392 SCHOLES: THE DETERMINATIOK OF TIN I N PURE IRON, MILD STEEL [VOl. 86 The Determination of Tin in Pure Iron, Mild Steel and Certain Low-alloy Steels by Cathode-ray Polarography BY P. H. SCHOLES (British Iron and Steel Research Association, Hoyle Street, Shefleld 3) A method is described in which a linear-sweep cathode-ray polarograph is used for determining tin in pure iron and mild steel. Tin is separated by co-precipitation with manganese dioxide and is determined in a base electrolyte containing peptone and ascorbic acid in 5 M hydrochloric acid. Molybdenum interferes, but, provided that the molybdenum content of the sample does not exceed 0-5 per cent., a correction can be made, so that the method may also be used for the analysis of low-alloy steels. The method is suitable for tin contents down to 0.001 per cent., and a standard deviation of 0.001 per cent.is claimed at the 0.02 per cent. level. A BRITISH Standard method for determining tin in iron and steel has been published.1 After isolation of the sulphide with molybdenum sulphide as carrier, tin is reduced with metallic aluminium in the presence of an antimony salt and is then titrated with potassium iodate solution. For irons, carbon steels and certain low-alloy steels, the sulphide separation is unnecessary and the filtered solution can be titrated immediately. This method has been shown2 to be suitable for tin contents in the range 0.01 to 0.25 per cent., and the reproducibility is stated to be within k0.003 per cent. when 0.05 per cent. of tin is present. A more precise method was required for the analysis of pure iron and mild steel containing 0.001 to 0-02 per cent.of tin, and it was considered that a polarographic approach might offer increased accuracy and sensitivity at these levels. Ferrett and Milner3 have described an attractive direct method for determining tin in steel by square-wave polarography and claim that it is suitable for tin contents as low as 0.0005 per cent.; chemical separations are unnecessary, and the valency state of the iron is unimportant. As base electrolyte, these workers used 5 M hydrochloric acid in preference to the 1 M hydrochloric acid - 4 M ammonium chloride mixture recommended by Li~~gane.~ Trials with this method in conjunction with a cathode-ray polarograph were disappointing; instrumental sensitivity for tin reductions did not approach that of the square-wave polaro- graph, and application of the method was limited to steels containing more than 0.01 per cent.of tin, with a probable standard deviation of 0.01 per cent. Some form of separation was therefore necessary for determining tin contents less than 0.01 per cent. As well as by precipitation as sulphide, tin can be separated from iron by precipitation as metastannic acid or distillation as chloride or bromide; all these techniques have been used as preliminaries to the polarographic determination of tin in steel. Allsopp and pamere115 described a method in which tin was co-precipitated with molybdenum sulphide and the reduction of stannic tin was recorded in Lingane’s base electrolyte.Mahr and Waff enschmidt6 distilled tin as tetrachloride by passing an azeotropic mixture of steam and hydrogen chloride through a solution of the sample in perchloric acid. An alternative separation procedure for alloy steel has been based on the method described by Kallmann, Liu and Oberthin,‘ in which a solution of the sample in sulphuric acid is distilled with a mixture of steam and hydrogen bromide; tin is co-precipitated from the distillate with aluminium hydroxide, and the wave for stannic tin is measured in 5 M hydrochloric acid. Co-precipitation of tin as metastannic acid with manganese dioxide from a solution of the sample in dilute nitric acid is a well established technique in non-ferrous analysis and has found particular application in the formulation of A.S.T.M.methods for analysing pig leads and special brasses and bron~es.~ As well as tin, arsenic and antimony are quantitatively precipitated, and iron, molybdenum, tungsten and bismuth are precipitated in part, GotG, Ikeda and WatanabelO first demonstrated the practical use of this separation combined with a polarographic finish for the analysis of steel. After separation, tin was reduced with metallic aluminium and was measured in Lingane’s base electrolyte ; antimonyJune, 1961 J AND CERTAIK LOW-ALLOY STEELS BY CATHODE-RAY POLAROGRAPHY 393 was also co-precipitated and was measured in a sulphuric acid base electrolyte. BrhGek later used powdered iron as reductant,ll and Rooney used the same technique for separating antimony from cast iron.12 Of the three separation procedures considered, co-precipitation with manganese dioxide is possibly the simplest and most suited to repetitive analyses.In this paper, a polarographic method is described for determining tin in pure iron, mild steel and certain low-alloy steels after preliminary separation of the tin by a co-precipitation method closely similar to that described by Rooney.12 EXPERIMENTAL SEPARATION OF TIN- Tests indicated that co-precipitation of tin with manganese dioxide was practically quantitative over a fairly wide range of concentrations of free acid. It seemed desirable, however, to adhere as closely as possible to conditions found by other workers to be suitable for the co-precipitation of antimony, so that, if necessary, this element could be determined at the same time.Babko and Shtokalo13 found that co-precipitation of antimony was complete in the pH range 1 to 7; this was confirmed by Rooney,12 who pointed out that, in order to obtain a pH in this range, the minimum amount of acid must be used to dissolve the sample and care must be taken to prevent loss of acid during solution so as to avoid hydrolysis of ferric iron. Trials with Got8, Ikeda and Watanabe’s procedurelo produced precipitates heavily contaminated with iron and difficult to filter. Rooney12 incorporated preliminary treatment with potassium permanganate to oxidise carbonaceous matter ; the eficess of permanganate was removed with hydrogen peroxide, and a controlled addition of permanganate was then made to precipitate manganese dioxide.This technique was more satisfactory, provided that the volume of solvent acid recommended was increased by 10 to 15 per cent. to prevent hydrolysis of iron salts. The degree to which iron was co- precipitated was largely dependent on the concentration of acid and the amount of perrnan- ganate used in the initial oxidation. Contamination by iron was considerably decreased when the volume of permanganate solution was kept to a minimum. The presence of man- ganese salts, as suggested by GotB, Ikeda and Watanabe,lo improved the coagulation of the precipitate, especially when it contained iron. Tests on samples of steel produced precipitates that could easily be separated on Whatman No. 541 filter-papers. Samples containing molybdenum produced precipitates that did not settle rapidly; for such samples it was necessary to pass the solution through the filter-paper two or three times to obtain a clear filtrate.POLAROGRAPHIC MEASUREMENT- Most previous workers reduced tin to the bivalent state after solution of the manganese dioxide precipitate in hydrochloric acid and hydrogen peroxide. It seemed more convenient, however, to determine tin in the stannic state, thereby avoiding the difficulties encountered in the reduction and in maintaining the element in the stannous state during polarography. The polarographic behaviour of tin in the presence of chloride has been investigated by Lingane.4314 Stannic tin produces an ill defined wave in 1 M hydrochloric acid at -0.47 volt, with a small pre-wave beginning at about -0.05 volt.As the concentration of chloride is increased, the waves are resolved into two well defined steps having half-wave potentials at -0.25 and -0.52 volt in a base electrolyte consisting of 4 M ammonium chloride and 1 M hydrochloric acid. The first wave involves a two-electron reduction of the hexachloro- stannate ion to tetrachlorostannate and is irreversible; the second wave is produced by the reduction of tetrachlorostannate to the metal. Ferrett and Milner3 reported inconsistent results for tin when Lingane’s base electrolyte was used; they used 5 M hydrochloric acid in order to obtain maximum definition. With use of a cathode-ray polarograph, I have confirmed this observation and found that peak heights are consistent for concentrations of hydrochloric acid in the range 3 to 6 M.Irreversible reductions produce characteristically rounded and rather drawn-out waves on a cathode-ray polarogram, and the second wave, which has a peak potential of -0.48 volt against the mercury-pool anode, was therefore chosen for quantitative measurement. The diffusion-current plateau of the first wave (peak potential at about -0.2 volt) provides a horizontal or nearly horizontal base line for measuring the height of the main wave for stannic tin. Antimony, if present, is reduced at a peak potential of -0.15 volt, and its wave is therefore superimposed on the first tin wave; it does not interfere with the second tin wave.394 SCHOLES: THE DETERMINATION OF TIN I N PURE IRON, MILD STEEL [VOl. 86 At first, waves were ill defined and had sloping base lines (see Fig.l), but the addition of 5 mg of peptone to the contents of the polarographic cell after de-oxygenation improved the definition considerably. The presence of a small unidentified wave at the foot of the tin wave still caused difficulties; however, when the solution was evaporated until fumes were evolved and fuming was maintained for a few minutes, this unidentified wave dis- appeared (see Fig. 2) and measurement of the peak height became much easier. Tests showed that, if fuming was continued for 5 minutes, there was some danger of loss of tin, presumably by volatilisation, and for this reason heating to fumes for 2 minutes is stipulated in the proposed procedure. Potential, volts Potential, volts Fig. 1. Typical polarogram for tin Fig. 2.Polarogram for tin in B.C.S. (0.02 per cent.) in B.C.S. No. 272 before No. 272 after heating t o fumes and inclusion of heating to fumes and addition of peptone The proposed method was tested by adding different amounts of a tin solution to 1-g samples of pure iron dissolved in 35 ml of dilute nitric acid (1 + 4) without heating. Some difficulty was experienced in preparing a standard tin solution in dilute nitric acid, but a satisfactory technique was established by dissolving AnalaR tin in dilute nitric acid (1 + 4) without heating. The tin dissolves completely in about 2 hours, and the solution is stable for several hours, provided that the room temperature does not exceed about 20" C. The percentage of tin recovered was determined by comparison with a calibration graph plotted from the results obtained with solutions of pure tin.The results are shown in Table I and indicate that the mean recovery is 97 per cent. in the range 0.01 to 0.10 per cent. of tin and between 90 and 110 per cent. for tin contents below 0.01 per cent. This is considered to be satisfactory for a method depending on co-precipitation ; it seems probable that complete recovery could be obtained by a second precipitation from the filtrate, but this would con- siderably increase the time required for a determination. In order that the method of calibration should be representative of the method of test, the calibration procedure in the pro- posed method is based on adding tin to a solution of pure iron. Similar tests were carried out on 5-g samples of pure iron dissolved in 135 ml of solvent acid, and the results are also shown in Table I.The mean recovery was 87 per cent., which can hardly be considered satisfactory for accurate determination of tin at the 0-01 per cent. level. On the other hand, use of the larger weight of sample permitted the extension of the method to a lower limit of 0.0002 per cent. of tin, and poor recovery may therefore be balanced by improved sensitivity in the range, say, 0.0002 to 0.002 per cent. of tin. METHOD REAGENTS- addition of peptone All reagents should be of the highest grade of purity obtainable. Standard tin solzttion-Transfer exactly 50 mg of AnalaR tin metal to a 500-ml calibrated flask, add 250ml of dilute nitric acid (1 + 4), and allow the metal to dissolve; maintain the temperature below 20" C.Dilute to the mark with dilute nitric acid (1 + 4), and use immediately. One millilitre of this solution is equivalent to 100 pg of tin (0.01 per cent. of tin in a 1-g sample or 0.002 per cent. in a 5-g sample).June, 19611 AND CERTAIN LOW-ALLOY STEELS BY CATHODE-RAY POLAROGRAPHY 395 PROCEDURE- Transfer 1 g of sample (see Note) to a 400-ml conical beaker, add 35 ml of dilute nitric acid (1 + 4), and allow to dissolve without heating. Add 150 ml of water and 10ml of a 10 per cent. w/v solution of manganese sulphate, insert a boiling rod, and heat to boiling- point on a hot-plate. Oxidise the solution with a 1 per cent. w/v solution of potassium per- manganate, added dropwise. When the solution has attained a deep-purple colour, add 5 drops of pennanganate solution in excess, and boil gently for 3 to 5 minutes.Reduce the solution by adding 20-volume hydrogen peroxide dropwise, and boil to remove excess of peroxide. Dilute the solution to 200 ml, if necessary, bring to the boil, and add 2 ml of the potassium permanganate solution. Boil gently for 15 minutes, remove from the hot-plate, allow the precipitate to settle, and filter the solution through a fluted Whatman No. 541 filter-paper into a clean beaker. (If the sample contains molybdenum, the precipitate settles slowly, and several filtrations may be necessary before a clear filtrate is obtained; set solutions from such samples aside for about 30 minutes before filtration.) The filtrate should be TABLE I RECOVERY OF TIN ADDED TO SOLUTIONS OF PURE IRON Tin recovered from- A r \ Tin added, l-g sample of iron, 5-g sample of iron, Yo % % 0-0005 0.001 0-002 0.004 0-005 0.01 0.02 0.10 - 0-0004, 0-0004 0.001, 0.001 0.0010, 0-0008 0.002, 0.002 - - 0.0034, 0-0036 0.0045, 0.0055 - 0.0095, 0.0095 0.0080, 0.0080 0.0095, 0.0105 - 0.0095, 0.0095 - 0.096, 0.098 - 0.095, 0-097 - 0.098, 0.099 - - 0-0168, 0.0170 perfectly clear; if it is not, re-filter it through the same filter-paper until a clear filtrate is obtained.Wash the precipitate thoroughly with hot water. Place the funnel containing the filter-paper in the neck of the original beaker, pierce a hole in the paper with a pointed glass rod, and wash the precipitate into the beaker with hot water and dropwise additions of diluted hydrochloric acid (1 + 1) and 20-volume hydrogen peroxide.To the solution add 5 ml of nitric acid and 10 ml of diluted sulphuric acid (1 + l), evaporate until fumes are evolved, and continue to fume for 2 minutes. Cool, add 10ml of diluted hydrochloric acid (1 + l), heat to boiling-point, and oxidise with 2 or 3 drops of a saturated solution of potassium chlorate. Concentrate the solution to about 10 ml, add 10 ml of water and 20 ml of hydro- chloric acid, sp.gr. 1.18, cool, and dilute to 50 ml in a calibrated flask. Add 1 g of ascorbic acid, and shake until solution is complete. Transfer 2 to 3ml of the solution to the cell of a cathode-ray polarograph (Southern Analytical Ltd.), remove oxygen by passing nitrogen for 5 to 10 minutes, add about 5 mg of peptone, and pass nitrogen until solution is complete.Measure the peak height of the second tin wave; use a start potential of -0.10 volt. (If the sample contains tungsten, a wave will be visible at about -0.6 volt, and it may be necessary to use the derivative circuit to resolve the tin wave; for such solutions, re-set the start potential at -0.30 volt.) If the sample contains molybdenum, a correction factor must be applied; multiply the apparent tin content of such a sample by the factor- 100 100 - (Molybdenum content, yo x 50) to give the true tin content. be multiplied by the factor- If the derivative circuit is used, the apparent tin content must 100 100 - (Molybdenum content, yo x 25) For other types of polarograph, the extent to which molybdenum interferes should be estab- lished as described under “Interfering Elements.”396 SCHOLES: THE DETERMINATION OF TIN IN PURE IRON, MILD STEEL [VOl.86 With each batch of samples, examine a reagent blank and a standard steel as a check The peak current of the reagent blank is normally 0.02 to 0.06 ph. NOTE-The sensitivity and lower limit of detection can be improved by using a 5-g sample dissolved Separation is then only about 87 per cent. complete, and on the calibration graph. in 135 ml of 20 per cent. v/v nitric acid. i t is therefore essential to plot a separate calibration graph based on a 5-g sample. CALIBRATION- Prepare solutions by dissolving l-g portions of pure iron having a low tin content in dilute nitric acid (1 + 4). To these solutions add amounts of standard tin solution suitable for the construction of two calibration graphs to cover the ranges 0.001 to 0.01 per cent.and 0.01 to 0.10 per cent. of tin. The volume of dilute acid used for dissolving the iron should be decreased by an amount equal to the volume of tin solution added, so that the total volume of final solution is always 35 ml. Treat these solutions as described under “Procedure.” INTERFERING ELEMENTS IRON- If the amount of co-precipitated iron exceeds a few milligrams, reduction of ferric iron at zero potential gives rise to a large standing current, which cannot be fully “backed-off.” This can be avoided by reducing the iron to the ferrous state with ascorbic acid, but even then, excessive amounts of ferrous iron tend to produce a sloping base line to the tin wave and may cause difficulties in polarographic measurement at high instrumental sensitivity. Potential, volts Potential.volts Fig. 3. Effect of molybdenum on tin Sample contained 0.01 per cent. Fig. 4. Effect of tungsten on tin wave. Sample (B.C.S. No. 277) contained 0.003 per cent. of tin and 0.12 per cent. of tungsten wave. of tin and 0.5 per cent. of molybdenum ARSENIC- There is no arsenic wave when the value selected for the start potential is more negative than -0.05 volt, but the presence of arsenic distorts the tin wave and causes high results; the extent of interference is dependent on the start potential. To prevent this interference, arsenic is oxidised to the quinquivalent state with potassium chlorate after the solution has been fumed; quinquivalent arsenic is not reduced a t the mercury electrode.MOLYBDENUM- In concentrated hydrochloric acid, molybdenum is reduced at about -0.25 volt, pro- ducing a small rounded wave ; its presence leads to base-line difficulties, which are particularly serious when the concentration of tin is low and the ratio of molybdenum to tin is high. Molybdenum distorts the shape of the first tin wave, which becomes more rounded and less well defined, and the slope of the base line of the second wave progressively increases with the concentration of molybdenum. This is indicated in Fig. 3, which shows the effect of 0.5 per cent. of molybdenum in a sample containing 0.01 per cent. of tin. To determine the effect of molybdenum on the peak height of the tin wave, various amounts of the element were added to solutions containing 1 g of pure iron and four different levels of tin.TheJune, 19611 AND CERTAIN LOW-ALLOY STEELS BY CATHODE-RAY POLAROGRAPHY 397 recoveries of the tin are shown in Table I1 and indicate that molybdenum exerts negative interference. At each level of tin, increasing amounts of molybdenum decrease the height of the tin wave by a value approximately proportional to the amount of molybdenum added, Le., for each 0.1 per cent. of molybdenum, the peak height is decreased by 5 per cent. A correction can therefore be applied to the apparent tin content to correct for interference from molybdenum, provided that the molybdenum content does not exceed 0.5 per cent. ; at higher contents of molybdenum interference is inconsistent and difficult to measure. A similar experiment, in which the derivative circuit was used, indicated that the peak height is decreased by approximately 2-5 per cent.for each 0.1 per cent. of molybdenum, but this decrease in interference is offset by a considerable reduction in sensitivity. TABLE I1 EFFECT OF MOLYBDENUM ON RECOVERY OF TIN ADDED TO 1-g SAMPLES OF PURE IRON % % % % Molybdenum added, Tin added, Tin found, Mean recovery, 0.019, 0.019 0.092, 0.094 0.0035, 0.004 0.009, 0.0085 0.018, 0.018 0.086, 0,088 0.003, 0.0035 0-0075, 0.008 0.0165, 0.017 0.080, 0.082 0-003, 0.003 0.008, 0.007 0.016, 0.016 0.076, 0.078 95 93 94 88 90 87 81 78 84 81 75 75 80 77 TUNGSTEN- Tungsten is reduced a t about -0-6 volt and produces a rather poorly defined wave, which tends to merge with the tin wave when the ratio of tungsten to tin approaches 10 to 1; Fig.4 shows the effect of tungsten when the ratio of tungsten to tin was 40 to 1. If the ratio of 10 to 1 is exceeded, the derivative circuit must be used to resolve the tin wave. With this circuit it has been found possible to determine 0.01 per cent. of tin in a sample known to contain 1 per cent. of tungsten. RESULTS The proposed method was applied to three British Chemical Standard samples of pure iron and one sample of B.I.S.R.A. vacuum-melted pure iron (AH iron). The results are shown in Table 111; at the level of 0.002 per cent. of tin, the standard deviation is 0.0005 per cent. The tin contents of three of these irons were also determined on 5-g samples; the results, expressed to the nearest 0.0002 per cent., were- Sample .. . . . . B.C.S. No. 149/1 B.C.S. No. 260/1 AH iron Tin content found, yo . . 0.0014, 0.0014 0.0018, 0.0018 0.0018, 0.0022 and the standard deviation was estimated to be about 0.0002 per cent. of tin. When it had been established that the method was suitable for analysing pure iron, a series of British Chemical Standard mild steels, containing alloying elements up to a maximum of 0.3 per cent., and two low-alloy steels containing molybdenum were analysed; the results are shown in Table IV. The peak heights for some of these samples were also measured by means of the derivative circuit. It is evident from Table IV that there is good agreement between results by the proposed and established procedures, with the possible exception of B.C.S.No. 276. It was not possible to measure the peak height for B.C.S. No. 277 with the direct circuit because of interference from tungsten, but resolution was perfect with the derivative circuit. The standard deviation is 0.001 per cent. at the 0.02 per cent. level and 0.005 per cent. a t the 0.10 per cent. level.398 SCHOLES: THE DETERMINATION OF TIN IN PURE IRON, MILD STEEL [Vol. 86 TABLE I11 TIN CONTENTS FOUND BY PROPOSED METHOD IN l-g SAMPLES OF IRON Each result is expressed to the nearest 0.0005 per cent. B.C.S. certificate Sample value for tin, Tin content found, % % B.C.S. No. 149 . . . . .. 0.0005, 0.001 < 0.002 { 0.0005, O*OOl} 0.0025, 0.002, 0.0015 B.C.S. No. 149/1 . . .. 0,002 { 0.002, 0.0025, 0.0015) 0.0015, 0.0015, 0.0015 { 0.002, 0.0025, 0-002 } < 0.002 B.C.S.No. 260/1 . . . . 0.001, 0.001, 0.001 (0.001, 0.002, 0.002 } - AH iron . . . . .. Mean, 0.001 % 0.002 0.002 0-0015 TABLE IV TIN CONTENTS FOUND BY PROPOSED METHOD I N 1-g SAMPLES OF MILD AND LOW-ALLOY STEELS Certificate values for tin in B.C.S. Nos. 271, 272, 273, 274, 275, 276 and 277 are quoted to the nearest 0.005 per cent. Results for tin in the range 0.02 to 0.10 per cent. are expressed to the nearest 0.001 per cent. and those below 0.02 per cent. to the nearest 0.0005 per cent. Tin content found- A r \ B.C.S. certificate value for- No. of Sample tin, moIybdenum, tungsten, determinations circuit), circuit), Yo % % % % B.C.S. No. 271 . . B.C.S. No. 272 . . B.C.S. No. 273 . . B.C.S. No. 274 . . B.C.S. No. 275 . . B.C.S. No. 276 .. B.C.S. No. 277 . . B.C.S. No. 239/1 . . B.C.S. No. 239/2 . . B.C.S. No. 218/2 . . B.C.S. No. 219/2 . . B.C.S. No. 256 . . 0.1 10 0.020 0.065 0.020 0.040 0.010 < <0.005 0.06 0.024 0.035 -0.025 -0.013 0.19 0.17 0-05 0.07 0.10 :0.01 0.02 0.03 0.03 0.03 0.43 0.54 0.02 0.01 0.28 0.04 0.05 0.20 0.12 - 6 0.110 - 3 0.020 0-019 3 0.068 - 3 0.025 0.023 3 0.044 - 6 0.017 0.018 3 - 0,003 3 0.049 - 3 0.022 0.020 3 0.033 - 3 0.026 0-023 0.012 6 0.012 CONCLUSIONS Separation of tin by co-precipitation on manganese dioxide is a suitable preliminary to the determination of this element by cathode-ray polarography. The method offers considerable improvement in sensitivity and accuracy over the British Standard volumetric procedure and was primarily designed for determining 0-001 to 0.02 per cent.of tin in pure iron and mild steel. Higher contents of tin can be determined in samples soluble in dilute nitric acid, provided that the molybdenum content does not exceed 0.5 per cent. Tungsten interferes with the measurement of the tin wave when the tungsten content is more than ten times the tin content. If this ratio is exceeded, it is possible to obtain reasonably accurate results by using the derivative circuit of the instrument to measure the tin wave. The sensitivity of the method may be further improved by using a larger weight of sample; the theoretical limit of detection is extended down to 0-0002 per cent. of tin in a 5-g sample. Poor recovery with large weights of sample is a disadvantage, but this is balanced by an improvement in sensitivity. Previous workerslO J1 $2 have shown that co-precipitation on manganese dioxide is also suitable for isolating antimony from iron and steel, and it should therefore be possible to determine this element in an aliquot from the solution after separation. Base electrolytes suitable for the polarographic determination of antimony have been described.lOJ1LYOl J AND CERTAIN LOW-ALLOY STEELS BY CATHODE-RAY POLAROGRAPHY 399 1. 2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. REFERENCES “Determination of Tin in Iron and Steel,” British Standard 1121 : Part 20: 1951. B.I.S.R.A. Report No. MG/D/185/59. Ferrett, D. J., and Milner, G. W. C., Analyst, 1956, 81, 193. Lingane, J. J., J . Amer. Chem. SOC., 1945, 67, 919. Allsopp, W. E., and Damerell, V. R., Anal. Chevn., 1949, 21, 677. Mahr, C., and Waffenschmidt, K., Arch. Eisenhuttelzzer., 1960, 31, 221. Kallmann, S., Liu, R., and Oberthin, H., Anal. Chem., 1958, 30, 485. “A.S.T.M. Methods for Chemical Analysis of Metals,” American Society for Testing Materials, American Society for Testing Materials, op. cit., p. 339. GotB, H., Ikeda, S., and Watanabe, S., Japan Analyst, 1954, 3, 320. BrhGek, L., Hutn. Listy, 1957, 12, 140. Rooney, R. C., Analyst, 1957, 82, 619. Babko, A. K., and Shtokalo, M. I., Zavod. Lab., 1955, 21, 767. Lingane, J. J., I n d . Eng. Chem., Aizal. Ed., 1943, 15, 583. Philadelphia, 1956, p. 456. Received Decenzbev 6th, 1960