(!|otneU Httluersttg Slihcarg atl^aca, Sitm lack BOUGHT WITH THE INCOME OF THE SAGE ENDOWMENT FUND THE GIFT OF HENRY W. SAGE 1891 Cornell University Library TS 320.G49 Heat treatment of soft and medium steels 3 1924 004 577 346 Cornell University Library The original of tiiis book is in tine Cornell University Library. There are no known copyright restrictions in the United States on the use of the text. http://www.archive.org/details/cu31924004577346 HEAT TREATMENT OF SOFT AND MEDIUM STEELS VIk Qmw^!)/ill Book (h 7m PUBLISHERS OF B O O K. S F O R_/ Coal Age ^ Electric Railway Journal Electrical World v Engineering News-Record American Machinist ^ Ingenierfa Intemacional Engineering g Mining Journal "^ Power Chemical 6 Metallurgical "Engineering Electrical Merchandising HEAT TREATMENT OF SOFT AND MEDIUM STEELS THEORY AND PRACTICE OF THE PRELIMINARY HEAT TREATMENTS DESIGNED TO GIVE MAXIMUM TOUGH- NESS TO STEELS USED FOR MACHINE PARTS BY FEDERICO GIOLITTI, Ph.D. BBBSEM£B MEDALLIST TRANSLATED BY E. E. THUM ASSOCIATE EDITOB, CHEMICAL AND MBTALLTJBGICAL ENGINEERING AND D. G. VERNACI METALLUBGICAL ENGINEER FiBST Edition McGRAW-HILL BOOK COMPANY, Inc. NEW YORK: 370 SEVENTH AVENUE LONDON: 6 & 8 BOUVERIE ST., E. C. 4 1921 Copyright, 1921, by the McGraw-Hill Book Company, Inc. THE MAPLE PKESS TOSK FA. TRANSLATOR'S PREFACE It has been often observed that man progresses from the simple and wasteful to the complex and economical. Progress in metallurgical art is no exception. From direct iron and steel (made from ore in simple forges, and in one operation) we have come as far as the mild steel of commerce, made from ore through an intermediate impure blast-furnace product, pig iron. Further progress may be expected along the same lines. A growing demand for the maximum of strength in the minimum of weight is inevitable, an end attainable by a further complication in process: either super-refining, perhaps in electric furnaces, or heat treatment carried far beyond the practices now commonly used on small parts of good quality. Metallurgists and metallographists have only recently been impressed with the fact that various impurities and addition agents may affect the properties of finished steel far in excess of that expected by their apparent amount. Precise data along these lines are almost entirely lacking; therefore Dr. Giolitti's present book, containing the first systematic discussion of their effect on commercial heat treatment, should prove a powerful stimulus toward their study; proving as he does the tremendous advantages to be gained by their elimination or suppression. Especially respectful attention will be given by those who have been bafHed by so-called "flakes" and "woody fractures" — a disease of metals well understood and under control in the Italian works managed by the au- thor — and by the many consumers and producers who will ultimately be dealing in alloy steel castings of high strength and toughness, such as have been under tonnage production abroad for several years. Previously published texts and theoretical treatises have illustrated the relationships between iron and carbon in the form of a diagram for complete equilibrium, giving all too scant attention to the difficulties of reaching this ideal. If for nothing else. Dr. Giolitti's volume should be of great usefulness for its insistence that commercial, heat treatment of steel depends primarily upon the diffusion of carbon and other soluble substances contained, gaseous or solid. In turn the approach to equilib- rium is profoundly affected by the speed of cooling, the spacing of centers of crystallization, the content of insoluble non-metallic inclusions and their physical and chemical constitution, the temperature of transforma- tion and the thermal hysteresis. He who would master his metal, must master these things. E. E. Thum. New Yokk City, March, 1921. Preface . CONTENTS Paqe V PART I THE PHENOMENA OF DIFFUSION IN PRIMARY SOLID SOLUTION Chapter I. General Characteristics of Crystallization of Soft and Medium Steels and of Its Preliminary Heat Treatment 1 II. Primary Crystallization of a Binary Solution under Complete Equilibrium . . . . ... . . 9 III. Actual Course of Crystallization of a Liquid Binary System 19 IV. Primary Crystallization of a Ternary System . 32 V. Primary Crystallization of More Complex Systems 37 PART II EFFECTS OF DIFFUSION UPON SECONDARY CRYSTALLIZATION VI. General Remarks Upon the Relations Between Diffusion and Secondary Crystallization . . 53 VII. The Beginning of Secondary Crystallization in Homogeneous AUSTENITE . 60 VIII. Completion of Secondary Crystallization of Uniform Austenite 75 IX. Secondary Crystallization op Non-homogeneous Austenite in the Binary Iron : Carbon System ..... 84 X. Secondary Crystallization in More Complex Systems . 93 PART III DIFFUSION IN AUSTENITE AS APPLIED TO THE PRELIMINARY HEAT TREATMENT OF STEELS XI. Fundamental Criteria for Practical Application of Preliminary Heat Treatments ... .... . . . 103 XII. Selection op Preliminary Heat Treatment According to the Characters of THE 7 —> a Transformation . . . . 113 XIII. Phenomena Which May Disturb the Course and Modify the Effects of Preliminary Heat Treatments .... .... 123 PART IV PRELIMINARY HEAT TREATMENT OF STEEL CASTINGS XIV. General Morphological Relationships Between the Secondary AND THE Primary Crystalline Systems. . . ... . . 145 viii CONTENTS Chapter Page XVI. Critbbia fob the Selection op Casting Steels . 193 XVII. Pkelimin ART Heat Tbeatmbnt OF Steel Castings. ... . . 216 XVIII. Effect of Solid Non-metallic Inclusions Upon Pkeliminabt Heat Tbeatment. . . 266 PART V PRELIMINARY HEAT TREATMENT OF FORGED AND ROLLED STEELS XIX. A Genebal Discussion of How the Effects of Heat Tbeatmbnt ABB Modified by Hot Wobk .... 277 XX. Effects of Peeliminabt Heat Treatment Upon Forged or Rolled Steels 298 XXI. Comparative Effect of Annealing and Quenching upon the Longi- tudinal Strength of Hot Wobked Steels 320 XXII. Compabative Effect of Annealing and Quenching upon the Tbans- VEBSE StBENGTH OF HOT WoBKED StEEL 330 Index 365 HEAT TREATMENT OF SOFT AND MEDIUM STEELS PART I THE PHENOMENA OF DIFFUSION IN PRIMARY SOLID SOLUTION CHAPTER I GENERAL CHARACTERISTICS OF CRYSTALLIZATION OF SOFT AND MEDIUM STEELS AND OF ITS PRELIMINARY HEAT TREATMENT 1. However extensive the series of mechanical and thermal treat- ments which may have been undergone during manufacture, every metallic alloy always retains a more or less evident imprint of many of the characteristic properties directly resulting from the special con- ditions of fusion and crystallization, those two successive steps in the formation of the alloy itself. Moreover, one can state that the great majority of the various manufacturing operations and manipulations upon metallic alloys — excepting those accomplished with the aid of cutting tools — have as their principal object the more or less complete elimination or transformation of those characteristics residual from circumstances of melting and crystaUization. The above statements may perhaps apply more broadly to steels than to other commercial alloys. As a result of this state of affairs, a piece of steel throughout its entire manufacture bears neatly impressed upon it those characteristics imparted by melting practice and method of casting. 2. The properties due to the first of these two phases — that is, due to fusion — are characterized by a greater stability in the sense that it is far more difficult to modify them by means of subsequent operations on the solidified metal. To appreciate this statement it is enough to remem,ber that the chemical analysis is the first and most important of the properties directly derived from the melting process. It is well known that, generally speaking, this chemical composition and the physical prop- erties strictly inherent to it are quite difficult to modify in sohdified metal; so difficult indeed that modifications of this kind are very seldom resorted to in industry, and then only for limited portions of the steel 1 2 HEAT TREATMENT OF articles under treatment. As examples of the special processes for this purpose, one may recall case-hardening processes, decarburization of malleable iron, sherardizing, and a very few others. Among other characteristic properties directly due to the melting practice, one may mention the various degrees of deoxidation of the metal, and the kind, quantity, subdivision, and distribution of the non- metallic inclusions, all of which are difficult to modify by manipulating the solidified steel. It is my purpose to study this first group of characteristic properties of steel only indirectly at this time and only in those cases in which they exercise an influence which cannot be disregarded upon the crystal- lization phenomena, thus substantially modifying those properties characteristic of the second group. As an example of the few cases to be discussed may be mentioned (a) the effect that the average con- centration of a given element in a bath of molten steel may exercise on the extent of intracrystalline and extra-crystalline liquation occurring during crystaUization, and (6) the energetic action exercised by the non- metallic inclusions in certain cases upon the crystallization of steel. 3. A great deal less persistent are the properties of the steel due to the solidification process. In fact these properties depend in great part directly and exclusively upon the crystalline state of the metal. Now, this crystallinity is nearly always easy to modify profoundly by means of operations on the totally solidified metal such as forging at a more or less high temperature, rolling either hot or cold, cold drawing, heat treating, etc. There are of course some exceptions to this rule; but these are not frequent in general metallurgy and are extremely rare in steel practice. As an example one can mention the unusual case where intracrystalline or extra-crystalHne hquation occurring during the cooling of a metallic alloy may reach such a state as to cause the formation of a new solid phase, whose existence is not consistent with the average composition of the alloy. In this case it sometimes happens in practice that the metastable equilibrium thus created cannot be destroyed and replaced by stable equilibrium by any mechanical operation or heat treatment short of remelting. A similar case sometimes happens in certain special kinds of high-speed tool steels. Leaving aside the rare exceptions mentioned, one can conclude that it is nearly always possible to greatly modify the crystalline structure of steel by means of mechanical or thermal treatments upon the metal in its solid state. The alterations which can thus be obtained necessarily belong to one of the two following groups, perfectly distinct one from the other: (a) Modifications of the crystalline structure of a purely morpho- logical nature. SOFT AND MEDIUM STEELS 3 ib) Chemical transformations in each crystaUine element or in the "colonies" in which they congregate. Modifications belonging to the first group can be obtained either by mechanical or thermal manipulation, although in practice the former is more frequently depended upon. Modifications of the second group, on the contrary, may be brought about almost exclusively by means of heat treatment. Whenever a complex treatment designed to effect chemical alterations includes one or more mechanical operations, the principal reason for the latter is usually to produce such a changed crystalUne form as to expedite and facilitate the desired transformation of the second group. We shall later have occasion to see several examples of these facts. My primary purpose in this volume is to investigate the heat treatments by which certain chemical modifications can be obtained within the crystals in a steel. I shall mention the application of mechanical work only in those cases where its purposes and consequences are strictly connected with those of the heat treatments. It is now opportune, therefore, to outhne in a general way the broad features of thermal transformations. 4. A concise statement may be made of the structure of the solid masses formed in a molten steel during slow cooling by means of an analysis of the process of solidification and of the properties of the resulting crystalline mass. Such reflection shows clearly that im- mediately after the crystallization process has been entirely completed the metal consists of a conglomeration of mixed crystals' containing the various elements which enter into the composition of our particular steel in a solid solution in 7 iron. This statement includes all the elements which appear in the com- position of commercial steels up to the limits imposed by specifications for metal adapted to the construction of machine elements. The separation of the most minute elementary crystals of solid solution from the molten steel and their rapid assemblage into colonies, forming the skeleton of the finally solidified metallic mass, takes place in substantial accordance with the well-known principles covering the formation of solid solutions. It is clear that a thorough study of these phenomena, based upon the direct apphcation of these principles to the various possible cases, would give precise and complete information which would be an in- valuable guide in the practical applications of heat treatment processes. However, a systematic, complete and exact study of the solidification of mixed crystals is so complex as to be practically impossible. This is easily understood when one stops to consider that on the one hand 1 Europeans use the terms "mixed crystals" and "solid solutions" in a synonymous manner — which will be the significance understood in this volume. 4 HEAT TREATMENT OF the great majority of the steels used in practice are alloys of six elements (iron, carbon, sihcon, manganese, sulphur and phosphorus), to say nothing of those which contain nine or even ten elements; and that, on the other hand, up to the present time complete equilibrium diagrams have been studied only for alloys containing but four elements at most. Nevertheless, from the standpoint of useful practical apphcations of theory the impossibiUty of making such a thorough and systematic research does not constitute an unsurmountable obstacle to an efficient study of the problem. The reasons follow: In view of the moderate concentrations of the various elements as they ordinarily occur in the large majority of the steels used in practice, it can be stated as a fact that the various solid solutions which segregate from the molten mass as soon as the crystallization starts all belong to the same series, or, in other words, all exhibit the same general characteristics. It should be remembered that we have experience with steels covering a considerable range of alloying elements, even with those which contain the minimima concentration which can be obtained for each different element or impurityin practice, approaching purest iron. Therefore we can be sure that there exists no discontinuity in the series of soUd solutions lying between the concentrations usually met with in practice and pure iron. Furthermore, excluding some very rare instances of slight practical importance., the mixed crystals appearing during the successive stages of crystaUization until the metal is totally solidified all belong to the same family as those which formed when soUdification began, again speaking in a physico-chemical sense rather than in a crystaUographic. This statement remains true for many steels regardless of the velocity with which the mushy stage is crossed in cooling, and for these steels we can limit our study by considering the metal as being formed of only one type of mixed crystals: — that is, that type of sohd solutions in which the iron acts as solvent. Some other steels present exceptions to this statement, since homologous solid solutions appear throughout the mushy stage only when the velocity of cooUng during that range is sufficiently low. Long stay at high temperatures permits diffusion to equalize somewhat the variations in composition in successive layers of the same crystals throughout the mass of the metal, thus preventing the building up of rich mother hquors from which are born anisomorphic phases. It should be recalled that the presence of a difference in con- centration, edge to center, existing in a single crystal is due to the difference between the solute concentrations of a solid solution and of a mother fiquor in contact with it, reciprocally in equihbrium, differences which vary with the successive temperatures assumed by the mass. In practice, diffusion is a powerful factor in by far the great majority of instances, and is especially potent in the solidification of soft and SOFT AND MEDIUM STEELS 5 medium machinery steels, to which grades the present study is strictly limited. 5. Immediately after the steel freezes it is generally composed of a solid solution of the various constituent elements in 7 iron. When the metal — now totally solidified — cools further until it reaches a suf- ficiently low temperature, it passes successively through the critical temperatures where 7 iron is transformed into /5 iron, and /3 iron into a iron. When the intervals of metastable equiUbrium existing at the lower region of each transformation range do not extend to temperatures so low as to cause the allotropic transformation of iron to take place at a velocity so small that it cannot be measured, such transformations actually occur at temperatures which depend upon the concentrations of the individual elements in sohd solution in the iron, and also upon the velocity of cooling. Now, among the various elements found ordinarily in soft and medium steels used in mechanical construction, even when considering the local concentrations which segregation may produce, only carbon possesses the characteristic that its solubility in solid iron or the solubility of its compounds with iron or with the other elements suffers marked variations because of the allotropic transformations of the iron itself — variations so noticeable as to give rise to phenomena of such importance as to make it necessary to take them into consideration. Consequently, the structural characteristics of such steels when cooled to ordinary temperatures depend upon the combined effects of two superposed orders of phenomena; namely, primary separation of mixed crystals of one series and a later segregation of carbides within entirely solidified metal caused by allotropic transformations of the iron. More complicated series of phenomena take place during cooling in those complex alloy steels in which other elements or compounds besides carbon and carbides are precipitated from saturated solid solution diu-ing allotropic transformations of the iron. Such instances occur in some hard steels such as chromium- and tungsten-steel (in which free double carbides segregate), and in some high nickel- or high silicon- steels; but they never appear in the ordinary mild or medium hard steels used for mechanical construction. This is the reason why it is highly desirable to limit the scope of this essentially practical study of preliminary heat treatment to low and medium carbon steels. In fact, if we wished to include in such a study the other principal types of steels, it would be necessary to introduce very compUcated theoretical considerations which yet lack for the most part any strong scientific foundation. On the other hand, such an extended study would not be of great commercial impor- tance, since the other types of steel usually are those to which the true pre- liminary heat treatments (above all, quenching for homogeneity) are 6 HEAT TREATMENT OF almost never applied in practice; nor, if applied, would influence the properties of the metal to an extent comparable to that which the same treatments cause upon machine steels containing, say, less than 0.60 per cent, carbon. The thoughts which I shall expound in the following pages will there- fore apply only to this last grade of steels. Here it is opportune to point out that in the extensive groups desig- nated in metallurgy by the terms "dead soft," "soft," "medium" and "medium-hard" steels are also included steels appHed to purposes very much different than for mechanical construction, but which, on account of their properties, are akin to those grades. As an example it is enough to remember that from a metallurgical standpoint we must include in the groups under discussion various types of steels used in the manufac- ture of projectiles, armor plates, etc. 6. Having thus briefly defined the tj^e of crystalline structure pos- sessed by the steels to which I intend to limit myself, it is possible to give in advance an outline of the fundamental characteristics of the heat treatments to be examined; that is to say, those heat treatments which result in permanent modifications in the constitution of crystalline structure and its chemical nature (see Sec. 3). And from what has already been said it is clear that these modifications can always be classi- fied as one of the two following fundamental types: 1. Modifications obtained when subjecting the metal to temperatures higher than that of its last critical point on heating. Inasmuch as at such temperatures our steels consist of mixed crystals of only one series, it is evident that the only phenomena active can be nothing else but processes of diffusion of the various elements in solid solution in the iron, from the regions in which their concentrations are greater toward the surroundings where they are lower. This diffusion may be limited in sphere to one crystalline element, or may extend beyond to various parts of one or more crystalline colonies. 2. Modifications obtained by subjecting the metal to heat treatment in an interval of temperature within which falls one or more of its critical points. Such modifications consist essentially in changes of the state of segregation of the various phases which form in totally solidified metal as a consequence of allotropic transformation. It is known that the character of the alterations of the second group is strictly dependent upon local variations in concentration of the nu- merous elements in solid solution in the iron. Therefore, modifications of the first classification to which the steel might have been subjected beforehand have a very strong action upon those of the second category. 7. From this brief summary it should be clear and definite that the operations I intend to study belong to those variously designated in metallurgical parlance as "preliminary heat treatments," "homogeneity SOFT AND MEDIUM STEELS 7 treatments" or "normalizing treatments" to distinguish them from the "final heat treatments" or "quality treatments," whose purpose is essentially different. It is not easy to indicate briefly the characters which distinguish the treatments belonging to one group from those of the other. Perhaps one can say that normahzing heat treatments are those whose principal purpose is to modify the distribution of various elements which appear in solid solution, and also as a consequence to modify the distribution of the soHd phases resulting from the decomposition of non-homogeneous mixed crystals at critical temperatures. The final state of aggregation of the equalized phases is in general immaterial to the successful comple- tion of a treatment for homogeneity. In other words, even though the process of diffusion takes place through solid solutions which are meta- stable under the ordinary conditions of use for the manufactured article, yet the preliminary heat treatment may produce homogeneous metal existing either as austenite, martensite, troostite, or any of the other metarals to be found in the complete scale from the unaltered solid solu- tion to the conglomerate finally produced by the complete separation of the various phases to which the solid solution itself gives place, and which are stable under the normal temperature of usage of the treated materials. The determination of this exact position in the transformation series, on the contrary, is the principal purpose of the final heat treatments. Consequently such operations are intended to bring the metal into the form of a properly selected metaral (or metallic entity) chosen in relation to the mechanical properties desired in the treated metal. In the case of soft and medium steels for machine construction the fundamental problem of the complete heat treatment is always the proper selection of the preliminary heat treatment. Contrary to the state of affairs for various other classes of steel, in these steels the final heat treatment usually does not present any special difficulties. In addition to this, the results which can finally be obtained depend largely upon the results of the preliminary treatments. A minute investigation of nor- malizing practice wiU not, therefore, seem inopportune. In the discussion of these matters, in order to be brief and to the point, I will presume that the reader is acquainted with the general theoretical laws concerning the formation and transformation of the solid solutions in metallic alloys. I may be obliged at times to mention these funda- mentals briefly, but only in order to present them under a form likely to simplify and clarify the conclusions which I shall deduce from them in relation to the special phenomena I propose to investigate. Finally, it is now opportune to mention the fact that the purposes of this treatise are essentially of a practical nature. In fact I propose to explain, as completely as the meager scientific information available in this field permits, the mechanism of the physical and chemical phenomena 8 HEAT TREATMENT OF SOFT AND MEDIUM STEELS upon which the thermal processes used in practice are based, and to elaborate the subject in such a way as to make possible the rational application of these processes to a certain number of cases and also their eventual extension to cover special cases arising from time to time. Once these purposes have been fixed, with a disclaimer of any inten- tion to expound theories which shall not be immediately' and directly apphed to sunplify and clarify the rationale of a technical process, it is clear that any considerations of a purely scientific character upon which my arguments must be based must be necessarily reduced to their siin- plest or diagrammatic representation. I must also omit all those rigid analyses and corollaries which should not be ignored in a complete and strictly theoretical treatise. Not finding an immediate and evident ap- plication to this study of technical operations, such exhaustiveness would only obscure the understanding of the nature of those phenomena. Naturally I must admit that such a method, neglecting that which remains outside the small circle of a given group of phenomena, would be absolutely unacceptable in a treatise whose principal aim would be the study of the phenomena as such, the more so in the present inquiry, where accurate and complete theoretical knowledge is still very scarce — so scarce that eUmination of any portion of that small stock can never be made without the chance of unconsciously altering the underlying principles even in their essentials. It is necessary to say this in order to be absolutely clear as to the extent to which the conclusions I propose to develop may be generalized. However, due to the practical end in view, such necessary restrictions or limitations cannot present any great inconvenience. Always having care that well-proved facts are used as a basis, any eventual modification found necessary on account of a new weight assumed by some of the elements at first considered unimportant should not alter in any way the actual relations established by current experimental observations and with the aid of this preliminary schematic but somewhat non- rigorous solution of the complex interrelations between the phenomena comprising technical processes. This is more than sufl&cient for the special practical ends I am aiming to reach. Besides, whenever the student desires to dig deeper into the various scientific aspects of the subject, and to construct an exact balance between the sum of all the known experimental facts and the approxima- tion to which I have arrived in their schematic interpretation, thus getting a precise notion of the real value of the latter, he can approach his task by reading the numerous existing treatises of scientific character, in which various features of the problem are thoroughly and rigorously studied, tak- ing into account all available experimental data. To aid this inquiry a list of the most important books will be found at the end of the present study from which it is possible to gather this complementary information. CHAPTER II PRIMARY CRYSTALLIZATION OF A BINARY SOLUTION UNDER COMPLETE EQUILIBRIUM 8. As with all other liquids, solidification of molten steel of any composition starts at certain points, named "centers of crystallization'' or "nuclei." Such tiny beginnings of the sohd solution, in equilibrium with the melt at that temperature, contain a materially less concentration of the alloying elements present. Step by step as the cooling of the metalUc mass proceeds, new layers of crystalline solid, each of slightly different composition, form around the primitive germs constituting the solid phase. At the same time new centers of crystalhzation originate in the residual Hquid. When solidification of tne metallic mass is complete, the final arrange- ment of the crystaUine structure is essentially determined by two factors: 1. By the position which the first crystallization centers assume and their influence upon the position of the subsequently forming centers. The spacing of nuclei largely governs the character of the crystalline colonies which constitute the steel and therefore controls the macro- structure of the metal. 2. By the laws governing the growth of the crystals as the tem- perature falls. These laws not only limit the boundaries of the result- ing crystals, but more especially do they control the chemical composition of the successive crystalline layers formed around each pre-existent solid particle whose surface was in equilibrium with the residual liquid at a shghtly higher temperature. It is clear that the nature of the individual allotriomorphic crystals, and therefore the microstructure of the metal, depends primarily upon these laws of growth. Inasmuch as the effects of the phenomena of the first group depend, at least in part, upon the laws which regulate those of the second group, it is logical to start by studying the latter. 9. I shall begin, therefore, by analyzing the happenings at only one center of crystallization. This can always be done by assuming the existence of an impenetrable wall around the nucleus under con- sideration. A quantity of the metallic mass is thus isolated small enough to exclude the presence of all other centers of crystalhzation, either those centers forming at the start of crystalliztaion or those originating afterwards during the mushy stage. It is evident that 9 10 HEAT TREATMENT OF this procedure corresponds to considering an elementary portion of the totally solidified metal which is occupied exclusively by the solid layers which will have accumulated around the germ of crystallization formed at its approximate center. In this way one can isolate and study a "crystalhne individual." I shall use this expression only in a chemical sense, and limit its meaning to the variation in composition in the various parts of the tiny mass considered, and exclude any reference to its crystallographic character. As a matter of fact it is known that usually such "chemical crystalline individual" does not correspond to a crys- talline element or crystallite considered from a crystallographic point of view. In order to represent these phenomena graphically, I shall use the familiar diagrams, where concentration is plotted against temperatures in rectangular Cartesian coordinates. As usual, temperatures are taken as ordinates along the Y axis (Fig. 1), and the concentrations of a given element X are plotted as abscissse. For a beginning, the effects of changing the concentration of any one of the various constituent elements in steel will be considered. In general, any one of these, such as carbon, manganese, sulphur, phos- phorus, silicon, nickel, chromium, or others, may enter as a solid solution to a certain extent in commercially pure 7 iron, and a discussion couched in general terms would apply with equal force to any of them. Limiting ourselves to a binary system will greatly facilitate the pre- hminary diagrammatic representation of the course of crystallization, and in view of the before-mentioned uniformity and continuity of the mixed crystals constituting the metallic mass even after its solidification, this limitation will not jeopardize in any way the preliminary "qualita- tive" conclusions which may be deduced. "Quantitative" data concern- ing the phenomena of solidification and transformation cannot well be synthesized by the same sunple method, because data for each one of the elements entering in the solid solution are modified numerically by the presence of each of the other elements. It would also be practically im- possible to construct a diagram or model which would represent simul- taneously the effects of all the elements entering into the constitution of the solid solutions under study. However, basing an argument upon the quahtative data obtained from the diagram showing the behavior of an isolated particle of the solid solutions under study, and taking into ac- count the results of many direct experimental studies, both chemical and metallographic, of the various alloys used in practice, it is possible to reach even quantitative conclusions exact enough for the great ma- jority of industrial purposes. 10. Let X be any one of the elements which form the mixed crystals constituting the steel under study, and let the concentrations of X be plotted as abscissse in the diagram. Fig. 1. Since 100 parts represent SOFT AND MEDIUM STEELS 11 the entire alloy, any abscissa x will represent a certain steel having x per cent, of element X. It is well known that for each concentration of the element X lying within the range of analyses which are completely soluble in iron either in the solid or in the liquid state, there are two distinct critical tempera- tures related to solidificational processes. Of these, the first and higher corresponds to the first separation of mixed crystals from the molten mass; while the second and lower corresponds to the end of the crystallization, where the system is completely solidified. Normally, the temperature of the beginning of sohdification cannot be lowered by a rapid chiUing of the Uquid metal; on the other hand, the temperature of complete solidi- FiG. 1^ — Solidification of complex solid solution. fication is greatly influenced by the speed at which the immediately pre- ceding cooling has taken place. •In Fig. 1, the point U represents the temperature of beginning of solidification of an alloy of iron and a constant amount of the other ele- ments in their usual proportions with the exception of element X. For this alloy, the concentration of X is evidently zero, and its thermal events would be plotted on X = 0, or the Y axis. U' is the temperature of end of solidification after very slow cooling for the same alloy of iron with X = 0. Another steel containing a material amount of X has a corre- sponding temperature for the beginning of solidification, which can be plotted on the temperature : concentration diagram. The curve joining 12 HEAT TREATMENT OF all such points of incipient solidification is UV, known to slope downward in case the solid solution separating out has a lower concentration in X than that in the molten mass. Since all the elements which enter into the composition of low carbon machine steels form soKd solutions in iron with the just mentioned characteristic, this investigation can be restricted to the case where curve UV slopes downward with increase in the con- centration of X in the alloy. All the points in the same diagram representing the temperature of the end of crystallization of the various alloys with increasing concentra- tions of the element X form a curve U'Z located entirely underneath the first curve UV. As already observed, the point U' corresponds to the temperature of completed solidification of the alloy of iron containing a constant amount of the other elements in their usual proportions with the exception of the element X, which has a concentration of zero. Again it is true that this second curve has a downward slope when the concentra- tion of the element X increases. The position of the curve U'Z, as shall be seen, changes noticeably with variation in the velocity of cooling during crystallization of the metallic mass. We shall also see directly that these perturbations in the position of line U'Z are due to variations in the extent of the diffusion of each component throughout the mixed crystals and between them and the molten mass. Now, these variations in cooling velocity cast a simulta- neous reflection in the amount of diffusion for all the elements alloyed with the iron as well as the element X. The same speed of cooling which even in the absence of the element X would cause the point U' to change a certain amount adds its influence to the displacement of U' caused by the element X, thus modifying the latter considerably. For sake of simplicity we will not consider these last variations for the time being. This we can do, for example, by limiting the present dis- cussion to the system in which only one element, X, is alloyed to piire iron. It is evident that in such case the two points U and U' will coincide in one, that one being the point of crystallization (and of fusion) of pure iron. This case is sketched in Fig. 2, and to this we shall now confine our attention. 11. Although, as already mentioned, the reader is considered to be familiar with the mechanism of the formation of solid solutions and of the subsequent segregation of the secondary constituents from them, never- theless now I deem it advisable to make an exception to the plan of the book in presenting this first equilibrium diagram. It is the fundamental basis for the arguments which I propose to develop later. It should also be developed or synthesized in a way which will simpUfy and clarify further applications of these general principles to the various practical cases. In Fig. 2 therefore, let us consider an alloy in which the element X SOFT AND MEDIUM STEELS 13 has a concentration D, and let us suppose it to have been heated to such a temperature that it is completely melted. D' then will be the point which represents the conditions as to temperature and composition ex- isting throughout this molten mass. It is well known that if the metal is allowed to cool until it reaches the temperature t' — corresponding to the intersection of the vertical DD' with the curve UV — ^mixed crystals start to separate from the liquid; in these crystals the concentration of the element X is that represented by the point F. For the reasons outlined above, F is situated at the left Fig. 2. — Solidification of binary solid solution. of the vertical DD' in all the systems which the alloys comprising the subject of our study can present. Solidification must start somewhere, and it is known that the first nuclei appear at determined points in the cooling melt. As I have al- ready mentioned, the formation of these first germs of crystallization is one of the most important phenomena among the many occurring during solidification. The properties characteristic of the solid alloy depend •to a very great degree upon the number of centers of crystallization and the time of their formation. So strong, in fact, does this state of affairs 14 HEAT TREATMENT OF impress itself upon the metal that vestiges remain even after all the heat treatment and mechanical work to which it will be afterwards subjected. I shall consider later these phenomena, which are dependent upon the sequence, propinquity and uniformity of distribution of the centers of crystallization, especially as influenced by the processes of under- cooling. Only the process of crystallization around a single nucleus will now be studied, which is always theoretically possible, provided the mind is attuned to contemplate a metallic mass sufficiently small in dimension. We shall see later that, although it is not possible to cool a piece of metal in the laboratory so that only one crystal results, the conclusions which may be drawn from a theoretical consideration of this limiting simple case can be easily extended to the more complex practical cases, in which the metallic mass under observation actually crystallizes around a large number of centers. 12. As soon as the temperature of the metallic mass falls a very little below t', reaching the value t", the separation of a certain quantity of mixed crystals F will cause an increase in the concentration of the element X in the residual liquid, because the solid phase contains a smaller pro- portion of the element X than that contained in the liquid E. Let us call E' the new concentration of X in the residual Uquid. Liquid E' is in equilibrium with solid of composition F' (greater than F) at tempera- ture t". In other words, the first solid to separate from the molten liquid when the temperature falls the least bit below t" will have an analy- sis in X equal to an abscissa t"F', richer in X than the nucleus about which it must precipitate. From the definition of the curve UV, it is clear that the point E' will lie upon it. If now we consider the conditions of our system containing homo- geneous liquid and heterogeneous crystal after it has reached temperature t" and take due account of the fact that the cooUng process in itself is essentially continuous, we easily see that this system cannnot be con- sidered as being in equilibrium from the standpoint of the phase rule. In fact, should no other reaction occur the crystallite forming around the nucleus F will consist of a series of shells of differential thickness, in each of which, as you pass from the center outward, the concentration of X would vary from that value corresponding to the point F to that value corresponding to the point F', showing a scale of all possible inter- mediate values between these two points. Diffusion must take place in such a compound crystaUite to allow an approach to a state of true chemi- cal equihbrium. In fact, diffusion of element X from a region of high concentration to one of low concentration does take place within the crystal of sohd solution and between this and the liquid masses in con- tact with its surface, in a manner similar to those movements which- render a Uquid solution homogeneous, although with lesser velocity and SOFT AND MEDIUM STEELS 15 intensity. It is known that the substances contained in a solution (either soUd or liquid) thus have a tendency to move about and equalize their concentration in all parts of a continuous volume. It is also known that such a process — which theoretically could not end in a finite lapse of time — actually takes place in liquid solutions with such velocity as to produce substantially complete uniformity of concentration of substances dissolved in very large quantity of solvent in a relatively short period of time, and this although the initial concentrations vary greatly in regions of liquid quite remote one from the other. The hnear velocity of diffu- sion in liquid solutions increases very rapidly with increase in tempera- ture, until at very hi^h temperatures it reaches a tremendous speed — as an instance, take the case of ferro-alloys shovelled through one door of an open-hearth furnace, but which seemingly spread throughout the entire bath very rapidly. As has already been mentioned, the velocity of diffusion in solid solu- tions, although increasing very rapidly as the temperature increases, is much lower than in the case of a liquid solution. Consequently in an- nealing practice the effects are noticeable only between regions separated by very short distances^ — ^usually by a few tenths of a milHmeter and but very seldom as great as a few millimeters. This is true even at high temperatures, such as for instance 800°C. to 1100°C. (1470°F. to 2010°r.) maintained for quite a long time, say 5 to 10 hours. Nevertheless we shall see later that even in solid solu- tions the processes of diffusion cause effects of the greatest industrial importance. 13. Returning to the thread of the argument, let us maintain our system at temperature t" for such a time that the composition of the sohd particle is able to become practically imiform at all points. This condition wiU ensue as an effect of diffusion of the element X from the higher concentration periphery to the lower concentration center. A somewhat more rigorous examination of this process would easily demonstrate that such equahzation is necessarily accompanied by a gradual diffusion of the element X from the liquid where its concentra- tion is E' into the external strata of the sohd portion, where the con- centration is being withdrawn to a figure slightly less than the abscissa t"F'. Only when the process of diffusion has proceeded substantially to completion wiU we possess a system in the state of chemical equihbrium at the temperature t". This system will be formed by a soUd crystallite of uniform composition F' and of mass if .' contained within a residual liquid of composition E' and of mass M]. Now if a mass of solid M[ has a concentration in X equal to the value fW', the total amount of X in the solid material is the mass times the concentration, or Ml X t"F'. Similarly the mass of hquid existing 16 HEAT TREATMENT OF at the same moment is M'i containing an X-concentration rep resen ted by t"E'; consequently the amount of X in the liquid is M', X t"E'. But the total mass of the system Mt = Ml + M'i. Since the average X- concentration of the whole system is t"R', the total amount of X existing equals Mt X t"R'. This is also represented by the sum of the X in solid plus X in liquid. In symbols Mt X iTR' = m; X ¥¥'+ m'i X ¥W' Substituting for Mt its value ilf j + M'i, multiplying and collecting. M; ITE' - iTR' M; WW Mi~ WW -¥W ""^ M ; ~ ~WF' ^' The same phenomena analyzed above will proceed upon further cool- ing the system below temperature t". Suppose that the cooling, although continuous, is slow enough so that complete uniformity of the concentra- tions of the element X is established at all times in each of the liquid and crystalline phases. Then when the temperature has fallen to lower tem- perature t'" our system will be formed of a quantity M" of mixed crystals of uniform X-concentration represented by the point Q, and of a quantity M"i of a residual hquid, in which the concentration of the element X will have reached the value represented by the point S. By symmetry it is known that the relation between the quantities M'\ and M"i and the length of the two segments QR" and W'S is as follows when complete equiHbrium exists : M^^ ^ W ,^. M'\ R"S ^ ' or (if Mt represents the total mass of the system) M"i QR" Mt - M"i QS - QR" (3) If the cooling is continued further under the same conditions, it is evident that crystallization must stop as soon as the quantity Mi of the residual liquid is reduced to zero. As can be readily seen from rela- tion (.3), this will happen when the length of the segment corresponding to QR" is reduced to zero — namely, at the point G. G is situated at the point of intersection between DD' and curve UZ. UG is the locus of all points such as F, F', Q and G, representative of the concentrations of the element X in the mixed crystals at the various temperatures t', i", <"' and t^^ which, at those temperatures, are in equilibrium with the liquid phases whose compositions are represented by the conjugate points of the curve UV (E, E', S and H, respectively). It has thus been established that the intersection G represents the temperature and composition of a mixed crystal of equalized X concentra- SOFT AND MEDIUM STEELS 17 tion at the instant when it has absorbed the last trace of residual mother liquor in equilibrium with it. The composition of the latter was repre- sented by the point H. Postulating a continuous progress of the cooling and crystallization, it is clear that the locus of all points analogous to the points F, F' Q and G is a curve UZ. This curve shall be hereafter given its usual name, the "solidus." It coincides in the special case of continual diffusion to exact equilibrium with the curve representing the actual end of crystallization. The curve UV takes the name of "liquidus." It is the locus of all points representing the composition of the mother liquors in equilibrium with the mixed crystals whose com- position is represented by abscissal intersections with the curve UZ. In the special case of complete equilibrium now under consideration, the liquidus coincides with the curve of the beginning of crystallization. 14. The validity of the conclusions which have been arrived at through the preceding pages is clearly dependent upon two principal conditions : (a) It must be possible for the element X to enter into true solution with the iron both in the solid or liquid state up to all concentrations of that element found in the alloys here under consideration. Solubility must be possible in the mixed crystals as well as in the mother liquor in equiUbrium with them — which, by the way, must carry a much higher percentage of element X. (In other words: None of the illustrative points which have been considered in Fig. 2 may encounter a region where true solution cannot take place during all the process of crystallization. As I have already said, this condition actually obtains in all the chemical systems to which this study refers, for all the higher intervals of temperature to which we have now restricted our investigation. Therefore I shall re- serve consideration of the occurrences when the temperature restriction is lifted, to the pages where we shall study the transformations taking place at temperatures below the end of crystallization.) (b) At every instant during the solidification of the mixed crystals, the concentrations of the element X must be uniform in all regions of the crystalline mass then in existence, and uniform at all points of the residual liquid — although the concentration differs, naturally, between the solid and liquid phases reciprocally in contact and equilibrium. Now, this second condition — especially as to the uniform concentra- tion of X in the solid phase— never exists in the alloys which form the object of the present study, at least at the time and under normal con- ditions of, solidification. In fact, the cooling velocity of the metallic mass throughout the entire interval of crystallization is ordinarily far greater than the snail's pace which would be necessary in order to allow the diffusion of the element X to take place in such measure as to render the various layers of the growing crystalline colonies practically uniform 18 HEAT TREATMENT OF SOFT AND MEDIUM STEELS in concentration. Intracrystalline heterogeneity is always present on first cooling, as I have already had occasion to remark, even when the dimensions of the crystallites formed around each crystallization-germ are extremely small, for instance, a few hundredths of a millimeter. However, the opposite extreme never occurs in the cases we are now studying — the extreme where the processes of diffusion in the mixed crys- tals are practically stationary and zero. Disregarding, therefore, this second extreme as well, we shall now start to analyze the process of crystalUzation as it really happens in in- dustry; namely, those cases in which diffusion of the substances dissolved in the mixed crystals takes place to a moderate amount, yet at a velocity lower than that necessary to permit complete uniformity in composition of the cooled crystalline masses to be attained. CHAPTER III ACTUAL COURSE OF CRYSTALLIZATION OF A LIQUID BINARY SYSTEM 16. Assuming that the last-mentioned hypothesis represents the ac- tual facts, when the temperature of our system falls until it reaches t" (Fig. 2) a certain quantity of the element X will have diffused from the external layers of the crystalUne mass where its concentration is F' to- ward the central nucleus of initial concentration F. At the same time new quantities of X pass into the external layers of the crystal from the hquid. The mother liquor will necessarily contain a larger percentage of X than that which should correspond to the conditions of complete equihbrimn. This excess quantity of X is the same as the deficiency in the material surrounding the nucleus' F, i.e., the amount of X over (in the hquid) or under (in the sohd) that amount which should correspond to equiUbrium at this temperature. In reality the diffusion in the liquid mass and between the hquid and the external layers of the solid solution crystals takes place at a much greater velocity than the process of diffusion between parts of the solid caused by the difference of concentration. Consequently for normal velocities of cooling the conjugate points determined by the same ab- scissa intersecting the Uquidus curve UV and sohdus UZ represent in reality and with a fair approximation the concentrations of the element X in the residual liquid and in the periphery of the mixed crystals in equilibrium with said liquid respectively. In consequence, it is clear that once the system has reached the tem- perature t" it will be composed of a residual liquid in which the concen- tration of the element X will have the value E' and by a crystallized mass which will possess the composition corresponding to the point F' only in its external layers, while the underlying strata will contain gradually decreasing proportions of the element X layer to layer as one proceeds toward the center. , Here it reaches a composition represented by a point F'l (see Fig. 3, where, to make these phenomena more clear, this part of the diagram is reproduced to larger scale) situated between F' and F'o. It will evidently be the nearer to the position F'o the smaller has been the amount of diffusion of X in the solid solution. It is, therefore, evident that the average composition of the total sohd phase which has actually formed down to the temperature t", will be represented by a point F'i situated between F'l and F'. ' "Nucleus" is used here in the sense of being ahnost a mathematical point. 19 20 HEAT TREATMENT OF The ratio between the mass M'a, of the solid phase of average com- position F'ii which has actually separated at the temperature t", and the mass M'21 of the residual liquid which is in equilibrium with it at the same temperature, is expressed in the following proportion similar to equation ( 1) (Sec. 13) : R'E' R'E' , (4) M'n F'iR' B'F' + F\F' By comparing (4) with (1) it seems that the total mass of the solid phase actually separated at the temperature t" is smaller than that which Fig. 3. — Crystallization of liquid binary system. would have separated if diffusion took place in it sufficient to make the concentration of X totally homogeneous. 16. Contiauing the cooling of the system, for each temperature two points may be determined corresponding respectively to F'l and F'i. From the definition of these two points as regards temperature t" it is clear that the locus of all points corresponding to F'^ will be a curve {F F'2 F"^ pa.ssing through F and lying totally under and at the left of the solidus. This curve represents the average composition of the solid phase at the various temperatures, and is marked ^ in Fig. 3. The locus of all the points corresponding to F'l will be a curve F' F'\ F"i also passing through F, and lying totally under and at the left side of SOFT AND MEDIUM STEELS 21 the preceding curve. This curve represents the minimum X concentra- tions in the solid phase at the various temperatures, or the X-concen- trations at the crystaUine nuclei at the various temperatures. This curve in Fig. 3 is indicated with a. As shown by equation (4) of Sec. 15, it is clear that the reacting mass will be totally soUdified at the moment it reaches the temperature t^^, or at F"2 where curve j3 intersects vertical EG representing the average analysis of the system. At temperature t^^ our crystalMne element will evidently have an average composition represented by the point F''^. At its nucleus it will possess the composition represented by the point i^"i, corresponding to the intersection between the abscissa through F"2 with curve a. Its external composition, however, will be that represented by the point A, determined by the intersection of the solidus {U F G A) with the hori- zontal line drawn through F''^. Afterwards when the temperature of the system continues to fall below the temperature of total solidification, t^^, diffusion of X between various parts of the crystalUne element does not stop suddenly, but its effect continues to be felt, although with a rapidly decreasing in- tensity as the temperature of the system lowers. Consequently the concentration of the element X decreases at the periphery and increases at the center of the crystal. This fact, in our diagram, corresponds to a gradual converging of' the points corresponding to F"i and to A toward the vertical EG as the temperature of the system falls. Therefore, starting from i'^, the temperature of the end of the crystallization, the diagram will have two descending curves— originating respectively at F"i and at A— marking out a band whose borders delimit the compositions of the solid phase at the edge and center of a single crystal. The width of this band (and the heterogeneity) decreases .with falling temperature. 17. We could have reached exactly the same conclusion starting from the following observation: Due to the fact that diffusion of X in the solid phase does not take place rapidly enough so that it could be considered as homogeneous at each instant, it follows that at a given stage in the freezing— for example, at temperature t"—the crystal previously formed contains a nucleus in which the concentration of element X is less than that which corresponds to complete equiUbrium in the system. From the point of view of the equiUbrium between the sohd and the hquid phase, this is equivalent to a condition where a certain quantity of the element X has been added to those parts of the system which are actively engaged in the process of crystaUization, namely, added to the metal forming the residual liquid and the outer surface layer of the crystal. Therefore, at the tempgrature t", the phenomena of crystaUization are actually 22 HEAT TREATMENT OF following such a course as though the chemical composition of the system under transformation were represented by a point R" which may be called the fictitious composition of our system, situated of course to the right of the point B'. It is clear that the fictitious composition will constantly move to the right all the time the temperature is falling, occupying at temperature t'", for instance, the position R'". Thus, it will follow a curve E R" R'", cutting the solidus UZ at the point A, the de- finition of which we had arrived at following another path. A is situated below G, the point corresponding to the end of crystaUization of the system in complete equihbriam. From the above explanation it follows directly that the points R", R'", etc., will be moved further towards the right, and therefore the point A will be also pushed further downwards with decreasing importance of the process of diffiision in the solid phase; in other words, the more rapid the cooling in the given system. From the meaning of the curve VZ, it is clear that at temperature f^ , corrresponding to the point A, the metallic mass will be completely solidified. Since F"i represents also in this case the composition reached by the center of the crystalline element at the instant in which solidification of the mass ends, we find again the same band of variant composition in the sohd field, bordered by two lines descending from Fi" and A, exactly as has already been defined from another premise. 18. Let us see now what is the meaning of the points in the diagram, contained within this band. I will refer again to Fig. 2, page 13, where a larger heterogeneous area is represented (F"i A F"'i A„). Let us consider the system at a temperature U, lower than the actual end of solidification {f). Remembering the expUcit limitation to the case in which the entire metallic mass has solidified around only one crystallization-germ, and thus is formed of only one crystalline element, we know by definition that the concentration of the element X at the nucleus of the crystalline mass will be represented by the point F"'i, while the entire surface of the same mass will have an X- content corresponding to the point A„. The successive layers of crystalline matter which build up around the nucleus during the successive steps in solidification evidently possess compositions in X represented by successive points along the line F"'iAo starting from the point F"'i. In otJier words, the successive points between F"'i and Ao represent the concentrations of the element X at similar points on any radius vec- tor, ranging from the center to the surface of the crystalline element. Inasmuch as the concentration of a given element in the various points of a given sohd is not, generally, a linear function of the distance which separates these points, it is impossible to say off-hand that the composi- tion at to of a point say half-way from center to superficies in the crystal is equal to the arithmetic mean of the abscissae F"'i and Ao. SOFT AND MEDIUM STEELS 23 19. Now, let us again take the same system at a temperature t, at which it is totally liquefied, and let us suppose that it cools more slowly than it did previously. In this case it is clear that the phenomena of diffusion will have oppor- tunity to proceed more nearly to copapletion during each given tempera- ture interval owing to the added time during which the system remains in it. Thus, for example, at the temperature ti the concentration of the element X at the nucleus of the solid phase will be enriched to a value 7', higher than the value 7 which it had reached at the same temperature in the previous case. At the same time — and for the same reasons — ^the average X-concentration of the solid phase will reach a value 5', larger than the previous value 6. Repeating the same analysis for successively lower temperatures, the conclusion is reached that in this slower cooling the two curves 7' and 5' will be^ displaced to the right of the respective positions (a and |3) which they occupied during quicker cooling. 7', as a, represents X- concentration at the nucleus, and 5' and j3 both denote the average com- position of the solid phase with falling temperature. As a consequence of slower cooling the temperature of the end of crystallization (^determined by the intersection of curve 5' with the ver- tical DD') will be higher than in the preceding case. At this temperature, the difference in composition between the nucleus (F^) and the peri- phery (A') will be smaller than before. Should further cooling be con- tinued at a slower velocity than before, the width of the band of variant concentrations in the soUd phase will narrow more rapidly. (See Sec. 18.) Summing up, we can establish the following propositions which regu- late the solidification of any given system capable of crystallizing in only one continuous series of mixed crystals: first. — For each given velocity of cooling the temperature of the end of crystallization has a determined value. Other conditions being equal, this is lower the faster the cooling. Second.— For each determined velocity of cooling the curve showing the composition of the crystalline nuclei and the curve showing the average composition of the solid phase assume two distinct positions. Other conditions being equal, the two curves are moved further to the left and downward, diverging more and more from the true solidus with the more rapid cooling. In the extreme case in which the coohng is so rapid (by drastic quench- ing) as to entirely prevent diffusion during solidification it is clear that the curve representing the composition of the nuclei becomes a vertical line. At the opposite extreme, where coohng is so slow that at each step the process of diffusion produces complete uniformity of concentration in 24 HEAT TREATMENT OF the crystalline mass, it is clear that the two curves under discussion will both coincide with the solidus. Third. — As a consequence of the above-indicated displacements of the two curves y and 5, the band representing the heterogeneity existing edge to center in the mixed crystals with slower cooling, is narrower at its top (that is, at the temperature of complete solidification) and con- tinually becomes narrower at lower temperatiu-es. 20. We will see later what great importance the width of this band has upon the structure and properties of steel, so that any treatment capable of producing modifications in it is capable of practical appHcations. However, it is even now easy to see from the foregoing brief analysis of the circumstances to which it is due, that any process capable of modi- fying the characteristics of the strip in question must be based upon a rational utilization of the phenomena of diffusion as they affect the va- rious elements entering into the formation of a given type of sohd solu- tion. In fact, such is the case. One or more fundamental principles are utihzed whose tendency is exclusively to modify the manner of diffu- sion — usually in the sense of rendering it more rapid and more nearly complete. All additional principles in the treatment are used to modify this primary action, by removal of factors which may halt or reverse its course. The criteria upon which these fundamental treatments are based can be grouped under two main heads : 1. Selection of such a chemical composition for the steel as will allow — other conditions remaining equal — the maximum velocity of dif- fusion for its elements in the mixed crystals. 2. Selection of a temperature at which the velocity of diffusion is at a maxinium, and maintaining the steel at that heat for a sufficient time. The second consideration is limited by the fact that a prolonged in- tense reheating might cause the metal to deteriorate for one or several reasons. The first criterion (of chemical composition) may influence diffusion not so much directly — in the sense of increasing the velocity — as indirectly by decreasing the distance over which equalization must be established. In this way the conditions under which the process of diffusion must operate are modified by selecting that chemical composition in which the number of crystaUization centers are maximum, and theMore the distances separating points of highest and lowest concentrations are minimum. 21. I will later have occasion to speak of the criteria of chemical com- position. Instead it is now opportune to investigate more closely the processes of the second group, which as noted have the inamediate purpose to heat the steel to a temperature at which the processes of diffusion take -SOFT AND MEDIUM STEELS 25 place with the greatest velocity, other conditions being equal; and second, to keep it at that temperature for a sufficient length of time to reduce the width of the band of variant concentration to an amount which experi- ence has shown to be sufficient to impart the desired physical properties for a particular case, or at least, to render it susceptible to acquire these properties by means of further treatments. As is well-known, these reheatings are widely used in metallurgical practice upon raw steel (as cast) as well as upon steel which has already undergone hot work such as forging, drop forging, etc. These constitute Fia. 4. — Effect of normalizing. the foundation of the so-called preliminary heat treatments for homo- geneity, or more briefly, normalizing. We will begin as we have been doing in the preceding pages by study- ing the course and the effect of heating for homogeneity on a metal gran- ule solidified around but one crystallization-germ, tracing the phenomena of dilifusion of only one of the elements (which we will again call X) to be found in solid solution in y iron. Let us suppose that the conditions under which the crystallization of such a liquid mass of composition D has taken place, and that the further cooling of the solid solution down to ordinary temperature to has been such, as to give place to the band of heterogeneity marked ABCE (Fig. 4). 26 HEAT TREATMENT OF Diffusion continues to take place in the solid mass even at ordinary temperatures, transporting element X from the points of maximum con- centration (E), to the points of minimum concentration (C). However, for the steels in which we are now interested, at atmospheric temperatures this occurs with a velocity so small as to produce no perceptible effects even after very long periods of time. Besides this, carbon, one of the most important elements in the machine steels, cannot be found any longer as a solid solution below the temperature corresponding to the last allotropic transformation of iron, except in conditions of metastable equihbrium. This fact further disturbs the course of diffusion for this element at low temperature. Leaving aside these considerations, suppose that immediately after being cooled to to, the temperature of the system be increased gradually. Diffusion of the various elements in solid solution again occurs at a velo- city which makes its results felt directly upon the properties of the steel at temperatures above approximately 800°C. only. At temperatures below this limit, it takes place in poticeable measure only across intervals of space extremely small. In this interval of temperature (whose lower limit is quite variable depending upon the composition of the steel, but can be es- timated for the ordinary analyses to be somewhere between 200°C. and 400°C.) the phenomena of intercrystalline diffusion only act indirectly by the segregation of metallic carbides contained in the steel. These last-mentioned carbide segregation, occurring at temperatures below the critical interval of the steel in question, takes place in limits of space far smaller than those which are directly affected by the true processes of diffusion through a solid solution — phenomena which we are now studying. Also the two are not connected in any way in practice. The study of carbide accumulation is therefore completely extraneous to our present investigation inasmuch as it is part of the "final" or "quality" heat treatments. 22. Let us inquire what happens in our system when after having passed from L to about 800°C. its temperature continues to increase gradually. It is clear that this first reheating has not caused the appearance of any new crystalline variety non-existent during cooling, as far as the physico- chemical properties of the solid solution in X are concerned. It also follows from the discussion immediately above, that the width of the band of heterogeneity on reheating to 800°C.will be but little smaller than the narrowest reached at the end of the first cooling. Heating at a uniform rate from 800° upward, the speed of diffusion continually increases with increasing temperature, so that the concentration, edge to center, of the crystallite will approach uniformity at an increasing rate. After having reached a certain temperature <2, A'B' representing the width of the band of heterogeneity, let the mass be maintained constantly SOFT AND MEDIUM STEELS 27 at that temperature for a given time t?. Diffusion will evidently continue, and the width of the band will decrease more rapidly the higher the temperature t^. Now suppose that at the end of the interval of time 5, the difference in X-concentration at the surface and center of the crystal be reduced to the width A"B". If the mass then cools, the phenomena of equalization will continue with appreciable though decreasing velocity until the tempera- ture reaches about 800°C., at which the width of the band will be reduced to A"'B"', an amount yet smaller than A"B". Continuing the cooling under 800°C., and excluding for the time being the intervention of allotropic transformation in elements, or of carbide, the width of the band does not appreciably decrease further. Hence, when the system again arrives at ordinary temperature, the heterogeneity will be reduced in value to abput the value C"E". It may then be concluded that by heating under the conditions indi- cated, it is possible to reduce the width of the band of variant composition; in other words, it is thus possible to decrease the heterogeneity of the system. 23. It is clear that the increase of homogeneity obtained in a solid solution by reheating to a given point will be greater the smaller are the distances separating the loci of the extreme concentrations (maximum and minimum), and the greater is the liriear velocity of diffusion of the element X. From what has already been said, the first condition depends directly upon the dimensions of the individual crystalline elements which build about each germ of crystallization. In fact, we have seen that in each crystallite of soHd solution the minimum concentration of an element is found at the nucleus, while its maximum concentration occurs at the periphery. Thence it can be deduced, all other conditions being equal in a deter- mined crystalline mass, that the gain in homogeneity induced by a given "reheating for diffusion" will be greater the larger has been the number of centers per unit of volume around which the crystalline elements have formed; or in other words, the greater has been the frequency of nuclei in the molten mass under observation. As I have mentioned, I shall have occasion to deal with these phe- nomena further on. In regard to the second property — the linear velocity of diffusion of a given element X in the solid solution in question — it is now opportune to observe that it depends essentially upon three conditions: {a) Upon the nature of the metals constituting the alloy. Upon this point, about which very uncertain data exist, I will have occasion to dwell later. (b) Upon the absolute difference in concentration of .element X at points separated a given distance. It has been thoroughly established 28 HEAT TREATMENT OF that the velocity of diffusion increases rapidly with the difference in con- centration according to known laws, but this is not the moment to enter into such details. It is enough to remember that Tick's law shows that the quantity of an element diffusing in given time through unit cube of differential dimensions is proportional to the difference of its concentra- tions at opposite faces. For routine appUcation to steels we are here interested in, it is suf- ficient to note in regard to condition (6) that a consequence of Fick's law is the fact that a given homogeneity reheating causes effects of great intensity only during the first steps of the treatment, when the materal subjected to it presents large discontinuity in concentration. When a certain point in the process of diffusion is reached, it is necessary to resort to other and much more drastic treatments to produce further noticeable improvements in properties. (c) Upon the temperature reached and maintained during reheating. In other words, the linear velocity of diffusion due to a given difference in concentration of an element in any solid solution increases very rapidly with the temperatures at which the phenomena takes place. This is a fact absolutely general in character. 24. While experimental researches have been made with various types of nonmetalUc solid solutions and for some metallic alloys (especi- ally those of mercury) in order to establish the relation connecting velocity of diffusion and temperature, to my knowledge analogous investigations with steel have not been made of such character as to permit the deduc- tion of general rules. Among all the elements in steels which are present as solid solutions in iron, only carbon has been studied with anything like precision from this point of view. Attention has been directed to carbon because the velocity of diffusion at various temperatures finds immediate industrial application in the case hardening process. In cementation, the diffusion of carbides (and especially cementite, FesC) which form true solid solutions with iron, produce effects to which are added the effects of the diffusion of gaseous carburated compounds, which follow special laws known to be very much different than those followed by the others. Moreover, in the great majority of industrial carburizing, gaseous diffusion produces effects far greater than the others, totally obscuring them in the end. These phenomena I have dwelt upon at length in another volume' and shall have occasion to touch in the following pages. In regard to the relations connecting temperature and velocity of diffusion of the other elements — such as manganese, silicon, nickel, or chromium, which with carbon frequently enter into the composition of steel — there are no experimental data available complete enough to permit general deductions to be drawn. 1 "Cementation of Iron and Steel." SOFT AND MEDIUM STEELS 29 To give an approximate idea of the influence of temperature on tfie velocity of diffusion in a metallic solid solution, I will mention briefly the results of some experimental determinations made by W. C. Roberts- Austen. A perfectly flat surface of an alloy of lead and gold containing 5 per cent, of the latter was compressed heavily against another perfectly true surface of a block of pure lead. Maintaining the system for 30 days constantly at the temperature of 165°C., Roberts-Austen found that a perceptible quantity of gold has passed from the gold-lead alloy placed below into the surface layers of the block of pure lead placed in contact above. He found, other conditions being equal, that even a small increase in temperatures at which the two blocks were maintained, corresponded to a very large increase in the quantity of the gold passing from the block underneath to the one above. Thus, taking as unity the velocity of diffusion of gold in lead at 100°C. under the experimental conditions mentioned above, this velocity reaches the values indicated in the second column, by increasing the temperature as indicated: Temperature Velocity of diffusion 100°C. 1 166°C. From 200 to 250 200°C. From 360 to 400 251°C. From 1150 to 1500 Increases in the velocity of diffusion of about the same order as those above tabulated, take place in a large number of metalHc solid solutions. It is also noteworthy as a nearly universal fact that the velocity increases very greatly the moment the temperature increases above a certain critical value. In the case of lead and gold this critical value Ues around 150°C. It is therefore clear that a heating for homogeneity is more effective the higher the temperature at which it is performed, and natu- rally, the longer this temperature is maintained. But, in practice, the attainment of the two indicated conditions designed to produce the maximum efficiency in a reheating for homo- geneity is subject to the following Umitations: It is desirable to reach as high a temperature as possible, yet a brief examination of the diagram shown in Fig. 4 is suflicient to define the necessary restrictions. In fact, if instead of arresting the increase of temperature at <2 in the first reheating considered in Sec. 22 it is continued beyond this limit, it is clear that a temperature ts will eventually be reached, which corresponds to the intersection of the solidus with the curve bordering the right hand side of the band of heterogeneity. At the instant at which temperature ta is reacned, fusion will start at those parts of the mixed crystals in which the concentration of element X is at its maximum, 30 HEAT TREATMENT OF that is to say, at the periphery of the individual crystallites. Surpassing this temperature even slightly, the crystalHne mass will be transformed to a conglomerate of little sohd grains each surrounded by a Hquid film of composition F. It is true that, when the temperature of the system is maintained at a value h + A«, a differential amount higher than tz, the band of variant composition applying to the mixed crystals decreases in width due to the process of diffusion, and this circumstance will in time cause the curve EB'B^^ to move sufficiently toward the left so that its intersection with the abscissa corresponding to the temperature h + A< will fall at the left of the solidus. As a consequence the system will again be totally solidified; but experience shows that in the great majority of cases in practice even the exceeding short time necessary for what may be called "re-soUdification" is amply sufficient to permit certain phenomena to transpire which generally modify profoundly the physical properties of the metallic alloy under treatment. Only rarely is the damage so slight as to have an inappreciable effect. It is not possible to describe in a general way the nature and the effects of these last phenomena. Further on, we shall see some practical examples of them. At the present time I will only mention that in many cases the damage results from processes of oxidation which take place with great rapidity in the partially molten mass under the action of the oxidising atmosphere in the furnace where heating takes place. In other cases the phenomena in question are due simply to changes in the oxygen tensions of liquid and crystalline phases in which the system is thus temporarily divided. These variations in equilibrium may at times cause the formation of new phases. Finally, sometimes these phenomena simply cause a different state of agglomeration in the non-metallic in- clusions contained in the alloy under treatment. In each case it is usually said that the metallic alloy which has suf- fered an alteration of the kind described has been "burned." It is easy to see from the examination of the diagram. Fig. 4, that, other conditions being equal, the highest temperature to which an alloy consisting of homologous solid solutions can be heated without burning varies greatly with the varying degree of homogeneity of the mixed crystals constituting it, and also, as a consequence of what has already been said, with variations in the conditions of solidification and further cooling, and with the efficiency of the homogeneity reheatings which it has undergone. Examination of Fig. 4 shows that when an alloy which has been given the heating for homogeneity described in Sec. 22, and has therefore attained a degree of uniformity represented by the hne A"B" , is heated above its annealing temperature h, it may be taken to a considerably higher temperature before burning proceeds (^4) than though the metal had been merely reheated. This is a necessary consequence of the SOFT AND MEDIUM STEELS 31 significance of the solidus and border-line B"B^ . Tlius, in the special case represented in Fig. 4, if — ^instead of continuing to increase the tem- perature immediately after temperature ^2 has been reached with the first heating — we proceed to this increase only after t^ has been maintained for some time, it is clear that burning will begin at temperature ti, considerably above h. It is evident that preliminary heat treatments for homogeneity exer- cise great influence upon the high temperature a metallic alloy can safely stand without burning. 26. Even in the case in which a reheating never reaches the minimum temperature B^^ , at which the alloy under treatment starts to burn, the maintenance of a very high temperature for a very prolonged time may cause troubles which greatly limit the industrial application of the rules mentioned in Sec. 23, aimed at increasing the efficiency of the prelim- inary heat treatments to its maximum. These troubles consist mainly in the fact that processes of agglomera- tion take place in the crystalline elements of all metallic alloys kept for a long time at a high temperature even if this temperature be below the solidus. The dimensions of each crystallite forming the alloy may some- times increase in very large measure; while at the same time phenomena of "orientation" may take place within the crystalline mass. We will have occasion to see many consequences of these phenomena later, such as an increase in size of all the crystalline units in the metal, and more especially an accentuation of the effects which the direction of testing has upon the physical properties found. Except in very special cases, the mechanical properties of an alloy which has been overheated are such that it must be considered to be in worse condition than it was before. CHAPTER IV PRIMARY CRYSTALLIZATION OF A TERNARY SYSTEM 27. Having outlined the manner of crystallization of binary alloys, we may now take the next step toward the more general case of multiplex systems already mentioned in Sec. 10 (Fig. 1) and which constantly occur in practice. This approach may be made by investigating a ternary alloy, which is the next more complicated than the one analyzed in the preceding pages. In other words, we will consider that portion of a ternary system rich in iron and containing two other elements, to be indicated by the symbols X and Y. Adopting the well known method of representing such ternary system by means of a diagram in three dimensions, this amounts to the study of that corner of a cube (represented in isometric projection in Fig. 5) closely adjacent to the edge AB corresponding to pure iron. As before, the study will be restricted to the alloys which form a homologous series of mixed crystals exclusively. There appears, then, on each of the two faces of the cube ABX and ABY a. binary equilibrium diagram between iron and elements X and Y respectively such as has already been studied in detail. The planes intersect in a common ordi- nate, AB, representing pure iron; CD and CE respectively represent the liquidus and solidus of the system Fe:Y (iron alloyed with element Y) and CF and CG those of the system Fe:X. All points in the XY plane represent ternary alloys in iron, element X and element Y. Ordinates erected vertically from these various points represent the condition of those alloys with changing temperatures. Each alloy may have such an ordinate, ending at the temperature where the metal is entirely melted, and a surface formed by these ordinate-ends passes through the liquidus curves CD and CF. It, also, may be called a "liquidus surface," or more shortly, "liquidus." In the same manner, the locus of all temperatures corresponding to to the end of crystalhzation of the various ternary alloys, such as the point L for the alloy I, is a surface passing through the two curves CE and CG. It may be called the sohdus. For the same reasons which apply in the case of the binary solidus, this last surface is also the locus of all the points representing the composition of the mixed crystals in equilibrium with residual liquids at the various temperatures. The compositions of conjugate sohd and Uquid phases in equilibrium are represented by the intersection of solidus and hquidus with a horizon- tal plane representing the temperature. 32 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 33 28. Let us now take any ternary alloy, for instance that one whose composition corresponds to the vertical IF, at a temperature high enough for it to be totally melted, such as that corresponding to the ordinate of the point I. If this alloy be allowed to cool to temperature H, the inter- section of the vertical//' with the liquidus surface CDF, mixed crystals will start to form around each nucleus of crystallization. Fig. 5. — Ternary system rich in iron. As has been done for the binary alloys, let us restrict our examination to only one crystallite. The composition of the first solid solution to separate will be represented by some point H' of the curve ss on the intersection of the solidus surface CGE with a horizontal plane passing through H. In the present state of experimental knowledge it is not yet possible to define the exact position of the point H'. 34 HEAT TREATMENT OF Continuing the cooling of the mass, and assuming that at every instant during the entire crystallization the diffusion of both the elements X and Y in the mixed crystals and between them and the residual liquid, takes place at such velocity as is sufficient to maintain a practically uniform concentration in the .entire solid mass, it will be inferred from what we have already seen in the case of binary alloys that the point H' will follow a curve which lies on the solidus CGE and stops at the point L, when the entire mass is solidified. At the same time the point H, the composition of the residual liquid, will follow a curve HM on the hquidus CDF, and stops at some point M of the curve rr, which represents the intersection of the surface CDF with the horizontal plane passing through the point L. In regard to the position of the point M upon the curve rr, same remark applies as was made in regard to the exact position of the point H' upon the curve ss. The discussion given in Sec. 12 covering the complete diffusion throughout the solid solution in binary systems might be repeated at this point as applying to the crystallization of a ternary system in complete equiUbrium. However, as already pointed out, one may say that this case never occurs in practice. In fact, the velocity of diffusion and the time at high temperature is so limited during cooling of hot metal in industrial practice — and for that matter usually even in specially slow coolings — that the equalization in concentration between the various points of the mass is far less than that which could cause even an approximately uniform sohd throughout the mushy stage. 29. This being the case, it is evident that phenomena will always appear analogous to those described for binary alloys. Let us outline the method whereby these phenomena can be represented in such a diagram as Fig. 5. Let us still consider the ternary alloy of composition I, whose solidifi- cation begins at H by the formation of mixed crystals whose nuclear composition is represented by the point H'. When the further cooling of the mass below temperature H proceeds with such velocity as to cause incomplete equihbrium in concentration, one can consider at each instant the three following characteristic points (situated, naturally, upon the same horizontal plane) which define the maximum, average and minimum concentrations of the two elements X and Y in each crystal of solid solution. 1. The point representing maximum concentration of the two ele- ments X and Y in each crystalhte. This concentration naturally occurs at its periphery, as is the case with binary systems, and it is clear that it is represented by some point on the true solidus ECG. If X and Y diffuse at about the same speed, it will be located on a curve such as H'P lying not far from the curve H'L. 2. The point representing the average concentration of the two metals SOFT AND MEDIUM STEELS 35 X and Y in each of the crystallites. For the same reasons which have been discussed for binary alloys, it is evident that — step by step as the temperature of the system goes down — this second point follows a curve H'N which is situated totally underneath the solidus ECG, and is located between it and the axis of coordinates AB. It is also clear both by definition and from what has already been said regarding binary alloys, that this curve must meet the vertical II' at a temperature N, cor- responding to the end of the crystallization of the system. This last conclusion depends upon the fact that at N the average composition of the crystalline mass is equal to the average composition of the alloy. Its vaUdity evidently depends upon the imphcation that in that portion of the tri-axial diagram examined, the liquidus and the solidus do not touch each other at any other point than C, the melting point of the major constituent. This hypothesis was made in the case of the binary alloys and is now to be extended to the ternary alloys. As a matter of fact, it represents the true state of affairs always existing in the alloys which are the subject of this treatise. 3. The point representing minimum concentration of the two elements X and Y in each crystallite. It is evident that this minimum con- centration is that of the nucleus. It may again be deduced from the reasons already examined in the binary system that, as the temperature of the system lowers, this point follows a curve H'O which is situated totally below the curve H'N of average compositions, and a fortiori below the curve H'P of maximum concentrations. It will be also located nearer axis AB than the two preceding ones. The third curve just defined will cut the horizontal plane passing through temperature N, at a point 0, representing the concentration of the elements X and Y in the nucleus of each crystal at the instant of total soUdification of the system. The same horizontal plane cuts the curve H'P in P, representing the concentration of the elements X and Y at the periphery of each crystal at the same moment. 30. From what has already been seen to happen with binary alloys, it is easy to understand that the position of the two curves H'N and H'O for an alloy of a given composition depends upon the velocity of cooling, precisely in the sense that the more rapid the coohng, the more the two curves diverge from each other and from the curve H'L. The hmiting position which curve H'O approaches under drastic quenching is the vertical through H' In other words, the analysis represented by H' could be preserved at the centers of the crystallites only in case the cooling had been so sudden as to prevent the smallest accession of elements X and Y by diffusion. Conversely, the slower the mass cools the more the three curves H'O, H'P and H'N approach each other and curve H'L. They finally 36 HEAT TREATMENT OF SOFT AND MEDIUM STEELS coincide with the latter only in the limiting case already considered in Sec. 28, where diffusion and equilibrium is complete at all times. Following the discussion given in Sec. 16 for binary alloys, it also follows that the temperature corresponding to complete solidification denoted by the horizontal plane ONP will be lower the more rapid the cooling. 31. From all these considerations and reasoning from analogies derived from the continuous nature of the phenomena under investi- gation, it may be stated that at the instant when crystallization ends,, the concentrations in X and Y of the successive concentric layers formed around the nucleus of each crystal, are represented in Fig. 5 by suc- cessive points of the curved segment ONP.^ In this segment, point represents the composition of the nucleus, and P that of the periphery of the crystallites. When the temperature of the system continues to decrease below the plane ONP corresponding to the end of the crystalhzation, diffusion of the two elements X and Y in the mixed crystals continues, and the two ends of the segment and P approach the vertical //'. Thus ONP, the "segment of variant concentration," in its successive positions and lengths assumed during cooling, describes a "surface of variant concentrations" defining the heterogeneity of the mixed crystals which gradually narrows with decreasing temperature. The significance, behavior, and characteristics of this surface which in Fig. 5 is indicated with a curved shading, are entirely analogous to those of the band of heterogeneity studied for binary alloy. Furthermore, one can apply to the present case the exact considerations developed in Sec. 18 to Sec. 26 concerning the effects of reheating processes; their limita- tions on account of burning the steel; the variations in the temperature at which steel starts to "burn" caused by heat treatments, etc. There- fore it is useless to repeat here all that has already been said for binary alloys. Nevertheless, we shall see presently in studying surfaces of variant concentrations for ternary and more complex alloys, that other phe- nomena specially characteristic of these more complex systems appear in addition to the phenomena analogous to those delineated for binary alloys. • The segment ONP would be a straight line only in case the diffusion velocities of the two elements X and Y rigidly followed a unique interrelation during the entire mushy stage. CHAPTER V PRIMARY CRYSTALLIZATION OF MORE COMPLEX SYSTEMS 32. In the most interesting cases which appear in practice, and especially in the machine steels under consideration, it is necessary to study the transformations of systems formed by a number of con- stituents much larger than three. In all these cases, graphic representa- tion of the phenomena becomes very difficult. Moreover, it can be stated that it is impossible to construct a diagram or space model for systems of more than four elements; indeed the complete graphic repre- sentation even for a quaternary system (studied especially by N. Parra- vano and G. Sirovich), is extremely complicated. Still for purely practical purposes it is nearly always possible to. reduce the study of the more complex technical problems in machine steels to the examination of relatively simple equihbrium diagrams sufficiently exact for the use intended. Such results may often be reached by first considering separately the qualitative and quantitative effects of one or two of the constituent elements as their content varies. Briefly, this may be done by selecting one or two of the elements which have the widest range in the analysis of a given type of steel, and whose effect is also the most pronounced. All the other elements composing the metal entering into solid solution with the iron, and whose analysis is substantially constant for the same type, may be lumped with the iron and considered as a "complex constituent." For instance, a plain carbon steel may be regarded, at least as far as the graphic representation of the phenomena of equilibrium is concerned, as a binary alloy. The minor constituent will be carbon, and the major constituent a "complex" containing a large amount of iron and small amounts of phosphorus, sulphur, manganese and silicon. Likewise, a nickel-carbon steel can be represented as a ternary system wherein the major constituent is the complex, and the two others are carbon and nickel. It is easy to see why such simplified graphic representation of these systems is possible only for those which comply with certain conditions. Thus, for example, it is clear that a suppositious "complex" constituent (in the sense above mentioned) would complicate instead of simpUfy the tracing of the diagram, should its components give rise to distinct and separate phases within the same range of temperatures and con- centrations encountered when examining the "fictitious" binary or ternary equilibrium. The necessity of securing such conditions naturally imposes many 37 38 HEAT TREATMENT OF restrictions upon the use of this simplified method. In particular, it is absolutely necessary to be acquainted with the various equilibrium diagrams of all the systems, binary and ternary, real and "fictitious," which can be formed with the constituent elements of the complex alloy in order that this artifice may be used with confidence. This is also true in case one wishes to examine only in a general way a hypothetical binary or ternary fictitious diagram. But since the fiction of complex constituents will not cause any new phases to appear in the heat treatment of the soft and medium steels under consideration so long as carbon is excluded from their number, it seems timely to develop an outline showing its usefulness in the study of the greater part of the instances which happen in practice. 33. As just remarked, one of the components in machine steels, viz. carbon, exhibits properties which completely differentiate its behavior and functions from those of the other components, mainly for the following reasons: (a) The linear velocity of diffusion of carbon due to a difference in concentration in the mixed crystals, is far greater than that of any of the other elements wliich normally join with it in the formation of the steels we are studying. It is known that this phenomenon is essentially due to the fact that in extremely large measure the process of diffusion of carbon in solid steel is by virtue of the action of gases dissolved in the metal. At this juncture it is certainly out of place to study the mechanism of this action. It has been described in recent studies upon the cementation and case- hardening of steel.i Sufiice it to say that all the experimental researches demonstrate the fact that the linear velocity of diffusion of carbon per se in solid solution in iron as cementite where the intervention of gases is practically excluded, is of the same magnitude as the velocity of diffusion of the other elements frequently found in steel. On the other hand, other things being equal, if diffusion of carbon in the solid steel takes place in the presence of the gases which are ordinarily dissolved in steel (notably the oxides of carbon, hydrogen, and hydrocarbons) its lineal velocity increases enormously, reaching values hundreds of times greater than those for all the other elements which with carbon com- prise the analysis of the steel under examination. (b) Other conditions being equal, variations in carbon concentration cause very much greater variations in the mechanical properties of the steel (especially under the action of certain heat treatments) than those produced by equivalent variations in concentration of most of the other elements contained in sohd solution in the steel (manganese, silicon, nickel or chromium, etc.) iThis subject has been studied in my volume on "The Cementation of Iron and Steel" (McGraw-Hill Book Company, Inc., New York, 1914). SOFT AND MEDIUM STEELS 39 It is true that very small variations in absolute concentration of some of the elements contained in steel, such as sulphur and phosphorus, are sufficient to produce much greater variations in mechanical properties of the metal than those caused by even greater variations of the absolute concentration of carbon. But it is necessary to bear in mind that these elements are contained in but minute proportions in steels suitable for industrial use. Therefore, a small absolute variation in their concentrations is actually a great relative variation, and it is nothing less than these relative values which have to be taken into account, as will be clearly seen presently. It will also be necessary in some cases to give these elements special consideration, exactly as we shall soon see it is useful to do in regard to carbon. (c) Iron carbides (FesC) in plain steel, or cementite mixed with other metaUic carbides in special steels, exists as solid solution in austenite immediately after freezing is completed. Yet while the temperature is dropping from this value^ to the transformation ranges of the solvent iron, they give rise to new phases by the segregation of conjugate sub- stances thrown out of the solid solution. This is due to the presence in the austenite area of a field of incomplete solubility, called for want of a better name an "insoluble field" (Italian: lacuna di miscibilitS, ; German: Mischungsliicke; French: lacune de miscibilit^). Such insolu- bihty does not occur with the other constituent elements, within the Umits of the concentrations occurring in machine steels. The phenomena mentioned are noted in the equilibrium diagram by special fields whose existence and significance are essentially related to other markings in the iron : carbon diagram. Similar insoluble fields do not occur in other binary equiUbria formed by the other substances taken two by two entering into the composition of any machine steels. Furthermore, the addition of these substances to the iron : carbon system may modify or obhterate some of those markings or fields representing the transformations which take place in the solid phases of such a binary system; but it can never produce new ones, provided their amounts be no greater than those in which they may appear in the steels which we are studying. 34. Therefore, from what has been said in Sec. 32, all existing conditions render it opportune to select carbon, present in the form of metallic car- bides in the soUd solutions, for one of the principal members of our fictitious binary diagram, and iron for the other— not pure iron, but iron containing all the other elements forming with it homogeneous molten masses or homologous mixed crystals. In the same way, the fictitious ternary diagrams to be considered in the study of some special steels shall always be made up of iron and carbon as two of the fundamental elements, and for the third the principal alloying metal in the steel under ' Defined in Sec. 17. 40 HEAT TREATMENT OF observation. For evident reasons of simplicity, we shall begin by considering a fictitious binary system in which the complex fundamental element is formed by the solution of but one element {X) in iron. Later on we shall see, for the alloys with which we are concerned, how it is possible to pass with relative facility from this simpler case to the real ones, which are a great deal more involved. Fig. 6. — Diagram for'steelscontaining a fixed amount of alloying element. With this end in view, let us again examine the complete ternary dia- gram already described in Sec. 27 to 31 from which the binary diagram is to be derived. It is formed by iron and the two elements X and Y, but with the modification that one of the two elements represents carbon. As has been often repeated, the resiilts of all the experimental re-r SOFT AND MEDIUM STEELS 41 searches made up to the present time, show that the crystallization of the various binary systems composed of iron and each one of the other nor- mal elements (except carbon) up to its maximum concentrations in ordinary specifications follows Roozeboom's first type. This com- prises the systems in which only one series of mixed crystals form, and their compositions are always higher in iron than that of the residual liquid with which they are in equilibrium. Up to the limiting concentrations in machine steels these binary systems do not possess any insoluble field either in the Uquid or in the soUd phase. Draw the two characteristic curves ("liquidus" and "solidus") of such a diagram between iron and another element Y upon the plane FeAX and let them be indicated respectively with a and b. Upon the other plane FeAC draw the well known equilibrium diagram of the binary iron: carbon system, or to be more precise, of the binary system iron: iron carbide (Fe:Fe3C), since this corresponds to that portion of this diagram to which our deliberations will be restricted. As has been done in the other cases, consider only the phenomena which take place at temperatures higher than the allotropic transforma- tions of solid iron, reserving for later consideration the occurrences at or below these points. Since the highest possible carbon concentrations in plain machine steels are a great deal less than 2 per cent., which is the point at which the insoluble field between iron and iron carbide intrudes itself at the temperatm-e of crystallization, it wUl be possible to restrict our present considerations to only two continuous curves, viz.: the "liquidus" c and "solidus" d. Passing through each pair of the four curves thus traced will appear the liquidus surface ac and the solidus bd in the ternary systems. Their meaning has been amply indicated and therefore is well known. Now consider a series of ternary alloys in which only the carbon concentration varies while the concentration of element X remains constant. Points representing the condition of these alloys at all pos- sible temperatures will be found upon the vertical plane parallel to the FeAY plane passing through the point B representing that binary alloy of iron and x per cent, of element X. BD and BE represent the intersections of this plane with the two faces FeAX and FeA Y and curves e and / the intersections of this plane with the liquidus surface ac and sohdus surface bd of the same system. Now consider all the alloys of this group as being part of a fictitious binary system whose constituents are iron carbide and the alloy FeX containing the quantity x of the original component X, and discover which of the phenomena occurring during crystaUizlation of such a system can be represented on the plane DBE. It will be easy to see from the following analysis what actual use such a sketch derived from the true ternary diagram may have. 42 HEAT TREATMENT OF 35. Consider any one of the alloys in the group as defined, such as the one whose composition is represented in Fig. 6 by points on the vertical GG'. Whatever may be the composition G, the initiation of crystalHza- tion will occur at the intersection of the vertical GG' with the liquidus surface ac, and, therefore, by the very definition of curve e, at the inter- section of the same vertical with the curve p. Hence, it is apparent that curve e in the fictitious binary diagram retains at least a part of the signifi- cance of the ternary surface ac from which it is generated; that is to say, it locates all points where 'crystallization starts. It will soon be seen that it can be denoted by the name liquidus only in this restricted sense. An analogous train of reasoning can be repeated in regard to the curve / by following the actual course of crystallization of the random alloy GG' located on plane DBS. In fact, in the limiting case where diffusion in the solid phase and between the solid and the mother liquor is able to equalize at all times the concentrations of all elements in the mixed crystals, it is clear that the end of sohdification will take place at a temperature when the composition of the solid phase in equilibrium with the last residual liquid is equal to the average composition of the entire system. Therefore, in the case under examination, freezing of the alloy GG' will be complete at point M, the intersection of the vertical GG' with curve / on the solidus surface bd. This last curve in the fictitious binary diagram, thus has a meaning limited to the curve of the end of crystallization. It is also easy to see why the two curves eand / in the fictitious binary diagram do not possess the entire meaning of the two surfaces ac and bd upon which they are located. In fact, the composition of the mixed crystals which start to separate from the molten mass at temperature H can be located on the solidus surface at I, in the manner already noted in Sec. 28. The carbon concentration in this nuclear solid solution is represented by a point L, the projection of I upon the plane DBE by a ray parallel to AX. This point L always lies above the curve / in the fictitious binary diagram, within the area between this curve and the liquidus e. This last fact is an evident result of the assumed position and curvature of the surface bd, which location is in agreement with all the known experimental results upon soft and medium steels used for machine parts. It is therefore clear that each of the points on the curve / in the ficti- tious binary diagram do not represent the carbon concentration of the mixed crystals at equilibrium with a molten mass whose analysis corresponds to a conjugate point on the surface ac somewhere along its intersection with the same horizontal temperature plane. As the temperature lowers step by step, the point representing the true composition of the ternary mixed crystals in complete equiUbrium will SOFT AND MEDIUM STEELS 43 follow a curve which starts at / and ends at M, lying entirely in the surface hd. Within this temperature interval, corresponding to the mushy stage, the point in the fictitious binary diagram representing the carbon concen- tration in the solid phase will follow a curved segment LM, lying entirely above curve / within the area between the two curves e and /.' In the fictitious binary diagram we may indicate this segment as the solidus for the alloy of initial composition GG'. At the same time, the point representing the true composition of the residual ternary liquid will follow a curved segment HN lying upon the ternary liquidus ac. This segment passes through the point H and the point N is its end as it meets the horizontal temperature-plane passing through M. Naturally, all this discussion has value only under the hypothesis that cooling takes place slowly enough to allow the concentration of the various elements to diffuse to practical uniformity at each instant in all parts of the solid phase. Having the positions of the two curves a and c given and assuming no discontinuity or singular points in the surface ac — it is easy to see that if iV be projected normally upon plane DBE of the fictitious binary diagram, its location N will lie below the curve e. Since by definition this projection must lie in the horizontal plane passing through M, it must be located in the zone limited by the two curves e and /. It is therefore clear that, during crystallization of the system, the point representing the carbon composition of the residual liquid will follow curved segment HN, situated in the space between the curves e and / of the fictitious binary diagram. HN' (on the plane DBE) is the trace of an element parallel to axis AX moving with curve HN as a directrix. In the fictitious binary diagram we may consider this segment as a liquidus for the alloy of initial composition GG'. Therefore it is seen that the curve e as well as the curve / does not possess the whole meaning of its generating surface ac in the true ternary diagram. This curve is apparently the locus of all points such as H, corresponding to the initial point of crystallization of wholly molten alloys of that average composition, but it is no longer the locus at all tempera- tures during the crystallization process of the composition of the mother liquor in equilibrium with the mixed crystals which have separated from it. 36. Once these points are established, let us see what useful conclu- sions can be deduced from the examination of the fictitious binary diagram which has just been described. It is evidently convenient to rotate plane DBEoi Fig. 6, which contains '■ These facts could be easily demonstrated in a rigorous way, on the basis of the data we have supposed to obtain, and which really do hold for the alloys to which the present book refei^. I have- deemed it better not to develop these demonstrations here, since they are somewhat long and not absolutely necessary. 44 HEAT TREATMENT OF the projection of the diagram, upon the plane of the drawing. In Fig. 7 this is done, using the same notation as in Fig. 6. It is clear that Fig. 7, being limited to a system of only three com- ponents (iron, carbon and X) is nothing more than a particular case of the diagram which was briefly investigated in Sec. 10 (see Fig. 1). The intervening study of the relations existing between hues and areas of Fig. 7 and the corresponding parts of the ternary diagram, from which the former are derived in a definite way, permits one to indicate precisely the real significance of the markings. Among the phenomena connected with these markings it is easy to distinguish clearly those which are Fig. 7. — Fictitious diagram derived from ternary system. directly due to the action of the principal constituent (in this case carbon) from those due to the influence of the other elements present in the liquid and solid solutions. The same study has estabhshed the position occupied by the Uquidus HN' and the solidus LM in the fictitious binary diagram for an alloy of a determined initial composition in its relation with the Uquidus and the solidus surfaces in the ternary diagram. From these considerations it is evident that the fictitious binary, diagram represents the real changes in the concentration of carbon, the fundamental element, both in the solid and in the liquid phase during the entire solidification of a given ternary alloy. On the other hand, the variations in concentration of the third element are not directly repre- SOFT AND MEDIUM STEELS 45 sented in the same diagram, and are only reflected in it through the effects which they produce upon the carbon, effects which we shall see are essentially among their most important ones. The practical uses of such a simpUfied graphic representation will be expounded immediately. 37. For this purpose consider the system next in increasing complexity. In other words, examine the quaternary system resulting by adding a fourth element Z to the ternary system iron : carbon : element X, which has been considered in the preceding paragraphs. Element Z is sup- posed to form solid solutions in iron exactly as does element X and is subject to all the conditions of equihbrium which we have assumed to exist for element X itself. It is easy to see. that to represent such a quaternary system it is only necessary to construct a "fictitious ternary diagram." From this it is possible to deduce new "fictitious binary diagrams" representing the various groups of alloys of the quaternary system, in a way perfectly analogous to that which has been done for the original complete ternary system. In detail, construct in space a true ternary diagram of the system iron : carbon : element Z. Then represent on the appropriate plane the group of aU ternary alloys containing a fixed quantity of the element. Z. We will thus be able to sketch, in the same way as for the sys- tem iron : carbon : element X, a fictitious binary diagram having X as the principal element and which will record the solidus and liquidus inter- sections for a ternary alloy with a constant proportion z of element Z. In a similar way, let us suppose that we have constructed the ficti- tious diagram of the system iron : carbon : element Z in which the element Z also appears in the constant proportion z, and as a solid solution in iron forms a principal element in the fictitious binary diagram. Now, let us take the two fictitious binary diagrams thus constructed and lay them out on the reference planes in a "fictitious ternary diagram" which will have as principal element the solution common to the two systems: that is to say, iron containing a constant amount of Z in solu- tion. This will be located at the origin of coordinates. Along the X axis will be laid the element X containing z per cent, of the element Z in solution, and along the Y axis will be found various proportions of carbon as carbide also containing the same constant proportion of element Z. The diagram thus constituted is represented in the usual way in Fig. 8. This can be considered in a limited sense as an equilibrium dia- gram comprising the alloys iron : carbon : element X, containing in addi- tion the fixed amount z of the element Z. In this diagram pass a plane parallel to the plane of coordinates FeAC passing through the point B on the X axis, which point corresponds to the proportion x of the element X. We will be able to trace upon this plane a fictitious binary diagram, in the same way we have already done 46 HEAT TREATMENT OF in Sec. 34. This diagram will have as principal markings lines which represent the temperatures of the start and of the end of crystaUization, together with intercepted portions of ordinates representing the tempera- ture range of the mushy stage for all alloys made up by iron and carbon in variable proportions, and containing the two elements X and Z m the constant proportions x and z respectively. ^...fS'xf^a;! Fig. 8. — Quaternary system having constant quantity of fourth element. It is easy to see what the principle characteristics of such new ficti- tious binary diagram will be from a consideration of the properties of the fictitious ternary diagram wherefrom it has been developed. 38. With this end in view, let us examine the "path" of crystallization, as shown in the diagram. Let GG' be the composition of the alloy SOFT AND MEDIUM STEELS 47 selected. Let us restrict consideration, for the time being, to the case in which coohng takes place slowly enough to allow diffusion to produce substantial uniformity of concentration in all parts of each of the phases. Curves e and / are located at the intersections of plane DBE with surfaces ac and hd. All points of these curves correspond respectively to the start and end of crystallization of the alloys whose average com- position is represented by the projection of the points in question on the XY plane. One can repeat nearly the same remarks made in Sec. 35 for the pre- ceding case, regarding the compositions of the sohds in equilibrium with the successive residual liquids as the crystalhzation gradually pro- ceeds. The only essential difference between the two cases (which may be distinguished as the "true" ternary system and the "fictitious" ternary system respectively) is entirely similar to the difference existing between the "true" and "fictitious" binary diagrams. It is this: the points rep- resenting the composition of the solid phases in equilibrium with mother liquor throughout the mushy stage do not lie any longer upon the surface bd representing the end of solidification. It can be demonstrated that these points (similar to point /, at the start of crystalhzation) are located upon a segment of curve IM lying entirely above the surface bd but intersecting it in the point M, corre- sponding to the end of crystallization of the alloy considered. Without entering into a rigorous demonstration — necessarily long yet superfluous here — it is easy to understand why this may be true. First bear in mind that the composition of the solid phase in the fictitious binary system represented upon the plane FeAC has been demonstrated to be located on a curve located above curve d. Plane FeAC corresponds to a; = 0. Evidently also the locus of the solid phase would be located above surface db in systems directly derived from x = simply by add- ing sufliciently small quantities of the element Y. Now, if any point such as on the segment IM be supposed for the moment to be located below the surface bd, such supposition would lead to consequences openly at variance with the definitions and with the known fundamental properties of the diagram. In fact, under this hypothesis, the vertical ordinate passing through would cut the surface bd in a point W corresponding to a temperature higher than that corresponding to point 0. But it is known that the point W corresponds to the end of crystalhzation of the alloy of average composition OW. It therefore represents the loioest temperature at which mixed crystals of uniform composition OW may be in equilibrium with the last traces of a mother liquor whose composition is located at some point on the intersection between the surface ac and the horizontal temperature plane passing through point W. But, by the very definition of segment IM, mixed crystals of the same composition in carbon and element X corresponding to the point are also in equilib- 48 HEAT TREATMENT OF rium with a residual liquid (corresponding to some point on the surface ac). This can occur only when is at or abo\e W, and it is evidently not possible that the point can be placed underneath as originally supposed. It is indeed true that the proportion of the element Z in the mixed crystals and W is not the same; but we positively know that in the crystals located at W this proportion has by definition the value z, while in the crystals it is certainly less than z, which it reaches as a matter of fact only when solidification is complete and the solid solution possesses the same composition as that represented for the aUoy GG'. This difference of the concentrations of the element Z adds therefore additional weight to the deduction made regarding the position of curve IM. At any rate, for reasons entirely similar to those mentioned when considering a true ternary diagram and depending upon the curvature of the surface bd, it is clear that the projection LM of the segment IM upon plane DBE will lie totally above curve/ within the area between this last curve and curve e. Similar considerations, which would be superfluous to repeat, will locate the pseudo-liquidus curves HN and HN' in a path conjugate to the one indicated for curves IM and LM respectively (see Sec. 35). 39. Therefore, even in the more complex case now examined, one may construct a "fictitious binary diagram" entirely like the one which was obtained in the simpler preceding case (Fig. 7, and Sec. 36). The same considerations made at that time can be repeated. Without entering into useless details or repeating similar reasons it follows that analogous conclusions may also be reached for more complex alloys containing three, four or even more elements, each one in a con- stant percentage, together with the two fundamental elements iron and carbon present in variable proportions. Naturally this remark is true only when the fundamental conditions as to solubility and continuity are taken as a necessary and essential basis. But as has repeatedly been said, these always obtain in practice for the steels belonging to the group to which we are Hmiting our study. It is now timely to remark that the preceding discussion and its accompanying sketches have visualized an idea cleverly enunciated and verified by Guillet.^ He postulates the existence of "equivalent values" for the various components entering into metallic aUoys, especially as regards the solid solutions and their decomposition products. In our graphic representation, these "equivalents" would be defined as the quan- tities of each of the half dozen or more component elements which cause 1 See L^on Guillet "Les aciers spiciaux" (Paris, H. Dunod and E. Pinat) 2 volumes, and "Etude industrielle des alliages m^talliques" in one volume, by the same author and publisher. The idea as applied to complex brasses has been ex- pounded at length in Chemical and Metallurgical Engineering, Vol. 24, p. 117. SOFT AND MEDIUM STEELS 49 equal displacements of the pseudo-solidus and -liquidus in a fictitious binary diagram for a series of alloys of given composition in the two principal elements selected. The experimental data available in regard to this subject at the pres- ent time do not permit the formulation of general conclusions, nor the quantitative application of the graphic method to the complete and pre- cise study of a practical case. Nevertheless we are going to see later how this representation may find convenient application in the partial study of some phenomena which take place in our steels. It would certainly be of a great industrial importance to gather a collection of experimental data, so as to define better the conceptions introduced by Guillet, and to coor- dinate his conclusions in a series of complete and well defined equilibrium diagrams. To reach this point, however, it will be necessary to gather a very large amount of experimental data. 40. Using the conclusions already attained for the more complex cases as a basis it is possible to represent graphically the phenomena which take place when the cooling velocity of the system is high, too high to permit diffusion to proceed in such measure as to equalize the concentrations of the various constituents at each instant in each phase. To do this, take as starting point the aggregate data which has been established in Sec. 29, 30 and 31 in regard to the true ternary system. A brief stirvey of this information in relation to the elements of the ficti- tious binary and ternary diagrams defined in the preceding pages shows instantly that all the considerations already made in the simpler cases may unquestionably be applied to the more complex cases represented in the corresponding diagrams. In fact we have seen that the fictitious diagrams of even the very com- plex systems always show in a well defined manner the segments of curves locating the points of initial and final crystaUization for every alloy of a determined average initial composition. Moreover, in the same diagrams, the curves or surfaces denoting the locus of the start and the end of crys- talUzation also bound the space within which must lie aU the curved seg- ments which determine the variation in composition throughout the mushy stage. Being merely an exact repetition of the same reasoning already de- tailed, I do not deem it opportune to again go over the minute analysis of the familiar processes connected with the incomplete development of diffusion, nor to a detailed explanation of the graphic representation of the phenomena. I believe it is enough to mention the fact that the fictitious diagrams of the more complex systems will also be able to show the "curve of average composition" of the mixed crystals and the "curve of minimum concentrations" of the principal elements in soUd solution, together with another sohdus, cori'esponding to the external part of each solidified crys- 50 HEAT TREATMENT OF talline element. These curves have a course entirely similar to those described in the simpler cases, and bound an area of variant concentra- tion whose significance is also Uke that analyzed for the preceding cases, when the former is interpreted according to the well defined values which have been assigned to the lines and fields of the fictitious diagram. Thus, diagrams may always be constructed entirely similar to those which were sketched when studying the binary and ternary systems (see Fig. 1, 2, 3, 4 and 5). Disregarding the difference in composition of the various phases inherent to the selection of different fundamental elements, simple and complex, the only difference in form which is found between the points, curves and surfaces of a true diagram of a simple system and those of a fictitious diagram of a complex system, lies in the fact that in the latter both liquidus and solidus vary upon changing the initial com- position of the alloy, and do not coincide any longer with the curves or surfaces representing the start and the end of crystallization. 41. Regarding the steels to which this study applies, we have already seen in Sec. 33 that for various reasons it is convenient to represent the various systems — always very complex — by means of fictitious diagrams belonging to two fundamental types: (a) A fictitious binary di9,gram between carbon and iron, the iron containing a certain amount of other chemical elements in solid solution. This diagram lends itself well to the study of the so-called " carbon steels, " in which none of the elements to be lumped together with the iron will give the steel any properties which notably overwhelm those produced by any of the others. (6) A fictitious ternary diagram between iron, carbon and another metal, iron again carrying the many elements in sohd solution. This diagram lends itself well to the study of those so-called " special steels " in which one of the constituent elements other than carbon has a prepon- derant influence upon the properties of the steel, both because it is con- tained in the steel in greater proportions and because its specific effect is more intense than those of the other components, even at parity of concentration. Sometimes it may be convenient when studying some practical cases to take into consideration another method of graphic representation of the more complex systems, founded directly upon the apphcation of Quillet's conception of equivalence already mentioned in Sec. 39. This method uses a three-dimension diagram which may be defined in the following way: (c) A diagram having three rectangular axes. The vertical axis rep- resents pure iron, and distances up from the origin represents tempera- tures. Upon one horizontal axis is taken the concentrations of the second principal element, which in our cases we have seen normally is carbon. The other horizontal axis represents the concentration of a hypo- SOFT AND MEDIUM STEELS 51 thetical equivalent of the other elements contained in the alloy taken all together. It is clear that a complete model thus constructed could lend itself to an exact graphic examination of the manner of crystallization of the various alloys and of the various phenomena of transformation which take place in them, only in case we knew with precision from complete ex- perimental data the "concentration equivalents" of each one of the ele- ments whose concentrations must be lumped together on the third axis of the diagram. In fact, it is clear that such detailed knowledge is necessary if we simply want to locate an alloy of given composition upon the dia- gram. This holds true even when the average value corresponds simply to the sum of the concentrations of each element. However, inasmuch as these equivalents have not been determined, nor is it known if an absolute value may be attributed to them with confidence, it is clear that diagrams constructed according to Guillet's ideas may not be used in any of the cases in which it is necessary to establish any quantitative datum. Such an approximate diagram may present some interest when one merely wants to represent graphically and but qualitatively the phe- nomena taking place in a given complex alloy. It is clear that a diagram of type (c) does not differ in construction from the true ternary diagram, which we have already studied (see Fig. 5 and 6). In fact, we find again in it all the characteristic lines and fields already analyzed in Sec. 27 to 31. Therefore it is not necessary to repeat its detailed examination, which anybody can easily accomplish on the basis of Fig. 5 and 6, by substituting the sum total of concentration "equivalents" for one of the principle elements such as Y in Fig. 5. The only difference between the two cases consists in the fact that points and hues in diagrams of thie third type (c) do not represent any definite analysis among the family of aUoys except in so far as the per- centage of iron and carbon are concerned. On the other hand, in the true diagram of a ternary system, as well as in the "fictitious" diagram of more complex systems constructed as specified just above in (a) and (6) the significance of each line, point and area representing a phase or having other meaning is always perfectly correlated to the initial compo- sition of each alloy. In the special cases to which this present study is limited, it is almost always of advantage to use one of the first two types of simplified diagram — either (a) or (6). PART II EFFECTS OF DIFFUSION UPON SECONDARY CRYSTALLIZATION CHAPTER VI GENERAL REMARKS UPON THE RELATIONS BETWEEN DIF- FUSION AND SECONDARY CRYSTALLIZATION 42. Let us now outline the phenomena which take place in a completely sohdified machine steel when its temperature falls below the value corresponding to an allotropic transformation of one of the constituents. In our steels, the only allotropic transformations which must be taken into consideration are those of iron. Only for special steels very high in elements other than iron (for instance high nickel steels) it may become necessary to take into account the inversions of some of those elements, but these can not be classed as soft or meduim steel for the construction of machine parts. Having already noted the inherent function of iron, in that its con- centration is always higher than that of all the other elements forming these steels, we may restrict our considerations to the simpler case of the allotropic transformation of the element which acts as "solvent" in the system under investigation. It is known that practically pure iron in cooling below the temperature of its solidification (about 1520°) undergoes three reversible transforma- tions, revealed by three arrests at 1410°, 910° and 790°C. Little data is available upon the first transformation point, cor- responding to the passage of delta to gamma iron. It has not yet been possible to isolate delta iron, and its properties are therefore not known. In carbon steels it seems that this transformation ceases to take place when the carbon concentration reaches or exceeds about 0.3 per cent. If a curve could be drawn in the iron carbon eqtdlibrium diagram represent- ing the beginning ofthe 5—^ transformation, it would intersect the solidus at this concentration. It is very probable that precise and complete knowledge about this first transformation of iron will throw great hght upon various occurrences frequently observed in industrial treatment of steel, and which have not been explainable by means of the lines and areas in the equilibrium diagram based only upon the other transforma- tion-points. At any rate, the data available upon this subject are too scarce and uncertain to be taken into the slightest account in a study of a practical. nature. 63 54 HEAT TREATMENT OF The nature of the second reversible transformation — corresponding to the passage of gamma into betairon^ — -has been the object of numerous researches, which, however, have not yet totally clarified the problem. Even supposing the reader to be acquainted with the recent literature upon this point, I must point out that I cannot dwell upon its nature, since prominent experimentors have even doubted the existence of such an allotropic transformation, basing their train of reasoning on the self same researches which others have made to estabhsh its nature. In fact, the results of the experimental researches now available are essen- tially negative in character, yet certain facts have been discovered which are difficult to harmonize with the older accepted iron : carbon diagram showing a definite phase called beta iron, yet it is not yet possible to substitute new lines and areas to take its place. Nor would I need to modify or substantially extend my conclusions should I include in this study what might be called the positive informa- tion now available regarding beta iron. This because as has been remarked often before I intend to limit consideration to those phenomena strictly dependent upon the chemical composition of the various parts of the crystalline elements of the steel; mentioning only in passing a brief outline of those facts which depend essentially upon the morphology of the metallic constituents — that is to say, upon their crystallography, using the term in its limited and precise significance. Various experimentors have reached much more concordant ideas (based upon a very large mass of experimental material) upon the nature of the third reversible transformation of iron, corresponding to the pas- sage of beta iron into the alpha modification or from gamma direct into alpha iron. 43. The rearrangement of phases accompanying allotropic trans- formations of iron take place according to laws entirely similar to those governing crystallization of the same alloys. This statment holds in the binary systems iron : carbon, iron : man- ganese or iron with any of the other elements entering into the- composi- tion of machine steels, as well as in the more complex alloys of the same elements within the limits of concentration to which I intend to confine my study. In fact, the transformation ranges are zones wherein new soHd solutions are forming. The concentration of the various elements existing in saturated solid solution with low-temperature modifications of iron as a solvent is always less at the temperature of equilibrium than the concentration of the same elements in saturated solid solution with another modification which is stable at the higher temperature. Evi- dently this statement holds for the temperatures at which both types of solutions, that is to say, both phases, are reciprocally in equilibrium. From this it follows that we may base our study of the transformations which take place upon cooling a given solid solution upon the same prin- SOFT AND MEDIUM STEELS 55 ciples already developed in the study of the first transformation (crys- tallization) of a liquid solution of the same composition. This under one condition however: that the original solid solution under consideration has been maintained at high temperatures during a suflBciently long period of time to cause the concentrations of its various components to be prac- tically uniform. When the last condition does not hold, it is clear that we will have to add to the reasoning developed in Part I the necessity of taking under examination a band of variant concentration in the region which repre- sents the phase stable at temperatures above the transformation point. Under absolute homogeneity, of course, such a band contracts into a vertical ordinate. I will analyze the two cases separately. 44. Beginning with the first and simpler case, in which the solid solu- tion stable above the transformation is exactly homogeneous, the appli- cation of the reasoning utilized when discussing crystallization will deduce some new conclusions. These are due, for all elements except carbon, to certain differences of a purely quantitative natiu-e in the way in which the different phenomena occur. In the case of carbon, a large proportion of the steels we will consider have an analysis which locates them in a region of insolubility in the equilibrium diagram extending downward from the eutectoid horizontal. Experimental data known at present are very scarce and uncertain concerning the phenomena connected with the alloying elements other than carbon. The only fact that can now be afiirmed with apparent cer- tainty is that the transformation range^ is very small. At least this remark applies in case the transformations take place without hysteresis of noteworthy amount. The consequences of a narrow transformation range are easily inferred from the reasoning already applied to the process of solidification ex- pounded in the preceding pages and may be summarized in the follow- ing remarks, which apply to the various elements usually present in machine steels, carbon excepted. (a) If as an effect of suitable diffusion, the homogeneity of the phase stable above the transformation is practically complete, so that the com- position of its individual crystallites (and of the whole mass) may be represented on the equilibrium diagrams by points, then the composition of each crystallite at temperatures below transformation can be repre- ' That is to say, the temperature at which the allotropy of iron first commences to make itself felt, either by evolution of heat, change in magnetism or some other physical property, is comparatively close to the temperature when these changes are at maximum intensities and to the temperature where the effects die out. Such transformation range, so-called, is evidently analogous to the mushy stage dursng olidification. 56 HEAT TREATMENT OF sented by a band of variant concentration so narrow that in practice these strips can be considered as lines. (6) When the composition in the crystaUites, edge to center, is appreci- ably non-uniform in the high-temperature phase, the band of heterogene- ity induced in this phase by the special treatments which it has undergone, is not very much modified after transformation to the phase stable at lower temperatures. Thermal hysteresis is frequently observed and is sometimes very intense in the transformations with which we are now concerned. It may sometimes substantially modify the course of the phemomena indicated above. But our actual knowledge in regard to the special effects of hys- teresis is too fragmentary to be taken into account In practice. However, it is very probable that a precise study of this problem might illuminate the nature of many occurrences now known and be utihzed in steel making and heat treating practice without being able to place them rationally in the outline diagrams of binary equilibria as experimentally determined. It is not possible, therefore, to present a complete graphical study of the group of phenomena indicated in the last paragraph. It is yet a blank page to be written in the future complete treatise on heat treatment. 45. Our knowledge regarding carbon is less incomplete on such points. As has been seen in the case of solidification of liquid metal into crystalline solid, it was unnecessary to take into account the presence of regions of insolubiUty when examining the actions due to any one of the elements occurring in machine steels up to the maximum concentrations there found. As a consequence, the partial or complete systems of any of these elements with iron (carbon included) could be represented during solidi- fication by means of schematic diagrams perfectly similar to each other. On the contrary, in the case of the transformations due to aUotropy in iron, areas appear in the iron-carbon equihbrium diagrams which denote a limited solubility or complete insolubility. Such insoluble areas, so- called, may be again found although substantially modified in the majority of the other more complex systems. Supposing the reader to be acquainted with the latest information about the equilibrium phenomena proceeding in an iron : carbon system possessing perfect homogeneity of concentration in all the mixed crystals of each series, I shall refer my remarks to that part of the familiar dia- gram which contains the low-carbon concentrations we are interested in, selecting among the alternative forms recently proposed by various investigators, that sketch which seems best to represent the occurrences which must be analyzed. Figure 9 reproduces this complete equilibrium diagram extended to the maximum carbon concentrations commercially possible. Therefore, SOFT AND MEDIUM STEELS 57 it includes cast irons as well as steels. As already remarked, it is not necessary to stop and explain the meaning of the various elements of Fig. 9. It is reproduced only in order to establish clearly which one of the proposed variations I intend to discuss. Perhaps it is superfluous Fig. 9. — Iron: carbon equilibrium diagram. to note that, inasmuch as we are concerned only with steels with low and medium silicon content, we may restrict our examination to the metastable iron : cementite diagram without extended reference to the stable iron : graphite system. 58 HEAT TREATMENT OF Let us Start, therefore, by investigating briefly the phenomena which take place during the more or less rapid cooling to room temperature of binary iron : carbon alloys ranging from zero to about 0.9 per cent, in carbon. In Fig. 10 is reproduced an enlarged portion of that part of the diagram containing carbon concentrations up to the eutectoid percentage, and covering the entire interval of temperature within which takes place all transformations of the three allotropic forms of iron — alpha, beta and gamma. It does not take into account some of the hypotheses recently proposed, such as Benedick's "ferronite," because the required experi- mental evidence certainly is far from complete. Fig. 10. — Transformation range in hypoeutectoid steels. Excluding the possibility of any heat treatment Hable to materially interfere with or arrest the transformation of beta into alpha iron, such as a drastic quenching from a temperature between A and 790°C., and considering only such slow cooling as to permit the complete allotropic transformations, it is easy to see that the effects of the two transitions which take place in the areas BEC and DCF simply accumulate upon each other. As a consequence, one may greatly simplify the considera- tions of general quahtative character without materially affecting the soundness of the conclusions, by merging the two above mentioned areas into one. By supposing that Oi is added to 02 we obtain a simplified diagram SOFT AND MEDIUM STEELS 59 (Section 46) in which are represented only the effects of the complete transformation. It would be easy to demonstrate that the conclusions reached for this case are entirely vahd for the complete diagram of Fig. 10. However the fundamental ideas upon which such demonstrations should be founded are so obvious that it is not necessary to linger upon them. To establish them it is simply sufficient to compare the method of crystaUization as sketched in the two diagrams. Therefore, the simphfied diagram Fig. 11 is to be used. In order that the diagram may be more legible, the distance IG has been increased much more than its correct length as determined by experiment. Thus, the width of that portion GH of the area of insolubility existing between alpha iron and iron carbide has been narrowed so that in the figure the concentration of carbon in saturated alpha iron appears a great deal larger than its actual value. It will be more clearly seen further on that this modification evidently cannot alter the value of a quaUtative discus- sion in any^way. CHAPTER VII THE BEGINNING OF SECONDARY CRYSTALLIZATION IN HOMOGENEOUS AUSTENITE 46. Let us, therefore, investigate the behavior of an iron : carbon alloy of composition V (Fig. 11) at a temperature L at which the iron is totally in the gamma form. Let us suppose that this alloy has been maintained at a sufl5.ciently high temperature during such a lapse of time as to allow the carbon concentration to become uniform in all points of the metallic mass. If we allow such an alloy to cool more or less slowly, the course of the phenomena which take place at temperatures above IGH (690°C.) will be entirely similar to that already studied for the liquid : crystalline trans- formation. Attention will also be restricted for the moment to the space occupied by a small crystalline mass forming around a single germ of transformed alpha iron. For this elementary solid we will also obtain in the diagram first, a curve of minimum concentrations of carbon in the alpha solid solution (curve d) ; second, a curve of the average composition of the alpha phase which has formed at each temperature (curve c) ; and third, curve h of maximum concentrations which coincides with what physical chemists call the "solidus of the transformation 7— > a." In regard to these three curves we may repeat the same observations already made in Sec. 16 to 19. The only difference between the two cases is quantitative. The velocity of diffusion of carbon in the alpha solid solution, as well as in gamma solid solution or austenite, at the temperatures at which the y-^a transformation takes place, is a great deal less than that existing at temperatures near melting. Consequently, given equal dimensions of the crystallites, it is necessary to have much slower coolings at temperatures below 900°C., and much more drastic reheatings and quenchings in order to obtain substantial displacements in the location of curves c and d. However, it should be noted that the hnear velocity of diffusion is not the only factor which must be reckoned with in any treatment to produce homogeneity in a steel. For instance, given quite small crystals of ferrite and cementite, equaHzation may be more rapid even at tem- peratures just below the transformation than at a considerably higher temperature, which produces coarse austenitic grains. This is for the reason that the distance between the points corresponding to the extreme concentrations of carbon in each crystalline element or colony is a great deal smaller in secondary crystallization. This distance factor often 60 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 61 reaches such an influence that, as remarked above, equalization occurs quicker at low temperatvu*es than at higher. When cooling reaches the eutectoid temperature IGH (690°C.), different events may take place depending upon differences in initial composition of the alloy and velocity of the cooling. These can be divided in two groups based upon the total composition of the alloys considered. The first group comprises those cases in which the line representing the average composition of the alloy is represented by a concentration-vertical o»o° D L N \ if N C^?^ ?-^ . i^v7**-\ s=^ ^F^5 J^ ^^^ fe •'^ L i.>^?^ r»^ ^a 1 l^^v^ -^--1^ •^"j-^ ^ JM-- i^l^^a^^i^ ^--f- •^Cti^ ^ II K ft Fig. 11. — Precipitation of carbon from ferrite. such as LL' which intersects IGE at the left of the point (?, corresponding to the carbon content of saturated mixed crystals of alpha iron at 690°. In this case the manner in which the transformations occur and the final result vary a great deal, not only quantitatively but quahtatively, with the variation of the cooling velocity. In fact, if the temperature of the real end of the transformation — corresponding to 0, the intersection of curve c with the vertical LL' — ^is higher than the eutectoid temperature IGE, we again have a system entirely similar to those which have been studied under primary crystal- Uzation. When the transformation is complete, each transformation germ will be surrounded by a structural element comprised of sohd solutions of 62 HEAT TREATMENT OF carbide in alpha iron, in which the concentration of carbon will vary from a minimum value V, at the center, to a maximum T, at the periph- ery of the crystallite. The only difference between the present case and the one studied in Part I Kes in the fact that the curve TR, which is the right boundary of the band of variant concentration, may meet the curve GQ, which is the left boundary of the region of insolubility of carbide in alpha iron, and thus require the formation of a new phase. However, with our steels, this intersection does not occur in practice, since the two curves TR and GQ approach verticals passing through T and G; or if it does occur, it is at temperatures so low as to prevent the theoretical transformations to go forward appreciably. 47. It is clear that the position of the point'O depends upon the initial composition of the alloy LL' and the velocity of cooling (that is to say, upon the position of curve c). If for a given cooling velocity the initial concentration of carbon is high enough (although remaining below the value G), or else if for a given initial carbon concentration the cooling velocity is high enough to keep curve c diverging considerably toward the left of the solidus b,'- it will be possible for the point to fall below the eutectoid temperature IGH. Let us suppose, for instance, that the same alloy LL' be cooled more rapidly than before so that the curve of average composition of the a phase assumes the position c' to the left of the preceding position c. In this case the temperature of the end of the transformation will be the one corresponding to the point 0'; in other words below the eutectoid temperature IGH. When this occurs it is evident that at the eutectoid temperature (690°C.), the periphery of the structural element under examination, will be constituted of saturated solid solution in alpha iron of composition G, in equihbrium with, a residuum of solid solution in gamma iron of com- position H. A further subtraction of heat from the metalic mass will cause this residual gamma iron to be transformed at constant temperature or isothermally into the eutectoid of the same total composition (H). Such decomposition results in an intimate mixture of small crystals of two constituents stable at lower temperatures. One phase consists of parti- cles of composition G and the other of particles of composition represented by the right hand end of the eutectoid horizontal. This horizontal repre- sents the top width of a band of insolubility of iron carbide in alpha iron, or a region of heterogeneous equihbrium. It is known that the right hand border is the ordinate representing the composition of cementite (iron carbide, or FesC) or a line near by representing a solid solution of alpha iron in cementite. It is useful to keep in mind the fact that the composition H of the 1 See Sec. 19. SOFT AND MEDIUM STEELS 63 eutectoid thus formed is practically independent not only of the initial composition of the alloy, but also of the cooling velocity. The signifi- cance of points P, U and H need not again be defined. Given their evident meaning, it may be easily demonstrated as has been done be- fore in Sec; 13 that the following relation holds when the transforma- tion is completed. Mp_UH ,.. M. ~ PU ^ ^ where Mp is the mass of alpha mixed crystals of average composition P, and Me is the mass of eutectoid of composition H. From this equation it is seen that the proportion of eutectoid will be greater the more the curve c' is shifted toward the left; namely, the more rapid has been the cooling. From the preceding remarks it also follows that it is always possible to select such a cooling velocity as to prevent the appearance of eutectoid in alloys of carbon content less than IG. Other rates of cooling will cause the appearance of a proportional quantity of eutectoid whose ratio may vary from zero to a maximum equal to SL" -^ UH. If, on the other hand, we start from an alloy in which the initial con- centration of the carbon be higher than that of the saturated mixed crystals G, it is clear that material of eutectoid composition will always be present. Each variation in cooling rate will cause the ratio between the quantity of alpha mixed crystals and the quantity of eutectoid to vary ; but it can never be slow enough to obliterate the pearlite completely. The preceding argument neglected the fact that the width of the insoluble band or the field of heterogeneous equihbrium is subject to change as the temperature drops below the eutectoid temperature IGH. These variations are partially represented by the inclination of the curve GQ away from the vertical but in reality are of a size very much smaller than any phenomena which cause changes in the characteristics of the alloy noticeable in practice. Besides this, the variations in solubility of cementite in alpha iron occur at temperatures where the velocity of diffusion is so small as to render their effects practically negligible in the great majority of cases. 48. The reasoning outhned in the preceding sections, has been based purely upon a simple extension of the conclusions which were obtained when studying the formation of a solid solution from a molten niass. Now it is true that such an extension is wholly justified, within the ap- proximate limits which are of interest in heat treating practice, when the train of thought is restricted to the nature of the constituents appear- ing in the transformed alloy. But it is no longer even approximately justified, when we wish to reach conclusions concerning the forms and the positions assumed by the phases produced by the decomposition of 64 HEAT TREATMENT OF austenite, nor the relation between the quantities of these structural, elements. This is due essentially to two series of phenomena, whose origins lie in entirely similar situations and whose consequences reinforce one another. Therefore they cannot be s"tudied separately. We know from the very definition of equilibrium that all the conclu- sions reached when studying the separation of a solid solution from a molten mass are valid only in case diffusion is rapid and substantially equalizes all heterogeneity in each phase. In practice, this takes place with a sufficient approximation to reality that the hypothesis may form a working basis. In other words, processes of diffusion due to differences in concentration do take place in the mother liquor during all that interval of temperature within which crystallization really occurs, with velocity sufficient so that the concentrations of the various components of the liquid phase may be considered as practically uniform at each instant in all the points of that phase itself. On the other hand, we also know from a somewhat detailed study that such an ideal homogeneity never occurs in solid solutions. In fact it has been seen how it is always necessary to draw a band of variant composition down through the field representing mixed crystals to visual- ize the amount of residual heterogeneity of this phase at various tempera- tures and after various heat treatments. Therefore, in the study of new phases resulting from allotropic transformations in iron, it is not only necessary to take into account the very slow velocity of diffusion at these relatively low temperatures, but also the additional heterogeneity inherited from the primary solidi- fication and uneffaced in the subsequent history of the metal. 49. With this in mind, I will restrict my remarks for the time being to the examination of the effects of retarded diffusion upon the forms and positions of the various constituents born in the solid iron-carbon alloy. This alloy, however, will be thought of as having undergone that heat treatment, very rare in practice, which will thoroughly equalize the con- centrations of the various components in solid solution in gamma iron. It is evident that the composition of such austenite, stable at the higher temperatures, may be represented by a point which traces a vertical line in the diagram as the temperature of the system lowers. Should completely homogeneous austenite occur in practice the identi- cal considerations already developed when discussing the crystalliza- tion process could be repeated for it, as far as the beginning of the y^a allotropic transformation is concerned. Again, during a slow cooling, the transformation point would be reached simultaneously in all parts of the mass. Separation of the phase stable at the lower temperatures would also take place in manner entirely similar to that of solidification. This secondary separation of alpha solid solution containing a lesser SOFT AND MEDIUM STEELS 65 concentration of the component elements than does austenite at the same temperature would start around centers of crystallization whose distribution depends directly upon the nature of the decomposing solid solution and upon the cooHng conditions of the metallic mass. Let us suppose for the moment that the secondary transformation might in reality start exactly in the way just indicated, and let us examine briefly how it should proceed further. Fig. 12. — Early stages in the decomposition of a particle of homogeneous austenite. When the iron : carbon alloy of composition LL' (Fig. 12) during uniform cooling has reached the temperature T of incipient transformation, mixed crystals will begin to form around certain nuclei. After a very short time t the temperature will drop further through the small interval^ T — Ti. ' In Fig. 12 the interval TTi has been taken very large as compared with the total interval of crystallization, with the aim of making the diagram clearer. The same remark may be repeated for the following interval T1T2. It is evident that this distortion or magnification does not in any way vitiate the accuracy of the con- clusions I shall develop further on. It is equally evident that it would not have been practicable to represent minute distances to exact scale. To do so, I should have taken the partial intervals so small as to render them practically invisible. 5 66 HEAT TREATMENT OF Upon reaching the temperature Ti, small masses of alpha solid solu- tion will have formed at those crystallization centers. In these minute masses the carbon concentration will have an average value E2, a value lying between E and Ei. Its composition will vary from the center to the periphery from a value corresponding to a point E^ (situated to the right of E, but very near to it) to the value Ei. At the same time the carbon concentration in the neighboring regions around each nucleus increases till it reaches the value Fi, corresponding to equilibrium be- tween austenite and the alpha crystals at temperature Ti. I have already discussed the reasons and the mechanism of this action in Part I. 50. If, now, we endeavor to determine the quantity of alpha mixed crystals which have separated from the gamma solid solution at tempera- ture Ti, we will see that it depends directly upon the diffusion velocity of carbon in austenite at that temperature. It has been postulated that diffusion is slow, a supposition which corresponds precisely to reality in the case now under study. Hence homogeneity of carbon concentration throughout the mass of austenite is impossible of attainment in the intervals of time used in any possible cooling, however slow. It is clear then, that at the end of the minute time interval t, the residual austenite will be composed of two distinct parts; as follows. (a) First, a heterogeneous region immediately surrounding the germs of alpha iron first to appear. In these regions the carbon concentration will vary from a maximum value Fi for the shell in immediate contact with the alpha solid solution, to a minimum value F corresponding to the external surfaces of the region which has been disturbed by difiFusion. It is clear that the distance of that surface from the alpha crystals, and therefore, the radius of the volume in which the carbon concentration is higher than F wiU be greater, the higher the velocity of carbon diffusion, other things being equal. In other words, given a certain velocity of diffusion, the volume of the heterogeneous portion will be the greater the longer the time t during which the temperature T falls to the temperature Ti. (b) A second part, formed by all the remaining mass of the gamma mixed crystals interposed between the regions defined in (a). In this portion in which the effects of the carbon accumulation in the central zones have not been felt, the carbon concentration still has its original value F. Even though that part of the austenite lying in contact with the alpha crystals has a carbon content Fi precisely equal to the value in equilibrium between mixed crystals of the two series at temperature Ti, the entire mass of untransf ormed gamma phase has an average value F^, smaller than Fi, but greater than F'. SOFT AND MEDIUM STEELS 67 The ratio between the quantity Ma of the alpha mixed crystals (ferrite) separated at the end of the time t to the quantity My of the resid- ual 7 mixed cr ystal s (au steni te) is expressed by the inverse ratio of the two segments E^F' and F'Fi. Ma ^ WWi My EJF' Indicating with M the entire mass of the metal under consideration, Ma ^ FFj M — Ma EF2' The quantity of ferrite of average composition E^ which will be separated from austenite after the time t at the temperature Ti, will therefore be the smaller, the nearer the point Fi will be to the point F'. After all, this is self-evident, since F^ is the average composition of the residual austenite and F' is its initial composition. Point F2 is moved toward the left and nearer to the point F', as the amount of diffusion becomes less. At the same time it is also evident that the quantity of alpha mixed crystals separated at temperature Ti will be larger, the more completely carbon has diffused throughout the residual gamma mixed crystals. Naturally, the tendency will be for carbon to diffuse through the solid solutions from the high-carbon interior- contact with carbon-poor ferrite, outward into the surrounding'austenitic material of original and medium concentration. Especially we see that the proportion of ferrite in the metallic mass will increase, if the temperature Ti be kept constant for a time. In this investigation we should also take into account the disturbances arising from the fact that similar diffusion processes take place in the car- bon of the alpha mixed crystals contemporaneously with the diffusion of carbon in the austenite. But the study of these disturbances would im- mensely complicate the investigation of the process with which we are concerned. Nor, on the other hand, would it present any practical inter- est, because the amount of carbon diffusing in austenite is so far greater than in ferrite that the effect of the latter disturbance is practically neg- ligible. This lack of proportion between the importance assumed by the phenomena occurring in the two solutions is due to the greater lineal velo- city of diffusion of carbon in austenite, other conditions being equal, as well as to the fact that both the absolute value and the differences of the carbon concentrations in ferrite are far below those in austenite. 51. Let us now see how the allotropic transformation y^a progresses when the temperature continues to fall below Ti, reaching T2 after an- other time interval h. Both at temperature Ti, reached after time t, as well as at temperature T2, reached after the additional lapse of time h, part of the unaffected 68 HEAT TREATMENT OF austenite which in the preceding section has been described in (6), is supersaturated with iron in comparison to the theoretical equiUbrium with the alpha mixed crystals. Supersaturation of that austenite also increases in amount with each decrease in temperature. From this it follows in accordance with the well known properties of supersaturated solutions, that step by step as the temperature of the systein is falling, the tendency grows toward the formation of new crystallization centers in that austenite preserving the initial concentration L. This much established, it is clear that during the interval of tempera- ture between Ti and T2 — which is supposed to have been traversed by all parts of our system at the same speed during time ti — the allotropic trans- formation 7 ^ a develops in three distinct ways : (a) Around ferrite of average composition E2, separated during the preceding interval of temperature, will form a new layer of mixed crystals of the same series, in which the carbon concentration at the external sur- face will reach the maximum value E3. Contemporaneously, diffusion of carbon inward from the periphery of each of the ferrite crystallites formed in the second short lapse of time will have increased the carbon concentra- tion at their center. As a result, the average carbon content of the new ferrite masses formed in this way around the original nucleus will be represented by a new point Ei situated at the right of E2 but at the left of £'3. At the same time the austenitic regions surrounding each mass of ferrite described immediately above has also increased in carbon content, reaching the maximum value F3 at the surface of contact between the gamma and alpha phases. Simultaneously the thickness of, the shells of austenite partially en- riched in carbon and surrounding the same ferrite crystal has also been increased as a result of the diffusion of carbon from the interface between phases back into the solid solution, a diffusion rendered more intense by the increase of the maximum concentration of carbon from Fi to F3. In regard to the ratio between the entire mass M of the part of our system under consideration and the mass M'a of the ferrite separated therefrom at temperature T2 around germs appearing at the start of the transformation, we may repeat the considerations made in reference to the previous interval of the same process. Thus, indicating with Ft the point representing the average composition of the 7 phase at temperature T2 (at the end of the second short time interval h) we may estabhsh the following relation: M'a W¥. M — M'a EiF" In regard to the position of the point Fi, the same remarks already made in reference to the corresponding point F2 apply (page 67). SOFT AND MEDIUM STEELS 69 (6) In that part of the austenite supersaturated in iron, in which the carbon concentration has remained unchanged at the original value L on account of the incomplete development of the diffusion processes, new centers of crystallization will form during the second minute period of time h, around which in turn new masses of ferrite will gather. The formation of alpha mixed crystals of this new series will proceed in a manner similar to that of the previous alpha crystals formed around the nuclei appearing in the former interval of temperature T — T\. But it is clear that, at the end of the second interval of time a, when the concentration in all parts of the initial phase is practically uniform. It has a well defined significance only when we consider the process of the allotropic trans- formation y — >Q! going forward around only one center of crystallization. B. Curves representing the concentration of carbon in the phase stable at the lower temperatures : (a) Curve of the minimum carbon concentrations in the a phase (line E Eo, Fig. 12.) Due to the small diffusion velocity of carbon in ferrite, this curve very nearly approaches a vertical, and in practice it can be considered as such. From what we have seen in the previous paragraphs, it is clear that it cannot have a precise significance unless it be considered in relation to the 7— >« transformation proceeding around only one center of crystallization. 74 HEAT TREATMENT OF SOFT AND MEDIUM STEELS (b) Curve of the average carbon concentrations in the a phase (curve E EiEi. . . .Em, Fig. 12). It has the same characteristics of the corresponding curve marked out in the diagram representing soUdifi- cation of molten steel. Note the important fact that a given reheating will cause a much smaller displacement of curve E E^Ei. . . .Em than in the corresponding line in the austenitic field (see Sec. 20 and following). Again this is a consequence of a low velocity of diffusion at temperatures near the critical. This curve has a well defined significance both when considering the entire a phase, as well as when considering only that portion formed around a single nucleus. (c) Curve of the maximum carbon concentrations in the a phase (line EE'E'iEiE'iEs, Fig. 12). This hne also has the same charac- ters as the corresponding curve studied in the case of solidification. It has also a definite significance both when the entirety of the a phase is considered, as well as when one takes into account only that particle of ferrite which has separated around a* given center of crystallization. CHAPTER VIII COMPLETION OF SECONDARY CRYSTALLIZATION OF UNIFORM AUSTENITE 57. Taking as a basis the principal elements of the binary diagram thus defined, it is not difficult to form an approximate idea of the course of the various phenomena when the system passses through the entire transformation range. It would not be so easy to determine in a quantitatively accurate way each step of the transformation for the various types of crystalline individuals forming each of the two phases, and so construct a precise equilibrium diagram. But inasmuch as such determinations would be superfluous for the practical aims in view, I deem it unnecessary even to mention the efforts which have been made from this angle, and of those lines of research which may be developed further in several directions. With the lowering of the temperature, it is clear that the 7— >a transformation will continue in the same manner and with the same mechanism which has been analyzed for the first two minute intervals of temperature T to Ti and Ti to T^. This is true for those regions of the y soHd solution surrounding each crystalline element of the a series^ in which the concentration of carbon due to the incomplete development of the process of diffusion has reached values higher than the original one and yet the diffusing carbon has not invaded the entire mass of the residual y phase. Instead, the development of the phenomena undergoes some modi- fications from the moment when diffusion extends to the supercarburated regions around the a crystals, and the same regions are multiplied in correspondence with new crystalline individuals continuously appearing in the residual austenite, causing the entire mass to be somewhat enriched in carbon. Evidently at that time the last trace of uniform 7 sohd solution disappears, such as was described in Sec. 61, (6), and in which the carbon concentration constantly remained the same as the initial composition of the 7 phase. From this point on during further cooling, it is evident that the curve of minimum carbon concentrations in the 7 crystals [see Sec. 56, A, (a)] cannot further coincide with the ordinate corresponding to the original concentration of carbon in the homogeneous 7 phase. Instead, the curve gets farther and farther away to the right as the temperature falls. In Fig. 13 is marked the course of this curve in ABC, supposing that the » See Sec. 51, (c), 75 76 FIE AT TREATMENT OF last residuals of y solid solution of the original concentration (AB) dis- appear at temperature T^. Although it is obvious after what has been said in Sec. 56 apropos of the meaning of each of the curves of the diagram, it should be remembered that the diagram in Fig. 13 is to be understood as referring to the concen- trations of the entire mass of each one of the phases which are present in the system at each instant. Starting from the moment in which the last portion of the y sohd solution of concentration AB disappears, it is just as clear that the course of the curve of the average concentrations of the y phase [see Sec. 56, A, (b)] shall undergo a modification, due to the interruption of phe- FiG. 13'. — Complete decomposition of massive austenite. nomena defined in Sec. 51, (b); phenomena of crystallization which, with equal intensity of the accompanying thermal phenomenon, would cause the smallest possible subtraction of carbon from the y phase. In fact the a phase which separated as a nuclear point at the birth of each one of these new crystalline individuals always contains the lowest carbon con- centration (Fig. 12, E) even when the process in question starts at a temperature much lower than that of the beginning of transformation of the system studied. ^ Therefore, it is easy to understand that when this gi'oup of phenomena stop, the fact must bring in consequence an increase in the quantity of carbon subtracted from the austenite caused by a given variation of the temperature of the system. This increase 1 Because of a local development of heat accompanying the transformation, reheating the region about the nuclear point substantially to temperature T. SOFT AND MEDIUM STEELS 77 IS in comparison to the quantity of carbon, which an equivalent variation of the thermal state of the system removed into the same 7 solution in the previous time interval during which that earher group of phenomena continued to exert its effects. Let us represent with ADE in Fig. 13 the thus defined course of the curve of average carbon concentrations in the 7 phase. 58. As a result of the very definition of the two curves ADE (average composition of the entire 7 phase) and FGL (average composition of the entire a phase), it is evident that at each temperature To, below A, the relationship Ma OH My ~ W exists between mass Ma of the a phase separated from the system, and the mass My of the 7 phase. The proportion between f errite and austenite evidently does not depend exclusively upon the velocity of diffusion of carbon in the phase stable at the lower temperatures, as it did in the case of crystallization of molten steel, but it depends also upon the velocity of diffusion in the phase stable at the higher temperatures. After all, this resulted from the considerations set forth in the previous paragraphs, and has been developed in Sec. 51, (a), as it applies to a particular portion of the a and 7 phase. In other words : during the solidification of molten steel the instanta- neous proportion of the two phases may be derived merely from a curve corresponding to FG of Fig. 13, yet during the 7 — > a transformation the same proportion is expressed by the relations existing between the course of same curve FG and that of another curve DEN. Finally, from what has been said in a previous chapter regarding solidification of molten steel, it is evident that the 7 — > a transformation will be completed at temperature Tp, at which point curve FG cuts the vertical AOL. 59. Having established these fundamental data, let us resume observ- ing the actual course of the phenomena which more especially interest us at present in view of the practical purposes in view: namely, those phenomena which accompany the closing stages of the reversible trans- formation in a pure binary hypoeutectoid steel. In Sees. 46 and 47 I have already briefly examined such a transforma- tion taking place in materials whose average analysis falls without the limits of the a solid solutions. However, the considerations developed in the last few pages now allow the study of the 7^0; change in the way it really happens in practice. Previous to this time we were obliged to work under the restriction that diffusion in the phase stable at higher temperatures (austenite) must be rapid and complete, since we lacked the necessary data to pass beyond this hypothetical condition. 78 HEAT TREATMENT OF Figure 14 contains merely the essential elements denoting the course of an actual transformation of initially homogeneous y mixed crystals. There is again found curve AB, the average composition of the y phase [Sec. 56, A, (b),] and curve CD, the average composition of the a phase [Sec. 56, B, (6)]. Suppose that the first series of a mixed crystals (that is to say, that part of the a crystal containing the lower percentage of carbon) is saturated with carbon at temperature Ti. Let F be the carbon concentration corresponding to this saturation. At temperature Ti, therefore, the region of insolubihty or of heterogeneous equilibrium (Italian, lacuna di miscibilitd; German, Mischungsliicke) in the a phase wiU extend from the point F to the right, reaching a point H' which in I T \ ^v CV- ^^ w 1 \ 1 A. 1 \^ 1 \ \ X » X 1 X \ X \ > \ > ^ X T, i — -jj.— 1 ""'/«\ 1 \ M CSj oCi Fio. 14. — Outline diagram at transformation range. Fig. 14 falls outside the sketch. It is known that the point H' cor- responds to pure cementite FeaC (6.67 per cent. C) in the special case of steel, now being studied. Any iron : carbon alloy with a carbon concentra- tion falling between F and Hi on arriving at Ti as 7 solid solution will separate on further cooling into conjugate solutions ai and a^ — two distinct series of crystals. Let us now see how the transformation 7^ a proceeds when the tem- perature of an alloy AE is lowered to Ti. Once this temperature is reached, it is clear that the maximum concentration of carbon cannot further increase in the saturated 7 mixed crystals. SOFT AND MEDIUM STEELS 79 To be more precise, once a temperature slightly below Ti is reached, the maximum concentration of carbon in the y crystals will have grown a small quantity above the value G. The point representing the compo- sition of this phase will have moved along the curve GL, a prolongation of the hquidus I AG. It is clear that at such a moment the system is undercooled and in a state of metastable equilibrium, being supersatu- rated in respect to exi phase. However, after a small lapse of time (whose duration changes with the general conditions of the system, and with the velocity of cooUng, but, which at any rate, is always very small) the undercooled state comes to an end; a certain quantity of the a^ phase (in our case cementite of composition H') separates abruptly with the evolution of latent heat so that the temperature of the mass increases suddenly, returning to Ti. From this point further abstraction of heat from the system causes a continued separation of the 0:2 phase along the curve GM, and thus the condition of metastable equilibrium returns, in this case the mother soHd being supersaturated with ai. When this undercooling reaches its limit, new quantities of the cti phase (ferrite of shghtly variable com- position in accordance with curve FN as concentration varies with temperature) separate abruptly with evolution of latent heat until the temperature of the alloy has again moved up to Ti. As heat is withdrawn from the system, the process of alternate separa- tion of small quantities of the ai and a2 phases continues. This is the mechanism of formation of the "eutectoid" structure, in steels known as "pearlite," an aggregate of minute portions of the ai sohd solution (ferrite), alternating , with small elements of the a^ phase (cementite). 60, I would not have stopped to describe in this brief and not rigidly accurate manner the mechanism of the formation of the eutectoid G were it not that I needed to clearly fix in the reader's mind the position of the points and hues in Fig. 14 representing various phases of the process. Now let us see how the concentrations of carbon in the eutectoid will vary during its formation. In the first place, our special point of view permits us to concern ourselves only with variations in composition of the ai constituent, inasmuch as experience shows that the second phase, a^, separates as a practically pure compound and, therefore, with a practically constant composition H', corresponding approximately to the formula FcsC. It also follows from what has been said immediately above that the eutectoid will form only in those regions of the 7 phase in which the carbon concentration has reached the value G. In fact, it will start along the interface between the «i mixed crystals (formed in the interval TTi) and the residual 7 solid solution. Concentrations F and G are in contact at this surface but the carbon concentration in untransformed solid solu- tion diminishes rapidly with the distance from each ai nucleus. 80 HEAT TREATMENT OF It readily appears how supersaturation of the 7 phase with constitu tent ao (cementite) will be reached at a later time, the more rapid the decrease in carbon concentration when one passes from the initial ai: a? interface into the layers of the solid solution contiguous. Consequently, the pro- portion of ferrite (more strictly the a solid solution of variable concentra- tion denoted by the curve NF) in the eutectoid will be greater, the steeper the carbon gradient existing at the instant of first formation of the eutectoid. Even when the formation of the eutectoid is more or less advanced the same apparent excess of ferrite will be produced if the carbon concentration rapidly decreases in the material immediately surrounding the already separated pearlite toward the layers of the same 7 solution farther away from the external surface of those nuclei. Fig. 15. — Peiirlite showing large segregation in ferrite. But the indicated variation of carbon concentration in the material surrounding a given crystallization-center of a. phase is the less abrupt, the earlier such nucleus has formed in the history of the transformation y:^a. This is because the time during which the processes of diffusion have had an opportunity to take place has been greater, and has rendered that variation less abrupt, and at the same time has started at higher temperatures at which the intensity of dispersion is greater. Fi'om this it follows that — while in practice the carljon analysis of the eutectoid pearlite always remains approximately constant and close to that represented by point G — the ratios between the quantities of its two structural constituents may appear under the microscope to undergo local variations, depending upon the different manner in which the dif- SOFT AND MEDIUM STEELS 81 fusion of carbon proceeds. This phenomenon in turn is bound to the occurrence of undercooHng accompanying the y^a transformation, and to the frequency of the a nuclei born in the steel. These observations explain the irregularities and seeming abnormali- ties which can be almost constantly detected in the pearlite constituents. Figures 15 and 16 show two examples. When considered in the light of the remarks set forward in this section, they are clear enough by them- selves so that it is superfluous to add any explanation. 61. Having thus examined the character of the eutectoid formed by the special conditions of diffusion supposed to hold for our sj'stem, let us briefly note how the transformation of the entire system proceeds. It will be noted that at temperature Ti the system has become invariant Fig. 16. — Pearlite showing large variation in texture. due to the appearance of a new phase, an, and when we continue to ab- stract heat from it, the temperature will remain constant until the y phase disappears. With this end in view, again observe the distinction between the three different regions of the y phase, described in Sec. 51 as being those in which the y<^a transformation occurs indifferent ways, and discover how the resulting eutectoid is modified by them. For the sake of brevity I shall indicate each one of the three regions with the letters (a), (b), and (c), respectively, used in Sec. 51 without repeating the definitions at length. (o) In region (a)— immediately surrounding the a mixed crystals formed at temperatures above Ti— the transformation fails to increase 6 82 HEAT TREATMENT OF the size of the ferrite crystals during the sojourn at temperature Ti. It therefore acquires a completely new character, and results in the forma- tion of masses of eutectoid of average composition G (Fig. 14). We have already outlined the characteristics of this new transformation in the preceding section. (6) In region (6) — characterized by a carbon concentration sub- stantially unaltered from AE, the original analysis of the system — the transformation proceeds in an absolutely identical manner to that already studied in Sec. 51 for the same region. All the observations made there apply to this transformation which takes place, by the way, in an under- cooled system. It will cause a gradual sidewise displacement of the point B, representing the average composition of the residual y phase, along the horizontal TiB G from B toward G. (c) In third region (c) — ^immediately surrounding region (a) and -possessing a variant carbon concentration which analyses greater than that of region (6) — the transformation y^^^a proceeds in the same way described in Sec. 51 when studying the corresponding region. Here the transition also causes a movement of the point B along the eutectoid horizontal toward G, as explained in Sec. 51. From definition of the region (&), and from what has been said in Sec. 57, it is clear that at a certain instant of time that region must dis- appear. Thereafter the transformation proceeds according to the two schemes peculiar to the remaining regions (a) and (c). It frequently happens also that region (6) has finally disappeared at a temperature above the eutectoid temperature Ti. Naturally, the entire eutectoid transformation then takes place according to the manner peculiar to (a) and (c) . In this case the ratio between the mass of the a mixed crystals actually separated in one of the reigons (6) or (c) and the mass of the a crystals which would have separated had there been total homogeneity in each phase, varies with each change of frequency of crystallization- germs. It is clear that one cannot further use considerations based upon the concentrations of the various elements in each phase when attempting to derive the ratio existing between the quantity of pearlite and that of ferrite which will constitute steel of composition AE (Fig. 14) when its temperature has lowered beneath the eutectoid temperature Ti. De- ductions based upon composition are valid only when the system is in complete equilibrium so that the composition of each phase is constant in all its parts. Nor can we start with an analysis based upon the average composition of the individual phases, because that reasoning could apply only when all the particles of the second phase may always be derived from the first one under the same conditions. Such a con- tingency was discussed and accepted in Part I because the liquid metal existing in equilibrium with solid austenite during the mushy stage SOFT AND MEDIUM STEELS 83 is substantially uniform in all parts. But from the deliberations of the previous sections, it clearly results that such a state never happens when austenite breaks up at Ar ranges. In order to deduce some positive indications even of a generic charac- ter from those same considerations in addition to the previous negative conclusion, let us briefly recount the conditions under which the eutectoid transformation is completed. It has already been seen that during the 7?^ transformation now being studied various reasons concur in producing displacements of point B toward the right. From the very definition of this point it is clear that it cannot be thrust in that direction beyond point G. On the other hand, it is known that transformation 7— >q: takes place at temperature Ti after the manner of a stable system within those regions in which the car- bon concentration in the 7 phase has the exact value G. In the other regions of the heterogeneous mass it occurs as the breakdown of a meta- stable state. But even in these undercooled regions, the transformation has always been accompanied since the arrival at Ti by the formation of eutectoid austenite (concentration G). It is therefore evident that at the end of the process the last residue of 7 phase will necessarily have a carbon concentration equal to G. Hence at that time point B will have reached position G. It may be superfluous to add that at the end of the eutectoid trans- formation the same point G will then become the limiting position of the intersection between horizontal TiG with curve BC of Fig. 13, repre- senting the minimum carbon-concentrations in the 7 phase. Point D also moves toward the right during the eutectoid transforma- tion. But, from the various observations previously made, especially in the end of Sec. 51 and Sec. 60 and 61, the amount of this perturbation and its consequent end-point depends upon the velocity of cooling, upon the mean distance between the alpha nuclei, and upon the importance of the undercooling which accompanies the formation of the new crystaUiza- tion-germs in the different regions of the 7 solid solution. If we admit as a sufficient approximation that the composition of the eutectoid is represented by the point G, the same ratio between its mass and that of the exi mixed crystals existing when the eutecoid transforma- tion is completed, is equal to the ratio between the distance DE and the distance EG (Fig. 14). It is therefore clear that all the causes which produce a right-hand movement of the location of D also produce a de- crease of the proportion of the eutectoid in comparison with ai mixed crystals, and vice versa. In addition to this, however, it is necessary to remember the remarks previously emphasized that the transformations which take place at tem- peratures below Ti in slowly cooled steels proceed with a velocity so small that in practice they can be totally neglected. CHAPTER IX SECONDARY CRYSTALLIZATION OF NON-HOMOGENEOUS AUSTENITE IN THE BINARY IRON : CARBON SYSTEM 63. In the study of the transformations which the binary iron: carbon alloy undergoes during cooling subsequent to its complete solidification, it has formerly been supposed^ that the austenite has been maintained at a high enough temperature for a sufficient lapse of time so that the diffusion of carbon would produce perfect uniformity. Let us now see what modi- fications result when we lift the restrictions imposed by the above hy- pothesis, which in practice — as I have already said many times — never occurs, not even approximately. In order to do this, one is obliged to consider the composition of an austenite crystal not as being represented by a "point," but instead by a line, a portion of an obscissa drawn to each successive temperature. Moreover, it has been demonstrated in Part I that the integration of such lines delineates a "band of variant concentration." Call AB (Fig. 17) the range of carbon concentrations in the 7 solid solution of an alloy whose average composition is H H', as it exists at some temperature T below the solidus (CD) but above the initial point of the 7— >a transformation. The band of variant concentration which must be taken into account when the temperature of the system has fallen below T, is bordered by two descending curves a and b which gradually approach one another. The distance between the two curves at a given lower temperature Ti, is less, the slower the cooling from T to Ti. Now let us see how the y^a transformation takes place in a system thus constituted. In order to visualize it in the clearest manner, we shall begin by restricting our examination to the phenomena which happen within the sphere of only one primary crystalline element. The laws upon which this restriction must be based have already been noted. We know that in such a case, the points of the curve a (Fig. 17) correspond to the concentrations of carbon at the center of the crystalline element under examination, while curve b denotes the concentrations of carbon at the periphery of the same element. When the system has reached temperature T2, at which the curve a cuts EF, the liquidus of the 7— >a transformation, crystals will start to 1 See Sec. 46 and 47. 84 HEAT TREATMENT OF SOFT AND MEDIUM STEELS ,85 separate having a percentage of carbon represented by the point G. This separation will start at the same point of the metallic mass occupied by the nucleus of primary crystallization. Here is an essential difference between the course of the phenomena in the more general and real case now under examination and in the occur- rences which would appear in perfectly homogeneous solution, already studied in the previous sections. While in that first ideal case any neces- sary relations between the position of the initial centers of crystallization of the first-appearing solid solution and the centers of primary crystalli- "*JJ Fig. 17. — Decomposition of non-uniform austenite. zation were not observed, in an actual case such a relation appears immediately and in the simplest form: the two points must coincide. A more complete study of the phenomena which I now wish to present in their simpler schematic form, would show that in practice the difference. between the two cases is never clearly defined. This is be- cause even in the ideal case, the presence of impurities, which are never possible to eliminate totally in steel producedin ordinary practice, estab- lishes a relationship between the network of the primary crystalhne structure and that of the secondary, even when the y solution has been brought to the highest degrees of homogeneity which can be attained in 86 HEAT TREATMENT OF industry. In the case of ordinary steel the presence of the same impuri- ties may so strongly modify the process of secondary crystaUization as to prevent the identity of position of the initial centers of this re-crystalliza- tion and of the primary solidification, although it does not ehminate (as it is easily understood) but rather accentuates the other relations obtain- ing between the two crystalline systems. For simpKcity, the effects produced by these complementary phenomena will not be taken into account at present, though it must be realized that their practical impor- tance may approach and sometimes greatly exceed that of the fundamental phenomena now under study, especially when very impure steels are under observation. 64. When after a very short lapse of time t, the temperature of the system has dropped further, passing from the value T2 to the value Ta,^ a small mass of a soUd solution will form around the center of transforma- tion defined in the preceding section. The carbon in this mass will have an average analysis lying between G and Gi and its specific concentration will vary from the center toward the periphery of the transformed mass from a minimum value corresponding to a point Go situated slightly at the right side of G to a maximum value equal to Gi. The reasons for all this have been seen in Part I. At the end of the same interval of time t, the carbon content in the region surrounding this ferrite mass, especially at the surface of contact between the residual 7 solution and the external surface of the small mass of a solution, has increased to the value Ai. Ai corresponds precisely to the conditions of equilibrium between the y mixed crystals and the a crystals at temperature T3. The same principles already indicated at the beginning of Sec. 50 allow two portions of the residual austenite to be distinguished at the end of the time t: (a) The first fraction forms something like a shell around the mass of a solid solution which precipitated in the temperature interval T^ to T3. As just remarked, its carbon concentration has the maximum value Ai at the internal surface of contact. Outwards it diminishes rapidly in the successive thin layers. The external limit of region (a) is marked- by the layer in which the concentration of carbon has remained unaltered from that which it possessed before the ferrite nucleus had separated. It is clear that the thickness of this "shell" or, in other words, the depth to which the 7 solid solution has been enriched in carbon by the separation of the first accumulation of a crystals, other conditions being equal will be greater, the greater has been the velocity of the diffusion of carbon, or — ^f or a given value of the velocity of diffusion — the longer has been the time t required for cooling the system from the temperature T2 to T3. ' The proportions between the different parts of the diagram are out of scale, as observed in the note at Sec. 49. SOFT AND MEDIUM STEELS 87 (6) The second portion of the residual y solution is that constituted by all the remaining mass of austenite, surrounding the region defined in (a). In this region, where the effects of diffusion of the carbon accumu- lated in zone (a) have had no time yet to be felt, the concentration varies from a minimum value (located between A' and Ai) in the layer cor- responding to the external limit of region (a) and the maximum value Bi at the external layer of region (?>),a layer which naturally coincides with the external surface of the primary crystalline element which is under observation. In regard to the quantitative ratio between the a mixed crystals formed at temperature T3, and the residual austenite then remaining unaltered, I could now repeat, except with a few perfectly obvious modifi- cations, what J have already said in Sec. 51, again reaching the conclusion that the quantity of ferrite will be the larger, the more complete the diffu- sion of carbon in the residual austenite. 65. Contrary to what was the usual occurrence in the preceding discus- sions, ^ this last conclusion is true for the non-uniform austenite under consideration only when the time interval t, as well as the temperature interval T2 to T3, is very small. In fact, one must also take into account the diffusion of carbon in the austenitic region (6) of Sec. 64, a region which in previous discussions had a uniform carbon concentration. Here the diffusion of carbon progresses in a contrary direction to that in which it proceeds in region (a) ; i.e., inward from the external layers of the primary crystalline element. It is easy to understand that this in- ward migration counteracts the effects of the analogous process taking place in region (a), the amount of the outward diffusion from ferrite crystallites and the effects originating from it. Among these last effects, one must recall the one which I mentioned at the end of Sec. 50, namely, the influence of cooUng velocity upon the proportion of a mixed crystals. Particularly it must be clear that the constant tendency of heterogeneous austenite to become uniform in carbon concentration counteracts the process by which a mixed crystals can be proportionately increased by slowing the cooling or prolonging the heating.^ In the more general case of heterogeneous austenite now under observation, the inward diffusion in zone (b) may reach such in- tensity with large differences in carbon concentration as to annul and sometimes even invert the effects which reheating the steel would be expected to produce upon the proportion of a mixed crystals separated from the 7 solid solution at a given temperature. Here is a first example of the marked structural differences which must occur in steel, according to whether the history of the piece has caused a minor heterogeneity in the primary 7 solid solution (approaching 1 Especially see Sec. 50 and 61. ' See the last lines of Sec. 50. 88 HEAT TREATMENT OF somewhat the first case we have studied) or a major non-uniformity (approaching somewhat the second). I shall have occasion to mention some examples shortly. 66. How does the 7— >« transformation proceed, as the coohng of the system continues? To answer this, observe the system when its temperature has reached the value Ti after a second short lapse of time h. Again it is convenient to consider separately three distinct regions in the residual austenite : (A) is the peripheral region of the primary crystalline element under observation throughout which the concentration of carbon is to the right of the liquidus EBg at temperature Tt (Fig. 17). It is bounded by the external surface of the crystalline element, where carbon has a concen- tration B2, and the surface at which the concentration has the value A2 exactly corresponding to a point on the liquidus. At A2 austenite is evidently saturated in respect to the a solid solution at the temperature From observation of the diagram, transformation 7— >a does not take place in any point of region (A) at temperatures above T^. But in this region, diffusion of carbon from the peripheral layers toward its innermost layers (Sec. 65) is continually in progress during all the time the metal is above temperature T^. (B) is the region immediately around the a mixed crystals formed in the preceding minute time interval required to cool from T2 to T.3. Its size is determined by the distance to which the influence of the ferrite crystallite built up about the original nucleus may be felt. Separation of new quantities of a solid solution during the temperature interval T3 to T4, takes place in it without undercooling. The newly born ferrite also forms a new layer around the crystallite formed in the previous interval T2 to T3. In this new layer however, the concentration of carbon at its external surface will reach the maximum value G2. But at the same time, the process of diffusion will have raised the concen- tration at the central region of the same ferrite individual by transport- ing carbon inward from the periphery. As a result, the average carbon concentration in the enlarged a crystal will be represented by a new point G3, situated at the right of Gq but at the left of G2. At the same time the carbon has also increased in the 7 solid solution surrounding this new layer of a solid solution, reaching the maximum value A2 at the interface. And contemporaneously the dimension of Zone (B) [which in Sec. 64 was defined as shell (a) of austenite] has in- creased due to the diffusion of carbon in the 7 solid solution occurring within the second time ti and rendered more intense by the concomitant increase in saturation with further cooling, represented by the change from point Ai to A2 (Fig. 17). This increase is also counteracted by the same phenomena studied SOFT AND MEDIUM STEELS 89 in Sec. 65; i.e., by the simultaneous diffusion of carbon from the external toward the internal layers of the original gamma mass. (C) Finally there is an intermediate region between (A) and (B) which also forms a part of the shell enveloping the latter. It is exactly limited on the outside by the surface at which the carbon concentra- tion has the value A2; a "surface which, as has already been seen, limits the inside of region (A). From the definition already given of region (5), it is clear that the internal limits of (C) cannot be indicated with equal precision, since it is determined only by the distance beyond which the action of the first particles of ferrite as a "crystallization-germ" does not "make itself felt. From all that was said in the previous paragraphs, it is clear that in region (C) the 7 solid solution is supersaturated in respect to a mixed crystals. Therefore, in general, new crystallization centers will form in this region around which new crystalline elements of the a series will develop. Owing to the schematic character of the diagram upon which our study is based — it would not be wise to afiirm that this last phenomenon actually takes place in the second time interval ii, rather than being initiated in a successive interval of time. But inasmuch as the course of the process continues identically in the manner just described until new centers of ferrite crystallization appear in region (C), it is clear that an argument will lose nothing in generality if, for simple reasons of brevity, it is supposed that the appearance of new germs actually occurs in the time fj which we now are studying. It is obvious that the phenomena accompanying the growth of such new crystaUine elements of ferrite in region (C), take place under con- ditions entirely similar to those accompanying the growth of the a crystallites of the first series, formed at the central region of the primary austenite crystal. The difference between the elements of the two series is purely quantitative and consists in the fact that the a mixed crystals last to appear are born in a solution of higher carbon content, so that their nuclei are also higher in carbon. A second quantitative difference is found in this l^st fact, whose effects accrue to those studied at the end of Sec. 65. Consideration will show that the consequences of transformation in region (C) counter- act those of the phenomenon explained at the end of the previous section, as far as concerns the proportion of a mixed crystals separated at a determined instant in an alloy of given initial composition. As formerly limited, all the new crystaUine elements which formed during a phase of the transformation corresponding to the one now under obeservation occurred in the region around the high carbon "shell" enveloping a crys- tals of the first series — region (6) of Sec. 51 — and were necessarily obliged to show a minimum carbon-concentration equal to that of the crystals of the previous series. 90 HEAT TREATMENT OF It is clear that the ratio between the carbon concentrations of the two phases, a and y, and, as a consequence, the ratio between the quanti- ties of the same two phases which are reciprocally in unstable equilibrium at a given instant of the transformation, are subject to further variations due to the different velocities at which the various transformation and diffusion processes take place. This is evidently true both for homo- geneous and for heterogeneous austenite. An analytical observation of these variations would not only take too long, but it should also presume at least an approximate knowledge of the relations between the diffusion velocity of carbon and the temperature of the system; also between the frequency of the f errite nuclei, the amount of undercoohng in the austenite when such germs are born, and the temperature of the system. The same observations apply in this regard which I outlined in Sec. 51, (6). 67. Now let us observe — as we did in Sec. 53 — how the reversible transformation progresses when the system cools to a temperature below Ti (Fig. 17). Tb is the intersection of liquidusS'i^ with curve 6, the right border of the band of heterogeneity in the y solid solution. It is also necessary in this analysis to bear in mind the observations contained in Sec. 52. It is clear that, in all the phases of the process which occur at tem- peratures above T&, the trend of the various phenomena does not differ qualitatively from that which we have briefly analyzed for the first phases of the transformation during the minute time intervals t and h. Note that Fig. 17 is badly out of scale, in that T4. to Ts represents a considerable temperature interval, while T2 to Ti is extremely small.' Having reached the temperature T^, the iron-carbon alloy is in a similar state to that of the same alloy at the same temperature in case the original austenite had a uniform concentration of carbon.^ The differ- ences have already been pointed out with a brief analysis of their causes and characters and consist only in the distribution of the secondary ferrite and in the carbon content. As a mere consequence of this last fact a different proportion exists between the mass of the two constituents themselves. But this certainly does not modify the further course of the trans- formation studied in the previous pages. It is easy to understand this point if the considerations which I have made from Sec. 53 up to Sec. 62 are observed from this special point of view. To justify this observation now would only result in repeating what was said there. In order to avoid misunderstanding as to the generaUty of the phe- nomena described in Sec. 53 to 62, it should be pointed out that it is perfectly possible to extend those principles to a mass of steel containing any number whatsoever of primary crystalline elements. In fact, a ' See Sec. 54 and following. SOFT AND MEDIUM STEELS 91 simple revision of the wording will make the text cover a sizable quantity of steel, maintaining the value of the remarks intact even when applied to an assemblage of any number of primary crystalline elements. The original analysis was restricted to the case of only one crystal for reasons of simphcity and clearness. However in this more general case it is necessary to consider the total quantities of each portion of the two con- stituents a and 7, which at each period in the process of transformation appear with the specific characteristics aheady studied. Also it is neces- sary to consider the average carbon concentrations in the total mass of said distinct portions. One could profitably repeat the observations made in Sec. 54 e.s. regarding the location of the curves representing the average and mini- mum concentrations of carbon in the y phase, with only this difference : that in the present case this curve may only accidentally coincide with the ordinate representing the average concentration of carbon of the alloy, while in the previous case this coincidence always occurs in a determined interval of the transformation. The specific characteristic of the curve discussed in Sec. 55 represent- ing the points where 7— >a transformation begins for each determined carbon concentration appHes to the present discussion for temperatures below T5. This curve will not represent the maximum concentration of carbon in the entire 7 phase except for temperatures below T^. It is also easy to see that the observation made in Sec. 56 concerning the curves denoting the carbon concentrations in the a phase can be apphed to the general case which we are now studying; also those of Sec. 57 and 58 in reference to the relations between the velocity of diffusion of carbon in austenite, and the proportion of the a and 7 phases co-existent in the system at each instant. Sections 59 and 62 discussed the limiting case where austenite, originally uniform in carbon concentration, decomposed at the eutectoid horizontal into a conglomerate of ai and a^ phases (f errite and cementite) . In general, this former discussion apphes to the more usual industrial condition where heterogeneous austenite is transformed. In applying the data of Sec. 60 regarding the anomaUes in the formation of pearhte due to slow diffusion of carbon in austenite, it is necessary to take into account what was said in Sec. 65 concerning the influence of heterogeneous austenite upon the carbon content of the f errite nuclei at the time of their birth, and also upon the ratio between the co-existent volumes of the two regions {B) and (A). 68. The schematic considerations already made up to this point would allow one to deduce some interesting practical consequences of the variations imposed upon the structural characters of a hypo-eutectoid steel by the conditions of its sohdification, of its subsequent cooling and of the heat treatment to which it is subjected after its complete solidifica- 92 HEAT TREATMENT OF SOFT AND MEDIUM STEELS tion. However, inasmuch as these structural variations are intimately connected with changes in the mechanical properties of steel, I deem it opportune to postpone its study to the successive parts, in which I shall briefly examine some examples of the effects which heat treatments, may produce in the mechanical properties of the metals comprising \ our investigation. Before leaving the study of the binary iron : carbon alloy to pass to more complex steels such as really appear in practice, I will only add a remark, negative in character, which avoids the necessity of taking under observation a series of phenomena, whose analysis would involve great complications and difficulties in our study. If one observes the whole of the iron : carbon equilibrium diagram comprising all the steels, properly speaking (namely, up to a content of 1.8 per cent, carbon), it may be seen how it is possible for some primary cementite to form even in hypo-eutectoid alloys, especially if containing- but little less than 0.9 per cent, carbon. Referring to Fig. 11, it is clear that if the band of variant concentration in the y mixed crystals of average analysis even below that of the point H is so large as to extend toward the right beyond the point H at the eutectoid temperature IGH (690°C.), then a portion of thq transformation y^a will take place in the richest austenite during cooling in correspondence to the points of the curve HN, and will cause the separation of primary cementite. The process Just mentioned explains one of the causes producing the paradoxical appearance repeatedly observed by many experimentors, of contemporary pro-eutectoid ferrite and cementite in hypo-euteotoid carbon steels. In some cases the same fact originates from a different cause, connected with the abnormal development of the processes of undercooling, seen in Sec. 59 to occur in the formation of the eutectic. This phenomenon does not happen in the great majority of that special class of machine steels which forms the object of our study. In those few cases in which it does occur it never attains noticeable importance either upon the microstructure of the alloy or upon the mechanical properties. It is, therefore, quite justifiable to disregard it in a summary study. 69. Although the observation presents a direct interest only in the apphcation of thermal analysis to the study of our steels, and is not apt to be utihzed for technical purposes, it is also opportune to point out that for an iron: carbon alloy of a given average composition, the "liquidus" of the reaction 7-^0; (thermally determined as the locus of the points at which, in cooling, the reaction now indicated starts) will appear to be moved toward the upper and the right side, the wider the band of variant concen- trations in the austenite due to incomplete diffusion of carbon. The correctness of such a remark will immediately be evident upon simply recollecting the definition of the point A' of Fig. 17. CHAPTER X SECONDARY CRYSTALLIZATION IN MORE COMPLEX SYSTEMS 70. Let us see now, how the course of the phenomena studied in the previous paragraphs can be graphically represented for the only cases which really appear in practice, namely, those in which the steel considered is no longer a binary alloy, but contains one or more of the other metals constantly entering into the composition of the ordinary machine steels. In other words, let us lift the third restriction which we adopted in Sec. 46 and 47 when beginning the study of the 7?=^ transformation. The first and second of these restrictions were eliminated in Sec. 59 and 63 respectively. In order to study this general case completely and intently, one ought to examine again the entire considerations already established in Part I regarding primary crystallization, in order to show how the various elements of the "real" diagrams in three dimensions have to be modified at the transformation range. This would show precisely how the so- called "fictitious" binary and ternary diagrams, representing the more complex systems, are derived from accurate sketches representing binary and ternary equilibria. However, such a method would be too lengthy, and would necessi- tate a large number of superfluous repetitions. Therefore, it is preferable to discuss these "fictitious" diagrams which in the simplest way represent the course of the transformations of om- system completely enough to suffice for all practical purposes. A few remarks will be added when necessary to establish the characteristics of the new lines and areas in them which must be considered. Especially this wiU be necessary to describe the position which these new markings assume in comparison with the one they occupy in true diagrams in three dimensions. Anyone who wishes to follow more precisely and closely in its details the phenomena following the 7<=^ transition in a more complex steel, has only to review the pages of Part I, taking into proper account the subject matter of the previous sections of Part II. He will easily realize the fact that all the other possible considerations beyond those before mentioned regarding the fictitious transformation diagrams in two and three dimensions, and their derivation from the true diagrams, are identical to those already devel- oped in part I in the case of primary solidification, or may be easily derived from them by means of arguments so simple and obvious as to appear entirely superfluous to dwell upon. 93 94 HEAT TREATMENT OF 71. For the reasons given in the study of the binary iron : carbon alloy we will start by examining the limiting case where the history of the crystalline mass has been such as to give complete equiUbrium of the concentration of the various elements entering into the steel. Furthermore, as has already been done in Part I when studying sohdi- fication, and for the same reasons advanced in Sec. 33 and 34, carbon and an iron complex will be taken as principal constituents, representing the carbon concentrations as abscissae of a fictitious binary diagram. All those conditions which apply to the elements which with carbon enter into the composition of our steels, apply to the y^a transition much the same as they affect the process of primary crystallization. In Sec. 32 to 34 are to be found some observations necessary and sufficient to justify the deduction of a simpHfied "fictitious" diagram from complete diagrams too comphcated to be directly applied to practical uses. Such remarks also apply to the y^>a transformation. Although the experimental data available is in reality very scarce and not always definite, it warrants the statement that the course of the y-^a transformation in some ways appears a great deal more suitable for representation by methods described in Part I and there applied to the study of sohdification. This is due to the fact that the transformation range for the binary iron alloys with the constituents of our steels beside carbon are always narrower than the intervals of the primary crystalliza- tion. Therefore the "deformations" of the "true" binary or ternary diagrams (by means of which we pass to the corresponding "fictitious" diagrams of the more complex alloys) are generally smaller. This happens in an especially relnarkable measure for all the constitu- ents other than carbon which enter into alloy with iron forming the so-called "carbon machine steels." In such steels, these constituents are ordinarily limited to silicon and manganese, beside minute quantities of the usual impurities (sulphur, phosphorus, copper, arsenic, etc.) Normally such impurities exist in extremely low percentages in machine steel, never much exceeding 0.2 to 0.3 per cent. Silicon and manganese also rarely go above 0.30 per cent, and 1 per cent, respectively. For these steels therefore, there cannot be any doubt that the best method of graphic representation is the fictitious binary diagram, with iron and carbon assumed as basic elements. If instead, we pass to the so-called "alloy steels," we find almost always in them an element beside iron or carbon preponderant both for its quantity as well as for its specific effects. And in this case — as we have already^ seen for the primary crystallization — it would generally seem best to adopt a fictitious ternary diagram, having as basic elements iron, carbon and this element X. However, the fact has been mentioned before that the transformation ranges are rather smaU for binary alloys of iron with one of the metals SOFT AND MEDIUM STEELS 95 which frequently enter into the composition of special machine steels such as nickel, chromium, tungsten, molybdenum, or vanadium. Con- sequently it is convenient to adopt the simpler fictitious binary diagrams for representing the greater number of those steels. Another reason is that the "deformations" caused even by a rather high proportion of one or more constituents are not such as to confuse this method of representation. For these reasons, I will refer in what follows only to the case of the fictitious binary diagram. Besides, it is very easy to extend the conclusions which we shall reach in regard to the binary diagram to the case of the fictitious ternary dia- gram, basing the work upon what we have seen in Part I. Fig. 18. — Decomposition of uniform austenite in a complex steel. 72. If curve AB in a fictitious binary diagram Fig. 18 represents the beginning of the y^a transformation and CM represents its ending, it will be easy here to dehneate the elements indicating the progress of the phenomena examined in Chapters VI to IX. First assume for the sake of simphcity that uniformity in concentra- tions has been reached and will be constantly maintained in the y solid solution. One may immediately assemble the experimental data concern- ing the transformation y^a in the diagram in the following principal lines : First. — The curve FGI, denoting the minimum carbon concentrations in the a imixed crystals. 96 HEAT TREATMENT OF Second. — Curve FL representing the average carbon concentrations in the same mixed crystals. L, its point of intersection with the vertical HH', corresponds to the end of the transformation. Third. — FN is the curve of maximum carbon in the ferrite. Its inter- section with the horizontal through L is the point N, determining the con- centration of carbon in the last traces of a mixed crystals to be formed. Therefore, it marks the lower extremity of the curve EN followed by the "fictitious representative point" of our system as defined in Sec. 17. Fourth. — Curve EP represents the concentrations of carbon in the 7 mixed crystals at various temperatures during the transformation range. Fifth. — The two curves GI and NO limit the band of variant carbon concentration in the a solid solution. From a practical point of view it would evidently be of no special interest to determine the present position of the curve EP, in comparison to the position which the same curve would occupy if diffusion in ferrite were speedy and complete. However, the same observation is not true for the three curves FG, FL, and FN in regard to their position with relation to curve FM, where as has been demonstrated, all these three curves would meet if complete equilibrium of concentrations in ferrite could be reached in all stages of the transition. In fact the temperature of the end of transforma- tion and the width of the band of heterogeneity in ferrite depend upon the position of the curves. Reflection will show that the reciprocal position of curves FN, FG and FL, in respect to curve FM, cannot be anything but the one indicated in Fig. 18. Naturally, their divergence from the location of FM will depend upon the nature of the alloy considered, the conditions under which solidi- fication has taken place, and upon the subsequent cooling velocity of the solid system. As already remarked, this may be thoroughly established by again examining the principles laid down in Part I regarding the method of passing from the "true" diagrams of simple systems to those "fictitious" ones corresponding to the more complex. 73. If now we analyze the y—^a transformation, taking into account the fact that the velocity of diffusion of carbon and of the other elements in the y phase is never sufficient to cause complete homogeneity, we see immediately that Fig. 18 contains all the elements necsseary to apply to this more complex case all the reasoning of the previous paragraphs concerning the 7;:^ transformaton in the binary iron : carbon alloy. On repeating those considerations we will again find the three new curves defined in Sec. 55 and 56 representing the concentration of carbon in austenite, as it varies with temperature and location. We may also simply repeat the considerations of Sec. 56 concerning the characteristics of the six curves which determine the course of the y^a transformation. Consequently all the conclusions reached in Sec. 45 to 69 may be ap- SOFT AND MEDIUM STEELS 97 plied to the study of the more complex real system. This statement applies both in regard to the simple uninterrupted transformation 7— >a, and to the instances where the effects of the insoluble range in the a mixed crystals are felt, resulting in the formation of peralite. However, it is necessary to remember that in all cases, but especially in the phenomena connected with eutectoid formation, the variations of concentrations of all the elements due to the incomplete diffusion give rise to new effects which reinforce those directly due to imperfect carbon diffusion. These last effects have been studied in the preceding pages only in the binary iron : carbon alloy and we have heretofore only considered the possibility of graphic representation of incomplete diffusion in our fictitious binary diagram. It is evident that the ternary diagrams studied in Part I would also contain the possibility of serving as a basis for graphic representation for these more complex phenomena. Such a study would be very long and complicated, and at present the quantita- tive experimental data are totally lacking which alone could give it some practical interest. It is clear that such missing data are just what would serve to construct the equilibrium diagrams of the binary systems formed by iron and each constituent of machine steels. It would be necessary to determine the value of the disturbances caused in such diagrams by the incomplete development of diffusion of each of those elements. Perhaps it is superfluous to point out that the effects of slow diffusion are felt in the fictitious binary diagram (Fig. 18) by sidewise displacements of the six fundamental curves defined in Sec. 55 and 56. However, the experimental data now available are insufficient to furnish indications, not only about the amount but even about the direction of these perturbations. Experience actually shows that in many ternary and quaternary special steels, the displacements above indicated reduce the width of the band of variant concentrations in comparison with the dimensions which the latter reaches in plain carbon steels of equal carbon percentage, other conditions being equal. 74. Finally, in order to eliminate the last restriction which has been imposed upon our schemes of graphic representation, it remains to investi- gate the case of a complex steel used for the construction of machine parts in which the course of the primary crystallization and of the succeeding cooling of the resultant 7 sohd solution has not been so slow as to produce complete uniformity of the various elements in the same solution. In order to sum up the conclusions which we have reached, it is con- venient to observe the entire fictitious binary diagram, including the in- terval of solidification of liquid steel as well as the 7^a transformation range. Such a diagram, reproduced in its essential parts in Fig. 19, does not contain any new elements in addition to those studied in the pre- vious cases, and therefore does not require further detailed explanations. 98 HEAT TREATMENT OF Both in Fig. 17 and 18 the same letters have been used to indicate corre- sponding points. Suffice it to point out that the curve OP' is the curve representing the maximum concentrations of carbon existing in the mixed crystals which still remain in the system at the various temperatures comprised in the same interval. It is superfluous to note that the same markings contained in the diagram of Fig. 19 could easily be modified so as to cover the more general FiQ. 19. — Constitution of complex steel during slow cooling. case regarding the formation of eutectoid pearlite; it would only re- quire an extension of the considerations developed in Sec. 59 to 62 and 67 to 69. 75. A most general study of the efi'ects which certain definite heat treatments may produce upon the transformation might now be based upon the ideas developed in the previous pages. These heat treatments would always consist in determined variations of the velocity of cooling of the metallic alloy, sometimes alternated with reheatings, or holdngs at a given temperature for some time (soaking). SOFT AND MEDIUM STEELS 99 However, such a general investigation would not present a real inter- est unless it was illustrated with an accompanying detailed and complete study of the modifications in physical and mechanical properties outlined above which accompany variations in the course of the transformations. But, under such conditions, it is clear that such an investigation could not find an adequate place in this volume both because of its magnitude and its character. Therefore, I plan to present later only a relatively few practical ex- amples illustrating how the previous deliberations can be employed both as an aid in the interpretation of the phenomena occurring in prac- tice, by deriving from them precise indications of how to modify their course rightly, and as a guide in searching for a heat treatment more suitable to impart the properties thought to be the best for the use of a given steel. In the following few pages I will only add some observations concern- ing two classes of phenomena which must be taken into account in the best practical applications of the principles derived in the previous sections. Both classes of these phenomena may be thought of as being "retards" in the start and in the development of the transitions. Therefore, in a cooling steel the transformations start at a lower temperature than would be expected from scaling the diagram of "normal" equilibrium. They would also proceed but incompletely. Published equilibrium diagrams are obviously based on experimental data determined under such condi- tions as to avoid the intervention of these very phenomena. As I have already said, they may be subdivided in two principal groups : (a) Retardations due to the fact that the given system can be cooled considerably below its transformation temperature, yet at the same time remain entirely unaltered. In other words, it is then in a metastable condition possessing the form normally stable only at a higher temperature. Such phenomena occur in the y—>a transformation as well as in the primary soUdification. In the later case they assume the characters of true "undercooling" phenomena, and may cause effects of the greatest practical importance, such as, for instance, great accidental segrega- tions or liquations. When phenomena of this group occur in the field of allotropic trans- formation, they may spread so much that the metastable system reaches temperatures so low that the transition takes place only at an extremely small speed. Sometimes the metastable state cannot revert to the stable; in this case we pass directly to retardations of the second group (6), below, whose effects can be connected directly to those of the first. Retardations due to undercooling are characterized, as is well known, by the fact that they cease abruptly at a certain temperature. Such 100 HEAT TREATMENT OF ' an action is similar to that caused by "primers" on an explosive, and is instituted by the appearance of "germs" of the phase stable at lower temperatures. If the reaction is "detonated" at a sufficiently high temperature to allow the transformation being completed at an observable velocity, once started, it continues with a development of heat until the system is converted to a condition of stable equilibrium. (6) Retardations of a given transformation upon cooling due to the fact that it normally starts and takes place in a range of tempera- ture so low that the velocity with which it proceeds is not sufficient to render the change noticeable in short intervals of time. Phenomena of this group appear with greater intensity in special steels with high percentages of alloying metals, as can be naturally expected if we remember the equilibrium diagrams of the complex steels. Other conditions being equal, their variations due to changing the nature and concentration of the various metals constituting the particular special steel, can be established with certainty a priori, when the equilibrium diagram of said steel is known in its more complete form studied in the previous pages. We have already seen clearly how the phenomenon can be represented graphically, when we studied the criteria by which we pass from the graphic representation of the simple systems to that of the more complex systems by the introduction of the "fictitious diagrams." This fact now constitutes indeed one of the properties which differen- tiate retardations of group (a) from those of (b). The former are not represented by any of the markings of the diagrams which we have studied. A second differentiation lies in the fact that recalescence naturally cannot appear in the latter. It is superfluous to point out that the retards belonging to this second group pertain not only to the continuous transformation 7— >q!, but also happen when the transformation generates pearlite. This appears evident from what we have seen in the previous sections. Here it is necessary to observe that, in practice, the two orders of retardations almost never appear separately. In general, those of the first group (a) appear frequently and to a greater measure in steels of low normal transformation points. Thus, for these steels, it almost always happens that when retardation due to undercooling ceases, the metal is already at a temperature at which the transformation velocity is extremely low. 76. Other conditions being equal it is well known that the amount of retardation is greater for speedy cooling. Again, the fa.cility in reach- ing temperatures at which an uncompleted transformation proceeds at unobservable rates permitting no noticeable segregation of new phases, increases with higher velocity of quenching. SOFT AND MEDIUM STEELS 101 It is also known that advantage of this fact is taken in the practice of what I call "quality" or "final" heat treatments, designed to impart determined mechanical properties to a given steel by means of a correct selection of the velocity of final cooling. Hardening without drastic quenching is effected in the well known processes of air hardening, hot oil hardening, etc. Such processes are based upon these very criteria, reinforced by thermal hysteresis occurring in greater measure in the special steels of low transformation intervals. With the aid of those processes it is possible to obtain the same effects by means of simple reheatings of steels of a suitably selected composition, which upon ordinary steels can be obtained only by means of drastic quenchings, an expedient which would often be disastrous. This is not the place to discuss the phenomena and processes just mentioned because their study should more appropriately be a part of a treatise on "quality" or "final" heat treatments than of a study of "homogeneity" or "preliminary" heat treatments, which are based essentially upon the utilization of the phenomena studied in the previous At any rate, I wanted to mention them here, because they are some- times taken advantage of in the "preliminary" heat treatments. Their existence must be known and noted, even if use is not made of their special characteristics. I propose to return later to this point in refer- ence to some concrete examples. PART III DIFFUSION IN AUSTENITE AS APPLIED TO THE PRELIMI- NARY HEAT TREATMENT OF STEELS CHAPTER XI FUNDAMENTAL CRITERIA FOR PRACTICAL APPLICATION OF PRELIMINARY HEAT TREATMENTS 77. The principles developed in the proceeding pages permit tracing the fundamental lines of a schematic chart in a broad manner. So-called preliminary heat treatments for homogeneity (normaUzing) are based upon the principal phenomena represented by such a chart. Our experimental knowledge which can actually be accurately placed in such a sketch is very scarce and often uncertain. At present it can therefore be stated that merely the first elements of a systematic doctrine exist which if known would permit the deduction of prac- tical rules for the preliminary heat treatment of a steel of a given chemical composition by the mere application of criteria directly inferred from that schematic chart. An enormous mass of detailed experimental material must be gathered and classified before the necessary theoretical studies can fully synthesize a diagram in such accurate particulars as to render it susceptible of easy, direct, and general applications in steel practice. From what has been seen in the previous sections, it appears clear that this enormous analytical experimental work shall comprehend all the problems, without exception, which I have been able only to mention during our summary and rapid examination. It would be superfluous to repeat a list of such problems ranging all the way from the accurate determination of all the possible binary equilibrium diagrams which may be formed by the elements entering into the composition of our steels, to the determination of the laws of diffusion of each one of these elements in the alloys in question. From what has been briefly said in the preceding chapters, it may easily be seen that the successful issue of such researches will require the use of a very great variety of experimental methods : ranging from thermal analysis to microscopic examination; from the purely chemical methods to the methods based on the study of electrical and magnetic properties of the alloys, and so on. Nor is it at all unlikely (as recent researches seem to prove) that new investigative methods, founded on principles which have never yet been used in such field of studies, may prove of great assistance. 103 104 HEAT TREATMENT OF 78. Since the deficiencies and the uncertainties are so great which render many parts of the proposed diagrams incompleted and undeter- mined, one might imagine that the considerations made up to this point would be entirely useless, constituting nothing more than a pure theo- retical exercise. This is not my opinion, and it may be useful to mention briefly my reasons. As has been said there were two purposes aimed at in compiling these notes — nor could they possibly have been different, for the reasons which are now evident at this point of our study. The first was to show how it is possible to rationally coordinate the meager experimental data now possessed. In my opinion, this is the best method to indicate, at least in its main lines, what may be the more desirable path for further experimental researches. Especially the Hght of past knowledge is necessary in order to draw results from fresh researches useful for the final purpose and to make them susceptible of easier and surer interpretation. So far as this first aim is concerned, I could have added nothing else to what has been said in the first two parts of this study. The actual state of our knowledge would not permit the theoretical investigation to be extended much further nor developed in greater detail, without totally abandoning the possibility of checking the results experimentally. I, therefore, consider that the present state of our knowledge has been reviewed as completely as the limits of an elementary treatise permit with the presentation of the considerations developed in the previous chapters . But my second aim depends a great deal more upon my conviction of the utility of the study previously developed in outline. This second aim was to place the real mechanism of the complex phenomena which are utilized as a basis for the preliminary heat treat- ment in clearer light by defining its nature with that precision which only a graphic representation permits. The observations made in the first two parts might also be sufficient upon this second point had the reader thoroughly known the technical nature of normalizing practice and had precise ideas regarding the nature and importance of the variations in structure and physical properties which may thereby be produced. Only in such an event might it be pos- sible for those observations to reach fully their ultimate purpose. Then the connection between numerous elementary experimental facts could be discerned each one of which was previously known as an isolated occurrence. A more precise idea of the mechanism of the more complex phenomena would also be obtained by an appreciation of the laws accord- ing to which the effects of various elementary facts accrue one upon the other. However, it does not seem logical that the main practical aim of a SOFT AND MEDIUM STEELS 105 treatise, elementary, or better preliminary, in character, should be exactly understood in its real value only by those M'ho are already con- versant with the phenomena to be coordinated. It seems especially illogical since this practical knowledge is apparently not widely scattered. Instead, beginners in this special branch of metallurgy should become familiar with the real practical value to be had from such a schematic study. Such knowledge will much facilitate the exact understanding of processes .whose direct technical study can furnish nothing else but confused and indetermined ideas, even though this was the only method of attack possible until a short time ago. From such a standpoint, it is evident that the discussion in the previ- ous pages is not complete, nor could it be sufficient to give a definite notion of its real practical value to the beginner. The best way to fill this gap would be to use that material in the development of a systematic treatise, as complete as present knowledge allows, dealing with the phenomena occurring in the principal metallur- gical processes ordinarily applied in making and working machine steels from tapping to the final cooling, at least for as many of those phenomena as enter in the schematic chart already drawn. However, such a re- search would necessarily be very lengthy and would lead far beyond the boundaries which we have definitely set for our study. For all these reasons, it is more desirable to follow a middle course. Consequently I intend to add to the formerly developed analysis a certain number of practical observations, which taken as isolated examples are suitable to illustrate, even to the beginner, the real nature and the practi- cal value of the phenomena which we have so far studied only in outline. This may also encourage some reader to deepen and to coordinate his theoretical and practical knowledge in this very interesting field of metallurgy. 79. However, before investigating such examples individually, some considerations of wide scope may be pointed out, a knowledge of which is necessary in order to understand the real practical value of those examples. All these observations would easily find their natural place in the course of a systematic and complete study of our general problem. But, inas- much as it is impossible to proceed with such a study for reasons already given, I am obhged to indicate them separately. To many readers they will appear tiresome, but they may be useful to others. 80. The so-called "preliminary heat treatment" of a low or medium steel used for mechanical purposes demanding toughness^ as the most im- portant physical property has for its essential if not exclusive purpose to minimize the heterogeneity of composition between the various parts of ' "Toughness" is a well known complex property which it is not necessary to define exactly at this time. 106 HEAT TREATMENT OF the primary and secondary crystalline constituents. Such treatments are therefore often called "homogeneity" heat treatments or normaliz- ing. When the maximum result of the preliminary heat treatment has been reached, the steel is also in the best condition to acquire the physical properties such as hardness, tensile strength, elastic limit, elongation, etc., required for its correct use, at the same time maintaining the maximum compatible tenacity. ^ Such additional attributes are acquired by means of "final" or "quality" heat treatments — largely quenchings followed by low temperature reheatings. In the first two parts of this volume, it has been said that the causes of heterogeneity in chemical composition belong essentially to two distinct series. The first includes liquation phenomena occurring during separation of austenite from the molten mass of steel during primary solidification. To the second, however, belong all the analogous phenomena which accompany the transformations of 7 iron into fi iron and j3 iron into a iron when the temperature crosses the critical ranges. In the previous general study of these phenomena it has been seen that the results of allotropic transformation depend closely upon the effects produced by the phenomena of the first series. Sections 63 to 75 demonstrated that the heterogeneity existing below the critical tempera- ture varied directly with the heterogeneity remaining in just-solidified austenite, other things being equal. Therefore, it is evident that the study of industrial applications of normalizing heat treatments must take into account all the phenomena depending upon conditions of solidification and of transformation, not only separately but also as the earlier affects the latter. This will appear clearer after the forthcoming examples have been presented. 81. From what has been said in the previous section and from the discussion contained in the first two parts of this volume, it appears that the principal groups of data determining the details of the preliminary heat treatment and the results thereby obtained, are the following; First. — Course of primary crystallization. Second. — ^Method of cooling the mass of y mixed crystals after total sohdification. Third. — ^Course of the y—*a transformation. Each preliminary heat treatment can consist in nothing else but a modification of the progress of one or more of the phenomena pertaining to these three groups. The field of practical application considered in its extension, as logically it should be, would therefore also involve the study of the phe- ^ Tenacity will be used as synonymous with "toughness" rather than as signifying ' 'tensile strength." SOFT AND MEDIUM STEELS 107 nomena accompanying solidification. From a theoretical point of view there can be no doubt of this, as we have seen in previous chapters the great influence which the course of primary crystaUization has upon the heterogeneity which the steel maintains in all its subsequent history. Yet in ordinary steel making practice, the study and the control of all the technical processess aimed at regulating casting practice are very seldom part of the duties of the person who has in charge the studying and the prosecution of all the other steps in the heat treatment. The current dimensions and type of ingot molds, teeming temperature, length of time the ingot remains in the mold before stripping, cooling of the ingot after stripping, etc., is with rare exceptions in exclusive charge of the steel maker. Its control is far away from the man who is responsi- ble for the further heat treatment of the metal. Here is not the best place to discuss the reasons for such a division of technical duties. Doubtless they depend essentially upon the peculiar necessities of individual works organization. At most it can be empha- sized that this division of routine duties requires a perfect accord between the criteria applied in the two series of processes, ordinarily in charge of different men. Also it follows that the man who has charge of the ingot pouring, for instance, should have a good knowledge of the ideas which are being appKed by the man in charge of the subsequent mechanical and thermal treatment. For instance, every time it is neces- sary to cast large masses of special steels, it is absolutely necessary to take the greatest care to prevent excessive wholesale segregation of its components. This is done daily, by the way, in the manufacture of armor-plates of great thickness, as well as of fittings and portions of large caliber guns. At any rate, the fact remains that at the present day the things which happen to a steel during solidification are not considered as being part of the "heat treatment" proper. Therefore, in selecting the examples further on, I shall pass by those more directly concerned with the effects of a determined mode of primary crystaUization, and the variations of the properties of the metal which it is possible to obtain by modifying this routine. Besides, we have already seen that the nature of the phenomena of solidification does not differ essentially from that of the second subdivi- sion, resulting from the method of cooling the austenite. 82. We have already seen that this second series of phenomena are of a very simple nature, consisting essentially in the formation of the band of variant concentration and in the variations of its size. We will presume that the primary crystallization has been regulated in such a way as to minimize the dimensions of this band of heterogeneity existing at temperatures immediately below that of the end of solidifica- tion. This is the best result which may now be expected in practice. 108 HEAT TREATMENT OF inasmuch as up to the present time it has been impossible to find a process analogous to quenching. Later we are going to use quenchings to avoid or reduce the formation of heterogeneities of concentration arising from the 7— ^a transformation. An analogous proceeding might also find industrial application as a means to regulate the process of primary- crystallization, suppressing the effects of liquation or segregation which usually accompany solidification. However, up to the present time all trials made in that line and known to the author have been a failure. Having once obtained the minimum heterogeneity in just-sohdified austenite, it follows from what has been seen in the previous chapters that the immediate problem of heat treatment as far as it is affected by the second phase of the process is concerned {i.e., the life history in the aus- tenitic region) is reduced to the determination of the more suitable conditions for a simple reheating. In fact in this phase of the process it is only a question of how best to facilitate and accelerate the diffusion of the various elements contained in solid solution in 7 iron, These latter processes take place exactly in such a direction as to cause a decrease of the dimensions of the band of variant concentration. The most favorable conditions for producing uniform masses of steel consist in maintaining the metal at the highest possible temperature as long as possible. However, in practice, the application of this simplified idea is subject to restrictions of the highest practical importance, some examples of which it may be timely to mention here. A first group of restrictions in the use of prolonged annealings at very high temperatures includes the tendency under those conditions to a modification of the crystalline structure, dangerous to the physical prop- erties of the metal. It is not exactly clear of what this modification really consists. Gen- erally it is thought to be due, at least in a large part, to a gradual re- arrangement in the orientation of the crystalline elements in such a way as to facilitate the formation and the extension of cleavage planes. An- other very important factor causing damage upon excessive annealing is the "growth" of the primary crystalline grains due probably to surface- tension. It is not necessary to dwell upon this first group of facts in these pages because they represent purely crystallographic traits and they have noth- ing to do with the physico-chemical phenomena which interest us now. They also have a prevalent negative importance from the standpoint of the heat treater inasmuch as they constitute essentially a mere restric- tion of the field of application of heat; nor does this restriction depend in any way upon the processes examined in the first two parts. Therefore, it is enough to note their existence and to point out that their dangerous effects — whose intensity increases very rapidly with increasing tempera- ture — eventually become very severe and cause extreme brittleness. SOFT AND MEDIUM STEELS 109 Up to the present time, it is impossible to give general rules regarding the time and the temperature beyond which a heating would cause a certain degree of damage. Therefore, this data can be obtained only by direct experiment upon each type of steel, teemed under determined con- ditions. Since it is so difficult to obtain precise data upon this point, it is usually preferred to maintain normalizing heats well within those limits of time and temperature indicated as dangerous bj^ experience upon anal- ogous steels. 83. A second important restriction to the use of high temperatures in homogeneity heats has already been mentioned in Sec. 25 and consists in the necessitj^ of keeping below the temperature at which steel starts "burning. " Fig. 20. — Slightly burned nickel-steel. X GO. Etched with alcoholic solution of picric aoid. It is well to point out immediately that in common parlance the ex- pression "burned steel" is often used to indicate a steel which has suffered dangerous but obscure modifications mentioned at the end of the previous .section due to an excessive heating. This use of the expression is not appropriate. Such a steel ought to be designated simply as "over-heated." It can be "restored" by means of a suitable heat treatment or hot-working and the previous qualities regenerated. A steel which has actually been "burned" is ruined and is good for nothing but remelting scrap. The term "burned steel" is used corrrectly to designate a steel which after its complete solidification has been reheated until it has reached the temperature of the final point of solidification in those of its parts which solidified last, i.e., where the concentration of the various constituents in soHd solution in iron is highest. It may be repeated with emphasis that no HEAT TREATMENT OF steel which has been "burned" cannot be restored by any mechanical or thermal treatment, short of remelting. In Chap. Ill, Sec. 25 and 26, the mechanism of the phenomenon of "burning" of steel has been already investigated quite fully. Only two actual examples of the characteristics shown in such material will be added here. Figure 20, at 60 diameters enlargement after etching with an alcoholic solution of picric acid, shows the microstructure of a 2 per cent, nickel steel only slightly burned during the heating subsequent to the first forging operation. Nevertheless, the deterioration of the physical properties due to such incipient "burning" is very remarkable. Such a steel as the type reproduced in Fig. 20 will give the following variations in physical prop- erties determined by tension test pieces cut longitudinally with the direc- tion of the forging: state of tlie metal Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elongation, per cent. (cylindrical test piece 13.8 X 50 mm.) Reduction of area, per cent. Appearance of tli6 fracture After first forging 89,300 54,000 24.0 37 Normal, at correct tem- fibrous perature and re- heating to 800°C, After incipient 45,600 Could not 2.5 Coarse, burning occurring be deter- shiny during heating for mined grains the second forg- ing. Figure 20 shows clearly the tenuous films of non-metallic matter — oxides and silicates — formed in the peripheral regions of the primary crystalline grains by oxidation of those portions which have suffered incipient fusion. When burning is more intense its effects upon the physical properties of the steel may become far more seribus until it produces such a disinte- gration in the metallic mass as could be actually characterized by the word "crumbling." From what we have seen in Chap. Ill, such a state of affairs naturally occurs more easily in the higher carburized steels and the decay is due to the fusion of the intergranular matter, annihilating crystalline adhesion. Figures 21 and 22 show one of these more serious cases occurring in a steel ingot containing 1.07 per cent, carbon and 2 per cent, chromium, charged by mistake into a heating furnace, together with other ingots of SOFT AND MEDIUM STEELS 111 Fig. 21. — Fragment of crumbling overheated steel. Half size. Fig. 22.— Badly overheated chromium steel. X 100. Etched with 5 per cent, picric acid in alcohol. 112 HEAT TREATMENT OF SOFT AND MEDIUM STEELS mild steel, and heated at the high forging temperature suitable only for the latter. The alloy ingot suffered true disintegration. Figure 21 at about half the natural size shows the appearance of one of the resulting fragments, and the "granular" structure assumed by the material may be clearly seen. The other Fig. 22 shows its microstructure at 100 diam- eters after etching with an alcoholic 5 per cent, solution of picric acid. Non-metallic inclusions may be clearly seen much more developed than in Fig. 20, where they were just forming. It is possible to increase the minimum burning temperature of a given steel by means of preliminary heat-treatrnents, as has already been described with sufficient precision in Sec. 25. CHAPTER XII SELECTION OF PRELIMINARY HEAT TREATMENT ACCORDING TO THE CHARACTERS OF THE y^a TRANSFORMATION 84. As mentioned at the end of Part II, there are large classes of steels whose critical interval on cooling occurs at very low temperatures, for a number of reasons. The same steels almost always present the phenomena of thermal hysteresis to a great degree. A combination of these two properties sometimes reaches such a sum that not only the end of the allotropic transformation on cooling but also the point at which it is initiated falls below the minimum temperature at which such reaction proceeds with measurable velocity. In other words, sometimes the speed of cooUng from above to below the transformation range as plotted on the equiUbrium diagram may be so high that the velocity of trans- formation may be considered as practically zero. In such cases we have seen that, for cooling velocities higher than a certain value, the y-^a transformation does not advance in perceptible measure. It follows that such metal retains at ordinary temperatures that degree of homo- geneity resulting from the conditions under which it solidified and from the subsequent homogeneity reheatings. Ordinary practice quite gen- erally allows a simple cooling in air after a heating. Yet in such practice the principal factor or, better, substantially the only factor determining whether the aUotropic transformation shall be completed or not, is the dimension of the different parts of the pieces treated. To allow all natural transitions to occur in pieces of irregular dimen- sions and special composition it is necessary to resort to special methods for slowing down the cooling after a normahzing treatment. Here it is necessary to observe that the existence of a greater or lesser lag in the 7— »a transformation depends in very large measure upon the effects reached by the same reheating. Other conditions being equal, the transformation will take place with more difficulty the greater the homogeneity produced by the preliminary reheating. It is easy to under- stand this point by referring to what was said in Chap. IX in connection with the relations between the dimensions and the position of the band of heterogeneous austenite and the temperature of the start of transforma- tion. Its commencement is at a lower temperature the narrower the band of variant concentration and the further it lies toward the right in the fictitious binary diagrams. For such steels, which are ordinarily austenitic at normal tempera- 8 113 114 HEAT TREATMENT OF tures, it is clear that the preliminary heat treatment is reduced to a re- heating for homogeneity. The third of the series of phenomena indicated in Sec. 81 is missing in them as one determining the characteristics of the heat treatment. Among the steels which belong to this category may be mentioned ternary steels high in manganese, containing from 9 to 14 per cent, of that element, or high in nickel, chromium, tungsten, and some other elements, besides many quaternary steels high in chromium and nickel, chromium and tungsten, etc. These steels find but limited use as machinery steels today mostly because their excellent properties are not well known by manufacturers. But there is no doubt that their applications will extend rapidly in that field, because their use proindes very great resources and very im- portant advantages, both in the manufacture of castings and in the manu- facture of rolled or forged parts. I will have occasion further on to refer to some examples of high alloy castings. 85. The great majority of the steels employed at present for machine parts possess a transformation range at equilibrium well above the blue heat, nor do they exhibit very wide thermal hysteresis. Therefore, the selection of the preliminary heat treatment in each case and the exact determination of the particular conditions for its execution, involves taking precise account of the phenomena belonging to the third of the series indicated at the beginning of Sec. 81; namely, of the phenomena which occur during the 7— >a transformation. Upon this point the general directive criterion apphed in the great majority of cases could not be simpler. It consists of an attempt to minimize the importance of the phenomena of liquation which we know always accompanies that transformation, causing fresh heterogeneity in the final mass, mainly of a iron. The practical difficulties in each case arise only in determining the limits within which one may work in compatibility with the other necessities of steel manufacture, as well as with the necessity of avoiding various troubles inherent in the process. Neglecting for the time being the practical difficulties just mentioned, and considering the problem from a more general and theoretical point of view, the general method to be followed to insure homogeneity by preventing liquation during transformation is easily found in the light of the considerations analyzed in Part II. I have already pointed out in Sec. 82 that it was not practicable to apply a process analogous to quenching which is capable of suppressing liquation during solidification. Instead, it is almost always possible to apply a true quenching process during allotropic transformation, and it is easy to understand that when applied so as to produce fully its characteristic effects, quenching must necessarily suppress any liquation effects accompanying the 7— >« transformation. In fact it is known that SOFT AND MEDIUM STEELS 115 in such cases the y—>a transformation described fully in Sec. 75 to 76 may be suppressed absolutely. Therefore, it appears that the simplest method to prevent hetero- geneity caused by allotropy consists in subjecting the steel to the most energetic quenching possible from a temperature above the initial point of the y—^a transformation. Other practical exigencies limit the selec- tion of the operating conditions. As a matter of fact it is necessary to resort to such so-called "homo- geneity quenchings" to prevent variation in concentrations simply be- cause it is not possible to resort to other means analogous to those which may be applied with good results to prevent or to minimize the hetero- geneities produced by Uquation during solidification. To apply processes similar to those applicable for the last-mentioned purpose, one ought to reheat the metal to a temperature below that of the end of the transformation. But this is impracticable first of all because at such a temperature, in all the steels to which this study applies — particularly steels having very low temperatures of transforma- tion — diffusion takes place at velocities so small that the effects cannot be felt in perceptible measure except when the reheating is prolonged for an extraordinary length of time. Under such conditions the method ceases to be applicable in manufacturing practice. Almost all machine steels decompose during the 7— >q: transformation into substances which cannot all enter into solid solution with the result- ing ferrite; that is to say, the composition lies within a field of insolubility (lacuna di miscibilitid) see page 39. In practice, this latter fact has a far greater importance than the one mentioned previously. Hence, trans- formation is accompanied by the segregation of the two phases stable at temperatures below the critical. Now we know that the properties of the resultant metal depend largely upon the forms and special struc- tures assumed by those phases, viz : saturated mixed crystals of a iron, or ferrite, secondly free cementite, and thirdly the conglomerate of the two, or pearUte. Those structures are determined mainly by the conditions surrounding the cooling of the system through the transformation inter- val. Also we know from the fundamental principles of metallurgy that they cannot be modified by means of reheatings at temperatures below the critical, except in very small measure, when the metal possesses those forms which it ordinarily assumes upon slow cooling. Reheating such material causes such effects as are due to phenomena proceeding at an extremely slow rate; for instance, the coalescence or balling-up of the structural elements constituting pearlite, into what is known as spheroidal cementite. We know, instead, that reheating to temperatures below the critical — a process more precisely indicated with the name of drawing — produces far more rapid and intense effects when applied to steels in which the 116 HEAT TREATMENT OF segregation of phases normally accompanying allotropic transformation has been partially or entirely prevented by quenching. 86. It is necessary to bear in mind some restrictions regarding the practical application of homogeneity quenchings. Due to the fact that the first step in a homogeneity quenching is nothing but a process of reheating to high temperature, which likewise constitutes the first essential phase of any reheating for homogeneity, it is clear that the same observations and restrictions indicated in previous sections apply to both. But other precautions and other restrictions essentially practical in character hold in the case of homogeneity quenchings. As clearly indicated in what has been said in the previous sections, the maximum efficiency of a preliminary quenching will be obtained when the fiijial phase, or quenching proper, is the most energetic possible. (This generally-used expression, "energetic," indicates that the steel has been cooled with the greatest possible velocity starting from the highest possible temperature.) This statement is on the basis that the coiiditions under which the first phase of the process has occurred may be disregarded. They have already been briefly investigated in the case of simple preliminary reheating. Now the indispensable practical restric- tions to the use of a too energetic quenching depend essentially upon the fact that in industry a piece of steel cannot be quickly cooled in a uniform way throughout the entire mass of the metal, especially if intricate in form or of large dimensions. In rapid cooling the metal is subject to certain temporary displacements, due to the different thermal contraction of the various parts of the piece, at each instant, or to perma- nent ones, due to the different degree of transformation of steel in the various parts cooled at different velocities, and therefore of permanent contraction or expansion. Such displacements in turn give place to so- called "internal stresses" which may reach such an intensity as to break the piece. It is clear that these phenomena have nothing to do with the inquiry under prosecution, therefore it would be out of place to investigate them at any length. It is enough to have indicated them in order to estabhsh the fact that, in practice, it is not always possible to make the final cooling at a velocity as high as would be required to derive the utmost benefit from the homogeneity quenching. 87. In practice it is often possible to eliminate, or at least to attenuate, many difficulties in the heat treatment of large or especially intricate pieces, by inverting the terms of the problem, so to speak. The minimum cooling velocity necessary to obtain and maintain a certain degree of homogeneity is not an absolute value, but depends upon the velocity of the transformations whose progress destroys pre-existing homo- geneity, or causes new heterogeneities. It is clear therefore that instead SOFT AND MEDIUM STEELS 117. of determining the most suitable quenching velocity for a given piece to obtain a required residual homogeneity, one may try to solve the problem of finding a steel of such a composition which gives the same residual heterogeneity after being subjected to a determined heat treatment. Having discovered such a composition, the quenching maybe accojnplished at the maximum velocity compatible with the dimensions, form, and nature of the piece under treatment. Such a problem appears constantly in the metallurgy of special steels, and it may be said that all the steel plants specializing in this field, solve the problem every day, but more or less approximately. Due to the fact that such a problem is very complicated in practice, it very seldom happens that it occurs in the explicit form under which I have thought useful to present it. At any rate, from what has been said in the previous sections, it is clear that the solution of the question is to be sought in the employment of steels with a low transformation point. The transition velocity in such alloys is in general the smaller, the lower the temperature at which it occurs. The right selection of one of those steels allows the operator to obtain a given degree of homogeneity with a heat treatment which does not include any quenchiiig whose "energy" (consequent to the heat from which the metal is quenched and to the velocity of cooling) exceeds a prede- termined limit. One may arrange steels of varying composition into a series whose prop- erties show a gradual change from those of the plain carbon steels to high alloy steels. The lower hmit will be a metal which can be normalized to the highest uniformity only when the treatments end with an energetic quenching. The upper limit will be the special steels in which the best obtainable results may be had by a simple reheating followed by a slow coohng. All gradations of cooling practice, such as quenching in cold brine, in water, in oil, or in molten lead; coohng in air blast, in still air, in ashes or in furnace, are used nowadays in steel making when employing these steels, whose varieties are extremely numerous. It may be noted here that steels with low transformation ranges have their critical temperatures lowered in direct proportion to the degree of homogeneity reached in the y sohd solution. This may be considered according to the conclusion reached in the preceding chapters, that is, that the beginning of the 7— >a transition is the higher the wider the band of variant concentration. Generally, no necessary connection exists between the most suitable temperature for homogeneity reheating and the correct temperature for the subsequent quenching, properly so-called. In fact, the first tempera- ture is limited only by a minimum value below which the diffusion takes place at a velocity too slow to cause effects which can be utiKzed in industry, and a maximum value resulting from the necessity of avoiding 118 HEAT TREATMENT OF over-heating and burning (see Sec. 82 and 83). The correct tempera- ture for quenching, on the other hand, is determined only by the position of the critical interval and by the velocity of the 7— >« transformation taking place within that interval. The independence of these two temperatures is a fact of great practical importance, worthy of the greatest attention in the study of normalizing. Especially is this true when dealing with special steel of low transforma- tion points. For the latter, only the suitable selection of two different temperatures may obtain the best results: namely, one for the reheating for homogeneity, and the other for subsequent quenching. In other words, it is always necessary to bear in mind that a homo- geneity quenching can never consist merely in a "quenching;" in the narrow meaning of the word, but its effect on the properties of a steel always results from the sum of the diffusion during annealing and the subsequent checking or attenuation of the liquation inherent in the 7— >a transformation, by means of a sufficiently rapid cooling. 88. It would be very easy to follow the gradual modifications that the equilibrium diagram of a given steel will undergo step by step with progressively increasing the cooling velocity. As was seen previously — in particular in Sec. 53 to 75 — a more rapid passage through the 7— >a transformation range results in a more incomplete attainment of the conditions of equilibrium in the mobile system. In other words, the heat treatment to which the steel is subjected gradually loses the charac- teristics of a true reheating and assumes those of a quenching proper. These considerations are not analyzed here, because such a develop- ment wotild consist of nothing else but a repetition of what has been said before regarding the perturbations which are caused by each variation of the cooling velocity in the principal lines of the diagrams reproduced in Fig. 12, 13, 14 and 17. 89. In the great majority of cases normalizing is practiced by taking essential account of the principal aim, i.e., to reduce segregation normally accompanying the 7— >q; transformation to the minimum value practically obtainable, without subordinating in any way the selection of the best means of reaching this purpose to the special physical and mechanical characteristics which the metal should eventually possess. In general, it follows that normalizing (which for this very reason is called a "prehminary heat treatment") is followed by others more or less complicated, called "final heat treatments," which have as their essential purpose the adjustment of the metal to the special and oftentimes very narrow limits of mechanical properties often specified as necessary for its intended employment. From what I have said several times, the analysis of the conditions under which these "final" or "quality" heat treatments must take plade, would go beyond the limits of this study. However, it is timely to point SOFT AND MEDIUM STEELS 119 out here that the conditions under which the "quality" treatment must be executed, are already subject to the restriction that said treatment must not obliterate totally or in too substantial amount the useful effects obtained by normalizing. In general, the result now indicated is almost always sought by operat- ing in such a way that the final heat treatment does not involve reheating the steel to a temperature above the Ac transformation points. The sub- sequent cooling can no longer be selected at will, since its velocity is determined by the final mechanical properties of the piece. If, then, at this point, the range where the a—>y transition proceeds is entered, the reverse reaction on leisurely cooling may allow sufficient segregation to neutrahze the effects of the previous equalization. In the practical application of this criterion, wide use is made of the phenomena of thermal hysteresis. It causes the a—yy transformation on heating to occur at a far higher temperatm-e than on cooling, and so largely widens the range of temperatiu-es which may be utilized in executing the final heat treat- ment under the best conditions. 90. Reheating to a temperature below the critical can be almost always utilized in practice in the simplest of all "quality" heat treatments. In this case there is no other purpose than more or less completely converting the metastable system to the stable one. In all but lowest carbon steels, a homogeneity quenching results in a complex, usually constituted mainly of austenite and martensite, but sometimes mixed with troostite. Reheating such material simply endeavors to pro- duce the required mechanical properties in the finished steel by main- taining the temperature as low as possible and yet allow a determined amount of segregation in the stable phases, constituting troostite, sorbite, and pearlite. Evidently in such a case it is always a prime en- deavor to avoid going above the transformation point on reheating. In other words, the final treatment is restricted to a true "drawing" operation instead of pushing the temperature up to the limits of a true "annealing." Rather than get a rapid rearrangement at high heat, it is then better to get the same result by prolonging the length of the heating necessary to impart to the steel the desired mechanical properties. In other cases, such a restriction may not be imposed upon the temperatures to be reached in the final operation. As an example, recol- lect what happens every time the final heat treatment must include a "hardening." Essentially, hardening imparts a determined structure to the solid solution constituting the steel after quenching, corresponding, generally, to a given ratio between the first products of the transformation of austenite; i.e., martensite and troostite. For the same reasons already given it is always the endeavor to limit temperature of heating as closely as possible and to maintain the metal at this "lowest maxi- mum " temperature for the shortest possible length of time. 120 HEAT TREATMENT OF The realization of the conditions now mentioned becomes very difficult in many instances on account of the nature of the steel em- ployed, the form of the pieces to be treated, and, finally, on account of different composition of the metal in various parts of the same piece. A piece which is thick in one section while thin in another, is delicate to handle since it may be impossible to reach determined heating or cooling velocities in some parts of them without causing those velocities to be far different in others. Variation in chemical composition may be the result of design, as happens, for instance, in partially case- hardened steels. On the other hand, a piece may be formed by various parts joined together in any way purposely, by welding or otherwise, or accidentally, as happens in raw steel which during solidification has undergone an intense segregation called "great liquation. "^ Discon- tinuities of this sort greatly increases the difficulties inherent to the selection of a final heat treatment rationally coordinated with the preliminary treatmient. A practical solution of these problems often necessitates the most complicated programs. 91. In making a complete heat treatment, it is possible only in ex- ceptional cases to disregard the considerations mentioned in the two previous sections. This happens when it is possible to select a pre- liminary heat treatment which, although being sufficiently efficacious as such, is also capable of imparting to the steel those special physical properties fitting it for use, alone and without any further treatment. This occurs very seldom. It may be due to an accidental coincidence between the most suitable conditions for the preliminary and the final treatment. Or less often it may be due to the less severe specifications which the metal must fulfill. In the last case, it sometimes actually happens that the wider tolerance allowable both for temperature and velocity of cooling . when executing each of the two treatments, may cause an overlap of determined regions in the two fields. In that case it is sometimes possible to obtain results sufficient for limited require- ments with only one treatment, by working within the limits of that common region. An accidental coincidence of optimum heat treatments is the most rare of the two cases just above indicated. It occurs almost exclusively for very mild steels, in which the physical properties of the mixture of metastable constituents resulting directly from even the most energetic prehminary quenching corresponds quite well to the require- ments which these steels must ordinarily satisfy. However, it is necessary to bear in mind that, even in this most favor- able case omission of the final drawing cannot be suggested except for a possible saving in expense. No matter how mild or low in carbon and manganese is the steel, its use after full quenching never insures the 'See page 150, paragraph 111-I>. SOFT AND MEDIUM STEELS 121 guaranteed reliability confidently obtained when a steel a little "harder" in chemical composition is used after being suitably drawn following a homogeneity quenching. This is due mainly to the fact that drawing, when suitably done, is the only means for efficiently eliminating internal stresses caused by quenching, especially if energetic, as must be a homo- geneity quenching for very mild steels with high critical ranges. These stresses may reach dangerous amounts especially in pieces of large di- mensions and of intricate, unsymmetrical forms, at times even causing spontaneous fracture. An insufficient or neglected drawing may often cause other troubles which, though less dangerous than breakage, represent a heavy loss. For this reason, false ideas of economy may result in net waste. As an ex- ample may be mentioned pieces of comphcated forms, such as a 6-cylinder crank shaft, which must be machined with great precision. If these pieces are subjected to an insufficient annealing after the preliminary quenching they sometimes retain internal stresses capable of warping them so badly during the finishing that they are rendered useless. Such machine work cuts away layers of metal which formerly partially bal- anced those tensions. 92. An overlap in the preliminary and final heat treatments may for instance, occur when a steel is used for a certain purpose requiring much less than its full possibilities. Thus, in working a steel belonging to a type harder than woiild be required, one could replace the true prelimi- nary quenching by an intermediate treatment lying somewhere between a quenching and a simple annealing. The nature of such so-called "incomplete quenching" has been ex- plained before at sufficient length when discussing special steels of low transformation-point. Therefore, it is superfluous to return to that argu- ment. Considering the process from the special point of view which we are now particularly concerned with, namely, that of the final physical properties of the steel, it must be added that other things being equal the greater difference between the results obtainable by an energetic pre- liminary quenching followed by drawing and the results obtainable by an "incomplete quenching" (more or less similar to a reheating) consists in the fact that in the second case the metal shows an elastic limit very much lower than in the first, yet without an equal difference be- tween the values for tensile strength. In other words, the double treat- ment produces a much higher value of the so-called "elastic ratio" of the metal (elastic limit divided by the ultimate strength, a ratio almost al- ways greater than one-half). When the uses to which the steel will be put do not require a high elastic limit, or when it is possible to use a harder steel than necessary without fear of dangerous consequences from a lower toughness, it may 122 HEAT TREATMENT OF SOFT AND MEDIUM STEELS be sometimes convenient to employ the process of "incomplete prelimi- nary quenching" just described. A comparison will be given further on, based upon experimental figures, of the differences between the physical properties and the structures obtained by each method. As far as the structures are concerned, it is even now easy to account for these diiferences, if one bears in mind the considerations developed in Part II. The relation between the degree of homogeneity and its influence upon the course of the phenomena of seg- regation accompanying the 7— >a transformation were discussed there at length. Perhaps it is superfluous to point out that, in the industrial applica- tion of the process of "incomplete quenching" just described an accurate study is necessary for each individual case to determine the more suitable "limitations" to the process of "normal quenching" in order to produce an "incomplete quenching." It is known that these limitations may consist essentially of taking a longer time to cross the transformation ranges; or else in interrupting the rapid cooUng of the quenching proper at a temperature below the initial point of the 7— ^a transformation, from which point follows a much slower cooling to room temperature. CHAPTER XIII PHENOMENA WHICH MAY DISTURB THE COURSE AND MODIFY THE EFFECTS OF PRELIMINARY HEAT TREATMENTS 93. Before closing the general discussion forming Part III, it may be said that the action of heat treatments is liable to many disturbances due to a great number of different causes, some of which produce effects of the greatest practical importance. It is not possible to make a complete study of these causes and effects in these pages. Two groups of facts may be mentioned as examples, selecting them from among those which occur more frequently in steel making and which have the greater practical importance. They have profound effects upon heat treatment processes; some concrete examples of their ultimate effects will be reserved for later mention. The first of these groups includes occurrences during hot-work, such as roUing, pressing, foi-ging, or drop forging the steel, evidently operations entirely normal to ordinary metallurgical practice. Yet they may cause important changes in the course of the phenomena underlying preliminary heat treatments and result in strong modifications in the properties which may be obtained in the same steels by means of a given heat treatment. On the other hand, the second group includes phenomena which ought, theoretically, to be considered as being accidental. They are connected with the presence of solid non-metallic inclusions or "sonims"' formed in the various phases of melting, refining, pouring, teeming and crystalhzation of steel. It has been remarked that the phenomena due to the presence of sohd non-metallic inclusions in steel ought to be considered as accidental. Sonims certainly cannot be considered as normal constituents. On the contrary, they are extraneous bodies held in a metallic mass which re- mained too short a time in a sufficiently fluid state to allow such inclu- sions to separate from the mass by reason of difference of density. But I added that this fact can be stated only "theoretically" because, in reality, it may be said that no steel exists if made under normal conditions of manufacturing practice which does not contain soMd non-metallic inclusions in larger or smaller proportions. 1 We shall use the accepted word "sonims" throughout in place of solid non- metaUic inclusions. 123 124 HEAT TREATMENT OF Some numerical data concerning special cases will be given later showing how the presence of slag inclusions may strongly alter the effects of preliminary heat treatments. 94. It may now be indicated how some of the general characteristics of hot-work place it within the sphere of the processes studied in the previous chapters. Later, a few particular cases will be cited to illustrate its technical effects. It would be easy to demonstrate on the basis of actual experimental data the fact dwelt upon later that the characteristic effects of hot-work are not due to compression, except perhaps in a very small measure which certainly may be entirely ignored in practice. There can no longer be a doubt about the correctness of such a statement although it is openly in disagreement with ideas generally accepted by the so-called "practical" men. By mere exclusion the effects produced by such mechanical operations are essentially due to the deformations which the steel suffers, a conclu- sion fully confirmed by experience. It is necessary to distinguish two series of effects caused by such defor- mations. In practice, the phenomena of both series always intervene contemporaneously in determining the net effect of rolling, forging, and the like. While one series has hardly any noticeable effects upon the course and results of a normahzing treatment, the other series deeply modifies the issue. Phenomena of a nature similar to those in Sec. 82, belong to the first series, namely, those connected with modifications of crystallographic structure, properly speaking. Numerous and very interesting observa- tions exist regarding this class and permit the mechanism to be clearly and definitely explained. They show characteristics entirely analogous to those already thoroughly studied by the science of crystallography. Movement of the hot metal under outside forces is essentially a " sliding " or "reciprocal shpping" of crystaUine elements or portions of such ele- ments along definite surfaces, and of more or less well developed "twin- ning. " The trend of these phenomena in the course of hot-work and the conditions of temperature under which they take place, exercise the greatest practical influence, causing intense modifications of the physical properties of the resulting piece. This statement holds regardless of the direction in which the article is tested, whether with or across the direc- tion of forging. Therefore it is evident how great is the practical im- portance of an accurate study of these circumstances. However, as has already been indicated, experience shows and simple theory readily supports the experimental conclusion that the mechanism of sUp does not perceptibly modify the operation of diffusion and other fundamentals upon which normahzing practice is based. It is, therefore, clear that their detailed investigation would be out of place in these pages, SOFT AND MEDIUM STEELS 125 although it would be indispensable as a basis of any study of mechanical hot-work. Conditions are different for phenomena of the second series. De- formations which steel suffers under the action of hot-work, produce large reductions of one or more of the dimensions of the primary and secondary crystalline elements. It is clear that this last fact must exercise a very strong action upon phenomena of diffusion, which indeed constitutes the main basis of preUminary heat treatment. From the discussion given in the first two parts of our study, and other conditions being equal, it is easily understood that a large deformation in any direction wiU correspondingly reduce the distances between the points corresponding to the maximum and minimum values of the con- centrations of the various elements in soUd solution in iron, and must lower the time necessary for a given homogeneity treatment to cause a determined effect. These expectations are fully confirmed, and we shall later see some examples which prove, by experience, that a given normahzation is far more efficacious when applied to a suitably forged or rolled steel than it is when applied to the same steel in the raw state. 95. Apropos of the phenomena which have been discussed in the pre- vious paragraph, one may point out that the observations regarding the crystallization of steel, indicated in Sec. 81, also hold true regarding the practical application of the processes based upon the phenomena accom- panying hot-working. It seldom happens except in very rare and commendable cases, that the study and the practical application of the processes aimed to suitably regulate the course of those phenomena, are a part of the duties entrusted to the man who has charge of studying and applying aU the other processes constituting the ulterior steps in the heat treatment. For instance, in nearly all steel plants the superintendent of the rolhng mill or forge shop has nothing to do with the heat treatment, and vice versa. This place is hardly suitable to discuss instances and methods whereby the hnes of authority could be advantageously modified. I only want to point out the fact in order to emphasize the necessity of perfect coordination in the plant as ordinarily organized between the operating methods followed in the mill and in the heat treatment department. To be convinced of this it is only necessary to remember the before mentioned facts, that the selection of the most suitable heat treatment is very strongly influenced not only by the special conditions under which the steel has been made and cast, but also the hot-work to which it has been subjected. From this special point of view in particular, the tem- perature at which mechanical working has been done has a preponderant practical importance beyond all the other considerations. Merely as an example, it is enough to remember that if a forging or rolling is executed 126 HEAT TREATMENT OF or even simply finished at too low a temperature, it gives the piece of steel certain pecuUar properties due to cold-work.' An overstrained steel cannot be reheated above a determined limit, generally very low, without seriously "coarsening" the grain (see Sec. 82), a structural occurrence always accompanied by a great increase in fragihty. The correct heat treatment of such material must be subordinated to special limitations of time and temperature whose determination is very dehcate and diffi- cult. Neither can it be made rightly except by a person possessing all the data concerning the conditions under which the mechanical work has taken place. Phenomena of plastic deformation assume great importance especially when working certain groups of special steels. Mechanical properties must be imparted to them which probably can be reached only by means of mechanical and heat treatments executed under rigorously determined conditions within extremely restricted tolerances. For reasons which have been repeatedly outlined, it is not possible to deal, in detail, with these phenomena here. They have been men- tioned only as an example of facts which may modify the normal train of events underlying normalizing practice. Merely as a matter of information, I wish to add that some steel plants, particularly those manufacturing special steels, have sometimes ehminated the evident inconveniences due to separate responsibility for the technical operations comprising the various steps of steel-making and working, by creating a kind of metallurgical supervisor-general, to observe all details of the manufacturing processes and report directly to the technical manager of the plant. 96. The second group of facts which have been mentioned in Sec. 93 as seriously modifying the trend of events analyzed in the first two parts of this study are due to the presence of non-metallic solid bodies included in the steel. I intend to give further on some actual examples of these phenomena, as well as of those of the previous groups, designed to give a sufficiently exact idea of their real practical importance. However, it is now advisa- ble to record some general observations concerning the particular charac- teristics by which they are more directly connected with the subject matter of the previous chapters. . Inasmuch as the topic is of the greatest practical importance but usually studied very cursorily in treatises on metallurgy and metallography, especially from the point of view which more directly connects it to the phenomena which we are studying, it will not seem amiss to dwell here a little longer upon it. A large variety of non-metallic solid inclusions are found in steel 'Following Howe, "plastic deformation" in the cold is identical with "cold- work" (French "ecrouissage," German "Kaltrecken," Italian "incrudimento") and "overstrain." SOFT AND MEDIUM STEELS 127 made by various processes now in operation. It is also known that the practical consequences resulting in a given steel from the presence of too large a quantity of any kind of nonmetallic solid inclusions may be very severe. Yet even extremely small quantities of inclusions are fatal if they belong to certain determined types. In order that the conceptions of the nature and characteristics of the influence of certain non-metallic inclusions may not be too undetermined and vague it is necessary to explain first some notions concerning the nature, origin and the principal properties of the various types of those inclusions which are more frequently found. In usual works' parlance, all visible non-metallic solid inclusions are usually indicated with the same generic designation of "slag inclu- sions." Indeed nothing prevents the adoption of such nomenclature provided it is clearly understood that the word "slag" shall not imply the notion that those inclusions are particles of that slag, properly so- called, which constantly forms upon the molten metallic bath in all steel-making processes. In fact, it now is known with certainty that many inclusions appearing in finished steel have nothing to do with the true slag whose formation is an integral and essential part in steel-making and refining. In order to avoid any misunderstanding, it is better, therefore, to adopt the expression "solid non-metallic inclusions" or "sonims." This explained, let us briefly note the principal types of inclusions, differentiating the ones which exercise an appreciable action upon the course of the phenomena with which we are concerned. Then we shall see further on how it is possible to formulate some general hypothesis to explain the mechanism of these effects and to deduce from it useful indications regarding the way of counteracting them. 97. Accurate microscopic examination of all steels — especially if made upon perfectly polished samples before being etched, and observed under high enlargements — always discloses a more or less large quantity of solid non-metaUic inclusions very small in dimension (of the order of a few hundredths of a milhmeter) and of uniform and characteristic forms and colors. They are corpuscle's of roundish form and grayish color, scattered in irregular manner in the various parts of the mass. Figure 23 reproduces at 200 dia. an example of the more frequent appearance which these inclusions present. This micrograph refers to an unetched section cut from an ingot of mild carbon steel made by the acid open-hearth process. Inclusions of the type illustrated are usually designated by the term "emulsified" inclusions. They occur in nearly all steels obtained by fusion in any of the commercial processes now used. The dimensions and the number of the grains is the only variation. We shall see soon that the relations between the reciprocal positions of the individual inclusions 128 HEAT TREATMENT OF as well as between those particles and the normal structural elements of steel, change with the conditions under which the steel has been melted, refined, poured, teemed and treated. We shall also see that these variations make their effects felt sometimes in very large measure upon the mechanical properties of the finished metal. For this reason emulsified inclusions may be considered as those which exercise the greatest modifications upon the phenomena accompany- ing heat treatment, and, therefore, indirectly upon the general physical properties of steel. From this point of view their industrial effects differ decidedtyfrom those produced by inclusions belonging to other types. Either because of their greater dimensions or difference in shape and constitution, these latter also exercise a direct action upon the properties Fig. 23. — Emulsified inclusions in acid open-hearth mild steel. X 200. Unetched. of steel, sometimes even a great deal more intense than those produced by the emulsified slags. However, this action is permanent in the sense that it is not appreciably modified by heat treatments, and in adcUtion has a localized influence. These last words should be understood in the sense that the solid non-metallic inclusions other than "emulsified inclusions" appear with a much smaller frequency (that is to say, the number of the particles is much smaller per unit of volume), and are ordinarily distributed in a much more irregular manner throughout the metallic mass forming isolated "groups" or "nodules." As a consequence it is always easy to detect localized modifications in mechanical proper- ties of the metal around the single inclusions or groups of them. These modifications almost always consist in a local decrease of resistance so SOFT AND MEDIUM STEELS 129 that when the metal is tested to destruction the fracture exhibits special characteristics and is located at the inclusion or groups of them. 98. Following the program which has already been indicated at the beginning, I should give a concrete idea of the specific characteristics of the modifications and alterations caused by emulsified inclusions, only after having tried to give a Uke concrete idea of the normal effects pro- duced by the preliminary heat treatment. However, inasmuch as I have alrcadj^ mentioned the external charac- teristics which distinguish emulsified inclusions from those of other types, I think it timely to briefly indicate the relative positions normally held by the individual particles of sonims and the ferrite, cementite, pearlite or other well-known structural elements in our steels. In this way, when Fig. 24. — Medium steel with inclusions enclosed in ferrite. X 80. Etched with picric acid. the modifications of mechanical properties shall be exhibited by means of actual examples of a determined preliminary heat treatment upon steel containing emulsified inclusions, the modifications produced by such inclusions upon the normal structure of steels will also l)e much clearer. Ordinary relations between particles of emulsified inclusions and the structural elements of steel may be easily seen in microscopical examina- tion of the raw steel, just as it has been poured. Polish the sample and etch slightly with one of the usual reagents, such as an alcoholic solution of 5 per cent, picric acid, or a solution of nitric acid in amyl alcohol. In the great majority of cases the emulsified inclusions are found inside crystals of ferrite. Ferrite totally surrounds those sonims follow- 9 130 HEAT TREATMENT OF ing the direction of their eventual alignments so that said inclusions are not in contact with pearlite. The two following figures show two characteristic examples of such structures. The first (Fig. 24) magnified 80 dia., reproduces the struc- ture of a medium-hard carbon steel in which the emulsified inclusions appear isolated and surrounded by a nimbus or halo of ferrite. Figure 25 reproduces at 80 dia. the structure of another medium hard carbon steel, in which the emulsified inclusions appear in orderly lines, and each string is entirely surrounded by ferrite. It happens only in rare cases that raw steels which have not yet been subjected to any work or heat treatment do not show either of these Fig. 2.5. — Emulsified inclusions in ranks directing disposition of ferrite veins. X SO. Etched with picric acid. characteristic arrangements. In such cases the emulsified inclusions are partly or entirely in contact with pearlite. An example of such is shown in Fig. 26, which reproduces (magnifica- tion 120 dia.) the structure of a medium hard 2 per cent, nickel steel. The figure does not require any explanation. A series of experiments^ makes it practically certain that the occur- rence of elements of ferrite surrounding the particles of slag or its inde- pendence from the position of such particles depends upon the state of oxidation in the system. Or, better, it depends upon the ratios between the oxygen pressure of the non-metallic particles and of the metal, respectively, and from the variations which this ratio suffers during crystallization and the subsequent cooling. When these variations are ' Chemical & Metallurgical Engineering, Vol. 24. SOFT AND MEDIUM STEELS 131 such as to give rise to a process of oxidation, it forms a layer of ferrite which evelops each non-metalhc particle during the cooling of the system. Such oxidation is due to an excess of CO2 in the atmosphere of dissolved gases above the amount in equihbrium with CO and C in the average sur- rounding metal. By the well known reaction COo + FeaC ^ 2C0 + 3Fe excess CO2 shifts the reaction to the right, decarburizing cementite and producing a layer of ferrite around each element of the inclusions at a given instant of the process. This first layer acts during subsequent coohng of the system through the transformation interval y-^a as a crystaUization germ attracting the further quantities of ferrite which separate during the critical range. Fig. 2G. — Medium 2 per nent. nickel steel containing emulsified inclusion. X 120. The reasoning which leads to this conclusion is based upon many observations of different nature which are not proper to develop here, but which are in part founded upon the study of the modifications which heat treatment produces in the relations between emulsified inclusions and ferrite. It is clear that such reactions impress an orientation to the ferrite and pearlite which differs greatly from the characteristic "normal" orientation whose genesis we examined in the second part of this study, because an oxidized inclusion surrounds itself with a thin sheath of ferrite, which in turn accumulates much more ferrite by acting as a series of crystallization centers. It is now seen that the presence of emulsified inclusions has immensely altered both the normal segregation of ferrite 132 HEAT TREATMENT OF accompanying the 7— >a transformation as we have studied it up to now, and the structures which are thereby produced. 99. Since "emulsified inclusions" are nearly always present in steel, they perhaps assume almost a "physiological" character, stretching this adjective to mean that their occurrence practically is a necessary conse- quence of normal steel-making processes. On the contrary, one may say that the presence of non-metallic solid inclusions belonging to all the other types constitutes a phenomenon of a "pathological" character, meaning that such inclusions are formed and have been retained in the metal by an abnormal and defective course of one or more of the steel-making operations. It is necessary to quaUfy these statements by remarking that the presence of the emulsified inclusions may also assume a pathological character in a steel too highly "oxidized "when they appear in thin lami- nations (Fig. 25) and when at the same time their effects upon the agglomeration and orientation of the ferrite are of exceptional intensity. Contrary to what is true for the other types of inclusions, the patholog- ical character of emulsified sonims is indirect, inasmuch as it is due not to the inclusions per se., but to their influence in the special disposi- tion or positioning ■ of the ferrite, which latter is of course a perfectly normal structural element of steel. In the greater majority of cases this effect is happily susceptible of strong modifications, and sometimes complete elimination, by means of suitable heat treatments. 100. Owing to the immense practical importance which the presence of non-metallic solids assumes in the study of the problems related with steel making, treatment and usage, and admitting the profound differ- ences indicated in the previous paragraph between the behavior and the technical effects of various types of inclusions, it is clear that an exact knowledge of the nature and origin of such inclusions would have a very great industrial importance. The problem has been already studied attentively, although perhaps not with the thoroughness which it deserves. It is now possible to describe precisely the nature and the origin of some types of inclusions, while of many other types it is not possible to say how they arrived nor what is their constitution. True enough, emulsified inclusions fall within this terra-incognita, even though as we have seen and shall see still better they exercise the greatest effects upon the course and results of homogeneity heat treat- ments. It would be impossible to present here with sufficient brevity a condensed discussion and criticism of the results obtained up to the present day by researches attempting to throw some light upon this question. Also this would be still more inopportune since the conclusions of such a discussion certainly would not add clearness nor precision to the approximate ideas which we may yet form of the entire problem. SOFT AND MEDIUM STEELS 133 Therefore, I prefer merely to indicate briefly the hypothesis concern- ing the nature and the origin of the emulsified sonims which appears most plausible at the present state of our knowledge. Opinions of various experimentors are extremely various concerning their chemical constitution. However, none is founded upon sure and direct experimental data such as could only be furnished by a chemical analysis of the material actually constituting the inclusions. It has not yet been possible to make such an analysis due to difficulties inherent to a mechanical separation from the surrounding metal in order to gather a sufficient quantity for testing. Chemical methods for the separation of inclusions from the metal are founded upon the emplojmient of solvents for the iron. Such a procedure has not yet furnished rehable results, because all the reagents suggested cause more or less profound alterations in the material forming the inclusions. It would take too long, as well as being out of place here, to list the conclusions reached by various experimentors on this subject based upon various methods of indirect investigation. It is enough to say that according to the opinion of many, emulsified inclusions are constituted exclusively or almost entirely by manganese sulphide or by mixtures of sulphides of manganese and of iron. According to others, these inclu- sions would be usually formed by oxysulphides of iron and of imanganese, while there is another who thinks that they also contain silicates. Fi- nally another opinion seems perhaps more likely, both because it conforms better with plausible hypotheses advanced to explain the origin of other types of inclusions, and because the more common emulsified inclusions show substantially the same properties in steels relatively high in sulphur content, as well as in those almost free from sulphur. According to this opinion, emulsified inclusions are essentially silicates in addition to sul- phides, sihcates of iron and of manganese, united to other silicates — as those of calcium, aluminum, magnesium, etc. — whose nature depends upon the kind of refractory materials with which the molten steel has come in contact. The proportionate amount of sulphides would vary with the average percentage of sulphur in the materials used in making the steel; but in general it would be somewhat low, so as to leave the preponderance of mass to the silicates. In other words: the composition of the emulsified inclusions would be fundamentally similar to that of the slag proper which forms during the operations of steel melting and refining, and which appear floating upon the molten metallic bath at given periods of the process. 101. Regarding the origin of non-metaUic inclusions, it is certain that it varies a great deal for the various type of inclusions. The more persistent, and to us more interesting, emulsified sonims are attributed by a very well founded hypothesis to an origin in phenomena similar to those occurring in aqueous suspensions of finely divided bodies, 134 HEAT TREATMENT OF or, more precisely, in true emulsions, formed by minute drops of a more or less viscous liquid suspended in another liquid with which the former cannot mix. It is from this analogy that the expression emulsified inclusions has originated. Inasmuch as it is impossible to enter upon a detailed discussion of the reasons why the hypothesis above indicated appears credible, we will only outline its content. During steel-making the formation of slag constantly accompanies the chemical reactions occurring along with fusion and refining, whatever the manufacturing process may be. These molten non-metallic substances forming the slag proper exist largely in the form of globules of different dimensions in suspension in the liquid steel, when it is more or less agi- tated by the energetic reactions which take place in it accompanied by evolution of gas. As soon as this period of agitation, which in certain measure always appears in all "fusion" processes, comes to an end, it is succeeded by a quiet period during which the fitial deoxidation reactions take place iii the metallic bath upon the addition of deoxidizing alloys, mainly ferro-manganese, ferro-silicon, or other alloys having silicon and manganese as basic elements. During these relatively quiet periods, all suspended slag globules tend to separate on account of a difference in density, a process analogous to that taking place when settling a turbid liquid. However in our case perhaps it is more exact to make a compari- son with an oily emulsion left in repose, when the oily drops gather little by little in groups, coalescing into larger droplets by the action of surface tension. These large drops ascend toward the surface of the denser supporting liquid, at a more rapid rate the larger their dimensions. Spring was one of the first to study this subject accurately. He indicated the causes of clarification of such mixtures under the general ex- pression floculations des milieux troubles (flocculation of turbid liquids). While flocculation is a pronounced reason, it is reinforced by the viscosity and some other properties of the dispersion medium very difficult to evaluate with the usual physical and chemical criteria. One needs only remember, in this regard, that the velocity of sedimentation is enormously increased in suspensions and watery emulsions by the addi- tion of very small quantities of some electrolites. It is also true that the entity and the nature of the phenomenon of sedimentation vary within wide limits with the nature of the added electrolite. 102. Now consider the mass of steel as an emulsion immediately after the period of agitation ceases; namely, after the bath ceases to "boil" in the Siemens-Martin refining process, or after the Bessemer " blow. " If we suppose that the process of separation of the suspended slag starts at this point, due to difference of density, and proceeds according to laws similar to those which regulate the corresponding separation occurring in aqueous emulsions, we may expect the following principal occurrences : SOFT AND MEDIUM STEELS 135 (a) Drops in suspension, of size larger than a certain " critical dimen- sion" will ascend with relatively high velocity toward the surface of the bath, there to unite with the floating layer of molten slag. (6) The same thing will happen for the drops resulting from the ag- glomeration of the smaller, step by step as their increasing dimension will have reached the critical. (c) Suspended droplets which have dimensions equal to or smaller than the above indicated limit will rise in the bath with velocities extra- ordinarily smaller than that reached by the larger drops mentioned in (a) and (6). Consequently, after a certain period of time, the undisturbed bath will no longer contain inclusions of molten slag, except those which are in the form of minute drops whose dimensions are equal to or smaller than the "critical" — a fact very well established for aqueous suspensions and emulsions. Under normal conditions of temperature and for steels of ordinary composition, the time necessary for substantially complete separation of the inclusions indicated in (a) and (6) is relatively short and of such an order as to render it possible in practice to maintain the molten metallic bath quietly so as to eliminate those inclusions. However, the velocity with which the particles of the dimensions speci- fied in (c) rise in the molten metalUc mass is so far below that of the others, that it would be practically impossible to keep the melt in quietude during a time long enough to allow their separation. Therefore, these remain in the metal and constitute those particles which we have named emulsified inclusions appearing in solidified steel. 103. The hypothesis briefly outhned in the two previous sections corresponds well to characteristics shown by the principal types of non- metaUic sohd inclusions. It explains the normal presence of the emul- sified inclusions in steel and their constant characters. It also explains the fact that the presence of other types has the characteristics of an acci- dental and abnormal phenomenon. A further similarity to liquid sus- pension lies in the fact, well known to the steel melter,. that with each variation in the quaUty' of the finished steel, the nature, state and number of the non-metallic solid inclusions also varies within very wide limits, other conditions being equal. This last fact is only in part due to the different "viscosity" which characterizes various steels under normal conditions of melting and pouring. In fact, in many cases it certainly appears that the viscosity of the metal is not competent to explain the extreme variation in numbers and dimensions of the inclusions. Therefore, it is necessary to admit the ' By the term "quality" is meant not only the chemical composition determinable by mean of analyses ordinarily made, but also in the larger sense, to include the state of oxidation, the nature of the gases dissolved in the metal, etc. 130 HEAT TREATMENT OF intcrvcnijVm of other causes which might be looked for in phenomena similar if) tliose canscd by electrolites upon colloidal suspensions. This is not the place to discuss the probable nature and the character- istics of these last phenomena as they apply to steel-making. Besides, very little accurate information is known on this subject. 104. The considerations briefly condensed in the previous sections assign a character and a definite field of existence only to the non-metalhc inclusions originating from the slag proper. Of these, it has already been said that only the inclusions whose dimensions are less than a certain definite limit (emulsified inclusions properly so-called) show pecuharities which might almost induce one to consider them as normal structural elements of the steels made by modern commercial processes. Therefore, Fig Large particle of slag in steel. X 50. it is necessary to keep account of these alone in a study of the general properties of steel. And inasmuch as their effects upon the object of our study may sometimes assume a great material importance since they are capable of modifying the general properties of steel obtained through normalizing treatments, I will mention, later on, some example of those effects. Instead, all other sonims without distinction must be excluded from the number of substances which normally modify the general properties of steel and therefore from the number which is necessary to take into account in a general study of the preliminary heat treatment of steels. 105. Merely as a matter of information, it may be timely to add some examples of those other non-metallic inclusions which simply represent accidental defects in the steel. SOFT AND MEDIUM STEELS 137 Bearing in mind the hypothesis formicated in the previous sections, the first to be taken into consideration anions fhe inchisions of this category are those originating; from the nietaUurgical slag but which, Fig. 28. — Slag inclusion showing two constituents. X 100. Fig. 29. — Globular slag inclusion. X 100. normally, should separate before the solidification of the metal — namely the inclusions which in Sec. 102 have been classified under the letters (a) and lb). 138 HEAT TREATMENT OF These inclusions sometimes reach hirge dimensions. An example is reproduced in Fig. 27, enlarged 50 dia. Often they show distinctly the two characteristic and well known constituents of the slag from Fig. .so. — Serpentine lamination of slag. X 100. Fio. 31. — Slag inclusion of unknown origin. X 100. whence they come. A very evident example of this fact is given by the inclusion reproduced in Fig. 28, enlarged 100 dia. Frequentlj' inclusions of this kind are found in the form of globules SOFT AND MEDIUM STEELS 139 or drops, a fact which confirms the cxphinatioii given above of their origin. An example is given in Fig. 29, enlarged 100 dia. Fig. 32. — Particles of refractory caught up by molten steel. X 100. Fig. 33. — Particle of refractory caught up by molten steel. X 100. It also often happens, especially in steels highly "oxidized" in melt- ing, that the large inclusions assume the form of long, thin laminations, more or less twisted as serpentine filaments in the section examined under the microscope. No satisfactory explanation is known of this 140 HEAT TREATMENT OF peculiar occurrence, an example of which is reproduced is Fig. 30, enlarged 100 dia. In order not to dwell too much upon this subject, which presents but a negative interest to normaliz- ing practice. Fig. 31, 32 and 33 at 100 dia. are given to show three ex- amples of the appearances which may be presented by inclusions hav- ing neither the characteristics of the emulsified inclusions, nor those of slag particles proper. The origin of such inclusions maj^ be widely different, nor do we need to investigate it here. In passing it may however be said that many reasons lend to the belief that inclusions of the type reproduced in Fig. 32 and 33 are only particles of some of the refractory materials with which the molten steel comes in con- tact during its manufacture, either in the furnace, in the spout or in the ladle, or, in some cases, after its teeming from the ladle, as happens for instance, when the ingots are bottom-poured through runners or have refractory tops or risers. 106 . In order to finish these, some- what random observations some ex- amples of those "local" effects may be added which, as has been said, are caused by other than emulsified inclusions. As has previously been inentioned, the latter are the only ones which maj^ be considered as possessing the property of altering the general characteristics of steel. In this way the effects produced will be more precisely defined, even before presentation of some examples which are more directly related to the specific results of preliminary heat treatments. The local effects produced by sonims other than emulsified inclusions show a great number of different aspects in steel-making practice. Such effects may be chemical in (■hara('ter. For instance, excessive corrosion Fig. .34. — Group of inclusions in heat- treated forging. X 70. Etched with 5 per cent, picric acid in alcohol. SOFT AND MEDIUM STEELS 141 may be locaUzed around the individual inclusions. On the other hand, the effects may be mechanical. Some of the latter, (for instance, the ab- rasions caused when the inclusions happen to occur upon a surface subjected to friction) show characters so different from those pro- duced by emulsified inclusions in relation to the results of heat treat- ment that really it is not necessary to illustrate with examples. Therefore, there are left the effects caused by non-emulsified inclusions upon the behavior of steel when subjected to the ordinary physical tests usually specified. Of these extraordinarily various examples I will give only two. The first is afforded by a medium hard 2-per cent, nickel steel contain- ing non-metallic inclusions of large dimension united in groups which forged down into elongated shapes looking like stripes in the micrograph. Figure 34, enlarged 70 dia. after etching with 5 per cent, alcohohc solution of picric acid, reproduces the appearance of such a group of inclusions taken after a very accurate and efficacious heat treatment has eliminated (by a process which we shall see further on in regard to the emulsified inclusions) the large strip of ferrite which had collected around the group in the forged steel, due to the phenomena seen in Sec. 98. Now, in spite of the efficiency of the homogeneity heat treatment, proved by the elimination of this ferrite accumulation, the inclusions continue to exercise very intense local effects easily detected in the tension tests made upon the metal by means of transverse test-bars taken with their longitudinal axis perpendicular to the direction of the elongation produced by forging. Some tension tests made upon a certain number of test-pieces whose fractures did not present any anomaly, gave satisfactory results varying between the following limits : TensUe strength 92,000 to 101,000 Elastic limit 64,000 to 70,000 Elongation 21 to 18 per cent. Reduction of area 43 to 35 per cent. Other test-pieces gave very poor results, extremely low tensile strengths, their elastic limits could not be ascertained and they showed very small elongations and contractions almost zero. The fractures of these last named test-pieces clearly showed local defects of the same type as that reproduced at natural size in Fig. 35. It is easy to find under the microscope that the heterogeneous white strip which shows upon the fracture is nothing else but the location of a series of lamellar inclusions. Due to their fragility and to their insufficient adhesion to the metal, they have caused the well known effects of a sharp notch made in the metal, acting as "primer" and provoking the premature fracture of the test-piece. 142 HEAT TREATMENT OF The test-piece reproduced in Fig. 35 gave the fohowing results: Tensile strength 47,000 lb. per square inch Elastic Umit Could not be determined Elongation 1.8 per cent. Reduction of area nil When testing such steel parallel to the direction of working, the presence of inclusions gathered in isolated strips such as the ones just illustrated causes great deal less perceptible effects than those which have been mentioned for the transverse test-pieces. A second example is afforded by a medium-hard carbon steel made in the acid open-hearth and forged from 7-ton ingots to I'ound bars 200 mm. in diameter. The test-pieces taken from all parts of the l)ars gave excel- FiG. .35. — Streaked fracture duo to lanicUar inr'lusion.=;. Natural size. lent results, except those from near the hottom of the ingot. These gave very low tensile strengths, the elongations and contractions were almost zero, even when the tests were made upon longitudinal test-pieces. The fractures of the latter showed appearances similar to the one reproduced at normal size in Fig. 36. Microscopic examination of these test-pieces demonstrated that they contained rmmerous non-metallic inclusions of the same appearance as that reproduced in Fig. 27. A more detailed study of the foreign bodies and a chemical analysis of a small quantity which could be separated mechanically from the steel, showed them to be constituted of acid slag formed during working the heat. 107. ■ Finally, it is necessary to call attention to a group of phenomena which sometimes occur as an effect of the same operations to which the steel is subjected during heat treatment, although they are not of a nature similar to the phenomena of diffusion and transformation upon which SOFT AND MEDIUM STEELS 143 normalizing treatments are based. Due to the fact that their effects accrue to those directly due to heat treatment, they often cause very noticeable modifications in the final phj^sical properties characterizing the various parts of the steel object treated. The phenomena which are to be mentioned are the following: When an object somewhat complicated in form is quenched, the cooling veloci- ties of its various parts are unequal : they reach the minimum values in the p-jQ 36. — Woody fracture due to inclusions of furnace slag. Natural size. thickest parts and the maximum values in the thinnest. Such differences of quenching speed sometimes cause very important internal stresses m the steel, which in turn, produce deformations, or warping. Slipping and twinning are produced in the crystalhne elements in certain parts of the piece, under the action of these deformations. These crystalline movements are analogous to those which have been already indicated in Sec. 94, when discussing the effects of hot-work. 144 HEAT TREATMENT OF SOFT AND MEDIUM STEELS The effects of these crystallographic phenomena sometimes are simply additive to those caused by heat treatment proper; but, in other cases, profound modifications of the results directly due to the heat treatment magnify this simple addition effect. For instance, this last statement holds when the effects of the deformations already mentioned continue to be felt after the piece has reached, during either rapid or slow quench- ing, temperatures very much below those which mark the limit called ordinarily, but erroneously, "hmit of plasticity." In these cases, a strain-hardening occurs analogous to that when steel is subjected to distortion at a temperature too low for correct working. The char- acteristics, the consequences, and the practical value of those processes have been also mentioned in Sec. 95, but it is not possible to dwell upon their study for the same reason then pointed out. It is suffi- cient to keep in mind that the practical importance of the phenomena is very great. Therefore it is necessary to take them into very strict account in the industrial application of heat treatment processes although, as we have seen, those phenomena are not related with the fundamental ones under study. They constitute rather one of the more delicate sides of the technical problems inherent to the preliminary heat treatment of steels. PART IV PRELIMINARY HEAT TREATMENT OF STEEL CASTINGS CHAPTER XIV GENERAL MORPHOLOGICAL RELATIONSHIPS BETWEEN THE SECONDARY AND THE PRIMARY CRYS- TALLINE SYSTEMS 108. The general characters assumed in practice by those complex phenomena comprising the foundation of normalizing technique were briefly indicated in the third part of this study. This was done mainly with the view of demonstrating that the effects of the fundamental phenomena studied in Chapters 1 and 2 are almost always modified in practice by the coincidence of other phenomena whose effects are some- times simply additive, while at other times they give rise to profound modifications of the course of the fundamental phenomena. In order to complete the development of our program all that is now needed is a certain number of examples designed to show the special technical aspects which characterize the more typical cases occurring in practice and to give some numerical data to define the quantitative, values of the results which can be obtained by the apphcation of preliminary heat treatments. We have already seen in Sec. 94 and 95 the immense practical im- portance of the effects of hot-work upon the proper development of the phenomena accompanying heat treatment. The apphcation of such treatments must therefore be conducted according to adjustable rules, depending upon whether the steel to be treated is in its cast condition (raw) or whether it has already undergone forging, rolling, or other work. Evidently there is an immediate opportunity for examining the illustrations derived from steel castings and from hot-worked metal separately. In this fourth part we wiU therefore study the former, reserving for the fifth part the study of the latter. 109. In order to place a complete schematic representation of the phenomena we are deahng with upon general foundations underlying every case taken into consideration it would be necessary to take into account the special characteristics shown by the crystalline structure of the steel, within whose elements occur the processes of diffusion and segregation which we have studied from a general point of view in the first two chapters. Such a program would eventually include the field of "general crys- 10 145 146 HEAT TREATMENT OF tallography" and would obviously oblige us to analyze a great number of phenomena similar to those which have already been incidentally mentioned, such as cold-working, plastic deformation, or twinning, which although of great technical importance among those factors determining the physical properties of steel, are not immediately related to the phenomena of diffusion. As has already been emphasized, diffusion is the only one, therefore, to which it is necessary that we restrict our study in order not to trespass the limits which we have set. But the facts which must be related as concrete examples illustrating the processes which up to the present time have been studied more or less abstractly could not be clearly comprehended should all considera- tions concerning their crystalhne morphology be disregarded. It is therefore necessary to mention in advance some brief observations con- cerning such properties, but they will be strictly limited to those necessary for the special and restricted scope in view. Thus, for our limited purpose, it is unnecessary to take into account the phenomena of "large crystallization" (so-called by Belaiew). Cer- tain factors characterize and accompany the formation of huge crys- taUites or the large primary crystalhne colonies separating from molten steel. They also determine the variations in structure, chemical com- position, and physical and mechanical properties in various regions of a large mass of steel, by reason of an accompanying wholesale segregation. Their analysis constitutes one of the most important parts of the general study of crystallization of steel. Yet they produce effects which gen- erally are spread over hmits of space much greater than the distances which in practice may be influenced by the processes of diffusion. It is proper, however, to clarify the limits of that group of phenomena as being those whose effects are not perceptibly modified by ordinary heat treatments. As an example, one might mention those causes which are responsible for the particular distribution of the large crystalline colonies characterizing the various types of steel ingots. It is known that the course of such phenomena is responsible for the forms and the custom- ary positions of the pipe, the presence and distribution of the cracks or "checks" so-called in various types of ingots, the value and distribution of chemical segregation, etc. 110. From what has been said in the previous paragraph it follows that for our particular purposes our considerations concerning the crystal- lography of steel may be restricted merely to a brief examination of the forms assumed in various instances by the first crystallites solidifying during primary crystallization, disregarding the particular way in which these elements arrange themselves to form the large dendrites. In order better to fix our ideas upon a concrete example, consider a steel ingot of normal form, tapering toward the top, like the one whose longitudinal section is reproduced in Fig. 37. The orientation of the SOFT AND MEDIUM STEELS 147 principal axis of the large dendrites is different in various regions of the metallic mass. Due precisely to that different disposition of the large dendrites various parts must be distinguished in the steel presenting special traits in some property, such as, for instance, the tendency to form forging or rolling checks, the distribution and the nature of the impurities, etc. Figure 37 shows schematically the arrangement of the principal, axis of the large dendrites in one of the innumerable cases which may happen in practice. Without engaging in a detailed examination of this case, which would be out of place here for reasons already mentioned, even a rapid and super- ficial examination would easily distinguish four regions, distinct one from the other in the respects just mentioned. These regions are the following : A. The principal axis of the large dendrites in the peripheral region marked with the letter A are arranged in concordant directions approximately normal to the ex- ternal surface of the ingot. This region has a quite variable depth, other conditions being equal, for steels of different composi- tion. In some steels, and especially in in- gots of small dimension in which the metal soUdifies rather rapidly, region A is limited in thickness to a few inches, and sometimes . even to a small fraction of an inch. In other steels, and especially in large ingots which sohdify slowly, it may reach a thick- ness of from several inches to one foot or more, even extending sometimes to the axis of the ingot,eliminating entirely the region 5. Figure 38 furnishes an example of this ^'Q- 37.— Diagram of erystalline X, , , i' , ,1 • 1 orientation in an ingot. case. It reproduces at about one-third natural size the appearance of the plane surface obtained by cutting along its axis a cold ingot of medium hard 2-per cent, nickel steel, subsequently etching the polished surface for about one hour with a 20 per cent, water solution of sulphuric acid heated to about 60°C. The etched surface was then slightly polished with very fine emery cloth, so that its more protruding parts, corresponding to the axis of dendrites left in relief by the etching appear more definitely in the photograph. It clearly shows large dendrites with their longer axis roughly parallel and at right angles to the periphery extending toward the axis of the ingot and 148 HEAT TREATMENT OF reaching region D with their extremities immediately neighboring the wall of the axial cavity of the ingot, or the pipe (Fig. 37) . For reasons which cannot be analyzed here, region A has the best developed tendency for the formation of cracks during hot-work (the so-called " checks") . It is the different development of this region that is responsible for the different tendency of the various steels to such cracks. Fk; 38. — Split ingot of medium hard 2-per cent, nickel steel. dilute H2S04. Pickled in hot B. The region immediately underneath the first one is beyond the columnar structure of region A. In it the orientation of the various primary crystallites is promiscuous, occurring in about the same amount along aU directions. Its random appearance apparently does not obey any determined law. Figure 39 (prepared and etched in the same manner as Fig. 38) reproduces the appearance of a plane surface in such a region as 5 in a medium hard 1.30-per cent, nickel steel ingot. It may be pointed out that regions A and B are not sharply separated SOFT AND MEDIUM STEELS 149 one from the other, so that a certain tendency can be noted in the border zone for the larger dendrites to be arranged with their longer axes normal to the periphery. C. The region C in Fig. 37 consists of crystalhtes surrounding centers of parasitic crystallization. We cannot enter into details of the nature of the reactions developing around these parasitic centers. It suffices Fig. .39. — Macrostructure of internal region of medium hard 1 M per cent, nickel steel. Pickled in hot dilute H2SO4. to remember that parasitic germs form in strongly undercooled liquids. Crystalhzation occurs around these centers under very different conditions than those holding for the other parts of the ingot. As a result the final crystalline conglomerate has quite different properties from those of the remaining normal metal, both in crystalfine structure and in chemical composition, due to wholesale segregation which characterizes this special mode of crystallization. 150 HEAT TREATMENT OF Wide heterogeneity is clearly illustrated in Fig. 40, which reproduces at about one-fifth of the natural dimensions, a surface' obtained by sawing a longitudinal section from an ingot of the same type as the one sketched in Fig. 37. Precisely, Fig. 40 represents an area similar to that marked Ci in Fig. 37, and includes the surface of separation between region C and region B. The different structure of two parts of the steel can be clearly seen, situated respectively at the right and at the left of the sepa- rating line crossing the middle part of the figure. In the region situated at the left of the line, the dendritic structure is a great deal better devel- oped than that on the right. The difference between the two structures appears "still more evident in the next figure (Fig. 41), reproducing at Fig. ■iO. — Junction between normal and parasitic crystallization, in hot dilute H2S04. One-fifth size. Pickled natural size, a more deeply etched part of the same area bordering the surface of separation between the two different crystalline masses. The steel missing from the upper part of the block photographed in Fig. 40 belonged to yet another crystalline mass of type C, of Fig. 37, formed around another center of parasitic crystallization. A great amount of impurities accumulated at their contact surfaces as a result of liquation of impure melts of low freezing temperature, and so modified the material immediately at that surface as to allow the metal to break easily along all of it. The study of the phenomena related to the formation of parasitic centers of crystallization has indeed a very great practical importance. D. Region D of Fig. 37 immediately surrounds the wall of the pipe. ' Polished and etched in the same manner as those reproduced in Figs. 38 and 39. SOFT AND MEDIUM STEELS 151 Metal constituting this region is characterized by the accumulation, as an effect of liquation, not only of higher proportions of all the normal constituents of steel, such as carbon, manganese, silicon, etc., but also the greatest part of the impurities which were contained in the molten metal, such as sulphur, phosphorus, solid non-metallic inclusions, etc. An example of the microscopic appearance which this region may assume is given in Fig. 38. Similar formations, with the same origin, composition and structure sometimes form in other regions of the ingot, more or less distant from the pipe. A particular example of such a case is given by the region im- mediately neighboring the surface of separation between the two crystal- line masses represented in Figs. 40 and 41. Fro. 41. — Portion of Fig. 40; more deeply etched. Natural size. 111. Hereafter we shall disregard the phenomena of "large liquation," with the resulting characteristics sufficiently explained by the examples mentioned above. In subsequent pages other examples will be given to illustrate the variations in normalizing practice induced exclusively by the phenomena of "small liquation;" meaning by this, those segregations which take place between the abutting portions of the primary dendrites. It may be pointed out immediately that dendritic structure can not be so neatly developed in all steels as in the illustrations presented in the previous paragraph. Nor is it so well marked in the same types of steel under all variations in the conditions under which crystallization and the 152 HEAT TREATMENT OF successive cooling have taken place. Thus, the development of dendritic structure is, in general, always very imperfect in the very mild carbon- steels. As an instance, the steel ingot shown in section in Fig. 42 was pohshed and etched under the same conditions as were the surfaces Fig. 42. — Center of 0.8-carbon steel ingot after pickling 1 hour in hot dilute H2SO4. reproduced in Fig. 38, 39 and 40. Its chemical composition was as follows ; Carbon . 08 per cent. Manganese . 58 per cent. Silicon . 07 per cent. No traces of dendrites appear upon the surface thus treated, and not even a much deeper etching is able to reveal well developed true dendritic structure. This last fact is shown clearly in Fig. 43, which reproduces the appearance of the same surface after the etching with hot dilute sulphuric acid has been repeated a second time, prolonging it for three hours. 112. Having established, although in this summary and rudimentary way, the various elements of crystalline texture among which the phe- SOFT AND MEDIUM STEELS 153 nomena of diffusion take place, it would seem logical that the selection of actual examples chosen to illustrate the etieets of normalizing practice should begin with those in which the phenomena studied in Part I would be isolated. In other words, it would be natural to study by concrete examples those phenomena which intervene only in the liquation processes accompanying primary crystallization of steel and the processes of dif- fusion in the austenite, without being modified by the phenomena con- comitant with the transformation of gamma to alpha iron. Fig. 43. — Same surface as Fig. 42 after etching for 3 hours. However, such a proceeding would immediately lead one outside the field which strictly limits the present study; namely, the field of the soft and medium steels normally used for machinery parts. In fact, in the present state of the art, mechanical construction never employs steels in which the y^a transformation has not taken place in a larger or smaller measure. Even those extremely rare cases in which some austenitic steels are used for certain machine parts do not constitute an exception' to this rule, for those parts are not mechanical parts properly speaking. Such special steels are then employed not on acount of their 154 HEAT TREATMENT OF mechanical properties but instead for other physical or chemical proper- ties, such as magnetism, electrical resistance, or resistance to corrosion. As has just been done for primary solidification, it will be necessary therefore to present some preliminary remarks relative to the morpho- logical characteristics manifested by the phenomena of the second group as they are related to those of the first. In other words, it is necessary to examine briefly in what forms the crystalline texture resulting from the 7<=±a transformation superposes upon the more or less imperfect system of dendrites resulting from soHdification. 113. This second aspect of the crystallography of steel, if studied in a thorough way and in all its details, would constitute one of the most important chapters of our knowledge. The relations between the phe- nomena due simply to the processes of diffusion and the phenomena originating from processes of purely crystallographic character are here no less rigorous than for primary crystallization. Therefore, one must evidently confine his discussion of this group to that which is strictly necessary for the right understanding only of those facts whch we will have to examine further on. Part II contained a general statement of the normal relations existing between the course of the primary separation of y mixed crystals from molten steel and the secondary separation of the /? and a mixed crystals from the austenite. Part III also mentioned in a general way, two of the principal groups of phenomena which interfere frequently in practice, and disturb those normal relations (see Sees. 93 to 107). Therefore, let us now see some examples of the morphological aspects which those relations may assume. It is clear that the simplest case which may be presented is that where- in the diffusion of the various elements throughout the austenite has taken place so completely as to cause perfect homogeneity of concentration. As already often noted, such an extreme and perfect case seldom or never occurs in industry. Nevertheless whenever it should happen, it is evident that the successive secondary crystallization of the /? and a crystals would take place freely, independent from the previous course of the primary crystallization and according to the laws studied in Part H, except, naturally, as a consequence of other eventual perturbative condi- tions such as these noted at the end of Part III. However, it is possible to mention some examples in which the uni- formity of concentration reached in the y mixed crystals approaches very near the extreme limit. In such cases the secondary crystallization will furnish very close approximation of the phenomena as they should take place in that extreme theoretical case. 114. Figure 44 reproduces at 30 dia., after pohshing and etching with a 5 per cent, solution of picric acid in alcohol, a section of a steel plate from a cementation muffle. This plate was maintained during more than ,SOFT AND MEDIUM .STEELS 155 five consecutive months ;ii a, teniporature between 900°C. and 1050°C. At the end of this period the furnace fires were drawn and the plate cooled, together with the entire mass of the furnace, so that the temperature dropped very slowly and uniformly, passing from about 1000°C. down to 150°C. in a period of time longer than one hundred hours. It is also necessary to observe that, while one of the faces of the plate was constantly in contact with the carburizing substances charcoal and carbon monoxide, the other face was bathed by oxidizing gases. Natu- rally, this circumstance provoked a continuous migration of carbon during the entire life of the muffle, due to the diffusion of dissolved carbon com- pounds through the entire thickness of the plate, from the inside to the outside. There is no doubt that this continual transfer must have con- FiG. 44. — Alpha iron in cementation muffle. X 30. Etched in alcoholic picric acid. tributed largely to the attainment of a high degree of homogeneity in the carbon concentration in each crystalline element of the y solid solution. It is clear that the circumstances just described may be considered as among the most favorable which can be realized in practice to obtain a thorough uniformity of the concentration of carbon in all parts of the individual crystals of austenite. In such a case, therefore, the secondary crystalhzation should proceed with the least interference from pre- existing conditions inherited from primary crystallization. It should place in evidence in the best possible way, the structural characteristics resulting from the mere separation of a ferrite during allotropic transformation. In the case taken as an example, this last result is obtained in a manner 15G HEAT TREATMENT OF quite close to tlia.t proceeding at the theoretical limit of perfect homogen- eity of austenite, inasmuch as the original average concentrations of the elements other than carbon contained in solid solution in the 7 ii'on, were very low. Consequently their eventual residual heterogeneities in the steel after its prolonged heating must necessarily be extremely small, an important point, since these heterogeneities have been seen to exercise appreciable effects upon the separation of ferrite. In fact, the composi- tion of the steel constituting the plate observed as far as the other four fundamental elements beside iron and carbon are concerned was as follows : Manganese . 40 per cent. Silicon 0.01 percent. Sulphur 0.01 per cent. Phosphorus 0.05 per cent. Fig. 4.5. — Rounded alpha iron and lamellar beta iron from homogeneous austenite. We may therefore, consider that the characteristic structure of the crystals of a iron reproduced in Fig. 44 is at least closely approxi- mate to that peculiar to ferrite separated from 7 mixed crystals of abso- lutely uniform concentration. It is known that such so-called Widmanstattian structure is built up of fiat plates of ferrite, oriented according to definite directions. They satisfy the crystallographic criteria showing them to be arranged in a a family of planes parallel to the surfaces of the octahedron, one of the simpler crystalline forms of the isometric system. Their mutual inter- sections vary according to the direction which the plane of the polished section happens to take. SOFT AND MEDIUM STEELS 157 Another observation may be made even upon Fig. 44, although the fact is shown still clearer in Fig. 45. The latter is a microphotograph of the structure at another point of the same section of plate, selected where the carbon is lower. In fact, the specific carbon content varies from a very high value at the internal surface of the muffle, to almost zero at its external surface, a state of affairs due to the history of this plate, which has been described. Fig. 46. — Fracture of a steel having highly developed A\'idman.stattian structure. Natural The observation which I want to make is the following: The two figures, but especially Fig. 45, reveal clearly an observation which recurs constantly in all analogous cases. In the regions of the metal in which the concentration of carbon (estimated according to the well known planimetric method of relative areas occupied in the section by ferrite and pearlite respectively) is lower than 0.4 per cent, the ferrite 158 HEAT TREATMENT OF appears in the form of the well known roundish kernels characteristic of soft steels. On the other hand in regions where the concentration of carbon exceeds the value of about 0.4 per cent., ferrite appears in the form of needles oriented along octahedral cleavage (Widmanstattian structure). For a series of reasons to which I cannot here refer/ the hypothesis appears justified that the granular appearance of the crystals in low-carbon areas is that characteristic of /3 ferrite. Carbon steels containing less than 0.4 per cent, carbon separates p iron directly from y mixed crystals (see Fig. 10, curve DF) and this is transformed afterwards at lower tempera- tures (along curve BC of Fig. 10) into a iron or ferrite without a sub- stantial change in its microscopic appearance. The lamellar form of the Fig. 47 -Widmanstuttian noei-Hos in grains defiMitely outlined l)y reticular structure. X 50. crystals in higher carbon steels, on the other hand, would possess the characteristics of the crystals of ferrite, or a iron, which separates directly from the y solid solution (see curve FH of Fig. 10). 115. From the point of view of intH'hanical properties Widmanstattian structure in hypoeutectic steels is known to be associated with high brittleness. Their fracture ajspears brilliant and coarsely crystalline. Fig. 46 reproduces at about natural size the appearance of the fracture of a soft steel having this structure highly developed. Such an abnormal development very seldom occurs and only in case of very prolonged and strong overheating, so as to approach the region of true burning (see Sec. 25, 26 and 83). Less severe instances of over- 'See "Crystallography of Alpha and Beta Iron," Chemical & Metallurgical Eng- ineering, V. 22, p 585 (March 31, 1920). SOFT AND MEDIUM STEELS 159 heating happen quite frequently in practice, especially in the heating operations preparatory to rolling or forging. Even this is always such as seriously to impair the mechanical properties of the steel. In such cases the structure of the resulting metal shows a combination of the pure needle-like Widmanstattian structure with the reticular network struc- ture which shall be examined shortly. An example of such a mixed struc- ture is that reproduced in Fig. 47, enlarged 50 dia. However, in such intermediate cases we are obviously confronted with phenomena of secondary crystallization occurring in a 7 solid solu- tion which is no longer even approximately homogeneous. These cases, therefore, do not belong in the category with which we are now concerned. We shall have occasion to refer to them further on. As far as concerns the extreme cases of homogeneous austenite men- tioned in the previous section, I may note that it is nearly always possible to modify the pure Widmanstattian structure, into a mixed structural type similar to that of Fig. 47 and of improved mechanical proper- ties, by disturbing or interrupting the crystallization of a iron, either by a rapid cooling through the critical interval, or by mechanical work through the upper part of the same range. The details of this process of disturbed crystallization does not concern us here because the extreme cases of uniform homogeneity such as the one related above have but a petty industrial importance. CHAPTEE XV ANALYSIS OF THE PRELIMINARY HEAT TREATMENT OF A SPECIAL STEEL 116. Passing by the theoretically important limiting case of com- pletely homogeneous austenite, let us examine the structural evidence appearing in an actual steel during the separation of 7 mixed crystals from the melt and its later disintegration into j8 and a. crystals. By exclusion this discussion will relate to the only industrially important and frequent case in which the 7 mixed crystals preserve to a larger or smaller degree the heterogeneities of concentration in their various constituents pro- duced during primary solidification in the manner indicated in Part I. To this effect let us first analyze one of the cases in which the effects of those relations are strongly apparent. Such conditions occur promi- nently in a medium hard 2-per cent, nickel steel, similar to the one whose microstructure is reproduced in Fig. 38. Let us start by examining this steel in the raw state, as cast; in other words in the condition of maximum heterogeneity in austenite. The composition of this steel was the following : Carbon . 42 per cent. Manganese! . 60 per cent. Silicon . 22 per cent. Sulphur . 008 per cent. Phosphorus . 02 per cent. Nickel 2 . 02 per cent. The dendritic structure revealed by warm dilute sulphuric acid attack was identical to the one reproduced in Fig. 38. Let such a section of the same metallic surface be accurately repolished and lightly etched with one of the usual metallographic reagents for the examination of hypo-eutectoid steels (for instance, 5 per cent, picric acid solution in alcohol). If it is then examined, that ensemble of struc- tura,l characteristics which H. M. Howe has called ingotism will dis- tinctly appear in the areas formerly exhibiting normal dendritic structure, corresponding to regions A and B of Fig. 37. Figure 48 reproduces at 3 dia. this exact structure as it appears near the surface of a 1-ton ingot of the steel in question, after pohshing and etching with a 5 per cent, solution of nitric acid in amyl alcohol. The micrograph reveals clearly the characteristic veins or shells of ferrite outlining grains or kernels of pearlite. In the interior of these 160 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 161 pearlitic meshes or kernels may be seen little islands of ferrite. It is easy to understand how such a distribution of the constituents corre- sponds to the course of secondary crystallization, where the a solid solu- tion precipitates from the non-uniform y mixed crystals, as has been carefully analyzed in Part II, Sec. 63 to 74. Dr. Howe's new word "ingotism" describing this structure is very suggestive of the charac- teristic structure of ingots, and it is a word which has therefore been generally adopted. The veins of ferrite, constituting a real network, are deposited by the successive nuclei of a iron, primarily formed at or very near to the primary crystallization centers — that is to say, region ji of Sec. 66, where the concentration of carbon and other elements was the minimum — and result from the successive deposit of lamellffi of a iron upon these germs. The islands of ferrite contained in the interior of the principal meshes (that is, in the kernels) ' are a growth around new crystallization germs of a iron, ap- pearing successively in the more advanced stages of the y^a trans- formation range. The mechanism was discussed under "region C" of Sec. 66. INIoreover, the structure which is now being examined con- stitutes a typical example of the forms assumed by the a crystals in the general case of their separation from the heterogeneous y solution. There is an evident tendency for crystalline elements which suc- cessively build themselves upon pre- existing germs of the same nature to arrange themselves in ranks along prevalent directions, so as to result in long tiny branches rather than massive accumulations. This fact is entirely similar in its causes and effects to that which is responsible for the formation of dendrites during primary crystaUization of the y solid solution. It also better defines from a morphological point of view the phenomena studied in Part II, especially Sec. 66 and 67. In order to exclude the hypothesis that the characteristic disposition of reticular structure which has just been examined might bo due to the presence of non-metalhc inclusions capable of locahzing and orienting the growth of the a crystals in the manner which has been indicated at the end of Part III,"^ I may inmiediately state that the islands or veins of ferrite do not contain non-mctalHc inclusions except in some rare instances ' See especially Sec. 98. 11 Fig. 48. — Thin ferrite network in nickel- steel casting. X 3. Etched by 5 per cent. HNO3 in amyl alcohol. 162 HEAT TREATMENT OF in the material whicli we are now concerned with and illustrated in Fig. 48. This can be seen clearly in Fig. 49, which reproduces at higher magnification (80 dia.) a spot on the same metalhc surface. 117. Two different methods of approach may be adopted in order to examine the relative disposition of the two crystalline networks, y and a, resulting respectively from the primary and the secondary crystalhzation. I think it opportune to mention an example of each of them, referring always to the same steel, whose characteristics have been indicated in the previous section. The first method consists in etching the pohshed surface with hot dilute sulphuric acid, until the dendritic system is well developed. The pitted surface is then shghtly pohshed to prepare the surface in relief Fig. 49. — Ferrite vein free from sonims. X 80. for a subsequent etching with a 5 per cent, solution of nitric acid in amyl alcohol, prolonged only long enough to place the ferrite in evidence. Figure 50 reproduces at 10 dia. the structure of a piece of the nickel steel doubly polished and etched in the manner just described. As is shown, there exists only a general concordance between the "habit" and the orientation of the two structural systems constituted respec- tively by the dendrites — crystals which originally were formed by 7 iron — and by the elements of ferrite — crystals of the (3 and a series. Certainly, the veins of ferrite in the network do not correspond uniformly to elements of the system of the dendrites as determined by deep etching. The conclusion is evident that the orienting forces responsible for the formation of the ferrite network developing in region B (Sec. 66) equals SOFT AND MEDIUM STEELS 163 and sometimes overwhelms the forces due to difference in concentration of the various elements in the heterogeneous austenite. For evident Fig. 50. — Nickel steel after double etching. X 10. Fig. 51.— Nickel steel etched with HNO3 in amyl alcohol. reasons, it is not possible to tarry and analyze these phenomena as they appear in Fig. 50 and in all the analogous cases. 164 HEAT TREATMENT OF The same facts are revealed even in a clearer way when the second of the two methods of observation is apphed. This consists in photo- graphing a given spot on a polished surface after etching it with the amyhc solution of nitric acid so as to reveal the ferrite network. Then exactly the same area is photographed a second time after being etched with the hot sulphuric acid solution, which develops the primary den- drites. Care is taken to make the two photographs at the same enlarge- ment. Then place the two negatives exactly one upon the other with precautions which are not necessary to describe here, and print through the two so as to get a single positive. This print reveals distinctly and H^^<^^^^^p^^ « IKSt ' »»''V*'>'^,,^„S - ^ fe^^ ' '^^ ;^'l '»''-'* /l ^k<-i,W. w^^ • - ^ m^^'^ ■^ »-.a transformation and the amount of ferrite contained in another steel in which these heterogeneities have been more or less smoothed out by means of a homogeneity heat treatment. In order to sum up in a few words the generic characters of all causes of ingotism, isolating the more important group of phenomena from the great number of the concomitant secondary facts analyzed in the preced- ing pages, this is what could be said : Sections 63 to 68 showed that as the heterogeneity of the concentra- tion of carbon and the other elements in the 7 mixed crystals is greater — that is to say, as the band of variant concentrations is wider — the lower is the concentration of carbon in the |3 and a mixed crystals which sepa- rate at the beginning of the y—^a transformation. Consequently, the lower will be the proportion of carbon which is subtracted from that residue of austenite which afterwards forms the eutectoid pearlite of practically fixed composition. It follows from this, that each decrease of the degree of heterogeneity of the concentrations in the 7 mixed crystals gives rise to a decrease in the carbon available for the formation of the eutectoid and, therefore, to a decrease of the proportion of the eutectoid itself. It is true that an increase in the width of the band of heterogeneity in the austenitic field may cause an increase in the proportion of a mixed crystals richer in carbon which form at the end of the transformation, but it is clear that there is no necessary proportional relation between the counteracting phenomena, so that the observations contained in the 176 HEAT TREATMENT OF previous section demonstrate simply that the influence of the first group exceeds that of the second, at least in the cases examined. Beside all this, it is necessary to take into account the demonstrated fact that the variations in concentrations of the various elements (princi- pally carbon) throughout the austenite may cause the formation of many extremely small crystallites of ferrite, as is clearly seen in the various microphotographs reproduced in Part II and in the last few sections. These crystalline elements remain scattered throughout the eutectoid kernels of the raw steel in many cases, thus escaping recognition and proper appraisement in a microscopic investigation made under ordinary conditions. Yet they are easily seen when they have agglomerated in less numerous but larger masses after a reheating. It is clear that such a fact must cause an apparent decrease of the proportion of ferrite in raw steel and an increase (also apparent), of the proportion of pearlite, a change in ratio which accrues to the actual increase which occurs for the reasons just analyzed. 122. Let us now see what variations are produced by normalizing treatments upon the physical properties and the accompanying structural changes in the nickel steel which has been under examination. This is Table II. — Mttltiple Heat Treatments Mark Opera- tion No. Heat treatment Brinell hard- ness Qorre- sponding macro- structure No. Mode of heating Quench- ing in water from Re- heat- ing Corre- sponding micro- structure 13 Cx 11 hours at 1090°C. fol- lowed by cooling during 2 hours to 800°C. 800°C. 2 hours at 580°C. 217 Fig. 64 Fig. 67 14 At-i 1 2 9 hours at 1000°C. fol- lowed by cooling during 1 hour 45 minutes to 800°C. 9 hours at 1050°C. fol- lowed by cooHng during 2 hours to 800°C. 800°C. 800°C. 2 hours at 580''C. 179 Fig. 65 Fig. 68 15 Ci 1 2 3 10 hours at IISO'C. fol- lowed by cooUng during 2 hours 25 minutes to 800°C. 11 hours at 1160°C. fol- lowed by cooling during 2 hours to 800°C. 10 hours at 1090°C. fol- lowed by cooling during 2 hours to 800°C. 800°C. soo-c. 800''C. 2 hours at 580°C. 196 Fig. 66 Fig. 69 SOFT AND MEDIUM STEELS 177 exactly what must be the principal purpose of all our study and most especially that of the concrete examples which I purpose to present in these last chapters. In order to complete this new group of observa- tions we shall observe the results obtained by subjecting the same steel to extremely energetic homogeneity heat treatments, made under conditions, Fig. 64. — Test-piece 13 after pickling in hot 20-per cent. H2SO4. X 2. Fig. fi.5. — Test-piece 14 after pickling in hot 20-per cent. H-SO4. X 2. which, in practice, are considered the most efficacious to strongly trans- form the original ciystalhne edifice of raw steel. These results will bo used as a fundamental term of comparison, and regarded as the extreme end of a series of most efficacious homogeneity treatments. In Table II are indicated such treatments impressed upon three test-pieces of the same dimensions and cut from the same 2-per cent, nickel steel ingot. The three test-pieces are numbered 13, 14 and 15 following the previous samples in Sec. 119. For the same reason the distinctive mark show- ing on each test-piece in later illustra- tions has been added in the second column. To avoid misunderstandings it may be pointed out that the two treatments indicated for test-pieces No. 14 were apphed successively to the same sample, which was observed after having been subjected to the double treatment. The same remark applies to the three treatments of No. 15. 123 Figures 64 65 and 66 show at 2 dia. the appearance oi the three respective test-pieces No. 13, 14 and 15, pohshed in the usual way, and etched with the hot aqueous solution of 20 per cent, sulphuric acid. Even after the drastic heat treatments to which the first two test-pieces 12 Fig. 66. — Test-piece 15 after pickling in hot 20-per cent. H2S04. X 2. 178 HEAT TREATMENT OF have been subjected, the system of the dendrites still appears upon the etched surface with a clearness practically equal to that with which it Fig. 67. — Test-piece 13 after double etching. X 65. Fig. 68. — Test-piece 14 after double etching X 65. appeared before the treatments. Only in the third test-piece on which the extremely energetic homogeneity treatment has been repeated three times — even very much prolonged etching does not show the dendritic system. SOFT AND MEDIUM STEELS 179 The appearance of sections of the same three test-pieces after the double etching! is reproduced at an enlargement of 65 dia. in Fig. 67 68 and 69. As could be foreseen, there exists no longer a true network of ferrite discernible under medium magnifications but only the usual conglom- erate due to reheating at a temperature below Aci, i.e., a segregation of the solid solution now rendered nearly homogeneous by the previous heat treatments, at least as far as the concentration of carbon is concerned. Here is a more definite confirmation of the fact that heat treatments designed to foster the phenomena of diffusion and to check or, at least to decrease the processes of segregation taking place during coofing through Test-piece 15 after double etching. X G5. the transformation interval, make their effects felt with enormously greater rapidity and intensity upon the network of ferrite (or upon the crystalline system which determines its form) than upon the dendritic system. Experimental data, which would be out of place to dwell upon here, seem to indicate that very remarkable modifications of form can be obtained in the structure of the dendritic system, even without exces- sively drastic heat treatments. However, for a steel of the type under study, it certainly is necessary to subject the metal to extremely energetic heatings and quenchings in order to obtain a perceptible smoothing out of the local heterogeneities, by which the dendritic system is revealed by chemical attack. ' Fu-st etching with hot dilute sulphuric acid, followed by polishing of the parts in relief and by a second etching with a solution of nitric acid in amyhc alcohol. 180 HEAT TREATMENT OF ■< 02 O m « H Oi o K P4 K H H 1 o 'A Stretched throughout its total length do do do Contraction well localized in a neck do do do do do do do do do do 1 g o>oo loooiomioioiooo eq ooi>t>o oo5oo(N'*'OOi>cD.-i (N (N oi r-i <*' CO CO CO 01 CO TO TO ■*'■*' ■*' Unit load on original section, lb. per sq. in. o oooo oooooooooo o oooo oooooooooo r^ rHlOOOi-l 10rHWDlO»OCOOOOI>CD o~ o" co" "f c3r m oo" cf co" c0 (M c^TtHoq_rH •;_oo>o^- of txroo"i>"od" 00 od" oT 00 oT oo" 00 i> 00 i> Elonga- tion at propor- tional ■ limit, mm. o loioioio loinmoioooooo o t^t~05c^ iNcqiMTfiTO-*-*-*-*"; (N o o o ci ooooooo o' o' o CO OiOeOO TO^OtOCDiHOOCOOO CO 1>COOOIN 05i-l-l O 00 ■* lO 00O5D (NOC0OO051-I050000 00 ^-^^>odtD t>-'i>tD'i>i>ot^>oio«i 1 6 rH IMC0-*'O tOl^00050TH(NM-*u5 rH tH »-( >-H rH 1-H SOFT AND MEDIUM STEELS 181 124. Having thus completed a typical series showing the results ol superposition of the two crystalhne systems, each sample showing clearly the characteristics of both systems, let us examine the corre- M 11 10 ■0^ s §M -^^ '^4 i€"s^ ^m ^\^ %-/ f^M il^ j^:,.>:.vj« BHaBiflH! 'M- ^ 1 ■f;:,;!^" ; ^ft'*^-, ^: 'Sw-' .. ^ fctattewS-' '4*^^ m " ' M' .. Hi;... Jk i ^P ■ 1 ^B|f:J ^R*'' # ■? ;;# ■ ^% ^mmM 1» '^"'' - wT Fig. 70. — Fractures of tension test-pieces subjected to simple heat treatments. Fig. 71. — Fractures of tension test-pieces subjected to multiple heat treatments. spending mechanical properties. Table III contains the results of a series of tensile tests made upon the fifteen test-pieces, treated as indicated in the previous paragraphs. 182 HEAT TREATMENT OF The test-pieces were of Fremont's cylindrical type with conical ends. In such test-piece the constricted region in which deformation previous to fracture is confined is a cylinder 10 mm. long and 8.20 mm. in diameter. Figures 70 and 71 reproduce at natural size the appearance of the fifteen breaks. It is necessary to note that the correlation between data of Table III and the appearances shown in the last two figures, is made by the conventional marks listed in the second column of the tables, because the numbers written serially upon the half-tones do not correspond to those of the first column of the table. Fig. 72. — Fractures after slow bending. Table IV contains the results of some slow-bending tests made upon the first si.x test-pieces. These tests were made upon prismatic bars 10 Table IV. — Flexure Tests Tot'il Deformation No. Mark maximum load, lb. corresponding to the maxi- mum load, mm. Appearance of fracture Observations 1 A, 6,370 2.80 Crystalline Test-piece stretched throughout. 2 4 3-1 7,000 3.00 Crystalline Test-piece stretched throughout. •H ^3 7,000 4,55 Crystalline Test-piece stretched only near the frac- ture. 4 Bs-1 6,4G0 4.70 Caystalline Te-st-pieee stretched only near the frac- ture. 5 D, 6,970 6.20 Fibrous Deformation well locahzed. 6 A, 7,740 7.75 Fibrous Deformation well localized. SOFT AND MEDIUM STEELS 183 mm. square in section, and 60 mm. long. Figure 72 shows the appearance of the six bars after failure. Finally, Table V contains the results of a series of impact tests made upon all the fifteen treated test-pieces. These tests were made by means of a Fremont drop hammer using bars 8 X 10 X 80 mm. with a semi- cylindrical notch (2 mm. radius) across the 8-mm. face. Table V. — Impact Tests No. Mark Total energy absorbed Appearance of the surface of fracture 1 A3 12 Coarse crystalline ' 2 As-. 18 Coarse crystalline 3 B, 20 Coarse crystalline 4 B3-. 20 Coarse crystalline 5 2)3 12 Crystalline 6 A, 12 Crystalline 7 B2 13 Crystalline 8 B2-1 13 Crystalline 9 D, 31 Fine crystalline 10 2>3-l 32 Fine crystalline 11 B4-1 30 Fine crystalline 12 o,_. 30 Fine crystalline 13 Ci 32 Short fibrous 14 A2-1 32 Fine, long fibrous 15 ' C2 32 Fine, long fibrous Figure 73 shows the appearance of the fractures at about natural size. 125. It is first of all necessary to observe that the gradually increasing intensity of the homogeneity heat treatments to which these test- pieces were subjected, may be considered as covering a substantially complete scale, both as concerns the constituent treatments — starting from a short reheating at a temperature below Aci and gradually ap- proaching the extremely energetic treatment to which the last was sub- jected — and also as concerns the effects of such treatments upon the structural properties of steel. In fact : First, as far as the dendritic system is concerned, we pass from the case where it is unaffected, to the case where it is entirely obliterated by the heat treatment. See for instance the last test-piece illustrated in Fig. 66. Second, as far as the network of ferrite is concerned, we also pass from the cases in which the heat treatment leaves it completely unaltered, to the cases in which it has undergone the greatest changes obtainable in practice. Third, and finally, as far as the surface of the fractured metal is con- cerned, we also find for each physical test an absolutely complete series of results starting from materials with a coarsely crystalline fracture 184 HEAT TREATMENT OF called "granular," and reaching materials of a finely fibrous structure. To such structure is attributed, and often with ]-eason, a very great im- portance in metallurgical practice. We may therefore find in this scries of experimental observations all the data regarding the relationships between structural characteristics and phj^sical properties. It is to be understood that this statement is limited to the special steel to which the experiments summed up in the last few pages refer. Fig. 73. — Impact fractures of a nickel-steel, variously heat treated. 126. As has been already mentioned, great importance is usually given in metallurgical practice to the appearance of the fracture, in forming a judgment of a steel's quality when considering its general suitability to withstand various ph3^sical tests. Thus a steel showing a fibrous fracture is considered to be endowed with good general mechanical properties, while a steel showing granular fracture is considered as being inferior. In the latter case, the ciuality of steel is thought to be poorer the more coarse and shiny the fracture. Without entering into a discussion of the more or less sure scienti- fic foundation upon which such a criterion may be based, nor upon the often times absolutely unquestionable practical value of the conclusions SOFT AND MEDIUM STEELS 185 which may rest upon such a simple test, it is interesting to examine briefly the experimental data presented in the last ten sections as they indicate relationships between the form of the fracture and the better defined microstructure, as well as between the character of the fracture and those mechanical properties which may be tested with greater pre- cision and expressed with numerical values. 127. A first conclusion which may be inferred from the test results on this special steel is the fact that for medium nickel-steel the so-called granular fracture is certainly not due in preponderant measure to the state of the dendritic system, but is instead very closely related to the state of the ferrite network. This is self-evident when comparing the series of structures represented by the photographs reproduced in the previous pages (see especially Sec. 119, paragraphs 2 and 3) and the corresponding data contained in Tables III, IV and V. From such a comparison it also appears very clearly that the rela- tion between the ferrite network and the fracture is quite different in case the fracture is caused by a gradual load or produced by impact. In fact, under gradual loading (Tables III and IV) this steel presents a fibrous fracture as early as test-piece No. 5; namely, as soon as heat treatment has obliterated ingotism — i.e. the wide ferrite network characteristic of the raw metal. From this point on, the appearance of the fracture in tension remains about constant, presenting but slight variation, except after a very energetic homogeneity treatment such as imposed upon test-pieces No. 13, 14 and 15. However, in the impact test, the granular crystalline fracture persists in all the test-pieces which were simply reheated, although always becoming finer step by step as the treatment is more elevated and prolonged, being replaced by the fibrous fracture only when the metal has been subjected to more complex and very energetic normalizing, such as those to which pieces No. 13, 14 and 15 were subjected. It is easy to see how these conclusions more definitely establish the limits and the characteristics of the general statement that the fibrous or the crystaUine fracture of a given steel depends in very large measure upon the conditions in which the fracture has been made. Confirmatory evidence will be indicated shortly. 128. If the results of the impact tests be more particularly examined, it must be admitted that another factor is present in addition to the very evident effect produced upon the fracture by variations of the state of the ferrite network. This second system of structural elements has a simultaneous action with the first one although of minor intensity. Thus, the variations in the ferrite and the fracture starting from the fifth test- piece (Ds) do not by any means proceed in step. For instance, the microstructure of test-piece An, reproduced at 65 dia. in Fig. 74 shows a much finer ferrite network than that of test-piece Z)4-i reproduced in 186 HEAT TREATMENT OF Fig. 58, while the fracture of the first is a great deal coarser than that of the second (see Fig. 73). The same observations can be repeated for many pairs. Close examination of all those figures previously shown giving the appearance of a plane section after etching with hot sulphuric acid, would seem to exclude the hypothesis that this second contributary system might be the dendritic system. In fact, a series of such photographs makes it plain that all the test-pieces which have been examined except only the last one, (d) possess a dendritic system developed by hot acid attack which has not been substantially modified, much less obliterated, by heat treatment. Fig. 74. — Test piece No. 6, after double etching. X 65. However, one should not deduce merely from this fact alone that the second structural system influencing the fracture cannot be the dendritic system. Nothing so far mentioned excludes the supposition that the etching figures might be qualitatively unaltered in form and disposition, although the absolute amount of those local differences in concentration of the various elements in sohd solution might be changed a great deal by the normalization. Granting the fact that such heterogeneities are respon- sible for the formation of etching figures, one cannot exclude the possibility that the heat treatments mentioned have simply attenuated by means of diffusion the characteristic differences of concentration in the various parts of each dendritic crystal. In this case it will be enough that such equalization certainly cannot cause any displacement of the primary crystalline elements, nor can it be easily revealed by a corresponding SOFT AND MEDIUM STEELS 187 variation in the rate of etching. This appears to explain rationally the persistence and apparent constancy in the etching figures reproduced in the preceding pages. It seems quite probable that this hypothesis is correct, for various reasons which cannot be analyzed here, but which are founded in part upon considerations inherent to the presence and distribution of non- metallic inclusions. 129. To sum up the conclusions which may be deduced from the cur- sory observations of the last few sections : As far as concerns the relations existing between the structural systems under consideration — and for the special case here under observation — the specific function of the network of ferrite appears to be as follows: (a) In the 2-per cent, nickel steel examined, the position and condition of the particles comprising ferrite network is independent of the position and condition of the particles comprising the dendritic system. This remark holds whether the excess constituent occurs totally or partially as a wide network, characteristic of steel as cast, as well as in the case where the network is totally destroyed. (6) If this hypoeutectoid steel be subjected to a series of heat treat- ments capable of breaking up the original ferrite network, the resulting fracture obtained either by gradual stress or by impact is most highly modified by those treatments which cause the substitution of a new net- work for the original reticulation characteristic of raw steel. (c) Having once destroyed the original ferrite network of ingotism, heat treatments which cause further remarkable transformations of the new ferrite system replacing the first one, produce no great variation in the fractured surfaces obtained by gradual stressing. It is necessary to have recourse to very energetic homogeneity treatments in order to obtain modifications of industrial value in the appearance of such test-pieces. The practical importance of this observation when studying heat treatment of steel castings of this type may be easily understood. (d) The fractures obtained by impact also vary a great deal when heat treatment destroys the original wide network of ferrite. Further modification in ferrite structure also is reflected in the fracture under impact. But in such cases variations in the fracture do not consist with variations in the ferrite, but rather appear to be a consequence of the added effect of another system of structural elements, which seems to be that of the dendrites. 130. Passing now to the examination of the effects which homogeneity heat treatments produce upon the most important among the physical properties of the steel under observation, which can be determined quanti- tatively and expressed numerically, let us examine briefly each one of the three series of test results tabulated in the preceding pages. 188 HEAT TREATMENT OF Start with the tension tests listed in Table III, Sec. 124. As far as the ultimate strength is concerned, it apparently oscillates between 94,500 and 112,500 lb. per square inch, without its variations presenting any apparent relation to changes in the heat treatment. This remark is fully confirmed by the examination of the Brinell hardness numbers, reported in the last column of Table I and the seventh column of Table II. However, the other properties show more regular variations, fully confirming the conclusions arrived at already in previous paragraphs. Thus, the first four test-pieces — in which the heat treatment has not destroyed the original wide network offerrite characteristic of cast steel-^ present all the well-known physical properties of metals formed by two structural constituents endowed with different plasticity, and built up of crystals of large dimension. The specific mechanical function of the ferrite network is well defined by said physical properties. As we have already seen in this case and more particularly in Sec. 118, the effect of the ferrite network still predominates by far over that of any other structural system. The same physical properties also give a well-defined significance to the granular and crystalline fracture of the first four test-pieces. The before-mentioned characteristics of the fracture persist through all the successive test-pieces which were simply reheated. However, reduction in area does not vary materially, and its fluctuations do not follow that of the ferrite network in any way. Instead, the reduction of area increases a great deal in the last three test-pieces. When the fact that the variation in the condition of the ferrite is very small in the last three test-pieces, when this fact is considered in conjunction with the others above mentioned, the intervention of modifications caused by the heat treatments upon another and different structural system is clearly indicated. Still better defined are the indications furnished by the analysis of the figures showing the elongation undergone when the metal has reached the first point of inflection in the stress-strain curves. This figure is the greatest for the first four test-pieces, which were strongly stretched on all their length, yet it falls immediately to very low values as soon as heat treatment has destroyed the ingot structure. From this point on, the materials are fibrous and have a well locahzed contraction and the elonga- tion increases step by step as the homogeneity heat treatment becomes more energetic; but as in the previous cases there cannot be observed any parallelism between the changes in ductility and that of the ferrite network. A fact which will again be found to be true for resistance to impact is worth noticing, namely, that the elongation shows a sudden increase at the ninth test-piece (D4), the first one heated at 900°C., without any cor- SOFT AND MEDIUM STEELS 189 responding change in existing variations in the ferrite network being noticeable. This last fact, together with the others just mentioned, also' seems to indicate clearly the intervention of another structural system in addition to that represented by the ferrite network. Finally, the maximum elongation shown by the test-pieces before fractures gives similar results. In fact, it is small for the first four test- pieces, and increases noticeably for the succeeding eight (also simply reheated) but without following in any way the variations in the ferrite. The maximum elongation before failure is still a little greater for test- pieces Nos. 13, 14 and 15, quenched and reheated. Therefore, the previ- ous conclusions are also confirmed in this case, especially those which have been indicated regarding variations in the reduction of area. 131. Examination of Table IV giving the results of static bending tests (Sec. 124) again illustrates the findings that variations in the physical properties are related to variations produced by the heat treatment in the state of two structural systems. The total loads at failure do not vary a great deal, while the maximum deflection increases very much as the heat treatments become more efficacious in the ehmination of ingotism. In this case also, the test-pieces containing wide network of ferrite are ex- actly the ones which deform by stretching along a considerable part of their length. Finally, in the impact tests (see Table V, Sec 124) the wide difference is no longer found between the physical properties of the test-pieces in which the large network of ferrite remains (the first four test-pieces of Table V) and those of the other test-pieces in which heat treatment has substituted a finer one. This fact — which finds its complete counterpart in the variations noted in the fracture of the metal — constitutes an important new pre- sumption in favor of the hypothesis already clearly indicated by the previ- ous observations covering the intervention of another structural system than ferrite. This last remark is supported by the fact that starting with test- piece No. 9 (Di), the first one heated at 900°C., a strong increase in the resistance to impact appears, and this without any corresponding substantial variation in the network of ferrite. From this point on the resistance to impact remains about constant although substantial varia- tions of the ferrite network still occur in the successive test-pieces. We have already seen in Sec. 130 that the same fact occurs in relation to the figures for elongation at the elastic hmit. In that case also, sudden variations were noticed starting from the first test-piece heated to 900°C. The practical importance of such a fact is evident. 132. Such considerations allow one to understand more exactly the relations between the different structural systems in the raw steel and the 190 HEAT TREATMENT OF path of fracture as it occurs in this steel under the action of gradual stress, as well as how these relationships are affected by a modification of the steel's structure upon normalizing. In fact, it is quite natural from what has been seen in Sec. 118 and subsequently, that the habitual path of plastic deformation in the steel under consideration, when the large network of ferrite has not been destroyed, is very much different from that which the same steel shows when this network is obhterated and its filaments are replaced by a fine and compact conglomerate. In the former case the reticular structure entirely subdivides the metallic mass along surfaces occupied by a more easily deformable system than the other superposed structural system. These webs constitute in their entirety the total resisting section and constitute a continuous plastic system. In the latter case, the small plastic elements are isolated and therefore no longer support the totality of stress to which the metal is subjected. In the first case it is clear that the average resistance of the ma- terial is determined by that of the ferrite constituting the large network. Observations demonstrate this to be the constituent of minor strength. The material will also necessarily exhibit the characteristic behaviour of a metal whose structure is a conglomeration of cellular nuclei bound to- gether with a continuous plastic cement; a supposition which aU the observations clearly confirm. Once, then, the heat treatment has destroyed the continuity of the wide network, the variations in the state of subdivision of the isolated plastic elements of ferrite cannot longer exercise a major action upon the average mechanical properties of the material. From this point on, therefore, the effects upon physical properties exercised by the modifica- tions undergone by other structural systems, principally that of fhe den- drites, begin to be felt in greater measure, inasmuch as they are no longer disguised by the presence of wide ferrite meshwork. It is well known that these arguments can be applied only in the cases in which the material is subjected to stresses whose intensity increases gradually and slowly. Sudden stress or "impact" profoundly modifies the specific action of the continuous network of plastic cement (in this case the wide network of ferrite), and it is no longer the plastic material constituting such network which necessarily must give way first. Besides, it is easily seen how all the phenomena which we have con- sidered in the last sections constitute merely a practical illustration of the reasoning developed in Part II regarding the relations between the mor- phological form assumed by the structural systems resulting from the y—^a transformation and the heterogeneity in concentration, remaining in the y mixed crystals at the initial moment of such transformation. 133. The 2-per cent, nickel steel to which the previous observa- tions may apply constitutes what might be called a typical instance of SOFT AND MEDIUM STEELS 191 the medium-hard special steels largely used in machine construction. This is the reason why I have thought it opportune to dwell a little longer upon the analysis of the properties which such a steel may assume by virtue of normalizing heat treatments, especially as con- nected with the structural characters corresponding to such properties. For the same reason it may be useful in this section to sum up briefly the principal conclusions which have been deduced from the experiments, as far as that particular type of steel is concerned: 1. Two structural systems, viz. a system of dendrites, and a wide network of ferrite, are in evidence in the metal as cast, when revealed by the two chemical methods of etching already indicated, and are inde- pendent of each other as far as the form and the position of their indi- vidual crystallites are concerned. 2. While the homogeneity heat treatments comprising reheatings above Acs (either followed by subsequent quenchings or not) make their effects felt immediately upon the ferrite network, suppressing and transforming it into finer network in the way and by virtue of the effects studied in Part II, the dendritic system does not disappear nor does it lose its initial characters except after extremely drastic homogeneity heat treatments. 3. No relation appears between the form of the respective elements assumed by the two systems after various treatments. It is timely to point out now that it might be thought that the fine network of ferrite might possess some singularities in connection with the cavities, whose images are contained in all the microphotographs reproduced in the previous pages. Direct examination of large polished surfaces of the various samples of the steel in question excludes this hypothesis. As an example, Fig. 75 shows the structure of test-piece A2 (already reproduced in Fig. 74 after the double etching) simply polished and etched with an alcoholic solution of picric acid. This third observation, as well as the two immediately preceding, proves that the process generally admitted to explain the genesis and transformations of the ferrite system in hypoeutectoid steels, cannot be accepted in the simpler form which identifies the system of the den- drites with that of the primary austenite. It is necessary at least to admit that the crystallites revealed by the hot sulphuric acid etching are individuals comprised of colonies of crystalline particles of primary austenite. 4. Examining the behavior of the steel under the action of gradual stresses after treatment in various ways, it clearly appears that the structural system of ferrite exerts an influence which is felt in an abso- lutely preponderant degree as long as this system appears in the form of the wide network characteristic of ingotism. When heat treatment has obhterated this reticulation, substituting other networks derived 192 HEAT TREATMENT OF SOFT AND MEDIUM STEELS from it, the influence of one or more other systems, endowed with charac- teristic critical intervals, are added to that of the first. 5. These observations sometimes allow one to attribute a signifi- cance sufficiently precise from the metallographic point of view to the Fig. 75. — Same area as Fig. 74, ropolished and etched with picric acid. two characteristic fractures, "granular "and "fibrous." Such significance may also be differentiated for the cases in which fracture occurs under action of suddenly applied stresses and for those in which it is caused by gradual loading. CHAPTER XVI CRITERIA FOR THE SELECTION OF CASTING STEELS 134. As has already been said in the previous section, the 2-per cent, nickel steel experimented with constitutes a typical case of the medium- hard special steels usually employed in present metallurgical practice for the construction of machine parts designed to resist intense stresses. As we shall see better further on, steels of such a type are well adapted to these practical applications. They may be given a great tensile strength and a high elastic limit, although retaining good reduction in area, a strong elongation and a high resistance to impact, by subjecting them to a suitable heat treatment after forging or other hot-work is completed. In other words: such steels, properly treated, may possess high tenacity and good hardness, although maintaining all those properties which are indicated with the general term toughness, a word which is in much use among workmen when passing a judgment based upon the fibrous appearance of the fracture. However, for various reasons of expediency it is not always possible to use special steels of that group. A first series of such reasons is of a purely economic nature. Some- times the higher cost of the special nickel steels in comparison with the cost of ordinary carbon steels, excludes the use of the former in all those cases in which the necessity of obtaining the superior mechanical proper- ties of which they are susceptible,' does not constitute so important a condition as to subordinate considerations of cost. On the other hand, a second series of reasons which sometimes makes it unadvisable to use the special steels of the indicated group, has a more technical nature. The manufacture, fabrication, treatment and employ- ment of these steels are a great deal more difficult and more delicate than plain carbon steels. While the best mechanical properties which the latter may possess may be obtained with relative faciUty and safety even by a technical personnel devoid of special ability and experience, when using alloy steel it is necessary to enlist the aid of very capable techni- cians, in order to obtain with certainty and consistency the superior results which those steels may furnish. And whenever such a staff is lacking or is imperfect, so that all the complex rules which must regulate the method of manufacture and treatment are not observed and strictly controlled, it may easily happen that the same steels show results actually inferior to those which the limited ability of the same personnel would 13 193 194 HEAT TREATMENT OF have been able to reach with certainty with other less expensive carbon steels. 135. In case the steel need not undergo hot work such as forging or rolling, but is to be used for castings, the commerical considerations just mentioned lead to the selection of very low carbon steels, contain- ing the proportion of silicon strictly necessary to avoid the formation of too many and too large blow holes, and also a minimum percent- age of manganese to avoid the strong tendency to the formation of intercrystalline cracks or checks. Steels of this type have a relatively low cost and do not present any special difficulty in manufacture and treatment. Also (and this perhaps is the property which renders their use advisable in the more ordinary cases) even large variations in the conduct of the melting, refining, deoxidizing, pouring, teeming, cooling and heat treating operations do not produce important variations in their mechanical properties. Thus, their use is relatively safe in the sense that it is very unlikely to cause unforeseen difficulties — even if in the hands of a second-class personnel. Naturally these steels are not susceptible of possessing mechanical properties, and especially elastic limits, as high as those which can be obtained with steels higher in carbon and manganese, therefore their use is limited to the construction of mechanical parts which will be subjected in service only to limited stresses. In order to give a concrete example of the characters of steels forming this group, I will select the typical case of a very soft carbon steel, in which the peculiarities are accentuated. It was made in the basic open-hearth furnace and bottom-cast in ordinary ingots about 8 in. square in section, and was of the following composition: Carbon . 08 per cent. Manganese 0. 52 per cent. Silicon . 07 per cent. Sulphur 0.01 per cent. Phosphorus . 03 per cent. Figure 76 reproduces at about natural size the appearance of the surface obtained by cross-cutting one of the ingots normal to its main axis at about three quarters the total height measured from the bottom. The surface was accurately polished and etched for one and a half hours with a 20 per cent, solution of sulphuric acid, maintained at 60°C. As is evident, such attack has not revealed a dendritic structure, properly speaking, nor did such a structure appear upon the surface even after prolonging the etching for many hours. This observation con- firms the correctness of the observations made in Part I regarding the behavior of the substances which are found in solid solution in the y crystals, and their responsibility for the local differences of concentra- SOFT AND MEDIUM STEELS 195 Fio. 76. — Macrostructure of 0.06-per cent, carbon steel ingot, cross cut. Etched with hot 20 per cent. H2SO4. Natural size. 196 HEAT TREATMENT OF tions in various portions of the primary dendrites. It is not pertinent to insist upon this point at this time. What is of more interest is to examine the modifications in physical n ™ ■ M^SI^^H^ "^^1 sSB^K^Bm : ^^H ^^BH^^ */jHI J^m ' 1 ^^^^HHbS|> ^ ■^■^^H '91 ^^^^^^^^^^^^^He7 * ^IJ^^^HJPR ^^1 mr /^a H m ■' 1 ^^H^Hm^^^^^H kS^ ^^^^H 1 L^l fl ^^n^^^^i .pu ^K' ^^1 ■ m^^ 1 I 1^1 fl f^^l j^ Ji ^H^,' ^ ^^H 'i^^^i H^^l ii^^^i 1 Fig. 77. — Tension test-piece of 0.06-pcr cent, carbon casting. properties caused by homogeneity heat treatments of average intensity in a steel of this type in which the dendritic system is so little developed. Two contiguous bars were cut from the piece whose section is shown SOFT AND MEDIUM STEELS 197 in Fig. 76, parallel to the same face of the ingot, so that their axis was about half way the distance from center to surface. A cylindrical tension test-piece of 13.8 mm. dia. and 50 mm. long between reference points was cut from one of the untreated bars. A test ' made upon this gave the following results: Tensile strength 51,000 lb. Elastic limit 31,000 lb. Elongation 26.0 per cent. Reduction of area 21.8 per cent. The test-piece was strongly stretched throughout its entire length at the instant of failure, giving the surface a dappled appearance as can be seen in Fig. 77, showing the appearance of the cylindrical surface and the fracture. The second test-piece was subjected to the following treatment: heated at 900°C. for four hours; quenched in cold water; reheated at 650°C. for two hours and allowed to cool slowly. '^■■jsm^i^ !i«H-. . i'K'f. "^■:i m-^i -Vy : v>^,<- l^' u. "M Tptft i? ">'," , ^r'-- ■■'■ ■ ^ff ^■mLj^-y&i) :s. H%'t M' :/^^,1/ Mj jTjQ 7g, — Microstructure of 0.06-per cent, carbon steel casting. X 100. From the bar thus treated was taken a cylindrical test-piece of the same dimensions as the previous one (13.8 X 50 mm.) which gave the following results in tension. Tensile strength 60,.300 lb. Elastic Hmit 42,.500 lb. Elongation 20 . per cent. Reduction of area 29.9 per cent. This test-piece showed a well locahzed neck. 198 HEAT TREATMENT OF From this data it seems evident that the phenomena of ingotism are a great deal less developed in the very soft carbon steel which we have now tested than in the medium nickel steel which we examined ■previously. This of course could be forseen when taking into account the known fact that those phenomena are due mainly to the wide ferrite network, which does not exist in very soft steels. In fact, ferrite consti- tutes the fundamental plastic mass of metal in low carbon steels, while the pearlite which fills the meshes of the ferrite network in the cast medium- hard steels forms but small islands in the former, which of themselves can not modify appreciably the continuity of the plastic mass. This fact is illustrated by Fig. 78, which shows at 100 dia. the structure of this low carbon steel as cast. Bearing in mind the observations already made in the previous sections regarding the specific function of the wide, open network of ferrite, it would be superfluous to insist further upon this point. As far as the numerical values of the physical properties are concerned, the data just given show clearly that the variations which may be ob- tained by homogeneity heat treatments are a great deal smaller in range for the very soft steels of the general analysis given than for the medium- hard nickel steel which was previously investigated. 136. From the point of view of steel casting practice the data which have been recorded up to this point constitute a clear confirmation of what has been already said in the previous sections on this same subject. It is possible by the use of low carbon steels, to obtain steel castings with relative facility which possess good mechanical characteristics, and are especially free from brittleness, even when all stages of their manu- facture are not supervised by exceptionally competent and experienced men. But in such cases, it is necessary to restrict the manufacture to products of low elastic limit. However, in the manufacture of steel castings with high tenacity and particularly with a high elastic limit, it is necessary to resort to the use of medium alloy steels, and special care in the selection, execution and control of all the phases of the manufacturing process becomes in- dispensable in order to avoid very dangerous accidental irregularities in the quality of the products. Therefore such can be conveniently manu- factured only in plants having the most perfect manufacturing equip- ment and control, with a thoroughly experienced organization at its disposal. Later on I shall have occasion to mention some examples showing how it is possible to obtain consistently uniform acid open-hearth steel castings of high tensile strength and elastic limit, whose mechanical properties indicating quality either singly or as a whole are equal and even superior to those of the best forged steels. This in case the castings are submitted to the necessary heat treatments. SOFT AND MEDIUM STEELS 199 It is not necessary to add words to explain what a wide field is thus opened to the application of steel products in the technique of jnechan- ical construction. 137. Notwithstanding the evident advantage of strictly conforming to the criteria indicated in the last sections when selecting the types of steels to be used for various steel castings, it often happens that reasons of economy prevent the application of those criteria. This is not the place to discuss whether the economy is real or ap- parent. It is enough to indicate that a very large proportion of the cases are based upon summary and often superficial calculations, and the re- sult is usually given so great an importance as to overshadow the purely technical advantages mentioned in the last sections. Yet a more pre- cise and detailed analysis of the real facts would lead to the conclusion that the economies shown by the use of low strength materials are en- tirely negligible and sometimes illusory. A credit is turned into a laTge debit if account is taken of the sometimes very great risks inherent in the use of steel castings incapable of guaranties against accidental de- fects and inferior to those upon which the use of better adapted mate- rials, even if apparently more expensive, would permit a reliance with perfect confidence. Without entering into details of this very important subject, as it would involve us too deeply in questions of cost which are out of place here, one might mention some examples to illustrate the characters of the phenomena which occur during heat treatment of the materials usually employed when it is thought best to give preponderant weight to questions of first cost. The most frequent illustrations may be found in the manufacture of steel castings to withstand relatively high static or dead loads; in other words when they are not to be subjected to shocks, to vibrations, nor to frequently repeated forces. In a general, and admittedly rather a theoretical way, it is thought that such a casting made of plain carbon medium steel may well answer the purpose, provided its elastic limit, tensile strength and elongation have satisfactory values. The piece is then designed so that the factor of safety usually required for the special conditions considered will not be passed when the piece is with- standing the maximum combined stresses calculated to occur during use of the mechanical detail which is to be constructed. Starting from this basis and taking into consideration the fact that plain steels cost less than the special steels of a type similar to the one we have already studied, both on account of the lesser value of the raw materials entering into its making, and for some economies which can be realized in the execution of the various processes of melting, hot-working, heat treating and machining, preference is often given to the former rather than to the latter, disregarding the quality of toughness as revealed by fibrous 200 HEAT TREATMENT OF fractures and measured by the great reduction of area, and by the high resistance to impact. All these properties may be constantly and surely obtained in suitably treated special steels, but cannot be given a pre- cise money value in the cost calculations for the piece to be fabricated. The opportunity of selecting our examples in this field is the larger, since medium carbon steels are at present still widely used in the manu- facture of castings for machine construction. 138. A first series of structural properties clearly differentiate the medium carbon steels from the medium special steels of the same type as the 2-per cent, nickel steel studied in the previous sections. Especially is this true from the point of view which interests us more particularly : that of the effects of heat treatment. These characters are caused by the greater tendency to form that plate-like arrangement of ferrite called Widmanstattian structure. Of this we have already seen some examples in Sec. 114 and 115 when discussing a carbon steel treated in such a way as to produce the maxi- mum degree of homogeneity practically attainable in the individual y mixed crystals. From the numerous microphotographs reproduced in the previous sections, it appears that the tendency to the formation of such structure is nil — or, at least, very small — ^in the 2-per cent, nickel steel. This char- acter is closely connected with that other property of this steel which easily allows it to acquire a high toughness under the action of the same heat treatments. The suppression of Widmanstattian structure is never found in equal measure in the medium carbon steels as an effect of properly selected homogeneity heat treatments. Experience shows that it is exactly to this difference of behavior in the crystalline texture to which may be ascribed the reluctance with which such steels acquire a high toughness. Therefore, it is evidently important to examine, at least summarily, the laws regulating the formation and the development of Widmanstattian structure in the medium carbon steels. During this rapid examination, the case which has been already indicated in Sec. 114 will not be further emphasized. In this the pre- liminary heat treatment has produced practically complete homogeneity of concentration of the various elements in the y solid solution. The characteristics of this limiting case have been set forth in the morpho- logical observations" contained in Sec. 114, and with the general argument in Part II, Sec. 46 to 62. Instead, the only case which appears in prac- tice will be considered. In this the heterogeneties of concentration which arose in the austenite during the primary crystallization of the steel, has only been partially reduced by later manipulation. 139. It is even possible to establish the following facts "a priori" on the basis of the considerations developed in Part II; facts which numer- SOFT AND MEDIUM STEELS 201 ous observations, omitted here for the sake of brevity, confirm regarding the morphological characters of the structures resulting from the actual y—>a transformation : 1. If the coohng takes place extremely slowly (with a velocity of the order indicated in Sec. 114), and the crystallization of ferrite is not dis- tiu-bed in any way, as it would be for instance during forging or rolling, it is clear that the first crystals of ferrite would form at the points where the carbon concentration is lowest, namely, coincident with the nuclei of the primary y mixed crystals. Reasons for this statement have been minutely analyzed in Part II. Besides, the slowness with which the cooling takes place through the entire interval of temperature corre- sponding to the segregation of ferrite, causes those first isolated portions to exercise their maximum influence as crystallization germs, extending to all other particles of ferrite crystallizing in that neighborhood. It follows that during quite slow cooling the crystals of ferrite start along the axes of the primary mixed crystals, and develop in branches constituting the meshes of the ferrite network. Thus originates the structure designated by Belaiew with the expression "structure of great crystals." 2. If the coohng still takes place slowly, but at a quicker rate than that indicated in Sec. 114 — so that, for instance, the passage from bright red (about 800°C.) to dark (below 550°C.) takes place in a lapse of time varying between 2 and 20 hours — and the crystallization of ferrite is not distxirbed mechanically, the morphological characters of the systems resulting from the 7— »a transformation are somewhat more complex. In fact, the marshalling power of the first separated crystalline elements of /3 and a ferrite cannot reach so important a value as to influence sub- stantially all of those last particles of ferrite forming in the successive steps of- the transformation. Thus a part of the last to be precipitated, not having time to migrate toward those earlier groups, orients itself around new centers of crystaUization forming in interior portions of the primary y mixed crystals. Ferrite will therefore divide in this most frequent case in the following distinct portions : (a) Ferrite of the open network, arranged in "branches" radiating from points on the axes of the primary mixed crystals. (6) Ferrite in independent crystals developed around secondary centers of crystallization, not located at points along the axes of the primary 7 crystals. We have already studied in Part II, and especially in Sec. 63 to 75, the manner in which the course of these phenomena is influenced by variant concentrations of the different elements in the primary solid solution. It is not necessary to repeat those considerations. However, 202 HEAT TREATMENT OF it is necessary to keep the details well in mind in order to interpret correctly the crystallographic phenomena we are concerned with now. 3. Cooling takes place rapidly, though without reaching the quench- ing velocity necessary to prevent the 7— »a transformation more or less completely. The marshalling action of the ferrite crystallization-germs formed at each given instant of the process upon the portions separating subsequently, now assumes smaller values the more rapid the cooling. Consequently it results that the development of the individual ferrite crystallites is reduced correspondingly, be they those forming the network or those independent of the meshes. Such cases will occur, for instance, when the steel is rapidly cooled in the air, in hot oil, or in a lead bath, as has been already indicated in Sec. 87, which processes are often incorrectly called "negative quenching." The crystalline elements of ferrite are broken up into minute fragments so that the result is a conglomera- tion of little crystals of ferrite, more or less uniform in size, scattered somewhat evenly throughout the kernels of pearlite. The ferrite crystals are smaller, the more rapid the cooling has been, other conditions being equal. 140. Separately examining the shape and disposition of the ferrite crystals belonging to groups (a) and (6) specified in the previous section, one might note the following facts : 1. The network appears similarly when it gathers all the free ferrite (case No. 1 of Sec. 139). In the case we are now considering however, the relative proportion of independent ferrite may increase a great deal at the corresponding expense of the network. Then the branches lead- ing away from the primary nuclei do not grow far enough to meet each other and close the meshwork. To be exact it would then be improper to speak of a network of ferrite, therefore, I usually denote it by saying that we have a "network with interrupted meshes." 2. Crystals of ferrite independent of the network ma^y appear under various aspects. We have already seen in Sec. 66 that such crystals may form in three distinct regions of the y solid solution existing on all sides of the ferrite comprising previously separated parts of the large network. Therefore, we must expect, and experience confirms this expectation, that they do not show uniform morphological characters. It would require too long a digression to analyze the various appearances caused by such diversity; it must be sufficient to say: (a) In those austenitic regions where the carbon is below 0.35 per cent., ferrite starts to crystallize in the /3 state assuming its characteristic "globular" form. This habit is well known because it constitutes almost the entire quantity of the ferrite in very soft steels. We have already spoken of such crystalline appearance in Sec. 114, giving a typical example. SOFT AND MEDIUM STEELS 203 Only when the precipitation of the ferrite in these regions has gone so far as to cause the carbon concentration in the neighboring austenite to rise above 0.35 per cent, does the newly separating ferrite form directly as alpha iron, taking the lamellar form which characterizes all the ferrite which forms in the regions to be examined in the following section, (6). (b) Alpha iron separates directly as such in the regions where the con- centration of carbon in austenite is greater than about 0.35 per cent. As has been remarked many times, alpha iron assumes a characteristic lamel- lar form, which appears in microscopic sections as commonly observed needles, more or less well developed. Such needles are exactly those which give rise to the so-called "Wid- manstattian structure" when they are highly developed. 3. As a deduction from the entire foregoing argument, it is clear that the relative division of ferrite between the network and the independent crystaUites depends upon the heterogeneity of concentration in the y soMd solution, upon the velocity of coohng of the metal and upon the nature of the steel. If the non-uniformity of chemical composition remaining in the aus- tenite after normalizing is still great, although less than that degree which would cause substantially all the ferrite to become part of the meshes of the reticular structure, the amount of ferrite in the network wiU be larger than that existing as independent crystals in the final complex structure. The contrary ratio will obtain if heat treatment has more effectively smoothed out the original heterogeneities in the y mixed crystals. In the same way, and other conditions being equal, the proportion of ferrite in the network will be larger than in independent crystals, the slower the cooling of the system. It is particularly necessary however that the heterogeneities of chemical composition in the austenite be exactly equal in comparative specimens. Finally, the relation between the two portions of the ferrite varies widely with the nature of the steel, always considering other conditions, are equal. Although the considerations of average chemical composition of the metal are of the highest practical importance, it is not possible to enter into a detailed discussion of them. I wiU content myseK with pointing out a fact which, after all, could clearly be deduced from what has been seen in Part II : that the proportion of ferrite finally existing in indepen- dent crystals is higher, the greater the frequency of the crystallization centers forming in the austenite under given conditions of cooling. Now, it is known that this frequency is a specific character of each sohd solu- tion, and therefore of each steel. 141. We have already seen a characteristic example of the facts in- dicated in the previous section in the study of the 2-per cent, nickel steel examined in the preceding chapter. This metal belongs to one of the cate- 204 HEAT TREATMENT OF gories of steels characterized by closely spaced centers of crystallization of ferrite in the y solid solution. After all one could easily infer this from the microstructure which has been examined. In fact, we have seen that numerous independent crystals of ferrite already appear in the ingot (for example sec Fig. 48 and 51) in which the austcnitic heterogeneity is a maximum. Also that all heatings to temperatures within the field of y solid solution as demarked in the equihbrium diagram even when followed by a relatively slow cooling, gave rise to a conglomeration of finely divided ferrite crystals, in which the last trace of the large network had practically disappeared. ^s B^^0 K^^^^ ^^4'M m^^ ^^^ 1 ^Mhi.j',-r'^\ **5K*S^^'T?*^ ^Iv ^^M^^^^Si *^&fi.' Fig. 79. — Dcudritic structure of 0.3.3-per cent, carbon steel. X 2. Etched with hot 2(1 per cent. H2SO4. In order to explain the phenomena mentioned in the last three sections, better basing the illustration upon concrete examples, one might compare the experimental data given for the 2-per cent, nickel steel, with some similar data concerning another steel, characterized by a lesser frequency of crystalhzation germs appearing in the 7 solid solution. To this effect let us examine a medium carbon steel of the following composition: Carbon . 33 per cent. Manganese 0.59 per cent. Silicon 0.21 per cent. Sulphur 0.03 per cent. Phosphorus 0.04 per cent. SOFT AND MEDIUM STEELS 205 Figure 79 reproduces at 2 dia. the appearance of a sample taken from the outer part of an unworked ingot weighing 1 ton. The sample was polished and etched for 3 hours with a hot 20 per cent, solution of sul- phuric acid in water. The dendritic structure of the metal is clearly seen although it is less prominent than in the 2-per cent, nickel steel. Figure 80 shows the same piece at the same enlargement after being repolished and then deepty etched with a solution of 5-per cent, nitric acid in amyl alcohol. The effects of the phenomena already studied in Sec. 120 and 121 can clearly be observed here. For the reasons there Fig. 80. — Large ferrito network of Fig. 79. X 2. amyl alcohol. Patched with 5 per cent. HNOa in given the proportion of ferrite is a great deal less than the amount which should appear in a steel of the indicated composition if the 7-^a trans- formation took place through a scries of systems in complete equihbrium. When reheating a number of large castings made of this steel, it happened that some of them cooled from above 850°C. at an extremely slow pace, requiring more than 40 hours to reach atmosphere — about 20°C. This happened because they were in the lower part of a pile charged upon the bottom of a very large furnace. The structure of the steel so treated is reproduced at an enlargement of 180 dia. in Fig. 81 after pohshing and etching with a solution of nitric acid in amyl alcohol. The proportion of ferrite existing as independent 206 HEAT TREATMENT OF crystals is clearly a great deal less than that comprising the network, which may be observed to be very broad. A tensile test-piece, 22.6 mm. in diameter and 100 mm. long between gauge marks, taken from one of the castings so treated, gave the follow- ing results: Tensile strength 8.3,500 lb. per square inch Elongation 20 per cent. Reduction of area 12 per cent. This test-piece was strongly stretched on all its length, as seen in Fig. 82 at the right. The coarsely crystalline fracture of the same test-piece is reproduced at the right in the next figure. Fig. 83. Fig. SI. -Broad network in 0.33-pGr cent, carbon .steel after extremely ^low cooling. lOtched with HNO3 in aniyl ulcoliol. X ISO. A prismatic test-piece of the same steel 20 mm. square in section bent gradually and broke after a very small deflection (angle at fracture about 140°). Figure 82 shows the appearance of this broken test-piece at the left, while at the left of the following figure is shown the coarsely crys- talline appearance of its fracture. Another casting of the same heat which was heated to between 830° and 850°C. and cooled less slowly than the previous, taking about 24 hours, possessed the microstructure shown at 180 dia. in Fig. 84. It is evident that the proportion of ferrite in independent crystals in com- parison with that of the network is largely increased, and the network is no longer of closed meshes, but its partially formed portions appear only as detached "islands" (see Sec. 140 and 141). SOFT AND MEDIUM STEELS 207 Fig. 82. — Bending and tension test-piece.s, slowly cooled 0.33-per cent. carVjon steel. ^:;,,^j|g -nw^ :/ ■.■WL£i. h k^ ^i ^^Ba^m ^:"a transformation is strictly bound to the course of the primary crystaHization, and to the distri- bution of various elements in the 7 mixed crystals (in the way we have 1 See Sec. 115 and 138. SOFT AND MEDIUM STEELS 213 Studied in detail in Chap. I and II). In particular, the just mentioned effects of the frequency of ferrite germs appearing in the y mixed crys- tals are added to those due to a larger or smaller frequency of the centers of primary crystallization which characterizes the sohdification of the molten steel (and therefore to the larger or smaller dimensions of the band of variant concentration in the austenite, which is one of the most direct consequences of this frequency). As we have seen in Part I and particularly at the end of Sec. 20 and the beginning of Sec. 23, the two phenomena show concordant characters in the sense that as the centers of primary crystallization are closer spaced, the homogeneity heat treat- ments are rendered more rapid and more efficacious. Austenite germs thus act in precisely the same way as closely spaced ferrite germs by decreasing the distances across which diffusion must take place. There- fore, we must not forget the important fact that the frequency of the centers of primary crystalHzation is an element of great importance which determines the tendency of a steel to respond to preliminary heat treatments. In this regard it is timely to observe that the frequency of the centers of primary crystallization is unique in that it makes itself felt directly upon the mechanical and physical properties of steels in which the y—>a transformation has been totally prevented by drastic quenching. However, the steels to which our study refers exclusively are very rarely and exceptionally used in this state. The above mentioned direct action has therefore a substantially lower importance than the influence it indirectly exercises upon the formation of ferrite, by virtue of its influence on the dimensions of the "area of variant concentrations" in the y mixed crystals. 144. In this last observation, we have a more precise explanation than the one given in Part III, especially in Sec. 84 and 85, of the fundamental differences between the effects produced by simple annealings and those possible to obtain with the aid of quenching when working the soft and the medium steels ordinarily used in mechanical construction. In fact, from the theoretical considerations developed in Part II and the concrete examples noted in the last chapters, it is evident that increasing the homo- geneity of the various chemical constituents in the austenite by a more or less prolonged reheating at a temperature above Aci does not prevent the appearance of new and strong heterogeneities in carbon concentration on subsequent cooling — provided it takes place with sufficient slowness. On the contrary, we have seen how these new heterogeneities may reach very high values, even though hmited to a variation in carbon concentration, evidencing themselves by the reappearance of the large network of ferrite or to the formation of Widmanstattian structure. Without going over those phenomena again to examine that which we have already studied with sufficient detail in Parts II, III and IV, 214 HEAT TREATMENT OF it may be useful to sum up briefly the characters by which the practical effects of preliminary annealing differ from those of preliminary quenching. This with the end in view of clarifying the exact understanding of the examples to be presented in the following pages. A. Preliminary annealing is defined by the process of holding the steel at a temperature above its last point of transformation upon heating (Aci) followed by a slow enough cooling to allow the substantial completion of the 7— >« transformation. This process has as a constant fundamental effect the decrease of the heterogeneities in the concentrations of the various constituent elements, except carbon. NormaUzation is the more pronounced the more prolonged the heating and the higher the annealing temperature. As far as carbon is concerned, it is true that we have seen that the conditions for its uniform concentration depend upon the time and temperature of the reheating, but afterwards during slow coohng this uniformity suffers such modifications as may completely alter and sometimes even reverse the effects of the previous anneal. We have seen in Part II that this unique behavior of carbon is due to the fact that within the limits of composition comprised by the steels to which our study is restricted, only the iron: carbon soHd solutions (or, more precisely, the sohd solutions between iron and iron carbide) enter insoluble regions of the, equilibrium diagram during coohng, cor- responding to the precipitation of j3 and a iron. B. Preliminary quenching consists in reheating steel to a temperature above its last point of transformation on heating, followed by a rapid enough cooling to prevent the 7-^0: transformation more or less com- pletely. It has as a fundamental effect the decrease of the heteroge- neities of concentrations of all the elements which were found in solid solution in 7 iron at the higher temperature reached by the treatment. Quenching is more efficacious the more prolonged and the higher the temperature of the heating and the more rapid the subsequent cooling. We have already seen that the velocity of cooling necessary to produce the effects above indicated varies within quite wide limits with the chemical composition of the steel treated. (See especially Sec. 86 to 88.) 145. And here I may mention again a consideration of important practical interest concerning the control of the preUminary quenchings by microscopic examination of the structure of the steel treated; a consideration which is founded upon observations already repeated many times during the first three parts of this volume. We have seen that, all other conditions being equal, the diffusion of carbon in the 7 mixed crystals takes place a great deal more rapidly than that of any of the other elements contained in steel, because of the aid given by gaseous carbon compounds. Corresponding com- pounds are lacking for the other elements. Besides we have also SOFT AND MEDIUM STEELS 215 seen that for well known reasons the only chemical heterogeneities directly revealed by the microscope are' precisely those of carbon. Non- uniformities in distribution of the other elements are revealed only indirectly through the heterogeneity of carbon remaining after heat treating in many different ways. They therefore do not appear in quenched and drawn steels. It consequently often happens in practice that a superficial microscopic examination, made only upon the steel as quenched and drawn, shows apparently a good homogeneity, when the observer has not taken the trouble to repeat it upon the same steel after having been subjected to a further true annealing at a temperature above Aca- The apparent homogeneity on first examination would then be shown to hold only for carbon, and the other elements may still be distributed so discordantly as to cause such inferior quahties in the steel as to approach actual ingotism. We have already seen very clear examples of these facts. Thus, for the 2-per cent, nickel steel which was studied somewhat in detail, it has already been pointed out that the data evidently indicates the much higher velocity with which a good uniformity of the carbon con- centration was reached by means of the heat treatments, in comparison to that which approached stable equilibrium for the concentrations of the other elements. The uniform distribution of carbon shown by the state of the ferrite disguised the heterogeneities of the concentrations of other elements, heterogeneities revealed essentially by the state of the dendritic system and by the mechanical properties — particularly by the reduction of area in tension and by the impact test. We shall see further on many other examples of the phenomena now indicated besides the one already given. However, inasmuch as the effects of such phenomena are felt in a particularly intense manner and with better defined characteristics for steels which have undergone hot-work such as forging or roUing, those examples will be reserved to the time when we shall study these last steels. CHAPTER XVII PRELIMINARY HEAT TREATMENT OF STEEL CASTINGS 146. The observations and experimental data referred to in the pre- ceding chapters of Part IV illustrate with sufficient clearness the funda- mental considerations developed in Parts I, II and III. Therefore it would be superfluous to repeat a detailed examination of their applica- tions to other practical examples. On the other hand, it is useful to show by means of some further illustrations the great variety of results possible to obtain on various types of steels in practice by applying the processes which we have studied in a theoretical outline, especially in connection with the dif- ferences existing between the effects of homogeneity quenchings and of mere homogeneity annealings. Therefore, I shall add to the data al- ready presented, giving only some figures regarding the practical results obtained by subjecting cast steels of various types to determined heat treatments. The reader himself can make the correct interpretation of the data, after all a very easy task, if founded on the basis of criteria such as those followed in the preceding chapters. In order not to multiply the examples unduly, some of the more characteristic and so to speak typical instances will be chosen. 147. A start will be made with a carbon steel for castings softer than the one examined in Sec. 141. The chemical composition of this steel was as follows : Carbon 0.23 per cent. Manganese . 66 per cent. SiMcon 0.21 per cent. Phosphorus . 05 per cent. Sulphur . 03 per cent. A casting of this steel was reheated at 850° to 900"C. and allowed to cool very slowly, requiring about 36 hours to reach room temperature. It showed the microstructure reproduced in Fig. 89 (at an enlargement of 100 dia.), and the following physical properties, determined by pull- ing a tensile test-piece 12 mm. diameter and 40 mm. long between ref- erence points. Tensile strength 61,600 lb. per square inch Elastic limit 34,300 lb. per square inch Elongation 13 per cent. 216 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 217 A prismatic test-piece 20 mm. square, tested liiuler static flexure broke at an angle of 130° with a coarsely crystalline fracture reproduced in Fig. 90. Fig. 89. — Microstructure of 0.23-per cent, carbon steel; annealed at 875°C. and very slowly cooled. X 100. I ■ i 1 1 1 ■'■1 f , %. Fig. 90. — Fracture under bending. Same steel as Fig. 89. The same steel, reheated during 12 hours at 800° to 850°C. and left to cool, requiring ten hours to reach atmospheric temperature, showed 218 HEAT TREATMENT OF the microstructurc reproduced in Fig. 91 (also enlarged 100 dia.) and the following physical properties, determined as before. Tensile strength 68,000 lb, per square inch Elastic limit 41,000 lb. per square inch Elongation 17.0 per cent. Reduction of area 13.9 per cent. Fia. 91. — Microstructure of 0.23-per cent, carbon steel; annealed at 825-°C. and slowly cooled. X 100. Fig. 92. — Part crystaUine and part fibrous tension fracture of steel shown in Fig. 91. The tension test revealed a mixed structure (part crystalhne and part fibrous texture) as shown in Fig. 92. The static bending piece bent into a U shape. SOFT AND MEDIUM STEELS 219 Finally, another casting of the same steel — after undergoing the first treatment above mentioned — was reheated at 800°C. during 20 minutes, quenched in boiling water, and finally drawn during one hour at 650°C. Fig. 93. — Microstructure of 0.23-per cent, carbon steel after complex heat treatment. X 100. Fig. 94. — Fibrous fracture of steel shown in Fig. 93. A test made under the identical conditions of the two previous ones gave the following results : Tensile strength 73,600 lb. per square inch Elastic limit 47,000 lb. per square inch Elongation 17.5 per cent. Reduction of area 24. 8 per cent. 220 UK AT THKATMENT OF The microstruciurc of the steel thus treated is j-cproduced in Fig. 93 at the same enlargement as Figs. 8!) and 91. The fracture of the broken test-piece reproduced in Fig. 94 was entirely fibrous. The prismatic flexure test bent into a U sinrilar to the previous one. It is easy to sec that all the above data are in complete accord with all the previously developed considerations, both regarding the observed structures and the test-results. 148. Another mild carbon-steel may be cited, made by the acid open-hearth process and having the following composition. Carbon . 2.5 per cent. Manganese . 8.5 per cent. Silicon 0.15 per cent. Sulphur . 02 per cent. Phosphorus . . ■ . 03 per cent. It was cast in ingots 15% inches (400 mm.) square. Figure 95 at 100 dia. shows the structure of this steel as cast. Fig. 9.5. — Ingot .structure of 0.2o-per cent, carbon acid steel. X 100. Rectangular bars .Sig X 5% X 11% inches (80 X 150 X 300 mm.) were taken from the lower half of those ingots, and subjected to the heat treatments indicated in the second column of Table VI. In the next four columns of the same table are noted the principal results of tension tests (13.8 mm. in diameter by 100 mm. long between reference points) taken from the bar after being subjected to the corresponding heat treatments indicated in the second column. Finally, the last SOFT AND MEDIUM STEELS 221 column lists the figures reproducing the microstructure of the variously treated steels after pohshing and etching with an alcoholic solution of picric acid. Table VI. — Tests on 0.25 Carbon Casting Heat treatment Physical properties No. Tensile strength, lb. per sq. in. Elastic linait, lb. per sq. in. Elonga- tion, per cent. Reduction in area, per cent. Structure (enlarged 100 dia.) 1 2 3 4 Annealed at 940°C. during 45 min. followed by slow cooling in the furnace. Annealed at 850°C. during 45 min. followed by slow cooling in the furnace. Heated and cooled as No. 1, followed by a drawing at 700°C. during one hour and subsequent slow cooling in the furnace. Heated and cooled as No. 2, followed by a drawing at 700°C. during one hour and subsequent slow cooling in the furnace. 79,600 76,800 73,200 74,500 42,700 44,200 39,800 42,700 17,5 14.0 21.5 26.0 18 13 27 35 Fig. 96 Fig. 97 Fig. 98 Fig. 99 The tabulated data are evidently in perfect accord with all the con- clusions previously reached regarding the variations in structure and physical properties of soft steels, pro.duced by mild homogeneity heat- . ings. ' Note particularly what has been said relative to new heterogenei- ties induced by recrystallization of ferrite from a system of austenite crystals in which the greatest part of the original chemical heterogeneities has been undisturbed by insufficient anneahng. Inherited non-uniformi- ties are especially evident in the large variations in ductility after having been tempered. To be convinced on this point, compare the figures for elongation and the reduction of area of the first and second test-pieces with the third and fourth. These physical properties find perfect cor- respondence in the structural characters, as can be clearly seen when comparing Fig. 96 and 97 with Fig. 98 and 99. 149. In order to confirm and extend the conclusions reached in Sec. 135 regarding the relatively limited effects which homogeneity heat 222 HEAT TREATMENT OF Fig. 96. — Microstructure of 0.2.5-per cent, carbon steel casting after heat treatment No. 1. X 100. Etched with picric acid in alcohol. Fig. 97. — Microstructure of same casting after heat treatment No. 2. X 100. Etched with picric acid in alcohol. SOFT AND MEDIUM STEELS 223 Fig. 98. — Microstructure of 0.25-per cent, carbon steel casting after heat treatment No. 3 X 100. Etched with picric acid in alcohol. FiQ. 99. — Microstructure of same casting after heat treatment No. 4. X 100. Etched with picric acid in alcohol. 224 HEAT TREATMENT OF treatments produce upon the physical properties of very low carbon cast steels which, by the way, never present the real phenomena of ingotism, I may add in Table VII some comparative data between the effects produced upon the physical properties of one of those steels by a pair of homogeneity quenchings. The steel examined was of the following composition: Carbon 0.09 per cent. Silicon 0.028 per cent. Manganese 0.47 per cent. Sulphur . 01 per cent. Phosphorus 0.02 per cent. Table VII. — Normalization op 0.09 Per Cent. Carbon Steel Heat treatment Physical properties No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elongation per cent. Reduction of area, per cent. 1 2 * Heated at 850°C. for 3 hours followed immediately by quenching in water; then drawn at 650° for 2 hours. Heated for 14 hours at 1100°C. followed immediately by quenching in water. Reheated at 850°C. for 3 hours followed immediately by quenching in water. Finally, drawn at 650°C. for 2 hours. 50,500 54,500 30,000 33,700 31.8 33.0 55.8 60.2 The tensile tests were made upon cylindrical bars 13.8 mm. in diameter, by 100 mm. between the reference points. After what has been said, these results- do not require further comment. When steel is cast in quite large masses, either in ingots or in formed castings, the phenomena of ingotism always appear with a great deal higher intensity than when it is cast in smaller sections. This is due to the well known fact that the slowness of cooling during the solidification range enhances the phenomena of intercrystalline liquation and segrega- tion. Ingotism also shows a greater intensity when massive castings contain large quantities of emulsified inclusions, for the reasons indicated at the end of Part III, and especially in Sec. 98. Evidently the steel then feels the effects of even a mild homogeneity heat treatment a great deal more than it would were it cast in smaller masses. \ SOFT AND MEDIUM STEELS 225 As an example of this fact take the case of an acid opon-hearth steel having the following composition: Carbon 0.47 per cent. Manganese 0,63 per cent. Silicon 0.27 per cent. Sulphur . 029 per cent. Phosphorus . 03 per cent. Rough castings for 381 mm. gun cradles were made with this steel weighing more than 10 tons each. One of these castings was heated for 12 hours at 750°C. Tension test-pieces (cylindrical bars 13.8 X 100 mm.) and Charpy impact samples (30 X 30 X 160 mm., with a cylindrical notch 4 mm. in diameter) were then cut from the body of this casting. The results of these first physical tests were as follows: Tensile strength 76,200 lb. per square inch Elongation 19.0 per cent. Resistance to impact (Charpy test) 1.2 kilogrammetcrs per square centimeter The pulled test-piece shown in Fig. 100 was stretched throughout all its length, and the fracture of the impact test revealed a coarsely crystal- FiG. 100. — stretched surface of tension test on a massive 0.47-per cent, carbon steol casting. line structure, Fig. 101. These properties correspond well to the micro- structure, reproduced at 100 dia. in Fig. 102 (etched with an alcohohc solution of picric acid), and are indicative of strong ingotism partly due to the presence of a laige number of inclusions. I have already spoken in Sec. 98 of the mechanism of such action by emulsified inclusions, appearing clearly in Fig. 102. The same casting reheated for 7 hours at 850°C. and then left to cool in the air, so that its temperature dropped to 400°C. in about 3 hours, revealed the structure reproduced in Fig. 103 (also at 100 dia.). Taking account of all the theoretical considerations presented up to this time and of the conditions in which the treatment has taken place, it can be seen that the effect of the heat treatment upon the structure has been very remarkable. IL/OtJ-l V V-'i J' i v./iAiC*x J.i.tjjr-' PJOtJ-l V VvJ. J' i \^J.JJ.t*X J.l.t^F-' 226 HEAT TREATMENT OF Fig. 101.— Coarsely crystalline fracture of impact test, massive 0.47-per cent, carbon steel casting. Fig. 102. — Microstructure of massive 0.47-pcr cent, carbon steel casting, revealing ingotism and inclusions. X 100. Etched with picric acid in alcohol. SOFT AND MEDIUM STEELS 227 Equally remarkable are the effects upon the physical properties. In fact, the same physical tests were made upon the steel subjected to the second heating, with the following results: Tensile strength 88,000 lb. per square inch Elongation 24 , 00 per cent. Resistance to impact (Charpy test) 3.61 kilogrammeters per square centimeter Fig. 10,3.— Same steel as Fig. 102 after second annealing at 850°C. X 100. Fig. 104. — Smooth .siirfar-e of tension test piefe. Steel of Fig. 100, Ijut twice annealed. The tension test-piece shown in Fig. 104 did not show the least stretching along its length which produces the characteristic dappled surface, and the impact fracture exhibited a finely crystalhne and partly fibrous structure. Fig. 105. This data well indicates the mechanical characteristics of ingotism appearing in thick masses of cast steel, especially when the metal contains large quantities of emulsified inclusions. 228 HEAT TREATMENT OF In fact we must conclude that the normal properties of this steel are still altered or disguised by the effects of ingotism even after a heating at 750°C., denoting by the term "normal properties" the average re- sulting from the superposition of the specific properties of its individual normal structural constituents. Were this not the case, there can be no doubt that the effect of the second heating at 850°C., followed by slow cooling, should have changed the physical properties of steel, in an opposite direction to that which in reahty it has produced. In fact, if the average properties of the steel would have been in large measure deter- mined by summation of the specific properties of the individual constituents, the effect of the second heating could not have been any- thing but a decrease in tensile strength and certainly not the remarkable increase revealed by the experiment. Fro. 105. — Fine cry-stailiiio and fibrous impact fracture. Steel of Fig. 101, but twice annealeti. Therefore the above example indicates clearly the specific fundamen- tal character of all the properties constituting that special state of steel which H. M. Howe has designated with the word "ingotism." It con- firms the idea that the average properties as usually determined on steel in that state are most probably due to the distribution and the continu- ity of the structural constituents rather than to the specific properties of the constituents themselves. 151. We should now sum up the conclusions we have reached regard- ing the results obtainable by various homogeneity heat treatments SOFT AND MEDIUM STEELS 229 when applied to castings for machine parts when manufactured of the types of steels more frequently used for such purposes. In order to give examples which conform to normal industrial con- ditions, I shall select steels of the average purity usually attained in steel foundry operation. This observation holds not only for structural or (improperly called "metallic") impurities such as sulphur, phosphorus, copper, etc.; but also for the foreign or non-metallic impurities, especially the solid inclusions. Results wiU be given for the usual physical tests, i.e., tensile, static bending, and impact upon a nicked bar. In addition to these it will be useful to add the results of an impact tension test, made by the well known method using Charpy's pendulum striking a fixture holding a cylindrical test-piece 10 mm. in dia. by 50 mm. long between the conical heads. The breaking-work will be indicated in kilogram-meters necesssary to rupture a 10-mm. round bar. AH the static tensile tests referred to in the following paragraphs of this chapter were made upon cylindrical test-pieces 13.8 mm. in dia. by 50 mm. between reference points. Static bending tests were made upon rectangular prismatic bars 9.5 X 20 X 160 mm. The test-pieces were placed horizontally upon one of its 20-mm. faces, upon two parallel hard steel rolls, 70 mm. in dia., spaced 140 mm. apart on centers, and free to turn around their axes. The test- piece was forced to bend by pressure imposed upon the upper face by means of a 40 mm. steel cyhnder, whose axis was placed parallel to the end supports, and so arranged that it could move up and down a vertical plane passing midway between. Pressure was maintained until the two ends of the test-piece were bent enough to slip between the two support- ing cylinders, so that at the end of the test, the piece was bent like a U with almost parallel legs. The examples to be mentioned in the following paragraphs are especially interesting when the physical properties obtainable by means of simple homogeneity reheatings and homogeneity quenchings are compared. 152. Starting with ordinary carbon steels for making castings, data relative to two steels containing 0.21 per cent, and 0.34 per cent, carbon are gathered in the following tables. Merely for comparison data ob- tained from forged test-pieces of the same steels have been added, data which will be returned to again in more detail in Part V. Working was done at about 1000°C., hammering a square bar 100 X 100 mm. down to 35 X 35 mm. Pieces for the various physical tests were always cut from the forged bars in a longitudinal direction; that is to say, with their long axis parallel to the direction along which the steel has been extended by forging. The importance of this last remark is well known, and we shall later emphasize it better. 230 HEAT TREATMENT OF ce the iken pieces t^ OS 1-H » o o i-t e tH T— ( T-H M bb bb s s s 0} 03 00 o o O O O TO (N lO {^ TO Tjf "* CO ?; o 4^ A .-g :; s «o t^ 00 (M ^ 0) iJ5>- • O o Q Q ago.-- o o H i> i> ZD CD J ^1=2'" lO" nT cn o' 1 1> t^ t~ 00 PI r-C 0) » T3 00 .fe :^ S OJ o "o — ^ o "o lO o 00 " T3 ^ ^ TO O "S "^ ^J e 03 ;.< ■B fl C3 c3 d OJ o 0) 1 1 -a 03 6°^ -« 8 O -° bS d 03 "m 6 rH (M TO ■* SOFT AND MEDIUM STEELS 231 s 1 ■* CO 00 lilli 1—1 1—1 1-H 1-1 bib 1-) bb 0) CO lO l> 05 &^ 1— < tH I-< tH T-< i-H 1— < i-t .§2 H .# .a= bb .a^ a P^ p^ E f^ |i' 01 g 13 o o IS o s| o o o ^ bll U) bO § 1—1 CO 1-1 1— ( I— ) 6 n hi !> 00 j^ TtH ^ 5 I— I (N CO a a s'i K ^ 9ijS . IC CO O o P^ 11? o> CO >o (N •* V V u c^ 00 CO CO CO c> HiJ< I— 1 r^ a fl •9S"§ o lO ■o o g lag 1-1 CO d eo' >< ■O-H !l! (N CO -* >o i (S^s, o O ■♦3 A .-s ^ s SPfl ® 00 lO -* o So" tH (N IN CO oq o °«fe H a t^ S +a o Q O o ^ .2 o o H -*a ■g.-B S.S (M 00 CO o "? sla^ T^ TlT co" oT 1 U3 "3 CO lo M M «ja t^ .' Q o o E§ a^fea o o g»B,.- >l^.,l^Ji.. », ^^..^L^^^rii^i^»i:^^^:...,„;_.. ._ JIJIIJIJ^.,, '" Fig. 118.— Tests on steel of Fig. 117. Fig. 119. — Same steel as Fig. 113. Forged and anaealed at 850°C. X 80. Etched with HNOa in amyl alcohol. 240 HEAT TREATMENT OF Tests No. 4, while the microstructure of Tests No. 3 given in Fig. 110 and 117 actually appear much more uniform than that of No. 4, shown in Fig. 112 and 119. Therefore, it is clear that the apparent homogeneity revealed by microscopical examination concerns only the distribution of carbon, and does not concern the other elements entering into the composition. 153. I said in Sec. 151 that T would later add complementary data to that for carbon steels in the form of figures for special steels more responsive to homogeneity heat treatments. This information will now be given for a nickel steel with composition similar to that studied in Chapter XV, but somewhat softer. Contrary to the experiments upon simple test-pieces described in that chapter, the present results were ob- /ff«J i+r— S^ - Fig. 120.— DetaUa of casting from which tests were taken. tained by subjecting a casting of intricate form to various heat treatments. For this reason the following data will give an idea of the results which can be really obtained in the practical application of homogeneity heat treatments. Steel used for these experiments was made in the acid open hearth and had the following composition : ^^'^""^ 0.26 per cent. Manganese 0.90 per cent. I'^f"^ 0.18 per cent. IfP^"" 0.02percent. Phosphorus 0.04|per:cent. Nickel 1 . 76 ^.g^t SOFT AND MEDIUM STEELS 241 Twenty-three pieces like that detailed in Fig. 120 were cast from the same heat. Pieces for physical tests were taken from the thicker part of the wall of the partly square partly cylindrical bore shown in section in the lower right hand part of the drawing. The whole casting had pre- viously been subjected to a given heat treatment. For information and as a datum, I might add that a bar taken from an ingot of the same heat, after being forged and reheated at about 850°C. gave the following results in tension when using a cylindrical test-bar 20 mm. in dia. and 200 mm. long between reference points: Tensile strength 99,500 lb. per square inch Elongation 17 per cent. The results of the physical tests made on castings subjected to various indicated heat treatments under conditions specified in Sec. 151 are gathered together in Table X. Table X. — Tests on Normalized Casting OP Mild Nickel Steel Treatment static tensile test Impact tensile test Resiliency No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Reduc- tion of area, per cent. Tensile strength, kg.-m. Elonga- tion, per cent. Kg.-m. cm2. (Charpy test- piece) Static bend- ing 1 Heated at SOO'C. for 4 hours fol- lowed by rapid cooling in air; completely cooled in 1 hour. 96,000 56,000 15.0 14.0 106 12 5.1 good 2 Heating and cool- ing as No. 1 ; then drawn 1 hour at 700°C., followed by cooling in 6 hours. 86,800 82,500 21.0 22.5 117 27 7.5 godd 3 Heated at 800°C. for 4 hours fol- lowed by a slow cooling in the furnace, requir- ing 18 hours. 92,200 52,000 28.5 39.8 145 35 8.22 good i Reheating and cooling as in No. 3 then drawn at 700°C. for 1 hovir followed by slow cooling (6 hours). 91,700 52,600 33.0 52.8 142 33 9.31 good 16 242 HEAT TREATMENT OF The numerical data contained in this table do not need explanation, and find easy interpretation in the body of considerations developed in the previous part of the book. However, it is useful to point out that these data, especially when compared with those of Sec. 152, clearly substantiate a phenomenon known to be related to the relative frequency of the centers of ferrite crystallization. Special steels of this type are characterized by a closely packed or a great frequency in centers of secondary crystallization. It is therefore a great deal less dangerous to cool special steels of this type through the transformation range slowly, after a given homogeneity heat treatment, than it is to do the same thing with an ordinary carbon steel. From this special point of view it is especially interesting to com- pare the differences between No. 2 and No. 3 of Table X with corre- sponding values contained in Tables VIII and IX. 154. It would be easy to record a great number of other examples similar to the ones already mentioned, and illustrate still better the re- sults which may be obtained on steel castings by means of rational nor- malizing. This, however, would trespass the limits within which this treatise must be kept. Those steel castings already mentioned in Sec. 136 would be especially interesting. In them the correct selection of the steel 's composition, the special precautions taken during the melting and pouring operations, and the rational application of homogeneity and final heat treatments, would procure articles possessing physical properties fully comparable to those of the best forged and treated steels. But an explanation in great detail is at present impossible owing to the fact that they have been brought to their present perfection in very recent times, and are still held as industrial secrets only partly protected by patents. Contrary to other similar cases, secrecy is not of illusory practical worth, because the value of the processes in question depends not only upon the rules for performing one or more isolated operations, but is strictly subordinated to the rational coordination of all the manu- facturing phases, starting from the selection of the raw materials used for making the steel and reaching to the final treatment of the fabricated piece. It does not seem necessary to suppress some data concerning typical industrial cases, even if we cannot enter into all the manufacturing de- tails. At any rate, it may be interesting to show the practical importance of our theoretical considerations by giving some examples of results selected from among the best which may be obtained by means of the rational application of homogeneity heat treatments to steel castings. Above all it should be interesting to show how castings if suitably manufactured and subjected to a rational heat treatment may be sub- stituted in many cases and with great advantage — especially as far as SOFT ANDZmEDIUM STEELS 243 rapidity of manufacture is concerned — for pieces of steel which formerly were necessarily forged and heat treated. Forged pieces naturally require very tedious and expensive machining operations in order to be brought to the final form, which in good casting practice is almost en- tirely obtained by the molding. Thus it has been thought interesting to select the examples just indicated from among those actual cases where special steel castings have been successfully substituted for forgings. 155. Even before 1915 a great deal of study was given to the fabrica- tion and heat treatment of high-strength special steel castings at the steel works of Gio. Ansaldo & Co. obtaining results so satisfactory as to permit castings being used for a great number of gun-carriage and gun- mount parts and of machine parts which had heretofore been fabricated exclusively by tedious machining from heat-treated forgings. The use of these special castings allowed that firm largely to increase their pro- duction, and the practice has therefore been greatly extended. Experi- ence with several thousand gun carriages under the severe and exacting use demanded during three years of the great war has fully confirmed the practical advantage and the reliability inherent in the substitution of heat-treated special steel castings for steel forgings. This substitution can fortunately be done in a great number of cases where the intensity and nature of the working stresses was such as to establish the conviction, held even by the Ansaldo Co. until a comparatively short time ago, that such substitution was practically impossible. It is from among these very cases that the following examples will be selected. The steels used in the Ansaldo works for the manufacture of such castings are made in the acid open-hearth, and the practice has been studied essentially with the end in view of obtaining perfect "deoxi- dation"of the metal, and the highest possible elimination of "emulsified" non-metallic inclusions which possess an oxidizing power upon the mass of the steel. The importance of the effects caused by such sonims has already been mentioned at the end of Part III (see Sec. 96 to 106) and in the previous chapter (Sec. 150). Other examples will be seen further on. Another characteristic of the steels in question, due in part to the specific action of the titanium and vanadium used in their manufacture, is the great "frequency" of the centers of alpha crystallization which form in austenite during allotropic transformation. The great practical im- portance of this fact has been seen from reasons advanced in the first two Parts. Numerous examples in Part IV illustrate how the potency and efficiency of the heat treatments are concerned and indicate the reliability and the uniformity of the effects produced upon such metal. The steels we are now concerned with belong to the group of the me- dium hard nickel steels. They are quite brittle when cast and tested 244 HEAT TREATMENT OF prior to anj^ heat treatment, in fact averaging the following physical properties -.^ Tensile strength 71,100 to 76,800 lb. per square inch Elastic limit ; . 54,000 to 61,200 lb. per square inch Elongation ' 3 to 5 per cent. Reduction of area 2 to 4 per cent. Charpy impact test (p) 1 to 2 kg.-m. per square centimeter A flexure test-piece breaks before even slight bending,' as is shown by Fig. 121, which also reproduces the appearance of the tension test and the fracture of the Charpy impact test. The coarsely crystalline struc- ture, typical of these steels as cast, appears still better upon the fractured surfaces obtained by bending untreated castings of medium or large dimensions in a press. Two typical examples of such structures are those reproduced in Fig. 122 and 123. The normal microstructure of these cast steels before any heat treat- ment is reproduced at an enlargement of 80 dia. in Fig. 124. Even simple reheating followed by slow cooling remarkably improves the properties of these steels. Thus annealing for about 3 hours at 850°C., imparts variable physical properties within the following limits, depend- ing upon the composition of the steel and the dimensions of the castings. Tensile strength 71,100 to 78,200 lb. per square inch Elastic Umit 54,000 to 61,200 lb. per square inch Elongation 15 to 22 per cent. Reduction of area 25 to 35 per cent. Impact tensile strength 80 to 110 kg.-m. Impact elongation 18 to 27 per cent. Transverse impact (Charpy test) (p). . 4 to 6 kg.-m. per square centimeter. A beam subjected to gradual loading usually breaks at an angle of about 90° to 110°, as shown in Fig. 125, which also reproduces the ap- pearance of the other fractures. The broken surfaces still show evident traces of the dendrites. Microstructure of the annealed steel is generally like that reproduced in Fig. 126 (enlargement 85 dia). It is easy to understand the origin of this internal structure when examining it in the light of all the considerations developed in the previous chapters. Mild homogeneity heat treatments such as those studied in the previ- ous chapters modify the structure which we have often noted as that normal for hypoeutectoid steels of medium carbon content. For instance, annealing followed by rapid cooling in air causes a microstructure of the type reproduced in Fig. 127 at an enlargement of 1 These numerical data, as well as all those which shall be presented for the special castings with which we are now concerned, result from tests made under the conditions specified in Sec. 151. SOFT AND MEDIUM STEELS ^KK^ . • ''^'"**''^^'**''«'-*^«».««»^ r , '^'1 ■" *«< .. > , .^ i» , ■:*■- Wt H ■mi rr ' ' .. » Fig. 121. — Test bars from untreated special nickel steel eastings. Fig. 122. — Fracture of untreated special nickel steel casting. 246 HEAT TREATMENT OF Fig. 123. — Fracture of untreated nickel steel casting. Fio. 124. — Microstructure of untreated nickel steel casting. X 80. SOFT AND MEDIUM STEELS 247 Fig. 125. — Test bars from annealed special nickel steel casting. FiQ. 126. — Microstructure of annealed special nickel steel casting. X 85. 248 HEAT TREATMENT OF 85 dia. It is superfluous to add detailed explanations of this con- figuration after what has ah'eady been said. Fig. 127. — Special nickel X 85. Fig. 128.— Metal of Fig. 127 after dra%Yi X 8.5, A drawing operation following such an air quenching will give rise to the structure reproduced in Fig. 128 at 85 dia. The slight ferrite segregation is entirely normal. As a comparison, Fig. 129 at the same magnification reproduces the SOFT AND MEDIUM STEELS 249 structure of the same steel as the one shown in Fig. 127, when quenched in air after forging. Evidenth' the difference between the structures appearing under the microscope is not remarkable. 156. In Parts I ajid II we have already noted the reasons why a simple homogeneity heat treatment consisting of a single annealing or quenching even if followed by drawing does not give results so good as those which can be obtained by more complex treatments. Without going over arguments alreadj' discussed at sufficient length, I will only recall that this depends upon the fact that the conditions under which the various phases of a complex treatment take place maj^ be so selected that each of them maj^ develop one of the desired characters to a maxi- mum degree, and up to a certain point without introducing into the Fig. 129. — Special nickel steel casting after forging, reheating and air quenching. X 85. resulting steel article other dangerous defects which otherwise could eventually accompany the development of that desired property. In general such a criterion may be applied only under condition that the successive phases of a complex treatment may be conducted in such a way as to eliminate, or at least sufficiently attenuate, such dangerous effects as may take place in the previous phase. We have seen numerous examples. Here it is sufficient to remember that even though a preliminary anneahng or quenching attended by a very prolonged reheating at a high temperature may give special steels of the type now under consideration a very remarkable homogeneity of chemical composition in the austenitic state, yet on the other hand, such practice favors grain growth. Then if the metal has to undergo a 250 HEAT TREATMENT OF slow cooling in the successive phases of the treatment, it may easily show Widmanstattian structure more or less well developed. This case is only another example added to those we have already seen, that the advantages of great uniformity in concentration obtained by means of very energetic homogeneity annealings or quenchings, cannot be con- veniently exploited in industry, except under condition that further phases of the treatment may eliminate the serious injury which always accompanies the above mentioned advantages in larger or smaller measure. It has already been pointed out that in practice the steel may be re- stored by means of hot work and also by means of certain heat treatments, consisting essentially of quenchings from relatively low temperatures — such as from within the transformation interval — properly conducted in such a way as to graduate their intensity. In these very cases mild quenching as in air, lead, hot oil, boiling water, etc.,^ present the highest practical interest. They permit one to adjust the quenching effect at any desired intensity in the series starting from the energetic real quenchings, and ending with quenchings so prolonged as to be comparable to actual annealings. It has already been noted that the special steels lend themselves most favorably to the application of these processes. A correct selection of chemical composition permits a large variation in several different properties, the normal location of y^a transformation temperatures, the metastable lowering of the limits of the extreme tem- peratures of said interval, and, finally, the amount of thermal hysteresis. Very numerous combinations can evidently be obtained by applying the criteria which have been studied to determine the procedure for each step of a complex heat treatment. 157. The physical properties which may be obtained by applying the more complete heat treatments to the cast special steels indicated in Sec. 155, range within the following limits, according to the composition of the metal and to the dimensions of the castings treated. Compare the values given in Sec. 155 for the raw casting and after annealing. Tensile strength 92,500 to 120,900 lb. per square inch Elastic limit 56,900 to 96,700 lb. per square inch Elongation 15 to 30 per cent. Reduction of area 40 to 65 per cent. Resistance to impact (Charpy test).. 9 to 15 kg.-m. per square centimeter Impact tensile strength 115 to 145 kg.-m. Impact elongation 25 to 35 per cent. Bending tests under gradual load performed under the conditions indicated in Sec. 151 always reach the maximum deformation into a "U" without showing any crack. Selection of steels possessing strong thermal hysteresis, relatively low transformation points and closely packed centers of ferrite crystalli- ' See Sec. 75, 76, and the first paragraphs of Part III. SOFT AND MEDIUM STEELS 251 zation obviates the necessity of sudden quenchings to preserve the homo- geneity obtained at the high temperatures reached in one or more of the phases of the heat treatment. The main practical consequence of this fact is that the complex heat treatments may be efficaciously apphed to castings of large dimensions and intricate form as well as to simple thin pieces, imparting uniform physical properties to all their parts even those of varying thickness. Moreover, the application of the desired heat treatments is safer because the ehmination of quenching avoids the formation of chance quenching craclcs. In the great majority Fig. 130. — Test-pieces from hcat-trciitcd special nickel steel casting. of cases the cooling velocities obtainable with simple air chilhng is sufficient. Figure 1.30 shows the appearance of the various test-pieces taken from a casting which gave the physical properties tabulated below. The bar broken bj^ static tension is shown at the right, the Charpy impact test in the center below, impact tension at the left, and gradual bending in the top center. Tensile strength 96,700 lb. per square inch Elastic limit 74,000 lis. per square inch Elongation 28 per cent. Reduction of area 49 per cent. Resistance to impact (Charpy test) 14.8 kg.-m. per square centimeter Impact tensile strength 122 kg.-m. Impact elongation 31 per cent. 252 HEAT TREATMENT OF Fig. 131. — Broken casting from which test-pieces of Fig. 130 were taken. imm wsm m^ tM ft!M W^ ^K'^ P"";. u f &' ' m k ^^AgA ^ i !■ ■ A HeI^^ L*. A § X ^M Ik; ^'' * ^^¥> '^ k Fig. 132.— Bending test-piece of Fig. 130 doubled flat to rupture. Fig. 133. — Localized fracture of Fig. 132. SOFT AND MEDIUM STEELS 253 All surfaces of fracture are shown to have a finely fibrous structure. The same casting when broken bj^ bending under a press, presents uniform fibrous fractures, from which every trace of crystaUine structure has totally disappeared. For instance. Fig. 131 reproduces such a frac- ture of the casting from which the test-pieces shown in the previous figure were taken. It clearly shows the characteristic dull lack-luster of fibrous fractures to be expected in the broken surfaces of tough metals. The test-piece bent like a U in Fig. 130 did not break until it was almost doubled upon itself, as shown in Fig. 132. Even then the fracture was perfectly localized, as appears better in the next Fig. 133, showing another view of the same test-piece. 158. Even in castings of harder steel the property of withstanding great deformations before breaking accompanies high tenacity to as great Fig. 134. — Cast, heat-treated yoke of hard nickel steel. degree as appears from the physical properties noted in the previous sec- tion. I would like to emphasize the fact that the castings which I shall now describe were all of high tensile strength steel, in which the tenacity was always above 108,000 lb. per square inch and the elastic limit above 70,000 lb. per square inch. This remark is of the utmost importance in order to appraise the true import of the deformations to be men- tioned shortly, deformations which could be easily obtained even after simple anneaHng with mild steel castings whose elastic limit would not be far from 42,000 to 50,000 lb. per square inch, but which represent ex- ceptional results for hard high-strength steels possessing physical proper- ties comparable only to those which can be obtained with forged and heat-treated steels. A first example is furnished by the piece shown in Fig. 134. The necessary load to produce such deformation as shown in Fig. 135 was 16)^ tons, applied by hydraulic press at the ends of the casting, held by a 254 HEAT TREATMENT OF single central support. After loading and bending not the least crack was found in the piece. On increasing the pressure the piece broke partly through in the man- ner shown in Fig. 136. Coinplete fracture was obtained only when the ends were pressed together in actual contact. Figure 137 shows the sur- face of fracture, which was of a finely fibrous, uniform structure, charac- teristic of the very toughest steels. Another cast trunnion made of high strength heat-treated steel is shown in Fig. 138. It was deformed by pressing against the two ends so as to squeeze them together in the way indicated in the next Fig. 139. During this test the high external ribs shown clearly in Fig. 140 (the same casting from the bottom) did not develop the least crack although they were badly stretched. This fact, as well as the existence of heavy stretch- FiG. 135. — Yoke after supporting 16.5 tons at the ends. ing in the ribs, is clearly seen in Fig. 141, viewed from the same point as the previous figure. The sector illustrated in Fig. 142, also made of the same kind of steel, was placed in a press so that the two points of the arch were squeezed together until it was deformed as shown in Fig. 143. The metal's exceptional toughness in spite of its hardness is demonstrated especially by the absence of fractures even at places where large differences of thick- ness occurs, and where the lines of deformation accumulated, due to well known causes. Finally, another large trunnion of the same kind of steel and manu- factured in the same way is illustrated in Fig. 144 as it appears before cutting the two risers. It was also loaded by a hydraulic press so as to push the ends toward one another. Once having reached the shape SOFT AND MEDIUM STEELS 255 f ■■ '1 fl J |Hb ^HRfflS^H^^U^^^^k '■' ^^^ . -,-^ Fig. 136. — Yoke bent in press to failure. ;.i? Fig. 137. — Texture of fractured yoke. 256 HEAT TREATMENT OF Fig. 138. — Cast trunnion of heat-treated hard nickel steel. FiQ. 139. — Trunnion of Fig. 138 bent together. SOFT AND MEDIUM STEELS 257 shown in Fig. 145, the two risers were sawed off on a hacksaw and the trunnion again placed under the press, as shown in Fig. 146. Locahzing the pressure upon the trunnions by means of a block of steel, placed between the hammer and the upper trunnion as shown in the figure, the bending was continued to the point shown without causing the smallest crack in the piece. n Fig. 140. — C'a.st trunnion from bottom; before bending. It may be superfluous to point out explicitly that all these flexure experiments were made upon completely cold castings, at normal tem- peratures between 10° and 20°C. In closing these observations regarding castings of heat-treated speciaj steel, it may be mentioned that even during machining such castings possess toughness and plasticity equal to and even exceeding that of very mild steel castings or those of the best forged and heat-treated steels. To illustrate this fact, I reproduce in Fig. 147 at about }e natural size the appearance of a turning cut from one of the hard castings in question. The figure needs no comment. Finally, in order to show that properly heat-treated cast special steels 258 HEAT TREATMENT OF act in a very similar manner to forged or rolled steels when subjected to cold work, I reproduce in Fig. 148 the appearance of two test-pieces taken from two castings employed in the tests mentioned above. The two pieces indicated with the letters A and B, gave the following results: A B Tensile strength 123,700 lb. per square inch 101,000 lb. per square inch Elastic limit 96,700 lb. per square inch 84,000 lb. per square inch Elongation 21 per cent. 22 per cent. Reduction of area 39 per cent. 50 per cent. r:^' ^ ' ■ %'■: \ 1 '^^ . mi J %■■■ i \^^^ ^p - ■ w^, ; \ '^fe,*SA; fm$^ -"rn^^.Tt^ ■^9 wm «*£* / .■.'^lA Fjg. 141. — Cast trunnion from bottom, after bending. The cylindrical part of the two test-pieces had been polished before pulling, and in their final shape show perfectly that the cold work revealed by the dull surface luster is locahzed in a relatively small part of the test- piece within the necked portion. It is well known that this appearance characterizes exactly those metals which have been subjected to some treatment after solidification (similar to hot-work) which has eliminated, at least in good part, the great sensitivity to cold-work characteristic of SOFT AND MEDIUM STEELS 259 FiQ. 142. — Cast sector of heat-treated hard nickel steel. Fig. 143. — Sector after severe local deformation. 260 HEAT TREATMENT OF Fig. 144. — Large trunnion, a heat-treated hard nickel steel casting Fig. 145. — Trunnion after bending. SOFT AND MEDIUM STEELS 261 Fig. 14G. — Trunnion under press; closed completely without fracture in ribs 262 HEAT TREATMENT OF Fig. 147. — Turning from heat-treated hard nickel steel casting. Fig. 148. — Test bars from castings illustrated in Fig. 134 to 146. SOFT AND MEDIUM STEELS 263 raw cast steels. Due to phenouicua which we cannot here dwell upon, it is already established that this sensitivity is responsible for the fact that when pulling test-bars made of conunon cast steels, even if annealed, the cylindrical test-pieces are stretched over their entire length, develop- ing ill-defined Ltider's figures throughout. 159. In the second as well as in the third and in the present part of this treatise, I have many times mentioned the fact that the phenomenon of thermal hj^steresis was a very important criterion for the determination of the most suitable processes for preliminary heat treatment, and have also indicated the manner of using such a phenomenon in order to obtain Fio. 149. — Expansion vs. temperature curves for annealed mild steel. the best results possible with a given steel by means of said treatment. Therefore it is opportune now to add some actual data regarding the values which thermal hysteresis may assume for mild carbon steels and medium hard 2-per cent, nickel steels. These steels comprise the two principal types whose thermal reactions we have carefully studied as determining the details of the preliminary heat treatment. Thermal hysteresis may be defined as the variation in the observed position of the transformation interval when determined upon a cooling specimen as compared to the same interval determined upon heating. Its value can be fixed experimentally by noting the rate of heat evolution 2G4 HEAT TREATMENT OF or absorption during heating or cooling at a uniform rate, particularly using the differential method devised by Le Chatelier and Saladin, or it can be found by noting the variations in the coefficient of expansion as the temperature passes the transformation interval. For the two examples to be given I have selected the second method. Figure 149 reproduces two curves — expansion vs. temperature above and contraction vs. temperature below — characteristic of annealed mild carbon steel. The two curves are reproduced in the figure exactlj^ as they were automatically traced upon the photographic plate in the apparatus ..„,^.KJ_i— -4-— !—' „^,..„^ ., : ■' • ■ ■ *' ■, . .. . ■, . ^ '. 1 i y K- i ■ ' ■ ■ 1 ■■. ,<>• 5oo* 6oo' 700* 8oo* 900" looo'iioo" Fig. 1.50. — Expansion vs. temperature curves for annealed medium hard 2-iDer cent nickel steel. designed by Le Chatelier and Broniewski. It would be superfluous to dwell upon the details of the tracing mechanism and the meaning of their various branches, because all this is comprised in elementary metallography. A comparison of these curves with those reproduced in the next Fig. 1.50, traced under identical conditions by an annealed medium hard 2- per cent, nickel steel, clearly shows how the hysteresis interval and its location differs for the two steels. From considerations repeatedly developed in the previous chapters, such comparison also shows the SOFT AND MEDIUM STEELS 265 greater opportunity of selecting suitable conditions for preliminary heat treatment offered by the 2-per cent, nickel steel. In order not to repeat things already said on this subject I will only observe that, while for the two steels the transformation upon heating ends at about the same temperature (between about 830°-850''C.) the transformation upon cooling requires the interval from about 740° to 620°C. for the carbon steel, yet the same change in the nickel steel is starting at a temperature somewhat below 700°C. and is not entirely ended even below 500°C. In the light of all the previous remarks it is evident that the temperature interval available for genuine preliminary quench- ings extends to temperatures far lower for this nickel steel than it does for the carbon steel, thus rendering these quenchings fully effective and practically applicable in many cases in which they cannot be made with- out great inconveniences upon carbon steel castings. The interval of temperature within which the so-called incomplete or negative quench- ing can be successfully noticed is also evidently the larger for such special steels. It is necessary to remember that the dimensions and the positions of the transformation intervals, as they are located by an expansion-tem- perature curve similiar to the ones here produced, do not have an absolute value, but, owing to the known phenomenon "of lag which we have already noted several times, depend upon the velocity with which the heating and the cooUng of the steel has taken place during the experiment. The study, of the changes which the dilation curves suffer with each variation in speed of heating or cooling is an important branch of the study of heat treatment of steel. CHAPTER XVIII EFFECT OF SOLID NON-METALLIC INCLUSIONS UPON PRELIMINARY HEAT TREATMENT 160. It has been my purpose to mention a series of actual examples illustrating the effects of homogeneity heat treatments on cast steels. To complete these illustrations it is useful to add another example clearly showing the magnitude of the perturbations in the course and effects of normahzing produced by the presence of certain types of non- metallic inclusions. True, in Sec. 150 I have ah'eady had occasion to Fig. 1.51. — Ranked inclusions in ingot of 0.49-carhon steel. X 100. mention an instance where the effects of heat treatment suffered un- doubted alterations from the presence of emulsified inclusions in the steel; but they were far from reaching values comparable to those which are very often reached in practice, even where it is certainly not possible to ascribe the trouble to faulty manufacture. Therefore one of these latter cases will be mentioned at the present moment, showing the effects of emulsified inclusions when disposed in ranks, ^ and the improvement still possible to obtain by means of homo- geneity heat treatment. Figure 151 reproduces at an enlargement of 100 dia. the structure ' See .Sec. 98. 266 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 267 of a medium hard carbon steel made by the acid open-hearth process, and of the following analysis: Carbon 0.49 per cent. Silicon 0. 28 per cent. Manganese 0. 63 per cent. Sulphur . 02 per cent. Phosphorus 0.06 per cent. The structure shown in the figure is that of the raw steel as cast in ingots weighing about 2 tons each. It clearly shows the character- istic disposition of "ranked inclusions," located in the central part of the ferrite shells (appearing as rough-bounded strips in a plane section). Some physical tests upon the steel in this state, made under the usual conditions indicated in detail in Sec. 151, and duplicated upon the same steel after it had been subjected to various heat treatments, gave the results indicated in the following table. Beside the results of each series of tests, the table indicates the preliminary heat treatment. Heat treatment Physical properties No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Resiliency (p) (Charpy te8t)kg.-m. per sq. cm. Micro- structure 1 Cast steel 94,600 101,700 99,600 122,300 98,400 could not be deter- mined 48,400 51,200 83,100 72,000 5.5 17.0 12.0 1.40 18.0 Zero 1.1 1.0 1.7 2.53 Fig. 151 Fig. 152 Fig. 153 Fig. 154 Fig. 155 2 3 4 5 Reheated at 800°C. during 4 hours followed by slow cooling in the furnace. Reheated at 900°C. during 4 hours followed by slow cooling in the furnace. Reheated at 900°C. during 10 minutes followed by quenching in water at 20°C. and by drawing at 600°C. during 4 hours. Reheated at 1100°C. during 4 hours followed by slow cooling in the furnace till 900°C. and then by sudden quenching in water at 20°C. Finally annealing at 700°C. during 4 hours. 268 HEAT TREATMENT OF Figures in tiie last column denote the illustration reproducing the structures assumed by the steel after each of the heat treatments indicated in the table (all enlarged 100 dia. after polishing and etching with an Fig. 152. — Steel of Fig. 1.51 after annealing at SOO°C. X 100. Fig. 15.3. — Steel of Fig. 151 after annealing at 900°C. X 100. alcoholic solution of picric acid) . It is perhaps superfluous to remark that the data contained in the table refer to the effects produced by certain- heat treatments executed upon the steel simply cast. SOFT AND MEDIUM STEELS 269 161. It is possible to make some interesting remarks about the data contained in this last table, which constitute a proof and an illustration of the propositions briefly developed at the end of Part III regarding Fig. 154. — Steel of Fig. 1.31 after quenching from 900°C. and drawing. X inO. Fig. 155. — Steel of Fig. 151 after complex heat treatment. X 100. the effect which the presence of the emulsified and strongly oxidized soUd non-metallic inclusions may produce upon the results of homo- geneity heat treatments. 270 HEAT TREATMENT OF Thus, if we examine the data, showing- the effects of the first treatment (No. 2 of the table, and Fig. 152) one sees that the metal simply reheated at 800°C. is still very brittle, although its static tensile strength is very much improved in comparison to that of the untreated steel. In ad- dition it is proper to observe that in this case as well as in the other example mentioned in Sec. 150, the reheating has produced an increase in the tensile strength, another instance that the fragility of steel before heating is to be considered as abnormal, and an indication of ingotism. Examination of Fig. 152, the raicrostructure of the steel after reheating for 4 hours at SOO^'C, shows the fact that the continuous ferrite shells developed in the unannealed steel along the ranks of inclusions are much diminished in dimension, being reduced to a few fragments gathered in the regions where the ranked inclusions are more numerous, while the remain- der of the ferrite appears in the form of crj^stals independent of the large network. In other words, the ferrite structure has changed from the form of "large continuous network" to that which I have called "large network with interrupted meshes." There has already been indicated in Sec. 98 the mechanism of the phenomena by which the ferrite tends to accumulate around strongly oxidized emulsified inclusions. Recalling the observations there made, it is not difficult to find an explanation of the above observed fact that an annealing prevents a full development of these continuous shells. In fact, it is sufficient to remember that diffusion always takes place during a reheating at a temperature above Acs. Such diffusion of the various elements then contained in solid solution in the steel we are consider- ing, cause their migration toward the regions occupied by the inclu- sions. Especially is it true of those elements whose chemical action is of a reducing nature, among which carbon predominates both be- cause of its own high reducing properties and of its greater velocity of diffusion, due to the action of its gaseous compounds. During solidifi- cation of the ingot the oxidizing action of the inclusions had locally decreased the concentration of various constituents by oxidizing or neutrahzing them. It is clear that a new influx of reducing and car- burizing bodies toward the inclusions will gradually decrease the oxi- dizing action of the latter and, therefore, at the same time will reduce the intensity of the action which causes the accumulation of ferrite around the inclusions during the y-^a transformation. We have already seen at the beginning of Sec. 106 an example of this process. As has been briefly outlined in Sec. 98, the oxidizing action which the inclusions may exercise upon the surrounding metal is due not only to the state of oxidation of this inclusion but also depends upon the ratio which is estabhshed between the characteristic oxidation tensions of the non-metalhc particles and the neighl^oring metal respectively, SOFT AND MEDIUM STEELS 271 in the Various steps during cooling through the transformation interval. Again, inasmuch as this last ratio depends i;pon a great many circum- stances which are very difficult to define (among which the velocity of the cooling certainly predominates in influence) it is easily explainable how the useful effects of a long heating at a high temperature may in part be impaired or counteracted by a succeeding too-slow cooling. An example of this last fact may be found in the effects of the second heat treatment indicated in the previous paragraph (line No. 3 of the table). In Fig. 153 after annealing at 900°C., one may in fact see a greater development of the elements of the large network of ferrite than in the steel annealed at SOCC. (Fig. 152). Physical properties are also inferior to those obtained by annealing at 800°C., poorer results which correspond to the microstructure of Fig. 153. The effects of the third treatment (see No. 4 of table in Sec. 160) demonstrate that the more pronounced development of the large network observed in case No. 3 is due, as has been said, to the reformation of said network during the slow cooling of the steel, and not to the fact (which, after all, would be unexplainable) that the heating at 900°C. had altered the ingot structure less deeply than a heating at 800°C. In the third heat treatment, in fact, 10 minutes at 900°C. were sufficient to disperse the greatest part of the accumulation of ferrite existing in the large net- work; this could not gather together again during cooling, since quenching took place at a velocity too great to develop the processes which formerly occurred during slow cooling. We must note in passing that the steel after quenching in water was subjected to a strong drawing. The phys- ical properties are far superior to those of No. 3, and thus confirm our conclusion. Another proof of the conclusions reached in this section is given by the results of the fourth treatment, line No. 5 of the table. Here the homo- geneity heat treatment was made under the best conditions, joining the advantages of an energetic annealing to those of a quenching, yet without faUing into the known imperfections due to fi quenching from too high a temperature. The structure of the steel thus treated is uniform, as shown in Fig. 155. Traces only of the large network of ferriteare re- tained, and the physical properties of the steel taken as a whole are better than those obtained by the previous treatments. However, the effects of the abundant inclusions are still felt, especially in the low value of resiliency (p) and, probably, are due to the persistency of hetero- geneities in elements other than carbon. Thus we have another example of the fact pointed out in Sec. 145 that an estimate of a steel's uniformity may be very erroneous if based exclusively upon a microscopic examina- tion of the distribution of carbon. 162. A comparison of the structural and physical characters of the same steel as modified by various annealings at 800°C., still better 272 HEAT TREATMENT OF demonstrates the fact that it is relatively insensitive to the effects of homogeneit}^ heat treatments. The following table contains all the necessary data for such comparison. Heat treatniont ]^hysical properties Tensile strength, lb. per sq. in. Annealed at SOO°C. during; 99,000 40 minutes followed by slow cooling;. Klastic limit, lb. per sq. in. Annealed at 800°C. during 97,400 48,400 5 minutes followed by slow ] cooling. Annealed at 800'C. during | 99,300 39,800 20 minutes followed by slow cooling. 37,000 Annealed at S00°C. during 92,700 48,400 1 hour followed bj' slow | cooling. I Resiliency Elongation' ,/-.u percent I 'Chappy per ceni. j^^^q kg.-m. per sq. cm. 12 14 12 10 1.1 1.1 0.9 1.2 Micro- structure X 100 Fig. 156 Fig. 157 Fig. 158 Fig. 159 From an inspection of these results, the effects produced by the four heatings differ very little one from the other, and the differences are irregular and do not seem to depend in any way upon the time of anneal- ing. This is true for the variations of the physical properties as well as for the microstructure of the steel. Regarding the irregularities in structural characters, these variations are particularly accentuated; thus, it is sufficient to note that in the steel heated during 1 hour at 800°C. (Fig. 159) the residue of the large network of ferrite is still more strongly developed than in the same steel heated at the same temperature but for only 40 minutes (Fig. 158). It is clear that these facts constitute a further proof of the statements made in the previous section, especially the one which pertains to the reconstruction of the large network of ferrite during slow cooling in steels strongly oxidized during the melting process and containing emul- sified inclusions. 163. Summing up the conclusions which it has been possible to reach in Parts III and IV regarding the presence of oxidized emulsified inclu- sions, and more especially when arranged along certain surfaces, I deem it opportune to insist upon the following points : SOFT AND MEDIUM STEELS 273 1. The presence of solid non-metallic inclusions of the type indicated in the last sections of Chap. XIII constitutes a severe obstacle to the regular course of those processes whose uninterrupted development is Fig. 156. — 0,4'J-curbon steel annealed 5 min. at S(IU°C'. X 100. Fig. 1.57.— 0.40-carbon steel annealed 20 min. at SOO°C. X 100. necessary to produce all the advantages which can be obtained by means of homogeneity heat treatments. 2. From the viewpoint of the graphic representation of the same proc- 18 274 HEAT TREATMENT OF esses, as studied in the Part I and II, the effects of these sonims consist essentially in producing a relative stability and immobility in the bound- aries of the band of variant concentrations existing in the y solid solution. Fig. 158.— 0.49-carbon steel annealed 40 niin. at S0O°C. X lOU. Fui. 1.5'.».— U.4!»-carl:»)n steel annealed 1 hr. at ,SU(I°C. X 10(1. We have already seen particularly at the end of Sec. 63 the consequences of this last fact. Equally important effects are due to the fact that the presence of oxidized emulsified inclusions and of other impurities of similar SOFT AND MEDIUM STEELS 275 nature and function establishes a strict bond between the network related to the primary crystalline edifice and that of the secondary structure. ' 3. From the point of view of the structural characteristics of steel, the last mentioned fact is responsible for a persistent retention of the elements of the large network of ferrite, and, therefore, of ingotism. From all the facts studied it may be concluded that unforged steels containing many oxidized sonims in ranks possess a large network of ferrite whose form and position is the result of two orders of facts; the first set of circumstances is related to the relative "distribution" of the concentrations of the various elements in the steel at the start of the 7— >a transformation and during the change, while the second is related to the variations in the equilibrium of oxidation and reduction reactions which occur between the inclusions and the metal surrounding them dur- ing the same temperature interval. 4. From the technical standpoint, the presence of large quantities of oxidized sonims, especially if they arrange themselves in ranks, has a tendency to widen the known differences between the effects produced upon a given steel by a simple preliminary annealing and a veritable pre- liminary quenching, even when the two treatments follow the same re- heating practice. It is perhaps unnecessary to remark that this is due to the fact that such inclusions facilitate the re-appearance of the elements of the large ferrite network, previously eliminated more or less completely by homogeneity annealings. 164. In the next and successive chapters we shall see how these four principal effects produced by oxidized sonims upon the properties which a given homogeneity treatment imparts to a given steel, may assume special characters and high importance should the steel have been sub- jected to the heat treatment after undergoing hot working such as forg- ing, rolling, stamping, etc. 165. Before closing this chapter, it is timely again to call attention to an observation of the greatest practical importance, which is evident when all the experimental data presented in the last chapters are compared. A great difference may thus be found in the physical and structural properties of the various steels; even the modifications which such prop- erties suffer by reason of certain homogeneity heat treatments do not appear to be consistent even in the cases in which the chemical composi- tions of the steels compared and the conditions under which the respective heat treatments have been made differ very little one from the other. Even though the experimental results which we have had occasion to mention are quite limited in number, the above remark holds especially when comparison is extended to steels containing emulsified sonims more or less strongly oxidized (see, for instance, the data referred to in Sec. 150, 161 and 162). Should larger and more complete groups of analo- 276 HEAT TREATMENT OF SOFT AND MEDIUM STEELS gous experimental data be examined (which cannot be done here) so as to extend the comparisons to an exhaustive series of experiments in which the chemical compositions and the conditions under which the heat treatment has been made would vary in a gradual and progressive way, very strong irregularities in the corresponding variations of physical and structural properties would be easily revealed in the steels so treated. All these facts, whose practical importance is evident, confirm and generalize the conclusion already reached in a precise way concerning some particular phenomena, such as the state of oxidation of steel, the nature of the sonims included in it, etc. I refer to the undoubted fact that the physical and structural properties of a given steel and the variations which said properties suffer by effect of predetermined heat treatments, may depend in very large measure upon many causes, which cannot always be easily determined. Among these causes, beside the others already mentioned, may be indicated as examples, the kind of raw materials used in the steel making, the refining and deoxidation of steel — whose effects sometimes are felt in a way inexpUcably remarkable and persistent — the temperature and the other conditions during teeming, the form and dimensions of the ingots and castings, etc. We shall see further on, how other irregularities and apparent ano- malies accrue in steels subjected to hot working before heat-treatment. PART V PRELIMINARY HEAT TREATMENT OF FORGED AND ROLLED STEELS CHAPTER XIX A GENERAL DISCUSSION OF HOW THE EFFECTS OF HEAT TREATMENT ARE MODIFIED BY HOT WORK 166. The paucity and inaccuracy of the knowledge even now possessed by many mechanical constructors regarding the properties which may be imparted to cast steels by means of a rational homogeneity heat-treat- ment, causes them to continue to specify forged or rolled steels for the construction of machine parts which in a great many cases could be manufactured with the same and even greater safety and sometimes with a great saving in time and cost from steels of correct composition, simply cast in molds and then subjected to a properly selected heat treatment. On the other hand, even when using forged or rolled steels, it very seldom happens that the mechanical constructor is able to avail himself fully of the advantages of homogeneity which the application of a rational preliminary heat-treatment would allow him to obtain. Consequently even in many modern machines are found parts made of forged or rolled steel blocks, and a great deal heavier than necessary. This is due only to the fact that the designer has taken into account exclusively the prop- erties of the metal resulting from a simple hot-working process such as forging, pressing, or rolling, and completed at best merely with a simple final heat treatment, intended to impart a higher tensile strength than that produced by the mechanical work. Such a final treatment for strength may be a quenching, followed or not, as the case may be, by a drawing. Sometimes it is the practice simply to eliminate the internal stresses caused by mechanical work by a final drawing or annealing. On the contrary, when the mechanical constructor and designer is able to avail himself surely and at will of the advantages obtainable in what we know as preliminary heat treatments for homogeneity, modern steel metallurgy offers him many other resources. This statement holds even in the still frequent cases in which it is not yet possible or conven- ient to substitute steel castings for forged, rolled, or stamped articles. We have already examined the structural origins of these metallurgical advantages. The examples which I propose to show in this last part of our study will give a more precise idea of the practical value of correct 277 278 HEAT TREATMENT OF heat-treatment. It is not less important in case the steel has been previ- ously subjected to a hot mechanical working than in the other cases which we have already studied where the homogeneity heat-treatment is applied to raw cast steels. On the contrary the practical advantages in ques- tion are sometimes greater in the former than in the latter instance. Just to mention an illustration of this remark, many examples of which shall be seen further on, it may be said that the principal purpose of homogeneity heat treatment often consists in the elimination of danger- ous transverse fragility — a term I have defined as the sum of all the properties which make a specimen weak transversely — whose presence would wholly prevent the construction of pieces such as large caliber gun tubes. Disregarding for the time being, however, the always very great and sometimes essential but purely technical advantages of the rational ap- plication of the preliminary heat treatments to forged or rolled steels, such practice may enable one in many cases to effect remarkable economies and savings (with the same factor of safety) by allowing the use of steels costing a great deal less than those steels now customarily specified to be used after a "final" heat treatment, such as indicated above. Finally, many of the most important modern machines, as for in- stance, aeroplane motors, light internal combustion motors for boats, and so on, are made possible only by the rational use of special steels subjected to eflicient normalizing after forging or rolling. 167. Usually, the nature and general course of the phenomena respon- sible for the effects of preliminary heat treatments are not much changed owing to the fact that this normalizing is done upon steels previously subjected to hot-work rather than upon unworked cast steels. There- fore, as far as concerns the general application of the arguments devel- oped in Parts I and II, what I have said in Part IV remains unaltered as it refers to the examples which will be given in this and subsequent chapters. However, the same thing cannot be said when one considers the effects produced upon the physical, mechanical and structural properties by a given homogeneity heat-treatment when appUed to a steel of a given composition, regardless of whether it had been worked or not, because if its ingot structure had been thoroughly broken up, strong quantitative and qualitative differences will appear in the results which are then ob- tained when compared to those when starting with cast steels. A general study of the source and the nature of these differences can- not be presented at this place because it would require a detailed treatise covering many groups of crystallographic facts which have no direct connection with those already studied in the previous sections, and upon which normalizing processes are based. Modes by which such differences manifest themselves in the mechanical and physical properties of the SOFT AND MEDIUM STEELS 279 heat treated steels will become sufficiently and clearly explained by the concrete examples which will be shown in the following sections. Therefore, in advance of these facts to be developed, I shall only mention some brief general observations, constituting simply an orderly enumeration of the principal groups of phenomena to which can be at- tributed the differences between the physical, mechanical and structural properties which a given preliminary heat treatment may produce in a forged or rolled steel and the corresponding properties which the same treatment produces when applied to cast steel. We have already mentioned in Part III^ the general characters of some phenomena with which we shall now be concerned, but at that time we did not have the terms of comparison, now furnished by the experi- mental data recorded in Part IV, necessary to complete those remarks. Therefore, it is now opportune to rehearse the argument then merely suggested. 168. First of all, in comparing the effects produced by a given heat- treatment in the two cases it must not be forgotten that during hot-work the steel is continuously subjected to an important homogeneity heat- treatment whose efficacy may be very great. Especially is this effective when the working, which usually starts at temperatures at which the iron is totally in the y state, is so prolonged that during its last periods the steel has cooled till it reaches and partially crosses the transformation interval. In case the mechanical work is entirely completed while the iron yet remains hot enough to be held in the 7 modification, the effects which the resulting plastic^ deformations produce upon the course and upon the results of the subsequent normalizing may be considered as resulting from the following groups of phenomena, all of which usually operate simultaneously : 1. Decrease of the original distances between points in the cast metal corresponding respectively to the maximum and minimum concentra- tions of the various elements contained in solid solution. 2. Variations of the orientation of the primary crystallites and of their colonies, with consequent variations of the orientation of the secondary crystalline particles resulting from the y—^a transformation. These groups of phenomena exercise a remarkable direct action upon the course of the phenomena utiHzed for normalizing practice, as we shall see later by concrete examples. 3. Modifications of the crystalline edifice of the steel by the known phenomena of "slip" and "twinning," caused by plastic deformation on a metallic mass. 'See Sec. 94 and 95. 2 It is unnecessary to distinguish here between the characters of a plastic and an elastic deformation. 280 HEAT TREATMENT OF As is well known and has been already mentioned in Sec. 94, the phenomena of this group -deeply influence the mechanical properties of steel. But the modifications so caused are in great part independent of the modifications produced upon the same physical properties by the phenomena of diffusion. Homogeneity heat-treatments are based upon diffusion alone, so that at least in good measure the effects of the two orders merely add one to the other. It is true that in some definite cases which will be mentioned later, the effects of homogeneity heat treatments may even be manifested princi- pally as modifications in the crystalline structure precisely similar to those mentioned in 1, 2, and 3 as depending strictly upon the previous hot work. But there is no doubt that in such cases the additional crystal- line modifications taking place during normalizing are only a continu- ation of the same phenomena of slip and twinning in the crystalline elements, which had been initiated and, so to say, "prepared" by previous mechanical working, and then continued by the more or less prolonged heating at high temperature. At any rate, phenomena of category 3 have nothing to do with those of diffusion, which we have studied, and upon whose commercial utilization are founded the criteria for the execu- tion of preliminary heat-treatments. Therefore these phenomena can be considered only as constituting collateral factors which accompany the diffusion processes, operating in such a way that their effects are added to those of the latter even for that part which takes place during the heating operations. This statement naturally does not exclude the evident fact that the phenomena now indicated must be taken into careful consideration in practical heat treatment, inasmuch as the magnitude of their effects may sometimes exceed those of diffusion. But it may be repeated that these are always phenomena whose course is not practically related to diffusion. Therefore their study must be conducted separately and cannot be an integral part of the study of the processes upon which normalizing prac- tice is founded. A fourth group of phenomena has to be added to these three, as has been indicated above, when at least a part of the hot work takes place within the interval of temperature where the 7— >a transformation occurs. The characteristics of this fourth group may be defined briefly in the following way: 4. Modifications of the forms, dimensions and arrangement of the a and |3 iron crystals, due essentially to a "disturbed crystaUization" of these structural elements, a process caused by the deformations im- posed upon the metallic mass during their separation. Let us npw examine separately, even if in a summary way, the general characteristics of the phenomena pertaining to each of the above four groups, in order to complete what has been said in Part III, Sec. 94 SOFT AND MEDIUM STEELS 281 to 95 upon the same subject, but in the less precise form permitted by the scant ideas upon which we could then base our arguments. 169. In regard to the phenomena of the first group, there is nothing to add from a general point of view to what has already been said in Sec. 94. Shortening the distance between maximum and minimum concentrations will result in modifications of the processes there explained at length, modifications so simple that the explanation of their effects is so intuitive as to render unnecessary any preliminary consideration to illustrate the numerical examples which we shall have occasion to exhibit further on in this chapter. Such data alone will be enough to place in full view the soundness of the conclusion already reached in Sec. 94 concerning the greater efficacy of a given homogeneity heat-treatment when appUed to a forged or rolled steel than to the same steel as cast, provided always that in judging the effects of the heat-treatment proper, the actual homogeneity-annealing which necessarily accompanies hot- work is always taken into account. In this manner a true comparison may be established with the modifications which a given heat-treatment produces upon the properties of a steel not worked in any way before being subjected to a suitable preliminary heating. Besides, when weigh- ing the comparative data it is necessary to take into account the possible perturbations due to the intervention of phenomena of the other three groups, or to the intervention of other disturbing causes, such as, for instance, the presence of abundant and strongly oxidized sonims. It is perhaps timely again to point out — although the observation is obvious — that the efficacy of this first group in aiding and completing the diffusion processes is the greater, the larger the primary crystalline elements and their colonies which characterize the raw steel, both on account of the chemical composition of the steel and the conditions under which it has solidified. The effects of the phenomena of this first group, therefore, other conditions being equal except that the metal has not been subjected to any mechanical work, are especially felt with greatest intensity upon steel cast in large ingots. 170. The phenomena of the second group mentioned in Sec. 168 are a great deal more complex. These phenomena depend upon the varia- tions in orientation of the primary crystalline elements and of aggrega- tions or colonies formed by them, and upon subsequent modifications of the effects of such changed orientation which a treatment for homo- geneity may cause. We have already presented in Sec. 110 an example taken from a par- ticular ingot showing that concordant orientation of the principal axes of the primary dendrites may occur even in certain parts of a given mass of cast steel. For obvious reasons such a condition of affairs is responsible for more or less marked anisotropy in the properties of the metal. Here- 282 HEAT TREATMENT OF tofore I have not thought it timely to dwell upon the characteristics of this condition because in practice it does not present sufficient importance in steels simply cast in ingots or in molded castings. On the other hand, the analogous phenomena appearing in forged or rolled steels are of quite different weight. We shall see later that their effects often assume a by far preponderant importance. In the greatest majority of commercial cases, the deformations pro- duced by the mechanical work obviously result in a great increase in either one or two of the dimensions of the original billets or blanks, or at least of some portion of them. Constituent elements of the steel — primary crystals and their colonies or products of their transformation — suffer a deformation which may in some cases be considered as limited essen- tially to a "stretching" in only one direction. This happens, for in- stance, to an ingot during rolling or forging into billets or bars and is accompanied by a large increase in its length and by an equivalent re- duction of the other two dimensions. On the other hand, when rolling into slabs, plates or sheets or in the forging of flat pieces or plates, the same constituent elements of the steel undergo deformations which still consist mainly in "stretching" in a longitudinal direction, resulting in a considerable increase of the corresponding dimension, but is also ac- companied by a smaller increase in the transverse dimension. Thus, when rolled into bars, the structural elements assume the general appearance of fibers all arranged with their length in the direc- tion along which the elongation produced by hot-work has taken place. Structural elements in plates however possess the general "likeness" of elongated flakes all arranged with their greater axis parallel to the "main" or "longitudinal" direction and with their width parallel to the transverse direction of rolling or forging. Prom the viewpoint of morphology of the structural elements and their colonies, a main direction must be distinguished in a rolled rod parallel to the elongation produced by hot-work, and is called "longitu- dinal" to differentiate it from other directions normal thereto, any of which may be called "transverse." In the case of plates, however, two principal directions must be distinguished, the first of which is parallel to direction of maximum elongation and is called the "main longitudinal," while the second is parallel to the widening deformation and is called the "main transverse" direction. Such morphological anisotropy of structural elements in rolled or forged steels produces an anisotropy of mechanical properties, and because of this, various mechanical tests imposing stresses directed along the main direction differ remarkably from those measured transversely. This result is obvious when we think of the close connections shown to exist between the mechanical properties and the forms and dimensions of the primary crystalline elements, the distribution of the chemical compo- SOFT AND MEDIUM STEELS 283 nents in them, and such Uke characteristics. In hot-worked steel we are confronted with the same facts taken under consideration in former chapters when studying the effect of heat treatment on ingotism, characteristic of castings. In a very rudimentary and purely intuitive way, we may think that the longitudinal stresses have a tendency to displace the structural elements of the steel one from the other, causing intracrystalline and intercrystalline sUp in a direction parallel to that of their best-developed surfaces of contact; while transverse stresses have a tendency to separate said structural elements without slipping. In the latter instance, the simple breaking of contact in a small portion of one of those adhering surfaces causes a fissure normal to the direction of the stress. That is to say, it is so placed as to be apt to become a "breakage primer" splitting open the elementary surface on a continuance of the stress, and causing a rapid and very large decrease in the adhesion between the two structural elements considered. I would be digressing a great deal from the principal theme of my study if I were to pause at this place and show how the differences be- tween longitudinal and transverse mechanical properties of rolled and forged steels may be very plausibly explained by the theory of inter- crystalhne amorphous cement proposed by Beilby, and accepted by many metallographists. Neither is this the right place to develop an analysis, no matter how interesting or important, of the distinguishing characters of the two groups of properties. Usually these are briefly indicated with the two expressions, which shall be used hereafter, of "longitudinal mechanical properties" and "transverse mechanical properties." Besides, the examples which shall be mentioned later on, will be sufficient to give a quite precise and complete idea of the differences between the two. It is necessary to remember that the mechanical properties of rolled slabs, plates and sheets or flat forged pieces or plates determined in the main transverse direction are numerically intermediate between those in the longitudinal and in the direction of squeezing, that is, the transverse direction, properly speaking. The ensemble of poor transverse mechanical properties — especially in a steel in which these differ considerably from the longitudinal prop- erties — constitutes a net result of the greatest practical importance. In shop parlance "transverse weakness" (Italian " transverso ") is usually given to this character of forged or rolled steel. In a general way it may be stated that a transversely-weak steel possesses transverse mechanical properties which are always a great deal inferior to the longitudinal. The inferiority shows particularly in the figures for elongation, and still more remarkably in the reduction of area and brittleness. Usually pronounced transverse weakness is evidenced by a character- 284 HEAT T HE ATM EN r OF istic "woody" fracture of the tension test-pieces, a typical example of which is reproduced in Fig. 160. As we shall clearly see later on, the same homogeneity heat treatments which arc utilized in cast steels to eliminate ingotism may be employed to decrease or eliminate transverse weakness in forged or rolled steels. It appears that the ultimate crystallographic causes of ingotism and transverse weaknesses are the same. 171. Inasmuch as the previous chapters have shown that the arrange- ment and orientation of neighbor- ing colonies of secondary crystal- line elements resulting from the y^a transformation are strictly bound to the mutual arrangement and orientation of the primary crystalline elements, it is easy to expect that the phenomena which we have briefly examined in the previous section must produce an- alogous phenomena of congruent orientation in the secondary crystal- lites of /? and a ferrite and pearlite. Experience fully confirms these obvious expectations. Hence, it follows that hot distortion produces effects upon the orientation of neighlioring secondary crystalline elements similar to those observed when working has ceased before the piece has reached A?'3, not only in consequence of the phenomena which are collected in group 4 of Sec. 168, but also because of those gathered together in group 2. In all commercial steels the dif- ferences in concentration of the various elements resulting from the primary sohdification of the steel per- sist (at least in part) in the y crystalhtes at the start of the transformation. From the considerations developed in the previous chapters of Parts II and IV in this regard it is therefore easy to understand that the arrange- ment of the colonies of /3 and a iron shall copy the general forms assumed by the primary crystalline elements, even in steels subjected to mechanical work at temperatures above Ar;,. In fact, in steels forged or rolled into billets or bars at high tem- peratures and afterwards loft to cool slowly without further deformation Fig. 160. — Typical woody fracture. SOFT AND MEDIUM STEELS 285 through the transformation interval, we find crystalHne elements of /5 and a ferrite gathered in groups like filaments or closel}' piled bars ("columnar" structure). On the other hand the same elements of ferrite are gathered in groups like parallel plates ("lamellar" structure) in case the steel has been rolled into slabs, plates or sheets or forged in flat pieces or plates. Thus, Fig. 161 shows at 100 dia. the lamellar structure of rolled mild steel. 172. It is obvious that the phenomena which have been considered in the previous section will manifest greater intensity the stronger the heterogeneities persisting in the 7 mixed crystals immediately above Afi, and the slower the cooling thi-ough the transformation interval. Fia. 101. — Lamellar structure in rolled mild steel. X 100. Likewise obvious is the explanation that the foi'mation of columnar ferrite and lamellar ferrite maj^ also be prevented — or at least, slowed down and rendered less complete — by means of the same homogeneity annealings and quenchings which are utilized to limit or prevent the formation of the wide network of ferrite. Figure 162 shows at an enlargement of 100 dia. the microstructure of a rolled soft carbon steel, in which the lamellar structure, at first highly developed, has been suppressed by means of a homogeneity heat treatment. The rationale of heat treating hot-worked pieces is plain, keeping in sight what has been attempted to be shown in the previous sections; namely, that the relations of form and position between the network of ferrite and the primary crystalline elements in cast steel are perfectly 286 HEAT TREATMENT OF analogous to those between the elements of the columuar or lamellar system of ferrite and the primary crystalhtes deformed by hot working. Thus it results that a columnar or lamellar ferrite system must add its influence upon the mechanical properties of rolled or forged steel to that of the deformation of the primary crystalhne elements, in a manner analogous to cast steels, where the influence of the larger network of ferrite accrued to that due directly to the primary crystalline elements. There- fore, as in the case of east steels where a highly developed wide network of ferrite aggravated the phenomena of ingotism, a high development of a columnar system of ferrite in forged or rolled steels causes a cor- responding deterioration in quality due to the phenomena of transverse weakness. FiQ. 162. — Partial suppresaion of lamellar structure in Fig. 161 by heat treatment. X 100. Consequently, one may foresee that the application of homogeneity heat treatments to rolled or forged steels may be regulated bj' criteria entirely analogous to those regulating the normahzation of cast steels. While for these latter the main result consisted in the elimination of ingotism, the preliminary heat treatment of rolled or forged steels mainly ehminates the troubles inherited from the sum of all those characters which we have indicated with the name of transverse weakness. The examples which will be given later will prove the correctness of these expectations. Meanwhile, it may be pointed out that the foregoing remarks will permit me to avoid dwelhng upon arguments purporting to show how the interpretation of reported data may be clearly based upon what has been SOFT AND MEDIUM STEELS 287 said in the first two parts of this volume, inasmuch as these arguments would only be a repetition of those developed in Part IV, and which may be completely apphed to the new cases under discussion with the only exception that account must be taken of the observations made immedi- ately above. 173. We can clearly predict that the presence of emulsified sonims in forged or rolled steel, especially when in ranks and strongly oxidized, facilitates the development of high transverse weakness and of columnar and lamellar ferrite, and stubbornly counteracts the effects of homo- geneity heat treatment. This is evidently due to the same reasons by which we ascribed to the presence of the same inclusions in cast steels the more tenacious persistency of the phenomena of ingotism, the thick shells of ferrite, and the strong tendency for the same network to re-form during coohng after an anneaUng.' As shall clearly be shown by examples, this prediction is also fully confirmed by experience. Therefore it will not be necessary to repeat later any special discussion of the causes of persistent transverse weakness due to sonims, nor of the special development of lamellar or columnar structure due to the same inclusions. The practical importance of the effects produced in rolled or forged steels by oxidized emulsified inclusions is greater than in cast steiels. This is due mainly to the fact that in the former the effects directly or indirectly produced by the non-metallic inclusions (especially upon the develop- ment and persistency of transverse weakness), are added to those pro- duced directly by the phenomena accompanying hot-work, belonging to the four groups indicated in Sec. 168, often accentuating their conse- quences very strongly. We shall also see how it sometimes happens that the just mentioned characteristic effects of oxidized sonims appear with great intensity in forged or rolled steels, yet are caused by inclusions so small in dimension as to be difficultly detected under the microscope, especially if more or less deeply etched surfaces are examined. In some of these cases it is difficult to certify the exact cause for defective metal. Doubts in this regard are often resolved by microscopical examination under high magnification of a perfectly polished, unetched section. Under these conditions the field often reveals extraordinarily numerous inclusions, but of exceedingly small dimensions. 174. Phenomena of the third group mentioned in Sec. 168 have al- ready been discussed briefly in Parts I, III and IV (see for instance Sec. 82, 93, 94). It is not appropriate to study peculiarities of the phenomena in question here because, as has been pointed out many times, their effects are not felt in a considerable measure in modifying diffusion phenomena ■See especially Sections 98, 150, 160, 161, 162,;.! 63. 288 HEAT TREATMENT OF although having a great practical importance upon the mechanical prop- erties of steel. As a rule, therefore, crystaUine slip and twinning do not intervene in noteworthy measure in studying modifications in heat-treatment made necessary by previous hot-work. This detailed crystallographic study should form the subject matter of a most im- portant chapter in any complete treatise on the metallurgy and metallo- graphy of steel. It is to be noted with regret that in some important works covering these branches of science, sufficient space has not been given to the study of plastic deformation and its effects, space which is deserved by the great practical importance which such phenomena present in numerous commercial processes such as rolling, forging, press- ing, drop forging, hot and cold drawing, etc. However, it is proper to add that this serious gap has been filled in many of the most important and recent pubUcations, as those by H. M. Howe, Albert Sauveur, and G. Tammann. More than one hypothesis based upon wide and sound experimental foundation have been proposed to explain the very remarkable direct effects produced by the phenomena of this group. We hardly have room here to dwell upon such hypothesis, prominent among which Beilby's ideas about amorphous cement — although subject to serious objections — appear to be founded upon concrete and direct observations. The experimental data to be mentioned later in this Part will furnish a sufficiently complete and precise idea of the great practical importance of plastic deformation and of the amount of its direct effects upon the mechanical properties of the steels with which we are concerned. This will be especially apparent when such data will be compared to those men- tioned in Part IV regarding the mechanical properties of cast steels, variously treated. It will then be clearly evident that the direct effects of these phenomena ordinarily appear as a remarkable improvement in toughness or resistance to impact. Such comparison will also especially emphasize the fact that the effects of crystallographic slip and twinning act directly upon the properties of the metal and do not practically consist in modifications upon the course of homogeneity heat treatment such as those produced by hot plastic deformations. In fact we shall see that in the great majority of normal roUed or forged steels, even before being subjected to any particularly energetic homogeneity heat treatments, there occurs a great decrease in fragility or in what amounts to the same thing, a great increase in the resistance to impact, when their physical properties are compared to those of the original ingot. It may be observed immediately that the resiliency is at a maximum when it is measured by an impact test taken normal to the elongation produced by forging or rolling. That is to say, a longitudinal test will give much higher values than a transverse test. If the impact flexure test-piece has been cut with its longer axis normal to the elongation pro- SOFT AND MEDIUM STEELS 289 duced by hot-work and, therefore, tested with an impact directed in a direction parallel to that of said elongation, the results will be low. Concerning other mechanical properties, including resistance to tensile impact, we shall see by comparison with the data referred to in Part IV that the improvements produced by the phenomena of the third group mentioned in Sec. 168 are not greater than those which can be obtained with cast steels of properly selected composition, after having been subjected to a rational homogeneity annealing, occupying no greater time than that required for rolhng or forging. In other words: as far as those other mechanical properties are concerned, a steel casting subjected to a rational homogeneity heat treatment will possess characteristics entirely due to the phenomena of slip and twinning we are now discussing equal to those which it is possible to obtain with forged or rolled steels. 175. Before passing on to the examination of the phenomena pertain- ing to the fourth and last group specified in Sec. 168, it is necessary to note some more precise information regarding the processes which take place in those cases, already mentioned in the same Section, in which the crystaUographic phenomena continue to develop, even after mechanical work has ceased exercising a new specific action. Such crystalline growth occurs during the heating processes constituting the preliminary heat treatment. As has been already pointed out, such cases could be errone- ously explained by saying that the continuing crystaUographic modifica- tions are also due to the phenomena of diffusion responsible for a veritable normalizing. In reality, grain growth is still a direct effect of the plastic deformation occurring during the previous mechanical treatment. On account of this possible misinterpretation it is now desirable to present some experimental data illustrating the specific characters of such crystal- line changes. Thus we shall avoid repeating a discussion in later chapters to clear up the apparent anomalies in the heat treatment of rolled steels due to the appearance of such phenomena. The fundamental result is a large increase of the dimensions of the primary crystalline elements of the steel or of their colonies, i.e., a large "grain growth. " This increase occurs when a steel is heated after it has , undergone plastic deformation responsible for intracrystaUine twinning and slip along the planes of cleavage, which occurrences have been gath- ered into the third group specified in Sec. 168. Within certain Hmits of temperature and deformation so well determined for each steel as to constitute actual critical intervals, the increase in grain size is greater the more intense has been the crystaUographic rearrangement, the higher the temperature of subsequent heating, and the longer this heating has been protracted. Inasmuch as the amount of slipping and twinning is greater, other conditions being equal, the lower the temperature at which plastic de- formation of a definite amount has been accomplished, whenever we ]9 290 HEAT TREATMENT OF consider only the above mentioned critical intervals it can be predicted that grain growth will be of greater importance when the hot-work has been continued to subnormal temperatures. Experience fully confirms this prediction, and shows that grain growth is the most notable, other conditions being equal, when plastic deforma- tion has taken place at ordinary atmospheric temperatures, as is done for instance in cold stamping, cold rolling or cold drawing. Furthermore, in the ordinary hot-working processes grain growth is greater the colder the piece has been finished. It is well known that if a steel undergoes plastic deformation at relatively low temperatures, temperatures whose upper limits vary widely according to the composition of the steels but which usually do not extend beyond incipient red, it simultaneously acquires quite different physical properties which are ordinarily thought of as being peculiar to "cold- worked metal." Normally it possesses a higher elastic limit and tensile strength than before. When the stresses which have caused plastic deformations have taken place either below or above the just mentioned critical temperature interval, or when an anneaUng temperature has reached values below or above the extreme critical points, the phenomenon of grain growth becomes less intense. 176. From the viewpoint of purely structural characters the effects produced by grain growth upon annealing cold-worked metal make them- selves evident only in very soft steels containing between 0.05 per cent, and 0.15 per cent, carbon. All the facts indicated in Sec. 175 are ex- hibited in the single plane section illustrated in Fig. 163, reproduced from a photograph published by Chappell. This represents a cylindrical test- piece of very soft steel (0.06 per cent, carbon) broken in tension and then annealed for about 2 hours at 600°C., then split down the axis, pohshed and etched. In fact, it is possible to determine by examination of this sample the effects produced by different specific loads by noticing the tapered form assumed by the test-piece under stress and knowing the total value of said stress. Thus we see at the bottom of the figure that grain growth has been practically zero for the sections larger than a certain limit, namely, for unit loads below a certain value. Immediately beyond this limit, the grain growth becomes the most pronounced, whereupon it decreases gradually step by step as the section decreases (and therefore step by step as the unit load increases) until again becomes practically zero for sections near the breaking point of the sample. Figure 16-3 therefore exhibits the whole critical interval of plastic deformations within which is produced the phenomenon of grain growth upon subsequent anneaUng. SOFT AND MEDIUM STEELS 291 The practical importance of these critical intervals as regards the selection of heat treatments for cold-worked steels is quite evident. Speaking from the standpoint of the mechanical properties of the steel, it is sufficient to notice that grain growth ultimately may be considered as a concordant re-orientation of the crystal- line elements constituting several con- tiguous grains, so that the new arrange- ment of the planes of cleavage of such adjusted grains brings them parallel with the prolongation of those of several adjoining grains. It is easy to foresee that such an enlargement of the planes of cleavage, an enlargement which may reach very great values, must give rise to noteworthy modifications in the mechanical properties. Again this last remark is fully con- firmed by experience. It has also been demonstrated that the characteristic modifications in the mechanical proper- ties of steel due to heating after cold- work may occur with great intensity even for those high carbon steels in which an increase of the grain is not re- vealed upon microscopic examination, or if visible, but imperfectly. For oft repeated reasons it is im- possible to pause and examine in detail the extremely various manifestations produced by the modifications under discussion. I will therefore confine my- self to mentioning an example which shows the effects of grain growth with particular clearness. Figure 164 illustrates at about half the natural size the characteristic ap- pearance of the long narrow fissures produced in a field-gun shield after being hit by rifle bullets fired against it in a direction normal to its faces. The thickness of the plate was 4 mm. and the bullet was the usual ItaHan service ammunition, clad in a nickel-copper sheath. The shield was hit at a velocity of about 670 meters per second. Fig. 163. — Grain-growth in annealed tension specimen. (Chappell.) 292 7/ £'^7' TREATMENT OF Many plates were obtained from ingots of the same heat by a similar and closely observed process of rolling, followed by homogeneity quench- ings, and then by final quenchings and drawings. When proof-tested they showed that the characteristic impact fissures of Fig. 164 only occur in plates whose rolling had proceeded at lower temperatures. The great Fig. 164. — Gun shic4d cracked by bullet inip.Tct. H.ilf size. development of planes of cleavage clearly appears in this instance to be due to the heatings during the various heat treatments after the cold-work (rolling at too low a temperature). 177. The facts which have been related in the last two sections make it evidently necessary to take into most careful account the influence of over-strain in cold metal. SOFT AND MEDIUM STEELS 293 For instance, in the manufacture of high tensile steel wire for musical instruments, or aeroplane construction, after the first cold drawing the wire is subjected to homogeneity heat treatments often alternated with further cold drawings. Excessive grain growth may thus occur often accompanied by very serious deterioration in the mechanical properties of the wire, primarily due to the fact that the deformation suffered by the metal during drawing has taken place just within the critical intervals described in Sec. 175. Sometimes it happens that for the particular requirements of the manufacturing process it is not possible to avoid or sufficiently suppress the deleterious consequences, except by executing all the homogeneity heat treatments in the best manner possible, before the steel has been subjected to any mechanical strains apt to cause the dangerous cold work, renouncing afterwards any further heat treatment. For instance, in the case of high tension wire this normalization, and other thermal manipulation would be done upon the hot-rolled rod. At any rate, it should be clear ere this how extraordinarily important it is to study accurately the critical intervals of cold work, in relation to the consequences which may be produced during heat treatment. Most especially important it is whenever there is reason to believe that cold straining would follow a given series of hot workings, in turn to be followed by a heat treatment. I have already briefly mentioned in Sec. 107 the fact that grain growth entirely analogous in microscopic appearance and practical consequences to those which we have studied, may even take place due to the so-called "internal stresses" carried by some parts of the steel pieces subjected to simple heat treatments. These stresses are due to the diiiferent variations in volume produced by heating and quench- ing in the different parts of the same object. As has been pointed out in Sec. 107, the phenomena of grain growth consequent upon internal stresses may, in a certain sense and in these cases, be considered as true effects of heat treatment. 178.' Finally, we must also briefly investigate the general characters of the phenomena gathered together under group four as defined in Sec. 168. These take place when the hot work partially or wholly occurs within the 7— >« transformation interval. As has already been indicated in the same section these phenomena may be compared to the so-called disturbed crystallization, so well known in salt solutions; phenomena which have as essential consequences the entire or almost complete elimination of any under -cooling and a con- siderable decrease in dimension of the resultant individual crystalline elements. Recalling the remarks made in Part IV, especially in Sec. 116 and Sec. 139 e.s., regarding the forms and characteristic arrangements as- 294 HEAT TREATMENT OF sumed by the crystalline elements of a and /? ferrite after undisturbed separation from austenite, I believe that the characteristics of the various phenomena which may appear during disturbed crystallization can be summed up in the following groups and sub-groups: (A) After its complete solidification the steel has been subjected to a homogeneity heat treatment similar, for instance, to that indicated in Sec. 114, designed to eliminate in the most complete way the differences in carbon concentration between the nucleus and the periphery of each primary mixed crystal. This case happens only very rarely in practice, as has been pointed out several times. In it the process of disturbed crystalhzation may result in different structures according to the composition of the steel and the velocity of coohng through the transformation interval: First. — If the cooling takes place very slowly and the steel contains less than 0.35 per cent, of carbon, the ferrite crystallization occurs in two periods of time, as we have seen in Sec. 140 (2,a). However, in the case of undisturbed crystallization as there discussed, the lamellae of ferrite forming directly in the a state appeared as prolongations upon the globu- lar masses of first-segregated /3 ferrite acting as germs of crystallization. On the other hand, in this present case of disturbed crystallization the nuclear orienting actioii exercised by the first massive crystals cannot be felt in substantial measure, and the two types of ferrite crys- tals {j3 globules and a lamellae) form separately from each other. Due furthermore to the known effects of disturbed crystallization, the dimen- sions of each crystalline element of ferrite appear much smaller, other conditions being equal, than the size of the same elements formed in a similar process of undisturbed crystallization. Finally, the deformations undergone by the entire metallic mass are accompanied by deformations of the crystalline ferrite, which often deeply change its appearance. The above mentioned limit of 0.3.5 per cent, of carbon refers to the common carbon steels, as must be clear from all the considerations devel- oped in the previous chapters. For special ternary and quaternary steels, this carbon hmit is lowered in accordance with the criteria and data pre- sented in the Part I and II; the indicated phenomena vanish for the special steels called " high alloy, " in which the separation of the two forms of ferrite is no longer distinct. Second. — If the cooling through the transformation interval takes place very slowly and the steel contains more than 0.35 per cent, carbon (or more than the corresponding amount when it is a special steel) the ferrite will be entirely of the lamellar type; but the laminations will be a great deal smaller than in the case of undisturbed crystalhzation. Further- more, simultaneous deformation with ferrite crystalhzation is reflected by corresponding deformations and contortions so as to change the microscopic appearance profoundly. SOFT AND MEDIUM STEELS 295 Third. — If the cooling through the y^a transformation takes place very rapidly, as for instance, in drop forging small pieces, the various crystalline elements appear in very small dimension, and imperfectly de- veloped as to form, so that it is difficult to distinguish one transformation product from the others. (B) The solidified steel has been subjected to a heat treatment in- sufficient to totally obliterate the differences in carbon concentration between the nucleus and the periphery of the individual primary mixed crystals. We have already seen that this case is far more common. In it the course of disturbed ferrite crystaUization may take place in the following Fig. 165. — Alpha and Beta ferrite in 0.23 per cent, carbon forging. X 100. different ways, according to the cooling velocity and the composition of the steel: First. — If the cooling is very slow and the steel contains'less than 0.35 per cent, carbon (or less than the corresponding amount in a special steel) the structure of the completely cooled metal is still similar to that which we have described for the case A, above, except that in the present case traces of the large network of ferrite may remain, more or less de- formed. The more or less marked tendency for the large network of ferrite to persist permanently depends upon the same causes which have already been examined in Part IV. Figure 165 reproduces at an enlargement of 100 dia. the structure of such a 0.23-per cent, carbon steel, in which no appreciable traces of the 29G HEAT TT-lEATMENT OF \&vgQ network (if forrito have remained. The fiimple draw. X 130 Fig. 172. — Niclvcl steel casting, after heat treatment enrliiig in a true anneal. X 131. ■ These last conclusions are completely confirmed by comparing the structure obtained by means of the first treatment, Fig. 171, with that produced by the second treatment, Fig. 172, clearly showing the segre- gation of ferrite. SOFT AND MEDIUM STEELS 305 All the other data contained in the table require no comment, since they are sufficiently clarified by previous discussions. It may be added that ferrite segregation is evidently the cause of the differences in the results of treatments No. 3 and 4. Fig. 173. — Nickel steel forging, annealed and drawn. X 130. Fig. 174. --Nickel steel forging, annealed, quenched and drawn. X 1311. 182. In order to complete this first group of comparisons between the effects of identical mechanical and heat treatments upon various steels, we will now examine the results of the manufacturing processes described 20 306 HEAT TREATMENT OF in the two previous sections upon the physical properties of a very soft carbon steel. As has already been pointed out, the structure of steels of this type is strikingly modified by the phenomena studied in the previous chapter. The steel in question had the following composition. Carbon . 08 per cent. Manganese 0.79 per cent. Silicon 0. 10 per cent. Sulphur 0. 02 per cent. Phosphorus 0.04 per cent. The following table contains data covering the treatments to which the steel was subjected and the resulting physical properties. As re- marked above, the forging, heat treatment and mechanical tests were static tensile test Impact resist- No. Mechan- ical work Heat treatment Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Reduc- tion of area, per cent. ance (Charpy test), kg.-m. per sq. cm. Static bend- ing Micro- struc- ture 1 non^ Reheated for 10 hoiirs at 900°C. followed by slow cooling in furnace and then drawn at 680°C. for 6 hours. 55,200 34,100 29.5 56 7.6 Un- broken Fig. 175 2 none Beheated for 10 hours at 900°C. followed by slow cooling in furn- ace, then by an annealing at 780°C. for 6 hours. Finally a slow cooling in furnace. 54,200 35,600 31.0 59 7.0 Un- broken Fig. 176 3 forged at gocc. to 1100°C. Reheated at 900''C. and drawn at eSCC. under identical condi- tions to those of test No. 1. 56,300 37,000 31.0 75 19.9 Un- broken Fig. 177 i forged at 900°C to llOCC. Heated at 900°C. during 10 hours followed by quenching in water at 20°C. and then by a drawing at 680°C. for 6 hours. 57,200 38,400 28.5 74 20.0 Un- broken Fig. 178 SOFT AND MEDIUM STEELS 307 made under absolutely identical conditions with those mentioned in the two previous sections. Fig. 175. — Low carbon steel casting after heat treatment ending in a draw. X 100. Comparison of the tabular data and the corresponding microstructures — reproduced in Fig. 175 to 178 at an enlargement of 100 dia. — furnish Fig. 176. — Large grained ferrite in steel like Fig. 175 after annealing. X 100. interesting illustrations paralleling those of the previous sections, espe- cially in regard to very low carbon steels. 308 HEAT TREATMENT OF In particular, and without repeating in detail what has already been said, the previously noted fact is confirmed that physical properties of Frri. 177. — Traces of cleavage plains intersecting polished surface iiarallel to extension by forging. X 100. Fig. 178. — Forged low carbon steel after complex heat treatment. X 100. very soft steels of this type do not differ a great deal for metal simply cast and subjected to heat treatments, and for metal forged before being sub- jected to the same heat treatments. Impact strength is an exception. SOFT AND MEDIUM STEELS 309 We have already pointed out Sec. 135 that this circumstance permits safe use of common steel castings even as substitutes for forged or stamped pieces whenever the requisite elastic limits or other mechanical properties are moderate enough to allow the use of very soft steels. Whenever very high tensile strengths and elastic limits are required, however, such substitution can be made only by using special steels sub- jected to accurately controlled heat treatments. One observes upon comparison of the microstructure after treatment No. 1 (Fig. 175) and that of the same steel after No. 2 (Fig. 176) that the former discussion of the two corresponding samples shown in Fig. 171 and 172 is confirmed. In particular large grained ferrite is quite notable whenever the heat treatment ends with a true anneaUng. Finally, Fig. 177 shows clearly some hght striations in the grains of ferrite revealed by the nitric acid etching. Sach lines correspond to in- tersections of planes of cleavage with the plane of polish, and are now closely parallel, yet they do not have concordant directions grain to grain in the unforged steel shown in Fig. 176. In Fig. 178 this evi- dence does not reappear because the microsection is normal to the direc- tion of elongation produced by forging, while the previous photograph was taken parallel to that direction. We have already analyzed the causes of all these phenomena in previ- ous sections. 183. The effects produced by homogeneity annealings and quenchings may thus differ greatly according to the velocity of cooling which termi- nates the heating, and this always in the sense that a more rapid cooling favors the maintenance of the homogeneity obtained by heating above Aca. This now famihar fact takes on peculiar importance when heat- treating pieces of very large dimension and massive form. In such cases, it is not only impossible to cool the pieces with the same velocity in all parts, but in some parts of the metal it is even impossible to reach very high cooling velocities. Such limitations depend essentially upon the high thermal conductivity of steel. It is clear that the maxi- mum coohng velocity at any point within the metallic piece will practi- cally be reached when the external surface of the object is constantly maintained at the temperature of the quenching medium. In commercial operation the most drastic quenches are given by water at atmospheric temperature, and for quickest cooling the process generally requires the use of enormous quantities of liquid. When treating pieces of very large dimension, the maximmn cooling velocity which can practically be reached in the deepest regions is often below the minimum cooHng velocity necessary to cause a true quench- ing. This is rather an indeterminate speed, but if quenching be de- fined as the more or less complete suppression of the normal processes which precipitate and segregate the metarals stable at ordinary tempera- 310 HEAT TREATMENT OF ture, then we know that this minimum cooling velocity for quenching may vary within quite large limits according to the composition of the steel. At any rate, it is clear that the effects produced by a homogeneity quenching will be greatest at the surface of any piece, large or small; they will diminish gradually toward deeper strata of underlying metal. Finally, in metal contained in the regions whose distance from the ex- ternal surface of the piece is above a certain figure, the results of quench- ing will correspond to those of a true homogeneity anneahng. Experience fully confirms these predictions in all the cases where subsequent heat treatments are missing or consist of such low drawings as will in turn cause no new heterogeneities which might accrue to the first ones. In practice, however, variations in the resulting structure are far from showing the regularity from one point to another which could be expected on the basis of a brief statement of the phenomenon, as such has just been outlined. Very little precise information is known regarding the causes of the strong heterogeneities produced by heat treatments in various parts of large steel pieces. In general they have been ascribed primarily to irregularities in the conduction of heat from the interior to the exterior part of the metallic mass. Variation in heat transfer would be caused by a lack of uniform dimension and orientation of the structural elements of the steel. This fact in turn depends upon abnormal phenomena due to undercooling which have been discussed when considering primary crystallization. In the second place it seems certain that the action of internal stresses set up within the piece during mechanical and thermal treatment are also partly responsible for variations in the resulting structure. We have already briefly spoken of the effects of these internal stresses upon heat- treatment. 184. I shall now mention some examples showing how the phenomena indicated in the previous sections make their presence felt, and the extent of their influence. These instances will be valuable not only as particular illustrations of the principles noted in Sec. 180 to 182, but also on account of their great practical importance. A large block of medium carbon steel of the following composition was studied. Oarbon 0.42 per cent. Manganese 0. 67 per cent. Silicon 0. 25 per cent. Sulphur 0.01 per cent. Phosphorus . 06 per cent. The piece was of rectangular section measuring 760 X 1400 mm. and was obtained by pressing a rectangular ingot 1000 mm. thick, weighing SOFT AND MEDIUM STEELS 311 65 metric tons. It is seen that the ingot was subjected to a very small reduction in cross-sectional area. After pressing the block was planed on all sides and then heated during about 20 hours at 800°C. It was then immediately quenched with water at 16°C. by an apparatus drenching the block from all sides with heavy sprays of water, a process lasting nearly 2 hours and requiring about 4000 cubic meters of cold water. ^ 7nn ■1 . 1 c 1 2 T T ,s o 10 11 12 ^ii o K 1 D 450 1| '■ •s 7 8 J. -■ i 1 i 1 ■' 1 1 1 1 uoo. 1 1 1 1 1 1 M ' M A y a B A A Fig. 179. — ^Location of test pieces described in Section 184, Section CD below, AB above. About 20 minutes after the quenching water had been turned off, the temperature of the surface metal had raised to 100° to 120°C., due to transmission of the residual internal heat outward by conduction. Therefore, it was thought superfluous to subject the block to a drawing operation. Mechanical tests were then made under the identical conditions already described in Sec. 151 upon longitudinal test pieces taken from the positions located in Fig. 179; the results of these tests are collected in the following table. 312 HEAT TREATMENT OF I*osition No. (sec Fig. 179) Tensile strength, lb. per aq. in. Elongation, per cent. Impact resistance (Charpy test), kg.-nj. per sq. cm. 3 60,400 16 4 67,700 10 5 69,000 16 6 65,400 17 6.66 7 64,000 18 7.14 s • 64,000 18 5.16 9 64,700 17 8.11 10 78,900 24 11 76,100 22 12 67,400 17.5 13 65,000 14 Fig. ISO. — Structure at interior of massive forging after heat treatment. X 100. A tension test taken from a 40 mm. bar forged down^ from a 110 X 110 mm. test ingot taken from the same heat gave the following results: TensUe strength 78,800 lb. per square inch Elongation 21 per cent. Comparison of this result with those furnished by samples 3 to 13, especially noting that test-pieces 10 and 11 give results closely ap- proximating those from the forged bar, clearly indicates that the effects of pressing the large block and its subsequent heat treatment have been felt only upon the surface regions, extending no deeper than 70 to 90 mm. All other test-pieces beside those located near the surface give erratic ' The hot bar cooled in air after forging. SOFT AND MEDIUM STEELS 313 results, especially in resistance to impact, a phenomenon of which we have spoken in the previous section. Microscopic examination of the various test-pieces confirms the de- duction that the characteristic effects of forging and heat treatment^ have not been felt beyond a depth of 90 mm. Thus Fig. 180 reproduces at an enlargement of 100 dia. the metallic structure of test No. 12. Such structure is that which a steel of the average composition of the block examined assumes after simply annealing a raw casting, the anneahng to be followed by a slow cooling. 185. When the forging has been carried further, causing a greater reduction of section of the ingot, its effects are felt to a greater depth, other conditions being equal. As an example, a block substantially identical to the one described in the previous section may be mentioned. This new piece was obtained by pressing an ingot 1200 mm. thick under entirely similar conditions as before, the reduction of the section due to pressing however having been considerably greater than in the previous case, whose original ingot was about 1 meter. The steel was of the same type of the previous block but a little harder grade, containing 0.46 per cent, carbon. The following table contains the results of a series of physical tests made upon samples cut from the same positions in the new block as shown in Fig. 179. Position No. (see Fig. 179) Tensile strength lb. per sq. in. Elongation, per cent. Impact resistance (Charpy test), kg.-m. per sq. cm. 3 81,500 18 7.40 4 78,400 16 6.75 5 70,800 16 6.10 • 6 66,800 10 7 74,700 10 8 68,600 11 9 69,700 10 10 92,400 23 11 94,200 21 12 77,100 20 7.15 13 77,800 15 8.02 It is clearly shown that the combined effects of pressing and heat treatment are more deep seated than in the previous case, noticeably affecting strata as far as 150 mm. from the external surface. Given the fact that the only great difference in the history of the two f orgings consists in the amount of pressing which the piece has undergone, it may be con- cluded that the higher ultimate strength of the metal within 150 mm. of the surface is due principally to the direct action of such hot working. ' In this case a homogeneity quenching. 314 HEAT r RE ATM EXT OF However, one is not justified in excluding tlie concomitant indirect in- fluence of an allrralion in the rates of diffusion caused by the plastic deformation under ilie jiress. Fici. ISl. — Microstracture of test-piece No. U). X 200. Fig. 182.— Microstriicture of test-pieeo No. 11. X 200. Microstructural examination of the different test-pieces confirms this precaution. The structures of the three test-pieces No. 10, 11, and 13 are reproduced at an enlargement of 200 dia. in Fig, 181, 182 and 183. SOFT AND MEDIUM STEELS 315 Taking into uccoiini their position in the block, it is clearly seen that the increased agglomeration of ferrite, evidently due to the lessor efficacy of the quenching when passing from the superficial strata to the deeper ones, accompanies the increase in crystalline dimensions. In the light of the historj' of the block, it is clear that this structural difference is due principally to the different degree of plastic deformation produced by the press upon metal at different distances from the external surface. 186. As has already often been pointed out, the phenomena briefly described in the last sections have great practical importance, which is in- creasing rapidly in step with the growing tendency of the mechanical engineer to require from metallurgical technique metallic pieces of con- tinually larger dimensions, to withstand ever greater stresses. While this Mir-rostriicture of test-piece No. Vi. X 2(10. last e.xigency presupposes the greatest uniformity in mechanical prop- erties in all the parts of the metallic piece, it truly exists only .after subjecting the steel to a suitable heat treatment. The importance of studying proper means to eliminate or at least to mitigate the consequences of the phenomena which we have just examined is therefore evident. When called upon to manufacture pieces of large dimensions which must be subjected to homogeneity heat treatments, best results may be obtained by using steels characterized by a closely packed frequency of crystallization centers, whose transformation range upon coohng lies at very low temperature and consequently which are very sensitive to homo- geneity heat treatments. Stable austenite at relatively low temperatures may bo had with special steels containing alloying metals which form solid .■uo HEAT TREATMENT 0/<' sohiiions with the iron and with the vsuioiis inctMUic carbides. Un- stable austenite on the other hand may be a i)ro(hict of retarded trans- formation and is associated with metastaldc^ eqiiilibriiim. In those oases the rational utilization of thei'mal hysteresis as mentioned in See. 159 assumes special importance since it causes a more or less noteworthy depression of the Ar range. For reasons already studied, it is possible by such means to prevent or at least largely subdue the ferrite segregation accompanying allotropic transformation, using relatively small cooling velocities in the deepest re- gions of the piece to be treated. With these precautions the efficiency of the homogeneity treatment is nearly uniform in all regions. The re- quired mechanical properties may afterwards be imparted in an equally Fiu 1S4. — Martciisite, troostitc aud massive ccmentilc in one si'i-liuii. X UIO. luiiform manner by means of a final heat treatment, which may consist in a simple drawing sufficiently prolonged. As a general example, it may be recalled that it is exactly the rational application of the indicated criteria that permits the fabrication of armor plates of great thickness (sometimes above 400 or 500 mm.) yet equally tough throughout their mass. 187. It does not seem to me superfluous to mention here an experiment giving direct proof of the possibility that an occasional steel "takes the quenching" in very different ways in closely adjacent regions. An explanation was noted in Sec. 183 of th(> non-uniformity sometimes exist- ing in normalized pieces. Yet lacking direct evidenc(> that I'easoniiig might appear unsatisfactory at first sight. SOFT AND MEDIUM STEELS 317 Confirmation is given by the examination of the structure of a high carbon steel (carbon about 1.45 per cent.) which was maintained at 1050°C. about 60 hours then left to cool slowly to about 730°C., whereupon it was quenched in water at 60°C. Figure 184 reproduces the structure of a por- tion of this steel at 100 dia. It shows a region where the metal retains the mixed structure of martensite and austenite, immediately surrounded by martensite already partially segregated into troostite, in which have even formed various large white crystals of cementite. Here- no doubt can arise that we have discovered two very small contiguous regions in the same mass of steel where the quenching produced very different results, with its corresponding changes in the constitution. 188. Finally, before closing this series of observations suggested by the study of the particular effects caused by homogeneity heat treatments when apphed to large masses, it is useful to add some experimental data better to indicate methods for practical utilization of thermal hysteresis. As a matter of fact we have just seen a characteristic example of this practical utilization in the case of the preliminary heat treatment of pieces of large dimensions, and I have already indicated the general criteria upon which it is based. We shall take as an example the heat treatment of a high nickel steel, in which thermal hysteresis is remarkably wide. Besides, this case is particularly interesting from a practical point of view, for the reason that forged pieces manufactured of such steel cannot be quenched from a high temperature without almost inevitably cracking to a greater or less extent and depth. Only the rational utilization of the phenomenon of hysteresis permits one to avoid such damage. A heat containing 0.33 per cent, carbon, 0.75 per cent, of manganese and 5.80 per cent, nickel was made in the electric furnace, and cast in square ingots 350 X 350 mm. The tests to be mentioned were made upon round bars about 120 mm. in diameter obtained from these ingots by forging under an eight-ton steam hammer. All test-pieces were longitudinal and from the forged bars. Figure 185 reproduces the differential temperature curves traced with the Le Chatelier-^aladin double galvanometer using a test-piece of the forged steel in question. The bottom curve represents conditions on heat- ing, while the top exhibits the lag on cooling. Without entering in a discussion of the shape of these curves — -a discussion which as a matter of fact would be very interesting, especially where it would concern the slow and gradual deviations of the differential galvanometer during the trans- formation on cooling — it is easy to see that thermal hysteresis is very marked, reaching an amphtude of about 300°C. under the conditions of cooUng obtained by the apparatus. 318 HEAT TREATMENT OF A])])lyiiis tlu' ideas developed in the previous ehaptei-.s to the numerical data i'urnislied by this curve, it seems evident that the characteristic advantages of a homogeneity quenching will be had in case the tempera- ture has been maintained for a certain time above the transformation temperature (in this case between 800° and 900°C.) and the piece is left to cool slowly and uniformly through an interval of temperature which may amount to almost 300°C., so that the true quenching operation may llien 1)0 executed when the entire mass has uniformly reached a relatively very low tenqjeratuie in all its parts. Whenever pieces of this analysis are to be heat treated and are not of exceptionally lai'ge dnnensions, the quenching temperature may be only slightl}^ superior t(j 600°C'. without ^ 100 joo 300 koo 50O 500 hoc Uqo aoo I'^n;. lS."j. — Differential eur\'e on heating and fooling a 5.X()-per cent, nickel steel. causing the homogeneity or properties retained by quenching to be noticeably decreased when compared to the physical properties which the same steel assumes after a simple final annealing. The just indicated fact permits one to utilize fidly the advantages of a C|uenching for homogeneity, even when the form of the pieces to be treated would make it impossible to quenchfrom a high temperature, which would have to be done for a steel having a smaller thermal hysteresis. In other words: proper utilization of the hysteresis phenomenon permits one to secure the advantages of a homogeneity quenching by means of a heat treatment which approaches more or less a simple homo- geneity annealing. This is interesting from the point of view of its applicability to hard steels and pieces of intricate form. The following table shows the results of mechanical tests made imder SOFT AND MEDIUM STEELS 319 the conditions already indicated in Sec. 151 upon test-pieces obtained in the described manner from 120 mm. forged rounds, quenched and then drawn at 600°C. for 3 hours. All homogeneity quenching were prepared by heating the bars to 860°C. during 1 hour. Then the bars were left to cool slowly in the fur- nace until they reached different temperatures indicated in the third column of the table, when they were quenched in oil at 30°C. No. Temperature of the pre- heating (during 1 hour), "C. Temperature of the steel at moment of the quenching in oil, "C Temperature of the 3-hour drawing, °C. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. 107,400 107,000 108,500 107,700 Elongation, * ^ per cent. 1 2 3 4 860 860 860 860 860 750 700 650 600 600 600 600 124,000 124,900 126,900 125,900 22 24 28 23 The practically complete constancy of results obtained in the four cases proves the foregoing observations to be correct, and demonstrates that a honjogeneity quenching will favorably affect the values of the elastic limit to the fullest extent even when the piece is cooled from a temperature only slightly above the minimum indicated by Fig. 185. CHAPTER XXI COMPARATIVE EFFECT OF ANNEALING AND QUENCH- ING UPON THE LONGITUDINAL STRENGTH OF HOT WORKED STEELS 189. As a deduction from the considerations which we have so far had occasion to develop and from the specimens which have been de- scribed, it can be concluded that the more notable practical advantages of homogeneity quenchings on rolled or forged steels as compared with the "equivalent" homogeneity annealings {i.e., including a heating for an equal period of time and at the same temperature) may be classed in the following two groups : (a) For steels which have to withstand longitudinal stresses a quench- ing increases the contraction in area more notably than will a slow cooling; the elastic limit and the resistance to impact are also improved. (b) Steels which have to withstand transverse stresses in service have much less tendency to exhibit the phenomenaof transverse weakness, and therefore the values for the transverse physical properties more nearly approach the corresponding properties when tested in a longi- tudinal direction. In other words: the physical properties — especially the elastic limit, contraction in area and resistance to impact — are improved much more by homogeneity quenchings than by the equivalent preliminary anneal- ings, not only when the piece is tested longitudinally but more especially when the properties are measured transverse to the direction of extension. 190. In order to illustrate item (a) of the previous section, I will mention some results of tensile and impact tests made as described in Sec. 151 upon longitudinal test-pieces, cut from round bars about 100 mm. dia. obtained by forging 300 X 300 mm. square ingots. The data refer to steels of four different heats of similar analysis containing about 0.20 per cent, carbon, 0.50 per cent, manganese, 1 per cent, chromium, and 2.5 per cent, nickel. The table below collects the results obtained by testing each steel after a simple anneal for about 1 hour at approximately 850°C., and after the same material had been quenched in oil after the saine period and temperature in the furnace, and then drawn at about 650°C. For reference, Fig. 186 reproduces the structure of steel No. 3 as annealed, and the next picture (Fig. 187) the structure of the same steel after quenching and drawing. Both figures are enlarged 500 dia. The 320 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 321 contrasting manner in which the ferrite and the pearhte are segregated gives ample evidence of the quenching efficiency. In accordance with the observation contained in paragraph («) of the previous section, the rela- FiG. 180. — Niekol-chroiiiium steel, forged aucl annealed. X .501). P'kj. 1S7. — Nic;kcl-ciironuuni .steel, forced, i|uenciied and drawn. X 50t). tive effects of the two heat treatments upon the contraction in area, elastic limit and resistance to impact can be clearly seen by comparing the numerical data contained in the table foi' each of the steels. 322 HEAT TREATMENT OF Heat treatment Results of the mechanical tests steel No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Reduction of area, per cent. Impact resistance (Charpy test) kg.-m. per sq. cm. 1 Annealed 92,900 100,800 97,400 115,600 94,700 112,800 94,300 116,200 54,500 85,500 57,300 84,500 56,200 86,900 55,000 88,400 18.2 18.6 20.0 18.5 19.1 17.9 20.2 19.1 48.2 12.3 2 Quenched and drawn 62.8 43.2 65.8 49.8 68.7 41.0 66.6 22.2 13.1 3 Quenched and drawn Annealed 21.8 9.8 4 Quenched and drawn 19.3 12.1 Quenched and drawn 20.7 191. Characteristic effects of the two treatments upon the elastic limit are still more evident when the chemical composition is such that the steel may be manipulated so as to obtain nearly constant tensile strength, while maintaining all or at least a good part of the efficiency of the heat treatment. The following table contains tensile properties of three such steels containing about 0.45 per cent, carbon, 0.50 per cent, chromium, 0.50 per cent, nickel and 0.60 per cent, manganese. Annealings were made between 800°C. and 850°C.; quenchings after an equal heating were in water at about 20°C. Drawing the quenched material was done between 600°C. and 630°C. for a length of time sufficient to bring the tensile strength back to a value near that of the same steel simply annealed. Usual test-pieces were obtained, cut exclusively in a longitudinal di- rection from forged bars subjected to the specified heat treatment. Heat treatment Results of the mechanical tests steel No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elongation , per cent. 1 Annealed Quenched and drawn 107,400 103,800 101,800 103,400 108,700 109,900 69,700 78,200 56,900 76,800 69,700 83,900 15.5 17.0 2 Annealed 19.0 19.5 3 Annealed 18.5 20.0 SOFT AND MEDIUM STEELS 323 As can easily be seen, large differences remain in the elastic limit of each steel even when the corresponding differences in the breaking load are very small. The facts indicated in subdivision (a) of Sec. 189 have such an impor- tant practical bearing that I think it well to illustrate them further with another example taken from routine operations. A hollow forging weighing 16 tons was obtained from an ingot of medium hard 2 per cent, nickel steel weighing about 50 tons, forged upon a mandrel. The walls of the hollow cylinder after rough turning and boring had a thickness of about 170 mm. Test-pieces for standard mechanical tests (specified in Sec. 151) were taken from about the middle of the wall, and in such a way that their principal axis was parallel to the tangent of the corresponding straight section of the cylinder. Considering the process of forging it is clear that these test-pieces must be considered as longitudinal test-pieces, in the sense being parallel to the elongation produced by forging. Tests made upon the piece after annealing for 6 hours at 850°C. and left to cool slowly in the furnace gave the following results : Tensile strength 87,500 lb. per square inch Elastic limit 44,100 lb. per square inch Elongation 20 per cent. Reduction of area 33 per cent. [Resistance to impact (Charpy test) 6 kg.-m. per square centimeter The structure of the metal thus treated, reproduced at an enlarge- ment of 100 dia. in Fig. 188, shows the heavy segregation of ferrite and pearlite, produced during the very slow cooling of the piece. The same piece was heated again at 800 °C. during 2 hours and then chilled in the open air. Tests made upon the piece thus treated gave the following results : Tensile strength 90,200 lb. per square inch Elastic limit 48,400 lb. per square inch Elongation 23 per cent. Reduction of area 44 per cent. Resistance to impact (Charpy test) . . 7. 15 kg.-m. per square centimeter Microstructure of the latter at 100 dia. (Fig. 189) is characterized by a great deal smaller dimension of the crystalline elements. In the dis- tribution of the structural elements, however, there remains a trace of the previous large network. Figure 190 reproduces the fracture of the tension test at about 5 times actual size. It seemed to be about half crystalline and half fibrous. The fractured surface of the Charpy impact test-piece reproduced in Fig. 191 at about twice the real size, shows similar characters to Fig. 190. 324 HEAT TREATMENT OF In this fracture the two struetiires, crystalline and fibrous, also appear mixed in about equal parts. Fig. 1S8. — Large crystals of ferrite in annealed gun tube. X 100. Fig. 189. — Microstrupture of gun tube of Fig. 188 after air quenching. X 100. Finallj^ the same piece was again heated at 900°C. for 4 hours; then left to cool slowly in the furnace to 800°C. and then quenched in water. Heat treatment was completed with a drawing at 640°C. -SOf r .lA'D MEDIUM STEELS 325 Fig. 190. — 'lY'iLsioii fracture, gun tul>e after air iiueucliiuK. X 5. m lg ■/^»m^^^S^^^^^ " '■ ■,>>■•; fe-V:--:'^^ ■ '■ -^^ ■>'.i-' ■■■ -vriV';.!'-'- '. vlw-';'-''-.. ; .V*^.i«i»i.»«i;»/i*in^^ _ *— '^'ikiji ■;:•:;._ J ^ "\i^^^i^nlnBB BMHh H^M^^B.' Fro. 1!")]. — C'liarpy frafdii'p fif truii lube, after air (inoiiciiiiiK. X -. 32() . HEAT TREATMENT OF Re.sulls of (■oinparativ(> mechanical tests made upon tlie metal thus treated were as follows: Tensile stveiigth 105,700 lb. per square iiicli Elastic limit 66,800 lb. per square inch Elongation 26 per cent. Reduction of area 57 . 3 per cent. Rcsistence to impact (Charpy test). . 14.2 kg.-m. per s(|uarc centimeter Fii 1!)2. — Gun tiilje, water ^^^^^^H SpSSi^«&&ii»GK^'; '-'/^^^^^H ^^^^^^^^^^^Br r**SW ^^Pt^pB^^I^/. i< '^^^^^H SBKi2S£@r^EB3Uid^ .^"'.s^^^^^^H ^^^^^^^^^u^S^ !lS^K^^&^ujwiP9f'^' ' "^l^^^^^^^^l ^^^^^^^^^^B^ra^ miiS^llltej^iMiLS^iBb^iEiL'r*''' 4!^^^^^^^^^^l ^^^^^^^^^^^^Ik^E^M HHI^^v '^^h|£EE^nBm>^''%^^^^^^^^^^^^| ^^^^^^^^^^^ H^y^^JB^^^^i^^iH^^^^^^^H ^^^^^^Hl^ S^^^^^'r^''V^I^^^H ^^^^^^^^^^^^^bS «e^S^^L^v«'- ^^^^^H ^^^^^^^^^^^^1 i^^^^;^^' I^^^^^H 1 ■j^^^^^^^^^^^^H Fig. 19.3. — Tension frarture oorre.sponding to Fi^. KILl. X 5. Fig. 194. — Charpy fracture corresponding to Fig. 192. X 2. 328 HEAT TREATMENT OF how an extremely mild quenching, such as chilling a piece of large dimensions in the open air, may give rise to much different effects from those caused by a true' annealing. This is especially evident when it is observed that the effects of the second treatment mentioned in Sec. 192 are similar to, but greater than, those of the first. Therefore, if, on account of the slowness of cooling due to the large dimensions of the piece, the second heat treatment had exhibited the features of a true annealing, it should have caused a lesser value for tensile strength and elastic limit; or at least, presenting the hy- pothesis least favorable to our argument, it should not have increased those values over the results from the first experiment. The fact that none of those two situations occurred, but instead a substantial increase of the tensile strength and of the elastic limit, proves that the observed effects are due to a true quenching process, although extremely mild and attenuated. We have seen that this conclusion is confirmed by the microstructure. For the reasons already explained, air chilling, quenching in molten lead, and other mild quenchings when utilized as homogeneity heat treatments, may affect the material far more profoundly when applied to special steels characterized by very low transformation points upon cool- ing. This is even more true of forged or rolled steels than of special steel castings, and consequently assumes a greater practical importance, a fact which is easily comprehended when one remembers that steels subjected to plastic deformation are a great deal more sensitive to homogeneity heat treatments, for various reasons indicated in Chapter XIX. The practical importance of these facts consists in this: that the use of soft and medium alloy steels possessing the distinctive properties pre- viously indicated permits one to attain excellent results on intricate forms or large dimensioned pieces which could not be subjected to the shock of energetic quenching without serious trouble due to warpage, checks, cracks and the like. In steels of such analyses, even the mildest quench- ing produces microstructures and improvement in elastic limit, contrac- tion and impact strength fully as good as the most drastic cooling and drawing would produce in plain carbon steels. It is seen that we are in command of another method by which it is even possible to obtain results similar to those described in Sec. 188, reached by low-temperature quenching of steels possessing a strong thermal hysteresis. Besides, the ideas underlying the two methods are clearly similar. 194. It is opportxme to give at least one example of the phenomena mentioned in Sec. 193 and of the processes derived from their commer- cial application. 1 That is to say, a long reheating at a temperature above the transformation range, followed by a very slow cooling. SOFT AND MEDIUM STEELS 329 A steel containing about 0.25 per cent, carbon, 0.80 per cent, chromium and 2.80 per cent, nickel, was made in the acid open-hearth furnace and cast in 550 X 550 mm. square ingots, which were afterwards forged under a press to flat blooms about 200 X 300 mm. in cross section. The following table contains a description of the various heat treat- ments to which four of these blooms were subjected and the results of tensile tests made upon the usual longitudinal test-pieces. It should be -remembered that in this case, as well as in the tests mentioned in Sec. 192, the coupons were cut from the central or deepseated portion of each bar; namely, from the region where the cooling velocity was necessarily moderate. Heat treatment Results of tension tests No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Reduction of area, per cent. 1 2 Heated at 850°C. for 6 hours, followed by slow cooling in the furnace Heated as above, followed by quench- ing in oil at 30°C 94,400 105,500 107,400 108,300 56,900 79,600 78,210 76,800 21 22 21 20 42 58 3 4 Heated as No. 1 followed by slow cool- ing to 700°C. and then by air chilling Treatment as No. 3 and then drawn at 600°C. for 2 hours 56 55 Number 3 shows that air chilling this steel gives practically identical results as when quenching in oil (No. 2), while these results differ substan- tially from those obtained by a true annealing (No. 1) especially as far as elastic limit and reduction of area are concerned. Data concerning test-piece No. 4 show that drawing this steel at 600°C. does not modify the mechanical properties substantially, because the differences between the values given by No. 4 and No. 3 are within the limits of experimental error. Therefore, we now have one of those cases which I have mentioned in another connection, in which an "attenuated quenching" (in this case in oil or in air) constitutes a complete heat treatment by itself, comprising in one operation the effects of normalizing and of a final heat treatment for quality. It has already been mentioned that in these cases a final drawing may sometimes be advisable (although it does not modify the mechanical properties imparted to the steel by the previous treatment) simply as a means to eliminate any internal stresses produced during the mechanical work or during the heat treatment. CHAPTER XXII COMPARATIVE EFFECT OF ANNEALING AND QUENCHING UPON THE TRANSVERSE STRENGTH OF HOT WORKED STEEL 195. Let us now see some examples showing the characters of the phenomena mentioned briefly in Sec. 189(6). They comprise the second group of main practical advantages obtainable by homogeneity quench- ings, as compared to the results made possible by a simple annealing. Essentially the advantage in quenching rests in the better suppression of all those characters which together are responsible for the transverse weakness occurring in greater or lesser measure in all forged or rolled steels. It is necessary to remind the reader of the observation already made in another connection: Generally, transverse and longitudinal properties do not differ greatly in very pure and well deoxidized steels, while serious and dangerous transverse weakness occurs in greater measure with in- creasing degree of oxidation at the time of solidification. Emulsified sonims are always contained in oxidized steels although their dimensions might be so small as to render them hardly visible under microscopic observation made without special care; yet when they appear in excep- tionally great quantities, transverse weakness-is sure to be pronounced. It follows that a practical study of transverse weakness in forged or rolled steels may be conducted at the same time one considers the effects produced upon them by emulsified and more or less oxidized sonims. In mentioning the experiments by which I propose to sketch the characters of transverse weakness, I shall follow this criterion exactly. Therefore, I shall begin by describing some very pure and well deoxidized steels, nearly isotropic, and shall then pass to highly oxidized steels con- taining many strongly oxidized emulsified inclusions. 196. First of all it is opportune to show by a concrete example that in some classes of work it is necessary to use the greatest care that trans- verse weakness does not exceed a certain value in relation to the stresses to be withstood by each individual part of the machine, and to study carefully the various aspects under which the defects may appear. In this way, the practical importance of the forthcoming data will be more evident. It will also be clearer that from this special point of view, the technical problems connected with mechanical working and with homo- geneity heat treatment of machine steels must be studied with great care. Figure 195 reproduces at about two-thirds actual size a sulphur print taken upon the surface obtained by cutting the neck of a crank-shaft for a 330 HEAT TREATMENT OF SOFT AND MEDIUM STEELS 331 six-cyliuder aeroplane motor along a plane passing through the axes of the adjacent bearings. Sulphur printing, as is perhaps well known, is done by pressing a sheet of silver bromide photographic paper moistened with dilute sulphuric acid upon the polished surface of the steel for a few seconds ; the sulphuric acid, in attacking metal and its inclusions reacts with the formation of hydrogen, mixed with a small proportion of hydrogen sul- -9XC. -T?*~ Fig. 195. — Sulphur print of longitudinal section of rrank shaft. phide, a proportion which is greater the higher the percentage of sulphur in the steel. The H2S thus formed, being in close contact with the silver bromide at the points where it is evolved, is immediately fixed in location by reacting to form silver sulphide. The sensitive paper is therefore tinted in its various points by a color which is denser the higher the sul- phur concentration in the corresponding points of the metallic surface. Inasmuch as the concentration of sulphur is highest in the regions 332 HEAT TREATMENT OF corresponding to the periphery of the primary crystals of steel (as we know from the discussion given in Parts I and II), the distribution of the darker regions of a sulphur print will furnish a picture of the form and distribution of the structural elements derived from the y mixed crystals. ' A roughly parallel alignment of these structural elements appears clearly in the case represented by Fig. 195. This alignment is about the same for the .various parts of the piece under examination, which is due to the method of manufacture. A flat bar was first produced under the hammer which was then shaped into three rectangular blocks connected at their centers by a neck of steel later to be turned to fit the intermediate bearings. From these square blocks were then machined the finished crank hangers and the piston pin. Observation of Fig. 195 confirms the conclusion evident from this description that the two throws of the crank shaft will be subjected to bending stresses in service in a direction trans- verse to the principal direction of forging. That is to say, the shaft will be loaded in a direction in which the effects of transverse weakness are felt in the most dangerous and pronounced way. It is clear that in cases similar to the one just described it is necessary to reduce this transverse weakness to the extreme minimum by means of special precautions during melting and pouring of the steel, and by means of a rational study and very accurate application of heat treatments for homogeneity. When the melting and heat treating practice were not so well under- stood, so that approximate equality in physical properties in all directions could not be attained, especially in the high tensile steels, it was necessary in many cases to modify the forging practice — even by greatly complicat- ing the manufacturing processes — so that the parts of the member which was to withstand the greater stresses would eventually be formed of material which had been forged longitudinally to the direction of said stresses. Thus, in cases similar to the one illustrated in Fig. 195, it was customary to start by forging a straight bar with diameter a little larger than the greatest diagonal of the finished square section of the crank hanger. The bar was then bent, often in expensive dies, into the form of the completed shaft. Again, whenever hollow cylindrical bodies designed to withstand tangential stresses were to be manufactured, as, for instance, rotors for high velocity steam turbines, guns, tubes, jackets, etc., recourse was made to fprging upon a mandrel, even for pieces of smaller diameter. At the present time this process, very difficult and ex- pensive as compared to forging solid sections, should be only applied to the fabrication of hollow cylindrical parts of large diameter. 197. As promised at the end of Sec. 195, I will start by giving some data regarding cases in which transverse weakness can be substantially eliminated with relative facility by means of homogeneity heat treat- SOFT AND MEDIUM STEELS 333 ments, primarily on account of steel well deoxidized when poured and containing a low proportion of emulsified sonims. The first of these cases is the large hollow cylindrical piece described in Sec. 192 made of 2-per cent, nickel steel and obtained by forging upon a mandrel. Due to the method of working, the radial direction is clearly transverse as regards direction of elongation. Therefore, I shall indicate the results of the physical tests made upon heat treated test-specimens of the dimensions indicated in Sec. 151 so cut that their axis corresponds to a radius of the cylinder. The three heat treatments are as indicated in Sec. 192. After annealing at 850°C. for 6 hours the results were the following: Tensile strength 84,800 lb. per square inch Elastic limit 44,100 lb. per square inch Elongation 16 per cent. Reduction of area 18 per cent. Resistance to impact (Charpy test) 4 kg.-m. per square centimeter After the second treatment (air chilling after heating at 800°C. for 2 hours) the results were as follows : Tensile strength 92,200 lb. per square inch Elastic limit 49,800 lb. per square inch Elongation 15 per cent. Reduction of area 16 per cent. Resistance to impact (Charpy test) ... 5 kg.-m. per square centimeter Finally, after the third treatment (heating at 900°C. for 4 hours followed by the slow coohng to 800°C.; then quenching in water and finally, drawing at 640°C.) tests gave the following results : Tensile strength 104,800 lb. per square inch Elastic limit 65,400 lb. per square inch Elongation 22 per cent. Reduction of area 48.5 per cent. Resistance to impact (Charpy test) . . . 12.8 kg.-m. per square centimeter Comparison of the data in the above three groups between each other and with the corresponding ones reported in Sec. 192 from longi- tudinal test-pieces, suggests the following observations: This annealed steel is but slightly deficient in transverse properties, proving that the steel used was remarkably pure and well deoxidized. Appreciable deficiencies appear only in the figures for reduction of area and resistance to impact. In the air-cooled steel (a mild quenching which, due to the dimensions of the piece, is very nearly a true annealing) the elongation and con- traction are practically unimproved, while there is a slight gain in the resistance to impact. Altogether, it can be observed that the improve- ment due to this second treatment in comparison with that of the previous 334 HEAT TREATMENT OF annealing, is substantially less in the transverse than in the longitudinal tests. Finally, in the steel heated at 900°C. quenched and drawn, the im- provement in the transverse physical properties is a great dealbetter than in the longitudinal; so much so that one may consider that this heat treatment has entirely eliminated the effects of any pre-existing trans- verse weakness. Here therefore is another example showing still more clearly that the efficiency of homogeneity q'uenchings is much higher than the corresponding annealings. 198. As has been already pointed out, and after all is evident from the arguments heretofore developed, the indications and quantitative measures of, transverse weakness vary within quite large limits, starting with very pure and perfectly deoxidized steels in which the effects are of small amount and can be eliminated nearly completely by proper heat treatment, and ending with some impure steels, strongly oxidized and strewn with abundant non-metallic inclusions, steels so weak transversely as to render the metal really unfit for the greater part of its intended uses. Such extreme cases are little improved by the most energetic heat treatment. One extreme of this series has already been illustrated immediately above. Now before describing some other more difficult cases, 1 think it best to describe a series of simple tension experiments, made upon steels of similar composition and forged under the same conditions, arranged according to the increasing amount of transverse weakness which they exhibit. In this way it will be easier for the reader to form a synthetic idea of the limits within which the phenomenon may vary due simply to variations in purity. In order to better emphasize the anisotropy of each heat, and in order to confirm the fact that even the worst of them may possess very satisfactory mecha,nical properties when determined in the longitudinal direction, 1 shall also give the results of longitudinal tension testpieces. The steels were made by the acid open-hearth process and cast in octagonal ingots weighing 12 to 15 tons. They were all medium-hard 2-per cent, nickel steels. The ingots were forged under the hydraulic press and reduced to cylindrical bars of about 350 to 400 mm. dia. The bars then were rough turned, bored and subjected to standardized annealings and quenchings for homogeneity, made in accordance with the criteria already mentioned. These preliminary heat treatments were followed by simple drawings, designed to bring the tensile strength and the elastic limit to the desired values. The transverse and longitudinal tests were made under identical conditions and upon test-pieces of the same size and shape, all as described in Sec. 151. SOFT AND MEDIUM STEELS 335 Transverse tests Longitudinal tests In tension In tension In bend- In No. Tensile Elastic Elonga- Reduc- tion Tensile Elastic Elonga- Reduc- tion bend- ing strength, limit. tion, of ing strength. limit, tion, lb. per lb. per per area, lb. per lb. per per sq. in. sq. in. cent. per cent. sq. in. sq. in. cent. per cent. 1 105,300 78,200 18 48 OK 101,000 75,400 19.5 52.0 OK 2 115,200 81,100 19 43 OK 118,100 79,600 20.0 55.0 OK 3 107,400 82,500 20 50 OK 109,500 81,100 22.0 61.0 OK 4 110,400 82,800 14 33 OK 112,200 76,100 23.0 52.0 OK 5 109,400 74,200 13 30 OK 106,700 75,400 21.5 48.0 OK 6 113,200 87,500 12 28 OK 110,900 82,500 22.2 62.0 OK 7 110,000 75,100 9 5 cracked 107,400 91,000 19.8 49.0 OK 8 114,400 84,700 7 11 broke at 80° 116,300 80,200 20.2 47.4 OK 9 116,100 85,600 8 14 cracked 118,100 94,200 19.6 46.4 slightly cracked 10 114,600 85,300 5 9 broke at 98° 112,200 75,100 20.0 38.9 slightly cracked 11 91,500 74,500 3 broke at 120° 114,200 89,600 18.0 43.5 OK. 12 82,500 49,800 1 broke at 150° 118,300 83,900 19.0 48.2 slightly cracked This table requires few comments. The data are gathered in four groups, the first of which includes heat treated steels which are nearly isotropic in physical properties. The succeeding groups cont'ain data relative to steels in which the dangerous transverse weakness appears with an ever increasing intensity in spite of the heat treatment. The three steels of each group are about equally deficient in toughness across the grain. 199. That homogeneity heat treatments eliminate, or at least suppress, transverse weakness is even more evident when using a more energetic heating and cooling program (as we shall see further on by some concrete example) or when a definite treatment is repeated. In general it is a fact that the efficacy of a prehminary heat treatment when based upon the use of quick coolings is very much greater when more than one quenching is made, than when the heating of the same degree and total duration is made continuously in one phase, and followed by only one quenching. In order to give a fully satisfactory explanation of this fact, it would be necessary to refer to ideas concerning the crystallography of steel which we have had no occasion to develop in the previous chapters. Therefore, I would only point out the fact itself as being one which has a great practical importance. 336 HEAT TREATMENT OF Heat treat- ment No. Description of repetitive heat treatment Transverse tests Steel In tension No. Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Reduc- tion, per cent. In bending 1 I II III Preliminary quench — followed by a draw. Two preliminary quenchings (identical to that of No. I ) followed by a draw- ing. Three preliminary quenchings {ident- cal to No. 1) followed by a drawing. 106,400 107,400 105,300 83,600 74,500 74,700 8 13 20 14.3 24.2 37.5 Broke at 90° Cracked OK 2 I II III IV Preliminary quench followed by a draw. Two preliminary quenchings (identical to that of No. 1) followed by a draw- ing. Three preliminary quenchings (identi- cal to No. 1) followed by a drawing. four preliminary quenchings (identi- cal to No. 1) followed by a drawing. 121,500 118,500 115,900 112,200 86,800 83,100 85,600 78,900 6 10 15 19 18.0 10.0 26.0 34.0 Broke at 110° Broke at 90° Cracked OK Instead it is more desirable to give some experimental data showing the enhanced effects obtained by several repetitions of a given heat treatment, in comparison with those that the same treatment produces when applied only once, or a smaller number of times. The figures contained in the above table, refer to steels of the same type as those mentioned in the previous section, manufactured, treated and tested under the same conitions. In this case, I deem it superfluous to add the results of longitudinal tests. Neither do the results require any comment. 200. Finally, it sometimes happens that steel of a certain heat appears "refractory" against normalization, at least as far as concerns an attempt to eliminate or mitigate the effects of transverse weakness. Two examples of this phenomenon are shown in the table on page 337, containing the results of testing 2-per cent, nickel steel forgings made under the same conditions as those described in the two previous sec- tions. In the greatest majority of the cases in which the steel is exception- ally weak when tested across the grain, and in which this defect cannot readily be eliminated by treatments ordinarily efficacious, such a condi- tion of affairs exists in all the forged or rolled pieces obtained from ingots of that heat. This is a clear indication that the trouble originates from the nature of the raw materials charged into the furnace, or from the con- ditions under which the melting, refining and deoxidizing operations have been carried out. Other conditions being equal, experience demonstrates SOFT AND MEDIUM STEELS 337 1 a ! M M M o o o M M M O O O a ^o a Reduc- tion of area, per cent. OS lO ,H M •* ■* »0 «D rt M W CO Elonga- tion, per cent. OS QO ,H ^ rH (N lO TJH >0 Elastic limit, lb. per sq. in. 82,500 79,600 82,500 83,900 83,900 83,900 Tensile strength, lb. per sq. in. 106,700 110,900 110,900 119,500 118,100 120,900 1 1 s fl3 Cracked Cracked Cracked Broke at 120° Broke at 120° Broke at 120° c o 1 0) Reduc- tion of area, per cent. l> O b- t^ IN OS Elonga- tion, per cent. OS 00 Oi r.. o o Elastic limit, lb. per sq. in. 72,500 81,100 71,100 86,800 93,900 91,000 Tensile strength, lb. per sq. in. 110,900 118,100 113,800 118,100 }20,900 120,900 Description of repetitive heat treatment One homogeneity quenching from 8oO°C., then drawn at 600°C. Two homogeneity quenohings from 850°C., then drawn at 600°C. Four homogeneity quenohings from 850°C., then drawn at 600°C. One homogeneity quenching from 900°C., then drawn at 600°C. Two homogeneity quenohings from 900°C., then drawn at 600°C. Four homogeneity quenohings from 900°C., then drawn at 600°C. Heat treat- ment No. " s 3 M S I-* So f-t IN 22 338 HEAT TREATMENT OF that the deoxidizing operations are those which if unskillfuUy conducted will leave the most serious after effects. However, there are cases in which the phenomenon of transverse weakness appears in quite different degree and persistency in pieces obtained from the different ingots of the same heat, and sometimes even in different parts of the same ingot, or in a single piece obtained from one portion of an ingot. It is clear that, in these last cases, the defect must originate during teeming or during the metal's solidification. This hypothesis is very frequently confirmed by microscopical analysis, revealing the presence of emulsified sonims. From their posi- tion in relation to the crystalline ferrite masses, these inclusions appear to be strongly oxidizing in their influence, in accordance with what was said in Sec. 97 and 98. Now a great many observations demonstrate that emulsified sonims of certain types may mix with the steel in very different degree during different phases of the teeming, and thus remain suspended in the metal in different proportion in various parts of the same ingot or of the same casting. ^ I cannot dwell upon the reasons for this segregation, but inasmuch as transverse weakness may be strongly accentuated by the presence of emulsified oxidized inclusions, the same reasons which explain the irregular distribution of those inclu- sions into different ingots of the same heat or in the different parts of the same ingot or casting will also explain the irregular results of physical tests. As an example, a steel of the following composition : Carbon 0.30 per cent Manganese . 62 per cent. Silicon . 26 per cent. Sulphur 0. 02 per cent, Phosphorus . 06 per cent. made in a 35-ton acid open-hearth furnace, was cast in octagonal ingots weighing about 7 tons each. The ingots were forged by a press into round bars about 300 mm. in diameter. After rough turning, the bars were subjected to a heat treatment for homogeneity. Those obtained from the first two ingots cast showed only a slight transverse weakness, easily eliminated by normaUzing. On the other hand, the bars obtained from the last three ingots were very brittle transversely, and they were improved but httle and with difficulty by the heat treatments. Microscopical examination demonstrated that the differences in the transverse properties were due to the content of oxidized sonims. Thus, Fig. 196 at 100 dia. reproduces the structure of the bar ob- ' These very facts emphasize the commercial importance of a detailed study of inclusions — ^a study which is closely bound up with an investigation of steel manu- facturing processes. SOFT AND MEDIUM STEELS 339 tained from the first ingot cast, after forging and annealing for 6 hours at 850°C. followed by slow coohng. The steel contains few non-metalhc inclusions, and these do not appear to have a strongly oxidizing influence. In fact, the various non-metallic particles to be seen in the figure are not surrounded by particularly well developed ferrite. On the other hand, some, as for instance the inclusion at the lower left side of the figure, and the long one appearing at the right, are actually in contact with pearlite. Fig. 19G. — Microstructure of 0.30-per cent, carbon steel forging containing harmless inclusions. X 100. A transverse tensile test upon the steel thus treated gave the following results : Tensile strength 81, .500 lb. per square inch Elongation 22 per cent. Reduction of area 26 per cent. The same bar, annealed again for 6 hours at 780°C. and left to cool in the air, showed the structure reproduced in Fig. 197 (enlarged 100 dia.) and gave the following results when testing transversely: Tensile strength 87,.500 lb. per square inch Elongation 21 per cent. Reduction of area 38 per cent. Aside from the reasons for the small dimensions of the crystals of fer- rite, which we know, we could repeat for Fig. 197 the same observations made about the structure of Fig. 196. The mechanical properties in the two cases also prove that the transverse properties are nearly equal to the longitudinal, especially after the mild quenching. 340 HEAT TREATMENT OF The steel in the fifth ingot of the same heat was quite different. Figure 198 reproduces at an enlargement of 100 dia. the structure of the bar obtained by forging in the same manner as described for the first Fig. 107. — Same .steel as Fig. 196 after air quenching. X 100. Fig. 198. — Banded fcrrite around oxidized inclu.sions. X 100. ingot, followed by an anneal at 850°C. for 6 hours. It was cooled slowly in the furnace. The inclusions are more abundant in this photograph than in the view showing the first ingot, and their oxidized state is indicated by the large SOFT AND MEDIUM STEELS 341 masses of ferrite which have gathered around them, forming lamellar masses appearing in section as bands. They are th(! principal cause of transverse weakness. Tension tests made upon this bar confirm results as follows: (he expectations, giving Tensile strength 82,500 lb. per square inch Elongation I3 pe,. cent. Reduction of area g per cent Transverse tests made upon the same bar after having been subjected to a second heat treatment (annealing at 780°C. for 2 hours followed by cooling in the air) gave the following results: Tensile strength . . . Elongation Reduction of area. 88,200 lb. per square inch 12 per cent. 11 per cent. Fig. 199. — Same steel an I'ig. 198 after air quenchiiiK- X 1(10. As seen, the physical properties remain relatively unaltered even after the second heat treatment, which also leaves almost entirely untouched the masses of ferrite gathered around the inclusions. This fact is shown in Fig. 199, (X 100) representing the microstructure of the bar taken from the fifth ingot, after having been subjected to the second treat- ment. 201. As has been pointed out, it sometimes happens that different parts of the same piece, often very near each othei-, give quite differ- ent results when tested_transversely. 342 HEAT TREATMENT OF In some cases microscopic examination of the metal immediately establishes the fact that these differences are due to a different amount of soUd non-metallic inclusions in the various parts of the piece; as an instance Sec. 106 may again be consulted. But it may also happen (as mentioned in Sec. 173) that very great differences in transverse properties may occur in the various parts of a forged or rolled steel piece even without being possible to detect note- worthy quantities of sonims with the ordinary means used in microscopy. A typical example of this follows. Part of an octagonal ingot one meter across flats was forged under the press into the form shown in Fig. 200 and then rough turned. The steel used was a soft steel containing 4 per cent, nickel and 0.16 per cent, carbon. C 1^ M B Fig. 200.' — Sketch of forging, showing location of test-pieces. After annealing at 800°C. for 6 hours followed by slow cooling, three longitudinal and three transverse test-pieces were cut at the points indi- cated in the figure by letters A, B, and C. Tension tests made under the usual conditions gave the results collected in the following table. Longitudinal test-piece Transverse test-piece Location (Fig. 200) Tensile strength, lb. per sq. in. Elastic limit, lb. per sq. in. Elonga- tion, per cent. Reduc- tion of area, per cent. Tensile strength, lb. per sq. in. Elastic limit, lb. per. sq. in. Elonga- tion, per cent. Reduc- tion of area, per cent. A B a 83,600 80,100 81,100 54,000 46,900 46,900 18 22 24 40 52 63 82,800 77,900 79,800 51,200 45,500 48,400 4 13 17 28 25 Flange A evidently shows the phenomenon of transverse weakness very markedly, while it is much less dangerous for test-pieces B and C. The piece was then heated for 8 hours at 850°C., quenched from 800°C. in water and drawn at 620°C. Transverse test-pieces were again cut from regions A, B and C in a position contiguous to the first ones. Re- sults of tension tests on these are tabulated in the following: SOFT AND MEDIUM STEELS 343 Location (Fig. 200) Transverse test-piece Tensile strength, lb. per sq. in. Elastic limit, lb. per s(i. in. A B C 82,500 92,400 91,000 66,800 69,700 66,800 Fjlongation per cent. 3 22,0 21.5 Uciiuction of ; per cent. 46 38 re B. It appears that quenching has actuallj^ decreased the ductihty gion -4., while it has entirely obliterated transverse M'eakness in reg m m region Microscopic examination of the metal after this last heat treatment reveals remarkable heterogeneities at region A but does not show excep- tional quantities of sonims (see Fig. 201 at 100 dia). Fig. 201. -Heterogeneity a.ssociated with transvor.se weakness in forging of Fig. 200. X 100. For reasons which we cannot analyze here it appears probable, even when microscopic examination of a polished surface etched with the usual reagents does not reveal the presence of the usual emulsified sonims in ab- normal quantities, that the phenomenon of transverse weakness strongly localized in certain parts of a forging is due to an oxidized condition of the metal, associated with the presence of numerous very small non- metallic inclusions in a relation analogous to that observed between the commonly observed emulsified inclusions and the state of oxidation which causes transverse weakness. Such very small particles are observed 344 HEAT TREATMENT OF only under very high magnifications when examining metaUic surfaces carefully polished and not etched. 202. It has already been said at the beginning of Sec. 199 that the eflficacy of homogeneity heat treatments may be increased not only by repetition (as has been seen in the last sections) but also by utilization of more drastic or energetic programmes, especially when it is designed to eliminate transverse weakness. This greater "energy" of normalizing practice must evidently consist in increasing the temperatures to which the various heatings comprising the program are maintained, in prolonging the time at such temperatures, and in increasing the velocity of subsequent cooling. Extremes must of course be avoided, such as would cause various perturbations in the course of processes naturally accompanying normalizing, which devia- tions have also beeii noted in the preceding pages. The following examples are illustrative : A steel of composition Carbon . 33 per cent Manganese . 68 per cent. Silicon . 33 per cent. Sulphur . 01 per cent. Phosphorus . 04 per cent. was cast in square ingots 550 by 550 mm., and forged under the press into round bars 250 mm. dia. One piece, after machining, was cut into two equal cylinders which were treated and tested separately. Transverse test-pieces were obtained from the ends of the two cylinders which originally were contiguous in the forging, in order to decrease the probabilities of accidental differences in quality of the metal. One half-bar was maintained at 820°C. for 2 hours, then quenched immediately in cold water and finally reheated at 700°C. The results of the transverse tension test upon the steel thus treated follow : Tensile strength 81,400 lb. per square inch Elongation 14 per cent. Figure 202 reproduces at 100 dia. the microstructure of the steel, while Fig. 203 shows the appearance of the fracture. As shown, the break is partly crystalline and partly woody in texture. The other end of the bar was heated for 6 hours at 900°C., then left to cool slowly in the furnace until it reached 800°C. and quenched in cold water. Finally it was reheated at 700°C. The transverse tension test made upon the steel thus treated gave the following results : Tensile strength 85,300 lb. per square inch Elongation 26 per cent. SOFT AND MEDIUM STEELS 345 Fic. 202.— Microstructure of 0.33-per cent, carbon forging after quenching and reheating. X 100. ^M ^^^^^^^^^^Hg^i^^^^^^^^^^^^^^^H ^^H ^^^^^^^^H^KT ' ''ISt^BHKBt^^^^^^ tip CThE H^H i-fl 1 y^y.-"^^^p ^^ IKi ^^E^^»^ij^j^MM[^^MWW 1 ^'fl..26:.i:.to.x;.ZJii*l*^ Fig. 203.— Tension fracture of steel of Fig. 202. 346 HEAT TREATMENT OF Fig. 204. — Mifroslrufdire nf sfcel of Fij;. 202 after normalizing;, ciucncliing and reheating. X 100. W>i' fc- v^' ' 'm^^^^^K «. fl 'gW| 1 1 ,^^^^^^^K^^k^ s^y'^ [ ^^y -' ^ -^ ,!^SH I' f 1 ^BaMMaK^ ^*a. ^^*ri^ ^A * ', ' ' ^'J^BWj f m lll ^N-f*^'' ,##ii '"'''saiP'^B??^^'' 2 kk- wm jH Fig. 205. — Tension fraeture of steel of Fig. 204. SOFT AND MEDIUM STEELS 347 Figures 204 (X 100) and 205 reproduce respectively the microstructure and the fracture (now entirely fibrous) . The greater efficacy of the second treatment for eliminating trans- verse weakness is evident by comparing this data. 203. An even more clear and complete example of the fact stated at the beginning of the previous section, is the following : A steel containing about 0.50 per cent, carbon and 0.50 per cent, chromium was cast and forged in the same way as the one just described. One of the bars was cut in four half-cylinders by dividing it along two planes, one passing through the axis of the bar and the other normal to the first one. The four parts were subjected to the heat treatments indicated in the second column of the following table. From each one of these, transverse test-pieces were obtained from positions which were contiguous in the whole bar, and when tested under the usual conditions specified in Sec. 151 gave the results tabulated below: Test- piece No. Heat treatment Tensile strength, lb. per sq. in. Elongation, per cent. Reduction of area, per cent. Heated for 10 minutes at 800°C. quenched in water; drawn at 640°C Heated for 30 minutes at 860°C. quenched in water; drawn at 670°C Heated at 880°C. for 1 hour quenched in water; drawn at 670°C Heated at 950°C. for 3 hours; cooled slowly to 850°C.; quenched in water; drawn at 670°C 99,800 96,700 92,400 101,400 12 16 22 18 47 The greater efficiency appears evident, step by step as the treat- ment becomes more energetic. But this is still clearer, especially as concerns the most complete elimination of transverse weakness as an effect of the more drastic treatment, when one examines the fracture of the test-pieces. Thus, Fig. 206 reproduces the break from test-piece No. 2; woody structure is highly developed. We have seen this defect to be character- istic of steels having poor properties when tested at right angles to the elongation produced by forging. The next figure (Fig. 207) gives the appearance of No. 3. Here also woody structure is still well developed although the mechanical properties are substantially improved, as can be seen from examining the table. 348 HEAT TREATMENT OF The inicrostniuture of the same test-piece is reproduced at 100 dia. in Fig. 208, showing that we now have a case analogous to that mentioned in Sec. 201; namely, where no inclusions apparently associated with the persistent grouping of ferrite elements into banded masses can be seen upon properly polished and etched sections when examined under mod- erate magnification, even though the ferrite segregation can be clearly seen. Finally, Fig. 209 shows the fracture of test-pieces No. 4. The woody structure has now disappeared almost entirely, and is replaced by a fibrous texture. Stratified accumulations of fei'rite have also been ehminated Fig. 206. — Woody structure in test-pieoe No. 2. by the energetic heat treatment, as is shown by Fig. 210, which repro- duces the microstructure of the same test-piece at an enlargement of 100 dia. 204. It sometimes happens that that summation of effects which we have grouped together in the term "transverse weakness" is felt in a very different measure upon the mechanical properties of steel, according to the special property considered. This may happen both for tests made upon longitudinal as well as transverse specimens, the latter, however, being more frequent. We shall start Ijy considering an example of the first case. In Sec. 160 we have studied the properties of cast metal after various heat treatments. Now we shall examine the behavior of the same steel after SOFT AND MEDIUM STEELS 349 Fig. 207. — Woody structure of test-piece No. 3. Fio. 208. — Ferrite banding associated with woody fracture of Fig. 207. X 100. 350 HEAT TREATMENT OF Fig. 209. — Fibrous fracture of test-piece No. 4. Fig. 210. — Microstructure corresponding to fibrous fracture of Fig. 209. SOFT AND MEDIUM STEELS 351 forging and various heat treatments. Inasmuch as we saw in the first place that this steel exhibited ingotism strongly and persistently due to the presence of emulsified oxidized inclusions, we must expect from reasoning advanced at the beginning of this chapter that, after forging, this metal would likewise present marked and persistent transverse weakness. Experience confirms fully such expectations. A 220 by 220-mm. square ingot of steel of the same composition as the one indicated in Sec. 160, was pressed into a 170 by 170-mm. square bloom. The cold bar was cut into four parts by sawing it along two planes passing through its axis and normal to the faces. Longitudinal test-pieces were then obtained in such a manner as to avoid the axial region of the ingot, usually the locus of segregated impurities. The successive heat treatments, however, were made upon the entire bar before the corresponding test-pieces were removed. The following table contains the results of a series of the usual physical tests upon longitudinal test-pieces, obtained after the heat treatments indicated in the second column of the table. No. Heat treatment Tension test Tensile strength, lb. ^er sq. in. ijlastic limit, lb. per sq. in. Elonga- tion, per cent. Resist- ance to impact (p) (Charpy test), kg.-m. per sq. cm. 1 2 I 3 None; steel as forged Heated for 4 hours at 800°C. followed by slow cooling Heated for 4 hours at 900°C. followed by slow cooling Quench in water from 900°C.; an- nealed at 600°C. for 4 hr Heated at 1100°C. for 4 hours; slowly cooled to 900°C.; quenched in water; annealed at 600°C. for 4 hours Heated and quenched as No. 5; annealed at 700°C. for 4 hours 105,500 103,300 101,000 118,200 123,700 102,700 57,600 59,000 61,900 74,800 84,300 64,700 20 17 20 20 18 26 1.7 2.5 4.1 5.7 4.9 6.4 Two transverse bars, one for tension and the other for impact were taken from the original forged bar, treated as indicated in line 4, and tested with the following results : 352 HEAT TREATMENT OF Tensile strength 115,500 lb. per square inch Elastic hmit 71,100 lb. per square inch Elongation 3 per cent. Resistance to impact (Charpy test). . 0.5 kg.-m. per square centimeter These results clearly show the existence of- high transverse weakness in accordance with expectations founded upon the behavior of the heat- treated ingot metal. If now the data obtained in the last table are compared among themselves and with those given in Sec. 160 for the same steel in the unforged condition, and also compared to the elongation and resistance to impact (both in a longitudinal direction) of forged well deoxidized normal steels not infected by overly large quantities of non-metallic inclusions,' we see immediately that the improvement in impact pro- duced by forging the impure steel is a great deal less than that caused by a like amount of work upon those other steels, especially when con- sidering the splendid longitudinal tensile strength and elongation ac- quired by the steel in question by reason of the forging and heat treatment (particularly No. 6 above) . 205. As I have already pointed out at the beginning of the previous section, the small improvement in one of the physical properties of a transversely weak forged steel, due to a heat treatment which simulta- neously produces marked improvement in the other properties of the same steel, may be exhibited in a far larger measure when the experimen- tation is done with transverse test-pieces. It is also opportune to indicate an example of this case. A medium hard steel, contained 0.40 per cent, carbon and 2 per cent, nickel, manufactured in the acid open-hearth furnace, was cast in octa- gonal ingots weighing about 10 tons each. The ingots were forged down to round bars about 300 mm. in diameter, which were afterwards bored and rough turned, and then subjected to a series of homogeneity quench- ings, followed by annealings. Inasmuch as the tests made upon the first pieces treated revealed high transverse weakness — due in this case to the presence of a great many very small highly oxidized sonims — it was necessary to apply repeated and very drastic homogeneity heat treatments to them before obtaining satisfactory results from static tension tests upon transverse pieces. The results of this very numerous series, made under the conditions indicated in Sec. 151, varied between the following limits: Tensile strength 93,900 to 1 12,400 lb. per square inch. Elastic limit 62,600 to 81,100 lb. per square inch. Elongation 18 to 23 per cent. Notwithstanding these results, which may be considered as entirely satisfactory, all bending tests were very poor. These were made under the conditions specified in Sec. 151 upon transverse coupons taken im- ' See for instance Sec. 179 to 183. SOFT AND MEDIUM STEELS 353 mediately alongside those used for tension. Nor was it possible to obtain an appreciable improvement of these results, even by subjecting the steel to further homogeneity heat treatments. In spite of all efforts, all the test-pieces subjected to bending broke with a strongly woody fracture at low deflections (included angles from 100° to 110°). The contrast between the satisfactory appearance, fibrous fracture, and good necking of test-pieces broken under tension and the woody frac- ture under bending is well shown in Fig. 211. 206. 1 have already pointed out the most important and probable causes of the fact that the reheatings contained in a more or less complex Fig. 211. — Bending and tension tests on medium hard nickel steel. homogeneity heat treatment cannot be prolonged beyond certain limits of time nor pushed bej^ond a certain limit of temperature without causing overheating. As then noted, the useful results of a slight over- heating^consist in the more perfect diffusion of the various chemical con- stituents into the y solid solution; as the treatment becomes more intense and prolonged the final effect becomes progressively less and less ad- vantageous and afterwards actually dangerous, even ending in a true burning (see Sec. 25, 26 and 83). It is not necessary to repeat here nor to dwell further upon the origin of phenomena of overheating. It forms an important restriction in all those cases where the most efficient possible homogeneity heat treat- ments must be applied. On the other hand, some experimental data may be presented which are apt to furnish a precise idea of the effects which these phenomena may produce. In order to exclude any doubt that "burning" instead of a simple 23 354 HEAT TREATMENT OF overheating might have taken place, I have selected a heating process in which the temperature was always maintained far below the minimum limit necessary to produce a true burning in a steel of that composition. In this case, therefore, the evil effects of over heating ensues only by reason of the length of heating. A low carbon nickel-steel of the following composition was forged in 200 mm. round bars: Carbon . 20 per cent. Manganese 0. 60 per cent. Silicon . 03 per cent. Sulphur 0.01 per cent. Phosphorus . 06 per cent. Nickel 2 . 98 per cent. It was subjected to the heat treatments noted in the second column of the table below. Inasmuch as these treatments do not include any heatings to a temperature above 1050°C. and inasmuch as a steel of the composition indicated does not pass the liquidus (suffer a true burn- ing) except at temperatures above 1250°C., especially after forging, no doubt may arise but that any dangerous effect produced by thermal manipulation is simply due to long stay at temperatures well above the transformation. Tests were made upon longitudinal test-pieces. The effects of overheating are shown by the decrease in elongation and still more by the lessened resistance to impact. Heat treatment Tension tests treat- ment No. Tensile strength, lb. per sq. in. Elongation, per cent. to impact (p) (Charpy test), kg.-m. per sq. cm. 1 Heated at 800°C. for 2 hours; quenched in water; drawn at 600°C 84,500 86,500 97,400 30 18 14 22 1 2 Heated at 1050°C. for 200 hours; followed by slow cooling in the furnace to 800°C.; quenched in water; reheated at 800°C. for 1 hour, quenched in water; drawn at 500°C 11 5 3 Heated at ]060°C. for 400 hours; slowly cooled in the furnace to 800°C.; quenched in water; re- heated at 1000°C.; slowly cooled in the furnace to 800°C.; quenched in water; drawn at 600°C 5.1 SOFT AND MEDIUM STEELS 355 207. Before ending this series of examples, I deem it opportune to present some observations concerning at least one instance not rare in practice, which is difficult to explain by means of the general notions which we possess at the present time and which I have tried to summarize in the previous pages. If nothing else, these observations will suffice to confirm further the fact that our knowledge about the phenomena upon which normalizing processes are based is still very incomplete and imperfect. In the case which I shall mention, the physical properties of the pieces examined are analogous to those of Sec. 201. In both instances the piece exhibits high transverse weakness in one part while this defect is practi- cally missing in the remaining parts. The steel in question contained 0.44 per cent, carbon, 0.57 per cent, manganese, and 0.50 per cent, chro- mixmi. It was cast in 550 by 550 mm. ingots. From the lower haK of each ingot, round bars 220 mm. dia. and from five to six meters long were obtained by forging under the press. A piece, 1 to l^-i m. long, was then cut from each end of the forged bars. One of such bars, cut, turned, quenched in water after heating for 10 hours at 850°C. and drawn at 650°C. was subjected to transverse tension tests taken from each of its ends, respectively indicated with the letters A and B. The results were as follows: Tensile strength, Elastic limit, Elongation, lb. per sq. in, lb, per sq, in. per cent. End A 102,800 74,000 21 End B 80,900 46,000 2 Having repeated the same heat treatment and new transverse test- pieces obtained from metal contiguous to that furnishing the preceding ones, tension tests gave the' following results : Tensile strength, Elastic limit, Elongation, lb. per aq. in. lb. per sq. in. per cent. End A 95,700 69,700 19 End B 66,700 45,500 2 The most drastic heat treatments, repeated upon the same bar, did not substantially change the mechanical properties of the metal at its two extremities, nor did accurate chemical analysis reveal appreciable differences in composition. Figure 212 reproduces at an enlargement of 150 dia., the structure at end A after the second heat treatment, while Fig. 213 at the same magni- fication is the structure of the metal at end B, also after the second heat treatment. These structures were not modified very much by all the subsequent heat treatments. It was demonstrated in this case that the permanence of high trans- verse weakness at end B is accompanied by a characteristic structure, which is quite different in microscopic appearance from the lamellar 356 HEAT TREATMENT OF structure which we have examined many times and which usually ac- companies to a certain extent the phenomenon of transverse weakness. Crystalline elements delimited by rectilinear boundaries appear even as Fig. 212. — Microstructure of sound out! of forged bar. X 150. Fig. 213.- -Characteristic and unexplained microstructure associated with transverse weakness, one meter from Fig. 212. X 150. much more abnormal structures, since they have been observed only in extremely rare cases, and always accompanied by defects in physical properties analogous to those now described. This remark is made as a SOFT AND MEDIUM STEELS 357 result of experience with an enormous number of pieces manufactured in practically the same way as that described for the bar to which the data above mentioned refer, and of steel of practically the same composition. An examination of the few cases analogous to the one indicated which have been discovered appears to justify the opinion that the laws govern- ing the formation and persistence of such crystalline aggregations as shown in Fig. 213 are quite different from those which are summarized in the ordinary equilibrium diagram, or in the general discussion developed in the previous chapters. As I have already pointed out, this fact constitutes a further proof of the actual imperfection of our knowledge in regard to the phenomena we are now discussing. 208. An accurate comparison of all experimental data presented in this last part of our study with those given in Part IV, a comparison which should be conducted under the guidance of the general considerations developed in the first three Parts, would allow one to get even a better idea of the significance and the technical value of such experiments. But inasmuch as this detailed comparison would mean a lengthy exposi- tion, I must permit the reader to work it out for each one of his problems. If the general considerations developed in this book are well un- derstood, one who attempts such analytical work to solve those of his every day problems as have clearly established fundamental data, will recognize that such work does not present great difficulties. It will then be easy to verify its utility on the basis of the actual results obtained. As I have already said, the data presented in this study are suf- ficient so that the great majority of the industrial problems may be analyzed and found to bear upon phenomena none of which are entirely new and such as to confuse the searcher with unforeseen facts. However, I deem it opportune again to call the attention of the reader to two characteristic sides of some observations, doubtless inherent to facts and reasoning already given, yet aspects which must be taken into especial account in the most general application of the ideas which we have developed and illustrated only by particular examples. Inasmuch as I wish to recall the attention to particular groups of ideas already stated, I shall endeavor to be very concise. 209. An accurate comparison of the variations caused by certain heat treatments in the physical properties of the types of steels under examination, with the corresponding variations which the same treatments cause in the fracture and the microstructure, estabhshes the fact that in the greater number of cases the appearance of the surface of fracture cannot be taken as an indication of the state of homogeneity and of the "quality" of the steel, except in a merely approximate sense. The widest experience confirms this conclusion. In fact it is not 358 HEAT TREATMENT OF possible under the present state of our knowledge to establish a precise and sure relation between the quality of a steel and its fracture. Nor at any rate could such relation be at all general since we have already seen examples of the fact that the appearance of the fractured surface varies a great deal for the same steel when broken in different ways. For some particular cases — especially soft and medium cast steels subjected to homogeneity heat treatments' — I have had occasion to show how granular crystalhne fracture exhibits itself as a consequence of the presence and the development of determined systems of metallog- raphically well defined structural elements, such as a large ferrite network. But the generality of such a co-relation is immediately excluded in the most absolute way by the frequent appearance of coarsely crystalline fractures in steels in which no trace of those microscopic systems exists. In particular, these observations demonstrate that the remarks in Sec. 127 to 129 have a value only for the type of steel there examined, and cannot be applied for instance to steels having very low trans- formation points, nor to many forged or rolled steels; in fact we know that in these steels the large network of ferrite does not exist, never- theless they very frequently give coarsely crystalline surfaces of fracture. Notwithstanding the lack of generality furnished by fracture examina- tions in the study of actual structure, such inspection may lead to useful practical conclusions whenever it is applied to numerous samples of one particular type of steel, constantly subjected to analogous treatments, especially if this examination is to be utilized essentially as a means of comparison between the results obtained in the various cases after a change in the history of the specimen. A trained eye may then rapidly gather very useful indications even by such an empirical method of observation. Beside the many examples which involve various appearances of fractured surfaces mentioned in the last two parts of our study, I shall also note here the extended use that it is commonly made of such a method of investigation in the control of armor plate manufacture. In this special branch of metallurgy the maker distinguishes a great number of types of fracture, different and well defined. As an example of the subtle distinctions. Fig. 214 shows at about ^■i of its natural size the mediocre structure termed "fibrous with granular traces," of an armor plate after it has been forged, cemented, and subjected to the preliminary heat treatment only. 210. In Sec. 208 I have observed that the individual reader may apply the principles developed in this book to solve his individual problems. For instance, certain criteria must be borne in mind whenever one must decide whether a simple cast steel (properly heat treated) can be used for the construction of a certain machine part, or whether it is necessary or ' As for instance, in Sec. 127 to 129. SOFT AND MEDIUM STEELS 359 more desirable to employ a steel which has been subjected to hot-work be- fore the heat treatment. This may be done by a comparison of the data given in Parts IV and V. Now, in the application of such criteria, it is of prime necessity to take careful account of the special eiJects caused by transverse weakness. In fact we have seen that, if it is true that in many cases normal and constant properties of high tensile strength together with high toughness may be obtained with greater ease in forged or rolled steels by means of homo- geneity heat treatments, it is also true that this greater facility is re- FiG. 214. — Fibrous frafturc, with granular traces, of armor plate. X stricted to the cases in which the mechanical properties are determined by longitudinal tests. On the other hand, when desiring good mechan- ical properties when tested "across the grain, " the data given in the last chapters when compared with those mentioned in Part IV demonstrate that the advantage remains all in favor of cast steels after proper thermal manipulation. Differences in this regard become still greater when we take account of the probabilities that dangerous accidental h-regularities in physical properties may come to light, as described for instance in Sec. 165, due principally to impurities contained in the steel. In fact, while such 360 HEAT TREATMENT OF SOFT AND MEDIUM STEELS irregularities may generally be efficaciously eliminated in forged or rolled steels when tested in a longitudinal direction, they ordinarily appear with high seriousness and persistency, as compared to steel simply cast and heat treated. Naturally this is true if other conditions are equal (especially the purity of the steel) and provided the steels selected are in both cases of suitable composition for the manufacturing processes and treatments to which they must be subjected. We know that this is mainly due in many cases to the fact that the non-metallic inclusions in the cast steel have a spherical form, while in forged or rolled steel they assume the form of threads or lamellse acting as sharp internal notches and therefore constitute dangerous "primers" for ultimate fracture. It follows that in many cases, and above all when it is impossible to forge the piece in such a way that the working stresses fall in a longitu- dinal sense, the rational employment of heat treated steel castings may preseht relevant advantages in comparison with steel forgings, not only from the viewpoint of rapidity of construction, but also of resistance to stresses and its immunity from accidental fragility. BIBLIOGRAPHY OF SOME BOOKS WHICH DWELL ON SUBJECTS RELATED TO THOSE DISCUSSED IN THIS VOLUME H. M. Howe.— The Metallography of Steel and Cast Iron. New York, McGraw- Hill Book Company, 1916. Thorough and very clear general discussion of all phenomena connected with the crystallization, hot working, and heat treatment of steels. It is certainly the most complete and profound treatise among those which dwell upon this branch of metal- lurgy. L. Grenet. — Trempe, Recuit, Cementation et Conditions d'emploi des Aciers. Paris, Ch. B6ranger, 1911. Complete technical study of steel heat treatment. It contains also some interest- ing practical data upon homogeneity heat treatments. Albert Sattveur. — The Metallography and Heat Treatment of Iron and Steel. Cambridge, Mass., U. S. A., Sauveur and Boylston, 1916, Second Edition, Fourth Thousand. General study of metallography with special reference to its applications to heat treatment of steel. It also contains some data on preliminary heat-treatment. GusTAV Tammann. — ^Lehrbuch der Metallographie. Leipzig, Verlag von Leopold Voss, 1914. Classical treatise on the elements of metallography. That part which bears upon the effects of mechanical working upon the properties of metals is practicularly interesting. Denison K. Bullens. — Steel and its Heat Treatment. New York, John Wiley and Sons, 1916. General study of the heat treatment of various types of steels, mainly from a practical point of view. The subject is treated very rationally and clearly, and is based upon a well defined scientific foundation. It mentions preliminary heat treatment. W. Guertler. — Metallographie. Berlin, Gebruder Borhtrager, 1912-13. See especially: Erster Band, Die Konstitution; Zweiter Teil; Heft 1, Die Konstitution des Systemes Eisen-Kohlenstoff, sowie der sonstigen binaren Kohlenstoff-Legierungen. A very extensive metallographie study of steels and cast irons. Especially inter- esting for its study of the structural elements comprising the iron-carbon alloys. A. Martens. — Handbuch der MateriaUenkunde fur den Maschinenbau. Zweiter Teil; Die Technisch wichtigen Eigenschaften der Metalle und Legierungen von E. Hbyn. Berlin, Verlag von Julius Springer, 1912. Complete study, crystallographic and metallographie, of the general properties of various metals and their alloys, particularly as related to their physical and mechanical characters. 361 362 BIBLIOGRAPHY L. GniLLET. — Traitements thermiques des produitsmetallurgiques: Trempe, Recuit, Revenu. Paris. H. Dunod et E. Pinat, 1909. General theoretical and practical study of the heat treatment of the principal metals and their alloys, with special reference to steels. It also contains remarks upon homogeneity heat treatments. It is especially valuable for numerous practical data, accurately classified. Otto Thallner. — Konstruktionsstahl. Freiberg i. S., Verlag von Craz & Gerlach, 1904. Essentially a practical study of the mechanical properties of machine steels as affected by heat treatment. The effects of homogeneity heat treatments are men- tioned, but only from a practical point of view. G. B. Upton. — The Structure and Properties of the more common Materials of Construction. New York, John Wiley & Sons, 1916. The second part, almost entirely devoted to steel, contains a very clear discussion of the heat treatment of plain carbon and special steels, with interesting remarks on homogeneity heat treatments. G. Mars. — Die Spezialstahle. Stuttgart, Verlag von Ferdinand Enke, 1912. Detailed study of the properties and uses of the special steels. Chapter V con- tains a brief but clear mention of the phenomena upon which homogeneity heat treatments are founded. The work contains a wealth of interesting technical data. W. RosBNHAiN. — An Introduction to the Study of Physical Metallurgy. London, Constable & Company, 1914. A general but very clear discussion of the fundamental principles of metallography and their applications. E. F. Lake. — Composition and Heat Treatment of Steel. New York, McGraw-Hill Book Company, 1911. General study, mainly practical, of fabrication and treatment. F. GiOLiTTi. — The Cementation of Iron and Steel. Translated from the Italian by Joseph W. Richards and Chas. A. Rouiller, New York, McGraw-Hill Book Co., 1914. Diffusion of carbon in soUd steel, particularly in reference to carburizing solid metal and to the heat-treatment of partly case-hardened steel. C. A. Edwards. — The Physico-Chemical properties of Steel. London, Charles Griffin & Co., 1916. An elementary study of metallography. The discussion is very clear, methodical, rational and complete, although brief. The bibliography is abundant. In a second edition of this very good work, it would be well to revise the sometimes incorrect practical data covering the composition, working, and properties of the various types of commercial steels. M. Lbvi-Malvano. — ^Tempera e cementazione dell'acciaio. Milano, Hoepli, 1917. Very clear and synthetical treatment of the phenomena upon which heat treat- ment is founded. It also contains interesting practical data related to homogeneity heat treatments. John Howe Hall. — The Steel Foundry. New York, McGraw-Hill Book Company 1914. General study of the manufacture of steel castings. Chapter IX contains interest- ing data (though mentioned only in outline) on the heat-treatment of steel castings. BIBLIOGRAPHY 363 Erik Oberg. — Heat Treatment of Steel. New York, The Industrial Press, 1917. Practical treatise on the heat treatment of steels in general, containing much interesting technical data. More details in regard to some of the principal arguments treated in this volume can be found in the following publications: 1. F. GiOLiTTi. — Sulla cristalUzzazione dell'acciaio. La Metallurgia Italiana, Vol. V, 1913, fasc. 3.° 2. F. GiOLiTTi and N. Boybr. — Sulla cristallizzazione dell'acciaio, Nota II. La Metallurgia Italiana, Vol. V, 1913, fasc. 5.° 3. F. GiouTTi. and P. Forcblla. — Sulla cristalUzzazione dell'acciaio, Nota III. Sulla relazioni che passano fra il sistema delle dendriti ed il reticolo di ferrite negli acciai ipoeutectice. La Metallurgia Italiana, Vol. VI, Novem- ber, 1914. 4. F. GiOLiTTi. — Sulla cristallizzazione dell'acciaio, Nota IV, Osservazioni intorno ad una delle cause che possono dar luogo a lacerazioni internenei lingotti d' acciaio di grandi dimensioni. L' Industria, Vol. XXIX, No. 16 and 17, Feb- ruary, 1915. 5. F. GiOLiTTi and S. Zublena. — Sul comportamento delle scorie occluse nell'- acciaio, Fabbricato su suola, Nota I, Annali di Chimica Applicata, I, Vol. II, No. 7-8, pp. 218-245. 6. F. GioiiiTTi and G. Tavanti. — Sul comportamento delle scorie occluse nell'- acciaio, Nota II, Annali di Chimica Applicata, Vol. II, No. 11 and 12. 7. F. GioiiiTTi and L. Sobbero. — Sul trattamento termico degli acciai speciali, Nota I, La Metallurgia Italiana, 28, February, 1917. 8. F. GiOLiTTi, — Crystallography of Alpha, and Beta Iron, Chemical and Metal- lurgical Engineering, Mar. 31, 1920, Vol. 22, p. 585. 9. F. GiOLiTTi. — Genesis of Ferrite, Chemical and Metallurgical Engineering, April 21, 1920, Vol. 22, p. 737. 10. F. GiOLiTTi. — ^Relationship between Dendritic Structure and Ferrite Mesh, Chemical and Metallurgical Engineering, May 19, 1920, Vol, 22, p. 921. 11. F. GioLiTTi. — Structural Effects of Slag Occluded in Open-Hearth Steel, Chemi- cal and Metallurgical Engineering, 1921, Vol. 24. INDEX Air-hardening steels, 114 Allotropic transformation. See Trans- formation. Alloy steel, /S and a ferrite indistin- guishable, 294 complex carbides, 5 equilibrium diagram, 50 soUdification, 37 suppression of transformation, 294 Alloy steels. See also Nickel steel, etc. advantages for castings, 198 constitution on slow cooling, 97 diagram for transformation, 94 high-strength, 199, 243 incomplete quenching, 121 manufacture, 242 mild quenching, 250 to match heat treatment, 117 Alpha iron. See Ferrite. Amorphous theory, 283 Anisotropy. See also Heterogeneity. causes, 283 in hot worked metal, 282 Annealing, definition, 328 effects, 25 impure steel, 272 limitations in time and temperature, 108, 115 preliminary, 214 prolonged, 272 temperature limits, 29 Anneal vs. draw, effect on final structure, 303 vs. quenchings, 333 Armor plate, 6 fracture studies, 358 heat treatment, 316 Austenite, carbon content, 71 e.s. composition of residuum, 83 decomposition of heterogeneous, 84 e.s., 91 decomposition of homogeneous, 60 e.s., 73 decomposition of uniform, 1'55, 213 Austenite, instantaneous quantity, 63 relation to dendritic system, 191 solid solution, q.v. supersaturated, 68, 91 thermal analysis, 92 B Band, of variant composition, 22 affected by cooling velocity, 24 relation to diffusion, 27 relation to sonims, 274 Bending tests, 229 impaired by inclusions, 353 special castings, 253 Beta iron, 53 crystalline habit, 157 indistinguishable in alloy steel, 294 Bibliography, 361 Binary system, 9 e.s., 19 e.s. end of solidification, 21 secondary crystallization, 60 e.s., 84 e.s. Burned steel, microstructure, 110 physical properties, 110 Burning, 30, 36 Carbon in iron. See also Cementite. diffusion, q.v. distribution in austenite, 71 e.s. ferrite saturation, 78, 89 specific influence, 39 Carbon oxides, aids to diffusion, 28, 38 Carbon steel, equilibrium diagram, 50 non-eubectoid, 63 Carbon steel castings 0.08 per cent, carbon for castings, 194 heat treatment, 194, 224 ingotism, 198, 224 macrostructure, 152, 196 microstructure, 198 physical properties, 194, 224 365 366 INDEX Carbon steel castings, 0.23 per cent, carbon forging, 230 heat treatment, 216 e.s., 230 heterogeneity, 221 microstructure, 222, 233 physical properties, 216 e.s., 230 thermal hysteresis, 263 0.33 per cent, carbon annealed, 205 forging, 231 heat treatment, 210, 230 ingot structure, 172, 205 macrostructure, 205 microstructure, 205, 233 physical properties, 211, 230 0.47 per cent, carbon emulsified inclusions, 227 heat treatment, 225 ingotism, 227 microstructure, 226 physical properties, 225 impure 0.49 per cent, carbon heat treatment, 267 ingotism, 270 microstructure, 268 physical properties, 267 prolonged anneals, 272 Carbon steel forgings 0.08 per cent, carbon heat treatment, 306 microstructure, 307 physical properties, 306 0.30 per cent, carbon transversely weak, 338, 344 0.47 per cent, carbon heat treatment, 351 impact test, 352 transverse properties, 351 Case hardening, 2, 28 Casting practice, influence on heat treat- ment, 107 relation to heterogeneity, 106 Castings, best compositions, 193 e.s. carbon steel, 152, 172, 196, 205, 225 cold worked, 263 compared to forgings, 230, 231 high strength, 243 homogeneity heat treatment, 145 e.s. massive, 225 Castings, nickel steel, 168, 241, 243, 263 nickel-chromium steel, 301 substituted for forgings, 243 suppression of ferrite, 171 Cementation. See Case hardening. Cementite, coexisting with primary fer- rite, 92 solubility in alpha iron, 62, 78, 89 Checks. See Cracks. Chromium steel forgings 0.50 per cent. Cr, 0.50 per cent. C heat treatment, 347 microstructure, 349 transverse properties, 347 woody structure, 348 Cleavage planes, appearance in micro- sections, 309 Coefficient of equivalence, 48 Cold work, castings, 263 causes grain growth, 290 grain growth ensuing, 126 gun shield, 293 Complex alloys, solidification, 37 e.s. transformation, 93 e.s. Complex carbides, 5 Complex constituent, 37 Composition, at nuclei, 23 changes in solid steel, 1 fictitious, 22 in ternary crystallite, 34 of austenite, 71 e.s. of ferrite, 62, 78, 89 of residuial austenite, 83 relation to diffusion, 27 variant, 22 variation within crystals, 19, 22 variations, 120 Cooling velocity. See also Quenching. affects solidification, 4, 23, 35 affects transformation temperature, 100 Cooling velocities, slow cooling and fer- rite aggregation, 201 through transformation, 241 variable in same piece, 143 Corrosion, 140 Cracks, hardening, 317 in forging, 148 Crank hanger, manufacture, 332 stresses, 332 Crystalline individual, definition, 10 variable composition, 19 Crystalline slip. See SUp. INDEX 367 Crystallization. See primary crystalliza- tion and secondary crystalliza- tion. Belaiew's "large crystallization," 146 columnar, 147 disturbed, 159 parasitic centers, 149 undercooled liquids, 150 D Dead-soft steels, heat treatment, 120 Decomposition of austenite. See Trans- formation and secondary crys- tallization. Delta iron, 52 Dendrites. See also Primary crystal- lization. in ingots, 162 persistency of form, 166, 179, 187 relation to slip bands, 165 suppression, 178 Diffusion, amount, 27 carbon in f errite, 67 critical temperature, 29 determines amount of ferrite, 66, 90 during heating and cooling, 26 effect of chemical composition, 24, 27 effect on secondary crystallization, 52 e.s. Pick's law, 27 from residuum, 4 gaseous aids, 28, 38 gold into lead, 29 impurities, 97 incomplete, in ternary system, 49 influence on sohdus, 12 lacking in dendritic structure, 187 only above Acs, 6, 27 range, 15 relation to heat treatment, 103 e.s. speed, 14, 26 e.s., 38, 60, 90 unhindered by twinning, 124 Draw, necessity of adequate, 120 Draw iiS. anneal, effect on final structure, 303 Drawing. See Wire drawing. Drop-forgings, crystallography, 295 Ductihty, reveals heterogeneity, 221 E Elastic ratio, 121 Emulsified inclusions, Inclusions. 127. See also Emulsified inclusions, flocculation, 134 physiological defects, 132 Equilibrium. See also Partial equilib- rium. in austenite, 60 e.s. in binary system, 9 e.s. in ternary system, 32 e.s. precedent conditions, 17 Equilibrium diagrams, binary solid solu- tion, 11, 21 fictitious, 40 e.s. for carbon steels, 50 iron : carbon, 56 locating quenching phenomena, 118 quaternary system, 40, 51 ternary system, 32 Equivalent values, 48 Eutectoid. See Pearlite. F Ferrite, after slow cooling, 201 banded, 340, 348 carbon diffusion, 67 coexisting with primary cementite, 92 columnar, 285, 302 crystalline habit, 155, 191 distribution after identical heat treatment, 302 easily modified, 166 from heterogeneous austenite, 87 from undercooled austenite, 76 genesis, 191 growth on annealing, 309 in alloy steel, 294 influence on ingotism, 165 in ingot, 161 instantaneous quantity, 63 interrupted meshes, 202 isolated particles, 176 lamellar, 285 meshes, 146 network in forgings, 295 partition, 203 relation to fracture, 165 saturated in carbon, 78 segregation, 203 sheaths about inclusions, 270 solubility of cementite, 62, 78, 89 striations, 309 suppression, J.71, 174, 316 surrounding inclusions, 130 ultimate quantity, 66 e.s. 368 INDEX Ferro-alloys, diffusion in liquid bath, 15 Ferronite, 58 Fiber in hot worked steel, 282 Fick's law, 27 Fictitious composition in binary system, 22 in complex system, 40 Fictitious equilibrium systems, 40, 47, 93 at transformation, 94 Fictitious representative point, 96 Final heat treaments, 118. See also Quenching and Drawing. , carbide segregation, 26 definition, 7 may coincide with prehminary 120 precautions to be observed, 119 principles, 101 Forging. See Hot work. Forgings, carbon steel, 306 compared to castings, 230, 231 cracks, 148 nickel steel, 303, 317, 323, 333 e.s. nickel-chromium steel, 298, 320, 339 on mandrels, 332 replaced by castings, 243 Fractures, armor plate, 358 granular, 185 ingot, 165 limitations to indications, 357 practical utility, 183 relation to microstructure, 184, 192 relation to physical tests, 185 utility, 358 woody, 348 Fragility. See Toughness. G Gas in steel, aid to diffusion, 28, 38 Grain fragmentation, by hot work, 125 Grain growth, 31, 108 cold work induces, 126 due to internal stress, 293 effect on mechanical properties, 291 in casting, 309 in hard steels, 291 laws, 289 limited to very soft steels, 290 Grain size, influence on anneal, 61 Gun shield, rolled too cold, 293 Gun tube, treatment and tests, 323 H Hard steel, disturbed crystallization, 294 grain growth, 290 Hardening, 101. See also Quenching, cracks, avoidance, 317 soft spots, 317 Heat treater, relationships with melter, 107 Heat treatment, annealing, q.v. definition, 98 hardening, q.v. soaking, 98 Heterogeneity. See also Segregation, aids transformation, 113 cause, 106, 283 effect of forging, 279 from slowly cooled uniform aus- tenite, 213 induced by quenching massive pieces, 310 intracrystalline, 18, 19, 22 cause, 22 in austenite, 84 e.s. in ternary alloys, 34 minimized by normalizing, 105 other than in carbon, 271 shown by ductility, 221 shown under microscope, 215 worked metal, 282 Heterogeneous equilibrium, 62, 96 Homogeneity, aided by quenching, 114 high temperature annealing, 115 hinders transformation, 113 Homogeneity heat treatment, affected by casting practice, 107 affected by diffusion, 103 e.s. affected by hot work, 124 affected by inclusions, 126 affected by nature of transforma- tion, 113 affected by segregation, 120 after wire drawing, 293 armor plate, 316 carbon steel forgings, 306, 351 e.s. castings, 145 e.s. chromium steel, 347 complex, 249 definition, 6 dirty steel, 338 disturbaiices in general, 123 e.s. effect of inclusions, 266 e.s. INDEX 369 Homogeneity heat treatment effect of plastic deformations, 279 forged and rolled steels, 277 e.g., 298 e.s. fundamental problem, 7 gun tube, 323 high temperature limitations, 30, 36 impure 0.49 per cent. C steel casting, 267 in one operation, 329 influence of grain size, 61 longitudinal properties, 320 e.s. massive pieces, 315, 328 matched by alloy steel, 117 modification by hot work, 277 e.s. most energetic, 176 nickel steel forgings, 303, 317, 323, 333, 342, 352 nickel-chromium steel, 299, 320 e.s., 329 operations, 214 overlap with final heat treatments, 120 principles, 24 repeated, 275 simplest type, 119 2 per cent. Ni steel casting, 160 e.s., 244 1.8 per cent. Ni steel casting, 241 0.47 per cent, carbon steel casting, 225 0.33 per cent, carbon steel casting, 210, 230 0.23 per cent, carbon steel casting, 210, 230 0.08 per cent, carbon steel casting, 194, 224 Hypoeutectoid steel, end of transforma- tion, 77 Hysteresis, high nickel steel, 317 in various steels, 263 thermal, 55, 101 utilization, 119, 317 Hot work, changes nuclear packing, 279 columnar structure, 285 crystalline movement, 124 disturbs recrystallization, 280 effect on heat treatment, 124 fragments structure, 279 grain fragmentation, 125 improves impact strength, 309 increases toughness, 288 24 Hot work, lamellar structure, 285 large pieces, 281 lessens apparent heterogeneity, 279 modification of heat treatment, 277 persistence of network, 295 pressing, 313 production of fiber, 282 re-orients crystals, 279 Impact strength, impaired by sonims, 362 improved by forging, 309 Impact tension test, 229 Impact tests, 229 Impaired quenching. See Quenching. Impurities. See also Inclusions. and inheritance, 86 flocculatipn, 134 imperfect diffusion, 97 Inclusions, 126 e.s. after excessive heating, 30 and inheritance, 86 attraction for ferrite, 130 cause corrosion, 140 effect on heat treatment, 126, 266 e.s. emulsified, 127 emulsified sonims, 287 ferrite sheaths, 270, 275 foreign, 229 harmless, 339 influence on microstructure, 129 in massive 0.47 per cent. C steel casting, 227 lamellar, 141 local effects, 140 metalUc, 229 nature, 127, 132 non-metallic, 229 oxidized, 336, 341 pathological defects, 132 produce ingotism, 275 produce transverse weakness, 287, 330 properties, 127 reduce bending strength, 363 reduce impact strength, 362 refractories, 140 slag, 137 structural impurities, 229 transverse weakness, 141 Incomplete quenching. See Quenching. 370 INDEX Ingots. See also Massive ingots. Ingot, columnar crystals, 147 ferrite, 161 ferrite suppression, 171, 174 solidification, 147 wall of pipe, 151 Ingotism, in 0.08 per cent. C steel, 198, 224 in 0.33 per cent. C steel, 172 in impure 0.49 per cent. C steel, 270 in massive 0.47 per cent. C steel, 227 nature, 160, 175 physical properties, 188 relation to sonims, 275 structural elements, 165 Ingotism and transverse weakness, 284 Inheritance, 1 due to impurities, 86 structural orientation, 284 Insoluble field, 39, 41 Internal stresses, 116 cause of grain growth, 293 effect on structure of massive pieces, 310 Intracrystalline heterogeneity. See Heterogeneity. Iron carbide. See Cementite. Liquation, 2, 99 excessive, 120 Liq nidus, definition, 17 in quaternary system, 42 in ternary system, 32 of binary system in equilibrium, 11 of transformation, 72 Longitudinal properties, effect of heat treatment, 320 e.s. M Machine design, limitations, 277 Machining, special castings, 257 Macrostructure, due to nuclear spacing, 9 Malleable iron, decarburization, 2 Mandrels, 332 Mass of piece affects transformation, 113, 225 e.s. Massive castings, properties, 225 Massive ingots, heat treatment, 315 internal structure, 310, 313 quenching, 309, 328 sensitive to forging, 281 variable structure with depth, 312 Mechanical treatment, effect on crystal structure, 2 Medium hard special steel. See 2 per cent, nickel steel. Melter, relationships with heat treater Microscopy, cleavage planes, 309 massive holohedral crystals, 356 transversely weak steels, 287 Microstructure due to crystalline growth, 9 influenced by sonims, 129 relation to fracture, 184 Mild steel, disturbed crystallization, 294 grain growth, 290 impact enhanced by forging, 309 See also Soft steels and carbon steels. thermal hysteresis, 263 Mixed crystals. See Solid solutions. Mother liquor, diffusion in, 15 instantaneous quantity, 16, 20 N Network, Belaiew's "large crystalliza- tion," 146, 201 interrupted meshes, 202 relation to sonims, 275 reticular, 159 Nickel-chromium castings, 301 2.53 per cent. Ni, 0.85 per cent. Cr, fracture, 302 heat treatment, 299 microstructure, 300 physical properties, 298 Nickel-chromium forgings, 0.50 per cent. Ni, 0.50 per cent. Cr, 0.45 per cent. C heat treatment, 322 physical properties, 323 2.5 per cent. Ni, 1 per cent. Cr, 0.20 per cent. C heat treatment, 320 microstructure, 321 physical properties, 322 2.53 per cent. Ni, 0.85 per cent. Cr, 0.28 per cent. C fracture, 302 heat treatment, 299 microstructure, 300 physical properties, 298 2.8 per cent. Ni, 0.8 per cent. Cr, 0.25 per cent. C INDEX 371 Nickel-chromium forgings, heat treat- ment, 329 physical properties, 329 Nickel steel, 5.8 per cent. Ni, 0.33 per cent. C quenching, 318 thermal analysis, 318 Nickel steel castings, 1.76 per cent. Ni, 0.26 per cent. C heat treatment, 241 physical properties, 241 2.0 per cent. Ni, 0.42 per cent. C homogeneity heat treatment, 160, 167 e.s., 176, 191 limitations, 193 physical properties, 180-183, 193 e.s. thermal analysis, 168, 263 medium hard, bending tests, 253 castings of great strength, 243 heat treatment, 244 machining properties, 257 physical properties, 243 Nickel steel forgings, 2.0 per cent. Ni, 0.34 per cent. C heat treatment, 303 microstructure, 304 physical properties, 303 2.0 per cent. Ni, 0.40 per cent. C transverse properties, 352 3 per cent. Ni, 0.20 per cent. C overheating, 354 4 per cent. Ni, 0.16 per cent. C heat treatment, 342 microstructure, 343 transverse properties, 342 medium hard, fracture, 325 heat treatment, 323, 333 e.s. microstructure, 324 physical properties, 323, 333 e.s. repeated quenching, 335 Non-eutectoid steel, 63 Normalization. See Homogeneity, Heat treatment and Annealing. Nuclear packing, and speed of cooling, 35 changed by hot work, 279 effect on diffusion, 24, 27 e.s. in secondary crystallization, 65 responsible for macrostructure, 9 Nuclei of crystallization, 9, 13 composition at, 23 orientative effect, 88 primary and secondary, 85 Orientation, 31 by pre-existing crystals, 88 changed by forging, 279 Overheated steel, 109 Overheating, 30, 36 cure, 211 physical damage, 353 Oxidation, causes transverse weakness, 336 Oxide in steel, 2 aids to diffusion, 28, 38 from excessive heating, 30, 36 Oxygen pressure, inclusions and metal, 130 Partial equilibrium, in binary system, 19 e.s. PearUte, laws governing structure, 82 location, 79 mechanism of formation, 79 variation in carbon content, 80 Pearlite kernels, influence on ingotism, 165 Phase changes. See Transformation. Physical properties, after grain growth, 291 at various depths, 312 average, 228 due to incomplete quenching, 121 ingotism, 160, 175, 188 normal, 228 relation to crystallization systems, 189 relation to fracture, 184 varying in one piece, 338 Physical tests, 229 Pipe, wall, 151 Plastic deformation. See Cold work. effect on normalizing, 279 Preliminary heat treatment. See Homo- geneity heat treatment and Annealing. Pressing. See Hot work, effects, 313 372 INDEX Primary crystallization, 9 e.s. complex systems, 37 e.s. general relation to secondary crystal- lization, 145 e.s., 189, 213, 275 independent of secondary network, 162 ingot, 147 involves segregation, 108 mechanism, 9 relation to dendrites, 191 relation to secondary crystallization, 86 relation to soniras, 275 solid solution in equilibrium, 11 solid solution in partial equilibrium, 19 e.s. suppression of dendritic system, 178 ternary system, 32 e.s. Primer of breakage, 283 Projectiles, 6 Q Quality, 135 Quality treatments. See Final heat treatments. Quaternary system, solidification, 37 e.s. transformation, 93 e.s. Quenching, alloy steel castings, 250 at variable speed, 143 definition, 309 energetic, 116 for longitudinal strength, 320 impaired, 202 incomplete, 121, 202, 215 limitations, 116 l6w-carbon steel castings, 232 massive pieces, 309 media, 117 mild, 250 phenomena on equilibrium diagram, 118 preliminary, 214 proper temperatures, 118 reduces secondary segregation, 114 repeated, 335 vs. annealings, 333 very mild, 323 R Reheating temperature, independence of quenching temperatures, 117 limits, 117 Recalescence, 100 Recrystallization. See Transformation. Refractory particles in steel, 140 Research, lacking metallurgical, 103 Re-solidification, 30, 36 Review of metallurgical knowledge, 104 Rolling. See Hot work. Secondary crystallization, beginnings, 60 completion, 75 complex systems, 93 e.s. ' disturbed by forging, 280, 293 e.s. effects of diffusion, 52 e.s. frequency of nuclei, 65, 212 general relation to primary crystal- lization, 145 e.s., 189, 213, 275 independent of dendrites, 162 laws, 54 location of nuclei, 85 on slow cooling, 241 relation to primary, 86 relation to sonims, 275 spontaneous, 293 supersaturated austenite, 69, 91 suppression, 294 uniform austenite, 60 e.s., 68, 75 e.s., 155 with rectilinear boundaries. 356 Segregation. See also Heterogeneity. affects heat treatment, 120 cause, 22 extensive, 99 ferrite disposition, 203 inherent to solidification, 108 of metallic carbides, 26 primary, 4 suppression in ferrite, 316 within crystals, 19 Shape of piece aifects transformation, 113 Sherardizing, 2 Slag inclusions, 137 cause woody fracture, 142 Slip, by forging, 279 by hot work, 124 by variable quenching, 143 Slip bands, relation to dendrites, 165 Soaking, 98 Soft spots, after hardening, 317 Soft steels. See also Dead-soft steels and Carbon steels. Solid solutions. See also Austenite. binary, 9 e.s. heterogeneous, 19 INDEX 373 Solid solutions, homologous, 4 incomplete solubility, 39 solidification, 3, 12 Solidification. See also Primary crys- tallization. end of, in binary solid solutions, 21 in ternary system, 35 influence of cooling velocity, 4 in quaternary system, 47 Solidus, definition, 17 in austenite transformation, 60 influenced by diffusion, 12 in quaternary system, 42 in ternary system, 32, 35 of binary system in equilibrium, 11 in partial equilibrium, 21 Solubility of carbon in iron, ferrite satu- ration, 78 incomplete, 39 influence of allotropy, 5 Sonims. See inclusions. Sorbite, by mild quenching, 202 Special steels. See Alloy steel. Spbntaneous recrystallization, 293 Static tests, 229 Strain hardening, 144. See also Cold work. Stresses, internal, 116 by variable quenching, 143 elimination, 121 Structure, 0.33 per cent. C ingot, 172 columnar, 285 crystals with rectilinear boundaries, 356 great crystals, 146, 201 lamellar, 285 massive pieces, 310, 313 reticular network, 159 variation after identical heat treat- ment, 302 Widmanstattian, 156, 200 Sulphur printing, 331 Supersaturation. See Undercooling. Suppression of expected phenomena, 39 Tension tests, 229 Ternary system, primary crystallization, 32 e.s. secondary crystallization, 93 e.s. Testing, direction in which test piece is cut, 282 longitudinal, 282 Thermal analysis, heterogeneous aus- tenite, 92 Thermal hysteresis. See Hysteresis. Toughness, increased by hot work, 288 Transformation, 5. See also Secondary crystallization, appearance of new phases, 64 beginning, 60 characteristics determine heat treat- ment, 113 completion, 113 effect of narrow range, 55 effect of stay within, 67 en-mass, 90 influence of mass, 113 influence of shape, 113 influence on solubility, 5 low temperature, 100 retarded, 99 velocity, 90 Transverse failures, 283 fracture of nickel-chromium steel forging, 302 main transverse, 282 testing, 282 weakness, crank hangers, 332 definition, 283 due to lamellar inclusions, 142 from emulsified sonims, 287, 330 furnace operations responsible, 336 in carbon steels, 338, 344, 351 e.s. in chromium steels, 347 in forgings, 330 e.s. ingotism, 284 in nickel steels, 333, 335, 342 microscopic detection, 287, 356 suppression, 320, 344 Twinning, by forging, 279 by hot work, 124 by variable quenching, 143 diffusion unaffected, 124 U Undercooled liquids, crystallization, 150 Undercooling, 14, 70 in austenite, 68, 91 rapid reactions, 100 Unstable equilibrium. See also Partial equilibrium. persistency, 2 374 INDEX Variant composition, 22 band of variant composition, q.v. in ternary alloys, 36 W Warping, 121 by variable quenching, 143 Widmanstattian structure, fracture, 158 in nickel steel, 200 nature, 156, 200 obstacles to growth, 212 physical properties, 158 Wire drawing, effect upon subsequent heat treatment, 293 Woody fracture, 286, 348 caused by slag particles, 142 Working. See Hot work or cold work. . .-jT^i.Tj^rx*!'-