Comparison APMP.QM-S2.1 Oxygen in nitrogen at atmospheric level

Transcription

1 APMP.QM-S2.1 report Comparison APMP.QM-S2.1 Oxygen in nitrogen at atmospheric level Final Report ByungMoon Kim 1, Kwangsub Kim 1, Jinsang Jung 1,*, Sanghyub Oh 1, Liu Hui 2, Hou Li 2, Teo Beng Keat 2, Chua Hock Ann 2 1 Korea Research Institute of Standards and Science (KRISS), Center for Gas Analysis, 267 Gajeong-ro, Yuseong-Gu, Daejeon , Republic of Korea 2 National Metrology Centre, A*STAR, #02-27 TUV SUD PSB building, 1 Science Park Drive, , Singapore * Corresponding to Jinsang Jung: jsjung@kriss.re.kr Field Amount of substance Subject Comparison of 0.2 mol/mol oxygen in nitrogen Participants KRISS(Korea), NMC/A*STAR(Singapore) Organizing body APMP Table of contents Introduction...3 Supported CMC claims...3 Schedule Process of the comparison..4 Measurement protocol....5 Measurement methods...5 1

2 Degrees of equivalence...6 Result Conclusion Appendix A. Verification of mixtures with GC-TCD...10 Appendix B. Report from each laboratory. 13 2

3 Introduction NMC/A*STAR has a schedule to start the calibration service of oxygen in nitrogen gas mixtures near atmospheric concentration in recent years. NMC/A*STAR is required to show its competence in measurement and calibration of oxygen at these concentration levels. KRISS and NMC/A*STAR agreed to collaborate in the area of gas metrology. KRISS organize a bilateral comparison between KRISS and NMC/A*STAR to show the comparability between them. This comparison was suggested and approved in the meetings of APMP TCQM in November 2013 and CCQM GAWG in April This document describes results of the bilateral comparison of an oxygen in nitrogen gas mixture. The nominal amount-of-substance fraction was 0.2 mol/mol oxygen in nitrogen. Supported CMC claims This comparison can be used to support CMC claims for oxygen in nitrogen matrix from 0.05 mol/mol to 0.3 mol/mol if same impurity analysis and uncertainty evaluation are performed based on participants reports. Schedule November 2013 Protocol issued by KRISS January 2014 Preparation of mixtures and first verification measurement April 2014 Shipment of sample cylinder to participating laboratory (NMC/A*STAR) 31 July 2014 Submission of measurement report to KRISS September 2014 Return of the sample cylinder to KRISS March 2015 Second verification measurement April 2015 Draft A report 3

4 April 2015 August 2015 October 2015 October 2015 Draft B report Draft B APMP TCQM review Draft B CCQM GAWG review Final approved Process of the comparison A set of mixtures of oxygen in nitrogen with nominal fractional amounts of 0.2 mol/mol was prepared gravimetrically according to ISO 6142 [1]. The mixtures were verified against primary reference mixtures. The pressure in the cylinders was approximately 100 bar and Luxfer cylinders of 10 dm 3 nominal were used. The amount-of-substance fractions were derived from gravimetry, molar mass, and purity verification of the parent gases. The gravimetric values were used as the Key Comparison Reference Values (KCRVs). Thus, each cylinder has its own reference value. The participating laboratory was requested to specify in detail which analytical method(s) were used and how the measurement uncertainty was evaluated. The participating laboratory was responsible for the calibration of its own equipment. For a proper evaluation of the data, it was necessary that the calibration method, as well as the way in which the calibration mixtures were prepared, were reported to the coordinating laboratory. The laboratory was asked to express the uncertainty on all results withthe evaluation of measurement uncertainty in accordance with the Guide to the express of uncertainty in measurement (ISO GUM). The participant was asked to provide a detailed description of the uncertainty budget, including: - method of evaluation (type A or B) - (assumed) probability distribution - standard uncertainties and sensitivity coefficients 4

5 After the measurement, the participating laboratory was requested to return the cylinder with sufficient amount of the gas (pressure at least 30 bar) to the coordinating laboratory for reanalysis. The cylinder was shipped to the participant in April The participating laboratory carried out a measurement from May to September The analysis report was received on September 30, Measurement protocol The measurement report requires per cylinder at least three independent measurements, obtained under repeatability conditions (at least) with three independent calibration, e.g. calibration (A) measurement (B) calibration (A) measurement (B) calibration (A) measurement (B) calibration (A) (etc.).this is a strict requirement to come to proper statistical analysis of the reported data. One single measurement result is usually obtained from multiple readings (sub measurements) without recalibration. Its standard deviation provides information about the performance of the measurement system. Measurement methods Table 1 shows the calibration method, traceability of calibration standards, and the measurement method at each laboratory. Table 1: Summary of the measurement methods of the participants Matrix of Measurement Laboratory Calibration Traceability standards technique KRISS ISO 6143 [2] Own Nitrogen GC/TCD 5

6 standards NMC/A*STAR ISO 6143 Own standards Nitrogen ABB paramagnetic oxygen analyzer Degrees of equivalence A unilateral degree of equivalence, D i, is adopted in this comparison. x i = D i = (x lab,i x ref,i ) The uncertainty of the difference, D i, corresponds to 95% level of confidence. Here, x lab,i is a reported value of the APMP.QM-S2.1 sample i from the participating laboratory and x ref,i is the reference value of the APMP.QM-S2.1 sample i and based on the gravimetric concentration determined by KRISS. The standard uncertainty of D i can be expressed as; u 2 2 (D i ) = u lab,i u ref,i where u lab,i and u ref,i are the uncertainties of x lab,i and x ref,i, respectively. The reference value, x ref,i can be expressed as; x ref,i = x prep,i + x ver,i + x lts,i where x prep,i is the amount of substance of a target component in APMP.QM-S2.1 sample i and obtained from gravimetric preparation. The Δx ver,i is the difference between the gravimetric value and measured one during verification analysis. The Δx lts,i is the difference between the gravimetric value and measured one during long-term stability study which was performed before and after the sample cylinder for NMC/A*STAR returned to KRISS. Results showed that Δx ver,i and Δx lts,i were smaller than the expanded analytical uncertainty, and thereby both Δx ver,i and Δx lts,i were set to zero. Assuming independence between errors, the uncertainty of x ref,i, u ref,i can be expressed as; 2 u ref,i 2 = u prep,i 2 + u ver,i 2 + u lts,i where u prep,i, u ver,i, and, u lts,i are the uncertainties of x prep,i, x ver,i, and x lts,i, respectively. In the gravimetric preparation, the amount of a target component is determined by the following

7 equation. x pre,i = x weighing,i + x purity,i where x prep,i is the fractional amount of substance of a target component in APMP.QM-S2.1 sample (i), x weighing,i is the fractional amount of substance of a target component in APMP.QM- S2.1 sample (i) gravimetrically prepared and Δx purity,i is the correction based on purity analysis. The uncertainty of the fractional amount is estimated as u prep,i = u weighing,i + u purity,i where u prep,i is the uncertainty from gravimetric preparation, u weighing,i is the uncertainty from gravimetric weighing process, u purity,i is the uncertainty from purity analysis. Results A complete set of results reported from each participant is described in Appendix B of this report. The results are summarized in Table 1. Table 1. Summary of measurement results for the comparison. The unit of each parameter is cmol/mol. The coverage factors, k lab, for both laboratories are 2. Laboratory Cylinder x prep u prep u ver u lts u ref x lab U lab cmol/mol KRISS D NMC D /A*STAR The parameters in Table 1 are defined as, x prep u prep amount of substance of target component in APMP.QM-S2.1 sample, from preparation (cmol/mol) uncertainty of x prep (cmol/mol) 7

8 u ver u lts u ref x lab U lab k lab x U( x) uncertainty associated with verification (cmol/mol) uncertainty associated with long-term stability test (cmol/mol) uncertainty of reference value (cmol/mol) reported result from each laboratory (cmol/mol) stated uncertainty of each laboratory, at 95% level of confidence (cmol/mol) stated coverage factor difference between laboratory result and reference value (cmol/mol) x, at 95% level of confidence (cmol/mol) Degree of equivalence, Δx, and its expanded uncertainty, U(Δx), of APMP.QM-S2.1 are summarized in Table 2 and plotted in Fig. 1. The results from the participants are consistent with the reference values as the deviations from the reference values are within the associated uncertainties. Table 2. Summary of Degree of Equivalence for the APMP.QM-S2.1 (k = 2). Laboratory Cylinder Δx U(Δx) Δx/x U(Δx)/x (cmol/mol) (cmol/mol) (%) (%) KRISS D NMC/A*STAR D

9 Figure 1. Degrees of equivalence (k = 2) Conclusion This bilateral comparison compares the measurement capability of oxygen in nitrogen matrix at 0.2 mol/mol. The results of both NMC/A*STAR and KRISS agree within 0.1 % with the KCRV. Reference [1] International Organization for Standardization, ISO 6142:2001 Gas analysis - Preparation of calibration gas mixtures - Gravimetric methods, 2nd edition. [2] International organization for standardization, ISO 6143, Gas analysis, Comparison methods for determining and checking the composition of calibration gas mixtures, ISO, Second edition, 2001(E) 9

10 Appendix A. Verification of mixtures with GC-TCD Four reference gas mixtures were prepared gravimetrically by KRISS according to ISO Table 1 shows a data set of gravimetric concentrations of each cylinder and their relative peak area of a GC-TCD (HP-7890) compared to a working reference mixture (~20% oxygen in nitrogen). Table 1. Analyzed results of four reference gas mixtures of oxygen in nitrogen prepared by KRISS using a GC-TCD analyzer. Mixture No. R x prep,r u prep,r (µmol/mol) (µmol/mol) y r u(y r ) D D D081136(KRISS) D081192(A*STAR) x prep : gravimetric concentration of reference gas mixtures, u prep : standard uncertainty of x prep, y r : corrected response relative to a QC cylinder (D155880, ~20% oxygen in nitrogen) of GC- TCD, u(y r ) : standard uncertainty of y r Linear regression result of x prep,r versus y r in Table 1 is shown in Table 2. The uncertainty of the linear regression fit is also shown in Table 2. Table 2. Parameters of a linear regression fit, y = a 0 + a 1 x Parameter Value a u(a 0 ) a u(a 1 )

11 Adj. R-Square After the regression analysis of table 2, the values of x ver,i in the following table 3 were calculated using the parameters and the equation in table 2 and the values of y r in table 1. Table 3. Comparison with gravimetric concentration and the result of analytical concentration. Mixture No. x prep,r u prep,r x ver,r u ver,r x ver,r U( x ver,r ) r µmol/mol D D D081136(KRISS) D081192(A*STAR) x ver,r : analytical concentration of reference gas mixture, (y r = a 0 + a 1 x ver,r ) u ver,r : standard uncertainty of x ver,r, x ver,r = x ver,r x prep,r : deviation of verification for mixture r. U ver,r : expanded uncertainty of y r (coverage factor, k=2). The corrected responses of GC-TCD, y r were obtained as follows. The responses of the analyzer were corrected with a quality control (QC) cylinder. The QC cylinder gas and other cylinder gases were injected sequentially into the analyzer using a multi-positioning valve. Cylinder gases were measured in the following order. QC(i =1) calibration standard 1 QC(i =2) calibration standard 2 QC(i =3) calibration standard 3 QC(i =4) calibration standard 4 QC(i =5) In each step of the cycle, measurement of the gas analyzer was repeated 5 times for each cylinder. The last 4 measured results were used for the calculation. This process ( QC(i=1) QC(i =5) ) were repeated j times (j = 3). The following calibration data set can be obtained at j th round (j = 1, 2, 3); 11

12 - Average values of responses to the QC cylinder, Y qc,i=1,j,., Y qc,i=4, j, - Average values of responses for calibration standards, Y 1, j, Y 2, j, Y 3, j, Y 4, j. The corrected response for calibration standard r at j th round, y r, j, was calculated as follows; y r, j = Y r, j / [(Y qc,i=r, j + Y qc,i=r+1, j )/2] (r = 1, 2, 3, 4) (1) The value of y r was calculated from the following equation; These standard uncertainties are [Ref. 1]; j=3 y r = j=1 y r,j /J (2) u 2 (y r ) = j=3 (y r,j y r ) 2 j=1 (3) J(J 1) The gravimetrically prepared mixtures have been verified by comparing the gravimetric composition value with its analytical measurement value (i.e., verification value) as shown in the following condition. x prep,r x ver,r 2 u2 prep,r + u2 ver,r (4) where x ver,r and u ver,r is the measurement result from verification and its the standard uncertainty, respectively. The uncertainty associated with the verification relies on the measurement capability and experiment design. In the comparison with gravimetric concentration and analytical concentration for each mixture, all values of x ver,r were smaller than those of U( x ver,r ). Reference [1] International Organization for Standardization, ISO 6142:2001 Gas analysis - Preparation of calibration gas mixtures - Gravimetric methods, 2nd edition. 12

13 Appendix B. Report from each laboratory Laboratory: KRISS (Korea Research Institute of Standards and Science), Korea Cylinder number: D Measurement 1 # Standard Date Result Number of Component deviation (dd/mm/yy) (cmol/mol) replicates (% relative) O 2 27/03/ Measurement 2 # Standard Date Result Number of Component deviation (dd/mm/yy) (cmol/mol) replicates (% relative) O 2 28/03/ Measurement 3 # Standard Date Result Number of Component deviation (dd/mm/yy) (cmol/mol) replicates (% relative) O 2 29/03/ Results Result Expanded uncertainty Component Coverage factor* (cmol/mol) (cmol/mol) O ( %) 2 *The coverage factor shall be based on approximately 95% confidence. 13

14 Calibration Standards Four reference gas mixtures were prepared by gravimetric method according to ISO Cylinder Number Assigned value (cmol/mol) Standard uncertainty (cmol/mol) D D D D Gravimetric preparation data Primary standard gas mixtures were prepared gravimetrically according to ISO6142. Specification of a balance Model No.: Mettler-Toledo Resolution: 1 mg, Capacity: 10 kg Uncertainty (k = 2): 3.2 mg Weighing method (A-B-A, substitution method) Substitution method, tare cylinder (A-B-A) -Purity Analysis Nitrogen source gas: %mol/mol Component Amount fraction Standard uncertainty Assumed (10-6 mol/mol) (10-6 mol/mol) distribution Hydrogen Rectangular Oxygen Normal Carbon monoxide Normal Carbon dioxide Rectangular Methane Normal Argon Normal 14

15 Water Normal Nitrous oxide Rectangular Hydrocarbons (CxHy) Rectangular Neon Normal Nitrogen Normal Oxygen source gas: %mol/mol Component Amount fraction Standard uncertainty Assumed (10-6 mol/mol) (10-6 mol/mol) distribution Hydrogen Rectangular Nitrogen Normal Carbon monoxide Normal Carbon dioxide Normal Methane Rectangular Argon Rectangular Water Normal Oxygen Normal Sample handling The sample cylinder was stored at a room temperature for 3 days before an analysis. The reference cylinder was also stored at the same condition. The room temperature of our laboratory was maintained at ~22 ± 2 C for all the period. A SS regulator was connected to the reference and sample cylinders. The reference and sample gases were directly introduced to the GC through a multi-positioning valve and a mass flow controller. The injection of gases was switched automatically using a multi-positioning valve. 15

16 Instrumentation -Analytical Instrument: HP7890A GC analyzer equipped with a TCD detector and sampling valve line without an injection port -Analytical Condition Condition Detector Thermal Conductivity Detector (TCD) Detector Temperature 250 C Carrier Flow rate 80 psi Reference Flow rate 45 ml/min Column Resteck Molesieve 5A, 4m, 1/8, SS Oven Temperature 60 C for 12min Valve Box Temperature 60 C Sample Flow rate 75 ml/min Sample Loop Volume 100 µl Instrument Calibration The corrected responses of GC-TCD, y r were obtained as follows. The responses of the analyzer were corrected with a quality control (QC) cylinder. The QC cylinder gas and other cylinder gases were injected sequentially into the analyzer using a multi-positioning valve. Cylinders were measured in the following order. QC(i =1) calibration standard 1 QC(i =2) calibration standard 2 QC(i =3) calibration standard 3 QC(i =4) sample gas QC(i =5) sample gas QC(i =6) sample gas QC(i =7) In each step of the cycle, measurement of the gas analyzer was repeated 5 times for each cylinder. The last 4 data were used for the calculation. This process was repeated 3 times during three different days. 16

17 During each measurement period (j =1, 2, 3), following calibration data set can be obtained. - Average values of responses to the QC cylinder, Y qc,i=1,j,., Y qc,i=5, j, - Average values of responses for calibration standards and sample gas, Y 1,j, Y 2, j, Y 3, j, Y s, j, Y s, j, Y s, j. The corrected response for calibration standard r at j th period, Y r, j, was calculated as follows. y r, j = Y r, j / [(Y qc, i=r, j + Y qc, i=r+1, j )/2] ( r = 1, 2, 3 ) (1) y s, j = Y s, j / [(Y qc, i=r, j + Y qc, i=r+1, j )/2] ( r = 4, 5, 6 ) (2) From the data set of X r=1 (reference value of calibration standard 1), X r=2, X r=3, y r=1, j, y r=2, j and y r=3, j, the linear regression parameters were obtained from a linear fit of y r, j = b 0 + b 1 X r. From the regression parameters, mixing ratios of sample cylinder, X s were calculated from y s, j. Uncertainty evaluation Typical evaluation of the measurement uncertainty of O 2 : Uncertainty [cmol/mol] Uncertainty [%] Gravimetric uncertainty -Purity analysis -Gravimetric method Molar mass Analytical uncertainty -Repeatability Reproducibility Combined uncertainty Expanded uncertainty (k=2)

18 Report Form oxygen in nitrogen Laboratory name: Gas Metrology Laboratory, National Metrology Centre, Singapore Cylinder number: KRISS Cylinder (D081192, 00T-3AL M-9905) Measurement 1 # Component Standard Date Result Number of deviation (dd/mm/yy) (mol/mol) replicates (% relative) O 2 29/07/ O 2 29/07/ O 2 29/07/ Measurement 2 # Component Date Result Standard Number of (dd/mm/yy) (mol/mol) deviation replicates (% relative) O 2 30/07/ O 2 30/07/ O 2 30/07/ Measurement 3 # Component Date (dd/mm/yy) Result (mol/mol) Standard deviation Number of replicates (% relative) O 2 31/07/

19 Results O 2 31/07/ O 2 31/07/ Component Result Expanded uncertainty Coverage factor*) (mol/mol) (mol/mol) O k=2 *) The coverage factor shall be based on approximately 95% confidence. Method description forms Please complete the following data regarding the description of methods and the uncertainty evaluation. Reference Method: The analysis was performed on three different days with ABB paramagnetic oxygen analyzer with the sampling box. The gas flow rate was set at 350ml/min. The mole fraction of the compared cylinder was calculated by interpolation of a calibration curve using CurveFit software. Calibration standard: The below standards were prepared by gravimetric method according to ISO6142. The purity of gases was analysed with GC PDHID. The cylinders used were 5L aluminum with Aculife 3 treatment from Scott Specialty Gases. The regulator used was SS verifo single stage without gauges purged 5 times according to operational procedure. PSM Number Mol fraction Standard uncertainty (relative) PSM E-05 19

20 PSM E-05 PSM E-05 PSM E-05 PSM E-05 Instrument calibration: The analyzer was adjusted in zero and span before every analysis. The above PSM were used as the calibration curve. Sampling handing: The received cylinders and NMC PSM were maintained inside the laboratory at room temperature for all the time. Modified Teflon was used for sample lines. The sampling to the analyzer and measurement were done under ambient pressure, and the pressure correction and response correction were included in the calculation. Detailed uncertainty budget: Please include a list of the uncertainty contributions, the estimate of the standard uncertainty, probability distribution, sensitivity coefficients, etc. Typical evaluation of the measurement uncertainty of O 2 : Quantity (Uncertainty source), X i Estimate, Evaluation x i type (A or B) Standard Distribution uncertainty, u(x i ) Sensitivity coefficient, c i Contribution, u(y i ) Gas standard Type B Normal E E-05 Gas standard Type B Normal E E-05 Gas standard Type B Normal E E-05 20

21 Gas standard Type B Normal E E-05 Gas standard Type B Normal E E-05 Repeatability Type A Normal Combined uncertainty (relative) Expanded uncertainty (relative) Expanded uncertainty (mol/mol)