Search Documents

By N. A. Warner

Dense cylindrical specimens of artificial hematite were reduced in hydrogen over a range 0-f total pressures between 0.1 and 1.0 atm and temperatures between 650" and 950°C. Hydrogen reduction at a total pressure of 1.0 atm in the presence of a chemically inert diluent and reduction in hydrogen/water mixtures were also studied Znter.face movement during reduction was followed using metallographic techniques. Substantiu1 partial-pressure gradients are shown to exist in both the gas film and the reduced iron layer. The results are consistent with a mixed-control mechanism involving an interaction of gaseous-difusion effects with a first-order reversible chemical reaction at the iron/wüstite interface. Over the hydrogen-pressure range of 0.1 to 1.0 atm, the rate is not directly proportional to the hydrogen pressure in the bulk gas stream. Nitrogen dilution of the reducing gas masks the true effect of hydrogen partial pressure by influencing the effective molecular dif-fusivity and gives an apparent first-order relationship. The transport of hydrogen and water vapor across the iron layer is consistent with molecular gaseous diffusion rather than a Knudsen diffusion mechanism. ALTHOUGH the literature on the reduction of iron oxides is very extensive, it is only comparatively recently that a definite quantitative model for the reduction of hematite has been proposed. This development has been largely due to the work of McKewan. An extensive literature review of previous work has recently been compiled by Themelis and Gauvin. Exclusive control by gaseous diffusion in the reduction of hematite has been recently advocated by Kawasaki and coworkers.' Although the results of the latter workers are of considerable practical significance, the extreme porosity (approximately 30 pct) of their specimens precludes their application to a kinetic evaluation of the reduction of dense polycrystalline hematite. The salient features of the kinetic model used by McKewan for the gaseous reduction of dense hematite at temperatures in excess of 560°C (when FeO is stable) are: 1) The reduction is a topochemical reaction taking place through 'the series FeO/FesO4/FeO/ Fe; 2) The gas-solid type of reaction takes place only at the FeO/Fe interface and the internal reduction: FeO proceeds by solid state diffusion; 3) The iron layer formed is quite porous and offers negligible impedance to gaseous diffusion; 4) The rate-controlling step in the reduction is a chemical process occurring at the Fe/FeO interface. In view of the relatively fast reaction rates involved when hematite is reduced in a stream of hydrogen, it seemed possible that the simplifying assumptions of negligible diffusional resistance in both the bulk gas phase and the reduced iron layer could introduce serious errors in the interpretation of experimental reduction data. The present work was initiated to elucidate the effect of gaseous diffusion on the reduction rate and to determine its possible influence in the formulation of the reaction mechanism at the metal/oxide interface. EXPERIMENTAL Procedure and Apparatus. Dense cylindrical compacts of artificial hematite were prepared by sintering pressed compacts of reagent-grade ferric oxide, containing 99.6 pct FezO, in air at 1200°C for 80 hr. The resulting compacts had a final porosity of less than 2 pct and a height and diameter of approximately 1 cm. Variations in sintering atmosphere from air to pure oxygen, in sintering time from 20 to 80 hr, and temperature from 1150" to 1250°C had no effect on the subsequent reduction behavior. The reduction of the samples took place in a vertical tube furnace in which was placed a 1-in. transparent silica tube. The 24-in. heated length

Jan 1, 1964

By W. M. McKewan

Magnetite pellets were reduced in flowing hydrogen at pressures up to 40 atm over a temperature range of 350° to 500°C. The rate of weight loss of oxygen per unit area of the reaction surface was found to be constant with time at each temperature and pressure. The reaction rate was found to be directly proportional to hydrogen pressure up to 1 atm and to approach a maximum rate at high pressures. The results can be explained by considering the reaction surface to be sparsely occupied by adsorbed hydrogen at low pressures and saturated at high pressures. PREVIOUS investigation1,2 have shown that the reduction of iron oxides in hydrogen is controlled at the reaction interface. Under fixed conditions of temperature, hydrogen pressure, and gas composition, the reduction rate is constant with time, per unit surface area of residual oxide, and is directly proportional to the hydrogen pressure up to one atmosphere. The reduction rate of a sphere of iron oxide can be described3 by the following equation which takes into account the changing reaction surface area: where ro and do are the initial radius and density of the sphere; t is time; R is the fractional reduction; and R, is the reduction rate constant with units mass per area per time. The quantityis actually the fractional thickness of the reduced layer in terms of fractional reduction R. It was found in a previous investigation2 of the reduction of magnetite pellets in H2-H,O-N, mixtures, that the reaction rate was directly proportional to the hydrogen partial pressure up to 1 atm at a constant ratio of water vapor to hydrogen. Water vapor poisoned the oxide surface by an oxidizing reaction and markedly slowed the reduction. The enthalpy of activation was found to be + 13,600 cal per mole. It was also found that the magnetite reduced to meta-stable wüstite before proceeding to iron metal. The following equation was derived from absolute reaction-rate theory4,8 to expfain the experimental data: where Ro is the reduction rate in mg cm-2 min-'; KO contains the conversion units; Ph2 and PH2O are the hydrogen and water vapor partial pressures in atmospheres; Ke is the equilibrium constant for the Fe,O,/FeO equilibrium; Kp is the equilibrium constant for the poisoning reaction of water vapor; L is the total number of active sites; k and h are Boltzmann's and Planck's constants; and AF is the free energy of activation. Tenenbaum zind Joseph5 studied the reduction of iron ore by hydrogen at pressures over 1 atm. They showed that increasing the hydrogen pressure materially increased the rate of reduction. This is in accordance with the work of Diepschlag,6 who found that the rate of reduction of iron ores by either carbon monoxide or hydrogen was much greater at higher pressures. He used pressures as high as 7 atm. In order to further understand the mechanism of the reduction of iron oxide by hydrogen it was decided to study the effect of increasing the hydrogen pressure on rebduction rates of magnetite pellets. EXPERIMENTAL PROCEDURE The dense magnetite pellets used in these experiments were made in the following manner. Reagent-grade ferric oxide was moistened with water and hand-rolled into spherical pellets. The pellets were heated slowly to 550°C in an atmosphere of 10 pct H2-90 pct CO, and held for 1 hr. They were then heated slowly to 1370°C in an atmosphere of 2 pct H2-98 pct CO, then cooled slowly in the same atmosphere. The sintered pellets were crystalline magnetite with an apparent density of about 4.9 gm per cm3. They were about 0.9 cm in diam. The porosity of the pellets, which was discontinuous in nature, was akrout 6 pct. The pellets were suspended from a quartz spring balance in a vertical tube furnace. The equipment is shown in Fig. 1. Essentially the furnace consists of a 12-in. OD stainless steel outer shell and a 3-in. ID inconel inner shell. The kanthal wound 22 in. long, 1 1/2, in. ID alumina reaction tube is inside the inconel inner shell. Prepurified hydrogen sweeps the reaction tube to remove the water vapor formed during the reaction. The hydrogen is static in the rest of the furnace. The sample is placed at the bottom of the furnace in a nickel wire mesh basket suspended by nickel wire from the quartz spring. The furnace is then sealed, evacuated, and refilled with argon several times to remove all traces of oxygen. It is then evacuated, filled with

Jan 1, 1962

By J. Bruce Wagner, Roger L. Levin

Integrated forms of two solutions of Fick's second law for the movement of a plane interface through a sample of wustite, and for diffusion into a semi-infinite slab of wustite, are shown to yield nearly identical values of the chemical diffusion coeffcients of cation vacancies in wustite, over the temperature range 900° to 1100°C. The chemical diffusion coefficients, the values of D obtained for the propagation of a concentration gradient through samples of undoped and doped wustite, are obtained from the equations It is becoming increasingly important in metallurgical practices to understand the role of the chemical diffusion coefficient, i.e., the diffusion coefficient of a species across a concentration gradient. During the first stages of sintering, one of the mechanisms by which the bonds between neighboring particles develop is by volume or bulk diffusion. This diffusion coefficient as deter- where ?m is the weight change at any time t, ?W is the total weight change at t =8, A is the area of the specimen, and AC is the total change in the cation concentration after the establishment of equilibrium, as determined by reduction experiments in different carbon monoxide -carbon dioxide mixtures within the limits of existence of the single-phase field. mined from the standard equations for volume diffusion during sintering is actually a chemical diffusion coefficient, and it has been observed to be many times larger than the self-diffusion coefficient as determined on the same materials by experiments using radioactive tracer techniques. Kuczyn-ski1 has noted that, if the controlling mechanism of sintering of A12O3 in an oxidizing atmosphere is the diffusion of oxygen, then this diffusion coefficient as measured from sintering rates is about 2.2 x 104 higher than the self-diffusion measurements obtained by tracer techniques. He has also reported the diffusion coefficient obtained from the rate of sintering of Fe2O3 powder to be about 10 6 times faster than the self-diffusion coefficient of

Jan 1, 1965

By B. M. Larsen

A discussion based on three commercial furnace tests and electrical analogue calculations is presented. It shows that while regenerator efficiency is mainly dependent on loading or relative amount of heat exchange surface plus the heat transfer coefficients as effected by flow velocity, the actual air preheat temperature can be high or low, largely independent of regenerator efficiency, mainly due to the diluting or unbalancing effects of cold air leakage into various parts of the furnace system. THERE are many signs of a renewal of interest in the heat regeneration aspect of the open hearth process, among them being the recent paper by Marsh,' the introduction of auxiliary checkers in the stacks of the Isley furnace design, and a number of recent installations of two-pass checkers in new and old furnaces. However, there are also signs of yielding to the usual temptation of oversimplifying a problem. Faced with the job of obtaining better draft on a furnace, for example, a good designer always recognizes that focusing attention on some one factor, such as height of the stack or capacity of the exhaust fan, does not make his problem any easier; rather he must look at the whole system, including such items as valve resistances, bends in the flues, and flow resistance in the waste heat boiler; then he must make his design in relation to the available draft so as to avoid any serious bottleneck in the system. Yet this example is really much simpler than the problem of air preheating, and again, the latter problem is not really made easier by concentrating on air temperature in uptakes and on the more obvious methods for making this air as hot as possible. Also to be considered are such things as the total amount of air needed theoretically for complete combustion at any selected fuel rate, together with that needed for oxygen absorbed into the bath and for burning the combustible gases given out from the bath, plus some excess air. Furthermore, how does this total air requirement relate to the regenerator efficiency at various loadings? How much of the total air equivalent in stack gases will come from preheated air passed through the whole length of the incoming regenerator and how much from air leakage into various zones of the whole furnace system? Does the effect of leakage air differ depending on whether it leaks into the ingoing regenerator, the furnace above floor level, the outgoing regenerator, or the stack zone? Over a number of years we have in this Laboratory experimented with methods for measuring hot gas temperatures; we have measured, approximately, the efficiency of a few regenerators on commercial furnaces and we have developed an electrical analogue' for calculations of regenerator efficiency. The whole problem is too complex to cover in one paper even with a more complete background of data than we possess, but we can attempt here to clarify certain aspects in a general way, hoping to assist specific plant studies on individual furnaces. Some of our earlier experience has been incorporated into chaps. 4, 19, 20, and 21 in the book "Basic Open Hearth Steelmaking,"" and on the assumption that this text is available to most readers, this discussion has been shortened by references to that book. The most recent design of a flowing-gas thermocouple which we are now using for measurement

Jan 1, 1955

By D. C. Hilty

It has long been known that in melting high-chromium steels, some of the carbon might be oxidized out of the melt without excessive simultaneous oxidation of chromium, and that higher temperatures favor retention of chromium. The advent of oxygen injection as a tool for rapid decarburization of a steel bath permits significantly higher bath temperatures, and it was quickly recognized that the use of oxygen injection facilitated the oxidation of carbon to low levels in the presence of relatively high residual chromium contents. Up to the present time, however, specific data pertaining to the chro-mium-carbon-temperature relations in chromium steel refining have not been available. Individual steelmakers have evolved practices more or less empirically, but there has been very little real basis for predicting how effective any given practice can be in permitting maximum oxidation of carbon with minimum loss of chromium. The current investigation, therefore, was undertaken in an effort to establish the fundamental carbon-chromium relationship in molten iron under oxidizing conditions. As reported below, the equilibrium constant and the influence of temperature on that constant have been derived for the iron-chromium-carbon-oxygen reaction in the range of chromium steel compositions with what appears to be a fair degree of precision. The practical application of the result will be obvious. Experimental Procedure The laboratory investigation was carried out on chromium steel heats melted in a magnesia crucible in a 100-lb capacity induction furnace at the Union Carbide and Carbon Re- search Laboratories. The charges for the heats consisted of Armco iron, low-carbon chromium metal, and high-carbon chromium metal, the relative proportions of which were calculated so that the various heats would contain from approximately 0.06 pct carbon and 8 pct chromium to 0.40 pct carbon and 30 pct chromium at melt-down. When the charges were melted, the bath temperatures were raised to the desired level, and the heats were then decarburized by successive injections of oxygen at the slag-metal interface through a ½-in. diam silica tube at a pressure of 30 psi. The duration of the oxygen injections was from 30 sec to 2 min. at intervals of approximately 5 to 30 min. It did not appear that length or frequency of the injection periods had any significant effect on the results; cansequently, no effort was made to hold them constant and they were controlled only as was expedient to the general working of the heats. Between successive injections, the heats were sampled by means of a copper suction-tube sampler that yields a sound, rapidly-solidified sample representative of the composition of the molten metal at the temperature of sampling. This sampling device is a modification of the one described by Taylor and Chipman.1 An attempt was made to vary bath temperatures between samples, but it quickly became evident that, unless the variations were small or unless the new temperature was maintained for a minimum of 15 min. during which an injection of oxygen was made in order to accelerate the reactions, a very wide departure from equilibrium resulted. For most of the runs, therefore, temperature was maintained relatively constant at approximately 1750 or 1820°C. A few reliable observations at other temperatures, however, were obtained. Temperature Measurement The high temperatures involved in this investigation were measured by the radiation method, utilizing a Ray-O-Tube focused on the closed end of a refractory tube immersed in the metal bath. The immersion tubes employed were high-purity alumina tubes specially prepared by the Tona-wanda Laboratory of The Linde Air Products Co. These tubes were quite sturdy under reasonable mechanical stress at high temperature. They were unusually resistant to thermal shock, and chemical attack on them by the melts was slow. With care, it was found possible to keep these tubes continuously immersed in a heat for as long as 5 hr at temperatures up to 1850°C, before failure by fluxing occurred. The Ray-O-Tube—alumina tube assemblage was similar to those supplied commercially for lower temperature applications. In operation, the alumina tube was slowly immersed in the molten metal to a depth of approximately 5 in., and the device was then clamped solidly to a supporting jig where it remained for the duration of the run. A photograph of the equipment, in operation with Ray-O-Tube in place and oxygen injection in progress, is shown in Fig 1. When in position in a heat, the instrument was calibrated by means of an immersion thermocouple and an optical pyrometer. For calibration through the range of temperatures from 1500 to 1650°C, a platinum -platinum + 10 pct rhodium thermocouple in a silica tube was immersed alongside the alumina tube. Output of the Ray-O-Tube in millivolts and the

Jan 1, 1950

By C. E. Sims, F. W. Boulger, H. A. Saller

Most of the data on equilibrium constant and the deoxidations potentialities of those elements, considered to be stronger deoxidizers for steel than is silicon, have been calculated from thermodynamic data. The reason for this is, primarily, the obvious difficulty of obtaining direct experimental evidence of equivalent accuracy. This is an excellent use of the principles of thermodynamics and has given valuable data not otherwise available. Such results, of course, can be no more accurate than the physical constants used in the calculations, and one can never be sure that the basic data are either complete or accurate. In fact, as in the case with silicon,1 there are not only discrepancies among the calculated theoretical values of the equilibrium constant for deoxidation of steel but also between the theoretical and experimental values. It is highly desirable, therefore, to obtain experimental values for checks on calculated results whenever possible. If they disagree, both cannot be right, but if there is good agreement, their value is enhanced. The present work was done in an effort to obtain experimental evidence in regard to some of the common alloying additions but more particularly the so-called "strong" deoxidizers for steel. The method used was to determine the minimum concentration of the deoxidizer that would effect a certain definite degree of deoxidation in steel. The criterion of deoxidation was the change from the large globular Type I sulphide to the eutectic Type II as described by Sims and Dahle.2 This change is sharp and definite, and inasmuch as it can be produced with equal facility by aluminum, zirconium, and titanium, it is considered a manifestation of a certain degree of deoxidation and not an alloying effect. Ostensibly such a procedure could give only a comparison of deoxidizing powers and no absolute values. Nevertheless, repeated observations have shown that, when increasing increments of aluminum are added to a steel, the residual aluminum content begins to increase simultaneously with the appearance of Type II inclusions. Thus it seems warranted to postulate that the Type II inclusions appear coincident with the virtual elimination of FeO as an active constituent of the steel. Experimental Procedure The data obtained were primarily from the microexamination of polished and unetched specimens and from chemical analysis. Experimental heats weighing 200 to 250 lb were made in a basic-lined high-frequency induction furnace. The base composition was nominally that of a medium-carbon casting steel to which the appropriate additions were made. Specimens were poured into sand-cast ingots 3 in. in diam as shown in Fig 1. Sand-cast ingots were used to prevent chilling and to allow sufficient time in freezing for normal inclusions to form of a size large enough to be studied readily. In the first few heats, the tapered wall ingot was used, but in the majority, the extra large riser was used to prevent piping in heavily deoxidized steels. Specimens for microexamination were taken from the location shown in Fig 1, and drillings for chemical analysis were taken from a similar location. The procedure was to melt the base composition and deoxidize with the usual manganese and silicon additions and then to pour an ingot. The furnace was then tilted back, and the first increment of strong deoxidizer or special alloy was added and allowed to disseminate through the melt, with enough power on to hold the temperature constant, for 45 sec. Then a second ingot was poured. After this, another increment was added, and after the same holding time another ingot was poured. In this way from 9 to 12 ingots were poured from each heat, each successive ingot having progressively larger total additions of alloy. Eighteen heats were made altogether, and the range of alloys used and additions made are outlined in Table 1. The three principal types of sulphide inclusions found are illustrated in Fig 2. The globular Type I sulphides are characteristic of silicon-killed steels, the eutectic Type II are characteristic of steels deoxidized with a small amount of aluminum, while the larger, angular Type III are usually found in steels with a residual aluminum content above about 0.02 pct. In all specimens studied, the transition from Type I to II either did not occur at all or was very abrupt and clear cut. There never was any doubt as to just which increment produced the change, although the individual additions were small, in the order of 0.01 pct. The change from Type II to Type III was considerably less sharp, and, in some cases, both types were found together. Inasmuch as the formation of Type III sulphides is apparently not a deoxidation phenomenon, they will not be discussed here.

Jan 1, 1950

By A. Simkovich, R. W. Lindsay

The possibility of using sodium sulfide slags to remove copper from ferrous alloys has been investigated by Jordan1 and by Langenberg.2, 3 In these studies, such slags were determined to be capable of removing copper and sulfur from the melt. The present work represents additional effort to clarify the effects of temperature on copper removal. The experiments were performed in a 17-lb induction furnace. Graphite crucibles contained the melts and kept the baths saturated with carbon. Temperatures were measured with a calibrated optical pyrometer and were controlled by manipulation of power input to the furnace. Estimated accuracy of temperatures in this investigation is ± 10°C (18°F) for measurements prior to slag additions, and + 20°C (36°F) after slag formation. The procedure consisted of melting 800 g of electrolytic iron. During this step, powdered graphite covered the exposed iron surface. After a predetermined temperature was reached, copper shot was added. A sample of the molten alloy for chemical analysis was then aspirated into a silica sheath. Next, a slag-forming mixture of sodium sulfite and graphite was added instantaneously to the melt. The sodium sulfite amounted to one-tenth the charge weight of iron; sufficient graphite was added to combine with oxygen in the sodium sulfite, assuming formation of carbon monoxide and reduction of the sulfite to sulfide. Subsequent to the slag addition, the molten alloy was sampled periodically, with the exception of heat A in which no intervening samples were taken between the slag addition and the end of the run. The iron was poured into a graphite mold, and the ingots sectioned and drilled for samples. Results of selected heats are presented in Table I. Analyses of samples drawn from the iron prior to slag addition are listed under zero time. Two samples from heat D were reported with copper contents greater than the initial concentration in the bath. Owing to the gradual but complete disappearance of slag during this heat, it is believed copper momentarily became more concentrated in the upper portion of the bath while reverting from the slag. This is the region from which samples were drawn. It should be noted that analysis of the ingot was equal to the copper content at the time of slag addition. The terminal temperatures of heats D and E, and the initial sulfur content of heat A are also to be noted. Because of the large temperature drop which occurred when slag was formed in heat D, power input to the furnace was increased in heat E after the slag addition, causing a higher terminal temperature. In heat A, the initial sulfur concentration was relatively high as compared to heats B through E owing to contamination by some slag remaining in the crucible from a previous heat. It is evident from Table I that copper was removed at the onset of slag formation. Roughly 30 pct of the copper was taken into the slag, with the exception of heat D, which had approximately 50 pct removed. For a comparatively short time of slag-metal contact, it appears that no gain is to be made in copper removal through use of high or low temperatures. If the slag initially formed remains in contact with the iron for an extended period, temperature has a marked effect upon copper removal, as can be seen by studying results for the two extremes in temperature. At about 1425°C, the copper level remained relatively constant after the initial removal by the slag. However, in the region of 1670°C, a definite reversion of copper occurred. Reversion was incomplete in heat D, and complete in heat E. The final temperatures of heats D and E differed by about 75°C. This temperature difference is thought to be the reason for only partial copper reversion in heat D. It is believed the effects of temperature noted above are related to the evolution of a white fume, which appeared in every run except heat A. (In the case of heat A, the fume was practically indiscernible.) After each slag addition, a yellow flame formed for about 5 sec. When the flame subsided, a white fume appeared. Upon contact with surrounding cooler surfaces, this fume deposited as a white solid. In the experiments made at 1425°C, evolution of fume continued unchanged to the end of the runs. However, heats D and E exhibited a different behavior. A very noticeable decrease in fume evolution from heat D was observed. Furthermore, this heat had much less slag remaining than did runs A through C when the experiments were terminated. No slag remained at the end of heat E; evolution of fume from this heat ceased prior to pouring. Spec-trographic analysis of the white deposit indicated sodium to be the major metallic element, with the maximum concentration of iron and copper as 0.1 and 0.01 pct, respectively. It is supposed the white fume observed in these experiments is principally sodium oxide (Na2O), formed by oxidation of sodium in the slag and subsequent sublimation. (Sodium oxide is a white to gray substance in the solid state; at 1275oC, it sublimes.4) According to this mechanism, elevated temperatures would accelerate removal of sodium from the slag, sulfur pickup by the

Jan 1, 1961

By G. F. Huff, G. R. Bailey, J. H. Richards

An improved bomb-sampling technique for obtaining samples for oxygen analysis from liquid steel is described. Analyses of samples taken from open-hearth furnaces by the improved method show sufficient agreement with laboratory data to indicate the accuracy of the method is better than that of other existing sampling methods. THE bomb-sampling method of determining the oxygen content of a liquid-steel bath was studied by obtaining and analyzing more than 200 samples from normal open-hearth heats. Of these samples, about 60 were obtained with a bomb of conventional design;' the remainder were taken with a somewhat modified bomb designed to give greater protection against slag contamination. In the new device, the contacting faces of the cast-iron mold halves are machined so that they fit closely; the opening at the top of the bomb is covered with a thin steel cap; and the top of the bomb is further covered with a wood block held in place by heavy steel wire. After the bomb has passed through the slag layer and has entered the steel bath, the wire melts and the wood block floats to the surface. Thereafter, the steel melts through the thin cap and flows into the bomb cavity where it is deoxidized with aluminum wire. An assembled bomb, ready for sampling, is shown in Fig. 1. To standardize sampling procedure, the samples were usually taken through the center door and at the bottom of the furnace; that is, the bomb was on the bottom when the metal cap melted. After each addition to the furnace, at least 20 min elapsed before a sample was taken. The oxygen contents of all the samples were determined by using the vacuum-fusion method, two or more specimens from each sample being analyzed. The specimens were cubes of metal cut from a region in the interior of the bomb sample, as shown in Fig. 2. Drillings for carbon analysis were obtained immediately above the oxygen-analysis region. Reproducibility of Results In determining the reproducibility of the results obtained with the modified bomb, the following four factors were studied: 1—variation in the oxygen concentration of duplicate specimens taken from a single sample; 2—variation in the oxygen concentration of duplicate samples taken in quick succession: 3—effect of the amount of aluminum wire

Jan 1, 1953

By G. Derge, Ling Yang, John Henderson

Self-diffusion coefficients of aluminum have been measured by the capillary reservoir technique in liquids of the CaO-SiO2-Al2O3 system containirg equimolar portions of CaO and SiO2, in the temperature range 1400o to 1520oC. For the melt containing 6.0 mole pct Al2O3 the results can be expressed by the equation D = [(4.3 * 0.3) x 104] exp [(-85,000 * 17,500)/RT]cm~sec-'; for the melt containing 12.5 mole pct Al2O3, by the equation D = (5.4 i 0.2) exp [(-60,000 * 10,00O)/RT] cm2sec-: A concept of the constitution of these melts has been developed which proposes that the ratios O/(Si +Al) and Al2 O3/CaO determine the dominant species present. In recent years considerable work has been done on the system CaO-SiO2-A12O3 with a view to elucidating the structure of liquids of this system. Conductivity,' electromotive force, 3 density,4 and activity5 measurements have all contributed to an understanding of the problem, but as yet there is no entirely satisfactory theory that can account for all the observed experimental results. In principle, measurements of self-diffusion coefficients should give further information which is not obtainable in other ways. Self-diffusion coefficients for calcium, silicon,6 and oxygen7 have been measured. The present investigation extends the series of self-diffusion coefficient measurements to include aluminum and develops a concept of constitution for liquids of this system which is compatible with all the observed experimental results. EXPERIMENTAL The technique used in the measurement of the self-diffusion coefficient of aluminum was the capillary-reservoir method,8 A126 being incorporated in the melt contained in the capillary and inactive aluminum in the reservoir melt. This method was chosen because it minimizes convection currents' in the diffusion zone. It has the further advantage that the boundary conditions are such that a slight modifica- tion of Anderson and saddington's8 solution of Fick's Law yields an equation which requires measurements only of the length and the initial and final average specific activities of the solidified melt within the capillary. where t is the diffusion time, Co is the average specific activity of the melt contained in the capillary at t = O, C is the average specific activity of the melt contained in the capillary at t = t, 1is the length of the capillary, and D is the self-diffusion coefficient. The equation is of somewhat different form than those previously applied to high temperature diffusion studies because the specific activity of the melt in the reservoir was negligible both before and after diffusion. The diffusion apparatus is shown schematically in Fig. 1. It consisted of a silicon carbide resistance furnace containing the inactive melt in a graphite crucible, internal dimensions 7/8 by 4 in., enclosed in a 1 1/4- by 1 1/2- by 30-in. McDanel tube. A 3/8-in. McDanel thermocouple sheath to which graphite radiation shields and a graphite protection tube were affixed, was used to hold the capillaries. By moving this sheath through a rubber seal at the top of the McDanel tube the capillaries could be lowered into or raised out of the melt. Temperature was measured with a Pt-Pt 13 pct Rh thermocouple which was calibrated periodically against a standard couple. The furnace temperature was controlled to ± 5°C through the thermocouple immersed in the melt. Temperature variation over the depth of the melt was less than ± 3°C. The inactive melts were prepared from pure native quartz, assaying better than 99.9 pct SiO2, A.R. grade A12O3 and A.R. grade CaCO,. The same quartz and CaCO3 were used in preparing the active melts, the

Jan 1, 1962

By M. T. Simnad, G. Derge, Ling Yang

STUDY of diffusion in molten substances is important in at least two respects. Diffusion data, combined with thermodynamic and kinetic information, throw light on the structure of the liquid state. Moreover, a knowledge of diffusion coefficients and their temperature dependence is useful in ascertaining the controlling step in many metallurgical reactions. Although diffusion measurements have been made on a number of molten salts, slags, metals, and alloys,'-'" the results available are still limited and widely scattered, especially for the high temperature melts. It is the purpose of this investigation to establish reliable procedures for the measurement of self-diffusion coefficients in high temperature melts and to use these procedures to determine the self-diffusion coefficients of iron in molten Fe-C alloys of different carbon contents. Experimental Measurement of self-diffusion coefficients in high temperature melts is usually carried out by using one of two methods. In the first, diffusion takes place between two cylindrical samples of the same material, one of which is made radioactive. The diffusion coefficient is calculated from the radioactivity penetration curve in the solidified sample after the diffusion. In the second method, samples contained in capillaries of an inert substance are immersed in an infinite reservoir of the same material. Either the material in the capillary or that in the reservoir is made radioactive and the diffusion coefficient is calculated from the change of the radioactivity of the sample in the capillary after the diffusion. The second method is simpler because a radioactivity penetration curve is not needed for the calculation of the diffusion coefficient; however,.it is valid only when no convection has occurred in the capillary during the diffusion. A radioactivity penetration curve is therefore always useful in indicating that the required conditions for the diffusion are ful- filled. Towers and Chipman" recently studied the self-diffusion of calcium and silicon in CaO-Al,O,-SiO, melts by using the capillary method. Instead of measuring the change of the radioactivity of the sample in the capillary, they autoradiographed the sample after the diffusion and calculated the diffusion coefficient from the intensity distribution of the autoradiograph. Their method is useful when the radioactive isotopes are weak ß-emitters, but is very difficult, if not impossible, to apply to cases involving strong ß or y-emitters. For such cases, sectioning of the sample according to the method of Morgan and Kitchener' may be used for the construction of the radioactivity penetration curve. The self-diffusion of iron in molten Fe-C alloys has been studied in this manner and the experimental procedures are described as follows: The diffusion apparatus is shown schematically in Fig. 1. The bath consisted of about 200 g of an Fe-C

Jan 1, 1957

By R. R. Webster, H. T. Clark

The side-blown converter has been investigated as a possible commercial process for the production of low nitrogen steel. During this work, two converters of 3-ton and 22-ton capacity were operated on a pilot plant basis for a total of 214 heats. The steel made in these converters was low in nitrogen and possessed good cold working properties. Some problems of converter operation remain to be solved. IN plants operating with a high iron capacity, sev-eral different refining methods are used in the conversion of the molten pig iron to steel. These include various ore practices in stationary and tilting open-hearths, the duplex process employing the Bessemer converter and open-hearth, and the Bessemer process. At J&L, a considerable part of the iron produced is handled by the Bessemer process, either alone or in conjunction with duplexing, and therefore an appreciable portion of the steelmaking research effort has centered about the method. This paper covers research work on the development of the side-blow converter for the commercial production of low nitrogen ingots and includes descriptions of the operation of a 3-ton and a 22-ton experimental converter at the Aliquippa Works. The refining of iron to produce steel requires the removal of a large portion of the carbon and silicon and the control of manganese, phosphorus and sulphur which are present in the iron in varying amounts. The first large-scale means of refining iron was the acid Bessemer process which was brought into use almost 100 yr ago. This method, using compressed air as the refining medium, accomplishes substantially complete removal of carbon, manganese and silicon. Phosphorus and sulphur are not affected but, by choice of an iron composition sufficiently low in these elements, a commercial product can be produced. Since the process will handle large tonnages rapidly, operates without external fuel and with a minimum of additional equipment, it quickly became the major tool in the early expansion of the steel industry. Later, the basic open-hearth process, by affording control of phosphorus and sulphur and by consuming the large quantities of steel scrap that were becoming available, forced the acid Bessemer process into a secondary position in the industry. During the past two decades the demand for steel to be used in cold forming and drawing operations has gradually increased. Bessemer steel, because of its work hardening and aging characteristics, is not as suitable for these applications as basic open-hearth steel, consequently the decline of the process was accelerated. More recently, because of changing economic conditions, this long range trend appears to have been arrested or perhaps reversed. Ingot production data for recent years furnishes only an incomplete picture of the importance of the converter in the American steel industry; open-hearth furnaces utilize large tonnages of blown metal for which no published statistics are available. Metallurgical Aspects The fundamental difference between Bessemer and open-hearth steels apparently lies not in the method of manufacture but, rather, in the differences in chemical composition of the two steels. It is further believed that the principal features distinguishing Bessemer from open-hearth steel are the higher nitrogen and phosphorus contents of the former. Evidence supporting this position is supplied by tests on laboratory induction furnace heats that were made to contain varying amounts of phosphorus and nitrogen but were otherwise similar to normal low carbon silicon-killed steels. Fig. 1, 2 and 3, summarizing the test results, are taken from G. H. Enzian's paper titled, "Some Effects of Phosphorus and Nitrogen on the Properties of Low Carbon Steels."' Fig. l indicates that phosphorus has a marked effect on the cold work embrittlement of steel as shown by the work brittleness test of Graham and Work.' In the low nitrogen steels, which as a group have the better cold working properties, the effect of phosphorus variations is the more pronounced.

Jan 1, 1951

By R. R. Webster, H. T. Clark

I. A. Sirel—I would like to ask Mr. Sims what would the preferred hot metal analysis be as far as manganese and silicon are concerned if you used specially made iron for this process instead of basic iron. C. E. Sims (authors' reply)—A preferred composition would contain somewhat less silicon than the basic pig iron used in these experiments. High silicon requires a large lime addition. Lime must be added in sufficient quantity to neutralize the silica formed and to leave an excess for dephosphorization. Manganese presents almost no problem in operation. It is almost completely oxidized and, in burning, develops considerable heat. The evidence indicates, however, that, in continuous commercial operation, there will be an excess of heat produced, and the bath will need to be cooled in some manner as by the addition of scrap or by utilizing the energy for the direct reduction of ore. There do not seem to be any serious limitations on composition of pig iron. One heat was made with an iron containing 2.27 pct silicon and the phosphorus finished at 0.007 pct. In another, the iron contained 0.45 pct phosphorus which was lowered to 0.019 pct. This indicates considerable leeway in composition. D. R. Loughrey—We are all very much interested in the recent discovery of very large deposits of iron ore both in Labrador and in Venezuela. I would like to ask Mr. Sims if the iron ore at either or both of these places is such that it would lend itself to the manufacture of iron suitable for use in the turbohearth. C. E. Sims-—According to my information, those ores are not essentially different from ores now used in this country for making basic pig. They should, therefore, be adaptable to this process. D. I. Brown—You have spoken about the leeway, or spread in analysis of the various irons you can use in the vessel. I would like to have you tell us, if you can, about the different kinds of steels which can be made in such a vessel. The turbohearth process, or it might more aptly be called pneumatic openhearth, is a little different from that used in most Bessemer shops and in the one in which I worked. If we had a blow in the air over 10 min, the boss was in the pulpit in a hurry

Jan 1, 1951

By N. A. Gokcen, John Chipman

SILICON is the most commonly used deoxidizer and an important alloying element in steelmak-ing; hence a detailed study of this element in liquid iron containing oxygen is of considerable interest. The equilibrium between silicon and oxygen in liquid iron has been studied by a number of investigators but generally with inconclusive or incomplete results. The variation of the activity coefficients of silicon and oxygen with composition is entirely unknown. Published investigations deal with the reaction of dissolved oxygen with silicon in liquid iron and the results are expressed in terms of a deoxidation product. For consistency and convenience in comparison of the published information, the deoxidation product as referred to the following reaction is expressed in terms of the percentage by weight of silicon and oxygen in the melt in equilibrium with solid silica: SiO (s) = Si + 2 O; K'l = [% Si] [% 012 [I] Theoretical attempts to calculate the deoxidation constant for silicon in liquid iron from the free energies of various reactions yielded results which were invariably lower than the experimental values. Thus, the deoxidation "constants" calculated by McCance,1,2 Feild,3 Schenck, and Chipman were of the order of 10, which is below the experimental values by a factor of more than 10. Experiments of Herty and coworkers" in the laboratory and steel plant resulted in an average deoxidation constant of 0.82x10 ' at about 1600°C. The technique employed in their investigation was crude and the reported temperature was quite uncertain. The concentration of silicon was obtained by subtracting silicon in the inclusions from the total. Since at least some of the inclusions resulting from chilling must represent a fraction of the silicon in solution at high temperatures, such a subtraction is not justifiable. Results of Schenck4 for K'1 from acid open-hearth plant data yielded a value of 2.8x10-5, which was later revised as 1.24x10 at 1600°C. Similarly Schenck and Bruggemann7 obtained 1.76x10-5 at 1600OC. The discrepancies and errors involved in the acid open-hearth plant data as compared with the results of more reliable laboratory techniques were attributed by these authors to the lack of equilibrium and the impurities in liquid metal and slag, and are sufficiently discussed elsewhere." Korber and Oelsen" investigated the relation between dissolved oxygen and silicon in liquid iron covered with silica-saturated slags containing varying concentrations of MnO and FeO. The deoxidation products obtained by their method scatter considerably, and their chosen average values of 1.34x10, 3.6x10-5, and 10.6x10-5 1550°, 1600°, and 1650°C, respectively, represent the best experimental results which were available until quite recently. Darken's10 plant data from a steel bath agree approximately with their data at 1575° to 1625°C. Zapffe and Sims" investigated the reaction of H2O and H2 with liquid iron containing less than 1 pct Si and obtained deoxidation products varying by a factor of more than 20. Inadequate gas-metal contact and lack of stirring in the metal bath should require a longer period of time than the 1 to 5.5 hr which they allowed for the attainment of equilibrium. Furthermore, their oxygen analyses were incomplete and irregular and confined to a few unsatisfactory preliminary samples. Their results did indeed indicate that the activity coefficient of oxygen is decreased by the presence of silicon, although they made no such simple statement. They chose to attempt to account for their anomalous data by the unlikely hypothesis that SiO is dissolved in the melt. Hilty and Crafts" investigated the reaction of liquid iron with acid slags under an atmosphere of argon, making careful determinations of silicon and oxygen contents at several temperatures. Despite erroneous interpretation of the data at very low silicon concentrations, their data represent the most dependable information on this equilibrium that has been published. In the range 0.1 to 1.0 pct Si, their data yield the following values for the deoxidation product: 1.6x10-5, 3.0x10- ', and 5.3x10 at 1550°, 1600°, and 1650°C, respectively. The purpose of the work described herein was to study the equilibrium represented by eq 1 as well as the following reactions, all in the presence of solid silica: SiO2 (s) + 2H2 (g) = Si + 2H2O (g);

Jan 1, 1953

By N. A. Gokcen, J. Chipman

A revised treatment of the authors' published data eliminates the complex relation previously proposed between concentration of silicon and activity coefficient of oxygen in liquid iron. Revised values of the thermodynamic properties of the liquid solution are presented. IN a recent experimental study of the reaction SiO2 (s) = Si+ 2O; Kf, = [% Si] [% O]² [1] the authors' found a substantially constant equilibrium product in liquid iron at 1600°C of 2.8x10-5 They also reported extensive data on the reactions: SiO2 (s) + 2H2 (8) = Si- + 2H2O (g); K'2= [% Si] (H2O/H2 [2] and H, (g) + 0 = H2O (g); K'3 =( H2 O [31 (H2) [%O] From the results on reaction 3 and earlier data of Dastur² on this same reaction in the absence of silicon, they determined the activity coefficient of oxygen, f0, on the basis of the definition K3 = (H2O)/ (H2)f0 [% 0] where K, is the equilibrium constant and f0, is taken as unity in the pure Fe-0 system. Similarly values of fsi were deduced from results on reaction 2. In a more recent study" of analogous reactions in the system Fe-A1-0, it was found impossible to reconcile the results on reaction 3 with Dastur's data; accordingly the latter were ignored and the equilibrium results were extrapolated to find a value of K, at zero concentration of aluminum. This procedure failed to locate the cause of the discrepancy but it did yield reasonable values of activity coefficients. It also avoided introduction of the complex empirical relation between the oxygen activity coefficient and the concentration of the added element. The same type of discrepancy exists for system Fe-Si-0.' In the earlier paper an attempt was made to fit both sets of data by a single curved line (Fig. 6 of ref. l), the form of which is contrary to the theoretical requirement of a finite slope at infinite dilution. In the light of experience on the Fe-A1-0 system the discrepancy must be recognized as one which can be resolved only by more refined measurements. Accordingly Figs. 6 and 10 are retracted. It is pointed out also that until the discrepancy is resolved Figs. 7, 8, and 11 are subject to some uncertainty. Qualitatively the following conclusions still appear valid: 1—The activity coefficient of oxygen is reduced by addition of silicon. 2—In dilute solutions the activity coefficient of silicon increases with its concentration. 3—With respect to equilibrium in reaction 1, the above effects are approximately compensating. The discussion of K'1 in the previous paper requires no revision. It was pointed out that the constancy of the product [% Si] [% 0]² ndicated a compensating effect of the activity coefficients of silicon and oxygen. Therefore, as a very good approximation, K1 = K'1 and the following average values are suggested both for K, and K', at the temperatures 1550°, 1600°, and 1650", respectively, 1.0x10-", 2.8~10-" and 5.5 ~lo-'. Revision of the thermodynamic treatment is necessitated by the recent appearance of new data, based on a combination of combustion and solution calorimetry,' which yields for the heat of formation of low-cristobalite from the elements, the value —209,330 ±250 cal per mol at 25°C. This is about 4000 cal larger than the value previously accepted. The new value for cristobalite is used, together with Kelley's tables of high-temperature heat contents" and entropies and with Korber and Oelsen's' heat of fusion of silicon to obtain the following equation for the standard free energy of cristobalite in the temperature range 1700" to 2000°K: Si (1) + 02 (g) = Si02 (crist.); ?F° = -217,700 + 47.OT [4] The free energy of solution of 0, in liquid iron is:8 O2 (g) = 20 (in Fe); AF° = -55,860 - 1.14T [5] and these two equations are combined to give: Si (1) + 20 = SiO, (crist.); AF° = -161,840 + 48.14T [6] ?F°1873 = -71,700 cal. From the experimental value of K, = 2.8x10-5, Si + 20 = SiO2 (crist.); ?F°1873 = -39,000 cal. [7] The combination of Eqs. 6 and 7 yields the free energy change when liquid silicon dissolves in iron to form the dilute solution of unit activity (1 pct). Si (1) = Si; ?F°1873 = -32,700 cal. [8] The heat effect in this process according to Korber and Oelsen' is an evolution of 28,500 cal per gram

Jan 1, 1954

By J. Chipman, J. C. Fulton

Reduction of silicon from blast-furnace-type slags by carbon-saturated iron is a very slow reaction even under conditions of rapid stirring. Equilibrium under atmospheric pressure of carbon monoxide was approached from both sides, and the silica-silicon relation was established at temperatures of 1425º to 1700°C for slags containing up to 20 pct Al2O3. Silicon carbide is formed by reaction of graphite with high silica slags and the conditions for its formation were determined. Activities of SiO2 and of CaO in their binary solutions at 1600°C were determined. The effects on the activity of SiO2 were established for additions of 10 to 20 pct AI2O3 and of 10 pct MgO. STUDIES of chemical reactions between blast-*>-5 furnace slags and metal have been focussed on the transfer of sulphur from metal to slag. The important role which silica reduction plays in this process' has led to a study of the transfer of silicon from slag to metal and the equilibrium represented by the equation SiO2 (in slag) + 2C (graphite) = Si + 2CO (g). [I.] The authors2 have published an account of the experimental method employed in attacking this problem and of three series of experimental results at 1600°C. This paper contains data on some additional compositions and extends the observations to cover the temperature range 1425° to 1700°C. The data are used to determine the activity of silica in the slags. As in the earlier experiments, the atmosphere was carbon monoxide which flowed through the furnace at 150 ml per min. In certain heats designated in the tables and figures, the gas was bubbled through the melt using a graphite tube. The latter increased in porosity during a run and in some cases bubbling ceased as indicated by cessation of surging. Some results were discarded for this reason. Experimental Results The new data are shown in Tables I through V and both old and new data are plotted in Figs. 1

Jan 1, 1955

By G. Britsianes, T. L. Joseph

THE reduction of a lump of iron ore is a complicated sequence of up to three reactions proceeding simultaneously in a gas-solid system. As the ore moves down the blast furnace into zones of higher temperature and higher reducing power, it is successively reduced through the three oxides of iron into metallic iron. The reduction process involves much more than chemical problems. Physical factors add to the complexity of the overall process. Under optimum conditions, reduction of the ore is completed at a level about one-half way down the blast furnace stock column. At this point, the ore undergoing reduction has attained a temperature of about 1000°C and has been in the furnace for about 6 hr. On the practical side, the behavior of the ore during smelting has been of great interest to operators. Unsatisfactory blast furnace operation on burdens containing magnetite ore or badly slagged sinter has often been attributed to poor re-ducibility. The question of reducibility has also been raised in formulating quality standards for agglomerates such as nodules, briquettes, and pellets. In the present investigation, the solid phases formed during reduction were studied as a step toward a better understanding of the overall process. Equilibrium Studies The iron-oxygen equilibrium diagram shown in Fig. 1 reveal; a number of facts pertinent to the gaseous reduction of iron ores. This diagram is from the work of Darken and Gurryl, 2 and represents a correlation of the best available data. Four solid phases may exist during the complete reduction of hematite to metallic iron. These are hematite (Fe2O3), magnetite (Fe3O4), wustite (FeO), and iron (Fe). The wustite phase is a solid solution which is not stable below 570°C. At this temperature the solid solution undergoes a eutectoid-type of decomposition into the phases, magnetite and iron. Thus above 570°C, the diagram dictates that a hematitic ore should pass through a four-phase sequence on reduction to metallic iron. Below 570°C, only hematite, magnetite, and iron should appear. Information on the iron-oxygen system has been derived largely from CO and H2 reduction equilibria. The Fe-C-0 relationships have been studied extensively by R. Schenck and his coworkers and well summarized by H. Schenck.3 More recent studies have been made by Darken and Gurry.1, 2 Data from these sources have been combined and plotted in Fig. 2. With respect to the Fe-H-O system, the works of Emmett and Schultz4 seem the most reliable, and these data have also been included in Fig. 2. Certain physical properties of the solid phases of the iron-oxygen system are summarized in Table I. The crystallographic information is of special interest as much of the present work has been concerned with the X-ray analysis of the products of reduction. Reduction with Hydrogen The reduction of ore with hydrogen is the net result of two or more gas-solid reactions. Above 570°C, the reaction sequence may be represented by stoichiometric stages as follows: 3Fe203+ H2e2Fe,O, + H20 [1] 2Fe3O, + 2H2 6FeOw + 2H2O [2] 6FeOw + 6H3 ^ 6Fe + 6H=O [3] Fe2O3 + 3H2 ^ 2Fe + 3H,O. [4] These reduction reactions follow the general form: A (solid) + B (gas) e C (solid) + D (gas). This type of gas-solid reaction has been investigated by Langmuirl' who has shown that such reactions can occur only at the boundary between the two solid phases. Furthermore, a nucleus of the second phase must initiate the reaction. Once such an interface exists, the reaction proceeds through a layer of the solid reaction product (C). The specific mechanisms involved will depend a great deal on the properties and condition of this particular layer. A number of heterogeneous reactions such as the dehydration of single crystals of copper penta-hydrate and the calcination of limestone follow this type of process. It should be noted that the inter-facial type of reaction also occurs even in dense polycrystalline material which simulates a mono-crystalline behavior.

Jan 1, 1954

By John F. Elliott, Robert D. Pehlke

The rate of reaction of nitrogen with liquid iron and with binary alloys of iron and Al, Cb, Cr, Ni, O, S, Si, and W were measured. The surface active elements, oxygen and sulfur, decrease the rate markedly. The other alloying elements have very little effect on the rate except when they function as deoxidizers, The rates of absorption and de-sorption were found to be equal. Several possible reaction mechanisms which may control the rate are considered. In order to describe completely the behavior of nitrogen in steelmaking processes, the mechanisms of mass transfer between the gaseous and liquid phases, nitrogen and iron, respectively, must be understood. A kinetic study to determine the nature of these mass transfer processes was undertaken, and the results of that investigation are presented in this paper. Specifically, the study was directed toward determining the influence of alloying elements on the rates at which pure nitrogen gas at constant pressure is absorbed and desorbed by liquid iron solutions: A previous paper1 reported measurements of the equilibrium solubility of nitrogen in liquid binary iron alloys. The experimental data are presented and discussed first and then several possible rate limiting mechanisms are considered. PREVIOUS RESEARCH The results of several investigations on the rate of absorption of nitrogen by liquid iron and iron alloys have been reported.2-14 There is little agreement on the rate coefficients among these studies because of the influence of geometry on the experimental system and of the purity of the liquid iron. The rate was found to be influenced by alloying content for the binary liquid iron systems: Fe-Al,2 Fe-C,4,6 Fe-Cr,4,5,8 Fe-Mn,5 Fe-Ni,5,8 Fe-O,9,10 Fe-O,9,10 and Fe-Si.2,6,7,11 Several investigators12"15 have reported that the rate of nitrogen absorption is increased in a reducing gas mixture and decreased in an oxidizing atmosphere. A mechanism for the retardation by oxygen of the absorption of nitrogen has been described by Naeser18 as the formation of an iron oxide film at the gas-metal interface. The film slows the absorption rate until it disolves in the metal, thus permitting nitrogen to enter the liquid iron. Karwat17 discussed the influence of oxygen in terms of the formation of gaseous molecules of nitrous oxide which represent a higher energy form than the nitrogen and oxygen or the oxidizing gas and consequently constitute an energy barrier for the process. Fischer and Hoffmann10 showed that the oxygen markedly reduces the rate of absorption of nitrogen, the rate constant being one order of magnitude smaller in liquid iron containing 0.15 to 0.24 pct 0 as compared to that in melts containing 0.001 to 0.02 pct 0. Although Chipman and Murphy2 reported that the rate of absorption of nitrogen in liquid iron was essentially independent of temperature, Eklund3 found that the rate increases strongly with increasing temperature as did Kootz4 and Saito.5,6 Karnaukhov and Morozov7 also reported a large temperature coefficient, but related it to increased stirring at higher power inputs in an inductively heated system. All of the authors above have treated their results assuming that the reaction is diffusion controlled, that the surface reaction is in equilibrium with nitrogen gas, and that a "first order" reaction takes place. Karnaukhov and Morozov7 and Humbert; who studied the rate of absorption in a closed system by observing pressure changes in the system as a function of time, concluded that their results best fit a diffusion-type analysis. This was further verified by von Bogdandy, Dick, and stranski18 and Fischer and Hoffmann10 who found the initial rate of

Jan 1, 1963

By R. D. Pehke, J. F. Elliott

The solubility of nitrogen in liquid pure iron has been measured as a function of pressure and temperature. Sieverts' Law is obeyed at all pressures up to 1 atm and the temperature coefficient of solubility is 8 x 106 %/°C. The solubility of nitrogen in liquid iron alloys has been studied by the Sieverts' method and by the sampling method. The solubility is decreased by Al, C, Co Cu, Ni, 0. Si atzd Sn, and it is increased by Cr, Cb, Mn, Mo, Ta, W, and V. Interaction coefficients eN, for the effect of these elements on nitrogen dissolved in liquid iron, are reported. 1 HE effect of nitrogen upon the properties of steel may be desirable or not, depending upon the composition, processing treatment, and the use of the product. To understand the behavior of nitrogen in the various steelmaking processes, it is necessary to have reliable fundamental data pertaining to the solubilities, rates of solution, and chemical reactions of nitrogen in steel, particularly in the liquid state. In an effort to extend our knowledge in these areas, an investigation of the thermodynamics and kinetics of the solubility of nitrogen in liquid iron alloys was undertaken. The thermodynamic aspects of the study are presented in this paper. The solubility of nitrogen in liquid pure iron has received the attention of several investigators whose results are summarized in Table I. Reasonable agreement among the more recent researches1-l8 is found for the solubility of approximately 0.044 wt pct at 1600°C and 1-atm pressure of N,, for the temperature coefficient of solubility being small but positive, and for adherence to Sieverts' law for pressures up to 1 atm. The solubility of nitrogen in a number of binary liquid iron alloys has been studied: Fe-A1,3 Fe-As,16 Fe-C,3, 71 18, 19, 20 Fe-Cr ,5,7, 10, 11, 15 21 Fe-Co,'4, '6 Fe-Cu,'6 Fe-Mn,10, 11, 18 Fe-Mo,13, 16 Fe-Ni,10, 11, 13, 15, 16 Fe-O,16, 17, l9 Fe-p,7 Fe-S,l6 Fe-Sb,16 Fe-Se,lg Fe-Si,4) 97 '77 187 19>2° Fe-Sn,16 and Fe-v.51 '3 A few ternary systems have been studied, most of which were reviewed by ~angenberg" who applied the methods suggested by WagnerPz3 Chipman,24 Darken and G~rry, and Morris and Buehl26 for predicting solubilities in multicomponent alloys. Langenberg22 and Humbert and Elliott15 compared the solubilities computed from simple binary alloy systems with experimental measurements of nitrogen solubilities in liquid iron containing two or more alloying elements. The interaction coefficients for nitrogen in

Jan 1, 1961

By O. D. Gonzalez, H. A. Wriedt

Alloys ranging from pure iron to pure nickel were saturated with nitrogen gas at 918°, 999°, and 1217°C and analyzed. The solubility of nitrogen at 1-atm pressure was obtained as a function of nickel content for the whole range of iron-nickel alloys and the temperature dependence of solubility is shown for the experimental alloy compositions. The heat of solution and the entropy of solution of nitrogen in the alloys as func-tiom of composition up to 41 pct Ni are given. 1 HE acquisition of more exact thermodynamic information for systems composed of iron with various alloying elements continues to interest ferrous metallurgists. The interstitial alloying elements, especially carbon, hydrogen, and nitrogen, comprise a field of particularly intense study. A recent investigation of carbon solubility in Fe-Ni alloys held at 1000°C in gases of fixed carbon activity showed an unpredicted minimum in the solubility near 72 pct Ni.1 The present investigation was undertaken to see if a similar anomaly in solubility existed for nitrogen in Fe-Ni alloys. The solubility of nitrogen in ? iron at known nitrogen activities has been measured several times.&apos;-&apos; Corresponding information for the solubility of nitrogen in solid Ni and Fe-Ni alloys is not available. Juza and sachsze9 reported that metallic nickel dissolved 0.07 pct N in equilibrium with Ni3N at 445°C. This result was regarded as dubious by Turkdogan and Ignatowicz10 who found less than 0.0004 pct N in nickel specimens equilibrated at 600°C with ammonia-hydrogen mixtures having fugacities of N2 gas up to 35,000 atm. Such fragmentary reports indicate only that nitrogen gas has a very much lower solubility in nickel than in iron. MATERIALS A purified iron from Battelle Memorial Institute was used for solubility measurements. This grade contained more than 99.95 pct Fe, the highest content of any measured individual impurity being 50 ppm for each of As, C, Cu, and Pb. The nickel and iron-nickel alloys, supplied by the International Nickel Co., contained 0.02 to 0.04 pct C, 0.17 to 0.19 pct Mn, 0.04 to 0.08 pct Si, < 0.01 pct P, <0.03 pct S, and <0.20 pct Co as principal impurities. The Fe-0.59 pct Mn alloy was of zone-melted quality, the highest concentration of any individual impurity being 15 ppm Al. The "prepurified" grade of nitrogen gas from Matheson Co. had a nominal nitrogen content of 99.996 pct. An "electrolytic" grade of hydrogen gas from National Cylinder Gas. CO. in which the principal impurity is water vapor was admixed with the nitrogen to prevent oxidation of these specimens. This grade was preferred to a purer grade because the presence of some oxygen helps in suppressing contamination of the specimens, for instance, by silicon from the porcelain furnace tube. EXPERIMENTAL PROCEDURE The specimens were equilibrated in vertical tube furnaces through which a 99 pct N2-1 pct H2 gas was passed at ambient pressure. For the 915", 918", and 999°C runs, a wound-resistance furnace was used, and for the 1217°C runs, a furnace with a tubular silicon carbide element was used. The temperature control system on both furnaces was similar: a Weston "Celectray" controller with a modulator, responding to the electromotive force of a thermocouple with its hot junction in the furnace, switched the furnace current between high and low settings. The temperature distributions in the working zones of the furnaces were determined with calibrated Pt/Pt-10 pct Rh thermocouples. The overall uncertainty in temperature was estimated to be ±3° at 915° to 999°C and ±5° at 1217°C . Consideration of the small temperature coefficients of the equilibria studied, together with the error limits in the method of analyzing for nitrogen, showed that further refinement in temperature uniformity would give no improvement in resultant accuracy. The reactive gas was generated by mixing together "prepurified" nitrogen and "electrolytic" hydrogen directly from cylinders commercially supplied. The flow rates of nitrogen and of hydrogen were fixed by setting with bleeders the pressure drops across the individual calibrated flow meters of constant resistance. For any given alloy, the solubility of nitrogen at 1-atm pressure (% N) was obtained from the analyzed nitrogen content of equilibrated specimens (% Na)according to Sieverts&apos; law, thus

Jan 1, 1962

By D. C. Hilty, W. Crafts

Determination of the solubility of oxygen in iron containing silicon, or manganese, or both, has confirmed the earlier work on silicon, shown that manganese is more effective than expected, and has demonstrated that the combination of silicon and manganese is a much stronger deoxidizer than might have been inferred from their individual effects. SILICON and manganese are of primary impor-tance in the deoxidation of steel, and a study has been made at the Union Carbide and Carbon Research Laboratories, Inc. of the solubility of oxygen in iron in the presence of silicon and manganese, singly and in combination, as a basis for more effective control of inclusion formation. The investigation was carried out with the same equipment and experimental procedure that were used in the determination of the oxygen solubility of iron containing aluminum&apos;. The heats were melted in a rotating crucible furnace under an argon atmosphere. Bath temperatures were controlled by platinum/platinum +10 pct rhodium thermocouples to attain equilibrium at 1550°, 1600°, and 1650°C, so that oxygen solubility isotherms were determined directly without requiring calculation from a temperature coefficient. A detailed description of the furnace and procedure is given in the paper on aluminum deoxidation1. Oxygen solubility in the presence of more than 0.10 pct silicon was found to approximate earlier determinations closely, but some deviation was observed at lower silicon contents. Manganese appeared to be more effective than had been anticipated. The combination of silicon and manganese lowered the solubility for oxygen markedy as compared with either silicon or manganese alone. Although the solubility for oxygen in the presence of both silicon and manganese represents only a part of the information needed to control inclusion formation, it suggests a partial explanation of the benefits of this combination that have been observed in practice. Manganese and Oxygen in Liquid Iron: The study of oxygen in iron in the presence of manganese was carried out in magnesia crucibles of the type furnished commercially for laboratory induction furnaces. Alumina crucibles were unsatisfactory, because the work on the aluminum-oxygen equilibria in iron&apos; had shown that merely melting iron in an alumina crucible limited the solubility of oxygen to a level considerably below that anticipated for moderate concentrations of manganese. Silica crucibles appeared unsuitable because of the dangers of excessive fluxing by the melt. Moreover, as will be shown later, melting iron in a silica crucible also limits the oxygen content below that in equilibrium with

Jan 1, 1951