Journal of Radioanalytical and Nuclear Chemistry, Vol. 245, No. 1 (2000) 157–161

k0 METHOD

k0 and comparator NAA: Influences and interactions F. De Corte* Laboratory of Analytical Chemistry, Institute for Nuclear Sciences, University of Gent, Proeftuinstraat 86, B-9000 Gent, Belgium

(Received November 22, 1999) A discussion is held on mutual influences and interactions between 0- and (relative) comparator-type NAA. Examples are given concerning: (1) the application of comparator-type NAA in the quality control/quality assurance of the IRMM-530 Al-0.1%Au neutron flux monitor developed for use in 0-NAA, (2) the utilization of the 0-method of calibration as a tool for the quality assurance of comparator-type NAA, (3) the introduction of corrections for detection efficiency and true-coincidence (of primordial importance in 0-NAA) in comparator-type NAA, (4) the development of 0-type standardization for use in prompt gamma neutron activation analysis, and (5) the renewal of insights in the traceability of 0- and comparator-type NAA. k

k

k

k

k

k

Introduction More

than

20

years

after

its

launching,

Comparator-NAA in QC/QA of IRMM-530 Al-Au alloy as a flux monitor for k0-NAA

the

k0-standardization

of

numerous activation analysis laboratories world-wide. The recognition of

k0-NAA as a valuable alternative for

reference material for the

activation analysis: IRMM-530 wires (diam. 0.5 and

NAA

is

nowadays

applied

in

comparator-type (i.e., relative) NAA is based on the stringent goals aimed at

in

the

development

of

its

In

1991,

IRMM-CEC

issued

an

aluminium-gold

k0-standardization of neutron

1 mm) and foils (thickness 0.1 mm), with a certified gold

13 The aim

mass fraction of 0.100±0.002 wt.% (95% CI).

methodology,1 is supported by the user-friendliness of its

was to have a material available that is adding to the

protocols

accuracy

software,

(including e.g.

the

2

“Kayzero”)

availability and

is

of

dedicated

reflected

in

the

k0 Users Workshops” 3 Ljubljana, 1996;4 Geel-Mol, 2001].5

and

determination of

continuing series of “International

of

(Gent, 1992;

characterization

In view of the above, it should not be surprising that this

coexistence

of

k0-

and

comparator-type

NAA,

NAA,

at

traceability

the

same

of

time

the

α,

ratio, and

the

for

the

(i.e.,

the

suitable

spectrum

E1+α

the

analysis

being

the epithermal 1/

standardization

many

of

monitoring of f, the thermal to epithermal neutron flux factor). In the context of the

involving

below)

neutron

confronting many NAA researchers with both types of and

(see

k0-factors and of the k0-standardization

“bare

triple

neutron flux shape

α-monitoring

monitor”-method

(with

according to

198Au

and

intercomparisons, has led to quite a number of mutual

95,97Zr),14

influences and interactions. Some relevant examples of

encountered,

these are given in the present paper, namely: (1) the

the IRMM-530 alloy. In order to solve this problem once

application

quality

for all, in a joint effort of three research groups (at the

control/quality assurance of the IRMM-530 Al-0.1%Au

neutron flux monitor developed for use in k0-NAA;6 (2) the utilization of the k0-method of calibration as a tool 7

NPI, Rez, the KFKI-AEKI, Budapest and the INW,

the introduction of corrections for detection efficiency

accuracy

and true-coincidence (of primordial importance in

comparator-INAA, involving dedicated techniques for

of

comparator-type

NAA

in

the

for the quality assurance of comparator-type NAA;

(3)

k0-

8 (4) the development of

NAA) in comparator-type NAA;

k0-type standardization for use in prompt gamma neutron activation analysis;9 and (5) the renewal of insights in the traceability of k0- and comparator-type NAA.10–12

∨

where

some

accuracy

problems

were

15 doubt was cast on the 0.1% Au content of

∨

Gent), the composition of the IRMM-530 alloyed wires and foils (and also of the R/X 602D Al-0.1124%Au wire)

was

carefully

6

investigated.

determinations

of

Au

Therefore,

were

high-

performed

via

the preparation of samples and gold standards so as to obtain the best possible precision and accuracy. The results

of

International

this

k0

work

were

discussed

at

the

2nd

Users Workshop, Ljubljana, 1996 and,

6 reveal an

as demonstrated in detail in the Proceedings,

excellent agreement between the participating laboratories.

* E-mail: [email protected] 0236–5731/2000/USD 17.00 © 2000 Akadémiai Kiadó, Budapest

Akadémiai Kiadó, Budapest Kluwer Academic Publishers, Dordrecht

F. DE CORTE:

k

0

AND COMPARATOR NAA: INFLUENCES AND INTERACTIONS

Results (in %) of Au determination, via comparator-type INAA, in Al-0.1%Au alloys Material Overall mean Catalogue value ± 95% CI IRMM-530 wire 0.0999 ± 0.00038 19 0.1000 ± 0.0020 IRMM-530 foil 0.1003 ± 0.00036 34 0.1000 ± 0.0020 R/X 602D wire 0.1073 ± 0.00065 24 0.1124* * No uncertainty specified. Table 1.

N

A summary of the Au-contents obtained is shown in

7 as:

the “Analysis of Precision”,

T = ∑ LMd k0 − k0 i

Table 1. It can be concluded that the mean value found

5

for the IRMM-530 materials is in perfect agreement with the certified value; on the other hand, the mean value obtained for the R/X wire is 4.5% lower than the declared value. In addition, excellent homogeneity was proven for both the IRMM and R/X materials (relative SD better than 0.5%) for sample masses down to 5 mg, i.e., less than the recommended minimum (10 mg for IRMM). In general, it could thus be stated that the IRMM-530

alloyed

specifications

given

reference in

the

material

certificate,

15

removing the above mentioned doubt

meets in

this

all way

with respect to

the accuracy of its composition.

Utilization of k0-method in QA of comparator-NAA

with

σ^ k

k0-NAA

7

with

certification

analysis by comparator-type INAA, where the accurate dispensing

of

comparator

standards

of

elements

or

stoichiometric compounds at the microgram level might well cause problems, proved that the

k0-method

of

calibration can be a valuable tool in the quality assurance of analytical traceability. As stated, this is possible because

simultaneous

comparator

irradiation

standards

yields

determination of the corresponding

k0 = k/ε

of

two

an

elemental

experimental

k0-factor, as:

(1)

k is the comparator factor defined according to IRARDI et al.,16 and ε is the peak detection efficiency

where G

(whereby

p

the

epithermal

activation

neglected). This experimental

contribution

is

k0-factor can then be k0-library (with

compared with the one published in the

k0-ratio”

gold as reference) and the “ expected

values

may

therefore

be

of observed and used

in

quality

assurance. Moreover, in the certification analysis for the BCR described by the above authors, with 5 replicates, the agreement between replicate

158

k0-ratios was tested by

degrees

7

of

the

if

no

freedom)

k0’s.

was the case of the

K (mean k0-

Whereas no problems were encountered for ratio = 1.029;

T = 4.14),

a significant excess variability

k0-ratios for Ca was established (mean k0 = 1.097; T = 17.71), which was proven to originate from

of the

differences among the comparator standards.

Corrections for detection efficiency and true-coincidence in comparator-NAA Intrinsically,

the

equation

calculation according to the terms

correcting

for

for

concentration

k0-standardization contains

detection

efficiency

and

true-

1 As a matter of fact, together with

coincidence effects. the

correction

for

the

contribution

of

epithermal

activation, this methodology is of primordial importance in

k0-NAA

and ensures its flexibility with respect to

source-detector geometry. Whereas

this

type

of

correction

is

usually

not

associated with comparator-type NAA, it was recently introduced

p

k0

–

and

should closely follow a

One of the examples given

was demonstrated, the

multielement

4

(2)

determination of K and Ca in BCR Haricots verts 383.

EYDORN and DAMSGAARD, in

dealing

(with

T

OP Q

k0-factor

the expected SD of each

χ2-distribution

Toronto, 1992. Indeed, H paper

s$ k2

significant difference exists between the individual

reverse has been reported as well, as done at NAC-II, their

N

1

weighted mean value. Thus,

If in the above the role of comparator-NAA in the quality assurance of

2

explicitly

in

the

software

IN XILEI

“MULTINAA” developed by L

package

8

et al.

This

VAX/VMS computer program was developed to be applied to comparator (i.e. relative), (modified

according

to

Westcott’s

k0-, monostandard formalism)

and

absolute methods. As to comparator-NAA, the equation for calculation of the concentration (ρ):

F N I GH W ⋅ S ⋅ D ⋅ t ⋅ C JK r = N F I GH W ⋅ r ⋅ S ⋅ D ⋅ t ⋅ C JK p

m

p

m

∗

(3)

F. DE CORTE:

k

0

AND COMPARATOR NAA: INFLUENCES AND INTERACTIONS

is transformed to:

F N I GH W ⋅ S ⋅ D ⋅ t ⋅ C JK r = N F I GH W ⋅ r ⋅ S ⋅ D ⋅ t ⋅ C JK

measured by cold and thermal neutron PGAA relative to titanium at 1381 keV. Finally, as reported in Reference

p

m

p

e ∗p

(4)

∗ e p COI

N

p

is

W is the mass, S is the saturation t D is t t C is [= (1–exp(–λt ))/λt ; t is the

the measured peak area,

factor [= (1–exp(–λ irr); irr is the irradiation time], the decay factor [= exp(–λ d); d is the decay time], counting

factor

measuring time],

εp

m

m

is

clear

that,

as

done

in

“MULTINAA”,

the

true-coincidence makes comparator-NAA more flexible practice,

since

it

allows:

(1)

for

differences

in

shape/size and or main components of standards and sample; and (2) for several different counting positions of the sample coupled to one single counting position of the standard. The above modification of comparator-NAA aiming at the increase of its flexibility with respect to the counting conditions, shows a distinct similarity to former KORTHOVEN,17

E BRUIN and IRARDI’s k-method16 in the same sense.

work of D

who extended

G

analysis

k0

method

is

a

valuable

New insights in the traceability of k0- and comparator-NAA In analytical chemistry traceability is a quite debated topic, and this is not different when the analytical method considered is neutron activation analysis and, obviously,

k0-standardized NAA.

In 1987, triggered by a controversy concerning the acceptability of

k0-NAA

for certification analysis, the

traceability of the method was evaluated at the INW, Gent by unraveling the procedure into its basic steps and

10

parameters.

Detailed attention was paid to the three

main topics: the coirradiated gold monitor, the nuclear data

k0

and

Q0

(resonance integral to thermal cross

section ratio), and the methodology of calibrating the neutron spectrum and the Ge-detector. It was concluded

k0-NAA provides adequate traceability, as well for

As to the gold monitor, it goes without saying that

13

institutes concerning the quantification of results of activation

the

the above mentioned introduction (in 1991)

has been performed at NIST and other cooperating neutron

that

create a complete and consistent nuclear data library.

that

Since the beginning of the 1990s, systematic research

gamma

concluded

standardization tool in PGAA, and it is intended to

routine analysis as for certification work.

k0-type standardization in prompt gamma neutron activation analysis

prompt

was

is the peak detection efficiency and

introduction of corrections for detection efficiency and in

(relative to the

m

COI is the true-coincidence correction factor. It

k0-factors

reasonable agreement, as graphically shown in Fig. 1. It

where * stands for the coirradiated standard, and

the

two laboratories was extended to PGAA as well: prompt measured at NIST and IKI, Budapest, and revealed a

m

t

k0-determination in 1951 keV line of 36Cl) were

9, the NAA basic methodology of

COI∗

(PGAA).

Recently, the evolution of ideas and methodologies was summarized in Reference 9, where the disadvantages of

IRMM-530

Al-Au

reference

material

for

of the

k0-

the

standardization of NAA, was even more ensuring the traceability of the method.

URNARI

In 1998, F

OHEN11

and C

discussed, in the

context of NAA, the application of the concept of traceability

to

the

management

of

nuclear

data.

the “classical” standardization methods of PGAA were discussed: the absolute (parametric) method, yielding insufficient accuracy; the relative method, which is not practical; the single-comparator (monostandard) method, leading to inflexibility. In Reference 9 and in previous

18,19 it was concluded that, in order to

papers from NIST,

eliminate analytical bias arising from the effects of neutron

absorption

and

scattering

in

samples

and

k0-type standardization is the method of choice. At the 1st International k0 Users Workshop, Gent, 1992,18 the model of k0-PGAA (emphasizing the standards,

use

of

cold

neutrons)

was

outlined

and

the

first

experimental results showed the constancy of the Br/K

k0-factors with sample mass and 1H content. At 2nd International k0 Users Workshop, Ljubljana, 19

relative the

1996,

attention was paid to the determination of the

detection efficiencies for both cold and thermal PGAA instruments as a function of energy up to 8 MeV, and the first

ever

k0-factors

for

14

selected

elements

were

Comparison of 0 values vs. the 1951 keV line of 36Cl, measured at IKI and NIST (±2s error bars are shown)

Fig. 1.

k

159

F. DE CORTE:

k

0

AND COMPARATOR NAA: INFLUENCES AND INTERACTIONS

In any method of NAA, they identified steps that imply

As to the above statements, three remarks can be

the use of some kind of data from literature: these

made here: (1) what has an “average” or “faithful” user

include time-dependent corrections (involving half-lives)

to do with the (un)traceability of a method?, (2) when it

and

comes to the methodology (including neutron spectrum

corrections

for

geometrical

configuration

of

samples, comparators and standards, such as neutron

characterization), what makes

self-shielding

“traceable”

parameters

(involving

etc.),

cross

attenuation

sections,

of

measured

resonance radiation

absolute

k0-NAA different from the

standardization?

[and

comes to nuclear data, why is the compound

when

it

k0,Au-factor

(involving attenuation coefficients) and summing effects

(gold

(involving

photon

intensity standard, and the analyses being performed

considering

the

emission

transition

probabilities).

from

When

comparator-type

being

an

ultimate

cross

section

and

gamma-

to

with the certified IRMM-530 Al-Au alloy) less traceable

parametric NAA (in other words, using less or no

than the cross sections, gamma-intensities, etc. for use in

comparators, resulting in the need for using more data

the absolute method?], and (3) why is not, when it

from tables), the following set of data is involved: –

k0-method: k0-constants; resonance to thermal

in the

–

concerns the

k0-method, the proof of the pudding in the

eating? (in the context of which reference can be made to

cross section ratios;

results submitted by various

in the traditional parametric method: relative atomic

certification rounds).

masses; isotopic abundances; emission probabilities of

the

radiations;

cross

sections;

resonance

integrals, Avogadro’s number.

11

opinion

Despite the above questions, it is more important, however, to conclude with the observation that the traceability of

In view of the above, and following the authors’ that traceability of data in practice can be

k0-NAA labs in recent CEC

k0-NAA

is at times subject to fire, with

arguments that are not always straightforward. In the author’s view, it is advisable that this topic be dealt with

interpreted as the possibility of renormalization, there is

by specialists at the forthcoming Third International

no doubt that in this respect

Users Workshop, Geel-Mol, 2001.

k0-NAA (with its welldocumented nuclear data library)20 is perfectly traceable, whereas

this

(absolutely

is

not

at

all

standardized)

the

case

NAA,

for

as

for

A

recent

review

12

paper

on

“Traceability

and

deals (among other analytical

k0-NAA.

*

instance

methodologies) with neutron activation analysis and, in more detail, with

k0

parametric

demonstrated in Reference 21. Analytical Chemistry”

5

Interesting to note is the

introductory reasoning that, in the sense of the Comité

The author is indebted to the Fund for Scientific ResearchFlanders for financial support and to Prof. Em. J. HOSTE for the many enlightening discussions concerning traceability. Special thanks are due to Stijn VAN LIERDE for his highly appreciated help in various ways.

Consultatif pour la Quantité de Matière (CCQM), the parametric

(absolute)

standardization

is

comparator (relative) method. Then, when it comes to the question which one of the methods of standardization is to be preferred for accurate measurement of the amount of the substance (from a strictly metrological point of view: the absolute method!), the strategy is changed to arrive at the remark: “the proof of the pudding is in the eating”. Thus, the reasoning is clearly ending up with a paradox: the most traceable method according to the CCQM definition is apparently not the

12

preferred one in any real practice. Next, the author discusses in detail the danger of using the intermediate

between

relative

k0-method (an

and

absolute

standardization) in certification analysis, for which the reasons given can be summarized as follows: the hidden complexity of the method, not always completely clear (“non-traceable”) to the average user; the tendency for general diffusion of potential systematic errors when faithful users apply the method as a black box; and the need for calibrating the neutron spectrum, a

well-

understood but complex matter, based on conventions with the use of non-fundamental (physical) constants.

160

References

potentially

traceable, whereas this is presently not the case for the

1. F. DE CORTE, A. SIMONITS, A. DE WISPELAERE, J. HOSTE, J. Radioanal. Nucl. Chem., 113 (1987) 145. 2. KAYZERO/SOLCOI User’s Manual, Version 4, DSM Research, December 1996. 3. F. DE CORTE (Ed.), Proc. Intern. 0 Users Workshop, Astene, 1992, Inst. Nucl. Sci., Gent, 1992. 4. B. SMODIŠ (Ed.), Proc. 2nd Intern. 0 Users Workshop, Ljubljana, 1996, J. Stefan Inst., 1997. 5. P. ROBOUCH, S. POMMÈ, private communication. 6. J. KUCERA, A. SIMONITS, F. DE CORTE, F. BELLEMANS, Proc. Intern. 0 Users Workshop, Ljubljana, 1996, J. Stefan Inst., 1997, p. 15. 7. K. HEYDORN, E. DAMSGAARD, J. Radioanal. Nucl. Chem., 179 (1994) 87. 8. X. LIN, F. BAUMGARTNER, X. LI, J. Radioanal. Nucl. Chem., 215 (1997) 179. 9. G. L. MOLNAR, Zs. REVAY, R. L. PAUL, R. M. LINDSTROM, J. Radioanal. Nucl. Chem., 234 (1998) 21. 10. F. DE CORTE, J. Trace Micropr. Techn., 5 (1987) 115. 11. J. C. FURNARI, I. M. COHEN, Appl. Radiation Isotopes, 12 (1998) 1523. 12. F. ADAMS, Accred. Qual. Assur., 3 (1998) 308. 13. C. INGELBRECHT, F. PEETERMANS, F. DE CORTE, A. DE WISPELAERE, C. VANDECASTEELE, E. COURTIJN, P. D’HONDT, Nucl. Instr. Meth., A303 (1991) 119. k

k

∨

k

F. DE CORTE:

k

0

AND COMPARATOR NAA: INFLUENCES AND INTERACTIONS

14. F. DE CORTE, K. SORDO-EL HAMMAMI, L. MOENS, A. SIMONITS, A. DE WISPELAERE, J. HOSTE, J. Radioanal. Chem., 62 (1981) 209. 15. N. ETXEBARRIA, P. ROBOUCH, J. PAUWELS, S. POMMÉ, F. HARDEMAN, Proc. 2nd Intern. 0 Users Workshop, Ljubljana, 1996, J. Stefan Inst., 1997, p. 137. 16. F. GIRARDI, G. GUZZI, J. PAULY, Anal. Chem., 37 (1965) 1085. 17. M. DE BRUIN, P. J. M. KORTHOVEN, Anal. Chem., 44 (1972) 2382. k

18. R. M. LINDSTROM, R. F. FLEMING, R. L. PAUL, E. A. MACKEY, Proc. Intern. 0 Users Workshop, Astene, 1992, Inst. Nucl. Sci., Gent, 1992, p. 121. 19. R. L. PAUL, R. M. LINDSTROM, Proc. 2nd Intern. 0 Users Workshop, Ljubljana, 1996, J. Stefan Inst., 1997, p. 54. 20. F. DE CORTE, A. SIMONITS, A. DE WISPELAERE, A. ELEK, J. Radioanal. Nucl Chem., 133 (1989) 3. 21. F. DE CORTE, A. SIMONITS, in: Proc. Nuclear Data for Science and Technology, 1988 Mito, S. IGARASI (Ed.), Saikon Publ. Co., 1988, p. 57. k

k

161