A comparison of nitrogen dioxide (NO 2 ) in nitrogen standards at 10 μmol/mol by Fourier Transform Infrared Spectroscopy (FT-IR)

Transcription

1 A comparison of nitrogen dioxide (NO 2 ) in nitrogen standards at 10 μmol/mol by Fourier Transform Infrared Spectroscopy (FT-IR) (Final report of the pilot study CCQM-P110-B1) Edgar Flores *1, Faraz Idrees 1, Philippe Moussay 1, Joële Viallon 1, Robert Wielgosz 1, Teresa Fernández 2, Andrés Rojo 2, Sergio Ramírez 2, Nobuyuki Aoki 3, Kenji Kato. 3, Lee Jeongsoon 4, Dongmin Moon 4 and Jin-Seog Kim 4, A. Harling 5, M. Milton 5, Damian Smeulders 6, Franklin R. Guenther 7, Lyn Gameson 7, Angelique Botha 8, James Tshilongo 8, Napo Godwill Ntsasa 8, Miroslava Valková 9, Leonid A. Konopelko 10, Yury A. Kustikov 10, Vladimir S. Ballandovich 10, Elena V. Gromova 10, Dirk Tuma 11, Anka Kohl 11 and Gert Schulz Bureau International des Poids et Mesures (BIPM), Pavillon de Breteuil, F Sevres Cedex. 2 Centro Español de metrología (CEM), Calle Alfar, 2, Tres, Cantos (Madrid), Spain. 3 National Metrology Institute of Japan (NMIJ), Umesono, Tsukuba Ibaraki, Japan. 4 Korea Research Institute of Standards and Science (KRISS),1, Doryong-Dong, Yuseong-Gu, Daejeon , Korea 5 National Physical Laboratory (NPL), Hampton Road, Teddington, Middx, TW11 0LW, UK. 6 National Measurement Institute Australia (NMIA), Bradfield Road, P.O. Box 264, NSW 2070 Lindfield 7 National Institute of Standards and Technology (NIST), 100 Bureau Drive, Gaithersburg, MD , USA 8 National Metrology Institute of South Africa (NMISA), CSIR, Building 4 West, Meiring Naude Road Brummeria, 0184, Pretoria, South Africa. 9 Slovak Institute of Metrology (SMU), Karloveská 63, SK Bratislava, Slovak Republic. 10 D.I.Mendeleyev Institute for Metrology (VNIIM), 19 Moskovsky pr., St. Petersburg, Russia. 11 Federal Institute for Materials Research and Testing (BAM), Unter den Eichen 87, Berlin, Germany. Coordinating laboratories: Bureau International des Poids et Mesures (BIPM) VSL Dutch Metrology Institute Study coordinator: Edgar Flores (BIPM) Correspondence to be addressed to: Edgar Flores edgar.flores@bipm.org (Tel: ) Field: Amount of substance Organizing Body: CCQM Page 1 of 76

2 Index 1. RATIONAL FOR COMPARISON 4 2. QUANTITIES AND UNITS 4 3. SCHEDULE 4 4. MEASUREMENT STANDARDS 4 Preparation and value assignment 4 Purity analysis 5 Stability of the mixtures 6 Deviations from the protocol 6 5. REFERENCE VALUES FOR CYLINDERS MEASUREMENT PROTOCOL MEASUREMENT METHODS 18 8 RESULTS DISCUSSION CONCLUSION 26 ANNEX 1- BIPM VALUE ASSIGNMENT PROCEDURE Description of the facility Measurement protocol of the BIPM BIPM measurement uncertainties and analyser response Covariance between two dynamically generated gas mixtures FT-IR analysis of gas standards FT-IR Spectra acquisition procedure Quantitative analysis of nitric acid 36 Page 2 of 76

3 8. Uncertainty budget Regression analysis Determination and validation of analysis functions 37 ANNEX 2 - MEASUREMENT REPORTS OF PARTICIPANTS 38 Centro Español de metrología (CEM) 38 Korea Research Institute of Standards and Science (KRISS) 41 National Measurement Institute Australia (NMIA) 44 National Metrology institute of Japan (NIMJ) 46 National Institute of Standards and Technology (NIST) 49 National Physical Laboratory (NPL) 52 National Metrology Institute of South Africa (NMISA) 57 Slovak Institute of Metrology (SMU) 60 Mendeleyev Institute for Metrology (VNIIM) 68 Federal Institute for Materials Research and Testing (BAM) 71 Bureau International des Poids et Mesures (BIPM) 74 Page 3 of 76

4 1. Rational for comparison This pilot study compares the performance of participants in analysing an unknown mixture of nitrogen dioxide in nitrogen (at a nominal mole fraction of 10 µmol/mol.) by comparison with their own standards using FT-IR spectroscopy. It uses the same standard gas mixtures as were used in the key comparison CCQM-K74 (1). The level of comparability between laboratories in this Pilot Study was evaluated and compared to that in CCQM-K74. In CCQM-K74, most participants used chemiluminescence, with a small number using UV absorption or FT-IR spectroscopy. These last two techniques are of particular interest because they do not exhibit any cross sensitivity to nitric acid (HNO 3 ), which was known to be present in the mixtures used for the comparison. 2. Quantities and Units In this protocol the measurand was the mole fraction of nitrogen dioxide in nitrogen*, with measurement results being expressed in mol/mol and its multiples μmol/mol or nmol/mol. (*the nitrogen balance gas contains nominally 1000 µmol/mol of oxygen) 3. Schedule The revised schedule of the project was as follows: June 2009 June August 2009 September 2009 October January 2010 February 2010 March 2010 May 2010 February 2010 May 2010 July 2010 September 2011 Shipment of cylinders to the BIPM Analysis of mixtures at the BIPM Shipment of cylinders from the BIPM to participants Analysis of mixtures by the participants Shipment of cylinders back from participants to the BIPM 2 nd set of analysis of mixtures at the BIPM Reports of the participants Distribution of Draft A of this report Distribution of Draft B V0.1 of this report 4. Measurement standards Preparation and value assignment The gas mixtures were prepared by the Dutch Metrology Institute (VSL). The nitrogen dioxide gas mixtures were contained in passivated aluminium cylinders of 5 L. The cylinders were pressurized to about 12 MPa. The nitrogen dioxide gas standards were produced by gravimetric preparation in accordance with the International Standard ISO 6142: ISO 6142:2001: Gas analysis-preparation of calibration gas mixtures-gravimetric method. Page 4 of 76

5 Each cylinder was value assigned by the BIPM, with its dynamic gas facility described in ANNEX 1, before and after the participant s measurements. The VSL and BIPM values and measurements are given in Table 1 and Table 2 where: x VSL u prep (x VSL ) u ver (x VSL ) x BIPM1 u(x BIPM1 ) x BIPM2 u(x BIPM2 ) is the value assigned by VSL based on gravimetric preparation; the standard uncertainty of the VSL values with contributions due to gravimetry and purity analysis; the standard uncertainty including contributions from verification associated with the assigned value x VSL ; the first BIPM measurement result (prior to sending out cylinders to participants); the standard uncertainty of the first BIPM measurement result, including losses and drift terms as explained in section 5; the second BIPM measurement result (on return of cylinders from participants); the standard uncertainty of the second BIPM measurement result; Purity analysis From previous studies carried out by the BIPM and VSL it was expected that the mixtures would contain quantifiable amounts of HNO 3. The analysis of the gas mixtures at the BIPM with FT-IR spectroscopy confirmed the presence of and permitted the quantification of nitric acid in the gas mixtures. Table 3 list the nitric acid mole fractions found in the gas standards. To verify the stability of the gas mixtures the purity analysis was repeated when the gas mixtures were returned to the BIPM provided that the participants returned the cylinders with the minimum gas pressure required as described in the comparison protocol (see Table 4). Table 3 lists: Cylinder x HNO3(1) u(x HNO3(1) ) x HNO3(2) u(x HNO3(2) ) the identification code of the cylinder received by the participating laboratory; the mole fraction of nitric acid measured in the standard by the BIPM (prior to sending standards to participants); the standard uncertainty associated with the nitric acid mole fraction measurement; the mole fraction of nitric acid measured in the standard by the BIPM (following return of standards to the BIPM); the standard uncertainty associated with the nitric acid determination by FT-IR spectroscopy after the participants measurements. Page 5 of 76

6 Stability of the mixtures The nitrogen dioxide mole fractions measured by the BIPM before and after measurements by participants are shown in Figure 1. The error bars in the first series of measurements represent the standard uncertainty associated with the BIPM measurement results including contributions from the dynamic preparation of nitrogen dioxide gas mixtures, NO 2 losses in the permeation system of the BIPM and an observed drift in the nitrogen dioxide mole fractions measured by the BIPM before and after the participant s measurements. For further information see the ANNEX 2 of the report International comparison CCQM-K74: Nitrogen dioxide, 10 μmol/mol. The error bars in the second series of measurements represent the standard uncertainty associated with the BIPM measurement results including the contributions from the dynamic preparation of nitrogen dioxide gas mixtures. The difference between the BIPM series of measurements for each standard is plotted in Figure 2.The NO 2 mole fraction in all cylinders was found to be in the range from μmol/mol to μmol/mol as measured by the BIPM. The amount of nitric acid found in each cylinder was consistent with the difference between the gravimetric preparation value and BIPM s analytical value for the nitrogen dioxide amount fraction, and accounts for the conversion of nitrogen dioxide to nitric acid (reacting with residual water and oxygen in the gas standards) and limited by the amount of water present. Figure 3 plots the nitric acid mole fractions measured in each gas standard before and after measurements by the participants. Changes in the mole fractions of nitrogen dioxide and nitric acid in each cylinder during the period of the comparison were well within the measurement uncertainties of these values. The uncertainty budget for the BIPM measurement result contains a component which covers any change in value due to instability of the gas transfer standard. The difference between the series on nitric acid mole fractions measured by the BIPM was plotted in Figure 4. The mixtures stability can be confirmed in the summation of nitrogen dioxide and nitric acid mole fractions of the first and second series of measurements shown in Figure 5. Deviations from the protocol The BIPM was unable to perform a second measurement of nitric acid content in gas mixtures and , as the participating laboratories that had made measurements on these cylinders had not followed the comparison protocol and returned the cylinders with insufficient gas to make these measurements. Cylinder was not returned on time to the BIPM and no additional measurements could be made on this cylinder. Cylinder was analyzed by the BIPM to provide a BIPM result in this comparison. This cylinder was sent after to NIST 2. Unlike other participants, the NIST performed its analysis on two different cylinders: cylinder during CCQM-K74 and cylinder during CCQM-P110. This should not impact the results as all reference values are given by the BIPM. 2 The cylinder , originally assigned to the BIPM, was sent to NIST before the second series of measurements due that the internal pressure of NIST cylinder, , was extremely low to take part in the present comparison after its participation in the Key comparison CCQM-K74. Page 6 of 76

7 VSL preparation values Certificate Preparation Number Assigned Gravimetric standard Certified standard number date of Cylinder NO 2 mole fraction uncertainty uncertainty x VSL uprep(x VSL ) uver(x VSL ) (μmol/mol) (μmol/mol) (μmol/mol) /02/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM /04/2009 # PRM Table 1. Characteristics of gravimetric mixtures as provided by VSL. Page 7 of 76

8 BIPM measurement results Number of Cylinder Measurement date 1 st measurement 1st BIPM Standard 2nd BIPM Measurement assigned Standard NO 2 mole fraction uncertainty NO 2 mole measurement date fraction uncertainty Δx= x BIPM1 u(x BIPM1 ) (x BIPM2-2 nd measurement x BIPM2 u(x BIPM2 ) x BIPM1 ) u(δx) 2u(Δx) μmol/mol μmol/mol μmol/mol μmol/mol μmol/mol μmol/mol μmol/mol # PRM 19/08/ * # PRM 18/08/ /04/ # PRM 18/08/ /03/ # PRM 18/08/ * # PRM 31/08/ /04/ # PRM 19/08/ /03/ # PRM 25/08/ /03/ # PRM 28/08/ /03/ # PRM 29/08/ /04/ # PRM 28/08/ ** # PRM 25/08/ /03/ Table 2. Results of BIPM NO 2 mole fraction measurements. * Insufficient gas for second measurement. *** Standard not yet returned to the BIPM. Page 8 of 76

9 BIPM HNO 3 Measurements Measurement x HNO3(1) u(x HNO3(1) ) Measurement x HNO3(2) u(x HNO3(2) ) Δx= (x HNO3(2) - Cylinder date date x HNO3(1) ) u(δx) 2u(Δx) (μmol/mol) (μmol/mol) (μmol/mol) (μmol/mol) μmol/mol μmol/mol μmol/mol # PRM 12/08/ * # PRM 30/07/ /04/ # PRM 28/07/ /05/ # PRM 11/08/ * # PRM 12/08/ /05/ # PRM 30/07/ /04/ # PRM 28/07/ /04/ # PRM 11/08/ /05/ # PRM 12/08/ /04/ # PRM 31/07/ *** # PRM 28/07/ /04/ Table 3. Nitric acid mole fraction measured in cylinder gas standards by the BIPM using FT-IR spectroscopy. * Insufficient gas for second measurement. *** Standard not yet returned to the BIPM. Page 9 of 76

10 Certification Number Date of pressure on pressure on Lab date of Cylinder return departure return Mpa Mpa NPL 24/02/2009 # PRM 26-Jan * SMU 09/04/2009 # PRM 02-Feb NMIA 09/04/2009 # PRM 26-Feb NMISA 08/04/2009 # PRM 24-Feb * NMIJ 19/03/2009 # PRM 19-Feb KRISS 02/04/2009 # PRM 26-Feb NIST 20/03/2009 # PRM 15-Mar CEM 10/04/2009 # PRM 02-Mar VNIIM 03/04/2009 # PRM 16-Feb BAM 08/04/2009 # PRM *** 9.8 BIPM 20/03/2009 # PRM In place - - Table 4. Departure and return pressure of the gas standards after being measured by the participating laboratories. * Insufficient gas for 2 nd series of BIPM measurements ( 5MPa). *** Standard not yet returned to the BIPM. Page 10 of 76

11 1st BIPM Standard 2nd BIPM Standard NO 2 mole fraction measurement uncertainty NO 2 mole fraction measurement uncertainty x BIPM1 2u x BIPM2 2u x BIPM1 u(x BIPM1 ) x BIPM2 u(x BIPM2 ) + (x BIPM1 + + u(x BIPM2 + (x BIPM2 + μmol/mol μmol/mol μmol/mol μmol/mol x HNO3(1) u(x BIPM1 + x HNO3(1) ) x HNO3(1) ) x HNO3(2) x HNO3(2) ) x HNO3(2) ) * * ** Table 5. Summation of Nitrogen Dioxide and Nitric Acid Mole fractions for each standard based on BIPM measurements. * Insufficient gas for second measurement. *** Standard not yet returned to the BIPM. Page 11 of 76

12 BIPM Nitrogen dioxide mole fraction / μmol/mol st 2nd Figure 1. First and second series of nitrogen dioxide mole fraction measurements by the BIPM prior to sending standards to participating laboratories. The second series of measurements was done after the return of standards from participating laboratories The error bars of the first series of measurements represent the standard uncertainty (k = 1) associated with the BIPM measurement results including contributions from the dynamic preparation of nitrogen dioxide gas mixtures, NO 2 losses in the permeation system of the BIPM and an observed drift in the nitrogen dioxide mole fractions measured by the BIPM before and after the participant s measurements. For further information see the ANNEX 2 of the report International comparison CCQM-K74: Nitrogen dioxide, 10 μmol/mol. The error bars in the second series of measurements represent the standard uncertainty (k = 1) associated with the BIPM measurement results including the contributions from the dynamic preparation of nitrogen dioxide gas mixtures. Page 12 of 76

13 Δx (xbipm1-xbipm2) / μmol/mol Figure 2. Difference between the BIPM series of measurements of nitrogen dioxide for each standard. The error bar represents the expanded uncertainty at a 95% level of confidence. Page 13 of 76

14 BIPM Nitric acid mole fraction / μmol/mol st 2nd NMI's Figure 3. First (red) and second (blue) series of nitric acid mole fraction measurements by the BIPM and nitric acid mole fractions found by the participating laboratories (green). The error bar represents the standard uncertainty (k=1). Page 14 of 76

15 Δx (x HNO3(1)-x HNO3(2)) mole fraction / μmol/mol Figure 4. Difference between the second and first series of nitric acid mole fraction measurements by the BIPM. The error bar represents the expanded uncertainty at a 95% level of confidence. Page 15 of 76

16 BIPM Nitrogen dioxide + Nitric acid mole fraction / μmol/mol Figure 5. Summation of Nitrogen Dioxide and Nitric Acid Mole fractions in each standard based on BIPM measurements. Red: First measurements. Black: Second measurements. The error bar represents the expanded uncertainty at a 95% level of confidence. Page 16 of 76

17 5. Reference Values for Cylinders During the 24 th and 25 th meetings of the CCQM GAWG it was agreed that the reference value for this comparison was to be based on BIPM measurement results prior to distribution of gas standards to participants. The BIPM s measurements clearly indicate the presence of nitric acid in the gas mixtures ranging from 100 nmol/mol to 350 nmol/mol. The nitrogen dioxide gravimetric preparation values provided by VSL were not used as the reference values for the comparison as they do not account for the presence of nitric acid in the standards, arising from the conversion of NO 2 to nitric acid through the reaction with oxygen and residual water in the cylinders. The current hypothesis is that the water must have been present on the cylinder coatings. In the current version of the report, laboratory results are compared to BIPM values, since they correctly account for the presence of nitric acid in the gas mixtures. Furthermore, the agreement between the summation of nitric acid and nitrogen dioxide mole fractions with the initial amount of nitrogen dioxide (prior to any reaction) expected from static gravimetric preparation values further confirms the hypothesis of the loss mechanism of NO 2 in the cylinders. For each cylinder, the reference value is the NO 2 mole fraction assigned by the BIPM (first measurement). Following the CCQM GAWG guidance, it was decided that the standard uncertainty of the reference value (x ref ) was to be calculated from the following equation u ref x ( u( x )) ( u( x )) ( u( x )) (1) NO2 BIPM NO2Losses Drift where u( x BIPM ) is the uncertainty associated with the value assigned by the BIPM, u( x NO 2Losses ) the uncertainty contribution due to NO 2 losses equivalent to 5.7 nmol/mol and u( x Drift ) the uncertainty contribution due to observed drift in NO 2 estimated to be 21 nmol/mol. This leads to an overall standard uncertainty of the reference value of μmol/mol. A full discussion of the uncertainty of the reference value is included in the report International comparison CCQM-K74: Nitrogen dioxide, 10μmol/mol. The permeation of nitric acid from NO 2 permeation tubes was detected and quantified by the BIPM, and the BIPM s values are corrected to avoid systematic errors caused by this issue. This is fully described in ANNEX 1- BIPM Value assignment procedure. 6. Measurement protocol The measurement protocol requested participants to use FT-IR spectroscopy methods calibrated and traceable to gas standards. Likewise, it was requested that participants provide the value and uncertainty of the nitrogen dioxide mole fraction measured by the laboratory, a complete uncertainty budget and a description of their gas analysis procedure. The participant s reports are included in ANNEX 2 - Measurement reports of participants. Page 17 of 76

18 7. Measurement methods As mentioned above, the protocol for this pilot study required participants to use FT-IR spectroscopy. Table 6 summarizes the main characteristics of the instrumentation and calibration methods need by each laboratory. From Table 6 we can conclude that: - five of the ten participating laboratories used Nicolet spectrometers and Omnic acquisition software; - of the participating laboratories nine used gas mixtures contained in high pressure cylinders as calibration references and just the BIPM used mixtures from its permeation facility; - of the ten laboratories using gas mixtures as standards nine were traceable to their own gravimetric standards - nearly all participants measured their gas mixtures near ambient conditions (297 K and ~100 kpa); - more than half of the participants performed water corrections; - half of the participants used a spectral region within the range ( ) cm -1 and half used the spectral region within the range ( ) cm -1 or both; - the resolution range used by the participants was (0.25 to 4) cm -1 ; and - eight of the ten participating laboratories used an optical path length between the range of (4.8-10) m, one 0.1m and one 20 m; and - two of the ten participating laboratories confirmed the presence of nitric acid in their gas mixture. 8 Results The results submitted by the participants are plotted in Figure 6. The evaluation of the level of consistency between the participating laboratories was done by comparison between the reported nitrogen dioxide mole fractions by participating laboratories and the BIPM FT-IR measurements (1 st series) for each cylinder. The consistency between the participating laboratory s results and the BIPM reference values is presented in terms of a difference (D) defined as: D x NMI x ref (2) where x NMI denotes the amount of substance fraction as measured by the participating national metrology institute (NMI) and xref the reference value given by the BIPM. The standard uncertainty in D is: u 2 2 ( D) u NMI uref (3) and the expanded uncertainty at 95% confidence level Page 18 of 76

19 U ( D) k u( D) (4) where k denotes the coverage factor, taken as k = 2 (normal distribution, approximately 95% level of confidence). The differences between BIPM and NMI s assigned values for each of the circulated gas standards are listed in Table 7 where: Laboratory Cylinder x ref u(x ref ) x Lab u(x Lab ) D u(d) is the acronym of the participating national metrology institute; the identification code of the cylinder received by the participating laboratory; the reference value (1 st series of BIPM measurement results); the uncertainty of the reference value;; the result as reported by the participating laboratory; the standard uncertainty of the reported value x Lab as reported by the participating laboratory ; the difference in amount of substance fraction as measured by the laboratory and x BIPM the BIPM value; and the standard uncertainty of the difference of amount of substance; The difference (D) in nitrogen dioxide reported values by participating laboratories and the BIPM are plotted in Figure 7. In this figure, the result attributed to the BIPM is based on the second analysis of a particular cylinder (# PRM), using the first measurement made on this gas mixture as reference value.the method used by the BIPM to calculated its uncertainty value, u(x Lab ), is carefully described elsewhere(1). Page 19 of 76

20 Laboratory NPL SMU NMIA NMISA NMIJ KRISS NIST CEM VNIIM BAM BIPM FT-IR Spectrometer Nicolet 6700 Varian Excalibur Nicolet 6100 Nicolet Magna IR 560 JASCO Bruker Nicolet 6700 PerkinElmer Monitoring Ldt, Russia Nicolet Nexus Nicolet Nexus Gas cell (m) Calibration method 2 mixtures peak area integration own gravimetric standards 5 mixtures (0.99, 2.52, 7.23, and 49.72) μmol/mol own gravimetric standards Bracketing (8-12) μmol/mol own gravimetric standards 4 mixtures at 10 μmol/mol own gravimetric standards 4 mixtures 1 mixture 1 mixture Normalized response of nine NO2 peaks (See pag. 49) own gravimetric standards own gravimetric standards own gravimetric standards 3 mixtures, GLS, linear, (5, 10, 15) μmol/mol NPL gravimetric standards 2 mixtures at 10 μmol/mol, CLS own gravimetric standards 2 mixtures (9.43, 12.2) μmol/mol own gravimetric standards GLS, linear, Bracketing, Permeation-dynamic mixtures own permeation tube system Traceability Average gas cell temperature (K) Average gas cell presure (kpa) Spectrometer software to: control acquire analyze Did the participant performed any water correction? if yes, how? Spectral range used for the mole fraction determination(cm -1 ) Omnic Omnic Omnic None 1600 ( ) Varian Resolutions Omnic Omnic Varian Resolutions Omnic Omnic Varian Resolutions Excel MALT Yes Peak selection Yes Using the spectral region cm Yes Omnic software procedure Spectra Manager OPUS Omnic Spectra Manager FCM Omnic Spectra Manager FCM Excel None None Yes, peaks selection Spectrum software Fspec Omnic Omnic-IMACC Spectrum software Fspec Omnic IMACC IFSS (Imacc Software) ASpec Omnic B_Least (ISO 6143) None Yes Vapor absorption elimination None Yes MALT interference correction by GLS and N 2 purge in FT-IR enclosure box Measurement resolution (cm -1 ) / u(x Lab) /% Table 6. Main FT-IR characteristics of the instrumentation and calibration methods submitted by the laboratories. Page 20 of 76

21 CCQM-P110 B Laboratory measurement results mole fractions / μmol/mol NPL SMU NMIA NMISA NIMJ KRISS NIST CEM VNIIM BAM BIPM Laboratory Figure 6. Nitrogen dioxide amount of fraction as reported by the participating laboratories for CCQM-P110 B1. The error bar represents the standard uncertainty (k=1). Page 21 of 76

22 BIPM Participants Laboratory Cylinder x ref u(x ref ) x Lab u(x Lab ) D( x Lab- x ref ) u(d) u(d) (k=2) NPL # PRM SMU # PRM NMIA # PRM NMISA # PRM NMIJ # PRM KRISS # PRM NIST # PRM CEM # PRM VNIIM # PRM BAM # PRM BIPM # PRM Table 7. Laboratory results for nitrogen dioxide measurements (μmol/mol). Page 22 of 76

23 D (μmol/mol) NPL SMU NMIA NMISA NIMJ KRISS NIST CEM VNIIM BAM BIPM Laboratory Figure 7. Difference between participants results and the reference value determined in CCQM-K74. The error bar represents the expanded uncertainty at a 95% level of confidence. Page 23 of 76

24 9. Discussion According to Figure 7 the majority of participants agree with the reference value. In order to draw more conclusions on the specificity of the FT-IR technique, the same results are plotted in Figure 8 together with CCQM-K74 results for those participants that took part in both comparisons, excluding the CCQM-K74 results obtained by FT- IR. In this figure, the similarity of the results confirms the good agreement between FT- IR and the two other techniques which are UV photometry and chemiluminescence. A different conclusion could have been expected, given that FT-IR analysers do not suffer from a cross sensitivity to nitric acid (unlike chemiluminescence analysers), which was observed in the circulated cylinders. As already concluded in CCQM-K74, the possible presence of nitric acid at the same level in the primary standards used to calibrate the FT-IR can explain this result, if the impurity was not detected in the FT-IR spectra and accounted for in the results. There is no evidence of any clear differences in the results due to the specific choices of FT-IR set-up parameters made by the participants. From Figure 8 it also appears that almost all participants attributed a larger uncertainty to the FT-IR measurement results than to the two other techniques. Based on all participant results in table 7, relative standard uncertainties ranging from 0.23% to 4.2% were attributed to FT-IR based measurements of NO 2 mole fractions. The majority (seven participants) however attributed a relative standard measurement uncertainty of less than 1.5 % to their measurement results, notably for participants that agreed with the reference values in both this and CCQM-K74. This is finally similar to the relative standard measurement uncertainties attributed to chemiluminescence and UV absorption techniques, which ranged from 0.4% to 1.7% for the quantification of NO 2 mole fractions at nominal 10 µmol/mol. Since the FT-IR measurement technique is potentially capable of detecting and quantifying impurities such as nitric acid in nitrogen dioxide gas mixtures. In principal this would mean that measurements made on the transfer cylinders by FT-IR and chemiluminescence should have shown a small bias in their results due to the nitric acid impurities present in the transfer standards, assuming that all nitric acid in the participants primary standards had been accounted for and that the nitric acid had not been scrubbed from the transfer standard gas prior to measurement by chemiluminescence. The magnitude of the bias that could be expected in these cases would be in the order of 100 nmol/mol to 350 nmol/mol or 1% to 3.5 % respectively of the nominal mass fraction of 10 µmol/mol. This level of bias was not observed, and would have been similar in magnitude to the measurement uncertainties of both techniques. Further interpretation of potential biases would require the nitric acid impurity content of all primary standards used by participants to be evaluated. In the case of measurements at the BIPM, a long pathlength FT-IR gas cell was used to quantify nitric acid and correct for its presence. The NIST chemiluminescence measurements were performed on a gas stream from the transfer standard that had been scrubbed of its nitric acid content using nylon filters. The BIPM and NIST measurement results agreed well within their measurement uncertainties and the reference value of the comparsion, confirming the agreement of two independent methods for correcting for the presence of nitric acid. Page 24 of 76

25 D (μmol/mol) UV CLD CLD CLD CLD CLD CLD NPL SMU NMISA KRISS NIST CEM VNIIM Laboratory CCQM-P110 B1 (FT-IR) CCQM-K74 Figure 8. Difference between participant results and the reference values in this pilot study and in the key comparison CCQM-K74. The error bars represents the expanded uncertainty at a 95% level of confidence. The measurement techniques used by participants for CCQM-K74 were: NPL; UV photometry. SMU, NMISA, KRISS, NIST, CEM and VNIIM; Chemiluminescence Page 25 of 76

26 10. Conclusion The results of this pilot study indicate good consistency for the measurements against in-house standards by FT-IR spectroscopy. The level of agreement between participants is very similar to that reported in CCQM -K74 for which CLD and UV methods were used. In addition, the relative standard uncertainties attributed to the FT-IR measurement results in this study reached similar values as those reported for the chemiluminescence and UV absorption techniques in CCQM-K74. This confirms that FT-IR can be operated as a comparison method when calibrated with appropriate gas standards for nitrogen dioxide measurements at the μmol/mol level, and can achieve similar measurement uncertainties to chemiluminescence and UV absorption techniques. The pilot study reported here addressed nitrogen dioxide at a relatively low amount fraction, the measurement of which can be subject to spectral interference. This choice was made in order to show the performance of FT-IR as a comparison method for a compound at a low amount fraction, and under conditions where some spectral interface is present. Very many measurements of standard gas mixtures are made at significantly higher amount fractions, and in mixtures for which there is no significant spectral interference. We expect the performance of FT-IR spectroscopy as a comparison method under such conditions to be as good as is reported here, and in some cases better. An additional pilot study (CCQM-P110-B2) was conducted on the same gas mixtures in parallel with this pilot study. The second study addressed FT-IR spectroscopy when used to measure the gas mixtures with respect to reference spectra. The results of this second study will be reported elsewhere. Page 26 of 76

27 ANNEX 1- BIPM Value assignment procedure 1. Description of the facility The BIPM-NO 2 primary gas facility combines gravimetry with dynamic generation of gas mixtures. The facility includes a magnetic suspension balance, a flow control system for the dynamic generation of gas mixtures and a flow control system for nitrogen dioxide gas standards in cylinders. Both, gas cylinder and dynamic sources of NO 2 mixtures are ultimately connected to a continuous gas analyzer ABB Limas 11 (AO2020), and to the spectrometer FT-IR Thermo-Nicolet Nexus (See Figure 9). The operation and automation of the ensemble of instruments (NO 2 FT-IR facility-abb Limas 11-FT-IR) is achieved through a LabView programme developed by members of the BIPM Chemistry Department. Through a graphical user interface the programme facilitates the setting and monitoring of all relevant instrumental parameters, automated control of complex procedures, the recording of mass measurements and NO 2 analyser readings and related data and the graphical real-time display of many of the instrument readings. Flow Control System for Rubotherm 1. Zero air generator 2. Nitrogen Generator 3. Nitrogen Cylinders 4. molbloc (0-1000) ml/min 5. SAES Nitrogen purifier 6. Mass flow controller (0-100) ml/min 7. Mass flow controller (0-1000) ml/min Rubotherm System 8. Magnetic suspension balance 9. NO 2 permeation tube V P Flow Control System for waste 11 V P Rubotherm System 8 9 P P V Flow Control System for NO2 Gas Standards 10. Mass flow controller (0-1000) ml/min 11. Multi position valve 16-position valve) 10 waste Flow Control System for NO 2 Gas P Figure 9: Schematic of the BIPM NO 2 facility The magnetic suspension balance. The magnetic suspension balance (MSB; Rubotherm, Germany) as depicted in Figure 10 is central to the system. An electromagnet is suspended from the base of the weighing pan. Below this electromagnet there is a long vertical glass vessel, the Page 27 of 76

28 measurement cell of the MSB. At the top of the glass vessel there is a permanent magnet which is held suspended by the electromagnet attached to the balance. Air buoyancy free basic load compensation Microbalance Electromagnet Glass Suspension Coupling Permanent Magnet Thermostating Chamber (for Circulating Liquid) Nitrogen Flow Measuring Load Decoupling Thermocouple Permeation tube Flange Connection Mixing chamber Nitrogen dioxide/ Nitrogen mixture Figure 10: Schematic of the BIPM NO 2 facility permeation tube chamber and magnetic suspension balance. The position of the permanent magnet is detected electronically and maintained by a servo-control of the current of the electromagnet. An NO 2 permeation tube is suspended from the permanent magnet. Thus, the balance measures the mass of the permeation tube without being mechanically in contact with it, since the balance and the weighing load are separated by a layer of glass. The coupling between the permeation tube and the balance is purely magnetic and the sensitive balance is protected from the highly corrosive NO 2 gas and the sometimes elevated temperatures and gas flows surrounding the permeation tube. This facilitates continuous monitoring of mass loss of the permeation tube, which is located in a temperature controlled environment by means of a double glass wall jacket containing water circulating at a constant temperature controlled by a remote thermostat. At constant temperature, the tube emits NO 2 through its permeable fluoropolymer membrane at a constant rate. The balance is a high resolution comparator (model AT20, Mettler, USA) with range of (0 to 22 g) and 2 μg resolution. The balance is configured with two mass pieces (see Figure 10) used to perform an external calibration of the balance. The term external calibration is used to distinguish it from the internal calibration of the balance performed with stainless steel mass standards. The two external calibration mass pieces have nominally the same Page 28 of 76

29 volume but different mass, as one is made of titanium (Ti) and the other of stainless steel (SS). Briefly, they are used to correct for an effect on the mass measurements arising from changes in the density of the ambient atmosphere surrounding the balance itself. Since it was important to know the mass difference of the Ti and SS pieces with a small uncertainty, they were calibrated in mass and volume in collaboration with the BIPM Mass Department. The flow control system for the magnetic suspension balance To generate primary mixtures using the MSB, a well characterized flow of NO 2 -free gas (nitrogen) is required. Once the flow controls system receives a pre-selected gas it delivers two well characterized flows to the balance. The total gas flow is characterized by means of a molbloc /molbox facility 3, which was calibrated at the LNE. An electronic digital pressure controller is used to maintain the pressure of the incoming gas entering the molbloc at about 2700 hpa that is the optimal pressure to minimize the uncertainty of the molbloc flow measurement (~0.1 %). The gas flow is then introduced to a gas purifier that removes the remaining water and oxygen that may leak into the gas. The gas flow is then divided in two streams, a carrier and a diluent, both regulated by two mass flow controllers (MFC s). The flow of the carrier stream is set at a constant value, 100 ml/min, mixing with the NO 2 emerging at constant rate from the permeation tube. The pressure conditions of the permeation chamber are controlled by an electronic digital pressure to avoid any buoyancy variation. The gas mixture of the carrier line is afterwards diluted by a larger flow, the diluent stream, varied within the range (0.3 to 5) L/min in order to, dynamically, generate primary NO 2 mixtures in nitrogen (or air) at various concentrations in the range (1-15) μmol/mol. Permeation tubes with permeation rates in the range ( ) ng/min are used for this purpose. The flow control system for NO 2 gas standards The third module, namely the flow control system for NO 2 gas standards, enables comparison between the dynamically generated gas mixtures and cylinder standards of NO 2 in nitrogen contained in high pressure cylinders (and, alternatively, comparison between various cylinder mixtures). This comparison is achieved via the response of the NO 2 analyser, whether ABB Limas 11 or FT-IR. The continuous gas analyzers ABB Limas 11 (part of the AO2020 series) operates according to the NDUV (Non Dispersive Ultraviolet Absorption) measurement principle. The measuring effect is specific radiation absorption of the measured gas component in the UV spectra region to detect NO 2. The FT-IR analyser is a Thermo Nicolet Nexus model enclosed in an isolation box fully described in Section A molbox facility is a support unit for making gas flow measurements using molbloc mass flow elements. The molbox hardware reads calibration data off the molbloc facility and measures molbloc upstream and downstream pressure using built-in high precision Reference Pressure Transducers (RPTs). The key molbloc L measurement is the differential pressure across the element, which is roughly proportional to the mass flow rate through it. The molbloc elements are calibrated to be used at an absolute pressure which remains nearly constant, while the differential pressure varies with flow rate. Page 29 of 76

30 The flow control system enables the sequential sampling of up to 15 standards contained in cylinders by means of a 16 position valve (MPV-16). V2 is a 4-port 2- position valve. It is used to select which sample stream, from either the MSB or from a cylinder, is directed to the analysers, the other stream being directed to waste, without perturbing the flow of either stream. 2. Measurement protocol of the BIPM On receipt by the BIPM, all cylinders were allowed to equilibrate at laboratory temperature for one week. All cylinders were rolled for 60 minutes to ensure homogeneity of the mixture. Each cylinder was connected to one inlet of a 16-inlet automatic gas sampler connected to the FT-IR spectrometer and to the BIPM NO 2 dynamic generation facility. The pressure reducers of each cylinder were flushed nine times with the mixture. The cylinder valves were then closed leaving the high pressure side of the pressure reducer at the cylinder pressure and the low pressure side of the pressure reducer at ~300 kpa. The cylinders were left to stand for at least 24 hours, to allow conditioning of the pressure reducers. Immediately prior to an analysis, each cylinder valve was opened again and the pressure reducer flushed three times. The suite of cylinders was analysed sequentially. For the FT-IR spectra acquisition 120 scans were co-added over a period of 2 minutes to provide one single beam spectrum of a sample. This single beam spectrum was then ratioed with a similar spectrum of ultra pure nitrogen collected under similar conditions to provide an absorbance spectrum of the gas sample (relative to ultra pure nitrogen). For each analyser, a calibration line was evaluated using the Generalised Least Squares approach described by ISO 6143: The assigned BIPM nitrogen dioxide value was then equal to the predicted value from a calibration line calculated from a set of dynamic nitrogen dioxide primary gas mixtures obtained from the BIPM Nitrogen Dioxide (NO 2 ) Primary Facility. 4 ISO 6143:2001: Gas analysis- Comparison methods for determining and checking the composition of calibration gas mixtures. Page 30 of 76

31 3. BIPM measurement uncertainties and analyser response The mole fractions of the dynamically produced gas mixtures obtained with the BIPM facility was calculated by the expression below: x NO2 P V qv M m NO2 M x 3 M HNO NO2 HNO3 M imp x M NO2 imp (5) where: x NO 2 is the NO 2 mole fraction in μmol/mol; P is the NO 2 permeation rate in ng/min -1 ; V m = L/mol, is the molar volume of air/n 2 at standard conditions ( K, kpa); M NO 2 = g/mol, is the molar mass of NO 2 ; q v is the total flow of N 2 given by the sum of carrier nitrogen (q v molbloc2 ) and the diluent nitrogen (F molbloc1 and) flows in Ml/min -1 at standard conditions ( K, kpa); x HNO3 is the HNO 3 mole fraction in μmol/mol measured by FT-IR spectroscopy; M HNO 3 = g/mol is the molar mass of HNO 3 ; x imp are the mole fractions of the impurities in μmol/mol measured by FT-IR Spectroscopy; and M imp are the molar mass of the impurities; Applying the uncertainty propagation law and assuming no correlation between the input quantities the following uncertainty budget was developed: u ( x 2 2 x NO xno xno x NO 2 ) u ( P) u ( V ) ( ) m u M ( ) NO u 2 P Vm M NO q 2 v 2 2 NO F xno NO 2 NO 2 2 NO u ( x HNO u ( M ( ( ) 3 HNO u x 3 imp u M (6) imp x HNO 3 HNO 3 imp imp x ) M x ) x 2 x ) M 2 The permeation standard uncertainty, considering a permeation device with a permeation equivalent to P 8357 ng/min, was estimated u P 4.18 ng/min where u P is the probability that the value of P lies within the interval ±6.17 ng/min with rectangular distribution. The uncertainty in NO 2 molar mass of g/mol, % relative, can be derived from the IUPAC Table of Atomic Weights. Page 31 of 76

32 The molar volume V m of a real gas at standard conditions (T = K, p = kpa) is given by the formula ZRT V m (7) p where Z is the compressibility factor and R is the gas constant, J/mol/K, with relative u(r) of Since they are defined by convention there is no uncertainty in T and p. The compressibility factor of nitrogen obtained from the NIST Refprop database is Z N2 = with relative u(z) of Thus the molar volume of nitrogen and its standard uncertainty are V mn2 = L/mol u(v mn2 ) = L/mol, or relative. The BIPM measured the flow in its system by using molblocs. These were calibrated by the LNE on 27 April The uncertainty of the BIPM s flow measurements is dominated by and based on calibration. The uncertainty in the flow measurements u(q v ) was taken from the LNE calibration certificate N K20869/1. No additional component for the stability of the flow instrument was added, since the time between calibration and the first measurements were short, and no significant deviation between the first and second series of BIPM measurement results was observed for stable cylinder gas standards. The expanded relative uncertainty (k=2) quoted in the calibration certificate is 0.2 % at the flows used in the comparison. In correspondence between the BIPM and the LNE, the LNE confirmed the relative expanded uncertainties quoted in their CMCs, comparison results and the calibration certificates to be as follows: % to 0.40 % in LNE s CMCs % to 0.26 % in the Euramet comparison reference (2) % to 0.27 % in the Calibration Certificate K20869/1. The uncertainty in the calculated nitric acid mole fraction, x HNO3, obtained by FT-IR spectroscopy, is given by: u x x 2 (8) x HNO3 where x is the mole fraction of nitric acid predicted by FT-IR into the gas mixtures. A future publication will give a detailed description of the measuring methodology and quantification process by FT-IR for the determination of nitric acid. As for NO 2, the uncertainty in nitric acid molar mass, g/mol ( % relative), was derived from the IUPAC Table of Atomic Weights. It follows that the uncertainty budget for a NO 2 mixture having a nominal concentration of ~10.0 μmol/mol is as tabulated below in Table 8: Page 32 of 76

33 Quantity P V m q v molbloc1 M NO2 x HNO3 x N2O4 x N2O3 x N2O5 x HONO x HO2NO2 M HNO3 Estimate x i Assumed distribution Standard uncertainty Sensitivity coefficient Uncertainty u(x i) c i= x NO2/ x u i(y) Index contribution % mol/mol g/min Normal 10 9 g/min Normal L/mol 10 6 L/mol L/min Normal L/min Normal g/mol 10 3 g/mol Normal mol/mol mol/mol Normal mol/mol mol/mol Normal mol/mol mol/mol Normal mol/mol mol/mol Normal mol/mol mol/mol Normal mol/mol mol/mol Normal g/mol 10 3 g/mol Quantity Value Standard Uncertainty x NO μmol/mol μmol/mol Table 8. Uncertainty budget for a NO 2 /N2 primary mixture generated with the BIPM facility. The degrees of freedom were numerous, so a coverage factor k = 2 was assumed appropriate for the expanded uncertainty. The main uncertainty contributors remain the mole fraction determination of nitric acid and the gas flow measurements. Figure 11 illustrates the new uncertainties in NO 2 x for the dynamic generation of NO 2 in nitrogen mixtures over the mole fraction range (8-12) μmol/mol, using a permeation tube with permeation rate of 8357 ng/min -1 and flows in the range ( ) ml/min. The uncertainty is almost a constant and can be fitted by a linear function of the mole fraction. A least squares fit was made using the Excel LINEST function. The standard uncertainties in NO 2 values in µmol/mol): x can be modelled by the following linear function (numerical Page 33 of 76

34 ux ( ) x (9) NO y = x R 2 = u (x NO2 ) / (μmol/mol) x NO2 / (μmol/mol) Figure 11. Standard uncertainty of dynamically generated NO 2 mixtures on the BIPM NO 2 facility over a range of x = (8-12) μmol/mol. NO 2 4. Covariance between two dynamically generated gas mixtures Non-zero covariances, ux ( NO 2, i, x NO 2, j) were included in the uncertainty calculations because all dynamic mixtures were derived from the same BIPM facility and an error in the analyte content of the one gas is considered to propagate to all gas mixtures in a positive correlated fashion. The covariance between two calibration gas mixtures i and j is described as follows: NO 2, i NO 2, j NO 2, i 2 ux (, x ) ux ( ), (10) Where u( x NO2, i ) is the standard uncertainty of the more concentrated mixture as given by equation 10, q j (11) q i is the dilution factor of the total gas flows q j and q i (with q j < q i ). Note that as the NO 2 calibration gas mixtures generated with the facility are distributed in a small range of mole fractions (typically 8 nmol/mol to 12 nmol/mol), the dilution factor is often close to 1, and the covariances often close to the variances u(x NO2,i ) 2. Page 34 of 76

35 5. FT-IR analysis of gas standards Analysis of all gas standards was undertaken to quantify nitric acid within the gas standards, and to compare these with the impurities and their uncertainties reported by the participating laboratories. 6. FT-IR Spectra acquisition procedure A ThemoNicolet Nexus FT-IR spectrometer was configured with a MCT-high D* liquid N 2 -cooled mid-infrared detector and a 6.4 m pathlength multipass White cell (Gemini Scientific Instruments, USA) for the purposes of quantitative analysis for gas reference standards. The White cell has wetted surfaces of only electropolished stainless steel and gold (mirror coatings) to minimise surface interactions with reactive gas phase species. To keep the internal optical path of the spectrometer free of any interference species this ensemble has been placed in stainless steel enclosure which is constantly purged with ultra high purity nitrogen (dewpoint ~-95 C, i.e. ~200 nmol/mol -1 H 2 O) flowing at ~15 L/min -1. The gas sample, from either the Rubotherm MSB or from a high pressure cylinder, flows from the NO 2 facility sampling manifold through the White cell, and then to waste. The sample flow rate is controlled immediately downstream of the White cell at ~400 ml/min -1. The sample pressure and temperature are measured on real time by means of a calibrated barometer (Series 6000 Digital Pressure Transducer, Mensor, USA) and a calibrated 100 Ω RTD temperature probe attached to the White cell. The spectrometer user interface is by means of the IMACC software. IMACC allows the automatic setting of all instrument parameters into Thermo's proprietary Omnic software for the control, spectra acquisition and on-line analysis. For the acquisition of high quality spectra suitable for quantitative analysis, 120 scans are co-added over a period of 2 minutes to provide one single beam spectrum of a sample. This single beam spectrum was then ratioed with a similar spectrum of ultra pure nitrogen collected under similar conditions to provide an absorbance spectrum of the gas sample (relative to ultra pure nitrogen). The White cell has a volume of ~750 ml and the sample flows at ~400 ml/min. Assuming perfect mixing in the cell we estimate that an initial sample at time t = 0 s has been 99.9 % replaced after 10 min of flow, and % replaced after 20 min. Accordingly, to ensure complete exchange of sample, spectrum acquisition started at t = 0 but only the measured spectra obtained after flowing the sample through the White cell for 35 min were used for the mole fraction determination. We also empirically verified that after 30 min of flow, the sample was completely exchanged, within the bounds of measurement uncertainty. The absorbance spectra of gas reference standards obtained following this procedure had a very high signal: noise ratio, with the level of noise in the baseline being typically ~ abs 10 peak-peak. By comparison the main NO 2 peak had absorbance in the range ( ) abs 10. Page 35 of 76

36 As the FT-IR was calibrated with our permeation device, its sole contributing uncertainty component is the type A uncertainty. From times series analysis the uncertainty in the response of the FT-IR spectrometer was estimated in 20 nmol/mol for 2 min averaging time. 7. Quantitative analysis of nitric acid The determination of nitric acid was assessed configuring the FT-IR facility with a multi pass white cells with an optical path of (48±1.2) m. Spectra were analysed by a non-linear least-square fitting of the measured absorption spectra with synthetic spectra using the program MALT4.4. This program included the calculation of synthetic spectra from the HITRAN database of infrared absorption line parameters using the core of the program MALT (an acronym for Multiple Atmospheric Layer Transmission) software developed at the University of Wollongong described in detail by Griffith in 1996(3). The program convolved a stick spectrum calculated from the line parameters with the temperature, pressure, path length, resolution and instrument line shape function specified by the user. Spectra were calculated iteratively from an initial estimate of all input parameters following a modified Levenberg-Marquart algorithm until a least squares best fit to the measured spectrum was obtained. Gas concentrations in the sample were iteratively adjusted during the fit. The quality of the fit could be improved by choosing a proper spectra window of the measured spectrum. Spectra which had been acquired across a total wavelength range of ( ) cm -1 were fitted on spectral windows according to the impurities of interest, in this case nitric acid. 8. Uncertainty budget Table 9 below summarises the uncertainty sources and presents the final combined uncertainty associated with the FT-IR/MATL/CLS measurements of nitric acid at a mole fraction (x) ranging from 100 nmol/mol to 250 nmol/mol with a FT-IR white cell with a 48 m optical path. Type A μmol/mol Stability Type B MALT 0.015x HITRAN 0.05x Combined uncertainty x x 2 (12) Table 9: uncertainty budget associated with the FT-IR spectrometer used as an absolute method of quantification to determine the concentration of HNO 3 in nitrogen. Page 36 of 76

37 9. Regression analysis The procedure outlined in ISO 6143:2001 (Gas analysis-comparison methods for determining and checking the composition of calibration gas mixtures) was used for the analysis of the data from the comparison. This required: - the determination of the analysis function x=g(y) which expressed analyte contents in relation to corresponding measured responses; - the validation of the analysis function; and - the prediction of the mole fraction values from the measured responses and comparison to VSL and NMI s values. 10. Determination and validation of analysis functions All calculations were performed with B_LEAST, a computer programme which implemented the methodology of ISO 6143:2001, and takes into consideration uncertainties in both axes for regression analysis. Validation studies performed by the BIPM to be published shortly will confirm the linearity of the FT-IR response in the x NO2 range ( ) μmol/mol. Page 37 of 76

38 ANNEX 2 - Measurement reports of participants Centro Español de metrología (CEM) B1-1. General information Institute CENTRO ESPAÑOL DE METROLOGÍA (CEM) Address CALLE ALFAR, TRES CANTOS (MADRID) SPAIN Contact person TERESA E. FERNÁNDEZ VICENTE Telephone Fax * tefernandez@cem.mityc.es Serial number of cylinder (D650059) received Cylinder pressure as received 95 bar B1-2. Results Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO 2 10,62 0,89 2 B1-3. Uncertainty Budget The mathematical mode used to calculate the uncertainty in the composition of mixture analyzed is a linear combination of the sources of uncertainty due to the instrument used and the repeatability of the measurements. This leads to: u ur uzero ul where u r is the standard deviation of the mean of the results obtained along the period of measurements, from the linear fit regression by means of the IFSS software; u is the largest uncertainty among the obtained uncertainties zero due to the linear fit regression by means of the IFSS software, that doesn t force zero point; and u L is the uncertainty of the path length due to the influence of the temperature and the pressure in the cell and the path length itself. Table 1 summarizes the uncertainties budget. Page 38 of 76

39 Uncertainty source r Assumed distribution Standard uncertainty / mol/mol Sensitivity coefficient Contribution to standard uncertainty / mol/mol u normal 0,41 1 0,41 u zero rectangular 0,14 1 0,14 u L normal 0,29-0,32 0,094 Combined standard uncertainty / mol/mol 0,45 Expanded uncertainty, k 5 = 2 / mol/mol 0,89 Table 1. Detailed uncertainty budget. B1-4. FTIR instrumentation and acquisition parameters Spectrometer Manufacturer Type Serial number Gas cell FTIR PerkinElmer GXI Manufacturer Specac Type SC Tornado T-20 Optical path (m) 20 m Operation software details Name of the software used to control the spectrometer Name of the software used to acquire spectra Name of the software used to analyse spectra Spectrum Software Spectrum Software IFSS (Imacc Software) B1-5. Description of the procedure used during the gas analysis Upon arrival the sample cylinder was rolled and stored in the laboratory under laboratory reference conditions. A pressure reducer was connected to the sample cylinder. The reducer was carefully flushed as prescribed in International Standard ISO 16664:2004 (Gas analysis Handling of calibration gases and calibration gas mixtures Guidelines). The sample cylinder was connected to a vacuum tree system connected to the FTIR. When the vacuum in the cell was kpa and the vacuum in the tree was kpa, 16 background scans were collected. With the pressure in the cell set next to 102,3 kpa, 100 scans were collected. Due to the fact that the system lacks a sensor for the measurement of the temperature inside the gas cell, the experimental temperature corresponds to the environmental conditions in the laboratory, next to 293 K. 5 The coverage factor shall be based on approximately 95 % confidence. Page 39 of 76

40 The sample results reported come from the data obtained along three consecutive working days, from February the 16 th to the 18 th. Three standards were used with the compositions specified in Table 2: Species Amount Fraction NPL1272 / mol/mol Amount Fraction NPL1273 / mol/mol Amount Fraction NPL1274 / mol/mol Nitrogen Dioxide (5,01 ± 0,10) (10,00 ± 0,15) (15,01 ± 0,22) Oxygen (not certified) Nitrogen Balance Balance Balance Table 2. Primary reference gas mixtures used. All mixtures were prepared gravimetrically and analysed by Non-Dispersive Ultraviolet (NDUV) technique. Protocol B1 and B2 were carried out simultaneously. B1-6 Complementary information on the cylinder The value of the pressure left in the cylinder before shipment to the BIPM was 65 bar approximately. Page 40 of 76

41 Korea Research Institute of Standards and Science (KRISS) General information Institute Address KRISS 1 Korea Research Institute of Standards and Science (KRISS), P.O.Box 102, Yusong, Daejeon, Republic of Korea Contact person Lee, Jeongsoon for P110 B1 and b2 Oh, Sanghyup for K74 Telephone Fax * Serial number of cylinder received Cylinder pressure as received Results Cylinder No.: D leejs@kriss.re.kr D About 65 bar Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO k = 2 Uncertainty Budget Please provide a complete uncertainty budget. Uncertainty Standard uncertainty factor μmol/mol Gravimetry uncertainty (PRM) 0.15 Analysis uncertainty 0.05 Total uncertainty 0.16 FTIR instrumentation and acquisition parameters Page 41 of 76

42 Spectrometer Manufacturer Bruker Type TENSOR 27 Serial number T Gas cell Manufacturer Type OTSUKA Optical path (m) MULTI PASS CELL Software Name of the software used to control the spectrometer Name of the software used to acquire spectra Name of the software used to analyse spectra FTIR COLLECTION MANAGER (FCM) OPUS FCM FCM Description of the procedure used during the gas analysis Please describe in detail the analytical method(s) used for gas analysis and gas standards used for calibration 6. We carried out ABA measurement in sequence to acquire the BIPM sensitivity comparable to that of the PRM (A) which was prepared with gravimetry by KRISS. Calibration of our FTIR analyzer was conducted at one point with PRM (A) cylinder with 0.30 ppm uncertainty (k = 2). Gas from cylinder was introduced as: Cylinder >> regulator >> MFC >> gas Cell >> vent to atmosphere or Vacuum during pumping Followings are the procedures to acquire the spectrum; VACUUM less than 1.0 mb >> GAS CELL FILL with N2 TILL 1013 mb >> measurement BG >> VACUUM less than 1.0 mb >> GAS CELL FILL with NO2 TILL 1013 mb >> measurement PRM (A) >> VACUUM less than 1.0 mb >> GAS CELL FILL with NO2 TILL 1013 mb >> measurement BIPM sample (B) >> VACUUM less than 1.0 mb >> GAS CELL FILL with NO2 TILL 1013 mb >> measurement PRM (A) B1-6 Complementary information on the cylinder Please report the value of the pressure left in the cylinder before shipment to the BIPM: 6 The choice of the procedure used for gas analysis is the responsibility of the participating laboratory. Nevertheless, for a proper evaluation of the data, it is necessary that the calibration method, as well as the way in which the calibration mixtures have been prepared is reported to the co-ordinators.. Page 42 of 76

43 PRESSURE LEFT: 52 BAR If any other component other than NO 2, nitrogen and oxygen was detected and/or quantified please report its mole fraction in the table below: N.A. Component Mole fraction / nmol/mol Expanded uncertainty Coverage factor Measurement technique Page 43 of 76

44 National Measurement Institute Australia (NMIA) General information Institute Address NMIA National Measurement Institute Australia Bradfield Road Lindfield NSW 2070 AUSTRALIA Contact person Damian Smeulders Telephone Fax * damian.smeulders@measurement.gov.au Serial number of cylinder received Cylinder pressure as received Results 100 bar Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO Uncertainty Budget Combined standard uncertainty: u = 0.32 μmol/mol Expanded uncertainty: U = 0.63 μmol/mol Contributions to uncertainty: Gravimetric uncertainty: Mixture stability and conversion to NO 2 : Instrument contributions: Repeatability: 0.30 Resolution: Difference due to spectral regions: FTIR instrumentation and acquisition parameters Spectrometer Manufacturer Type 6100 Thermo Nicolet Page 44 of 76

45 Serial number Gas cell Manufacturer Type Optical path (m) 10 Operation software details AHR Thermo Nicolet Nickel plated aluminium. KBr windows Name of the software used to control the spectrometer Name of the software used to acquire spectra Name of the software used to analyse spectra Omnic Omnic Excel Description of the procedure used during the gas analysis A Nicolet FTIR was used to acquire the spectra of the standards and unknown sample. The spectra were run at resolutions of 0.5cm -1 and 0.25cm -1 with an aperture setting of scans were obtained for each analysis. The background spectra was collected on the evacuated cell. Spectra were collected on a static gas sample with a temperature of 60ºC at a pressure of 650 Torr. The strong bands in the region cm -1 and the weaker bands in the region cm -1 were both used for quantitation. The analyses of the standards and sample were repeated three times at each resolution with evacuation and flushing of the cell between tests. The analysis procedure was repeated on several occasions over a two week period. Four closely bracketed calibration standards containing NO2 over the concentration range 8-12 µmol/mol were used to determine the concentration of NO2 in the cylinder from the BIPM. Standards were made in uncoated, but passivated 5L Luxfer aluminium cylinders with SS valves. Standards were manufactured from nitrogen oxide that was converted to nitrogen dioxide in the presence of oxygen. Oxygen in the final mixtures was present at approximately 1000 µmol/mol. B1-6 Complementary information on the cylinder Please report the value of the pressure left in the cylinder before shipment to the BIPM: 78 bar If any other component other than NO 2, nitrogen and oxygen was detected and/or quantified please report its mole fraction in the table below: Component Mole fraction / nmol/mol Expanded uncertainty Coverage factor Measurement technique Page 45 of 76

46 National Metrology institute of Japan (NIMJ) General information Institute National Metrology institute of Japan Address Umesono, Tsukuba Ibaraki Contact person Nobuyuki Aoki Telephone Fax * Serial number of cylinder received APEX Cylinder pressure as received Results 10MPa Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO Uncertainty Budget Nitrogen dioxide mol fraction was estimated from calibration curve obtained by measuring four primary standards. Calibration curve was regression line (x =G=b 0 +b 1 y) and calculated according to the Deming's generalized least-squares method. The standard uncertainty of the analyte content, u(x) using the propagation of uncertainty on the measured response and on the parameters of the analysis function, as follows. u 2 2 G y G b x u y u b u b 2 u b, b 0 G b G G b b u(y): the standard uncertainty of the response y u(b 0 ), u(b 1 ): the variance of the parameter b 0 and b 1 of the analysis function u(b 0, b 1 ): the covariance of the parameters b 0 and b 1 of the analysis function Page 46 of 76

47 Table 1. Parameter of the Deming's generalized least-squares method Parameter Value b b u(y) u(b 0 ) u(b 1 ) u(b 0, b 1 ) FTIR instrumentation and acquisition parameters Spectrometer Manufacturer Type Serial number JASCO Corporation FTIR-6100 Gas cell Manufacturer Type Optical path (m) 10cm Software JASCO Corporation 10cm gas cell Name of the software used to control the spectrometer Name of the software used to acquire spectra Name of the software used to analyse spectra Spectra Manager Spectra Manager Spectra Manager Description of the procedure used during the gas analysis Calibration standards: Four calibration standards were used for the determination of nitrogen dioxide concentration in nitrogen. The standards were prepared from pure nitrogen dioxide, pure nitrogen, and pure oxygen in accordance with ISO6142:2001 (Gas analysis-preparation of calibration gases-gravimetric method. Pure nitrogen dioxide was from Sumitomo Seika Chemicals Company Limited and pure nitrogen and oxygen from Japan Fine Products. Two-step dilution was used to make the mixtures, with nitrogen dioxide Page 47 of 76

48 concentration of (2000~4000) mol/mol, and (9~18) mol/mol. Oxygen was added in the first-step dilution. Table 2 shows characteristics of the calibration standards. Table 2. Gravimetric value and expanded uncertainty in calibration standards Gravimetric value of Cylinder number O Gravimetric Value of 2 ( mol/mol) NO 2 ( mol/mol) Expanded uncertainty [k=2] (nmol/mol) YAO YA YA YA Instrument calibration: The following measurement cycle was repeated 7 times for the determination of nitrogen dioxide concentration in air: STD 1 - STD 2 sample STD 3 STD 4 The response (y i ) to analyte content in each cycle was the mean value of five individual responses. 5 l 1 l y i yi, j 5 j 1 Calibration data ( y STD, y 1 STD, y 2 STD, y 3 STD4 seven cycles response ( 7 1 y i, 2 y i, 3 y i, 7 y i )., y sample ) were obtained from the mean value of 1 yi 1 y l i 7 The calibration data was used in order to determine the concentration of nitrogen dioxide in synthetic air. The determination process was according to ISO 6143 using the Deming s generalized least-square method. The four calibration standards listed in table 2 were used for instrument calibration. Sample handling: The sample cylinder was stood at room temperature after arrival. Samples were transferred to a gas cell with a stainless steel pressure regulator. the regulator was evacuated before samples were introduced to gas cell. in addition, the measurement was performed about ten minutes after sample introduction to the gas cell. Page 48 of 76

49 National Institute of Standards and Technology (NIST) General information Institute Address National Institute of Standards and Technology 100 Bureau Drive, Gaithersburg, MD , USA Contact person Franklin R. Guenther, Lyn Gameson Telephone Fax * Serial number of cylinder received Cylinder pressure as received Results APEX Mpa Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO Uncertainty Budget Please provide a complete uncertainty budget. Uncertainty Source, X I Assumed Distribution Standard Uncertainty (% Relative), u(x i ) Sensitivity Coefficient, c I Gravimetric Standard or Analytical Component Certified working standard Gaussian Analytical Instrument Reproducibility Gaussian Analytical Change in gas matrix Gaussian Analytical Non linear Gaussian Analytical Cell pressure Gaussian Analytical Cell temperature Gaussian Analytical Page 49 of 76

50 FTIR instrumentation and acquisition parameters Spectrometer Manufacturer Type Serial number AEQ Nicolet Nexus 670 (MCT Detector) Gas cell Manufacturer Type Optical path (m) 10 Specac Cyclone 10C (Quartz) Software Name of the software used to control the spectrometer OMNIC V7.1 Name of the software used to acquire spectra OMNIC V7.1 Name of the software used to analyse spectra In-House Description of the procedure used during the gas analysis Please describe in detail the analytical method(s) used for gas analysis and gas standards used for calibration 7. FTIR Collection Parameters: Parameter Value Number of Scans 512 Apodization Boxcar Resolution cm -1 Standard used (in balance air): Cylinder# NOx (µmol/mol) HNO 3 (µmol/mol) NO 2 (µmol/mol AAL ± ± ± FTIR Response summation of the following NO 2 peak areas: Start (cm -1 ) End (cm -1 ) 7 The choice of the procedure used for gas analysis is the responsibility of the participating laboratory. Nevertheless, for a proper evaluation of the data, it is necessary that the calibration method, as well as the way in which the calibration mixtures have been prepared is reported to the co-ordinators.. Page 50 of 76

51 Analytical methodology: Sample flow (200 ml/min) through the gas cell was controlled by a needle valve. Each sample was purged through the cell for 60 minutes followed by coadding 512 scans (total collection time was 50 minutes). At least two spectrums were collected per sample. The cell pressure (measured by Mensor Series 6000 Pressure Transducer, Serial# ) was auto collected every minute throughout the FTIR acquisition; cell temperature was monitored manually. The nine NO 2 FTIR peaks above were identified as being well formed and free from water interference. They were summed to give a raw instrument response. The cell temperature and pressure were averaged for each spectrum acquired. Using the standard gas equation, a ratio was calculated for converting these cell conditions to standard temperature (25 o C) and pressure (760 mm Hg). The normalized FTIR response was achieved by multiplying the raw response by this ratio. B1-6 Complementary information on the cylinder Please report the value of the pressure left in the cylinder before shipment to the BIPM: 6.1 MPa If any other component other than NO 2, nitrogen and oxygen was detected and/or quantified please report its mole fraction in the table below: Component Mole fraction / nmol/mol Expanded uncertainty Coverage factor Measurement technique Page 51 of 76

52 National Physical Laboratory (NPL) General information Institute Address National Physical Laboratory (NPL) Hampton Road, Teddington, Middlesex, TW11 0LW, UK Contact person * Serial number of cylinder received Cylinder pressure as received Results Alice Harling / Martin Milton alice.harling@npl.co.uk / martin.milton@npl.co.uk D bar Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO K = 2 Uncertainty Budget Source of uncertainty Estimation Method Standard uncertainty Nominal value Relative standard uncertainty Gravimetric preparation A 13 nmol/mol nmol/mol 0.13% of standard Drift in gravimetric value B 10 nmol/mol nmol/mol 0.10% of standard Repeatability of A 0.35% 0.35% measured area Interference of HNO3 in B 0.5% 0.50% analyser Cell pressure B 1 mbar 1050 mbar 0.1% Cell temperature B 0.5K 298 K 0.2% Combined uncertainty 0.67% The expanded uncertainty (k=2) is 1.4 %. Note: since the optical path length of the cell remained unchanged during the experiment, it is not incorporated as a source of uncertainty. Page 52 of 76

53 FTIR instrumentation and acquisition parameters Spectrometer Nicolet 6700 Manufacturer Thermo Scientific Type 6700 Serial number AHRO Gas cell Manufacturer Type Optical path (m) Software 8 m Specac Cyclone C5 Name of the software used to control the spectrometer Name of the software used to acquire spectra Name of the software used to analyse spectra OMNIC OMNIC OMNIC Description of the procedure used during the gas analysis Please describe in detail the analytical method(s) used for gas analysis and gas standards used for calibration 8. Calibration Standards Gravimetric standards were prepared to compare against the unknown (D ). Standard A (NPL1275R) contains ppm NO 2 in N 2. Standard B (NPL1126R2) contains ppm NO 2 in N 2. Parents Daughter pure NO + N2 50 mmol/mol NO/N2 74R2 50 mmol/mol NO/N2 + 9% O2/N2 + N mol/mol NO2/N2 464R 4000 mol/mol NO2/N2 + N2 800 mol/mol NO2/N2 1206R2 800 mol/mol + 9% O2/N2 + N2 100 mol/mol NO2/N2 1117R4 8 The choice of the procedure used for gas analysis is the responsibility of the participating laboratory. Nevertheless, for a proper evaluation of the data, it is necessary that the calibration method, as well as the way in which the calibration mixtures have been prepared is reported to the co-ordinators.. Page 53 of 76

54 100 mol/mol NO2/N2 + N2 10 mol/mol NO2/N2 1275R FTIR Acquisition Conditions The system was setup to identify the optimum resolution, path length and pressure necessary for this analysis. The optimised conditions chosen for the measurement were Detector MCT (liquid N 2 cooled) No of scans / Scan time 64 Apodisation Happ-Genzel Nominal Resolution 0.5 cm -1 True Resolution ( Data point Spacing ) cm -1 Path length 8m Temperature of gas cell 298K Measurement Procedure 1. The gas cell was evacuated to 5 x 10-4 mbar, and the FTIR system purged with Metrology Grade nitrogen for several hours. 2. The gas cell was filled with BIP N When the pressure in the cell (indicated on the Baratron pressure gauge) reached 1050 mbar, the exhaust to the cell was opened and the BIP N 2 left to flow through and flush the cell. During this time the reduction in carbon dioxide in the background spectrum was monitored. 4. After 10 minutes the background was collected (over 64 scans) 5. The BIP N 2 was stopped and the gas line and regulator from the FTIR system to the cylinder to be measured were purged and evacuated. 6. The gas cell was then re-evacuated to 5 x 10-4 mbar. 7. The sample was allowed to flow into the gas cell until the pressure read ~1050 mbar, then the exhaust was opened and the sample flowed through the cell during the measurements. 8. Five consecutive measurements were made for each sample (over 64 scans), using the same background. Results and spectroscopy The nitrogen dioxide peak at 1600 cm-1 was used for the measurements. The results from a second peak at 2860 to 2940 cm-1 are also given for information. Page 54 of 76

55 The samples were run in the following sequence on 9th Dec 2009: Background Standard NPL 1275R Background Unknown D Background Standard NPL 1126R2. The integration of the peak areas was carried out using the Thermo Scientific software OMNIC. The results were validated using a MATLAB code developed at NPL. The value of the unknown was calculated by means of a ratio of the FTIR response against one of the standards. In each case, the area counts given in the table are after background subtraction. The result given is the mean of the analysis carried out against the two standards. Results of analysis 9/12/2009 AREA COUNTS using Modified area counts OMNIC software (taking into account pressure ) NO2 (1) NO2 (2) NO2 (1) NO2 (2) Run # Matrix Cylinder Grav conc (ppm) File name Pressure (mbar) Temperature (K) cm cm cm cm-1 1 BKG_ Air NPL1275R NPL1275R_091209_ BKG_ Air NPL1275R NPL1275R_091209_ Air NPL1275R NPL1275R_091209_ Air NPL1275R NPL1275R_091209_ Air NPL1275R NPL1275R_091209_ AV %STDEV 6.56% 0.35% 6.56% 0.35% 8 BKG_ N2 D D650042_091209_ N2 D D650042_091209_ N2 D D650042_091209_ N2 D D650042_091209_ N2 D D650042_091209_ AV %STDEV 10.96% 0.28% 10.96% 0.28% 14 BKG_ N2 NPL1126R NPL1126R2_091209_ N2 NPL1126R NPL1126R2_091209_ N2 NPL1126R NPL1126R2_091209_ N2 NPL1126R NPL1126R2_091209_ N2 NPL1126R NPL1126R2_091209_ AV %STDEV 5.80% 0.36% 5.80% 0.36% Calculated conc of D (ppm) Standard NO2 (1) NO2 (2) NPL1275R NPL1126R Note: the modified area counts are corrected from the measured pressure to a standard pressure of 1056 mbar. B1-6 Complementary information on the cylinder Page 55 of 76

56 Please report the value of the pressure left in the cylinder before shipment to the BIPM: 40 bar If any other component other than NO 2, nitrogen and oxygen was detected and/or quantified please report its mole fraction in the table below: Component Mole fraction / nmol/mol Expanded uncertainty Coverage factor Measurement technique HNO nmol/mol K=2 FTIR (library spectra) Page 56 of 76

57 National Metrology Institute of South Africa (NMISA) General information Institute National Metrology Institute of South Africa (NMISA) Address CSIR, Building 5 Meiring Naudé Road Brummeria, 0184 Pretoria South Africa Contact person A Botha Telephone +27(0) Fax +27(0) * abotha@nmisa.org Serial number of cylinder received Cylinder pressure as received D bar Results Nitrogen dioxide mole fraction Expanded uncertainty Coverage factor x / μmol/mol U ) / μmol/mol NO2 ( x NO 2 10,62 0,22 2 Uncertainty Budget Parameter Temperature Pressure Verification Stability Gravimetry Bias Standard uncertainty(u) 0,11% rel 0,09% rel 0,18% rel 0,10% rel 0,13% rel 1% rel Page 57 of 76

58 FTIR instrumentation and acquisition parameters Spectrometer Manufacturer Type Serial number: ADU Gas cell Manufacturer Type Optical path (m):10 Software Nicolet Instrument Corporation Magna IR 560 E.S.P Gemini Scientific Instruments Gemini-Mars Name of the software used to control the spectrometer Omnic version 8 Name of the software used to acquire spectra Omnic version 8 Name of the software used to analyse spectra Malt5 version 5.2 Description of the procedure used during the gas analysis After the arrival of D cylinder in the laboratory, the cylinder was stabilised at room temperature (22 ºC ± 2 ºC) and humidity of (50 % ± 10%) before checking the pressure and doing measurements. The standards and sample were transferred directly to the FTIR using a system composed of pressure regulator, mass flow controller and control valves. The measurements of D received from the coordinator (BIPM) were made during January 2010 by direct comparison with four gravimetric preparation standards by NMISA containing 10 μmol/mol concentrations. The results reported all follow from a bracketing comparison strategy. For FTIR measurements a NICOLET-Magna-IR-560 was used. A fixed optical path length gas cell of 10 m from Gemini-Mars was used. Measurement conditions: 0.5 cm -1 resolution, 64 scans, phase correction (Mertz), Happ-Genzel apodization, water correction performed by the OMNIC-Software, zero level of zero-filling, liquid nitrogen cooled mercury cadmium telluride (MCT/A) detector. Nominal measurement pressure for all samples: kpa. The optical bench set up was continuously purged with BIP TM + N 2 (6.0). Page 58 of 76

59 For every measurements of cylinders a fresh Background BIP TM + N 2 (6.0) was used. Cylinders were measured randomly at continuous flow (i.e 1000 ml/min) controlled by mass flow controller. The quantification was performed by MALT version 5.2 software. B1-6 Complementary information on the cylinder Please report the value of the pressure left in the cylinder before shipment to the BIPM: 23 bar If any other component other than NO 2, nitrogen and oxygen was detected and/or quantified please report its mole fraction in the table below: Component Mole fraction / nmol/mol Expanded uncertainty / nmol/mol Coverage factor Measurement technique HNO FTIR Page 59 of 76

60 Slovak Institute of Metrology (SMU) Page 60 of 76

61 Page 61 of 76

62 Page 62 of 76

63 Page 63 of 76

64 Page 64 of 76

65 Page 65 of 76

66 Page 66 of 76

67 Page 67 of 76

68 Mendeleyev Institute for Metrology (VNIIM) Page 68 of 76

69 Page 69 of 76

70 Page 70 of 76