Search Documents

By L. S. Darken

The manufacture and treatment of metals comprises operations whose effectiveness depends in large measure upon diffusion phenomena. The significance of such phenomena has, for a few simple cases, long been understood, but in more complex cases it has not been adequately recognized. This paper presents a mathematical analysis of some of the cases in which diffusion leads to the formation of a new phase within the metal, and this analysis provides a basis for interpreting quantitatively many apparently diverse experimental observations recorded in the literature, some of which are outlined in the following pages. OBservt1ons Recorded in Literature An outstanding instance is the general type of phenomenon in which diffusion of an element into an alloy is followed by a phase change, which most frequently is a reaction (precipitation) of the diffusing element with one or more components of the alloy to form a compound. Thus in certain alloys heated in contact with oxygen not only the usual oxide scale appears but, in addition, beneath the surface, a well-defined zone in which free oxide particles are dispersed with notabIe uniformity. This zone, commonly referred to as the subscale or zone of internal oxidation, is remarkable in that the visible boundary between it and the unaltered metal is usually sharp. Such a subscale, which appears to have been recognized first in oxidized copper alloys, has been described and studied by C. S. Smith,l by Rhines,2 and by Rhines, Johnson and Anderson;3 but it appears also in other alloys when they are exposed to conditions that bring about the precipitation of some compound, not necessarily an oxide. Another clear case of this same kind is the diffusion of carbon into a steel containing an element such as molybdenum, chromium, vanadium or tungsten, whose carbide is less soluble in iron than iron carbide, the result being that some part of that element is precipitated as a dispersed carbide. A similar phenomenon may occur on the welding of two alloys each of which contains one of a pair of elements that react to form a compound or a new phase of any sort. But in addition to the precipitation of a compound formed from the diffusing element, other effects may result merely from the presence of the diffusing element. For instance, if homogeneous austenite, nearly saturated with respect to carbon, is heated in contact with silicon in the absence of oxygen, one would expect to find, behind the several iron-silicon phases analogous to a scale, a zone of austenite containing dispersed carbide, because of the fact that silicon decreases the solubility of carbon in iron. As still another example, oxidation of an alloy may result in diffusion toward the surface of one or more of the constituents; for instance, during oxidation of a steel containing an element such as "copper, which is less prone to oxidation than iron, there is a notable enrichment of copper just beneath the scale-metal interface. AS an example of a yet more

Jan 1, 1942

By L. S. Darken

The manufacture and treatment of metals comprises operations whose effectiveness depends in large measure upon diffusion phenomena. The significance of such phenomena has, for a few simple cases, long been understood, but in more complex cases it has not been adequately recognized. This paper presents a mathematical analysis of some of the cases in which diffusion leads to the formation of a new phase within the metal, and this analysis provides a basis for interpreting quantitatively many apparently diverse experimental observations recorded in the literature, some of which are outlined in the following pages. OBservt1ons Recorded in Literature An outstanding instance is the general type of phenomenon in which diffusion of an element into an alloy is followed by a phase change, which most frequently is a reaction (precipitation) of the diffusing element with one or more components of the alloy to form a compound. Thus in certain alloys heated in contact with oxygen not only the usual oxide scale appears but, in addition, beneath the surface, a well-defined zone in which free oxide particles are dispersed with notabIe uniformity. This zone, commonly referred to as the subscale or zone of internal oxidation, is remarkable in that the visible boundary between it and the unaltered metal is usually sharp. Such a subscale, which appears to have been recognized first in oxidized copper alloys, has been described and studied by C. S. Smith,l by Rhines,2 and by Rhines, Johnson and Anderson;3 but it appears also in other alloys when they are exposed to conditions that bring about the precipitation of some compound, not necessarily an oxide. Another clear case of this same kind is the diffusion of carbon into a steel containing an element such as molybdenum, chromium, vanadium or tungsten, whose carbide is less soluble in iron than iron carbide, the result being that some part of that element is precipitated as a dispersed carbide. A similar phenomenon may occur on the welding of two alloys each of which contains one of a pair of elements that react to form a compound or a new phase of any sort. But in addition to the precipitation of a compound formed from the diffusing element, other effects may result merely from the presence of the diffusing element. For instance, if homogeneous austenite, nearly saturated with respect to carbon, is heated in contact with silicon in the absence of oxygen, one would expect to find, behind the several iron-silicon phases analogous to a scale, a zone of austenite containing dispersed carbide, because of the fact that silicon decreases the solubility of carbon in iron. As still another example, oxidation of an alloy may result in diffusion toward the surface of one or more of the constituents; for instance, during oxidation of a steel containing an element such as "copper, which is less prone to oxidation than iron, there is a notable enrichment of copper just beneath the scale-metal interface. AS an example of a yet more

Jan 1, 1942

By M. A. Grossman

The harden ability of most steels can be predicted within 10 to I5 per cent provided the complete chemical composition is known, including "incidental" elements; and provided the as-quenched grain size is known; and provided, finally, that the composition and heating temperatures for hardening are such as to result in austenite free from carbide particles. In the method proposed herein, the steel is considered as having a base hardenability due to its carbon content alone (the hardenability of a "pure steel" of the given carbon content, without any other elements), and this base hardenability is multiplied by a multiplying factor for each chemical element present. After multiplying all these together, the final product is the hardenability. Grain size may be taken into account either in the base hardenability or after the multiplication. Hardenability is stated in terms of "ideal critical diameter", namely, the diameter of bar, in inches, that will just harden all the way through (absence of un-hardened core) in an "ideal" quench, and the calculation may also be related to the Jominy test The data bring out certain features rather clearly. For example: i. It is quite useless to attempt to predict hardenability unless all elements, including "incidentals,,) are known; thus an "incidental,, chromium content of 0.20 per cent increases the hardenability by about 50 per cent. 2. The relative effectiveness of different alloys is sometimes unexpected; thus molybdenum when calculated in this way appears to be of the same order of effectiveness as manganese, rather than much more powerful, as would appear to be common experience; observe that an increase from no manaanese to 0.20 per cent Mn provides a multiplying factor of 1.67, and an increase from no molybdenum to 0.20 per cent Mo provides a factor of 1.63, an increase of over 60 per cent in each case. However, if to a steel containing 0.50 per cent Mn there is added "20 points of manganese," the factor is raised from 2.65 (for 0.50 per cent Mn) to 3.35 (for 0.70 per cent Mn), or an increase of only 26 per cent. Thus the fist small addition of an element has a much more powerful percentage effect than an equal further addition when some is already present, and in most cases the effect of molybdenum is considered in relation to a steel in which molybdenum is absent. 3. If two elements are equally effective, a greater hardenability will be obtained by using, for example, 0.5 per cent of each than by using I.o per cent of either of them alone. 4. A knowledge of the as-quenched grain size is essential for precise work, since a difference of only one grain size number (say No. 7 instead of No. 6) makes a difference of almost 10 per cent in hardenability (in these units); this, however, does not apply to certain It should be emphasized that in chromium steels (over 0.30 per cent Cr) and chrome-molybdenum and chrome-vanadium steels, un-dissolved carbides are likely to be present in the steel as quenched, and that in such cases the charts can indicate only a maximum Passible hardenability, whereas the extent of hardening actually obtained may be much less. Thus tests on a number of chrome-molybdenum steels have indicated a degree of hardening amounting to only 50 to 65 per cent of the maximum possible, and in chromium steels from full hardening (100 per cent) down to as low as 70 per cent. On the other hand, when the amount of such elements is small (Cr under 0.30 per cent, Mo up to 0.25 per cent in the

Jan 1, 1942

By M. A. Grossman

The harden ability of most steels can be predicted within 10 to I5 per cent provided the complete chemical composition is known, including "incidental" elements; and provided the as-quenched grain size is known; and provided, finally, that the composition and heating temperatures for hardening are such as to result in austenite free from carbide particles. In the method proposed herein, the steel is considered as having a base hardenability due to its carbon content alone (the hardenability of a "pure steel" of the given carbon content, without any other elements), and this base hardenability is multiplied by a multiplying factor for each chemical element present. After multiplying all these together, the final product is the hardenability. Grain size may be taken into account either in the base hardenability or after the multiplication. Hardenability is stated in terms of "ideal critical diameter", namely, the diameter of bar, in inches, that will just harden all the way through (absence of un-hardened core) in an "ideal" quench, and the calculation may also be related to the Jominy test The data bring out certain features rather clearly. For example: i. It is quite useless to attempt to predict hardenability unless all elements, including "incidentals,,) are known; thus an "incidental,, chromium content of 0.20 per cent increases the hardenability by about 50 per cent. 2. The relative effectiveness of different alloys is sometimes unexpected; thus molybdenum when calculated in this way appears to be of the same order of effectiveness as manganese, rather than much more powerful, as would appear to be common experience; observe that an increase from no manaanese to 0.20 per cent Mn provides a multiplying factor of 1.67, and an increase from no molybdenum to 0.20 per cent Mo provides a factor of 1.63, an increase of over 60 per cent in each case. However, if to a steel containing 0.50 per cent Mn there is added "20 points of manganese," the factor is raised from 2.65 (for 0.50 per cent Mn) to 3.35 (for 0.70 per cent Mn), or an increase of only 26 per cent. Thus the fist small addition of an element has a much more powerful percentage effect than an equal further addition when some is already present, and in most cases the effect of molybdenum is considered in relation to a steel in which molybdenum is absent. 3. If two elements are equally effective, a greater hardenability will be obtained by using, for example, 0.5 per cent of each than by using I.o per cent of either of them alone. 4. A knowledge of the as-quenched grain size is essential for precise work, since a difference of only one grain size number (say No. 7 instead of No. 6) makes a difference of almost 10 per cent in hardenability (in these units); this, however, does not apply to certain It should be emphasized that in chromium steels (over 0.30 per cent Cr) and chrome-molybdenum and chrome-vanadium steels, un-dissolved carbides are likely to be present in the steel as quenched, and that in such cases the charts can indicate only a maximum Passible hardenability, whereas the extent of hardening actually obtained may be much less. Thus tests on a number of chrome-molybdenum steels have indicated a degree of hardening amounting to only 50 to 65 per cent of the maximum possible, and in chromium steels from full hardening (100 per cent) down to as low as 70 per cent. On the other hand, when the amount of such elements is small (Cr under 0.30 per cent, Mo up to 0.25 per cent in the

Jan 1, 1942

By G. V. Smith, R. F. Mehl

The decomposition of austenite in steels, because of its immerse practical importance, has been subjected to extensive study in recent years from the point of view of the mechanism of the process.1"3 Although much information has been acquired, a number of questions, principally of theoretical interest, remain. One is that of the atomic-crystallographic mechanism by which each of the various reaction products forms. The isothermal reaction products of the eutectoid decomposition in plain carbon steels (to which this study is restricted) have conveniently been classified in three general groupings—pearlite, bainite, and martensite. Of these several products, pearlite and martensite have been studied extensively, but bainite has received only scant attention. -As a result, the mechanism of the formation of pearlite and that of martensite are now fairly well understood while a great deal of uncertainty exists regarding bainite. Pearlite Pearlite forms from austenite by the precipitation of alternate platelets of cementite and ferrite, which grow into the austenite matrix.2--5 The rate of reaction may be described in terms of a rate of nucleation and a rate of growth.2.3 Mehl and Smith6 have attempted to determine the lattice relationships existing between the ferrite in pearlite and the' parent austenite. Using a pole-figure method of analysis, it was found that a limited number of ferrite orientations resulted from the decomposition of a single grain of austenite; it was not possible, however, to decide between two assumed ideal relationships, both of which appeared to account for the experimentally determined pole figure. It was definitely shown, however, that the relationship obtaining was quite different from that determined for the proeutectoid separation of ferrite.7 No attempt was made to determine the habit plane (matrix plane to which the precipitate lies parallel) of the pearlite lamellae. Belaiew8 earlier had stated that the number of directions assumed by the pearlite lamellae in a single grain of austenite was 12 (no details were given) and concluded wrong!:!, as Mehl and Smith6 have pointed out, that the lamellae delineate the {2I0} planes of ferrite.* Martensite It is known that martensite is a transiti0nal constituent of body-centered tetragonal lattice,' and that it forms suddenly

Jan 1, 1942

By R. F. Mehl, G. V. Smith

The decomposition of austenite in steels, because of its immerse practical importance, has been subjected to extensive study in recent years from the point of view of the mechanism of the process.1"3 Although much information has been acquired, a number of questions, principally of theoretical interest, remain. One is that of the atomic-crystallographic mechanism by which each of the various reaction products forms. The isothermal reaction products of the eutectoid decomposition in plain carbon steels (to which this study is restricted) have conveniently been classified in three general groupings—pearlite, bainite, and martensite. Of these several products, pearlite and martensite have been studied extensively, but bainite has received only scant attention. -As a result, the mechanism of the formation of pearlite and that of martensite are now fairly well understood while a great deal of uncertainty exists regarding bainite. Pearlite Pearlite forms from austenite by the precipitation of alternate platelets of cementite and ferrite, which grow into the austenite matrix.2--5 The rate of reaction may be described in terms of a rate of nucleation and a rate of growth.2.3 Mehl and Smith6 have attempted to determine the lattice relationships existing between the ferrite in pearlite and the' parent austenite. Using a pole-figure method of analysis, it was found that a limited number of ferrite orientations resulted from the decomposition of a single grain of austenite; it was not possible, however, to decide between two assumed ideal relationships, both of which appeared to account for the experimentally determined pole figure. It was definitely shown, however, that the relationship obtaining was quite different from that determined for the proeutectoid separation of ferrite.7 No attempt was made to determine the habit plane (matrix plane to which the precipitate lies parallel) of the pearlite lamellae. Belaiew8 earlier had stated that the number of directions assumed by the pearlite lamellae in a single grain of austenite was 12 (no details were given) and concluded wrong!:!, as Mehl and Smith6 have pointed out, that the lamellae delineate the {2I0} planes of ferrite.* Martensite It is known that martensite is a transiti0nal constituent of body-centered tetragonal lattice,' and that it forms suddenly

Jan 1, 1942

By Robert A. Colten, Frederick C. Hull, Robert F. Mehl

It is known that pearlite forms from austenite by a process of nucleation and growth, and that the rate of formation of pearlite may be described by a rate of nucleation and a rate of growth.l,2 The manner in which these two quantities operate to produce the observed isothermal reaction curve has been analyzed in a paper published in I939.3 Two types of process were described: (I) that in which nucleation occurs at the grain boundaries of the parent phase, with the nodules growing toward the centers of the grains, initially as hemispheres and then after impingement as roughly radial-columnar structures; and (2) that in which nucleation occurs generally, with the nodules growing to full spheres until impingement with other growing nodules restricts growth. Both cases are met in the formation of pearlite from austenite (Fig. I);2,4 the former, occurring near the knee of the S-curve, has been named "grain-boundary transformation," and the latter, occurring at temperatures near Ael, has been named "group-nodule transformation." In developing our knowledge of the factors that determine the rate of formation of pearlite from austenite, it is important1,2 to obtain data on the rate of nucleation and the rate of growth of pearlite. Such data, when obtained near the knee of the S-curve, will furnish information of use in understanding the factors that affect hardenabil-ity; when obtained at higher temperatures, such data will be useful in understanding the behavior of steels that during heat-treatment react to pearlite.2 It is the purpose of this paper to present data on the rate of nucleation and the rate of growth of pearlite in dependence upon composition, grain size, temperature, and the degree of homogeneity of the parent austenite. The studies are restricted chiefly to plain carbon commercial steels." Previous Data on Rate of nucleation and rate of Growth Data on the rate of nucleation and the rate of growth of pearlite have been published by Scheil and Lange-Weise,5 by Mirkin and Blanter,6,7 by Meh1,l and by Dorn, De Garmo and Flanigan.2, It is difficult to interpret the data, for in most cases too little information is given on austenitizing temperatures and times, the state of homogeneity of the austenite (presence or absence of carbide, etc.), austenite grain size, and mode of formation of pearlite at the temperatures given. Except for the approximations of Mehl and the work of Dorn, austenite grain size was disregarded, though it is well known

Jan 1, 1942

By Robert A. Colten, Frederick C. Hull, Robert F. Mehl

It is known that pearlite forms from austenite by a process of nucleation and growth, and that the rate of formation of pearlite may be described by a rate of nucleation and a rate of growth.l,2 The manner in which these two quantities operate to produce the observed isothermal reaction curve has been analyzed in a paper published in I939.3 Two types of process were described: (I) that in which nucleation occurs at the grain boundaries of the parent phase, with the nodules growing toward the centers of the grains, initially as hemispheres and then after impingement as roughly radial-columnar structures; and (2) that in which nucleation occurs generally, with the nodules growing to full spheres until impingement with other growing nodules restricts growth. Both cases are met in the formation of pearlite from austenite (Fig. I);2,4 the former, occurring near the knee of the S-curve, has been named "grain-boundary transformation," and the latter, occurring at temperatures near Ael, has been named "group-nodule transformation." In developing our knowledge of the factors that determine the rate of formation of pearlite from austenite, it is important1,2 to obtain data on the rate of nucleation and the rate of growth of pearlite. Such data, when obtained near the knee of the S-curve, will furnish information of use in understanding the factors that affect hardenabil-ity; when obtained at higher temperatures, such data will be useful in understanding the behavior of steels that during heat-treatment react to pearlite.2 It is the purpose of this paper to present data on the rate of nucleation and the rate of growth of pearlite in dependence upon composition, grain size, temperature, and the degree of homogeneity of the parent austenite. The studies are restricted chiefly to plain carbon commercial steels." Previous Data on Rate of nucleation and rate of Growth Data on the rate of nucleation and the rate of growth of pearlite have been published by Scheil and Lange-Weise,5 by Mirkin and Blanter,6,7 by Meh1,l and by Dorn, De Garmo and Flanigan.2, It is difficult to interpret the data, for in most cases too little information is given on austenitizing temperatures and times, the state of homogeneity of the austenite (presence or absence of carbide, etc.), austenite grain size, and mode of formation of pearlite at the temperatures given. Except for the approximations of Mehl and the work of Dorn, austenite grain size was disregarded, though it is well known

Jan 1, 1942

By J. K. Stanley, R. F. Meel

The recrystallization of cold-worked metals is studied ordinarily by determining the temperatures required for complete recrystallization to occur within a given arbitrary time period, usually within 15 min. to 2 hr. The results obtained are conventionally assembled in a three-dimensional diagram relating this temperature to the amount of cold-work (percentage deformation) and the resultant grain size, though sometimes other modes of representation are emp1oyed.l Such three-dimensional diagrams are not complete, for the number of variables exceeds the number represented. The process of recrystallization is characterized by a time-rate, a fact that is not portrayed upon ordinary recrystallization diagrams. It is known that recrystallization consists in the formation of new nuclei and the growth of these nuclei, a process that continues until the matrix is completely consumed and the original set of grains entirely replaced by the new. This knowledge is old.2 Truly heterogeneous reactions invariably proceed by nucleation and growth; the decomposition of eutectoid solid solutions provides an example about which much is known,3 and freezing, originally studied by Tammann, provides the classic example. Although recrystal- lization is not a heterogeneous reaction within the orthodox meaning of the term, the mechanism of the process is the same. As in eutectoid inversion and freezing, it should be possible to study the rate of isothermal recrystallization as a function of time, and it should also be possible to analyze the isothermal reaction curve into the component rates of nucleation and growth that determine them.* Other methods, not employing the rates of nucleation and growth, have been used to represent the rates of such reactions, but these in general have been artificial and have not led to progress. 4,5 The theory of recrystallization is not in advanced state; if recrystallization data were obtained in terms of the rate of isothermal recrystallization, and in terms of the rate of nucleation and the rate of growth of recrystallized grains, and in particular if the temperature coefficients of these rates were measured accurately, an adequate theory should be more readily developed.27 Such an effort requires proper experimental technique, but especially requires suitable methods of mathematical analysis to convert mcasurements made upon a surface of polish into values for the rate of nucleation and the rate of growth and to combine these into a rate of isothermal recrystallization. These methods are now available for the case of three-dimensional recrystallization (recrystallization of

Jan 1, 1942

By R. F. Meel, J. K. Stanley

The recrystallization of cold-worked metals is studied ordinarily by determining the temperatures required for complete recrystallization to occur within a given arbitrary time period, usually within 15 min. to 2 hr. The results obtained are conventionally assembled in a three-dimensional diagram relating this temperature to the amount of cold-work (percentage deformation) and the resultant grain size, though sometimes other modes of representation are emp1oyed.l Such three-dimensional diagrams are not complete, for the number of variables exceeds the number represented. The process of recrystallization is characterized by a time-rate, a fact that is not portrayed upon ordinary recrystallization diagrams. It is known that recrystallization consists in the formation of new nuclei and the growth of these nuclei, a process that continues until the matrix is completely consumed and the original set of grains entirely replaced by the new. This knowledge is old.2 Truly heterogeneous reactions invariably proceed by nucleation and growth; the decomposition of eutectoid solid solutions provides an example about which much is known,3 and freezing, originally studied by Tammann, provides the classic example. Although recrystal- lization is not a heterogeneous reaction within the orthodox meaning of the term, the mechanism of the process is the same. As in eutectoid inversion and freezing, it should be possible to study the rate of isothermal recrystallization as a function of time, and it should also be possible to analyze the isothermal reaction curve into the component rates of nucleation and growth that determine them.* Other methods, not employing the rates of nucleation and growth, have been used to represent the rates of such reactions, but these in general have been artificial and have not led to progress. 4,5 The theory of recrystallization is not in advanced state; if recrystallization data were obtained in terms of the rate of isothermal recrystallization, and in terms of the rate of nucleation and the rate of growth of recrystallized grains, and in particular if the temperature coefficients of these rates were measured accurately, an adequate theory should be more readily developed.27 Such an effort requires proper experimental technique, but especially requires suitable methods of mathematical analysis to convert mcasurements made upon a surface of polish into values for the rate of nucleation and the rate of growth and to combine these into a rate of isothermal recrystallization. These methods are now available for the case of three-dimensional recrystallization (recrystallization of

Jan 1, 1942

By J. T. Eash, N. B. Pilling

As a group, the alloys of iron, nickel and carbon are, in application, one of the most versatile of the ferrous alloy family, and while many investigations have been made of their properties and structure only a few attempts have been made to show the microstructural relation over a wide range of composition. It is neccssary at once to be clear about the kind of structural relation with which we are concerned. The equilibrium diagram provides quantitative data regarding structural make-up and its variation with temperature, but equilibrium structures often are far different from the technologically significant ones. This is particularly true of steels and cast irons, and some of the most important cornponents of their structure may have no place in the diagram of equilibrium. The fugitive phases that result from transFormation are dealt with in a highly detailed and descriptive manner in. the S-curves originated by Bain. The present effort lies in a still different field, which is to portray the relation between sweeping changes in composition and in resultant structure, under controlled but nonequilib-rium conditions of thermal treatment. The Guilletl diagram presented in 1903 was the initial attempt at such a correlation and frequent references to this diagram can be found in the literature even though later work has shown it to require considerable modification. Kase2 published in 1925 a diagram that was based upon observations of furnace-cooled melts: At low carbon contents this diagram agreed fairly well with experience in showing pearlite to exist in alloys containing up to 6 per cent nickel, and completely austenitic steels at 24 to 28 per cent nickel. At high carbon contents, however, irons were indicated as being martensitic at around 2 per cent nickel and austenitic at 12 per cent nickel; these interpretations are open to serious question. Based upon experience with steels and cast iron, Wickenden3 suggested a diagram that agreed fairly well with Kase's in the steel range but which presented a wide departure from it in the higher-carbon, high-nickel regions. In its broad picture, the Wickenden diagram is an approach to "practical" structures. Reference may be made at this point to the efforts of Marsh5 to construct an iron-nickel-carbon equilibrium diagram by the coordination of scattered data, which are described at length in the monograph cited. None of these preceding attempts has appeared to meet the need for an exposition of the structural sequence that is encountered in traversing the nickel-steel nickel-cast-iron range under what might be called ordinary departures from equilibrium. ExperimentaT. Conditions Since the intent has been to establish structures developed under arbitrary conditions, experimental details have perhaps more than usual importance. Three of these conditions should particularly be noticed: (I) the material was cast and not subjected

Jan 1, 1942

By J. T. Eash, N. B. Pilling

As a group, the alloys of iron, nickel and carbon are, in application, one of the most versatile of the ferrous alloy family, and while many investigations have been made of their properties and structure only a few attempts have been made to show the microstructural relation over a wide range of composition. It is neccssary at once to be clear about the kind of structural relation with which we are concerned. The equilibrium diagram provides quantitative data regarding structural make-up and its variation with temperature, but equilibrium structures often are far different from the technologically significant ones. This is particularly true of steels and cast irons, and some of the most important cornponents of their structure may have no place in the diagram of equilibrium. The fugitive phases that result from transFormation are dealt with in a highly detailed and descriptive manner in. the S-curves originated by Bain. The present effort lies in a still different field, which is to portray the relation between sweeping changes in composition and in resultant structure, under controlled but nonequilib-rium conditions of thermal treatment. The Guilletl diagram presented in 1903 was the initial attempt at such a correlation and frequent references to this diagram can be found in the literature even though later work has shown it to require considerable modification. Kase2 published in 1925 a diagram that was based upon observations of furnace-cooled melts: At low carbon contents this diagram agreed fairly well with experience in showing pearlite to exist in alloys containing up to 6 per cent nickel, and completely austenitic steels at 24 to 28 per cent nickel. At high carbon contents, however, irons were indicated as being martensitic at around 2 per cent nickel and austenitic at 12 per cent nickel; these interpretations are open to serious question. Based upon experience with steels and cast iron, Wickenden3 suggested a diagram that agreed fairly well with Kase's in the steel range but which presented a wide departure from it in the higher-carbon, high-nickel regions. In its broad picture, the Wickenden diagram is an approach to "practical" structures. Reference may be made at this point to the efforts of Marsh5 to construct an iron-nickel-carbon equilibrium diagram by the coordination of scattered data, which are described at length in the monograph cited. None of these preceding attempts has appeared to meet the need for an exposition of the structural sequence that is encountered in traversing the nickel-steel nickel-cast-iron range under what might be called ordinary departures from equilibrium. ExperimentaT. Conditions Since the intent has been to establish structures developed under arbitrary conditions, experimental details have perhaps more than usual importance. Three of these conditions should particularly be noticed: (I) the material was cast and not subjected

Jan 1, 1942

By Blake M. Loring

Recently the S-curves for 30 to 40 alloy steels have been pub1ished.l.2 These steels show individual characteristics, which make each additional S-curve of great interest. There are important differences in the mechanism of the transformation process, the type of transformation product, and the influence of alloying elements. Steels of the same general alloy composition may possess such variable characteristics with regard to the length of time of isothermal transformation that an S-curve for each heat of steel may be desirable. In the present work the isothermal transformation of a steel having the following composition and grain size was studied: 0.29 per cent carbon; 0.21 manganese; 0.026 phosphorus; 0.017 sulphur; 0.056 silicon; 1.45 chromium; 3.25 nickel; austenite grain size, 7. The progress of the isothermal transformation was followed by several different methods, including hardness measurements, dilatometer observations, and examinations of the microstructure. It was not expected that these methods would be in precise agreement.3 Accordingly, the changes in microstructure and hardness were given additional weight in determining the initial stage of transformation and the dilatometer measurements were relied on in estimating the completion of the process. The beginning of transformation was indicated by I per cent of transformation products and the end of transformation by about 99 per cent. The reproducibility of the various methods in the ranges of highest sensitivity was checked by several observations at each time and temperature. The steel used in this investigation was annealed for 24 hr, at 925°C. Macroetching revealed no serious segregation. The specimens for the metallographic and hardness investigations were 1¼ by ½ by ? in. Wires attached to the samples were used for handling. instead of tongs, to avoid chilling. The specimens were heated at 845°C. for ½ hr. and quenched into constant-temperature salt baths held at various temperatures from 205' to 650°C. The time required to transfer the specimens was found to be about 1½ sec., and nearly 5 sec. more was required for them to come to the temperature of the bath. After various lengths of time in the constant-temperature baths the specimens were quenched in water at 2I°C. The specimens were cut in half for the metallographic examination, to provide a surface free from decarburization. To determine the beginning and end of the transformation the percentage of transformation determined metallographically and by hardness was plotted against time. A planimeter was used in the metallographic determination. The hardness and metallographic determinations made at room temperature were confirmed reasonably well by dilatometer measurements taken during the process of transformation at the various temperatures. Two dilatometers were used to

Jan 1, 1942

By Blake M. Loring

Recently the S-curves for 30 to 40 alloy steels have been pub1ished.l.2 These steels show individual characteristics, which make each additional S-curve of great interest. There are important differences in the mechanism of the transformation process, the type of transformation product, and the influence of alloying elements. Steels of the same general alloy composition may possess such variable characteristics with regard to the length of time of isothermal transformation that an S-curve for each heat of steel may be desirable. In the present work the isothermal transformation of a steel having the following composition and grain size was studied: 0.29 per cent carbon; 0.21 manganese; 0.026 phosphorus; 0.017 sulphur; 0.056 silicon; 1.45 chromium; 3.25 nickel; austenite grain size, 7. The progress of the isothermal transformation was followed by several different methods, including hardness measurements, dilatometer observations, and examinations of the microstructure. It was not expected that these methods would be in precise agreement.3 Accordingly, the changes in microstructure and hardness were given additional weight in determining the initial stage of transformation and the dilatometer measurements were relied on in estimating the completion of the process. The beginning of transformation was indicated by I per cent of transformation products and the end of transformation by about 99 per cent. The reproducibility of the various methods in the ranges of highest sensitivity was checked by several observations at each time and temperature. The steel used in this investigation was annealed for 24 hr, at 925°C. Macroetching revealed no serious segregation. The specimens for the metallographic and hardness investigations were 1¼ by ½ by ? in. Wires attached to the samples were used for handling. instead of tongs, to avoid chilling. The specimens were heated at 845°C. for ½ hr. and quenched into constant-temperature salt baths held at various temperatures from 205' to 650°C. The time required to transfer the specimens was found to be about 1½ sec., and nearly 5 sec. more was required for them to come to the temperature of the bath. After various lengths of time in the constant-temperature baths the specimens were quenched in water at 2I°C. The specimens were cut in half for the metallographic examination, to provide a surface free from decarburization. To determine the beginning and end of the transformation the percentage of transformation determined metallographically and by hardness was plotted against time. A planimeter was used in the metallographic determination. The hardness and metallographic determinations made at room temperature were confirmed reasonably well by dilatometer measurements taken during the process of transformation at the various temperatures. Two dilatometers were used to

Jan 1, 1942

By R. W. Gurry

In the course of a series of measurements of the rate of diffusion of carbon in austenite at about 960°C. (1760°F.) and III0°C. (2030°F.), it became necessary to determine carbon concentration when austenite is saturated with graphite, because of indications that the hitherto accepted value of the solubility of graphite in this temperature range is too low. Specimens of carbonyl iron were saturated at temperature in a hydrogen-toluene atmosphere which was in equilibrium with graphite. The carbon conteDt of the steel was measured either by combustion or by the loss of weight after substantially complete decarburization in an atmosphere of hydrogen saturated at room temperature with water vapor. The absence of carbide or graphite as a phase at temperature was demonstrated by microscopic examination of saturated specimens that had been drastically quenched, The mean result of concordant ments-—taeats-is-^that gamma iron saturated with respect to graphite contains 1.39 per cent carbon at 957°C .(i755°F.) and 1.89 per cent at III0°C. (2030°F.). These two values, plotted along with what appears from the literature to be the best value (0.69 per cent) at the iron-graphite eutec-toid temperature (738°C. or I360°F.), yield a curve that is almost linear, and which, continued upward through a short interval, leads to a limiting solubility of 1.98 per cent at the iron-graphite eutectic temperature II35°C. (2075°F.) instead of the commonly accepted value of about 1.7 per cent. Experimental Procedure The samples of carbonyl iron, with the exception of those for microscopic examination, were cylinders 3.8 cm. long, 0.475 cm. in diameter. They were placed in a graphite holder and suspended in a vertical furnace, the temperature of which was controlled precisely by means of the resistance of the platinum winding.' In the empty furnace the range of temperature over a distance of 4 cm. at the heat center was 9°C. (I6°F.) but at a given location the temperature remained within +o.5°C. (o.9°F.); it was measured by inserting a platinum-rhodium couple calibrated at the melting point of palladium, 1555°C. Since the presence of the sample probably improved the uniformity of temperature, it was decided to consider the average temperature of the sample to be 3°C. lower than the maximum measured in the empty furnace with the same setting of the controller. Data were obtained in the vicinity of two temperatures, 960°C. (I760°F.) and III0°C. (2030°F.). The samples for microscopic examination, 2.2 cm. long, 0.475 cm. in diameter, were suspended individually on pure iron wire; for these no temperature correction was applied. The carburizing gas passed into the furnace was dry, oxygen-free hydrogen saturated at room temperature with toluene; this quickly produced on the

Jan 1, 1942

By R. W. Gurry

In the course of a series of measurements of the rate of diffusion of carbon in austenite at about 960°C. (1760°F.) and III0°C. (2030°F.), it became necessary to determine carbon concentration when austenite is saturated with graphite, because of indications that the hitherto accepted value of the solubility of graphite in this temperature range is too low. Specimens of carbonyl iron were saturated at temperature in a hydrogen-toluene atmosphere which was in equilibrium with graphite. The carbon conteDt of the steel was measured either by combustion or by the loss of weight after substantially complete decarburization in an atmosphere of hydrogen saturated at room temperature with water vapor. The absence of carbide or graphite as a phase at temperature was demonstrated by microscopic examination of saturated specimens that had been drastically quenched, The mean result of concordant ments-—taeats-is-^that gamma iron saturated with respect to graphite contains 1.39 per cent carbon at 957°C .(i755°F.) and 1.89 per cent at III0°C. (2030°F.). These two values, plotted along with what appears from the literature to be the best value (0.69 per cent) at the iron-graphite eutec-toid temperature (738°C. or I360°F.), yield a curve that is almost linear, and which, continued upward through a short interval, leads to a limiting solubility of 1.98 per cent at the iron-graphite eutectic temperature II35°C. (2075°F.) instead of the commonly accepted value of about 1.7 per cent. Experimental Procedure The samples of carbonyl iron, with the exception of those for microscopic examination, were cylinders 3.8 cm. long, 0.475 cm. in diameter. They were placed in a graphite holder and suspended in a vertical furnace, the temperature of which was controlled precisely by means of the resistance of the platinum winding.' In the empty furnace the range of temperature over a distance of 4 cm. at the heat center was 9°C. (I6°F.) but at a given location the temperature remained within +o.5°C. (o.9°F.); it was measured by inserting a platinum-rhodium couple calibrated at the melting point of palladium, 1555°C. Since the presence of the sample probably improved the uniformity of temperature, it was decided to consider the average temperature of the sample to be 3°C. lower than the maximum measured in the empty furnace with the same setting of the controller. Data were obtained in the vicinity of two temperatures, 960°C. (I760°F.) and III0°C. (2030°F.). The samples for microscopic examination, 2.2 cm. long, 0.475 cm. in diameter, were suspended individually on pure iron wire; for these no temperature correction was applied. The carburizing gas passed into the furnace was dry, oxygen-free hydrogen saturated at room temperature with toluene; this quickly produced on the

Jan 1, 1942

By R. W. Gurry

When a steel in the austenitic state, with all its carbon in solution, is maintained, at constant temperature, in contact with a gas that removes the carbon from the surface, yet without otherwise altering the composition of the steel, the change in weight of the specimen, after any period of exposure, is a direct measure of the amount of carbon it has then lost. The purpose of the present paper is to show how, by application of established mathematical reasoning based on the fundamental diffusion law and on certain postulates, this loss of weight yields a complete picture of the distribution of carbon throughout a specimen of regular shape after any period of exposure, with all the accuracy required in any practical occurrence of decarburization. Precisely the same is true for the gain in weight during the reverse process of carburization, or for the gradual gain or loss of any element, so long as the steel comprises only a single phase at temperature, provided always that the experimental conditions conform to the postulates. The change in carbon content from surface to center of the specimen is always continuous, when everything is in solution, even though under the microscope the decarburized zone (or carburized rim) may appear to be sharply defined. The experimental procedure is described and some typical results are presented, which are then interpreted by application of the diffusion law. Experimental Procedure Initially an endeavor was made to follow the gain of weight during the process of carburization in a hydrogen-toluene atmosphere, but it was soon found that the specimen became coated with graphite, which could not be removed without some loss of metal; therefore it was decided to continue carburization until the steel was saturated throughout, to remove all graphite from the surface of the specimen, and then to follow the loss of weight during decarburization in a moist hydrogen atmosphere at the same temperature as that used for carburizing. The carburizing procedure has been described in an earlier paper;' it suffices to state here that the carburization lasted about I40 hr. at 760°C. (1760°F.) and 44 hr. at iiio°C. (2030°F.), these periods being long enough-—on the basis of the calculations presented later—to ensure that the specimen was about 99.9 per cent saturated at each temperature. The specimen was a cylinder about 3.8 cm. long and 0.475 cm. in diameter; the materials were carbonyl iron, a commercial low-metalloid iron, and a steel with 3.6 per cent nickel and the usual proportion of other elements. The fully carburized specimen, after cooling at the top of the furnace tube, was placed in a lathe and sandpapered to remove all graphite; then the diameter was measured with a micrometer caliper, and the initial weighing (corresponding to complete saturation) was made. By means of a

Jan 1, 1942

By R. W. Gurry

When a steel in the austenitic state, with all its carbon in solution, is maintained, at constant temperature, in contact with a gas that removes the carbon from the surface, yet without otherwise altering the composition of the steel, the change in weight of the specimen, after any period of exposure, is a direct measure of the amount of carbon it has then lost. The purpose of the present paper is to show how, by application of established mathematical reasoning based on the fundamental diffusion law and on certain postulates, this loss of weight yields a complete picture of the distribution of carbon throughout a specimen of regular shape after any period of exposure, with all the accuracy required in any practical occurrence of decarburization. Precisely the same is true for the gain in weight during the reverse process of carburization, or for the gradual gain or loss of any element, so long as the steel comprises only a single phase at temperature, provided always that the experimental conditions conform to the postulates. The change in carbon content from surface to center of the specimen is always continuous, when everything is in solution, even though under the microscope the decarburized zone (or carburized rim) may appear to be sharply defined. The experimental procedure is described and some typical results are presented, which are then interpreted by application of the diffusion law. Experimental Procedure Initially an endeavor was made to follow the gain of weight during the process of carburization in a hydrogen-toluene atmosphere, but it was soon found that the specimen became coated with graphite, which could not be removed without some loss of metal; therefore it was decided to continue carburization until the steel was saturated throughout, to remove all graphite from the surface of the specimen, and then to follow the loss of weight during decarburization in a moist hydrogen atmosphere at the same temperature as that used for carburizing. The carburizing procedure has been described in an earlier paper;' it suffices to state here that the carburization lasted about I40 hr. at 760°C. (1760°F.) and 44 hr. at iiio°C. (2030°F.), these periods being long enough-—on the basis of the calculations presented later—to ensure that the specimen was about 99.9 per cent saturated at each temperature. The specimen was a cylinder about 3.8 cm. long and 0.475 cm. in diameter; the materials were carbonyl iron, a commercial low-metalloid iron, and a steel with 3.6 per cent nickel and the usual proportion of other elements. The fully carburized specimen, after cooling at the top of the furnace tube, was placed in a lathe and sandpapered to remove all graphite; then the diameter was measured with a micrometer caliper, and the initial weighing (corresponding to complete saturation) was made. By means of a

Jan 1, 1942

By William L. Fink, H. R. Freche

The constitution of the aluminum-cobalt alloys has been the subject of relatively few investigations. Brunckl analyzed the residue which remained after treating the 17 per cent cobalt alloy with dilute hydrochloric acid. The results of the analyses indicated the existence of a compound Co3Al13. Guillet2 published a liquidus for the aluminum-cobalt system which contained two maxima corresponding to CoA16 and CoAl2 and two eutectic points at 30 and 74 per cent Co. The most extensive investigation of the system was carried out by Gwyer13 whose results are summarized in Fig. 1. It will be noted that at the aluminum end there is a suppressed eutectic at 644' C. By extrapolation of the duration of the eutectic arrest to zero, the composition of the compound was placed at Co3Al13. Another thermal arrest was noted at 550" C., which Gwyer attributed to a peritectic reaction. The present investigation of the aluminum end of the aluminum-cobalt system is the eleventh of a series dealing with the constitution of aluminum-rich alloys prepared from electrolytically refined a1uminum.l Preparation of Alloys.—The alloys used in this work were all prepared from electrolytically refined aluminum, the analysis of which is shown in Table 1. In preparing some of the alloys, metallic cobalt (98 to 99 per cent) was used either directly or in the form of an intermediate alloy, the composition of which is given in Table 1. In order to obtain alloys of higher purity, cobalt was introduced into some of the alloys by means of chemically pure cobaltous chloride which was readily reduced by the aluminum. A small electric furnace with nickel-chromium heating elements was used in the preparation of the alloys. Chemical Analysis.—An analysis sample from each melt was chill-cast in the form of a plate 3/16 in. thick. In all cases the casting temperature was well above the liquidus. Cobalt was determined electrolytically in all alloys containing more than 0.05 per cent cobalt and by the nitroso-

Jan 1, 1932

By E. H. Dix

The work of Merical and other investigators indicates that the phenomenon of age-hardening in alloys of the duralumin type is primarily dependent upon the variation in the solubility of copper with temperature. In addition to copper, however, these alloys contain small amounts of magnesium, silicon, manganese and iron, which may affect the age-hardening either directly or indirectly. The effect of these elements on the solubility of copper and on the manner in which this solubility varies with the temperature is of considerable interest, therefore. The effect of magnesium is of especial interest because apparently it exerts a marked effect on the natural or room-temperature aging of alloys of the type described.2 A series of investigations3 has been in progress in the research laboratories of the Aluminum Company of America to determine the functions * Metallurgist, Aluminum Research Laboratories. f Metallurgical Division, Aluminum Research Laboratories. Formerly with Metallurgical Division, Aluminum Research Laboratories. 1 P. D. Merica, It. G. Waltenberg and H. Scott: Heat Treatment of Duralumin. U. S. Bur. Stds. Sn. Paper 347 (1910). 2 M. Gayler: Constitution and Age-Hardening of Alloys of Aluminum with Copper, Magnesium, and Silicon in the Solid State. Jnl. Inst. Met. (1922) 28, 213. W. Fraenkel: Vergutbare Aluminiumlegiengen. Ztsch. Melallkunde. (1920) 12, 427. E. H. Dix, Jr.: Note on the Microstructure of Aluminum-Iron Alloys of High Purity. Proc. Amer. Soc. Test. Mat. (1925) 26, Pt. 2, 120. E. H. Dix, Jr. and H. H. Richardson: Equilibrium Relations in Aluminum-copper Alloys of High Purity. Trans. A. I. M. E. (1926) 73, 560. E. H. Dix, Jr. and W. D. Keith: Equilibrium Relations in Aluminum-manganese Alloys of High Purity. Proc. Inst. Met. Div., A. I. M. E. (1927) 315. E. H. Dix, Jr. and A. C. Heath, Jr.: Equilibrium Relations in Aluminum-silicon and Aluminum-iron-silicon Alloys of High Purity. Proc. Inst. Met. Div., A. I. M. E. (1928) 164. E. H. Dix, Jr. and F. Keller: Equilibrium Itelations in Aluminum-magnesium Alloys of High Purity. Proc. Inst. Met. Div., A. I. M. E. (1920) 351. E. H. Dix, Jr., F. Keller and R. W. Graham: Equilibrium Relations of Aluminum-magnesium Silicide Alloys of High Purity. Trans. A. I. M. E., Inst. Met. Div. (1931) 404. W. L. Fink and K. R. Van Horn: Constituents of Aluminum-iron-silicon Alloys. Trans. A. I. M. E., Inst. Met. Div. (1931) 383.

Jan 1, 1932