NEACRP-A-1080 Session B 3.6

International Comparison on Measuring Techniques of Tritium

Production Rate for Fusion Neutronics Experiments

--- Summary of Additional Questionnaire and Result for ANL Samples ---

Summarized by

Hiroshi MAEKAWA

Department of Reactor Engineering

Japan Atomic Energy Research Institute

Contributors :

ANL/U.S.A.

AECL-CRNL/Canada

IGA-EPFL/Switzerland

ENEA/Italy

CEN Cadrache/France

Univ. Tokyo/Japan

Osaka Univ./Japan

JAERI/Japan

K. G. Porges W. Workman, J. M. Miller, I. J. Hastings

S. Azam, M. Schaer

A. Moauro, P. L. Carconi

P. Michailli

T. Iguchi, M. Nakazawa

S.,Yoshida, A. Takahashi

H. Maekawa, F. Maekawa, T. Nakamura

The 33rd NEACRP Meeting

Oct. 15 - 19, 1990 Paris, France '

1. Introduction

The idea of the program, "International Comparison on Measuring

Techniques of Tritium Production Rate for Fusion Neutronics Experiments"

by using the 14 MeV neutron source, FNS, was first proposed from the

JAERI representative at the 30th NEACRP meeting. The representative of

Switzerland offered to use in addition the LOTUS facility to strengthen

the program. It was advised at the meeting that a work plan would be

prepared by a cooperation of the experimental groups at JAERI and

IGA/EPFL. The plan thus prepared through the exchange of opinions between

FNS and LOTUS staffs was endorsed at the 31st NEACRP meeting, and

participation in the program was solicited via the representatives of the

committee for research institutes in the member countries. Eight

institutes and Universities from six countries applied to this comparison

program. They are shown in Table 1.

Each participating group prepared Li-containing samples and sent to

the neutron source facilities. The irradiation experiments were performed

on April 27 and May 18, 1989 at FNS and LOTUS, respectively. The

irradiated samples were sent back to the participants to be measured the

tritium production rate. For the normalization of tritium counting system

used by each participating group, JAERI provided a blind sample and

distributed to all participants. Six out of eight groups submitted the

full set of results to JAERI by the middle of September, 1989. JAERI

examined and summarized the results with the assistance of LOTUS group.

At the 32nd NEACRP meeting, the summarized result was presented as an

interim report (NEACRP-A-1021).

From the inter-comparison described in the report, the followings

facts can be summarized:

(1) For the irradiation level of LOTUS, typically several tens Bq of

tritium per samples, agreement of tritium production rates from

different methods turned out to be around 10 % for one standard

deviation without any adjustment. The agreement for several Bq of

tritium production in FNS irradiation went worse to several tens %.

This shows that the degree of agreement depends on the tritium

amounts. The error assigned by the participants are not consistent

with the observation.

( 2 ) The interim result is against the requirement and expectation for a

benchmark experiment, i.e., the measured data are deviated among

the participants over the target accuracy of 5 %. The observation

suggests that there exist unidentified and organization-dependent

systematic errors not directly related to counting statistics nor

difference in tritium standards.

(3) It is most important to identify the causes that bring forth the

discrepancy by comparing and examining the whole procedures of the

participants before mentioning on the true accuracy of the current

measuring techniques. The examination includes the selection of Li-

containing probe, physical and chemical processing of irradiated

samples, preparation of liquid scintillation samples, counting of

foreground and background, efficiency determination and calibration,

tritium standard used, and error assignments on each factor.

According to the "Next Step" proposed to the 32nd NEACRP meeting,

the following two "Actions" have been done.

Action 1

JAERI asks the items relevant to tritium counting techniques to each

participant by a questionnaire, and examines and summarizes them.

Action 2

(1) Three types of diluted tritium water samples, i.e., around 10,000

Bq/g, several 10 Bq/g and several Bq/g, are made from the certified

HTO standard with high accuracy, and are distributed to each

participant.

(2) The participants measure the tritium concentration of the three

samples and report it to JAERI. JAERI examines and summarizes the

results reported.

Section 2 describes the results of questionnaire mentioned above as

Action 1. Following Sections describe the results of Action 2, i.e., the

preparation and measured results of diluted tritium water samples. A

recommendation to "Next Step" is shown in Section 6. Additional results

reported after the 32nd NEACRP meeting are described in Appendix.

2. . Summary of Additional Questionnaire

Additional information based on the questionnaire are summarized in

a Tables 1 and 2. Three groups employed the Dierckx method to obtain a

liquid scintillation sample. The rest five groups adopted quite different

methods. In the case of Dierckx method, almost all tritium produced were transferred to the liquid scintillation sample except the tritium remained

in gas-phase. In the cases of ANL and JAERI, most part of tritium

produced was collected to make the liquid scintillation sample. While AECL

sampled only about one percent of tritium produced.

As a dry method was adopted at ANL, it was not necessary to consider

the tritium in gas-phase. CEN/Cadrache and JAERI considered it but the

other five groups ignored. In the case of JAERI, the quantity of tritium

in gas-phase was estimated separately using samples irradiated near the

target. Because the quantity of tritium in gas-phase for a Li20 pellet

irradiated in the simulated blanket assembly was too small to measure.

a The minimum of signal to background ratio was 1.6 shown in Table 2.

Though this value is not so high, we can still measure the tritium concentration with adequate error under a long measuring time.

From the additional questionnaire, we could not deduce any concrete reason why the tritium production-rate reported was deviated so large.

Table 2 Average counting rates of raw data during tritium counting for the samples irradiated by FNS and LOTUS.

[ Unit : cps / liquid scintillation sample I

ANL AECL/CRNL IGA/EPFL ENEA CEN/Cadr . U. Tokyo Osaka U. JAERI U.S.A. Canada Switzerland Italy France Japan Japan Japan

I Background I 0.08 1.8 0.896 0.15s0.41 0.190 0.485 0.254

FNS outer inner

LOTUS outer inner

*1 Data for liquid scintillation sample for the distillated part (about 80 % of total). *2 counts per min. / decay per min. *3 counts per min. / decay per sec. *4 Average signal to background ratio for the raw data of FNS experiment.

17 1.35 17 1.467 20 7.4 16 8.24

3. Preparation of Diluted Tritium Water Sample

The Analytical Chemistry Laboratory of Argonne National Laboratory

(ANL) prepared a set of three tritium solutions for this NEACRP

International Tritium Comparison, which were distributed on or about

March 5, 1990 from ANL. Each set was prepared by diluting NIS SRM 4926-D

tritium standard with deionized water. The data of tritium standard is

shown in Attachment. All dilutions were done by weight using an

analytical balance with a 1000 g ( 2 0.1 mg) capacity. All weighings were

performed in triplicate and the average used for the final value. The

solutions were prepared in Teflon weighing bottles and were mixed for

over 45 minutes prior to aliquoting. Ten ml aliquots were heat sealed a into engraved glass ampoules and distributed. Twenty ml aliquots of the

deionized water used in the preparation were also sent to the

participants. Aliquots of each prepared tritium solution were asseyed by

liquid scintillation counting to assure the quality of the solutions.

The specific activities (Bq/g) as of 1200 EST, July 25, 1989, are as

follows:

The uncertainties involved with the preparation of the standards

(weighings) were negligible in comparison to the uncertainty assigned to 0

the NIS standard, namely 0.86 percent.

The activity levels of Type "B" and "C" are corresponding to the

expected tritium concentration in the samples irradiated by FNS and LOTUS

at the 1989 experiment of this project, respectively. While the level of

Type "A" was selected to be high enough comparing any error sources.

4. Results for ANL Samples and Discussions

In the beginning of April 1990, JAERI sent a set of return form for

the report of measured tritium concentration in the diluted ANL samples

to the participants. The reports were sent back to JAERI from the

participants except the University of Tokyo by the end of July. Some of

ampoules were broken when they arrived at the University of Tokyo. JAERI

will receive the data soon after the new samples arrive at the University

of Tokyo. The results reported are summarized in Tables 3 and 4. The data

in the first column of Table 3 are the tritium concentrations on April 1,

1990 for three type samples and are corrected for. the decay from the date

shown in Section 3. Because the results reported by all participants are

the tritium concentrations of samples on the same date, i. e., April 1,

1990. The value in parentheses is the ratio of concentration measured to

that assigned by Dr. R. R. Heinrich (the value in the first column).

Though they used quite different tritium standard for calibration,

very good agreement is observed for three types of samples within 1 * 2 %

except some participants. The data of Osaka University deviate systemati-

cally by about 6 %. Their errors given in the Table 3 seem not to include

the error of tritiated water standard which is 4.7 %. The data of

background sample used at ENEA is very low comparing with that of ANL

Blank sample. This fact suggests that the correction for the contribution

of tritium (?) in the dilution water (Blank) used at ANL is necessary for

the data reduction. In the case of IGA/EPFL, it is difficult to

understand why the data of "B" and "C" are 6 % higher than the assigned

values.

From Table 4, the signal to background ratio for the sample "C"

distributes from 6.9 to 55.8 among the participants. This means that the

tritium concentration of a few Bq/g can be measured with high accuracy

even by the liquid scintillation counting system of the lowest signal to

background ratio.

Table 5 summarizes the conditions of measurements for each partici-

pant. The volume of ANL tritium water used and the volume percentage of

cocktail are distributed from 1 ml to 5 in1 and from 50 % to 95 %,

respectively for a liquid scintillation sample. Two groups used the

internal standard method for the calibration and three groups used the

most popular method, i. e., external standard method with a fitting

curve. The total measuring time for a sample was 1 to 9 hours.

Table 3 Measured tritium concentration on April 1, 1990 for ANL samples

[ Unit : Bq/g ]

Sample ID

Background Same 0.132'0.002 Same 0.110~0.007 Same - - - - - - - 0.459+0.030 0.049+0.009 as above as above as above

ANL AECL/CRNL IGA/EPFL ENEA CEN/Cadr . U. Tokyo Osaka U. JAERI U.S.A. Canada Switzerland Italy France Japan Japan Japan

I m I

* Data are assigned by Dr. R. R. Heinrich (ANL) and corrected for decay. ** Ratio to the assigned value.

C 3.81 t 0 . 0 3

Blank

3.893'0.021 3.88 '0.09 4 . 1 '0.13 4.69 20.38 3.198t0.023 ------- 4.172+0.090 3.898t0.057 (1.006) (1.003) (1.059) (1.212) (0.981) (1.078) (1.007)

0.134+0.003 0.142+0.006 ( 0 . 2 3 3 ~ ~ ~ ) 0.505+0.076 0.292t0.009 ------- 0.508+0.033 0.052t0.012

Table 4 Raw data of tritium concentration measurement for ANL samples

[ Unit : cps / vial 1

I Sample ID I ANL AECL/CRNL IGA/EPFL ENEA CEN/Cadr . U. Tokyo Osaka U. JAERI U.S.A. Canada Switzerland Italy France Japan Japan Japan I

I

+ Average ratio of the raw data "C" to background.

Blank

Background

0.075 0.130+0. 009 0.3 0.77350.1 0.327+0.010 ------- 0.195'0.013 0.096'0.007 * 0.085 0.200t0.013 0.099'0.003

Same 0.122t0.003 Same 0.119t0.006 Same - - - - - - - 0.197'0.013 0.090i0.007 as above as above as above 0.194'0.013 0.090i0.011

0.092i0.016

5. Conclusion of Present Comparison for ANL Sample

We can summarize the following conclusion from the international

comparison of tritium counting for the ANL diluted tritium water

samples. :

(1) Using an appropriate tritium water standard with high accuracy

for the calibration of liquid scintillation counting system, we can

expect a good agreement for a blind sample among the participants.

This is also supported by the result of JAERI's blind sample

described in Appendix.

( 2 ) All participants have the ability of measurement for the tritium

concentration level of several Bq/g. Namely, we can measure the

tritium production rate in a simulated fusion blanket under the

irradiation level of 7 x 1015. This total neutron yield can be

obtained by the D-T neutron yield rate of 2 x 1011 n/s and 10 hours

irradiation. This irradiation level is corresponding to the last experiment at FNS.

(3) It is important to obtain a liquid scintillation sample keeping the loss of tritium produced as low as possible without any contamina- tion. Namely, the method and technique are essential for the

chemical treatment of irradiated sample with lithium. If all

participants adopt an appropriate method to extract the tritium

produced from the irradiated sample and a good liquid scintillation

sample, a good agreement is expected among the results of tritium

production rate measured by the participants.

6. Recommendation to Next Step

From the results of present international comparison for the ANL

diluted tritium water samples, JAERI recommends the same type experiments

as previous irradiations. JAERI will continue the organization of this

project. JAERI and IGA-EPFL will provide the same types of neutron fields

as previous ones by FNS and LOTUS facilities, respectively. The recom-

mended procedure and schedule for the next step are as follows:

1990 Nov. The member of NEACRP meeting will confirm the attendant of this

project to the participant(s) of their countries again and

inform the results to the host organization, JAERI by the end 0

of NOV. 1990.

1990 Dec. JAERI will ask and confirm the material and size of sample to

be used by each participant.

1991 Feb. All participants will send their Li-contained samples to JAERI

by the middle of March, 1991.

1991 Apr. Irradiation at FNS/JAERI.

1991 May Irradiation at LOTUS/IGA-EPFL.

a It should be a necessary condition that all participants will use a

suitable method to extract the tritium from an irradiated sample when we

proceed to the next step.

Acknowledgement

Author would like to thank Dr. Y. Kaneko of JAERI for his promotion

of this project. He is also grateful to Drs. R. R. Heinrich and K. G.

Porges of ANL for their preparation and distribution of the diluted

tritium water samples.

Appendix Additional Results Reported After the 32nd NEACRP Meeting

A participant has revised the data of blind sample supplied by

JAERI after the 32nd NEACRP meeting. Table A.l shows results of measured

tritium concentration in the blind sample. The reported data distribute

from 65.63 to 71.0 with average of 67.39 Bq/g, the standard deviation

being 2.8 %. Most of them agree each other within the experimental

errors.

One participant has reported their additional results after the

32nd NEACRP Meeting. The revised tables of measured tritium production

rates are shown in Tables A.2 and A.3 in the samples irradiated at

FNS/JAERI and LOTUS/IGA-EPFL, respectively. The C/E values (ratio of

calculated to experimental values) are presented in Table A.4.

Table A.l Measured tritium concentration in the blind sample

distributed by JAERI (Revised).

Organization Tritium concentration [Bq/g] (error)

Average*3 67.39 i 1.88 (2.8 %)

* Errors are assigned by each participant.

*1 Statistical error only (30). *2 Statistical error only (lo).

*3 Equal weight average.

Table A.2a Measured tritium production rate for outer samples

irradiated at FNS/JAERI (Updated).

Tritium production rate [T-atoms/Li-atom/source]

Organization

# 1 # 2 # 3

A 4.394 * E-29 lost in processing 4.432 * E-29

B not available 1.570'0.090 E-26 3.730t0.153 E-26

(5.7 %) (4.1 %)

C 1.612' ** E-29 1.804' ** E-29 1.808' ** E-29

D not available

F missing of sample 4.00 t0.13 E-29 3.94 t0.13 E-29

(3.3 %) (3.3 %) 0

* Experimental error is not assigned by participant. ** Statistical errors are bigger than 15 %.

Table A.2b Measured tritium production rate for inner samples

irradiated at FNS/JAERI (Updated).

Tritium production rate [T-atoms/Li-atom/source]

Organization

# 4 # 5

D not available

* Experimental error is not assigned by participant. ** Statistical errors are bigger than 15 %.

Table A.3a Measured t r i t i u m product ion r a t e f o r o u t e r samples

i r r a d i a t e d at L(YTUS/IGA-EPFL (Updated).

Tr i t ium product ion r a t e [T-atoms/Li-atorn/source]

Organiza t ion

# 1 # 2 # 3

A 2.378 E-29 2.415 E-29 2.420 E-29

D no t a v a i l a b l e

*Neutron y i e l d of 7.56 x 1016 i s used.

The e r r o r of neutron y i e l d i s not included i n above e r r o r .

Table A.3b Measured tritium production rate for inner samples

irradiated at LOTUS/IGA-EPFL (Updated).

Tritium production rate [T-atoms/Li-atom/source]

Organization

# 4 # 5

A 2.651 E-29 2.608 E-29

D not available

*Neutron yield of 7.56 x 1016 is used.

The error of neutron yield is not included in above error.

Table A.4 Ratio of calculated to experimental values

for the PNS experiment (Updated).

Sample number

Organization

# 1 # 2 # 3 # 4 # 5

SUPPLEMENT TO THE NEACRP REPORT

The International Comparison on Measuring Techniques of

Tritium Production Rate for Fusion Neutronics Experiments

- Summary of Additional Questionnaire and Result for ANL HTO Samples -

JAERI Nakamura, Tomoo

The content of this supplement should have been integrated in the

main paper. However, lack of time in adjustment the styles and some

what different approaches has resulted in a separate treatment.

I) Introduction

As explained in the main text, the international comparison

program was undertaken on the tritium production rate measurement

techniques for fusion blanket neutronics, as one of the benchmark

problems of the NEACRP activity.

The interim results summarized from the data reported

from the participating organization has shown large deviation each

other as shown in the Table 6.6 of NEACRP-1021. This is rather

astonishing because the accuracy in determining the tritium production

rates had been reported to be within lo%, at most 20%, in the papers

of individual experiments so far reported. And it has been generally

accepted that the irradiated Li-containing sample and liquid

scintillation counting method is well-matured technique in the fusion

neutronics experiments.

Whole procedure is divided roughly into three stages: Irradiation of

Li-containing sample pellets, chemical/physical processing of the

sample pellets into the HTO-containing samples and counting of the

samples combined with the calibration by tritium standard. The first

stage has been conducted in a way to assure an irradiation under

identical neutron spectrum and fluence, and there is little reason to

cause such a large difference. The second and third steps must

responsible for the cause(s) of the difference, since they were

conducted separately at individual organizations by different manners

of each own. The second stage has little common factors each other,

and most probable for the cause of the discrepancy. But before seeking

the cause of the difference in it, it is needed to confirm the

accuracy by a use of common standard, in the third stage which is

different each other yet follows the similar principle or procedure.

This is the motivation for the intercomparison by HTO samples provided

by ANL. Samples of three different concentrations were prepared to

examine the dependence on the tritium amount, because the results of

the last experiment had shown such dependence.

It is generally supposed or accepted that the technique to

measure the tritium production rate by the neutron irradiation of

Li-containing samples and liquid scintillation counting

11) Comparison of HTO samples provided by ANL, COM-ANL

The reported results from the participants are tabulated in Table

3 of the main text summarized by Maekawa. In the table it seems that

the error assignments have not been conducted on a same basis, because

the inclusion of error due to tritium standard is not clear in some

cases. In fact the errors from some organizations are apparently

smaller than the error component from tritium standard.

i) Simple Comparison

In the first place, a simple comparison was made on the reported

tritium concentrations for ANL HTO samples. An arithmetic mean and

deviation were calculated assuming an equal weight among the organiza-

tions rather than weighted mean taking error amplitude into considera-

tion.

Table S-1

Min. Max. Average D (in Bdg)

A: 253.6 - 270.9 261.8 t 6.5 (2.5%) 6.6%

B: 48.10 - 52.02 49.8 A 1.5 (3.1%) 7.9%

C: 3.80 - 4.69 4.06 t 0.31 (7.6%) 21.9%

D is the range of the deviation relative to the average defined by

Max. - Min. D =

Average

For the high tritium concentration samples of which errors due to

counting statistics are negligible, the 2.5% uncertainty as a whole is

by no means a good value. The deviation between the largest to the

smallest amounts to near 7%. For low concentration samples the

corresponding values rise to 7.6% and 21. 9%.

This indicates that a more close examination and discussion is

necessary on the the reported values.

ii) Absolute Value Comparison

Each experimenter assigned absolute values using the tritium

standard owned by his organization. Originally the problem of absolute

values rigorously is not an easy matter. I could be admitted, however,

that the tritium concentrations and errors assigned by Dr. Heinrich in

the Table 3 is a highly reliable measure of the absolute values of the

HTO samples distributed. They are based on the well certified value of

NIST standard and on a careful treatment in the dilution.

Of course this does not necessarily mean the COM-ANL values is

the most probable ones. When the reported values are close enough to

this value yet different it is hard to say which is closer to the true

value, since the other standards from different origins should be

treated likewise reliable within the assigned accuracy. The accuracies

of the tritium standards are given below. Abbriviations are used to

0 denote the organizations

Table S-2

COM-ANL ANL AECL IGA/EPFL ENEA CENCAD UOT OU JAERI

In that meaning, the ratios relative to the Heinrich's values in

Table 3 do not necessarily mean the degree of correctness. In fact,

ANL and JAERI use the tritium from the same supplier, NIST, so that it

is natural, that their results are close to unity, if the counting has

been performed correctly.

However, the ratios help relative comparison within the uncer-

tainties in the standards. In other words, an appreciable deviation of

ratio from unity would mean something exist in the reported values.

As seen from Table 3, the results from ANL, AECL, CENCAD and

JAERI give agreement for the three samples of different concentrations

within t 2%, while IGA/EPFL, ENEA and OU give larger deviations in

positive direction except for A value of IGA/EPFL which is close

enough to value.

iii) Relative Comparison among Three Different Concentrations

Comparison of the relative ratios within an organization provides

interesting information. The request for three different concentra- .

tions of ANL HTO samples has intended to separate the issues depending 0

on the tritium content in the counting procedure. The relative

ratios among each organization are summarized in the Table S-3 taking

the highest concentration sample A as the denominator. It is based on

the natural assumption that A value can be determined most exactly

from the S/N viewpoint.

Table S-3

COM-ANL ANL AECL IGA/EPFL ENEA CENCAD OU JAERI

The ratios of COM-ANL are exact reference values, different from

absolute value issue, because the only concerned was the dilution

procedure which has been done with extreme care and high accuracy in weighing. From the table, ANL, AECL, CENCAD and JAERI give good

agreement with the reference values. No concentration dependence was

observed for them. The OU results give a little high values. Much

larger ratios were observed for the CIA of ENEA and B/A and C/A of

EPFL showing a dependence on the concentration

iv) Summary of HTO Comparison

From the comparison of COM-ANL samples, it is concluded that,

1) Simple comparison with equal weighting gives appreciable discre-

pancy beyond the presumed level.

2) Examination on absolute values and relative ratios discussed in

ii) and iii), show that the tritium counting procedures of ANL, AECL,

CENCAD and JAERI have consistency and good accuracy down to several

Bq/g samples.

3) ENEA and OU gave about 5% higher absolute values: this suggests the

calibration procedure including the tritium standards be re-examined.

The OU results might partly be attributed to the relatively large

error of 4.7% for the tritium standards. The IGA/EPFL and ENEA results

for relative comparison also indicate an improvement is needed in the

counting, data processing or else for low concentration samples. In

fact, the treatment of background results is very sensitive to several

Bq/g class samples.

4) For reference, the averages, in the same way as i), of above four

are given below. Reasonable agreement is observed.

Averaged value COM-ANL ( Bdg)

A: 258.18 + 3.57 (1.385%) 257.0 ?: 2.2

B: 48.66 i 0.482 (0.99%) 48.54 + 0.42 C: 3.868 i 0.0472 (1.22%) 3.87 + 0.03

5) Reasonable agreement of four different organizations proves that

the part of the liquid scintillation counting, if carefully

conducted, is reliable.

6) Hence, it is concluded that the main cause of the discrepancy in

the previous Li-containing sample experiment exists, as inferred, in

the stages prior to the preparation of samples for the liquid

scintillation counting.

111) Summary of the Questionnaire

The results of the questionnaire for the participants are

tabulated in the main text. They show a variety of basic information

on the measuring techniques for the tritium production rates. Detailed

comparison of sample, processing, counting, calibration procedure,

data deduction, correction, error assignment etc. on a unified basis

would provide valuable data each other on the final results and

uncertainty assessment for individual measurements, though the

analysis has not been completed yet unfortunately. However, while the

data are useful to discuss the cause of the discrepancy in reasonable

range and to adjust it, the deviation observed in the experiments is

beyond this. In other words, the answers to the questionnaire have not

given a hint to the possibility of such a large deviation

One item to be commented is on the sample of LiAL02. As is well

understood, the accuracy in measuring tritium production rates depends

on the tritium amount produced in the sample pellet, i.e., irradiation 0

fluence. The LiA102 sample gives low specific activity for counting . .

sample because large amounts of water is needed to dissolve them. For

that reason, LiAl02 is applicable only higher fluence irradiation

compared to the other samples used which can produce higher specific

activity counting samples. Hence it is concluded that LiA102 was not

an appropriate material for the neutronics experiment of the level of

LOTUS or FNS fluences.

IV) Interpretation of the irradiation experiment

Since the last NEACRP meeting, new experimental results were

added by CEN/Cadarache. The Table 5-4 is the simplified revision of

the Table 6.6 of the Interim Report. This does not change the general

observation given in the interim report. The 1u deviation for LOTUS

irradiation is around lo%, while that for FNS irradiation amounts more

than 30%, both being inappropriate to examine the deficiency in the

nuclear data or methods.

Table S-4

Case ANL IGA/EPFL CENCAD UOT OU

( x 10-I 3 T Atom/Li Atom)

FNS(outer) 3.06 1.21 5.28 2.76 2.03

FNS(inner) 3.62 1.48 4.43 3.15 2.34

( x 10-l2 T Atom/Li Atom)

LOTUS(outer) 1.82 1.44 1.91 1.68 1.59

LOTUs(inner) 1.99 1.60 2.07 1.81 1.72

JAERI Average

* CENCAD data were converted from the values of Table A2.a.

So far, there have been no positive responses or modification

from the participant organizations on the results reported in the

interim report other than the addition of Cadarache data. Hence I have

to recognize that this is the reality of the current status in the

technique when a comparison is made with minimum limitation in the

whole procedure.

It should be pointed out this fact does not mean that any of the

means applied is not reliable. Somehow a measure has to be given to

the evaluation of the results in order to proceed to the next step,

even if there is no absolute standard, different from the common HTO . comparison For a rough screening, a comparison with calculation would

help judging the degree of the deviation. Of course, a calculated

value does not mean most probable value because it depends on the

nuclear data, calculation method, modeling and absolute neutron yield

normalization, and it is themselves that should be confirmed by the

experiments. Still, a deviation of as much as 50% in a very simple

experimental arrangement would be quite improbable. In that meaning, a

C/E comparison given for FNS irradiation in the Table A.4 in the main

text furnish some index. For the LOTUS results, similar comparison can

be made, but for smaller deviation judgement would be not easy as for

the FNS case. The stabilify is another measure of the reliability of

the results. This is inferred from the dispersion among the plural

number of the samples in the common irradiation and the ratio between

the LOTUS and FNS irradiations. Judging from these two measures, ANL,

UOT and JAERI get a better points down to the level of FNS

irradiation.

V) Summary and the next step

i) If carefully conducted, the accuracy of the liquid scintillation

counting is good enough down to the level of Bq/g from the results

of comparison of HTO samples.

ii) No evidence has been found to explain the cause of the

discrepancy in the irradiation experiment at FNS and LOTUS from

the the summary of the answers to the questionnaire.

iii)The main cause of the discrepancy must be in the stages prior

to the preparation of samples for the liquid scintillation

counting, most possibly in the processing of Li-containing

pellets.

iv) The present average level of tritium measurements seen from this

international comparison, gives an accuracy of about 10% for the

neutron fluence level of producing several tens Bq per sample.

V) When well prepared and cautiously conducted, an accuracy of 5-10%

will be achieved in the irradiation of neutron level producing

several Bq per sample.

vi) If the lesson learned and expertise in this comparison study is

properly reflected, all participants should be qualified for the

demonstration of the accuracy denoted above. I, vii)Upon the agreement of the all.participants, it is advised that

each to re-examine the whole procedures before the preparation of

the liquid scintillation samples including material selection,

chemical/physical processing appropriate for it and data process-

ing including necessary corrections. This is followed by a new

irradiation experiment to determine the ultimate value of the

uncertainty on the Li-sample and liquid scintillation counting

technique.

This supplement is a subjective view as one of the coordinators of

the program. To encourage the discussion on this subject, critical

comments or different opinions are welcomed. 0