SIST CR 10321:2003
(Main)Chemical analysis of ferrous materials - Recommendations for the drafting of standard methods of analysis employing flame atomic absorption spectrometry for the chemical analysis of iron and steel
Chemical analysis of ferrous materials - Recommendations for the drafting of standard methods of analysis employing flame atomic absorption spectrometry for the chemical analysis of iron and steel
Revision of ECISS Information Circular No 8.
Chemische Analyse von Eisenwerkstoffen - Empfehlungen für die Entwickling von Standard-Analyseverfahren für die chemische Analyse von Eisen und Stahl unter der Anwendung der Flammen-Atomabsorptionsspektrometrie
Analyse chimique des matériaux sidérurgiques - Recommendations pour la rédaction de méthodes d'analyse normalisées employant la spectrométrie d'absorption atomique dans la flamme pour l'analyse chimique des fontes et des aciers
Kemična analiza železovih materialov – Priporočila za razvoj standardnih analiznih metod za kemične analize železa in jekla z uporabo spektrometrije s plamensko atomsko absorpcijo
General Information
Standards Content (Sample)
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.Chemische Analyse von Eisenwerkstoffen - Empfehlungen für die Entwickling von Standard-Analyseverfahren für die chemische Analyse von Eisen und Stahl unter der Anwendung der Flammen-AtomabsorptionsspektrometrieAnalyse chimique des matériaux sidérurgiques - Recommendations pour la rédaction de méthodes d'analyse normalisées employant la spectrométrie d'absorption atomique dans la flamme pour l'analyse chimique des fontes et des aciersChemical analysis of ferrous materials - Recommendations for the drafting of standard methods of analysis employing flame atomic absorption spectrometry for the chemical analysis of iron and steel77.040.30Kemijska analiza kovinChemical analysis of metalsICS:Ta slovenski standard je istoveten z:CR 10321:2003SIST CR 10321:2003en01-november-2003SIST CR 10321:2003SLOVENSKI
STANDARD
SIST CR 10321:2003
CEN REPORTRAPPORT CENCEN BERICHTCR 10321March 2003ICSEnglish versionChemical analysis of ferrous materials - Recommendations forthe drafting of standard methods of analysis employing flameatomic absorption spectrometry for the chemical analysis of ironand steelAnalyse chimique des matériaux sidérurgiques -Recommendations pour la rédaction de méthodesd'analyse normalisées employant la spectrométried'absorption atomique dans la flamme pour l'analysechimique des fontes et des aciersChemische Analyse von Eisenwerkstoffen - Empfehlungenfür die Entwickling von Standard-Analyseverfahren für diechemische Analyse von Eisen und Stahl unter derAnwendung der Flammen-AtomabsorptionsspektrometrieThis CEN Report was approved by CEN on 3 October 2001. It has been drawn up by the Technical Committee ECISS/TC 20.CEN members are the national standards bodies of Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece,Hungary, Iceland, Ireland, Italy, Luxembourg, Malta, Netherlands, Norway, Portugal, Slovakia, Spain, Sweden, Switzerland and UnitedKingdom.EUROPEAN COMMITTEE FOR STANDARDIZATIONCOMITÉ EUROPÉEN DE NORMALISATIONEUROPÄISCHES KOMITEE FÜR NORMUNGManagement Centre: rue de Stassart, 36
B-1050 Brussels© 2003 CENAll rights of exploitation in any form and by any means reservedworldwide for CEN national Members.Ref. No. CR 10321:2003 ESIST CR 10321:2003
CR 10321:2003 (E)2ContentsForeword.31Introduction.32Instrumental criteria.43Adjustment of atomic absorption spectrometer.64Preparation of solutions.75Calibration procedure.86Interferences.97Background correction.98Expression of results.99Safety.1010Miscellaneous.1011Summary of recommendations.1012Literature.11Annex A (informative)
Chemical analysis of ferrous material - Determination of calcium in steels -Flame atomic absorption spectrometric method.12Annex B (informative) Precision data.18SIST CR 10321:2003
CR 10321:2003 (E)3ForewordThis document (CR 10321:2003) has been prepared by Technical Committee CEN/TC 20 "Methods of chemicalanalysis of ferrous products", the secretariat of which is held by SIS.Standard methods of chemical analysis have traditionally been based on the techniques of classical chemistry.Generally, by specification of quantities and purity of reagents, such methods may be described completely andthereafter followed exactly in any laboratory.Today, however, the bulk of analysis is done by instrumental methods for which different laboratories may useinstruments from various manufacturers with different configurations and operational settings. It becomes ofpressing importance to consider how such methods may be specified in standard documents with regard to bothinstrument quality and procedure.This CEN Report considers these matters for methods employing flame atomic absorption spectrometry and makesrecommendations. These prescribe instrument quality in terms of limit of detection, curve linearity and precision.Recommendations concerning the preparation of solutions, the calibration procedure and the expression of resultsare also included. Methods for the determination of calcium in steel and for cobalt in iron steel are included asillustrative examples of the drafting recommendations.NOTE - Attention is also drawn to CR 10322, Chemical analysis of ferrous materials - Operational guidelines for the applicationof flame atomic absorption spectrometry in standard methods for the chemical analysis of iron and steel.1 IntroductionAn acceptable analytical procedure has to be written in such a form and in such detail that after faithful executionby a well trained analyst an unquestionable result will be obtained. Neither the individual analyst nor the make ofapparatus used should have any influence on the final result. These requirements are particularly stringent in theformulation of standard methods used for reference purposes and for arbitration in cases of dispute.Traditionally, standard methods of analysis have been based on the techniques of classical chemistry. Whiletechnically sound, they do not reflect modern developments such as optical emission spectroscopy, X-Rayfluorescence and atomic absorption spectrometry, now in widespread use. New and revised standard proceduresnow in the course of preparation will increasingly embody these techniques and it becomes pressingly important toconsider how such methods may be specified.With traditional chemical methods there was little problem. Generally, by specification of quantities and purities ofreagents, the methods could be described completely and thereafter followed exactly in any laboratory. Withinstrumental methods, however, the situation is different. Different makes of instruments may be employed in thevarious laboratories, with different configurations or operational settings, and it becomes more difficult to specifythe procedure precisely. Clearly, the standard should not refer to a particular make of instrument, to the impliedexclusion of other satisfactory equipment, nor should it include operational settings or measurement times orsolution concentrations which may be appropriate to a given instrument only.On the other hand, if matters are left in the hands of the individual analyst just to ’follow the manufacturer’sinstructions’ the document can no longer be regarded as a standard nor is there any guarantee that the instrumentemployed will have a quality of performance suitable for the purpose.The way forward is for the standard document to specify instrument suitability in terms of fundamental performancecriteria such as precision, limit of detection and linearity, according to agreed definitions. Individual manufacturerswould then be free to meet these requirements in any way they choose. Also, since even a good instrument maydeteriorate, the standard should contain reference to practical procedures for the determination of these criteria, toensure that the instrument meets the specification and continues to do so during its functional years. OtherSIST CR 10321:2003
CR 10321:2003 (E)4essential aspects of the standard include instrument optimization, preparation of the test sample, calibrationprocedures, calculation of results and, where appropriate, how these should be related to the actual instrumentused.This CEN Report considers these matters in relation to standard methods of analysis employing flame atomicabsorption spectrometry and makes recommendations for the drafting of such methods. Non-flame atomicabsorption spectrometric methods involving instrumental parameters whose definitions and descriptions requirefurther study are not included in this document. The determination of calcium in steel for which performance dataare available is included as an illustrative example of the drafting recommendations. A method for thedetermination of cobalt in steel covering a wide concentration range has also been included in order to illustratedilution procedures, optimization for minimum matrix interference and correction procedures where cobalt may bepresent in the iron used for the calibration solutions.2 Instrumental criteriaIn the past, attempts have been made to specify instrument performance in a variety of different ways under suchheadings as minimum sensitivity, curve linearity and minimum stability. It is important that terms used in standarddocuments should be in accordance with agreed definitions. The term ’minimum sensitivity’, for example, has beenused to imply a stipulation that the absorbance of the most concentrated calibration solution measured in a flamepath length of 10 cm must be at least 0,3. Useful as this requirement may be as a practical guide, it is not theaccepted definition of sensitivity in instrumental analysis which is, more properly, the change in instrument readingfor unit change in concentration (dx/dc), or the slope of the analytical curve at a given concentration. In atomicabsorption spectroscopy, the term ’sensitivity’ ha also been used very widely in manufacturers’ literature andelsewhere to imply the concentration of the analyte which will produce a change in the absorbance reading,compared with that of pure solvent, of 0,0044, i.e. 1 % net absorption. The IUPAC Commission has drawn attentionto this misuse of the term sensitivity and while recognizing the usefulness of the value it represents, has suggestedreplacement of the name by the term characteristic concentration.This document recommends that instrument suitability as specified in a given standard document should be basedon three performance criteria, according to agreed definitions – limit of detection, acceptable curvature of thecalibration graph and precision.2.1 Limit of detectionThe limit of detection is smallest concentration in solution of the element of interest which may be detected withconfidence. It is obtained from the calibration graph and is generally taken in instrumental analysis as that value onthe concentration axis which corresponds to an instrument reading of two or three times the standard deviationabove the mean reading of the blank (analytical background).The IUPAC proposal recommends that this background measurement should be derived from a sufficient numberof replicate determinations on the blank solution. Purely to facilitate the practical measurements, the present reportsuggests a minor modification, i.e. that the measurement should be based on a concentration value selected togive an absorbance of just above the zero and that the standard deviation should be calculated on the basis of 10replicate determinations. In taking these measurements, as with all other measurements, it is very important thatthe instrument should be fully optimized. It is not sufficient that it should be adjusted merely to meet the detectionlimit quoted in the standard. Instrument optimization is considered in some detail in CR 10322.It is not unusual for the experimentally derived limit of detection to be a few times higher than that quoted by themanufacturer because the latter will have been derived, in all probability, from a new instrument under particularlycarefully controlled conditions using pure solutions. If the discrepancy is large, however, the instrument should beoverhauled.The value placed on this detection limit in the standard should be specified with reference to the lowest concen-tration of the element of interest likely to be encountered in the application envisaged. In the ideal case, thespecified limit of detection should be less than one-tenth of the lowest concentration level to be determined andshould be measured in the same matrix. The analysis of drinking water which is subject to legal requirementsprovides a convenient example of this doctrine. If, for example, the maximum legally permitted concentration oflead in potable water were, say, 0,05 ppm Pb (mg/l) and a result very close to this were obtained, it would benecessary to know what confidence could be placed on the result. It would be little comfort to the analyst to feelSIST CR 10321:2003
CR 10321:2003 (E)5that his instrument was working at the very limit of its capabilities. Since the limit of detection of many flame atomicabsorption spectrometers is only of the order of 0,01 to 0,02 mg/l Pb, special techniques would be required to givethe necessary confidence when operating near the statutory limit. In the case of manganese in potable water, onthe other hand, for which the maximum permitted concentration is of the same order, the flame atomic absorptionspectrometric technique is very sensitive (limit of detection typically 0,005 mg/l or better), thus meeting the criterionmentioned above and being eminently suitable for direct application.It is of course recognized that there may be situations where flame atomic absorption spectrometry might still bethe method of choice even though the above requirements could not be met. Such situations might exist where thetechnology does not require residual element determination with a high degree of precision, and a much higherlimit of detection relative to the concentration being determined would be acceptable. In this sense, each analyticalrequirement could be considered on its own merits. Nevertheless, for a standard referee method, as distinct from aroutine procedure, specification of a limit of detection of less than one-tenth of the lowest concentration to bedetermined is a useful guideline.The exception would be if, despite not meeting this requirement, the method were still the most accurate oneavailable.2.2 Curvature of the calibration graphIt is widely accepted in instrumental analysis that the ideal analytical curve is one which is linear throughout theconcentration range of interest. Not only does this situation imply that the measurement system is showing itshighest sensitivity but the graph is easy to locate accurately either by a line drawn manually through all the pointsor by a simple linear ’least squares’ mathematical treatment. Most practising analysts feel instinctively that bestresults are obtained under these conditions.In practice, however, non-linear calibration graphs are commonplace in atomic absorption spectrometry, eventhough the inherent curvature may in some cases be concealed by electronic manipulation of the signal beforedisplay, i.e. with inherently slightly curved calibration graphs there is the risk that a linear graph is forced throughthe calibration points which clearly introduces errors. Potential risks associated with a heavily curved region ofcalibration are that absorbance measurements are insensitive and there may be problems in defining the shape ofthe curve on the basis of relatively few points in the curved region. It is therefore necessary to consider whatconstraints should be written into the standard to safeguard analytical accuracy under the conditions on non-linearcalibrations graphs.It has been proposed that the slope of the calibration curve covering the top 20 % concentration range (expressedas a change in absorbance) shall not be less than 0,7 of the value of the slope for the bottom 20 % concentrationrange determined in the same way. It is recognized that this criterion is not a fundamental requirement in the samesense as an adequate precision. It is quite possible that an instrument just failing to comply with the maximumpermitted curvature requirement would give a satisfactory performance if the precision were adequate. Indeed,precision, being a function of sensitivity, curvature, and signal to noise ratio, and the basis on which limit ofdetection is calculated, is a fundamental performance criterion which, in effect, renders the others redundant.Ringbom plots have shown that acceptable precision can be obtained at absorption values considerably greaterthan those commonly recommended. Detailed study of such plots enable an analyst to decide on a working rangefor his instrument which will be appropriate to the precision requirements of his particular application.However, bearing in mind that accuracy presupposes not only an adequate precision but also freedom from biasand likewise bearing in mind possible difficulties in locating exactly the position of a curved graph, i.e. wheredeparture from linearity becomes significant and how the ensuring shape may be accurately described manually ormathematically in the absence of an inordinately large number of calibration solutions, it is recommended that theabove requirement defining the maximum permitted curvature be incorporated in the standard as a guideline.It is also recommended that information on curvature of the calibration graph should be available for inspection at apoint before data processing takes place. While modern developments are welcomed, it is important to ensure thatcurve straighteners and other electronic devices should not be operating on poor quality input information such asa near horizontal part of the curve, with possibly misleading results of the less experienced analyst.Recent work has shown that if curvature is due to stray light, linearity requirements will be met if the mid-pointcalibration absorbance does not exceed 0,55 times full scale absorbance. This provides a particularly convenientand practical way of judging an acceptable linearity.SIST CR 10321:2003
CR 10321:2003 (E)62.3 PrecisionAs mentioned in previous clauses, precision is a performance criterion of fundamental significance. A value ofprecision stated in the standard provides a requirement which proposed instrumentation must meet if it is to beconsidered suitable for operation of the standard. It provides an assurance against inadvertent use of poor qualityequipment.In some methods, the acceptable precision of measurement at the two ends of the calibration curve has beendefined as a stated proportion of the standard deviation of the absorbance of the most concentrated calibrationsolution.Typical wording is ’the standard deviation of the absorbance of the most concentrated calibration solution, and thestandard deviation of the absorbance of the zero calibration solution, each being calculated from a sufficientnumber of repetitive measurements, shall be less than 1,5 % and 0,5 % respectively, of the mean value of theabsorbance of the most concentrated calibration solution’.Although perhaps somewhat obscure at first sight, the recommendation has the advantage of relating the requiredprecision to the concentration being determined. Thus assuming for a moment a linear calibration, the precisioncalled for over the concentration range 0,0 % to 1 % of a given element in the above example would be 0,005 %and 0,015 % respectively or, at the 95 % confidence level (2s), 0,01 and 0,03 % respectively. The introduction ofprecision limits based on absorbance measurements in this way also has the advantage of eliminating poor qualityequipment at an early stage. The reproducibility of the entire method in terms of concentration units will, however,be affected by other factors, for example curvature of the calibration graph.The exact values placed on these relative precisions (0,5 % and 1,5 % in the example given above) may well varyfrom element to element or from standard document to standard document. These values must take account of thepurpose of the analysis, i.e. the technical or commercial implications, but will be limited ultimately by the capabilitiesof the instrument.Obviously, specified instrumental precisions should not be so generous that the method is incapable of yielding therequisite technical information but it would be equally pointless to set limits in excess of requirements which noinstrument is capable of meeting. A comparison of these values with what is available by alternative chemicalmeans will be an important factor in assessing the acceptability of the AAS procedures for standard refereepurposes. This point may well be illustrated by reference to some earlier work concerned with the determination oflow levels of titanium in steel. It was found that with the instruments at that time available, the precision attainableat low concentrations of titanium was so inferior to the performance of an alternative spectrophotometric procedureas to rule out flame atomic absorption spectrometry for referee purposes, in this particular case.Two modifications to the above precision requirements may be recommended. Firstly, the number of replicatemeasurements on which the precision is based should be specified. For consistency with the proposed definition oflower limit of detection, 10 replicate determinations are recommended. Secondly, it is suggested that the secondprecision measurement should be taken on the lowest calibration standard, instead of on the zero solution. In thisway, distortion of the measurement by instrument noise may be minimized. The calibration procedurerecommended in clause 5, taken in conjunction with the above precision requirements provides a warning ofinstrument drift.3 Adjustment of atomic absorption spectrometerIt is assumed that the standard will specify the type of hollow cathode lamp, the wavelength of measurement, andthe type of flame, e.g. air-acetylene or nitrous oxide-acetylene, with appropriate burner. Problems with gas purityhave been reported. Impure supplies of acetylene have been encountered in which oily drops formed on the burnerand the flame was coloured to the extent that it was impossible to see the ’red feather’ of the nitrous oxide-acetylene flame. With a satisfactory supply of gas, the air-acetylene mixture will burn with a clear blue transparentflame. It has not so far been found possible to quantify the acetylene purity in percentage terms but in the UK thestipulation ’phosphine-free’ has been found to result in a satisfactory gas. Cylinder acetylene is dissolved inacetone. Manufacturers have warned that other solvents may damage plastic components of some instrumentsand constitute a safety risk. Manufacturers’ instructions must be closely followed. Air quality has also been foundvariable, ranging from an excessive oxygen content to a gas which would not support combustion. The standardshould therefore, include a gas quality specification.SIST CR 10321:2003
CR 10321:2003 (E)7Optimization of instrument settings is of vital importance but is one of the more difficult aspects of the method tospecify in the standard document. Lamp current, alignment of lamp and burner, warming up time and adjustment ofthe nebulizer should be covered by the manufacturer’s instructions. Burner height and flame composition, however,may be adjusted in various ways. Commonly, the fuel flow and burner position are adjusted for maximumabsorbance while aspirating the calibration solution of highest element content. This is a widely used technique butis not favoured as a general procedure because the increase in absorbance may be due to the flame itself.A preferred method is to make the adjustments so that the difference in absorbance between the highest calibra-tion solution and the blank is at a maximum. When setting for the determination of elements very near theirdetection limits, however, a procedure which gives a minimum absorbance while water is being aspirated may beadopted, as it simultaneously ensures that the noise level is at a minimum. An alternative optimization proceduremay be adopted to minimize severe matrix interference, if this is expected. Optimization of the instrument isconsidered further in CR 10322.4 Preparation of solutionsMethods of chemical analysis traditionally recommend that calibration solutions should be prepared to a series offixed dilutions. However, there is a danger with this approach that a given instrument may not be being used underoptimum conditions. It is possible to envisage, for example, that with the stated concentrations, a highly sensitiveinstrument could give a reading beyond the acceptable upper limit of curvature or even off scale. A differentapproach is therefore necessary and the optimum concentration range must be determined experimentallybecause it will depend upon the characteristics of the instrument used.It has been stated in the literature that the optimum working range of an atomic absorption spectrometer is from 20to 125 times the characteristic concentration.1 Assuming a linear calibration, this corresponds to an absorbance of0,1 to 0,55. This may be used as a basis for the preparation of sample and calibration solutions as follows:(i) Determine the characteristic concentration1 of the element of interest for the given instrument.(ii) Prepare an initial sample solution, ideally to have not more than 1 % salts concentration (in order to avoidproblems with burner blockage, although higher concentrations may be permissible with some instruments).From this point there will, in general be four possibilities:(i) The solution is too concentrated to be used directly. In this case proceed as follows:Test the solution at various approximate dilutions to find the solution which falls near the mid-way of theoptimum working range – say – 0,25 to 0,45 absorbance.(a) Accurately dilute the sample solution by this factor.(b) Prepare a series of five or more calibration solutions containing the principal matrix elements in the sameconcentration as the diluted sample solution with element concentrations to cover the required range, forexample, 0 to 140 times the characteristic concentration value (i.e. 0 to 0,6 absorbance).The above procedure is intended to give optimum performance for referee analysis as befits a standardmethod.In routine batch operation, it is recognized that it would be more convenient to make additions to thediluted sample solutions in order to match a predetermined matrix used in preparation of the calibrationsolutions.(ii) The solution is already within the working range of the instrument. In this case dilution is obviously notrequired and the calibration solutions should be prepared directly as in paragraph (i).
1 N.B. Characteristic concentration, formerly termed ’sensitivity’ in atomic absorption spectrometry.See clause 2.SIST CR 10321:2003
CR 10321:2003 (E)8(iii) The solution is somewhat below the optimum working range of the instrument but cannot be concentratedwithout causing burner blockage problems or introducing unacceptable separation stages. In this case, anapproximate estimation of the element concentration should be made first. Calibration solutions should thenbe prepared having a similar matrix but with element concentrations of 0, 0,5, 1,0, 1,5, and 2 times the firstestimate, or some other suitable multiples.(iv) The element concentration is very low indeed, near the detection limit, and cannot be concentrated. If theprocedure in paragraph (iii) were followed, the concentration of the highest calibration solution may not besufficient to give adequate precision. In this case, a range of calibration solutions should be prepared suchthat the most concentrated will contain sufficient of the element to give full scale deflection when using themaximum scale expansion which is compatible with an acceptable noise level.5 Calibration procedureA recommended procedure is to aspirate the calibration solutions in ascending order repetitively until each givesthe specified precision, thus showing that the instrument has attained stability. Two calibration solutions are thenselected, one of absorbance just lower than the sample and one just higher. These are aspirated first in ascendingand then in descending order of concentration, with the sample as the middle solution in each case. The completerange of calibration solutions is then aspirated again.A widespread practice is to obtain the net absorbance of each calibration solution by subtraction of the meanabsorbance of the zero calibration solution. At a later stage, a similar procedure involving subtraction of absor-bances is followed for the blank and test solutions. When calibration graphs are curved, however, this procedurecan introduce significant error. Strictly speaking, absorbances should be converted to concentration values beforesubtraction, both in the case of the calibration and the
...
Questions, Comments and Discussion
Ask us and Technical Secretary will try to provide an answer. You can facilitate discussion about the standard in here.