316 VAN WYK, CUYPERS, FITE AND WAINERDI: [Analyst, Vol. 91 A Study of the Macroscopic Distribution of Oxygen in a Steel Rod Neutron-activation and Vacuum Fusion Techniques BY JUSTUS M. VAN WYK (Basic Research Division, Research and Process Development, South African Iron and Steel Industrial Corporation, Pretoria, South Africa) MARC Y. CUYPERS,* LLOYD E. FITE AND RICHARD E. WAINERDI (Activation Analysis Research Laboratory, Texas A and M University, College Station, Texas, U.S. A . ) The distribution of oxygen was determined along the length of a steel rod. Neutron-activation and vacuum fusion techniques were used alterna- tively, and the relevant pieces of apparatus and methods are described. The over-all average oxygen content determined by neutron-activation analysis was 129 p.p.m., in excellent agreement with 128 p.p.m.found by vacuum fusion. The results further show the continuity between the two sets of results, and also a definite inhomogeneity in the macroscopic distribution of oxygen. THE importance of oxygen determination in modern steelmaking has become increasingly apparent as the need for cleaner steels arose during the last few years.192y3 It is the purpose of this study to show the variation of oxygen content along the length of a steel rod. Thereby, a picture is obtained of the degree of homogeneity that can be expected in a finished steel product. Two methods for oxygen determination have been applied, viz., neutron-actil-ation and vacuum fusion analysis. Concurrently with the oxygen distribution, an evaluation could therefore be made of the respective merits of the two methods.MATERIAL AND SAMPLING SELECTION OF STEEL- Because a high manganese content interferes with the vacuum fusion determination of oxygen in steel by gettering of carbon m ~ n o x i d e , ~ ? ~ ~ ~ ~ ~ ~ ~ a steel with a very low manganese content was selected for this study. The composition of the steel was: 0.08 per cent. carbon, 0.16 per cent. manganese, 0.020 per cent. phosphorus, 0.021 per cent. sulphur and 0 4 1 per cent. silicon, and it was produced as a normal semi-killed grade in an l8-ton basic electric furnace from an all-scrap charge at the ISCOR steelworks in Pretoria. From the rolled stock of this heat a 50-mm square billet, corresponding to the central part of a 4-ton ingot, was selected, and one end was hot forged to a rod approximately 14 mm in diameter a d 2-5 m in length. SAMPLING- The 14-mm rod was divided into 10 bars, numbered consecutively from 1 through to 10, and each bar was machined to 12-mm diameter.From each bar, 8 samples of approximately 18 g (19 mm) each, and then 10 samples of approximately 7 g (7 mm) each, were cut consecu- tively. The 18-g samples were analysed by neutron activation, and the 7-g samples by vacuum fusion techniques. Each sample was numbered according to its original position i n the 2-5-m rod and the method of analysis utilised. Thus, the first sample cut from the rod would be 1x1, the second 1N2, and the last two samples 1OV9 and 10V10, respectively. Just before analysis, each sample was etched for 1 minute in concentrated hydrochloric acid to remove surface oxide, rinsed consecutively in distilled water, ethanol and acetone, and finally dried in an air stream.* On leave from the University of Liege, Belgium.May, 19661 DISTRIBUTION OF OXYGEN I N A STEEL ROD 317 APPARATUS- For the neutron-activation part of this research, an apparatus for general activation analysis at the Activation Analysis Research Laboratory (AARL), Texas A and M University, was adapted for possible routine use in a steel production plant for on-line analysis and in-process control. The analysis is based on the l60 (n,p) 16N reaction that has previously been proposed and used for oxygen determination.* to l7 The apparatus consists basically of a neutron source, a sample transfer sytem and detecting equipment (see Fig.1). NEUTRON-ACTIVATION ANALYSIS I I I -L L A = Programmer G,, G, = Transfer boxes B,, 6 2 = Scalers H,, H, = Solenoid air valves C, = Amplifier I = Flux monitor counter C, = Amplifier discriminator J = Sample positioning pin D,, D, = Photocells K = Neutron generator E = Detecting head L = Compressed air F,, F, = Detectors with pre-amplifiers Fig. 1. Block diagram of neutron-activation analysis system Nezdroiz source-A Texas Nuclear Corporation EO-kV, 1-mA maximum, Cockcroft - Walton deuteron accelerator with a tritiated titanium target is used as a neutron source. The 3T (d,n) 4He reaction yields an essentially isotropic and mono-energetic flux of 14.7-hIeV neutrons. At full output, the generator, K, in the present case, is capable of producing a total flux of 10lo 14-MeV neutrons per second.To extend the target life the acclerator is, however, operated at a beam current of only 0-2 mA. A boron trifluoride counter, I, coupled through an amplifier, C,, to a scaler, B,, is used to monitor the neutron flux during irradiation. Transfer system-The short half-life (7.35 seconds) of the product nuclide, nitrogen-16, makes rapid detection imperative. A pneumatic system was constructed for this purpose. I t consists of a compressed air source which connects at L, a 16-mm i.d. plastic tube, solenoid air valves, H, and H,, transfer boxes, G, and G,, to direct the air flow,18 a pin, J, and a detecting head, E, to position the sample at the irradiation and counting sides, respectively, and photocells, D, and D,.Detecting eguifiunent-The 6-13 and 7.12-MeV y-rays of nitrogen-16 are detected by two horizontally opposed and matched Harshaw Integral Line 3-inch x 3-inch thallium-activated sodium iodide crystal and photomultiplier assemblies with pre-amplifiers, Fl and F,. The remainder of the system consists of an amplifier discriminator, C,, a scaler, B,, and a pro- grammer, A. Shielding-The neutron-generator target is surrounded by 25 cm of paraffin wax. The generator is situated in a room with 60-cm concrete walls. The y-ray detectors are enclosed318 VAN WYK, CUYPERS, FITE AND WAINERDI: [L-lrzaZyst, 1'01. 91 in a lead castle of wall thickness 7-5 cm. Between the detectors and the generator room, another 25 cm of paraffin wax and 60 cm of concrete are used to minimise activation of the crystals during irradiation.PROCEDURE- At the beginning of an analysis cycle, a sample is introduced into the pneumatic system at the transfer box, G,, (Fig. 1). The rest of the cycle is controlled by the programming unit, A, in the following sequence. The sample positioning pin in the detector head, E, is raised for a short time and the sample blown through the pneumatic tube (air pressure 80 to 100 p.s.i.) against the sample stop pin, J, at the target. A signal from photocell D, turns on the neutron generator, K, and the flux monitor scaler, B,, for an irradiation time of 20 seconds. At the end of irradiation the air flow is reversed by transfer boxes G, and G,, and the activated sample is blown back against the positioning pin in the detector head, E.After a set delay time of 2 seconds, scaler B, is turned on for 20 seconds only if photocell D, has signalled the arrival of the sample at the detector. The amplifier discriminator, C,, is set to pass only pulses corresponding to 7-ray energies above 4-6 MeV. A continuation of the air flow for a few seconds after arrival of the sample at both the target and the detector eliminates the possibility of the sample bouncing back from the respective positioning pins. The flow of air during the 2-second delay period purges the system of any activated air still present. Standards-Weighed amounts (ranging from 3 to 1750 mg) of dry Specpure ferric oxide (Fe,O,), sealed in polythene vials, were used as standards. Sealed vials filled with air were run to determine the blank value.The possible increase in the blank value as a result of the absorption on the polythene of recoil nitrogen-16 nuclei, stemming from the activation of the air layer around the vial,13 could be ignored at the lel7el of activity in this experiment. The error introduced by assuming a constant volume of enclosed air in all the standards was insignificant. General-The sharp edges of the steel samples were rounded with emery paper to facilitate their transport in the pneumatic system. Each sample and standard was irradiated and counted five times, and a background count, followed by a check determination on a standard, was made at hourly intervals. The linear distance between the detectors and the neutron generator is about 7 m with the shielding as described.There is still, however, a slight build-up of activity in the crystals during a day of continuous operation, which gradually increases the background. On a routine basis a single determination can be made in 1 minute and, allowing 1 to 2 minutes for the manual calculation, a result can be reported every 4 minutes lor duplicate analyses. I VACUUM FUSION ANALYSIS APPARATUS- A slightly modified conventional vacuum fusion apparatus was constructed in the Research and Process Development Department of ISCOR, Pretoria. A schematic diagram is given in Fig. 2. The furnace assembly, A, consists of a water-cooled quartz tube joined, via a ground-glass joint, to a Pyrex-glass cross. A graphite crucible, thermally insulated from the quartz with coarse graphite powder (about 0-5-mm granular size), is heated inductively by a 15-kIl' radio- frequency unit.A four-stage, high speed, high backing pressure, mercury diffusion pump, B, is used for gas extraction, circulation and collection. The all-glass gas analysis equipment consists of a manometer, C, with dibutylphthalate as manometric fluid, a furnace-heated (400" C) vessel, D, containing cupric oxide (CuO) in wire form, a vapour trap, E, filled with granular magnesium perchlorate (Mg(ClO,),), and a cold trap, F. A Dewar flask with liquid nitrogen for cooling F is raised and lowered by means of a motor-driven hoist. Stopcocks S1 to S, and the mechanical fore-pump, which connects as indicated, complete the apparatus. METHOD- Primiple of anaZysis-The oxygen in the sample is reduced by carbon at 1600" C and evolved as carbon monoxide together with nitrogen and hydrogen.The carbon monoxide and hydrogen are oxidised to carbon dioxide and water, respectively, by cupric oxide at 400" C. The water vapour is trapped in magnesium perchlorate and the pressure of theMay, 19661 DISTRIBUTION OF OXYGEN IN ,4 STEEL ROD 319 carbon dioxide and nitrogen mixture measured. The carbon dioxide is then frozen out at -190" C by liquid nitrogen, and the pressure of the nitrogen alone is measured. The pressure of the carbon dioxide is determined by a ddference calculation. f t E To fore-pump A = Furnace assembly B = Diffusion pump C = Manometer E = Vapour trap (magnesium per- ch lo rate I F S,, S,, S,, S, = Stopcocks = Cold trap iliquid nitrogen) D = Furnace heated copper oxide con- tainer Fig.2. Schematic diagram of vacuum fusion analysis equipment Calibration-The collection volume (1') was calibrated by introducing a known volume of pure nitrogen at room temperature and pressure at the cross arm of the furnace assembly, and expanding it through the diffusion pump into the volume 1'. The pressure in V was measured on the manometer, which has a scale calibrated in millimetres of mercury. By arbitrarily assuming normal room temperature (25" C) in volume V, most of which is in- corporated in the lower part of the diffusion pump at an unknown temperature, V was calcu- lated from the ideal gas equations. PROCEDCRE- A maximum of fifteen 5 to 10-g samples at a time are etched, weighed and introduced into the sample loading arm of the furnace assembly, together with about 30 g of nickel.The whole system is then evacuated with the crucible at 1800" C until the blank l-alue is less than 1 per cent. of the expected gas content of a steel sample; this takes about 2 hours. The crucible temperature is then lowered to 1600" C, and the nickel (in part or in total) introduced into the crucible through the quartz funnel (Fig. 2) with a magnet. This nickel is necessary for quick de-gassing and quantitative reduction of aluminium oxide.5 After blank measurement the samples are dropped into the crucible and analysed. The evolved gas is extracted for 2 minutes and then circulated through the hot cupric oxide and the magnesium perchlorate for 1 minute to remove hydrogen. A pressure reading (nitrogen and carbon dioxide) is taken to the nearest 0.02 mm of mercury.The cold trap is subsequently immersed in liquid nitrogen and the final pressure (nitrogen alone) read after 3 minutes. To conclude the analysis, the cold trap is heated to room temperature in an air stream and the whole system evacuated by the diffusion pump (backed by the fore-pump) for 2 minutes, during which time the oxygen content of the sample is calculated. The determination, complete with calculation, is thus completed in 8 minutes.320 VAN WYK, CUYPERS, FITE AND WAINERDI: [Analyst, VOl. 91 RESULTS AND DISCUSSION ACTIVATION ANALYSIS- After correcting for background, the observed counts for the five activation analyses of each sample were normalised with respect to the neutron flux and the average and standard deviation calculated.Oxygen, mg Fig. 3. Neu tron-activation analysis calibration curve The calibration curve obtained with the ferric oxide standards is shown in Fig. 3. The straight line, calculated by the least squares method, gives a calibration constant of 189 counts per mg of oxygen, and fits the results points very well throughout the entire range, i.e., from 1 to 525mg of oxygen. TABLE I Bar : Sample 1 2 3 4 5 6 7 8 Average of bar (U in p.p.m.) Bar : Sample 1 2 3 4 5 6 7 8 OXYGEN FOUND BY NEUTRON-ACTIVATION ANALYSIS 1N 2N 3N 4N 5N & r_----h----, r---J-, r----h--- (--A-, Oxygen, 0, Oxygen, u, Oxygen, u, Oxygen, u, Oxygen, 0, p.p.m. percent. p.p,m. percent. p.p.m. percent. p.p.m. percent. p.p.m. percent. 137 137 136 145 145 151 172 136 10 5 4 2 9 12 13 5 135 6 115 125 5 119 103 0 113 106 4 118 116 6 107 120 4 129 122 4 113 131 7 102 14 114 8 123 2 122 4 116 3 111 5 122 4 114 10 117 125 138 139 143 132 145 151 147 145 f 12 120f 11 115 -J= 8 117 + 4 140 f 8 TABLE I-continued GN 7N 8N 9N 1 ON 7+ r----h--7 r----A-, ( P A - - - - , f-h-, p.p.m.per cent. p.p.m. per cent. p.p.m. per cent. p.p.m. per cent. p.p.m. per cent. Oxygen, a, Oxygen, u, Oxygen, u, Oxygen, a, Oxygen, 0, - 152 7 - - - - - - - - - - 139 2 - 163 12 - - 157 12 136 3 133 5 136 3 147 4 135 4 Average of bar (0inp.p.m.) 149 f 11 136 f 1 Over-all average : 170 9 145 143 4 112 140 6 107 125 8 121 123 4 111 145 11 111 111 10 102 112 8 114 134 f 20 115 f 13 129 p.p.m. f 17 p.p.m. 166 20 129 6 120 6 123 9 122 7 131 G 143 12 130 3 133 f 15May, 19661 DISTRIBUTION OF OXYGEN IN A STEEL ROD 32 1 The oxygen content and percentage standard deviation for each sample are listed in Table I, and are grouped according to the sampling sequence described in the paragraph on sampling.The oxygen content varies from 102 to 172 p.p.m. Samples with visible cracks due to the forging of the original steel billet were not analysed (cf. bars 6N and 7N). Typically, 625 counts and a background of 64 counts were observed per sample. The standard deviations quoted for the samples in Table I compare favourably with an expected value of 6 per cent., and the reproducibility is sufficient for routine analysis. In a few instances, somewhat higher standard deviations were observed. This fact is attributed to local inhomogeneity in the samples concerned. In Table I a value is also given for the average oxygen content of each bar.The standard deviation in this case was calculated from the oxygen contents of the samples constituting the particular bar, and it is given directly in p.p.m. of oxygen. To test the validity of the calibration with ferric oxide, 26 apparently homogeneous steel samples were analysed independently by two other laboratories, namely at Texas Nuclear Corporation and at Kaman Nuclear. In the first laboratory, titanium with a known oxygen content (certified by the N.B.S.) was used for calibration, whereas a synthetic sample consisting of a stack of alternating mylar and steel discs, was used as a standard at Kaman N~c1ear.l~ The average oxygen content of the 26 samples (5 determinations on each) found by the Texas Nuclear Corporation and the Kaman Nuclear were 124 pap.m. 11 p.p.m.and 123 p,p.m. 1 13 p.p.m., respectively, which are in excellent agreement with the value of 125 p.p.m. 5 14 p.p.m. found in our system (Table II), thus justifying the calibration with ferric oxide. TABLE I1 COMPARISON OF RESULTS OBTAINED IN THREE LABORATORIES FOR 26 HOMOGENEOUS SAMPLES Laboratory : AARL Sample 1N2 137 1N4 145 1N8 136 2N4 106 2N6 120 Oxygen, h p.p.m. TNC KN 128 138 132 106 119 120 138 124 103 125 2N7 122 132 130 3N3 113 113 120 3N4 3N5 3N8 4Nl 4N4 4N8 5N2 5N5 5x8 6N4 6N7 7N6 7N7 8N2 8N3 9N3 9N6 10N3 10N5 Mean . . .. 118 107 102 11.1 116 117 135 132 147 139 133 136 136 143 140 107 111 120 122 125 f 14 116 120 109 124 111 I24 127 127 153 138 128 131 123 137 12s 106 112 122 131 124 f 11 119 112 107 121 128 132 145 135 161 133 118 125 12.2 124 121 95 106 117 124 123 f 13 In the three systems concerned there are significant differences in irradiation geometry.In the Kaman nuclear system a double-axis rotator is used to minimise the effect of in- homogeneity in the sample, whereas no sample rotation during irradiation occurred in the cases of the Activation Analysis Research Laboratory and the Texas Nuclear Corporation. Further, the minimum distance between sample and target is about 0.4mm in the Texas Nuclear Corporation system (which was constructed for maximum sensitivity) as compared with about 6mm for the other two. One would therefore expect sample inhomogeneity to322 VAN WYK, CUYPERS, FITE, AND WAINERDI: ‘,41zalyst, Vol. 91 have a greater affect in the Texas Nuclear Corporation system and a smaller effect in the Kaman nuclear apparatus in comparison with the Activation Analysis Research Laboratory system.This is reflected clearly in the different standard deviations for 5 determinations on the same sample as found in the three laboratories, namely 13 per cent. (Activation Analysis Research Laboratory), 25 per cent. (Texas Nuclear Corporation) and 5 per cent. (Kaman nuclear) for an inhomogeneous sample, and 4, 8 and 2 per cent., respectively, for a homo- geneous sample. VACUUM FUSION- As stated in the paragraph on the principle of analysis, one oxygen atom in each resultant carbon dioxide molecule originates from the sample being analysed.Thus, one gram mole of carbon dioxide (22,400ml at S.T.P.) contains one gram atom (16g) of sample oxygen. Assuming validity of the ideal gas equations, the oxygen content of a sample with mass M g is thus related to the carbon dioxide pressure reading by- P 273 lo6 16 p.p.m. of oxygen = - V - ~ - - M 298 760 22,400 where V = calibrated collection volume in ml, P = pressure of carbon dioxide in V (mm of mercury) (typically 4.2 mm of mercury). The collection volume ( V ) was found to be 253 ml. Therefore, the oxygen content of the analysed sample is- P p.p.m. of oxygen = 218 - n/r ’ The results of the vacuum fusion oxygen determinations are given in Table 111. Again, the results are listed according to the sampling sequence, and a value for the average oxygen content of each bar is given, together with the standard deviation.There was only one sample containing a visible crack, and this was not analysed. The values range from 96 to 172 p.p.m. TABLE 111 OXYGEN FOUND BY VACUUM FUSION ANALYSIS Oxygen, p.p.m. Bar : Sample 1 2 3 4 5 6 7 8 9 10 Average of bar Bar : Sample 1 2 3 4 5 6 7 8 9 10 Average of bar iv 2 v 3 v 117 131 127 131 172 136 136 136 151 145 138 f 15 101 111 121 128 131 125 122 127 117 119 120 f 9 110 110 109 102 99 96 106 119 128 125 110 & 11 TABLE TII-contiizzied Oxygen, p.p.m. h -_ r 6V 7 v 8V 132 122 108 142 123 105 143 132 109 144 130 116 137 129 109 142 127 133 - 125 115 156 127 127 130 128 122 133 123 109 140 f 8 127 f 3 115 f 9 Over-all average: 128 f 15. 4 v 10s 123 114 115 126 135 147 171 134 126 130 18 5 v 138 166 154 134 137 133 139 136 144 138 142 & 10 9 v 114 125 118 137 134 108 110 139 151 169 131 f 19 1 oG 126 124 119 134 128 117 140 120 121 145 127 f 9May, 19661 DISTRIBUTION OF OXYGEK IN A STEEL ROD 323 COMPARISON AND CONCLUSIONS- 17 p.p.m., and is in striking agreement with 128 p.p.m.15 p.p.m. found as the average for the 99 vacuum fusion analyses (Tables I and 111). In Fig. 4, the two sets of results are combined and plotted as the oxygen content along the length of the original steel rod. The continuity between the two sets of results is evident. The continuous curve, drawn in Fig. 4 through the points representing the average values of the 20 sub-groups, illustrates the large-scale distribution of oxygen content around the over-all average for the complete rod.I t clearly indicates an almost periodic fluctuation of the oxygen content around the over-all average for the complete rod, instead of the random scatter that would normally be expected in such a case. The average of the 73 neutron-activation determinations is 129 p.p.m. I --T-’ -- 0 50 I00 I50 200 250 Distance from end of steel rod, cm Fig. 4. Distribution of oxygen along the length of a steel rod The figure shows that, in this rolled and forged steel rod which is a fair example of a finished steel product, there exists a definite inhomogeneity in the macroscopic oxygen dis- tribution. This distribution can be determined with equal accuracy by both neutron-acti1.a- tion and vacuum fusion techniques, but faster, non-destructively and more conveniently by neutron activation.The authors wish to express their thanks to the Activation Analysis Research Labora- tory, Texas A and 1c.I University, and the Research and Process Development Department of the South 14frican Iron and Steel Industrial Corporation, as well as their appreciation to Texas h’uclear Corporation and Kaman Xuclear for their kind co-operation dexribed above. I . 3. 4. ti. 7. S. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 7 1. 9. REFERENCES Chepn. Engizg News, 1965, 43, KO. 11, 38. Etterich, O., Taxhet, H., and Thoniich, I\’., Arch. EisenAuttTYes. 1964, 35, 613. Kraus, T., Ibid., 1962, 33, 527. Sloman, H. .4., J . Iron Steel Inst., 1941, 143, 298. Kraus, T., Frohberg, 31. G., and Gerhardt, -%., Avch. EisenhzittWes., 1964, 35, 30. Beach, A. L., and Cruldner, W. G., A n a l j ~ t . Clzcin., 1959, 31, l i 2 2 . Sperncr, F., and Koch, I<.-H., iWetrrlZ, 1964, 18, i01. Koch, R. C., “-tctivation Analysis IIandbook,” A4cademic Press Inc., Xew 170rk, 1060, p. 3 0 . Steelc, E. L., and Meinke, 121. W., Aual\d. Chem., 1962, 34, 185. Veal, D. J., and Cook, C. F., Tbid., 19G2, 34, 178. Coleman, I<. F., arid Perkin, J. L., Analyst, 1959, 84, 233. F3-ud’homnie, J. T., “Texas Nuclcar Corporation Neutron Generators,” Tesas Nuclear Corporation, Anders, 0. U., and Briden, D. W., A ?zalvt. CIwiz., 1965, 37, 530. Wood, D. E., and Pasztor, L. C., “Proceedings of the 1965 International Conference on JIoclern Nickel, H., Rottmann, J., Stdcker, 13.- J., Koster-Pflugmacher, A., and Frohberg, Rf., Avch. Kopineck, H.-J., Sommerltorn, G., Bass, R., and Presser, G., Ibid., 1964, 35, 957. Coleman, R. F., Analysl, 1962, 87, 590. Fite, I>. E., Steele, E. L,., and Wainerdi, R. E., Report No. TEES-2671-2, U.S. Department of Corn- Received Sepfeinber 28th, 1965 Austin, 2962, p. 91. Trends in Xctilration Analysis,’ 31cGraw-Hi11, New York, z ?a the press. Eisetihiitt W e s . , 1964, 35, 637. merce, Office of Technical Services, 1962.