboratory has obtained 10375Bq/l,which is a result within the incertitude.

The electrolytic enrichment step was verified in the

Danube River samples (Table 3), the same sample were

measured twice, without any enrichment procedure and

using the electrolytic enrichment technique. The activity

of the sample from upstream Danube, non affected by

Kozloduy NPP, is below the Critical Level if no

enrichment method is used. The electrolytic concentra-

tion diminishes the MDA an order of magnitude, thus,

the activity concentration for that sample is 1.7 Bq/l.

The results with and without enrichment are in

agreement, within the incertitude, for the three recol-

lected samples.

ARTICLE IN PRESS

Table 2

Tritium activity concentration in samples from the Tagus river.

Sampling location Collecting date U. Sevilla MDA U. Extremadura

A (Bq/l) 7s Bq/l A (Bq/l) 7s

Downstream (60Km) 01-07-01 15 1 1,8 15 2

Downstream (60Km) 01-09-01 19 1 20 2

Downstream (40Km) 01-11-01 17 1 21 6

Downstream (1Km) 01-07-01 121 6 123 2

Downstream (1Km) 01-10-01 36 2 36 2

Outlet/cooling reservoir 01-11-01 271 12 271 13

Outlet/cooling reservoir 01-12-01 251 11 251 12

The results correspond to the measurements carried over by the University of Sevilla and the University of Extremadura.

Table 3

Tritum activity concentration in samples from the Danube river.

Sampling Collecting date Without electrolysis With electrolysis

location A (Bq/l) 7s MDA (Bq/l) A (Bq/l) 7s MDA (Bq/l)

Downstream 01-01-03 2,0 0,4 1,9 3,0 0,2 0,2

Upstream Not detected 1,7 0,1

Outlet 10,3 1,5 10,7 0,7

The results were obtained without any enrichment processes and with electrolytic enrichment.

Table 4

Tritium activity in mineral waters from several springs in Spain,

tap and rain water and minimum detectable activity.

Origin Collecting

date

A (Bq/l) 7s MDA

(Bq/l)

Zambra — 0,28 0,08 0,23

Fontvella — 0,55 0,08 0,20

Lanjar !on — 0,46 0,07 0,19

Fuente Primavera — Not detected 0,20

Sol!an de Cabras — Not detected 0,21

Tap water May-02 0,47 0,05 0,45

Jul-02 0,15 0,04 0,20

Sep-02 0,32 0,03 0,15

Oct-02 0,31 0,07 0,21

Mar-03 0,21 0,07 0,23

May-03 0,26 0,07 0,21

Rainwater Nov-01 0,32 0,08 0,22

Mar-02 0,37 0,10 0,28

Mar-02 0,16 0,07 0,22

Apr-02 0,19 0,16 0,51

Apr-02 0,13 0,13 0,20

Nov-02 0,14 0,07 0,18

All the measurements where done using the electrolytic

concentration method.

M. Villa, G. Manj !on / Applied Radiation and Isotopes 61 (2004) 319–323322

3.3. Mineral water samples

Table 4 presents the activity concentration of 3H in

commercial mineral waters. In this case, the application

of the electrolytic enrichment is always necessary.

However the results were under the LC, in two cases:

Fuente Primavera and Sol!an de Cabras. The springs of

those does not receive any supply from the exterior, for

that reason, in the waters Fuente Primavera and Sol!an

de Cabras the natural tritium would have decayed.

Actually, water from Sol!an de Cabras spring has been

checked to be used as a blank for the electrolytic

processes. So, the tritium electrolytic enrichment proce-

dure was applied to six aliquots of Sol!an de Cabras

water and the count rates obtained were compared to

the results obtained with a background sample. The

average values for the six aliquots and the background

sample are, respectively, 1.08 and 1.10, with a standard

deviations of 0.04 and 0.03. This values are equivalent

within the error. Confirming the availability of Solan de

Cabras spring water as a blank simple.

3.4. Other samples

The low limit of detection obtained with this tritium

enrichment method make possible to measure the

activity concentration of 3H in othersamples, like

rainwater or tapwater. Thus, Table 4 presents some

results of 3H in rainwater samples collected in Seville

since November 2001.

Water public supply network (EMASESA) is usually

monitored by our laboratory as a participant in

REVIRA-REM, a radiological surveillance coordinated

ARTICLE IN PRESSM. Villa, G. Manj !on / Applied Radiation and Isotopes 61 (2004) 319–323 323

by Spanish Consejo de Seguridad Nuclear. The presence

of 3H in these samples must be well established. Some

results in samples collected in 2002 and 2003 are listed in

Table 4.

In both tables, the results are presented despite that

sometimes they are lower than the MDA. In all the

measurements carried over with rain and tap water, the

results were over LC.

4. Conclusions

We have developed an electrolytic enrichment method

to measure low-level tritium concentrations. We have

tested Currie’s discrimination levels to be used in

measurements with very low contents of tritium, the

activity concentrations obtained when the counting rate

is over the Critical Level are reliable within the

incertitude. Thus, we manage to measure successfully

river samples affected by NPP releases, but also natural

levels of tritium in mineral, tap and rainwater.

References

Currie, L.A., 1968. Limits for qualitative detection and

quantitative determination. Anal. Chem. 40 (3), 586–593.

Baeza, A., Garc!ıa, E., Mir !o, C., 1999. A procedure for the

determination of very low activity levels of tritium in water

samples. J. Radioanal. Nucl. Chem. 241 (1), 93–100.

Taylor, C.B., 1977. Tritium enrichment of environmental

water by electrolysis. Procedings of the International

Conference of Low-Radioactivity Measurements and Appli-

cations. Bratislava, Slovenske pedagogicke nakladatielstvo,

pp. 131-140.

UNSCEAR, 1982. Sources and effects of ionizing radiation:

sources and biological effects, 1982. Report to the General

Assembly, with Annexes, United Scientific Committee on

the Effects of Atomic Radiation. UN, New York, 1993.