CEN/TC 132

Date: 2025-11

EN 14726:2025

CEN/TC 132

Secretariat: AFNOR

Aluminium and aluminium alloys — Determination of the chemical composition of aluminium and aluminium alloys by spark optical emission spectrometry

Aluminium und Aluminiumlegierungen — Bestimmung der optische Funkenemissionsspektra-lanalyse der Aluminium und Aluminiumlegierungen bei optische Funkenemissionsspektral-analyse

Aluminium et alliages d'aluminium — Détermination de la composition chimique de l'aluminium et des alliages d'aluminium par spectrométrie d'émission optique à étincelles

ICS:

Descriptors:

Contents Page

European foreword 3

Introduction 4

1 Scope 5

2 Normative references 5

3 Terms and definitions 5

4 Symbols and abbreviations 5

5 Principle 6

6 Apparatus 6

6.1 Spark optical emission spectrometer 6

6.2 Equipment for sample preparation 6

7 Consumables and reference materials 7

7.1 Consumables 7

7.2 Reference materials and recalibration samples 7

8 Samples 8

8.1 General case 8

8.2 Sampling of finished and semi-finished products 8

8.3 Sample preparation 8

9 Operating conditions of the spectrometer and measurements 9

10 Calibration procedure 11

10.1 General 11

10.1.1 Calibration process 11

10.1.2 Range of calibration 11

10.1.3 Number of sparks on calibration samples 11

10.2 Calibration 11

10.3 Recalibration 11

10.4 Type recalibration 12

11 Accuracy (precision and trueness) 12

12 Controls 12

13 Test report 12

Annex A (informative) Representative sparking area 13

Annex B (informative) Detailed information on calibration 14

Annex C (informative) Detailed information on recalibration 18

Annex D (informative) Detailed information on accuracy and uncertainty 20

Annex E (informative) Guidance for controls 23

Bibliography 25

European foreword

This document (prEN 14726:2026) has been prepared by Technical Committee CEN/TC 132 “Aluminium and aluminium alloys”, the secretariat of which is held by AFNOR.

This document is currently submitted to the CEN Enquiry.

This document will supersede EN 14726:2019.

prEN 14726:2026 includes the following significant technical changes with respect to EN 14726:2019:

— deletion of the sentence about the determination of mercury content from Scope as being a recommendation;

— rewording of Subclause 10.2;

— rewording of Subclause 10.3;

— deletion of Subclause 10.5;

— minor text updates in Clauses 7.2, 9 and 13;

— introduction of requirements regarding reference materials in Table B.1;

— rewording of Subclause B.3.1;

— rewording of last paragraph, Subclause C.1;

— deletion of Subclause C.3;

— updated Bibliography.

Introduction

In spark optical emission spectrometry (S-OES), a small portion of the sample is thermally vaporized through the erosion of an electric spark. In the spark discharge, the aerosol is vaporized, partially ionized and excited to emit optical radiation. The characteristic radiation of each element is used in spark optical emission spectrometry for its detection and for its quantitative determination.

Optical emission spectrometry (OES): A technique that measures the emission characteristic of a material in the ultraviolet, visible, or infrared wavelength regions of the electromagnetic spectrum. Atomised particles are excited, and each element emits a characteristic radiant energy. This characteristic radiation is detected using either a photomultiplier tube or a solid state detector; appropriate software is used to record the presence of elements and to quantitatively determine elemental content.

Spark optical emission spectrometry (S-OES): A technique that utilizes a high voltage capacitor discharge to ablate and atomise a section of the tested material in an inert atmosphere. The excited atoms and ions emit electromagnetic radiation, which is detected and analysed by an optical emission spectrometer.

Spark optical emission spectrometry is suitable for determining the chemical composition of alloys before the manufacturing and casting processes: in these cases, samples are taken from the liquid metal at different stages of the casting process. Spark optical emission spectrometry is also used to determine the chemical composition of final products.

The method covered by this document is primarily for the analysis of aluminium or aluminium alloy chill cast solid samples, as described in EN 14361, although other samples forms are acceptable.

This document describes the criteria and the procedure for analysing aluminium and aluminium alloys with spark optical emission spectrometry (S-OES). This document specifies the following:

— sample preparation;

— operational guidelines for an optical emission spectrometer (including maintenance);

— traceability of the analytical results to the International System of units: mass (kg);

— assessing the uncertainty associated with each analytical result.

This document refers to simultaneous spark emission spectrometers for the analysis of solid samples.

This document applies to the determination of silicon, iron, copper, manganese, magnesium, chromium, nickel, zinc, titanium, boron, gallium, vanadium, beryllium, bismuth, calcium, cadmium, cobalt, lithium, sodium, phosphorus, lead, antimony, tin, strontium and zirconium in aluminium and aluminium alloys.

This document is applicable to the determination of elements other than those listed above with the following conditions:

a) suitable reference materials are available; and

b) the instrument is suitably calibrated and equipped.

The test result obtained from a spark optical emission spectrometer generally concerns an amount of less than one milligram per spark spot. The result can be used to refer to the laboratory test sample, to the aluminium or aluminium alloy melt or to the cast product.

The following documents are referred to in the text in such a way that some or all of their content constitutes requirements of this document. For dated references, only the edition cited applies. For undated references, the latest edition of the referenced document (including any amendments) applies.

EN 12258‑2, Aluminium and aluminium alloys - Terms and definitions - Part 2: Chemical analysis

EN 14361, Aluminium and aluminium alloys - Chemical analysis - Sampling from metal melts

For the purposes of this document, the terms and definitions given in EN 12258‑2 apply.

ISO and IEC maintain terminology databases for use in standardization at the following addresses:

— ISO Online browsing platform: available at https://www.iso.org/obp

— IEC Electropedia: available at https://www.electropedia.org/

Symbols are defined for each formula. Abbreviations are put in brackets immediately after a term first appears in the text (e.g. S-OES).

Measurement of the intensity of the radiation, whose wavelength is characteristic of each element, generated by a spark resulting from the application of an electrical discharge between the sample, as one electrode, and an inert counter-electrode, after mechanical preparation of the surface of the sample, which is in general taken from the metal melt.

The content of each element is determined by relating the measured intensities of the samples to calibration curves.

Signals are evaluated using:

— alloy calibration and universal calibration: reference materials with similar chemical compositions are used to prepare evaluation functions;

— master curve calibration: reference materials with known chemical compositions are measured, and evaluation functions are recalculated.

Evaluation of the accuracy of the results, in accordance with quality assurance procedures, to be defined by each laboratory.

The optical emission spectrometer shall utilize excitation by spark discharge and be suitable for the determination of the chemical composition of aluminium and aluminium alloys.

Spark optical emission spectrometers are composed of the following main functional devices:

a) system for atomization and excitation:

1) spark generator (spark source);

2) spark stand with counter electrode.

b) optical system (system for spectral radiation splitting);

c) system for radiation intensity measurement (radiation detectors);

d) system for acquisition of the measured values, data processing and evaluation.

The documentation of spark optical emission spectrometers should be in accordance with the requirements of EN ISO/IEC 17025.

Lathes, milling machines, circular and band saws, grinders or any other suitable device can be used for the preparation of the samples. Equipment used for the surface final preparation stage shall be capable of machining both reference samples and test samples to the same condition.

NOTE Adjustable cutting speeds are advantageous for alloys of different hardnesses.

Consumables are generally specified:

— in the laboratory analytical procedures,

— in the manufacturer equipment instructions, or

— according to preliminary tests.

Consumables include, but are not restricted to the following:

— feed gas of specified purity (argon for spectrometry, high purity; see instrument manufacturer recommendation);

— gas purification cartridge for the feed gas (if required to meet the instrument manufacturer specifications);

— cleaning brushes for the counter-electrode, if needed (the filaments should not contaminate the electrode);

— particle trap for filtering the metal condensate out of the waste-gas stream;

— spare and expendable parts for the spark optical emission spectrometer in accordance with the manufacturer's instructions (e.g. counter electrode, insert for sample table, etc.).

The certified reference materials, the reference materials (see ISO Guide 30) and the recalibration samples should be listed and documented in a laboratory procedure and/or in the validation or verification report:

— certified reference materials for calibration (see Clause 10);

— blank sample: high-purity aluminium or aluminium alloy prepared from high-purity constituents (e.g. Al Sn30) (see Clause 10);

— binary samples (if required e.g. for line interference correction (see Clause 10));

— control samples for checking the accuracy of the calibration; they shall not be included in the calibration functions (see Clauses 10 and 11);

— samples for the control of the spectrometer drift (see Clause 12);

— recalibration samples for drift correction (see Clause 10).

Sampling plays an essential role in the accuracy of the analytical results. Sampling allows obtaining laboratory samples whose dimensions are suitable for the preparation of test samples for S-OES and whose chemical composition shall represent that of the material to be tested.

Sampling of molten metal shall be carried out according to EN 14361.

Test samples, shall present a defined area which represents their average chemical composition. This area shall be sufficiently homogeneous across the test section. The position and size of the representative sample area varies with the sampling conditions, as well as with the type of alloy and the analytes.

NOTE 1 The test result only refers to the effective test area which is the vaporized fraction of the sample.

For the simultaneous multi-element analysis of different alloys, a mean analysis zone should be defined.

For S-OES, there is an additional requirement: as far as possible, the metallographic structure in the test sample and in reference materials should be similar.

NOTE 2 Cylindrical samples with ∅ 40 mm × 30 mm (∅ 55 mm × 30 mm) and disc samples with a central sprue, e.g. ∅ 50 mm × 10 mm or ∅ 55 mm × 4,5 mm (also called plate or mushroom sample) are frequently used.

A piece, suitable for use on the spark stand, is mechanically separated (e.g. by sawing) from the part to be analysed (see NOTE). Such piece shall have a minimum thickness of approximately 1 mm, cut in such a way that allows the plane surface to be machined or otherwise prepared. Additionally, it should be large enough for a sealing edge to protrude over the opening of the sample table (exception: air stand). When using small samples, care should be taken to ensure that no overheating occurs due to sparking.

NOTE A small piece of any finished or semi-finished product can never truly represent the whole and a sample of this type cannot be used to certify a cast.

To check the homogeneity of a sample using spark spectrometry, the piece of metal or ingot can be analysed at various locations (e.g. along the diagonal through the part); special attention should therefore be given to areas susceptible to segregation.

Attention should be given to possible systematic deviations due to structural differences to the reference materials during evaluation.

A compromise for samples of sufficient size is to re-melt them in a suitable furnace under inert gas to produce a sample similar to those normally used for S-OES. However, volatile elements, such as sodium, magnesium can be partially lost during a re-melting operation.

For spark optical emission spectral analysis, a plane, flat surface in the representative area is produced on the sample by machining. Lathes and milling machines are used.

During final machining, the cutting speed, cutting angle and cutting tool should be chosen in such a way that no sample material is raised above the machined surfaces and that no single hard grains are torn from a soft microstructure. A certain residual roughness promotes the formation of electric sparks (see manufacturer's instructions). The machined surface shall not be touched directly with the fingers or otherwise contaminated, especially for the determination of sodium, calcium and phosphorous.

For the sample preparation of reference materials, check the homogeneity of a sample using the same technique with the same machining parameters shall be used so that a similar surface condition is achieved.

A compromise for samples of sufficient size is to re-melt them in a suitable furnace under inert gas to produce a sample similar to those normally.

The operating conditions of the spectrometer shall be optimized.

NOTE 1 For simultaneous spectrometers equipped with photomultipliers, the detector channels for the individual elements are pre-set on the base of the spectral line table, which is generally fixed by the apparatus manufacturer according to each task definition (system requirements/specifications). Changes and expansions are only possible by modifying the spectrometer. The space requirement for a detector channel does not allow the combination of just any lines in a spectrometer of a given design. As a result, several optical spectrometer units are sometimes used in one instrument.

NOTE 2 Information about possible interferences due to line overlap can be deemed from the spectral line reference table, whereby interferences as a result of lines being in a different order can also be considered. In general, the measuring signal of an analytical line is related to a line of the matrix element taken as a reference line. For trace analyses, sometimes the intensity of a background position is used as reference.

NOTE 3 Other measuring conditions, such as spark parameters for pre-sparking and measurement sparking, flush time, pre-spark time, delay time, measurement period, time-resolved intensity measurement, masking-out of the radiation from the plasma, high-voltage adjustment of the photomultiplier tubes, are as a rule adjusted by the manufacturer of the apparatus according to the measuring task or are optimized in accordance with the manufacturer's instructions.

As instrument and computer software design differ, information on measurements, spectrometer controls, the auxiliary equipment and maintenance operations shall be carried out in accordance with the manufacturer's instructions and other relevant documents.

These instructions should be transposed into laboratory procedures describing individual analytical programs and operational processes. To that, the following items should be taken into account:

a) start-up (restart):

1) check before start-up (e.g. argon feed, exhaust-gas duct, cooling water, vacuum pump oil);

2) switch on the spectrometer and all units generally in the following order: cooling water pump, vacuum pump (if any), instrument electronics and high voltage power supply;

3) start computer and analytical program;

4) check the instrument status stability (e.g. vacuum, temperature, instrument profile);

5) check the analysis stability (e.g. measurement of suitable spectrometer control samples).

b) switching off:

1) back-up of data;

2) close vacuum valve or purge spectrometer with inert gas, if necessary;

3) switch off the instrument units in the following order: high voltage power supply, instrument electronics, vacuum pump (if any), cooling water pump;

4) switch off the instrument power supply;

5) shut off gas feed.

c) sparking (independent of sample type):

1) in case of manual operation: place sample, start measuring cycle, exclude bad sparking (e.g. intensity of the reference line < 90 %; memory effect), clean counter electrode and spark stand, number of sparkings (may vary according to sample type), method of averaging;

2) in case of instruments with sample handling systems (robots): position sample, start.

d) drift correction or recalibration:

1) list of recalibration samples (setting-up samples);

2) procedure, criteria of decision for outliers, maximum allowable deviations;

3) responsibilities.

e) carrying out an analysis or checking control analysis:

1) selection of the analytical program (according to alloy or alloy group);

2) sample identification;

3) validation of the results (criteria, responsibilities);

4) test report.

f) fault detection and backdated corrective actions in the event of instrument malfunction:

1) spectrometer failures;

2) computer and software failures;

3) analytical failures (certified control samples);

4) rejecting (suspensions) of released analytical results (backdated before the recognition of the discrepancies).

g) tests on spectrometer system:

1) status measured values (voltage stability, temperature, vacuum, gas flow);

2) spark parameters (electrode gap);

3) intensity during drift correction;

4) line profiles;

5) light transmittance of the entrance window;

6) limits of detection and background equivalent contents;

7) short and long term precision.

h) maintenance and cleaning:

1) at fixed intervals in relation to the consumables, depending on usage

2) by external service technician;

3) data back-up.

The calibration process is subdivided into calibration and drift compensation by recalibration.

The range of calibration for an element shall extend well below the minimum content reported in the relevant products standards and above the maximum content reported in the same products standards, taking into account that the lowest limit should be at least three times the detection limit.

The number of sparks carried out on each reference material for calibration shall be not less than four. The spark areas shall be distributed over the prepared surface. Centre and border of the sample shall be avoided. All measurements shall be examined; if any measurement is obviously defective, further sparks shall be carried out to obtain the minimum four acceptable measurements. The intensity average of the four acceptable measurements is used for calibration provided that the related relative standard deviation is < 5 %. For further information on representative sparking area, see Annex A.²

The calibration of the spectrometer is performed by using a series of suitable materials of known composition (preferably RM and CRM), which shall have the same or at least a similar matrix and metallurgical structure as the samples to be analysed, in order to calculate the calibration functions from which the analysis of the analytical samples can be obtained. The content range of the materials used shall cover that of all the samples to be analysed within each specific analytical program. For each element in each reference material, the mean intensity is correlated to the corresponding certified content and a regression is calculated.

Calibration is usually performed when the device is installed. Calibration shall be carried out in accordance with the spectrometer manufacturer's operating manual.

The trueness of the analytical procedure is checked by measuring a set of certified reference materials or - if not available - a set of reference materials not used in the calibration. These reference materials shall cover at least the low, mid and high points of the calibration range for each element.

NOTE For further information on calibration, see Annex B.

Drifts of the spectrometer readings shall be corrected using a recalibration procedure (often described in the manufacturer's instruction manual). Recalibrations can be done either for all analytical channels (global recalibration), or only for individual analytical channels (selective recalibration).

Recalibration can either be done periodically or due to a deviation from statistical process control (SPC) limits (see Annex C). When a periodical recalibration procedure is used, the period depends on the stability of the spectrometer and has to be established after stability tests of the spectrometer. The stability check shall be repeated at appropriate intervals.

NOTE 1 The same set of check samples can be used both for drift control and for statistical process control of the spectrometer.

After recalibration, the trueness shall be checked by measuring reference materials not used in the recalibration procedure.

NOTE 2 For further information on recalibration, see Annex C.

Type recalibration offers a further possibility of correcting instrument drift and, in addition for matrix effects. Here, one or two certified reference materials of the same material type and of comparable composition are analysed together with the test sample. Subsequently, by means of linear correction, the contents of the test sample are related to the certified values of the certified reference materials used.

NOTE This practice cannot be used if the calibration function is a second degree one, unless if the element to be calibrated has a content very close to this of the type calibration certified reference material.

The aim is to state a reliable range of uncertainty [see also ISO/IEC Guide 98‑3:2008] which characterizes the accuracy at a given confidence level.

Precision is determined by random error while the trueness reflects possible systematic deviations from the true value. In the case of precision, it is distinguished between the scattering under repeatability (ISO 5725‑1) conditions (short-term variations in one laboratory with one measuring instrument, one operator) from the scattering under reproducibility (ISO 5725‑1) conditions (between different laboratories, different measurement equipments, different operators, long-term variations).

NOTE For further information on accuracy, see Annex D.

Guidance for controls is given in Annex E.

The test report shall contain the following information:

a) identification of the test sample;

b) reference to this document (i.e. EN 14726);

c) test method used;

d) results;

e) unusual characteristics noted during the determination;

f) any operation that is not included in this European Standard or in the document to which reference is made or that is regarded as optional;

g) date of the test and/or date of the preparation and signing (if applicable electronically) of the test report;

1. 
   (informative)
   
   Representative sparking area

For the determination of the representative sparking area of a sample taken from a molten metal, several samples are investigated using wet-chemistry methods and spark optical emission spectrometry. Additionally, at least two samples are machined in layers (e.g. each 0,5 mm) and the results of the spark optical emission spectrometry for each level are plotted (depth profile) as a function of the sample depth (e.g. measured from the sample side facing the mould bottom). For setting the radial area, a V-shaped stop on the sample table is recommended. The average contents determined by wet-chemistry methods serve as nominal values, for the optimization of a sparking area.

1. 
   (informative)
   
   Detailed information on calibration
   1. Fundamental principles

The principle is to determine the relationship between the measured signals and the element contents by means of representative reference samples.

NOTE Since the optical spectral measuring signals depend on the structure and composition of the sample, their relation to the element contents is determined by calibration using reference materials which are related to the laboratory samples. In general, the intensity of an analyte spectral line is related to the intensity of one spectral line of the base metal of the sample (internal reference) and thus the intensity ratios of the sample can be evaluated further. The evaluation functions, which are needed for subsequent analyses, are calculated by means of simple regression calculations.

A long term drift of the spectrometer is superimposed on and usually exceeds the short-term measured value variations. The calibration process is subdivided into two or more parts depending on the calibration system used. The alloy group calibration has as basic calibration one or more sets of evaluation functions and a drift compensation (setting up), or recalibration procedure. If only one set of evaluation functions with appropriate interference corrections is suitable for all alloy groups of interest, this basic calibration is also called universal (global) calibration. Small differences between the calibration for an alloy group and the response of samples belonging to a deviating alloy type may be corrected by a type of recalibration. The master curve calibration has a single base calibration with a recalibration procedure which compensates for drift and alloy matrix effects for each alloy, either as a single procedure, or can use setting up samples and alloy type samples as separate procedures.

The instrument calibration – which is valid for several years - is carried out in either of two ways:

a) For the alloy group calibration and universal calibration: a series of reference samples corresponding to the alloy types to be analysed is required. For the determination of influence coefficients, where spectral (e.g. line interferences) and non-spectral interferences (inter-element influences) are taken into account with the evaluation functions, additional and special samples are frequently used, e.g. binary samples;

b) For the master curve calibration: a series of reference samples corresponding to the alloy types to be analysed is required. Additional interference correction samples are not usually required for the binary system but can be if alloy reference samples are used.

Drift compensation can be performed in the form of intensity corrections or corrections to the element contents, determined in a preceding calculation stage.

For the intensity correction of the instrument drift, a sample set containing the appropriate elements in a low mass fraction or a high mass fraction according to the defined measuring range, is used.

In accordance with the accuracy requirements, the correction of the instrument drift may be repeated after a few hours or after several days.

In an upper recalibration sample for multi-element procedures, as many elements as possible should be included to limit expenditure during recalibration.

The homogeneity of the recalibration samples or setting-up samples should be carefully evaluated.

Parallel displacements and/or rotation of the evaluation curves often occur.

• 1. Reference materials

A classification of the reference materials is given in Table B.1.

Table B.1 — Classification of reference materials

• 1. Base calibration
     1. Alloy group calibration and universal calibration

For this system of calibration of the spark optical emission spectrometer, similar alloy types are put together in one alloy group, each with its own set of evaluation functions which are laid down in an analytical program. The coefficients of the evaluation functions (usually polynomials) are calculated from the measured intensities or intensity ratios of the calibration samples (RM and CRM) and their corresponding specified element contents. Matrix corrections can be taken into account in the evaluation according to different models: additive correction, e.g. for the compensation of line interferences, and multiplicative correction, mainly for the compensation of non-spectral interferences. The corrections can be made using the intensities or using the contents of the interfering elements as influence quantities.

Calibration shall comprise the following steps (see also the equipment and software documents supplied by the manufacturer):

— selection of calibration samples having the required content range for each alloy group;

— selection of recalibration samples (setting-up samples) for drift correction;

— if necessary selection of special samples (e.g. binary samples, blank samples) for the determination of correction factors;

— definition of the measurement parameters and documentation (electrode gap, residual vacuum, high voltage of the photomultiplier tubes, internal settings of spark generator, etc.);

— measurement of the nominal intensities of the recalibration (setting up) samples;

— measurement of the intensities of the samples for the acquisition of function interferences (if additive intensity corrections are provided for), calculation of the correction coefficients and insertion in the analytical programs for the calibration samples;

— measurement (defined number of sparks per sample) recording and storage of the intensities of the calibration samples;

— testing the stability of instrument between the beginning and end of the calibration measurements using the recalibration (setting up) samples;

— calculation of the evaluation functions.

For optimization, different model functions (polynomial order, correction functions), computational variants (e.g. weighted regression) and the mode of association of alloy groups have often to be compared with one another. The standard deviation of the single values from the regression function (residual standard deviation, standard deviation of the method) and the standard deviation of the calculated coefficient are the criteria used for evaluating the quality of the line fit. Should an individual sample show particularly large deviations between the certified value and the value calculated using the evaluation function, the following shall be checked:

— scatter of individual measurements;

— sample with unconsidered interfering element;

— content at limit of calibration range;

— deviating sample structure;

— reliability of the certified value (trueness and confidence limits).

These investigations may lead to corrective actions or to the rejection of some results:

— definition of the range of validity of the evaluation functions (detection limit, quantification limit, upper limit);

— verification of the trueness with certified references materials;

— documentation and data record of the calibration.

NOTE The calibration of a spark optical emission spectrometer is often carried out by the instrument manufacturer.

• • 1. Master curve calibration

For this system of calibration of the spark optical emission spectrometer, a series of certified materials shall be selected covering the entire content range of each element to be analysed for all alloys. In the frame of a master curve system, each channel of spectrometer has its own evaluation function within a single master analytical programme. Binary samples are used. Spectral interference corrections shall be taken into account as for the alloy calibration system.

When binary samples are used, matrix effects do not occur; if non-binary samples are used, such effects should be taken into account, when evaluating the data.

Spectral interference corrections shall be considered.

Calibration shall include the following:

— selection of reference materials covering the entire content range for each element to be determined. No additional setting-up samples are used during binary calibration;

— definition of measurement parameters not included in the analytical programme;

— measurement of the intensity ratios of the samples using the defined number of sparks per sample, and recording these values;

— calculation of the evaluation functions and any correction coefficients by linear regression. The criteria for evaluating the quality of line fit are as for the alloy group calibration system;

— documentation and data record of the calibration;

NOTE This master analytical programme, and any correction coefficients, are sampled or duplicated to create all alloy analytical programmes to be used subsequently.

Low content samples may be common to many elements and can be used to check for drift during calibration with the master curve system.

Certified reference materials are used to check the trueness of the evaluation functions.

1. 
   (informative)
   
   Detailed information on recalibration
   1. Intensity drift correction

The intensities recorded for the setting up samples at the same time as the base calibration measurement, are used as reference data.

NOTE The same set of check samples can be used both for drift control and for statistical process control of the spectrometer.

Drift corrections are calculated according to the formula:

(C.1)

where

I_Ak is the (corrected) intensity of the laboratory sample traced back to the nominal state;

I_Aa is the actual measured intensity of the laboratory sample;

I_Tn is the nominal intensity of the lower recalibration sample measured during base calibration;

I_Hn is the nominal intensity of the upper recalibration sample measured during base calibration;

I_Tr is the intensity of the lower calibration sample at the last recalibration;

I_Hr is the intensity of the upper recalibration sample at the last recalibration.

NOTE Many computer programs indicate the intercept of the ordinate α and the slope β of the Formula (C.2), transformed to a linear function.

(C.2)

If a recalibration (setting-up) sample is nearly consumed, nominal intensities for a new setting-up sample, which show a slight deviation in composition, can be determined by means of statistically ascertained comparing measurements on the old and new setting-up samples (see manufacturer's software instructions).

After recalibration, the trueness shall be checked by measuring reference materials not used in the recalibration procedure.

• 1. Master curve recalibration

The recalibration will vary within this system depending on the software configuration and which type of certified reference materials was used in base calibration.

The binary calibration system requires alloy programmes created from the master analytical programme to use two or more certified reference materials to correct simultaneously for instrument drift and for alloy matrix effects. The high level sample is similar in alloying element content to the test sample. The low level sample has similar alloying element levels but near zero levels of trace elements, or a high purity aluminium certified reference material may also be used. The high purity aluminium sample (or a dilute version of either of the above samples containing a lower level alloying element,) is used to locate the low point of the calibration curve for the alloying elements. If additional setting-up samples are applied to this calibration system they shall be sparked initially with the recalibration samples to provide the relational data to allow their subsequent use as replacement recalibration samples.

1. 
   (informative)
   
   Detailed information on accuracy and uncertainty
   1. Introduction

The present annex details some guidance for the assessment of accuracy and uncertainty.

• 1. Possibility of error

For the assessment of precision, repeated measurements are carried out in order to include the influence of sampling in the case of samples taken from molten metal, since the representative area in the sample can shift due to random influences (e.g. cooling rate). If repeated measurements are carried out over long periods of time (quality control charts), by various instrument operators and samplers, using different auxiliary devices (moulds, cutting tools), the reproducibility precision is assessed.

NOTE Sampling error also includes a contribution arising from the inhomogeneity of the melt which belongs to the random process variations and not to the analytical errors.

The assessment of systematic deviations is of particular importance for achieving a reliable analysis since they, compared with precision, frequently cannot be neglected. The difficulty is to identify the causes.

Well-known causes of possible systematic deviations in spark spectrometric analysis are:

— differences in the preparation between the test samples and the certified reference materials;

— variation in structure between the test samples and the certified reference material;

— instrument drift which was not or insufficiently corrected;

— inadequate fitting of the evaluation function to the true calibration function of the spectrometer (e.g. linear evaluation function for, in reality, a curved functional relation);

— test samples and certified reference materials of very different composition (matrix);

— insufficient or neglected correction of interferences;

— sparked areas not lying in the representative area of the sample

To reveal, minimize or quantify possible systematic deviations, preliminary tests or periodic controls, which accompany the analysis, are essential. Where structural influences have to be minimized and when the average of a larger representative amount of the sample has to be covered, wet chemical analytical methods are recommended.

• 1. Measuring and calibration uncertainty

For determining the uncertainty of the results (with regard to precision and trueness), the contributions of several random and systematic influence factors shall be worked out separately, as far as possible.

The contribution made by calibration and drift correction to the uncertainty of an analytical result can only be calculated from the calibration measurements with difficulty, in particular when using nonlinear evaluation functions and taking matrix interferences into account. Therefore, to estimate the uncertainty, at least two certified reference materials are used, which matches the matrix and brackets the element content.

Besides the general requirements of the certified reference material, such as good homogeneity, the certified reference materials used for the monitoring of the analytical procedure should be as similar as possible in structure, grain size and chemical composition to the laboratory samples since the measured value of the analytes can also depend on the constitution and the other components of the sample.

In relation to a sparked area and without taking structural influences into account, the uncertainty of an analytical result can be stated in the form w_Element ± u:

(D.1)

where

u is the half uncertainty range of a single analysis in relation to the sparking area, if structural influences are negligible;

t is the significance limit of the t distribution (Student factor) in two-sided boundary, dependent on the degree of freedom and on the confidence level P (see ISO 3534‑1);

P is the statistical confidence level P = 1 – α. In case P is not given, P = 95 % is often used;

s_w is the standard deviation of n mean values from k sparkings at a time. The standard deviation s_w is ascertained from measurements in accordance with set analytical procedure on laboratory samples taken in parallel (number: n). The value of k arises from the prescribed method of averaging. The procedure to be used for the rejection of the first spark value shall also be prescribed (e.g. in case of memory effects). Should the analytical result with the related uncertainty merely apply to the supplied sample without taking the precision of the sampling into account, the sample surface area shall be subdivided n times with enough space for k sparkings each time or, if the sample is too small, another larger sample of the same homogeneity and composition shall be analysed several times;

s_m1 is the standard deviation of n mean values from each time k sparkings of reference sample 1;

s_m2 is the standard deviation of n mean values from each time k sparkings of reference sample 2;

u_r1 is the stated uncertainty (half range) in the certificate of the certified reference material 1.

u_r2 is the stated uncertainty (half range) in the certificate of certified reference material 2.

The scatterings s_m of the certified reference material can be obtained from regular measurements (control charts, etc.).

• 1. Investigation of accuracy

A fundamental examination of spark analytical accuracy refers to comparison measurements using alternative methods. This investigation should be carried out for all alloy types at least at the upper and lower end of each content range covered by the analytical program.

For this purpose, several laboratory samples — taken in a period as short as possible — are analysed by wet chemistry and spark optical emission spectrometry alternately. In both methods, the same (at least two) certified reference materials are analysed. The analyses of the certified reference materials confirm the agreement of both methods or form the basis for a correction prior to further evaluation.

The statistical evaluation gives the:

— repeatability of both methods for the reference materials (analysis number: n ≥ 5; for spark analytical precision s_w, see D.3);

— traceability to the certified reference materials used;

— systematic deviation exceeding precision between both methods in regard to the certified reference materials;

— repeatability of both methods on laboratory samples including sampling (for n ≥ 6);

— systematic deviation exceeding precision between both complete methods (including sampling, sample preparation, calibration, instrument drift). Since trueness was proved from the reference material (see above) and the wet-chemical method registered the average of the total laboratory sample, (this is to be established separately), the observed deviation is the systematic deviation of the spark optical emission spectrometry method.

The statistical investigation also frequently gives pointers as to the causes of the observed systematic errors. Optimization of the analytical procedures is to be strived for so that systematic error compared with random errors are negligible. Since systematic deviations, however, do not go in the same direction for all analytes and the measuring parameters mostly represent a compromise for several alloy types, either a verifiable correction should be made for individual analytes or the uncertainty limits should be increased accordingly.

1. 
   (informative)
   
   Guidance for controls
   1. General

All parameters and influence quantities co-determining the results and their uncertainty shall be checked periodically corresponding to the requirements and should be documented. Likewise, the requirements on the supplied materials (e.g. laboratory sample, reference material) and the services (e.g. calibration service, maintenance service) shall be reviewed.

• 1. Continuous control

Examinations which accompany analyses, in particular of the long-term drift of the spectrometers, should be carried out, represented (control charts) and documented by means of:

— the documentation and the assessment of the intensity measurements of the setting-up samples which gives information on the instrument stability without drift correction and about the extent of the linear correction;

— the drift control samples which indicate instrument stability before and after drift correction;

— the certified reference materials with matched matrix and structure which trace the actual calibration state back to the certified values.

— another laboratory sample analysed by wet-chemistry methods which allows a verification of the complete spark optical emission spectrometry method including the influences due to sampling.

• 1. Checking the analytical sample

The suitability of each sample and the fixed sparking area for spectral analysis should be established in preliminary tests or by means of periodic reference analyses of parallel samples using a procedure which includes a representative quantity of the whole sample and minimizes structural influences. For checking the precision of sampling which accompanies the process, the regular spark spectral analysis of parallel samples taken by a second sampler using a different mould is suitable. The differences between the analytical results of the parallel samples can be statistically evaluated and give the precision of the spark spectral analysis including sampling under reproducibility conditions.

• 1. Analytical capability

Taking the scattering of the total process (that means also possible systematic deviations of the analytical procedure) into account, an assessment for the process capability on a confidence level of P = 1 – α = 99,7 % can be given as process performance index [ISO 3534‑2]:

(E.1)

where

C_pk is the critical process capability index (ISO 3534‑2);

F is the tolerance range of the element contents (alloy specification);

d is the systematic deviation of the mean value from the target value; in d, the systematic deviation of the analytical results from the true value is taken into account;

s is the standard deviation of the complete process (production and analytical).

The total standard deviation is subdivided into production and analytical contributions:

(E.2)

where

s_p is the standard deviation of the production process;

s_a is the standard deviation of the analytical process.

The sampling error is normally attached to the process scattering (e.g. heterogeneity of the melt). Since sampling also has an effect on the position of the correct sparking area in the sample, its influence has also to be taken into account in the analytical error if the result shall refer to the mean content of the sample.

Making process control independent of the analytical influences demands that analytical precision is negligible with respect to the process scattering.

(E.3)

where

δ is the ratio of process scattering to precision of analysis.

Bibliography

[1] EN ISO 10012, Measurement management systems - Requirements for measurement processes and measuring equipment (ISO 10012)

[2] EN ISO/IEC 17025, General requirements for the competence of testing and calibration laboratories (ISO/IEC 17025)

[3] ISO 3534‑1, Statistics — Vocabulary and symbols — Part 1: General statistical terms and terms used in probability

[4] ISO 3534‑2, Statistics — Vocabulary and symbols — Part 2: Applied statistics

[5] ISO 5725 (all parts), Accuracy (trueness and precision) of measurement methods and results

[6] ISO 11095, Linear calibration using reference materials

[7] ISO 33401, Reference materials - Contents of certificates, labels and accompanying documentation

[8] ISO 33403, Reference materials - Requirements and recommendations for use

[9] ISO Guide 30, Reference materials — Selected terms and definitions

[10] ISO/IEC Guide 98‑3, Uncertainty of measurement — Part 3: Guide to the expression of uncertainty in measurement (GUM:1995)

[11] DIN 5030‑3, Spectral measurement of radiation — Part 3: Spectral isolation, definitions and characteristics

[12] DIN 32645, Chemical analysis — Decision limit, detection limit and determination limit — Estimation in case of repeatability — Terms, methods, evaluation

[13] DIN 51008‑1, Optical atomic emission spectral analysis (OES) — Part 1: Terms for systems with sparks and low-pressure discharges

[14] DIN 51009, Optical atomic spectral analysis — General principles and definitions

[15] Bertil Magnusson, Teemu Näykki, Håvard Hovind, Mikael Krysell, Eskil Sahlin, Handbook for calculation of measurement uncertainty in environmental laboratories (NT TR 537 – Edition 4)

[16] Müller A., Staats G., Tröbs V. Dynamic calibration and automated instrumental analysis. Fresenius J. Anal. Chem. 1994, 348 pp. 615–625

[17] Lührs C.H., Kudermann G. Spark Spectrometry — Guidelines for the chemical analysis of metals by optical emission spectral analysis with spark excitation, Ed. Chemists committee of the GDMB, Clausthal-Zellerfeld, 1996

[18] Slickers K. Automatic Atomic-Emission-Spectroscopy, Brühlsche Universitätsdruckerei, Sec. Edition, Gießen, 1993

[19] ASTM E716, Standard practices for sampling aluminium and aluminium alloys for spectrometrical analysis

[20] ASTM E1251, Standard Test Method for Analysis of Aluminum and Aluminum Alloys by Spark Atomic Emission Spectrometry