APRIL. 1968 Vol. 93, No. I105 THE ANALYST The Application of Atomic-absorption Spectrophotometry to the Analysis of Iron and Steel A Review“: BY P. H. SCHOLES (BISRA , The Inter-Group Laboratory of the British Steel Corporation, Chemical Analysis Section, Metallurgy Division, Hoyle Street, Shefield S3 7EY) Atomic-absorption spectrophotometry is a useful technique for the determination of many of the minor elements commonly present in iron and steel. It is rapid and relatively free from most of the troublesome inter- element effects associated with alternative techniques such as colorimetric and polarographic analysis. This paper reviews published and certain unpublished information dealing specifically with the analysis of iron and steel by atomic-absorption spectrophotometry .SINCE 1961 about twenty-five papers have been published concerned with the use of atomic absorption for the analysis of iron and steel. Procedures have been proposed for the deter- mination of twelve elements with the main emphasis on low-level concentrations, but some authors have also suggested possible application to elements present in alloying concentra- tions. Most of the elements commonly found in iron and steel have been determined. Notable exceptions are carbon, sulphur and phosphorus, the resonance lines of which lie in the vacuum ultraviolet region of the spectrum, and silicon, arsenic, zirconium, niobium, tantalum and tungsten which form refractory compounds in the flame that are not easily dissociated into atomic form. The determination of aluminium is being studied in the author’s laboratory, and this will be considered in a separate publication.In atomic-absorption procedures it is usual for the sample to be dissolved and the solution then sprayed into the flame; hence the outstanding advantage of the method is its simplicity. Most of the published methods have aimed at perpetuating this‘simple approach ; occasionally , suppressing agents are added but the original solution is more often used without any time-consuming separations. The technique can, of course, be used as a means of completing certain determinations, for example, after concentration of the elements sought by the use of selective organic reagents. On many occasions, this approach offers advantages to the analyst but chemical pre-treatment should be avoided when possible.This review is restricted to papers dealing specifically with iron and steel. In addition, reference is made to unpublished work in the author’s laboratory and to information made available by colleagues in industry. Each element is dealt with separately and a table is included in which are listed the instrument, resonance line, flame, gas flow-rate and absorption path length used. Similarly, the composition of the solution sprayed into the flame is given, together with analytical sensitivity in terms of element concentration (in p.p.m.) giving rise to 1 per cent. absorption (0*0044 optical density). The deficiencies in the use of the term “sensitivity” have been pointed out by Slavin, Sprague and Manning1 and Elwell and Gidley.* Sensitivity is dependent upon many variables; most authors only provide details of gas flow-rates, but position of the absorption path in the flame, monochromator slit width and rate of solution uptake into the atomiser are equally important.When a stable flame such as air - acetylene is used to determine easily dissociated elements, the actual limit of detection may be several times better than the quoted sensitivity value, but in the determination of elements such as aluminium and titanium with the relatively unstable nitrous oxide - acetylene flame, the signal-to-noise ratio is un- favourable, and the detection limit may be of the same order as the sensitivity. When the noise level is high a chart recorder is an essential accessory in order to permit accurate measurement of absorption signal superimposed on the background noise.The principles of atomic absorption and the techniques of measurement will not be discussed in this review, as they have been adequately covered by Elwell and Gidley2 and Robins~n.~ Elwell and Gidley in their monograph also provide a tabulated list of instruments For details see Summaries in advertisement pages. *Reprints of this paper will be available shortly. 0 SAC and BISRA. 197198 SCHOLES : THE APPLICATION OF ATOMIC-ABSORPTION [AutdySt, VOl. 93 that were commercially available in mid-1966. Since then several new instruments have appeared, notably the Techtron AA4 and the Hilger and Watts Atomspek. Of special interest is the Techtron AR-200, in which the conventional monochromator is replaced by a resonance detector of the type described by Sullivan and Walsh.4 MAGNESIUM Belcher and Bray6 of B.H.P.Ltd., in Australia, have proposed a simple procedure for the determination of magnesium in iron to replace traditional procedures, most of which are tedious and require separation of matrix elements. Measurements are made by absorption of the 285-2-mp resonance line in an air - acetylene flame. The l-g sample is dissolved in 30 ml of hydrochloric acid (1 + 1) and 5 ml of concen- trated nitric acid. After evaporation to dryness the residue is baked at 200" C for 5 minutes, then re-dissolved in 10ml of concentrated hydrochloric acid. The solution is diluted with 50 ml of water, and silica removed by filtration, washing with dilute hydrochloric acid (5 + 95).After the addition of 5ml of 18 per cent. w/v strontium chloride solution, the filtrate is diluted to 200 ml and absorption measurements made. The addition of strontium is made to suppress interference from aluminium present in up to 2 per cent. concentration. No interference is encountered from 0.5 per cent. each of phosphorus, titanium or zirconium, 1 per cent. of zinc, 2 per cent. each of vanadium, man- ganese or silica, 5 per cent. each of nickel, copper or molybdenum and 10 per cent. of chromium. Variations of k25 per cent. in the concentration of iron, strontium chloride and hydrochloric acid are without significant effect. In 1961, the Standards Association of Australia Committee (S.A.A.) CH/4 on Sampling and Analysis of Ferrous Metals tested Belcher and Bray's method by analysing six samples of iron containing 0 to 0.17 per cent.of magnesium.6 Satisfactory results were obtained by four analysts, and the method was published as Australian Standard Method, AS KI : Part 20: 1964. The Australian procedure has been tested by Sprague and Slavin,' and Clarke (in un- published work at B.C.I.R.A.). The interference effect of aluminium was confirmed by Clarke, who found that the addition of strontium was effective as a suppressive agent, except for a sample of iron containing 22 per cent. of aluminium. An apparent interference effect caused by a large amount of nickel was also noted by Clarke and Cooke (in unpublished work). Sprague and Slavin, and Clarke, considered the procedure to be satisfactory for the analysis of cast iron.Clarke, who used a rapid method in which removal of silicon by dehydration was omitted, obtained a satisfactory result on a standard sample of nodular cast iron, the time required to complete a determination being less than 7 minutes. TABLE I INSTRUMENT CONDITIONS AND SENSITIVITIES FOR MAGNESIUM Author and instrument Belcher and Bray6 Special Clarke (unpublished work) Evans Electro- selenium Ltd. (E.E.L.) Clarke Southern A1760 Clarke Optica Densatomic Clarke and Cooke Unicam Model SP90 Sprague and Slavin' Perkin-Elmer Model 303 Sensitivity Equivale;l t content, weight Flame conditions Solution composition p.p.m. per cent. Air - acetylene, 286.2 mp. 0.6 per cent. w/v of sample Not stated - 10-cm path in dilute HCI Air - acetylene, 285.2 mp 0.6 per cent.w/v of sample Not stated - (6 + 96) + Sr in dilute HCl + Sr Air - acetylene (6 and 0.1 per cent. w/v of sample 0.007 0.0007 1.8 litres per minute), 286.2 mp, 12-cm path 3 cubic feet per hour), in dilute HCI (1 + 99) + Sr in dilute HC1 Air - acetylene (16 and 0.1 per cent. w/v of sample 0.012 0.00 12 285.2 mp, 5-cm path Air - acetylene (6 and 0.1 per cent. w/v of sample 0.012 0.0012 (1 + 99) + Sr 1.8 litres per minute), 285.2 mp, IO-cm path Air - acetylene, 286.2 mp 0.1 per cent. w/v of sample Not stated - in dilute HC1 (1 + 99) + Sr in dilute HC1 (1 + 99) + SrApril, 19681 SPECTROPHOTOMETRY TO THE ANALYSIS OF IRON AND STEEL 199 CALCIUM In preliminary, unpublished work in the author’s laboratory, it has been shown that calcium can be determined in stainless and maraging steel after a preliminary separation by mercury-cathode electrolysis.Previous work on the development of a method for the deter- mination of lime in steel-making slag has suggested that, in addition to phosphorus, certain other elements, such as vanadium and titanium, which would be present after electrolytic separation, might interfere in absorption measurements. For this reason sufficient strontium solution was added as a releasing agent to give a concentration of 2000 p.p.m. in the final solution. In the proposed procedure the 1-g sample is dissolved in aqua regia and the solution evaporated to incipient fumes, after the addition of 20ml of perchloric acid. The solution is then electrolysed and the electrolyte diluted to 100m1, after the addition of strontium.Rigid precautions must be taken to prevent contamination from the calcium present as a filler in the rubber used in laboratory apparatus. Calibration graphs, prepared by adding calcium to high-purity iron and electrolyshg the solution, were in good agreement with those prepared by adding calcium to dilute solutions of perchloric acid. Further work is required to confirm the non-interference of molybdenum and tungsten present with the calcium after electrolysis. More recent work has suggested that small amounts of calcium can be determined in alloy steel, without chemical separation, by using a nitrous oxide - acetylene flame. The 1-g sample is dissolved in 15 ml of concentrated hydrochloric acid and 5 ml of concentrated nitric acid, the solution filtered and diluted to 100 ml.A “standard addition” technique is recommended for calibration purposes. TABLE I1 INSTRUMENT CONDITIONS AND SENSITIVITIES FOR CALCIUM Sensitivity r t content, weight Author and instrument Flame conditions Solution composition p.p.m. per cent. Thulbourne and Air - acetylene, 422.7 mp, 1 per cent. w/v of sample 0.2 0.002 Scholes 13-cm path in dilute HClO, Hilger AA2 Thulbourne and Scholes 422.7 mp, 6-cm path in a mixture of HCl, Hilger AA2 HNO, and water (1 + 4) + Sr after elec- trolytic separation Nitrous oxide - acetylene, 1 per cent. w/v of sample 0.15 0*0016 (3 + 1 + 16) TITANIUM Amos and Willis8 have demonstrated the feasibility of titanium determinations by atomic absorption with a nitrous oxide - acetylene flame.Headridge and Hubbard9 used this flame to determine titanium in steel, magnet alloy and cast iron without preliminary separation. Hydrofluoric acid was chosen as solvent in order that niobium and tungsten present in alloy steel would be held in solution. Headridge and Hubbard found that the sensitivity of titanium absorption in the nitrous oxide - acetylene flame was low (6.5 p.p.m.), but a considerable improvement could be achieved by the use of aqueous - ethanolic sample solutions. In a mixture of M hydrofluoric acid and ethanol (1 + 9) the sensitivity was 2 p.p.m.; in the presence of iron, the authors estimated that the actual limit of detection would be about 1 p.p.m. of titanium. With the exception of aluminium, alloying elements commonly present in steel do not affect absorption measurements.Aluminium present at the 10 per cent. level causes a slight enhancement, but this effect can be prevented in the analysis of magnet alloys by omitting the ethanol addition. Analytical results reported by Headridge and Hubbard on four alloy steels, and one cast iron with a slightly different sample solution composition, agreed closely with accepted values by other techniques. Mostyn and Cunninghamlo determined titanium in nickel and iron alloys with a nitrous oxide - acetylene flame by direct dissolution of the sample in aqua regia, avoiding the use of hydrofluoric acid. Interference from other elements present in alloys was significant, and200 SCHOLES : THE APPLICATION OF ATOMIC-ABSORPTION [Andyst, vol. 93 it was found that almost complete matching of calibration solutions and test sample was necessary in order to achieve reasonable accuracy.Interference effects could, however, be stabilised by the presence of an alkali metal such as potassium, and an addition of 1000 p.p.m. of this element was adopted in the proposed procedure. Bowman and Willisll preferred to use a sulphuric acid sample solution. The presence of iron was found to depress titanium absorption at sulphuric acid concentrations less than N, but to enhance it at higher acid con- centrations, maximum absorption being shown at an acid concentration of 4 N. This enhance- ment effect depends not only on the concentration of iron and acid, but also on the type of flame used and the position of the absorption path in the flame.The method finally adopted was based on matching calibrations with respect to the major constituents present in the test samples. Satisfactory results are given for several standard samples of low alloy and stainless steel. TABLE I11 INSTRUMENT CONDITIONS AND SENSITIVITIES FOR TITANIUM Sensitivity --t At1 thor content, weight and instrument Flame conditions Solution composition p.p.m. per cent. Headridge and Nitrous oxide - acetylene 0-6 per cent. w/v of sample 3 0.06 Hubbards Hilger AA3 Mostyn and Cunning- Nitrous oxide - acetylene, 0.1 per cent. w/v of sample 3.5 0-35 hamlo 364.3 mp, 5-cm path in dilute HCl(2 + 98) Perkin-Elmcr Model 303 Bowman and Willisl' Nitrous oxide - acetylene, 0.8 per cent. w/v of sample 3 0.04 Techtron AA4 (6.26 and 3.8 litres per minute), 364-3 mp, 6-cm path in a mixture of 0.1 M HF and ethanol (1 + 1) + 1000p.p.m.of K 364.3 mp, 5-cm path in dilute H,SO, (1 + 9) VANADIUM There is relatively little published information on the determination of vanadium by atomic 'absorption. In 1965, Trent and Manning12 described vanadium determination with a pre-mix, oxy-acetylene burner. In this work clogging of the burner was encountered when aspirating steel-sample solutions. Although the method of iron separation used to avoid this difficulty made the procedure longer than desirable, the vanadium determination could be made on a routine basis, provided that the burner was controlled to prevent flashback. Capacho-Delgado and Manning13 used a high intensity lamp and made absorption measure- ments at 318.4 mp in an acetylene - nitrous oxide flame. With a sample solution containing dilute sulphuric and phosphoric acids satisfactory results are reported for vanadium contents o f 0.02 to 0.05 per cent. present in alloy-steel samples.Both sulphuric and phosphoric acids interfere in absorption measurements of vanadium, but this effect is eliminated by calibrating in the presence of both acids. TABLE IV INSTRUMENT CONDITIONS AND SENSITIVITIES FOR VANADIUM Sensitivity -t Author content, weight and instrument Flame conditions Solution composition p.p.m. per cent. 0-026 Capacho-Delgado Nitrous oxide - acetylene, 0.6 per cent. w/v of sample 1.3 and ManningIa 318-4 m p in a mixture of H,SO,, Perkin-Elmer Model HaPo, and water 303 (3 + 3 + 94) CHROMIUM Kinson, Hodges and Belcherl* determined chromium by using the 359*4-mp resonance line in an air - acetylene flame.Interference from nearby argon lines prevents the use of the inore sensitive chromium line at 35'7.9 mp, but this line may be used with advantage if lamps are available filled with a gas other than argon.April, 19681 SPECTROPHOTOMETRY TO THE ANALYSIS OF IRON AND STEEL 201 The effects of different solvent acids were examined; a mixture of sulphuric and phos- phoric acids was selected because of its advantage in retaining tungsten and other acid- hydrolysable elements in solution. Belcher used this same solution for the determination of manganese, nickel and copper (see following sections). In the proposed procedure, the l-g sample is dissolved in 30 ml of a mixture of sulphuric acid, phosphoric acid and water (3 + 3 + la), the solution oxidised with nitric acid and then evaporated to fumes.After filtration the solution is diluted to 100ml for absorption measurements. Iron causes a major reduction in absorption but this may be maintained at a constant level by using a sample solution containing sulphuric and phosphoric acids. Interference from molybdenum, tungsten and nickel varies with flame type and height of the absorption path above the burner. The most useful flame type is a slightly rich mixture (10.4 litres of air per minute and 2-4 litres of acetylene per minute with the absorption path 8 mm above the burner top. Under these conditions there is no interference from 5 per cent. each of nickel, manganese, copper or tungsten, 2 per cent.each of aluminium or vanadium, and 1 per cent. of molybdenum. Higher percentages of nickel, tungsten and molybdenum cause interference; at the 0.5 per cent. chromium level, 20 per cent. of nickel and 20 per cent. of tungsten give slightly high results, and 5 per cent of molybdenum gives slightly low results. Giammarise,ls who used the 357-9-my line, investigated the effects of ammonium chloride and strontium chloride for the suppression of iron interference in chromium determination. When iron is present at concentrations of 0.1 g per 100 ml, or less, it is recommended that 20,000 p.p.m. of ammonium chloride should be added to eliminate interference at the 1 to 4 p.p.m. chromium level. Strontium chloride also exhibits a suppressive effect but it is less effective than ammonium chloride.Barnes1* has also shown that the addition of ammonium chloride doubles the sensitivity for chromium determination in the presence of iron. Barnes' findings have been confirmed by Clarke and, by using the resonance line at 357.9 mp and 1 per cent. w/v sample solution in dilute hydrochloric acid (1 + 9), Clarke and Cooke obtained a curved calibration graph over the range 5 to 100 p.p.m. (0-05 to 1 per cent. of chromium). The curvature may be caused by the presence of the nearby argon lines. Results on cast irons and low alloy steels were satisfactory. Further unpublished work by Clarke, with a Southern A1750 instrument and a 0-1 per cent. w/v sample solution, gave an almost linear calibration graph over the range 0.1 to 10 p.p.m.Reid and Goldrich, at Colvilles Ltd.,l' also noted a pronounced curvature when using the 357.9 mp line. In the 0 to 0.2 per cent. range these workers found that the 95 per cent. confidence limits of the procedure was similar to that of the British Standard spectrophotometric method (B.S. 1121 : Part 24 : 1952), i.e., +O.Ol per cent., but precision deteriorated at higher chromium levels, possibly because of instrumental difficulties in obtaining a sufficiently fuel-rich flame when using an air - acetylene mixture. These workers also used an air-propane flame in an attempt to improve precision with an addition of ammonium chloride to the sample solution. They found that sensitivity was reduced by a factor of three and that ammonium chloride as a suppressant for iron interference was not effective when using air - propane.TABLE V INSTRUMENT CONDITIONS AND SENSITIVITIES FOR CHROMIUM Sensitivity Author and instrument Kinson, Hodges and Belcherl, Special Clarke and Cooke Unicam SP90 Clarke Southern A1760 Reid and Goldrich17 Hilger AA2 Flame conditions Air - acetylene (10-4 and 2.4 litres per minute), 10-cm path, 359-4 mp Air - acetylene (7.6 and 2.0 litres per minute), 10-cm path, 357.9 mp Air - acetylene (7 and 2-6 litres per minute), 12-cm path, 367-9 mp Air - acetylene (9 and 1.2 litres per minute), 357.9 mp Solution composition 1 per cent. w/v of sample in a mixture of H,SO,, H,PO, and water (4.6 + 4.5 + 91) 1 per cent. w/v of sample in dilute HCl (1 + 9) + 1 per cent. w/v of NH,C1 0.1 per cent, w/v of sample in dilute HC1 (1 + 99) + 2 per cent.w/v of NH,Cl 1 per cent. w/v of sample in dilute HCl (1 + 9) r- 1 Equivalent content, weight p.p.m. per cent. 0.5 0.005 0.6 0.2 0.7 0.005 0.02 0.007202 SCHOLES : THE APPLICATION OF ATOMIC-ABSORPTION [Analyst, Vol. 93 MANGANESE A procedure has been described by Belcher and Kinsonla for the determination of 0*001 to 2 per cent. of manganese in low and high alloy steels with a solution prepared in a similar manner to that described for chromium determination, but with the addition of a little sulphurous acid to reduce any oxidised manganese salts. For best sensitivity the resonance line 2796mp is used, but for higher manganese contents one of the less sensitive lines at 279.8 or 280.1 mp lines is preferred. Iron interferes with manganese absorption in an air - acetylene flame, but the effect may be largely eliminated by using a stoicheiometric gas mixture (15 litres of air per minute Cphs 2.2 litres of acetylene per minute).Chromium at the 20 per cent. level causes positive interference but this effect may be reduced by confining the absorption path to a small section of the flame and by increasing the slit width of the burner from 0.04 to 0.075 cm. Under these conditions, Belcher and Kinson reported freedom from interference from 30 per cent. each of nickel or chromium, 10 per cent. each of tungsten or cobalt, 5 per cent. each of molybdenum or copper, 3 per cent. of silicon, 2 per cent of vanadium and 0 6 per cent. of aluminium. Results reported on standard steels show excellent precision and accuracy, particularly at low levels of manganese concentration.Sprague and Slavin' used aqua regia to dissolve steel samples by a procedure similar to that described by Belcher and Kinson. A significant interference was experienced due to molybdenum, silicon and tungsten, but on replacing the acetylene cylinder, which was nearly empty, the interference due to molybdenum and tungsten disappeared. Silicon interference was eliminated by modifying the equipment so that only radiation passing through a restricted area of the flame was received by the detector. Results obtained on N.B.S. samples analysed by Sprague and Slavin show poor precision, and in certain cases the agreement with certificate values is unsatisfactory. Tyou and CatouP9 studied the effect of monochromator slit width on analytical sensi- tivity.They used a sulphuric acid sample solution and found slight interference with changes in acid concentration. An enhancement effect due to iron was also noted but this was eliminated by adding iron to the calibration solutions. Excellent results are reported for standard steels, and the coefficient of variation of the procedure is about 1-5 per cent. Clarke, by using the E.E.L. instrument, obtained low results for cast irons when using a calibration prepared by adding manganese to pure iron. Tests indicated that this might be caused by phosphorus interference and that satisfactory results could be obtained by adding strontium chloride to the samples. This apparent interference was not confirmed by tests made by using the Southern and Unicam instruments.Other elements present in cast irons and low alloy steels did not interfere. On the Southern instrument a markedly curved calibration graph was obtained with the 279.5 mp line. At 403-1 mp a linear calibration graph was obtained but sensitivity was considerably reduced. Reid and Goldrich17 omitted the addition of strontium and obtained satisfactory results on standard samples of mild and low alloy steel. TABLE VI INSTRUMENT CONDITIONS AND SENSITIVITIES FOR MANGANESE Sensitivity Equivaledt Author content, weight and instrument Flame conditions Solution composition p.p.m. per cent. Belcher and Kinsonl* Air - acetylene (15 and 1 per cent. w/v of sample 0-08 0.0008 Sprague and Slavin7 Air - acetylene, 279.5 mp 1 per cent.w/v of sample Not stated - Special 2.2 litres per minute), in a mixture of H,SO,, 279.6 mp, IO-cm path H,PO, and water (4.5 + 4.6 + 91) in a mixture of HCl, Perkin-Elmer Model 303 HNO, and water Clarke and Cooke Air - acetylene (6 and 0.1 per cent. w/v of sample 0.11 Unicam SP90 1.6 litres per minute), 279.6 mp, 10-cm path Clarke Air - acetylene (7 and 0.1 per cent. w/v of sample 0.09 Southern 1.8 litres per minute), A1750 279-6 mp, 12-cm path Reid and Goldrich17 Air - acetylene (9 and 1 per cent. w/v of sample 0.11 HiIger AA2 1-25 litres per minute), 279-5 m p (12 + 3 + 86) in dilute HC1 (1 + 99) in dilute HCl (1 + 99) in dilute HCl (1 + 9) 0.01 1 0.009 0.00 1 1April, 19681 SPECTROPHOTOMETRY TO THE ANALYSIS OF IRON AND STEEL COBALT 203 McPherson, Price and Scaife, at J.Lysaghts (Aust.) Ltd.,20 proposed a method (applicable to all types of steel) for the determination of less than 1 per cent. of cobalt. By using a modified solution procedure the method may also be applied to cobalt contents up to 12 per cent., at which level the tentative “reproducibility” is stated to be k0.2 per cent. Typical results for the low level cobalt range were reported by McPherson at the 17th Chemists’ Conference.21 Both procedures, which are simple and apparently free from interference, are being considered by the Australian CH/4 Committee as a possible S.A.A. method to replace existing spectrophotometric techniques. Lockyer22Js has also proposed a procedure, with special application to the analysis of small amounts of cobalt in alloy steel.In an attempt to achieve maximum sensitivity, a concentrated sample solution (10 per cent. w/v) was used initially, but results indicated some apparent absorption caused by light scattering by particulate matter in the flame. A less concentrated sample solution (1 per cent. w/v) was preferred with maximum electrical amplification, and with these conditions absorption measurements at 240.7 mp showed satis- factory agreement with conventional procedures over the range 0-005 to 0.2 per cent. of cobalt. Further successful tests were made in which cobalt was added to the test sample and the percentage recovery determined. Sprague and Slavin,’ who determined cobalt with a 1 per cent. sample solution, reported a limited number of satisfactory results on low and high alloy steels.Preliminary work in the author’s laboratory has shown that instrumental sensitivity is not adequate for determination of cobalt at levels below 0-005 per cent.; for this range a preliminary solvent extraction with isobutyl acetate is necessary to remove most of the iron and to permit concentration to give a 10 per cent. w/v sample solution. Large amounts of nickel and chromium remaining in solution after iron removal will enhance cobalt absorption measurements, and it is recommended that calibration solutions should be prepared containing about the same amount of those elements as the test samples. Careful calibration and a sample concentration technique permits results to be produced that are significant to the nearest 0.0001 per cent. of cobalt.The interference effects of large amounts of nickel and certain other elements have been confirmed by the Australian CH/4 Committee. Interference, which occurs when a fuel-rich air - acetylene flame is used, can be eliminated by using a lean flame. Author and instrument McPherson et aE.2‘J Techtron TABLE VII INSTRUMENT CONDITIONS AND SENSITIVITIES FOR COBALT Sensitivity Lock ye@ Hilger AA2 Sprague and Slavin7 Perkin-Elmer Model 303 Thulbourne and Scholes Hilger AA2 r Equivaledt content, weight Flame conditions Solution composition p.p.m. per cent. Lean air - acetylene, (a) 4 per cent. w/v of sample Not stated - 240.7 mp, 10-cm path in dilute HCl(1 + 4) (b) 0.5 per cent. w/v of sam- Not stated ple in a mixture of H,SO,, H,PO, and water (1 + 1 + 23) (for alloy steel) 0.4 or 1 per cent.w/v of sample in dilute HC1 Air - acetylene (10.5 and 1.22 litres per minute), 240.7 m p Not stated Air - acetylene 1 per cent. w/v of sample Not stated - in a mixture of HC1, HNO, and water (12 + 3 + 85) 10 per cent. w/v of sample in dilute HCl (5 + 95), after solvent extraction Air - acetylene (10 and 1.25 litres per minute), 240-7 mp, 13-cm path 0.25 0.00028204 SCHOLES : THE APPLICATION OF ATOMIC-ABSORPTION [Arta@st, VOl. 93 NICKEL With a dilute phosphoric acid solution, Kinson and B e l ~ h e r ~ ~ determined nickel in steel by absorption of the 232-01-mp line in a lean air - acetylene flame. Sensitivity is markedly dependent upon slit width because of the difficulty in obtaining resolution from the non- absorbing nickel line at 231998mp; the resultant non-linear calibration graph limits the maximum content of nickel that can be determined to 2 per cent.It seems probable that such difficulties will be minimised by the use of high intensity lamps of the type described by Sullivan and Walsh,26 or other types of high output lamps. Kinson and Belcher report no interference from iron, 30 per cent. of chromium, 20 per cent. each of manganese or tungsten, 10 per cent. each of copper or cobalt and 5 per cent. each of vanadium, molybdenum or aluminium. Clarke and Cooke report that nickel can be determined in cast iron and low alloy steel over the range 0-05 to 0-5 per cent. With a conventional hollow-cathode lamp they recom- mend use of the line at 341*5mp, which gives a nearly linear calibration graph. With a ‘‘high spectral output” lamp the line at 232.01 mp can be used with an appreciable gain in sensitivity.Work by Clarke on other instruments also indicates the advisability of using the 341-5 m p line. However, on the Optica instrument the calibration graph at 232-01 mp was appreciably less curved, even although a combined nickel - cobalt - iron lamp was used. Reid and Goldrich,l7 who used standard hollow-cathode lamps and the 232.01-mp line, reported satisfactory results on standard steels containing up to 1 per cent. of nickel. At higher nickel levels poor precision was noted, possibly because of the curvature of the cali- bration graph. TABLE VIII INSTRUMENT CONDITIONS AND SENSITIVITIES FOR NICKEL Sensitivity r t Author content, weight and instrument Flame conditions Solution composition p.p.m.per cent. Kinson and BelcheP Air - acetylene (10.6 and 1 per cent. w/v of sample 0-52 0*0062 Special 1.6 litres per minute), in a mixture of H,PO,, Clarke Air - acetylene, 341.6 mp 0.6 per cent. w/v of sample 0.70 0.014 E.E.L. Clarke and Cooke Air - acetylene (5 and 1 per cent. w/v of sample 0.44 0-0044 Unicam SP90 232-01 mp, 10-cm path H,SO, and water in dilute HCl (1 + 9) 1-20 litres per minute), in dilute HC1 (1 + 9) 232.01 mp, 10-cm path 341.6 mp, 10-cm path 1 per cent. w/v of sample 2.6 0.026 in dilute HCI (1 + 9) Clarke Air - acetylene (16 and 0.1 per cent. w/v of sample 0.28 0.028 Optica in dilute HCl (1 + 99) 3 cubic feet per hour), 232.01 mp, 5-cm path 341.6 mp, 6-cm path 0.1 per cent. w/v of sample 0.9 0.09 in dilute HCl(1 + 99) in dilute HCl (1 + 9) Reid and Goldrich17 Air - acetylene (9 and 1 per cent. w/v of sample in 0.21 0.0021 COPPER The determination of small amounts of copper in steel by atomic absorption is a relatively simple procedure, first demonstrated by Kinson and Belcher.26 Absorption measure- ments at 324.8 mp in a lean air - acetylene flame are free from interference from 20 per cent.each of nickel, chromium, manganese or tungsten, 10 per cent. each of cobalt or vanadium, 5 per cent. of molybdenum and 1 per cent. of aluminium. Variations in the iron content of the sample solution from 0 to 10,000 p.p.m. cause only a small decrease in copper absorption measurements. Kinson and Belcher’s work has been confirmed by other workers. Wollerton and Nall, at the Bragg Laboratory (unpublished work), used a similar procedure to analyse low alloy steels and stainless steel after a fuming treatment with sulphuric acid.Results by Sprague and Slavin,’ unpublished work at B.C.I.R.A., and Reid and Goldrich,17 all show satisfactory agreement with results obtained by more conventional procedures. Hilger AA2 1.26 litres per minute), 232.01 mpApril, 19681 SPECTROPHOTOMETRY TO THE ANALYSIS OF IRON AND STEEL TABLE IX INSTRUMENT CONDITIONS AND SENSITIVITIES FOR COPPER 205 Author and instrument Flame conditions Kinson and BelcherB8 Air - acetylene (10.5 and Special 1.5 litres per minute), 3244 mp, 10-cm path Wollerton and Nall Hilger AAl Air - coal gas, 324.8 mp Sprague and Slavin' Perkin-Elmer Model 303 Air - acetylene Clarke E.E.L.Clarke and Cooke Unicam SP90 Clarke Southern A1760 Clarke Optica Reid and Goldrich" Hilger AA2 Air - acetylene, 324.8 m p Air - acetylene (6 and 1.5 litres per minute), 324.8 mp, 10-cm path Air - acetylene (7 and 1-6 litres per minute), 324.8 mp, 12-cm path 327.4 mp, 12-cm path Air - acetylene Air - acetylene (9 and 26 litres per minute), 32443 mp Sensitivity - content, weight Solution composition p.p.m. 1 per cent. w/v of sample in a mixture of H,PO,, H,SO, and water (4-5 + 4-6 + 91) (a) 1 per cent. w/v of 0.10 Not stated sample in dilute HNO, (5 + 95) (b) 2 per cent. w/v of sample in dilute H,SO, 1 per cent. w/v of sample in a mixture of HCl, HNO, and water 0-5 per cent. w/v of sample Not stated in dilute HCI 0.1 per cent. w/v of sample in dilute HCI (1 + 99) Not stated (1 + 9) Not stated (12 + 3 + 85) 0.12 As in previous sample 0.09 As in previous sample 0.2 0-1 per cent.w/v of sample in dilute HCI (1 + 99) 1 per cent. w/v of sample in dilute HC1 (1 + 9) 0.15 0-17 per cent, 0.001 - - - 0.012 0.009 0.02 0.015 0.0017 MOLYBDENUM The original work of David,27 who examined various factors affecting the determination of molybdenum with the 313.3-mp resonance line in an air-acetylene flame, has been evaluated by Mostyn and Cunningham.2s Absorption is extremely sensitive to small changa in solution composition, and it is possible that a complex interfering ion system is formed of the type described by Firma11.2~ David added 2000 p.p.m. of aluminium to suppress interference from such elements as manganese and iron, but Mostyn and Cunningham found that the addition of ammonium chloride (2 per cent.w/v) in solution was more efficient. David, and Mostyn and Cunningham, used fuel-rich luminous flames which tend to give unfavourable signal-to-noise ratios. Mostyn and Cunningham had to use a less sensitive resonance line (379-8 mp) in order to obtain an acceptable noise level. Kirkbright, Smith and West30 also examined molybdenum determination in an air- acetylene flame. Several resonance lines were used but it was found for each that either the noise level was high or the sensitivity was inadequate. They pointed out that the sensi- tivities reported earlier by David, and Mostyn and Cunningham, benefit from scale expansion and from the use of relatively long absorbance path-lengths.To overcome the disadvantage of the air - acetylene flame, Kirkbright, Smith and West proposed the use of nitrous oxide - acetylene for special application to the analysis of alloy steel. The noise level is considerably reduced, thus permitting absorption measurements with the most sensitive 313-3-mp line. Only the matrix element iron interferes, causing severe depression of molybdenum absorption, but, provided that the solution aspirated contains about 1 per cent. w/v of tervalent iron, variation in the ratio of molybdenum to iron is not critical. In the proposed procedure, the 1-g sample is dissolved in 20 ml of concentrated hydro- chloric acid plus 2 ml of concentrated nitric acid and diluted to 100 ml without filtration.206 SCHOLES THE APPLICATION OF ATOMIC-ABSORPTION [Analyst, VOl.93 For high alloy steel, iron(II1) chloride is added before dilution to make the total amount of iron present about 1 g. The sensitivity of the procedure is comparable to those based on air - acetylene flames, but the main advantage lies in freedom from interference caused by alloying elements present in the sample solution. It has been tested with standard steels containing 0-3 to 5 per cent. of molybdenum with good correlation against certificate values. Samples containing tung- sten were included in these tests, but no instructions are given for removal of hydrolysed precipitates in order to prevent atomiser blockage. For the determination of smaller con- centrations down to 0.01 per cent., the authors suggest a preliminary concentration stage involving extraction of molybdenum from the iron into an organic solvent, either as an 8-hydroxyquinolinate or thiocyanate complex.TABLE X INSTRUMENT CONDITIONS AND SENSITIVITIES FOR MOLYBDENUM Author and instrument Davida? Special Mostyn and Cunning- ham2* Kirkbright, Smith and Westao Unicam SP9OOA Flame conditions Luminous air - acetylene, 10-cm path, 313.3 mp Luminous air - acetylene, 10-cm path, 379.8 m p Nitrous oxide - acetylene, 5-cm path, 313.3 m p Sensitivity --&zz content, weight Solution composition p.p.rn. per cent. 0.25 per cent. w/v of sample 3.3 0.13 + 2000 p.p.m. solution of aluminium volumes of dilute HCI + HNO, + 2 per cent. w/v NH,C1 1 per cent. w/v of sample 3.3 0.033 in 2 M HC1 Sample in various 1.3 - CADMIUM Wilson31 has proposed a simple method for the determination of cadmium in alloy steels.Sensitivity is high with an air - coal gas flame; the limit of detection of 04003 per cent. can be achieved without chemical separation. The l-g sample is dissolved in aqua regia and diluted to 100 ml without filtration. If tungsten is present in the sample, 10 ml of phos- phoric acid must be added at the dissolution stage. Absorption measurements are made with the 228-8-rnp resonance line ; corrections are made for apparent absorption effects caused by iron by subtracting measurements made at the cadmium line 231.1 mp, which is known to be non-absorbing. The presence of iron in the sample solution gives rise to a small positive interference, but 20 per cent. each of manganese, nickel or chromium, 10 per cent.each of cobalt or copper and 5 per cent. each of molybdenum, titanium, vanadium, lead or aluminium have no effect on cadmium absorbance. TABLE XI INSTRUMENT CONDITIONS AND SENSITIVITIES FOR CADMIUM Sensitivity r 1 Equivalent and instrument Flame conditions Solution composition p.p.m. per cent. Author content, weight Wilsonal Air - coal gas, 228.8 mp, 1 per cent. w/v of sample 0.03 0.0003 Special 10-cm path in a mixture of HCI, HNO, and water (10 + 2 + 88) LEAD Progress in the determination of lead was reviewed by Thulbourne and Scholes in 1966,32 and this section is based on their paper. Elwell and Gidley33 determined lead in free-cutting steel and at the 0.005 per cent. level in chromium steel by using a d.c. operated instrument, an air - coal gas flame and the 283-3-mpApril, 19681 SPECTROPHOTOMETRY TO THE ANALYSIS OF IRON -4ND STEEL 207 resonance line.While their method appears to be free from direct interference, in the absence of lead a significant background absorption from iron was noted equivalent to 0.05 per cent. of lead. This effect could be allowed for by calibrating with solutions containing high purity iron as well as lead. Nall and Wollerton based their method on Elwell and Gidley’s work, extending its application to the analysis of stainless steel. Analytical sensitivity is increased by the addition of an organic solvent to the sample solution before spraying into the flame: the minimum detection limit is about 2 p.p.m. in the solution, equivalent to 0.005 per cent.of lead. In Australia a similar procedure based on an air - acetylene flame, is the subject of a S.A.A. pr0cedure.~4 The method is suitable for lead contents between 0.01 and 0.5 per cent.; it is stated to be free from interference from 1 per cent. each of nickel, chromium, molybdenum or silicon, and 2 per cent. of manganese. In more recent collaborative work in Australia, two highly sensitive procedures are being studied for the determination of trace amounts of lead. In the first of these procedure^,^^ the 2-g sample is dissolved in a mixture of hydrochloric, nitric and hydrofluoric acids. Thallium is added and lead and thallium precipitated as thiourea complexes. The precipitate is dissolved in acid, hydrogen peroxide added and the solution evaporated.Salts are dissolved in nitric acid and the solution diluted to 10ml. Absorbance measurements are made at 217-0rnp with an air-coal gas flame. The method is stated to be satisfactory for all types of irons and steels; the range of application is O*OOOl to 0.1 per cent. of lead. In the second procedure (unpublished), the 1-g sample solution is evaporated to fumes with perchloric acid, hydrogen peroxide added and the lead extracted by solvent extraction with a solution of dibenzyldithiocarbamate in chloroform following the method of Stobart.86 After wet oxidation of organic material the solution is diluted to 10 ml and sprayed into an air - propane flame. The proposed method is suitable for virtually all types of steel but it is not quite as free from interference as the thallium - thiourea method.Lockyer’~=,~7 procedure was tested at Hilger and Watts Ltd. with a solution preparation similar to that used for cobalt determination.lg For complex alloy steels good agreement with results by a spectrophotometric method was found, except for two samples that gave high results. These anomalous results were explained by what Lockyer has termed “a smoke effect” in the flame caused by the effect of matrix elements. Following Elwell and Gidley’s work,= Thulbourne and Sch01es~~ examined more closely the application of the original method. By using an a.c. instrument and air - acetylene flame, interference from iron is much less marked when compared with the results reported with air - coal gas and a d.c. instrument. Excellent results were obtained for free-cutting steels, with standard deviations of about &0*002 in the range 0.1 to 0.3 per cent.of lead. By increasing the sample solution concentration from 0-5 to 2 per cent. w/v the procedure may also be used to analyse mild and low alloy steel containing 0.001 to 0-5 per cent. of lead. Thulbourne and Scholes also examined the suitability of Lockyer’s procedure for the analysis of complex alloy steels. Sensitivity was found to be inadequate for the determination of contents less than 0-005 per cent., and for this purpose a concentration technique was proposed. The major part of the iron in a 2-g sample is removed by a single solvent extraction with isobutyl acetate, and the aqueous extract concentrated to give a 10 per cent. w/v solution.Chromium and nickel, which are present in samples of stainless steel after extraction, interfere in lead-absorption measurements, causing positive errors of up to 0.001 per cent. These effects have also been noted in collaborative Australian work. An attempt by Thulbourne and Scholes to determine the actual degree of interference and to apply an appropriate correction was unsuccessful. The effect of large amounts of nickel and chromium added either separately or together was inconsistent and difficult to interpret. As a practical solution to the problem it is recommended that the effect should be compensated for by calibrating with solutions containing high purity nickel and chromium in similar amounts to those present in the test samples. Alternatively, a “standard addition” technique can be used.In principle, the influence of matrix elements can be eliminated by extracting lead as lead iodide into isobutyl methyl ketone following preliminary removal of iron by solvent extraction. Several workersa ,39 have proposed procedures based upon this principle, but a double solvent extraction makes the method tedious, thus eliminating the advantage of atomic absorption over other techniques.208 SCHOLES : THE APPLICATION OF ATOMIC-ABSORPTION [AfldjJSt, VOl. 93 Clarke found an air - propane flame more sensitive than air - acetylene when using the E.E.L. instrument. The sensitivity was adequate for lead in leaded steels but was still insufficient for the direct determination of lead in cast iron (at the 0*0001 to 0-001 per cent.level). By using a scale expansion satisfactory results were obtained on steels and cast irons at the 0.002 per cent. level with the Unicam SP90 instrument, but the calibration graph showed a high background absorption, probably caused by iron and equivalent to 0-006 per cent. of lead. Similar results were obtained by Clarke on the Southern A1750 instrument with the resonance line at 21700mp and an air-acetylene flame. In this case the background absorption was equivalent to about 04035 per cent. of lead. TABLE XI1 INSTRUMENT CONDITIONS AND SENSITIVITIES FOR LEAD Author and instrument Elwell and Gidleys8 Hilger AA1 modified Nall and Wollerton Hilger A A 1 Australian Draft Standards4 Australian Draft Standards6 Techtron AA3 Australian CH/4 Committee Lockyer28 Hilger AA2 Thulbourne and Hilger AA2 ~ch01e~33 Clarke E.E.L.Clarke Southern A1750 Clarke and Cooke Unicam SP90 Flame conditions Air - coal gas, 283.3 mp Air - coal gas, 283.3 mp Lean air - acetylene, 283.3 mp, 10-cm path Air - coal gas, 217-0 mp, 10-cm path Air - propane Air - acetylene (10-6 and 1.2 litres per minute) , 283.3 mp Sensitivity -t content, weight Solution composition p.p.m. per cent. 2 per cent. w/v of sample Not stated - in dilute HCI (a) 2 per cent. w/v of Not stated sample in a mixture of HNO,, isopropyl alcohol and water (2 + 5 + 3) (for mild steel) (b) 2 per cent. w/v of sample in a mixture of HNO,, HC1, isopropyl alcohol and water (4 + 16 + 50 + 30) (for stainless steel) 2 per cent. w/v of sample in a mixture of HCI, HNO, and water (3 + 1 + 17) 20 per cent.w/v of sample Not stated - in dilute HNO, (5 + 95) after separation 10 per cent. w/v of sample 0.2 0.0002 in dilute HCIO, (1 + 3) after separation 2 per cent. w/v of sample Not stated - in dilute HC1 Not stated Not stated Air -acetylene (9 and 1-25 (a) 2 per cent. w/v of sample 0.6 0.003 litres per minute), in dilute HCl either 283.3 mp, 13-cm path (2 + 98) or (1 + 9) (b) 10 per cent. w/v of sample 0.6 0.0006 in dilute HCI (5 + 96) after solvent extraction Air - acetylene, 283.3 mp, 2 per cent. w/v of sample 0.9 0.0046 10-cm path Air - propane, 283.3 mp, As in previous sample 0.66 0.0033 10-cm path Air - acetylene, 283.3 mp, As in previous sample 0.6 0.0026 10-cm path Air - propane (6 and 0.4 As in previous sample 0.66 0.0028 litres per minute) , 283.3 mp, 10-cm path in dilute HCI (1 + 9) 217.0 mp, l0cm path As in previous sample 0.2 0.001 CONCLUSIONS Many of the more common elements present in steel can be determined easily by dissolving the sample in acid, removing insoluble matter by filtration, and diluting to a fixed volume as a preliminary to absorption measurement. There are few problems in the determinationApril, 19681 SPECTROPHOTOMETRY TO THE ANALYSIS OF IRON AND STEEL 209 of such elements as magnesium, chromium, manganese, cobalt, nickel, copper and lead when present in small amounts in cast iron, mild and low alloy steel.Molybdenum, vanadium and titanium can also be determined provided that a nitrous oxide - acetylene flame is used instead of air - acetylene. Difficulties arise in the determination of trace elements below 0.1 or 0.005 per cent., when concentration techniques are often necessary and in the determination of contents exceeding about 2 per cent., mainly because of the unsatisfactory precision of atomic-absorp- tion measurements. Alloying elements, such as nickel and chromium, present in large amounts also cause problems in trace-element determination.For widest application, it is important to select an instrument that has a high sensitivity, and a measurement precision adequate for the determination of elements in high concen- trations. Most manufacturers quote sensitivities based on results obtained with aqueous solutions free from anions and other metals. Of more importance is the performance of the instrument when dealing with solutions containing about 1 g of the steel sample in 100 ml of dilute acid.Measurement precision and calibration drift, which are both largely functions of flame stability, may be readily assessed by making repetitive measurements of the same sample solution over a period of about 1 hour. FUTURE DEVELOPMENTS Atomic-absorption spectrophotometry is an established technique for the determination of elements present in small amounts. It is capable of replacing more traditional colorimetric and titrimetric methods of analysis used in iron and steelworks laboratories. With refinement of instrument precision and careful calibration technique, it should be possible to extend the level of content measured up to about 10 per cent. Success in the determination of trace elements will depend on the development of rapid separation techniques so as to permit the maximum degree of sample concentration.Multi-separations of the type often necessary in colorimetric and polarographic procedures must be avoided, otherwise the inherent advantage of atomic absorption as a simple measuring technique will be lost. Extension of the range of elements determined will depend on instrumental develop- ments in the case of carbon, sulphur and phosphorus, and the use of hotter flames, such as nitrous oxide - acetylene, for the determination of aluminium, silicon, arsenic, niobium, tin, tantalum and tungsten. In the first group of elements an interesting indirect approach has been suggested by Kirkbright, Smith and West,40 who complexed phosphorus (and silicon) as molybdates and then measured the molybdenum content of the complexes.Work has been reported on most of the elements in the second group but most sensitivities claimed are rather poor. Brief details of a solution technique suitable for the determination of silicon in cast iron and steel have been reported by McA~life,~~ but no details are given of instrument sensitivity. This review is concerned with the analysis of iron and steel but there are other applications of atomic absorption in the iron and steel industry. Chief among these is the analysis of slag and, while there has been reference to the analysis of silicate materials in the literature, little attention has been given to the analysis of steel-making slag. In the author’s laboratory magnesium42 is determined on a routine basis after prior removal of silicon, and some success has been achieved in the determination of calcium45 and magnesium.At Broken Hill Pro- priety Co. Ltd.44 and at C.N.R.M. in Belgiuml9 procedures are being developed for the deter- mination of calcium and magnesium in sinters, slags, flue dust and a variety of raw materials but no published method details are available. It seems probable that manganese, iron and silicon might also be determined in oxide materials. The preparation of this review was made possible by the kind co-operation of Mr. B. Bach of Colville’s Ltd., Mr. W. R. Nall of the Bragg Laboratory, Department of Chemical Inspection, and Mr. W. E. Clarke of B.C.I.R.A., in making available unpublished material, also of Mr.L. McPherson of Lysaghts (Australia) Ltd. for information on the current work of the S.A.A. CH/4 Committee. Mrs. C. Thulbourne of B.I.S.R.A. also made many helpful suggestions. REFERENCES 1. 2. Slavin, W., Sprague, S., and Manning, D. C., Atomic Absor+tion Newslettev, 1964, No. 18, 1. Elwell, W. T., and Gidley, J. A. F., “ Atomic-Absorption Spectrophotometry,” Second Edition, Pergamon Press, Oxford, London, Edinburgh, New York, Toronto, Sydney, Paris, and Braun- schweig, 1966.210 SCHOLES Robinson, J. W., “Atomic Absorption Spectroscopy,” Edward Arnold (Publishers) Ltd., London, Sullivan, J. V., and Walsh, A., Spectrochim. Acta, 1965, 21, 727. Belcher, C. B., and Bray, H. M., Analytica Chim. Ada, 1962, 26, 322. Belcher, C. B., Proc. R . Aust. Chem. Inst., 1963, 30, 111.Sprague, S., and Slavin, W., Atomic Absorption Newsletter, 1964, No. 23, 8. Amos, M. D., and Willis, J. B., Spectrochim. Acta, 1966, 22, 1326. Headridge, J. B., and Hubbard, D. P., Analytica Chim. Ada, 1967, 37, 151. Mostyn, R. A., and Cunningham, A. F., Atomic Absorption Newsletter, 1967, 6, 86. Bowman, J. A,, and Willis, J. B., Analyt. Chem., 1967, 39, 1210. Trent, D. J., and Manning, D. C., in “Pittsburgh Conference on Analytical Chemistry,” 1965. Capacho-Delgado, L., and Manning, D. C., Atomic Absorption Newsletter, 1966, 5, 1. Kinson, K., Hodges, R. J., and Belcher, C. B., Analytica Chim. Acta, 1963, 29, 134. Giammarise, A., Atomic Absorption Newsletter, 1966, 5, 113. Barnes, L., Analyt. Chem., 1966, 38, 1083. Reid, J., and Goldrich, D., Colville’s Research Report, V 1212, 1967. Belcher, G., and Kinson, K., Analytica Chim. Acta, 1964,30, 483. Tyou, P., and Catoul, P., La Metallurgie, 1966, 98, 271. McPherson, G. L., Price, J. W., and Scaife, P. H., Nature, 1963, 199, 371. McPherson, G. L. , in “Proceedings of the 17th Chemists’ Conference, Scarborough, 1964,” British Iron and Steel Research Association, London, 1964. Lockyer, R. L., Hilger and Watts Research Report BR 26, 1966. -, BISRA Subcommittee Report MG/DA/241/66. Kinson, K., and Belcher, C., Analytica Chim. Acta, 1964, 30, 64. Sullivan, J. V., and Walsh, A., Spectrochim. Acta, 1965, 21, 721. Kinson, K., and Belcher, C., Analytica Chim. Acta, 1964,31, 180. David, D. J., Analyst, 1961, 86, 730. Mostyn, R. A., and Cunningham, A. F., Analyt. Chem., 1966, 38, 121. Firman, R. J., Spectrochim. Acta, 1966, 21, 341. Kirkbright, G. F., Smith, A. M., and West, T. S., Andyst, 1966, 91, 700. Wilson, L., Analytica Chim. Ada, 1966, 35, 123. Thulbourne, C., and Scholes, P. H., BISRA Committee Report MG/D/344/66. Elwell, W. T., and Gidley, J. A. F., Analytica Chim. Acta, 1961,24, 71. Draft Australian Standard for the Determination of Lead in Low Alloy and Carbon Steels. Draft Australian Standard for the Determination of Lead in Iron and Steel. Stobart, J. A., Analyst, 1966, 90, 278. Lockyer, R. L., Hilger and Watts Research Report BR 26, 1966. Chakrabarti, C. L., Robinson, J. W., and West, P. W., Analytica Chim. Acta, 1966,34, 269. Dagnall, R. M., West, T. S., and Young, P., Analyt. Chem., 1966, 38, 358. Kirkbright, G. F., Smith, A. M., and West, T. S., Analyst, 1967, 92, 411. McAulife, J. J., Atomic Absorption Newsletter, 1967, 6, 69. Thulbourne, C., and Scholes, P. H., Rept. BISRA, MG/D/621/1/67. -- , Ibid., MG/D/453/1/68. Clift,, B., B.H.P. Technical Bulletin No. 29, 1967, 31. 1966. Received October 31st, 1967 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 16. 16. 17. 18. 19. 20. 21. 22. 23. 24. 26. 26. 27. 28. 29. 30. 31. 32. 33. 34. 36. 36. 37. 38. 39. 40. 41. 42. 43. 44.