Вы находитесь на странице: 1из 46

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of


Chlorinated Paraffins in Environmental
Matrices
Phase 1

Ike van der Veen

IVM Institute for Environmental Studies, VU University, Amsterdam,


the Netherlands

Wim Cofino

Wageningen University and Research Centre, Alterra, QUASIMEME


Laboratory Performance Studies, The Netherlands

Steven Crum

Wageningen University and Research Centre, Alterra, QUASIMEME


Laboratory Performance Studies, The Netherlands

Jacob de Boer

IVM Institute for Environmental Studies, VU University, Amsterdam,


the Netherlands

Report W-12/11, version 1


11 September 2012
Accredited under No. L476 (RvA)

This report is released by: Prof. Dr. J. de Boer


Head of the department C&B

It was internally reviewed by: Prof. Dr. J. de Boer

IVM
Institute for Environmental Studies
VU University Amsterdam
De Boelelaan 1087
1081 HV AMSTERDAM
The Netherlands
T +31-20-598 9555
F +31-20-598 9553
E info@ivm.vu.nl

Wageningen University and Research


Centre, Alterra, QUASIMEME Laboratory
Performance Studies, The Netherlands
P.O. Box 47
6700 AA Wageningen
The Netherlands
T +31 (0)317 486546
F +31 (0)317 419000
E quasimeme@wur.nl

Copyright 2012, Institute for Environmental Studies


All rights reserved. No part of this publication may be reproduced, stored in a retrieval
system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the copyright
holder
IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Contents
Summary

Introduction

Materials and methods

2.1
2.2
2.3
2.4

Study design
Material preparation
Methods used by participants
Data Assessment

9
9
9
11

Results

17

Discussion

19

4.1
4.2

Laboratory performance
Comparison with other ILSs

19
22

Conclusions

25

Acknowledgements

27

References

29

Appendices

31

Appendix 1

List of Participants

32

Appendix 2

Results and graphical presentation

33

Appendix 3

Numerical z-score values per matrix

38

Appendix 4

Consistency of data

39

Appendix 5

Graphical output of the Cofino Model statistics

40

Appendix 6

Additional method information

44

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Summary
The first round of an international interlaboratory study (ILS) on chlorinated paraffins
(CPs) was organised by the Institute for Environmental Studies (IVM) in cooperation
with the proficiency testing scheme of QUASIMEME. This study was initiated because it
was generally agreed at a workshop on the analysis of CPs in Ostend, Belgium,
organised by QUASIMEME in March 2010, that there was a clear need for an
interlaboratory study, preferably designed in a step-wise way. The first objective of the
study was to assess the intercomparability of the data produced and secondly to
initiate improvements.
In total 15 laboratories participated in the present study on the analysis of CPs. The
participants were requested to quantify the total concentration of CPs and the
concentration of the three individual CPs 1,1,1,3,9,10-Hexachlorodecane (1,1,1,3,9,10HCD) , 1,1,1,3,11,12-Hexachlorododecane (1,1,1,3,11,12-HCDD), and
1,1,1,3,11,13,13,13-Octachlorotridecane (1,1,1,3,11,13,13,13-OCTD) in an iso-octane
solution of CPs. Two solutions containing CPs in known concentrations were provided
to the participants for calibration purposes. Participants used their in-house
quantification methods and techniques. The results were collected and statistically
evaluated using the Cofino statistics. Between lab coefficients of variation (CVs) and zscores were appointed to the laboratorys results as an expression of accuracy.
The majority of the laboratories (57-71%) obtained satisfactory z-scores for the
analysis of the three individual CPs. For individual CP analysis the model between lab
CVs varied between the compounds from 22 to 46%. For total CP a between lab CV of
56% was found. Since no other intercomparison studies were known for the analysis of
individual CPs, no comparison could be made with previous intercomparison studies.
Even though a variety of quantification techniques were used, the difference between
the median value and the target value of the total CP concentration has decreased
compared to a study of Tomy et al. (1999), in which all participants used the same
quantification technique.
The present study also showed that the overall performance of participants in the
analysis of total CPs has improved compared to the interlaboratory study of Pellizzato
et al. (2009) although the present study only dealt with a solution of CPs, while the
study of Pellizzato et al. (2009) dealt with an oil extract.
The overall experience in CP analysis in the present study varied between the
participants, but no significant difference could be calculated between the
performance of participants with different levels of experience.

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Introduction

Chlorinated paraffins (CPs), also known as Polychlorinated alkanes (PCAs), are complex
mixtures of chlorinated n-alkanes with carbon chain lengths of 10 to 30 and a
chlorination degree between 30% and 70% by mass. Characterization of CPs is
performed by their alkane chain lengths. They are divided into three groups: shortchain (C10-13) (SCCP), medium-chain (C14-17) (MCCP) and long chain (C18-32) (LCCP)
CPs.
CPs are used in several industrial applications like flame-retardants in the rubber
industry, as high temperature lubricants and cutting fluids in the metalworking
industry and as additives in liquids, in paints and textile. The analysis of CPs is highly
complicated. There tens of thousands of congeners which make separation by gas
chromatography (GC) and even by GCxGC difficult. Alternative methods are scarce and
may suffer from false positive results. Therefore, the data reported on CPs up to the
present include very high uncertainties (100% or more).
Chlorinated paraffins are currently of utmost importance to the environmental analyst.
They are being produced in high volumes, they are under discussion in the United
Nations Environmental Programme (UNEP) for possible definition as a persistent
organic pollutant (POP), and they are a mandatory determinand in the European Water
Framework Directive. However, their analysis is subject to very large variation, as the
CP patterns are extremely complex. In March 2010, QUASIMEME organized a workshop
on the analysis of CPs in Ostend, Belgium. A number of critical issues in the analysis of
these CPs were discussed. It was generally agreed that there was a clear need for an
interlaboratory study, preferably designed in a step-wise way and accompanied by
expert comments.
The present study focuses on the analysis of CPs in a solution of undisclosed
concentration, to assess the possibility of producing precise and accurate data for a
range of CPs in a mixture.
In total 15 laboratories participated in the present ILS, of which 11 submitted data. The
ILS study focussed on the CPs listed in table 1.
Table 1

Full names, abbreviations and chemical formulas of individual CPs covered


in this study and provided.

Full name

Abbreviation

1,1,1,3,9,10-Hexachlorodecane
1,1,1,3,11,12-Hexachlorododecane
1,1,1,3,11,13,13,13-Octachlorotridecane

1,1,1,3,9,10-HCD
1,1,1,3,11,12-HCDD
1,1,1,3,11,13,13,13-OCTD

Chemical
formula
C10H16Cl6
C12H20Cl6
C13H20Cl8

Chlorination
degree (% Cl)
61.0
56.4
61.7

This study was carried out by the Institute for Environmental Studies (IVM) in
collaboration with QUASIMEME (www.QUASIMEME.org).

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Materials and methods

2.1

Study design

The aim of the sample preparation was to select individual CPs as representative for
CPs in the environment. Unfortunately, only commercial standards with chlorine atoms
at the end carbon atoms of the CP backbone were available. Due to this lack of
standards, three congeners with CPs on the ends of the backbone were selected.
Three Ampoules were prepared and sent to the participants with the aim of quantifying
in triplicate the three individual CPs, mentioned in table 1, and the total amount of CPs
in ampoule C by using ampoule A and B (see chapter 2.2).
The first invitation for participation in the study was sent out in November 2010 and
the samples were distributed in August 2011.

2.2

Material preparation

Ampoule A
Ampoule A contained an iso-octane solution of three individual CPs of a known identity
and in a known concentration (see Table 2).
Table 2

Content of Ampoule A

Abbreviation

Concentration (ng/mL)

1,1,1,3,9,10-HCD

1366

1,1,1,3,11,12-HCDD

1389

1,1,1,3,11,13,13,13-OCTD

1009

Ampoule B
Ampoule B contained an iso-octane solution of a technical mixture of SCCPs (C10-C13,
51,5% Cl) in a disclosed concentration (53 g/mL).
Ampoule C
The iso-octane solution in ampoule C consisted of a technical mixture of SCCPs (C10C13, 55.5% Cl) spiked with the three individual CPs, which are mentioned in table 1.
Details on the concentrations of the CPs can be found in Table 3 and in Appendix 2.

2.3

Methods used by participants

A short description of the methods reported by each individual participant is provided


in Appendix 6. One participant (CPP15) handed in two data sets, obtained with two
different analysis methods (marked with m1 and m2). Although, participants used
different detection methods for the analysis of CPs, all participants used a GC method
for the separation (see Figure 1). The preferred method was GC-mass spectrometry
(GC-MS) (69%), and a few laboratories used GS-MS/MS (23%). Only one participant did
not use an MS for the detection. A GCxGC- electron capture detector (ECD) system was
used instead. The ion sources used by the participants who did use an MS technique
IVM Institute for Environmental Studies

10

Materials and methods

varied. Almost half of the participants (46%) used chemical ionization (CI), while 31%
used electron impact (EI) and 23% used electron capture chemical ionization (ECNI) (see
Figure 2). Whenever EI was used the polarity used was positive, while the polarity used
with CI and ECNI was negative.
Ionization

Analysis method
GCxGC-ECD
8%
ECNI
23%

GC-MS/MS
23%
CI
46%

GC-MS
69%
EI
31%

Figure 1

Methods used for analysis of


CPs, for details see Appendix 6.

Figure 2

Ionization techniques of the


MS methods used.

For the GC separation often a DB-5MS column was used (46%), but also other columns
were utilised, such as Rtx-5MS and Rtx-1614 (see Figure 3). Splitless injection was used
(69%) as well as on-column injections (23%) and cooled injection with solvent venting
(8%).
DB-1MS xColumn
innowax
8%
Rtx-200
8%

Rtx-1614
8%
DB-5MS
46%

Restek
15%

Rtx-5MS
15%

Figure 3

2.3.1

Columns used for analysis of CPs on a GC,

Quantification

For the quantification of the concentration of total CPs in ampoule C participants were
requested to use ampoule B. Since the mixture in ampoule B contained CPs with 51.5%
Cl, and the mixture in ampoule C contained 55.5% Cl some participants reported that
an in-house mixture was a better match to quantify CPs in ampoule C. Participant CPP4
and CPP8 handed in two data sets. The first data set, marked with m1, was obtained by
quantification of total CP with ampoule B and the second data set, marked with m2,

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

11

was obtained with quantification with in-house mixtures. In-house mixtures were also
used by participants CPP3 for the quantification of total CPs in ampoule C.
The majority of the participants (64%) used in addition an internal standard, like 13C
Chlordane, 13C12-138-HxBDE, d10 anthracene, and 13C HCB, for the quantification of
CPs.

2.4

Data Assessment

The data assessment was carried out according to the principles employed in the data
assessment of the QUASIMEME laboratory performance studies. All data received from
the participants were entered into a database and assessed using a standard
procedure enabling direct comparison between participants. The approach to the
assessment is based on standard ISO 13528 (Thompson et al., 2006). Additions or
differences in the assessment from these standards are given or referred to in this
report. However, the assigned value and the laboratory assessment using z-scores is
calculated using the model developed by Cofino et al. (2000).
The standard, (ISO 13528), includes statistics for proficiency testing schemes, and
uses robust statistics as a basis for the assessment. However, it is generally
acknowledged that robust statistics cannot cope with more than 10% extreme values,
particularly with a skewed distribution. The Cofino model is able to routinely handle
these types of distribution and provide the best estimate of the consensus value,
which may be used as the assigned value.
The Cofino model is used routinely for QUASIMEME assessments. In this study the
Normal Distribution Approximation (NDA) has been employed (Cofino et al., 2000).
This approach includes all data in the evaluation and no subjective truncation or
trimming is made. This model has been further developed to include Left Censored
Values (LCV, which is the correct nomenclature for less than values) The development
of these models has been fully documented and published (Cofino et al., 2000, Cofino,
et al., 2005, Wells, et al., 2004) and accepted by the Dutch Accreditation Body. An
overview of the assessment with explanation and examples is given in the Assessment
Rules for the Evaluation of the QUASIMEME LP Studies Data (Wells and Scurfield, 2004).
The details of the Cofino Model were provided elsewhere (Wells et al., 2004, Wells and
Scurfield, 2004), but in summary the approach is as follows:
All data included in the assessment;
No data trimmed or down weighted;
Assigned values (AV) based on Cofino NDA model;
All LCV are also included, provided certain criteria are met (see chapter 2.4.2).

2.4.1 Plots
The performance of the laboratories in this study is illustrated in the z-score
histograms in Appendix 5, and an example of the plots is shown in Figure 4. Where
the assigned value for a determinand is indicative, the values are plotted as their
original reported concentrations. The rules for confirming whether the consensus
value should be an assigned value or an indicative value are given in the Assessment
Rules for the Evaluation of the QUASIMEME LP Studies Data (Wells and Scurfield, 2004)
with appropriate examples.
Normally, four plots are given for each determinand (see Figure 4). The upper left plot
provides an impression of the probability density function (PDF) model for all data
(black) and for the first mode (PMF1) model of the data (blue dotted). Superimposed on

IVM Institute for Environmental Studies

12

Materials and methods

these PDFs is a histogram of the individual measurements (grey bars). This plot shows
the distribution of the data as a whole, and of the data in the main mode (PMF1) model
on which the assigned value is based.

Figure 4. Examples of the graphical output of the Cofino Model statistics for
1,1,1,3,11,13,13,13-OCTD in ampoule C.
The Kilt Plot (Overlap Matrix) (upper right plot) provides an overview of the degree of
overlap of each pair of data. It gives a clear indication of the degree of homogeneity of
the data. As a key, the white areas indicate maximum overlap of the PDFs and,
therefore, highest agreement (an overlap of one implies that the two laboratories of
the pair report exactly the same results), while the black area show the pairs in poor
agreement.
The lower left plot is a ranked overview of all data with an error bar of 2 SD. The
numerical values are given in blue and the LCVs are given in red.
The ranked z-score plot (lower right) is based on the mean of the data, which is
normally also the assigned value. However, if there is any adjustment required to the
assigned value as a result of the assessment, e.g. use of the nominal concentration or
a trimmed value, then the final z-score given in the z-score histograms will reflect
these changes.

2.4.2 The assigned values and indicative values


The assigned value (AV) is obtained from the main mode model of the data using the
Cofino Model (bleu dotted line in upper left panel in Figure 4), and is centred around
the highest density of values. Unless otherwise stated, the assigned value is based on
this consensus value of all data. Although all data are included in the assessment,
those values that lie some distance from AV contribute less to the mean than values
which occur at or near the mean.

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

13

In some instances it is not possible to set an AV, and an indicative value is given. No
assessment of laboratory performance is given where an indicative value is set. An
overview of the assessment, with explanation, decision flowcharts and examples, is
given in the paper Assessment Rules for the evaluation of the QUASIMEME Laboratory
Performance Studies Data, available on the QUASIMEME website
(www.quasimeme.org). A summary of the categories is given below:
Category 1
For data with the number of numerical observations 7
An assigned value is based on the mean when 33% of values have a z-score of |Z| < 2.
Where < 33% of the data has |Z| < 2 the value is indicative. i.e. at least 33% must be in
good agreement.
Category 2
For data with the number of numerical observations > 3 and < 7
An assigned value is based on the mean when 70% of values have a z-score of |Z| < 3
and a minimum of 4 observations have |Z| < 2. Otherwise the value is indicative. i.e. for
small datasets, n > 3 and n < 7, there need to be very good agreement and a
maximum of one extreme value before an assigned value can be given.
Category 3
For data with the number of numerical observations < 4
No assigned value is given. Normally the median value is given as an indicative value.
Category 4
For data with the high Total Error% >100% in combination with bad performance
No assigned value is given.

2.4.3 The Z-score Assessment


A z-score (Thompson and Wood, 1993) is calculated for each participants data for
each matrix / determinand combination which is given an assigned value. The z-score
is calculated as follows:

z - score =

Mean from Laboratory - Assigned Value


Total Error

It is emphasized that in many interlaboratory studies the between-laboratory standard


deviation obtained from the statistical evaluation of the study is used as total error in
the formula above. In the QUASIMEME data assessment, the total error is estimated
independently taking the needs of present-day international monitoring programs as
starting point. For each determinand in a particular matrix, a proportional error (PE)
and a constant error (CE) have been defined. The total error depends on the
magnitudes of these errors and on the assigned value:
Total Error =

Assigned Value x Proportional Error (%)


+ 0.5 x Constant Error
100

The values for the PE and CE are set by the QUASIMEME Scientific Assessment Group
and are monitored annually. The values are based on the following criteria:
- Consistency of the required standard of performance to enable participating
laboratories to monitor their assessment over time.

IVM Institute for Environmental Studies

14

Materials and methods

- Achievable targets in relation to the current state of the art and the level of
performance needed for national and international monitoring programmes.
The assessment is based on the standard ISO 17043 and z-scores. The QUASIMEME
model is designed to provide a consistent interpretation over the whole range of
concentration of analytes provided, including an assessment where LCVs are reported.
The proportional error is set at 12.5% for all matrices. This applies to all determinands.
The constant error (CE) has been set for each determinand or determinand group. This
value was initially set to reflect the limit of determination, but is at present more
closely related to the overall laboratory performance. The magnitude of the CE is set to
provide a constant assessment in terms of z-score regardless of concentration. The CE
is set at 0.025. Therefore at low concentrations the level of accuracy required to obtain
a satisfactory z-score is less stringent than at high concentrations.
Following usual practices e.g. ISO 17043, the z-scores can be interpreted as follows for
laboratories which take part in QUASIMEME to assure the quality of their data for use in
international marine monitoring programmes:
|Z| < 2 Satisfactory performance
2 <|Z| < 3 Questionable performance
|Z| > 3 Unsatisfactory performance
|Z| > 6 frequently points to gross errors (mistakes with units during reporting,
calculation or dilution errors, etc.).
The following schematic presentation illustrates the interpretation of the z-scores:

Questionable
performance

Unsatisfactory
performance

Satisfactory
performance

AV-3*TE

Assigned
Value (AV)

AV-2*TE

Unsatisfactory
performance

AV+3*TE
AV+2*TE

TE : total error

It is not possible to calculate a z-score for LCVs as LCVs represent a cut-off value
rather than continuous data. However, Quasimeme provides a simple quality criterion:
LCV/2 < (concentration corresponding to |z|=3): LCV consistent with assigned value

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

15

LCV/2 > (concentration corresponding to |z|=3): LCV inconsistent with assigned value,
i.e. LCV reported by laboratory much higher than numerical values reported by other
laboratories.
Z-score key:

S Satisfactory
Q Questionable
U Unsatisfactory

LCV key:

C Consistent
I Inconsistent

No data:

B - Blank

IVM Institute for Environmental Studies

17

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Results

The submitted results have been evaluated statistically and whenever the data met the
requirements (as described in chapter 2), an assigned value was established. Z-scores
were calculated based on the assigned value. Summary statistics are presented in
Table 3. A summary of the assigned values and the percentage of satisfactory to
unsatisfactory z-scores are presented in Table 4. Whenever less than values (LCV) were
submitted, the percentage of consistent and inconsistent LCVs with the assigned value
is given. The submitted data is presented in Appendix 2. Tables with individual zscores are presented in Appendix 3, consistencies of the individual results are
presented in Appendix 4 and z-score plots in Appendix 5.
Table 3

Summary of results of CPs in ampoule C (results in ng/mL).


Target Assigned Indicative Model
Median Min*
value
value
Value
Mean

Determinand

Max**

Model
Between
Lab CV%

Percentage
n>LOQ
in PMF1

1,1,1,3,9,10-HCD

129

138

NA

138

148

51

810

22

57

1,1,1,3,11,12-HCDD

132

177

NA

177

197

122

2800

46

76

1,1,1,3,11,13,13,13OCTD

96

111

NA

111

120

90

1500

28

78

12000

NA

18500

18180 18500 8100 49300

56

76

14

Total CP

* Min: lowest value submitted > LOQ


** Max: highest value submitted > LOQ

Table 4

Summary of laboratory performance for CPs in ampoule C (results in


ng/mL).

Determinand

Total
Assigned Indicative
Error
value
Value
%

% data
received

% |Z|<2
(Satisfactory)

% 3>|Z|>2 % 6>|Z|>3
% |Z|>6
(Questi(Unsatis(Extreme)
onable)
factory)

1,1,1,3,9,10-HCD

138

NA

12.9

50

57

14

29

1,1,1,3,11,12-HCDD

177

NA

12.8

50

57

14

14

14

1,1,1,3,11,13,13,13OCTD

111

NA

13.0

50

71

14

14

Total CP

NA

18500

100

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

19

Discussion

In total 15 laboratories from all over the world participated in the present study. Of
those laboratories, 11 submitted data, of which three laboratories handed in two data
sets obtained with either two different analyses methods(see chapter 2.3), or with
different calibration solutions (see chapter 2.3.1). Seven participants were already
working on CP analysis for over 3 years, while two laboratories were analysing CPs
between 1 and 3 years. Only one laboratory had an experience in CP analysis of
shorter than one year. The experience of one participant was unclear. No significant
difference is observed between the reported concentrations of the less experienced
participants and the more experienced participants (Figure 5).

Total CP
50000
45000
40000
35000
30000

(ng/mL)

25000
20000

Indicative value

15000

Target value

10000
5000
0

experience
< 1 year

experience
1-3 years

experience
> 3 years

experience
unknown

Participating laboratories

Figure 5 Experience of participant vs total CP concentration reported.

4.1

Laboratory performance

As shown in Table 3, the model between lab CVs varied between the compounds from
22-56%. Satisfactory z-scores (i.e. Z|2|) were obtained from the majority (57-71%)
(depending on the determinand) of the participants that submitted results (see Table
4).
Comparing the assigned values (or indicative value in case of total CP) with the target
values (see Table 3 and Figure 6) shows that all assigned values (and indicative values)
are 1.11.5 fold higher than the target values, which might be due to coeluting CPs,
since the deviation is not observed by Participant CPP3 who used high resolution (HR)
MS.
Due to the deviation between assigned value and target value some participants, who
reported concentrations which agreed well with the target concentrations, received
higher, though still satisfactory, z-scores. For example participants CPP3, CPP10 and
CPP15m2 received z-scores of -1.9, -2.0 and -1.8 respectively for 1,1,1,3,11,12-HCDD,

IVM Institute for Environmental Studies

Discussion

while the reported concentrations were 135, 131, and 136 ng/mL and the target value
was 132 ng/mL.

300

1,1,1,3,9,10-HCD (n=21)

250
200

(ng/mL)

150

Assigned value
Target value

100
50
0

Participating laboratories

300

1,1,1,3,11,12-HCDD (n=21)

250

200

Assigned value
150

(ng/mL)

20

Target value
100

50

Participating laboratories

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

200

21

1,1,1,3,11,13,13,13-OCTD (n=21)

180
160
140
120

Assigned value

(ng/mL)

100

Target value

80
60

40
20
0

Participating laboratories

50000

Total CP (n=42)

45000
40000
35000
30000

(ng/mL)

25000
20000

Indicative value

15000

Target value

10000
5000
0

Participating laboratories

Figure 6 Comparison of assigned values (or indicative value) with the target values.
In Figure 7 a comparison is given between the different ionization techniques used and
the reported concentration of total CP. Although only limited data is available per
method and no hard conclusions can be made, it looks like EI with positive polarity
gives results, which are closer to the target value. Only one participant reported results
on the concentrations of total CPs obtained with ECD, which resulted in a deviation of
only 4.7% of the target value. Two of the participants, marked with a asterisk in figure
7 used HR MS.

IVM Institute for Environmental Studies

Discussion

Total CP
50000

45000
40000
35000

HR MS

30000
25000

(ng/mL)

22

20000
15000

Indicative value
HR MS
Target value

10000
5000

CI negative
quantified with
in-house std.

CI negative
quantified with
ampoule B

ECNI

EI positive

ECD

Participating laboratories

Figure 7 Methods used by participant vs total CP concentration reported.

4.2

Comparison with other ILSs

No previous intercomparison studies are known on the analysis of individual PCAs. In


an interlaboratory study of Tomy et al. (1999) two SCCP containing solutions (PCA-1
and PCA-70) were quantified for the total SCCP concentration by 7 participants who all
used HRGC-ECNI-MS. Concentrations of the mixtures in that study were 6-10 fold
higher than the concentration of the total CP concentration in ampoule C of the
present study. Target CP concentrations of the mixtures were 74 and 118 ng/L
respectively, while the means of the determined concentrations were 99.3 and 297
ng/L respectively, which is a difference of 34 and 152%. CVs were 20 and 44%. The
model between lab CV for the total CP determination in the present study was 56%,
which is higher than the CVs of the study of Tomy et al. (1999). The difference
between the mean of the reported concentrations and the true value in the present
study is 35 %, which is equal to the performance on PCA-1 but better than the
performance on PCA-70 of the study of Tomy et al. (1999). In the study of Tomy et al.
(1999) all participants used the same quantification technique, while in the present
study a variety of techniques is used, which might explain the higher variance in the
present study.
In another intercomparison study (Pellizzato et al., 2009) the total SCCP concentration
was determined in an extract of industrial oil. Reported concentrations varied between
8.5 and 3200 mg/L, with 4 of the 6 participants reporting concentrations between 8.5
and 16.1 mg/L, which are in the same range as the concentrations of total CPs in
ampoule C of the present study. In the study of Pellizzato et al. (2009) the CV was
209%. When the results of one participant were excluded, the CV decreased to 82%.
The better performance in the present study compared to the study of Pellizzato et al.
(2009) is most likely due to the fact that a standard solution is cleaner than an oil
extract.
Possible source for data variance
Although participants were requested to use Ampoule A and B for the quantification of
total CPs in ampoule C, some participants used a better fitting in-house mixture for

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

23

the quantification (see chapter 2.3.1), which resulted in a better performance of those
participants and a lower model between lab CV (see Figure 8).
Total CP
50000

CPP4 M1

45000
40000
35000
30000

(ng/mL)

25000
20000

CPP4 M2

CPP8 M1

Indicative value

15000

Target value

10000

CPP8 M2

5000
0

Quantiification with ampoule B

Quantiification with
in-house mixture

Participating laboratories

Figure 8 Quantification solution used by participant vs total CP concentration


reported.

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

25

Conclusions

For individual CP analysis the between lab CVs varied between 22 and 46%. Since no
other intercomparison studies were known for the analysis of individual CPs, no
comparison could be made with previous intercomparison studies. For total CP a
between lab CV of 56% was found.
A larger number of laboratories participated in this international interlaboratory study
than in previous intercomparison studies on total SCCP, showing that an increasing
interest in quantitative analysis of CPs, as well as the intention to evaluate laboratory
performance so as to improve the data produced in the field. This study showed that
even though a variety of quantification techniques were used the difference between
the median value and the target value has decreased compared to a study of Tomy et
al. (1999), in which all participants used the same quantification technique.
The present study also suggests that the overall performance of participants in the
analysis of total CPs has improved compared to the interlaboratory study of Pellizzato
et al. (2009), although the present study only dealt with a solution of CPs, while the
study of Pellizzato et al. (2009) dealt with an oil extract.
Obviously, further developmental work in laboratories and more interlaboratory
comparison exercises are needed to improve the analysis of CPs. Analytical standards
of CP congeners that occur in environmental samples are badly needed.
A second step of this exercise including analysis of clean extracts will be organised
shortly.

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

27

Acknowledgements
The Chlorinated Paraffins Sector Group of Eurochlor is gratefully acknowledged for
their financial support of this interlaboratory study, and their technical input.

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

29

References
Cofino, W.P., Stokkum, I.H.M. van, Steenwijk, J. Van & Wells, D.E. (2005) A new model for
the inference of population characteristics from experimental data using uncertainties.
Part II. Application to censored datasets. Anal. Chim. Acta, 533, 31-39.
Cofino, W.P., Wells, D.E., Arise, F., Stokkum, I. van , Wegener, J.W. & Peerboom, R. (2000). A
new model for the inference of population characteristics from experimental data
using uncertainties. Application to interlaboratory studies. Chemometrics. Intell. Lab.
Syst. 53, 37-55.
Pellizzato, F., Ricci, M., Held , A., Emons, H., Bhmer, W., Geiss, S., Iozza, S., Mais, S.,
Petersen, M. & Lepom, P. (2009). Laboratory intercomparison study on the analysis of
short-chain chlorinated paraffins in an extract of industrial soil. Trends Anal. Chem.
28, 1029-1035.
Thompson, M., Ellison, S.L.R. & Wood, R. (2006). The international harmonised protocol for
proficiency testing of analytical chemistry laboratories. IUPAC Technical report.
Interdivisional working party for harmonization of quality assurance schemes. Pure
Appl. Chem. 78 , 145-196.
Thompson, M.& Wood, R. (1993). International harmonised protocol for proficiency testing
of (chemical) analytical laboratories. J. AOAC Int, 76, 926-940.
Tomy, G.T., Westmore, J.B., Stern, G.A., Muir, D.C.G. & Fisk, A.T. (1999). Interlaboratory
study on quantitative methods of analysis of C10-C13 polychloro-n-alkanes. Anal.
Chem., 71, 446-451.
Wells, D.E., Cofino, W.P. & Scurfield, J.A. (2004). The application of the Cofino model to
evaluate laboratory performance study data using bandwidth estimator. UK. Report
nr. 04/04. FRS Marine Laboratory, 375 Victoria Road, Aberdeen 388 AB11 9DB.
Wells, D.E. & Scurfield, J.A. (2004). Assessment rules for the evaluation of the QUASIMEME
laboratory performance studies data. UK, version 2. FRS Marine Laboratory, 375
Victoria Road, Aberdeen AB11 9DB, pp.390 -391.

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

31

Appendices
1.
2.
3.
4.
5.
6.

List of participants
Results and graphical representation
Numerical z-score values per matrix
Consistency of data
Graphical output of the Cofino Model statistics
Additional method information

IVM Institute for Environmental Studies

32

Appendix 1

List of Participants

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Appendix 2

33

Results and graphical presentation

Participant code:
Date Received:
Date Analyzed:
Weight Received:
Weight Analyzed:

Concentration in ng/mL
1,1,1,3,9,10-HCD
1,1,1,3,11,12-HCDD
1,1,1,3,11,13,13,13-OCTD
Total CP

Target
value

Assigned
value

129
132
96
12000

138
177
111
NA

Indicative Model
value
Mean Median
NA
NA
NA
18500

138
177
111
18180

148
197
120
18500

MIN

MAX

51
122
90
8100

810
2800
1500
49300

Model
Model
Between percentage
Lab CV%
in PMF1
n>LOQ
22
46
28
56

57
76
78
76

7
7
7
14

ND: not detected


NA: not analyzed

IVM Institute for Environmental Studies

34

Participant code:

CPP1
1

Date Received:
Date Analyzed:
Weight Received:
Weight Analyzed:

08/08/2011
10-14/10/2011
7.0712g 7.3768g 7.1024g
7.0716g 6.8108g 7.1030g

Concentration in ng/mL
1,1,1,3,9,10-HCD
1,1,1,3,11,12-HCDD
1,1,1,3,11,13,13,13-OCTD
Total CP

NA
NA
NA

CPP1
2

NA
NA
NA

CPP1
3

NA
NA
NA

29600 28000 29300

CPP2
1

CPP2
2

CPP2
3

CPP3
1

CPP3
2

CPP3
3

02/08/2011
12/08/2011
7.2569g 7.2209g 7.3395g
7.2566g 7.2214g 7.3400g

NA
NA
NA
NA

NA
NA
NA
NA

NA
NA
NA
NA

CPP6
1

CPP6
2

CPP6
3

CPP4
m1
1

CPP4
m1
2

CPP4
m1
3

04/08/2011
11/09/2011
7.2241g 7.2393g 7.0773g
200 l
200 l

CPP4
m2
1

CPP4
m2
2

CPP4
m2
3

04/08/2011
11/09/2011
7.2241g 7.2393g 7.0773g
200 l
200 l

138
132
139
NA
NA
NA
NA
NA
NA
135
130
139
NA
NA
NA
NA
NA
NA
91
96
94
NA
NA
NA
NA
NA
NA
11400 12200 12050 49300 48500 40300 21600 20300 17100

ND: not detected


NA: not analyzed

Participant code:

CPP5
1

CPP5
2

CPP5
3

Date Received:
Date Analyzed:
Weight Received:
Weight Analyzed:

Concentration in ng/mL
1,1,1,3,9,10-HCD
1,1,1,3,11,12-HCDD
1,1,1,3,11,13,13,13-OCTD
Total CP
ND: not detected
NA: not analyzed

IVM Institute for Environmental Studies

CPP7
1
01/11
1.2267g
0.3597g

NA
NA
NA
NA

NA
NA
NA
NA

NA
NA
NA
NA

NA
NA
NA
NA

NA
NA
NA
NA

NA
NA
NA
NA

CPP7
2

CPP7
3

04/08/2011
28/10/2
27/10
011
1.4115g 1.2328g
0.5237g 0.3420g

CPP8
m1
1

CPP8
m1
2

CPP8
m1
3

05/08/2011
28/10/2011
7.2778g 7.3531g 7.1863g
2 l
2 l
2 l

51
51
60
640
810
700
220
210
210
2600 2700 2800
110
130
120
1400 1500 1300
12000 12000 12000 18000 17000 17000

CPP8
m2
1

CPP8
m2
2

CPP8
m2
3

05/08/2011
12/09/2011
7.2778g 7.3531g 7.1863g
2 l
2 l
2 l

NA
NA
NA
8700

NA
NA
NA
8100

NA
NA
NA
8200

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Participant code:

CPP9
1

CPP9
2

CPP9
3

Date Received:
Date Analyzed:
Weight Received:
Weight Analyzed:

20/08/2011
20/10/2011
7.2042g 7.3329g 7.2475g
7.2042g 7.3329g 7.2475g

Concentration in ng/mL
1,1,1,3,9,10-HCD
1,1,1,3,11,12-HCDD
1,1,1,3,11,13,13,13-OCTD
Total CP

NA
NA
NA
NA
NA
NA
NA
NA
NA
31900 33500 33080

CPP10
1

CPP10
2

CPP10
3

CPP11
1

CPP11
2

CPP11
3

CPP12
1

01/08/2011
22/09/2011
1
1 ampoule
1 ampoule
1
1 ampoule ampoule
1 ampoule
ampoule

126.6
129.2
92.20
22600

123.5
131
97.02
23130

126.6
133.3
97.9
23030

NA
NA
NA
10300

NA
NA
NA
11000

35

CPP12 CPP12
3
2
14/11/2011

NA
NA
NA
11100

NA
NA
NA
18000

NA
NA
NA
26000

NA
NA
NA
28000

CPP15
m1
3

CPP15
m2
1

CPP15 CPP15
m2
m2
2
3

ND: not detected


NA: not analyzed

Participant code:

CPP13 CPP13
1
2

Date Received:
Date Analyzed:
Weight Received:
Weight Analyzed:

15/11
7.2471 g
1.3518 g

Concentration in ng/mL
1,1,1,3,9,10-HCD
1,1,1,3,11,12-HCDD
1,1,1,3,11,13,13,13-OCTD
Total CP

145.2
292.4
153.0
11942

CPP13
3

CPP14
1

CPP14
2

CPP14
3

CPP15
m1
1

09/11/2011
11/11
12/11
7.3842 g
7.1574 g
1.4382 g
1.3758 g

147.0
270.8
133.8
12399

152.95
254.2
148.3
11504

CPP15
m1
2
01/08/2011
13/10/2011

NA
NA
NA
NA

NA
NA
NA
NA

NA
NA
NA
NA

308
196
118
11200

308
200
124
11200

01/08/2011
13/10/2011

301
195
123
11900

153
147
101
31200

143
122
90
30200

159
139
101
31600

ND: not detected


NA: not analyzed

IVM Institute for Environmental Studies

36

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

37

IVM Institute for Environmental Studies

38

Appendix 3

Numerical z-score values per matrix

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

Appendix 4

39

Consistency of data

S = Satisfactory; Q = Questionable; U = Unsatisfactory; B = Blank

IVM Institute for Environmental Studies

40

Appendix 5

Graphical output of the Cofino Model statistics

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

41

IVM Institute for Environmental Studies

42

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

43

IVM Institute for Environmental Studies

44

Appendix 6

Additional method information

IVM Institute for Environmental Studies

Interlaboratory Study on the Analysis of Chlorinated Paraffins in Environmental Matrices

45

IVM Institute for Environmental Studies

46

IVM Institute for Environmental Studies

Вам также может понравиться