482 WHITE AND SCHOLES: DETERMINATION OF CARBON I N [Ana&St, VOl. 91 Determination of Carbon in Steel by a Dynamic Infrared System BY G. WHITE AND P. H. SCHOLES (Bvitish Iron and Steel Research Association, Metallurgy Division, Hoyle Street, Shefield 3) A simple, automatic apparatus has been developed for the rapid deter- mination of carbon in steel. I t is based upon the continuous measurement of carbon dioxide evolved during the high temperature combustion of steel in oxygen by using a specially designed infrared gas analyser and integration system. When this is used in conjunction with a conventional resistance-tube furnace, the spced of the determination varies from 40 to 55 seconds for mild and low alloy steels, and slightly longer for highly alloyed materials. CONSIDERABLE progress has been made during recent years in the development of com- bustion apparatus for determining carbon in steel.There are now several analysers available commercially in Europe and North America.1 These new instruments are entirely automatic. The operator has only to place a weighed sample into a refractory container and insert this into a combustion furnace. From this stage the determination proceeds automatically, the operator being required only to read a meter and perform a simple conversion to the percentage of carbon. The commercial analysers can be classified into three groups according to the technique used to measure the carbon dioxide evolved during the high temperature combustion of steel samples in oxygen. These measuring systems are (a) electrochemical, mainly those of German origin, (b) thermal conductivity (used in instruments made in North America), and (c) infrared absorption (used in a British and a French instrument).Current developments in the tech- niques available for the rapid sampling of molten steel permit the use of these analysers for process control purposes, and another rather arbitrary grouping is possible in terms of instru- ment time. For the electrochemical instruments this is normally 2& to 4 minutes, being rather longer for the French infrared analyser, and 60 to 90 seconds for the thermal-con- ductivity instruments and the British infrared analyser. With instruments of the latter group it is, therefore, possible to determine the carbon content of, for example, large open- hearth furnaces at 5-minute intervals.With the exception of the Canadian Thermocarb,2 the commercial process analysers incorporate a static measurement system in which the combustion gas is first collected before measurement. One of the disadvantages of this approach is that conditions must be carefully pre-arranged to ensure that combustion is completed before measurement of carbon dioxide takes place. For maximum speed it seemed preferable to use a dynamic approach, in which the carbon dioxide content of the combustion gas is monitored continuously by using a fast-response detector with electronic integration. Previous work by the authors had shown that remarkably fast combustion of samples, in the form of turnings and solid pins, could be obtained with an inexpensive resistance-tube furnace.In the system used, combustion takes place in the presence of excess oxygen, and the combustion gas is pumped from the furnace at a fixed rate. Measurement of the evolved carbon dioxide by infrared absorption is an attractive alternative to thermal-conductivity measurement, and is free from many of the difficulties arising from the instrumental instability of the latter technique. The principle is not new, it was first proposed by Lay3 in 1955, Le Controle de Chauffe of Paris published a brochure in 1962 describing a prototype carbon analyser based upon infrared measurements, and in the following year, Tipler4 described an analyser designed principally for determining low carbon contents in steel and silicon - iron.The latter instrument has now been modified by Hilger-I.R.D. Limited for more general appli~ation.~ Both instruments incorporate static measurement systems of the type mentioned earlier. In early 1963, an infrared analyser was kindly made available to the authors by Hilger-I.R.D. Limited for initial experiments on the dynamic measurement of the evolved carbon dioxide. In this paper, the development of a dynamic carbon-in-steel analyser, the Dynacarb, is described. The primary consideration in design is that the apparatus should be inexpensive, have a rapid throughput time and be capable of application to all types of steel. Sometimes this is unnecessary as the percentage of carbon is shown directly.August , 19661 STEEL BY A DYNAMIC INFRARED SYSTEM 483 EXPERIMENTAL Preliminary experiments were carried out with a standard single range, direct-reading infrared analyser supplied by Hilger-I.R.D. Limited. Combustion patterns for various types of steel were recorded with a moving-chart potentiometer, and it was found that, with furnace temperatures in excess of 1300" C and high oxygen flow-rates, combustion of most samples was complete in less than 1 minute. The recorded pattern of carbon dioxide concentration versus time was extremely symmetrical, but from attempts to relate carbon content to integrated peak area it was clear that the response of the analyser was not adequate to detect all of the carbon dioxide present in the gas. Experiments designed to slow down the rate of carbon dioxide evolution and its subsequent introduction to the analyser were not successful, and the obvious requirement was an analyser with a much more rapid response- time.A second infrared analyser was obtained that had a response-time of about 200 milli- seconds, and was suitable for measuring concentrations of carbon dioxide up to 12 per cent. by volume. This instrument, which incorporated a high speed recorder suitable for obtaining combustion patterns, had previously been used as a prototype for experimental work in determining carbon dioxide in respiratory gases. ( a ) 0.29 percent. of carbon Time, seconds Fig. 1. Typical combustion patterns. Temperature 1350" C, flow-rate 900ml per minute In general, the recorded patterns showed a reasonable degree of symmetry, but, in a few instances, irregular combustion conditions caused minor rapid fluctuations in the carbon dioxide content of the evolved gases.The maximum carbon dioxide concentration at any one instant did not exceed 8 per cent. by volume. Typical combustion patterns are presented in Fig. 1. Measurement of the area under the peak gave reproducible values, provided that the oxygen flow-rate through the analyser was maintained at a constant level. Mixing tubes of various capacities were inserted into the flow system at the entrance side of the analyser, in order to decrease the proportion of carbon dioxide in the combustion gas. The values obtained were similar to those obtained in the absence of expansion tubes, confirming that instrument response was sufficient to detect all of the carbon dioxide passing through the analytical cell.484 WHITE AND SCHOLES: DETERMINATION OF CARBOX IS [L4naZyst, Vol.91 These experiments showed that dynamic infrared measurement of carbon dioxide was feasible, and an apparatus based upon these principles was designed in conjunction with Hilger-I. R.D. Limited. 10- Resistance tube O ~ I I I I I I Infrared anal yser Electronic linearising and integrating device Flow controller Fig. 2. Block diagram of prototype apparatus AFPARATUS- A block diagram of the Dynacarb apparatus is shown in Fig. 2. Gas analyser-The gas analyser is a modified, general-purpose instrument incorporating a specially designed analytical cell for the measurement of high velocity gas streams. I t is suitable for use with carbon dioxide concentrations up to 10 per cent.by volume in oxygen, and has a response-time better than 200 milliseconds. As the electrical output of infrared analysers is logarithmic in function, it was necessary to incorporate a linearising circuit in order to produce a signal that could be integrated and displayed on a meter. This was accomplished by dividing the curve relating meter-reading to voltage-output into a number of segmenk6 Each segment has an associated printed circuit card with a silicon diode and an adjustable potentiometer. Bias is applied to each diode to ensure that it will pass only current higher than the voltage of selected ordinates of the meter reading - voltage relationships; the voltage of each segment is then adjusted by its potentiometer to give an over-all linear relationship.During combustion of a sample, the linearised signal is stored by a suitable condenser, and is continuously displayed on a meter throughout the determination. OxygenJEow-A pump maintains a constant flow of gas through the apparatus. Fluctu- ations in flow-rate, caused by irregular combustion, are avoided by the supply of a large excess of oxygen during ignition of the sample. Excess oxygen escapes from the mouth of the tube, and provides an adequate seal against the atmosphere without the need for any form of closure. Large fluctuations in the percentage of carbon dioxide evolved are smoothed by passing the gas through a small pre-mixing vessel attached to the analyser. The effect of changes in flow-rate was studied by injecting identical volumes of carbon dioxide into the flexible tubing of the apparatus by means of a Hamilton gas-tight syringe, and recording the meter deflections at various flow-rates. In Fig.3, it can be seen that there is a plateau region between 900 and 1000 ml of oxygen per minute. The reasons for this areAugust, 19661 STEEL BY ii DYNAMIC IKFRARED SYSTEII 485 obscure, but Bartley (in a private communication) has suggested that it may be caused by a change in the nature of the gas flow through the analytical cell. It seems probable that, at flow-rates below 900 ml per minute, the flow is laminar, but above this value a certain amount of turbulence is introduced and, while the over-all flow-rate is unaffected, there is some slowing down in the passage of carbon dioxide through the cell.A flow-rate of 950ml per minute was, therefore, used in subsequent work; this was achieved by means of a rotary pump and adjustable flow regulator. FORMATION OF IRON OXIDES- When combustion of samples takes place at high temperatures with fast oxygen flow- rates carry-over of iron oxide becomes troublesome and it is essential to filter the gas stream efficiently before passing it through the analyser. Filters made from cotton-wool and glass- wool become blocked and cause fluctuations in the flow-rate. The problem was overcome by using a calico filter-cloth in a small glass container of the type used with American high frequency furnaces. This proved to be most effective and may easily be cleaned by tapping the container or gently brushing the filter-cloth.CHOICE OF FLUX AND BLANK DETERMINATION- The pick-up of carbon from extraneous sources arises from ( a ) the oxygen supply, (b) the combustion boat and (c) the fluxing materials. After purification of the oxygen supply and pre-ignition of the combustion boat, the contribution from ( a ) and ( b ) is negligible at the carbon levels investigated.’ Pre-ignition of the boats is best performed in one tube of a twin-tube furnace. After cooling in air for a few minutes, the boat is loaded with the sample and flux and then ignited in the second tube. Under these conditions the blank value can be attributed solely to the flux. Two fluxing materials, tin powder and lead foil, have proved satisfactory. The addition of lead foil to carbon and low alloy steel gives a fairly smooth combustion and a minimum carry-over of iron oxide.With careful handling the blank is of the order of 30 p.p.m. of carbon. For more complex steels and solid-pin samples, lead is not suitable and additions of tin powder must be made. Combustion is not as smooth and carryover of iron oxide is greater, but the blank values are lower, and are in the range of from 10 to 20 p.p.m. of carbon. CALIBRATION OF THE ANALYSER- The infrared analyser is initially set up with a gas mixture nominally containing 8.6 per cent. of carbon dioxide in oxygen. With the prototype, it is not possible to adjust the inte- grator meter to obtain a direct reading of the percentage of carbon for a specified sample. Scale calibration may be achieved by burning a number of standard steels in oxygen, and constructing a graph relating meter divisions to carbon content.I t is preferable, however, that calibration should be independent of standardised steels and, therefore, it is better to inject varying volumes of high purity carbon dioxide (volume corrected to S.T.P.) into the system with a Hamilton gas-tight syringe. The relationship between integrated meter reading and carbon dioxide is linear and remains constant over long periods, provided that the infrared analyser is free from electronic drift. This possibility must be checked several times during the course of each working day by using the setting-up gas mixture. A simple slide-rule conversion after each test is all that is necessary to obtain the percentage of carbon. The instrumental precision of the analyser was assessed by injecting a series of identical volumes of carbon dioxide into the system. The coefficient of variation of measurements, obtained for repetitive injections, is about 1 per cent.: this value is, however, limited by the precision of the injection syringe used in these tests. ANALYTICAL PERFORMANCE WITH RESISTANCE AND HIGH FREQUENCY COMBUSTION FURNACES For reasons of economy and simplicity, the Dynacarb apparatus is primarily intended to be used with a resistance-tube combustion furnace. Tests have also been made with a high frequency furnace and a comparison of analytical speeds is given in Table I.486 WHITE AND SCHOLES: DETERMIXATION OF CARBON I N (Analyst, Vol. 91 TABLE I COMPARISON OF ANALYSIS TlMES WlTH RESISTANCE AND HIGH FREQUENCY COMBUSTION FURNACES Instrument time, seconds f-y------ Resistance High frequency Millings and drillings heating heating Mild and medium-carbon steel .. . . . . 40 to 50 35 to 45 Low alloy and high-carbon steel . . . . .. 45 to 55 40 to 50 Stainless steel . . . . . . . . . . . . 45 to 50 30 to 40 High-carbon alloy steel . . . . . . . . 65 to 76 50 to 60 Pin sambles- 1 5-mm diameter Mjld steel . . . . . . 55 to 70 35 to 45 3-mm diameter {Carbon and low alloy steel . . 50 to 65 35 to 45 1-High-carbon steel . . . . 65 to 80 45 to 55 40 to 50 { High-carbon steel . . . . 60 to 80 RESISTANCE HEATING- There was no difficulty in obtaining the complete combustion of samples of even the most complex alloy steel in the form of millings and drillings, with a conventional tube furnace.The best results were obtained by using 26-mm i.d. aluminous-porcelain combustion tubes that were maintained at a temperature of 1350" to 1400" C by silicon carbide heating rods. Instrument time, from the time of inserting the loaded refractory boat into the tube up to the time taken for reading the integrated signal from the meter, varied from 40 to 55 seconds for carbon and low alloy steel, to 65 to 75 seconds for alloy steels containing about 1 per cent. of carbon. Samples of stainless steels ignited quite readily with analysis times rarely exceeding 55 seconds, and a limited number of tests indicated that nickel-base alloys could be analysed within about 60 seconds. For maximum speed in steelworks process control, suction samples are taken in preference to a small cast sample which requires milling or drilling before analysis.One method is to insert the tip of an evacuated tube into a spoon sample of de-oxidised molten steel. The high temperature leads to fusion at the tip, and the sudden suction that is produced causes the metal to enter and fill the tube, so producing a solid rod of 3 to 4-mm diameter. A suitable alternative procedure is to aspirate the molten steel into a glass tube by suction from a rubber bulb. A piece, weighing approximately 1 g, is then cut from the cooled rod or pin and used for analysis. A number of pin samples have been analysed with the Dynacarb apparatus. Preliminary tests on samples of low alloy steels from Samuel Fox & Company indicated that complete combustion was obtained in a resistance furnace, provided that the sample was covered with tin powder.Analysis time, for 3 and 5-mm diameter samples, was related to the carbon content of the sample, and varied from 60 to 80 seconds. Modern tube furnaces are capable of continuous operation at temperatures well in excess of 1400" C ; it was possible to reduce the minimum analysis time to about 50 seconds by increasing the furnace temperature to 1450" C. During subsequent trials of the instrument at the English Steel Corporation Ltd., suction samples were taken alongside the conventional samples that were intended for direct- reading spectrographic analysis. Excellent comparisons with spectrographic results were obtained on steels of different steel-making compositions.HIGH FREQUENCY HEA~ING- Comparative tests were made with a Radyne 5-Mc generator, fitted with a quartz- glass combustion chamber containing a movable refractory pedestal to support the com- bustion crucibles. The speed of combustion, by high frequency heating, is dependent upon the method of passing oxygen through the combustion chamber. When oxygen entered the bottom of the chamber below the refractory crucible, combustion times were found to be of the same order as those obtained with resistance heating. When the flow was reversed and passed through a jet directly above the crucible, extremely rapid ignition was achieved and theAugust , 19661 STEEL BY A DYNAMIC INFRARED SYSTEJI 487 total analysis time was approximately 30 seconds. With such rapid evolution of gas, however, the localised concentration of carbon dioxide may exceed 10 per cent.by volume, thereby overloading the infrared analyser, particularly when high-carbon steels are being analysed. It was, therefore, necessary to design a special oxygen-delivery jet that would ensure adequate mixing and dilution of the carbon dioxide in the combustion chamber. This was made from 4-mm glass tubing with a tapered jet of 2 to 3-mm diameter. Several 1-mm holes, bored into the wall of the tube above the orifice, created sufficient turbulence in the chamber to maintain the carbon dioxide level at below 10 per cent. As described earlier, the combustion gases were pumped through the analyser, and excess of the oxygen was allowed to escape into the atmosphere through one arm of a T-piece situated in front of the combustion chamber.\f'ith high frequency heating, all types of steel in the form of millings or drillings could be analysed within 50 seconds. Solid-pin samples required a further 5 seconds for complete combustion. KESULTS Many standard steels have been analysed; the results of the analyses are presented in Table 11, and are based upon instrument calibration by injection of high purity carbon dioxide. Good agreement was obtained with certified values, and analytical precision varied from about 1 per cent. coefficient o f variation at the 1 per cent. level of carbon content, to 3 or 4 per cent. at the 0.05 per cent. carbon level. There was no significant difference in precision for samples analysed with both the resistance and high frequency combustion furnaces.TABLE I1 RESVLTS OBTAINED iV1TH KESISTASCE AND HI(;H FREQUENCY C 0 31 B U ST I 0 N FLl R X ACE S Certificate value, B.G.S. percentage s o . Type o f carbon 260/1 Iligh purity iron . . . . 0.014 265/1 Carbon steel . . . . 0.043 295 Carbon stcel . . . . 0.263 23X/1 Carbon stcel . . . . 0.21 193 Carbon steel . . . . 0 . 6 3 I N / S Carbon stcel . . . . 0.54 215/l Carbon steel . . . . 0.925 221/1 Chrome - vanadium steel . . 0-50 290 13 per cent. manganese steel 1.17 241jl 1Iigh speed steel . . . . 0.8.5 E 0 / 1 High speed steel . . . . 0.93 :!3.5/1 Stainlcss steel . . . . ct.042 2 3 5 / 2 Stainless steel . . . . 0 . 0 5 2 310 Ximonic 90 . . . . . . 0.098 "1 Stainless steel . . . . 0.083 247/3 \2'hite cast iron . . . . 3 4 0 203/l Ferrochrome alloy .. . . 0.045 203j2 Ferrochrorne alloj- . . . . 0.027 ?39/2 Carbon steel . . . . 0.295 138/1 Carbon steel . . . . 0.81 Rcczrlts olitaiiit)d ztz tvzals at E.S.C. Ltd- 293 Carbon stcel . . . . 0.63 Resistance heating - ~ 2 ~~~~ l'ercentage Standard of carbon tie\ iation 0.01 0 0.002 0.043 0.001 0 . 2 1 0.004 0.64 0.00Fi 0.54 0.009 0.93 0.006 0.50 0.008 1-17 0.016 0.85 0-01 1 0.94 0.008 0.042 0.002 0.071 0.00 1 0.098 0.00 I 0486 0 a 003 2.95 0.028 0.042 0.001 0.02t.i 0.002 7 - __ 0.294 0.008 0.207 0-010 0.64 0.008 High frequency heating f-----l l'ercentage Standard o f carbon deviation - 0-26 0.63 0-54 0.92 0-5 1 1-15 0.84 -~ - 0.054 0.097 - 0-004 0.004 0.003 0.015 0.012 0*010 0.009 - - - 0.002 0.003 X s a further confirmatory test of performance under steelworks conditions, 15 samples of stainless steel, 4 nickel-base alloys and 10 samples of high-speed tool-steel were analysed at the English Steel Corporation, and the results were compared with those obtained by using the !$'ostoff Carmhograph conductimetric analyser for low carbon contents,' and the British Standard gravimetric methodB for the high-carbon tool-steels.The results in Table 111 are in favourable agreement with these alternative procedures. During these tests, 3 standard steels were each analysed by 3 operators at frequent intervals over a period of 2 weeks, in order to provide a more realistic assessment of precision, which ~ o u l d include variation between different operators (see Table TI).488 WHITE AND SCHOLES: DETERMINATION OF CARBON IN [A?tUlJt5t, VOl.91 RESULTS OBTAINED Cast No. Stainless-steel samdes- 5157 . . .. 2509 . . . . 4085 . . . . 2507 . . . . 2603 . . . . 2719 . . . . 2497 . . . . 4080 . . . . 4090 . . . . 4091 . . . . 4092 . . .. 4093 . . . . FW 14 .. FW 2515 . . FK 07 . . .. Nickel-base alloys- Nimonic S207. . NimonicS85 . . Nimonic H8707 Nimonic Y1041 . . . . . . . . . . . . . . .. . . . . . . . . .. .. .. . . .. . . . . TABLE I11 DURING STEEL WORKS ROUTINE OPERATION .. . . . . . . . . . . . . . . . . . . . . . . .. . . .. . . . . . . .. High speed tool-steel samples- R A 4 4 . . . . . . . . RA 45 . . .. . . . . RA 46 . . . . . . . . RA 47 . . . . . . . . RA 53 . . . . . . .. RA 54 . . . . . . . . RA 55 . . . . . . . . RA 56 . , . . . . .. RA 57 . . . . . . ..RA 58 . . . . . . . . .. .. .. . . . . .. . . . . . . . . . . . . .. .. . . Wosthoff Carmhograph, percentage of carbon 0.017 0.035 0.068 0.049 0.057 0.095 0.078 0.027 0.04 1 0.052 0.056 0.27 0.037 0.059 0.051 .. 0.042 . . 0.012 . . 0.029 . . 0.28 Standard method, percentage of carbon .. 0.74 . . 0.74 . . 0.73 . . 0.75 . . 0.78 . . 0.83 . . 0.83 . . 0-8 1 .. 0.83 .. 0.83 I) ynacarb, percentage of carbon 0.019 0.037 0.069 0-055 0.059 0.098 0.080 0.037 0.044 0.053 0.053 0.35 0.036 0.065 0.060 0.045 0.008 0.03 1 0.28 0.75 0.72 0.72 0.74 0.80 0.84 0.84 0.83 0.8 1 0.84 The full-scale meter deflection of the Dynacarb is nominally equivalent to 1.2 per cent. of carbon on the basis of a l-g sample weight, and a “ x 4” scale expansion is provided giving full-scale deflection, equivalent to 0-3 per cent.of carbon. For samples containing more than about 1 per cent. of carbon, a smaller sample weight must be used. The lower level of detection is of the order of 0-005 per cent. of carbon and, a t this percentage, the 95 per cent. confidence limits for the mean of two results is +0.0015 per cent. of carbon. CONCLUSIONS The application of infrared absorption to the determination of carbon in steel by a dynamic system has been shown to be feasible. Precise and accurate results may be rapidly obtained on a wide range of steel-making compositions, by using samples in the form of millings, drillings and solid pins. The proposed apparatus may be used either with a high frequency or resistance combustion furnace, but as the advantages of the former are marginal, a simple resistance-tube furnace is recommended. Capital costs are lower than with alternative instruments that incorporate high frequency heating and automated gas collection. The system is continuously flushed with oxygen, and there is no possibility of the carry-over of carbon dioxide from one sample to the next. Combustion processes can be followed, and completion of the analysis is immediately obvious. A further advantage is the use of an open tube, excess of oxygen providing a seal against the atmosphere during combustion. We acknowledge with gratitude the full co-operation of Mr. W. Bartley, Managing Director, and his colleagues of Hilger-I.R.D. Limited. In addition, thanks are given to Mr. L. Kidman for his permission to make trials at English Steel Corporation, to Mr. R. Staham, Samuel Fox and Company, who supplied pin samples, and to Mr. P. Barker, who performed some of the experimental work. The chief advantages of a dynamic system are simplicity and economy.August, 19661 STEEL BY A DYNAMIC INFRARED SYSTEM 489 REFERENCES 1. Scholes, P. H., “Proceedings of the Sixteenth Chemists’ Conference,” British Iron and Steel 2. 3. 4. 5. Waldock, P., Hilger J . , 1965, 9, 18. 6. 7. 8. British Standard 1121 : Part 1 1 : 1948. Research Association, London, 1963, p. 26. Hines, W. G., Addinall, R. L., and Orten, J . P., J . Metals, 1964, 165. Lay, J. O., Metallurgia, 1955, 51, 109. Tipler, G. A., Analyst, 1963, 88, 272. Rippon, K. P., and Smith, F., Brit. Irzsttz Radio Euzgrs J . , 1962, 24, 127. Scholes, P. H., Rep. B.I.S.R.A., MG/D/244/62. Received January 24th, 1966.