Slovak Institute of Metrology

Karloveská 63, SK-842 55 Bratislava 4, Slovakia

Key comparison CCQM-K34

Assay of potassium hydrogen phthalate
Final report

Michal Máriássy

With participation of:

Martin Breitenbach and Christa Oberroeder (BAM); Euijin Hwang and Youngran Lim (KRISS);
Kenneth W. Pratt (NIST); Akiharu Hioki and Toshihiro Suzuki (NMIJ); Ma Liandi, Wu Bing
and Shen Yu (NRCCRM); Alena V. Skutina, Gennady I. Terentiev (UNIIM)

Bratislava, October 2005

Abstract
The CCQM-K34 key comparison was organised jointly by the inorganic and electrochemistry
working groups of CCQM as a follow-up to pilot study CCQM-P36 to test the abilities of the
metrology institutes to measure the amount content of acid in solid weak acids. Slovak Institute
of Metrology acted as the coordinating laboratory, seven NMIs expressed interest in
participation. All participants used constant current coulometry at different sophistication level.
In general very good agreement of the results was observed; some possible problems were
highlighted.

2
1 INTRODUCTION
The CCQM-K34 key comparison “Assay of Potassium Hydrogen Phthalate” has been proposed
and discussed at the October 2002 CCQM Inorganic and Electrochemical Analysis Working
Group meetings in Ottawa as a follow-up for the pilot study P36. The aim of the comparison is to
demonstrate and document the capability of interested National Metrology Institutes to measure
the amount content of acid in a pure weak acid, in this case potassium hydrogen phthalate.
Assays of acids are made almost exclusively by titration methods. Potassium hydrogen phthalate
(KHP) is the most used reference material for these measurements. The reliability of its assay is
therefore of prime importance for chemical producers and analytical chemistry in general. It is
also of interest regarding its use in pH standardisation, where the composition (specifically, the
presence of trace phthalic acid or potassium phthalate) affects the pH value of the buffer
solution. The main impurity present is usually water from mother liquor occluded in the crystals.
There are several producers, who offer this material as a reference material. In our previous
experience, the assays can have a bias an order of magnitude higher than the declared
uncertainties.
The objective of CCQM-K34 was to determine the total amount content (mol·kg-1) of acid in
a sample of potassium hydrogen phthalate. The participants were free to choose the analytical
procedure.

2 LIST OF PARTICIPANTS

Five institutes originally indicated interest in participating in the comparison. At last, seven
institutes took part. Table 1 contains the full names of all participating NMIs and contact
persons.

Table 1 List of participants

Contact person
Institution Country
BAM Germany Martin Breitenbach
Bundesanstalt für Materialforschung und Prüfung
KRISS Korea Euijin Hwang
Korean Research Institute of Standards and Science
NIST USA Kenneth W. Pratt
National Institute for Standards and Technology
NMIJ (AIST) Akiharu Hioki
National Metrology Institute of Japan Japan

NRCCRM China Ma Liandi

National Research Center for Certified Reference Materials
SMU Slovakia Michal Máriássy
Slovak Institute of Metrology
UNIIM Russia Gennady I. Terentiev
Ural Scientific Research Institute of Metrology

3
3 SAMPLES
A batch of old commercial material was selected for comparison. The material was sieved
through plastic sieves and the middle fraction was homogenised in a large bottle and filled into
10 glass bottles closed with silicone lined plastic caps. Four bottles were tested for homogeneity
by analysing each bottle in triplicate by coulometry. Data were treated using ANOVA one-way
analysis [1]. The results indicate that the between bottle variation is negligible compared to the
repeatability of the measurement. The assay did not change on analysing fractions with small or
large crystals nor crushed sample, thus it can be assumed that water content is below 0.005%
(crushing is an effective means of releasing occluded water from KHP).
The samples were sent to the participants by Fedex on May 7, 2004 (except KRISS, where the
sample was handed over personally, and UNIIM, where the sample was sent on May 10 by DHL
due to import regulations). All samples arrived to their destination without damage. The receipt
dates and the responsible persons are given in Table 2.
The deadline for reporting results was set to 31 August 2004 in order to prepare draft A report
for discussion at the CCQM WG meeting in October 2004. All participants reported their results
in time.

Table 2 Sample receipt dates and report dates

Institute Sample receipt Date report sent

date
BAM 10 May 2004 July 15
KRISS 28 April 2004 August 27
NIST 11 May 2004 August 31
NMIJ (AIST) 10 May 2004 August 30
NRCCRM 11 May 2004 August 27
SMU –– pilot
UNIIM 26 May 2004 August 31

4 INSTRUCTIONS TO PARTICIPANTS
The instructions sent to the participants by e-mail consisted of technical protocol and results
report form.
The technical protocol (appendix A) contained background information, timing of the
comparison, and information on the participating institutes. Information on sample homogeneity
and sample preparation for measurements was given. The participants were free to choose the
measurement procedure. Some possible problems with measurement were highlighted.
Participants were requested the results as amount content of acid and to provide uncertainty
evaluation according to ISO Guide [2].
The results report form contained entries relating to the measurement results, detailed uncertainty
evaluation and description of the measurement procedures.

5 METHODS OF MEASUREMENT

4
The methods of measurement were left free to be selected by the participating institutes. The
potential pitfalls for different methods were mentioned in the protocol. They included significant
interference of carbon dioxide (the pH of the solution is alkaline during the final titration) for
volumetric and coulometric titrations. For coulometric titration, there is in addition the
possibility of electrochemical reduction yielding low results.
Indirect methods had to take into account the dependence of the assay on the H/K (or other
metals) ratio and the water content. Probably in view of the difficulties associated with the
indirect assay via impurities, no one of the institutes used this approach; some impurities were
determined, however.
All participants used coulometric titration for assay determination and reported more or less
details on their procedure in their reports or additional information. Some details on
measurements as derived from the reports are given in Table 3. Six participants introduced the
solid samples directly into the coulometric cell, NMIJ used a special way to get the result for
solid sample by extrapolation of the relationship of the result for solution versus inverse of the
sample volume (with constant volume of liquid added). This procedure seems to minimise the
influence of the gas and electrolyte impurities at the expense of more experimental work.

Table 3 Details on measurement methods used

Approx. Procedure details

Institute sample mass Cell volume Main current
/g Cell type IC rinse
/mL /mA
BAM 0.6 vertical, Yes 180 200
1 intermediate
chamber (IC)*

KRISS 0.5 horizontal, 2 IC Yes ~100 100

NIST 0.4 horizontal, 2 IC Yes ~80 100

NMIJ 0.05 horizontal, 2 IC Yes 120 50

NRCCRM 0.5 horizontal, 2 IC Yes 180 100

SMU 0.5 vertical, 1 IC* Yes 250 200

UNIIM 0.5 vertical, 1 IC No 400 100

* - continuous flow into the working chamber during main titration

5
 Table 3 Details on measurement methods used (continued)

Institute EP estimation Initial CO2

titration correction

BAM nonlinear regression Yes -

KRISS max. slope Yes -

NIST 3rd order polynomial regr. Yes -

NMIJ 3rd order polynomial regr. Yes -

NRCCRM 3rd order polynomial regr. Yes Yes

SMU nonlinear regression Yes -

UNIIM 3rd order polynomial regr. No -

6 RESULTS AND DISCUSSION

After receiving all the results, BAM and UNIIM were asked to check their results for numerical
errors. No numerical errors were reported and thus the values are given as originally reported,
but UNIIM indicated that a technical fault in the measurement could lead to errors as high as
0.2%.
The reason for higher result of BAM was not clear. In order to exclude material fault the sample
was sent back to the coordinating laboratory; and a comparison with the original sample revealed
a difference of about 0.023%, which indicates a change in the sample composition. This was
confirmed by ion chromatography, which detected sample contamination with chlorides and
another anion, presumably formate. As it is not exactly known whether the contamination
occurred after BAM received the sample (BAM processed the whole sample), it was agreed to
exclude BAM result from the comparison due to “travelling standard failure” and to do
a subsequent bilateral comparison.
The reported values and uncertainties are summarised in Table 4 and also displayed graphically
in Figure 1.

Table 4 Results (amount content of weak acid, relative standard deviation, relative
combined standard uncertainty and number of measurements)

Result
Institute Measurement date RSD uC,r n
/mol.kg-1
AIST June 7 - July 30 4.89192 0.0146% 0.0151% 2*
KRISS Aug 18 - 25 4.89259 0.0017% 0.0034% 6
SMU July 1 - 14 4.89269 0.0035% 0.0033% 6
NRCCRM July 8 - Aug 19 4.89272 0.0057% 0.0056% 11
NIST July 22 - 29 4.89298 0.0049% 0.0034% 10
UNIIM Aug 9 - 26 4.89458 0.0213% 0.0155% 6
* NMIJ used an extrapolation from several results with different solution sample sizes

6
Several approaches to estimate the key comparison reference value (KCRV) were considered.
Except of arithmetic mean, all gave comparable values (table 5). The use of median and its
uncertainty based on median of the absolute deviations (MAD) was agreed as the KCRV at the
IAWG+EAWG meeting in October 2004. The formula used for calculation of the uncertainty of
the median is as follows [3]:

1.858 ⋅ MEDIAN (xi − KCRV )

u KCRV =
n −1

Table 5 Arithmetic mean, MM-median, median and the weighted mean of the reported
CCQM-K34 results

Possible KCRV Value Standard Rel. stand.

uncertainty uncertainty
Arithmetic mean 4,89291 0,00034 0.0069%
Variance weighted mean 4,89276 0,00013 0.0026%
MM-median 4,89275 0,00020 0.0042%
Median 4,89270 0,00016 0.0033%

4,8950 expressed as
mass fraction
-1
Amount content /mol.kg

4,8942 99,950%

4,8934 99,934%

4,8926 99,917%

4,8918 99,901%

4,8910 99,885%
KRISS

SMU

NRCCRM

UNIIM
AIST

NIST

Figure 1 Results of CCQM-K34 (error bars correspond to expanded uncertainties (k=2);

median and its uncertainty are also given)

The higher uncertainty of NMIJ result is understandable in view o the about 10 times smaller
samples used in the measurement, compared to the other participants.
The higher deviation of UNIIM result can be in part attributable to instrumentation and cell
construction; the high uncertainty is dominated by a component attributed to endpoint
determination.
Five of the six results are in excellent agreement. Most of the results overlap within the expanded
uncertainty.

7
Significant differences can still be observed regarding the uncertainty evaluations, as
summarised in Table 6. The chemical sources of uncertainty were in many cases not taken into
account, thus leading to smaller uncertainty estimates. Some inconsistencies were also noted in
uncertainty calculation.
Impurity determination (Table 7) and moisture (see above) do not explain the lower assay
compared to the theoretical value. The lower assay therefore seems to be due to the presence of
dipotassium phthalate impurity present in the sample.

Table 6 Summary of uncertainty evaluation

Institute Major sources chem. CO2 other phthalate

considered uncertainties influence impurities reduction
considered

BAM voltage, mass, Y N N N

current efficiency

KRISS EP determination N N N N

NIST phthalate reduction, Y Y Y Y

weighing

NMIJ CO2 correction Y Y N

NRCCRM CO2 effect, EP Y Y N N

determination

SMU electrolyte&gas Y Y Y Y
impurities,
diffusion, voltage

UNIIM EP detn., weighing N N N N

7 SCOPE OF THE COMPARISON (HOW FAR THE LIGHT SHINES)

The comparison tested the capabilities and methods used for assay of high purity materials. For
coulometric methods, good results will indicate good performance in assaying both strong and
weak solid acids with pKa<7.

8
Table 7 Impurities determined

Institute (method) UNIIM (ICP MS) NIST (ICP MS) BAM (ICP/OES)
value uncertainty value uncertainty value uncertainty
Impurity mg/kg mg/kg mg/kg
Ca 5.7 (AAS) 5%
Na 35.1 3% 188 5%
Fe 6 47%
Rb 8.5 1% 7.1 50%
Al 1.9 53%
Cu 1.05 114% 1.8 100%
Zn 0.5 60% 9.1 100%
Mn 0.2 50%
Ni 0.13 46%
Mg 0.22 45%
Ba 0.14
B <1
Si <50
Pb <0.1 1.3 100%
Cr, Co, Sr, Cd, Sn, <0.1
Sb, Te, Bi <0.1

8 CONCLUSIONS

Good agreement between the participating laboratories for measurement of potassium hydrogen
phthalate was observed. Median of the results was chosen as the reference value (amount content
of acid 4,89270 mol/kg, associated expanded uncertainty 0.00032 mol/kg).
The comparison demonstrated that great care must be taken if the assay of compounds is based
only on 100%-impurities concept, as even the most important impurities may remain undetected.

9 ACKNOWLEDGEMENTS
The coordinating laboratory gratefully acknowledges the contributions of all participants and of
David L. Duewer (NIST) for critical evaluation and provision of a copy of PDF_maker macro
spreadsheet to enable the calculation of the MM-median.

10 REFERENCES
[1] Van der Veen A, Linsinger TPJ, Pauwels J, Uncertainty calculations in the certification
of reference materials. 2. Homogeneity study. Accred. Qual. Assur. 6, 26-30 (2001)
[2] BIPM, ISO, IEC, OIML, Guide to the expression of Uncertainty in Measurement (1995)
1st ed., ISO, Geneva.
[3] Müller J.W., Possible Advantages of a Robust Evaluation of Comparisons. J. Res. Natl.
Inst. Stand. Technol. 105, 551-555 (2000)

9
Appendix A – Technical Protocol

CCQM-K34 Assay of potassium hydrogen phthalate

Technical protocol
Introduction

Assays of acids are made almost exclusively by titration methods. Potassium hydrogen phthalate
(KHP) is the most used reference material for these measurements. The reliability of its assay is
therefore of prime importance for chemical producers and analytical chemistry in general. It is
also of interest regarding its use in pH standardisation, where the composition (specifically, the
presence of trace phthalic acid or potassium phthalate) affects the pH value of the buffer
solution. The main impurity present is usually water from crystallisation.
There are several producers, who offer this material as a reference material. In our previous
experience, the assays can have a bias an order of magnitude higher than the declared
uncertainties.
After the successful pilot study, CCQM approved a key comparison to underpin the claimed
calibration and measurement capabilities of the institutes.

Scope:
The comparison will test the capabilities and methods used for assay of high purity materials.
For titration or coulometric methods, good results will indicate good performance in assaying
both strong and weak solid acids.

Time schedule

Dispatch of the samples: beginning of May 2004

Deadline for receipt of the report: 31 August 2004
Distribution of Draft A for comments: end of September 2004
Draft A discussion: IAWG meeting in October 2004.

Samples

Each participant will receive one numbered bottle containing about 20 g of material. Shipment to
all participants will be performed at the same time. The bottles are shipped in a cardboard box by
courier and the airwaybill/consignment number1 is reported by email to the contact person of the
receiving laboratory for tracking purposes. The contents will be marked “potassium hydrogen
phthalate” for research purposes and value 1 USD; please be attentive of possible customs
delays, etc. The measurement protocol is sent by e-mail.
The homogeneity of the sample material was measured based on assay using sample size of
about 500 mg and found to be adequate for the key comparison.
The assay is in the range of 99.9 – 100% of the theoretical value.

1
aiwaybill/consignment number, the carrier identification of the shipment allowing detailed tracking of the
shipment. If you have not received the shipment within 3 days of our notice, please use the tracking facility to
monitor whether your shipment is being held up in customs or similar.

10
Actions at receipt of samples
Please inspect the received bottles for damage. Please inform the contact person of receipt and
report any mishaps to the coordinating laboratory. The sample should be stored at laboratory
temperature in the original container until used.

Sample preparation for measurement

The material should be dried at 110°C for 2 h without crushing or grinding the material.

Measurement method

Any method or method combination can be used for this comparison. The results will be
reported as amount content [mol/kg] of monoprotic weak acid, to be accompanied by a full
uncertainty budget. Information on the assay dependence on sample mass is also welcome. At
least four determinations should be performed (where applicable).
Indirect methods must take into account the dependence of the assay on the H/K(or other metals)
ratio and the water content.

Reporting

The report should be sent to the coordinating laboratory before August 31, 2004, preferentially by
e-mail. The coordinator will confirm the receipt of each report. If the confirmation does not arrive
within one week, contact the coordinator to identify the problem.
A template for the report will be enclosed (Excel sheet). If possible the requested data should be
entered into the corresponding boxes, if not the format can be modified or the data can be reported
in another form.
Information requested:
1. Report the results as amount content [mol/kg] of weak monoprotic acid, accompanied by a
full uncertainty budget. Information on impurities is welcome also from participants not
using (100% - impurities) approach.
2. If the assay is determined from impurity analysis, results for all the elements/compounds
sought must be included.
3. A detailed description of the measurement procedure is to be given (for coulometry this
should include also: cell description, volume of electrolyte in working chamber, endpoint
evaluation procedure, example titration curve for initial and final titration), and of the
equipment used.
4. The complete measurement equation has to be given, as well as the values of the constants
used and variables (raw data) for at least one measurement. The data should enable the
recalculation of the result of this measurement.
5. State all the individual results, not only the final mean value.
6. The uncertainty budget has to include instrumental sources of uncertainty (mass, time,
voltage, volume, ...) as well as chemical ones (endpoint estimation, equilibria, CO2
interference, impurities, purity of calibration standards, ...). The uncertainty calculations
should conform to the ISO document: Guide to the expression of Uncertainty in
Measurement (1995) 1st ed., ISO, Geneva. Both Type A and Type B uncertainty components
and a summary of how they are calculated have to be included.
7. 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.
8. Report the details of the procedure used (a separate text file can be used).

11
 Reference value
The reference value will be agreed upon on the meeting of IAWG.

Participation
Participation is open to all institutes eligible for a key comparison in this field.

The Draft A Report, based on the reported results will be prepared and sent to the participants for
comments and will be discussed at the autumn 2004 meeting of CCQM Working Groups on
Electrochemical Analysis and on Inorganic Analysis. The individual reports will also be
distributed among the participants.

Coordinating laboratory and contact person

Michal Máriássy
Slovenský metrologický ústav (Slovak Institute of Metrology, SMU)
Karloveská 63
SK-84255 Bratislava 4
SLOVAK REPUBLIC
Tel.: +421 2 602 94 522 Fax: +421 2 654 29 592
E-mail: mariassy@smu.gov.sk

12