V ACUUM INDUCTION MELTING: •

the revolutionary influence ln steelmaking by F. N. Darmara

THE HISTORY and antecedents of vacuum induction melting have been discussed adequately in various papers. ' -4 Briefly, the first known industrial use of vacuum induction melting was by W. Rohn during the 1920's. Unfortunately, the vacuums utilized were crude by today's standards (2 to 5 mm). This effort seems to have been unsuccessful in initiating an industrial trend. It may be that the economics of the crude process were too poor for the returns obtained. However, vacuum induction furnaces were introduced in well-tailored laboratories for the edification and occasional use of researchers, but for most industrial pyrometallurgists, it was a gadget. The impetus of atomic energy requirements during and shortly after World War II led to immense strides in the development of pumps with comparatively high capacities at adequate vacuums (1 to 5 microns), enabling the construction of semi-industrial furnaces. This was not, however, enough to kindle the firethere had to be a need and this was supplied by the advent of the jet engine with its requirements for materials of extreme cleanliness and strength. In 1945-1946, J. Nisbet at General Electric Co. and Darmara and Huntington at NACA had small furnaces for the melting of cobalt and nickel-base alloys. About the same time, J. W. Moore at NRC had a furnace in operation for the production of pure iron and copper. By 1950-1951 the demands of the jet age and the newer higher-thrust engines strained the possibilities of conventional air melting methods to the limit. Vacuum induction melting was tried again to resolve the difficulties in meeting engine requirements. ' -4 This time the seed fell on fertile ground, and there has been continued, ever accelerating growth and progress ever since. By 1958 a 2%-ton furnace" was in operation and smaller furnaces were quite commonplace. It was in 1961 that F. N. DARMARA is president of Special Metals Corp., New Hartford, N. Y. This article is published by permission of the Vacuum Metallurgy Div. of the American Vacuum Society. 42-JOURNAL OF METALS, DECEMBER 1967

J. Nisbet showed foresight and courage by venturing into the construction of a 6-ton furnace. Given the circumstances, it was a great act of faith and was greatly instrumental in minimizing the skepticism of operating personnel as to the value of vacuum induction melting as a production tool. A 15-ton cold charge furnace is in operation and one of the following papers at this Symposium gives design criteria of another 15-ton furnace.

Theoretical foundations Vacuum induction melting as a pyrometallurgical process has been an ever-growing success because it is based on sound thermodynamics, allowing the control of two thermodynamic parameters. Non-vacuum processes can only control temperature. In vacuum, however, the control of both temperature and pressure is possible. As thermodynamic calculations predict, this opens up whole new regions of permissible metallurgical manipulations. However, operations at low pressure also bring concomitant reactions which do not occur or are not as severe under atmospheric conditions. Melt attack on crucible materials is one such problem. During the last decade there has been much work done on the thermodynamic foundations of vacuum induction melting.5-17 The most lucid expositions of the subject are in Schaffer's'· and Winkler's" papers. The various reactions occurring in vacuum induction melting such as deoxidation, desulfurization, degassification, dissociation, volatization, and crucible-melt reactions, are covered in these works and thermodynamic calculations are given. It can be said that the thermodynamics of the process are well understood and explored. However, as KinglO • ll points out, "Thermodynamics is essentially concerned with equilibrium states and is not concerned with considerations of mechanisms or path. While this is its greatest strength, it is also its greatest weakness; the information it affords is not on reaction rates and mechanisms."

In vacuum metallurgy and specifically in vacuum melting, it is the kinetics that are determining. How close to equilibrium the reaction may approach during a reasonable time is a matter of kinetics. Furthermore, most conditions prevailing during melting, holding, or refining are only metastable or steady state conditions. It is also important to realize that most classical kinetic information is concerned with homogeneous reactions, whereas in vacuum melting the most important reactions are heterogeneous. It is for these reasons that vacuum induction melting is the most flexible vacuum melting process, for it allows the independent control of time, temperature, pressure, and mass transport (by means of stirring). During the last ten years there have been many theoretical studies on the kinetics of vacuum metallurgy.,o-15 There has also been an extensive amount of experimental data obtained on the same subject, which will be discussed later. As is pointed out in the references cited, most vacuum metallurgical reactions are heterogeneous, taking place at two- or three-phase boundaries. Since the temperatures involved are high, reaction rates are so fast that the rate-limiting steps in all probability are diffusion, mass transport, and nucleation of a new phase. In vacuum induction melting most reactions occur in fluids (gas and melt). (Neglecting the depletion of surface elements in crucibles, which would be transport in a solid and governed by Fick's Law.) Mass transport can be speeded up by forced convection and modern furnaces do in effect provide stirring independent of the heating power, which means that temperature can be held constant at a desired value while mass transport is accelerated. The maximum forced convection possible would be the point at which metal is thrown out of the crucible. It is easily seen that even with stirring, the surface-area-to-volume ratio of a crucible will play an important part in the speed of the over-all reaction. In practice this reviewer feels that the role of stirring in vacuum melting has been underestimated in spite of the highlighting it has received in theoretical discussions. The disparity between various experimentalists is, in all probability, due to variations in mass transport caused by geometry and forced convection differences. However, it is not felt that the situation is as hopeless as Kingll states: "It is to be noted that kinetic data obtained from one experimental set-up are not normally useful for other set-ups because of lack of knowledge of the convection conditions."* There is no question that this statement is in great measure true today. However, it can be conceived that in the near future some means of calibration could be devised so that kinetic

data from vacuum induction melting experiments could be interchangeable between various laboratories. This calibration could possibly take the form of obtaining experimental results on a known reaction in one installation and calibrating another installation to these values. Thus, differing experimental set-ups could be normalized to the same boundary conditions. This semi-empirical approach, if feasible, could be of great practical importance in the exchange of data and their utilization in practice. It should be cited, as Machlin does,13 that kinetic calculations must assume a mechanism or path on which calculations are based. As he points out, even for the simple reaction C+O = CO(g), many models of the path of the reaction can be imagined and arguments presented to justify each, but only experiment can distinguish the real mechanism. Machlin, applying a streamline flow model to describe the behavior of the melt in the vicinity of reaction surfaces, analytically investigates the kinetics of vacuum distillation, vacuum-melt surface reactions, crucible-melt reactions, and boiling. His quantitative predictions are in good agreement with the then available data and also some of the data obtained since.' • He is also in agreement with Samarin's'7 observation that the limit to which a full boil can reduce the oxygen content for a melt depth of 50 cm is insensitive to pressure variations when the pressure is below about 0.05 atmosphere. Samarin arrives at his conclusion from a thermodynamic argument and Machlin from a kinetic one. But, Samarin concludes that very high vacuum is not necessary for complete deoxidation, whereas Machlin points out that after cessation of the boil the deoxidation proceeds via the surface effusion of CO until a steady state is achieved between oxygen pick-up from the crucible and oxygen elimination from the surface. Both of these arguments on the efficacy of the boil rest on the energy necessary to generate a bubble of CO. It is well agreed that the bubbles formed during the boil are not homogeneously nUcleated.'",21 The source of nucleation can possibly be either at the cavities already present in the crucible or at the negative pressure regions in vortices induced by violent stirring. It is most probable that in most cases the stirring is not violent enough and the nucleation occurs at the crucible metal interface. However, nucleation in vortices cannot be ignored. Some preliminary data at this reviewer's laboratory seem to indicate that such is indeed the case and low oxygen values are obtained rapidly by extreme agitation. However, the results are not conclusive. A diligent search in the literature has not revealed any papers dealing with this subject.

Experimental results * This statement is based on the assumption that the effective boundary layer thickness cannot be calculated as its value depends on the diffusion coefficient and the flow velocities past the boundary.

A review of all the experimental work done during the last ten years is certainly a task beyond the DECEMBER 1967, JOURNAL OF METALS-43

powers of this reviewer. What will be attempted is a review of papers either in significant areas or utilizing techniques that may have influence on future practice and future research. The equilibria of MgO-Fe-C-O has been investigated by several researchers."5 - 2 " They all agree that there is a minimum oxygen level below which it is not possible to deoxidize the bath. Turillon's"9 experiments seem to substantiate Machlin'slO predictions based on theoretical calculations of this steadystate reaction. These results seem to indicate that in a magnesia-alumina crucible (MgO = 65w/o, * AIO s = 25 wlo, Si02 = 5.8 wlo, CaO = 1.5 w/o) the value of oxygen remains at 10 to 15 ppm as long as the carbon is above 0.006 w/o. When the carbon falls below this value, the oxygen begins to rise. The minimum value of oxygen is almost identical with that of Fisher and Hoffmann."8 It should be noted that both crucibles contained SiO, and this minimum value may be reduced further with pure fused MgO crucibles. Bennett et al. '1 obtained results that are slightly lower (5 to 10 ppm). If these values are indeed meaningful, i.e., not due to analytical discrepancies, it may be surmised that their crucibles were purer than the previously mentioned investigators. One of the predictions made by theoretical calcUlations18 was that the steady-state oxygen values would be independent of the carbon content, providing the final carbon stayed above a minimum value (about 0.006 for pure iron). This is borne out by the results 18- 21 providing, however, that steadystate is reached. The deoxidation of other metals and alloys by carbon in magnesia and other crucibles has been explored by numerous workers.""-21l In general, the mechanism of oxygen removal is analogous to that in iron. The results of Linchevskii'" can be taken as an example. He studied the melting of 18 wlo chromium steels at two temperatures, 1500 and 1600° C, in magnesia, alumina, and zirconia crucibles under pressures of 50, 1, and 0.02 mm Hg. In general, the data indicate a steady and relatively pressure-independent but decreasing loss of carbon with time. On the other hand, the oxygen content reaches a minimum in less than 30 minutes and begins to rise. The minimum steady-state oxygen value is strongly dependent on the pressUre and crucible material, and somewhat insensitive to temperature. The lower the pressure and temperature, the lower the oxygen value. However, at the higher pressures (50 and 1 mm) the steady-state minimum is of short duration and the oxygen content begins to rise. The effect of temperature at the lowest pressure is strongly influenced by the crucible material. In alumina, an oxygen content minimum is attained in 25 to 30 minutes, and these values remain about the C

• w /0 = weight percent.

44-JOURNAL OF METALS, DECEMBER 1967

same up to 70 minutes for both 1500 and 1600° C.; whereas in zirconia at 1500° C. the minimum value of 30 ppm is attained in 20 minutes and rises to about 40 ppm at the end of 70 minutes. On the other hand, at 1600° C. there is a sharp minimum of 70 ppm at 15 minutes and the oxygen begins to rise immediately and attains about 120 ppm in 70 minutes. Linchevskii attributes the deterioration of the deoxidation at the higher temperature to the increased solubility of oxygen in iron-chromium alloys, but there is undoubtedly an added factor related to crucible composition. These data are of further interest in showing the importance of kinetics in these reactions and the absolute necessity of understanding the kinetic mechanisms involved. In the Fe-O-C system previously reviewed, there seemed to exist general agreement that the oxygen value reached a minimum and stayed there as long as sufficient carbon remained to sustain the deoxidation reaction. Unfortunately, Linchevskii does not give simultaneous carbon-oxygen values. But, if the data given were all generated within the same experiment, the minimum carbon values can be estimated roughly. For example, at 1600° C. and 50 mm in both alumina and zirconia crucibles, the value would be about 0.25 wlo after 30 minutes. At the same temperature in zirconia at 0.02 to 1.0 mm, the minimum carbon would be about 0.05-0.08 wlo for the same time period which corresponds with minimum O2 content. At the end of 70 minutes these values have dropped to a minimum of 0.02 wlo carbon at 0.02 to 1.0 mm while the oxygen has risen to a maximum of about 120 ppm. In alumina, however, for the same conditions of pressure and temperature the minimum carbon corresponding to the minimum of oxygen is 0.17 wlo for 1 mm, for 0.02 mm the carbon at 30 minutes is 0.18 wlo, at 70 minutes 0.16 wlo, while the oxygen keeps dropping slowly, showing no minimum. Some general observations on the effect of pressure on the reactions involved may be made based on these rather incomplete data. The lower the pressure the more efficient the deoxidation reaction versus the pollution reaction. The rate of attaining the minimum oxygen is not too dependent on pressure; however, the absolute value of this minimum oxygen is strongly pressure dependent. The lower the pressure the lower the carbon necessary to attain minimum oxygen values. Higher temperatures or less crucible stability increases the efficiency of the pollution reaction and presumably the higher the carbon necessary for minimum oxygen values. It is obvious also that chromium as a fairly strong oxide former has decreased the efficiency of the carbon deoxidation reaction in iron-chromium alloys as compared to pure iron. One of the most original papers on deoxidation in vacuum is the paper by Moore 30 who investigated

the deoxidation of iron with methane. As Meysson and Rist31 point out in their paper dealing with theoretical calculations of the same subject, this technique is especially suitable for vacuum induction melting since at the end of the deoxidation period the metal is saturated with hydrogen (calculated values 24 ppm at 1 atm and 2.4 ppm at 0.01 atm). In vacuum this would be no problem because at 76 microns this value would drop to 0.2 ppm. Moore observed that a temperature of 1600° C. was required for pyrolysis and the rate of carbon deposition on the melt increases rapidly above this temperature. If the gas is not shut off at the precise end-point, carbon pick-up in the melt is very rapid. This end-point was determined by the change of color of the burning gas at the exit end of the pumps. By using this end-point, iron heats containing 10 to 20 ppm of oxygen with 10 to 20 ppm carbon were obtained. Pure nickel and cobalt heats contained 10 ppm oxygen and 10 to 20 ppm carbon. The pressure of the methane in the chamber is not mentioned, but since reference is made to the gas being pumped one can surmise that it was below 1 atm. The great advantage of this method was total lack of boil, which in practice would be of major importance. Nitrogen removal. The removal of nitrogen in vacuum melting has been studied by numerous investigators.'8,!lO,Z7,33 In metals or alloys not containing strong nitride formers, removal of nitrogen at pressures of 5 to 20 microns occurs rather fast and the nitrogen drops to fairly low values-20 ppm. This removal is quite rapid during the carbon boil period and slows down thereafter. In iron-chrome and ironchrome-nickel stainless steels, on the other hand, there is more than an order of magnitude difference between calculated nitrogen equilibrium values and values actually obtained. The removal of nitrogen is fast at first and then slows down and proceeds asymptotically. Presumably, if melts could be held a long time very low values of nitrogen should be attainable. Desulfurization. The problem of desulfurization during vacuum melting has preoccupied metallurgists since the advent of vacuum melting. This is easy to understand as the elimination of sulfur is most certainly one of the important problems of pyrometallurgy. In the electric arc furnaces, suitable slagging techniques can allow the reduction of sulfur to low values (0.003 w/o) rendering the element more or less innocuous to properties. These techniques, it was felt, were not readily applicable to vacuum induction melting; therefore, during the last decade various approaches to desulfurization have been investigated. Machlin'" studies the elimination of sulfur in nickelbase alloys by the generation of sulfur dioxide and its subsequent removal from the melt. The oxygen content of the melts varied from 0.12 to 0,48 wlo, and

sulfur was reduced from 0.03 wlo to 0.005 wlo in 60 minutes. Theoretically this value should drop to about 0.0006 to 0.0008 wlo in the next hour; however, no data are available to substantiate this. The technique is quite cumbersome in practice and reduction of this amount oxygen by the standard solid carbon addition is quite tedious, leading to undesirable production problems. It is quite possible, though, that this technique could be coupled with methane reduction, eliminating the danger of violent boil. Still, the times involved might be quite long and economically disadvantageous. The sulfur dioxide reaction is difficult to apply to iron melts due to relatively high value of the wlo sulfur in equilibrium with the usual levels of S02 partial pressure attainable. The desulfurization of nickel and iron by dissociation of the sulfides and evaporation of elemental sulfur normally would not be expected to take place; however, additions of alloying elements such as carbons and silicon which raise the activity coefficient of sulfur allow the reduction of sulfur in vacu0 3 ,,35 in cast iron and 3 to 4 % silicon transformer steels. In the case of the transformer steel," it is not quite clear whether the loss is by the elemental sulfur evaporation mechanism or through the sulfur dioxide reaction. Desulfurization by formation of a sulfide of a metal with higher affinity for sulfur than the base metal, such as calcium and cerium, has been studied:7 ,38 Ward and Hall" were successful in reducing sulfur by the use of lime. They used calcined lime, a crucible coated with a paste of commercial slaked lime, and a crucible coated with a paste of freshly slaked quicklime. The freshly slaked quicklime worked the best and sulfur was reduced to 0.003 to 0.002 wlo levels quite easily. Their conclusion is, "Sulfur removal at all pressures can be attributed to a reaction with silicon in the melt and lime, but it is also possible that some desulfurization takes place at low pressures by a reaction with carbon and lime." Volkov 38 conducted experiments not only in a laboratory furnace but in a large 500-kilogram commercial furnace. He concludes that the most expedient method of desulfurizing in a vaucum induction furnace is by addition of a mixture of 90% freshly burned lime and 10% fluorspar in the form of finely ground powder from which dust has been separated. This powder is placed at the bottom of the crucible before charging and amounts to 2 to 3 % of the mass of the charge. Desulfurization proceeds during the melt-down period under vacuum. The sulfur content of the metal after melt-down is reduced to 0.002 to 0.004 wlo in 10 to 15 minutes and is practically independent of the initial sulfur content of the metallic charge. He found that the degree of desulfurization is dependent on the oxygen content, and that in low carbon steels where the carbon content is DECEMBER 1967, JOURNAL OF METALS-4S

less than 0.05 wlo high vacuums had to be used. When vacuums of the order of 1 mm Hg were used the desulfurization did not proceed very well. In other words, oxygen had to drop to 20 to 30 ppm before the reaction could proceed. When oxygen rose above these levels desulfurization stopped and even reversed itself and the metal picked up sulfur from the slag. These data indicate that in this lime-induced desulfurization reaction low oxygen is absolutely necessary to allow the desulfurization to proceed. However, if the oxygen is low, desulfurization is extremely rapid and very efficient. The other element that has, of course, a high affinity for sulfur is cerium. Unpublished work done in our laboratory indicates that cerium is a very effective reducer of sulfur in practically all base metal charges. Since cerium itself is also an effective deoxidizer, the melt has to be properly deoxidized just prior to the addition of the cerium; otherwise, the cerium is used up as a deoxidizer and seems to deposit on the walls of the crucible as cerium oxide. On the other hand, if the cerium is added at the point of lowest oxygen, cerium sulfide is formed and again seems to deposit on the walls of the crucible. Inasmuch as cerium is volatile, it is somewhat difficult to add to the melt in vacuo since the tank has to be back filled with argon during the addition of the cerium. When the vacuum is reapplied the unused portion of the cerium evaporates. However, since any remaining cerium in the metal-even in parts per million-has profound effects on the properties, the permissible limits of cerium content have to be controlled very tightly. The control of residual cerium in the melt is quite a chore. The technique, though, offers great hopes for commercial production when the quality control problems associated with maintaining very close limits on the ppm cerium residual levels are solved. Janiche and Beck36 investigated the desulfurization of pure iron by means of the addition of elements that form volatile sulfides and themselves are volatile. Thus, both the sulfide and the residual metal would evaporate. They used bismuth, tin, and antimony. Their experimental procedure apparently did not have all independent variables under control as the results were quite erratic. The experiments do show a certain measure of success in desulfurization, but the levels of residual elements remaining in the melt might be considered intolerable. Unless future experiments are more successful, it is doubtful that this method of desulfurization will prove of value in practice. Evaporation of impurities. In the early days of the vacuum induction melting of superalloys, it was felt that the great improvements in property-especially stress rupture and creep-were due mainly to the efficiency of the carbon deoxidation process which did not tie up and produce any solid oxides. It was soon discovered, however, that there were 46--JOURNAL OF METALS, DECEMBER 1967

anomalous behaviors. Virgin heats that had been properly deoxidized with carbon exhibited low stressrupture ductilities. It had been known also for some time that in air-melted high temperature alloys there had been small additions of boron; however, it was thought that the action of the boron was as a final deoxidizer and degassifier. About ten years ago it was discovered that 30 to 40 ppm of boron was absolutely necessary in developing adequate stressrupture ductility in high temperature alloys. At about the same time, it was learned that heats that contained 10 to 15 ppm of lead exhibited a high degree of brittleness. It became apparent that impurities in the ppm range could not only exert a beneficial effect but also an extremely pernicious one. It then became obvious that a part of the so-called refining cycle, in addition to lowering the oxygen, was actually useful in boiling out volatile impurities. The first work actually dealing with the effects of trace contents of impurity elements on the properties of nickel alloys were conducted by Wood and Cook." They showed that some elements in extremely minute quantities could affect creep properties of nickel-base alloys to a great extent. A good deal of work by many investigators was carried out'·...7 on the rate of evaporation of trace elements during vacuum induction melting. As would be expected, the behavior of impurity elements during vacuum induction melting is to a certain extent dependent on the matrix material. Turillon'" finds that in 80-20 wlo nickel-chromium alloy, arsenic, tin, and antimony are not lost at all, whereas selenium, copper, bismuth, lead, and tellurium are reduced to very low levels by evaporation. On the other hand, Fischer and Hoffmann'" in their investigation of the behavior of iron melts containing phosphorus and arsenic found that arsenic could be gotten rid of somewhat easily. Phosphorus apparently can be evaporated as long as there is sufficient oxygen. This presumably indicates that the phosphorus is eliminated by means of a mechanism similar to the loss of sulfur by the formation of a volatile oxide. There are also anomalous behaviors among the results obtained by various investigators. It should be noted that distillation or evaporation reactions are controlled by the kinetics of the melting conditions. The rate of these reactions will be strongly controlled and influenced by the surface-to-volume ratio of the crucible geometry, the diffusion to the surface layer, the stirring power of the crucible, and possibly more important, the condition of the surface. It is known that certain impurities poison the surface of the melt so that evaporation or degassing reactions are slowed down; for example, the presence of oxygen limits the effusion rate of nitrogen. It has also been observed empirically that when stirring power is increased, the evaporation rate of elements in the melt increases remarkably. It is therefore not surprising that results are different between different

investigators in this field dealing with rather subtle phenomena. It should be emphasized that this field of impurity elimination is probably one of the most important ones in the future exploration of vacuum induction melting. This pyrometallurgical technique of vacuum melting allows all the controls necessary to understand and facilitate the elimination of impurities from metals and, subsequently, allows the proper "doping" of the alloys as required for specific properties (via controlled trace element additions). Effect of vacuum induction melting on properties.

Many papers have been published during the last decade on the subject of vacuum induction melting and its effect on properties. It is in no wayan exaggeration to say that there was hardly any type of steel or alloy that was melted by this technique whose properties did not benefit from it. It is beyond the scope of this article to review all that has been investigated. The main purpose of citing this is to point out that much work has been done successfully in this field.

Advances in design and engineering Another paper at this meeting discusses in detail the advances in engineering that have made possible the large furnaces being constructed today. Only some of the high lights of these advances will be touched upon here. The development of triplers or transformers that triple the line frequency of 60 cycles to yield 180 cycles has been instrumental in allowing the reduction of power cost in the construction of large furnaces. It must be conceded, however, that until the state of the art had made large furnaces practicable and the demand had made them desirable, 180cycle current could not have been utilizable. Nevertheless, these triplers have been of great importance in reducing the capital cost involved in large installations. Another area of importance has been the improvement in insulation of the induction coils, allowing the utilization of higher voltages and lower currents. Without this, corona effects would have mitigated against higher voltages. Large furnaces would probably have still been built, but at a higher cost. The development of coil designs allowing a controllable degree of stirring regardless of the heating power utilized is also of great importance. Previously it had been necessary to utilize dual frequency and dual generators to accomplish this. Previous discussions in this paper have shown the crucial role that stirring can play in the control of the kinetics of the reactions in vacuum induction melting.

Capacity and economics The economics of vacuum melting have undergone a radical reduction in the last decade. Ten years ago furnace size determined the direct cost of melting.

The bigger the furnace, the lower the direct cost. In a half-ton furnace this cost is around 50¢/lb.; however, this cost curve is asymptotic to furnace size. Beyond a given size the difference in cost between two furnaces, one twice the size of the other, becomes negligible. The difference in cost of melting between a 15-ton and a 30-ton furnace is probably, at the most, no more than l¢/lb. At this point other considerations become much more important, such as the relation between scrapped heats and raw material cost. Today vacuum melting technology has reached the point where melting cost will no longer be the deciding factor in determining furnace size. Such factors as scrapped heats, melting demand in terms of production which, in turn, determines the efficiency of utilization, size of orders to size of furnace, and so on are relatively much more important in determining economic profitability than mere size of furnace. Some years ago Nisbet projected a yearly vacuum induction capacity of about 1,000,000 tons for the U. S. by 1970. For 1967, his prediction was about 200,000 tons per annum. It is rather hard to know what U.S.A capacity is at present. A rough estimate is about 90,000 tons per year. Considering that his estimate is based on an extrapolation from roughly 750 tons per year 10 years ago, this should be considered a very good prediction. It should be realized that when this estimate was made most people considered it visionary to say the least.

Future progress and trends To prophesy the events of the next decade is certainly an audacious task for anyone to undertake. This reviewer is emboldened in this attempt by the comforting thought that 10 years from now he probably will be away in retirement and safe from the consequences of prophecies that did not materialize. As furnace capacity increases and the economic stake becomes larger, problems that were trivial 10 years ago will become crucial in the coming years. Problems of scrap utilization, refining of impure and cheaper raw materials, and desulfurization will become of pressing economic importance. The solution of these problems will require the laying of a vast foundation of experimental data. It is obvious in reading the literature quoted in this paper that some of them still exhibit a non-understanding of the reactions taking place in vacuum melting. Important parameters are not controlled, the kinetics of the reactions are not appreciated, data are not translatable from one laboratory to another. This is a problem of insight and familiarity, and the next few years will see an improvement. A knowledge of gas-metal reactions (an extension of Moore's work) and its application and utilization DECEMBER 1967, JOURNAL OF METALS--47

will find the importance it deserves. Slag-metal reactions-their understanding and utilization in production-will become essential. There will be an extensive exploration of the effects of impurities and their elimination. This field is now handicapped by the lack of chemical or other analytical methods accurate to levels less than 5 ppm. Either analytical methods or methods of measuring the properties of grain boundaries where these impurities congregate will have to be developed. After the elimination of impurities has been accomplished, the effects of "doping" will have to be learned. If in the last 10 years we have taken some of the ppm's out, we have to learn to put them back again in controlled amounts. Problems of refractory life and contaminations will have to be explored. Our knowledge of ceramics for vacuum induction melting is quite limited and refractory problems are the most costly items in commercial production. What is really required is a stable refractory with the thermal stability and shock resistance of silica. Complex ceramics tailored to meet the requirements of this technique will have to be developed; if not, techniques of composite crucible wall construction will have to be evolved. Even today there is no limit to the size of furnace that can be constructed. From now on, furnace size is not a limitation. No doubt furnaces of 5G or even 100 tons will be constructed in the future. By 1975 capacity may approach 1,000,000 tons. A large proportion of stainless steels, electrical and transformer steels, tool steels, and other specialty alloys will be melted by the vacuum induction process. As the past 10 years have been rewarding, we can see continued excitement for the next 10 years in this field which not so long ago was classified as a laboratory curiosity.

References t F. N. Darmara and J. S. Huntington: Experiences in Production Vacuum Melting and Nickel-Base Alloys, ASME Paper No. 55-A198, Nov., 1955. • F. N. Darmara, J. S. Huntington, and E. S. Machlin: Vacuum Induction Melting, Journal of Iron & Steel Institute. vol. 191. p. 266, March, 1959. 3 W. E. Britton, et al.: Vacuum Melting, Harvard Business School, April, 1957. • J. H. Moore.: Metal Progress, vol. 64, pp. 103-105, Oct .. 1953. 6 W. J. Pennington: Thermodynamics of the Vacuum-Induction Melting Process, paper from 1956 National Symposium on Vacuum Technology, pp. 209-214. of. C. Langenberg and S. Beer: Thermodynamics in Vacuum Metallurgy, lecture No.3, Second Vacuum Metallurgy Course, New York Univ., June 23, 1958. 1 T. R. Meadowcroft: Deoxidation at Low Pressures, Part 1. Ind. Heating, vol. 33, No.7, j)P. 1285-1288, July, 1966. 8 T. R. Meadowcroft: Deoxidation at Low Pressures, Part 2, Ind. Heating, vol. 33, No.8, pp. 1485-1488, 1490, 1492, 1525, August, 1966. • D. S. Kamentskaya: Some Theoretical Questions about Vacuum Metallurgy, paper from Use of Vacuum in Metallurgy, Akademiya Nauk, SSSR, Moscow, pp. 49-53, Henry Brutcher Translation No. 4364, 1958. ,. T. B. King: Kinetics in Vacuum Metallurgy, lecture No.4, Vacuum Metallurgy Course, New York Univ., June 23-27, 1958. 11 T. B. King: Thermodynamics and Kinetics in Vacuum Metallurgy, paper from Vacuum Metallurgy, Reinhold Publishing Corp., New York, pp. 35-38, 1958. 12 A. M. Aksoy: Thermodynamics and Kinetics in Vacuum Induction Melting, paper from Vacuum Metallurgy, Reinhold Publishing Corp., New York, PP. 59-78, 1958. 13 E. S. Machlin: Kinetics of Vacuum Induction Refining Theory, Metallurgical Society of AIME, Transactions, vol. 218, pp. 314-326, April, 1960. 14 O. Winkler: The Theory and Practice of Vacuum Melting, Metallurgical Reviews, vol. 5, No. 17, pp. 1-117, 1960.

48-JOURNAL OF METALS, DECEMBER 1967

"A. S. Darling: Low Pressure Metallurgy, Metallurgia, vol. 64, pp;" ~~-7~: AS~t~J:;:1. Production of Superior Quality Ultra-HighStrength Steel Castings by Vacuum-Induction Melting, paper from Vacuum Metallurgy Conference. Transactions. 1960, Intersciencc Publishers, Inc., New York, pp. 151-190, 1961. 17 A. M. Samarin: Deoxidation of Steel in Vacuum. paper from Vacuum Metallurgy, Reinhold Publishing Corp., New York, pp. 255265, 1958. "w. A. Fischer and A. Hoffmann: Properties of Iron and St0ri Melts in High Vacuum, Archiv flir das Eisenhiittenwesen, vol. 29, PP. 339-349, June, 1958. 19 P. P. Turillon and E. S. Machlin: Deoxidation of Pure Iron by Vacuum Induction Melting, Transactions of the Vacuum Metallurg:,>Conference, 1959, New York Univ. Press, New York, pp. 81-90, 1960. 20 W. A. Fischer: Effect of a High Vacuum on the Metallurgy of Iron, Archiv flir das Eisenhiittenwesen, vol. 31, pp. 1-9, Jan., 1960. "G. H. Bennett, H. T. Protheroe, and R. G. Ward: The Role of Carbon as a Deoxidizing Agent in the Production of Vacuum· Melted Steel, Journal of the Iron & Steel Institute, vol. 195. Part 2, p. 174-180, June, 1960. "B. D. Nikolayev, B. V. Linchevskii, A. A. Simkin, F. P. Edncl'al. et al: Investigation of the Deoxidation of Iron-Carbon Alloys During Production in Vacuum Induction Furnaces. Izv VUZ Chernaya Met, No.5, pp. 61-65, May, 1966. 23 Syan-Khua, Shao, Chzhi Shui, Chzhao Zhen'-Chuan' and Tsui Pei-Syun: Oxidation of Nickel With Oxygen From the CrucIble Material During Melting in a Vacuum Induction Furnace, Acta Met. Sinica, vol. 8, No.3, pp. 292-294, 1965. 24- Sawamura, Hiroshi, Toshisada Mori, Masao Yakushiji, and Hiromasa Inoue: Fundamental Study on Vacuum Melting of Pure Iron and Stainless Steel, Par 1, Crucibles, Ingotism and Recovery of Alloy Additions, Kyoto Univ., Faculty of Engineering, Memoirs. vol. 22, Part 2, pp. 236-248, April, 1960. 25 Tetsuya Watanabe: Deoxidation Equilibrium Caused by Carhon in Vacuum Induction Melting, Tetsu-to-Hagane (Iron & Stecl Institute of Japan, Journal), vol. 47, pp. 1670-1675, Nov. 1961. :!o V. I. Krasnykh, and V. 1. Sokolov: Production of Precision Alloys in the Vacuum Induction Furnace with Refining by Hydrogen, Stal, No.3, pp. 206-208, 1965. 21 B. V. Linchevskii: Behavior of the Components of Stainless Steel During Vacuum Melting, Izvest. VUZ-Chern. Met. pp. 70-76. Henry Brutcher Translation No. HB 5553, March, 1963. 28 B. D. Nikolayev. B. V. Linchevskii, F. P. Edneral, A. A. Simkin, et al.: Deoxidizing Iron-Chromium Melts in Vacuum, Izv. VUZ Chernaya Met, No.5, pp. 51-55, May, 1966. 2\> K. A. Matsarin, S. V. Kraskovskii, and F. P. Edncral: InHuence of Alloying Elements on Deoxidizing Capability of Cnrbon in Nickel Under Vacuum Conditions, Izv. VUZ Chernaya Met. No. 3, pp. 68-72, 1966. 30 W. F. Moore: Deoxidation Techniques for Vacuum-Induction Melting, Journal of Metals, vol. 15, No. 12, pp. 918-921, Dec., 196:). 31 N. Meysson and A. Rist: Theoretical Calculation of the Deoxidation of Liquid Steel by Bubbling of Methane, Rev. Met., vol. 62, No.2, pp. 121-126, Feb., 1965. 32 A. Simkovich: Variables Affecting Nitrogen Removal in the Vacuum Induction Melting of Iron- and Nickel-Base Alloys, JOU"NAL OF METALS, vol. 18, No.4, pp. 504-512, April, 1966. "G. A. Garnyk anI A. M. Samarin: Vacuum Metallurgy lof Steel), Deoxidation and Desulphurization in Vacuum, Izvestiya Akademii Nauk SSSR, Otdelenie Teknicheskikh Nauk, pp. 77-84. Henry Brutcher Translation No. 4054, May, 1957. 3.J. A. M. Samarin and R. A. Karasev: Transactions of 5th National Symposium on Vacuum Technology. San Francisco, 1959. Pergamon Press, New York, p. 35, 1959. 35 Ohno, Reiichi: Thermodynamic Considerations on DesulphurizCltion of Cast Iron in Vacuum Melting, Tohoku Univ., Science Reports of the Research Institutes, series A, vol. 12, pp. 353-367. Aug .. 1960. :16 W. Janiche and H. Beck: Desulphurization and Deoxidation of High-Purity Iron in the Vacuum Melting Process, Archiv fUr das Eisenhiittenwesen, vol. 29, pp. 631-642, Oct., 1958. 31 R. G. Ward and R. Hall: Desulphurization of Molten Steel by Solid Lime During Vacuum Melting, Iron and Steel Institute Journal, vol. 195, pp. 75-78, May, 1960. as S. E. Volkov, et al.: Desulphurization of Steel in Vacuum Induction Furnaces, Stal, in English, PP. 115-118, Feb., 1965. 39 G. M. Gill and G. W. Austin: The Behavior of Various Elements in Vacuum Steelmaking, Iron and Steel Institute, Journal. vol. 191, pp. 172-175, Feb., 1959. '0 R. G. Ward and T. D. Aurini: Mechanism of Alloying Elements During the Vacuum Induction Melting of Steel, Iron & Steel Institute, vol. 204, No.9, pp. 920-923, Sept., 1966. u M. Olette: Vacuum Distillation of Minor Elements From Liquid Ferrous Alloys, paper from Physical Chemistry of Process Metallurgy, Part 2, vol. 8, Metallurgical Society of AIME, Interscience Publishers, Inc., New York, pp. 1065-1087, 1959. Cl W. A. Fischer and A. Hoffmann: The Behavior of Iron Heats With Phosphorus and Arsenic Content in High Vacuum, Archiv fur das Eisenhiittenwesen, vol. 30, pp. 199-204, April. 1959. 43 A. Franklin: The Potentialities of Extremely Pure Materials. Engineers' Digest, vol. 23, pp. 78, 121, May, 1962. .. W. E. Duckworth and B. Appleby: Removal of Elements From Steel by Vacuum Treatment, BISRA Paper No. MG/C/l07/62. 26 pages, March-April, 1963. 4' P. P. Turillon: Evaporation of Elements From 80/20 NickelChromium During Vacuum Induction Melting, Transactions of Vacuum Metallurgy Conference, 1963. .J.6 W. A. Fischer and M. Derenbach: Investigation of Distillation Effects in the Vacuum-Melting of Iron Alloys, Part I, Theoretical Derivations and Investigations of Binary Alloys of Iron With Arsenic, Manganese, Copper, and Tin, Archiv filr das Eisenhilttenwesen, vol. 35, No.4, pp. 307-316, April, 1964. 47 D. R. Wood and R. M. Cook: Effects of Trace Contents of Impurity Elements on the Creep-Rupture Properties of Nickel-Base Alloys, Metallurgia, vol. 67, pp. 109-117, 1963.