Accred Qual Assur (2000) 5 :437–440Q Springer-Verlag 2000 GENERAL PAPER

M. MáriássyL. VyskocilA. Mathiasová

Link to the SI via primary direct methods

Received: 17 March 2000Accepted: 21 June 2000

Presented at 3rd EURACHEMWorkshop “Status of Traceability inChemical Measurement”, 6–8 September1999, Bratislava, Slovak Republic

M. Máriássy (Y) 7 L. VyskocilA. MathiasováSlovak Institute of Metrology, Karloveská63, 842 55 Bratislava, Slovakiae-mail: mariassy6smu.gov.skTel.: c421–7–6029 4522Fax: c421–7–6542 9592

Abstract The possible approachesto realising a link to the SI systemand the status of primary directmethods in the traceability chain ofchemical measurements are dis-cussed. Some results obtained withthe new coulometric standard sys-tem are presented.

Key words Traceability 7Coulometry 7 Uncertainty 7Primary methods

Introduction

There is general agreement that an artefact for the real-isation of the unit mole as the top of the traceabilitychain is not needed/rational. However, there is a largevariety of opinions on the nature of the link to the SI.They range from primary methods through pure ele-ments to commercial substances. Often we hear objec-tions that the uncertainty at this level is negligible com-pared to that in routine measurements, so that work atthis level is unimportant. This is usually true for traceconstituents, but for analysis of major or minor compo-nents standards may be a significant source of error.

Is there a difference between chemical and physicalmeasurements? Most chemical analytical methods arerelative, i.e. they compare the signal generated by asample to the signal generated by the same quantity ofa standard. In physical measurements, the standard isoften incorporated into the measuring instrument,therefore the instrument may be calibrated for longertime periods. In contrast, in chemical measurements theinstruments usually serve as comparators between thesignals of unknown samples and external standards. In

addition, chemical assay leads either to an intensivequantity or the sample (and standard) is used up.

The values of extensive quantities (in contrast to in-tensive quantities) depend on the system size (theamount of solute depends on the volume of the solu-tion taken; its concentration does not).

Partial quantities are related to some part of the sys-tem; they have their opposite in integral quantities,whose value relates to the system as a whole(likelength, volume, mass, voltage,...). As an example: hav-ing a mixture of two components, like sand and seeds –the mass of the mixture of both is an integral (and ex-tensive) quantity, the masses of the individual compo-nents are partial in nature (extensive, too). Only inte-gral extensive quantities are those that can be mea-sured directly, partial or intensive ones are calculatedfrom the results of other measurements.

In amount of substance measurements, we almostnever face the need to determine the sum of all compo-nents. We try to determine specific substances thatform a part of the whole. Amount of substance, whichis of course an extensive quantity, can be considered ashaving partial nature – this is supported by its defini-tion, too (entities must be stated). Any prepared stand-

94 M. MÆriÆssy• L. Vysko�cil • A. MathiasovÆ

ards cannot be dosed by amount of substance – for anydosing we need to measure an integral quantity likemass or volume. That means the certification of refer-ence materials is best done in terms of concentration(mol/dm3) or amount content (mol/kg).

The Consultative Committee for Amount of Sub-stance (CCQM) has set up a definition of primarymethods [1, 2] and has selected some methods with thepotential of being “primary”, from the viewpoint of theend user. From the point of view of metrology, meth-ods used for linking the chemical measurements withthe SI system at the highest level should not refer toother amount of substance standards. This requirementexcludes methods which are relative in their principle.Some other methods identified as having the potentialto be primary yield information expressed as amountfraction. This is essential for evaluation of purity, but inorder to convert it to a value useful for transfer of theunit, additional information on the identity (molarmass) and content of the impurities is required. Thisadditional information is needed to convert the resultinto amount content or similar quantities.

From practical considerations, for the area of inor-ganic analysis, there are two methods, the results ofwhich are not dependent on a known amount of sub-stance in some form of reference material (RM) (some-times called absolute methods):– Coulometry– Gravimetry.

Coulometry is based on direct or indirect electro-chemical transformation of the determined substance.For a complete electrochemical transformation ofamount of substance n of the substance determined, weneed electric charge Q quantitatively described by theFaraday law:

npQ

z7Fwhere n is amount of substance, Q is electric charge, zis charge number of the electrochemical reaction and Fis the Faraday constant (96485,3415B0,0039) C/mol(CODATA 1998).

Determination of the amount of substance is thus indirect relation to basic units of the SI system and doesnot need a RM for comparison. The Faraday constantis one of the fundamental constants (it can be ex-pressed as the product of the electron charge and theAvogadro constant). It enables the attainment of highprecision and accuracy and is independent of theatomic weights of the elements in the sample. Its draw-back is lower selectivity, a feature common to titrationmethods. This makes coulometry especially suitable fordetermination of relatively pure substances used asstandards by other (relative) methods. The Faradayconstant has been proposed as an ultimate standard inchemistry [3].

Gravimetric analysis is one of classical analyticalmethods. It is based on chemical transformation of thesample using excess of a reagent to a substance, whichis weighed after processing. The weight of the sub-stance obtained serves as a base for calculation ofamount of substance.

The advantage of the method is its feasibility withcommon laboratory equipment. The disadvantage liesin lower selectivity and in the integral character of massmeasurements, i.e. they determine a property of the ob-ject as a whole. Moreover, for attaining the highest ac-curacy needed at this level, the need of determinationof actual atomic weights cannot be overlooked in someinstances.

Strictly speaking gravimetry belongs to the relativemethods. The only difference is in the position of theRM in the measurement process: the weighed productitself serves as an RM – usually with the a priori as-sumption, that its purity is 100% and stoichiometry iscorrect. The mass fraction of the weighed substanceshould be taken into account in the equation used.Problems associated with the formation of solid phase[4], e.g. surface adsorption effects (ionic species andwater [5]) are significant in analyses aiming at relativeuncertainties at about 10–4.

Another possible approach, which is broadly used, isto use high-purity substances (indirectly assayed) asstandards. For use at the highest level, this approachrequires the determination of all important impurities inthe sample. This means not only metallic impurities,commonly stated in the manufacturers certificates, butalso non-metals as oxygen, carbon, etc. The content ofimpurities is not always known in advance. If the totalcontent of impurities is very low, the uncertainty oftheir determination does not affect the required uncer-tainty of the sample assay. Some other problems arediscussed in Ref.[5]. The need of determining the molarweight may equally apply here.

A schematic view of the traceability and uncertain-ties in inorganic analysis is depicted in Fig. 1.

The Slovak Institute of Metrology (SMU) has re-cently rebuilt its high-precision coulometric equipmentto be used as a national standard for amount of sub-stance measurements. Its main purpose is certificationof primary reference materials of composition with di-rectly determined main component. These primary ref-erence materials can be subsequently used for dissemi-nating traceability into different chemical measure-ments.

Results and discussion

The new, computer-controlled equipment was con-structed based on past experience with coulometrygained at SMU. Some parts were constructed specially

Link to the SI via primary direct methods 95

Fig. 1 A schematic view of the traceability and uncertainties ininorganic analysis (w –amount content)

Table 2 Uncertainty budget for 0.5 g samples (relative contributions in 10–6)

Amidosulfuricacid

PotassiumHydrogen Phthalate

PotassiumDichromate

PotassiumChloride

ArsenicOxide

Incomplete rinsing 2 4 2 2 2Spray losses 2 2 0 0 0Sample introduction 0 0 0 0 5Current efficiency 1 1 1 30 1Electrolyte impurities 10 10 10 10 1Inert gas impurities 5 15 5 5 2Diffusion 1 10 10 5 10Adsorption – – – 50 –Total: 11.6 21.1 15.2 59.6 11.6Weighing 6.5 6.5 0.5 6.5 6.5Air buoyancy correction 0.8 1.3 8.5 0.9 0.5Voltage measurement 9.8 9.8 3.0 9.8 8.8Electric resistance 3.0 3.0 3.0 3.0 3.0Uncertainty of mass and charge: 12.2 12.2 11.1 12.2 11.4Total Type B uncertainty 17 24 19 61 16

Table 1 Selected assays made on the new system

Material Analyte Amount contentw/mol7kg–1

wexp/wtheor/% RSD/% uC/% Remark

K2Cr2O7 CrVI 6.79732 99.9832 0.0010 0.0019 RM6.79694 99.9777 0.0024 0.0022 CCQM-P7

HSO3NH2 Hc 10.29878 99.9960 0.0032 0.0019 RM10.29652 99.9740 0.0015 0.0019 COOMET

KCl Cl– 13.4113 99.9722 0.0033 0.0062 CCQM-P7NaCl Cl- 17.1061 99.9829 0.0029 0.0066 CCQM-P7

0.0250224 – 0.032 0.021 Solution

for this purpose. The control program was written inTurbo Pascal.

The entire constant current coulometric system con-sists of several instruments, completed by auxiliaryequipment. The main parts of the coulometric systemare given in the following list:– Current source 7961 (I~ 1 A) (Applied Precision)– Indication unit 8971 (potentiometric, amperometric)

(Applied Precision)– Relay/valve unit (Applied Precision)– Digital voltmeter Solartron 7071 (Schlumberger)– Microbalance S4 (Sartorius)– Analytical balance AE240S (Mettler)– Standard weights– Standard 1 V resistors (Metra)– Piston burette 665 Dosimat (Metrohm)– Coulometric cells– PC Pentium, 266 MHz– Auxiliary equipment

The main metrological parameters of the new sys-tem are as follows:– Typical RSD 7 0.002 %– Typical Type A uncertainty (for np10) 0.0007%– Typical Type B uncertainty 0.002–0.006%– Combined uncertainty 0.002–0.006%

96 M. MÆriÆssy• L. Vysko�cil • A. MathiasovÆ

– Expected change per year (drift of components)–0.0015%

– Expected change per year (after drift correction inprogram) –0.0002%Results of some measurements made on the new

coulometric system from November 1998 are given inTable 1.

The main problem in evaluating the uncertainty ofmeasurements in coulometry lies in identification ofimportant uncertainty sources and estimation of theircontribution (Table 2). With very low instrumental un-certainty, other factors become limiting to the achieva-ble uncertainty, mainly those connected to the chemicalprocesses in the cell and the homogeneity of the mate-rial.

The situation is seldom favourable enough to enablethe use of a single method for establishing a link to theSI system. The information on the content of impurities(by means of relative methods) is needed in most cases.Except for coulometry, determination of molar weight

is of importance for highest level work. Thus seekingfor a “method that stands alone” seems to be over op-timistic. In order to link an amount of substance meas-urement of a given species to the SI system at the high-est level, the choice of a particular method will dependon the nature of the species and possibilities of themethods under consideration.

The “ultimate” link to the SI system is currently pos-sible in some cases only by using pure substances (e.g.elements) and their corresponding atomic weights;however, a more straightforward way is to use directmethods like coulometry, which eliminate the problemsassociated with the dissolution step. Their use can fulfilboth the role of a standard (they incorporate the unitmole into the measurements) and the role of measure-ment capability.

Acknowledgements This work has been supported by the U.S. –Slovak Science and Technology Joint Fund, Grant No 94 086.

References

1. CCQM (1995) Report of the 1st meet-ing. BIPM, Sèvres

2. CCQM (1999) Report of the 4th meet-ing (February 1998). BIPM, Sèvres

3. IUPAC, Commission on Electroanaly-tical Chemistry (1976) Pure ApplChem 45 :127-130

4. Beck II CM, Watters RL (1996) Docu-ment CCQM/96-13

5. Yoshimori T (1975) Talanta 22 :827-835

6. Moody JR, Greenberg RR, Pratt KW,Rains TR (1988) Anal Chem60:1203A-1218A