Analyst, April, 1968, Vol. 93, @. 211-213 21 1 The Spectrographic Determination of Nickel in Molten Steels BY J. B. HEADRIDGE AND A. K. LAMBERT (Department of Chemistvy, The University, Shefield 10) A prerequisite for the direct spectrographic analysis of molten steels in industrial furnaces is the knowledge that good quality spectra can be obtained from molten steel surfaces, and that precise quantitative determinations can be made by using such spectra. The emission-spectrographic determination of nickel from a molten steel surface at 1600" C under an argon atmosphere has, therefore, been investigated. When a condensed spark between a graphite electrode and the molten steel surface was used, spectra of good quality were produced. The standard deviations in the error for the determination of 0.7 to 1.8 per cent.of nickel in molten and solid steel samples, analysed under similar conditions, were 0-045 and 0.022 per cent., respectively. THE routine analysis of molten steel is usually made by withdrawing a sample from the furnace, allowing it to solidify and, after suitable machining, subjecting it to emission- spectrographic analysis on a direct-reading instrument. This whole operation takes about 5 to 10 minutes, and, from an economic standpoint, it would be desirable to reduce the time required for an analysis. This could probably be done by carrying out an emission-spectro- graphic analysis on the molten steel directly, but a search of the chemical literature reveals that only a few studies have been made on the quality and possible analytical application of emission spectra from a molten steel surface, and most of these are of a preliminary nature.Balandin and Mandel'shtaml investigated the possibility of determining the composition of molten steel in an arc furnace without sampling, and state that this method should be practicable. The electric arc of the furnace was used as an excitation source. In experiments on a laboratory scale, Shaevich and Shubina have determined silicon in molten pig-iron2 and carbon in molten iron - carbon alloys3 by using emission spectrography with relative standard deviations of about 5 per cent. With a pulsed laser source and a large Littrow spectrograph, Runge, Bonfiglio and Bryan4 have obtained satisfactory calibration graphs for the deter- mination of 9 to 24 per cent.of nickel and 13 to 25 per cent. of chromium in three samples of molten stainless steels. The preliminary studies on the spectrographic analysis of molten steels made by Hilger and Watts Ltd. and by the British Iron and Steel Research Association are reviewed by Scholes and Williams6 As few results are given in the above papers on the precision of spectrographic methods of analysis of molten steels, the authors have investigated the spectrographic determination of 0.7 to 1.8 per cent. of nickel in five steels at room temperature and in the molten state at 1600°C. These results are now reported. EXPERIMENTAL A molten steel surface was produced by the induction melting of a 1-8-lb cylindrical sample of steel in a magnesia crucible. An inert atmosphere was maintained above the molten surface by passing argon into a Vitreosil hood positioned over the crucible.This refractory cover was constructed with three holes in it. Light from the molten surface was directed through one hole to a continuous optical pyrometer, which allowed the temperature to be kept constant at 1600" C. Visual inspection of a spark could be made through the second hole and the third hole permitted light to enter a large quartz spectrograph. With the electrode gap set at 3*8mm, 15-kV condensed sparks were struck between the molten surface and two graphite electrodes. Light from one of the sparks, after passing through a quartz lens, was deflected by a surface-aluminised mirror along the optical axis of the spectrograph. Before entering the spectrograph the light beam passed through another quartz lens, which produced a de-focused image of the spark on the slits.0 SAC and the authors.212 HEADRIDGE AND LAMBERT : THE SPECTROGRAPHIC [ArtUlySt, VOl. 93 The molten steel surface was half an inch below the top of the crucible and, when the solid samples were analysed at room temperature, a cylinder of steel was placed in the crucible so that its surface was the same distance below the crucible top; otherwise the apparatus was identical. The spectra from the solid and molten samples were photographed on Kodak B10 plates over the range 2800 to 5000 A. The exposure time for the solid samples was 3 minutes but, with the molten samples, the intensity of the spark was less and the exposure time was increased to 5 minutes.The spectra of the five samples, and that from an iron spark in conjunction with a rotating stepped sector, were recorded on the same plate. After developing, fixing, washin and drying in the recommended manner, the optical densities of the nickel line at 3414-77 1 and the iron line at 3417.84 were recorded with a Joyce Loebl microdensitometer for each spectrum from the steel samples. The optical densities for each step of the iron line at 3440.61 A on the spectrum from the iron spark were recorded, and the plate-calibration curve constructed. The log intensity ratios for the line pair nickel 3414.77 - iron 3417.84 A for the five samples were obtained from the plate-calibration curve and plotted against nickel concentration to obtain a calibration graph.The same procedure was used for both solid and molten samples. RESULTS The five steels and their nickel contents are shown below. Steel .. .. .. .. EN361 EN362 EN363 EN24 EN364 Nickel content, per cent.. . . . 0.77 0.98 1.23 1.40 1-74 Analysed samples of these steels were a gift from the English Steel Corporation Ltd. The calibration graphs for both the solid and molten steel samples are shown in Fig. 1. I / / I *60 t Log of nickel concentration Fig. 1. Calibration graphs of log(relative in- tensity for nickel/ (relative intensity for iron) against log (concentration of nickel): 0 solid samples, 0 liquid samplesApril, 19681 DETERMINATION OF NICKEL IN MOLTEN STEELS 213 DISCUSSION It can be seen from Fig. 1 that the method is more sensitive for molten than for solid steel, as the slope of the calibration graph for the molten steels is the steeper of the two.When the results for the molten steels were re-plotted as intensity ratio against nickel con- centration, a straight line passing through the origin was obtained. The standard deviation in the error in nickel concentration was 0.045 per cent., which corresponds to a relative standard deviation of 4.5 per cent. at a nickel concentration of 1 per cent. The standard deviation is based oh five results, one for each of the five standards. The error in nickel concentration was expressed by c (observed) - c (calculated), where the values of c (calculated) were points exactly on the straight-line calibration graph of intensity ratio against nickel concentration.A similar type of plot for the solid steels must also pass through the origin and give a curve concave to the nickel-concentration axis. This curvature probably resulted from self- absorption for the nickel line in the spectra from the solid samples. When a similar statistical treatment was made of this curve, the standard deviation in the error in nickel concentration was 0.022 per cent. This study has shown that the spectrographic determination of nickel in molten steel is feasible and that the precision of the results is acceptable. The precision of the results for the molten steel samples could no doubt be improved if a satisfactory optical system could be devised to allow the exposure time to be appreciably reduced. There is every reason to believe that many other elements with low volatilities at 1600" C could be determined in molten steel by emission spectrography (Note).If this method were to be applied to the spectrographic analysis of molten steel in an industrial furnace, some way would have to be found for isolating a slag-free pool of molten steel under an argon atmosphere on the side of the furnace. This, however, is a problem for the mechanical engineer rather than the analytical chemist. The authors are now hoping to obtain satisfactory calibration graphs for carbon in molten steels with similar apparatus and a vacuum spectrograph. NOTE-The authors started their investigations by trying to obtain a suitable calibration graph for 0-4 to 1.7 per cent. of manganese in molten steel with a medium quartz spectrograph.Although a satisfactory calibration graph was obtained for manganese in five solid steel samples, consistent results for the determination of manganese in the molten samples could not be obtained. It was found that the concentration of manganese in the molten steels was, in fact, decreasing with time, for the vapour pressure of manganese above molten steel at 1600" C is appreciable, and the manganese vapour was being removed from the system by the argon stream. We are indebted to Mr. Faine of this department and his workshop staff for building the electrode assembly. We gratefully acknowledge the receipt of grants for this work from the Science Research Council, English Steel Corporation Ltd., Firth-Brown Ltd., Guest, Keen and Nettlefold Steel Co. Ltd., Steel Company of Wales Ltd. and United Steel Companies Ltd. REFERENCES 1. 2. 3. 4. 5. Balandin, V. N., and Mandel'shtam, S. L., Zav. Lab., 1967, 23, 645. Shaevich, A. B., and Shubina, S. B., Ibid., 1962, 28, 447. Shaevich, A. B., Mel'nikov, S. I., and Danilevskaya, V. V., Ibid., 1966, 31, 169. Runge, E. F., Bonfiglio, S., and Bryan, F. R., Spectrochim. Actu, 1966, 22, 1678. Scholes, P. H., and Williams, R. V., British Iron and Steel Research Association Report MG/D/ Received September 18th, 1967 335/66.