Development of a harmonised method for specific migration into the new simulant for dry foods established in Regulation 10/2011

Transcription

1 Development of a harmonised method for specific migration into the new simulant for dry foods established in Regulation 1/11 stablishment of precision criteria from an U interlaboratory comparison organised by the URL- Food Contact Materials for the quantification from and migration into poly(,6- diphenyl phenylene oxide) Giorgia Beldì, Natalia Jakubowska, Aurélie Peychès Bach and Catherine Simoneau 1 Report UR 568 N

2 uropean Commission Joint Research Centre Institute for Health and Consumer Protection Contact information Catherine Simoneau Address: Joint Research Centre, Via nrico Fermi 749, TP 6, 17 Ispra (VA), Italy -mail: Tel.: Fax: This publication is a Reference Report by the Joint Research Centre of the uropean Commission. Legal Notice Neither the uropean Commission nor any person acting on behalf of the Commission is responsible for the use which might be made of this publication. urope Direct is a service to help you find answers to your questions about the uropean Union Freephone number (*): (*) Certain mobile telephone operators do not allow access to 8 numbers or these calls may be billed. A great deal of additional information on the uropean Union is available on the Internet. It can be accessed through the uropa server JRC77515 UR 568 N ISBN ISSN doi:1.788/7684 Luxembourg: Publications Office of the uropean Union, 1 uropean Union, 1 Reproduction is authorised provided the source is acknowledged. Printed in Italy

3 C-JRC-IHCP, CAT Unit action programme URL Food Contact Materials No SANCO/11/FOOD SAFTY7-Food Contact Materials Giorgia Beldì, Natalia Jakubowska, Aurèlie Peychès Bach and Catherine Simoneau

4 Table of content 1. Summary Introduction Scope Time frame Test materials Preparation Homogeneity assessment Stability test Distribution Instructions to participants Approaches for statistical evaluation of results valuation of the BHT, BP, DiBP, DHA and DINCH method performance characteristics methods for determination of the consensus value, repeatability ( r ) and reproducibility ( R ) standard deviation Identification of modes using Kernel density plotting Mandel s h- and k-statistics valuation criteria for laboratory performance z-scores Results and conclusions Preliminary considerations Method performance characteristics from the precision experiment z-scores and laboratories performance Final conclusion Acknowledgements References Annexes

5 1. Summary Regulation (U) No 1/11 establishes (poly (,6-diphenyl phenylene oxide, PPPO)) as food simulant for testing specific migration from plastics into dry foodstuffs. Although research data has been available, there was to date no precision or validation data on the ability of a method both for the testing of specific migration and for the subsequent quantification of representative substances from PPPO. Previous research done in 1-11 by the URL-FCM focused the development and the optimisation of a standard operation procedure for using the new simulant for testing specific migration into dry foodstuffs. A preliminary part of the research led to the selection of substances that responded to criteria of being able to migrate into PPPO from a spiked film, be amenable to be homogeneously spiked into PPPO and into a plastic film, be stable in the film and in PPPO, and be representative a of a range of substances that are common food contact additives. The outcome of the preliminary research led to the selection of substances chosen as models for the subsequent interlaboratory comparison (ILC) exercise due to their migration properties and stability characteristics in LDP foil and the PPPO. The models for the exercise were butyl hydroxytoluene (BHT), benzophenone (BP), diisobutylphthalate (DiBP), diethylhexyladipate (DHA) and 1,-cyclohexanedicarboxylic acid diisononyl ester (DINCH). An efficient analytical method was developed for the quantification of a range of substances investigated from PPPO including extraction of the PPPO followed by GC-MS analysis. A method for fortifying PPPO and polyethylene film (LDP) with a range of model substances (surrogates) was also developed. An experimental design was also developed for the migration tests using smaller migration test cells than the larger conventional ones for overall migration. This work presents the results of the subsequent ILC which involved 17 laboratories at U level. The general aim of the ILC was to evaluate the laboratory performance and precision criteria of the harmonised method for the extraction and quantification of the models substances from PPPO and also for the migration test from a fortified plastic film into PPPO followed by their subsequent quantification. The URL prepared and proposed a SOP describing the exposure of a plastic material or article (intended for use with dry foods) to PPPO and the subsequent extraction of the PPPO and analysis of the extract. The exercise foresaw: 1) one part using a spiked PPPO for which the ILC focused on extraction from PPPO itself and quantification of surrogate substances ) another part using a film specifically contaminated with a selection of substances for which the ILC consisted of the exposure of a film to PPPO and subsequent extraction and quantification. This allowed to obtain data on precision and laboratory performance for both migration and quantification. The test materials used in this exercise were spiked PPPO samples prepared by URL-FCM with three levels of model substances and a spiked LDP film to perform the migration part of the exercise. The homogeneity and stability studies were performed by the URL-FCM laboratory. There were 17 volunteer laboratories participating from 15 countries to whom samples were dispatched and 16 of which submitted results. From the URL-NRL network 14 laboratories out of 15 reported results. There were guests from Germany that provided results as well. Since the aim of the ILC was evaluation of the laboratory performance and to obtain precision criteria there was a communication to the participants to the use harmonised protocol developed. Participants were invited to report four replicates measurements in repeatability conditions. This was done by - 5 -

6 most of the participants. The results of analyses were received and statistically interpreted. The assigned values were obtained as a consensus values after applying the robust statistics to the results obtained from the participants. results were also rated with z-scores in accordance with ISO 1358 [4]. The results showed that the validation ILC was successful with more than 8 valid results even taking into account the complexity of the exercise. The results showed that the method performance could be estimated to obtained 7-9% repeatability standard deviation (sd) for most substances (regardless of concentration), with 1% rsd for the high level of BHT and for DiBP at very low levels. The reproducibility results resulting from the ILC involving the 16 U laboratories were summarised as performance indicators of the method for the quantification from PPPO (for the 3 levels of concentrations of the 5 substances) as well as from migration experiments from the fortified plastic at 6C for 1 days and subsequent quantification. Reproducibility S.D. (%) calculated by ISO for the different exercises (LL: legislative limit, i.e. SML) are summarised below. Substance LL value mg/kg R sd % at 1/1 * LL R sd % at the LL R sd % At * LL R sd % From FILM BHT 45% 8% 19% 41% BP.6 (/) 1% % 17% % DiBP 4% 3% 6% 5% DHA 18 9% 1% 16% 35% DINCH 6 4% 1% 16% 16% Considering the lack of the data previously available in the literature, this interlaboratory comparison exercise provided a protocol method description which was tested and evaluated, and most importantly the first successful attempt to derive precision data at U level for a specific migration experiment. It also represented a challenging scenario and was very extensive with 3 levels of concentrations and 1 film, 5 substances, 4 replicates and results from 16 laboratories, i.e 136 data points. This exercise consequently provided a great breadth of valuable detailed and traceable precision data, which can be used for a more refined information and proficiency test of official controls planned for

7 . Introduction ILC studies are an essential element of laboratory quality assurance and allow individual laboratories to compare their analytical results with those from other laboratories while providing them objective standards to perform against. It is one of the core duties of the uropean Reference Laboratories to organise interlaboratory comparisons, as is stated in Regulation (C) No 88/4 of the uropean Parliament and of the Council [5]. In accordance with the above requirements the uropean Reference for Food Contact Materials (URL-FCM) organised in 11-1 interlaboratory comparison tests for the network of appointed National Reference Laboratories (NRLs). 3. Scope The scope of this ILC had 3 objectives: 1. to establish precision criteria of the analytical method (standard Operating protocol, SOP) proposed for the testing specific migration into PPPO for model substances.. to test the laboratory performance of the appointed NRLs to analyse butyl hydroxytoluene, benzophenone, diisobutylphthalate, diethylhexyladipate and 1,-cyclohexanedicarboxylic acid diisononyl ester in PPPO 3. to establish precision criteria and laboratory performance for both migration and quantification using a film specifically contaminated with a selection of substances for which the ILC consisted of the exposure of a film to PPPO and subsequent extraction and quantification. The concentration levels were chosen in accordance with the existing legislation [6]. The assessment of the measurement results was undertaken on the basis of requirements laid down in international standards and guidelines [4, 7, 8, 9]. 4. Time frame The URL-FCM prepared a SOP (Annex 1) based on modified and refined CN and CN methods [, 3]. Invitation letters were sent to the laboratories (Annex ). Laboratories were invited to fill in a letter of confirmation of their participation (Annex 3). The samples were dispatched to participants together with a letter accompanying the samples (Annex 4), the SOP of the analytical method to be used for the exercise - 7 -

8 (Annex 1), a format for the compilation of results to be eventually sent in nonelectronic format (Annex 7) and electronic files where the result should be inserted. The participants were asked to fill in a letter of confirmation of the receipt of the samples (Annex 5). The participants were given 1.5 month to complete the work. 5. Test materials Samples of spiked LDP at one concentration level and PPPO at three different concentration levels of five model substances were prepared by the URL-FCM as presented in Table 1. To perform the migration tests the 5 Petri dishes, 5 glass o-rings and 5 filters were sent to the laboratories by the URL. The standard of DINCH (CAS: ) was sent by the URL to ensure the same representative mixture of the isomers. Table1. Samples for ILC 11 Name Sample TNX1 One bottle with 1 g of the spiked PPPO at level 1 TNX One bottle with 1 g of the spiked PPPO at level TNX3 One bottle with 1 g of the spiked PPPO at level 3 MTNX 4 LDP discs for migration tests Blank TNX One bottle with 5 g of clean PPPO 5.1 Preparation Preparation and homogenization of the test materials were done by the URL-FCM laboratory according to the procedure described in Annex 6. After spiking and homogenization the PPPO were divided into glass amber bottles and the LDP foil were wrapped into aluminium foil and hermetically closed together in plastic bag. 5. Homogeneity assessment The samples were tested for homogeneity by the URL. Ten randomly selected test specimens for each sample (TNX1, TNX, and TNX3 and MTNX) were analysed in duplicate for butyl hydroxytoluene, benzophenone, diisobutylphthalate, diethylhexyladipate and 1,-cyclohexanedicarboxylic acid diisononyl ester using the method description. Homogeneity was evaluated by the ProLab Software [1] according to IUPAC International Harmonised Protocol [9], F-test and to the method proposed in the ISO 1358 [4]. The results together with their statistical evaluation are given in Annex 8. All test materials showed sufficient homogeneity

9 5.3 Stability test Randomly selected specimens for each sample (TNX1, TNX, and TNX3) were stored at 3 different temperature conditions (4ºC, C and 4ºC). The test samples were monitored for stability by the URL laboratory for all model substances from the date of preparation and homogeneity assessment to the closing of the ILC. The stability was evaluated as described in ISO GUID 35:6 [11]. The evaluation of data was carried out by performing a linear regression on the determined concentrations of BHT, BP, DiBP, DHA and DINCH (mean values) vs time. For a stable material it is expected that the intercept is (within uncertainty) equal to the assigned value, whereas the slope does not differ significantly from zero. Using the linear regression equation: Y (BHT, BP, DiBP, DHA and DINCH mg/kg) = b +b 1 X (time, weeks) the slope is not significantly different from zero if the following requirement is respected; where b1 is the slope obtained from the linear regression, t.95,n- is the Student s t-factor for n- degrees of freedom and p =.95 (95% level of confidence) and s(b 1 ) is the uncertainty associated with the slope. This can be calculated as follows: The value of s (standard deviation of the time-points) can be obtained from: where n is the number of points of the linear regression. The results together with their statistical evaluation are given in Annex 9 It should be pointed out that no significant trend was observed for the test samples at all temperature conditions (4ºC, C and 4ºC) for the time of the ILC. It was concluded that the stability of the samples was adequate for at least two months following their preparation

10 5.4 Distribution The sample kits were dispatched to the participants by the URL-FCM. ach participant received a box containing: a box containing the test materials: TNX1, TNX, TNX3 Three amber glass bottles with 1 g of the spiked PPPO (three different concentration level for each compound); MTNX Amber bottle with 5 g of the clean PPPO for the migration tests and calibration curves; 4 LDP discs for migration tests; 5 Petri dishes + 5 glass o-rings - to perform the migration tests (1 days 6 C); 5 filters to extract the PPPO ; 1 vial with Cyclohexane dicarboxylic acid diisononyl ester (DINCH): CAS: to use as standard in the exercise; The letters: an accompanying letter with instructions on sample handling (Annex 4); a form to sent back after receipt of the samples to confirm their arrival (Annex 5); a form for reporting results when not possible to report in the software form (Annex 7). 6. Instructions to participants Instructions were given to all participants in the letters that accompanied the samples (Annex 4). Laboratories were asked to report 4 results for each concentration level and 4 results for the migration experiment. Participants were asked to follow the SOP provided and to avoid / report any eventual deviation. The results had to be reported using the unit of measure indicated in the instruction letter. The results were reported in a special ProLab software [1] form or in a form sent to the participants (Annex 7). 7. Approach for statistical evaluation of results 7.1. valuation of the method performance characteristics consensus value, repeatability ( r ) and reproducibility ( R ) standard deviation The statistical evaluation of the results was performed using the ProLab software [1] applying different algorithms for the determination of the consensus value, repeatability (r) and reproducibility (R) standard deviation. ISO 575- [1] is a standard approach for the statistical analysis of method validation interlaboratory studies. It uses specific assumptions as follows: all laboratories (apart from only a very few outlier laboratories) must have equal analytical performance in order to guarantee that distribution of the test results is close to the normal distribution (this assumption cannot be made for proficiency testing); all laboratories must use the same analytical method; the method requires replicates

11 The standard ISO 575- applies the Grubbs test for the outlier identification of individual test results and laboratory mean values. Additionally tests for the identification of exceeding intra laboratory standard deviations are applied (Cochran test and F-test, respectively). It is common when analysing data from precision experiments to find data that are borderline between stragglers and outliers, thus requiring judgments that may affect the results. The robust methods described in ISO [13] allow to assign less weight to extreme results, rather than eliminate them from the data set. In this case it is not required to use decisions potentially affecting the results of the calculations. The algorithm of ISO (Algorithm A+S) is similar to one in ISO 1358 [4]. Where results are normally distributed the statistics in ISO 575- give results that are very similar to robust statistics in or Q/Hampel algorithm in DIN 384 A45 [14]. 7.. Identification of modes using Kernel density plotting Kernel density plots were additionally used to identify multi modality in the distributions of the values and identify outlying results. Analytical results from a collaborative study are not always normally distributed or can contain values from different populations giving rise to multiple distribution modes. These modes can be visualised by using Kernel density plots [15, 16]. In case the results are not normally distributed the classical statistics from ISO 575- should not be applied. Kernel density plots were computed by the ProLab software from the analytical results by representing the individual numeric values each as a normalised Gaussian distribution centred on the respective analytical value. The sum of these normal distributions formed then the Kernel density distribution Mandel s h- and k-statistics Mandel's h-statistic and Mandel's k-statistic [1] represent measurements for graphically surveying the consistency of the data. They are helpful for method and laboratory assessment. Mandel's h-statistic address whether there are differences between the mean values of the laboratories. Mandel's k-statistic is useful to assess the variance of each laboratory compared to the variances of the other laboratories. Mandel s h- and k- values were calculated by ProLab software following ISO 575. The examination of the plots of Mandel's h- and k-statistics may indicate that specific laboratories exhibit patterns of results that are markedly different from the others. This is indicated by (compared to the other laboratories) consistently high or low variation and/or extreme (high or low) mean values. Various patterns can appear in the plot of Mandel's h-statistic. All laboratories can have both positive and negative values. Individual laboratories may tend to give either all positive or all negative values. If one laboratory stands out on the k-statistic as having many large values, the respective laboratory has a poorer repeatability precision than the other laboratories. A laboratory could give rise to consistently small k-values because of such factors as excessive rounding of its data or an insensitive measurement scale

12 7.4. valuation criteria for laboratory performance z-scores Individual laboratory performance was expressed in terms of z and z -scores in accordance with ISO 1358 and the International Harmonised Protocol: ( x lab X assigned ) z = σ p z' = σ ( x p lab X ) + u assigned Where: x lab X assigned σ p u assigned is the measurement result reported by a participant; is the assigned value; is the target standard deviation for proficiency assessment; is the standard uncertainty of the assigned value. The z- and z -scores can be interpreted as follow: z < z 3 z >3 satisfactory result; questionable result; unsatisfactory result. The z-scores compared the participant's deviation from the assigned value with target standard deviation accepted for the interlaboratory comparison σ p. z -scores can be used when the assigned value is not calculated using the results reported by the participants. z -score takes in consideration the uncertainty of the assigned values. In case the guidelines for limiting the uncertainty of the assigned value u assigned <.3 σ p [4] are met, then z-scores will be similar to z -scores. The present ILC uses the mean values of all results reported by the participant as an assigned value. Therefore the z -scores could not be used for assessment of the laboratory performance. For results reported as "smaller than" (<-values), the reported value was not used in any calculations and no evaluation of the measurement results was made. No scores were given. 8. Results and conclusions 8.1 Preliminary considerations There were 17 participants from 15 countries to whom samples were dispatched. The ILC was closed permanently in the end of January 1 for statistical interpretation

13 Sixteen laboratories submitted results. From the URL-NRL network 14 laboratories out of 15 reported results. There were guests from Germany that provided results as well. As requested, most of the laboratories reported four replicate results under repeatability conditions. The participation of the laboratories was regarded as satisfactory for the aim of the precision experiment with regards of the numbers of received results. Absolute minimum of participating laboratories for conducting a precision experiment is 8. Since the aim of the ILC was evaluate the laboratory performance and precision criteria of the harmonised method there was a communication to the participants that different techniques were not fit to the scope of this ILC. Fifteen participants followed strictly the SOP and one laboratory used a minor modification which did not influence the results, so all the data were evaluated together. 8. Method performance characteristics from the precision experiment The mean values, reproducibility and repeatability standard deviations were calculated according to 3 different algorithms: DIN 384 A45 Q-Hampel algorithm, robust ISO and classical ISO There are shown in Tables -4. In DIN 384 A45 Q-Hampel algorithm and robust ISO all the data are taken for the calculations as foreseen. In the ISO 575- algorithm, the Grubs and Cochran test showed a few outliers and stragglers. The outliers were taken out from further statistical elaboration performed with that method. Their number and type per samples are shown in Table 5. Table.: Assigned values and repeatability/reproducibility S.D. calculated by DIN 384 A45 algorithm Sample Compound TNX1 TNX TNX3 MTNX Assigned value [mg/kg] Reprod. S.D. [mg/kg] Rel. reprod. S.D. [%] Repeat.S.D. [mg/kg] Rel.repeat. S.D. [%] DIN 384 A45 BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH

14 Table 3: Assigned values and repeatability/reproducibility S.D. calculated by ISO (A+S) algorithm Sample Compound TNX1 TNX TNX3 MTNX Assigned value [mg/kg] Reprod. S.D. [mg/kg] Rel. reprod. S.D. [%] Repeat.S.D. [mg/kg] Rel.repeat. S.D. [%] ISO (Algorithm A+S) BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH Table 4: Assigned values and repeatability/reproducibility S.D. calculated by ISO 575 algorithm Sample Compound TNX1 TNX TNX3 MTNX Assigned value [mg/kg] Reprod. S.D. [mg/kg] Rel. reprod. S.D. [%] Repeat.S.D. [mg/kg] Rel.repeat. S.D. [%] ISO 575- BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH BHT BP DiBP DHA DINCH

15 Table 5: Number of Grubs and Cochran outliers for all the samples Sample Compound Grubs outliers (mean value) TNX1 TNX TNX3 MTNX Cochran outliers (standard deviation) BHT - LC56, LC9 - BP - LC34 - DiBP LC8 - - DHA - LC41 - Grubs + Cochran outliers DINCH LC5 - - BHT - LC56, LC9 - BP DiBP LC8 - - DHA LC5 - - DINCH LC56, LC5 - - BHT LC5 LC9 - BP DiBP LC8 - - DHA DINCH LC5 - - BHT - LC13 - BP - LC13 - DiBP - - DHA DINCH - LC5, LC13, - Mandel's h- and Mandel's k-statistics are shown in the following pages as calculated according to ISO 575 (Figures 1-5). Values differing statistically significantly from values of other laboratories are marked in a different colour. For a 1% significant level the indicative Mandel s h value is.33 and k-value (for n=4 replicates) is 1.88 Laboratories with higher values are marked in red. For a 5% significant level the indicative Mandel s h value is 1.86 and k-value (for n=4 replicates) is Laboratories with higher values are marked in yellow. The outcome of the Mandel's h- and Mandel's k-statistics were similar to and in correspondence with the Grubs and Cochran outliers tests according to ISO The calculated mean/assigned values from the three different algorithms gave similar results. Slight differences could be observed in the estimation of the repeatability/ reproducibility standard deviations. To allocate less weight to extreme results rather than eliminate them from the data set, the robust method ISO was chosen for further calculations. The experiments conducted with BHT, BP, DiBP, DHA and DINCH in PPPO allowed to derive precision parameters (Table 3). Factors that were relevant in the values obtained for repeatability and reproducibility standard deviation were the sample preparation procedure, which included two extraction steps with filtration and evaporation which could influence the reproducibility of the method. Another factor could be the difficulty to deal with the model analytes (e.g. DiBP could cause blank problems when analysed at low concentrations, while DINCH was a mixture of isomers). It should be also pointed out that it was the first exercise organised using PPPO as simulant which is a complex porous polymer difficult to manage

16 3.6 Mandel's h statistics for BHT-butylated hyrdotoluene Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC Mandel's k statistics for BHT-butylated hyrdotoluene Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX1 LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Figure 1: Mandel s h and k statistics for BHT

17 Mandel's h statistics for benzophenone. 1.8 Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Mandel's k statistics for benzophenone Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Figure : Mandel s h and k statistics for BP

18 Mandel's h statistics for diisobuthylphthalate Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX1 LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Mandel's k statistics for diisobuthylphthalate.3..1 Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Figure 3: Mandel s h and k statistics for DiBP

19 Mandel's h statistics for diethylhexyladipate Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Mandel's k statistics for diethylhexyladipate Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Figure 4: Mandel s h and k statistics for DHA

20 Mandel's h statistics for DINCH Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Mandel's k statistics for DINCH Bar 1: TNX1 Bar : TNX Bar 3: TNX3 Bar 4: MTNX LC5 LC1 LC13 LC16 LC5 LC8 LC31 LC34 LC38 LC41 LC48 LC49 LC5 LC55 LC56 LC9 Figure 5: Mandel s h and k statistics for DINCH - -

21 8.3 z-scores and laboratories performance For calculation of the z-score, the most important decisions are the choice of the assigned value and that of the target standard deviation (SD) against which the performance will be assessed. For the assigned values a robust mean was chosen as a consensus assigned value. For the target SD the choice impacts the correct assignment of satisfactory, questionable and unsatisfactory. While Horwitz SD can be a good compromise, it does not reflect various levels of complexity of analytical methods. Thus in this case it was more adequate to use the reproducibility SD as target SD. In Table 6-1 z-scores calculated against two different target SDs are presented. The z-score calculated by reproducibility SD as target SD were more realistic and were used to assess the laboratories performance. Table 6. Summary of the z scores against different target S.D. for BHT z-scores for BHT classical Horwitz ILC Reproducibility SD TNX1 TNX TNX3 MTNX TNX1 TNX TNX3 MTNX LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC Table 7. Summary of the z scores against different target S.D. for BP z-scores for BP classical Horwitz ILC Reproducibility SD TNX1 TNX TNX3 MTNX TNX1 TNX TNX3 MTNX LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC

22 Table 8. Summary of the z scores against different target S.D. for DiBP z-scores for DiBP classical Horwitz ILC Reproducibility SD TNX1 TNX TNX3 MTNX TNX1 TNX TNX3 MTNX LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC Table 9.Summary of the z scores against different target S.D. for DHA z-scores for DHA classical Horwitz ILC Reproducibility SD TNX1 TNX TNX3 MTNX TNX1 TNX TNX3 MTNX LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC

23 Table 1. Summary of the z scores against different target S.D. for DINCH z-scores for DINCH classical Horwitz ILC Reproducibility SD TNX1 TNX TNX3 MTNX TNX1 TNX TNX3 MTNX LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC LC Five sets of figures for each of the four samples are presented in Figures 6-5. ach set included: individual laboratories values and their mean and standard deviation (a), Kernel density plot (b) and z- scores (c). 8.4 Final conclusion The participation in the ILC was satisfactory for the purpose of the study with 16 out of 17 laboratories which reported results; The ILC on method performance was successful with more than 8 valid results, even taking into account the complexity of the exercise. Considering the lack of the data previously available in the literature, this ILC exercise provided the first successful attempt to derive precision data at U level for a specific migration experiment. It also represented a worst case and was very extensive with 3 levels of concentrations and 1 film, 5 substances, 4 replicates and 16 laboratories, i.e 136 data points. This exercise consequently provided a great breadth of valuable detailed and traceable precision data, which can be used for a more refined validation of the method and proficiency test of official controls planned for

24 Sample: TNX1 Measurand: BHT-butylated hyrdotoluene Method: ISO (Alg. A+S) No. of laboratories: Assigned value:. mg/kg (mpirical value) Rel. target s.d.: 44.6% (mpirical value) Rel. repeatability s.d.: 6.71% Limits of tolerance: mg/kg ( Z score <.) mg/kg.6.4 Assigned value Mean LC5 LC8 LC5 LC31 LC56 LC55 LC34 LC41 LC38 LC13 LC48 LC9 ProLab Assessm ent=iso (Alg. A+S); Assigned value= M ; Target s.d.= S; Z score<= Z score LC5 LC8 LC5 LC31 LC56 LC55 LC34 LC41 LC38 LC13 LC48 LC9 ProLab 1 Figure 6. Summary graphs of the laboratories test results for BHT with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 4 -

25 Sample: TNX1 Measurand: benzophenone Method: ISO (Alg. A+S) No. of laboratories: 16.5 Assigned value:.51 mg/kg (mpirical value) Rel. target s.d.: 1.1% (mpirical value) Rel. repeatability s.d.: 7.5% Limits of tolerance: mg/kg ( Z score <.).4 mg/kg.3 Assigned valu e Mean..1 LC5 LC5 LC56 LC8 LC1 LC48 LC13 LC38 LC5 LC31 LC49 LC9 LC55 LC41 LC34 ProLab 1 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC5 LC56 LC8 LC1 LC48 LC13 LC38 LC5 LC31 LC49 LC9 LC55 LC41 LC34 ProLab 1 Figure 7. Summary graphs of the laboratories test results for BP with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 5 -

26 Sample: TNX1 Measurand: diisobuthylphthalate Method: ISO (Alg. A+S) No. of laboratories: Assigned value:.89 mg/kg (mpirical value) Rel. target s.d.: 39.51% (mpirical value) Rel. repeatability s.d.: 1.14% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean LC5 LC48 LC13 LC31 LC5 LC1 LC9 LC49 LC5 LC38 LC56 LC41 LC34 LC55 LC8 ProLab Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC48 LC13 LC31 LC5 LC1 LC9 LC49 LC5 LC38 LC56 LC41 LC34 LC55 LC8 ProLab 1 Figure 8. Summary graphs of the laboratories test results for DiBP with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 6 -

27 Sample: TNX1 Measurand: diethylhexyladipate Method: ISO (Alg. A+S) No. of laboratories: Assigned value:.868 mg/kg (mpirical value) Rel. target s.d.: 8.98% (mpirical value) Rel. repeatability s.d.: 7.49% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean LC34 LC31 LC8 LC5 LC49 LC38 LC5 LC56 LC9 LC13 LC55 LC1 LC41 LC5 ProLab 1 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC34 LC31 LC8 LC5 LC49 LC38 LC5 LC56 LC9 LC13 LC55 LC1 LC41 LC5 ProLab 1 Figure 9. Summary graphs of the laboratories test results for DHA with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 7 -

28 Sample: TNX1 Measurand: DINCH Method: ISO (Alg. A+S) No. of laboratories: Assigned value: mg/kg (mpirical value) Rel. target s.d.: 4.46% (mpirical value) Rel. repeatability s.d.: 7.38% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean 5.5 LC8 LC5 LC34 LC31 LC9 LC55 LC38 LC48 LC49 LC5 LC1 LC13 LC56 LC41 LC5 ProLab Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC8 LC5 LC34 LC31 LC9 LC55 LC38 LC48 LC49 LC5 LC1 LC13 LC56 LC41 LC5 ProLab 1 Figure 1. Summary graphs of the laboratories test results for DINCH with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 8 -

29 Sample: TNX Measurand: BHT-butylated hyrdotoluene Method: ISO (Alg. A+S) No. of laboratories: Assigned value: mg/kg (mpirical value) Rel. target s.d.: 8.4% (mpirical value) Rel. repeatability s.d.: 8.7% Limits of tolerance: mg/kg ( Z score <.) 8 6 mg/kg 4 Assigned value Mean - LC5 LC8 LC34 LC5 LC38 LC48 LC55 LC5 LC13 LC31 LC41 LC56 LC9 ProLab Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC8 LC34 LC5 LC38 LC48 LC55 LC5 LC13 LC31 LC41 LC56 LC9 ProLab 1 Figure 11. Summary graphs of the laboratories test results for BHT with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 9 -

30 Sample: TNX Measurand: benzophenone Method: ISO (Alg. A+S) No. of laboratories: 16.8 Assigned value:.488 mg/kg (mpirical value) Rel. target s.d.: 19.58% (mpirical value) Rel. repeatability s.d.: 7.8% Limits of tolerance: mg/kg ( Z score <.).7 mg/kg.6.5 Assigned value Mean.4.3. LC5 LC49 LC8 LC5 LC56 LC34 LC13 LC9 LC48 LC5 LC38 LC31 LC1 LC55 LC41 ProLab 1 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC49 LC8 LC5 LC56 LC34 LC13 LC9 LC48 LC5 LC38 LC31 LC1 LC55 LC41 ProLab 1 Figure 1. Summary graphs of the laboratories test results for BP with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 3 -

31 Sample: TNX Measurand: diisobuthylphthalate Method: ISO (Alg. A+S) No. of laboratories: 16 6 Assigned value: mg/kg (mpirical value) Rel. target s.d.: 6.7% (mpirical value) Rel. repeatability s.d.: 9.13% Limits of tolerance: mg/kg ( Z score <.) 5 4 mg/kg 3 Assigned value Mean 1 LC5 LC48 LC13 LC49 LC31 LC5 LC5 LC34 LC1 LC9 LC38 LC41 LC55 LC56 LC8 ProLab 1 Z score Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= LC5 LC48 LC13 LC49 LC31 LC5 LC5 LC34 LC1 LC9 LC38 LC41 LC55 LC56 LC8 ProLab 1 Figure 13. Summary graphs of the laboratories test results for DiBP with their repeatability SD (a), Kernel density plot (b) and z scores (c)

32 Sample: TNX Measurand: diethylhexyladipate Method: ISO (Alg. A+S) No. of laboratories: Assigned value: mg/kg (mpirical value) Rel. target s.d.: 1.3% (mpirical value) Rel. repeatability s.d.: 6.71% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean LC34 LC5 LC48 LC31 LC5 LC13 LC41 LC38 LC55 LC9 LC49 LC8 LC56 LC1 LC5 ProLab Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC34 LC5 LC48 LC31 LC5 LC13 LC41 LC38 LC55 LC9 LC49 LC8 LC56 LC1 LC5 ProLab 1 Figure 14. Summary graphs of the laboratories test results for DHA with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 3 -

33 Sample: TNX Measurand: DINCH Method: ISO (Alg. A+S) No. of laboratories: Assigned value: mg/kg (mpirical value) Rel. target s.d.:.69% (mpirical value) Rel. repeatability s.d.: 5.9% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean LC34 LC5 LC13 LC5 LC55 LC8 LC31 LC41 LC48 LC49 LC9 LC38 LC1 LC56 LC5 ProLab 1 13 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC34 LC5 LC13 LC5 LC55 LC8 LC31 LC41 LC48 LC49 LC9 LC38 LC1 LC56 LC5 ProLab 1 Figure 15. Summary graphs of the laboratories test results for DINCH with their repeatability SD (a), Kernel density plot (b) and z scores (c)

34 Sample: TNX3 Measurand: BHT-butylated hyrdotoluene Method: ISO (Alg. A+S) No. of laboratories: Assigned value: 3.58 mg/kg (mpirical value) Rel. target s.d.: 19.31% (mpirical value) Rel. repeatability s.d.: 1.1% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean.5 LC5 LC8 LC56 LC5 LC55 LC13 LC48 LC38 LC5 LC34 LC41 LC31 LC9 ProLab 1 Z score Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= LC5 LC8 LC56 LC5 LC55 LC13 LC48 LC38 LC5 LC34 LC41 LC31 LC9 ProLab 1 Figure 16. Summary graphs of the laboratories test results for BHT with their repeatability SD (a), Kernel density plot (b) and z scores (c)

35 Sample: TNX3 Measurand: benzophenone Method: ISO (Alg. A+S) No. of laboratories: 16 Assigned value:.941 mg/kg (mpirical value) Rel. target s.d.: 16.59% (mpirical value) Rel. repeatability s.d.: 7.3% Limits of tolerance: mg/kg ( Z score <.) Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= 1.5 mg/kg 1 Assigned value Mean.5 LC49 LC5 LC41 LC8 LC1 LC9 LC38 LC48 LC5 LC55 LC5 LC56 LC34 LC13 LC31 ProLab 1 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC49 LC5 LC41 LC8 LC1 LC9 LC38 LC48 LC5 LC55 LC5 LC56 LC34 LC13 LC31 ProLab 1 Figure 17. Summary graphs of the laboratories test results for BP with their repeatability SD (a), Kernel density plot (b) and z scores (c)

36 Sample: TNX3 Measurand: diisobuthylphthalate Method: ISO (Alg. A+S) No. of laboratories: Assigned value: 1.11 mg/kg (mpirical value) Rel. target s.d.: 9.89% (mpirical value) Rel. repeatability s.d.: 13.55% Limits of tolerance: mg/kg ( Z score <.).5 mg/kg Assigned value Mean.5 LC5 LC49 LC31 LC1 LC5 LC9 LC5 LC48 LC38 LC13 LC55 LC56 LC41 LC34 LC8 ProLab 1 Z score Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= LC5 LC49 LC31 LC1 LC5 LC9 LC5 LC48 LC38 LC13 LC55 LC56 LC41 LC34 LC8 ProLab 1 Figure 18. Summary graphs of the laboratories test results for DiBP with their repeatability SD (a), Kernel density plot (b) and z scores (c)

37 Sample: TNX3 Measurand: diethylhexyladipate Method: ISO (Alg. A+S) No. of laboratories: 16 6 Assigned value: mg/kg (mpirical value) Rel. target s.d.: 16.8% (mpirical value) Rel. repeatability s.d.: 7.35% Limits of tolerance: mg/kg ( Z score <.) mg/kg 4 35 Assigned value Mean LC41 LC5 LC34 LC31 LC5 LC49 LC55 LC13 LC9 LC48 LC38 LC56 LC8 LC1 LC5 ProLab 1 Z score LC41 LC5 LC34 LC31 LC5 LC49 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= LC55 LC13 LC9 LC48 LC38 LC56 LC8 LC1 LC5 ProLab 1 Figure 19. Summary graphs of the laboratories test results for DHA with their repeatability SD (a), Kernel density plot (b) and z scores (c)

38 Sample: TNX3 Measurand: DINCH Method: ISO (Alg. A+S) No. of laboratories: 16 Assigned value: mg/kg (mpirical value) Rel. target s.d.: 15.96% (mpirical value) Rel. repeatability s.d.: 7.69% Limits of tolerance: mg/kg ( Z score <.) 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= mg/kg Assigned value Mean LC13 LC34 LC55 LC41 LC5 LC5 LC9 LC31 LC49 LC8 LC48 LC38 LC1 LC56 LC5 ProLab 1 Z score LC13 LC34 LC55 LC41 LC5 LC5 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= LC9 LC31 LC49 LC8 LC48 LC38 LC1 LC56 LC5 ProLab 1 Figure. Summary graphs of the laboratories test results for DINCH with their repeatability SD (a), Kernel density plot (b) and z scores (c)

39 Sample: MTNX1 Measurand: BHT-butylated hyrdotoluene Method: ISO (Alg. A+S) No. of laboratories: Assigned value: mg/kg (mpirical value) Rel. target s.d.: 41.8% (mpirical value) Rel. repeatability s.d.: 14.5% Limits of tolerance: mg/kg ( Z score <.) mg/kg 5 15 Assigned value Mean 1 5 LC5 LC56 LC34 LC49 LC5 LC55 LC31 LC41 LC5 LC8 LC38 LC48 LC9 LC13 ProLab 1 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC56 LC34 LC49 LC5 LC55 LC31 LC41 LC5 LC8 LC38 LC48 LC9 LC13 ProLab 1 Figure 1. Summary graphs of the laboratories test results for BHT with their repeatability SD (a), Kernel density plot (b) and z scores (c)

40 Sample: MTNX1 Measurand: benzophenone Method: ISO (Alg. A+S) No. of laboratories: 16 4 Assigned value: mg/kg (mpirical value) Rel. target s.d.: 1.84% (mpirical value) Rel. repeatability s.d.: 15.44% Limits of tolerance: mg/kg ( Z score <.) 35 3 mg/kg 5 15 Assigned value Mean 1 5 LC5 LC49 LC55 LC5 LC41 LC56 LC1 LC31 LC8 LC38 LC5 LC34 LC48 LC9 LC13 ProLab 1 4 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC49 LC55 LC5 LC41 LC56 LC1 LC31 LC8 LC38 LC5 LC34 LC48 LC9 LC13 ProLab 1 Figure. Summary graphs of the laboratories test results for BP with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 4 -

41 Sample: MTNX1 Measurand: diisobuthylphthalate Method: ISO (Alg. A+S) No. of laboratories: 16 6 Assigned value: mg/kg (mpirical value) Rel. target s.d.: 5.43% (mpirical value) Rel. repeatability s.d.: 16.45% Limits of tolerance: mg/kg ( Z score <.) 4 mg/kg Assigned value Mean LC5 LC31 LC41 LC49 LC1 LC5 LC56 LC38 LC8 LC55 LC34 LC5 LC13 LC48 LC9 ProLab 1 3 Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC5 LC31 LC41 LC49 LC1 LC5 LC56 LC38 LC8 LC55 LC34 LC5 LC13 LC48 LC9 ProLab 1 Figure 3. Summary graphs of the laboratories test results for DiBP with their repeatability SD (a), Kernel density plot (b) and z scores (c)

42 Sample: MTNX1 Measurand: diethylhexyladipate Method: ISO (Alg. A+S) No. of laboratories: Assigned value: mg/kg (mpirical value) Rel. target s.d.: 34.84% (mpirical value) Rel. repeatability s.d.: 18.6% Limits of tolerance: mg/kg ( Z score <.) 4 35 mg/kg 3 5 Assigned value Mean LC13 LC56 LC31 LC41 LC49 LC5 LC1 LC5 LC55 LC38 LC34 LC48 LC9 LC8 LC5 ProLab Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC13 LC56 LC31 LC41 LC49 LC5 LC1 LC5 LC55 LC38 LC34 LC48 LC9 LC8 LC5 ProLab 1 Figure 4. Summary graphs of the laboratories test results for DHA with their repeatability SD (a), Kernel density plot (b) and z scores (c) - 4 -

43 Sample: MTNX1 Measurand: DINCH Method: ISO (Alg. A+S) No. of laboratories: 16 3 Assigned value: 7.45 mg/kg (mpirical value) Rel. target s.d.: 6.67% (mpirical value) Rel. repeatability s.d.: 15.74% Limits of tolerance: mg/kg ( Z score <.) mg/kg Assigned value Mean 5 LC56 LC31 LC41 LC5 LC55 LC38 LC49 LC8 LC5 LC34 LC13 LC48 LC1 LC9 LC5 ProLab Assessment=ISO (Alg. A+S); Assigned value= M; Target s.d.= S; Z score<= Z score LC56 LC31 LC41 LC5 LC55 LC38 LC49 LC8 LC5 LC34 LC13 LC48 LC1 LC9 LC5 ProLab 1 Figure 5. Summary graphs of the laboratories test results for DINCH with their repeatability SD (a), Kernel density plot (b) and z scores (c)

44 9. Acknowledgements The laboratories who participated in this exercise - listed below - are kindly acknowledged. NRLs CZCH DNMARK STONIA FINLAND FRANC GRMANY IRLAND LUXMBOURG POLAND PORTUGAL SLOVNIA SPAIN TH NTHRLANDS UNITD KINGDOM NIPH- NRL for Food Contact Materials and for Articles for children under 3 years old, National Institute of Public Health (SZU ) Department of Food Chemistry, National Food Institute Technical University of Denmark Health Protection Inspectorate - Central of Chemistry Finnish Customs SCL Laboratoire de Bordeaux-Pessac Bundesinstitut für Risikobewertung (BFR) (Federal Institute for Risk Assessment) Public Analyst - Sir Patrick Duns Hospital Laboratoire National de Santé, Division du Controle des Denrées Alimentaires of Department of Food and Consumer Articles Research, National Institute of Hygene SB-S (Portuguese Catholic University - Biotechnology College Packaging Department) National Institute of Public Health of Republic of Slovenia, Department of Sanitary Chemistry Centro Nacional de Alimentación, Agencia spanola de Seguridad Alimentaria y Nutrición (ASAN) Food and Consumer Product Safety Authority (VWA), Ministry of conomic Affairs, Agriculture and Innovation Food nvironment Research Agency (FRA) GUSTs GRMANY GRMANY LAV Sachsen-Anhalt, Fachbereich Lebensmittelsicherheit, Halle Bayerisches Landesamt für Gesundheit und Lebensmittelsicherheit, rlangen 1. References [1] Regulation C1/11 of the uropean Parliament and the Council of 14 January 11 on plastic materials and articles intended to come into contact with food. [] CN : , Materials and articles in contact with foodstuffs Plastics, Part 13 Test methods for overall migration at high temperature, December [3] CN 14338:3, Paper and board intended to come into contact with foodstuffs Conditions for determination of migration from paper and board using modified polyphenylene oxide (MPPO) as simulant, March 4 [4] ISO 1358:5; Statistical Methods for Use in Proficiency Testing by Interlaboratory Comparisons [5] Regulation (C) No 88/4 of the uropean Parliament and of the Council of 9 April 4 on official controls performed to ensure the verification of compliance with feed and food law, animal health and animal welfare rules [6] /7/C Commission Directive of 6 August relating to plastic materials and articles intended to come into contact with foodstuffs. (Plastics: Unofficial consolidated version

45 including /7/C, 4/1/C, 4/19/C, 5/79/C, 7/19/C, 8/39/C) [7] M. Thompson, Analyst, (), 15, [8] T. Linsinger et al., Accreditation and Quality Assurance in Analytical Chemistry (1), 6, 5 [9] The International Harmonised Protocol for the Proficiency Testing of Analytical Chemistry Laboratories by M. Thompson et al., Pure and Applied Chemistry (6), 78, [1] ProLab Software QuoData, Drezden [11] ISO GUID 35:6; Reference materials General and statistical principles for certification. [1] ISO 575 :1994 () Accuracy (trueness and precision) of measurement methods and results Part : Basic method for the determination of repeatability and reproducibility of a standard measurement method [13] ISO Accuracy (trueness and precision) of measurement method and results. Part 5: Alternative methods for the determination of the precision of a standard measurement method. [14] DIN 384 A45 Ringversuche zur externen Qualitätskontrolle von Laboratorien. [15] AMC, Representing data distributions with kernel density estimates. AMC Technical Brief, 6, pdf. [16] P.J. Lowthian and M. Thompson, Bump Hunting for the proficiency tester searching for multimodality. The Analyst,. 17: p. 1359, Annexes Annex 1. Standard Operating Procedure: Determination of buthyl hydroxytoluene, benzophenone, bis(-ethylhexyl)adipade, diisobutylphtalate and 1,- cyclohexanedicarboxylic acid diisononyl ester in PPPO Annex. Invitation letter to laboratories ILC1 11 Annex 3. Format for confirmation of participation to ILC1 11 Annex 4. Letters accompanying the sample ILC1 11 Annex 5. Letter of confirmation of receipt of ILC1 11 Annex 6. Annex 7. Annex 8. Annex 9. Procedure for the preparation of the spiked PPPO and LDP foil Form for the compilation of the results in non-electronic format Results of the homogeneity study Results of the stability study

46 Annex 1. Standard Operating procedure ILC DTRMINATION OF BUTYL HYDROXYTOLUN, BNZOPHNON, BIS(-THYLHXYL)ADIPAD, DIISOBUTYLPHTALAT AND 1,-CYCLOHXANDICARBOXYLIC ACID DIISONONYL STR IN PPPO INTRODUCTION Butyl hydroxytoluene (BHT) is an antioxidant, benzophenone (BP) is a light stabilizer (UV absorber), bis(-ethylhexyl)adipade (DHA), diisobutylphtalate (DiBP) and 1,- cyclohexanedicarboxylic acid diisononyl ester (DINCH) are plasticizers for the manufacture of plastic articles in application area such as food packaging. This standard operation procedure (SOP) represents an analytical method for the determination of buthyl hydroxytoluene, benzophenone, bis(-ethylhexyl)adipade, diisobutylphtalate and 1,-cyclohexanedicarboxylic acid diisononyl ester in PPPO a simulant for fatty and dry foods. 1 SCOP This protocol describes the methods for the cleaning of the PPPO, preparation of migration tests and the extraction procedure with following quantification of buthyl hydroxytoluene, benzophenone, bis(-ethylhexyl)adipade, diisobutylphtalate and 1,- cyclohexanedicarboxylic acid diisononyl ester from the PPPO. The method is appropriate for the quantitative determination of BHT, BP, DHA, DiBP and DINCH in PPPO in approximate analyte concentration range of.5 to 75 mg/kg,.5 to 5 mg/kg, 1 to 5mg/kg,.5 to 5 and from 5 to 1 mg/kg respectively. PRINCIPLS Prior to each using, the PPPO is cleaned with acetone in a Soxhlet apparatus for 6h and than dried in the oven at 16 C for 6h. Migration tests are performed using 1g of the PPPO, which is in one side contact with the spiked LDP foil inside closed Petri dish and incubated for 1 days in 6 C. BHT, BP, DHA, DiBP and DINCH are extracted from the PPPO manually with hexane ( x ml) at room temperature. The extracts are filtrated and dried under nitrogen to around 5 ml. Determination is carried out by means of gas chromatography-mass spectrometry (GC-MS). Quantification is achieved using benzyl butyl phthalate (BBP) as an internal standard

47 3 RAGNTS 3.1 Reference material, reagents and solvents ,6-Di-tert-butyl-4-methylphenol (BHT): Sigma-Aldrich, Cat no. 4.-8, purity 99%, CAS no Benzyl butyl phthalate (BBP): Sigma, Cat. no. P-6699, Purity 98%, CAS: ; Benzophenone (BP): Sigma-Aldrich, Cat no. B-3634, CAS no ; Bis(-ethylhexyl)adipate (DHA): Fluka, Cat. no. 138, Purity 99%, CAS: ; Bis(-methylpropyl)benzene-1,-dicarboxylate (DiBP): Fluka, Cat. no. 813, Purity 98%, CAS: ; Cyclohexane dicarboxylic acid diisononyl ester (DINCH): Industrial supplier, CAS: ; Acetone for HPLC, 99.8% Hexane for HPLC, 97% 3. Solutions 3..1 Standard stock solutions of each compound: buthyl hydroxytoluene, benzophenone, bis(-ethylhexyl)adipade, diisobutylphtalate and 1,-cyclohexanedicarboxylic acid diisononyl ester in hexane with an accurately known concentration of approximately 1 µg ml -1 of each compound. Weight accurately about 1 mg of each compound (3.1.1, ) into a 1 ml volumetric flasks. Fill volumetric flask up to the mark with hexane (3.1.8) and mix. Calculate the exact concentration of the substances in µg/ml. 3.. Intermediate standard solutions of buthyl hydroxytoluene, benzophenone, bis(- ethylhexyl)adipade, diisobutylphtalate and 1,-cyclohexanedicarboxylic acid diisononyl ester in hexane (1 µg/ml) Pipette 1. ml of the standard stock solutions (3..1) into a 1 ml volumetric flasks and fill the flasks up to the mark with hexane. Calculate the exact concentration of the substances in µg/ml Intermediate standard solutions of buthyl hydroxytoluene, benzophenone, bis(- ethylhexyl)adipade, diisobutylphtalate in hexane (1 µg/ml) Pipette 1. ml of the standard stock solutions (3..) into a 1 ml volumetric flasks and fill the flasks up to the mark with hexane. Calculate the exact concentration of the substances in µg/ml Internal standard solution of benzyl butyl phthalate (BBP) in hexane with an accurately known concentration of approximately 1 µg ml

48 Weight accurately about 1 mg of BBP (3.1.) into a 1 ml volumetric flask. Complete the volume with hexane (3.1.8). Calculate the exact concentration of BBP in µg ml PPPO extraction solution of buthyl phthalate (BBP) in hexane with accurately know concentration of approximately.5 µg ml -1. Pipette.5 ml of the standard stock solution (3..4) into a 1 ml volumetric flask and fill the flask up to the mark with hexane. Calculate the exact concentration of BBP in µg ml -1. NOT 1: All the solution shall be storage in refrigerator (app. 4ºC) and protected from light (till 3 months). 4 LABORATORY QUIPMNT Cleaning of the PPPO : 4.1 xtraction thimbles; 4. Apparatus for Soxhlet extraction; Migration tests: 4.3 Calibrated balance accurate to.1 mg; 4.4 Oven or incubator; 4.5 Migration cells: Petri dishes (Duran 11.8; 6 x mm) and glass O-rings; Calibration curve and extraction of the PPPO: 4.6 Syringes or digital pipettes, 5-5 µl; 4.7 Funnels with the frit or Buckner (Ø 45mm); 4.8 Glass microfiber filters (Whatman, GF/A, 47mm Ø circles); 4.9 Nitrogen evaporator; 4.1 Normal laboratory glassware and apparatus. 5 GC-MS apparatus 5.1 Gas chromatograph equipped with an autosampler and in connection with mass selective detector. 5. GC column, capable of delivering reproducible peaks of buthyl hydroxytoluene, benzophenone, bis(-ethylhexyl)adipade, diisobutylphtalate and 1,- cyclohexanedicarboxylic acid diisononyl and benzyl butyl phthalate (BBP) as an internal standard, and capable to separate this peaks from interference peaks

49 originated from samples used. NOT : For guidance, the instrument parameters which are found suitable for the analysis, using the selected column are given in Annex. 6 PROCDUR 6.1 Cleaning of the PPPO procedure NOT 3: Sent PPPO is already cleaned. Put PPPO into the Soxhlet thimble; Add acetone (3.1.7) into the boiling flask (5 ml of Acetone for g of PPPO); Turn on the heater and clean (extract continuously) the PPPO for 6h; After 6h turn off the heating system take out the PPPO from the cartridge into the backer or big Petri dish; Place the covered Petri dish under the fumehood to evaporate the solvent while mixing; Put Petri dish into the oven at 16 C for 6h; After heating, store the PPPO into closed rlenmeyer flask. 6. Migration test sample preparation Place LDP foil into the Petri dish (4.5); Place the glass O-ring in the middle of the LDP foil; Weight 1. g of the clean PPPO and put inside the O-ring; NOT 4: LDP foil, Petri dish, O ring and cleaned PPPO are provided by JRC Petri dish; glass O-ring; 3 LDP foil; 4 PPPO Close the Petri dish; Shake gently the closed Petri dish to cover all the LDP foil by the PPPO Wrap the Petri dish by aluminum foil carefully to keep the LDP foil surface covered by PPPO; Put the Petri dish inside the oven at 6 C for 1 days; After migration test take out the Petri dish from the oven;

50 Transfer carefully the PPPO into 4 ml vial or the rlenmeyer flask using funnel; 6.3 xtraction of the PPPO Transfer 1. g of the PPPO (or all the PPPO from the Petri dish after migration) into the 4 ml vial; Add ml of extract solution (3..5); Shake manually for 1 min; Leave it for 5 min without shaking; Insert the funnel with the filter into a new 4mL vial; Decant the hexane from the PPPO through the filter into the new vial; Repeat this extraction procedure using once again ml of the extract solution (3..5); Collect the extracts in the same 4mL vial; vaporate the hexane extract under the nitrogen to around 5mL; Inject the concentrated extract into GC-MS. 6.4 Preparation of the calibration curve in PPPO for GC-MS analysis Calibration curve for spiked PPPO samples (TNX1, TNX, TNX3): Weight 1. g of clean PPPO in 4 ml vials or rlenmeyer flasks; Spike the PPPO using the syringe or digital pipette to obtained concentrations of BHT from.5 to 1 µg/g, BP and DiBP from.5 to.5 µg/g, DHA from 1 to 5 µg/g and from 5 to 1 µg/g for DINCH (see table below); xtract the PPPO as in 6.3; Use the PPPO without the spiking as a blank and extract it as in 6.3. Concentration level in PPPO [mg/kg] BP PPPO weight Intermediate stock solution of: [µg/ml] BP Spiking volume [µl] BP BHT and DHA DINCH [g] BHT and DHA DINCH BHT and DHA DINCH DiBP DiBP DiBP

51 Calibration curve for PPPO samples after migration tests (MTNX): Weight 1. g of clean PPPO in 4 ml vials or rlenmeyer flasks; Spike the PPPO using the syringe or digital pipette to obtained concentrations of BHT from 1 to 75 µg/g, BP and DiBP from.5 to 5 µg/g, DHA from 1 to 5 µg/g and from 5 to 1 µg/g for DINCH (see table below); xtract the PPPO as in 6.3; Use the PPPO without the spiking as a blank and extract it as in 6.3. Concentration level in PPPO [mg/kg] BP PPPO weight Intermediate stock solution of: [µg/ml] BP Spiking volume [µl] BP BHT and DHA DINCH [g] BHT and DHA DINCH BHT and DHA DINCH DiBP DiBP DiBP GC-MS analysis Before starting measurements, examine the base line stability. Maintain the same operating conditions of the GC-MS system throughout the measurements of all samples and calibration solutions. NOT 5: For guidance, the instrument parameters suitable for analysis using the selected column are given in Annex. 7. CALIBRATION 7.1. Analysis of calibration standard solutions Inject the relevant calibration solutions (6.4). Integrate peaks and measure peak area for BHT, BP, DHA, DiBP, DINCH and BBP as an internal standard. Construct a calibration function by plotting the analytes concentration against the ratio of the peak area of the analytes and the peak area internal standards in the calibration solutions. Calculate the regression parameters and correlation coefficient. The calibration curve shall be linear and the correlation coefficient shall be,99 or better. If either of the two requirements is not met, fresh standard solutions shall be prepared from the original standard solutions. Analysis of the solutions and construction of the calibration graph have to be repeated

52 7. Analysis of samples Inject the sample solutions prepared in 6.3 under the same conditions used for the calibration solutions. Measure the area of the peaks for calculation of the substance concentrations against established in 7.1 regression. Internal standard is used to compensate for the losses of analytes caused by sample handling, adsorption effects etc. 7.3 valuation of data Calculation of the BHT, BP, DHA, DiBP and DINCH in the PPPO sample in mg/kg. Concentration in PPPO [mg/kg] = C/W Where: C is the substance mass in µg derived from the calibration curve, and W is the weight of sample taken in g. 8 CONFIRMATION For confirmation of peak identity the retention time and a ratio between target and qualifier ions shall be taken into account. 9 VALIDATION RSULTS To be included after performing interlaboratory comparison study FCM ILC

53 BHT chemical information Common name: IUPAC name: PM reference number: Butylhydroxytoluene,,6-di-tertbutyl-p-cresol,6-bis(1,1-dimethylethyl)-4- methylphenol 4664 CAS no.: Molecular formula: C 15 H 4 O Molecular weight:.35 Boiling point ( C): 65 Melting point ( C): 7-73 SML value: Water solubility: 3 mg/kg 1.1 mg/l ( C) Log P: 5.31 D structure:

54 BP chemical information Common name: IUPAC name: PM reference number: Benzophenone Methanone, diphenyl 384 CAS no.: Molecular formula: C 13 H 1 O Molecular weight: 18. Boiling point( C): 35.4 C Melting point( C): 47.9 C SML value: Water solubility: Log P:.6 mg/kg Insoluble, 137 mg/l (estimated) 3.18 (estimated) D structure:

55 DiBP chemical information Common name: IUPAC name: DiBP, Diisobutylphtalate Bis(-methylpropyl)benzene-1,- dicarboxylate PM number: reference CAS no.: Molecular formula: C 16 H O 4 Molecular weight: Boiling point ( C): 3 Melting point ( C): -37 SML value: Water solubility:.5 mg/kg mg/l Log P: 4.11 D structure:

56 DHA chemical information Common name: IUPAC name: DHA, diethyladipate Bis(-ethylhexyl)adipate PM reference number: 319 CAS no.: Molecular formula: C H 4 O 4 Molecular weight: Boiling point ( C): 417 Melting point ( C): SML value: Water solubility: 18 mg/kg <.5 mg/l Log P: > D structure:

57 DINCH chemical information Common name: IUPAC name: PM reference number: 1,-cyclohexanedicarboxylic acid, diisononyl 1,-bis(3,5,5-trimethylhexyl) cyclohexane-1,- dicarboxylate 4575 CAS no.: Molecular formula: C 6 H 48 O 4 Molecular weight: Boiling point ( C): Melting point ( C): SML value: Water solubility: Log P: 4-5 C -54 C 6 mg/kg <. mg/l (5 C) 1 (5 C) D structure:

58 GC-MS Conditions (for information) GC conditions System Analytical Column: Oven temperature program: GC Agilent 789A DB-17 HT, 3 m,.5 mm fused silica provided with a.15 μm thick layer of (5%-Phenyl)- methylpolysiloxane 7 C 1 min; C/min 5 C; 1 C/min 3 C 5min Carrier gas Flow rate Helium 1.3 ml/min Injector temperature: 8 C Injection mode: splitless Injection volume: 1µL MS conditions Mass-selective detector MSD Agilent technologies, inert XL I/CI MSD 5975C Transfer line temperature 3 C Source temperature 3 C Solvent delay Detection: 4 minutes SIM mode Compound Target ion Qualifier ion RT (minutes) BHT BP DHA DiBP DINCH BBP

59 Annex. Invitation letter to laboratories

60 Annex 3. Format for confirmation of participation to the ILC - 6 -

61 Annex 4. Letters accompanying the samples

62 Annex 5. Letter of confirmation of receipt - 6 -

63 Annex 6. Procedure for the preparation of the spiked PPPO and LDP foil PPPO: 1. Washing of the 1 L round bottom flask (spiking PPPO): - 1 L mixing round bottom flask was washed in a dish washer, then filled with the deionised water (resistivity 18 5 C) and mixed; - The water was removed; - The mixing bottle was filled with.5 L of pentane (Aldrich, HPLC grade + 99%) and shaked; - Pentane was removed.. Preparation of the spiking solutions: a) Stock solution of BHT, BP, DiBP, DHA and DINCH in hexane (approx. 1 mg/l) - 1 mg of each compound was weighted separately into a 1 ml class A volumetric flask and dissolved in hexane; - the flask was filled up to the mark by hexane. b) spiking solution of BHT, BP, DiBP, DHA and DINCH in pentane (1 level) - 1µl of BHT, 15 µl of BP, 87.5 µl of DiBP, 16 µl of DHA and 4 µl of DINCH (stock solutions) were pipetting into the 1L volumetric flask; - the flask was filled up to the mark by pentane; c) spike solution of BHT, BP, DiBP, DHA and DINCH in pentane ( level) - 15 µl of BHT, 1 µl of BP, 175 µl of DiBP, 63 µl of DHA and 175 µl of DINCH (stock solutions) were pipetting into the 1 L volumetric flask; - the flask was filled up to the mark by pentane; d) stock solution of BHT, BP, DiBP, DHA and DINCH in pentane (3 level) - 1 µl of BHT, 4 µl of BP, 35 µl of DiBP, 16 µl of DHA and 35 µl of DINCH (stock solutions) were pipetting into the 1 L volumetric flask; - the flask was filled up to the mark by pentane. 3. Preparation of the spiking PPPO: - 35 g of the cleaned PPPO was weighted and transferred into the 1 L round bottom flask; - 1 L of the spiking solution in pentane was add to the PPPO; - The flask was rinsed by pentane and all liquid was collected in the round bottom flask; - The flask was closed and the PPPO was mixed continuously for hours; - The flask was open and the PPPO was mixed till the complete evaporation of the pentane. - The same procedure was repeated for the other two spiking levels. LDP foil: 1. Washing of the 1L reaction chamber (spiking the LDP) - 1 L reaction chamber was washed in a dish washer, then filled with the deionised water (resistivity 18 5 C) and mixed; - the water was removed; - the mixing bottle was filled with.5 L of ethanol (Aldrich, %) and shaked; - ethanol was removed.. Preparation of the spiking solution: a) spiking solution of BHT, BP, DiBP, DHA and DINCH in ethanol: g of BHT, 1.5 g of BP,.5 g of DiBP,.5 g of DHA and 1 g of DINCH were weighted and transferred into the 1 L volumetric flask; - the flask was filled up to the mark by ethanol. b) spiking of the LDP foil: - the LDP foil (size: 3x1 cm) was placed into the reaction chamber; - the foil was rolled and did not touch the chamber walls; - to avoid the contact between the foil itself and the chamber walls the glass bars were used; - the chamber was filled by the ethanol spiking solution to cover up the LDP foil; - the reaction chamber was placed into the oven at 6 C for days; - after the exposure time the foil was taken out from the chamber and rinsed carefully with the fresh ethanol; - the foil was dried using paper and cut into the discs

64 Annex 7. Form for the compilation of the results in non-electronic format

65 Annex 8. Results of the homogeneity study Sample Measurand Unit Mean s(analytical) [%] Mode s(target) HORRAT s(target) [%] F-test Check for significant heterogeneity ISO 1358 Check for sufficient homogeneity Harmonised Protocol - test on significant heterogeneity BHT mg/kg HORWITZ OK OK OK TNX1 TNX TNX3 MTNX1 BP mg/kg HORWITZ OK OK OK DiBP mg/kg HORWITZ OK OK OK DHA mg/kg HORWITZ OK OK OK DINCH mg/kg HORWITZ OK OK OK BHT mg/kg HORWITZ OK OK OK BP mg/kg HORWITZ OK OK OK DiBP mg/kg HORWITZ OK OK OK DHA mg/kg HORWITZ OK OK OK DINCH mg/kg HORWITZ OK OK OK BHT mg/kg HORWITZ OK OK OK BP mg/kg HORWITZ OK OK OK DiBP mg/kg HORWITZ OK OK OK DHA mg/kg HORWITZ OK OK OK DINCH mg/kg HORWITZ OK OK OK BHT mg/kg HORWITZ OK OK OK BP mg/kg HORWITZ OK OK OK DiBP mg/kg HORWITZ OK OK OK DHA mg/kg HORWITZ OK OK OK DINCH mg/kg HORWITZ OK OK OK

66 Annex 9. Results of the stability study A) BHT

67 B) BP

68 C) DiBP

69 D) DHA

70 ) DINCH - 7 -

71 F) Stability spiked PPPO level 1 (TNX1) Sample TNX1 Compound BHT BP DiBP DHA DINCH Test condition Intercept (b) Slope (b1) s(b1) t(α=.95,n-) s(b1) b1 < t(α=.95,n- ) s(b1) 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK G) Stability spiked PPPO level (TNX) Sample TNX Compound BHT BP DiBP DHA DINCH Test condition Intercept (b) Slope (b1) s(b1) t(α=.95,n-) s(b1) b1 < t(α=.95,n- ) s(b1) 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK

72 H) Stability spiked PPPO level 3 (TNX3) Sample TNX3 Compound BHT BP DiBP DHA DINCH Test condition Intercept (b) Slope (b1) s(b1) t(α=.95,n- ) s(b1) b1 < t(α=.95,n- ) s(b1) 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK 4ºC OK ºC OK 4ºC OK - 7 -

73 uropean Commission UR 568 Joint Research Centre Institute for Health and Consumer Protection Title: Development of a harmonised method for specific migration into the new simulant for dry foods established in Regulation 1/11; stablishment of precision criteria from an U interlaboratory comparison organised by the URL- Food Contact Materials for the quantification from and migration into poly(,6-diphenyl phenylene oxide) Author(s): Giorgia Beldì, Natalia Jakubowska, Aurèlie Peychès Bach and Catherine Simoneau Luxembourg: Publications Office of the uropean Union 1 7 pp. 1. x 9.7 cm UR Scientific and Technical Research series ISSN (online) ISBN (pdf) doi:1.788/7684 Abstract This report presents the results of the U Interlaboratory comparison (ILC) of the URL-FCM towards validation criteria for the using the new simulant for specific migration into dry foods, poly(,6-diphenyl phenylene oxide), PPPO. Regulation U 1/11 has established PPPO as new simulant for testing compliance of plastics in contact with foods. The enforceability of the Regulation required the validation of a method effectively to measure the quantification of a range of substances to use as models into PPPO as well as the validation of the migration test for long term storage which is typical for dry foods. This was accomplished under the work programme of the URL with 16 volunteer National Reference Laboratories (NRLs). The laboratories performance were successful with more than 8 valid results, even taking into account the complexity of the exercise. Considering the lack of the data previously available in the literature, this interlaboratory comparison exercise provided the first successful attempt to derive precision data at U level for a specific migration experiment. It also represented a worst case and was very extensive with 3 levels of concentrations and 1 film, 5 substances, 4 replicates and 17 labs, i.e 136 data points. This exercise consequently provided a great breadth of valuable detailed and traceable precision data, which could be used for a more refined validation of the method and proficiency test of official controls planned for

74 z LB-NA-568-N-N As the Commission s in-house science service, the Joint Research Centre s mission is to provide U policies with independent, evidence-based scientific and technical support throughout the whole policy cycle. Working in close cooperation with policy Directorates-General, the JRC addresses key societal challenges while stimulating innovation through developing new standards, methods and tools, and sharing and transferring its know-how to the Member States and international community. Key policy areas include: environment and climate change; energy and transport; agriculture and food security; health and consumer protection; information society and digital agenda; safety and security including nuclear; all supported through a cross-cutting and multidisciplinary approach.