APRIL 1950 Vol. 75, No. 889 Method for the Estimation of Small Amounts of Carbon in Steel BY K. GARDNER, W. J. ROWLAND AND H. THOMAS SYNOPSIS-A brief survey is given of the methods available for the determina- tion of small amounts of carbon (less than 0-03 per cent.) in steel. The method described utilises a combustion procedure in which the carbon dioxide produced is absorbed in baryta solution. The decrease in electrical conductivity of this solution is related to the carbon content. The difficulties which arise are discussed and necessary precautions are described. A device for the introduction of the sample into the combustion zone, without ingress of air, is shown. Adsorption of carbon dioxide by the pre-ignited boats is avoided by storing over soda-asbestos. A high efficiency of absorption in the baryta solution is attained by the use of a cell having a long absorption path and by addition of a non-ionic wetting agent, Lissapol N, to the absorbing solution.For steels containing less than 0.03 per cent. of carbon, the average deviation from the mean value does not exceed f0.0005 per cent. The determination occupies 40 minutes, half the time needed for similar methods previously published. The apparatus can be constructed from laboratory equipment, and is robust and comparatively inexpensive. The method has been in satisfactory operation for over two years, and the results compare favourably with those obtained by the standard low-pressure method. IN the course of an investigation in these laboratories, it was necessary to develop a method for the accurate determination of the carbon content of steel containing less than 0.03 per cent.of carbon. Three types of method were considered. (a) Classical microgravimetric procedures-These have the disadvantages that a skilled operator is required and that the time for an estimation is long. ( b ) Low-pressure combustion method-This method may be considered the standard procedure at the present time for the determination of carbon in low-carbon steels. The basic principles were devised by Yensenl in 1920. A system is evacuated and the sample of 173174 GARDNER, ROWLAND AND THOMAS: METHOD FOR THE ESTIMATION v o l . 76 steel then heated and ignited in purified oxygen. The carbon dioxide formed is collected in a liquid air trap and subsequently determined by expanding the gas into a known evacuated volume and observing the pressure. Improvements in the technique were made in the methods of Wooten and Guldner,Z Murray and Ashley3 and Murray and Niedrach.4 In the method of Stanley and Yensen: it is claimed that the average time for an estimation is 20 minutes and that an accuracy of f0.0005 per cent.is obtained. Nesbitt and Hendersons use a modification of this method in which the steel is burned in a stream of purified oxygen and the evolved carbon dioxide collected in a special absorber containing caustic soda. The solution is then acidified and the volume of gas evolved is measured at low pressure. All these methods have the disadvantage that the apparatus required is complicated and relatively expensive.Attention was, therefore, focussed on the third type of method. (c) Combustion of the sample in 0xyge.n followed by absorption of the carbon dioxide in barium hydroxide solution-The reduction in concentration of the barium hydroxide can then be measured, either by titrating the residual base with standard acid, as in Kalina and Joseph’s method,’ or by measuring the decrease in conductance of the solution. Cain and Maxwells used the conductimetric method for the rapid determination of carbon with an accuracy of _tO.OI per cent. Bolliger and Treadwellg used a conductimetric method for the estimation of carbon in aluminium, the carbon dioxide being absorbed in caustic soda solution. After much work had been carried out by the authors on this method, their attention was drawn to a reference in “Reports on the Progress of Applied Chemistry’’ to the work of Ericcson.lo Using an absorption cell similar in design to that of Kalina and Joseph, he was able to determine carbon contents with a n accuracy of f0.0005 per cent.The average time for each analysis, however, was greater than 1 hour. APPARATUS Oxygen is admitted from a cylinder, through the flow gauge A and thence is passed through sodium hydroxide B and sulphuric acid C. Carbon monoxide and hydrocarbons are converted to carbon dioxide and water by catalytic combustion of the gas in furnace D, which is packed with palladised asbestos and operated at 400” C. Any water and carbon dioxide formed are absorbed in U-tubes E and F, which contain anhydrous calcium chloride and soda-asbestos respectively. These are followed by a Pregl type bubble-counter G, containing sulphuric acid, which serves to indicate any leakage in the train.It is followed by a glass wool filter H and a 2-litre aspirator bottle J, which is connected to the combustion furnace by tap K. The aspirator provides a reservoir of oxygen for use during combustion of a sample. All joints between the purification furnace D and the combustion furnace are glass to glass, the .rubber tubing merely acting as a seal. A mullite combustion tube, L (type triangle H6/T3411), is used, and after combustion the gases pass through a manganese dioxide tube M, which absorbs any sulphur dioxide formed. This tube is connected through rubber tubing and clip N to the absorption vessel P.Morgan 28 combustion boats are used. The entrance to the combustion tube is illustrated in Fig. 2. This part of the apparatus provides a gas-tight joint from the train to the furnace and at the same time minimises the volume of air introduced into the tube on insertion of the sample of steel. The steel cone R is fixed to the end of the tube by means of a ring of large-diameter heavy-section rubber tubing U. The steel cone is fitted to a B40 standard cone end-cap S, which is sealed at the extremity by glass tube V with rubber tubing. Extending to the narrow end of this end-cap is a steel tube T, which slides into the side-arm of the furnace tube. The sample boat W may be inserted into the position shown in the diagram, by removing the glass cone. Oxygen can then be passed through tube T until any atmospheric carbon dioxide is completely removed.The glass tube V can then be quickly removed and the boat pushed into the hot zone of the furnace tube with a steel rod, while oxygen passing out through T prevents admission of any air. The tube V is replaced and the combustion of the sample carried out. The absorption vessel, details of which are shown in Fig. 3, was prepared by modifying a Friederich type absorption vessel (Pyrex catalogue type S31). The gas passes through the central spiral which contains the conductivity electrodes (A, A) and bubbles into the absorbent through a sintered-glass bubbler €3. The bubbler was prepared by the method of Stone and Weissll using 40 to 60 mesh Pyrex glass. The small gas bubbles produced follow a 26-inch path round the spiral and out through C.In order to measure the conductance, air is forced through a soda-asbestos tube into C by means of a rubber bulb, forcing the A diagram of the apparatus is shown in Fig. 1.April, 19501 OF SMALL AMOUNTS OF CARBON IN STEEL 7 t U m d 176176 GARDNER, ROWLAND AND THOMAS: METHOD FOR THE ESTIMATION [VOl. 75 liquid back through the sinter into the measuring cell. Bubbles of purified air rising through the sinter serve to stir the liquid completely. The clip N is closed and the conductance measured with a Cambridge conductivity bridge. The cell is maintained at 20.0 f 0.1" C. by means of a water-bath. REAGENTS In order to obtain a suitable range on the conductivity bridge, for steels containing about 0.03 per cent.of carbon, it is desirable to use a concentration of 1 g. of hydrated baryta, Ba(OH),.$H,O, per litre. In addition, to ensure complete absorption, the absorbent contains 10ml. of 2 per cent. v/v Lissapol N* solution per litre. The exact concentration of the baryta is not important. To facilitate delivery of a standard volume into the absorption vessel, the absorbent is stored in an aspirator bottle connected to an autornatic pipette. The bottle is protected by a guard tube containing soda-asbestos. PROCEDURE A suitable volume of reagent in the absorption vessel is 50ml. Place 2 g. of the sample in a combustion boat and insert into the cool end of the com- Empty and dry the outer surfaces bustion tube. Close the tube by means of the glass cone.\ R U w \ Fig. 2. Details of entrance to combustion tube of the absorption vessel and blow purified air through the sinter to remove any liquid from the inside of the cell. Connect the cell to the apparatus and pass oxygen through at: a rate of 2.7 to 3-0 litres per hour. Adjust the temperature of the water-bath to 20 f 0.1" C. and measure the conductivity at 5-minute intervals until the bridge reading is constant. Isolate the absorption vessel by means of clip N. Remove the stopper V from the end of the glass cone and push the boat into the hot zone of the furnace with a clean steel rod. The temperature of the furnace should be 1200" to 1250" C. Close the end of the cone and increase the oxygen flow to prevent the pressure in the apparatus falling unduly. If the sample consists of drillings, combustion will start almost immediately; with sheet samples there is a lag of about 1 minute.In both cases the time for burning a 2-g. sample is 2 .to 3 minutes. During combustion the rate of oxygen flow increases to 7 to 10 litres per hour. When combustion is complete, as shown by a decrease in oxygen flow, open the clip which isolates the absorption cell and reduce the oxygen flow to 2.7 to 3.0 litres per hour. Determine the conductivity of the baryta solution after 15 minutes and thereafter at 5-minute intervals until the bridge reading is constant. With carbon contents of less than 0.01 per cent., a constant reading will be obtained in 20 minutes. The method of correlating the difference in bridge reading with carbon content is given below.Barium carbonate may be removed, when necessary, from the sintered glass bubbler by means of dilute hydrochloric acid, followed by washing with water. Run in 50 ml. of baryta solution from the pipette. With higher carbon contents 30 minutes is usually required. Lissapol N is an ethylene oxide condensation product prepared by Imperial Chemical Industries Ltd.April, 19503 OF SMALL AMOUNTS OF CARBON IN STEEL 177 EXPERIMENTAL The efficiency of the absorption was first checked with a known quantity of calcium carbonate as a carbon standard. This method proved successful for carbon contents greater than 0.02 per cent. For lower carbon contents it was difficult to weigh out the small amounts of calcium carbonate required and, therefore, sucrose was used. A standard solution of sucrose was made such that 1 ml.was equivalent to 0.4mg. of carbon. Known volumes of this solution were delivered from a micro-burette either on to a porous boat lid or into a non-porous boat and these were placed inside an ordinary combus- tion boat. Water was removed, either by desiccating over phosphorus pentoxide or drying in an oven at approximately 90” C. These samples were then ignited as for a steel. The first experiments were carried out with an absorbent containing 1 g. of hydrated barium hydroxide per litre with no wetting agent. With gas flows below 2.5 litres per hour, the time for sweeping out carbon dioxide from the combustion tube was excessive. When the flow was increased to 2.7 to 3.0 litres per hour, absorption was incomplete as shown in Fig.4 (A) in which the difference in conductivity for this absorbent is plotted against carbon content. A theoretical curve for complete absorption is also shown, Fig. 4 (C). This theoretical curve was constructed from data relating conductance and concentration obtained from Gmelin’s Handbook12 and the “Handbook of Chemistry and Physics.”13 These data were checked by measuring the con- ductance of baryta solutions of known concentra- tions, the cell constant being determined by measurement of standard potassium chloride solutions. The absorption was increased by the introduc- tion of butyl alcohol into the absorption reagent, and by doubling the concentration of barium hydroxide, but was still incomplete. Finally, a non-ionic surface-active agent, Lissapol N, was incorporated in the reagent.The relation between conductivity difference for this solution and percentage carbon is shown in Fig. 4 (B). I t is seen that for carbon contents below 0.03 per cent., the difference between this curve and the theoretical curve does not exceed 0.0005 per cent. of carbon. It was considered that curve B, Fig. 4, was sufficiently close to the theoretical curve to warrant its use as a calibration curve for the met hod. Fig. 5 shows the percentage absorption of the two reagents as a function of carbon concentration. With carbon contents greater than 0.03 per cent., the absorption falls off rapidly. Fig. 3. Details of absorption vessel Owing to the small amount of carbon estimated, care had to be taken in the preparation of the samples of steel, e.g., all surface grease had to be carefully excluded and samples were never touched by hand after preliminary cleaning by abrasion.It was found that if the combustion furnace was switched off and allowed access to the air, a considerable period (2 to 3 hours) was required, on re-heating, to attain a steady reading on the conductivity bridge. This was probably due to carbon dioxide adsorption on the walls178 GARDNER, ROWLAND AND THOMAS: METHOD FOR THE ESTIMATION [vd. 75 CARBON CONTENT (PERCENTAGE ON 2 g.) Fig. 4. Relationship between carbon content Fig. 5. Effect of carbon content on absorption. and conductivity difference. Curve A, Lower curve, solution containing 1 g. of solution containing lg. of Ba(OH),.SH,O Ba(OH),.8H20 per litre.Upper curve, per litre. Curve B, solution containing solution containing lg. of Ba(OH),.8H20 lg. of Ba(OH),.8H20 plus 0.2 ml. of plus 0.2 ml. of Lissapol N per litre Lissapol N per litre. Curve C, theoretical relationship DETERMINATION OF CARBON CONTENT OF B.C.S. CARBON STEELS B.C.S. Steel No. Weight of sample, No wetting agent 218 0.1 200 0.2087 0-2377 156 0-2147 0.2620 213 0-2264 158 0.2289 5- Carbon expressed on basis of assumed 2-g. sample Calculated, Found, A I 7 % Yo 0*0092 0.0090 0.0161 0.0155 0.0183 0.0185 0.0245 0-0245 0.0300 0.0292 0.0414 0.0412 0.0531 0.0635 0-02 per cent. of Lissapol N present 0.2010 0.0155 0.0 160 218 0.2045 0.0234 0.0230 156 0.2062 0.0377 0.0380 213 158 0.2010 0.0467 0.0470April, 19501 OF SMALL AMOUNTS OF CARBON I N STEEL 179 of the combustion tube.During routine analysis, this difficulty was avoided by maintaining the combustion furnace overnight at a temperature of 700" to 800" C., while passing a very slow trickle of oxygen through the apparatus. The combustion boats had to be ignited in oxygen before use and stored in a desiccator over soda-asbestos. It was found that, if the boats were exposed to the air for an appreciable time after ignition, a blank of 0.01 mg. of carbon (0-0005 per cent. on a 2-g. sample) was obtained. Because of the difficulty in obtaining low-carbon standard steels, a number of determina- tions were carried out on small weights of B.C.S. carbon steels. The results are shown in Table I. Table I1 shows the results of ten analyses on a sample of commercial low-carbon steel.These results show the expected deviation caused by heterogeneity of the steel. The average deviation from the mean is very close to the value obtained by Stanley and Yenson,6 who analysed a similar steel by the low-pressure combustion method. TABLE I1 CARBON CONTENT OF A COMMERCIAL LOW-CARBON STEEL Carbon, % . . 0.0067 0.0070 0.0080 0.0072 0.0070 0.0067 0,0075 0.0075 0.0077 0.0080 Mean 0.0073 & 0*0004(1)% Duplicate determinations were carried out on a standardised sample of low-carbon The results were steel which had been analysed by the low-pressure combustion method. as follows- Carbon, yo Low-pressure method . . . . .. . . 0.0060 f 0-0005 Conductimetric method .. .. .. 0.0065 0.0062 SUMMARY With this conductimetric method for the determination of carbon in low-carbon steels, containing less than 0.03 per cent.of carbon, the average deviation from the mean value does not exceed ~t0-0005 per cent. The determination occupies about 40 minutes, about half the time needed for similar methods previously published. The apparatus can be constructed from laboratory equipment and is robust and comparatively inexpensive. This method has been in satisfactory operation in these laboratories for over two years. The authors are indebted to the Director of the Nelson Research Laboratories for permission to publish this work. REFERENCES 1. Yensen, T. D., Trans. Amer. Electrochem. Soc., 1920, 37, 227. 2. Wooten, I. A., and Guldner, W. G., Ind. Eng. Chem., Anal. Ed., 1942, 14, 835. 3. Murray, W. M., and Ashley, S. E. Q., Ibid., 1944, 16, 242. 4. Murray, W. M., and Niedrach, L. W., Ibid., 1944, 16, 634. 5. Stanley, J. K., and Yensen, T. D., Ibid., 1945, 17, 699. 6. Nesbitt, C. E., and Henderson, J . , Ibid., 1947, 19, 401. 7. Kalina, M. H., and Joseph, T. L., Blast Furnace Steel Plant, 1939, 27, 347. 8. Cain, J . R., and Maxwell, L. C., Ind. Eng. Chem., 1919, 11, 852. 9. Bolliger, H. R., and Treadwell, W. D., Helv. Chim. Acla. 1948, 31, 1247. 10. Ericsson, G., Jernkontor. Ann., 1944, 128, 579. 11. Stone, H. W., and Weiss, L. C., Ind. Eng. Chem., Anal. Ed., 1939, 11, 220. 12. Gmelin, Handbuch anorg. chem., 8 Aufl. System No. 30: Barium, 1932, p. 123. 13. Hodgman, C . D., "Handbook of Chemistry and Physics," 31st Edition, Chemical Rubber 14. Naughton, J . J., and Uhlig, H. H., Anal. Chem., 1948, 20, 477. Publishing Co., Cleveland, Ohio, 1949, p. 1995. ENGLISH ELECTRIC COMPANY NELSON RESEARCH LABORATORIES STAFFORD September, 1949