SYNTHESIS REPORT: INTERCOMPARISON TEST FOR

THE DETERMINATION OF LOW-LEVEL TRITIUM

ACTIVITIES IN NATURAL WATERS FOR AGE DATING PURPOSES
(TRIC2012)

19 August 2014

SYNTHESIS REPORT

Prepared by

D. J. Hillegonds, L.I. Wassenaar, P.M. Klaus, P.K. Aggarwal

Isotope Hydrology Laboratory

Division of Physical and Chemical Sciences
Department of Nuclear Sciences and Applications
International Atomic Energy Agency

Vienna, Austria

1
 Abstract

A world-wide intercomparison was undertaken to evaluate the performance of laboratories

conducting low level tritium (3H) assays of surface and ground waters. Eight test samples were
prepared by dilution of a high-level tritium standard with tritium-free groundwater. The test water
samples were comprised of one tritium-free water (T20) and five water samples (T21-T25) with low
ratios between 0.43 - 7.51 TU. These low-level tritium test samples encompass 3H concentrations
currently observed in modern precipitation, surface and ground waters, whereupon each
participating laboratory employed routine pre-treatment or electrolytic enrichment procedures and
3
H counting methods. Two higher-level samples (T26, T27) were to be measured without any pre-
treatment or enrichment to evaluate performance of laboratories utilizing decay counting.
Fifty-eight laboratories reported test data to the IAEA for all, or a sub-set, of the eight test
samples. Forty-three laboratories used liquid scintillation counting (LSC) with electrolytic
enrichment, 7 laboratories utilized LSC without electrolytic enrichment, 6 laboratories utilized 3He
accumulation and mass-spectrometry, and 2 laboratories used gas proportional counting. Of these
laboratories, approximately half demonstrated a good ability to generate tritium results sufficiently
accurate for groundwater age dating purposes, with an additional 25% reporting results that were
reasonable within their reported uncertainties, but inadequate for high-precision age dating
applications.
The following general recommendations may improve the performance of many laboratories for
low-level 3H measurements: 1) increased vigilance to the overall electrolysis progression, and
possibly seeking 3H enrichments >50x will likely provide an immediate performance improvement
especially for water samples below 5 TU, ii) periodically re-evaluate the determination and reporting
of analytical uncertainties based upon measured instrumental data, and iii) because systematic bias
appeared to be more common than random errors, we strongly recommend that calibrations using
primary 3H standards are performed every 4-5 years for decay counting and for electrolysis
performance.

2
 Introduction
The IAEA Isotope Hydrology Section has conducted world-wide tritium (3H) inter-comparison
(TRIC) and performance testing since 1965 (Table 1). TRIC testing provides an ongoing assessment of
the quality of 3H data produced by laboratories worldwide, primarily aimed at those conducting
groundwater age dating applications in hydrogeology, but also for industrial environmental
monitoring purposes. As in previous IAEA-led tritium intercomparisons, test sample waters were
prepared that covered a 3H concentration range representative of the current environmental waters
in the post-bomb era. In 1965, 1970 and 1995, the TRIC sample 3H values were known
approximately; in 1975 and 1985 TRIC samples were prepared by gravimetric dilution of tritiated
standard water NBS4926 (now SRM4926) with tritium-free (dead) water. In TRIC 2000, 2004, and
2008, test samples were prepared by gravimetric dilution of tritiated standard water (SRM4927F)
with tritium-free water, and pre-measured by selected laboratories to obtain the assigned 3H
reference values before distribution to the participants. In TRIC2012, samples were similarly
prepared by careful gravimetric dilution using a high-level tritium stock and tritium-free “dead
water” but the reference values were not pre-determined, and thereby provided a blind inter-
comparison in which the final values and uncertainties were determined by the participant’s tritium
measurements. All previous TRIC test reports [1-8] are available on the Isotope Hydrology
Laboratory website (http://www-naweb.iaea.org/napc/ih/IHS_programme_ihl_tric.html).
In 2013, all of the TRIC tests results were submitted by the laboratories and were evaluated
in accordance with previously used performance assessment criteria and methods, and individual
laboratories privately received scores for their performance results. While the assessment criterial
for TRIC2012 individual laboratory scores provided a basis upon which laboratories could improve
their outcomes, these evaluation assessments depended upon the laboratories’ reported
uncertainties. The difficulty with this approach was that laboratories could technically pass by
reporting accurate but unacceptably high measurement uncertainties. Moreover, when we
examined the entire pool of TRIC2012 laboratory 3H results, we noted that a significant proportion of
laboratories with Acceptable evaluation assessments were not performing well enough to meet the
measurement uncertainties required for the purposes of high-precision ground water age dating.
Thus, for this synthesis report we updated our assessment criteria to more appropriately take into
account the i) current analytical limitations of declining 3H levels to near background levels, ii) the
need for significantly improved 3H analytical performance required to obtain high-precision
radiometric groundwater ages by 3H and by 3H/He methods, and iii) to keep in line with international
norms for performance evaluations that focus on the purpose of tritium measurements in the
context of hydrology, as outlined by ISO and IUPAC [9, 10]. Although some laboratories participating

3
in TRIC do not make tritium measurements for the purposes of high-precision age dating (e.g., most
direct counting laboratories), we included all of the laboratories in our synthesis assessment in order
to be consistent, and importantly to maintain an overall picture of 3H performance within the
context of high-precision 3H and 3H/He assays used for critical groundwater resources and
contamination evaluation.

Materials and Method

Participating Laboratories
Participating volunteer laboratories in TRIC2012 were solicited through email and various
online list servers. Overall, participation in TRIC testing decreased from a high of 86 laboratories in
2000 to 58 laboratories representing laboratories from 37 countries (Table 1). Laboratories were
advised of the approximate 3H concentration range of the test samples in order to help inform local
analytical procedures. Each laboratory correspondingly provided the IAEA with detailed information
concerning their laboratory techniques, measurement instrumentation, and analytical procedures
(Appendix A). Each participating laboratory was assigned a randomized ID in order to preserve
anonymity. All laboratory analytical reports were received in 2013, and individualized grading
reports using criteria similar to prior TRIC tests were distributed in 2013. Of nine laboratories that
received TRIC samples but did not report results, three were mass-spectrometry laboratories
currently setting up new tritium-He accumulation capabilities, and six reported technical problems
that prevented participation. The total estimated annual 3H sample throughput for all measurement
techniques by all participating laboratories is about 24,000 3H assays per year (Table 2).

Test Sample Preparation

TRIC test samples were prepared from 1200 L of filtered ground water obtained from a deep
observation well near Grafendorf, Austria; the radiocarbon determined age of this ground water was
>5ka. This was the same well used for obtaining tritium-free water in previous TRIC tests (e.g.,
denoted G4 in 2004, G3 in 2008). This “dead water” stock was referred to as G1, according to the
borehole number. The tritium-free content of G1 water was verified previously, and by repeated 3H
analyses at the IAEA tritium laboratory and by one external laboratory prior to TRIC test sample
preparation. A 10 L subsample of G1 was distilled and used for the preparation of a high-level tritium
stock solution, and for one of the high-level direct counting test samples (T27).
To improve the gravimetric accuracy in obtaining target low-level 3H concentrations of the
test samples, a high-level stock solution was made by adding the primary tritium standard NIST SRM
4361C to 6 L of distilled G1 dead water. The tritium ratio of this high-level solution was determined

4
to be 597 ± 16 TU (see below, and Appendix B). For the low-level TRIC samples, the first test sample
(T20) was comprised of G1 dead water. All other TRIC samples were made by careful gravimetric
additions of the high-level 3H stock solution to batches of G1. The low-level TRIC samples (T21-T25)
were made by weighing 70 L of G1 water into a stainless steel container, and then adding pre-
determined gravimetric amounts of the high-tritium stock solution, and then adding another 70 L of
the G1 dead water. For the two higher activity samples for direct counting, T26 was made by mixing
2 L of the high-level stock solution with 0.5 L of G1 water. T27 was made from distilled G1 water and
the high level stock solution, and was targeted to attain a tritium ratio of approximately 100-150 TU,
which is roughly what most laboratories electrolytically enrich their samples to for LSC counting
purposes. The two high-level samples (T26-27) were to be measured directly without any processing
by the laboratories (i.e., no distillation or electrolytic enrichment) in order to remove sample
processing artefacts as an additional variable, and to directly assess the performance of the
instrumentation.
All TRIC test samples were mechanically mixed for about one hour and stored in tightly
sealed in steel barrels until dispensation. The low-level TRIC samples (T20-25) were dispensed into
labelled 0.5 L HDPE liner-sealed bottles for shipment to participants, the two high-level samples for
direct counting (T26, T27) were dispensed as 25 mL aliquots into 50 mL HDPE bottles with cap liners.
A supplementary stock of TRIC sample sets was produced in order to facilitate future testing needs
by laboratories. Notably, the starting date of the TRIC test for all subsequent decay calculation
purposes was 1 December 2012. Full details of the TRIC sample preparation procedures are
documented in Appendix B.
Tritium concentrations were reported as TU (Tritium Units) in terms of the tritium/hydrogen
isotope ratio, where 1 TU corresponds to a 3H/1H ratio of 1x10-18 [11, 12]. The massic tritium activity
is related to the tritium ratio via the following relationship: 1 Bq/kg corresponds to 8.390 ± 0.015 TU
[12, 13], using the newest value of 4500 ± 8 days for the tritium half-life [14] and the CODATA 1998
value of 6.022 141 99(47)x1023 mol-1 for the Avogadro constant [15].

Reference Values and Target Uncertainties

In order to obtain reference 3H values and uncertainties for TRIC samples T20-T27, three top
performing laboratories were used to determine a high-precision 3H value for the high-level stock
solution, from which all of the TRIC test samples were made by gravimetric dilutions. The reference
laboratories included a liquid scintillation counting (LSC) and a gas proportional counting (GPC) lab
both utilizing enrichment, and a mass-spectrometry (MS) laboratory. Multiple measurements were
made by each of these laboratories (n=7) to determine the tritium content of the primary stock

5
solution. These 3H measurements were normally distributed; we therefore used the mean and 1
standard deviation which was 597 ± 16 TU, n=21 (Appendix B).
All of the TRIC test samples T20-T27 were made by careful gravimetric dilutions of this high-
level stock solution with pre-determined masses of G1 dead-water (Appendix B). The gravimetrically
obtained 3H reference values for the TRIC samples along with their propagated 1 uncertainties
were summarized in Table 1 as follows; T27 = 111.7±2.9 TU; T26 = 475±13 TU; T25 = 7.51±0.20 TU;
T24 = 4.37±0.12 TU; T23 = 2.741±0.074 TU; T22 =1.121±0.033 TU; T21 = 0.430 ±0.018 TU and T20 =
0±0.02 TU. The main source of preparative uncertainty was from the high-level stock solution. For
low-level samples T21 and T22, additional uncertainty resulted from the negligible amount of tritium
in G1 water. The uncertainty of repeated gravimetric weighing was insignificant. Because T20 was
known to be tritium-free, a “less than” <0.02 value could be considered appropriate, but we
reported +0.02TU / -0TU for situations where a laboratory is capable of making such high-precision
low-level 3H measurements.
For the TRIC2012 laboratory performance evaluations, we used the mean values listed
above and in Table 1 as the target 3H concentration values to be achieved. However, using the
propagated 1  uncertainty around the reference values for performance assessment would require
an unnecessarily high level of analytical measurement precision (and associated cost) which is not
required for most groundwater 3H age dating applications. For this reason, we established a fit-for-
purpose “standard deviation for proficiency assessment” (p) that targeted groundwater age
determination accuracy to <± 2 years for the lowest 3H test samples (<4 TU), and tightening this
requirement to ± 1 years for samples with higher (>4 TU) amounts of 3H. It is important to note what
these precision values represent : the σp values define how close a reported mean value must be to
satisfy the requirements of high-precision groundwater age dating, whereas the reference value
uncertainties are the one sigma analytical error margins determined by error propagation. An
appropriate value of σp assigned to each TRIC test sample allows age dating of that water sample to
plus or minus one year (extending to plus or minus 2 years for older waters still containing
measurable tritium).1
Accordingly, for TRIC samples above 4 TU, we established a p of ±3 % of the known
reference value (e.g. close to the propagated 1  reference uncertainties; Table 1), which targeted
age determination performance to a precision of ± 1 year. This resulted in the following target 3H
values and p: T27 = 111.7±3.4 TU; T26 = 475±14 TU; T25 = 7.51±0.23 TU and T24 = 4.37±0.13 TU.

1 3
Note: In circumstances where the end user of H data does not require high measurement precision (e.g.,
environmental contamination or radiation monitoring), participants are recommended to self-evaluate using
an appropriately increased value of σp; conversely, if a radiometric age determination with precision better
than 1 year is required, a modified z-score should be determined with a smaller value of σp than used here.

6
For the <4 TU test samples T21-T23, a p of 3 % would also target uncertainties of ± 1 year, but this
was deemed analytically unrealistic, and in most cases not needed since tritium levels this low
normally indicate older waters (decades) where such dating precision is not required. Thus, for the
<4 TU test samples we established a fixed p of ±0.1 TU. This p should be achievable by well-
performing enrichment laboratories, but importantly is fully in line with expectations for
hydrogeological age determinations corresponding to a resolution of <± 2 years by modern 3H and
3
H/He methods. This resulted in the following target values and p for samples T23-T20 as
summarized in Table 1: T23 = 2.741±0.1 TU; T22 =1.121±0.1 TU; T21 = 0.430 ±0.10 TU and T20 =
0±0.1 TU.

Data Evaluation Procedure

Criteria used for evaluating laboratory performance were based on established international
proficiency testing guidelines [9, 10, 16]. For each TRIC test sample, to determine whether a
laboratory measurement was sufficiently close to the assigned value (i.e., the accuracy or trueness),
a z-score was calculated:

z = (x-xa)/ (σp) (1)

where x was the reported tritium ratio, xa was the assigned 3H reference value, and σp was the
standard deviation for proficiency assessment (Table 1). Accordingly, z-scores were Acceptable when
-2 ≤ z ≤ 2, Questionable for z-scores in the range of 2 to 3 and -2 to -3, and Unacceptable when z was
larger than 3 or smaller than -3.
Zeta scores () were also calculated for each laboratory test sample measurement, enabling
an assessment of both the reported mean value and associated uncertainty margin:

 = (x-xa)/(u2(x)+u2(xa))1/2 (2)

where x was the reported tritium ratio, xa was the assigned value, u(x) was the reported one sigma
uncertainty and u(xa) was the propagated one sigma uncertainty of the assigned value in Table 1.2
Zeta scores may be evaluated in a similar way to z-scores, but only in situations where the
uncertainty budget of participating laboratories could be validated [9]; accordingly, we did not
evaluate or grade zeta scores as Acceptable/Questionable/Unacceptable in this synthesis report as
we did in the individualized reports (as shown in Appendix C), except for making some general

2
Note: u(xa) ≠ σp, as shown in Table 1.

7
conclusions concerning laboratories that were reporting less accurate, but statistically reasonable
data. All data and statistical calculations were conducted using Microsoft Excel.

Results
Individual Laboratories
The TRIC test results submitted by each laboratory were synthesized and tabulated in
Appendix D, with columns including the anonymized laboratory identification number (identifiable
to each laboratory), the reported tritium activity (in TU units) and reported 1 σ measurement
uncertainty for each of the eight TRIC samples. Results were sorted by laboratory ID number. In
cases where the laboratory 3H data reporting was incomplete, the cells were left empty.
Additionally, results for each test sample from participating laboratories were depicted graphically
using the calculated z-scores (Figs.1-8), and were sorted by score magnitude. These plots also
depicted the Acceptable and Questionable z-score ranges; however the laboratory reported error
margins were not shown, since these were not considered in the grading criteria. For each TRIC
sample, the percentage of Acceptable results (-2<z<2) ranged from 47 % to 76 %, the percentage of
Questionable results (2 < |z| < 3) ranged from 6 % to 13 %, and percentage of Unacceptable results
(|z| > 3) ranged from 14 % to 47 %.

High-level 3H Sample Results (T26, T27)

The two high-level samples were provided with tritium concentrations high enough to
enable direct counting without any sample processing, which allowed for the assessment of
laboratory performance in the absence of potential variations contributed by sample handling,
distillation, and preparative electrolytic enrichment process. As noted, the tritium ratios for T26 and
T27 were targeted to the range of where environmental water samples are typically enriched to by
most laboratories.3 Accordingly, the accuracy of these two high-level samples reflected the quality of
each laboratory’s instrumentation and its standardization and data processing procedures, while the
precision was a measure of the laboratory’s decay counting time, as well the method of propagating
all uncertainties.
Overall, the results for these two high-level direct counting samples were surprisingly
disappointing; only 56 % of laboratories results had Acceptable z-scores for both T26 and T27. The
Zeta scores were somewhat improved, with 76 % reporting Acceptable for both test samples.
Laboratories that had z-scores between 2 and 3 for one or both of these high-level samples should

3
As calculated from the self-reported laboratory information in Appendix A, the enrichment factor
(approximated by the ratio of pre- and post-electrolysis sample weights) is 30 for most laboratories (range =
10-120 x, median = 26 x).

8
consider reviewing their analytical and data processing procedures, including the age and type of
cocktail used, their calibration standards, and better control over factors such as fluorescence, static
charges on the vials, and final sample impurity leading to quenching. We also noted that these
higher level samples showed a small but distinctive negative bias (<3 %), which might be attributable
to analytical problems at participating labs, but could also reflect the fact that T26 was made by
diluting 2 L of distilled high activity sample with 0.5 L of un-distilled G1 water (although still very
dilute). Accordingly, it is possible some quenching could have caused this sample to be imperfectly
comparable to the 100 % distilled T27 sample.
Given that a 100 TU test sample was well within the linear range of typically enriched low-
level tritium samples (and no sample processing was required), it was troubling to observe that 13 of
50 (26 %) of labs that reported results for T27 were Unacceptable (i.e., the returned value was >9 %
different from the reference value). In some cases, there was an explanatory reason: for two mass
spectrometry labs the 25 mL high activity samples were both outside of their analytical range, for
laboratories with major problems in many/most of the TRIC samples, and for laboratories with
apparent calculation errors wherein enriched sample results scored significantly better than for T27.
In any event, laboratories conducting routine 3H analyses that were not able to produce an
Acceptable result for T27 are strongly advised to re-evaluate their data analysis and processing in
order to seek a clearer understanding of how an unprocessed direct counting sample returned an
Unacceptable value.

Low-level 3H Sample Results (T20-T25)

A total of 54 laboratories reported more than one low-level result. Of these, only 7
laboratories had fully Acceptable z-scores, 9 had Acceptable plus Questionable z-scores, and 10 had
only one Unacceptable z-score. This amounts to approximately half of the participating laboratories
being capable of making low level tritium measurements that are sufficiently appropriate for
groundwater age dating purposes. Of the remaining 28 laboratories with two or more Unacceptable
z-scores, 13 were characterized as laboratories making measurements that are accurate to within
their reported uncertainties (as evidenced by reasonable zeta scores), but these reported
uncertainties were clearly inadequate for high-precision groundwater age dating, based on their z-
scores.

Assessment of Error Sources

In order to assess the most prevalent sources of error in this TRIC dataset, pairs of z-scores
were plotted against each another (Figure 9). The red box at the centre of each plot encompassed

9
the results for within an Acceptable grade was given. Results that plotted outside the red box
indicated increasing error up to very divergent results with z-score magnitudes of nearly 10 (or
higher, not plotted). Results that plotted away from the red box and along the diagonal dashed line
suffered from systematic errors, or generally those within the shaded quadrants for each plot.
Results that fell within the unshaded quadrants showed increasing random errors towards the
outside of each plot. All six charts revealed that both random and systematic errors occurred within
the tested population of laboratories. Random and systematic errors were approximately equally
distributed in the lowest 3H level pair (T20/T21), while the remaining pairwise comparisons all
suggested a greater prevalence of systematic errors than random errors. We encourage laboratories
that produced results that fell outside the Acceptable ranges to use the information in Figure 9 to
determine whether their results were predominantly random (inconsistent practices or
performance) or systematically biases (bias in instrument performance/setup, problems with
standardization/background corrections, calculation errors).

Data Summary
Overall performance for each laboratory was summarized in Figure 10, wherein absolute
values of all z-scores were summed using the following:

Σ|z-score|=|z-scoreT20|+|z-scoreT21|+…+|z-scoreT27|. (3)

In this histogram, laboratories were sorted in order of increasing net deviation from the assigned
reference values, and were separated into two groups of laboratories; (1) those using enrichment or
mass spectrometry, and (2) those not using enrichment (i.e. direct counting). In cases where a
laboratory submitted fewer than 8 results, the cumulative z-scores were projected by:

projected Σ|z-score| = Σ|z-score|*8/(# of submitted results) (4)

with (# of submitted results) given parenthetically following the laboratory ID code in Figure 10.
Additionally, the 13 laboratories reporting less accurate but relatively high reported uncertainty data
(mentioned in the Low-level 3H Sample Results section) were marked with an asterisk. Performance
was graded in accordance with individual z-scores, where Acceptable had a Σ|z-score|<16 (i.e., the
product of 8 samples and maximum z-score of 2), and Questionable had Σ|z-score|<24.
According to this analysis, of the 51 enrichment and mass spectrometry labs, only 20
provided a high enough level of measurement accuracy appropriate for high-precision groundwater

10
age dating (denoted with green fill). Overall, scores of Questionable category were determined for
10 laboratories (orange fill), and the remaining 21 laboratories showed Unacceptable performance
for the purposes of high-precision groundwater age dating.
Laboratories using direct counting for 3H samples provided acceptable and accurate results
only for the highest 3H concentration samples (T26, T27), or they provided uncertainties that were
statistically reasonable with respect to their mean reported values. These direct counting
laboratories were not graded in Figure 9, since their results were not sufficiently precise for ground
water age dating purposes. Nevertheless, these laboratories are encouraged to determine
appropriate σp values and to self-assess using appropriate criteria that are fit-for-purpose.

Conclusions
The measurement and interpretation of low levels of 3H in ground and surface waters
remains an important and valuable assay in hydrologic studies and in water resource and
contamination assessment, despite increased analytical challenges with globally declining 3H
concentrations towards background levels in some parts of the world. Furthermore, the potential
issue of 3H releases from nuclear facilities and research activities (biomedical waste, nuclear
research) around the world into the air, soil, and ground water requires ongoing vigilance to
monitoring this important radionuclide, as currently done by the IAEA through its Global Network of
Isotopes in Precipitation (GNIP). The low 3H levels currently observed in nature, and the importance
of tritium monitoring, require that all laboratories making 3H measurements in environmental
samples are able to produce reliable, accurate, and legally defensible results. Inter-comparison tests
like TRIC help laboratories to identify prevailing and current weaknesses and potential solutions.
This TRIC synthesis report identified three key areas for improving the performance of
tritium laboratories that are involved in high resolution groundwater age dating. The following
general recommendations could improve the performance of many laboratories for low-level 3H
measurements: i) increased vigilance to the overall electrolysis progression, and possibly seeking 3H
enrichments >50x will likely provide an immediate performance improvement especially for water
samples below 5 TU, ii) periodically re-evaluate the determination and reporting of analytical
uncertainties based upon measured instrumental data, and iii) because systematic bias appeared to
generally be a problem, we strongly recommend that calibrations using primary 3H standards are
performed every 4-5 years for decay counting and for electrolysis performance.

11
 Acknowledgements

We thank all of the participating laboratories for their generous participation and contribution to
this TRIC test, as well as two laboratories which shared their revised data sets. We thank C.
Belgacem, M. VanDuren, T. Chavez, and G. Kainz for technical and logistical assistance. We thank
Uwe Morgenstern for helpful advice. Funding for TRIC was provided by the IAEA.

Inventory of TRIC2012 Samples

The IAEA purposely produced a considerable stock of excess TRIC2012 test water sets. These test
sets may be used by 3H laboratories for future testing and evaluation. An additional TRIC set can be
obtained from the IAEA cost-free, but the requester must cover the shipping costs. Please email
[email protected] for more information.

12
 REFERENCES

[1] CAMERON, J.F. Survey of systems for concentration and low background counting of tritium in
water. In: Radioactive Dating and Methods of Low-level Counting. Vienna: International
Atomic Energy Agency. (1967) p. 543-573.
[2] FLORKOWSKI, T., PAYNE, B.R., SAUZAY, G., Interlaboratory comparison of analysis of tritium in
natural waters. Int. J. Appl. Rad. Isot. 21 (1970) 453-458.
[3] TAYLOR, C.B., Interlaboratory comparison of low-level tritium measurements in water. Int. J.
Appl. Rad. Isot. 29 (1978) 39-48.
[4] HUT, G., Intercomparison of low-level tritium measurements in Water. International Atomic
Energy Agency, Vienna (1987).
[5] ÖSTLUND, G., SCOTT, E.M., TAYLOR, C.B., Fifth IAEA intercomparison of low-level tritium
measurements in water. International Atomic Energy Agency, Vienna (1995).
[6] GRÖNING, M., TAYLOR, C.B., WINCKLER, G., AUER, R., TATZBER, H.,Sixth IAEA intercomparison
of low-level tritium measurements in water (TRIC2000). International Atomic Energy Agency,
Vienna (2001).
[7] GRÖNING, M., DARGIE, M., TATZBER, H., Seventh IAEA intercomparison of low-level tritium
measurements in water (TRIC2004). International Atomic Energy Agency, Vienna (2007).
[8] GRÖNING, M., TATZBER, H., TRINKL, A., KLAUS, P., VAN DUREN, M., Eighth IAEA interlaboratory
comparison on the determination of low-level tritium activities in water (TRIC2008).
International Atomic Energy Agency, Vienna (2007).
[9] THOMPSON, M., ELLISON, S.L.R., WOOD, R., The international harmonized protocol for the
proficiency testing of analytical chemistry laboratories (IUPAC Technical Report). Pur Appl.
Chem. 78(1) (2006) pp. 145-196.
[10] International Organization for Standardization. ISO 13528: Statistical methods for use in
proficiency testing by interlaboratory comparisons, Geneva, Switzerland (2005).
[11] TAYLOR, C.B., ROETHER, W., A uniform scale for reporting low-level tritium measurements in
water. Int. J. Appl. Rad. Isot. 33 (1982) 377-382.
[12] GRÖNING, M., ROZANSKI, K., Uncertainty assessment of environmental tritium measurements
in water. Accred. Qual. Assur. 8(8) (2003) 359-366.
[13] ROZANSKI, K., GRÖNING, M. Tritium assay in water samples using electrolytic enrichment and
liquid scintillation spectrometry. In: Proceedings of a Consultants Meeting on Quantifying
Uncertainty in Nuclear Analytical Measurements, 11-14 May 1998 in Vienna: International
Atomic Energy Agency.(2004) p. 195-217.

13
[14] LUCAS, L.L., UNTERWEGER, M.P., Comprehensive review and critical evaluation of the half-life
of tritium. J. Res. Nat. Inst. Stand. Technol. 105 (2000) 541-549.
[15] MOHR, P., TAYLOR, B.N., CODATA recommended values of the fundamental physical constants:
1998. Review of Modern Physics 72(2) (1998) 351-495.
[16] THOMPSON, M., ELLISON, S. L. R., Fitness for purpose – the integrating theme of the revised
Harmonised Protocol for Proficiency Testing in Analytical Chemistry Laboratories. Accred.
Qual. Assur. 11 (2006) 373-378.

14
Table 1. Reference tritium ratios and uncertainties of the TRIC samples, the target
uncertainties for evaluating performance, Sample ID, date, and the number of participating
laboratories in previous TRIC tests.
Year Participants Reference Sample Reference Tritium Target uncertainty
(Sets Mailed) Date Ratio & Uncertainty, TU margin, TU (σp)
2012 58 (68) 2012-12-01 T20 0+0.02/-0 0.1
T21 0.430 ± 0.018 0.1
T22 1.121 ± 0.033 0.1
T23 2.741 ± 0.074 0.1
T24 4.37 ± 0.12 0.13*
T25 7.51 ± 0.20 0.23*
T26 475 ± 13 14*
T27 111.7 ± 2.9 3.4*
Previous Tests
1965 12 (29) Sample A < 100
Sample B 200 - 800
Sample C 1000 – 6000
1970 35 (38) A ~ 10
B ~ 50
C ~ 250
1975 41 (79) A < 0.2
B 8.37 ± 0.04
C 33.1 ± 0.1
D 678.5 ± 2.2
1985 57 (85) A < 0.2
B 5.94 ± 0.03
C 25.66 ± 0.13
D 100.2 ± 0.5
1994 57 (91) A ~0
B ~1
C ~5
D ~ 19
2000 86 (102) 2000-07-01 T1 10.112 +0.073 / -0.037
T2 2.152 +0.044 / -0.008
T3 0 +0.036 / -0
T4 5.252 +0.055 / -0.019
T5 26.118 +0.131 / -0.095
T6 505.83 +1.88 / -1.84
2004 70 (74) 2004-07-01 T7 0 +0.050 / -0
T8 10.920 +0.064 / -0.040
T9 1.738 +0.050 / -0.006
T10 5.433 +0.054 / -0.020
T11 25.408 +0.105 / -0.092
T12 1011.58 +3.68 / -3.68
2008 63 (74) 2008-12-01 T14 1.54 ± 0.05
T15 4.07 ± 0.05
T16 7.74 ± 0.06
T17 14.46 ± 0.08
T18 0.67 ± 0.05
T19 568.7 ± 2.3

* Calculated as 3% of the reference value.

15
Table 2: Estimated number of tritium sample measurements made by participating
laboratories in this TRIC test.
Analytical method Laboratories Samples/year

LSC with electrolytic enrichment 43 16915

LSC no enrichment 7 4050

3
He accumulation/mass spectrometry 6 2105

Gas Proportional Counting 2 1150

all 58 24220

LSC – liquid scintillation counting.

16
  Acceptable  Questionable  Unacceptable

Figure 1. Laboratory z-scores for TRIC sample T20 (dead water), sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-
scores (Equation 1), with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green
dashed lines). Outliers are in red symbols. The value of σp for this sample is 0.1TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed
with an asterisk provided results outside the scale of the Figure.

17
  Acceptable  Questionable  Unacceptable

Figure 2. Laboratory z-scores for TRIC sample T21, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1), with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are in red symbols. The value of σp for this sample is 0.1TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed with an
asterisk provided results outside the scale of the Figure.

18
  Acceptable  Questionable  Unacceptable

Figure 3. Laboratory z-scores for TRIC sample T22, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1) with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores between ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are the red symbols. The value of σp for this sample is 0.1TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed with
an asterisk provided results outside the scale of the Figure.

19
  Acceptable  Questionable  Unacceptable

Figure 4. Laboratory z-scores for TRIC sample T23, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1), with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are in red symbols. The value of σp for this sample is 0.1TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed with an
asterisk provided results outside the scale of the Figure.

20
  Acceptable  Questionable  Unacceptable

Figure 5. Laboratory z-scores for TRIC sample T24, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1), with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are in red symbols. The value of σp for this sample is 0.13TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed with
an asterisk provided results outside the scale of the Figure.

21
  Acceptable  Questionable  Unacceptable

Figure 6. Laboratory z-scores for TRIC sample T25, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1), with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are in red symbols. The value of σp for this sample is 0.23TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed with
an asterisk provided results outside the scale of the Figure.

22
  Acceptable  Questionable  Unacceptable

Figure 7. Laboratory z-scores for TRIC sample T26, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1 with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are in red symbols. The value of σp for this sample is 14TU, as shown in Table 1. The highest quality data is near zero.

23
  Acceptable  Questionable  Unacceptable

Figure 8. Laboratory z-scores for TRIC sample T27, sorted from lowest to highest value and showing laboratory ID on the x-axis. Y-axis values are unitless z-scores
(Equation 1), with Acceptable scores being within ±2*(σp) (dashed green lines), and Questionable scores within ±2-3*(σp) (e.g. between the orange and green dashed
lines). Outliers are in red symbols. The value of σp for this sample is 3.4TU, as shown in Table 1. The highest quality data is near zero. Laboratories numbers listed with an
asterisk provided results outside the scale of the Figure.

24
Figure 9. Pairwise comparison of z-scores. Laboratories falling inside the red box provided acceptable z-scores for both samples. For measurements falling outside the red
box, there was greater uncertainty with increasing distance away from the origin. Results within the shaded areas or falling along the diagonal line reflected a prevalence
of systematic errors, whereas results within the unshaded areas denoted a prevalence of random errors.

25
 enrichment & decay counting or helium accumulation & mass spectrometry no enrichment

Figure 10. Net deviation from known values for all samples combined, expressed as the sum of the absolute value of all z-scores. Green bars denote laboratories with sums less
than 16 (i.e., Acceptable), orange bars denote laboratories with sums less than 24 (i.e., Questionable). Laboratories that did not enrich samples are shown on the right and are
not graded since these measurements were likely not made for high precision age dating purposes. Laboratory listed with an asterisk (*) submitted results that were
insufficiently accurate for high precision groundwater age dating purposes, but provided results that were accurate to within the reported uncertainty.

26
APPENDIX A: Analytical Parameters as Reported by Laboratories
sample size, sample size volume of minimum accum-
lab enrich- before after cocktail background count rate detection ulation
lab # type ment standard used working standard activity halflife used instrument(s) used enrichment enrichment LSC cocktail used (Q=Quantalus, T=TRICARB) limit in TU error for a 5TU sample time method for degassing
1 LSC yes NIST SRM 4926E 10 TU 4500 days Quantulus 1220 (two of them) 500 mL 10-16mL Ultima Gold uLLT, HiSafe 3 12 mL 0.7 - 0.9 cpm 0.6 TU 0.4 TU
2 MS 2.3 TU 4500 days MAP 215-50 500 0.2 cps 0.02 TU 0.25TU 100 - 200 mechanical shaking
3 LSC yes CMI, ER 25X 370 TU 12.33 years Quantulus 1220 500 mL Ultima Gold LLT 10 mL 1.2 7 2.4
4 MS NIST 15 TU 4500 days MAP215-50 500 0.1 cps 0.1 0.2 TU 60 mechanical shaking
5 LSC yes NIST, SRM 4361 180 TU 4500 days TRICARB 3100TR 250ml 15 Ultima Gold LLT 12 3 0.6 0.7
6 LSC yes NIST, SRM 4361B-21 30 TU 4500 days Quantalus 1220, TRICARB 1550, 1500, 2500 250 ml 8 ml Ultima Gold LLT 13-15ml Q=1, T=2.5, 1.9, 2.3 cpm 0.8 0.8 TU
7 GPC yes NIST, SRM4927F 28,000 TU 12.32 years 275 ml 5 to 6 ml Hydrogen 0.1TU 0.175 TU
8 MS SRM 4361-B 13.9 TU 12.43 and 12.32 VG5400 500 <0.1 cps 0.3 TU <0.4 TU 21 heat/chill cycles
9 LSC yes NIST SRM 4926E 220 TU 4500 days Quantalus 1220 1L 30 mL Ultima Gold LLT 10 mL 1.89 cpm 0.5 TU 10 %
10 LSC yes PTB 4453 1000 Bq/L 12.33 years Quantulus 1220 500 mL 17 mL Zinsser Quicksafe 400 12 mL 1.2 cpm 0.84 5%
11 LSC yes PTB, 2005-1660 338696 TU 4510 days TRICARB 3180 TR/SL 250 ml 10 ml Ultima Gold LLT 10 ml 1.5 TU 10 %
13 LSC no NIST SRM 4927 12.32 Quantulus 1220 7-8 mL Ultima Gold LLT 12 mL 0.7 cpm 6 TU Below MDA
14 LSC yes NIST, NBS SRM 4927F 33956.9 +/- 93.4 TU 4496.8 days 2x Quantalus 1220 250 or 500ml 8ml Zinsser Quicksafe 400 13.5ml 0.82 +/- 0.07 0.4 TU
15 LSC yes GSF - Germany 2012 dpm in 1-6-1997 12.32 years Quantulus 1220 500mL 20mL Ultima Gold LLT 7.5mL 0.5 cpm 0.6 TU 0.3 TU
16 LSC yes NIST; SRM 4927F ~300 cpm 4500 days Quantalus 1220, 500mL ~11mL Ultima Gold LLT 12 mL 0.94 0.38 0.4
17 LSC yes Amersham TRY64 43.13 TU 4500 days Quantulus 1220 950 mL 14 mL Ultima Gold uLLT 12 mL 0.55 & 0.98 cpm 0.15 TU ≤ 5%; 0.25 TU
18 LSC yes ER X (Czech Metrology Institute) 10 dpm 4500 days Quantalus 1220, TRICARB 3170SL 500 ml 11-12 ml Ultima Gold LLT 12 mL 0.8÷1.0 0.1 0.4
20 LSC yes NIST, SRM 4927 16952 TU 4497 days TRICARB 2250CA 250 g 10 g Ultima Gold uLLT 12 ml 1.8 cpm 1.0 TU 1.3 TU
21 LSC yes NIST SRM4926-D 700 TU 4500 TRICARB 2770 TR/SL 500 ML 12 ML ULTIMA GOLD LLT 12 ML 1.4 0.75 0.47
22 MS no standard natural groundwater calibrated 12.33 years GV Helix SFT 200 ml 0.1 cps 0.1 TU 0.2 TU 200 days ultrasonic
23 LSC yes NIST - SRM 4926E 498.5 TU in 01st Aug. 2008 12.32 yrs TRICARB 3170 TR/SL 500 ml 20 ml Ultima Gold LLT 10 ml ~ 1.6 cpm 0.6 TU ~0.25 TU for 500ml
24 LSC yes NIST, SRM 4926E 18806,64 TU 4500 days QUANTULUS 1220 250 ml 17 ml Ultima Gold LLT 10 ml 1,5 cpm 1,7 TU 0,45
25 LSC no PerkinElmer, 1210-121 12,312 years Quantulus 1220 8 ml Ultima Gold uLLT 12 ml 0,6 - 0,8 cpm 7 TU
26 LSC yes NIST, SRM 4361C ~2.7cpm/g and ~3.5 cpm/g 12.33 yrs Quantulus 1220, TRICARB 3180TR/SL 200 g 10 g Ultima Gold LLT 10 ml T=1.29, Q=1.31-1.33cpm 0.6 TU ±0.2 TU
27 LSC yes NIST SRM 4361C 0.46 Bq/g 4500 days Quantalus 1220 250 ml 13-15 ml Ultima Gold LLT 9 ml 1,0-1,2 cpm 0,4 0,3 TU
28 LSC yes SRM 4926E 4500 +- 8 days TRICARB 1600-TR, Quantulius 1220 520 ml 10 - 12 ml Ultima Gold uLLT 20 ml T=2.3 cpm, Q=0.9 cpm 0.3 TU 0.4 TU
29 LSC yes SRM 4927F 12.32 years Quantulus 1220 1000 mL 8.3 mL Quicksafe 400 (Zinsser) 12 mL 0.45 cpm 0.02 TU 0.07 TU
30 LSC yes NIST SRM 4926E 50TU 4500 days Perkin Elmer TRICARB 2500TR 500, 200, 50mL 9mL Ultima Gold LLT 9mL 1.0 cpm 0.3 TU 0.3 TU for 500mL
31 LSC yes SRM 4427 F 19672 ± 152TU on 19/10/2012 4500 days TRICARB 3100 TR 500 ml 22 ml Ultima Gold LLT 12 ml 1.5 cpm 0.6 TU 0.3 TU
32 LSC yes Wallac 2E05 dpm 100x diluted 12.32 years Quantulus 1220 250 ml 20 ml Ultima Gold LLT 10 ml 3.60 cpm 1 TU 0.5 TU
33 LSC yes CERCA LEA : 42.1 kBq/g 29/01/2007 6400TU 12.32 years TRICARB 3180 TR/SL 500 mL 15 to 20 mL Ultimagold ULLT 12 mL 0.35 0.4 0.5
35 LSC yes T17 from Tric2008 11.603 TU 12.32 Year TriCarb 3180TR/SL 250 ml 10 ml Ultima Gold LLT 12 ml about 4 cpm
37 LSC yes NBS4361, PTB2005-1466 1696 TU, 298214 TU 12.43 years Quantalus 1220 100-400ml 10 ml Ultima Gold LLT 11 ml 0.7 0.7 0.5 (100ml), 0.2 (400ml)
38 LSC yes CMI Praha, ER 3 H HWR 2 40 Bq/mL 4500 days TRICARB 2900TR 250 mL 10 mL Ultima Gold LLT 10 mL 10 cpm 4 TU 1,3 TU
39 GPC no SRM 4361 ca. 150 TU 12.32 yr 9 ml no CH4-Ar 0.4 1 1
40 LSC no SRM 4926 E 15000 TU 4500 days Quantalus 1220 10ml Ultima Gold LLT 10ml 0.7 8 4%
41 LSC yes NIST SRM 4926 E 1000, 500, and 100 TU 4500 +/- 8 days Quantulus 1220 500 mL 10-15mL Ultima Gold LLT 12 mL 0.58 cpm 0.10 TU 0.15 TU
42 LSC yes SRM 4927 4500 Days TRICARB 3170SL 500ml 10 to 15 ml Ultima Gold LLT 10 ml counts per minute 0.6 TU 0.85
43a MS 12.32 years modified-VG5400 SFT 100 mL 0.1 cps 3TU ca. 5% 30 days ultrasonic
43b LSC yes 12.32 years ALOKA LB5 500 mL 50 mL Ultima Gold LLT 50 mL
47 MS NIST 7-13TU 4500 days VG5400 50-2500 ml 0.15 cps 0.012 TU 0.12 60 mechanical shaking
48 LSC yes SRM 4361C ~8 TU 12.32 years TRICARB 2770TR/SL 500 mL ~20 mL Ultima Gold LLT 11 mL 1.07 cpm 0.2 0.4
49 LSC yes SRM 4927 103.4 TU 4500 days Quantulus 1220 400 g 9 mL Ultima Gold LLT 11 mL 0.61 0.5 0.3
50 LSC yes NIST, SRM 4926 11552 TU 12.32 yr Quantulus 1220 500 ml 8 ml Ultima Gold LLT 12 ml 0.9 counts/min 0.5 TU 0.8 TU
52 LSC yes Czech Metrology Institute, ER 2 10TU 4510 days Tri-Carb 3170TR/SL 250 ml 9 ml Ultima Gold LLT 11 ml 1.855 0.9 0.18
53 LSC no SRM 4361C 4500 days Quantulus 1220, Tricarb 3170,3180 NA NA Ultima Gold LLT 12ml 0.48 7.8 NA
54 LSC yes SRM 4927 12.32 TRICARB 1900TR 500 20 Ultima Gold LLT 11 ml 4.89 cpm 1.1 TU 0.05
55 LSC yes NIST, SRM 4361C 12.32 years TRI-CARB 2000CA/LL 200 mL 10 mL Ultima Gold LLT 10 mL 1.8 0.5 TU 0.25
56 LSC yes SRM 4927 4500 days Quantalus 1220 500 ml 10 g Ultima Gold LLT 12.5 g 0.7 cpm 0.2 cpm 0.5
57 LSC yes NIST SRM 4361C 2g NIST SRM 4361C 4500 d Quantulus 1220, ALOKA LB5, Tricarb 3180 250ml, 500ml 9ml or 70ml UG LLT 13,5-70ml 1 cpm 0.3 TU 0.5 TU
58 LSC no PerkinElmer, product no. 1210-121 3.3 kBq 4500 days Quantulus 1220 8 ml Ultima Gold uLLT 12 ml 0.65-0.74 CPM 4 TU 2.2 TU (1000 min. count)
59 LSC yes SRM4926E 10291 TU 12.32 years TRICARB3180TR/SL 500 ml 20 ml Ultima Gold LLT 12ml 1.1 cpm 0.5 TU 0.2 TU
60 LSC yes SRM 4926E 24000 TU 4500 days TRICARB 2000 CA/LL 250 ml 15 ml Ultima Gold LLT 12 ml 4.9 +/- 0.2 cpm 1TU 0.5 - 0.6 TU
61 LSC yes SRM4926 10000 TU 4540 days Quantalus 1220 250 ml 11 ml OPTIPHASE HISAFE 14 ml 0.60 cpm 0.25 TU 0.5 TU
63 LSC yes NIST SRM 4927F 20000TU 12.32 years Quantulus 1220 250 ml 20ml Perkin Elmer Optiphase Supermix 12ml 0.90cpm 0.50 TU 0.40 TU
64 LSC yes NIST-SRM 4926 D-14 10537TU(03-01-2013) 4500 DAYS TRICARB 3170SL 250 ml 15 ml Ultimagold llt 12 ml 1cpm 1 TU 0.7 TU
65 LSC no NIST SRM 4926 E 17 Bq/ml(143945.8 TU) 12.33 years Beckman & Coulter LS 6500 8 ml Beckman & Coulter Ready Gel 10 ml 15.594 CPM 44.54 TU -
66 LSC yes NPL E5492/03 60 dpm/g 12.43 Quantulus 175 ml 7 ml HiSafe 2 16 ml 1 1 TU ± 0.3 TU

27
APPENDIX B: Details of TRIC Sample Preparation
Some details of the test sample preparation were provided in the main synthesis report. The
gravimetrics of T21-T27 are given in Table 5 below. G1 water was collected from the Grafendorf well
in two new 600 L plastic tanks. Distilled G1 water was produced using a closed system glass still
(Barnstead Mega-Pure MP-6A). Samples T21-T25 were prepared in a 200 L stainless steel tank by
filling a 3 L plastic beaker with G1 water, weighing it, pouring the water into the tank, and re-
weighing the beaker for amount of water added to the tank; this process was repeated between 45
and 49 times to achieve the desired final amount, with the tritiated water solution being similarly
added with a smaller beaker after the first 70 L of the G1 additions. The 200 L tank had not
previously held high-level tritium before, and was carefully rinsed and dried prior to use; preparation
of TRIC test samples proceeded from lowest tritium ratio to highest, in order to avoid contamination
of the low-level tritium test samples. Given smaller needed quantities, T26 and T27 were produced
in glass containers by directly adding the high-level sample followed by G1. Where necessary, a
funnel was used to eliminate unaccounted for loss, with careful rinsing of the funnel to assure
quantitative addition of the tritium-containing solution, and a final correction for water loss on the
funnel itself.

Table 5: Amounts of G1 and tritiated water solutions used to produce TRIC2012 test samples. T20 is
not listed, because this sample was G1 water. Uncertainty margins in the Table reflect propagation
of the balance reproducibility (Kern model PLE 4200-2N, stated reproducibility ± 20mg).
G1 added, g Tritiated water solution added, g
T21 143479.4 ± 0.1 103.34 ± 0.02
T22 139670.4 ± 0.1 262.84 ± 0.02
T23 139653.6 ± 0.1 644.17 ± 0.02
T24 139189.1 ± 0.1 1025.76 ± 0.02
T25 138484.4 ± 0.1 1764.37 ± 0.02
T26 509.83 ± 0.02 1992.73 ± 0.02
T27 2080.58 ± 0.02* 478.72 ± 0.02
*distilled G1

The tritiated water solution contained approximately 450 Bq 3H, and was made up with NIST SRM
4361C and distilled G1 water. In order to assign reference values for T21-T27, we chose to use three
reference laboratories that submitted high precision results, were consistent among one another,
were near the centre of the distribution, and were each internally consistent with the gravimetric

28
dilutions in Table 5. Interestingly, these three laboratories consisted of one LSC and GPC lab (both
utilizing enrichment), and one MS laboratory. Each of the 7 measurements from these three
laboratories (i.e., 21 in all) was used to calculate individual estimations of the original tritium ratio in
the tritiated water solution, using the weights from Table 5. The resulting 21 values for the tritiated
water solution were normally distributed, and we used the mean and standard deviation to arrive at
the final tritium ratio of this solution, which was 597 ± 16 TU. We chose not to utilize the weighted
average of these 21 measurements to avoid imparting bias based on reported precision of these
three laboratories. The final reference values (Table 1) were arrived at using the calculated tritium
ratio of the tritiated water solution and the weights in Table 5.

29
APPENDIX C: Information Gathering Forms and Example Score Report

30
Example of an evaluation sheet sent to participating laboratories (IAEA lab ID 14).

31
32
33
Example Initial Score Report

34
APPENDIX D: Individual Laboratory Results – white/black lettering =
Acceptable, grey = Questionable, white lettering = Unacceptable, no value
given = results not submitted.
T20 T21 T22 T23 T24 T25 T26 T27
lab
1 0.03±0.015 0.194±0.046 1.4±0.14 2.43±0.29 2.96±0.25 5.6±0.59 475±16 106.8±3.7
2 0.01±0.02 0.38±0.04 1.14±0.04 2.8±0.08 4.53±0.12 7.72±0.2
3 0.018±0.014 0.5±0.08 0.94±0.08 2.8±0.2 4.33±0.24 7.01±0.33 480±18 116±7
4 0.05±0.02 7.42±0.24
5 -0.5±0.6 0.3±0.6 0.6±0.6 2.9±0.6 4.5±0.7 7.2±0.8
6 0.14±0.2 0.51±0.2 1.2±0.2 3.52±0.28 7.89±0.38 3.93±0.27 475.31±2.67 109.14±1.97
7 -0.02±0.04 0.46±0.04 1.1±0.06 2.83±0.08 4.34±0.11 7.36±0.16 471±6 111.9±2.7
8 0.05±0.08 0.25±0.29 0.97±0.12 2.47±0.37 3.37±0.43 7.03±0.47
9 1.02±1.01 0.03±0.18 -0.11±0.33 1.39±1.18 3.21±1.79 6.06±2.46 468.3±21.5 106.4±10.3
10 0±69% 0.7±0.14 1.4±0.168 2.9±0.232 4.5±0.32 7.6±0.46 468±9.36 116±3.48
11 0.37±0.55 0.34±0.55 1.01±0.53 2.48±0.57 3.99±0.57 7.83±0.64 495±18 117±13
13 9.9±1.7 421.9±6.6 103.5±3.2
14 0.2±0.1 0.5±0.1 1.2±0.1 2.6±0.1 4.9±0.3 8.3±0.4 463.4±11.5 102.4±4.3
15 0±0.2 0.5±0.2 1±0.2 3.1±0.3 5.1±0.3 8.4±0.4 509.9±21 103.1±7
16 0.18±0.23 1.27±0.25 1.91±0.26 4.93±0.31 7.81±0.35 14.09±0.43 479±17 120±9
17 0.03±0.03 0.4±0.05 1.04±0.07 2.42±0.12 4.15±0.2 7.06±0.33 463±10 113±3
18 0.03±0.07 0.48±0.08 1.23±0.11 2.68±0.21 4.43±0.31 7.98±0.52 469.5±14.2 112.4±4.1
20 -0.7±0.42 0.64±0.46 1.39±0.52 2.87±0.26 4.53±0.47 8.4±0.62 465.48±20.55 112.15±9.55
21 0.084±0.0086 0.991±0.099 2.735±0.265 1.077±0.105 4.237±0.404 7.142±0.668 474.04±26.49 133.29±8.22
22 0.07±0.03 0.65±0.05 1.63±0.1 3.41±0.22 6.26±0.4 11.03±0.7 164±10
23 0.13±0.37 0.36±0.32 1.49±0.45 3.68±0.45 5.78±0.35 8.46±0.43 341.9±18.5 174.1±14.3
24 0.14±0.27 0.39±0.28 1.02±0.29 2.66±0.31 4.43±0.35 7.27±0.39 439.05±2.76 103.71±2.78
25 466±24 109±11
26 -0.22±0.14 0.34±0.15 1.33±0.16 2.58±0.16 4.12±0.17 7.34±0.2 464±7 111±3
27 0.05±0.08 0.45±0.1 1.2±0.15 2.9±0.2 5.05±0.3 7.1±0.35 470±8 117±6
28 -0.03±0.22 0.45±0.22 0.95±0.22 2.7±0.3 4.8±0.3 8.1±0.3 470±8 112.5±5
29 0.018±0.011 0.437±0.016 1.115±0.024 2.702±0.042 4.45±0.06 7.45±0.09 466.1±3.2 113.4±1.5
30 -0.05±0.18 0.24±0.2 1.04±0.2 2.73±0.23 4.13±0.26 7.05±0.33 471±8 114±4
31 0.5±0.3 0.36±0.3 1.76±0.3 2.64±0.3 4.1±0.3 7.9±0.3 456.27±27.7 126.63±16.6
32 0.3±0.2 0.8±0.3 1.5±0.3 2.9±0.3 6±0.4 7.8±0.5 478.7±10.8 130.1±5.8
33 0.2±0.3 0.6±0.1 0.9±0.1 2.5±0.4 4.6±0.6 7.1±0.5 454±14 97±10
35 40.22±0.733 3.482±0.268 0.752±0.081 4.548±0.093 6.008±0.131 13.357±1.477 465.81±15.87 105.43±10.121
37 0.1±0.1 0.9±0.4 1.2±0.4 2.8±0.4 4.5±0.4 7.9±0.5 495±17 140±6
38 -0.46±1.5 1.1±1.5 2.13±1.5 3.29±1.5 5.83±1.5 7.9±1.7 443.4±77.7 109.4±25.3
39 0.19±0.91 0.19±0.94 0.27±0.92 2.77±1.04 8±1.03 7.14±1.03 441.45±9.12 109.42±2.69
40 425±31 112±18
41 0.01±0.12 0.62±0.12 1.29±0.12 2.99±0.17 4.48±0.17 7.51±0.27 451.12±7.5 107.91±3.47
42 0.72±0.38 2.73±0.41 5.52±0.64 13.52±1.71 5.02±0.59 8.29±1.03 576.27±5.41 2045.55±9.15
43a 0.2±0.2 0.2±0.2 1.2±0.3 1.8±0.2 3.1±0.2 7.5±0.4 370±10 55±2
43b -0.4 (<LOD) -0.4 (<LOD) 1.17±0.49 2.26±0.5 3.85±0.47 7.63±0.34
47 -0.015±0.028 0.414±0.031 1.081±0.039 2.873±0.064 4.348±0.091 7.382±0.132 476.2±9.5 114.2±2.6
48 0.6±0.2 0.2±0.2 1.2±0.3 2.5±0.3 4.5±0.4 7.2±0.5 455±20.1 113.8±9
49 -0.06±0.11 0.34±0.12 1.19±0.13 2.58±0.18 4.56±0.25 7.1±0.37
50 0.19±0.2 0.45±0.23 1.2±0.3 2.2±0.3 4.3±0.7 7.5±1.1 477±9 122±5
52 0.82±0.21 0.94±0.12 1.7±0.13 2.5±0.21 4.59±0.26 7.49±0.39
53 -2.59±0.5 -1±1.5 3.22±0.73 0.5±1.6 4.38±0.49 6.63±0.48 467.6±3.1 108.48±0.52
54 0.27±0.58 0±0.5 1.35±0.61 6.24±0.69 4.76±0.69 5.92±0.68 443.6±29.6 99.69±7.68
55 -0.29±0.19 0.57±0.22 1.01±0.21 2.09±0.21 3.76±0.21 6.68±0.24 414±8 95.7±3.7
56 0.07±0.2 0.41±0.2 1.15±0.3 2.86±0.3 6.07±1 7.39±1 472.8±17 107.1±8
57 0±0.2 0.4±0.2 1.3±0.2 3±0.2 4.8±0.3 7.9±0.4 455±20 115±6
58 1.6±2.6 0.6±2.6 -1.7±-2.6 2.9±2.6 4.4±2.6 8.7±2.7 474.4±6.5 109.5±4.8
59 0.1±0.1 0.5±0.1 1.1±0.2 2.6±0.2 4.5±0.3 6.9±0.5 463.8±28.9 113.6±7.5
60 1.2±0.9 1.1±0.9 0.2±0.8 2.3±0.8 3±0.8 8±0.8
61 0.16±0.11 0.58±0.08 1.35±0.14 2.85±0.16 4.45±0.19 7.87±0.26 467±5 107±3
63 -1.32±0.3 0.37±0.33 0.82±0.34 2.64±0.38 3.52±0.39 7.72±0.46 450±28 106±9
64 4.7±0.5 6.6±0.5 11.8±0.6 6.4±0.5 8.9±0.5 11.6±0.6 452±8.6 105.4±4.2
65 414.89±7.05 113.83±1.71
66 0 1±0.3 1.6±0.3 3.1±0.3 5.2±0.3 8.3±0.4 480±20 125±6

35
APPENDIX E: List of Participants
ROBERT CHISARI Rudolf ENGELBRECHT
AUSTRALIAN NUCLEAR SCIENCE & TECHNOLOGY ORGANISATION Seibersdorf Labor GmbH
NEW ILLAWARRA ROAD, LUCAS HEIGHTS Radiation Safety and Applications
SYDNEY, NSW, 2234 Seibersdorf, Lower Austria, 2444 Seibersdorf
AUSTRALIA AUSTRIA
[email protected] [email protected]

Philipp Klaus Zildete Rocha

IAEA - NAPC -IHS CDTN/CNEN - Centro de Desenvolvimento da
Wagramerstrasse 5 Tecnologia Nuclear

Vienna, Vienna, A-1400 Avenida Presidente Antônio Carlos, 6627

AUSTRIA Belo Horizonte, Minas Gerais, 31270-901
[email protected] BRAZIL
[email protected]

Ratan Mohapatra Richard Heemskerk

University of Ottawa Environmental Isotope Lab, University of Waterloo
140 Louis Pasteur 200 Univerisyt Ave. W.
Ottawa, Ontario, K1N 6N5 Waterloo, Ontario, N2L 3G1
CANADA CANADA
[email protected] [email protected]

Robert Drimmie Nada Horvatincic

Isotope Tracer Technologies Inc Rudjer Boskovic Institute
695 Rupert St., Unit B Bijenicka 54
Waterloo, Ontario, N2V 1Z5 10000 Zagreb
CANADA CROATIA
[email protected] [email protected]

Jiri Bruthans Tarja Heikkinen

Charles University in Prague Radiation and Nuclear Safety Authority
Albertov 6 Laippatie 4
Praha 2, Praha, 12843 Helsinki, FI-00880
CZECH REPUBLIC FINLAND
[email protected] [email protected]

Travi Yves Jürgen Sültenfuß

Université D'Avignon et des Pays de Vaucluse EMMAH-LHA University of Bremen, Institute of
33 rue Louis Pasteur Environmental Physics

Avignon, 84000 Otto-Hahn-Allee

FRANCE Bremen, 28359
[email protected] GERMANY
[email protected]

36
Stephan Weise Jürgen Dr. Herrmann
Helmholtz Centre for Environment Research - UFZ GmbH Federal Maritime and Hydrographic Agency
Theodor-Lieser-Strasse 4 Wuestland 2
Halle (Saale), 6120 Hamburg, 22589
GERMANY GERMANY
[email protected] [email protected]

Werner Aeschbach-Hertig Axel Schmidt

Institute of Environmental Physics, Heidelberg University Federal Institute of Hydrology
Im Neuenheimer Feld 229 Kaiserin-Augusta-Anlagen 15
Heidelberg, 69120 Koblenz, 56068
GERMANY GERMANY
[email protected] [email protected]

Petra / Christine Seibel / Stumpp Miklós Süveges

Helmholtz Zentrum München - Institute of Groundwater Ecology HYDROSYS LABOR Ltd.
Ingolstädter Landstrasse 1 Botond u. 72.
Neuherberg, Bavaria, 85764 Budapest, H-1038
GERMANY HUNGARY
[email protected] [email protected]
[email protected]

Laszlo Palcsu Róbert Janovics

Institute of Nuclear Research Institute of Nuclear Research of the
Bem ter 18/c Hungarian Academy of Sciences

Debrecen, 4026 18/c Bem tér

HUNGARY Debrecen, Hajdú Bihar Megye, H-4026
[email protected] HUNGARY
[email protected]

T B JOSEPH Satrio
BHABHA ATOMIC RESEARCH CENTRE PATIR-BATAN
BB04, BARC HOSPITAL, ANUSHAKTINAGAR, Jl. Lebak Bulus Raya 49, Jakarta Selatan
MUMBAI, MAHARASHTRA, 400094 Jakarta, Jakarta, 12440
INDIA INDONESIA
[email protected], [email protected] [email protected]

Saadi Al-Naseri Barbara Stenni

Ministry of Science and Technology Department of Mathematics and Geosciences
Al-Jadiriya, Karrada via E. Weiss 2
Baghdad Trieste, 34128
IRAQ ITALY
[email protected] [email protected]

37
Naoto Takahata Hirochika SUMINO
Atmosphere and Ocean Research Institute, The University of Tokyo
the University of Tokyo 7-3-1 Hongo, Bunkyo-ku
5-1-5, Kashiwanoha Tokyo, 113-0033
Kashiwa-shi, Chiba, 277-8564 JAPAN
JAPAN [email protected]
[email protected]

Ismail Abdeldin Arunas Gudelis

Laboratory & Quality Sector Of Water Center for Physical Sciences and Technology
Authority/Environmental Isotopes Savanoriu Ave. 231
Al-Khansa'a Street, Biader Wadi Esseer, Beside civil defense Vilnius, LT-02300
Amman, 11183 LITHUANIA
JORDAN [email protected]
[email protected]

Rimantas Petrosius Tomislav Andelic

Nature Research Centre, Institute of Geology & Geography Center for Ecotoxicological Research of Montenegro
T. Shevchenkos str. 13 Put Radomira Ivanovica 2
Vilnius, LT-03223 Podgorica, 81000
LITHUANIA MONTENEGRO
[email protected] [email protected]

Hamid MARAH Uwe Morgenstern

National center for Nuclear Energy, Science and Technology GNS Science
(CNESTEN) 1 Fairway Drive
Madinate El Irfane street FAR Lower Hutt, 5010
Rabat, 10001 NEW ZEALAND
MOROCCO [email protected]
[email protected]

MUHAMMAD RAFIQ SHEIKH TADEUSZ KUC

PINSTECH/ PAEC AGH University of Science and Technology
P.O. NILORE al. Mickiewicza 30
ISLAMABAD, CAPITAL, 44500 Krakow, 30-059
PAKISTAN POLAND
[email protected] [email protected]

Paula Carreira Paquete Yoon Yeol Yoon

Campus Tecnológico e Nuclear, IST - UTL Korea Institute of Geoscience and Mineral
Estrada Nacional Nº 10 Resources (KIGAM)

Sacavém, 2686 - 953 Sacavém Gwahang-no 124, Yuseong-gu

PORTUGAL Daejeon, 305-350
[email protected], [email protected] REP. OF KOREA
[email protected]

38
Carmen Varlam Alena Belanova
Institute of Cryogenics and Isotopic Technologies Water Research Institute
Uzinei Street, no.4 Nabr. L. Svobodu 5
Rm. Valcea, 240050 Bratislava, 812 49
ROMANIA SLOVAKIA
[email protected] [email protected]

Jasmina Kožar Logar Michael Butler

Jožef Stefan Institute iThemba LABS
Jamova cesta 39 Empire Road (between Yale Road and Jan Smuts Ave)
Ljubljana, SI-1000 Johannesburg, Gauteng, 2050
SLOVENIA SOUTH AFRICA
[email protected] [email protected]

Díaz Teijeiro Maria Fe BEATRIZ ROMERO DEL HOMBREBUENO

Centro de Estudios de Técnicas Aplicadas (CEDEX) UNIDAD DE RADIACTIVIDAD AMBIENTAL. CIEMAT
C/ Alfonso XII, 3. AVDA COMPLUTENSE 40
Madrid, 28014 MADRID, 28040
SPAIN SPAIN
[email protected], [email protected] [email protected]

Viraj Edirisinghe Quality Assurance Office

Atomic Energy Authority Atomic energy Commission of Syria (AECS)
60/460, Baseline Road Kafar Sousah- 17th Nisan
Orugodawatta, Wellampitiya, Damascus, 6091
SRI LANKA SYRIAN ARAB REPUBLIC
[email protected] [email protected]

Kiattipong Kamdee Kamel ZOUARI

Thailand Institute of Nuclear Technology Laboratory of Radio-Analysis and Environment
National School of Engineering of Sfax
Rangsit-Nakornnayok Route de Soukra
Ongkarak, Nakornnayok, 26120 Sfax, 1173-3038
THAILAND TUNISIA
[email protected] [email protected]

Alime TEMEL DİLAVER N.Nur OZYURT

General Directorate of State Hydraulic Works Tech. Research and Hacettepe University
Quality Control Dept. Isotope Lab. Geological Eng. Dept.
Esenboğa Yolu Çankırı Yol Ayrımı DSİ TAKK Dairesi Başkanlığı Beytepe, Ankara, 6800
İzotop Laboratuvarı Altınova TURKEY
ANKARA, 6291 [email protected]
TURKEY
[email protected]

39
Jill Walker Shari Fanta
RCD Lockinge Illinois State Geological Survey
The Old Stables, East Lockinge 615 E. Peabody Drive
Wantage, Oxfordshire, OX12 8QY Champaign, IL, 61820
UNITED KINGDOM USA
[email protected] [email protected]

Keith Hackley Ate Visser

Isotech Laboratories, Inc. Lawrence Livermore National Laboratory
1308 Parkland Court 7000 East Ave
Champaign, Illinois, 61821-1826 Livermore, CA, 94550-9234
USA USA
[email protected] [email protected]

Bradley K Esser Megan Young

LLNL Environmental Monitoring Radioanlytical Laboratory U.S.Geological Survey
7000 East Avenue 345 Middlefield Road MS434
Livermore, CA, 94551-9234 Menlo Park, CA, 94025
USA USA
[email protected] [email protected]

James Happell D. Kip Solomon

University of Miami Tritium Laboratory University of Utah
4600 Rickenbacker Cswy 1460 East 130 South, Room 383
Miami, Florida, 33149 Salt Lake City, Utah, 84112
USA USA
[email protected] [email protected]

ANH VO THI
INSTITUTE FOR NUCLEAR SCIENCE AND TECHNOLOGY
179 HOANG QUOC VIET
HA NOI, CAU GIAY, 84
VIETNAM
[email protected]

40
APPENDIX F: Revised Data for Laboratories 16 and 24

Initial analysis of the results from Laboratory 16 showed that while the directly counted samples
were reasonable, the enriched samples exhibited a consistent and pronounced bias. Subsequently, a
calculation error was discovered by this laboratory, and the following corrected data was provided
for enriched samples, which improved the z-scores considerably, but appear to have been an over-
correction. The revised data was not used in any of the main report evaluation or Figures.

T20 T21 T22 T23 T24 T25 T26 T27

lab
16 0.09±0.22 0.61±0.24 0.92±0.25 2.37±0.3 3.75±0.34 6.76±0.42 479±17 120±9

Following the distribution of our individual reports, Laboratory 24 inquired as to possible causes for
sample T26 not being given an Acceptable score. Our analysis suggested that the uncertainty margin
for this sample was anomalously low, and this laboratory discovered a mathematical error in the
uncertainty calculation for this sample, revising from ± 2.76 TU to ± 9.03 TU. The revised data was
not used in any of the main report evaluation or Figures.

41