Search Documents

By G. B. Gibbs

Equations are developed which describe the growth of graphite nodules during a-phase graphi-tization under various limiting conditions. When carbon diffusion is rate controlling a modified Birchenall-Mead equation predicts an "apparent activation energy" of 31 keal mole-1 for the parabolic growth constant. If reaction at the graphite or cementite interface is the slowest process growth kinetics are linear and the activation energy must be >50 keal mole-1. A comparison with the experimental observations confirms that carbon diffusion determines the rate of reaction. Apart from minor approximations associated with Lou! carbon solubility in a iron the theory may also be applied to I-phase graphitization. It is well-known that cementite exists as a meta-stable phase in Fe-C alloys and that if a specimen is held at diffusion temperatures for a sufficiently long period the cementite decomposes and graphite is precipitated. Graphitization occurs by the nu-cleation and growth of carbon nodules and several impurity elements catalyze the reaction when present in minute amounts.1 The growth kinetics of nodules have been extensively investigated in the fee ? range2, 3 and recently in the bee a range.1, 4 Experimental growth curves (nodule radius vs time) have slopes which decrease with time, but the scatter of points is generally too great for the precise shape of the curves to be determined with certainty. Birchenall and Mead5 concluded that they are parabolic in ? iron, indicating diffusion control, and that the migration of carbon atoms determines the growth rate. Harris et al, 1 advanced similar ideas for a-phase graphitization. On the other hand, Burke3 chose to regard the initial growth region as a transient followed by a linear law indicating interface control (i.e., cementite decomposition or graphite crystallization as the slowest process). Jeminson and Higgins6 originally considered the initial steep portions of "parabolie" to represent a linear growth law also. In support of these ideas it has been noted that the apparent ac-

Jan 1, 1965

By F. V. Lenel, G. S. Ansell, M. J. Salkind

An experiment is described in which the growth of interparticle necks in an array of loose spherical silver powder at temperatures near 300°C was determined by measuring changes in the electrical resistance of the array. The activation energy for neck growth of contacts was derived by abruptly changing temperature and noting the change in the rate of neck growth. Since the experiments were run under an external load, the changes in activation energy with stress could he determined. The activation energy was found to decrease rapidly with increasing stress. It is concluded that the material transport in the experiments is one of plastic flow by dislocation motion. The variation of activation energy with stress is typical for cross slip of dislocations, an important material transport mechanism in the temperature range investigated. THE process in which powder compacts or aggregates of loose powder are held at an elevated temperature under an external load is called hot pressing or sintering under pressure by powder metallurgists. In this investigation the early stages of this process as applied to loose spherical silver- powder aggregates were studied. The powder particles were approximately 100µ in diameter and the temperature was near 300°C, i.e., below one half of the melting point of silver in degrees Kelvin, where volume diffusion is not expected to contribute to sintering. In the process, the contact or neck areas between powder particles grow. Since this involves plastic deformation of the particles under a constant external load (even though not under constant stress) it would be termed creep by mechanical metallurgists. On the other hand, the growth in neck area observed in compacts or loose-powder aggregates held at elevated temperature without the application of an external load is termed sintering by powder metallurgists. During sintering the major driving force is the stress due to surface tension. However, even when no load is placed on the compact or powder aggregate, an external stress, i.e., gravity, plays an important role in sintering in addition to those stresses arising from surface tension, as was shown in two previous investigations.'" It is therefore postulated that no sharp dividing line can be drawn between the phenomena of "creep" and "sintering". Different material transport mechanisms may be involved in either of these phenomena, but they will depend upon the temperature and the level of stress, but not necessarily upon the nature of the driving force: surface tension or external load. At the low temperatures of this investigation, an external stress much higher than the stress due to surface tension is necessary to obtain measurable rates of neck growth. In order to avoid misunderstanding this phenomenon will be called "creep". The rate of growth in contact area in the loose silver-powder aggregates was determined by meas-

Jan 1, 1965

By G. R. Speich, L. Zwell, H. A. Wriedt

HORNBOGEN1 has recently shown that it is possible to measure the degree of segregation accompanying cellular precipitation in ferritic Fe-Zn alloys by measuring the lattice parameter of the ferrite phase within the cell. In extending this work to study a wider range of composition and to evaluate the free-energy change accompanying the growth of the cells, the lattice parameter and the boundary of the a! phase in Fe-Zn alloys were redetermined to check the earlier work of Schramm.2 This appeared to be necessary since the Fe-Zn diagram has been drastically revised since the work of Schramm. Stadelmaier and Bridgers3 have recently shown that, contrary to prior belief: the y iron field is closed and the a phase boundary extends to an a + T + L peritectic at 782°C where peritectic a contains 42 at, pet Zn. The phase T in equilibrium with the a solid solution has a composition near Fe3Zn10 and a cubic 081 type structure.' The alloys were prepared by two different methods. In the first method carbonyl iron powder (400 mesh, 99.8 pet pure) and high-purity zinc powder (200 mesh, 99.99 pet pure) were mixed in a ball mill and then pressed in a die at 40,000 psi pressure to make compacts with a diameter of 5/8 in. and a thickness of l/4 in. These were sealed in evacuated quartz capsules, gradually heated to 1100°C over a period of 1 week (3 days at 600°C) 3 days at 800°C, 1 day at (1100°C), and then quenched in iced brine. For the alloys containing 23.4 and 30.1 at. pet Zn, the final homogenization anneals were performed at 1000" and 850°C) respectively, to prevent melting. In the second method, alloys were prepared by saturating high-purity iron millings (0.001 in. thick, 99.9 pet pure) in zinc vapor. For each alloy the impregnation was conducted at a fixed vapor pressure of zinc and a fixed temperature between 750' and 800°C. When saturated, the millings were quenched in water. The compositions of the alloys prepared by both methods are shown in Table I. The lattice parameters of the ferritic solid solutions were determined by comparing the 26 values of their 310, peaks (using Co Ka) or of their 211 , peaks (using Cr Ka) with those of corresponding peaks for high-purity a iron. The 20 values were measured at room temperature with an X-ray dif-fractometer (G.E.-XRD5). By using a fiducial value of 2.8664Å for the lattice parameter of pure a iron at 25oC,5 the lattice parameters of the zinc-bearing

Jan 1, 1964

By H. T. Clark

Within the last twenty years at least three sets of lattice constants for the room temperature (alpha, hexagonal close packed) phase of titanium have been reported in the literature. These values, summarized in Table 1, differ one from another by amounts far greater than can be explained on the basis of inaccuracies in parameter measurements. This paper presents new values determined for high purity titanium. It also details the effect of two elements, oxygen and nitrogen, on these parameters and shows that two of the previously reported values are probably in error because of the presence of one or both of these elements in the material used for examination. Preparation of Samples The titanium+ used for this study was prepared by the thermal decomposition of titanium tetraiodide7 on a hot titanium filament. This titanium rod was prepared for X ray examination by melting about 10 g in purified argon in an electric arc furnace on a cold hearth. This cast button was cold rolled to 0.040 in. thick sheet, cleaned, annealed in vacuum at 1000°C and cold rolled to 0.020 in. thickness. About 0.003 in. of surface material was then removed by hand grinding on papers to remove any contamination caused by the rolls and the sample was bent to the radius of the focusing camera. The sample was then vacuum annealed at 600°C, a temperature just high enough to cause recrystallization, producing a fine grained structure. Such a structure is essential in sheet specimens prepared for focusing cameras in order to have a sufficient quantity of grains to develop continuous diffraction lines. The use of sheet specimens rather than powder samples appears to be essential for the X ray examination of titanium. This is because of titanium's exceptionally high gettering power and its rapid absorption of adsorbed gases, water -vapor, and the like at annealing temperatures. While the effect of this contamination may be neglected for sheet specimens, its effect on 200-300 mesh powder is appreciable. Titanium alloys of oxygen and nitrogen were prepared by a technique simi- lar to that described above except that the required amount of the highest purity titanium oxide or nitride available was added prior to melting in the arc furnace. These materials were added by placing them in holes drilled in the titanium. A small weight loss invariably occurs during the melting of these oxygen and nitrogen alloys. This is believed to be due entirely to the loss of some of the powdered compound in the blast of the electric arc, an assumption which is justified on the basis that there is no weight loss during the melting of titanium metal and that wet analyses for nitrogen on representative samples have checked with the calculated analyses when it was assumed that the entire loss during melting was due to the loss of titanium nitride powder. These alloy buttons were reduced to sheet in the same manner as used for the unalloyed titanium. Experimental Procedure A North American Phillips 12 cm diam symmetrical precision focusing vacuum camera and copper radiation

Jan 1, 1950

By A. C. Losekamp, R. M. Fincel, J. B. Conway

HIGH-TEMPERATURE thermal-expansion data for tungsten, rhenium, molybdenum, Mo-50Re, and W-25Re were obtained recently by means of a paired

Jan 1, 1965

By D. Mizer, J. B. Clark

In a recent investigation of the entire magnesium-yttrium phase diagram, Gibson and Carlsonl report the maximum solid solubility of yttrium in magnesium as 8.0 wt pct Y at 1050°F (565°C). However, independent studies of the precipitation in magnesium-yttrium alloys, 2 conducted prior to the publication of the above phase diagram, indicated a greater solid solubility of yttrium in magnesium. Accordingly, the magnesium-rich region of the diagram was re-investigated to determine more accurately the liqui-dus, solidus, and solvus of the magnesium solid solution in this system. Metallography and thermal analysis were chosen as the best corroboratory methods for accurately determining this region of the diagram. A magnesium-yttrium master alloy was prepared from yttrium of 99.9 pct purity and singly sublimed magnesium of 99.9 pct purity. Subsequently, the twelve magnesium-yttrium alloys shown in Fig. 1 were prepared using standard alloying procedures for magnesium. Metallographic samples of the alloys were heat treated under an SO, atmosphere in an electric furnace capable of control within ±3oF of the desired temperature. Alloy samples for the metallographic determination of the solvus, and for the precipitation studies were solution treated at 1000° F for 16 hr. Subsequent aging times for the solvus determination varied from a minimum of 6 hr for the higher temperatures up to 816 hr for lower temperatures. Periods of 30 min at temperature were used to generate liquation for the metallographic placement of the solidus and eutectic horizontal. The liquidus and the eutectic horizontal were determined by thermal analysis. Simple time-temper- ature heating and cooling curves were obtained from 10-g alloy samples heated and cooled in a well-insulated electric furnace. An Iron-Constantan thermocouple was placed in direct contact with the sample. The thermocouple did not contaminate the alloy sample since metallographic examination showed that the alloy melt did not wet or attack the thermocouple. The heating and cooling rates were approximately 5 to 8OF per min. The standard metallographic procedures for magnesium were used in polishing and etching the magnesium-yttrium alloys studied. Glycol etchant* was etchant was used to reveal precipitate in the solvus determination. The redetermined magnesium-rich region of the magnesium-yttrium diagram is shown in Fig. 1. The

Jan 1, 1962

By Y. L. Yao

THE NaT1-type compound may be considered as the penetration of two diamond lattices in such a way that a superstructure of the bcc lattice is formed. Examples of the NaT1-type compound of lithium are LiZn, LiCd, LiA1, LiIn, and LiGa. Except for LiZn and LiCd, each molecule of these equiatomic compounds contains a total of four valence electrons. In LiZn and LiCd it was suspected (for example, see Ref. 1) that the fourth electron was withdrawn from the 18-shell of zinc or cadmium. An incomplete 18-shell would have to show strong paramagnetism. In 1955 Klemm and Fricke2 determined the magnetic susceptibility of LiZn, LiCd, LiA1, and LiIn (also NaTl and NaIn). The molar susceptibility of the outer electrons of LiZn and LiCd was found to be not strongly but weakly positive; that of LiAl and LiIn not negative but weakly positive. Thus they questioned the hypothesis about LiZn and LiCd in particular and the description of the NaT1-type structure in general. However, they did not determine the magnetic susceptibility of LiGa. This note is to report results for all five compounds of lithium. The starting materials are 99.9 pct pure lithium sticks containing less than 0.005 pct Na. 99.9 pct Zn, 99.9 pct Cd, 99.99 pct Al, 99.999 pct In, and 99.999 pct Ga. Weighed amounts of two components were wrapped in tantalum foil and placed in a graphite crucible. It was found that the tantalum sheath minimized the vaporization losses of lithium during melting. The sheath was easily stripped from the alloy rod and the tantalum in the alloy was barely detectable (less than 0.01 wt pct). The melting was performed in a resistance furnace under an argon atmosphere. Alloy specimens were annealed for 1 hr at 400°C in vacuum. The magnetic susceptibility at room temperature was determined by the Faraday Method, using the procedures previously described.' The results are shown in Table I. The interpolated values at 50 at. pct of Li and the average values at 48 to 52 at. pct of Li at 20°C determined by Klemm and Fricke2 are also shown. It is seen that the present value for LiZn and LiCd is slightly more positive but the present value for LiAl and LiIn is slightly less positive than their corresponding values. Also, it is noted there is an abrupt decrease of susceptibility of LiAl from 51.2 to 52.5 at. pct of Li. Klemm and Fricke2 found that the total susceptibility of LiZn, LiCd, LiA1, and LiIn is almost independent of temperature. slater4 gave a working expression for the diamagnetism of the atomic core as follows:

Jan 1, 1964

By K. Winterton

The effect of a near-saturation unidirectional magnetic field was to decrease slightly the endurance life in Armco iron in the annealed condition. This is expected since the energy-absorbing mechanism associated with domain wall movemenfs is inhibited bv the strong field. THE factors determining failure under dynamic loading are not yet clearly recognized. One of the properties of a metal which might be thought to influence fatigue life is its damping capacity. Foppl1 conducted an alternating torsion test at a frequency of 700 cpm continuously for a period of 3 years, without the test piece breaking or showing any sign of deterioration after 1100 Million cycles of stress; the rate of dissipation of energy was sufficient to maintain the test piece at about 100°C above the ambient temperature. cazaud2 concludes from this, and others follow him, that ('the capacity for dissipation in the form of heat by internal damping is unlimited" and that therefore "damping is not directly related with fracture by fatigue." Future work may necessitate qualification of this rather uncompromising view. It is shown later that in all

Jan 1, 1960

By C. M. Wayman, J. F. Breedis

The crystallography of the martensitic transformation in the Fe-30 Ni alloy was reinvestigated. The scatter in the habit Plane as determined from measurements on the mid-rib plane was smaller than reported previously, and the orientation relationship was determined from a martensite plate having known direction cosines. From agreement with crystallographic theory and from observations of the morphnlogy of the martensite plates, it is suggested that the mid-rib plane represents the starting region of the transformation. THE crystallography of the martensitic transformation in the iron-nickel system has been the subject of many investigations.&apos;-7 However, for iron alloys containing approximately 30 wt pct Ni, a wide scatter in the reported habit planes exists. In only one instance has a Laue determination of the orientation relationship been made, but the habit plane of the martensite plate studied was not determined experimentally. Because the most successful thermodynamic and kinetic analyses8 of the martensite transformation have been based on Fe-30 Ni alloys, it was considered desirable to reinvestigate the crystallographic features of the transformation in this alloy. Lieberman9 suggested that the most significant information for use in a theoretical treatment involves the determination of all the crystallographic observables from a single plate of the product phase. This was first done in essence by Greninger and Troiano10 for Fe-22 wt pct Ni-0.8 wt pct C, and is the method followed in this investigation. Some theoretical approaches11-13 in describing the martensitic transformation consider the overall transformation distortion to be comprised of a Bain distortion, or lattice deformation (PI, which produces the structure change, and a lattice invariant deformation (G), such as twinning, which together leave the interface or habit plane un-distorted. The incorporation of a rigid body rotation (R) results in a transformation distortion (E) such that the interface plane is left both undistorted and unrotated. In matrix notation, E = RPG where the lattice invariant deformation has "taken place" in the austenite. Bowles and Mackenzie14-16 in formulating a theoretical approach proposed an adjustable scaler quantity, denoted as 6, which represents a uniform dilation in the interface plane. This dilation acts to adjust the lattice spacings of the parent and product structures at the plane of contact. Using the dilation parameter as a variable, these authors were able to account for a continuous variation of habit planes in steels along a curve in the stereogram between {225}A and {259}A. she invariant line strain14 of the transformation is 6 RP, and the corresponding transformation distortion is E/6.) With the dilation parameter known, the invariant line is computed as the triple coincidence of the habit plane, the shear plane of the lattice invariant deformation, and the initial Bain cone.14 The contraction axis for the Bain distortion can be taken9&apos;17 as [010]A, and the correct permutation of the habit plane indices used. Kelly and Nutting 18 have established the shear plane (of twinning) in the tetragonal martensite of Fe-22 Ni-0.8C to be {112)M by means of transmission electron microscopy. The reasonable shear system is then {112}M which is equivalent to {011}A <01l>A. The as-

Jan 1, 1962

By J. G. Early, F. V. Lenel, G. S. Ansell

The isothermal shrinkage rates of copper-powder compacts were determined in the temperature range from 760o to 1060oC. The rates for compacts fabricated from a pure spherical copper powder were compared, on the one hand, with those fabricated from a spherical powder containing a dispersion of aluminum oxide in the copper matrix and, on the other hand, with those fabricated from a pure copper powder having irregularly shaped particles. The irregular copper-powder compacts shrank much faster than the spherical copper-powder compacts and these, in turn, shrank faster than dispersion -strengthened copper -powder compacts. The experiments were run under a small external load. By varying this load, the stress dependence of the isothermal shrinkage rate could be measured for compacts from spherical-copper powders. This stress dependence was found to be approximately proportional to the 4th (dispersion-strengthened alloys) or 4.5th (pure metals) power of the stress. It is concluded that the material-trans port mechanism is due to dislocation climb rather than to Herring-Naharro creep. In this investigation the isothermal shrinkage rates of compacts made of copper powder with an average particle size of 70 g in the temperature range from 780o to 1060°C were studied. The aim of the investigation was to determine the mechanism of material transport responsible for the shrinkage of these compacts. Since the phenomenon observed in this investigation was shrinkage of compacts at high temperatures, it will be necessary to consider all the material-transport mechanisms possible in sintering. It is unlikely that evaporation and condensation, surface diffusion, or even volume diffusion of vacancies, in which surfaces of small radius of curvature are vacancy sources and those of larger radius of curvature vacancy sinks, will materially contribute to a sintering process involving shrinkage, i.e., one in which the centers of adjacent particles move towards each other. The two principal mechanisms which were considered were the so-called Nabarro-Herring mechanism and a plastic deformation mechanism by dislocation movement. Both of these mechanisms were originally conceived to explain creep, i.e., the plastic deformation of metals at elevated temperatures under a constant applied stress. The shrinking of metal-powder compacts which was studied in this investigation can be readily treated as a creep process. This means that the shrinkage rates observed are considered as creep rates under the influence of an applied stress. It is true that the stresses in sintering experiments are generally not constant but decrease continually with time, but, by considering creep rates or shrinkage rates over a relatively short time interval, the small changes in stress during this time interval can generally be neglected. Nabarro-Herring microcreep is a high-temperature deformation mechanism which is caused by a diffusional flow of vacancies from boundaries under a tensile stress towards those under a compres-sive stress. Herring1 has shown that in this type of microcreep the strain rate e is proportional to the applied stress and follows an equation of the form ma, aexp(-Q/RT) [l| where <j is applied stress, Q is the activation energy for self-diffusion, R is the gas constant, and T is the absolute temperature This type of creep can explain the shrinkage of a porous compact. The surface of the pores are subjected to the tensile stress of surface tension, while the grain boundaries are under a compressive stress also due to surface tension. Vacancies diffuse from the pore surface through the lattice to the grain boundaries, where they are eliminated. In order to maintain vacancy equilibrium at the pore surface the vacancies leaving the region of the pore surface are replenished from the volume of the pore, which means that the pores shrink.

Jan 1, 1964

By J. E. Hilliard, J. W. Cahn

In this paper it is suggested that the degree of contiguity (or contact) between adjacent grains be described by three parameters which can be rigorously determined from measurements on a random plane of polish. An expression is also derived for the number of contacts per unit volume and it is shown that this quantity can-not, in general, be estintated from measurements on a random plane. IN a recent contribution1 Gurland has proposed various parameters for describing the degree of contiguity (or contact) between grains of the same phase in an alloy. To relate these parameters to measurements on a random two-dimensional section, it was necessary in most cases to make some assumption about the form of the structure. Since Gurland was specifically interested in sintered carbide structures, he chose, quite reasonably, to assume that the grains were constant-size spheres and that all the intersections were flat areas of constant radius. Since, as we will show, the relationships between the parameters and the measurements are strongly dependent on the grain shape and size distribution, it is evident that these assump tions will not be satisfactory approximations for a

Jan 1, 1960

By R. M. Asimow

The magnetization curves for specimezs containing superparmagnetic particles are considered. It is shown that the curves may differ from a Langezlin function because of particle anisotropy, particle interactions , and the presence of a size distribution of particles. The various causes of particle anisotropy are considered, and magnetization curves for specimens with cubic and uniaxial anisotropy are presented. The magnetization curves depend on the In the particle-size below 100A it is quite difficult to obtain quantitative size and shape information on precipitate particles. All of the available techniques have their limitations; however, quantitative applications of superparmagnetism have not been developed as fully as possible due to a lack of analysis of the complete magnetization curve. With such an analysis it appears possible to obtain from a single magnetization curve of a specimen containing randomly oriented SPM (superparamagnetic) particles a quantitative volume distribution and a description of the particle shape.* sign of the anisotropy energy and in the case of a negative uniaxial anisotropy (prolate spheroids) the resulting curves may differ very considerbly from a Langevin function. A graphical technique is developed whereby it is possible to separate out the effects of the size distribution, the particle interaction, and the particle anisotropy. Thus it is possible to obtain quantitative values for both the particle shape and the size distribution. Langevin function2 I/Is = ctnh(HVIs/k T)-k T/HVIS [ 1 ] where I, is the saturation magnetization, V the particle volume, H the applied field, T the particle temperature, k the Boltzmann constant, and I the magnetization in the direction of the applied field. Such behavior is rarely if ever observed in actual systems. One explanation for this is that the SPM particles do possess a significant anisotropy and thus the magnetic energy associated with the particle depends on the direction of magnetization in a more complicated way than for the isotropic particle of Eq. [I]. In general the magnetic energy E of an isolated SPM particle embedded in a nonmagnetic matrix can be expressed as a sum of several contributions: E = Eh + Ed+Es+Ec [2] where Eh is the external field energy, Ed the energy resulting from the particle demagnetizing factor, E, the magnetostriction energy, and E, the crystalline anisotropy energy. In each term we consider only the part dependent on the orientation of the magnetization vector and the particle axes as the shape of the magnetization curve will be determined by these contributions. Following the same procedure used in deriving the Langevin function,

Jan 1, 1965

By D. M. Schuster, E. Scala

The elastic effects occurring in the matrix of a composite reinforced by discontinuous fibers were studied by means of photoelastic techniques. A hirefringent plastic was employed as the matrix material with high-strength a Al2O3 whiskers as the reinforcing fibers. It was found that for whiskers aligned parallel to the tensile load direction: (I) Matrix reinforcement occurred along the length of the whisker to within about 5 diameters of the whisker tip for a length to diameter ratio, L/D, of 40. (2) Significant stress concentrations were created in regions immediately surrounding the tips; the fine as-formed tapered ends exhibited the minimum stress concentration whereas the square-ended whisker produced relatively high values, K r 2.5 for L/D = 40. (3) The axial and radial stress distributions could he determined quantitatively; the stress distribution at the whisker-matrix interface was in general agreement with theoretical calculations. (4) The highest source of stress concentration occurred at points of fracture in whiskers which had ruptured after incorporation into the composite. Whiskers aligned perpendicular to the load direction neither reinforced nor caused appreciahle stress concentrations in the matrix. ThE properties of composites have generally been formulated empirically on the basis of macroscopic tests. There are still many questions concerning the mechanical and chemical interactions occurring at the interface between the reinforcing fiber and its matrix. Continuously reinforced plastics and dispersion-hardened metals represent the two extremes of composite reinforcement. Whether artificially produced or formed in situ, the whisker composite falls between these two extremes and is an example of discontinuous fiber reinforcement. The effect of fiber shape, size, surface condition, and end configuration on the stress distribution in the matrix is important since the presence of stress concentrations, especially in any high-strength, thin-wall structure, could become the cause of catastrophic failure. DOW,&apos; in his theoretical discussion of discontinuous fiber reinforcement, has pointed out that appreciable stress concentrations occur at the tips or ends of the reinforcing fibers. It is the purpose of this study to examine directly, by means of the photoelastic technique, the stresses which occur in the vicinity of a discontinuity or whisker tip and to measure qualitatively and quantitatively the effects of whisker geometry when embedded in a birefringent matrix material. The use of sapphire (a A1,O3) whiskers was decided upon as the reinforcing agent because of their high elastic modulus (E- 60 X 106 psi) and in some cases strengths in excess of one and a half million psi. Sapphire whiskers are of further practical interest since they retain large fractions of their strength up to temperatures approaching the melting point (3720oF).2 The whiskers were grown, by the vapor-phase reaction commonly employed,3 as part of this study to insure a physically uniform supply. Tensile and bend tests were performed to verify that the whiskers were of the expected ultra-high strength. The high-strength whiskers can be identified and selected from the combustion boats by surface perfection, shape, and size; otherwise, strength values can vary by a factor of ten or more. Macroscopic defects are quite common on the surface of large bladed whiskers and the principal experiments were conducted with hexagonal whiskers of about 0.002 in. diameter and lengths ranging from 0.1 to 0.4 in. Upon completion of these two phases of the experiment, several birefringent resins were evaluated. A plastic, developed by Zandman,4 which is supplied by the Budd Co., was found most suitable and was used as the matrix material for the whisker composite throughout the study. It has a sensitivity of 60.5 lb per in. order, calibrated in tension and cures without introducing residual stresses. Young&apos;s modulus for this matrix material is 430,000 psi in the fully cured condition. Since good wetting and bonding between the sapphire and the matrix material is essential to the interpretation of the results, a wettability test was performed. The contact angle was measured to be 168 deg where perfect wetting is defined as 180 deg; therefore, the Budd Plastic wets the sapphire

Jan 1, 1964

By J. W. Freeman, R. F. Decker

A microstructural investigation was pursued to establish the mechanism of the pronounced benefits of traces of boron and zirconium on creep properties of a 55 Ni-20 Cr-15 (20-4 MO-3 Ti-3 A1 alloy at 1600°F. The mechanism of improvement of creep properties was found to be a pronounced stabilizing effect of traces of boron and zirconium on the grain boundaries of the alloy. The alloy with low boron and zirconium was subject to rapzd agglomeration of M23 C6 and in the grain boundaries, followed by depletion of y &apos; and intergranular microcracking at the grain boundaries transverse to applied stress. Brittle fracture then occurred by linking of microcracks. However, additions of zirconium, boron, and boron plus zirconium decreased this tendency in that order, allowing longer creep exposure and higher creep defor.mations before fracture occurred. VERY little information has been published describing the mechanism by which trace amounts of boron or zirconium improve the creep properties of complex nickel-base heat-resistant alloys, although the effect is now well recognized.1-4 Because the phenomenon is so important to properties and the effects are so striking, a study was made of the mechanism in a 55 Ni-20 Cr-15 Co-4 Mo-3 Ti-3 A1 alloy. It was believed that such a mechanistic approach would increase knowledge in theory of alloying and give a more basic understanding of alloy design to resist creep. The basic premise of the study was that boron or zirconium would alter the microstructural characteristics controlling creep and fracture to the extent that the effects could be identified by light and electron microscopy and by electron diffraction techniques. Characteristic of the microstructure of the Ti + Al-bearing nickel-base alloys is the dispersed precipitation of the intermetallic compound Ni,(Al, Ti), the ? &apos; phase, which is generally con- sidered to be the major source of high creep resistance.5-7 In addition, carbide phases have been identified8-10 and these can alter creep properties markedly.11 Since both the ?&apos; and carbide transformations have been found to control properties, careful attention was paid to the effect of boron and zirconium on these transformations.

Jan 1, 1961

By R. A. Vandermeer, Paul Gordon

On the basis of a unified concept, theoretical erPressions for grain boundary migration in recrys-tallization are derzved for impurity-controlled and impurity-independent migration. The expression in the former case is analogous to that Previously derived by Lucke and Detert, and in the latter is shown to be equivalent to that usually obtained on the basis of absolute reaction rate theory. The equations are tested with quantitative experimental data for boundary migration obtained for the recrystallization of poly crystalline zone-refined alumimnz and for binary "alloys" of this aluminum with 0.00021, 0.00043, 0.0017, 0.0034, 0.0068, 0.0124, and 0.0256 at. pct Cu, respectively. All the essential characteristics of the e@erimentally determined boundary migration rates check both qualitatively ard quantitatively with the predictions of the theoretical equations. THE rates of transformations are customarily analyzed phenomenologically in terms of the two component rates—N, the rate of nucleation and, G, the rate of growth of the nuclei. The scientific aim of both theory and experiment is to describe the fundamental atomic phemmena involved in these two component processes with the hope that parameters calculated from theoretical equations will check those found experimentally. In solid state transformations, the theoretical and experimental problems are both generally of almost prohibitive complexity, for in addition to changes in composition and in such properties as density which may accompany transformations in amorphous, nonrigid materials, transformations in crystalline substances also involve structural changes, crystal reorientations, and the development of stresses. In studying such transformations it thus becomes; highly desirable, if not essential, to focus on a transformation and a set of conditions which minimize the number of active variables. This principle has been applied with considerable success by Cohen and his associates&apos; in their evaluation of nucleation theories using the martensite transformation as the basis of study. A similar possibility exists for the evaluation of growth theories by the use of recrystallization after cold-deforming. If an unalloyed metal is employed as the subject of investigation, the recrystallization transformation involves basically only a change in strain-conditions and orientations. In spite of this possible simplilfication, however, until recently little quantitative success has attended attempts to correlate recrystal1ization parameters measured experimentally with those calculated theoretically. It seems highly probable that this lack of success may be attributed, both in the experimental and theoretical treatments, to one or more of the following factors: a) conditions have usually been studied under which apparently* both nucleation and growth are active;

Jan 1, 1962

By John C. Sawyer

The mechanism of catastrophic oxidation of chromium and 446 stainless steel is examined. Data are presented to show that accelerated oxidation of these two materials, as caused by lead oxide, can occur in the absence of a liquid layer contrary to presently accepted theory. An alternate theory is proposed in which the rate of accelerated oxidation is a function of the rate at which lead oxide destroys the protective oxide formed on the base metal. An example of the application of the theory is given for the catastrophic oxidation of chromium in the presence of lead oxide. WHEN stainless iron-, nickel-, or cobalt-base alloys are heated in air to moderate temperatures in the presence of certain metallic oxides, oxidation will proceed at an accelerated rate. This phenomenon, often called "catastrophic oxidation", is most pronounced for the stainless steels. With these alloys the condition is so severe that large masses of oxide will form on the surface of the alloy in 1 hr or less at temperatures of 1200o to 1700oF. While a number of oxides are known to cause this effect, PbO, V2O5, and Moo3 are the most familiar, having been the subject of one or more investigations which have appeared in the literature.1-7 In presenting the results of these investigations, many of the authors have offered possible explanations to account for the more rapid rate of oxidation observed; however, the liquid layer theory as proposed by Rathenau and Meijering 2 has been the most commonly accepted mechanism. The liquid layer theory proposes that a low-melting oxide layer is formed on the surface of the alloy as the result of the interaction of the alloy oxide and the contaminating oxide. When the temperature of oxidation is above the melting point of the oxide on the surface, a liquid layer will form and oxidation will proceed at an accelerated rate. At temperatures below the melting point of the surface oxide, oxidation will proceed more slowly in the normal manner. It is argued that the rates of diffusion of oxygen and metal ions through the liquid layer are extremely rapid thereby accounting for the high rate of oxidation. Various experimental data have been presented to show that the temperature at which accelerated oxidation first becomes apparent coincides with the melting point of the eutectic oxide which would be present on the surface. Some exceptions have been observed, e.g., silver will oxidize in the presence of Moo3 at temperatures below the lowest melting eutectic; on the other hand, stainless steel will not be catastrophically oxidized at 1500oF in a molten bath of PbO and SiO2. In reviewing the various theories which have been used to explain catastrophic oxidation, Kubaschewski and Hopkins 8 favor the liquid layer theory, but note that, ".. .as experimental observations are not altogether in agreement with this theory (liquid layer theory), one should consider it a necessary but not a sufficient condition." In contemplating the liquid layer theory, it appears that sufficient evidence has not been presented to establish the theory beyond question. As a means of further clarification, a program of research was undertaken to determine in greater detail the mechanism of accelerated oxidation as caused by lead oxide. The first part of the program deals with a comparison of the oxidation of both AISI 446 stainless steel and chromium metal in the presence of lead oxide, vs the oxidation of these two materials in air alone. These comparisons are made at a number of different temperatures, most of which are below the melting point of the surface oxides. The second part of the program is concerned with a presentation of an alternate theory of accelerated oxidation exemplified by the system Cr-PbO-Air. PROCEDURE AND RESULTS Several experimental methods are commonly used to follow the progress of oxidation. One of these, the weight-gain method, was chosen for this work. This procedure requires that a specimen of the alloy be weighed, oxidized for a given period of time at an elevated temperature, and reweighed—the difference between the two weights being noted. The weight gain of the specimen represents the amount of oxygen acquired from the atmosphere to transform a portion of the specimen to oxide. In those cases where there is a tendency for the specimen or oxide to volatilize at the testing temperature, additional data must be collected so that a correction factor can be determined. This factor must be applied to the weight change in order to ascertain the actual amount of oxidation which has taken place. The specimens used for this work were 1 1/2 in.

Jan 1, 1963

By W. D. Robertson, A. S. Tetelman

The technique of decorating dislocations was employed to investigate deformation and fracture resulting from Precipitation of hydrogen in Fe-3 pet Si single crystals. It is shown that cracks are produced on (100) planes inside crystals, as a consequence of the precipitation of hydrogen gas, either when crystals are quenched from a hydrogen atmosphere at elevated temperatures or cathodically charged with hydrogen at room temperature. Plastic deformation in the vicinity of cracks, observed as arrays of decorated dislocations, is in conformity with the calculated stress distribution about a crack containing an internal pressure, and the observations provide information permitting a detailed analysis of the mechanics of crack growth. The fracture characteristics of crystals containing internal cracks were evaluated at 25o and -196°C and the results are related to the mechanism of hydrogen embrittlement in terms of the growth of preexisting cracks. HYDROGEN in bee iron or steel produces two characteristic changes in mechanical properties: I), a decrease in ductility (macroscopic strain at fracture) that depends on temperature and strain rate, and 2), spontaneous fracture under static load that depends on time and apparent stress. Resolution of these phenomena into the microscopic processes that combine to produce the macroscopic effects is especially difficult in the Fe-H system in which the active component cannot be directly observed and only maintained in equilibrium with great difficulty in the temperature and pressure range in which embrittlement phenomena appear. As a result of these complications, verification of the numerous hypotheses&apos;-5 that have been proposed to explain the microscopic mechanism of embrittlement depends on self-consistency with measured changes in macroscopic properties that are averaged over the volume of a specimen. It is obviously important to obtain a more direct evaluation of the microscopic processes that combine to produce embrittlement. Accordingly, the present investigation was designed to identify and observe, as directly as possible, the microscopic effects of introducting hydrogen into a bee crystal. It was concluded from a previous investigation6 of X-ray line broadening, produced by introducing hydrogen into pure iron at room temperature, that the only effect that could be associated with the introduction of hydrogen was plastic deformation; in particular, dilation of the lattice was not observed, within ± 0.01 pet, and no evidence was found that could be interpreted to support the idea7&apos;* that hydrogen occupies tetrahedral interstitial sites in the iron lattice. The absence of lattice dilation is consistent with the extremely small solubility of hydrogen in bee iron9,10 (0.015 at. pet at 900oC and 1 atm pressure), and the evidence of plastic deformation is consistent with the fact that hydrogen, introduced by cathodic charging, eliminates the carbon yield point.11-13 Following this investigation, the immediate problem was to devise a method by which plastic deformation could be observed directly. The technique employed by Hibbard and Dunn14 to observe the movement of dislocations during bending and subsequent polygonization of Fe-3 pet Si single crystals was adopted and modified for the present purpose to include the observation of dislocation arrays in the volume of the crystal, and their association with the fracture process. EXPERIMENTAL PROCEDURE Single crystals of a high purity Fe-3 pet Si alloy, containing 0.007 pet C, were grown in vacuum by

Jan 1, 1962

By A. R. Troiano, A. B. Greninger

There is need for an adequate working hypothesis that would describe at least qualitatively the crystallographic mechanism for the transformation from austenite to martensite in steel. A general theory would not only be of great assistance in the correlation of existing data and in planning future experimentation, but would also form the basis for an eventual explanation of the property changes that accompany lattice transformations. The mechanism of martensite formation described in the following pages is the outcome of two new experimental determinations: (1) the accurate evaluation of the lattice relationship between austenite and individual crystals of martensite, and (2) the measurement and analysis of the change in position that a volume of austenite undergoes when it transforms into a crystal of martensite. For this latter study, the stereographic analysis of shear was employed; this method greatly simplifies the solution of lattice-shear problems, and some space is devoted to its description. Lattiice Relationships and Habits PREVIOUS WORK The lattice relationships between austenite and martensite in carbon steels have been studied by Kurdjurnow-and Sachs,&apos; and by Wassermann;2 and in iron-nickel alloys by Nishiyama.3 Wassermann,2 and Mehl and Derge.4 The studies by Kurdjumow and Sachs and by Wassermann were made on single (austenite) crystals of quenched 1.4 pct carbon steel, and the results of these two independent studies agree closely with the postulated lattice relationships: (III)r||(0ll)a [101]r [111]a [1] Both Nishiyama and Wassermann worked with an alloy of iron plus 30 pct nickel (all austenite at room temperature), and "martensitic alpha" was formed by cooling in liquid air. They agree that the lattice relationships for these conditions are: (lll)r||(011)a [112]r[011]a [2] Mehl and Derge worked with a range of iron-nickel alloy compositions centered about 30 pct nickel. They concluded that when the transformation takes place near room temperature or above, relationship [l] holds, but if the same specimen is made to transform at -195°C, relationship [2] obtains;intermediate temperatures evidently produced a combination of these two relationships. Sachs and co-worker~,1,7 as well as Nishiyama,3 have proposed transformation mechanisms for martensite formation.* Both of the mechanisms picture the transformation as initiated by shearing movements along the octahedral [Ill] plane of austenite, followed by suitable expansion and contraction to complete the transformation. These mechanisms have been described and commented upon by Mehl, Barrett, and Smith,8 Mehl and Derge,4 Wassermann,2 and others, and generally accepted by investigators in this field. Greninger and Troiano9 have recently published the results of a detailed study of the shapes and orientation habits of martensite crystals in plain-carbon and nickel steels. They found that, contrary to previous assumptions, martensite plates form not along the octahedral planes but along austenite planes of very high indices; these results were in line with those of previous studies of martensite reactions in nonferrous alloy systems by Greninger and coworkers.6,10 Greninger and Troiano concluded that the transformation theories of Kurdjumow and Sachsl and of Nishiyama3 are untenable. Phragmén11 reached a similar conclusion. EXPERlMENTAL METHODS The object of this particular study was the evaluation of (1) lattice relationship between austenite and martensite, and (2) relationship between the martensite lattice and the martensite plate. The technique used for solving these problems was similar to that used by Greninger6 on martensitic structures in 8 copper-aluminum alloys. Briefly, the method consists of grinding and polishing a specimen on a surface parallel to an individual martensite plate and thus exposing a single martensite crystal for an area of l to 3 mm. The orientation of this exposed martensite crystal is then determined by means of a back-reflection Laue X ray pattern.12 Another pattern is obtained from the matrix crystal, and a stereo-graphic plot of data from these two patterns then provides the solution to the problem. It is necessary to repeat this process on several crystals in order to determine whether or not the solution is unique.

Jan 1, 1950

By C. Ernest Birchenall

TWO recent papers1,2 cite measurements of surface contour changes on copper which, when attributed to surface self-diffusion, can be interpreted to yield activation energies for surface self-diffusion that approach1 or exceed2 those reported for volume self-diffusion in the same metal.3 It is acknowledged2 that AS must be assigned much too large a value to be physically reasonable in order to make the theoretical Do&apos;s agree with the experimental values, if the jump distance is of the order of the interatomic distance. However, long jumps would probably lead to marked anisotropy on planes that consist of ridges and valleys (like (110) in fcc) whereas only mild

Jan 1, 1963

By P. W. Gilles, B. D. Pollock

THE pioneering work of Steinitz1 and Steinitz, Binder, and Moskowitz2 has shown conclusively the existence at high temperature of two additional phases in the molybdenum-boron system and thus brings to a total of six the number of structures appearing in this system. To the structures Mo2B, MOB, and Mo2B5 they have added MO3B2, a new -MOB form, and have shown that MOB,, which has the same range of composition as Mo2B5, is only a high temperature structure of the latter. This solid solution, interestingly enough, includes neither of the compositions corresponding to the stoichiomet-ric compounds, MOB, or Mo2B5, but rather at all temperatures has intermediate values of composition. These workers have also, in the course of their work, measured melting points, transition temperatures, eutectic and peritectic points in the system and have shown that Mo3B2, because of its dispro-portionation at low temperature to Mo2B and MOB, is stable only in a limited high temperature range. During the course of the present work on the vaporization properties of the molybdenum-boron compounds, a few transition temperatures were observed. When the report of the other workers appeared, it was decided to repeat, in part, their study of the system. As a result, considerable evidence has been obtained that substantiates the specific kinds of melting processes they report as well as the general features of their diagram. However, a marked difference was found between the temperatures they report and the ones observed in this study, with the latter being higher. The purposes of this paper are to present the evidence obtained in this laboratory that verifies their diagram of the system, to give some important temperatures in the system, to compare them with those previously published, and to seek an explanation of the difference. Samples The metal starting material was 400 mesh molybdenum powder with a purity stated by the manufacturer to be 99.9 pct. The initial treatment, designed to remove volatile contamination, consisted of heating in a vacuum for 10 min to a temperature of from 800" to 1000°C during which a loss of 0.3 to 0.4 pct occurred. An assay following this treatment showed it to be 99.4 pct pure, with the principal impurity probably being oxygen. The boron starting material was obtained from the Cooper Metallurgical Laboratories and the Fair-mount Chemical Co. as 325 mesh powder with manufacturers&apos; analyses of 99 pct or better. Initial treatment consisted of heating in molybdenum in a vacuum at about 1700°C for 10 min. During this time a loss of 3.5 pct occurred. An assay following this treatment showed the different samples to have purities ranging from 95.5 to 99.0 pct with iron and carbon as the principal impurities. Following the initial treatment, the elements were combined to form stocks of Mo2B and MOB by heating pressed mixtures in a vacuum to 1100" to 1200°C to accomplish reaction and to 1500" to 1900°C for a few minutes to evaporate the more volatile impurities. Analysis of the two compounds for boron by a modification of the method of Blumenthal3 and for molybdenum by the lead-molybdate method indicated them to have purities greater than 99 pct. The individual samples to be studied had compositions in the Mo2B-MOB range and consisted of mixtures of the stock compounds. Procedure As is usually the case in high temperature work the selection of containers for the samples posed some problems. For vapor pressure studies tantalum crucibles, allowing little contact with the pressed samples, were used and some of the observations made during these experiments are pertinent to the study of the phase diagram. Most of the experiments, however, were performed in graphite containers, as were those of the previous authors. Two kinds of spectroscopic grade graphite crucibles were used. One was a % in. cylinder, 3/4 in. high, containing seven 3/16 in. holes drilled 1/2 in. deep into which were packed samples of the different mixtures weighing 250 to 500 milligrams. The other, consisting of separate crucibles, was prepared by drilling 3/16 in. holes, 1/2 in. deep into 1/4 in. graphite rods % in. long. The 7/8 in. cylinder was heated directly by induction while the small crucibles were packed in a tantalum heating element for induction heating. All heating was done in a high vacuum system in which the pressure was generally less than 1x10-5mm and never rose above 2x10-5mm when the samples were hot. The general pattern of the heating in graphite was to heat rapidly to a temperature somewhat below the desired one, then to raise the temperature slowly. The samples were held for 2 to 5 min at the maximum temperature, which in all cases was far higher than that needed to produce reaction. The short time was employed to reduce possible contamination by the crucible material and to reduce composition changes that would occur because of vaporization. After examination following the heating, the samples were reheated to a higher temperature. Temperatures were measured with a Leeds and Northrup disappearing filament optical pyrometer, certified by the National Bureau of Standards, by sighting through a window at the top of the vacuum

Jan 1, 1954