SIST CR 10322:2003

2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.MHNODChemische Analyse von Eisenwerkstoffen - Richtlinie für die Anwendung der Falmmen-Atomabsorptionsspektrometrie bei Standardverfahren der chemischen Analyse von Eisen und StahlAnalyse chimique des matériaux sidérurgiques - Recommandations 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 - Operational guidelines for the application of flame atomic absorption spectrometry in standard methods for the chemical analysis of iron and steel77.040.30Kemijska analiza kovinChemical analysis of metalsICS:Ta slovenski standard je istoveten z:CR 10322:2003SIST CR 10322:2003en01-november-2003SIST CR 10322:2003SLOVENSKI
STANDARD



SIST CR 10322:2003



CEN REPORTRAPPORT CENCEN BERICHTCR 10322March 2003ICSEnglish versionChemical analysis of ferrous materials - Operational guidelinesfor the application of flame atomic absorption spectrometry instandard methods for the chemical analysis of iron and steelAnalyse chimique des matériaux sidérurgiques -Recommandations 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 - Richtlinie fürdie Anwendung der Falmmen-Atomabsorptionsspektrometrie bei Standardverfahren derchemischen Analyse von Eisen und StahlThis 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 10322:2003 ESIST CR 10322:2003



CR 10322:2003 (E)2ContentsForeword.31Introduction.42Equipment.43Systematic and random errors.74Setting and checking of the atomic absorption spectrometer.95Literature.11Annex A (informative) Instrumental settings and optimization of the atomic absorption spectrometerfor application in accordance with the standard method for analysis of iron and steel.12Annex B (informative) Reporting the measurement results.15Annex C
(informative) Safety.16Annex D (informative) Directions for the determination of some instrument parameters.18SIST CR 10322:2003



CR 10322:2003 (E)3ForewordThis document (CR 10322:2003) has been prepared by Technical Committee CEN/TC 20 "Methods of chemicalanalysis of ferrous products", the secretariat of which is held by SIS.Flame atomic absorption spectrometry is playing an increasingly important role in the analysis of iron and steel.However, the quality of results attainable, including the sensitivity, the detection limit and the precision depends notonly on the instrument used but also, and markedly, on its optimization. This CEN Report considers the essentialfeatures of an atomic absorption system, identifies possible sources of error and gives verification procedures toensure optimum performance. The Report is intended to be the work of reference during the performance ofindividual standard methods.NOTE - Attention is also drawn to CR 10321, Chemical analysis of ferrous materials - Recommendations for the drafting ofstandard methods of analysis employing flame atomic absorption spectrometry for the chemical analysis of iron and steel.SIST CR 10322:2003



CR 10322:2003 (E)41 IntroductionThe introduction of flame atomic absorption spectrometry into the analysis of iron and steel has greatly facilitatedthe determination of many of the elements. Indeed, some determinations which were scarcely practicable by theolder methods and subject to serious inaccuracies may now be determined with relative ease. Understandably,atomic absorption spectrometry is being increasingly adopted as the preferred procedure for many future standardmethods of analysis.As in the case of molecular absorption spectrophotometry, atomic absorption spectrometry is a relative method ofanalysis which depends on a comparison of absorbance values with those obtained from a series of calibrationsolutions. Whereas in spectrophotometric methods, absorbance values are relatively independent of the instrumentused and the settings are not very critical. In atomic absorption spectrometry absorbances are influenced by manyfactors. Not only the slope of the calibration graph but also its curvature, the random and systematic errors, thedetection limit and the occurrences of interferences are dependent on the instrument and on its many variables.This means that sample and calibration solutions have to be measured under identical conditions and that muchcare has to be exercised in the correct setting and regular checking of the instrument. The successful application ofatomic absorption spectrometry in standard methods therefore depends on defining the operational parametersand achieving them in practice.The differences between available instruments implies that specific operational settings may vary from one toanother. In order to meat the precision requirements of the standard methods, however, the equipment will have tosatisfy certain performance criteria. These criteria and the appropriate adjustable variables of the equipment arediscussed in the ensuring sections of this document. Parts of the spectrometer are also discussed where it is feltthat unfamiliarity with certain aspects may give rise to erroneous results. For general aspects of atomic absorptionspectrometry, reference is made to ISO 6955:1982, ISO/DIS 6956:1981, and to the literature mentioned in section5.ECISS/TC 20 is of the opinion that operational parameters cannot as yet be sufficiently well defined and controlledfor flameless atomic absorption spectrometry to warrant its use in standard methods. This Report is thereforeconfined to consideration of the flame technique.2 Equipment2.1 The radiation sourceFor most elements, hollow cathode lamps are available as radiation sources of sufficient intensity, stability and life.In general, separate single element lamps are preferable and the manufacturer’s literature often includes arecommended lamp current. When multi-element lamps are employed, a higher current may be required to ensuresufficient intensity of the resonance line and a sufficiently low detection limit.The linearity and sometimes the slope of the calibration graph usually increase if a slightly lower lamp current ischosen, especially for volatile elements such as cadmium, lead and zinc. This has the added advantage ofextending the life of the lamp. When the emphasis is on the attainment of low detection limits, a relatively high lampcurrent, within the specified maximum, will have to be chosen.The quality of the hollow cathode lamp should be regularly checked by scanning the emission spectrum in thevicinity of the wavelength used under the normal conditions of lamp current and slit width. The background signalon each side of the line should be less than 2 % of the maximum intensity. The intensity is considered zero whenthe light path is blocked.Some of the more volatile elements are less suitable for hollow cathode lamps and a better intensity is obtained bythe use of electrodeless discharge lamps. These, however, require a special high frequency power supply.After the manufacturer’s prescribed warning up time, the signal from each radiation source should not deviate bymore than 0,5 % from the maximum value (equivalent to 0,002 absorbance units) over a period of 15 minutes.Many different devices are available to correct for any non-specific absorbances (background correction).SIST CR 10322:2003



CR 10322:2003 (E)52.2 Atomization of the sample2.2.1 NebulizationThe quantity of the element reaching the flame per unit of time. And thus the absorbance measured, depends onthe rate of aspiration and the fraction of the nebulized sample which is carried forward into the flame.Consequently, all the physical properties involved, such as density, vapour pressure, surface tension, viscosity(and thus also temperature) should be identical as far as possible for the calibration and analyte solutions. This willbe taken into account in the drafting of the standard method. It must also be realized by the analyst should he beforced to make minor adjustments in the execution of the method.Most nebulizers have an adjustment device to optimize the setting. The optimum setting depends on the solvent(water, alcohol etc.) so that re-adjustment is required for each new solvent or mixture, in order to obtain themaximum signal to noise ratio.The main effect of the nebulizer setting is on the fraction of the nebulized sample reaching the flame. With sometypes the fuel gas/oxidant ratio will also be influenced. For a given solvent, it is advisable to adjust the nebulizerwith the aid of an element which is not sensitive to the gas ratio (such as silver, magnesium, copper) and to do soonly with an air-acetylene flame. The manufacturer’s instructions, however, should always be followed.Corrosion and wear of the nebulizer may lead to a decrease in efficiency and increasing instability, as well as adeterioration in the characteristic concentration (4.1), the limit of detection (4.2) and the relative standard deviation(3.2).2.2.2 Atomization with a flameOn reaching the flame, part of the fine sample mist, the aerosol, is converted into atoms. The extent of the atomi-zation which differs for each element depends on the temperature, the oxidizing or reducing properties of the flameand the measuring height above the burner head. These factors may also influence the effect which othercomponents in the solution have on the atomization – the matrix effect.The gas mixture consists of an oxidant (air or nitrous oxide) for burning the fuel gas which is usually acetylene. Thetemperature and the nature of the flame is determined by the ratio of the two gases. While the components of thegas mixture can be specified in a standard method, the composition and the measuring height can only beindicated and the optimum conditions have to be determined. After optimizing the equipment, all solutions inclu-ding the calibration solutions, the test solution and the blank should be measured in one series, without inter-ruptions or intermediate adjustment of the flame.Safety regulations (see also Annex C) must be strictly followed. Nitrous oxide/acetylene flames require specialcare.The nitrous oxide/acetylene flame has a much higher burning velocity than the air/acetylene flame which meansthat either the overall gas rate has to be increased or a special burner with a slot length of about 5 cm has to beused to avoid flashback. This ’nitrous oxide’ burner may also be used for air/acetylene but gives a lower absorp-tion than the usual 10 cm long air/acetylene burner.Use of a too reducing nitrous oxide/acetylene flame may cause the formation of soot-like encrustation on theburner slot and undesired changes in the absorbance measurements. Similarly, when solutions with high saltconcentrations are aspirated, various types of burner head may become gradually blocked due to deposition ofsalts in the slot. This results in undesired changes in measurements, and may be cause of flash back of the flame,especially when using solutions containing perchloric acid or its compounds, Such deposits may be minimized byfrequent aspiration of pure solvent between measurements.After the ignition of the air/acetylene flame and also after switching to the nitrous oxide/acetylene flame, sufficienttime must be allowed for the burner head temperature to stabilize before measurements are started.The gases are subject to the following requirements:(i) the acetylene must be sufficiently pure to burn with a clear blue transparent flame. In some countries a’phosphine-free’ grade is sold which meets these requirements. The pressure in the acetylene cylinder shouldSIST CR 10322:2003



CR 10322:2003 (E)6not be allowed to fall below 600 kPa (approximately 6 bar) or as otherwise recommended by themanufacturers in order to avoid risk of contamination with acetone;(ii) the air supply should be compressed air of high purity;(iii) Nitrous oxide cylinders should be fitted with an appliance to prevent excessive cooling and freezing of theregulator. Failure to do this could result in a decrease in gas flow, which in turn may lead to erroneous resultsor even a flash back.2.3 The monochromatorTwo adjustments are available – the slit width and the wavelength. For a correct setting of the instrument, the slitwidth should be set first and then the wavelength setting optimized until maximum light intensity is obtained. Over aperiod of time, a rise in temperature of the monochromator, as a result of flame radiation, may cause an alterationin the wavelength setting. This setting should therefore be checked and reset just before measurements arestarted. Suitable wavelengths are normally given in the manufacturer’s manual and will be further specified in thestandard method.The slit width determines the spectral bandwidth (the section of the spectrum transmitted by the monochromator)as well as the quantity of radiation transmitted, and thereby the detection limit. It is therefore advantageous toselect the largest possible slit width. However, the permissible slit width is limited by the presence of non-absorbingemission lines in the spectrum of the radiation source and possibly by interfering emissions from the flame,especially the nitrous oxide/acetylene flame.The spectral bandwidths (or slit widths) recommended in the manufacturer’s manual are usually based on theassumption that single element lamps are used. In the case of multi-element lamps, the resonance lines with thehighest sensitivity are free from overlap from other lines in the more complex spectrum but this is not always thecase for alternative lines. For this reason, it is necessary to use only the resonance lines of highest sensitivity or toscan the emission spectrum in the region of the relevant lines with a narrow slit and, if necessary, adopt a narrowerslit width than prescribed.2.4 The detector, the signal processing system and the displayThe radiation transmitted by the monochromator is converted by the detector into an electrical signal which isproportional to the intensity of the incident radiation. In the signal processing system, the output from the detector istransformed into a value T for the transmission (i.e. the ratio between transmitted and incident intensity) or into theabsorbance A, where A = log T.With the advent of microprocessors, there are now many further possibilities for processing the signal. Suchelectronic devices have simplified the operating procedure and the reliability but for reasons explained later duecaution must be exercised in having blind faith in the resluts obtained unless appropriate verfication has beencarried out beforehand. The following points are reviewed in some detail:(a) Scale expansionEssentially this implies electronic amplification of low absorbance signals or minor differences in absorbance. Theread-out errors is reduced in this way and the sensitivity is apparently increased but the signal to noise ratioremains unchanged. Scale expansion changes the apparent signal, but the absorbance remains the same.Scale expansion can be used until the noise observed is greater than the read-out error and is always recom-mended for absorbance below 0.1. If scale expansion has to be used and the instrument does not have the meansto read the value of the scale expansion factor, the value can be calculated by measuring a suitable solution withand without scale expansion and simply dividing the signal obtained.(b) Concentration read-out and curvature correctionFor concentration read-out, the signal is amplified or reduced so that the numerical read-out value is equal to theconcentration of the analyte in solution. This direct method can only be used if the calibration curve is linear. Fornon-linear calibrations, the curvature has to be corrected electronically so that for a limited range an apparentlylinear relationship is formed, which permits a correct concentration readout.SIST CR 10322:2003



CR 10322:2003 (E)7Equipment without a built-in microprocessor has been encountered which use correction methods which do notalways lead to reliable results. Built-in microprocessors typically compute a mathematical function from thereadings through a few calibration points only. In general, this gives better results but has the drawback of makingeven strongly curved calibration graphs seemingly usable, while in the upper part of the calibration graph, the slopemay be so low that a minor error in the absorbance measurement can result in substantial deviations in thecalculated concentrations. Errors in the preparation as well as in the measurement of the reference solutions havean additional and relatively large influence on the calculated calibration graph.As the calibration graphs in atomic absorption spectrometry are usually not linear, the application of concentrationread-out and curvature correction may give rise to errors in the ultimate results which are easily overlooked.Therefore these auxiliary procedures are only permissible for application in standard methods of iron and steelanalysis if it has been explicitly verified that the results obtained do not deviate significantly from those obtainedwith a very carefully hand-drawn calibration graph of absorbance against concentration.The following suggestions are helpful in drawing a correct (curved) calibration graph:(i) the curve should be drawn smoothly and without bends;(ii) the data points should be spread evenly and at random around the drawn curve. By looking at grazingincidence along the graph paper, these two conditions can easily be checked;(iii) for unknowns with low absorbance values, enlarging of the lower part of the calibration graph is advisable.(c) Damping/integrationDamping suppresses the noise but has the side effect of making the system respond slower to changes of theconcentrations of solutions being measured. Integration means that the average signal is measured by integrationover a pre-selected time. This averaging leads to a reduction of the random error in the final result. Forconvenience of operation, integration times less than 10 seconds are usually to be preferred, while times in excessof approximately 30 seconds are inadvisable because of the risk of drift. It is advisable to carry out two readings inorder to be able to ensure that the signal has stabilized before the first integration period is initiated.3 Systematic and random errorsAs with all analytical measurements, the results obtained by AAS will deviate from the true value due to randomand systematic errors. Random errors are made apparent by differences in results when the measurements arerepeated. Averaging results reduces the random errors. Systematic errors are not reduced by repeating themeasurement. The acceptable error in the final results depends on the purpose of the determination and theconcentration of the element to be determined. In general, there is a tendency to keep the systematic error smallerthen the random error. ECISS/TC 20 considers its task to be to formulate standard methods so that both types oferror are within acceptable limits. Since the apparatus may have a major influence on the derived measurementsand minor adjustments to the standard may be necessary with respect to the sample and the available instrument,it is important to have a clear understanding of the systematic and random errors which may occur.3.1 Systematic errorsSince systematic errors are difficult to detect, much attention has to be given to this aspect in drafting and testingthe standard methods. However, there are some sources of systematic errors which require particular attention.3.1.1 InterferencesSubject to the composition of the solution, the following interferences or matrix effects may occur:(i) physical effects influencing the quantity of the element which reaches the flame;(ii) chemical interference influencing the number of atoms released in the flame;(iii) non-specific absorption by undissociated molecules in the flame;SIST CR 10322:2003



CR 10322:2003 (E)8(iv) spectral line overlap, e.g. iron on zinc, although such effects are rare.Physical effectsThe quantity of the sample that is aspirated and the portion of it which reaches the flame for a given flow of the gasthrough the nebulizer, depends on the viscosity, surface tension, density and vapour pressure of the solution. It istherefore necessary to ensure that the physical properties of the sample and the calibration solutions are as similaras possible.Chemical interferencesChemical interferences are mostly due to reactions of the element of interest with other elements present in theflame. The effects may be very different and for this reason much attention will have been paid to avoiding and/orcorrecting these interferences during the drafting of the standard. Therefore it is undesirable to deviate from thestandard method without a good reason. Since instances have been known an interference is peculiar to aparticular instrument and not to others, the user is advised to check for these effects on his own instrument the firsttime at which that type of sample is analysed. A simple test is to add a known quantity of the element beingdetermined to a portion of the sample and to check for complete recovery.Non-specific absorptionCompounds present in the matrix may cause a decrease in the intensity of the reasonance radiation reaching thedetector. This is due to molecular absorption or scattering and it increases with decreasing wavelength. Using aflame, this interference only occurs at wavelengths below about 325 mm and with high salt concentrations in thesolution. Solutions containing high concentration of iron (in excess of about 1 000 mg/ml) should always be checkedfor background absorption.The occurrence of non-specific background absorption can easily be detected with the aid of the different devicesavailable. These have to be used under the same conditions applied to the atomic absorption measurement.Simultaneous use of the atomic radiation source makes it possible to correct for background absorptionautomatically.3.1.2 Errors as a result of curvature of the calibration graphThe concentration of the sample solution is determined by comparison with calibration solutions used to plot thegraph. The accuracy with which the graph can be established depends on the number of points and its shape. Ifthe graph is a straight line, a reliable calibration can be easily calculated from a limited number of points. However,the calibration graph in atomic absorption spectrometry is usually not a straight line and there is not yet a generalagreement about the best mathematical function which describes it. This does not mean that one is restricted to thelinear part but that extreme care has to be exercised in plotting since an erroneously placed line will lead tosystematic errors in the results.The uncertainty in positioning the graph increases as the curvature increases and as the number of pointsdecreases. Therefore an upper limit has to be set to the permissible curvature in each standard method. This limitis stated explicitly or it is taken into account when the maximum permissible absorbance or concentration are given.It must be emphasized that these considerations apply not only to manually plotted graphs but also for calibrationcalculated mathematically. Most modern atomic absorption spectrometry instruments are fitted withmicroprocessors which calculate the best fitting curve through a number of calibration points. This number is oftenvery small and it may therefore not be possible to obtain the optimum result when the graph is curved. Unless ithas been previously verified that reliable results are attainable, this facility should not be used for most accuratework.When the graph is not a straight line, it is not permissible to subtract absorbances, for example of blank solutions,from the calibration solution or test solution absorbances directly. All measurements of calibration and analytesolutions should be carried out with reference to a solution which does not contain the elements of interest inmeasurable concentration (for instance pure water or other solvent). Only after conversion of the measuredabsorbances into concentration units may corrections be made for the reagent blank.SIST CR 10322:2003



CR 10322:2003 (E)93.2 Random errorsIn the first place, random errors are apparent from the fluctuations of the absorbance values when measurementsare repeated and may be characterized by the relative standard deviation. For the analytical determinationhowever, it is the relative standard deviation of the concentration which is important and this correlates with therelative standard deviation of absorbance via the slope of the calibration curve at the point concerned. The relativestandard deviation of the absorbance measurement depends on the instrument used, damping/integration time,element, flame and the absorbance level. It is at a minimum of about 0,5 to 1,5 % at an absorbance value of about0.5. At greater absorbances the relative standard deviation of the concentration increases, either as a result ofincreased variability of the measurements or decreased slope of the graphs. In practice it appears that aboveabsorbance values of 0,8 to 2,0 depending on the instrument, the setting and the element, the relative standarddeviation of the concentration is too high to be acceptable for most purposes.Generally the usable part of the calibration graph is limited by the desired relative standard deviation (coefficient ofvariation) in the concentration. If a large working range is required (e.g. when analysing batches of samples toavoid unnecessary dilutions), then its acceptable limits should be determined in concentration units for eachselected combination of measurements conditions.ECISS/TC 20, aiming to formulate the best standard method for a single sample, has opted for a simplerprocedure, where for each standard method the upper limit of the measurement range is restricted in terms ofabsorbance, or concentration, the permissible curvature of the calibration graph is limited (3.1.2) and the allowablestandard deviations at the top and the bottom of the working range are defined in absorbance units. Thus a checkon the suitability of the instrument can be made (at an early stage) without performing the whole calibrationprocedure.If the specified standard deviations are exceeded then the setting and operation of the instrument should bechecked. If they are still exceeded, it may be possible to achieve them by reducing the working range. If not, theinstrument is not suitable for the method.As noted above, the corresponding relative standard deviations in concentration units will be affected by graphcurvature. When required, these may be determined by following the procedure given in Annex D3.4 Setting and checking of the atomic absorption spectrometerInstructions for the setting of and measurements with the spectrometer are given in Annex A. During the settingprocedure, two properties of the instrument – sensitivity and detection limit – are determined. An explanation ofthese terms is given below, while directions for their determination are given in Appendices D1 and D2respectively. Annex B gives a survey of the instrument parameters and settings which have to be defined andwhich may be stated in an Annex to the analysis report. Annex C deals with the safety aspects which should beobserved.4.1 SensitivityThe sensitivity of a relative analytical method is defined as the slope of the calibration graph i.e. the relationshipbetween concentration and signal. In atomic absorption spectrometry, sensitivity depends not only on the elementbut also on the instrument and on its settings.The variations in sensitivity, combined with maximum permissible absorbance or concentration specified in thestandard method make it necessary to adjust the concentrations of calibration solutions and dilution of the samplesolutions to suit the particular instrument used.Each standard method should include performance criteria that the instrument must meet, as described in CR10321. As a check for correct functioning of the instrument, however, it is useful to compare achieved sensitivityw
...