Search Documents

By Theodore B. Winkler, John Chipman

In order to understand more fully the complexities of the reactions occurring between the liquid steel and the slag in the basic open-hearth steelmaking furnace, investigations in this country and. abroad have turned to laboratory equilibrium studies. This paper is one of a series of reports on investigations being conducted at the Massachusetts Institute of Technology. The experimental work was completed early in 1942. The slag systems found in the basic open hearth are very complex, being made up of at least eight major components and as many others whose effect might be classified as minor. The first two papers of the series, by Fetters and Chipman,l,2 as well as the paper by Taylor and Chip-man,3 dealt with the simplest possible slag system that would resemble those found in the basic ope), hearth—the slag system consisting of CaO-MgO2-SiOz-FeO. (Taylor's rotating induction furnace eliminated the necessity of contamination of the slag by MgO.) Having arrived at some reasonable conclusions regarding the physical chemis- try of the reactions between molten iron and the simple slags at equilibrium, the next step was to investigate more complex slag systems. The major purpose of the present investigation was to determine what factors control the distribution of phosphorus between molten iron and the more complex basic slags at equilibrium, and to what extent these factors are influential. The slags used contained CaO, MgO, SiO2, FeO, Fe2O3, P2O6, MnO, Al2O3 and CaF2, but not all of the slag samples contained all of these components at one time. Review of Literature Although a number of interesting papers have been published on the subject of dephosphorization of an iron bath, there are as yet no results from laboratory studies that give thoroughly reliable information on the dephosphorization reaction. Even for the supposedly simple Fe-O-P system involving iron-phosphorus melts and iron oxide-iron phosphate slags, the experimental results are contradictory and no formulas have been established that adequately fit the equilibrium conditions. The Fe-O-P system has been studied by Schackmann and Krings,4 von Samson-Himmelstjerna,6 Oelsen and Maetz,6 Bischof and Maurer,7 Diepschlag and Schürmann,8 and Herty,9 and the results of their investigations appear in Table I. A number of different equilibrium constants and distribution . Ratios

Jan 1, 1947

By J. S. Marsh

Joseph Priestley prepared oxygen on Aug. I, 1774, and noted with great surprise "that a candle burned in this air with a remarkable brilliant flame." On Aug. 2, 1774, some ironmaker possibly began to wonder if this substance could be used to increase his rate of production; at any rate, there is no disputing the fact that the notion of using oxygen for steelmaking in concentrations greater than that available in great abundance—as air—has been in existence for years. Action was delayed by the lack of enough oxygen at a cost within reach of the steelmaker. Within the last several years enough has been available to permit large-scale tests of its use for both combustion, with which this paper is concerned, and for decarburization. The literature is now enormous and easily available; it is sufficient here to cite three items of historic importance in that all are about 25 years old and in that all are remarkably consistent as far as they go, with what is known today.1,2,3 Significance of High-temperature Heat The principal function of the open-hearth furnace is to supply sufficient high-temperature heat to melt the charge and then to increase its temperature to a range suitable for tapping, ordinarily in the neighborhood of 2900°F. That heat is required to make steel is understood by everybody; however, there seems to be some confusion now and then about the significance of high-temperature heat. Transfer of energy—and heat is energy— is always from a point of higher potential to a point of lower potential and, other things being equal, the greater the potential difference, the greater is the rate of transfer of energy. In the open hearth the flame, of course, is the high-potential source. Perhaps the most satisfactory approach is through the notion of theoretical flame temperature, which is the ratio of heat input to the heat content of the combustion products,* the former is composed mainly of heat of combustion of the fuel plus sensible heat of the fuel plus sensible heat of the combustion air. Thus, flame temperature can be increased by increasing the value of the numerator, by decreasing the value of the denominator, or both. The first method is as old as the open-hearth process and, of course, makes use of preheated air; its possibilities have been by no means exhausted. The second method can be put into effect only by changing the composition of the combustion air, since the combustion of a given quantity of fuel produces a definite quantity of product, and since the denominator includes the non-reactive fraction of the air, principally nitrogen. In other words, the denominator consists of the volume of exit gas multiplied by its heat capacity and that volume consists of carbon dioxide and water vapor as reaction

Jan 1, 1949

By J. S. Marsh

Joseph Priestley prepared oxygen on Aug. I, 1774, and noted with great surprise "that a candle burned in this air with a remarkable brilliant flame." On Aug. 2, 1774, some ironmaker possibly began to wonder if this substance could be used to increase his rate of production; at any rate, there is no disputing the fact that the notion of using oxygen for steelmaking in concentrations greater than that available in great abundance—as air—has been in existence for years. Action was delayed by the lack of enough oxygen at a cost within reach of the steelmaker. Within the last several years enough has been available to permit large-scale tests of its use for both combustion, with which this paper is concerned, and for decarburization. The literature is now enormous and easily available; it is sufficient here to cite three items of historic importance in that all are about 25 years old and in that all are remarkably consistent as far as they go, with what is known today.1,2,3 Significance of High-temperature Heat The principal function of the open-hearth furnace is to supply sufficient high-temperature heat to melt the charge and then to increase its temperature to a range suitable for tapping, ordinarily in the neighborhood of 2900°F. That heat is required to make steel is understood by everybody; however, there seems to be some confusion now and then about the significance of high-temperature heat. Transfer of energy—and heat is energy— is always from a point of higher potential to a point of lower potential and, other things being equal, the greater the potential difference, the greater is the rate of transfer of energy. In the open hearth the flame, of course, is the high-potential source. Perhaps the most satisfactory approach is through the notion of theoretical flame temperature, which is the ratio of heat input to the heat content of the combustion products,* the former is composed mainly of heat of combustion of the fuel plus sensible heat of the fuel plus sensible heat of the combustion air. Thus, flame temperature can be increased by increasing the value of the numerator, by decreasing the value of the denominator, or both. The first method is as old as the open-hearth process and, of course, makes use of preheated air; its possibilities have been by no means exhausted. The second method can be put into effect only by changing the composition of the combustion air, since the combustion of a given quantity of fuel produces a definite quantity of product, and since the denominator includes the non-reactive fraction of the air, principally nitrogen. In other words, the denominator consists of the volume of exit gas multiplied by its heat capacity and that volume consists of carbon dioxide and water vapor as reaction

Jan 1, 1949

By B. M. Larsen, T. E. Brower

In two previous paperslJ dealing with the carbon-oxygcn reaction, and the simultaneous content of each, in liquid steel in the furnace, we have made use of the quantity A[O], defined as the excess oxygen beyond what would be in equilibrium with carbon then present if the pressure of carbon monoxide were I atm.; that is, if [C] X [O] is taken as 0.00222, A[O] = [O] - 0.0022[C] where [O] is the percentage of oxygen by weight as measured by the method described earlier.3 This quantity ?[O] proves to be substantially independent of carbon content over the range 0.08 to I,z per cent, as well as of the basicity of the slag, and is nearly always within the limits o.o15 to o.oz5 per cent when the bath is boiling normally and is in what we have termed a steady state, Apparently it must be at least about 0.014 per cent before boiling occurs; but this threshold value seems to depend upon the ease of bubble formation at the metal-hearth interface, hence upon the condition of the surface of the hearth. In any case, ~(01, on the average, changes considerably following any disturbance of the bath, as by addition of ore. The present paper deals with the average changes in A[O] level following addition of spiegel or crop ends, or after procedures to promote stirring, or by tapping of the steel from the furnace. The observations were made on a large number of commercial heats, and consequently are subject to the influence of various disturbing factors; the results have the compensating merit of showing the consistency obtainable under the limitations imposed by practical operation. Variability of Excess Oxygen in Normal Open-hearth Operation The preceding genera' statements refer to the average value of A[0] particularly in heats that are boiling actively; on individual heats, which are less active, there may be a variable time lag between the disturbance and its effect, which complicates the interpretation of what is going on. For instance, if in the normal refining process boiling is sluggish, the slag usually is gaining Oxygen from the gas phase above it, Or from added Ore$ faster than it is losing oxygen to the metal; Yet when the bath starts to boil vigorously, the opposite condition may prevail for Some time. The consequence of this lag is a variation of A[o], the interpretation of which in a sing1e heat may not be immediately obvious. This condition is illustrated in Figs. I and 2. In Fig. 1, A[0] at the beginning of the test period was near the lower limit of its usual range (dashed lines) in the steady state as a consequence of the lime boil, which had ended about as indicated. The bath is now cold, carbon drop is slow, and the first small orc feed causes only a small rise in A[O]; as the temperature rises and as lime dissolves in the slag, and the bath becomes more active, the

Jan 1, 1948

By B. M. Larsen, T. E. Brower

In two previous paperslJ dealing with the carbon-oxygcn reaction, and the simultaneous content of each, in liquid steel in the furnace, we have made use of the quantity A[O], defined as the excess oxygen beyond what would be in equilibrium with carbon then present if the pressure of carbon monoxide were I atm.; that is, if [C] X [O] is taken as 0.00222, A[O] = [O] - 0.0022[C] where [O] is the percentage of oxygen by weight as measured by the method described earlier.3 This quantity ?[O] proves to be substantially independent of carbon content over the range 0.08 to I,z per cent, as well as of the basicity of the slag, and is nearly always within the limits o.o15 to o.oz5 per cent when the bath is boiling normally and is in what we have termed a steady state, Apparently it must be at least about 0.014 per cent before boiling occurs; but this threshold value seems to depend upon the ease of bubble formation at the metal-hearth interface, hence upon the condition of the surface of the hearth. In any case, ~(01, on the average, changes considerably following any disturbance of the bath, as by addition of ore. The present paper deals with the average changes in A[O] level following addition of spiegel or crop ends, or after procedures to promote stirring, or by tapping of the steel from the furnace. The observations were made on a large number of commercial heats, and consequently are subject to the influence of various disturbing factors; the results have the compensating merit of showing the consistency obtainable under the limitations imposed by practical operation. Variability of Excess Oxygen in Normal Open-hearth Operation The preceding genera' statements refer to the average value of A[0] particularly in heats that are boiling actively; on individual heats, which are less active, there may be a variable time lag between the disturbance and its effect, which complicates the interpretation of what is going on. For instance, if in the normal refining process boiling is sluggish, the slag usually is gaining Oxygen from the gas phase above it, Or from added Ore$ faster than it is losing oxygen to the metal; Yet when the bath starts to boil vigorously, the opposite condition may prevail for Some time. The consequence of this lag is a variation of A[o], the interpretation of which in a sing1e heat may not be immediately obvious. This condition is illustrated in Figs. I and 2. In Fig. 1, A[0] at the beginning of the test period was near the lower limit of its usual range (dashed lines) in the steady state as a consequence of the lime boil, which had ended about as indicated. The bath is now cold, carbon drop is slow, and the first small orc feed causes only a small rise in A[O]; as the temperature rises and as lime dissolves in the slag, and the bath becomes more active, the

Jan 1, 1948

By B. M. Larsen

Jan 1, 1948

By B. M. Larsen

Jan 1, 1948

By J. J. Stokes

Literature on the electrolysis of aluminum from cryolite melts and on the structure of these melts is surveyed critically. Data on density, freezing point, and other properties are reviewed. Theories of the electrolysis are examined in the light of these data. Two theories are presented which account equally well for the observations. ALTHOUGH the production of aluminum is a large-scale industrial process, consuming approximately 3 pet of all of the electric current generated in this country, the mechanism of the Hall Process remains a 70-year-old mystery. It is a remarkable fact that the aluminum industry, which was established largely through research and owes much of its growth to research, still employs as its basic process the one developed by Charles Martin Hall while he was still a young man working in a woodshed. There are, of course, modifications in detail and a tremendous expansion of the size of the units used, but basically the process employs carbon anodes and a carbon-lined iron vessel as a cathode for the electrolysis of alumina dissolved in molten salt. Of the almost infinite variety of salts available, the relatively minor Greenland mineral, cryolite, 3NaF.A1F3, is still the only one which gives good performance in the electrolysis. In the early days, the process was thought of as a simple one. The dissolved alumina dissociated and, during the electrolysis, the aluminum was carried to the cathode where it was discharged and collected in a pool of metal. The oxygen was carried to the carbon anode where it reacted to form primarily carbon dioxide together with some carbon monoxide. About the only virtue that can be claimed for this hypothesis is that of simplicity. The actual mechanisms must be tremendously more complex. This paper is devoted to a rather narrow part of the whole subject, namely, what ions are present in the bath, and by what mechanism the current is carried. The complexity of the literature on this system is demonstrated in Table I, which shows the ions postulated as being present during the process. This system is not easy to investigate. Cryolite melts at about 1000°C, so that all the investigations are at relatively high temperatures. Molten cryolite fumes in air, particularly if excess aluminum fluoride is present. The cryolite changes composition on heating at 1000°C, principally by reaction with moisture in the air, but apparently also by reaction with oxygen. Cryolite reacts with most of the customary materials of construction, and those that are not attacked by cryolite seem inevitably to react with molten aluminum. Graphite is the only common material that is inert to the electrolytic bath. Pure cryolite can be held in platinum. There is hope that some of the newer refractories will stand up in bath, and their development for this purpose would be a major advance in the field. The techniques to be discussed, then, are not the most advanced ones. Rather, they are examples of old standardized methods applied under difficult conditions. The principal methods that shed real light on the mechanism are density, electrical conductivity, freezing-point lowering, transport number, viscosity measurements, and reactivity of bath with carbon dioxide. The first measurement to be discussed is density. This procedure is relatively straightforward. A hollow platinum bob is suspended by fine platinum wire from an analytical balance into the melt being investigated. Density1-4 of the system sodium fluoride-aluminum fluoride is shown in Fig. 1. There is a maximum in the density near the cryolite composition, but actually the maximum is on the NaF-rich side of cryolite, close to the composition NaF - Na3AlF8. Density measurements are also useful in giving a clue to the nature of cryolite-alumina melts.1, 4 On the first addition of alumina, density decreases, Fig. 2, although the density of alumina, either solid or liquid, is higher than that of cryolite. These density results indicate that the systems are not simple. A great deal of effort has been spent in measuring the electrical conductivity of molten cryolite and of cryolite with various additives. Conductivity measurements are quite easy for fused salts that can be measured in a dip cell, but no successful dip cell has been made available yet for cryolite. Recourse is had

Jan 1, 1959

By S. Senderoff, D. E. Couch

A PROCESS for the electrolytic preparation of molybdenum from molten salts has been described previously. This previous work centered on electrowinning and electroplating characteristics of the process. In this paper the properties of the process as a refining medium are discussed. In producing metals from ores, it frequently has proven most economical to obtain a comparatively impure metal by smelting, and to purify the product by electro refining. Direct electrowinning from molten salts is rather difficult when a porous diaphragm is required in the process, as in the case of molybdenum, since satisfactory refractory materials for such a diaphragm in molten salts have not yet been developed. In electrorefining, however, a soluble anode is used. By suitable control of anode and cathode current densities, the lower-valent molybdenum compound used in the electrolyte is not oxidized, and a diaphragm is not needed, thus avoiding difficulties which are encountered with other electrowinning processes. Electrorefining may result from two different processes in the cell: Dissolution of the molybdenum and the less noble impurities with a separation of the more noble impurities as an anode sludge; the electrodeposition of molybdenum and the more noble metals at the cathode, leaving the less noble metals in the electrolyte. The amount of separation which occurs in each process is determined by the potential of the operating electrode and the relative concentration of the various species at (or in) that electrode. Since molybdenum has a comparatively noble position in the EMF series in molten alkali halides,2 most refining would be expected to occur at the cathode. In this work only the cathodic process was studied. The objective was to determine to what extent various metallic ions, if they dissolve from the anode or are otherwise introduced, were concentrated in the electrolyte. The necessary conditions for the separation of each and the limiting concentration which may be tolerated in the electrolyte for effective refining were studied. The deposits were chipped from the cathode and analyzed spectrochemically for metallic impurities. Besides determining the major contaminant, a general qualitative analysis was obtained to be sure that other metallic ions were kept at a very low level during each series of runs and to be sure that the baths were not being contaminated accidentally. Samples of the bath, for analysis, were taken at the temperature of operation so there was no chance of selective crystallization during cooling. In this way the ratio of molybdenum in the bath to the other metal could be determined readily. These data permitted a check on the anode efficiency and any volatilization that took place during the electrolysis. Experimental Apparatus—The apparatus as shown in Fig. 1 was used to determine the effect of the various foreign ions upon electrodeposited molybdenum. The entry chamber A allowed the cathodes to be removed without cooling the bath and without allowing oxygen or moisture to enter." The cathodes were made Prom 0.25-in. tungsten rods and were rotated during the electrolysis. The balloon in the lower arm B was inflated with helium after the cathode was drawn up into the upper section for removal. A new cathode was then placed into the entrance chamber A after which the chamber was flushed

Jan 1, 1959

By K. A. Semlak, F. N. Rhines

NFILTRATION is a term used to designate that i- process by which the pores of a metal powder are filled with a relatively low-melting liquid metal through the action of capillary forces. This is accomplished in practice either by heating the porous body in contact with the solid infiltrant metal above the melting point of the latter, or by preheating the porous body separately and then dipping it into the liquid metal. The liquid, as it sweeps through the porous body, replaces the gases progressively, thus yielding a relatively dense product in a short time.' Liquid-phase sintering, which is related to infiltration, is a process by which a mixture of two or more metal powders is sintered at a temperature between the liquidus and solidus of the ultimate alloy.' According to Cannon" liquid-phase sintering can be divided into three stages which he calls liquid flow, accommodation, and solid sintering. The largest density increase occurs during the first stage, where the lower melting component melts and, driven by capillary forces, flows among the particles of the higher melting component, where, after dissolution of the bonds between the solid particles, it moves them into close packing. This happens so quickly that little has yet been learned of its mechanism. A study of the rate of capillary rise of a liquid metal in a higher melting metal-powder compact may be expected, therefore, to contribute not only to a firmer understanding of the process of infiltration as such but also to a more extensive knowledge of what occurs during the initial stage of liquid-phase sintering. Theoretical Basis In order to derive a mathematical expression of the rate of infiltration, it has been assumed that a powder compact consists of a great many parallel capillary tubes and that the rate of rise of a liquid in an average capillary channel represents the rise in the entire compact. For a straight capillary, assuming that Poiseuille's law obtains for the non-stationary state and that wetting is very rapid, Ligenza and Bernstein' have derived the following differential eauation of motion where h is the height of rise of the column of liquid at time t, R, is the radius and 1 the length of the capillary, t) and 7, the viscosities of liquid and air, respectively, p and p, the density of the liquid and air, respectively, y the liquid-gas surface tension, 0 the liquid-solid contact angle, and g the acceleration due to gravity. The term on the left of Equation [I] represents the rate of change of momentum of the contents of the capillary tube. On the right the terms represent, respectively, the forces due to a) surface tension, b) viscous resistance, c) gravity, and d) end-drag effect. Equation [I] is not amenable to explicit solution, but for a tube of sufficiently small radius, the rate of rise may be slow enough so that the rate of change of momentum and end-drag terms may be neglected. The validity of this simplification was established by Ligenza and Bernstein' by graphical differentiation of observed data obtained from organic liquids rising in glass tubes. The solution of this simplified case is:

Jan 1, 1959

By A. P. Young, F. C. Holden

IN many of the early investigations of the aging of titanium alloys, it was observed that the retained beta phase could be aged to a high hardness without an apparent change in microstructure; moreover, in some alloys it was found that even a very rapid quench produced beta having an anomalously high hardness. What appeared to be hard beta phase was designated as "beta prime," to distinguish it from the softer normal retained beta phase. Frost and co-workers'.' demonstrated that the increased hardness is caused by the formation of a metastable phase, omega, which can be identified from its X-ray diffraction pattern. This reaction also has been studied by Brotzen, et a1.,3 and the structure of the omega phase has been published by Silcock, et al.," and by Austin and Doig.3 The X-ray diffraction analysis by Austin and Doig indicated that omega is low in alloy content, approaching the composition of the stable alpha. The aging process proceeds by the path Thus, overaging results in the formation of alpha and the ultimate disappearance of omega. This is accompanied by a decrease in hardness and a recovery of ductility. The present work represents a study of the aging process in a high-purity Ti-6.4 Mn alloy, in which optical and electron microscopy, X-ray diffraction, and hardness data were correlated. The specimens, 1/8-in. sections cut from 1/4-in.-diam rod, were heated in argon to a temperature just into the beta field (810°C), held for 1 hr, and water-quenched. The aging samples were heated in air for 1 hr at 300 to 500°C. The specimens were ground and mechanically polished, finishing with alternate etching and polishing on a slow-speed wheel to remove disturbed metal from the surface. The surface finally was swab-etched, rinsed with tap water, and air-dried. The etchant used was an aqueous solution containing 1 1/2 pct HF and 3 1/2 pct HNO3. This procedure is identical with that used for preparing metallographic specimens, except that the optimum degree of final etching required for electron microscopy may vary from that used for light microscopy. Metal replicas, backed with silica, were prepared from plastic strip-pings for examination with the electron microscope. Complete details of the procedures used for preparing these replicas are available elsewhere." Fig. 1 shows a comparison between the structures revealed by the electron microscope and those obtained by conventional light microscopy. Hardness and X-ray diffraction data are listed in Table I. No evidence of a second phase is apparent in the optical micrographs for the as-quenched sample, or for the specimens aged at 300 or 400°C; fine alpha particles can be seen in the specimen aged at 500°C. In the electron micrographs, visible equiaxed pocks appear in the as-quenched specimen, and, with increasing density, in the samples aged at 300 and 400°C. After aging at 500°C, the structure becomes markedly coarser and develops an oriented appearance of a second phase, coincident with the appearance of visible alpha particles in the light micrographs. These observations are confirmed by X-ray diffraction and hardness data. In these specimens, aged for 1 hr, the intensity of the omega pattern increases to a maximum at the 400°C aging temperature, corresponding to maximum hardness and the maximum density of pocks in the electron micrographs. At this point the beta phase shows a definite enrichment in Mn. At 500°C, little or no omega re-

Jan 1, 1959

By J. E. Hove

The high temperature physical properties of graphite are reviewed and interpreted in the light of present day knowledge of the mechanisms affecting these properties. The thermal and mechanical behaviors only are discussed and, whenever possible, comparisons are made with other refractory materials. Possible further studies are indicated, including some carbide work. AS long as the term high temperature implied only temperatures up to about 1000°C, the materials problems which arose could usually be handled by fairly conventional metal alloy types, such as the Co-Cr-Ni superalloys, for which there exists a great deal of technology. Perhaps this temperature can still be considered an upper limit for normal applications, but it is certainly true that the number of abnormal applications is increasing rapidly. The advent, in recent years, of ram-jet and rocket missiles and of high power nuclear reactor heat sources has raised a host of questions concerning the basic problem of what material to use in the temperature range up to 2000°C and higher. While there are, of course, many metals, in the second and third transition series, which melt at considerably higher temperatures than this, these metals are, at present, pretty well excluded from practical use by other considerations, such as re-crystallization, chemical activity, or excessive plastic deformation. The behavior of metals, from the standpoint of dislocation theory, is just beginning to be understood and thus there is some hope for the future development of very high temperature metals, but the immediate problems would most logically appear to have solutions involving the nonmetals, such as the refractory ceramics and graphite. For this reason, there is presently a great deal of engineering and experimental research being performed on the latter materials, much of this research being exploratory in the sense of gathering new property data. The situation, so far as graphite is concerned, is somewhat more fortunate than with the other refractory solids in the sense that a great deal is already known about its basic properties. This stems both from the fact that the carbon-carbon bond has been of interest to chemists for a long time (and graphite can be considered as a very large aromatic molecule, if desired) and the fact that its properties, both as a function of temperature and of radiation damage, are of critical importance to nuclear reactor designers. It is still true, of course, that such fundamental questions as why graphite remains solid to such a high temperature and why it has such a high thermal conductivity cannot entirely be answered at the present time. It is, nonetheless, meaningful and instructive to consider such problems in the light of existing knowledge. This is what the present paper will attempt to do. Before going on, it may be appropriate to classify graphite and justify its discussion before readers primarily interested in metals. Graphite is comparatively unique among materials in that there is always a property or group of properties which precludes calling it either a metal, a semiconductor, or a ceramic. It has the high electrical and thermal conductivities of a metal, but the artificial, poly-crystalline types show a negative thermal coefficient of electrical resistivity, generally characteristic of semiconductors. On the other hand, semiconductors, by definition, show an ever increasing resistivity as the temperature is lowered, whereas graphite approaches a finite, and, indeed, a rather low resistivity in the region of 10°K and, furthermore, a good single crystal of graphite has a positive temperature coefficient, as for a metal.' On still another hand, its porosity and brittleness at lower temperatures would put graphite in the ceramic class although, unlike most ceramics, it is readily machinable and has a high resistance to thermal shock. All things considered, it is probably more nearly appropriate to call graphite a metal than anything else. Although certainly outstanding in some ways, graphite has its peculiarities and, especially if the reader is unfamiliar with the data, it is probably valuable to review some representative property variations at high temperature. This review is meant to be mainly illustrative and no attempt has been made to be exhaustive. Review of High Temperature Properties At ordinary pressures, graphite does not melt, but sublimes directly into the gaseous phase at about 3700°C. Although the phase equilibrium diagram is still in some doubt, graphite will melt, at a slightly higher temperature, at pressures in excess of about 100 atm. The chief difficulty of using graphite in an oxidizing atmosphere is that the reaction rate becomes quite high at fairly low temperatures. If a threshold oxidation temperature2 is defined as the temperature at which graphite loses 1 pet of its weight in 24 hr, the value in air is 450°, the value in steam is 700°, and the value in carbon dioxide is 900°C. Efforts are presently being made to raise this threshold temperature either by impregnation with a retardant of some type (sodium tung-state and phosphoric acid, for example) or by a suitable metal or oxide coating. It is probably fair to say that, to date, these attempts have not shown any outstanding success in all respects. Most commercial graphites are fabricated by impregnating a carbon flour (say of coke or lampblack particles) with some type of hydrocarbon

Jan 1, 1959

By M. Cohen

THERE seems to be a natural urge for the human mind to wonder about the beginning of things. When an explosion occurs, we immediately inquire "what set it off?" If a person contracts a disease, we are likely to ask "how did it happen?" On a much loftier plane, when we reflect about the universe, there is the inevitable question "what, or rather Who, started it all?" In principle, these varied queries have at least one attribute in common: they are attempts, whether earnest or casual, to search out the origin of events in nature. Accordingly, they involve not only scientific, but philosophical and even re1igious matters of deep concern. It is evident that, in a broad sense, the phenomenon of nucleation is enmeshed in this enticing concept of "the beginning", but we must extract the scientific component of the problem for our present purposes. Nucleation, as in the case of phase transformations, is best regarded from the standpoint of kinetics. There are then two main steps, which may come in either order: (a) Experimentally, one attempts to measure the frequency with which transformation centers spring into existence; and (b) theoretically, one seeks a model that will account for this observed birth-rate. Temperature, composition, strain, and prior history are some of the important variables that must be considered both experimentally and theoretically when dealing with nucleation in the solid state. Such interaction of theory and observation is an integral part of science in general, but it is particularly significant in the nucleation problem because the very beginning of a phenomenon invariably occurs on too localized a scale in both space and time to be detected with available techniques. Thus, there is little hope of making direct measurements on the nucleus during its birth or to ascertain the exact nature of the process; one can only infer the details of nucleation from the model whose predicted behavior seems to harmonize best with the measured rate of nucleation and its dependence on known parameters. In the field of solid-state nucleation, both theory and experiment call for much ingenuity, but somehow the conceptual schemes have spawned faster than they can be tested by observation. This is not to say that there are many basically new ideas in modern nucleation theory; rather, as we shall see, it is because the required fine-scale measurements are of immense technical difficulty. Quite likely, some of the nucleation models existent today have survived only because of the latitude permitted by meager data. There is great need to bring theory and experiment into better balance in this important area of physical metallurgy. Although one can

Jan 1, 1959

By W. R. Hibbard

Evidence is presented which confirms previous findings that models of solution strengthening depending solely on lattice parameter changes are incomplete. Direct evidence for the Suzuki interaction of chemical effects was obtained from alloys with constant lattice parameter, for which solution strengthening was nearly independent of temperature from 300° to 20°K. SOLUTION hardening principles have been reviewed recently by Parker and Hazlett.' In addition, the interactions of dislocations and solute atoms were reviewed by Cottrell.' It is not proposed to review this information again here, but to refer only to those details which are pertinent to the developments which have occurred since these papers were published. Literature Survey To set the stage, it has been known for some time that solid solutions are stronger than the pure parent metals, and that the strength increases with increasing amounts of alloying element. One of the most extensive earlier investigations was that of Lacy and Gensamer,3 who reported that the solid solution strengthening of iron is a power function of the composition, that the relative strengthening is a function of the solubility limit, and that the rate of strain hardening is a function of the amount of solution hardening. Thus, the slope of the plastic portion of the stress-strain curve increases with increasing yield strength. An approach to the development of principles resulted from work by Gulyeav,4 who reported that the strengthening effect of solid solution alloys is periodic in character when plotted as a function of the atomic number of the solute. A more quantitative approach was to relate the amount of solution hardening to the lattice parameter of a solid solution. This was done quite early by Norbury,5 and later by Brick, Martin, and Angier,6 Frye and Hume-Rothery,7 and French and Hibbard. This relationship is only approximate in character. The next development was to relate the amount of solution hardening to lattice stress, as proposed by Frye and co-workers.9 They calculated what they called ionic overlap from the lattice parameter changes represented as a change in volume, using Bridgman's hydrostatic compression equation. The resulting correlation was somewhat improved as compared to the simple correlation with lattice parameter. The next step was carried out by Dorn, Pietrow-sky, and Tietz,10 who suggested an effect of valence differences in addition to the effect of lattice parameter. Their proposed parameter shows flow stress as a single valued function of the concentration and a relationship involving these two atomic variables. In addition, they contributed new data on the temperature dependence of solution hardening which indicated that it was a strong function of temperature for aluminum alloys. Allen, Schofield, and Tate11 carried the valency effect even further by reporting data indicating that with a given solvent element the stress-strain curve is a function of electron density, entirely neglecting the size factor, which varied by a factor of two. This evidence led to a quantum mechanical calculation by Cottrell, Hunter, and Nabarro,12 based on the formation of a dipole around an edge disloca-

Jan 1, 1959

By S. B. Twiss, P. R. Basford

The relationship between areas of iron powders, briquettes, and sintered compacts and tensile strength has been determined. It has been found necessary to distinguish between two types of areas which exist in such powders—a macroscopic area due to the shape and size of particles and a microscopic area due to pores and cracks in the surface. Only the macroscopic area, specifically the area due to particle irregularity, contributes to tensile strength; the microscopic area cannot be forced into contact with adjacent particles ond does not participate in bond formation. Of the two methods of area measurement employed in this study—permeability to liquids and gas-phase adsorption—only the former is useful for determining the roughness factor. A measure of this roughness, (area measured by permeability to liquid) -f- (orea of powder with the same size distribution, but made up of spherical particles), is shown to be uniquely related to tensile strength at constant density. The other important factors in determining the strength of sintered-metal compacts are the density and presence or absence of surface oxide. THE manufacture of metal parts by the process of powder metallurgy is an important and rapidly growing industry. Progress has been rapid and continuous, largely because of the excellent research in the field. Nevertheless, the process remains inherently complex and empirical. Several investigators have shown that the properties of the final compact must depend on many variables; namely, shape,' size distribution,"-" nd compositiona of the original particles, the pressure necessary to compress to the desired density, the density of the compact, the time and temperature of sintering, and so on. Through the application of some new techniques to the study of powders and compacts, and the correlation of a number of observations of known variables, the present paper attempts to clarify and simplify the underlying theory of the powder-metallurgy process. Any theory must start with the obvious fact that the strength of a compact, its most useful property, is due to the formation of welded areas between adjacent particles, and attempt to deduce what effect other variables will have on (a) the area and (b) the intrinsic strength of these welds. Such welded areas can be demonstrated by electron micrographs, as in Fig. 1. It was believed that (a) could be most directly studied by means of surface-area measurements on the original powders and on the compacts made from them; the area decrease should be related closely to the welded area. A representative group of eleven iron powders from commercial sources, and made by a variety of processes, was chosen for this study. In addition to the surface-area measurements, correlations were established between the tensile strength and all the known variables which might possibly affect it. The results are quite unexpected. They indicate that the situation is much simpler than the authors had anticipated. The strength of the sintered compact is determined, indeed, by no more than three variables; viz., the irregularity of the original particles, the density of the compact and, in exceptional cases, the amount of surface oxide. All three had been recognized previously as having some effect on the tensile strength, but in the absence of the special techniques and interpretations to be developed here, there could be no suspicion that they were the only relevant variables. Materials and Methods Material—All the iron powders were commercial products obtained from the Amplex Division of Chrysler Corporation. Powders A, B, C, and J were very pure electrolytic iron. Powder D was made from oxidized scrap reduced by heating with an excess of cast iron. Powder E was a high-grade iron ore reduced by hydrogen. Powder I was prepared by atomizing molten cast iron in an air blast, quenching the droplets in cold water, and crushing the coarse powder. Mill scale (oxide) is added and the mixture heated to reduce the high carbon content. The other powders were all prepared by reduction of iron oxides, such as mill scale. They

Jan 1, 1959

By J. Gurland

A method of measuring the degree of contact between grains of 1 phase in multiphase alloys is derived. It is shown that the number and the areas of contacts between grains can be determined by metallographic analysis. Relations between the area of contact, the dihedral angle, and the mean free path of spherical grains are developed. It is often desirable to interpret the mechanical behavior of alloys in terms of structural features observed under the microscope. A quantitative description of microstructures of multi-phase alloys requires consideration of geometrical and topological factors such as the amount, size, and shape of grains and the continuity of the constituent phases. The number and dimensions of grains may be metallographically determined by applying the methods described by Smith and Guttman1 and by Fullman2 in their papers on the measurement of particle size and internal boundaries in opaque bodies. The present work intends to show that the same methods can be used to evaluate some of the parameters of phase continuity; namely, contiguity, number of contacts, and areas of contact of grains of 1 phase. Application of these concepts may be found in such problems as the calculation of electrical conductivity of 2-phase alloys, or the evaluation of the length of fracture path through grains of a brittle phase more or less dispersed in a matrix. I. DEFINITION The following definitions and derivations are based on 2-phase alloys, of which the units of the microstructure are grains. Contiguity is defined as the average fraction of surface area shared by 1 grain of a phase with all neighboring grains of the same phase. In a 2-phase alloy of constituents A and B, the contiguity of phase B is given by: where Cb is the contiguity of phase B aBß is the average area of contact between a grain of phase B and adjacent grains of phase B, aAB is the average area of contact between a grain of phase B and adjacent grains of phase A. The contiguity is a measure of the average degree of contact between grains of phase B in the 2-phase mixture. The ratio varies from 0 to 1 as the distribution of phase B changes from a completely dispersed to a fully agglomerated structure. Smith and Guttman1 have shown that the relative areas of 2 kinds of interfaces in a volume are directly proportional to the number of interfaces of each type intersected by a random line. The lines can be drawn on metallographically prepared surfaces as long as these are random sections through the sample. The contiguity ratio of phase B then becomes: Cb = 2NBB/NAB + 2N BB [2] where: NBB is the number of interfaces between grains of phase B intersected per unit length of random line. Nab is the number of interfaces between grains of phase B and grains of phase A intersected per unit length of random line. The factor 2 arises in Eq. [2] because each interface B-B is associated with 2 grains of B-phase, whereas the interface A-B involves only 1 grain of B-phase. The calculations of contiguity do not require any assumption as to volume fraction, particle size, or particle shape. Fullman2 gives an expression for the determination of the mean free path between particles. The mean free path PB between grains of phase A, i.e., through grains or agglomerates of phase B, may be found from the relationship: PB =2fB/NAB [3] where fB is the volume fraction of phase B. The calculation of the mean free path is independent of grain shape and grain size distribution. The calculation of the number of contacts and of contact area requires assumptions of grain shape, grain size, and contact shape. In the case of uni-

Jan 1, 1959

By P. G. Shewmon

THIS paper is concerned with the redistribution of a second phase when a 2-phase solid alloy is annealed in a temperature gradient. An examplewhich is discussed in detail is the redistribution of carbides in a low-carbon steel annealed in a temperature gradient below the A, temperature. As a result of such a treatment, the carbides at the cooler side disappear, while those at the hotter side grow in size. In a 2-phase zirconium-hydrogen alloy, the amount of hydride will tend to be increased at the cold end and decreased at the hot end. That such an effect can occur in solid metals is not generally appreciated. However, with the trend toward higher operating temperatures in several fields, this phenomenon will become more important in the future. Actually, such effects should be rather common in the core of nuclear power reactors where the op- erating temperatures are high and temperature gradients are an order of magnitude higher than in previous high-temperature applications of alloys. The purpose of this paper is to point out the existence of such an effect, to discuss the manner in which one predicts whether a redistribution of the second phase will occur in a given system, whether the second phase will go to the hot or cold face, and how rapidly the redistribution will occur. One aspect of this problem has been discussed by pfam and Wernick2who considered the migration of a small liquid region through a solid. In their case, the diffusivity in the liquid is much larger than in the solid and they have interpreted their results successfully by using Fick's first law as the flux equation in the liquid, and taking the concentration gradient from the phase diagram. For all the cases they have studied, this type of analysis predicts that the liquid region migrates from the cold end to the hot end, and this is what was found ex-

Jan 1, 1959

By J. R. Mihalisin

THE application of electron-microscopic techniques to the study of high-temperature alloys has met with considerable success.' This is due, in part, to the submicroscopic nature of the phases present in this type of alloy making high resolution methods necessary. Among the newer electron-microscopic techniques, the extraction replica technique' has proven to be of great value. This method allows phases to be examined in situ in the microstructure and permits their identification by means of electron diffracticn. This technique eliminates the main disadvantage in using electrolytic extraction of residues as a means of phase identification; that is, the uncertainty of relating constituents identified in the residue to what is observed microscopically. In the course of work relating microstructure to physical properties in high-temperature alloys, it was found that the formation of grain-boundary carbide phases and intragranular precipitation of an age-hardening phase greatly influence strength and ductility at high temperature. In particular, it was found that the formation of a unique type of precipitate in Nimonic "80" (a nickel-chromium-aluminum-titanium type superalloy), termed cellular or discontinuous precipitate, at grain boundaries generally results in poor ductility at high temperature. However, this mode of precipitation can be avoided by proper heat-treatment. It was assumed that this precipitate was either a carbide or age-hardening type of phase. Using various etching procedures with the extraction-replica technique, it was found to be possible to extract the carbide and age-hardening phases preferentially in this type of alloy. Previous work has shown that the age-hardening phase can be extracted from superalloys merely by

Jan 1, 1959

By A. H. Daane, J. F. Haefling

BECAUSE of the wide use of platinum in industry and research, the physical properties of this metal, including its vapor pressure, have been studied in some detail.' The other members of the palladium-platinum group of metals have not been studied as extensively, and only qualitative information is available on the vapor pressure of palladium. Data given in literature by Vines1 and Brewer2 on the vapor pressure of palladium are based on the qualitative observations by Holborn and Austin3 and Crookes4 on the rate of volatilization of palladium filaments in various atmospheres. The general observation that palladium is the most volatile metal of the palladium-platinum metal group was corroborated by Bell, Love and Normand who studied the separation of isotopes of the platinum-group metals by a mass spectrometric method. We have measured the vapor pressure of palladium using the Knudsen effusion technique. In this work, the effusion vessel was suspended from a quartz-fiber microbalance into an induction furnace, and the amount of vapor effusing from the vessel was determined by direct weighing of the vessel during the measured heating times. This technique has been described in a report from this laboratory." The vapor pressure of palladium was calculated from these measurements using the equation P...... = 17.14 -W- (77m)'u [1] a t where ? is the weight loss in grams of vapor of molecular weight m from a vessel having an orifice of a sq cm, during t sec. The derivation of this equation from kinetic theory has been described in a previous report from this laboratory: as well as in texts discussing the kinetic theory of gases.%" Test Vessel—In the search for a suitable material for the effusion vessel, it was found that tantalum reacted with palladium metal to form intermetallic compounds, and graphite was readily dissolved in palladium metal to form what appeared to be a liquid phase at a temperature several hundred degrees below the melting point of palladium. On cooling, the graphite gathered into spherical balls in the palladium, with crystals extending radially from the center of these balls, some of which were a millimeter or more in diameter. However, palladium metal vapor did not attack the graphite vessel, so graphite could be used where it did not contact condensed palladium. Tungsten did not appear to react with palladium vapor or molten palladium at its melting point, but it was not convenient to use a tungsten-vapor pressure vessel because of the difficulty of fabricating this metal. Since the vapor-pressure range to be studied (10 2 to 10 ' mm) did not appear to extend above the melting point of palladium, a combination of tungsten and graphite served to provide a very satisfactory effusion vessel. The lower half of the graphite vessel was lined with some strips of tungsten foil and the sample of palladium consisting of a coil of sheet was placed in the tungsten liner; in this way, no contamination of the palladium by carbon occurred. Procedure After outgassing at 1500C, the tungsten-lined graphite effusion vessel was loaded with 300 mg of

Jan 1, 1959

By A. H. Cottrel

SINCE metallurgy exists to provide strong, tough, engineering materials it must inevitably be perpetually concerned with the problem of brittle-ness. The steel-making industry was created because chemically unrefined iron is brittle. The steel-working and alloy-steel industries exist partly because hardness and ductility are mutually exclusive qualities in structurally unrefined steel. Yet, in spite of these immense industries, the brittle failure of steel ships hulls, bridges, pressure vessels, and pipelines, is still a contemporary problem. In fact, brittleness is a normal property of most solids, including metals and alloys, at low temperatures. Only face-centred cubic metals are commonly ductile at the lowest temperatures, and even here exceptions are known.'," Expensive are-melting processes. under inert gases or vacuum, have been developed to overcome brittleness in titanium, zirconium, columbium. and molybdenum. Polycrystal-line zinc, magnesium, and uranium have little ductility at room temperature. Beryllium, chromium, and tungsten have even less; similarly for antimony, bismuth, germanium, silicon, intermetallic compounds, and metallic carbides, nitrides, silicides, and borides. Oxides and other ceramics would be ideal creep-resistant materials but for their extreme brittleness when cold, and great efforts have been made to overcome this problem by mixing ceramics with metals. In fact, the traditional use of the name metallurgy for what is really the science of engineering materials is a recognition that most non-metallic solids have so far been precluded, by their extreme brittleness, from use as major structural materials in mechanical engineering. Most of what we know about brittleness in metals has come from studies of structural steel, and it is this material that shall mainly be considered. The things we have learned from it have a wider application, at least to other body-centred cubic transition metals, although the extent to which similar ideas can be applied to hexagonal metals and other materials is not yet clear. Fracture and Plastic Deformation The theoretical breaking strength of an ideal solid, about E/10 where E is Young's modulus, has been approached reasonably closely in experiments on fibres. But large specimens break at much lower stresses. For example, brittle cracks in large structural steel assemblies have been observed to spread catastrophically at speeds of 6000 ft per sec under stresses of about 10,000 psi, i.e. about E/3000. There are two sources of weakness; stress concentrations, and chemical agents in grain boundaries or on crack faces that lower the surface energy y of the material. Although we shall deal mainly only with the first of these, much of the discussion is also applicable to the second when different values of y are used. Griffith's well-known formula p = [ 2Ey/p(1 - v2)c]1/2 [1] gives the smallest tensile stress p able to propagate an atomically sharp surface crack of length c or interior crack of length 2c through a thick plate (compared with c) of elastically isotropic material of

Jan 1, 1959