Search Documents

By C. Zener, John H. Hollomon

It is commonly recognized that a given material may be described as ductile or brittle only with reference to the conditions of test. Thus under the usual test conditions quartz is brittle, but under high pressures it is ductile. Salts that are brittle at room temperature become ductile at elevated temperatures. Pitch, brittle with respect to rapid loads, flows at low rates of loading. Pearlitic steel, ductile under the usual conditions of test, may be embrittled under the proper conditions of combined stresses, temperature, and rate of loading. Earlier Investigations A clear analysis of the conditions necessary for plastic flow and for fracture was first given by LudwikI1 who introduced two types of stress, the "flow stress" and the "stress at fracture." These stresses may refer to tensile or to shear stresses, or, in fact, to any type of stress. Ludwik considered these two stresses to be independent of one another, functions of the conditions of test and of the prior history of the material. For the purpose of analysis. both a Flow stress and a stress at fracture can be envisaged as functions of strain over the entire range from zero to infinite strain. Thus, the material is brittle if the stress at fracturc is less than the flow stress and ductile if the reverse is true. According to Ludwik's analysis, the problem of determining the effects of combined stresses, of temperature and of strain rate upon the behavior of metals reduces to a problem of determining the effect of these variables upon the flow stress and upon the stress at fracture. In the present paper, methods are described for determining the effects of two of these variables, temperature and strain rate, and examples are given of the dependence of tlow stress and stress at fracture, upon these two variables. The effect of combined stresses will be discussed in a succeeding paper. The flow stress may be determined from a load-diameter curve. Load-diameter curves were obtained over a wide range of temperature and over the limited range of strain rates obtainable with the lever-type Riehle machine used. The curves were then converted to an extremely wide range of strain rate by making use of the equivalence of the effects of changes in strain rate and changes in temperature, previously proposed2 and demonstrated3 by the authors. According to this principle of equivalence, strain rate e and temperature T affect the flow stress only through a single parameter-p of the form: p = eeQ/RT 111 In this equation, R is the gas constant (2 cal. per gram mol per deg. C.), and Q is a heat of activation, which varies from material to material, and may depend upon the strain, but not upon the strain rate or the temperature. There is associated with every value of p a flow-stress curve. article of strain rate and of temperature that give the same p have the same flow-stress curve. Therefore, merely by changing the temperature of test, the effect of changing the strain rate may be obtained. Thus, under isothermal conditions, the material will behave the same at

Jan 1, 1944

By Eugene F. Poncelet

This paper attempts to analyze the phenomena involved in the fracture of brittle solids by simple compression. Glass squares standing on edge, and compressed between two parallel steel jaws, developed jagged fractures, roughly parallel to the compression. Polarized light disclosed the presence of high local stress concentrations indicating poor contact all along the contact surfaces with the steel jaws. Teszar has shown that partial contact between the jaws and such a compressed solid causes tensile stresses in the free surfaces of the solid normal to the compression. Furthermore, Scoble has shown that in brittle solids fractures are caused by, and develop normally to, a sufficient tensile stress. It follows that such "partial-contact" cracks must be expected. With optically flat jaws and specimens, no "partial-contact" cracks developed. Instead, very smooth cracks vertical to the compression appeared 'upon release of the pressure. Investigation showed that the friction with the rigid steel opposed the return of the glass to its original width, after creeping on the steel as compression built up, thus setting up tensile stresses in the glass surfaces. It was not found possible to eliminate these "release" cracks. Compressed to destruction, such specimens flew to dust. A Microflash photograph taken during the explosive disintegration revealed: (I) oblique parallel fragmentation cracks, 30" off the direction of the compression; (2) transverse forking fragmentation cracks, normal to the former; (3) release cracks parallel to the compression; (4) a disintegration cloud. Given a distribution of submicroscopic cracks of random orientation throughout the glass, Griffith has shown the highest tensile stresses to be located at the tip of the cracks 30" off the direction of the compression. Although the preexistence of such cracks or flaws in unstressed glass is doubtful, they can be generated by the distorting forces set up by the compression itself. Atoms, in a structure held together by Morse-type forces or bonds, have under tension both a stable and an unstable state-of equilibrium. The difference in energy level between these two states is a function of the tension, and this difference must be borrowed from the thermal energy level of the bonds to reach the unstable state and result in a broken bond. As the fraction of the bonds having at any instant a sufficient thermal energy level is a known function of that level, it is possible to determine, in function of the prevailing tension, the fraction of the bonds that are broken. Upon breaking of the bond uniting two atoms at the tip of an advancing crack, the two atoms accelerate away from each other, causing in the solid propagation of a

Jan 1, 1944

By Eugene F. Poncelet

This paper attempts to analyze the phenomena involved in the fracture of brittle solids by simple compression. Glass squares standing on edge, and compressed between two parallel steel jaws, developed jagged fractures, roughly parallel to the compression. Polarized light disclosed the presence of high local stress concentrations indicating poor contact all along the contact surfaces with the steel jaws. Teszar has shown that partial contact between the jaws and such a compressed solid causes tensile stresses in the free surfaces of the solid normal to the compression. Furthermore, Scoble has shown that in brittle solids fractures are caused by, and develop normally to, a sufficient tensile stress. It follows that such "partial-contact" cracks must be expected. With optically flat jaws and specimens, no "partial-contact" cracks developed. Instead, very smooth cracks vertical to the compression appeared 'upon release of the pressure. Investigation showed that the friction with the rigid steel opposed the return of the glass to its original width, after creeping on the steel as compression built up, thus setting up tensile stresses in the glass surfaces. It was not found possible to eliminate these "release" cracks. Compressed to destruction, such specimens flew to dust. A Microflash photograph taken during the explosive disintegration revealed: (I) oblique parallel fragmentation cracks, 30" off the direction of the compression; (2) transverse forking fragmentation cracks, normal to the former; (3) release cracks parallel to the compression; (4) a disintegration cloud. Given a distribution of submicroscopic cracks of random orientation throughout the glass, Griffith has shown the highest tensile stresses to be located at the tip of the cracks 30" off the direction of the compression. Although the preexistence of such cracks or flaws in unstressed glass is doubtful, they can be generated by the distorting forces set up by the compression itself. Atoms, in a structure held together by Morse-type forces or bonds, have under tension both a stable and an unstable state-of equilibrium. The difference in energy level between these two states is a function of the tension, and this difference must be borrowed from the thermal energy level of the bonds to reach the unstable state and result in a broken bond. As the fraction of the bonds having at any instant a sufficient thermal energy level is a known function of that level, it is possible to determine, in function of the prevailing tension, the fraction of the bonds that are broken. Upon breaking of the bond uniting two atoms at the tip of an advancing crack, the two atoms accelerate away from each other, causing in the solid propagation of a

Jan 1, 1944

By Nicholas Kowall, K. L. Clark

Maximum furnace efficiency and close control of final steel composition demand that the steel melter be able to follow closely the variations in the carbon content of the bath. For many years, the fracture test has been used by melters for estimating carbon, because it requires only a very short time to pour a sample into a chill mold, cool it in water, break it with a sledge hammer, and observe the fracture. After long experience gained from the observation of such fractures, most melters are able to estimate the carbon content of steel with amazing accuracy, particularly in the lower carbon ranges. Notwithstanding the utility of the fracture test, there has been widespread interest in other methods for rapid carbon determination to supplement it and to ensure against possible errors in human judgment. Several methods, based upon measurement of magnetic properties as functions of carbon content, have been described in the literature1-6 and have been found to give reasonably reliable results in plant practice when scrap composition is known and controlled. A difficulty encountered in the use of analyzers of the magnetic type is that alloying elements do change the relationship between carbon content and magnetic properties and it becomes necessary to establish calibration curves for all the expected combinations of alloy residuals that are not oxidized in the bath. The work reported here is of an investigation of a method that is independent of alloying elements normally present in constructional steels, being based upon the relatively well-known fact that in plain carbon and low-alloy steels martensitic hardness is a function of carbon content alone. Work oF Other Investigators In I938, Burns, Moore and Archer published & curve? that showed that the hardness of martensite is a function of carbon from 0.10 to 0.50 per cent and is independent of grain size or alloy content, at least within the rallge found in construc-tional S.A.E. compositions. Some of the deeper hardening types investigated, however, contained carbon to a maximum of only o.2o per cent. Using this principle, Kern8 developed a method of rapid carbon determination for S.A.E 43xx steels, His procedure consisted of chill-casting a test bar that was large enough to give a a that could be prepared for Brinell indentations. The test piece was quenched immediately after solidi fication was complete and hardness readings were taken, so that the carbon content could he read from an experimentally determined curve giving the relationship between as-quenched hardness and carbon. Experimental Work Although Kern's work had been done with a relatively deep-hardening steel, it

Jan 1, 1944

By K. L. Clark, Nicholas Kowall

Maximum furnace efficiency and close control of final steel composition demand that the steel melter be able to follow closely the variations in the carbon content of the bath. For many years, the fracture test has been used by melters for estimating carbon, because it requires only a very short time to pour a sample into a chill mold, cool it in water, break it with a sledge hammer, and observe the fracture. After long experience gained from the observation of such fractures, most melters are able to estimate the carbon content of steel with amazing accuracy, particularly in the lower carbon ranges. Notwithstanding the utility of the fracture test, there has been widespread interest in other methods for rapid carbon determination to supplement it and to ensure against possible errors in human judgment. Several methods, based upon measurement of magnetic properties as functions of carbon content, have been described in the literature1-6 and have been found to give reasonably reliable results in plant practice when scrap composition is known and controlled. A difficulty encountered in the use of analyzers of the magnetic type is that alloying elements do change the relationship between carbon content and magnetic properties and it becomes necessary to establish calibration curves for all the expected combinations of alloy residuals that are not oxidized in the bath. The work reported here is of an investigation of a method that is independent of alloying elements normally present in constructional steels, being based upon the relatively well-known fact that in plain carbon and low-alloy steels martensitic hardness is a function of carbon content alone. Work oF Other Investigators In I938, Burns, Moore and Archer published & curve? that showed that the hardness of martensite is a function of carbon from 0.10 to 0.50 per cent and is independent of grain size or alloy content, at least within the rallge found in construc-tional S.A.E. compositions. Some of the deeper hardening types investigated, however, contained carbon to a maximum of only o.2o per cent. Using this principle, Kern8 developed a method of rapid carbon determination for S.A.E 43xx steels, His procedure consisted of chill-casting a test bar that was large enough to give a a that could be prepared for Brinell indentations. The test piece was quenched immediately after solidi fication was complete and hardness readings were taken, so that the carbon content could he read from an experimentally determined curve giving the relationship between as-quenched hardness and carbon. Experimental Work Although Kern's work had been done with a relatively deep-hardening steel, it

Jan 1, 1944

By Herbert H. Uhlig

Although the mechanical properties of high-carbon manganese-iron alloys, particularly the Hadfield manganese steels, have been established, the literature discloses discrepancies in the reported properties for the low-carbon alloys. Hadfieldla summarized the existing data up to 1927, including the investigations of Guillet, Arnold, Strauss, Burgess and Aston, and reported his own values for alloys containing for the most part from 0.06 to 0.I5 per cent carbon. He noted: From these comparisons it wil! be seen that while there is a broad and general agreement as to the character of the alloys at the different manganese percentages, there are disparities in individual cases, especially as regards the degree of toughness or brittleness displayed by an alloy with certain percentages and with particular types of treatment. These differences are not always readily explainable.lb His data showed that alloys containing 4 to 10 per cent manganese water-quenched from Iooo°C. were brittle and hard with maximum brittleness at about 7 per cent. Slight improvement in ductility was obtained with compositions from 10 to 17 per cent manganese. Alloys containing up to 17 per cent manganese slowly cooled rather than quenched from 87o°C., he found, were slightly improved in ductility with significant improvement in alloys analyzing 4.1 and 4.8 per cent manganese. Recently Walters, Kramer and Loring2 reported mechanical properties for alloys made up from electrolytic manganese and ingot iron and containing from 0.02 to 0.04 per cent carbon. They showed that a tempering treatment at 54o°C. (Iooo°F.) for one hour significantly improved the ductility of compositions ranging from 6.7 to &apos; 20.7 per cent manganese. The differences of these compared with Hadfield&apos;s results they considered to originate in this tempering treatment plus the fact that Hadfield&apos;s alloys contained higher carbon and phosphorus (0.06 to 0.I5 per cent C; 0.06 to 0.07 per cent P compared with 0.02 to 0.04 per cent C, <o.ooI per cent P). In this paper it is shown that of the low-carbon alloys (0.02 to 0.03 per cent) containing from 3.3 to 21.5 per cent manganese moderately or rapidly quenched from IOOO°C. some are brittle or ductile, depending on presence or absence of hydrogen. Hydrogen is readily absorbed at high temperatures, and much of it is retained when the alloys are cooled to room temperature. The presence of hydrogen in the range of composition from 10 to 18 per cent manganese is a major factor in determining relative ductility. At 22 per cent manganese, and also at compositions below 10 per cent manganese, factors other than or in addition to hydrogen play a role. These are discussed. Also data are presented to show that hydrogen absorbed by the high-carbon alloys of Hadfield composition is without effect on their mechanical properties.

Jan 1, 1944

By Herbert H. Uhlig

Although the mechanical properties of high-carbon manganese-iron alloys, particularly the Hadfield manganese steels, have been established, the literature discloses discrepancies in the reported properties for the low-carbon alloys. Hadfieldla summarized the existing data up to 1927, including the investigations of Guillet, Arnold, Strauss, Burgess and Aston, and reported his own values for alloys containing for the most part from 0.06 to 0.I5 per cent carbon. He noted: From these comparisons it wil! be seen that while there is a broad and general agreement as to the character of the alloys at the different manganese percentages, there are disparities in individual cases, especially as regards the degree of toughness or brittleness displayed by an alloy with certain percentages and with particular types of treatment. These differences are not always readily explainable.lb His data showed that alloys containing 4 to 10 per cent manganese water-quenched from Iooo°C. were brittle and hard with maximum brittleness at about 7 per cent. Slight improvement in ductility was obtained with compositions from 10 to 17 per cent manganese. Alloys containing up to 17 per cent manganese slowly cooled rather than quenched from 87o°C., he found, were slightly improved in ductility with significant improvement in alloys analyzing 4.1 and 4.8 per cent manganese. Recently Walters, Kramer and Loring2 reported mechanical properties for alloys made up from electrolytic manganese and ingot iron and containing from 0.02 to 0.04 per cent carbon. They showed that a tempering treatment at 54o°C. (Iooo°F.) for one hour significantly improved the ductility of compositions ranging from 6.7 to &apos; 20.7 per cent manganese. The differences of these compared with Hadfield&apos;s results they considered to originate in this tempering treatment plus the fact that Hadfield&apos;s alloys contained higher carbon and phosphorus (0.06 to 0.I5 per cent C; 0.06 to 0.07 per cent P compared with 0.02 to 0.04 per cent C, <o.ooI per cent P). In this paper it is shown that of the low-carbon alloys (0.02 to 0.03 per cent) containing from 3.3 to 21.5 per cent manganese moderately or rapidly quenched from IOOO°C. some are brittle or ductile, depending on presence or absence of hydrogen. Hydrogen is readily absorbed at high temperatures, and much of it is retained when the alloys are cooled to room temperature. The presence of hydrogen in the range of composition from 10 to 18 per cent manganese is a major factor in determining relative ductility. At 22 per cent manganese, and also at compositions below 10 per cent manganese, factors other than or in addition to hydrogen play a role. These are discussed. Also data are presented to show that hydrogen absorbed by the high-carbon alloys of Hadfield composition is without effect on their mechanical properties.

Jan 1, 1944

By H. A. Schwartz, W. Kenneth Bock

In this paper it is shown that when the tensile strength of a given steel in various states of heat-treatment is plotted against its elongation, a straight line results. The equation of this straight line can be computed with considerable accuracy from its chemical composition. The appropriate equations are given. The logarithm of the slope of the tensile-strength locus is shown to be proportional to the cumulative effect of the total alloy addition and the vosition of the line to be directly proportional to the cumulative effect of the alloy content. It is further pointed out that heat-treatment is more effective than composition in determining yield ratio. The latter enters mainly by its effect on hardenability, which is briefly discussed. Reduction of area and Brinell number are given some incidental attention. The data are applicable only to steels that have been normalized and/or quenched. The steels, if normalized, may or may not have been tempered. If quenched they must have been tempered at or above 4oo°C. (75o°F.). Introduction Cast steels vary in composition and frequently are used in various states of heat-treatment. It is obviously important to be able to predict the compositions that might accompany specific engineering properties by suitable selection of heat-treatments. For cast steels, which are of necessity produced heat by heat in accordance with the particular requirements of the castings to be poured, there has been, happily, much less tendency to standardization than for rolled products, whose destination is not necessarily known when particular ingots are made or rolled. It is of distinct advantage to the steel foundry-men to be able to make steel of given properties, using alloys that can be recovered from scrap, or most conveniently or cheaply bought, rather than a particular formula adopted in the interest of standardization. The heat-treatment appropriate to securing a given result in a given steel is well known to be closely related to composition. For example, molybdenum steels are not much softened by tempering except at rather high temperatures. Highly alloyed steels are "air-hardening," and on cooling in air behave like carbon steels that have been quenched. With these matters, we do not concern ourselves in this paper, although we recognize their importance. The entire picture evidently will be very complex and can be brought within accurate understanding only if our knowledge of test results can be coordinated in terms of general principles.

Jan 1, 1944

By H. A. Schwartz, W. Kenneth Bock

In this paper it is shown that when the tensile strength of a given steel in various states of heat-treatment is plotted against its elongation, a straight line results. The equation of this straight line can be computed with considerable accuracy from its chemical composition. The appropriate equations are given. The logarithm of the slope of the tensile-strength locus is shown to be proportional to the cumulative effect of the total alloy addition and the vosition of the line to be directly proportional to the cumulative effect of the alloy content. It is further pointed out that heat-treatment is more effective than composition in determining yield ratio. The latter enters mainly by its effect on hardenability, which is briefly discussed. Reduction of area and Brinell number are given some incidental attention. The data are applicable only to steels that have been normalized and/or quenched. The steels, if normalized, may or may not have been tempered. If quenched they must have been tempered at or above 4oo°C. (75o°F.). Introduction Cast steels vary in composition and frequently are used in various states of heat-treatment. It is obviously important to be able to predict the compositions that might accompany specific engineering properties by suitable selection of heat-treatments. For cast steels, which are of necessity produced heat by heat in accordance with the particular requirements of the castings to be poured, there has been, happily, much less tendency to standardization than for rolled products, whose destination is not necessarily known when particular ingots are made or rolled. It is of distinct advantage to the steel foundry-men to be able to make steel of given properties, using alloys that can be recovered from scrap, or most conveniently or cheaply bought, rather than a particular formula adopted in the interest of standardization. The heat-treatment appropriate to securing a given result in a given steel is well known to be closely related to composition. For example, molybdenum steels are not much softened by tempering except at rather high temperatures. Highly alloyed steels are "air-hardening," and on cooling in air behave like carbon steels that have been quenched. With these matters, we do not concern ourselves in this paper, although we recognize their importance. The entire picture evidently will be very complex and can be brought within accurate understanding only if our knowledge of test results can be coordinated in terms of general principles.

Jan 1, 1944

By John H. Hollomon

The interpretation of notched-bar impact results has been a matter of controversy since the introduction of more or less standard tests by Fremont,&apos; Charpy2 and others at the turn of the century. Many investigators3-l1 have contributed to the understanding of the significance of such tests. Several symposiums12,13 have been held with the express purpose of discussing the significance and interpretation of the impact tests. It is not the purpose of this paper to review all these contributions to the development of the knowledge on this subject, but rather to present an interpretation of notched-bar impact tests that appears to be in agreement with the available published data. Fundamentally, the interpretation discussed in this paper of the brittle failure of some steels in notched-bar impact test is similar to the qualitative analysis presented by Ludwik3 in the I92o&apos;s. Essentially, Ludwik stated that the function of the notch was to increase the tensile stress necessary for yielding, and that brittle failure occurred when the stress necessary for yielding exceeded the normal stress required for fracture (with no plastic deformation). Ludwik, however, had no means of measuring this normal stress necessary for failure of a plastically* undeformed metal, therefore he could not extend his interpretation beyond the qualitative stage. Kuntze14 and recently Mc-Adam15 attempted to determine the effect of deformation on the stress necessary for fracture by testing, in tension, specimens notched to various depths and angles. Because of the uncertainties in Kuntze&apos;s (and McAdam&apos;s) analysis of the stresses (which have been pointed out by Sachs16) and the inherent nonuniformity in the stress distribution in notched bars, the results of such investigations are difficult to interpret. A more direct method of measuring the effect of deformation on the fracture stress was suggested in previous papers"-18 and has been utilized in the present investigation. The present paper concerns itself primarily with the behavior of notched impact specimens. The effects of strain rate, temperature, and stress distribution, which are discussed with reference to the impact specimen, apply just as well to the behavior of metal at the bases of notches in any engineering structure. All that is necessary is to determine the stress distribution in the structure at the base of the notches.

Jan 1, 1944

By John H. Hollomon

The interpretation of notched-bar impact results has been a matter of controversy since the introduction of more or less standard tests by Fremont,&apos; Charpy2 and others at the turn of the century. Many investigators3-l1 have contributed to the understanding of the significance of such tests. Several symposiums12,13 have been held with the express purpose of discussing the significance and interpretation of the impact tests. It is not the purpose of this paper to review all these contributions to the development of the knowledge on this subject, but rather to present an interpretation of notched-bar impact tests that appears to be in agreement with the available published data. Fundamentally, the interpretation discussed in this paper of the brittle failure of some steels in notched-bar impact test is similar to the qualitative analysis presented by Ludwik3 in the I92o&apos;s. Essentially, Ludwik stated that the function of the notch was to increase the tensile stress necessary for yielding, and that brittle failure occurred when the stress necessary for yielding exceeded the normal stress required for fracture (with no plastic deformation). Ludwik, however, had no means of measuring this normal stress necessary for failure of a plastically* undeformed metal, therefore he could not extend his interpretation beyond the qualitative stage. Kuntze14 and recently Mc-Adam15 attempted to determine the effect of deformation on the stress necessary for fracture by testing, in tension, specimens notched to various depths and angles. Because of the uncertainties in Kuntze&apos;s (and McAdam&apos;s) analysis of the stresses (which have been pointed out by Sachs16) and the inherent nonuniformity in the stress distribution in notched bars, the results of such investigations are difficult to interpret. A more direct method of measuring the effect of deformation on the fracture stress was suggested in previous papers"-18 and has been utilized in the present investigation. The present paper concerns itself primarily with the behavior of notched impact specimens. The effects of strain rate, temperature, and stress distribution, which are discussed with reference to the impact specimen, apply just as well to the behavior of metal at the bases of notches in any engineering structure. All that is necessary is to determine the stress distribution in the structure at the base of the notches.

Jan 1, 1944

By Myron A. Pugacz, Frank S. McKenna, Clarence E. Jackson

The notched-bar impact test has proved worth while in certain applications as a test for control of the quality or the heat-treatment of steel. In view of the serious thought that even so simple a test as the tensile test has evoked during the past years, it is not surprising that the notched-bar impact test should have inspired much experimental work. The correlation of the great number of data has not been successful, although steady progress has been made toward a better understanding of this test. Progress has been made, especially, in the isolation of the effects of the many variables that control the results. The stress system imposed on a specimen with a notch is so complicated and unpredictable that no simple evaluation of the test has been possible. Because of this uncertainty, many engineers hesitate to use the notched-bar impact test, and this attitude may well continue until a better understanding of the effects of the variables in the test are available. Without doubt, any additional data that may be presented should be accompanied by as complete a description of the - details of test as possible, so that it may be added to the general fund of information that is to serve as the basis for the theory of the notched-bar impact test. The many factors that may control the test results have been pointed out repeatedly. The importance of the effect of the type and dimensions of the specimen has been recognized. Most of the work reported by American investigators has been limited to small-size (Charpy) specimens tested as a simple beam; two types of notches (Fig. I) have been commonly used. Considerable data have also been presented by many investigators using a small specimen tested as a cantilever beam; this commonly being called the Izod test. Here, however, a shear component of stress makes the stress system even more complex than that present in the simple beam type of test, hence a critical analysis is difficult. Much work has also been reported by investigators using the tension type of specimen. A second variable that must be controlled is the testing temperature. An indication that the tests were performed at room temperature is not sufficient, since in many steels a variation of a few degrees in the actual testing temperature may be sufficient to give an entirely different range of results. Other factors, such as the composition and condition of the material being tested and the speed of test, should also be considered. Many investigators have contributed data and theoretical discussions on the problems encountered in notched-bar test-

Jan 1, 1944

By Frank S. McKenna, Myron A. Pugacz, Clarence E. Jackson

The notched-bar impact test has proved worth while in certain applications as a test for control of the quality or the heat-treatment of steel. In view of the serious thought that even so simple a test as the tensile test has evoked during the past years, it is not surprising that the notched-bar impact test should have inspired much experimental work. The correlation of the great number of data has not been successful, although steady progress has been made toward a better understanding of this test. Progress has been made, especially, in the isolation of the effects of the many variables that control the results. The stress system imposed on a specimen with a notch is so complicated and unpredictable that no simple evaluation of the test has been possible. Because of this uncertainty, many engineers hesitate to use the notched-bar impact test, and this attitude may well continue until a better understanding of the effects of the variables in the test are available. Without doubt, any additional data that may be presented should be accompanied by as complete a description of the - details of test as possible, so that it may be added to the general fund of information that is to serve as the basis for the theory of the notched-bar impact test. The many factors that may control the test results have been pointed out repeatedly. The importance of the effect of the type and dimensions of the specimen has been recognized. Most of the work reported by American investigators has been limited to small-size (Charpy) specimens tested as a simple beam; two types of notches (Fig. I) have been commonly used. Considerable data have also been presented by many investigators using a small specimen tested as a cantilever beam; this commonly being called the Izod test. Here, however, a shear component of stress makes the stress system even more complex than that present in the simple beam type of test, hence a critical analysis is difficult. Much work has also been reported by investigators using the tension type of specimen. A second variable that must be controlled is the testing temperature. An indication that the tests were performed at room temperature is not sufficient, since in many steels a variation of a few degrees in the actual testing temperature may be sufficient to give an entirely different range of results. Other factors, such as the composition and condition of the material being tested and the speed of test, should also be considered. Many investigators have contributed data and theoretical discussions on the problems encountered in notched-bar test-

Jan 1, 1944

By Alden B. Greninger, Alexander R. Troiano

The crystallographic mechanism for the austenite-to-martensite trensformation has been deduced from the results of the following new experimental determinations: (I) the accurate evaluation of the lattice relationship between austenite and individual crystals of martensite—and thus the relationship between the martensite lattice and the martensite plate, and (2) the measurement and analysis of the change in positions that a volume of austenite unclergoes when it transforms into a crystal of martensite. For the latter study, a new tool was developed, the stereographic analysis of shear. Lattice relationship.—Thirty-five separate 50-gram melts of an alloy containing about 22 per cent nickel and 0.8 per cent carbon were prepared in a high-vacuum furnace. This alloy is all-austenite at room temperature; grain size was about I cm. Specimens of the alloy were cooled to about — 70°C. to form a few martensite crystals, and suitable specimens were ground and polished on a surface parallel to a single martensite plate, exposing the martensite crystal for an area of 1 mm. or more. Four martensite cystals from four diffierent ingots were prepared in this manner; the largest crystal of this group is illuslratecl in Fig. I. Back-reflection Laue patterns were obtained from martensite crystals (and the matrix austenite) both before and after tempering. The locations of the basal-plane pole of martensite was checked by means of oscillating crystal X-ray patterns. All four crystals studied yielded the same solution ( ± 0.5°), and these relationships are expressed in gnomonic projection in Fig. 3.* The narrow dark-etching bands often visible in martensite "needles" were found to be parallel to (112) and thus are undoubtedly twin bands in the martensite crystal.

Jan 1, 1941

By Alden B. Greninger, Alexander R. Troiano

The crystallographic mechanism for the austenite-to-martensite trensformation has been deduced from the results of the following new experimental determinations: (I) the accurate evaluation of the lattice relationship between austenite and individual crystals of martensite—and thus the relationship between the martensite lattice and the martensite plate, and (2) the measurement and analysis of the change in positions that a volume of austenite unclergoes when it transforms into a crystal of martensite. For the latter study, a new tool was developed, the stereographic analysis of shear. Lattice relationship.—Thirty-five separate 50-gram melts of an alloy containing about 22 per cent nickel and 0.8 per cent carbon were prepared in a high-vacuum furnace. This alloy is all-austenite at room temperature; grain size was about I cm. Specimens of the alloy were cooled to about — 70°C. to form a few martensite crystals, and suitable specimens were ground and polished on a surface parallel to a single martensite plate, exposing the martensite crystal for an area of 1 mm. or more. Four martensite cystals from four diffierent ingots were prepared in this manner; the largest crystal of this group is illuslratecl in Fig. I. Back-reflection Laue patterns were obtained from martensite crystals (and the matrix austenite) both before and after tempering. The locations of the basal-plane pole of martensite was checked by means of oscillating crystal X-ray patterns. All four crystals studied yielded the same solution ( ± 0.5°), and these relationships are expressed in gnomonic projection in Fig. 3.* The narrow dark-etching bands often visible in martensite "needles" were found to be parallel to (112) and thus are undoubtedly twin bands in the martensite crystal.

Jan 1, 1941

By A. H. Geisler, R. F. Mehl, C. S. Barrett

The complicated nature of the property changes that accompany age-hardening has made it necessary to reconsider and to elaborate the simple dispersion theory.l It has been apparent for some time that more direct information is needed concerning the atomic mechanism of precipitation from solid solution if the theory of age-hardening is to be developed past a purely qualitative stage.‡ Fortunately, in recent years information has been developed which allows us to trace in detail the lattice changes accompanying the process of precipitation from solid solution. This information relates chiefly to the sequence of the several stages of lattice alteration by which the matrix lattice is transformed to the equilibrium precipitate lattice, and permits reasonable inferences to he drawn concerning the states of internal strain that accompany these stages. This information is most complete for the solid solution of copper in aluminum. The equilibrium precipitate is the 0 phase, "CuAl2." This forms in plates, parallel to the {100} matrix planes. The atom pattern and atom spacing on the plane in 0 lying parallel to this plane are roughly similar to those opposite it in the matrix.3 The formation of 0, however, is preceded by the formation of a transition lattice 0&apos;: which again lies parallel to the matrix {100} plane, but in this case the matching of atom pattern and spacing across the matrix-precipitate interface is nearly periect; thus there is lattice coherency across the interface and the interface is a plane of separation in only a very limited sense.&apos;8 Particles of 0&apos; may be recognized microscopically and form a well-defined Wid-manstatten figure.* The 0&apos; lattice is recognized by the occurrence of diffraction lines on powder photo-grams or of spots on Laue photograms which may be analyzed in the usual ways to determine the lattice type and the lattice spacings. Very recent research 6, 7, 1 has shown, however, that streaks occur on the Laue photogram at an earlier point in time during the aging process. These streaks have been attributed to a preferential clustering of copper atoms into small platelike elements of only a few atoms in thickness and a lew hundred atoms in diameter, probably of the composition of " CuA12," which may be designated as Guinier-Preston zones or aggregates.1 It is difficult if not impossible to determine the lattice type and lattice spacing in these aggregates. It is possible1 that these aggregates are in fact ?&apos; in a high State of dispersion; the inclination to argue this is strong, as we shall see later. Never-

Jan 1, 1941

By R. F. Mehl, C. S. Barrett, A. H. Geisler

The complicated nature of the property changes that accompany age-hardening has made it necessary to reconsider and to elaborate the simple dispersion theory.l It has been apparent for some time that more direct information is needed concerning the atomic mechanism of precipitation from solid solution if the theory of age-hardening is to be developed past a purely qualitative stage.‡ Fortunately, in recent years information has been developed which allows us to trace in detail the lattice changes accompanying the process of precipitation from solid solution. This information relates chiefly to the sequence of the several stages of lattice alteration by which the matrix lattice is transformed to the equilibrium precipitate lattice, and permits reasonable inferences to he drawn concerning the states of internal strain that accompany these stages. This information is most complete for the solid solution of copper in aluminum. The equilibrium precipitate is the 0 phase, "CuAl2." This forms in plates, parallel to the {100} matrix planes. The atom pattern and atom spacing on the plane in 0 lying parallel to this plane are roughly similar to those opposite it in the matrix.3 The formation of 0, however, is preceded by the formation of a transition lattice 0&apos;: which again lies parallel to the matrix {100} plane, but in this case the matching of atom pattern and spacing across the matrix-precipitate interface is nearly periect; thus there is lattice coherency across the interface and the interface is a plane of separation in only a very limited sense.&apos;8 Particles of 0&apos; may be recognized microscopically and form a well-defined Wid-manstatten figure.* The 0&apos; lattice is recognized by the occurrence of diffraction lines on powder photo-grams or of spots on Laue photograms which may be analyzed in the usual ways to determine the lattice type and the lattice spacings. Very recent research 6, 7, 1 has shown, however, that streaks occur on the Laue photogram at an earlier point in time during the aging process. These streaks have been attributed to a preferential clustering of copper atoms into small platelike elements of only a few atoms in thickness and a lew hundred atoms in diameter, probably of the composition of " CuA12," which may be designated as Guinier-Preston zones or aggregates.1 It is difficult if not impossible to determine the lattice type and lattice spacing in these aggregates. It is possible1 that these aggregates are in fact ?&apos; in a high State of dispersion; the inclination to argue this is strong, as we shall see later. Never-

Jan 1, 1941

By Anson Hayes, John Chipman

The quality of sheet and strip products made of rimming steel is closely related to the structure and chemistry of the ingots. The variation in composition throughout the ingot, as affected by segregation, is of great importance, as are also the form and distribution of the voids that result from the gases evolved. It is because of this importance that the considerable literature in this field has accumulated. No attempt is made in the present paper to compile a bibliography of the excellent work that has already been done, as this has been adequately covered in the many excellent reports on the heterogeneity of steel ingots prepared by a joint committee and published by the Iron and Steel Institute.&apos; From time to time, in any field of investigation new tools and new technique are developed, which make further progress possible. In the field with which this paper is concerned, this is true. The development of the vacuum-fusion method as a means of following oxygen segregation, and a technique for obtaining the volume and composition of the gases evolved during solidification, have opened the way to obtaining data by means of which some considerable progress has been realized in developing more of the detailed mechanisms of solidification and segregation in rimming-steel ingots. General Theory of Segregation The basic cause of segregation lies in the fact that the impurities are less soluble in solid than in liquid iron. The solid that forms from the impure liquid is more nearly pure than the liquid itself. If solidification proceeds slowly enough, or if the liquid and solid phases are maintained in contact with one another for a sufficient time, a condition of equilibrium is set up, which may be represented by a phase diagram. Portions of the

Jan 1, 1939

By Anson Hayes, John Chipman

The quality of sheet and strip products made of rimming steel is closely related to the structure and chemistry of the ingots. The variation in composition throughout the ingot, as affected by segregation, is of great importance, as are also the form and distribution of the voids that result from the gases evolved. It is because of this importance that the considerable literature in this field has accumulated. No attempt is made in the present paper to compile a bibliography of the excellent work that has already been done, as this has been adequately covered in the many excellent reports on the heterogeneity of steel ingots prepared by a joint committee and published by the Iron and Steel Institute.&apos; From time to time, in any field of investigation new tools and new technique are developed, which make further progress possible. In the field with which this paper is concerned, this is true. The development of the vacuum-fusion method as a means of following oxygen segregation, and a technique for obtaining the volume and composition of the gases evolved during solidification, have opened the way to obtaining data by means of which some considerable progress has been realized in developing more of the detailed mechanisms of solidification and segregation in rimming-steel ingots. General Theory of Segregation The basic cause of segregation lies in the fact that the impurities are less soluble in solid than in liquid iron. The solid that forms from the impure liquid is more nearly pure than the liquid itself. If solidification proceeds slowly enough, or if the liquid and solid phases are maintained in contact with one another for a sufficient time, a condition of equilibrium is set up, which may be represented by a phase diagram. Portions of the

Jan 1, 1939

By R. C. Dalzell, A. E. Wells

In 1866, Thomas Graham1 called attention to the volume of gases, three times the volume of the sample, evolved from a meteoric iron heated in an evacuated porcelain tube. From that time to this the occurrence and significance of gases in metals have been the subject of investigation by many workers, among whom Sieverts,2 Iwasé3 and 0. W. Ellis4 stand preeminent. Sieverts and Iwasé concerned themselves with the solubility of a number of gases in metals in the liquid and solid states. They determined the absorption coefficients of these gases, and in all cases found a sharp change in solubility at the melting point of the metals. Ellis analyzed the gases evolved by solidifying copper and those occluded by the solidified metal. Lobley and Jepson5 endeavored to determine the effects of hydrogen, carbon monoxide and nitrogen on the porosity of copper that was allowed to solidify in the presence of each gas after 30 min. exposure to it in the

Jan 1, 1930