- Email: [email protected]
Contribution of the Community Bureau of Reference (CBR) of the Commission of European Communities (CEC) to quality improvement and certification of reference materials for PCDD and PCDF analysis
Chemosphere, Vol.20, Nos.lO-12, Printed in Great Britain
pp 1291-1298,
1990
0 0 4 5 - 6 5 3 5 / 9 0 $3.00 + .00 Pergamon Press plc
CONTRIBUTION OF THE COMMUNITY BUREAU OF REFERENCE (CBR) OF THE COMMISSION OF EUROPEAN COMMUNITIES (CEC) TO QUALITY IMPROVEMENTAND CERTIFICATIONOF REFERENCEMATERIALSFOR PCDD AND PCDFANALYSIS. T.A. Rymen*(a), B. Griepink (b) and S. Fachetti (c). Community Bureau of Reference, Commission of the European Communities, Brussels, Belgium.
(a) SCK/CEN, Boeretang 200, Mol, Belgium. (b) CEC, BCR, Wetstraat 200, Brussels, Belgium. (c) CEC, JRC, Ispra, Italy. Interlaboratory comparisons on PCDD- and PCDF-analysis resulted in an appreciable improvement of analytical quality. As a result c e r t i f i c a t i o n of a f l y ash for both PCDD and PCDF became feasible. Introduction After the positive i d e n t i f i c a t i o n of PCDD and PCDF in the most different types of samples (1) (2), the importance of the dioxin problem in relation to public health became evident (3) and so did the need for more accurate quantitative measurements. Following this need, native and labelled PCDD- and PCDF-congeners were made available as synthetic standards.
Actually, this a v a i l a b i l i t y , has reached a point of general satisfaction.
Most of the laboratories involved in PCDD- and PCDF-analysis are now in the position to perform adequate instrument
calibrations and to control
their respective methods for
sample clean-up by spiking.
Nevertheless, an effective control of trace analysis (e.g.
extraction efficiency), is often impossible due to the lack of appropriate and c e r t i f i e d reference materials.
In order to f u l f i l
this need, the BCR has set up a program aiming at
the c e r t i f i c a t i o n of the PCDD- and PCDF-content in different types of samples.
Design of the c e r t i f i c a t i o n program Prior to c e r t i f i c a t i o n , interested laboratories carried out various collaborative analytical
studies
in order
to detect and consequently eliminate systematic
errors.
Thirtheen laboratories, from six European countries, voluntarily participated. I t was agreed to focus a l l efforts upon the determination of the so-called " d i r t y dozen" (tetra-,
penta- and hexa-congeners with the 2,3,7,8-chlorination configuration) and to
select an incinerator f l y ash as a test case. combined gas chromatography - mass spectrometry.
All analyses had to be performed using At the other hand, i t was found to be
absolutely imperative that each participant would use i t s own analytical methodology, thus covering a broad range of experimental conditions regarding sample preparation (extraction, clean-up), chromatographic separation (stationary phases), detection parameters (high and low resolution mass spectrometric detection), calibration
standards) and data evaluation
(automatic versus manual integration). Taking into account the r e l a t i v e l y large discrepancies
n results for interlaboratory
comparisons in the f i e l d of PCDD- and PCDF-analysis, which are often cited in l i t e r a t u r e , 1291
1292
(4) (5) (6), i t was decided that the actual c e r t i f i c a t i o n should be preceeded by a series of preliminary exercises. All i n i t i a l tests were restricted to the analysis of PCDD. They were arranged in such a way that they allowed the systematic detection of analytical problems in each stage of the procedures applied by the respective laboratories (Figure
I). Figure I : overall sheme of collaborative studies
J SYNTHETICMIXTURE 14 EASY RESOLUTION r
[ •
I
REPEAT I NOT OK
I
ADJUST
SYNT~TIC MIXTURE I
I REPEAT ADJUST I
, CRITICALPAIRS
I.OT oK J
I
~ I
CLEANED-UP ~ EXTRACT
I
REPEAT
~ / ~ ' ~ N O T O K I
REPEAT ADJUST
i
~
ADJUST I
bq CRUDEEXTRACT I
rOT OK I IFLY AS" li"
I
REPEAT
l.oT OK
i
ADJUST
I PARTICJF'ATIONIN
, [ERTIF(ATION
The implementation of the corrective measures, evolving from these preliminary investigations, into the respective methodologies, always remained the responsability of each individual participant.
1293
Results and discussion Synthetic samples For reasons of e f f i c i e n c y only, tests 1 and 2 were combined into one and the same exercise.
The samples f o r test I were synthetic mixtures of some of the " d i r t y f i v e "
PCDD-congeners, dissolved in separation problems.
In
the
n-nonane.
The composition was such that
samples f o r
test
2 the
there were no
chromatographic complexity was
increased by mixing a l l " d i r t y f i v e " PCDD and by adding 1,2,3,4-T4CDD, 1,2,3,7-T4CDD, 1,2,3,8-T4CDD and 1,2,3,4,7-P5CDD. The concentration levels
for
the unknowns ranged from 30 to
compounds were used as internal standards.
50 ng/ml.
C13-1abeled
Their i d e n t i t y and concentration, which was in
the same range as that of the unknowns, were communicated to each p a r t i c i p a n t . The samples were analyzed by each of the p a r t i c i p a t i n g l a b o r a t o r i e s , using t h e i r instrumentation and r e g u l a r l y applied methodology f o r
identification,
own
i n t e g r a t i o n and
calibration. A simple s t a t i s t i c a l evaluation of the results led to the set of data summarized in table I .
Table
I : means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence l i m i t f or the samples of test i and 2
Test Mean
CV
2,3,7,8
44
12
. . . .
1,2,3,7,8
29
14
. . . .
1,2,3,6,7,8
. . . .
PCDD-congener
Mean
CV
34
1,2,3,7,8,9 1,2,3,4,7,8
The
resulting
28
relative
11
. . . .
standard deviations
9
Mean
CV
Mean
CV
31
18
31
16
46
12
27
10
49
9
42
8
42
10
34
9
33
10
. . . .
were concidered rather
promising.
Nevertheless, when aiming at a f i n a l c e r t i f i c a t i o n with the best possible precision and taking into account that the uncertainty w i l l increase when going from pure solutions to real world samples, i t
was decided to perform a more detailed investigation of all
experimental data. I t became clear that some laboratories showed large systematic deviations from the true value.
Others were much closer to the true value but the deviations s t i l l
systematic.
remained
The fact that these problems occurred for a l l mixtures, immediatelyexcluded
1294
separation errors as a possible cause. Furthermore, there were no systematic differences between laboratories using high or low resolution mass spectrometric detection, nor between those using different types of capillary columns. A detailed investigation of the integration and calibration procedures applied by the respective participants revealed that they formed the main sources of error and therefore that a more stringent control of both of these quantification parameters was paramount, Following these findings, i t was decided to repeat test 2 on a new synthetic mixture, for which special care should be taken regarding signal integration.
Additionally, the
BCR made available a primary standard solution against which the working standards, used in the respective laboratories, couldbe referenced. table
The resulting data are summarized in
II. Table
II
: means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence l i m i t , obtained upon r e p e t i t i o n of test 2.
PCDD-Congener
It
Mean
CV
2,3,7,8
49
2.5
1,2,3,7,8
50
4.0
1,2,3,6,7,8
51
7.4
1,2,3,7,8,9
49
4.0
1,2,3,4,7,8
51
7.5
can be concluded from table
II
that indeed a better control of integration and
calibration was necessary to obtain acceptable results. Cleaned-up f l y ash extract Following this conclusion, the quantitative determination of PCDD in a cleaned-up incinerator f l y ash extract was attempted.
A batch of f l y ash was treated with HCl, washed
with water, dried and extracted in a soxhlet with toluene.
Sample clean-up was performed
by atmospheric pressure liquid chromatography, using s i l i c a g e l , modified silicagel (H2SO4, NaOH and AgNO3) and basic alumina. The resulting sample was then spiked with the appropriate C13-1abeled internal standards and homogenized thoroughly. The data submitted are summarized in t a b l e I I I . The obviously erroneous results obtained for 2,3,7,8-TCDD and 1,2,3,7,8,9-H6CDD were from laboratories using non polar capillary columns, which apparently are not suited to obtain the chromatographic resolution required for the determination.
At the other hand
1295
there
were no differences
between columns of
appropriate
polarity,
as obtained from
d i f f e r e n t manufacturers.
Table I I I :
means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence i n t e r v a l , obtained upon analysis of a cleaned-up f l y ash e x t r a c t for PCDD.
A l l data i n c l . Congener
Mean
CV
2,3,7,8
15
60
1,2,3,7,8
92
8
1,2,3,6,7,8
213
1,2,3,7,8,9
193
1,2,3,4,7,8
81
6.8 18 6.6
Outliers excl. ( ' ) Mean
CV
n (*)
I0.4
7.7 (b)
11
91
2.3 (a)
11
210
4.3 (a)
10
169
6.6 (a)
11
80
4.9 (a)
10
( ' ) A l l data out of the four sigma confidence i n t e r v a l were considered to be outliers. (a) : single step e l i m i n a t i o n (b) : two step e l i m i n a t i o n , which was j u s t i f i e d for the 2,3,7,8-TCDD data, because of the presence of one extremely deviation r e s u l t (*) n : the number of data a f t e r e l i m i n a t i o n of o u t l i e r s .
Crude f l y ash e x t r a c t The next exercise involved clean-up of a crude f l y ash e x t r a c t , as well as analysis for both PCDD and PCDF. The sample had been spiked previously with the necessary C13-1abeled internal standards. In addition to the already e x i s t i n g primary standard for PCDD's ( d i r t y f i v e ) , an analogous reference material for PCDF's ( d i r t y seven) was made available by the BCR. The sample clean-up procedures that were applied by the respective p a r t i c i p a n t s can be grouped as follows : 1. (7) - mixed modified s i l i c a (H2SO4, NaOH and AgNO3) -
alumina
2. (8) - mixed modified s i l i c a - carbon (Amoco PX21)/glass f i b r e s (Whatman GF/D) 3.
- mixed modified s i l i c a
1296
-
alumina
- carbon (Amoco PX21) + C e l i t e 545 4. (9) - alumina B, super I (Woelm), macro column The a n a l y t i c a l results are summarized in table IV. The results obtained for the PCDD's, when compared with those found in the previous t e s t (Table l l l ) ,
indicate that the c o n t r i b u t i o n of sample clean-up to the i n t e r l a b o r a t o r y
variance is small (maximum 4 %).
The overall c o e f f i c i e n t s of v a r i a t i o n are well w i t h i n
acceptable l i m i t s from the point of view of f i n a l c e r t i f i c a t i o n .
Table IV: means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence i n t e r v a l obtained upon analysis of a crude f l y ash extract for PCDD and PCDF.
PCDD ( ' ) (a) Congener
Mean
CV
n (*)
PCDF ( ' ) (a) Mean
CV
n (*)
2,3,7,8
37
5.4
Ii
15
11
9
1,2,3,7,8
29
5.6
10
44
11
10
1,2,3,6,7,8
143
5.4
11
1,2,3,7,8,9
103
8.0
I0
17
36
8.9
11
89
9.0
9
146
8.3
I0
2,3,4,7,8
1,2,3,4,7,8 (+) 2,3,4,6,7,8
47
7.3
9
90
8.3
11
28
12
( ' ) (a) and (*) : cfr. t a b l e I I I . (+) : 1,2,3,4,7,8 + 1,2,3,4,7,9 for the furanes
The agreement of the results for PCDF's, clearly reflects the experience gained in the preceeding exercises.
Nevertheless,
the coefficients of variation for some PCDF
congeners are d e f i n i t e l y larger than that for PCDD. A detailed investigation of a l l experimental data led to the following conclusions : the partial separation of 2,3,7,8-T4CDF and 2,3,4,8-T4CDF, which is possible on polar columns, varies considerably from one column to another, even when comparing new columns of the same type obtained from the same manufacturer. The same effect was found for the separation between 1,2,3,7,8,9-H6CDF and an unknown accompanying impurity.
In this case, about half of the participants obtained a single
peak where the others found a reasonably well separated doublet.
1297
The separation of 1,2,3,7,8-P5CDF from 1,2,3,4,8-P5CDF is impossible on the a c u t a l l y
-
available polar columns but can e a s i l y be performed on a nonpolar column.
The problem for
1,2,3,7,8,9-H6CDF, described in the previous paragraph, can also be circumvented by using a non polar column. As a consequence, i t still
be improved,
*
polar
if
was agreed that the q u a l i t y of the results for these PCDF's could
columns could be tested,
before use, as to t h e i r
capability
to separate
2,3,7,8-T4CDF from 2,3,4,8-T4CDF, *
and i f
and 1,2,3,7,8,9-H6CDF would be determined using a suitable
1,2,3,7,8-P5CDF
nonpolar column. Conclusions and future planning The systematic preliminary t e s t s , set up in t h i s BCR-programme, and the dedicated care and commitment of the p a r t i c i p a t i n g analysts made i t possible to obtain good q u a l i t y data for the analysis of PCDD and PCDF in f l y ash extracts. Fly ash e x t r a c t i o n i t s e l f
has not yet been attempted.
However, t h i s step has been studied extensively in the past and the results were published in great d e t a i l .
Therefore, i t is not expected to be a c r i t i c a l
step in view of the f i n a l
c e r t i f i c a t i o n exercise, which is planned for the very near future. Additionally,
it
is envisaged to produce, as a surplus, a cleaned-up f l y
ash e x t r a c t
c e r t i f i e d for PCDD and PCDF. This reference material is proposed as a response to the problem of preliminary column testing,
which was proved
to
be an absolute
condition
especially
in
the
field
of
PCDF-analysis. F i n a l l y , the experience gathered during the exercises described in t h i s paper may be the basis f o r the c e r t i f i c a t i o n of a number of other important matrices in the future. Acknowledgements.
The authors wish to thank a l l members of the Dioxin Working Group for
t h e i r commitment and f r u i t f u l l P a r t i c i p a t i n g laboratories :
cooperation. - Cherchar
(France)
- Ist.
(Italy)
-
Inq. Atmosferico
I s t . Mario Negri
(Italy)
- University of Amsterdam
(Netherlands)
- TNO, D e l f t
(Netherlands)
-
Duphar, Weesp
(Netherlands)
- Shering Agrochem. Ltd.
(England)
- SCK/CEN
(Belgium)
-
-
-
Bayer AG, Leverkusen
(GDR)
Centre Nat. de Recherche
(France)
BASF AG
(GDR)
- Ecole Polytechnique, Paris
(France)
- Warren Spring Lab.
(England)
1298
References (1) (2) (3) (4)
(5) (6) (7) (8) (9)
Bumb, R.R. et a l . , Science 210, 385 (1980) Hutzinger, 0., Fiedler, H., VDI Berichte 634, 17 (1987) Sander, H.P., VDI berichte 634, 37 (1987) McMillin, C.R. et a l . , Chlorinated Dioxins and Dibenzofuranes in the Total Environment, Choundhary G., Keith L. and Rappe C. eds, Butterworth Publ., Ann Arbor Science, vol I, 165 (1983) Gross, M.L. et a l . , Anal. Chem., 53, 1902 (1981) Brumley, W.C. et a l . , J. Agric. Food Chem., 29, 1040 (1981) Kook, R., Lustenhouwer, J., Olie, K., Hutzinger, 0., Anal. Chem., 53, 461 (1981) Smith, L.M., Anal. Chem., 53, 2152 (1981) Hagenmaier, H. et a l . , VDI Berichte 634, 61 (1987)
pp 1291-1298,
1990
0 0 4 5 - 6 5 3 5 / 9 0 $3.00 + .00 Pergamon Press plc
CONTRIBUTION OF THE COMMUNITY BUREAU OF REFERENCE (CBR) OF THE COMMISSION OF EUROPEAN COMMUNITIES (CEC) TO QUALITY IMPROVEMENTAND CERTIFICATIONOF REFERENCEMATERIALSFOR PCDD AND PCDFANALYSIS. T.A. Rymen*(a), B. Griepink (b) and S. Fachetti (c). Community Bureau of Reference, Commission of the European Communities, Brussels, Belgium.
(a) SCK/CEN, Boeretang 200, Mol, Belgium. (b) CEC, BCR, Wetstraat 200, Brussels, Belgium. (c) CEC, JRC, Ispra, Italy. Interlaboratory comparisons on PCDD- and PCDF-analysis resulted in an appreciable improvement of analytical quality. As a result c e r t i f i c a t i o n of a f l y ash for both PCDD and PCDF became feasible. Introduction After the positive i d e n t i f i c a t i o n of PCDD and PCDF in the most different types of samples (1) (2), the importance of the dioxin problem in relation to public health became evident (3) and so did the need for more accurate quantitative measurements. Following this need, native and labelled PCDD- and PCDF-congeners were made available as synthetic standards.
Actually, this a v a i l a b i l i t y , has reached a point of general satisfaction.
Most of the laboratories involved in PCDD- and PCDF-analysis are now in the position to perform adequate instrument
calibrations and to control
their respective methods for
sample clean-up by spiking.
Nevertheless, an effective control of trace analysis (e.g.
extraction efficiency), is often impossible due to the lack of appropriate and c e r t i f i e d reference materials.
In order to f u l f i l
this need, the BCR has set up a program aiming at
the c e r t i f i c a t i o n of the PCDD- and PCDF-content in different types of samples.
Design of the c e r t i f i c a t i o n program Prior to c e r t i f i c a t i o n , interested laboratories carried out various collaborative analytical
studies
in order
to detect and consequently eliminate systematic
errors.
Thirtheen laboratories, from six European countries, voluntarily participated. I t was agreed to focus a l l efforts upon the determination of the so-called " d i r t y dozen" (tetra-,
penta- and hexa-congeners with the 2,3,7,8-chlorination configuration) and to
select an incinerator f l y ash as a test case. combined gas chromatography - mass spectrometry.
All analyses had to be performed using At the other hand, i t was found to be
absolutely imperative that each participant would use i t s own analytical methodology, thus covering a broad range of experimental conditions regarding sample preparation (extraction, clean-up), chromatographic separation (stationary phases), detection parameters (high and low resolution mass spectrometric detection), calibration
standards) and data evaluation
(automatic versus manual integration). Taking into account the r e l a t i v e l y large discrepancies
n results for interlaboratory
comparisons in the f i e l d of PCDD- and PCDF-analysis, which are often cited in l i t e r a t u r e , 1291
1292
(4) (5) (6), i t was decided that the actual c e r t i f i c a t i o n should be preceeded by a series of preliminary exercises. All i n i t i a l tests were restricted to the analysis of PCDD. They were arranged in such a way that they allowed the systematic detection of analytical problems in each stage of the procedures applied by the respective laboratories (Figure
I). Figure I : overall sheme of collaborative studies
J SYNTHETICMIXTURE 14 EASY RESOLUTION r
[ •
I
REPEAT I NOT OK
I
ADJUST
SYNT~TIC MIXTURE I
I REPEAT ADJUST I
, CRITICALPAIRS
I.OT oK J
I
~ I
CLEANED-UP ~ EXTRACT
I
REPEAT
~ / ~ ' ~ N O T O K I
REPEAT ADJUST
i
~
ADJUST I
bq CRUDEEXTRACT I
rOT OK I IFLY AS" li"
I
REPEAT
l.oT OK
i
ADJUST
I PARTICJF'ATIONIN
, [ERTIF(ATION
The implementation of the corrective measures, evolving from these preliminary investigations, into the respective methodologies, always remained the responsability of each individual participant.
1293
Results and discussion Synthetic samples For reasons of e f f i c i e n c y only, tests 1 and 2 were combined into one and the same exercise.
The samples f o r test I were synthetic mixtures of some of the " d i r t y f i v e "
PCDD-congeners, dissolved in separation problems.
In
the
n-nonane.
The composition was such that
samples f o r
test
2 the
there were no
chromatographic complexity was
increased by mixing a l l " d i r t y f i v e " PCDD and by adding 1,2,3,4-T4CDD, 1,2,3,7-T4CDD, 1,2,3,8-T4CDD and 1,2,3,4,7-P5CDD. The concentration levels
for
the unknowns ranged from 30 to
compounds were used as internal standards.
50 ng/ml.
C13-1abeled
Their i d e n t i t y and concentration, which was in
the same range as that of the unknowns, were communicated to each p a r t i c i p a n t . The samples were analyzed by each of the p a r t i c i p a t i n g l a b o r a t o r i e s , using t h e i r instrumentation and r e g u l a r l y applied methodology f o r
identification,
own
i n t e g r a t i o n and
calibration. A simple s t a t i s t i c a l evaluation of the results led to the set of data summarized in table I .
Table
I : means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence l i m i t f or the samples of test i and 2
Test Mean
CV
2,3,7,8
44
12
. . . .
1,2,3,7,8
29
14
. . . .
1,2,3,6,7,8
. . . .
PCDD-congener
Mean
CV
34
1,2,3,7,8,9 1,2,3,4,7,8
The
resulting
28
relative
11
. . . .
standard deviations
9
Mean
CV
Mean
CV
31
18
31
16
46
12
27
10
49
9
42
8
42
10
34
9
33
10
. . . .
were concidered rather
promising.
Nevertheless, when aiming at a f i n a l c e r t i f i c a t i o n with the best possible precision and taking into account that the uncertainty w i l l increase when going from pure solutions to real world samples, i t
was decided to perform a more detailed investigation of all
experimental data. I t became clear that some laboratories showed large systematic deviations from the true value.
Others were much closer to the true value but the deviations s t i l l
systematic.
remained
The fact that these problems occurred for a l l mixtures, immediatelyexcluded
1294
separation errors as a possible cause. Furthermore, there were no systematic differences between laboratories using high or low resolution mass spectrometric detection, nor between those using different types of capillary columns. A detailed investigation of the integration and calibration procedures applied by the respective participants revealed that they formed the main sources of error and therefore that a more stringent control of both of these quantification parameters was paramount, Following these findings, i t was decided to repeat test 2 on a new synthetic mixture, for which special care should be taken regarding signal integration.
Additionally, the
BCR made available a primary standard solution against which the working standards, used in the respective laboratories, couldbe referenced. table
The resulting data are summarized in
II. Table
II
: means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence l i m i t , obtained upon r e p e t i t i o n of test 2.
PCDD-Congener
It
Mean
CV
2,3,7,8
49
2.5
1,2,3,7,8
50
4.0
1,2,3,6,7,8
51
7.4
1,2,3,7,8,9
49
4.0
1,2,3,4,7,8
51
7.5
can be concluded from table
II
that indeed a better control of integration and
calibration was necessary to obtain acceptable results. Cleaned-up f l y ash extract Following this conclusion, the quantitative determination of PCDD in a cleaned-up incinerator f l y ash extract was attempted.
A batch of f l y ash was treated with HCl, washed
with water, dried and extracted in a soxhlet with toluene.
Sample clean-up was performed
by atmospheric pressure liquid chromatography, using s i l i c a g e l , modified silicagel (H2SO4, NaOH and AgNO3) and basic alumina. The resulting sample was then spiked with the appropriate C13-1abeled internal standards and homogenized thoroughly. The data submitted are summarized in t a b l e I I I . The obviously erroneous results obtained for 2,3,7,8-TCDD and 1,2,3,7,8,9-H6CDD were from laboratories using non polar capillary columns, which apparently are not suited to obtain the chromatographic resolution required for the determination.
At the other hand
1295
there
were no differences
between columns of
appropriate
polarity,
as obtained from
d i f f e r e n t manufacturers.
Table I I I :
means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence i n t e r v a l , obtained upon analysis of a cleaned-up f l y ash e x t r a c t for PCDD.
A l l data i n c l . Congener
Mean
CV
2,3,7,8
15
60
1,2,3,7,8
92
8
1,2,3,6,7,8
213
1,2,3,7,8,9
193
1,2,3,4,7,8
81
6.8 18 6.6
Outliers excl. ( ' ) Mean
CV
n (*)
I0.4
7.7 (b)
11
91
2.3 (a)
11
210
4.3 (a)
10
169
6.6 (a)
11
80
4.9 (a)
10
( ' ) A l l data out of the four sigma confidence i n t e r v a l were considered to be outliers. (a) : single step e l i m i n a t i o n (b) : two step e l i m i n a t i o n , which was j u s t i f i e d for the 2,3,7,8-TCDD data, because of the presence of one extremely deviation r e s u l t (*) n : the number of data a f t e r e l i m i n a t i o n of o u t l i e r s .
Crude f l y ash e x t r a c t The next exercise involved clean-up of a crude f l y ash e x t r a c t , as well as analysis for both PCDD and PCDF. The sample had been spiked previously with the necessary C13-1abeled internal standards. In addition to the already e x i s t i n g primary standard for PCDD's ( d i r t y f i v e ) , an analogous reference material for PCDF's ( d i r t y seven) was made available by the BCR. The sample clean-up procedures that were applied by the respective p a r t i c i p a n t s can be grouped as follows : 1. (7) - mixed modified s i l i c a (H2SO4, NaOH and AgNO3) -
alumina
2. (8) - mixed modified s i l i c a - carbon (Amoco PX21)/glass f i b r e s (Whatman GF/D) 3.
- mixed modified s i l i c a
1296
-
alumina
- carbon (Amoco PX21) + C e l i t e 545 4. (9) - alumina B, super I (Woelm), macro column The a n a l y t i c a l results are summarized in table IV. The results obtained for the PCDD's, when compared with those found in the previous t e s t (Table l l l ) ,
indicate that the c o n t r i b u t i o n of sample clean-up to the i n t e r l a b o r a t o r y
variance is small (maximum 4 %).
The overall c o e f f i c i e n t s of v a r i a t i o n are well w i t h i n
acceptable l i m i t s from the point of view of f i n a l c e r t i f i c a t i o n .
Table IV: means in ng/ml and c o e f f i c i e n t s of v a r i a t i o n of the means (CV in %) at the 95 % confidence i n t e r v a l obtained upon analysis of a crude f l y ash extract for PCDD and PCDF.
PCDD ( ' ) (a) Congener
Mean
CV
n (*)
PCDF ( ' ) (a) Mean
CV
n (*)
2,3,7,8
37
5.4
Ii
15
11
9
1,2,3,7,8
29
5.6
10
44
11
10
1,2,3,6,7,8
143
5.4
11
1,2,3,7,8,9
103
8.0
I0
17
36
8.9
11
89
9.0
9
146
8.3
I0
2,3,4,7,8
1,2,3,4,7,8 (+) 2,3,4,6,7,8
47
7.3
9
90
8.3
11
28
12
( ' ) (a) and (*) : cfr. t a b l e I I I . (+) : 1,2,3,4,7,8 + 1,2,3,4,7,9 for the furanes
The agreement of the results for PCDF's, clearly reflects the experience gained in the preceeding exercises.
Nevertheless,
the coefficients of variation for some PCDF
congeners are d e f i n i t e l y larger than that for PCDD. A detailed investigation of a l l experimental data led to the following conclusions : the partial separation of 2,3,7,8-T4CDF and 2,3,4,8-T4CDF, which is possible on polar columns, varies considerably from one column to another, even when comparing new columns of the same type obtained from the same manufacturer. The same effect was found for the separation between 1,2,3,7,8,9-H6CDF and an unknown accompanying impurity.
In this case, about half of the participants obtained a single
peak where the others found a reasonably well separated doublet.
1297
The separation of 1,2,3,7,8-P5CDF from 1,2,3,4,8-P5CDF is impossible on the a c u t a l l y
-
available polar columns but can e a s i l y be performed on a nonpolar column.
The problem for
1,2,3,7,8,9-H6CDF, described in the previous paragraph, can also be circumvented by using a non polar column. As a consequence, i t still
be improved,
*
polar
if
was agreed that the q u a l i t y of the results for these PCDF's could
columns could be tested,
before use, as to t h e i r
capability
to separate
2,3,7,8-T4CDF from 2,3,4,8-T4CDF, *
and i f
and 1,2,3,7,8,9-H6CDF would be determined using a suitable
1,2,3,7,8-P5CDF
nonpolar column. Conclusions and future planning The systematic preliminary t e s t s , set up in t h i s BCR-programme, and the dedicated care and commitment of the p a r t i c i p a t i n g analysts made i t possible to obtain good q u a l i t y data for the analysis of PCDD and PCDF in f l y ash extracts. Fly ash e x t r a c t i o n i t s e l f
has not yet been attempted.
However, t h i s step has been studied extensively in the past and the results were published in great d e t a i l .
Therefore, i t is not expected to be a c r i t i c a l
step in view of the f i n a l
c e r t i f i c a t i o n exercise, which is planned for the very near future. Additionally,
it
is envisaged to produce, as a surplus, a cleaned-up f l y
ash e x t r a c t
c e r t i f i e d for PCDD and PCDF. This reference material is proposed as a response to the problem of preliminary column testing,
which was proved
to
be an absolute
condition
especially
in
the
field
of
PCDF-analysis. F i n a l l y , the experience gathered during the exercises described in t h i s paper may be the basis f o r the c e r t i f i c a t i o n of a number of other important matrices in the future. Acknowledgements.
The authors wish to thank a l l members of the Dioxin Working Group for
t h e i r commitment and f r u i t f u l l P a r t i c i p a t i n g laboratories :
cooperation. - Cherchar
(France)
- Ist.
(Italy)
-
Inq. Atmosferico
I s t . Mario Negri
(Italy)
- University of Amsterdam
(Netherlands)
- TNO, D e l f t
(Netherlands)
-
Duphar, Weesp
(Netherlands)
- Shering Agrochem. Ltd.
(England)
- SCK/CEN
(Belgium)
-
-
-
Bayer AG, Leverkusen
(GDR)
Centre Nat. de Recherche
(France)
BASF AG
(GDR)
- Ecole Polytechnique, Paris
(France)
- Warren Spring Lab.
(England)
1298
References (1) (2) (3) (4)
(5) (6) (7) (8) (9)
Bumb, R.R. et a l . , Science 210, 385 (1980) Hutzinger, 0., Fiedler, H., VDI Berichte 634, 17 (1987) Sander, H.P., VDI berichte 634, 37 (1987) McMillin, C.R. et a l . , Chlorinated Dioxins and Dibenzofuranes in the Total Environment, Choundhary G., Keith L. and Rappe C. eds, Butterworth Publ., Ann Arbor Science, vol I, 165 (1983) Gross, M.L. et a l . , Anal. Chem., 53, 1902 (1981) Brumley, W.C. et a l . , J. Agric. Food Chem., 29, 1040 (1981) Kook, R., Lustenhouwer, J., Olie, K., Hutzinger, 0., Anal. Chem., 53, 461 (1981) Smith, L.M., Anal. Chem., 53, 2152 (1981) Hagenmaier, H. et a l . , VDI Berichte 634, 61 (1987)