Repositorio Institucional

Key Comparison CCQM-K73 Amount Content of H+ in Hydrochloric Acid (0.1 mol·kg-1)

Final Report: 26 October 2012

Kenneth W. Pratt1,2 Jose Luis Ortiz-Aparicio, Francisco Javier Matehuala-Sanchez, Monika Pawlina, Władysław Kozłowski, Paulo P. Borges, Wiler B. da Silva Junior, Mónica B. Borinsky, Ana HernandezMabel Puelles, Nadia Hatamleh, Osvaldo Acosta, João Nunes, M. J. Guiomar Lito, M. Filomena Camões, Eduarda Filipe, Euijin Hwang, Youngran Lim, Wu Bing, Wang Qian, Wei Chao, Akiharu Hioki, Toshiaki Asakai, Michal Máriássy, Zuzana Hanková, Sergey Nagibin, Olexandra Manska, Vladimir Gavrilkin

Abstract This key comparison (KC), CCQM-K73, was performed to demonstrate the capability of the participating National Metrology Institutes (NMIs) to measure the amount content of H+, νH+, in an HCl solution with a nominal νH+ of 0.1 mol·kg-1. A parallel Pilot Study, CCQM-P19.2, was performed for NMIs that did not desire to participate in the KC. The comparison was a joint activity of the Electrochemical Working Group (EAWG) and Inorganic Analysis Working Group (IAWG) of the CCQM and was coordinated by NIST (USA) and CENAM (México).

The method of determination of νH+ was left to the individual participant. All participants used either coulometry or titrimetry with potentiometric determination of the endpoint.

The agreement of the results was not commensurate with the claimed uncertainties of the subset of participants that claimed small uncertainties for this determination. A workshop on technical issues relating to the CCQM-K73 measurements was conducted at the joint IAWGEAWG meeting at the Bureau International des Poids et Mesures (BIPM), Paris (Sèvres) in April 2010. Several possible sources of bias were investigated, but none could explain the observed dispersion among the participants’ results.

In the absence of a specific cause for the dispersion, the IAWG and EAWG decided to assign a Key Comparison Reference Value, KCRV, and standard uncertainty of the KCRV, uKCRV, based on the DerSimonian-Laird statistical estimator. The uKCRV is dominated by the between-laboratory scatter of results in CCQM-K73. The uncertainty estimates from the participants with the lowest reported uncertainties remain unsupported by this KC.

Metrology Area Amount of Substance

Branches Electrochemistry, Inorganic Analysis

Subject Determination of the amount content of H+ in hydrochloric acid solutions.

1 Study Coordinator 2 Organizational affiliations of all authors listed in Table 1.

Page 1 of 22



Time schedule Dispatch of the samples: Deadline for receipt of the report: Discussion of results: Draft A Report



26 June 2009 30 September 2009 EAWG/IAWG Joint meeting, 4 November 2009 December 2009



All dates in this Report are 2009, unless noted otherwise.



Participants The list of participants is given in Table 1 for CCQM-K73. VNIIFTRI originally registered for CCQM-K73 but withdrew after it proved impossible to ship the samples to them. DFM originally registered for CCQM-K73 and received samples, but withdrew owing to equipment difficulties.



Table 1. Table of participants, Key Comparison CCQM-K73.



Acronym Participant (NMI) CENAM Centro Nacional de Metrología



Country México



DFM



Dansk fundamental metrology



GUM



Główny urząd miar



INMETRO



Instituto Nacional de Metrologia, Normalização e Qualidade Industrial



INTI



Instituto Nacional de Tecnología Industrial



Denmark Poland Brazil

Argentina



IPQ

KRISS NIM NIST NMIJ



IPQ/UMCA-LCM, Instituto Português da Qualidade - Scientific and Applied Metrology Unit



Portugal



Korea Research Institute of Standards and Science

National Institute of Metrology of P. R. China

National Institute of Standards and Technology



Korea China USA



National Metrology Institute of Japan



Japan



Analyst(s) Jose Luis Ortiz-Aparicio, Francisco Javier MatehualaSanchez Pia Tønnes Jakobsen Monika Pawlina, Władysław Kozłowski Paulo P. Borges, Wiler B. da Silva Junior Mónica B. Borinsky-Ana Hernandez-Mabel Puelles-Nadia Hatamleh-Osvaldo Acosta

M. João Nunes, M.J. Guiomar Lito, M. Filomena Camões, Eduarda Filipe

Euijin Hwang, Youngran Lim

Wu Bing; Wang Qian; Wei Chao

Kenneth W. Pratt

Akiharu Hioki, Toshiaki Asakai



SMU



Slovenský metrologický ústav



UMTS VNIIFTRI



Ukrainian State Research and Production Center of Standardization Metrology, Certification, and Consumers’ Rights Protection

All-Russian Scientific Institute for Physical-Technical and Radiological Measurements



Slovak Republic

Ukraine

Russia



Michal Máriássy, Zuzana Hanková

Sergey Nagibin, Olexandra Manska, Vladimir Gavrilkin

Viatcheslav Kutovoy



Page 2 of 22



Coordinating Laboratories

National Institute of Standards and Technology Analytical Chemistry Division Gaithersburg, MD 20899-8391 USA

Centro Nacional de Metrología Carretera a los Cués, km 4,5 El Marqués, Querétaro, C.P. 76246 MÉXICO

Study Coordinator: Kenneth W. Pratt (Telefax: +1 301 869 0413; e-mail: kenneth.pratt@nist.gov)

A copy of the Technical Protocol distributed to all participants is attached as an Appendix to this Report.

Sample preparation and bottling The CCQM-K73/P19.2 solution was prepared from concentrated reagent-grade hydrochloric acid (Mallinckrodt AR3, Lot H611 T13A01) and high-purity deionized water. The water had an indicated electrolytic conductivity of 18.2 MΩ·cm at delivery. A 150 mL portion of this acid was added to ≈18 L H2O and the solution was mixed by agitating and subsequently by bubbling house N2 (mass fraction of CO2 < 1 µmol·mol-1) through the solution for > 1 h. Preliminary titration of this solution yielded νH+ ≈ 0.9958 mol·kg-1. To permit a larger number of bottles to be filled, a 2 L additional portion of solution was prepared from ≈ 2 L H2O and 17 mL of the same reagent grade hydrochloric acid. This second portion was added to the original batch to yield the final batch. The composite solution was thoroughly mixed by agitation and bubbling with house N2 for 2 h. This final batch (the “CCQM-K73 solution”) had a volume of ≈ 20 L, with a nominal νH+ of 0.1 mol·kg-1.

Sixty-four 250 mL high-density polyethylene (HDPE) bottles were cleaned by rinsing three times in water and soaking in water for 3 h or longer. This cycle was performed a total of three times. The bottles then were rinsed a final three times with water and were dried at 50 °C to 70 °C in an oven. The clean, dry bottles were filled with the CCQM-K73 solution and numbered/labeled consecutively in filling order on 23 June. The caps of the filled bottles were hand-torqued and allowed to sit overnight. They were re-torqued the following morning and re-weighed. All masses remained constant within 0.010 g. Heat-shrink seals were then applied to the necks of the re-torqued bottles. The sealed bottles were labeled, weighed, and bagged in transparent polyethylene (PE) bags. Then, the PE-bagged, labeled bottles were sealed in aluminized polyethylene terephthalate (PET) bags, which were numbered and labeled identically to the bottles. Finally, the PET-bagged bottles were weighed. The PETbagged bottles were delivered to the NIST Shipping Department on 26 June, and were shipped to participants from 26 June through 1 July. The first (1) and last (64) bottles were retained at NIST for verification of homogeneity of the CCQM-K73 solution.

3 Certain commercial equipment, instruments, or materials are identified in this paper to specify adequately the experimental procedure. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.

Page 3 of 22



Solution homogeneity and stability Coulometric titrations of samples from bottles 1 and 64 were performed to verify the homogeneity of the CCQM-K73 solution with respect to the set of bottles. Coulometric titrations of solution directly taken from the carboy were also performed immediately before and following this bottling step. Results of these titrations are shown in Table 2.

The sample was not deaerated after the main titration [1] in the preliminary titrations. However, as the preliminary titrations were performed from deaerated solution (see details of preparation, above), the solution may be assumed to be largely free from dissolved CO2. This assumption was verified by noting that the maximum slope of the titration curve in the final titration was only about 8 % lower than the maximum slope of the titration curve in the initial titration (both slopes normalized to 1 L, to account for the increase in volume on addition of the sample).



Table 2. Measurements of Homogeneity and Stability at Coordinating Laboratory (25 °C).



Source carboy (before filling bottles) carboy (before filling bottles) carboy (before filling bottles) carboy (before filling bottles) a carboy (after filling bottles)

CCQM-K73 Bottle 64 CCQM-K73 Bottle 1 CCQM-K73 Bottle 64 CCQM-K73 Bottle 1 Mean Standard Deviation (SD) SD of Mean



Date of titration 22-Jun 22-Jun 23-Jun 23-Jun 23-Jun 24-Jun 24-Jun 24-Jun 24-Jun



νH+/(mol·kg-1) 0.100 922 9 0.100 922 2 0.100 923 9 0.100 916 6 0.100 916 8 0.100 918 3 0.100 917 6 0.100 913 7 0.100 914 4 0.100 918 5 0.000 003 7 0.000 001 2



aCarboy was vigorously swirled prior to this titration to assure homogeneity of contents.



The data in Table 2 provide an evaluation of homogeneity and stability of the CCQM-K73 solution to the precision of the coulometric method as implemented at the preparing laboratory (NIST). The agreement between the preliminary titrations of Bottles 1 and 64 indicates that the CCQM-K73 solution as bottled was homogeneous. The ≈ 0.007 % decrease noted in the preliminary titrations of the carboy solution (prior to bottling) in Table 2 likely resulted from the incorporation of condensate water in the carboy neck into the bulk solution at the time of mixing. The agreement between the results of the titrations of Bottles 1 and 64 and the results of the carboy titrations indicate that no significant change in νH+ resulted from the bottling procedure. Finally, the agreement between all these preliminary titrations (performed in June) and the NIST CCQM-K73 titrations (performed at the end of September) indicate that any drift that occurred over the entire designated period [2] of the CCQM-K73 measurements was < 0.007 % (cf. the reported mass changes documented in Table 3).

Page 4 of 22



Although the lack of sample deaeration in the preliminary titrations compromises this assessment of drift to some extent, the fact that the preliminary titrations were performed on substantially deaerated solution reduced the component of uncertainty for dissolved CO2 from ca. 0.011 % (calculated for non-deaerated solution in equilibrium with atmospheric CO2) to ≈ 0.005 % for these preliminary titrations.

Sample delivery The sample bottles were shipped from NIST on 26 June through 1 July. Each participant received two or three 0.25 L HDPE numbered bottles, packaged as described above, according to the quantity requested.

Following the initial shipment, VNIIFTRI reported that delivery of the CCQM-K73 samples required two special licenses, as Russian regulations classify hydrochloric acid (of any concentration) a precursor for narcotics. VNIIFTRI was not able to obtain these licenses in the required time frame and requested to withdraw from CCQM-K73 on 3 August.

Verification of mass stability of shipped bottles Participants were requested to measure the mass of the PET-bagged bottles soon after receipt (storing at least overnight to assure temperature equilibrium with the weighing laboratory). These masses are listed in Table 3 as mbag, along with the weighing dates4.

Immediately before performing the measurement of νH+, participants were requested to measure the mass of the bottle outside the bags (removed from the external PET and internal PE bags). These masses are listed in Table 3 as mbottle, along with the weighing dates.

Participants also reported the balance reading, mW; the ambient pressure, p, and temperature, T, at the time of each weighing. If the relative change in mass using the submitted mbag or mbottle differed by > 0.02 % from the initial pre-shipment masses, m0,bag and m0,bottle, the coordinating laboratory independently calculated separate values of mbag and mbottle from the participant-supplied values of mW, p, and T using the formula used to calculate m0,bag and m0,bottle. The coordinating laboratory then contacted the participant to verify the weighing and calculation for buoyancy.

The data in Table 3 demonstrate that except for one value of mbag reported by IPQ, the bagged bottles remained constant in mass to within 0.015%, relative, between the initial sealing in the aluminized PET bags on 25 June and the respective date of measurement. The values of mbottle outside the bags also remained constant to within 0.013 %, relative, although a small loss in mass was evident in almost all cases.

The relative changes in mbag and mbottle listed in Table 3 are referenced to m0,bag and m0,bottle, respectively. If the relative changes are referenced instead to the mass of solution, the values of Δmbag/m0,bag and Δmbottle/m0,bottle in Table 3 should be multiplied by 1.11 and 1.12, respectively.

4 Masses of the PET-bagged bottles for NIST in Table 3 were measured immediately prior to opening the PET bags.

Page 5 of 22



Table 3. Mass Changes on Shipping: PET-Bagged Bottles and Sample Bottles.



Sample information



NMI



Bottle No.



4



Data Pertaining to Bagged Bottle



25-Jun pre-ship

mass, m0,bag/g



Weighing Date at NMI



mbag/g (NMI)



Relative change in

mass Δmbag/m0,bag



314.553 3-Aug 314.552 -0.0003%



Data Pertaining to Bottle outside Bags



24-Jun pre-ship

mass, m0,bottle/g

300.944



Weighing Date for

mbottle (NMI)



mbottle/g (NMI)



Relative change in mass

Δmbottle/m0,bottle



1-Sep 300.926 -0.0060%



CENAM 44 308.011 3-Aug 307.997 -0.0044% 294.485 1-Sep 294.456 -0.0097%



49 309.536 3-Aug 309.531 -0.0015% 296.170 1-Sep 296.152 -0.0059%



20 313.624 2-Jul 313.635 0.0034% 300.158 ---



---



---



DFM



31 311.777 2-Jul 311.783 0.0018% 298.196 ---



---



---



25



GUM



39



60



21 INMETRO

56



19



INTI



42



53



8



IPQ



36



45



10 KRISS

33



9 NIM

40



3



NIST



28



57



7



312.708 309.474 312.755 311.778 307.128 312.102 309.613 312.035 304.238 309.285 309.393 312.968 310.722 309.891 307.165 309.554 312.398 310.558 312.014



14-Jul 312.710 14-Jul 309.460 14-Jul 312.760 27-Jul 311.775 27-Jul 307.119 6-Jul 312.120 6-Jul 309.620 6-Jul 312.050 3-Jul 304.380 3-Jul 309.250 3-Jul 309.340 7-Jul 312.986 7-Jul 310.740 6-Jul 309.894 6-Jul 307.172 24-Sep 309.562 24-Sep 312.406 24-Sep 310.569 30-Jun 312.015



0.0005% -0.0045% 0.0015% -0.0011% -0.0027% 0.0059% 0.0022% 0.0049% 0.0468% -0.0113% -0.0171% 0.0059% 0.0056% 0.0008% 0.0024% 0.0025% 0.0025% 0.0034% 0.0003%



299.083 295.896 298.170 298.598 293.662 298.342 296.300 298.325 290.943 295.654 295.835 299.435 296.987 296.477 293.480 295.818 299.184 296.981 298.317



8-Sep 299.071



15-Sep 295.886



16-Sep 298.155



6-Aug 298.583



6-Aug 293.631



5-Aug 298.340



---



---



20-Aug 298.310



29-Jul 290.938



14-Aug 295.634



17-Aug 295.815



15-Sep 299.408



29-Sep 296.949



23-Sep 296.461



23-Sep 293.470



25-Sep 295.809



24-Sep 299.174



28-Sep 296.969



11-Aug 298.308



-0.0040% -0.0033% -0.0050% -0.0052% -0.0107% -0.0007%

---0.0051% -0.0015% -0.0069% -0.0065% -0.0090% -0.0127% -0.0055% -0.0035% -0.0027% -0.0033% -0.0040% -0.0029%



NMIJ SMÚ



27 314.019 30-Jun 314.020 0.0003% 300.207 ---



---



---



52 308.259 30-Jun 308.264 0.0016% 294.547 ---



---



---



6 306.600 7-Jul 306.606 0.0019% 293.054 23-Sep 293.042 -0.0040% 59 306.273 7-Jul 306.277 0.0015% 292.457 23-Sep 292.446 -0.0038%



UMTS



14 309.970 21-Jul 309.954 -0.0050% 296.607 25-Aug 296.583 -0.0082% 62 310.056 21-Jul 310.042 -0.0044% 296.026 25-Aug 295.999 -0.0090%



Timetable of Measurements and Submission of Reports The dates of receipt of the samples, the dates of measurements, and reporting dates are given in Table 4.



Page 6 of 22



Table 4. Dates of Sample Receipt, Measurement Period, and Report Date.



NMI Study Sample Received Measurement Period Report Received



CENAM K73



20-Jul



1-6 Sep



30-Sep



DFM K73



1-Jul



---



---



GUM K73



14-Jul



8, 16 Sep



30-Sep



INMETRO K73



24-Jul



24 Aug-18 Sep



24-Sep



INTI K73



1-Jul



5, 19 Aug



7-Oct



IPQ



K73



2-Jul



29 Jul-25 Aug



30-Sep



KRISS K73



29-Jun



28-30 Sep



1-Oct



NIM K73



6-Jul



2-17 Sep



27-Sep



NIST K73



25-Jun



24-28 Sep



1-Oct



NMIJ K73



29-Jun



12-13 Sep



30-Sep



SMU K73 UMTS K73



30-Jun 9-Jul



23-25 Sep 26-27 Sep



29-Sep 30-Sep



VNIIFTRI K73



---



---



---



All CCQM-K73 reports were received by the deadline, excepting INTI, who requested (and was granted) an extension within the allotted timeframe. The results were distributed to all participants on 20 October.

Correspondence with Participants Each participant was contacted as the weighing and reported results were received by the coordinator. One participant also received specific correspondence from the coordinator prior to the report deadline as noted below.

CENAM requested advice regarding the correct procedure to follow if analysis is performed using more than one method. On being advised that the selection of method(s) and whether to combine results is left solely to the participant, CENAM elected to submit the coulometric result as their CCQM-K73 result and to submit the separate titrimetric result as an information value. CENAM was also advised to recheck the weighings of their bottles and did so. The results in Table 3 reflect the revised mass values.

In none of the above cases was any information released or implied in any correspondence regarding the value of νH+ for the CCQM-K73 solution, prior to the release of the summary of results to all participants on 20 October.

Results and Discussion The results are summarized in Table 5 for CCQM-K73. In addition to the official results, several participants also reported “information only” values for the CCQM-K73 solution based on other determinations. The information only results are listed in Table 6 for comparison with the official results.

Page 7 of 22



Table 5. Results of Key Comparison CCQM-K73 Sorted by Value.



NMI



Amount Content of H+ νH+/(mol·kg-1) U/(mol·kg-1)



IPQ



0.100 143 0.000 092



UMTS



0.100 42



0.000 48



INTI



0.100 87



0.000 30



CENAM



0.100 894 0.000 013



KRISS



0.100 906 0.000 018



NIM



0.100 917 0.000 080



NIST



0.100 922 4 0.000 004 5



SMÚ



0.100 936 7 0.000 007 8



NMIJ



0.100 941 5 0.000 005 8



INMETRO 0.100 974 1 0.000 004 9



GUM



0.101 04



0.000 26



Table 6. Results of “Information Only” Values for CCQM-K73 Solution Sorted by Value.



NMI

CENAM NIM NIM



Measurand j

H+ Cl– H+



Amount Content νj/(mol·kg-1) Uj/(mol·kg-1) 0.100 884 0.000 038 0.100 922 0.000 102 0.101 070 0.000 114



The above-tabulated values are presented in graphical form in Figures 1a and 1b, with the results sorted by value within the given study. The Key Comparison Reference Value (KCRV), calculated as described below, is shown in Figure 1a as a reference. Figure 1b is identical to Figure 1a except that the y-axis is expanded by a factor of 10 around the KCRV.



Page 8 of 22



Figure 1a. Results of Key Comparison CCQM-K73 and Submitted “Information-Only” Values. Error Bars Correspond to Standard Uncertainties (k = 1).

Figure 1b. 10-Fold Expansion of Figure 1a around KCRV.

Page 9 of 22



Measurement Techniques Table 7 lists the methods used by the participants. Participants are listed in alphabetical order within each set of results.



Table 7. Participant Methods.



Set of Results

K73

Info values



NMI

CENAM GUM INMETRO INTI IPQ KRISS NIM NIST NMIJ SMÚ UMTS CENAM – H+ NIM – Cl– NIM – H+



Method (for titrimetry – reference compound) coulometry – vertical titrimetry (Trisb) coulometry – vertical titrimetry (KHPc) titrimetry Cl− (KCl)

coulometry – horizontal, 2 IC coulometry – horizontal, 2 IC coulometry – horizontal, 2 IC coulometry – horizontal, 2 IC coulometry – vertical coulometry – agar membrane, 1 IC

titrimetry (Tris)

coulometry Cl – horizontal, 2 IC

titrimetry (Na2CO3)



a Certified reference material b Tris(hydroxymethyl)aminomethane c Potassium hydrogen phthalate



CRMa (for titrimetry), Comments (coulometry) SMÚ A0701509 NIST 84j NIST 999b

NIST 723d Result uncorrected for Br NIM GBW06101A



For participants who used coulometry, Table 7 notes the configuration of the coulometric cell and the number of intermediate compartments (IC) used. Coulometric cells use two general configurations: vertical and horizontal [3]. Vertical cells have the counter-electrode half-cell (anode for titrations of H+) situated vertically along the center axis of the sample half-cell (cathode for titrations of H+). An intermediate compartment (IC) along this same axis connects the two half-cells. To avoid losses of sample or titrant, the IC usually is filled with solution withdrawn from the sample cell prior to sample introduction and returned during the course of the main titration. Horizontal cells have the counter-electrode half-cell and the sample half-cell situated parallel (side-by-side), with one or (generally) two ICs connecting the two half-cells near their bottoms. In both configurations, any sample or titrant that diffuses into the IC(s) is returned to the sample cell by emptying and filling the IC(s) pneumatically (piston burette or gas pressure/vacuum).



All participants that used coulometry used a silver anode in the coulometric cell. Anode geometries varied. A silver anode has the advantage that the pH of the anolyte (electrolyte in the anode compartment) remains virtually constant during the electrolysis, owing to the absence of H+ in the anodic half-reaction:

Ag + Cl− → AgCl(s) + e−.



Participants that used titrimetry generally used gravimetric delivery of the major fraction of the titrant, with volumetric additions near the end point.



Page 10 of 22



IPQ performed an argentimetric titration of Cl− in the CCQM-K73 sample. No correction was performed for Br−. The reported value of νH+ was obtained via the assumption of equality between the reported νH+ and the amount content of total halide (result of the Ag+ titration) in

the HCl sample.



Traceability of Titrimetric Results Titrimetric methods are (in principle) primary ratio methods [4]. A titrimetric method compares an amount of analyte to a standard of the same quantity. Therefore, each titrimetric result must be traceable to a reference compound. For each participant that used titrimetry, Table 7 notes the reference compound that was used and the specific certified reference material (CRM) that served as the acidimetric standard. Two exceptions are noted. The IPQ result was not traceable to an acidimetric standard, but rather to a KCl standard.

Each CRM noted in Table 7 (other than the pH CRM with no certified assay) was certified by coulometry. The Tris CRMs were certified using HCl as the excess added substance in a coulometric back-titration. Hence, the issues discussed below with respect to the CCQM-K73 coulometric results apply indirectly to those participants that used Tris as the reference CRM. Although an existing KC is available for KHP [5], certain problems associated with the coulometric determination of νH+ in the CCQM-K73 solution apply also in the case of KHP. The NIM Na2CO3 CRM was certified by coulometry with electrogenerated H+ [6] and does not involve the coulometric standardization of HCl in its certification.



Statistical Evaluation of the Results A Birge analysis [7] of the CCQM-K73 results was performed. The weighted mean of the

participants’ results, R (H ) , for CCQM-K73 was calculated using Eq 1:



n



R (H )



wi i (H ) .



(1)



i1



The wi are given by Eq 2, where u(xi) is the reported standard uncertainty for participant i:



1



wi



u2[ i (H )]



n



1



.



(2)



i 1 u2[ i (H )]



The uncertainty of the weighted mean for CCQM-K73 determined by the external consistency method, uE[ R (H )], was calculated at each temperature using Eq 3:



n

wi i (H )



R (H ) 2



uE[ R (H )] i 1



n1



.



(3)



In Eq 3, n is the number of participants, wi is the normalized weight for participant i, and i (H ) is the reported result for participant i.



Page 11 of 22



The uncertainty of the weighted mean for CCQM- K73 determined by the internal consistency method, um[ R (H )], was also calculated. The value for um[ R (H )] is given by Eq 4:



1



um[ R (H )]



n



1



.



(4)



i 1 u2[ i (H )]



The Birge ratio, R = uE/um, was then calculated at each temperature. The result including all CCQM-K73 participants was R = 7.923. The Birge ratio including all CCQM-K73 participants except IPQ was 6.024. Each of these R values indicate that in each case (including or excluding IPQ), the external (between-participant) uncertainty greatly dominates over the claimed uncertainties of the participants. The value compares to R values between 1.6 and 3.6 for pH KCs [10-15], in which the claimed uncertainties were far more consistent with the between-participant scatter.



The CCQM-K73 results were subsequently submitted for statistical evaluation by S. Ellison, LGC. His analysis (slightly rephrased from [8]) is as follows.

The χ2 on all the CCQM-K73 participants came out over 600. This value is far too large. Usually, the problem goes away after eliminating the obvious outliers (in this case, IPQ). Omitting IPQ, the χ2 is still ≈300. To accept the hypothesis of mutual consistency, the χ2 would have to be not greater than about 20.



Other explorations make it clear that this is not down to any one or two labs. IPQ aside, it is mostly a matter of substantive disagreement among the labs with the smallest uncertainties.



With such marked disagreement among those labs, there is no real point in seeking a KCRV at all, unless the differences turn out to be attributable to test material inhomogeneity. Otherwise, the participants with large uncertainties can mostly take some comfort from it – they mostly agree with each other and with the participants with small uncertainties. On this evidence, the participants with large uncertainties have no additional work to do. The participants in the middle, particularly those with the four smallest uncertainties, need to work to find out what is going on. In other words, about half the data set are in disagreement and the other half mostly have such large uncertainties that they all agree with everyone.



[Ellison’s] recommendation is not to put a KCRV on CCQM-K73, and it is even more difficult to assign an uncertainty to the KCRV. If an uncertainty is assigned to the KCRV, it would be large.

[Ellison’s] current favorite estimator for this kind of situation (an MM-estimate, which is uncertainty-respecting, very robust, and uses a Birge-like scale expansion) comes in at 0.100 928 2 mol·kg-1 with a standard deviation of the mean of 0.000 005 8 mol·kg-1. The only CCQM-K73 participants that are heavily downweighted in this calculation are IPQ and INMETRO.



Even if the MM-estimate model is used, [Ellison is] still not at all sure what should be used for the uncertainties of the degrees of equivalence.



Calculation of the KCRV and Its Uncertainty Calculations of candidate KCRVs, x, and corresponding combined standard uncertainties of the candidate KCRVs, uKCRV, were performed using two data sets: (1), the set of nine CCQM-K73 results (excluding IPQ and UMTS), n = 9; and (2), the central set of seven



Page 12 of 22



coulometric results from CENAM through INMETRO (see Table 5, Figures 1a and 1b), n = 7. The n = 9 set contains titrimetric results from two participants: INTI and GUM. The n = 7 set includes only the coulometric results.

Calculations were performed for each data set using the following statistical estimators: mean (unweighted), median, Graybill-Deal, and DerSimonian-Laird. The results are shown in Table 8 for the n = 9 set and Table 9 for the n = 7 set.



Table 8. Statistical Estimators for CCQM-K73 Results. IPQ and UMTS Excluded (n = 9).



Estimator

Mean Median Graybill-Deal DerSimonian-Laird



Candidate Value,

x/(mol·kg-1) 0.100 933 6

0.100 922 4

0.100 941 1

0.100 929 6



u(Value)/(mol·kg-1)

0.000 016 4 0.000 011 8 0.000 001 4 0.000 010 8



Table 9. Statistical Estimators for CCQM-K73 Results. Coulometric Results Only, Excluding UMTS (n = 7).



Estimator



Value/(mol·kg-1) u(Value)/(mol·kg-1)



Mean



0.100 927 5



0.000 009 9



Median



0.100 922 4



0.000 011 2



Graybill-Deal



0.100 940 9



0.000 001 4



DerSimonian-Laird 0.100 929 2



0.000 010 9



The best estimator for the present data should consider the degree of confidence in the reported uncertainties of the participants. The mean and median each neglect the reported uncertainties in obtaining the final value. The Graybill-Deal estimator overestimates the effect of results with low uncertainties and is inappropriate for the present set of data.



Preliminary results were presented to the April 2011 joint IAWG-EAWG meeting of the CCQM. At this meeting, it was felt that the uncertainties reflect the differences in the procedures, yet many of the reported uncertainty intervals do not overlap. Accordingly, it was decided to use the DerSimonian-Laird estimator for the KCRV and uKCRV for CCQM-K73.



Based on the above analysis, the Degrees of Equivalence, di, and uncertainties of the Degrees of Equivalence, u(di), were calculated for each participant. Values of di were calculated using Eq 5, where the KCRV is given by the DerSimonian-Laird mean, xDL, and xi is the reported result of participant i:



di xi xDL .



(5)



Page 13 of 22



For the nine participants for which the results were used in calculating the KCRV, the u(di) were calculated using Eq 6, where ui is the standard uncertainty reported by participant i, λ is the interlaboratory variation, and u(xDL) is the standard uncertainty of xDL:



u di



ui2



u2 xDL .



(6)



For the two participants (IPQ and UMTS) whose results were not used in calculating the KCRV, the u(di) were calculated using Eq 7:



u di



ui2



u 2 xDL .



(7)



Table 10 gives the xKCRV = xDL, xi, uKCRV, di, u(di); and the corresponding relative quantities, di/xKCRV, u(di)/xKCRV, for CCQM-K73. The u(di) for CCQM-K73 have a minimum value,



equal to



u2 xDL = 0.000 024 mol·kg-1, that is set by the between-laboratory scatter of



results. This circumstance limits the utility of the results of this KC to support claimed



uncertainties that are smaller than this value.



Table 10. Key Comparison Reference Valuea for CCQM-K73, xKCRV; its Standard Uncertainty, uKCRV; Degrees of Equivalence, di; Standard Uncertainties of the di, u(di); and Relative Values of these Quantities Referred to xKCRV.



Participant (NMI)

KCRV



Value, xi/(mol·kg-1)



di/(mol·kg-1)



u(di)/(mol·kg-1)



Relative value, xi/xKCRV



0.100 930



---



0.000 011



1.000 00



di/xKCRV ---



u(di)/xKCRV 0.000 11



IPQ



0.100 143 -0.000 786 0.000 054



0.992 21 -0.007 79 0.000 54



UMTS 0.100 424 -0.000 506 0.000 240



0.994 99 -0.005 01 0.002 38



INTI



0.100 871 -0.000 059 0.000 153



0.999 41 -0.000 59 0.001 52



CENAM 0.100 894 -0.000 035 0.000 025



0.999 65 -0.000 35 0.000 25



KRISS 0.100 906 -0.000 024 0.000 026



0.999 77 -0.000 23 0.000 26



NIM



0.100 917 -0.000 012 0.000 047



0.999 88 -0.000 12 0.000 46



NIST



0.100 922 -0.000 007 0.000 025



0.999 93 -0.000 07 0.000 24



SMÚ



0.100 937 0.000 007 0.000 025



1.000 07 0.000 07 0.000 25



NMIJ



0.100 942 0.000 012



0.000 025



1.000 12 0.000 12 0.000 24



INMETRO 0.100 974 0.000 044 0.000 025



1.000 44 0.000 44 0.000 24



GUM



0.101 039 0.000 109



0.000 131



1.001 08 0.001 08 0.001 30



a First line of table presents xKCRV, uKCRV, and uKCRV/xKCRV.



The above-tabulated relative values are presented in graphical form in Figure 2, with the results sorted in order of increasing xi/xKCRV.



Page 14 of 22



Figure 2. Relative Degrees of Equivalence, di/xKCRV, for CCQM-K73. Error Bars Denote Expanded Uncertainty of di, U(di)/xKCRV, with k = 2.

Determination of Impurities CENAM and NIM determined impurities in the CCQM-K73 solution. The reported determinations, in terms of mass fraction of the given element, are listed in Table 11. Inductively coupled plasma – mass spectrometry (ICP-MS) was used in all determinations.

Page 15 of 22



Table 11. Reported Impurities in CCQM-K73 Solution.



NMI CENAM

NIM



Impurity i

Na K Br



Mass Fraction



wi/(µg·kg-1)



Ui/(µg·kg-1)



27.3, 27.5, 14.08 1.0, 2.2, 0.79



16.9, 22.2, 10.6 1.5, 1.0. 1.5



336



17



Comments

wi and Ui values listed for bottles 4, 44, 49, in that order

Agilent 7500CE, standard addition



How Far the Light Shines Key Comparison CCQM-K73 directly covers the dissemination of the direct measurement of νH+ at those NMIs that provide this service to customers. The value of νH+ typically measured in such instances is on the order of 0.1 mol·kg-1. However, standardization of strong acids via the determination of the anion of the strong acid (e.g., νCl− by argentimetry in HCl solutions) is not covered by CCQM-K73. In addition, the measurement of significantly smaller νH+ values (e.g., acid rain samples) is not covered, unless the larger uncertainties for the measurement in dilute solutions are taken into account.

In addition to the direct dissemination of the measurement of νH+, CCQM-K73 also provides information on two other major uses of standard HCl solutions at NMIs.

The first use is in the certification of those base CRMs in which an HCl solution with νH+ on the order of 0.1 mol·kg-1 is used as the titrant (or as the excess added substance in a coulometric back-titration) for the acidimetric certification of the CRM. In such cases, CCQM-K73 informs only regarding the attainable uncertainty for the standardization of the titrant (or the excess added substance), not as to the overall certification of the base CRM itself. In most such cases, the uncertainty of the CRM titration (or coulometric back-titration) significantly exceeds that of the standardization of the HCl.

The second use of standard HCl solutions at NMIs is in the determination of the standard electrode potential, E°, for Ag|AgCl electrodes used in the primary measurement of pH [3,9]. For this application, the value of νH+ is usually fixed at 0.01 mol·kg-1. The relevant uncertainty for the disseminated pH measurement is given by the pH KC for the corresponding buffer [10-15]. However, CCQM-K73 can inform as to the expected uncertainty for the standardization of the HCl solution itself. The claimed uncertainty must take into account any increase in the uncertainty associated either with the titration of a solution with νH+ = 0.01 mol·kg-1 or with the gravimetric dilution of a solution with νH+ ≈ 0.1 mol·kg-1 to νH+ = 0.01 mol·kg-1.

Conclusion Eleven NMIs participated in Key Comparison CCQM-K73, Amount Content of H+ in Hydrochloric Acid. The agreement among participants with the lowest uncertainties was not as good as would be expected, although agreement among participants with large reported uncertainties was acceptable. Candidate values for the KCRV and uKCRV were calculated using several statistical estimators.

In an effort to resolve the observed discrepancy in the results, a Coulometric Workshop was conducted in April 2010 at the joint meeting of the EAWG and IAWG. Although several



Page 16 of 22



NMIs reported detailed investigations of possible sources of bias at that Workshop and in subsequent work, no specific cause for the dispersion of results was evident.

A part of the excess variation originates from the change in mass of the bottles during shipment. However, this source of uncertainty was evaluated by a controlled protocol of mass checks of the bottles on receipt. The results of these mass checks indicate that the change in mass associated with shipment (see Table 3) only accounts for a small fraction of the observed excess variation of the results among those participants with the lowest reported uncertainties.

In the absence of a specific cause for the dispersion, the IAWG and EAWG decided to assign a KCRV and uKCRV based on the DerSimonian-Laird statistical estimator. The uncertainty uKCRV is dominated by the between-laboratory scatter of results in CCQM-K73. Owing to the high between-laboratory scatter, the claims of the participants with the lowest reported uncertainties cannot be supported by this comparison.

Acknowledgment The coordinating laboratory gratefully acknowledges the contributions of all participants and of the members of the CCQM Working Groups on Inorganic Analysis and Electrochemical Analysis for their valuable suggestions concerning the measurement protocol and the evaluation process. The contributions of David Duewer (NIST), Hugo Gasca-Aragón (NIST and CENAM), and Steve Ellison (LGC) in the statistical evaluation are also gratefully acknowledged.

References 1. Pratt K W 1994 Anal. Chim. Acta 289, 125-134 and 135-142. 2. Technical Protocol for CCQM-K73 and CCQM-P19.2, distributed 29 June 2009. 3. Máriássy M, Pratt K W and Spitzer P 2009 Metrologia 46 199-213. 4. BIPM (1998) Proceedings of the 4th meeting of CCQM, Bureau International des Poids et

Mesures (BIPM); see also Taylor P, Kipphardt H and de Bièvre P 2001 Accred. Qual. Assur. 6 103-106; Milton M J T and Marschal A 2001 Accred. Qual. Assur. 6 270-271. 5. Máriássy M et al 2006 Metrologia 43 08008. 6. Certificate for NIM CRM GBW06101A, Document GB 10735–200X (updates GB107351989). See also Taylor J K and Smith S W 1959 J. Res. Natl Bur. Stand. (US) A 63 153159. 7. Birge R T 1932 Phys Rev. 40 207-227. 8. Ellison S, personal communication (email) to Pratt K W and Duewer D L, 23 October 2009. Used with permission. 9. Buck R P et al. 2002 Pure Appl. Chem. 74 2169-2200. 10. Final Report for Key Comparison CCQM-K9, 12 December 2001, available at http://kcdb.bipm.org/AppendixB/appbresults/ccqm-k9/ccqm-k9\_final\_report.pdf 11. Final Report for Key Comparison CCQM-K17, available at http://kcdb.bipm.org/AppendixB/appbresults/ccqm-k17/ccqm-k17\_final\_report.pdf 12. Final Report for Key Comparison CCQM-K18, available at http://kcdb.bipm.org/AppendixB/appbresults/ccqm-k18/ccqm-k18\_final\_report.pdf 13. Final Report for Key Comparison CCQM-K19, available at http://www.bipm.org/utils/common/pdf/final\_reports/QM/K19/CCQM-K19.pdf

Page 17 of 22



14. Final Report for Key Comparison CCQM-K9.2, available at http://kcdb.bipm.org/AppendixB/appbresults/ccqm-k9/ccqm-k9.2\_final\_report.pdf

15. Final Report for Key Comparison CCQM-K20, available at http://www.bipm.org/utils/common/pdf/final\_reports/QM/K20/CCQM-K20.pdf

Page 18 of 22



Appendix



Technical Protocol



CCQM-K73/CCQM-P19.2 Amount Content of H+ in Hydrochloric Acid



Introduction

Key Comparison CCQM-K73 and the parallel Pilot Study CCQM-P19.2 are being performed to evaluate the degree of equivalence of national measurement procedures for the assay of hydrochloric acid. The measurand is amount content of hydrogen ion, νH+. The nominal value of νH+ is 0.1 mol·kg-1.

The measurement procedure is left to the participant. Any method or combination of methods is acceptable. It is anticipated that the majority of participants will use coulometry or titrimetry.

Information on impurities, particularly bromide, is also of interest. This information will be provided as an annex to the Key Comparison and Pilot Study results.



Time schedule



Dispatch of the samples: Deadline for receipt of the Data Reporting Form Draft A report distributed Discussion of results and Draft A report Draft B report Approval of Draft B report



30 June 2009 30 September 2009 1 November 2009 EAWG/IAWG joint meeting, Nov. 2009 1 March 2010 EAWG/IAWG joint meeting, April 2010



Description of the sample and details of shipment

The HCl comparison solution is prepared from deionized water and reagent-grade hydrochloric acid. This solution is bottled in separately-numbered, 250 mL high-density polyethylene (HDPE) bottles. The cap of each bottle is sealed with a transparent, heat-shrink plastic seal that extends to the neck of the bottle. Each bottle is sealed inside two bags: a transparent, interior polyethylene bag and an exterior aluminized polyethylene terephthalate (PET) bag. Each bottle and its exterior PET bag are labeled with the number of the bottle and a description of the contents.

Each participant will receive two (three, if requested) separately-numbered bottles filled with the comparison solution. The bottles will be shipped in a cardboard box or other container by air courier. The coordinating laboratory will send the tracking number by email to the contact person of the corresponding receiving laboratory. The contents will be marked “aqueous solution” with value 1 USD per bottle. Please be attentive to possible customs delays, etc. Shipment to all participants will be performed at the same time.

The Coordinating Laboratory will verify the homogeneity of the material before shipment and will perform a stability check in the course of the comparison.



Page 19 of 22



Two spreadsheets are being distributed simultaneously with this Technical Protocol. The file K73\_P19.2\_bottle\_masses.xls is for reporting the masses of the bottles and related information, for verification of integrity of the shipment. The second file, K73\_P19.2\_Data\_Reporting\_Form.xls, is used to report the results of the measurements of νH+ and measurements of impurities.

Actions at receipt of samples

1. Inspect the bagged bottles for visible damage or leakage. 2. Confirm the sample receipt (by e-mail), report any damage, and send the weighing

data for the bagged bottles to the Coordinating Laboratory using the worksheet “(1) Bagged Bottles”. Please send this information as soon as it is available, to permit timely reshipment of a new bottle if necessary. 3. Store the bottles at room temperature in their original PET bags. Refrigeration of the bottles is not necessary.

Instructions for participants

1. Weigh each bagged bottle with a resolution of 0.01 g or less to verify its integrity during shipping. Allow the bagged bottle to equilibrate in the weighing laboratory overnight before performing the weighing. Report both the weighing result (balance reading) and the bottle mass (corrected for air buoyancy) on the worksheet “(1) Bagged Bottles”. Use an assumed aggregate density of 1000 kg·m-3 for the bagged bottle in correcting for air buoyancy. Also report on this worksheet the ambient atmospheric pressure and temperature at the time the bottle was weighed.

2. Store the bottles in their original exterior bags until you are ready to perform the measurements of νH+.

3. When you are ready to start the measurements of νH+, open the exterior bag by cutting it between or above the red lines on the outside the bag. These red lines correspond approximately to the upper edge of the interior polyethylene bag.

4. Remove the bottle with its transparent, interior bag from the exterior bag. Verify that no drops of liquid are present on the inside surface of the interior bag. If drops are noted, leakage has occurred. In this case, please contact the Coordinating Laboratory immediately to obtain a replacement bottle.

5. Remove the bottle from the interior bag. 6. Weigh the bottle, removed from the interior bag, including its label and heat-shrink

plastic seal. Repeat the procedure in Step 1 of this section for the bottle removed from both bags. Record these data on the worksheet “(2) bottles” and send the data to the Coordinating Laboratory. Use an assumed aggregate density of 1000 kg·m-3 for the unbagged bottle. 7. Remove the heat-shrink plastic seal before opening the bottle. Make a small cut (1 mm to 2 mm) in the top portion of the seal with a scissors or knife. Then, peel the remaining seal off the bottle. 8. Please send the file K73\_P19.2\_bottle\_masses.xls to the Coordinating Laboratory a second time with the masses of the unbagged bottles entered in the second worksheet. 9. Perform the assay measurements of νH+ using your selected procedure. To the extent possible, your selected procedure should conform with that which you use to disseminate this measurement capability at your institute.

Page 20 of 22



Reporting

The Data Reporting Form should be sent to the Coordinating Laboratory before 30 September 2009, preferentially by e-mail. The Coordinating Laboratory will confirm the receipt of each Data Reporting Form. If the confirmation does not arrive within 1 week, please contact the Coordinating Laboratory to identify the problem.

The Data Reporting Form has three worksheets, Summary, H+ results, and Method+Example. The requested data should be entered into the corresponding boxes on each sheet. Certain items are automatically copied from one sheet to another. These items only have to be entered once. If you cannot enter your data into the cells as supplied, please change the format as necessary or submit the data in another form. A separate text report in place of the description “Analytical method - detailed description” is equally acceptable.

Please be sure to indicate in the Summary sheet the comparison in which you are participating, by entering an “X” in the applicable box of the Summary worksheet. This X will automatically cause the corresponding comparison to be entered into the worksheets and graph titles.

The uncertainty calculations should conform to the ISO document Guide to the Expression of Uncertainty in Measurement, 1st ed., ISO, Geneva, Switzerland, 1995. Both Type A and Type B components of uncertainty and a summary of how they are calculated must be included. Use the coverage factor, k = 2, to calculate the expanded uncertainty of your result.

Information requested: 1. Report the results as amount content of H+, accompanied by a full uncertainty budget, on the worksheet “H+ results”. 2. Give a detailed description of the measurement procedure. For coulometry, this should include the following: description of the coulometric cell, volume of electrolyte in the working chamber, endpoint evaluation procedure, and the equipment used. 3. Give the complete measurement equation in the designated space in the worksheet “Method+Example”. Include in the space below the values of the input quantities (raw data) for a representative measurement. The data should enable the Coordinating Laboratory to recalculate of the result of that measurement. 4. In the worksheet “H+ results”, report all the individual results, not only the final mean value. 5. Provide a complete uncertainty budget for your measurement. This uncertainty budget must include instrumental sources of uncertainty (mass, time, voltage, volume, ...) as well as chemical ones (endpoint estimation, CO2 interference, side-reactions, purity of calibration standards, ...) 6. In order to facilitate comparisons of your measured masses (for assay measurements), please also provide either (1) the air density used for each buoyancy correction, or (2) the air temperature, humidity and pressure in your laboratory at the time of each mass measurement. 7. Report the details of the procedure used (a separate text file can be used). 8. Information on impurities is welcome, especially for Br−, owing to the influence of trace Br− on the standard potential of the Ag|AgCl electrode in pH metrology.

Page 21 of 22



Reference value The reference value will be agreed upon on the joint meeting of the EAWG and IAWG in November 2009. Participation Participation in CCQM-K73 is open to all institutes eligible for a key comparison in this field. Participation in CCQM-P19.2 is additionally open to other designated laboratories. National Metrology Institutes that desire to use the results of the present comparison to support Claims of Measurement Capability (CMCs) should participate in CCQM-K73, not in CCQM-P19.2. Coordinating Laboratory and contact person: Kenneth W. Pratt NIST (National Institute of Standards and Technology) 100 Bureau Dr., Stop 8391 Gaithersburg, MD 20899-8391 USA Tel.: +1 301 975 4131 Telefax: +1 301 869 0413 Email: kenneth.pratt@nist.gov

Page 22 of 22