CCQM-K55.b (Aldrin)
Final Report: October 2012
CCQM-K55.b Key Comparison on the Characterization of Organic Substances for Chemical Purity
Coordinating laboratory:
Steven Westwood*, Ralf Josephs, Tiphaine Choteau, Adeline Daireaux, Charline Mesquida and Robert Wielgosz Bureau International des Poids et Mesures (BIPM, Coordinating laboratory) Sèvres, France
With contributions from:
Adriana Rosso and Mariana Ruiz de Arechavaleta Instituto Nacional de Tecnologia Industrial (INTI) Buenos Aires, Argentina
Stephen Davies and Hongjie Wang National Measurement Institute (Australia) (NMIA) Pymble, NSW, Australia
Eliane Cristina Pires do Rego, Janaína Marques Rodrigues, Evelyn de Freitas Guimarães, Marcus Vinicius Barreto Sousa, Tânia Maria Monteiro, Laura Alves das Neves Valente, Fernando Gustavo Marques Violante, Renato Rubim Ribeiro Almeida, Maria Cristina Baptista Quaresma and Raquel Nogueira Instituto Nacional de Metrologia, Qualidade e Tecnologia (INMETRO) Xerem, Brazil
Anthony Windust Institute for National Measurement Standards - National Research Council Canada (NRC-INMS) Ottawa, Canada
Xinhua Dai, Xiaomin Li, Wei Zhang, Ming Li, Mingwu Shao, Chao Wei National Institute of Metrology (China) (NIM) Beijing, China
Siu-kay Wong Government Laboratory of Hong Kong SAR (GLHK) Kowloon, Hong Kong SAR
Julie Cabillic, Fanny Gantois Laboratoire National de Métrologie et d’Essais (LNE) Paris, France
Rosemarie Philipp, Dietmar Pfeifer, Sebastian Hein, Urszula-Anna Klyk-Seitz Bundesanstalt für Materialforschung (BAM) Berlin, Germany
Keiichiro Ishikawa National Metrology Institute of Japan (NMIJ) Tsukuba, Japan
CCQM-K55.b Final Report October 2012
Page 1 of 41
Esther Castro and Norma Gonzalez Centro Nacional de Metrologia (CENAM) Querétaro, Mexico
Anatoly Krylov D.I. Mendeleyev Scientific and Research Institute for Metrology (VNIIM) St Petersburg, Russia
Teo Tang Lin and Lee Tong Kooi Health Sciences Authority (HSA) Singapore
M. Fernandes-Whaley, D. Prévoo, M. Archer,R. Visser, N. Nlhapo and B. de Vos National Metrology Institute of South Africa (NMISA) Pretoria, South Africa
Seonghee Ahn Korea Research Institute of Standards and Science (KRISS) Daejeon, South Korea
Preeyaporn Pookrod, Kanjana Wiangnon, Nittaya Sudsiri, Kittiya Muaksang and Chainarong Cherdchu National Institute of Metrology (Thailand) (NIMT) Bangkok, Thailand
Ahmet Ceyhan Gören and Mine Bilsel National Metrology Institute of Turkey (UME) Gebze-Kocaeli, Turkey
Thierry LeGoff LGC Limited (LGC) Teddington, United Kingdom
Dan Bearden, Mary Bedner, David Duewer, Diane Hancock, Brian Lang, Katrice Lippa, Michele Schantz and John Sieber National Institute of Standards and Technology (NIST) Gaithersburg, MD, USA
Comparison coordinator: Steven Westwood, BIPM (steven.westwood@bipm.org)
CCQM-K55.b Final Report October 2012
Page 2 of 41
Introduction
The CCQM-K55 comparison was undertaken by the CCQM Organic Analysis Working Group (OAWG) for National Measurement Institutes (NMIs) and Designated Institutes (DIs) which provide measurement services in organic analysis under the CIPM Mutual Recognition Arrangement (MRA). The ability to perform suitable purity assessment on the materials that an NMI either makes available to external users as pure substance reference materials or that are used by an NMI as their primary calibrators for the assignment of the property values either of solution or matrix reference materials or for their reference measurement services is a core technical competency for the provision of measurement results in organic analysis that are traceable to the SI. The purity property value (generally reported for applications in organic analysis as the mass fractiona of the main component) assigned to the primary calibrator in a measurement hierarchy underpins the traceability chain for all results linked to that calibrator. All NMIs with ongoing programs in organic analysis were encouraged to participate in this series of comparisons.
The comparisons allow NMIs and DIs to demonstrate that their procedure(s) for assignment of a purity property value and its associated uncertainty are fit for purpose for their intended application(s).
Summary of Previous Studies
The CCQM-P20 multi-round pilot study on purity determination was completed prior to the CCQM-K55 comparison. Studies were undertaken on the purity assessment of tributyl tin chloride (CCQM-P20.a), xylene (CCQM-P20.b), atrazine (CCQM-P20.c), chlorpyrifos (CCQM-P20.d), theophylline (CCQM-P20.e)1 and digoxin (CCQM-P20.f)2.
The “mass balance” or “summation of impurities” method for purity assessment, which aims to identify and quantify on a mass fraction (mg/g) basis all the orthogonal classes of impurity present in the material and by subtraction provides a measure of the mass fraction of the main component, was the most widely used approach by participants in the CCQM-P20 pilot studies. However the use of quantitative nuclear magnetic resonance (qNMR) approaches to obtain a direct measure of the content of the main component were also increasingly being used.
The BIPM coordinated the final two rounds of the CCQM-P20 pilot study and developed a “molecular weight v. polarity” model to map the analytical space for comparisons in this area. This model provided the criteria for the selection of the measurands for each of the four consecutive rounds – respectively CCQM-K55.a, CCQM-K55.b, CCQM-K55.c and CCQM-K55.d – that make up the initial CCQM-K55 key comparison. The relation based on this model between the proposed CCQM-K55 comparison materials, potential future measurands and major areas of calibration and measurement capability (CMC) claims for the provision of primary calibrators and calibration solutions for organic analysis currently accepted or under consideration under the CIPM Mutual Recognition Arrangement is shown in Annex A. The OAWG meeting at Sèvres in April 2008 accepted this overall strategy for the comparison as well as the specific measurand, 17β-estradiol, proposed for the first comparison round, CCQM-K55.a. A pilot study, CCQM-P117.a, was undertaken in parallel with the key comparison. The CCQM-K55.a comparison was completed in 2009 and the Final Report was published in September 2012 in Appendix B of the BIPM Key Comparison Database.3
The BIPM proposed aldrin as the measurand for CCQM-K55.b. The proposal was discussed at several OAWG meetings and approved at the April 2009 meeting at Sèvres. The comparison samples were distributed in May 2010. The individual results were communicated to the comparison coordinator in August 2010 and the results were first discussed at the November 2010 meeting of the CCQM OAWG in
a For the purposes of this comparison, the mass fraction of both the main component and associated impurities are expressed in units of mg/g. The upper limit value of 1000 mg/g would correspond to a “100 %” pure material.
CCQM-K55.b Final Report October 2012
Page 3 of 41
Singapore. Further studies were subsequently undertaken to resolve a disparity between the results obtained by mass balance approaches and those obtained by qNMR. This difference, which was finally ascribed to the presence of a significant non-volatile organic impurity (ca. 1% on a relative mass fraction basis) which was not detected by the majority of participants was further discussed at the April 2011 OAWG meeting at Sèvres and an assignment of a KCRV proposal for aldrin in CCQM-K55.b based on combination of separate KCRV estimates for contributing orthogonal impurity classes was accepted.
Aldrin
Aldrin was selected as the measurand for the second round of the comparison because it:
provides an analytical challenge representative of a laboratory’s capability for the purity assignment of organic compounds of medium structural complexity and low polarity (see “How Far The Light Shines” statement);
represents a sector for general CMC claims on the “analysis space” model (Annex A) which is distinct from the area already covered by the CCQM-K55.a measurand, estradiol.
is a chlorinated pesticide for which there are a number of specific CMC claims in Appendix C of the BIPM Key Comparison Database;
is classified as a Persistent Organic Pollutant (POP) and an important analyte in its own right in the areas of environmental and food residue analysis
is safe and stable for transport in the amounts involved for the comparison and was available in sufficient amount to allow the preparation of a relatively large batch of the comparison sample.
The structure of aldrin (1) is shown in Figure 1 with the conventional ring numbering in bold. The structure of other compounds related to aldrin and referred to in this report are given in Annex B.
Cl
Cl
11
H5
H4 H3
6 H7
2 H1 H Cl
12s
12a
HH
8 Cl Cl
9
10
Cl
Figure 1 – Structure and ring-numbering of Aldrin
Pure aldrin is a colourless solid with a reported melting point of 104 °C. It is insoluble in water, moderately soluble in alcohols and generally soluble in non-polar organic solvents. CMC claims for the measurement of aldrin, disseminated as both pure substance and standard solution CRMs, are listed in the BIPM Key Comparison Database (KCDB) Appendix C.
CCQM-K55.b Final Report October 2012
Page 4 of 41
KEY COMPARISON – MATERIAL AND CONDUCT OF STUDY
For the preparation of the study material for the CCQM-K55.b comparison and the parallel pilot study CCQM-P117.b, a bulk sample of technical grade aldrin donated by the NMIA was used as the source material. This material was a dark, aggregated solid. It contained less than 90% aldrin as determined by relative GC-FID response. A number of related structure or residual starting material impurities were present at significant levels in the material and which were identified by GC-MS.
Trial purification of the material by recrystallisation was investigated using various organic solvent combinations. The most suitable solvent for scaling up the production of the source material was determined to be aqueous methanol. Three successive recrystallisations, which included some recombination of crops obtained from earlier recrystallisation rounds, were required to produce a bulk material having aldrin content greater than 95% by relative GC-FID response.
This material was dried under vacuum at room temperature, ground in a planetary ball mill to give a free running grey powder and redried to constant weight under vacuum at 40 °C.
The ground, dried bulk material was subdivided into a batch of 150 individual units given the BIPM identifier OGP.014. Each unit of BIPM OGP.014 contained a minimum of 500 mg of aldrin in a glass storage vial (5 mL capacity) fitted with a rubber insert and crimped with an aluminium cap.
The batch of candidate material vials were evaluated for impurity profile, homogeneity and stability at the BIPM. The mass fraction of aldrin in the comparison material was assessed by the BIPM to be greater than 950 mg/g while the homogeneity and stability of the aldrin and the associated impurity components were suitable for the purposes of the comparison.
A summary of the results for aldrin content and for characterization of the material’s impurity profile reported by the study participants are contained in this report.
“How Far The Light Shines” Statement for CCQM-K55.b
The comparison is intended to demonstrate a laboratory’s performance in determining the mass fraction of the main component in a relatively pure organic material. The measurement results should be indicative of the performance of a laboratory’s measurement capability for the purity assignment of organic compounds of medium structural complexity (relative molecular mass range 300-500) and low polarity (pKOW < -2) where KOW is the octanol-water partition coefficient4 and is representative of compounds for which related structure impurities are generally quantified by capillary gas-phase chromatography. The expected overall outcome of the rounds making up the CCQM-K55 comparison is to evaluate through a series of strategically planned exercises the scope, applicability, limitations and appropriateness of the various procedures used to assign mass fraction property values to organic materials.
CCQM-K55.b Final Report October 2012
Page 5 of 41
Characterisation study The methods used to investigate, assign and confirm the quantitative composition of the CCQM-K55.b candidate material by the BIPM are summarised below. Characterization studies Related structure impurity content was evaluated by:
a. GC-FID b. GC-MS c. LC-UV d. 1H and 13C NMR Water content was evaluated by: a. coulometric Karl Fischer titration with oven transfer of water from the sample b. coulometric Karl Fischer titration using direct addition of the sample c. thermogravimetric analysis (TGA) as a consistency check for the assigned value d. microanalysis (% C,H content) as a consistency check for assigned value Residual solvent content was evaluated by: a. GC-MS by direct injection b. Headspace GC-MS c. 1H NMR d. Thermogravimetric analysis as a consistency check for the assigned value e. Microanalysis (% C,H content) as a consistency check for the assigned value Non-volatile/ inorganics content : a. ICP-MS for common elements (Na, K, Ca, Mg, Si, Fe, Al) b. Microanalysis (% C, H content) as a consistency check for the assigned values Main component (Aldrin) content a. qNMR b. Differential scanning calorimetry (DSC) Homogeneity studies i. Related structure components The homogeneity of minor components related in structure to aldrin were assessed by sampling ten sub-units selected from across the candidate material batch with analysis by GC-FID and separately by LC-UV with detection at 220 nm. The minimum sample size used to prepare each analysis sample was 2.5 mg. ii. Water The homogeneity of the material relative to water content was assessed by coulometric Karl Fischer titration using oven transfer and a minimum sample size of 50 mg per analysis on five subunits representative of the candidate material batch iii. Residual solvent The homogeneity of the material relative to methanol content was assessed by direct injection GCMS analysis using a minimum sample size of 10 mg per analysis on five sub-units representative of the candidate material batch.
CCQM-K55.b Final Report October 2012
Page 6 of 41
iv. Inorganics content Three units selected from across the production batch were analysed by ICP-MS and by elemental microanalysis for carbon and hydrogen. All gave metal content levels below the detection limits (25 ppm) for each element. Results for % C, H content were in accord with the molecular formula of aldrin
v. Aldrin As a consistency check, the homogeneity of the aldrin content in the material was assessed using the ten sub-units selected for the related structure impurity study by the same GC-FID and LC-UV methods developed for the related structure impurity characterisation. In addition a limited qNMR study was undertaken (two samples from two units of CCQM-K55.b), using dimethyl terephthalate as the internal standard, which gave a value of 958 mg/g with standard uncertainty 2 mg/g for the aldrin content of the sample
The uncertainty contributions due to the inhomogeneity of each related structure impurity component were evaluated by ANOVA. This provided an estimate of the variation due to inhomogeneity of each impurity at a stated sampling size both between and within sample units. The uncertainty contributions due to the inhomogeneity of the major related structure components detected by GC-FID (ubb(rel)) were evaluated by ANOVA. This provided an estimate of the variation due to inhomogeneity of related structure impurities at the stated sampling size both between and within sample units. Acceptable uncertainty contributions due to inhomogeneity were observed for each of the resolved impurities present in the sample. Table 1 shows the estimated content, ubb(rel) and ubb(abs) for each of the resolved related sttructure impurities, and a combined value for the overall uncertainty contribution from between unit inhomogeneity (ubb) of the related structure impurities content of the material. This was calculated as 0.42 mg/g by quadratic combination of the absolute inhomogeneity uncertainties for each impurity.
Impurity Isodrin
Content (mg/g) from homogeneity study*
27
ubb(rel) (%) 1.36
ubb(abs) (mg/g) 0.37
Dieldrin
3.0
1.08
0.03
Dechlorane
4.0
5.0
0.20
Minor impurities
1.5
2.0
0.03
Combined related
35.5
structure impurities
1.2
0.42
* From relative GC-FID peak area – uncorrected for differential response factors or other impurities Table 1: Homogeneity assessment for related structure impurities in CCQM-K55.b
Karl Fischer titration was used to assess variations in water content of the material. As the level of water in the material was below the limit of quantification of our method, it was not possible to estimate the uncertainty of between unit inhomogeneity of the water content of the material in a rigorous manner. There was however no evidence of significant inter-sample differences.
CCQM-K55.b Final Report October 2012
Page 7 of 41
The contribution to the overall uncertainty from between unit inhomogeneity of the methanol content of the material was estimated at ubb(MeOH) = 0.1 mg/g from comparison of the within unit and between unit repeatability of the analysis of two 10 mg replicates taken from each of five units of the candidate material. The within unit repeatability (sr) under these conditions was 0.1 mg/g.
Stability studies
An isochronous stability study was performed using a reference storage temperature of -20 °C and test temperatures of 4°C, 22 °C and 40 °C. A set of units from the production batch were stored at each selected temperature over 8 weeks, with units transferred to reference temperature storage at 2-week intervals. Trend analysis of the data obtained by GC-FID analysis of the stability test samples under repeatability conditions indicated no significant change in the relative composition of aldrin or of the related structure components over this time at any of the test temperatures. No significant changes in water content, which in any case were all below the level of quantification of our method, were observed after storage at 4 °C or 22 °C. There was some evidence of minor uptake of water but only after prolonged storage at 40 °C. The methanol content of the material was stable on storage at 4 °C and 22 °C but did decrease significantly after storage beyond 6 weeks at 22 °C or 2 weeks at 40 °C. The effect of storage temperature on methanol content of the comparison material is shown in Figure 2.
4C
MeOH in CCQM-K55.b: Temperature Stability
22 C 40 C 3.0
MeOH content (mg/g)
2.5
2.0
1.5
0
2
4
6
8
Time (weeks)
Figure 2
Effect of storage temperature on methanol content of comparison material
CCQM-K55.b Final Report October 2012
Page 8 of 41
On the basis of these studies it was concluded that for the purposes of the comparison the material was suitably stable for short-term transport at ambient temperature, provided it was not exposed to temperatures significantly in excess of 40 °C, and for longer term storage at 4 °C. To minimise the potential for changes in the material composition due to loss of methanol, participants were instructed to store the material at 4 °C.
Sample distribution
Two units of the study sample, each containing a minimum of 500 mg of material, were distributed to each participant. Participants were asked to return a form acknowledging receipt of the samples and to advise the co-ordinator if any obvious damage had occurred to the vials during shipping. Recipients were asked to confirm that a monitoring strip inside the shipping container had not registered a temperature in excess of 37 °C during the transport process. The monitor strips indicated that the units originally supplied to Singapore (HSA) were exposed to temperatures in excess of 40 °C during shipping. A replacement set was provided in this case which was delivered without incident. There was also a prolonged delay in delivery of the original sample sent to CENAM due to customs clearance problems. A replacement sample set was issued, which was delivered without difficulties. Otherwise all samples were delivered to the comparison participants without incident. Every registered participant in the CCQM-K55.b comparison provided a result for their sample.
Quantities and Units
Participants were required to report the mass fraction of aldrin, the major component of the comparison sample, in one of the two units supplied to them. The additional unit was provided for method development and trial studies. In addition all participants who used a mass balance (summation of impurities) procedure to determine the aldrin content were required to report the combined mass fraction assignment and associated uncertainty for some or all of the following sub-classes of impurity.
i. total related structure organic substances ii. water iii. total residual organic solvent / volatile organic compounds (VOCs) iv. total non-volatile organics & inorganics
Participants were encouraged to also provide mass fraction estimates for the main impurity components they identified in the comparison sample.
CCQM-K55.b Final Report October 2012
Page 9 of 41
Reported Mass Fraction of Aldrin in CCQM-K55.b
The values reported by participants for aldrin content in CCQM-K55.b are given in Table 2 with a summary plot in Figure 3.
Participant
NRC-INMS NMIJ HSA BAM NIMT NMISA CENAM * LGC NIST KRISS UME LNE NMIA BIPM GLHK INTI NIM INMETRO VNIIM
Aldrin
Standard
Coverage
(mg/g) Uncertainty (mg/g) factor
947.8 950.1 950.4 953 958.5 960.2 960.7 960.9 961 961.6 961.7 962.0 962.1 962.6 963.1 963.7 964.6 966.8 972.6
3.0 3.23 1.3 4 1.4 1.9 3.5 1.9 NA 0.9 1.0 0.47 2.4 +0.65 / -0.75 1.7 9.5 1.2 0.57 0.9
2 2 2 2 2 2 2 2 NA 2 2 2 2.07 2 2 2 2 2.45 2
Expanded Uncertainty (mg/g)
6.1 6.5 2.6 8.0 2.8
3.7 7.0 3.8 +1.6 / -1.8 1.8 2.1 0.93 4.9 +1.3 / -1.5 3.4 19 2.4 1.4 1.8
Table 2 : Aldrin content for participants in CCQM-K55.b * Revised value from initially reported result, received prior to circulation of all results
CCQM-K55.b Final Report October 2012
Page 10 of 41
Aldrin (mg/g)
Aldrin content in CCQM-K55.b
1000 990 980 970 960 950 940
VNIIM INMETRO NIM INTI GLHKSAR BIPM NMIA LNE UME KRISS NIST LGC CENAM* NMISA NIMT BAM HSA NMIJ INMS-NRC
Figure 3
Aldrin content reported by participants in CCQM-K55.b (plotted with 95% confidence interval expanded uncertainties) * Revision of initially reported result, received prior to the circulation of all results
Disparity between Mass Balance and qNMR estimates for Aldrin in CCQM-K55.b
Inspection of the reported values for aldrin content in CCQM-K55.b shows that the comparison results can be divided into two main groups – one having a consensus value for aldrin in excess of 960 mg/g and a second where the assigned value is in the vicinity of 950 mg/g. The higher estimate was obtained by participants using a mass balance approach in which hyphenated chromatography - and qNMR in the case of NIST - was used to determine the total related structure impurity levels with contributions from other potential classes of impurities obtained by complimentary techniques. The aldrin content of CCQM-K55.b was then assigned by subtraction of the combined impurity estimate from the theoretical maximum of 1000 mg/g. The group of results with the lower assigned value for aldrin were based on qNMR techniques to directly determine the aldrin content either as the stand-alone approach or undertaken in conjunction with a mass balance approach.
CCQM-K55.b Final Report October 2012
Page 11 of 41
The direct assignment of aldrin content by qNMR was achieved by integration of the signal due to the equivalent olefinic H4- and H5-protons (Fig. 1), which are distinct from interfering signals due to the main impurities - isodrin, dieldrin, dechlorane and methanol. Representative 1H NMR spectra for CCQMK55.b including an expansion of the main region containing signals due to the various impurities are reproduced in Annex C. The NMIJ and the HSA were the only participants to report a mass balance assignment below 955 mg/g. In each case they obtained direct qNMR values for aldrin in the vicinity of 950 mg /g. Their mass balance estimates were in agreement within their stated uncertainties with this value, and were markedly lower than those reported by other participants, because they reported a significant level (ca. 10 mg/g) for nonvolatile organic impurity, which was not identified by other participants. It should be noted that the NRCINMS in CCQM-K55.b and the SIRIM, who participated in the parallel CCQM-P117.b pilot study, both using qNMR as a stand alone technique, also assigned the aldrin content at a level consistent with those reported by the NMIJ and HSA. The situation however was not simply a choice between a low qNMR estimate and (generally) a higher mass balance value for aldrin. The NMIA, BIPM and UME all obtained qNMR assignments for aldrin content that were in excess of 955 mg/g and were consistent within their associated expanded uncertainty estimates with their independent mass balance values. A combined comparison plot of aldrin content obtained by participants by direct qNMR compared with the mass balance value obtained by the same participant is shown in Figure 4. In several cases the qNMR data shown in Figure 4 was not included in the participant’s report but was made available to the study coordinator subsequent to the circulation of the comparison results and the initial discussion at the November 2010 OAWG meeting. The additional information on qNMR conditions and results, with comparison with the mass balance value reported where available, are summarised in Table 3. Note that in Figure 4 results are plotted with their standard uncertainties or the standard deviation of the qNMR estimate rather than the expanded uncertainties as shown in other data graphs.
Figure 4
Comparison of direct qNMR and mass balance values (± uc or std. devn.) provided by the same participant for aldrin content of CCQM-K55.b (*CCQM-P117.b for SIRIM).
CCQM-K55.b Final Report October 2012
Page 12 of 41
Participant
BAM LGC NRC-INMS INMETRO NIST NMIJ UME BIPM NMIA
HSA NIM SIRIM*
Solvent(s)
CDCl3 CDCl3 CDCl3, CD3OD CDCl3 CD3OD CD2Cl2 d6-Acetone d6-DMSO d6-DMSO, CD3OD, CDCl3 CDCl3 CDCl3 CDCl3
Quantification Standard
Benzoic acid Trioxane
Benzoic acid Benzoic acid Benzoic acid Benzoic acid
1,4-DCB DMTP DMTP
Aldrin (mg/g) by Aldrin (mg/g) by “direct” qNMR “mass balance”
944.6
961.4
947 (σ = 2.6)
960.9 (uc = 1.9)
947.8 (uc = 3)
948.5 (σ = 2.1) 966.8 (uc = 0.6)
949 (uc = 0.7)
961 (uc = 0.8)
952.9 (uc = 2.9) 947.2 (uc = 1.1)
957.4 (uc = 0.2) 961.7 (uc = 1.1)
958 (σ = 2)
962.6 (uc = 0.7)
963 (σ = 1.9)
962.1 (uc = 2.4)
Benzoic acid Benzoic acid Chlorpyrifos
950 946 (uc = 3.5) 950.1 (σ = 6.9)
950.4 (uc = 1.3) 964.7 (uc = 1.2)
Table 3 – qNMR and mass balance estimates for aldrin in CCQM-K55.b (1,4-DCB = 1,4-Dichlorobenzene ; DMTP = Dimethyl terephthalate)
At the OAWG discussions in November 2010, the NMIJ presented supporting data from a GC-FID external calibration using a certified sample of aldrin as the primary calibrant. Their value for aldrin in CCQM-K55.b obtained in this manner was consistent with their reported value for the comparison.
There was considerable discussion on possible reasons for the discrepancy between mass balance and qNMR results at the November 2010 OAWG meeting. To attempt to resolve the issue additional studies were requested in the following areas:
homogeneity of aldrin in CCQM-K55.b by qNMR
influence of solvent and internal standard on qNMR result for aldrin in CCQM-K55.b
additional evidence for the presence of a significant non-volatile organic impurity
At the April 2011 OAWG meeting the results of qNMR studies on five samples of CCQM-K55.b selected from across the production batch and undertaken by the NRC-INMS were reported. They prepared triplicate samples from each unit and determined the aldrin content of solutions in CD3OD with benzoic acid as the internal quantification standard. The results obtained, which also include estimates for the associated impurities isodrin and methanol, are summarised in Table 4 and are consistent with an aldrin content of CCQM-K55.b of 950 mg/g with no evidence of significant inhomogeneity between the individual units. In addition, no effect on qNMR values due to use of different solvent (comparing CD3OD with d6-DMSO) or internal standard (benzoic acid v. dimethyl terephthalate) was found.
CCQM-K55.b Final Report October 2012
Page 13 of 41
Bottle 003-A 003-B 003-C 030-A 030-B 030-C 059-A 059-B 059-C 090-A 090-B 090-C 125-A 125-B 125-C
Aldrin 950.5 948.0 951.6 953.0 948.2 951.9 951.9 952.5 952.0 951.7 946.9 952.2 952.8 945.4
Isodrin 25.5 26.3 26.4 26.5 25.5 26.3 26.8 26.4 26.5 26.4 25.2 26.6 26.5 25.9
MeOH 2.2 2.1 2.0 2.0 1.9 2.1 2.0 2.3 1.8 1.9 2.1 2.0 2.0 2.2
Mean 950.6 26.2
2.0
sd
2.5
0.48
0.1
rsd
0.3% 1.8%
5%
Table 4 – qNMR results (mg/g) for homogeneity check of aldrin, isodrin and methanol in CCQM-K55.b (data provided by NRC-INMS for solutions in CDCl3 using benzoic acid as quantification standard)
The NMIJ reported size-exclusion chromatography studies that indicated the presence of high molecular weight impurities equivalent to approximately 1.2% of the peak area response by refractive index detection relative to the response of the aldrin component. Peaks due to higher molecular weight species were not observed when higher purity reference standards of aldrin were analysed under the same conditions. A representative chromatogram is shown in Figure 5.
Dechlorane
Aldrin
“Polymer”
MW: 1000-2600
R Ｉ
UV-VIS (400 nm)
0
5 min 10
Figure 5 Gel permeation chromatogram of CCQM-K55.b (SHODEX KF401HQ column), elution with 100% THF; detection by refractive index (black) and UV (blue) response.
15 - reproduced from data provided by
the NMIJ
CCQM-K55.b Final Report October 2012
Page 14 of 41
The NMIJ also reported that cycles of dissolution and selective precipitation could be used to furnish purified samples of the non-volatile impurity.
After consideration of the additional data the study participants agreed that the presence of a non-volatile organic impurity that was not detected by the majority of mass balance approaches had been established and that this resolved the apparent difference between the aldrin values assigned by the two approaches.
The cases where higher values for aldrin were originally obtained by qNMR were explained as arising from software processing issues – in particular the applicability or otherwise of automated baseline correction algorithms commonly used in processing the raw FID data. The NMIA reported that when they used the automatic baseline correction option with their software they obtained a value for aldrin above 960 mg/g, but when this option was not used the result was 948 mg/g.
No further information is currently available as to the nature and structure of the impurity. Despite being present at appreciable levels it has still not been observed directly by MS or NMR methods and did not perturb the elemental analysis data for CCQM-K55.b significantly. This indicates that it is likely to be a polychlorinated compound structurally related to aldrin or its synthetic precursors. One suggested possibility is a mixture of oligo- or poly-(hexachlorocyclopentadiene), potentially formed as a sideproduct of technical grade aldrin synthesis from hexachlorocyclopentadiene monomer.
Impurity Profile and Key Comparison Reference Values (KCRVs) for Impurity Classes in CCQM-K55.b
All participants in CCQM-K55.b using a mass balance procedure to assign the aldrin content were required to give estimates for the mass fraction of the sub-classes of impurities they quantified to obtain their final aldrin mass fraction estimate. At the April 2011 WG meeting it was agreed that, as was done for CCQM-P20.f and CCQM-K55.a, the comparison coordinator would propose an overall KCRV for the aldrin content of CCQM-K55.b based on the combination of individual KCRVs for the mass fraction of each of the orthogonal classes of impurity in the CCQM-K55.b comparison sample. This required the assignment of separate KCRVs for:
structurally related impurities;
water;
volatile organic solvent;
non-volatiles/inorganics.
i. KCRV for Aldrin-related impurity content The structures of the individual impurities reported by two or more participants are shown, with their associated numbering, in Annex B. The major compounds identified by participants as present at levels above 0.1 mg/g in CCQM-K55.b were, in decreasing order of mass fraction, isodrin (2), dieldrin (3), dechlorane (4), dihydroaldrin (5) and endrin ketone (6). The detection in the sample at low levels (all below 0.1 mg/g) of hexachlorocyclopentadiene (7), octachlorocyclopentene (8) and chlordene (9) was also noted by several participants. A GC-FID chromatogram obtained by NIST using a polar RTX-OP column was highly representative of GC chromatograms obtained by all other participants and shows the relative retention time of each of the main related structure impurities reported by participants and their structural assignment. It is reproduced
CCQM-K55.b Final Report October 2012
Page 15 of 41
in Figure 6 below. It was noted by several participants that dihydroaldrin is not resolved from isodrin if a lower polarity column (eg DB-5) is used.
Figure 6
GC-FID chromatogram for CCQM-K55.b (reproduced from NIST report)
All participants reported and identified isodrin (2) and dieldrin (3) in the study material. The majority also reported the presence either by GC or LC or both of an impurity with a long retention time relative to aldrin, although only five participants identified it as dechlorane (4).
Many participants reported a peak corresponding to dihydroaldrin (5), eluting between aldrin and isodrin on GC using a DB-17 column or equivalent and at slightly longer retention time than aldrin by LC on reverse-phase columns. While several participants tentatively identified it as dihydroaldrin based on its mass spectrum, the NIST were the sole participant to establish the structure assignment by comparison with an independently synthesised sample. Although the reporting requirements only required a value for total related structure impurities, all but one participant also provided estimates for the major individual impurities. The reported values for individual and combined related structure impurities are listed below in Table 5.
CCQM-K55.b Final Report October 2012
Page 16 of 41
Participant INTI VNIIM NIM INMETRO * UME LNE BIPM HSA NIMT NMIA GLHKSAR KRISS NIST BAM LGC NMISA NMIJ CENAM
Isodrin 21.8 21.3 26.47 * 25.62 28.38 25.5 25.9 28.98 26.35 26.71 27 26.88 26.1 28.08 27.77 26.6 31.9
Related substance impurities (mg/g)
Dieldrin
Dechlorane
2.62
5.2
3.4
*
*
2.94
2.68
3.53
2.03
4.1
4.5
2.9
4.4
3.72
2.1
3.16
3.48
3.39
1.44
3.15
3.64
3.19
4.13
3.2
3.7
4.08
3.05
3.46
4.19
3.46
4.81
3.6
2.65
Total 24.4 27.3 32.5 32.7 32.85 33.94 34.4 34.6 34.8 34.8 35.3 35.5 36.1 36.2 36.45 36.9 37 38.9
uc 0.5 0.9 1.1 0.57 0.047 0.46 0.65 0.72 1.38 0.32 1.69 0.11 0.8 0.05 1.83 0.93 0.5 2.6
Table 5 – Estimates of total and individual aldrin-related impurities in CCQM-K55.b * Presence of isodrin, dieldrin and dechlorane reported without individual estimates
For assignment of a KCRV for total organic impurities, it is proposed to use the values reported by the fifteen participants who observed and included all three major impurities (isodrin, diledrin and dechlorane) in their total value and exclude the three results that did not report the presence of the long retention time impurity due to dechlorane.
The mean of the selected results was chosen as the estimate of the KCRV for related structure impurity content (KCRVRel Subst.). The associated standard uncertainty of the KCRV (uRel Subst.) is the standard deviation of the mean of the selected data set.
KCRVRel Subst. =
uRel Subst
.=
35.4 mg/g; 0.42 mg/g
The results reported by participants with their associated standard uncertainties (k = 1) plotted against the proposed KCRVRel Subst. are shown in Figure 7.
CCQM-K55.b Final Report October 2012
Page 17 of 41
50
Total Related Substances (mg/g)
40
30
20 INTI VNIIM
NIINMMETRO UME
LNE BIPM
HSA NIMT NMGILAHKSAR KRISS NIST BAM
LGCNMISA NMIJCENAM
Figure 7
Estimates of total related structure impurities in CCQM-K55.b plotted with their reported
standard uncertainties ( uc , k = 1). The proposed KCRV for this impurity class (KCRVRel. Subst.) (solid red line) is 35.4 mg/g. Dashed red lines show the KCRV uRel. Subst. (k = 1).
ii. KCRV for water content in CCQM-K55.b
The values for water content in CCQM-K55.b provided by the participants are set out below in Table 6, along with a short description of the variant of the Karl Fischer titration method used.
Participant VNIIM NMIJ KRISS CENAM HSA
BAM NIST GLHK LGC BIPM INMETRO
Water content (mg/g) U95% (mg/g)
Karl Fischer Method outline
0.11
0.02
Direct addition of solution, 3 x 100 mg
0.31
0.04
Direct addition of solid, 3 x 55 mg
0.32
0.14
Direct addition of solid, 3 x 20-30 mg
0.4
0.1
Direct addition of solid, 2 x 100 mg
0.4
0.28
Direct addition of solid, 5 x 8-20 mg
Oven transfer @ 170 C
0.42
0.09
Oven transfer @ 130 C, 3 x 100-150 mg
0.44
-0.09/+ 0.17 Direct addition (of soln. ?), 5 x 20-50 mg
0.46
0.1
Oven transfer @ 160 & 230 C, 50 mg
0.47
0.98
Oven transfer @ 100 C, 3 x ?? mg
0.5
- 0 /+ 0.58 Oven transfer @ 120 C, 3 x 50 mg
0.54
0.026 Direct addition of solution, 2 x 100 mg
Table 6 – Estimates of water content for CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 18 of 41
Participant Water content (mg/g) U95% (mg/g)
Karl Fischer Method outline
NMISA
0.6
0.22
Direct addition of solid, 3 x 50 mg
NMIA
0.91
3.6
Direct addition of solid, 3 x 20-30 mg
NRC-INMS
1
- 1 /+ 0
NIM
1.24
0.36
Direct addition of solid, 3 x 50 mg
LNE
1.31
0.07
Oven transfer @ 100 C, 3 x 100 mg
UME
2
0.12
Direct addition of solution, 1 x 50 mg
NIMT
6.43
0.4
Oven transfer @ 120 C, 5 x 60 mg
INTI
11.9
0.2
Direct addition of solid, 25 mg
Table 6 (ctd) – Estimates of water content for CCQM-K55.b
All participants used coulometric Karl Fischer titration as the minimum method to obtain the water
content estimate. A mixture of procedures including direct addition into the titration cell as a solid, direct
addition as a solution and sample oven transfer to deliver the water content from a sample of
CCQM-K55.b into the titration cell were used. There was generally good agreement for the presence of a
small but detectable amount of water in the material. No ready explanation is available for the two results
that were appreciably in excess of the median value of 0.5 mg/g. Several participants used other
techniques (TGA, qNMR, elemental analysis) as confirmatory analyses or consistency checks.
Because of the asymmetric spread of results, the median of the combined results for water is proposed as
the KCRV for water content (KCRVH2O) in CCQM-K55.b. The associated standard uncertainty of the
KCRV (uH2O) is assigned as the robust statistical estimate of the standard deviation of the median
(MADe/√n) of the data set
KCRVH2O
= 0.47 mg/g;
uH2O
.= 0.05 mg/g
The results reported by participants with their associated standard uncertainties (k = 1) plotted against the
proposed KCRVH2O are shown in Figure 8.
Figure 8 Estimates for water in CCQM-K55.b plotted with reported standard uncertainties (k = 1). The proposed KCRV for water content of CCQM-K55.b (solid red line) is 0.47 mg/g. Dashed red lines show the proposed KCRVH2O uH2O (k = 1).
CCQM-K55.b Final Report October 2012
Page 19 of 41
iii. KCRV for VOCs in CCQM-K55.b
Fourteen of the eighteen participants provided estimates for the volatile organics content of CCQM-K55.b. Each participant who investigated residual solvent content reported a significant level of methanol in the sample. GC approaches with detection by headspace and direct MS analysis (GC-MS) or headspace FID (, in several cases independently cross checked by NMR methods, were used to identify and quantify the methanol content of the sample. The number of participants now incorporating rigorous methods to detect residual solvent was encouraging and a significant increase compared with the earlier CCQM-P20 and CCQM-K55.a comparisons. Given the volatile nature of methanol, which makes it a challenging compound to quantify, the overall agreement with the proposed KCRV was also encouraging. HSA were the only participant to report a methanol content significantly higher than the values found by other participants. They also commented that the samples they received displayed a significant level of inhomogeneity. No comments on the variability or instability in the methanol content of the sample were received from other participants. The results obtained by the study participants of CCQM-K55.b are listed in Table 7 and shown graphically in Figure 9.
Participant
GLHK NIM NRC-INMS
BAM LGC NMIA NIST NMISA KRISS BIPM LNE UME NMIJ HSA
Methanol content (mg/g)
0.9 1.67 1.9 2.5 2.0 2.14 2.32 2.33 2.33 2.44 2.5 2.74 3.0 3.49 5.2
U95% (mg/g)
0.17 0.15 0.6 0.3 0.1 0.46 1.2 0.28 0.0004 0.26 0.1 0.18 0.12 0.08 0.5
Method(s)
HS-GC-MS & GC-MS HS-GC-MS & HS-GC-FID qNMR GC-MS HS-GC-MS & qNMR HS-GC-MS & qNMR HS-GC-MS & qNMR qNMR, check by HS-GC-MS HS-GC-TOF & HS-GC-FID TGA GC-MS, check by TGA & NMR HS-GC-FID HS-GC-MS & qNMR HS-GC-MS & qNMR HS-GC-FID
Table 7 – Estimates of residual solvent (methanol) content in CCQM-K55.b
The mean of the results is proposed as the KCRV for VOC content (KCRVVOC) in CCQMK55.b. The associated standard uncertainty of the KCRV (uVOC) is assigned as the standard deviation of the mean of the data set
CCQM-K55.b Final Report October 2012
Page 20 of 41
KCRVVOC uVOC
= 2.30 mg/g .= 0.16 mg/g
Figure 9 Estimates for methanol content reported by participants in CCQM-K55.b plotted with their associated standard uncertainties (k = 1). The proposed KCRV for residual solvent in CCQM-K55.b (2.30 mg/g, uVOC 0.16 mg/g) is indicated by the solid red line. Dashed red lines show the KCRVVOC uVOC (k = 1).
iv. KCRV for non-volatile organics/inorganics content in CCQM-K55.b
The estimates reported for non-volatiles/inorganics content provided by the comparison participants are listed in Table 8. Using various methods including TGA, ICP-MS, ICP-OES or XRF spectrometry participants essentially found very low to negligible levels of inorganic material in the sample. The HSA observed a significant residue after high temperature TGA under non-oxidising conditions. NMIJ found no significant residue after high-temperature TGA but did detect a significant residue after exhaustive distillation (100 °C, 90 h, 2.5 x 10-4 Pa). The NMIJ ascribed this distillation residue to a non-volatile organic residue present in the CCQM-K55.b. As mentioned already, in further studies subsequent to the original discussion of results the NMIJ provided additional evidence through size-exclusion chromatography for the presence of an oligomeric organic-based impurity in CCQM-K55.b. NIST reported the presence of a small level of Si-related material by XRF and GLHK reported Al-related impurity present at 0.25 mg/g by ICP-MS. GLHK noted that AlCl3 is a commonly used catalyst in the manufacture of aldrin, which may account for its presence. In general it appeared that the methods used by the other participants in this case while suitable for detecting inorganic impurities were insensitive in this case to a significant level of non-volatile organics.
CCQM-K55.b Final Report October 2012
Page 21 of 41
Participant
BIPM KRISS NIMT NMIA
NMISA VNIIM INMETRO NIM NIST NMIJ (Inorganics) LGC GLHK UME HSA NMIJ (Organics)
Non-volatiles U95% (mg/g) content
(mg/g)
Method outline
0
0 / 0.06 ICP-MS, cross check by % C,H analysis
0
1.8
TGA (160 - 500 °C, 1 hr)
0
0.28
TGA (200 - 1000 °C, 20 min under O2)
0
2
TGA (120 - 850 °C, 20 min [under O2?])
cross check by % C,H analysis
0
0.22
TGA (90 - 800 °C, 10 min [under air])
0
ICP-MS
0.00008
0.000007 ICP-MS
0.038
0.04
ICP-MS
0.05
0.02
XRF
0.12
TGA (35 - 650 °C, 1 hr under air)
0.15
0.08
ICP-MS, ICP-OES, TGA (20 - 1000 °C)
0.25
0.28 / 0.25 ICP-MS
0.5
0.007 TGA (50 - 1000 °C)
9.9
2.1
TGA (120 - 850 °C, 30 min under air)
12
2
Exhaustive vacuum distillation (110 °C,
90 h, 2.5x10-4 Pa)
Table 8 – Estimates of non-volatiles/inorganics content for CCQM-K55.b/CCQM-P117.b
Given that the follow up studies by the NMIJ demonstrated the presence of a significant non-volatile impurity in CCQM-K55.b, the information available directly from the study is limited to the mean of the two results for non-volatile organics. It is proposed to use this value as the KCRV for nonvolatiles content (KCRVNV) in CCQM-K55.b. The associated standard uncertainty of the KCRV (uNV), assigned assuming a rectangular distribution in the range between the two values used for the KCRV, is 0.58 mg/g (= half range/√3).
KCRVNV uNV
= 11 mg/g .= 0.58 mg/g
The proposed value for total non-volatiles is consistent with the data obtained subsequently by NMIJ using gel permeation chromatography. The reported individual values for each participant are given in Table 8 and are shown graphically in Figure 10 along with the proposed KCRV.
CCQM-K55.b Final Report October 2012
Page 22 of 41
Figure 10 Estimates for combined non-volatiles/inorganics content reported by participants in CCQM-K55.b plotted with their associated standard uncertainties (k = 1). The proposed KCRV for non-volatiles in CCQM-K55.b (11 mg/g, uNV = 1 mg/g) is indicated by the solid red line. Dashed red lines show the proposed KCRVNV uNV (k = 1).
Key Comparison Reference Values (KCRVs) for Aldrin in CCQM-K55.b
It was agreed by the participants during the discussion of results that, as for the CCQM-K55.a comparison, assignment of a KCRV for aldrin in CCQM-K55.b based solely on the combined reported results for overall aldrin content was not justified and would be misleading. After initial discussion at the November 2010 CCQM OAWG meeting, follow-on studies were undertaken to investigate the reason for the approximate 10 mg/g discrepancy between mass balance and qNMR results. Subsequent discussion of this data continued at the CCQM OAWG meeting in April 2011 where the NMIJ presented convincing additional data using in particular gel permeation chromatography to demonstrate that a high molecular weight impurity (MW > 1000 ) was present at around the 10 mg/g level in CCQM-K55.b. The comparison coordinator was asked to follow the precedent of the approach used in the CCQM-P20.f and CCQM-K55.a comparisons to propose individual KCRVs for the mass fraction of each of the orthogonal classes of impurity present in the comparison material and to use these values to assign an overall KCRV for aldrin content. Information on the mass fraction assignments of these major impurities was requested in the comparison protocol and reporting form from those participants using the mass balance approach to assign aldrin content.
CCQM-K55.b Final Report October 2012
Page 23 of 41
Assignment of KCRV for Aldrin in CCQM-K55.b
The measurement equation (Eqn. 1) to assign the KCRV of aldrin in CCQM-K55.b (in mg/g) is:
w Aldrin
w Ald rin
=
wRel.Subst.
=
wWater
=
wVOC
=
wNonVol
=
H Rel.Subst.
=
HVOC.
=
1000 [wRel.Subst wWater wVOC wNonVol. H relsubst HVOC ] KCRV for mass fraction of aldrin in CCQM-K55.b KCRV for mass fraction of aldrin-related impurities in CCQM-K55.b
(Eqn. 1)
KCRV for mass fraction of water in CCQM-K55.b
KCRV for mass fraction of volatile organic solvents in CCQM-K55.b
KCRV for mass fraction of non-volatile organics/inorganics in CCQM-K55.b
Correction for between unit inhomogeneity of aldrin-related impurities in the
CCQM-K55.b material. Assigned 0 with associated uncertainty (uH Rel Subst.) Correction for between unit inhomogeneity of volatile organic content in the
CCQM-K55.b material. Assigned 0 with associated uncertainty (uH VOC)
Units for reporting mass fraction ( w ) are mg/g throughout.
The standard uncertainty associated with the mass fraction estimate was calculated from equation (2):
u (u ) (u ) (u ) (u ) (u ) (u ) wAldrin
2 wRel Subst
2 wWa ter
2 wO rg S o lv.
2 wNo n Vo l.
2 H Re lSubst
2 H OrgSolv
(Eqn. 2)
The KCRVs for the impurity classes used for calculation of a mass balance KCRV for aldrin in the CCQM-K55.b comparison are summarised in Table 9.
Input factor w
Related structure organics Water Volatile organics Non-volatiles/inorganics Homogeneity - related structure impurities Homogeneity – volatile organics Combined value
Proposed KCRV (mg/g)
35.4 0.47 2.30
11.0 0.0
0.0
49.17
n
u(KCRV)
(mg/g)
15
0.42
17
0.05
14
0.16
2
0.58
large
0.42
large
0.1
0.85
Table 9: KCRV values for impurities used for calculation of aldrin KCRV and associated combined standard uncertainty in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 24 of 41
When substituted into the equations (1) and (2), the KCRV for aldrin content becomes:
KCRV Aldrin
1000 [KCRVRel.Subst KCRVWater KCRVOrg.Solv. KCRVNonVol. H RelSubst HOrgSolv ] mg/g
= 1000 – [35.4 + 0.47 + 2.3 + 11]) mg/g
= 950.8 mg/g
uKCRVAldrin (uKCRVRel Subst )2 (uKCRVWater )2 (uKCRVOrgSolv. )2 (uKCRVNonVol. )2 (uHRe lSubst )2 (uHOrgSolv )2 mg/g
(0.42)2 (0.05)2 (0.16)2 (0.58)2 (0.42)2 (0.1)2 mg/g = 0.85 mg/g
This is regarded as a conservative estimate for the standard uncertainty of the KCRV. Figure 11 shows the participant results with their reported standard uncertainties plotted against the proposed KCRV (solid red line) and its associated standard uncertainty (k = 1). Figure 12 shows the same results with their expanded uncertainty and the KCRV with the corresponding expanded uncertainty for an approximately 95% coverage range (dashed red lines).
Aldrin content in CCQM-K55.b
1000
990
980
Aldrin (mg/g)
970
960
950
940
VNIIM INMETRO NIM INTI GLHKSAR BIPM NMIA LNE UME KRISS NIST LGC CENAM* NMISA NIMT BAM HSA NMIJ INMS-NRC
Figure 11: Mass fraction estimates by participants for aldrin in CCQM-K55.b with their reported combined standard uncertainty (u). Key Comparison Reference Value for CCQM-K55.b (solid red line) is 950.8 mg/g . The calculated combined standard uncertainty of the KCRV is 0.85 mg/g. Dashed red lines show KCRV uc (k = 1)
CCQM-K55.b Final Report October 2012
Page 25 of 41
Aldrin (mg/g)
Aldrin content in CCQM-K55.b
1000
990
980
970
960
950
940
Figure 12: Mass fraction estimates by participants for aldrin in CCQM-K55.b with reported expanded uncertainty corresponding (U95%). Key Comparison Reference Value for CCQM-K55.b (solid red line) = 950.8 mg/g . The expanded uncertainty for 95% coverage range of the KCRV (dashed red lines) is 1.7 mg/g.
The degree of equivalence of a participant’s result with the KCRV (Di) is given by: Di = wi – KCRVAldrin The expanded uncertainty Ui at the approximately 95% coverage level associated with the Di was calculated as: U95% (Di ) 2 * u(wi )2 u(KCRVAldrin )2 Table 10 records the degree of equivalence (Di) of each key comparison participant’s result with the proposed KCRV for aldrin. These results are shown graphically in Figure 13.
VNIIM INMETRO NIM INTI GLHKSAR BIPM NMIA LNE UME KRISS NIST LGC CENAM* NMISA NIMT BAM HSA NMIJ INMS-NRC
CCQM-K55.b Final Report October 2012
Page 26 of 41
Di (mg/g)
25.0 20.0 15.0 10.0
5.0 0.0 -5.0 -10.0
2.5% 2.0% 1.5% 1.0% 0.5% 0.0% -0.5% -1.0%
VNIIM INMETRO NIM INTI GLHKSAR BIPM NMIA LNE UME KRISS NIST LGC CENAM* NMISA NIMT BAM HSA NMIJ INMS-NRC
Figure 13:
Degree of equivalence (absolute and relative) with the aldrin KCRV for each participant. Points are plotted with the associated expanded uncertainty in the degree of equivalence corresponding to an approximately 95% coverage range.
CCQM-K55.b Final Report October 2012
Page 27 of 41
Participant
NRC-INMS NMIJ HSA BAM NIMT NMISA CENAM LGC NIST KRISS UME LNE NMIA BIPM GLHK INTI NIM INMETRO VNIIM
Di (mg/g)
-3.0 -0.7 -0.4 2.2 7.7 9.4 9.9 10.1 10.2 10.8 10.9 11.2 11.3 11.8 12.3 12.9 13.8 16.0 21.8
UD (mg/g)
6.2 6.7 3.1 8.2 3.3 4.2 7.2 4.2 1.7 2.5 2.6 1.9 5.1 2.1 3.8 19.1 2.9 2.0 2.5
Table 10: Degrees of equivalence Di and expanded uncertainties UD at approximately 95% coverage range in mg/g for aldrin in CCQM-K55.b
Degree of equivalence plots for impurity KCRVs in CCQM-K55.b
The motivation for assigning KCRVs for the contributing impurity classes in CCQM-K55.b was to assess the fitness of the individual mass balance methods and to confirm that an overall value for the main component in agreement with the KCRV for aldrin did not arise simply through cancellation of errors in the contributing impurity assignments. The derived DoE plots also allow for a visualization of specific problem areas in cases where agreement with the KCRV for aldrin was not achieved.
The combined DoE plots by participant for each impurity class quantified are shown on the following pages. To aid in assessment and comparison, the DoE of the overall result for aldrin (cf Figure 13) is plotted at the right (green data point). Where a participant used a mass balance approach but provided no information on a particular class of impurities a “pseudo” DoE is shown in this case as a red data point. This provides information on the validity of the participant’s implicit assumption that the particular impurity component does not make a significant contribution to the overall purity.
CCQM-K55.b Final Report October 2012
Page 28 of 41
Mass Balance KCRV DoEs by Participant:
BAM
15.0
10.0
BAM
Di (mg/g)
5.0
0.0
-5.0
-10.0
BIPM
-15.0
Organics
Water
VOC Non-volatiles
15.0
BIPM
10.0
Di (mg/g)
5.0
0.0 -5.0
-10.0
-15.0
Organics
Water
VOC Non-volatiles
Aldrin Aldrin
Di (mg/g)
CENAM
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
CENAM
Water
VOC Non-volatiles
Aldrin
Key: ♦ = DoE for reported impurity; ■ = “DoE” when no value reported; ■ = DoE for aldrin in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 29 of 41
Di (mg/g)
GLHK
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
GLHKSAR
Water
VOC Non-volatiles
Aldrin
Di (mg/g)
HSA
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
HSA
Water
VOC Non-volatiles
Aldrin
Di (mg/g)
INMETRO
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
INMETRO
▲?1166..00
Water
VOC Non-volatiles
Aldrin
Key: ♦ = DoE for reported impurity; ■ = “DoE” when no value reported; ■ = DoE for aldrin in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 30 of 41
INTI
15.0
10.0
5.0
Di (mg/g)
0.0
-5.0
-10.0
-15.0
KRISS
Organics
15.0
10.0
5.0
Di (mg/g)
0.0
-5.0
-10.0
-15.0
Organics
INTI
Water
VOC Non-volatiles
Aldrin
KRISS
Water
VOC Non-volatiles
Aldrin
Di (mg/g)
LGC
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
LGC
Water
VOC Non-volatiles
Aldrin
Key: ♦ = DoE for reported impurity; ■ = “DoE” when no value reported; ■ = DoE for aldrin in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 31 of 41
LNE
15.0 10.0
Di (mg/g)
5.0 0.0 -5.0
-10.0 -15.0
Organics
Di (mg/g)
NIM
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
Di (mg/g)
NIM
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
LNE
Water
VOC Non-volatiles
Aldrin
NIM
Water
VOC Non-volatiles
Aldrin
NIMT
Water
VOC Non-volatiles
Aldrin
Key: ♦ = DoE for reported impurity; ■ = “DoE” when no value reported; ■ = DoE for aldrin in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 32 of 41
Di (mg/g)
Di (mg/g)
NIST
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
NMIA
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
NIST
Water
VOC Non-volatiles Aldrin
NMIA
Water
VOC Non-volatiles Aldrin
NMIJ
15.0
NMIJ
10.0
Di (mg/g)
5.0
0.0 -5.0
-10.0
-15.0
Organics
Water
VOC Non-volatiles
Aldrin
Key: ♦ = DoE for reported impurity; ■ = DoE for aldrin in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 33 of 41
Di (mg/g)
Di (mg/g)
Di (mg/g)
NMISA
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
UME
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
VNIIM
15.0 10.0
5.0 0.0 -5.0 -10.0 -15.0
Organics
NMISA
Water
VOC Non-volatiles
Aldrin
UME
Water
VOC Non-volatiles
Aldrin
VNIIM
▲? 2211..88
Water
VOC Non-volatiles
Aldrin
Key: ♦ = DoE for reported impurity; ■ = “DoE” when no value reported; ■ = DoE for aldrin in CCQM-K55.b
CCQM-K55.b Final Report October 2012
Page 34 of 41
CONCLUSIONS AND HOW FAR THE LIGHT SHINES
Aldrin was selected to be a representative low polarity, moderately complex organic compounds capable of analysis by GC. It was anticipated to provide an analytical measurement challenge representative for the value-assignment of compounds of broadly similar structural characteristics.
For such a relatively impure material, there was excellent agreement between the majority of participants in both the identification and the quantification of the related structure impurity content of the sample, confirming the conclusions of CCQM-K55.a that measurement of this general class of impurities is performed satisfactorily by most NMIs. Where relatively low values for total organic impurity were reported by an individual participant these were consistent with underestimation of either the isodrin (2) content or a failure to detect the long retention time impurity dechlorane (4).
There was also a good level of agreement on the quantification of the water, residual solvent and inorganics content of the material. Unlike the previous CCQM-P20.f (digoxin) and CCQM-K55.a (estradiol) comparisons, the majority of participants sought to specifically quantify each of these impurity components.
As discussed in detail in the main report, the disparity in the final results reported for the aldrin content, as evidenced from the DoE plot which shows a set of results which overestimate the aldrin content by approximately 10 mg/g, arose primarily from a failure of most participants using a mass balance approach to detect the presence of a significant level of non-volatile organic impurity in the sample. This class of impurity had not been present to any significant extent in any previous comparisons.
Inspection of the DoE plots by impurity sub-group obtained by participants using the mass balance approach provides a clear indication of the laboratories that are applying this approach satisfactorily, apart from missing the non-volatile impurity in this particular instance, and those which have more general issues to address.
Nevertheless the follow-up discussions revealed that several participants did have an indication from qNMR data that their mass balance procedure had overestimated the aldrin content, but decided to accept the mass balance result and assume a hidden bias in the NMR data. By contrast the laboratories that placed greater value on the qNMR result, notably the NMIJ who invested significant effort into resolving the discrepancy, were able to find evidence for the presence of non-volatile organic impurity at the requisite level to bring the mass balance and qNMR results into agreement.
A secondary issue was that some laboratories need to review their NMR data processing procedures and software as they obtained a high qNMR value for aldrin, which resulted in a false degree of confidence in their mass balance results.
In summary, the major conclusions coming out of the comparison were:
experienced laboratories obtained good levels of agreement in the mass fraction assignments for related impurities, water and VOC content in CCQM-K55.b. This could be demonstrated clearly through the equivalence plots relative to the individual KCRVs for individual participants for these classes of impurities;
CCQM-K55.b Final Report October 2012
Page 35 of 41
a weakness in direct methods for detecting the presence of significant non-volatile organic impurities. The inferential methods that had been relied on in previous comparisons (elemental analysis, NMR, TGA) were not suitable for CCQM-K55.b;
the varying perceptions prior to the comparison between laboratories of the reliability (or otherwise) of qNMR for direct assignment of the purity of the main component should be reviewed in the light of the actual results;
a need in some cases for improved QC procedures and/or understanding of software processing issues such that laboratories can have confidence that qNMR data is not biased as a result of their processing parameters
The results of the comparison provide a strong example of the need to undertake further investigations when notionally independent methods (mass balance v. qNMR in this instance) give significantly different main component purity results. A valid explanation for the cause of such differences should be established before one or other result is discarded. This is particularly so in a situation where the qNMR result is lower than the mass balance result, as it is not possible in this case to explain the discrepancy as arising simply from a contribution of NMR signals from unidentified impurities to the NMR signal being integrated to quantify the main component.
The study shows that in a case where a material is relatively impure both in the level and range of the impurity components, a capability to assign purity through either a mass balance approach or a direct qNMR approach with a moderate expanded uncertainty ( U95% < 0.6 % relative) can be achieved.
“How Far The Light Shines” Statement for CCQM-K55.b
The comparison was intended to demonstrate a laboratory’s performance in determining the mass fraction of the main component in a high purity organic material. The measurement results should be indicative of the performance of a laboratory’s measurement capability for the purity assignment of organic compounds of medium structural complexity (molecular weight 300-500) and low polarity (pKOW < -2) and for which related structure impurities are best quantified directly by capillary gas-phase chromatographic techniques.
CCQM-K55.b Final Report October 2012
Page 36 of 41
Annex A : Analysis Space for Organic Primary Calibrators
Molecular Weight
pKOW
CCQM-P20 & CCQM-K55 measurands
CMC claims for pure substance calibrators or calibration solutions
CCQM-K55.b Final Report October 2012
Page 37 of 41
Annex B – Aldrin-related compounds identified in CCQM-K55.b
H
H Cl
Cl
H
H
H
H
Cl
Cl
H Cl
H
Cl
Cl
Cl
H H
OH
H
H H Cl
HH
Cl
Cl
Cl Cl
Cl
Cl
Cl
Cl Cl
Cl Cl
Cl Cl
Cl Cl
Cl
Cl
H HH
H H
H
H H Cl
HH
Cl Cl
Cl
Isodrin (2) Dieldrin (3) Dechlorane (4) Dihydroaldrin (5)
CCQM-K55.b Final Report October 2012
Page 38 of 41
Cl Cl Cl
Cl
Cl
O
Cl
Cl Cl
Cl
Cl
Cl
Cl
Cl Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl Cl
Cl
Cl
Endrin ketone (6) Hexachlorocyclopentadiene (7) Octachlorocyclopentene (8)
Chlordene (9)
CCQM-K55.b Final Report October 2012
Page 39 of 41
Annex C – 1H NMR spectra of CCQM-K55.b
1H NMR
Cl
Cl
11
H5 6 H7
H4 H3
2 H1 H Cl
12s
12a
HH
8 Cl Cl
9
10
Cl
H-2,7
CDCl3
H-4,5
H-3,6
H-12a H-12s
1H NMR spectrum of OGP.014 in CDCl3 : full scale
SpinWorks 3: BIPM Aldrin OGP-014 005 (10mg) in 0.6ml CDCl3 (over K2CO3) QNP600
Methanol
H-3,6
H-2,7
Isodrin H-2,7
Isodrin H-3,6 Dieldrin H-4,5
Dieldrin H-2,7 & 3,6 Dechlorane?
PPM 3.50
3.40
3.30
3.20
3.10
3.00
2.90
2.80
2.70
2.60
2.50
file : ...O AW G\\ K55 stuff\\ NM R \\ swe 091008\\ 2\\ fid e x pt: <zg> tra nsm itte r fre q.: 600.133001 M H z tim e dom a in size : 32768 points width: 6613.76 H z = 11.0205 ppm = 0.201836 H z/ pt num ber of scans: 32
fre q. o f 0 ppm : 600.130013 M H z pro ce sse d size : 32768 co m ple x po ints LB: 0.200 GF: 0.0000 Hz/cm : 26.962 ppm /cm : 0.04493
1H NMR spectrum of OGP.014 in CDCl3 : expansion of δ 2.5 – 3.5 ppm
CCQM-K55.b Final Report October 2012
Page 40 of 41
References
1 Westwood, S., Josephs, R. D., Daireaux, A., Wielgosz, R., et al.: An international comparison of mass fraction purity assignment of theophylline: CCQM Pilot Study CCQM-P20.e (Theophylline), Metrologia, 46 (2009) 1A, 08019 2 Westwood, S., Josephs, R. D., Choteau, T., Mesquida, C., Daireaux, A., Wielgosz, R., et al: An international comparison of mass fraction purity assignment of digoxin: CCQM Pilot Study CCQM-P20.f (Digoxin), Metrologia 48 (2011) 1A, 08013. 3 Westwood, S., Josephs, R. D., Daireaux, A., Wielgosz, R., et al; Final report on key comparison CCQMK55.a (Estradiol): An international comparison of mass fraction purity of estradiol, Metrologia, 49 (2012) 1A, 08009 4 J. Sangster, Octanol-water Partition Coeffcients: Fundamentals and Physical Chemistry, John Wiley & Sons, Chichester (1997).
CCQM-K55.b Final Report October 2012
Page 41 of 41
Ver+/-
