Search Documents

By Nicholas J. Grant, Oliver Preston

PUBLICATION of data by Irmann' indicating outstanding thermal stability and elevated-temperature strength properties in a sintered aluminum powder product (SAP) stimulated interest in the strengthening of metals by means of a highly dispersed oxide phase. This material was produced by compacting, sintering and extruding a fine flake aluminum powder, resulting in a dense product containing 10 to 16 volume pct alumina, which arises from the thin oxide film on the starting powder, highly dispersed in a matrix of pure aluminum. The method of producing SAP, and the stable high-temperature properties obtained from it, introduced the possibility of developing superior materials for high-temperature service by powder metallurgy methods. The composition of the materials of this type which are presently under consideration is predominantly metallic, containing from less than 1 pct up to about 20 volume pct of the oxide. As a result, other properties of these materials, such as electrical and thermal conductivity, thermal shock resistance, and to a certain extent ductility, notch sensitivity, machinability and hot working properties, are also predominantly metallic. These alloys do not constitute, therefore, a completely new class of materials as do the cermets, but rather present a new means of modifying the properties of a metallic matrix. In this sense, SAP type alloys are more closely analogous to the alloy aging systems. While the strength properties realized in the oxide-pure metal systems at the lower temperatures are substantially less than the best attainable in the aging systems, the thermal instabilities in the latter make them of less value for use at temperature much above their optimum aging temperatures, as indicated in Fig. 1. While the means of obtaining an oxide dispersion are more difficult and costly than conventional aging processes, in the light of the as yet limited understanding of such systems there is considerable promise of improved high temperature properties arising from the thermal stability of these structures. Since the development of SAP in 1946, its properties and the properties of a series of similarAl,O,-A1 products containing from about 1 to 20 pct aluminum oxide have been fairly extensively studied and reported.'- "articular attention has been given to

Jan 1, 1958

By E. W. Hart, W. R. Hibbard

The room temperature flow characteristics of a series of Cu-Cr alloys are found to be related to the amount and characteristics of the chromium-rich precipitate. The results are consistent with the theory of Fisher, Hart, and Pry, which considers the trapping to dislocation loops around precipitate particles. THE use of dispersed phases for hardening metals, particularly high temperature alloys, has been practiced for several years. Current theories describing the effects of dispersions on the basis of dislocation theory have been proposed by Mott and Nabarro.l-4 rowan," and Fisher, Hart, and Pry.8 These theories recently have been discussed by Hart,' who pointed out that the experimental data available for comparison with theoretical calculations for overaged alloys are confined to studies by Gensamer, Pearsall, Pellini, and Low,8 Shaw, Sher-by, Starr, and Dorn,9 and some hitherto unpublished data on Cu-Cr alloys which will be described herein. Certain difficulties were associated with the interpretation of the first two studies. Gensamer, Pearsall, Pellini, and Low were concerned principally with interparticle spacing and did not document particle sizes. Shaw, Sherby, Starr, and Dorn included all the data in their report which was necessary for the theoretical evaluation, but their study is subject to some question1" due to difference in prior history of various specimens and due to systematic differences in particle characteristics" which appeared to be associated with differences in the magnification of their microscopic examination. The current study was undertaken to obtain critical data in a manner which avoids some of these limitations. Specimens of eight Cu-Cr alloys remaining from a previous investigation'' were available in the form of 100 mil sheet of the compositions shown in Table I. At 200 mil this material had been previously an- nealed ½ hr at 1000°C, water quenched, and cold rolled 50 pct. Samples were further cold rolled to a thickness of 20 mil for a total reduction of 90 pct. Sheet tensile specimens of standard geometry with a gage length 2½ in. long and 0.20 in. wide were machined from this sheet. These specimens were then annealed in a dried N2-H2 atmosphere for the times and temperatures indicated in Table I. This anneal served two purposes: 1—to obtain a fairly uniform and nearly constant grain size of 0.015 to 0.022 mm average diameter, and 2—to precipitate a fairly uniform, nearly spherical over-aged dispersion of essentially pure chromium. A typical microstructure illustrating the grain size and the uniformity of the dispersion is shown in Fig. 1. A typical dispersion illustrating the range of sizes and shapes of the particles is shown in Fig. 2. The tensile specimens were tested in the Instron equipment described in ref. 11 at room temperature at a strain rate of 0.04 in. per in. per min. Three tests were run on each alloy with sufficient repro-ducibility that a single true stress-true plastic strain curve could be drawn through the three sets of data for each alloy. Typical curves are shown in Fig. 3.

Jan 1, 1956

By N. J. Grant, K. M. Zwilsky

A series of copper-alumina dispersion strengthened alloys were prepared using three different copper and two different alumina powder sizes. Improvements in strength of up to ten times that of pure copper were found in room-temperature as well as elevated-temperature stress-rupture tests. Conductivity values remained from 70 to 90 pct that of pure copper. The best alloys were based on the finest metal powder (1 p) and contained from 7.5 to 10-0 vol pet of 0.018 or 0,033 p alumina. DISPERSION strengthening of metals consists of strengthening a metal matrix by the addition of oxides or other refractory constituents, followed by strain hardening, in order to reinforce the base material and extend its temperature range of application. Since the discovery of SAP by lrmannl and Von zeerleder2 in 1949, many investigators have reported data obtained on a variety of metal-inert oxide alloys.3-7 The present work deals with the mechanical mixing of copper and alumina powders. While several techniques of introducing the second phase have been investigated to date, mechanical mixing so far has been recognized as the simplest, least expensive, and most universally applicable technique that can be used to obtain an ultrafine dispersion of inert particles. In the present case, the effects of changing the metal particle size, the oxide particle size and the volume percent of oxide were investigated. This work is an extension of a previous study.6 EXPERIMENTAL PROCEDURE The starting materials were three different grades of copper powder and two different alumina powders. The three metallic powders were -74 µ(- 200 mesh) electrolytic copper powder consisting of irregular equiaxed particles having a normal screen distribution and analyzing 99.5 pct Cu, and two powders having a 5-µ and 1-µ particle size, average, respectively. The latter powders were produced by the hydrogen reduction of purified sulfate solutions at elevated temperature and pressure and were furnished by the Sherritt Gordon Mines, Ltd. The alumina used was obtained from Godfrey L. Cabot Co., under the trade name of Alon C, one batch of powder having a particle size of 0.018 µ, the other a particle size of 0.033 µ. The crystal structure of these alumina powders was found to be cubic gamma. Each batch of powder was mixed in a Waring Blendor for a total of 15 min, followed by a low-temperature reduction in hydrogen and vacuum compacting in cylinders which were hydrostatically pressed at 35,000 psi to give a compact that could be handled conveniently. Six of the compacts were hydrogen sintered at 900°C while the remaining four were sintered at 1000°C. Subsequently, all compacts were hot extruded at 760°C with a reduction of area ratio of approximately 20 to 1. Evaluation of alloys included density and resistivity measurements, room-temperature tensile tests, stress-rupture tests at 450°C plus one other temperature per alloy, hardness tests, microstruc-tural examination, X-ray analysis of the dispersed phase after dissolving the metal phase, and oxidation studies on two of the alloys. A vacuum-melted, wrought copper rod was tested together with these alloys for comparison purposes. RESULTS Table I is a summary of the compositions, sintering temperatures, densities and conductivity values of the ten alloys prepared. All alumina additions are given in volume percent, and will be referred to as

Jan 1, 1962

By Nicholas J. Grant, Oliver Preston

A series of dilute solid solutions of a1uminum and silicon in copper, in powder -form, were internally oxidized, compacted, and extruded, to produce Cu-A12O3 and Cu-SiO2 alloys with 0.1 to 12 vol pct of oxide. Room-temperature tensile tests, creep-rupture tests from 450oto 850°C. conductivity measurements, and recrys-tallzzation studies were made. Oxide particle size and interparticle spacings were determined in an effort to relate the structure to measured strength values by means of known dispersion strengthening theories. A new dispersion strengthening theory, based on the effect of the dispersed phase on the retention of strain energy from the deformation step, zs proposed for this type of alloy, THE study of dispersion strengthening is of partitular interest in high-temperature materials technology in view of the excellent creep-rupture Properties and thermal stability achieved with SAP.1-5 Considerable effort has been devoted to this study in recent years, and encouraging results have been obtained in a number of systems.6 In general, the most

Jan 1, 1962

By R. A. Wood, R. I. Jaffee, H. R. Ogden, D. N. Williams

Dispersion-containing titanium-copper alloys were prepared having mean free paths varying from 1.0 to 9.7 P. Tensile studies at room temperature and at 1000°F showed that little or no strengthening occurred when the mean free path exceeded 3 p. However, at lower values of mean free path appreczable strengthening was observed. A considerable amount of strengthening also resulted from copper in solid solution. DURING the past several years, an increasing amount of study has been devoted to dispersion hardening, an alloying process in which increased strength is obtained through the addition of a ran-

Jan 1, 1961

By V. R. Thompson, R. C. Westgren

In a recent paper,1 the solid-solution strengthening of tungsten and tantalum in a portion of the W-T;-MO-C~ alloy system was described. Additions of tantalum and columbium to tungsten led to significant strengthening, and the introduction of dispersions into these alloys was suggested as a means of further increasing strength. Dispersion strengthening of columbium and molybdenum alloys has been studied extensively.2-4 The approach used to dispersion-strengthen molybdenum alloys appeared equally suitable for tungsten alloys, and, therefore, a brief study was undertaken to determine to what extent the introduction of carbide dispersions would strengthen a solid-solution tungsten-base alloy. A base composition of tungsten and 10 to 12 pet Cb was selected; previous work1 had shown that the tensile strength of this solid-solution alloy was 50,000 psi at 3000°F. For the study of the effects of small additions of

Jan 1, 1964

By R. M. Goldhoff, J. W. Spretnak

IN connection with establishing the mechanism by which boron enhances the hardenability of heat treatable steels, this research work has been undertaken. Spretnak and Speiser1,2 indicated the need for studying high temperature effects in boron steels and, chief among these, the possible nonhomogeneous distribution of boron in iron grains. The occurrence of this effect may be of significance in explaining the mechanism of enhancement of hardenability by boron as well as in explaining some of the observed characteristics of boron steel behavior. Primarily, then, this work was undertaken a) to establish whether or not boron is adsorbed in ? solid solutions, b) to determine the nature of any such adsorption effects, i.e., whether it is positive or negative as well as its temperature coefficient, and c) to establish the magnitude of any such effect. While past literature on this specific subject is rather sparse, a number of authors have postulated that the observed temporary loss of hardenability as well as the occurrence of the characteristic boron constituent in these steels is due to the nonuniform distribution of boron in austenite.3-0 Experimental Techniques Attempts to obtain direct evidence of grain boundary effects in boron steels have been unsuccessful, both by using the technique of Chalmers' in which small concentrations of solute made radioactive by irradiation are detected by autoradiographic means, and by means of studying cleavage fractures with the electron microscope.8 The present study attempts the design of experiments to measure the effect of boron on properties which might be subject to variation because of grain boundary effects. A study of the theoretical aspects of grain growth indicates that such studies may be pertinent. Another valid type of experiment involves the measurement in polycrystalline materials of properties whose values depend on the bulk of the grain, being negligibly influenced by grain boundaries. One such property is the lattice parameter measured by X-ray diffraction. The value of the lattice parameter reflects the average conditions which obtain over a relatively large number of atoms in a fixed crystalline lattice, and is therefore mainly dependent on bulk grain material. Measurement of lattice parameter variation over the austenite range of temperature in iron with and without boron should give evidence of concentration variations in the bulk grain material if adsorption effects are significant. Finally, use is made of the Grange and Garvey3 metallographic test for boron in which a characteristic boron constituent can be developed in austenite grain boundaries. The very occurrence of this phenomenon suggests grain boundary enrichment of boron and a study of the occurrence of this constituent and its varation with austenitizing temperature should help to detect grain boundary effects. Materials One lb ingots of pure iron, an Fe-B alloy, an Fe-C alloy, and an Fe-C-B alloy were prepared. The melting technique included an initial melting of electrolytic iron under vacuum followed by remelting under hydrogen and, finally, by melting under argon with the addition of alloying elements. Each ingot was surface machined between meltings. A quantitative spectrographic analysis was obtained on the pure ingot and, in addition, oxygen, nitrogen, and carbon were determined by the vacuum fusion method. The results of these analyses are given in Table I. Semiquantitative spectrographic analyses were performed on the alloy ingots, while quantitative determinations of the alloy additions were obtained. These analyses are shown in Table 11. The cast ingots were subjected to the following sequence of fabrication treatments: forging to 5/8 in. bars, swaging to 0.300 in. rounds, drawing to 0.025 and 0.010 in. wire. Portions of each ingot were retained in the swaged condition for experimental use. For certain portions of the work typical commercial steels were used. These materials were supplied by the Republic Steel Corp., Canton, Ohio, and correspond to AISI 8640 and AISI 86B40 compositions.

Jan 1, 1958

By Voyle R. McFarland, Robert A. Burmeister, David A. Stevenson

The distribution of lead between phases in the Ag-Sb-Te system was studied using microautoradio -graphy. Two compositions were investigated, both containing an intermediate phase Known as silver antimony telluride as the major phase, and one containing AgzTe and the other SbzTes as the minor phase. For both compositions, two thermal treatments were used: nonequilibrium solidification from the melt and long equilibration anneals of the as-solidified structure. For each composition, lead was segregated in the minor phase of the as-solidified structure, but was distributed in the matrix after anneal. The electrical resistivity and carrier type were insensitive to the distribution of lead in the two-phase structure. ThERE has been considerable interest in the Ag-Sb-Te system because of its thermoelectric properties. The major interest has been in compositions on the vertical section between AgzTe and SbzTes, particularly the 50 mole pct SbzTes composition AgSbTez (compositions are conveniently expressed as mole percent SbzTes along the AgzTe-SbzTes section). One of the major problems in the proper evaluation and utilization of this material is the inability to control the electrical properties through impurity additions: all alloys prepared to date have been p-type, even with the addition of large amounts of impurities. It has been shown Wit all the compositions previously studied contain an intermediate phase of the NaCl st'ructure as a major phase (denoted by b) and a second phase, either AgzTe or SbzTe3, as a minor phase.'-3 One explanation for the unusual electrical behavior of this material is that the impurity additions have a higher solubility in the second phase than in the matrix; the impurity would segregate to the second phase, leaving the bulk matrix essentially free of impurity.4 In order to investigate this mechanism with a specific impurity element, the distribution of lead between the two phases was determined using autoradiography. Lead 210 was chosen because of the suitability of its 0.029 mev 0 particle for autoradiography and also because of the interest in lead as an impurity in this system.5'6 EXPERIMENTAL PROCEDURE Two compositions were taken from the vertical section between AgzTe and SbzTes, 50 mole pet SbzTes (Viz. AgSbTez) and 75 mole pct SbzTes, in which AgzTe and SbzTes appear, respectively, as the minor phase. Lead containing radioactive lead (pb210) was added to the above compositions to provide a concentration of 0.1 wt pct Pb. The material was placed in a graphite crucible in a quartz tube which was then evacuated and sealed. The samples were melted and solidified by cooling at a rate of 8°C per min and then removed and prepared for microa~toradiography. After autoradiographic examination of these samples, they were again encapsulated and annealed in an isothermal bath at 300°C for a number of days and prepared for examination. An alternate method of preparation employed a zone-melting furnace; the molten zone traversed the sample at a rate of 1.2 cm per hr and the solid was maintained at a temperature of 500°C both before and after solidification. This treatment had the same effect as solidification at a slow rate followed by an anneal for several hours at 500°C. In order to obtain the best resolution, thin sections of the alloy were prepared by hand lapping to a thickness of approximately 20 p. Other samples were prepared for examination by lapping a flat surface on the bulk sample. The resolution, although somewhat better in the former procedure, was adequate in both instances and the majority of the samples were treated in the latter fashion. A piece of autoradiographic film (Kodak Experimental SP 764 Autoradiographic Permeable Base Safety Stripping Film) was stripped from its backing, care being taken to avoid fogging due to static-electrical discharge. A small amount of water was placed on the sample, the film applied emulsion side down on the surface of the sample, and the sample and the film dipped into water in order to assure smooth contact. After drying, the film was exposed for 2 to 5 days, the period of time selected to give the best resolution. The film was developed on the specimen and fixed and washed in place. Two major factors must be considered in establishing the reliability of an autoradiograph: the in-

Jan 1, 1964

By L. F. Mondolfo, W. T. Collins

A study of the relationship between undercooling for nucleation and structure in Sn-Cu alloys with 0.1 to 5 pct Cu has shown that in hypereutectic allojls the halo of tin that surrounds the primary crystals of Cu3Sn5 is larger, the larger the undercooling for nucleation o,f the tin. This increase of halo size results in a decrease of coupled eutectic, and, in alloys far from the eulectic composition, may produce its complete disappeavance, with the formation of a divorced eutectic structure. This was confirnred by the excrrnination of other alloys in which divorced eutectic slructuves are formed, and leads to the conclusion that they ave only an extrenle case of halo forrtzalion , which results when the two phases freeze one at a time and solidification of the first is completed Defove the second starts. It was also found that under proper conditions of nucleation all types of eutectic structures can be formed in the sartte system , and therefore divorced eutectics, like normal and anomalous, are not characteristic of the syslett~, but are mainly controlled by nucleatiorz. Dizlovced eutectics are formed when the phase that tutcleates the eulectic vequires a large undevcooling for ils nucleation and when the cotnpositiorz of the alloy is far from the eutectic., on the side of the primary phase that does not nucleate the other phase. It is recommended that the tevm "divorced" be used in preference to degenerate because it is more desct-iptice of the morphology and mode of forinalion of the structures. ThE variety of structures found in eutectic alloys has been extensively investigated and classified. The most accepted classification is the one by ~cheil,' in which three different types of eutectic were distinguished: 1) normal, 2) anomalous, 3) degenerate (divorced). ATornlal eutectics are typified by the simultaneous growth of the two phases ("coupling") by which the two phases appear as interpenetrating crystals. The presence of a crystallization front, in which the two phases grow side by side, creates the eutectic grains, with the boundaries where the fronts meet. The presence of eutectic grains is the .distinguishing feature of a normal eutectic, according to Scheil. Straumanis and Brakss2 examined the Cd-Zn system and showed that there was a crystallographic relationship between the phases. Later, others4 also investigated additional systems and found definite crystallographic relationships in the coupled eutectics. The anornalous eutectic shows much less coupling than the normal; the two phases are intimately mixed but 'grow without a uniform crystallization front—a consistent crystallographic relationship— and the eutectic grain is conspicuously absent. As in the normal eutectics faster rates of growth result in a finer structure, but there is not the typical uniform spacing of normal eutectics. The degenerate eutectic shows no coupling; in fact the two phases attempt to minimize their area of contact and to form separate crystals. It has been suggested5" that slow cooling may favor this type of structure. Scheil believes that normal eutectics are formed when the two solid phases are present in more or less equal proportions, whereas both anomalous and degenerate eutectics form when one of the phases is present only in small amounts. spengler7 extended much farther this qualitative relationship between the eutectic type and the ratio of the two phases, and added a relationship to the melting point of the constituents. On this basis he proposed two equations for determining into which of Scheil's classifications an alloy belongs. The first equation is: where TI is the melting temperature of the lower-melting component, Tp of the higher-melting component, and Te the eutectic temperature. The second equations is: where is the volume percent of the lower-melting phase and $2 of the higher-melting phase at the eutectic composition. If 0 and/or 4 are in the range 0.1 to 1, a normal eutectic is formed; if in the range 0.01 to 0.1, anomalous; if less than 0.01, degenerate. Although the examples given by Spengler show a good agreement with the formulas, chadwick found that the Zn-Sn eutectic is normal to all growth rates, even though the volume ratio is 12/1, and Davies9 reports that the A1-AlgCo2 eutectic is normal, with a volume ratio of more than 30/1. Many more discrepancies of this type can also be found. Neither Scheil nor most of the other investigators have considered nucleation as a factor in the formation of divorced eutectics. Daviesg states that divorced eutectics form when neither phase acts as

Jan 1, 1965

By J. B. Cohen

EARLY investigators1-4 reported that additions of beryllium to silver caused embrittlement. The in-termetallic phases which are now known to exist5,' were not known at the time. Solidification appeared to occur by a simple eutectic reation.4,7 (The primary beryllium-rich phase appeared globular, when the liquidus temperature was near the melting point of pure beryllium, and angular when the liquidus was much lower; in discussion of Sloman's work,4 it was suggested that this difference was due to an intermetallic compound which was not revealed by the polishing technique.) In the investigation reported here, it was found that if the compounds now known to exist were avoided, certain beryllium-silver alloys containing large amounts of the primary beryllium phase possessed sufficient ductility to be cold formed. Alloys containing 14 and 20 at. pct Ag were prepared from vacuum-cast beryllium containing 1.0 to 1.2 pct impurities and silver, 99.99+ pct pure. The elements were melted together by induction in beryllia or alumina crucibles under argon at less than 1 atm pressure. The cooling rate was adjusted by regulating the temperature of the melt and by reducing the power to zero without stopping the water flow in the induction coils, which were in contact with the crucible. Ingots were about 0.75 in. in diam with a grain size of 0.12 to 0.25 in. Microscopic investigation of transverse or longitudinal sections of an ingot cast from a temperature well above its melting point showed that, in general, the compounds did not form, presumably because of rapid cooling; the alloys consisted of beryllium-rich particles in a silver-rich matrix as shown in Fig. 1. A eutectic structure was not observed, probably .due to the low beryllium concentration of the eutectic composition and eutectic divorcement. Within a given grain, the small globules of the primary phase were all oriented in the same direction, as indicated by examination under polarized light; the orientation varied from grain to grain. This suggested that the globules were branches of the same dendrites. If cooling was too slow, a third constituent was observed in the microstructure; because it formed around the primary beryllium-rich globules it was Fig. 1—Ag-Be alloy, 18.6 at. pct Ag, as-cast. Bright field illumination. Etched with an aqueous solution of NH,OH and H2O2. X500. Reduced approximately 21 pct for reproduction. probably the 6 compound. A thin ring around an ingot often contained this constituent. When this compound was present in large quantities, or the silver segregated so that large regions occurred with the beryllium-rich globules in contact, sections from the ingot were brittle. When the compound was initially absent, it was possible to carry out reductions in thickness of 50 pct by cold rolling with only limited edge cracking; failure did not occur until reductions of 80 pct. A typical cross section after cold rolling is shown in Fig. 2. After simple bend tests, microcracks were found; these apparently were difficult to propagate in the two-phase structure, as the specimens continued to behave in a ductile manner after the cracks appeared. Similar cracks would probably be closed during rolling. It is apparent in Fig. 2 however, that the primary beryllium particles undergo extensive plastic deformation. Hardness values of the alloys indicated that some work-hardening occurred during cold working. Fig. 2—Ag-Be alloy, 18.6 at. pct Ag, reduced 83 pct in thickness by cold rolling. Bright field illumination. Etched with an aqueous solution of NH,OH and H2O2. X500. Reduced approximately 21 pct for reproduction.

Jan 1, 1961

By W. A. Backofen, G. Y. Chin, W. F. Hosford

The ductile fracturing process was studied in single-crystal and poly cvystalline aluminum deformed in tension over a temperature range from 295° to 4.2°K. At temperatures as low as 77°K, the fracture of "inclusion-free" material, including zone-refined aluminum, was by rupture (-100 pct RA). At 4.2 OK, fracture was brought on by adia-batic shear. Metallographic examination did not disclose any voids or slip-band microcracks, thus negating for inherently ductile metals any mechanism of void nucleation by vacancy condensation or of cracking due to dislocation pile-ups. In Izigh-purity aluminum not treated to be inclusion-free, fracture at temperatures as low as 45°K was of the double-cup type and a result of void formation. The reduction-of-area decreased as temperature was lowered, corresponding to the earlier appearance of voids. Such behavior was rationalized in terms of a larger increase, with decreasing temperature, in the .flow stress relative to the strength of the inclusion-matrix interface. Evidence for low-temperature adiabatic shear was found in discontinuous flow at 4.2"K, in the transition to a localized shear fracture at low temperatures, and in the suppression of shear fracture with an elastically hard pulling device. A simple analysis for the initiation of adiabatic shew permitted a general correlation of the various contributing factors. It has been pointed out that the duration of shear depends upon effective mass and elastic stiffness of the deformation system. IT has long been recognized that fracture* may Throughout this paper, the term "fracture" is taken to mean any process that results in the separation of a material into two (or more) parts. Thus rupture as it may be encountered in a tension test leading to 100 pct reduction-of-area is included in this category. occur in a ductile mode, and that the process can be of great practical as well as general interest. Much information about ductile fracture has also been accumulated over this period, but only recently has an understanding of mechanism begun to appear. Ludwik,' in 1926, first reported fracture in a tensile specimen starting with a central crack in the necked section. Since then, other studies have disclosed that such cracks may form by the coalescence of voids nucleated in this region where hydrostatic tension is highest.2-4 Rogers and Crussard et al.' have emphasized void formation and reori-entation along localized shear bands as a mode of crack propagation. pines6 has considered the tensile rod as a bundle of fibers joined by weak interfaces, which subsequently separate to allow individual fiber contraction. The notion of cavity growth and coalescence by purely plastic processes was discussed by Cottrell: who added that the tensile reduction-of-area ought not to be sensitive to temperature. On the other hand, it has been observed that the reduction-of-area is greatly increased if tests are carried out at high temperaturesa or under high hydrostatic pressure.' Fracturing anisotropy in wrought products lends support to the idea of void formation from preexisting flaws strongly aligned by earlier processing.''-l2 There is evidence that many voids result from the fracturing of inclusions or separation at the inclusion-matrix interface Another possibility is that voids grow out of pore volume produced in the initial solidification and never fully removed in later working. In general, a structure 3f particles, pores, and weak interfaces can be expected, at least in materials of engineering interest. Vacancy condensation has been suggested as an alternative mechanism of void formation for materials considered to be inclusion-free.13 Yet experience has shown that tensile reduction-of-area increases with purity, to the extreme of rupture as so often observed in single crystals. Adiabatic shear has an important bearing on ductile fracture. It occurs when the decrease of flow stress, as a result of local temperature rise from heat generated during straining, becomes larger than the increase due to strain and strain-rate hardening. As demonstrated by experiments on punching of plates,14 a large temperature rise may be brought about by rapid straining. Adiabatic flow as a result of the high strain rate reached in an ordinary tensile specimen just prior to separation may account for the cone formation in cup-and-cone fracture;14 evidence of such local heating has been presented.15 For geometrical reasons, however, pure sliding along the conical surfaces is unlikely, and separation under tensile forces is probably an important accompanying feature of the shear.7 In deformation processing operations, a high shear-strain rate may exist at boundaries between plas-

Jan 1, 1964

By J. Greenspan

The anisotropy of fracture and slip, that is, the brittleness and ductility of the beryllium single crystal, is characteristic also of po1ycrystalline beryllium in which the grains are oriented in a preferred manner. Beryllium with grain orientations resulting from hot working either unidirectionally or bidirectionally is ductile or brittle in directions given by the anisotropy and orientation of the grains. Textures developed from various hot-working sequences are given, and tensile results are correlated with textures. Not only texture, but fine grain size is necessary for obtaining high tensile elongation. Ultimate tensile strength and elongation are both correlated with grain diameter. Of particular interest is the case in which unusually high tensile elongations of 30 to 40 pct are obtained when basal planes are oriented to carry THE beryllium crystal (of nominal purity) is probably unique among metals with respect to its unusually acute anisotropy of slip and fracture. While extensive slip can occur on the system (1010) [ll20] starting at about 19,000 psi, the (0001) plane is particularly vulnerable to fracture at 4500 psi or less. Fracture occurs also on the (11zo) plane at about 26,000 psi.1"3 Further, there seems to be no other slip System which contributes significantly to

Jan 1, 1960

By Erhard Hornbogen

Twins were propagated into large, well-annealed crystals of a, iron-phosphorous and a, iron-molybdenum solid solutions. Strain fields caused by interaction of these twins were made visible by precipitation. These strain fields can be explained by assuming that transmitted and reflected plastic waves are created when a twin strikes an obstacle with high velocity. The twin front itself can be regarded as a shear wave that may be able to develop a one-dimensional shock front. The occurrence of {WO} fracture by the reflected stress is also discussed. The described effects were not found when twins propagated at lout velocity into disturbed crystals, for example, into crystals previously deformed by slip. DEFORMATION twinning leads to shearing of a crystal lattice under external load. The shear appears to be homogeneous over a large number of crystal layers. In a iron, shear occurs on the (112) planes and in the <1ll> directions. The formation of a twin proceeds by the movement of partial dislocations that change the stacking at the twinning plane.l, In the bcc lattice, this partial dislocation can be created by dissociation of a sessile a/2[111] dislocation that lies in the (112) plane: Cottrell and Bilby&apos;s spiral mechanism can explain the homogeneous shear, but has not been proven experimentally. Nevertheless, this mechanism can explain the possible high velocity of formation of a twin. In Cottrell and Bilby&apos;s model, as in other models of twinning or phase transformation, a much higher energy is required to spread the first stacking fault than to add new layers by further movement of the partial dislocations. The dislocations will therefore become accelerated proportional to the difference between the stress, tN, to nucleate and tp to propagate the twin. It has been assumed3,4 that tn is the stress required to make two partial dislocations of apposite sign pass for the first time. After the first layer of new stacking has been formed, the maximum stress is TN=Gb/4pd where G is the shear modulus, b is the Burgers vector of the partial dislocation, and d is their perpendicular spacing so that the amount of shear is S = b/d. From this formula, a value of 25 kg per mm2 has been estimated for twinning in Zn3. The measured values are smaller, but local internal stresses may help to provide the necessary stress. For a iron, a higher value can be expected, mainly due to the higher value of G. That agrees with the observation that "burst" twinning occurs if the yield strength is in-

Jan 1, 1962

By N. K. Chen, R. B. Pond

IN the study of slip band* formation, there have been many examples to show that they do not always appear as lines traversing the entire crystal, but as segments whose ends seem to vanish in their path through the crystal. This characteristic appearance of slip bands has been witnessed under the optical microscope at various magnifications&apos;-&apos; and also under the electron microscope.&apos; A typical example of this behavior seen with the optical microscope is shown in Fig. 1. However, the slip band studies were generally conducted on polished surfaces of single or poly-crystalline metals which had been previously deformed; i.e., the load was generally released so that the observation could be made. Any picture of slip bands so obtained can represent the surface phenomenon only in a static state of the strained material. The conditions prior to their formation cannot be definitely and clearly assigned. Thus, while a segmented slip band may suggest that slip is a growth process as supposed by the theory of the nucleation of slip,V he usual appearance of suddenly and fully developed slip bands around the crystal has generally been considered as a consequence of uniform shear of the entire slip plane akin to a cataclysmic process. Little clear-cut information is available with regard to the speed at which a slip band forms, its direction of motion, the geometry of its position with regard to its neighbors, and its dependence on orientation. A description is given in this paper of experimental apparatus by which the progressive formation of slip bands can be recorded while the specimen is undergoing deformation. Qualitative and quantitative data on the dynamic formation of slip bands will be presented with special interest concerning the propagation of slip bands, the spacing of slip bands, and their relations to strain hardening. Views on the formation of slip bands are discussed and a mechanism of the unit process involved in the formation of a slip band is proposed. Preparation of Specimens Single-crystal specimens of high purity aluminum (99.997 pct), 1/8 in. square in cross-section and 1% in. long in gage length, were made by the method of gradual solidification from the liquid state. Since no machining work could be introduced in preparation of crystals of such small size, a special mold was designed for casting them to final shape. The mold consisted of two, separate, high purity graphite blocks. Generally, 20 molds were packed together in one container so that 20 specimens could be obtained by one casting operation. This was desirable since it was possible by this method to obtain groups of crystals with similar orientations. The as-cast crystals were carefully clipped from the gate, etched, and homogenized for 24 hr at 600°C. They were then very gently polished using a 4/0 paper, re-etched, and finally electrolytically polished after the method previously described by Chen and Mathewson.6 The crystallographic orientations were determined using a back-reflection, Laue method. Tensile Testing and Photographic Method The tensile testing equipment for these tests was composed of a specially designed microtensile machine, microload cell and microclip gage with necessary appurtenances. The members of the microtensile machine consisted of three parts, as shown in Fig. 2. The chassis is equipped with an oil cylinder, A, and piston, the piston being part of the movable cross-head, B. Pressure in the oil cylinder is controlled and regulated by an external pneumatic-hydraulic cell, C, which is connected to the cylinder by a Vs in. high pressure copper tube. This cell is half filled with hydraulic oil and has a needle valve, D, on the oil exit side as well as a needle valve, E, on the gas inlet side. A quick-acting valve, F, as well as a pressure gage is provided for the gas side. The oil exits into the load piston on the microtensile machine. By connecting a tank of inert gas to the gas inlet, regulated pressures were provided over the oil so that the oil would leave the cell at a rate determined by the setting of the exit needle valve, the gas pressure, and the pressure of the oil in the load piston of the microtensile machine. With this pneumatic-hydraulic appurtenance it was possible to

Jan 1, 1953

By Harry L. Brown, Philip E. Armstrong

Young&apos;s modulus doto ore presented for W, Mo. Ta. V, Cr. Ni, Ti, and Zr as a function of temperature up to about 0.7 of the melting points. A plot of reduced temperature us reduced modulus produces a general correlation except for titanium and zirconium. Young&apos;s modudlus data for some representatiue commercial graphites are given up to 2400°C. All results were obtained from longitudinal resonance-frequency measzrrements of 4-in. long samples, corrected for simultaneously measured thermal expansion. Comparison of poly crystalline and single-crystal tungsten behavior indicates that all the observed modulus values are frequencv -dependent at the higher temperatures. STATIC measurements of elastic properties of materials at elevated temperatures are extremely difficult to make since the elastic behavior is obscured by relatively large first-stage creep effects. Dynamic methods which avoid this difficulty by utilizing the measurement of the resonance frequency of a simple vibrational mode have been described in the 1iterature.l-3 The authors have extended the operating temperature range of this type of measurement to above 2000°C and have described the equipment elsewhere.~ The Young&apos;s modulus data presented here were obtained with this equipment, using samples about 4 in. long and 1/4 in. diam, supported horizontally at the center and excited to the fundamental longitudinal resonance frequency. Young&apos;s modulus was calculated by the formula: E = 4f2f2p where E = Young&apos;s modulus, / = frequency, 1 = length, and p = density. Dilatometric data were obtained simultaneously by optical measurements of the change of length of the sample. For the metals tested, an attempt was made to obtain modulus values that would be representative of the pure, strain-free, randomly oriented material. All of the graphites studied exhibited a high degree of elastic anisotropy, while the method assumes an isotropic homogeneous material. However , when the values for Young&apos;s modulus obtained for AUC graphite were compared with those taken from static measurements at room temperature they were found to be only about 4 pet higher. This is excellent agreement considering the low precision of modulus calculations from stress-strain curves. The difficulty in using data of apparent Young&apos;s modulus for anisotropic materials arises when the attempt is made to use these data and shear-modulus data to obtain Poisson&apos;s ratio. The results are usually meaningless and sometimes absurd.&apos; RESULTS Table I lists the sources, impurity content, thermal history, and dilatometric constants of the metal samples that were investigated. All the metal samples were tested under a vacuum of 1 to 5 x 10-5 Torr. Where our dilatometric data agreed with previously published data, the table lists the previous values and the literature reference. Where no reference is given, the constants were determined from our dilatometric data. Table II presents the modulus data in units of millions of psi as a function of temperature in degrees centigrade. The numbers were obtained from curves fitted to the experimental data. For some types of calculations, such as those attempting to correlate creep data, the slope of the modulus curve is of interest, and this information also is reported in Table 11. It is always tempting to try to fit physical-property data into some general pattern and then use this pattern to predict the behavior of other similar materials. The general pattern for an elastic modulus of a metal seems to be a linear deoro- with increasing temperature Over the range between 0.1 and 0.4 of the melting temperature, then a progressive1y decreasing value up to the melting point. Metals appear to retain finite moduli

Jan 1, 1964

By R. L. Fleischer, W. F. Hosford

Tensile data for coarse grained aluminum Polycrystals suggest that the "grain size" effect is not due to dislocations piled up at grain boundaries but rather is primarily a relative size effect due to surface crystals being weaker and less confined. STUDIES directed at interpreting hardening of poly-crystalline metals normally identify their strain hardening properties with those in some particular type of single crystal.1"4 The recent recognition in face centered-cubic metals of a nearly linear stage with rapid hardening occuring at comparable rates for both polycrystals and single crystals, suggested that the same process or processes determine both cases and hence that there exists some justification for the use of single crystals to understand polycrystals. Further evidence for the above view may be found by an approach initiated by Chalmers:5 By using bicrystals of controlled orientation it is possible to begin to assemble a polycrystal and also to study grain boundary effects in detail. In this way it has been found that a single grain boundary affects easy glide but not the subsequent stage II hardening.6 This result suggests that a sensitive way to observe grain boundary effects in polycrystals would be to vary grain size and measure easy glide. As will be seen, easy glide is only possible for coarse-grained samples, and hence the results will serve to fill in the gap in measurements between single crystals and bicrystals on one hand and fine-grained polycrystals on the other. One problem inherent in comparing single crystals with polycrystals is the uncertainty as to what slip systems are acting in a polycrystal. To compare the two types of samples, rates of shear hardeninn---L. on the acting -planes are needed. and these may be computed only if it is known what particular systems are active. The acting systems were examined for a coarse-grained polycrystal and it will be shown that the systems supplying the preponderance of slip can be determined with little ambiguity. EXPERIMENTAL PROCEDURE Twelve samples of aluminum were prepared by chill casting into a heated graphite mold, followed by annealing at 635° ± 5°C for 24 hr with an 8-hr fur- nace cool, and finally either etching7 or electropol-ishing.&apos; The samples, with a 7 to 10 cm length between grips and 4.4 by 6.6 mm in cross section, were deformed at a strain rate of about 3 10 -3 . per min in a tensile device which has been described elsewhere.5 The composition was reported by Alcoa as 99.992 pct Al, 0.004 pct Zn, 0.002 pct Cu, 0.001 pct Fe, and 0.001 pct Si; nine samples were deformed while immersed in liquid helium and three in air at room temperature. The stress-strain curve for one of the samples (P-1) deformed at 4.2 "K has been reported previ~usl~.~ This sample was selected for determination of active slip systems. Eighteen of the crystals were examined by optical microscopy to determine the angles of slip line traces and by X-ray back reflection to determine orientation. By this means the slip planes were determined and the resolved shear stress factors for possible slip systems could be computed. Finally each sample was sectioned so that after etching, the number of crystals could be counted for each of ten newly exposed surfaces. The average of these ten values will be termed n, the number of crystals per cross section. Values of 11, varied from 1.9 (nearly bamboo structure) to 12.7. Sketches of typical cross sections appear in Fig. 1. RESULTS AND DISCUSSION: SLIP SYSTEMS 1) Determination of Acting Slip Planes—The stress axis orientation and operative slip planes in eighteen crystals of sample P-1, as determined by slip line traces and crystal orientation, are summarized in Fig. 2. For one of the crystals two planes had a common trace. so that the traces alone did not distinguish which plane or planes were slipping. However it was found that the stress resolving factor for the primary system was 0.386, .while that for the most stressed system in the other plane (indicated bv the dotted arrow) is 0.138. It will be assumed tgerefore that only the primary plane acted. Since the orientations were determined after extending the samples 4 pct, the stress axes may be rotated from their original value by as much as 2 deg in some cases. It is interesting to note that in five crystals only one slip plane acted, in eight two acted, and in five three planes were observed—an average of two slip

Jan 1, 1962

By R. A. Vandermeer, P. Gordon

The study of recrystallization in cold-worked metals is complicated by the impossibility of making direct three-dimensional optical measurements. Metallo-graphically, theinternal three-dimensional features of samples must be inferred from their intersections with the two-dimensional plane of polish. The inherent difficulties involved in this procedure have led— particularly during the last few years—to some instances of rather doubtful interpretations of experimental data in terms of nucleation and growth rates in recrystallization. In the present paper an analysis ; is given of the recrystallization characteristics in a polycrystalline alloy containing 0.008 wt pct copper in ZONE-refined aluminum *and an effort is madf to reemphasize the necessity in such studies for attention to the proper statistical relationships between one- two- and three-dimensional features of opaque samples. It is shown that, in common with several recently described cases for other types of transformations,1-5 recrystallization in this alloy can best be rationalized on the basis that all nucleation took place essentially at time zero—i.e., there is no significant time-dependent nucleation during annealing—and that the kinetics are controlled exclusively by the growth-rate parameters. This is true despite the fact that the number of

Jan 1, 1960

By M. J. Marcinkowski, A. Szirmae, R. M. Fisher

Room -temperature hardness measurements obtained from single and polycrystalline samples of a 47.8 at, pet Cr-Fe alloy which were aged for various times al 500°C show a two-fold increase over that of the unaged alloy after annealing for 1000 hr. A detailed examination of the deformation markings in the neighborhood of the hardness imbressions reveals that twinning becomes an increasingly more important mode of deformation as aging proceeds. this observation is shown to be inconsistent with the Proposals that the transformation is eve of order-disorder On the other hand, the results are in agreement with previous observations of Fisher et al. of a coherent chromium-rich precipitate which forms in an iron-rich matrix as a result of a miscibility gap in this alloy system as proposed by Williams . Using the coherent precipitate hypotlzesis as a model, a detailed analysis of the various possible strengthening contributions to both the slip mid twinning stresses is made. In the case of&apos; slip, lattice friction within the chromium-rich phase and the chemical energy associated with the interface between the two phases contribute about 60pet to the total strength. The contributions from coherency strains and modulus differences are thought to contribute the remaining 40 pet of the strength but are difficult to evaluate because of uncertainties regarding the flexibility of the dislocation line. All of the factors have nearly the same or else a smaller effect on the twinning stress in the aged alloy. Twinning is never observed in the unaged alloy because the twinning stress is much higher- than that for slip. With increased aging times, the various contributions to the total stress for both twinning and slip increase, but most of those for slip increase much more rapidly, so that, in the fully aged alloy, it surpasses the stress for twin propagation. When a twin is nucleated by the chance occurrence of an internal stress concentration during a test, the subsequent twin burst results in a low hardness reading. FeRNTIC Fe-Cr alloys from about 15 to 80 at. pet Cr content show considerable change in properties such as hardness, electrical resistivity, saturation magnetization, and so forth, after aging in the vicinity of 500°C.* The most marked change is a large increase in hardness accompanied by a sharp decrease in ductility, so the phenomenon has often been referred to as the 475°C (885°F) em-brittlement problem. Changes in microstructure are rather subtle and most investigators have not been able to observe any X-ray diffraction or metal-lographic changes during aging. There is little doubt that the phenomenon is inherent to the Fe-Cr binary system and is not directly related to the formation of a phase. Two distinct schools of thought have developed during the past decade concerning this problem. Masumoto, Saito, and sugiharal have concluded, from their own specific-heat measurements, that atomic ordering occurs in this temperature range. Pomey and Bastien2 have also attributed the changes in physical properties with aging in the neighborhood of 500°C to atomic ordering. In addition, Takeda and Nagai3 claimed to have found X-ray verification for superlattices corresponding to the compositions FesCr, FeCr, and FeCr3. However, all known attempts to observe superlattices in Fe-Cr alloys using neutron diffraction, which should be much more sensitive, have been unsuccessful. These have been reported by Shull et al.4 for 25 at. pet Cr, williams5 for 75 at. pet Cr, and Tisinai and samans6 for a 28.5 at. pet Cr. steel. The second and more prevalent opinion ascribes the 475°C embrittlement phenomenon to the formation of a coherent precipitate due to the occurrence of a miscibility gap in the Fe-Cr system below about 550°C. This latter concept was first indicated by the results of Fisher, Dulis, and Carroll,7 who were able to extract fine particles about 200Å in diameter from samples of 28.5 at. pet Cr steels aged from 1 to 3 years at 475°C. The extracted material was found to be nonmagnetic, to have a bee structure with a lattice parameter between that of iron and chromium, and to contain about 80 at. pet Cr. Williams and paxton8 and williams5 confirmed these results and first explicitly proposed the existence of a miscibility gap below the a region in the equilibrium diagram. Alloys aged within this gap

Jan 1, 1964

By L. J. Dijkstra, R. J. Sladek

IN earlier work the effect of manganese on the general behavior of nitrogen in iron was the subject of a careful examination by Fast.&apos; Part of the investigation was made, in collaboration with one of the present authors (L.J.D.), by using the internal friction method. This technique, which has been described elsewhere,&apos; showed some significant changes in the properties of Fe-N alloys by the addition of a small amount of manganese namely: 1—Under standard conditions the maximum of the nitrogen peak in pure iron appears around 20 °C and the half width is about 27 °C. By addition of 0.5 atomic pct Mn a general broadening of the internal friction peak occurs. The half width is increased by about 9°C. Moreover, a shift of the maximum of about 5°C toward a higher temperature takes place, Fig. 2. 2—The precipitation of nitrogen from the supersaturated solution, measured, e.g., at 200 °C, was inhibited by the addition of the amount of manganese given in effect 1. The broadening of the nitrogen peak was confirmed in more recent work3 on samples which were prepared from Westinghouse Puron iron. The effect of manganese on the rate of precipitation, however, was much less striking than in the earlier work&apos; based on Philips high purity iron. Theory To explain the first change in properties just referred to, effect 1, the following two assumptions were made:&apos; First, every manganese atom in sub-stitutional solid solution creates around it six interstices located between an iron and a manganese atom, as is evident from Fig. 1. These interstices, conveniently called Fe-Mn interstices, have for nitrogen a free energy level which is AG* lower than the normal Fe-Fe interstices between two iron atoms. For such low concentrations of manganese as 0.5 atomic pct, the possibility that two manganese atoms will be nearest neighbors can be disregarded. Second, it is assumed that in addition to the normal relaxation time TFe at least one new relaxation time, Mn, Fe, enters in the phenomena which is associated with the elementary diffusion jumps between a Fe-Mn interstice and an adjacent Fe-Fe interstice. Fig. 1 gives schematically the free energy for a nitrogen atom as a function of its position along the line ABC. A tentative analysis of the nitrogen peak in a 0.5 pct Fe-Mn alloy based on these assumptions has been made in Fig. 2. The dotted curve with a maximum at 22°C is caused by jumps between Fe-Fe interstices, the dotted curve with maximum at 32°C by jumps between an Fe-Mn and an Fe-Fe interstice. The two effects added together give the dash-dot internal friction curve for the alloy. In the following these two components will be referred to as the normal peak at 22°C and the abnormal one. (The small secondary peak at

Jan 1, 1954

By A. T. Robinson, J. E. Dorn

The electrical resistivities of aluminum alloys containing CU, Ge, Zn, Ag, Cd, and Mg were found to increase linearly with the atomic percentage of the solute atoms. Application of Linde&apos;s rule to these data suggests that each aluminum atom contributes 2.5 electrons to the metallic bond. ALTHOUGH aluminum invariably exhibits a Jt\. valence of three in ionic solids, it does not follow necessarily that its valence in the metallic state is also three. As a matter of record, many properties suggest that the number of bonding electrons in metallic aluminum is less than three. For example, as shown in Fig. 1, the atomic radii1 of the metallic elements in the solid state decrease with increasing atomic number at the beginning of any one period. The atomic radius of aluminum, however, is greater than that which would result from a regular decrease in atomic radius with increasing atomic number, suggesting that the number of bonding electrons in aluminum is somewhat less than three. Correlated with the anomalously large atomic radius of aluminum is its abnormally low melting temperature. As the atomic numbers increase in any one period, the melting temperatures of the metallic elements increase.&apos; As shown in Fig. 2, however, the melting temperature of aluminum (13) is only slightly greater than that of magnesium (12). Again the number of bonding electrons in metallic aluminum appears to be nearer two than the expected value of three. In addition Ageev and Ageeva3 alculated the electron density distribution in the aluminum lattice. A comparison of such electron density curves for various assumed valence states of aluminum with experimentally determined curves suggests that the actual number of bonding electrons in metallic aluminum is between two and three. Hume-Rothery and Raynor&apos;, " have suggested that the above-mentioned abnormalities are associated with the overlapping of the Brillouin zones of metallic aluminum when it is in the usual face-centered cubic structure. The importance of the number of bonding electrons in metals has been extended recently into the field of mechanical properties. While formerly it was thought that the plastic properties of alpha solid solutions were primarily dependent upon the lattice strain induced in the solvent by the solute, it is now known that the difference in the number of bonding electrons of the solute and solvent also has its influence on the plastic properties."1&apos; In order to achieve a satisfactory correlation of the effect of alloying elements on the plastic properties of aluminum alloys,"&apos;T however, it was necessary to assume that the number of bonding electrons contributed by aluminum was about two rather than the expected value of three. It appeared to be desirable, therefore, to seek other criteria for evaluating the number of bonding electrons contributed by aluminum to its alpha solid solutions, especially in view of the practical as well as the theoretical importance of the effect of valency on the plastic properties. Norbury&apos; was perhaps the first to illustrate that the increase in resistance of alpha solid solutions is associated with the differences in valence of the solute and solvent metals. In a series of illuminating investigations Linden-&apos;&apos; has shown that the electrical

Jan 1, 1952