Search Documents

By L. W. Davies

A problem often encountered is the provision of materials which have impurity contents below a certain specified level. This problem is in some cases solved by making use of the segregation of the impurity with respect to the liquid and solid states of the material: an ingot of the material is subjected to directional solidification and the ends of the ingot. which are richer in solutes which respectively raise and lower the melting point of the solvent, are removed. An important extension of this process is the zone-refining process invented by pfann,' in which molten zones are passed successively along the length of an ingot and displace solutes towards the ends. Pfann's process is particularly appropriate in cases for which the segregation coefficient k does not differ greatly from unity, or when the impurity content must be kept extremely low (-=,6 atom fraction); it is widely used, particularly in the purification of materials for use in semiconductor electronic devices. It is the purpose of this paper to investigate theoretically the efficiency of the above refining processes; that is! we determine the most economic method of attaining a given yield of solvent of specified purity, taking into account the variables of the process. One variable of interest is the effective value of the segregation coefficient. which depends on the rate of movement of the liquid-solid interface and on the degree of stirring in the liquid; in addition to this, in the zone-refining process we are also interested in the choice of zone length for most efficient operation. A mathematical determination is made here of the lengths of successive molten zones which give, for an initially uniform distribution, the most efficient removal of a solute of arbitrary distribution coefficient. In order to be able to give a quantitative treatment of efficiency, we first define figures of merit applicable to a given distribution of solute. FIGURE OF MERIT Let an ingot of length L contain a distribution y(x) of solute: we confine our discussion to mono-tonic distributions, in order to avoid inconsistencies in the subsequent analysis. The first moments of the distribution about the origin and x = L are respectively L F = hy{x)dx [I] G = f(L -x)y(x)dx = YoL2-F, [21 where yoL is the total amount of solute present. We shall see that the quantities F and G are useful figures of merit for a discussion of the efficiency of zone-refining processes. In the first place, it is

Jan 1, 1960

By R. W. Fenn, H. A. Lepper, W. R. Hibbard

THIS paper reports the results of static tension and torsion tests made on single crystals of alpha brass for the purpose of determining its elastic coefficients. 70-30 alpha brass was chosen because it has been used extensively in the study of the plastic deformation and because it has the following desirable characteristics: (1) face-centered cubic structure, (2) marked elastic anistropy, (3) low elastic modulus. In 1928 Masima and Sachs' using static methods reported the following values for the elastic coefficients of 72-28 pct alpha brass: S11 = 19.4 X 10-13' cm2 per dyne S12 = —8.35 X 10"' cm2 per dyne S44 = 13.9 X 10-13= cm2 per dyne These values were obtained from the measurements shown in fig. 1. The experimental points for both tension and torsion lie consistently below the lines drawn to represent them. Fig. 2 shows the stress-strain curves for their specimens. Only a few points in the elastic range of these curves were available to obtain their values for Young's modulus. Therefore, it was considered that a re-evaluation of these data was desirable. In addition a variation of from 1.13 to 1.485 kg per mm2 was found in the values of the critical resolved shear stress for 70-30 alpha brass reported by Burghoff,² Treuting and Brick: Martin,4 Maddin, Mathewson and Hibbard5,6 and von Goler and Sachs.' Therefore, it was desired to evaluate the conclusion of Boas and Schmid8 that a constant resolved shear stress and not a constant elastic shear initiates plastic deformation. Theoretical Considerations*: The elastic deformation of cubic crystals may be defined by three elastic coefficients (S11, S12, S44,) or three elastic parameters (C,,, C,, C,,). These coefficients or parameters are sufficient to define the stress and strain relations for cubic crystals in the generalized form of Hooke's law referred to Cartesian coordinates. Elastic parameters and coefficients in the cubic system are related by the expressions: C11 —S11+S12/(S11-S12)(S11+2 S12) C12=-S12/(S11-S12)(S11+ 2 S12) (S11 — S12) (S11 + 2 S12) [2] C44 =1/S44 PI It is possible to determine all the elastic coefficients for a cubic metal by measuring two of the three elastic constants: Young's modulus (E), torsion modulus (G), or compressibility (S). The following relations exist between crystallographic orientation and E, G and S for cylindrical crystals of the cubic system: 1/E = S11 — 2 [(S11 - S12) — Su] (?1² ?2² + ?2²?83 + ?3² ?1²) [4]

Jan 1, 1951

By R. W. Fenn, H. A. Lepper, W. R. Hibbard

A. J. Shaler—I should like to congratulate the authors on the presentation of this paper, which we have been awaiting a long time. The view they have taken of the sintering process, namely that voids are comparable to particles of a second phase capable of dissolving to a slight extent in the metal phase, is an interesting but dangerous one. Pines has given a first-order solution of the problem and Has predicted a decrease in the concentration of pores near the surface. The authors of the present paper have experimentally demonstrated the second-order effects, that is, the diffusion of the space phase between voids of different sizes, and also the movement of atoms from highly curved parts of a void to less curved regions of the same void. I have recently calculated the rate of vapor-phase transport within odd-shaped pores in copper at 850 °C and found that probably at that temperature volume flow is the more rapid shape-changing mechanism. The authors of the present paper have found that at 1000°C, vapor-phase transport is dominant. There is no disagreement here, I think. The dominant phenomenon can easily be different in different temperature ranges and with different pore shapes: I used a cylinder going to a sphere, not a tricuspid going to a "triple cuspidor." But there is some danger in the other art of the explanation, that is, in the study of the diffusion of vacancies from a pore to its neighbors and to the outside. A vacancy is very different in dimension from an atom. When it moves, it carries a field of elastic strain. This has to be taken into account. A concentration gradient of vacancies can cause them to diffuse. But in the absence of such a gradient, they can be made to diffuse by a field of stress. What happens in some of our experiments, and it is most noticeable in the shrinkage of synthetic cylindrical pores, like the tubes we have been working with, may be broken down into the following steps. (1) The presence of a surface establishes a vacancy concentration gradient, as Dr. Kuczynski and the present authors have shown. (2) Vacancies then diffuse in order to decrease this gradient. (3) In so doing, they cause an elastic field to be built up in such a direction as to oppose the diffusion, so that it slows down until it reaches a steady-state rate. (4) The steady-state flow is due to the fact that the surface tension acting on curved surfaces applies a stress to the system which itself mechanically causes vacancy diffusion. What I have said is that the diffusive types of flow of the authors and of Dr. Kuczynski are transient phenomena which soon stop if the pores are large. We are then left with a mechanical problem only, and our researches have not yet disclosed just what the mechanism of flow is; it may be a slip process, or it may take place by a vacancy diffusion of the type Nabarro has de-

Jan 1, 1951

By John G. Simmons

A tutorial account of the tunnel effect between metal electrodes separated by a thin insulating film is presented. Energy diagrams of metal-insulator -metal sandwiches are briefly discussed, and the isothermal and thermal J-V characteristics for symmetric and asymmetric tunnel junctions are derived from a generalized expression. Experimental data from tunnel junctions are presented, and it is shown how, correlation between theory and experiment permits a determination of some of the oxide properties. Finally, the barrier heights in AT-Al2O3-A1 junctions determined by various inr3estigators are discussed. The object of this paper is to give a tutorial account of the electric-tunnel effect between metal electrodes separated by a thin insulating film and to illustrate its utility in the study of metal-oxide interfaces. This paper will emphasize the theoretical aspects of the electric-tunnel effect: however, a brief account of the more recent experimental work will be presented. A more comprehensive experimental study of the tunnel effect will be presented in the following paper by Drs. Pollack and Morris, see p. 497. We will open with a discussion of the energy diagrams for metal-insulator-metal junctions and what we mean by the electric-tunnel effect. Following this is an account of the theoretical studies of Sommerfeld and Bethe, and Holm, and the recently reported calculations on symmetric and asymmetric junctions using the generalized formula. The theoretical presentation is concluded with calculations on the temperature dependence of the tunnel J-V characteristic. In the experimental section, it is shown how barrier heights and thicknesses are determined by comparing the experimental data with the theory. Finally, the results of various investigators on the determination of the barrier heights in Al-Al2O,-A1 systems are discussed. I) POTENTIAL BARRIERS BETWEEN CLOSELY SPACED ELECTRODES Since it is necessary to supply energy to a metal in order to observe electron emission from its surface, it is evident that the potential energy of an electron within a metal is lower than that of an electron at rest outside the metal. It follows, then, that the electron must undergo a change in potential as it crosses the metal-vacuum interface. In the Sommerfeld1 free-electron theory of metals it is assumed that the free electrons, the electrons responsible for the electrical characteristics of metals, move in a constant potential environment in the metal, and are contained within the metal by a potential barrier at the metal-vacuum interface. It will be evident that the idea of a constant potential environment is a very idealized model, since the potential energy of an electron is influenced by all of the atoms and the other electrons within the metal; however, its use is justified by the many excellent results it gives.2 Fig. l (a) is the energy diagram of the model envisaged by Sommerfeld; the metal is represented by a potential-energy well of depth V,; the vacuum level represents the energy of an electron at rest outside the metal. At 0°K the electrons, which exist in pairs in discrete energy levels, fill up the well to an energy ?, the Fermi level. The distance, Vo - ?, from the Fermi level to the vacuum level is known as the work function of the metal, and represents the minimum energy required to free electrons from the interior of the metal at 0°K.

Jan 1, 1965

By J. E. Coyne, L. E. Tanner, J. E. Boyd

In recent investigations of the isothermal transformations of the retained 0 phase in a titanium alloy having 13 pct V, 11 pct Cr, and 3 to 4 pct Al, there have been indications that the w transition phase is the initial decomposition product at temperatures below about 950°C (510 c).' This metastable phase, found in a number of titanium-base alloys rich in stabilizing elements and known to have an adverse effect on their ductility and toughness, is not easily observed by the usual experimental techniques. Since wis coherent with the /3 matrix,495 distinct interphase boundaries are not revealed metallographically. However, in some cases, an etching effect (i-e., a "mottled" /3 surface) poorly resolved by light microscopy has been found to correspond with the presence of w.6 X-ray diffraction examination has generally been the most suc- cessful means for identifying this phase and crys-tallographic determinations have been made.4'5 Dynamic elastic modulus7 and electrical resistivity measurementse have also aided in the identification of w. In the current study, the electron microscope was utilized to examine microstructures in aged Ti-13V-11Cr-3A1 that were unresolvable by light microscopy techniques. Using negative parlodion replicas (shadowed with chromium at 27 1/2 deg) examination was made of various transformation

Jan 1, 1962

By C. Hirayama

THIS investigation was undertaken to determine some of the factors contributing to the embrittle-ment of glass-coated silver wire on heat treatment. The materials used in this investigation were 22 and 23 AWG silver wires obtained from Handy and Harman as "fine quality, soft temper." A wet analysis of one sample showed 99.8 pct Ag. The 23-gage wire was covered with a double layer (2.5 mils thick) of electrical grade fiber glass (aluminum borosilicate glass containing a relatively high concentration of calcium, and a very small amount of

Jan 1, 1960

By A. S. Appleton, M. B. Bever

The stored and expended energies of an Au-Ag alloy deformed in torsion were investigated as functions of strain, strain rate, and temperature. At room temperature, the stored energy, measured by tin solution calorimetry, increased linearly up to an intermediate strain, above which it was nearly constant; at 78 °K the relation was linear up to fracture. Approximate values of the energy stored at 4 OK indicate a rapid increase below 78 OK. The energy stored by specimens deformed successively at 4o and 78oK changed slowly with increasing second strain from the curve for energy stored at 4 OK (measured after annealing at 78oK) to that for deformation at 78°K. The ratios of the stored to the expended energy at room temperature and 78 oK were unusually large. The additional amount of energy stored at the lower temperature is attributed mainly to an increase in the number and energy of dislocations. The stored energy as a function of strain rate at 4oK (measured after annealing at 78oK) increased to a sharp maximum over a narrow range of strain rates, whereas the expended energy changed only slightly. These effects are explained by the generation of point defects resulting from the enforced nonconservative movement of jogs in fast moving screw dislocations. The effect of straining successively at two strain rates at 4oK was also investigated. Stress-strain curves for 4o, 78oK, and room temperature and the results of micro examination of specimens deformed at these temperatures are also reported. THIS investigation is a continuation of a research program primarily concerned with the effects of variables on the energy relations involved in the deformation of metals. The variables investigated in earlier work have included strain, strain rate, temperature of deformation (room temperature and 78oK), and the type of deformation process. Most of this work has been carried out with Au-Ag alloys. In the investigation reported here, the deformation process was torsion, which had not been used previously in the program. The stored and expended energies involved in the deformation process were measured as functions of strain, strain rate and the temperature of deformation in the range 4" to 373" K. The microhardness and microstructure of deformed specimens were also investigated. The relation between stored energy and strain differs for different deformation processes.1,2 In general, the stored energy increases with strain and may attain a nearly constant value at large strains. This value and the rate at which it is reached, however, depend on the process of deformation. This dependence has been attributed, at least in part, to differences in the effective strain rate and rise in temperature of the specimen during deformation. The effects of the temperature of deformation on both the stored energy and the energy expended in the working process have been the subject of only a few investigations.3-6 Values of these quantities at low temperatures are of special interest as they yield information on point defects which, for many metals, is not obtainable at higher temperatures. The effect of the strain rate on the stored and expended energies has received little attention The strain rate has usually been measured only in relative terms, and hence the interpretation of its effect on the stored energy has been mainly comparative.

Jan 1, 1963

By B. A. Joyce, R. W. Bicknell, R. J. Bennett, P. J. Etter

Epitaxial layers of silicon have been grown on single-crystal quartz and alumina by two types of chemical reaction the hydrogen reduction of trichlorosilane and b) the pyrolysis of silane. These reaction systems are described. The cqsfallography of the two systems, as determined by electron and X-ray diffraction, is discussed, and epitaxial mechanisms suggested. The electrical behavior is briefly considered, and the design and properties of field-evert devices fabricated in the epilaxial material are also discussed. RECENT work carried out in these laboratories" has shown that it is possible to obtain epitaxial deposits of silicon on single-crystal quartz and alumina. The purpose of this paper is to describe the method of preparation of these layers using both the hydrogen reduction of trichlorosilane and the pyroly-sis of silane, to compare the crystallographic behavior of films on the two substrates, and to present some initial results of the electrical behavior of silicon layers on alumina. The structural study has been carried out using electron and X-ray diffraction and electron microscopy. The electrical assessment has been made in terms of conductivity and mobility determinations by the van der pauw3 technique, and by measurements on metal-oxide-semiconductor assemblies. 1) FILM PREPARATION Both quartz and alumina substrates, approximately 1 sq cm in area and 0.4 mm thick. were prepared from slices cut parallel to within a few degrees of (11.0), (10.0), and (00.1), and polished mechanically down to about 0.3 p diamond grade. Chemical polishing of alumina slices, using phosphoric acid at elevated temperatures, has also been used; epitaxial silicon layers have been grown on these substrates, but the process is not entirely reliable, possibly because the large strain present in the Verneuil-grown alumina crystal affects the chemically prepared surface. The two reaction systems are fairly conventional and are described briefly below. a) Trichlorosilane-Hydrogen System. The apparatus is shown schematically in Fig. 1. The substrate was heated by induction, with a fused-quartz encapsulated graphite hemicylinder as susceptor. The procedure for carrying out a deposition was to heat the substrate to reaction temperature in a stream of

Jan 1, 1965

By R. T. DeHoff

A new approach to the measurement of the size distribution of particulate structures imbedded in opaque bodies is developed. The method is based upon the assumption of a general two-parameter distribution curve, and the estimation of these two parameters from three well-known counting measurements made upon metallogvaphic sections. The total number of particles in the system is also a product of the analysis. Application of the method with the assumption of the logarithmiconorma1 disfribution function is developed in detail. A complete description of the geometry of a particulate structure requires the determination of the size distribution of particles. The problem of estimating the size distribution of particles of a particular phase in a multiphase system, or the grain-size distribution in a homogeneous system, has been approached by many investigators.'-'0 All of the solutions proposed by these investigators share the following characteristics. a) A minimum of several hundred particle sections in the microstructure must be examined, and some characteristic dimension recorded, such as area,3'5 intercept length,6"9 or diameter.1'2'4'10 b) The resulting data must undergo a series of fairly complex calculations to generate the empirical histogram which describes the size distribution in three dimensions. c) In order to manipulate the result, the histogram must be fitted to some assumed mathematical formula, using the attendant statistical procedures. d) With two notable exceptions,8,10 the particles must be spherical. The measurements and calculations involved are in general very tedious and time-consuming, with the logical result that much effort has been expended in the development of equipment which will perform these tasks automatically and semiautomati-cally.9,11-15 considerable success has been achieved in this area, but the equipment is in general expensive, and usually places stringent requirements On the quality of microstructure which can be reliably analyzed. In laboratories which do not possess such equip- ment, the analysis of geometry is usually limited to measuring total properties of the microstructures, such as volume fraction and interfacial area, because the measurements required are much simpler to make. The determination of size distributions rarely is undertaken. The development which is presented in this paper permits the estimation of size distributions of three-dimensional structures from the same simple counting measurements which are required to determine the surface area and volume fraction of the particulate structure, and one other counting measurement. The basic approach that is used is to assume a general mathematical form for the distribution, which is described by two parameters (as the normal distribution is described by assigning values to the mean and the standard deviation), and then estimate these two parameters from the results of the counting measurements. The introduction of a third counting measurement also permits the evaluation of the number of particles per unit volume, Nv.

Jan 1, 1965

By Y. C. Liu, R. H. Richman

Deformation and recrystallization textures of the a solid solutions of Cu-P, Cu-As, and Cu-Sb alloys are examined as a function of composition. It is found that the deformation texture of copper is unckanged up to a composition of 0.43, 0.41, and 0.24 at. pct of phosphorus, arsenic, and antimony, respectzvely, and then the transition to the 70 : 30 brass type of deformation texture proceeds linearly as the logarithm of the solute content. Recrystallization textures consist of strong, new components that do not all bear the usual<111> rotational relation to the deformation components. The dependence of the deformation texture transitions and the recrystallization orientations upon composition and solute-element type is discussed. A systematic evolution of individual recrystallization components in the annealing texture of copper was portrayed as a function of germanium or tin content in a previous study.&apos; Similar patterns of tex-tural change have been observed with copper-zinc alloys,2 and in less detailed studies of other systems.3,4 In fact, most of the literature dealing with preferred orientations in copper alloys indicates that the influence of solute elements upon textures in copper is of one general type. As the alloy content of copper is increased, the deformation texture changes from the copper type to the 70:30 brass type, and the annealing texture undergoes a comparable change from the (001) [ loo] orientation of unalloyed copper to the {113} <211> type of 70:30 brass, with the notable exception of copper alloyed with Periodic Subgroup V-B elements.4-8 In almost every case it is possible to describe the reorientations after annealing as 30 to 40 deg rotations about <1ll> poles of the deformation components. Since there is apparently not much difference in the final textures of the copper binary systems already investigated, only a limited variety of preferred orientations has served as the basis for current theories of the origin of annealing textures. In order to test the validity of the proposed theories, and the possibility that the <111> rotations are not unique, it would be most desirable to study the reorientation relationships of additional, and heretofore unanalyzed, recrystallization components if such components exist. Examination of the literature reveals a good deal of uncertainty concerning preferred orientations in copper alloyed with elements in Subgroup V-B of the Periodic Table. Phosphorus is known to inhibit the cube texture in copper when present in amounts as low as 0.01 wt pct,8 but the recrystallized matrix that occurs instead of the cube orientation was reported as either random or weakly textured, with orientations not observed in Cu-Zn, Cu-Ge, or Cu-Sn alloys.4-9 A Cu-0.75 at. pct As alloy apparently contained, along with the (001)[100] and (113) <211> type of recrystallization components, {110} <112> and (110) <001> components, whereas an almost random distribution of orientations was found in alloys of higher arsenic content.4 A {110} <001> component also appeared in rolledand annealed Cu-Sb alloys, but otherwise the recrystallization textures of Cu-Sb alloys seem to con-

Jan 1, 1962

By Y. C. Liu, R. H. Richman

The effects of gel,manium and tin on the deformation and 9-e-crystallization textures of copper have been explored in detail with in the copper-rich terminal solid solutions. Addition of solute to copper produces a change in deformation texture from the copper type to the 70 :30 brass type. This transition has been measured quantitatively The types mid proportions of recrystal-lization components have been determined as a function of com-position, and correlation of recrystallization components with specific deformation components is discussed. On the basis of information available in the literature, it has been suggested that the effect of solute elements on the recrystallization texture of copper

Jan 1, 1961

By Linda Lee, R. E. Reed-Hill, C. R. Smeal

Four fcc metal surfaces, etched to make them responsive to polarized light, have been studied with an electron microscope. Jones&apos;prediction that these surfaces are grooved has been verified. Optical-goniometer measurements made on commercially pure nickel indicate that the groove walls are poorly defined (100) facets. Surface mientations close to either 4001 or {ill).- show no extinctions on the polam&apos;zid-light microscope. An explanation is offered for this orientation dependence. It has also been deduced that polarized-light extinctions on a grooved cubic-metal surface should not be used directly in crystallographic-orientation determinations. The nature of the etching solutions that produce these surfaces is considered. A cubic-metal surface may be made responsive to polarized light in two basic ways."&apos; In one, an anodic film, believed anisotropic, is formed on the surface. A plane-polarized-light beam, at normal incidence, should be reflected from this type of film with an ellipticity that varies with grain orientation. Under crossed polarizers of a polarized-light microscope, each grain may be distinguished from its neighbors by a difference in reflected-light intensity. An alternate treatment,&apos; more generally applicable to cubic metals and of principal concern here, involves etching of grain surfaces. The characteristic features of this surface were first deduced by Jones3 from light-microscope observations. She concluded that, in general, grain surfaces were furrowed so that light was reflected from two parallel sets of etched facets. A grooved surface produces elliptical polarization of a plane-polarized beam because the beam does not strike groove walls at normal incidence. Jones also observed that the furrows must have faces inclined to each other by approximately 90 deg, since the light returns along the incident path, and that a grooved surface should show four maxima and minima of reflected-light intensity during a 360-deg stage rotation of the polarizing microscope. Positions of maximum extinction were predicted to occur when groove axes were parallel to either polarizer or analyzer vibration directions. Because of the limited resolving power of Jones&apos; optical microscope, her deductions were primarily indirect. Proof of the correctness of her conclu-sions, as demonstrated by the electron microscope, will be given as well as evidence concerning the crystallographic nature of the etch furrows and the types of etching solutions that produce them. EXPERIMENTAL PROCEDURE During a comprehensive study of hot plastic deformation in nickel and nickel alloys, an etch was evolved that produced a surface with an excellent polarized-light response on Nickel 200 ("A" Nickel). The etching solution and associated metal-lographic procedures are given in Table I. In evaluating this etch, a study was made of the topological features of the etched surface and their relation to the underlying crystalline structure. As part of this investigation, large crystals (2 mm average diameter) were grown in a 1-cm-sq Nickel 200 specimen. After the surface was etched, the crystallographic orientations of ten grains were determined by the standard back-reflection Laue technique of Gren-inger.4 Maximum extinction positions during a microscope stage rotation were also measured for the ten grains. Groove-wall positions on the surfaces of the ten grains were measured with a two-circle optical goniometer. The technique was essentially that of Barrett and Levenson.5 Several grain surfaces were photographed with a Philips Model 100A electron microscope. All specimens were replicated with collodion, or collodion backed with formvar and chromium shadowed at 18 to 20 deg. The basic material was Nickel 200. However, an electron-micrographic study was also made of surfaces developed by polarized-light etches on other fcc metals (90-10 a brass, Monel 400, and lead). All etching procedures are given in Tables I and 11. EXPERIMENTAL RESULTS Fig. 1 shows typical electron micrographs of three different fcc metal specimens and an optical micrograph of a fourth. All photographs show a grooved structure corresponding closely to Jones&apos;3 predictions. Also, as noted by Jones, extinctions were always observed when groove axes were either parallel or perpendicular to the microscopes&apos; vertical cross hair. The symmetry of the grooves, with respect to the twin boundaries in Fig. 1, implies that furrows have a crystallographic basis. The coarse-grained Nickel 200 specimen was used to study this basis. Facet Orientation. The poles of the ten Nickel 200

Jan 1, 1964

By T. E. Leontis, R. S. Busk

WORK was initiated several years ago at The Dow Chemical Company to determine the applicability of powder metallurgy processes to magnesium and its alloys. Although it was found possible to apply the normal techniques of compacting and sintering, it was discovered early that the technique of extrusion offered special promise. This technique and the alloys produced were therefore studied extensively, and the present paper is a partial report of those studies. It is expected that a second paper will discuss more extensively some of the alloy systems simply mentioned here. Although the literature on powder metallurgy is very extensive, there are only two articles dealing with magnesium13,14 and there is no reference to the direct extrusion of the powder of any metal without the preliminary step of compacting. Furthermore, no mention has been made of the possibility of designing alloys by the technique of powder extrusion, particularly alloys that cannot be made in any other way. Sauerwaldl3 discusses the extrusion of magnesium and aluminum powder to produce alloys containing nonequilibrium phases, but again he emphasizes the importance of pre-compacting before extrusion. There seems to be general agreement that some form of hot-pressing is essential to obtain in a compact the full theoretical density of the metal used.&apos;-" The nearest approach to the technique of direct powder extrusion is that of extrusion of pre-sintered compacts,&apos;-" for which the best review through 1944 is that by Kieffer and Hotop.11 Ensign* discussed a specialized technique of "extruding" already compacted and sintered gears for the purpose of sizing. Sauerwa1d13 discussed the compacting and extrusion of magnesium and magnesium-aluminum alloy band-saw chips. The properties of the extruded compacts were found to be comparable with those of extrusions prepared from solid billets. He also extruded a compacted mixture of magnesium and aluminum powder and found the properties to be somewhat different from those obtained on an extrusion made from a cast billet of the same composition. The second reference dealing specifically with the powder metallurgy of magnesium is Stern&apos;s patent," in which is described the extrusion of scrap chips and turnings following a hot-pressing and sintering step. Considerable emphasis in most of these articles is placed on the necessity of atmospheric protection to minimize the danger of oxidation. Experimental Procedure The first experiments were made with the small extrusion press shown in fig. 1. With this press, billets of about 1/2 in. diam are used to produce wire

Jan 1, 1951

By A. G. Metcalfe, A. Siede

The hardening of annealed copper during fatigue testing appears to be independent of the applied stress and to occur largely within the first 4000 cycles. Copper hardened by fatigue is more resistant to annealing than copper hardened by static means, and softens by a multistage process. Creep studies at 300°C show that copper hardened by fatigue also has greater resistance to creep than statically hardened copper. Some theoretical explanations are discussed. RECENT interest in the behavior of metals subjected to fatigue stems from observations of phenomena which could not be explained by current theories of the structure of metals. In 1953 Bullen et al.l showed that the hardening caused by fatigue at low stresses did not cause severe bending of the lattice as did hardening by static methods. This was later confirmed by Wood, &apos;and Broom and Ham.3 Clarebrough et al4 showed that as the stress level used in fatigue hardening was increased, bending of the lattice was introduced, eventually approaching that of metals deformed by conventional cold-working methods. The type of hardening introduced by fatigue has other interesting characteristics. Kemsley5 showed that copper fractured by fatigue at low stresses resisted annealing as compared to copper hardened an equal amount by static means. The work of Clarebrough etal, shows that the manner of energy release during annealing differs for fatigue-hardened and statically hardened metal in both magnitude and temperature range of release. In addition to hardening of the lattice, fatigue has been shown to cause softening in materials previously hardened by static methods,5 and also in age-hardened alloys.7 The superposition of a small fatigue stress on the static loading of high-temperature alloys has been shown to increase their resistance to creep.&apos; The present work was an attempt to determine the relationship between fatigue hardening, resistance to annealing, and the creep properties of pure copper. EXPERIMENTAL PROGRAM The program described here was designed to answer three questions: 1) Can stress-cycle combinations well below the failure values give significant increases in hardness? 2) Does hardening, under different stress-cycle combinations, give increase resistance to annealing? And if so— 3) Does superior resistance to annealing confer superior creep properties over statically hardened material? Three-quarter hard OFHC copper was annealed in vacuo at 550°C for 1 hr to produce a strain-free material having a fine grain size and uniform hardness. (The standard deviation of the mean hardness of the annealed samples was less than + 1.9 Dpn). Annealing was performed after machining the fatigue specimens so as to avoid work-hardened surfaces. Jigs were used to avoid distortion. The fatigue properties of the annealed copper were determined using an accurately aligned Baldwin-Sonntag SF1-U machine operating at 1800 cycles per min. This limited the minimum number of cycles which could be examined to about 4000 stress reversals. Since failure resulted in specimen damage, hardness increase with number of cycles was measured by testing specimens fatigued to different fractions of the failure values. The results are shown in Fig. 1, and are superimposed on the S-N curve of the copper in Fig. 2. The resistance to annealing of fatigue-hardened material was compared to that of material hardened by uniaxial tensile straining. Specimens hardened to equal values by both methods were progressively annealed for 1 hr at temperature in vacuo. After each annealing treatment the hardness was measured using a Leitz Durimet hardness tester with a 500-g load. At least five tests were made for each hardness determination. The standard deviation for individual specimens varied from 0.26 to 1.48 hardness points. The results are shown in Figs. 3 to 5. The discontinuous nature of the annealing process has been brought out by drawing idealized curves through the data points. This can be justified by the consistency of the results as well as by general agreement with results of Kemsley.5 The creep resistance of fatigue-hardened and

Jan 1, 1960

By W. Vickers

Single-point rotating cantilever fatigue tests have been carried out at 550" to 650°C on beryllium produced by a variety of fabrication routes. All the specimens gave similar plots of stress against number of cycles to failure but failure occurred at a lower stress range in the hot-pressed material than in the extruded material. A discontinuity or "kink" appears to exist in the curves and this is considered. A true endurance limit was not evident up to 10 cycles. When Brush hot-pressed warm-extruded beryllium was tested at 600oC a relationship where Nf = number of cycles to failure, ?ep = plastic strain range, and c = constant, was shown to be applicable for values of ?ep > 6.3 x 10-4. Further for this moterial and temperature and for frequencies between 150 and 4500 cycles per min, the variation of number of cycles to failure with frequency obeyed the expression N = Af0.44 where N =number of cycles to failure, f = frequency, and A = constant. Examination of the fracture surfaces indicated that they were generally intergranular and in all cases the higher the stress the more irregular was the fracture. Microexamination has shown that grain boundary cavities and fine intergranular cracks were associated with the fractures. REVERSING or cyclic stresses set up in fuel elements can lead to failure by fatigue. Fatigue stresses may be developed by instability of the fuel element in the coolant flow leading to rattling against the channel wall or fin vibration. Thermal cycling also sets up reversing stresses due to differential expansion of fuel and can. Fatigue resistance is of some importance therefore in the evaluation of canning materials. At the time the work described in this paper was initiated, beryllium was being considered for the can and endcaps of an advanced gas-cooled reactor fuel element. With this in mind, beryllium obtained from different manufacturing routes was tested on single-point rotating cantilever fatigue machines at temperatures the same as those envisaged in the reactor to obtain appropriate fatigue data. The results obtained indicated a discontinuity in the S-N curves coinciding with the yield stress in the specimens. Although in some cases the data is insufficient to warrant a detailed quantitative analysis, a qualitative analysis of this effect was thought to be justified. 1) EXPERIMENTAL DETAILS The materials used are given in Table I together with the production route and available analyses. The specimen shape is shown in Fig. 1. The fatigue specimens were machined slightly oversize, after which they were given a stress-relieving anneal of 1 hr at &apos;750° C in vacuum. This was followed by an etch in sulfuric-phosphoric-chromic acid mixture down to the finished dimensions given in Fig. 1. Etching served to remove the interference colored oxide film formed on the surface during annealing and very small recrystallized surface grains too. The specimen was mounted in the grips of a cantilever machine and rotated inside a furnace controlled at the testing temperature. The specimen was loaded in increments as the speed was increased until the required load was applied at 5000 cycles per min. A revolution counter was used to record the total number of cycles.

Jan 1, 1964

By Harry M. Bendler, W. A. Wood

Electron metallography has been applied to the study of the torsional fatigue of polycrystalline OFHC* copper. The development of slip bands, fis- sures, and microcracks has been followed as a function of strain amplitude and number of cycles. Those factors that have been found to be relevant to a geometrical theory of the fatigue process are described. From the point of view of this theory it is important to distinguish between the effects of cycling at large amplitudes and at small amplitudes (true fatigue deformation). EARLY investigations showed metal fatigue cracking is caused by continued reversal of a small plastic strain. Fundamental study of fatigue, therefore, has been largely concerned with the cumulative effects of these plastic reversals on the metal structure. In particular, recent work has shown that the plastic reversals produce slip movements that are finer than the slip movements produced by unidirectional increments of plastic strain; individual slip movements in fatigue, for instance, are too fine to see by optical microscopy.2 It has also been shown that the fine slip movements of fatigue tend to concentrate in bands and that in this way they can build up, after enough cycles, a progressive surface disturbance. It is this disturbance which, in effect, makes the slip movement visible. It is also in this disturbance that the first cracks appear to form. From this point of view, therefore, the problem of fatigue becomes one of finding why the fine slip movements should concentrate in bands and why the. bands should finally open into cracks. Thus far, study of the slip bands has been limited to the changes they produce at the surface only. These studies have given interesting results. Ewing and Humfrey3 many years ago showed that the bands in iron could extrude debris. Forsyth found that bands in aluminum alloys could extrude thin tongues of material. Thompson showed that, after lightly dissolving away the surface of copper specimens, some bands persisted as superficial cracks might, and that the life of a specimen under fatigue deformation might be prolonged by dissolving away the surface at intervals. Thus these and related results have suggested that the metal structure in fatigue bands may disintegrate and that this may happen long before obvious cracking. These surface studies, however, have not shown definitely how cracking may begin, for they do not show sufficiently what is happening in depth. However, some advance in this direction has been made recently by a new approach, summarized later, which reveals slip bands and their surface disturbances in section. In the work done thus far, the sections have been studied by optical microscopy only. In the present work they have also been studied by electron microscopy with extended results. This paper also describes some observations on the surface disturbances prior to sectioning. These observations are relevant to a suggested theory of fatigue cracking. REVERSED SLIP MOVEMENTS Slip movements in fatigue deformation must reverse direction in each cycle. The question arises: DO the "backward" and "forward" movements take

Jan 1, 1962

By D. P. Kedzie, R. A. Dodd

The fatigue strength, fatigue hardening, and effect of fatigue deformation on subsequent age hardening of supersaturated Al(Cu) solid solutions have been determined as functions of alloy composition and temperature. The fatigue strength/tensile strength ratios, determined at 150°, 25°, and -195°C, decreased with increase in alloy content for all temperatures, but the F.S./U.T.S. ratios at -195°C decreased much more rapidly than did the ratios for 150" and 25°C. This suggested that strain aging and/or age hardening occurred during tests at higher temperatures. Additionally, the F.S./U.T.S. ratios at 150°C exceeded those at 25°C for all compositions, indicating greater strengthening during fatigue at 150°C. The effect of fatigue or tensile deformation at 150°, 25°, and -195°C on subsequent age hardening showed that the deformatzon increased the rate of precipitation and indicated that mechanically produced vacancies were probably formed during deformation. Fatigue hardening was studied at 150°, 25°, and -195°C, and the effect of room-temperature rests after 10 and 100 cycles was examined. The results confirmed that strain aging occurred at the higher temperatures . DURING the last 15 years various mechanisms of fatigue crack nucleation and growth based on dislocation and vacancy interactions, operating singly or collectively, have been proposed. The probable consensus of present opinion is that the fatigue process in pure metals essentially involves dislocation interactions, and that vacancies formed by such interactions play a minor or inconsequential role. However, there is some evidence that age-hardened alloys tend to overage during fatigue, probably by local vacancy-enhanced diffusion, and strain aging also might be important in selected cases. Furthermore, it has been established that the behavior of quenched-in vacancies in Al(Cu) and other solid solutions is composition-sensitive. Therefore, it seemed worthwhile to investigate various aspects of the fatigue behavior of supersaturated Al(Cu) alloys and to examine the results in terms of vacancy-enhanced effects. EXPERIMENTAL The alloys used in this investigation were prepared from 99.994 wt pet A1 and OFHC copper, the latter containing 0.04 pet 0 as the principal impurity. Six alloys were made, containing, by actual analysis, 0.58, 0.96, 1.96, 2.85, 4.45, and 5.51 wt pet Cu. They were induction-melted in air in graphite crucibles, cast as 7-in. by 7/8-in.-diam rods in graphite molds, and hot-rolled to 5/8 in. diam. The experimental work was a three-part program involving the determination of a) 10&apos; cycle fatigue stresses as a function of alloy composition and temperature; b) the effect of fatigue deformation on subsequent aging of the supersaturated alloys; and c) fatigue hardening as a function of alloy composition and temperature. For determining the 105 cycle fatigue stresses, a portion of the 5/8-in. stock was machined into Krouse rotating cantilever beam fatigue specimens, 2 in. in length by 1/4 in. minimum diameter. These were tested at +150°, 25°, and -195°C (liquid nitrogen) on a Krouse high-speed machine, with special weights to provide lower -than-normal loading ranges, this being necessitated by the small load requirements of those alloys of lower copper content. For the high-temperature tests a small resistance heater was designed to clear the collets and fit in between the chucks, while for the low-temperature tests a hollow nylon cylinder was used, having closed ends drilled to pass a fatigue specimen, and positioned similarly to the heater. A plexiglass container completely enclosed the fatigue machine; rubber gloves fixed to ports in the walls enabled the machine to be operated from outside the box. A tray of magnesium perchlorate dried the air in the container and prevented both atmospheric corrosion fatigue at room and elevated temperatures and troublesome ice build-up at low temperatures. A total of ten to fifteen specimens was used for each combination of alloy composition and temperature. The result of each test was plotted on a standard S-N diagram, and the next stress was selected on the basis of the trends indicated by previous specimens. In this manner a small portion of the S-N diagram was constructed, and the 105 cycle fatigue stress obtained. Small Krouse fatigue specimens were also used to study the effect of cyclic prestrain on subsequent

Jan 1, 1964

By William R. Patterson

ACCORDING to the phase diagram for the Mn-Ni system,&apos; the high-temperature, fcc y-manganese solid solution is stabilized to below room temperature, in the range approximately 22 to 28 wt pct Ni. However, pearson2 cites evidence in the literature showing that for manganese contents greater than about 75 wt pct the quenched phase in Mn-Ni alloys is fct. This is in keeping with zener&apos;s3 suggestion that in manganese-rich alloys the fcc high-temperature phase undergoes a diffusionless or martensitic transformation to a fct structure upon quenching. Work by Basinski and christian4 on manganese-rich compositions in the similar alloy system Mn-Cu established that in alloys of greater than about 82 wt pct Mn the y-manganese phase underwent a fee fct martensitic transformation upon quenching to room temperature. Striking evidence for this martensitic transformation was the highly twinned or banded microstructure, an example of which is shown in Fig. 1. Similar microstructures have also been found in quenched Mn-Cr5 and Mn-Au6 alloys. In the present investigation, we attempted to obtain similar evidence of a fcc — fct martensitic transformation in manganese-rich Mn-Ni alloys. Manganese alloys in the range 10 to 30 wt pct Ni were prepared by arc-melting, under argon, at atmospheric pressure. The alloys all were 50 g in size, and negligible weight loss occurred during melting. The starting materials were electrolytic manganese and nickel, both of 99.9+ pct purity. The alloys were subsequently homogenized 48 hr at 1750°F, hot-rolled 50 pct at 1700°F, annealed 2 hr at 1700°F, and water-quenched to room temperature. All heat treatment was done in purified argon. Both X-ray and metallographic techniques were used to examine the structures of the quenched alloys. X-ray diffractometer studies were conducted on powder specimens obtained from filings rean-nealed 2 hr at 1700°F in argon and water-quenched. Identification of the intensity peaks from the diffractometer counter traces showed that below about 10 to 12 pct Ni the y phase decomposed partially upon quenching to give an a- + ?-manganese structure. In the range approximately 12 to 25 pct Ni, the structure was found to be fct ? manganese with a c/a ratio approaching 1.0 at 25 pct Ni. Beyond approximately 25 pct Ni, the structure remained fcc. These results, including lattice-parameter determinations, are given in Table I. Metallographic examination. was carried out on specimens polished and etched after quenching, and also on polished specimens reannealed and quenched. In the first case, the distinctive twinning or banding so prevalent in Mn-Cu alloys was sought, while in the second, the surface relief effect generally regarded as being truly characteristic of a martensitic transformation was sought. The microstructure of representative alloys in the range 12 to 25 pct Ni is shown in Fig. 2. The structure of the Mn-15 pct Ni

Jan 1, 1965

By J. H. Brophy, J. Wulff, H. W. Hayden

The mickel-activated sintering of tungsten occurs in two stages of densification. A kinetic analysis of the second stage has been based on an appropriate model of the process. As in the first stage, the rate-controlling mechanism for the second stage is also the solution of tungsten in the nickel carrier phase. The transition from the relatively rapid first-stage results from the spherozdization of pores and the initiation of grain growth. Finer grain size due to higher-nickel content leads to more rapid and higher degrees of densification 1 HE sintering of tungsten is accelerated by small additions of nickel, cobalt, and iron.&apos; Experimental evidence in detail has shown that nickel-coated, hydrogen-reduced tungsten powder may be sintered to more than 90 pet of theoretical density at l100o C.2, 3 Simultaneously, a kinetic analysis of the process showed that it took place in two distinct stages. In the first stage, nickel evidently served as a carrier phase through which tungsten atoms could move. This stage was kinetically similar to the "solution-precipitation" step postulated when the carrier phase is a liquid,4 although in the Ni-W case, there is no evidence of the existence of a liquid phase either in equilibrium phase relationships or in microstruc-tures at the temperatures under consideration. It has been shown that this stage corresponds to a process which is rate controlled by a phase boundary reaction.3 More recent evidence has been presented5 showing that after the initial rapid stage of sintering, densification continues in the final stage but at a slower rate, although no mechanism for this stage of densification was defined. An analysis of later stages of densification in solid state sintering has been presented,6 but it is not directly applicable in the present case of activated or carrier phase sintering. It is the purpose of this investigation to study the final stage of nickel activated tungsten sintering in greater detail The effect of varying nickel content has been examined, and a possible kinetic mechanism defined. EXPERIMENTAL PROCEDURE AND RESULTS Commercial hydrogen-reduced tungsten powder from Wah-Chmg Corp., having a nominal particle size of1 µ and a B.E.T. specific surface area of 5500 sq. cm per g was mixed with aqueous solutions of hydrated nickel nitrate in proportions sufficient to produce nominal compositions of 0.25, 1, 3, and 5 wt pet Ni. This solution was then evaporated to dry-ness at approximately 150" C for 15 hr. Following evaporation, the resulting powder cake was broken with a mortar and pestle and reduced in hydrogen at 800oC for 1 hr. Following reduction, the powder

Jan 1, 1962

By G. Krauss, W. Pitsch

The fine structure of the bcc martensite formed in an Fe-33 wt pct ATi single crystal of arrstenite is sho~on by transmission electron microscoPy to consist of combinations of transformation twins, stacking faults, deformation twins, and regular arrays of parallel screw dislocations. These structures constitute evidence for the multiple lattice-invariant deformations which operating during the formation of martensite could produce the real habit-plane scatter measured by a two-surface analysis of the plates formed in the single crystal of this investigation and reported in the literature for other Fe-Ni rnartensites. CRYSTALLOGRAPHIC theories1,2 of martensitic transformation show that the habit plane of martensite in a parent lattice is dependent in part upon an inhomogeneous distortion or lattice-invariant deformation which takes place on a fine scale within a martensite plate during its formation. Several recent theoretical papers3,4 have addressed themselves to an analysis of a wide variety of conceivable lat-tice-invarient deformations and the habit planes which they produce, while experimental investigation have been concerned with either the measurement of habit planes or the description and identification of the martensitic fine structure which reflects the nature of the lattice-invariant deformation operating during transformation. In Fe-Ni alloys with subzero Ms temperatures, the group of alloys with which this paper concerns itself, habit planes have often been found to scatter an amount greater than might be expected from possible experimental errors,5-7 and fine twinning has been identified as a major constituent of the fine structure of martensite.8-11 It has been suggested3,4 that more than one type of invariant shear occurs during martensitic transformation. This possibility has been experimentally supported12,13 by the observation of both dislocation configurations and twinning in a single martensite plate. The purpose of this paper is to report additional evidence for multiple lattice-invariant deformations in martensite and so to account for the real scatter in the habit planes of the martensite plates formed in Fe-Ni alloys. EXPERIMENTAL PROCEDURE The Fe-Ni single crystal was produced by pulling a high-purity iron and nickel charge through a single-crystal vacuum furnace in an alumina crucible. The crystal was double-melted to promote homogeneity and to increase its size by further additions on the second pass. In its final form the crystal was 4 cm in diam and 5 cm long. The nickel and carbon contents were analyzed at 32.9 and 0.006 wt pct, respectively. The austenite of this alloy first transformed to martensite by bursts at about -120°C, and, to preserve as much of the austenite as possible, all transformation was performed just below -120°C. Some observations were made on transformed samples which had been heated for 2 min at 340°C. It is assumed that the features of the martensite of these samples, Figs. 1 and 4, are the same as those of the as-quenched martensite. Orientation of the crystal by X-ray diffraction established 10.735 0.609 0.3161? as the axis of the crystal, an orientation that was checked within 2 deg by neutron diffraction. Further checks by electron diffraction of samples cut normal to the axis confirmed this orientation within the larger limits of error inherent in electron diffraction of thin foils. The X-ray orientation was the one used for the two-surface analysis of the martensite habit planes. A two-surface analysis was performed on the quadrant of the single crystal which had been oriented by both X-ray and neutron-diffraction techniques. Photomicrographs at X50 were made on two surfaces along an edge 2 cm long. Fiducial marks and the fact that many of the plates were almost completely surrounded by retained austenite made good matching of individual plates on two surfaces possible. The habit-plane trace on a surface was taken as the best line parallel to the long axis of a plate. A measure of the accuracy afforded by this criterion was provided by a family of very large plates which appeared at intervals along the entire 2 cm length of the edge. The plates all had habit-plane traces within 2 deg of one another. Many of the plates did not show midribs and, therefore, the use of midribs7 to represent habit-plane traces was not feasible in this investigation. The over-all experimental accuracy is estimated to be better than ±2 deg. Samples for transmission examination in a Siemens Elmiskop I at 100 kv were prepared by cutting 2-mm-thick discs from the single crystal, removing about 0.5 mm by chemical polishing,14 trans-

Jan 1, 1965