Search Documents

By Michael Hoch, Seong Kwan Rhee

The ternary system Zr-Cr-0 was investigated at 1200°, 1500°, and 1700°C. The isotherms at these temperatures were determined by metallographic and X-ray diffraction analysis of carefully selected alloys. AS a part of a series of investigations1-4 of the ternary and quaternary phase diagrams on refractory metals and their compounds, the ternary system Zr-Cr-O was investigated at 1200°, 1500°, and 1'700°C. The binary diagrams Zr-Cr, Zr-O, and Cr-0 as reported by Hansen5 and Levin et a1.' and the existence of a ternary coompound ZrsCrsO (q-carbide structure a. = 11.96A) provided a good base for this investigation of the Zr-Cr-O phase diagram. MATERIALS The materials used and their suppliers were: Zirconium powder, 99.4 pct pure, Charles Hardy, New York; Zirconium dioxide, Fairmount Chemical Co., Newark, N.J.; Chromium powder, 99.85 pct pure, Fairmount Chemical Co., Newark, N.J.; Chromium sesquioxide, 99.86 pct pure, J. T. Baker Chemical Co., Pillipsburg, N.J. Though the materials used were not "hyperpure", the impurities present do not affect the results (lattice parameters, phase boundaries) within the experimental accuracy. EQUIPMENT AND PROCEDURE The detailed procedures and equipment used for this investigation were similar to those described elsewherel' "ith the exception that in this study an induction-heated argon-atmosphere furnace was used. The latter furnace was used to prevent the vaporization of chromium metal in samples in the higher temperature range (above 1200°C). The tungsten crucible was induct ion-heated in a vertical quartz tube filled with argon gas. The quartz tube was 1-1/2 in. in diameter and 10 in. long; it was connected at the top and bottom to water-cooled copper plates. The argon gas of a purity 99.995 pct was further purified by being passed through zirconium metal chips which were maintained at 800°C in a silica tube 1 in. in diameter and 30 in. in length. The furnace was flushed with argon gas for an hour before starting heating; during heating, a slight (3 cm oil) overpressure of argon was maintained in the system. Power was supplied by a 20-kw Ther-Monic Induction Generator. The samples rested on a tungsten-wire basket; the wire basket was connected to a tungsten wire which passed through the tungsten crucible lid, and entered the system on top through a Wilson-type seal. By pulling on the wire, the sample was removed from the crucible and dropped to the water-cooled copper base of the furnace. Because of the presence of argon gas, the cooling of the sample was even more rapid than in the vacuum furnace.' EXPERIMENTAL RESULTS AND DISCUSSION 1) Isothermal Section at 1200°C. The compositions investigated have been numbered within the phase diagram, Fig. 1, and all the pertinent information about the samples can be found in Table I. This table lists the composition of the samples, the phases present in the room-temperature diffraction patterns, the phases detectable in the microstruc-ture, and calculated lattice parameters of the a-zirconium phase. The shapes of the various phase

Jan 1, 1964

By J. H. Brophy, M. H. Kamdar, J. Wulff

A constitutional diagram for the Ta-W-Re alloy system is presented. Rhenium dissolves in the complete range of solid solutions between tungsten and tantalum up to 48 wt pct in tantalum 'to about 30 wt pct in tungsten. Intermediate x and u phases only are found. X-ray diffraction and fluorescence, metallography and melting-point measurements were employed. THE constitution diagram of ternary alloys of tungsten, tantalum, and rhenium for temperatures above 1000°C is reported in this paper. Alloys in this system rich in tantalum and tungsten are of interest for high-temperature applications, and those rich in rhenium for electronic applications. No ternary phase diagram appears to be available in the literature. The three-component binary diagrams have been previously established1-3 and the present ternary data are consistent with these. EXPERIMENTAL TECHNIQUES Commercially pure tantalum (99.7 pct), rhenium (99.99 pet), and tungsten (99.9 or 99.95 pct) powders were purchased from Fansteel Metallurgical Corp., Chase Brass and Copper Co., and Fansteel or Westinghouse Electric Corp., respectively. These powders were pressed without binder and arc-melted under helium. The preparatory and analytical techniques employed were identical to those used for the Ta-Re system investigation carried out in this laboratory.324 These included X-ray diffraction and metallographic phase identification, X-ray fluorescent and wet chemical analyses, solidus determinations and heat treatments in a resistance-heated tantalum-filament vacuum furnace. For microstructural examinations standard mechanical polishing techniques were used through diamond laps. Polished speciments of all compositions were etched by swabbing with 4 parts HI?, 4 parts HNO3, and 1 part H20. Since the microstruc-tures of alloys containing one or two phases closely resembled their counterparts in the respective binary systems, typical examples are omitted in this presentation. In all the alloys examined, it was possible to observe the number of phases present. Particularly when there were small amounts of one phase involved, it was possible to identify it only by a combined conclusion based on X-ray diffraction, position of an alloy in a composition series, and the phase rule. A total of 100 different alloy compositions was arc-melted and studied in the as-cast condition metallographically and by X-ray diffraction. Specimens of each of these compositions were heat-treated at 2680, 2530, and 2020°C for 1, 2, and 4 hr, respectively. Those which were treated at 2020 were given a preliminary homogenization treatment at 2500°C for 1 hr. Specimens of 25 alloys were heat-treated at 1200o C for 168 hr. Solidus temperature measurements were made by the detection of incipient melting in eighteen alloy compositions. The metallographic and X-ray diffraction results are summarized in Table I. Graphically, this information appears in the isothermal plots of Figs. 2 and 3. Within the accuracy of the experimental methods, two such plots are sufficient to locate these data, since the results for 1200°C were similar to 2020 and those at 2530 resembled 2680

Jan 1, 1962

By R. J. Jackson, D. E. Williams, W. L. Larsen

The phase diagram of the Ta-Zr system is presented with a discussion of the methods used in its determination. The solidus contains a minimum at 1820°C near 25 wt pet Ta. A complete solid solution exists at elevated temperatures between tantalum and ß zirconium. This solid solution decomposes monotectoidally at 800°C and 12 wt pet Ta. The maximum solubility of tantalum in a zirconium is slightly less than 2 wt pet, and decreases with decreasing temperature. The solubility of zirconium in tantalum is 5 wt pet at the monotectoid temperature and about 2 wt pet at room temperature. THIS investigation was part of a program of research on refractory metal alloy systems having potential application in nuclear reactor technology. Preliminary prop-ess has been reported in the Ames Laboratory annual research reports.l Most of the previously reported work on the Ta-Zr alloy system consisted of partial phase diagram investigations and the determination of the mechanical properties of the zirconium-rich alloys. Anderson eta1.2 studied alloys containing up to 30 wt pet Ta as part of an investigation of several zirconium alloy systems. These investigators also determined some of the mechanical properties of the alloys at room temperature and 650°C. In a discussion of Anderson's work, Pfei13 suggested that the bright phase in the microstructures of the intermediate alloys is a tantalum-rich solid solution or an intermediate phase in a matrix of ß zirconium solid solution. Litton,4 Keeler,5 and Chubb6 determined the mechanical properties of zirconium-rich alloys at ambient and elevated temperatures. Van Thyne and McPherson7 reported the tensile properties

Jan 1, 1962

By D. E. Hartman, J. M. Roberts

A detailed study of the temperature dependence of the critical stress, TB . necessary to cause a damping loss of 1.41 x 10-6 g mm per mm3 in pre-strained magnesium single crystals has been carried out in the temperature range 82" to 320°K. Tb is found to be linearly related to the reciprocal of the ahsol~cte temperature in this temperature range. The temperature dependence of the anelastic limit, TA, defined as the stress necessary to form open unidirectional stress-cycle damping loops has been measured between 82" to 320°K. ta is found to be independent of temperature in this range. It is suggested that TA is determined by the stress necessary to activate Frank-Read dislocation NUMEROUS studies1-9 have been reported concerning the details of yielding and microstrain near and below the generally discussed macroscopic flow stress. The work of Kramer and coworkers10-13 is helpful in determining certain qualitative facts. concerning preyield phenomena. The absence of a stress-dependent delay time in annealed copper and aluminum single crystals between 83" and 298 suggests that the preyield mechanism is temperature-independent in this range. This is not sources. The temperature and stress dependence of the effective activation volume (Veff) has been measured in the damping-loop region in prestrained magnesium single crystals. The data suggests that Veff/kT is approximately constant, equal to 2 4.5 sq mm per g in the temperature range 130°to 298°K, and that Veff is almost independent of stress at stress levels greater than 10 g per sq mm. A restricted model is developed to predict the stress (rb) necessary to cause a small fixed damping loss as a function of temperature, effective activation volume, and dislocation density. Various possible dislocation mechanisms controlling TB are discussed. true for zinc, p brass, and iron crystals in the same temperature range, however, since the existence of a stress-dependent delay time suggests a thermally activated preyield mechanism in these materials.10-13 The work by Vreeland et a1.14 confirms the work by Liu et al.ll on delayed yielding in zinc single crystals. A stress-dependent delay time between 82" and 298°K was found12 in prestrained aluminum and copper crystals. These results suggest that the preyield phenomena may be different between prestrained and unstrained or annealed crystals. The temperature dependence of the yield point at 2 x 10-8 plastic strain found by Rosenfield and Averbach3 in annealed copper and aluminum is not in agreement with the delay-time data, unless possibly the delayed yielding experiment only de-

Jan 1, 1964

By B. G. Reisdorf

This paper describes the microstructural changes that occur when quenched ultrahigh-strength steels containing OA pet C and various amounts of nickel, silicon, and cobalt are tempered. The changes that occur in hardness, yield strength, and martensite diffraction-peak widths when these steels are tempered are related to the precipitation of the carbide phases. An unea$ectedly high silicon content (about 30 pct) was found in the E -carbide phase in a steel containing only 1.5 pct Si. TEMPERED martensitic steels have such wide-spread utility that they have been the subject of numerous investigations, and a sizable amount of literature has developed on this subject.M This investigation, which is a part of a larger program to correlate microstructure with mechanical properties at a variety of strength levels, is limited to the microstructural changes that occur when a quenched 0.40 pct carbon steel (USS Airsteel X-200 steel) is tempered, and to the effects of the alloying elements—nickel, cobalt, and silicon—on these microstructures. Used in this investigation were electron microscopy, electron-diffraction identification of the carbide phases, electron-probe micro-analysis of the carbides, X-ray determinations of retained austenite, and measurements of martensite X-ray diffract ion-peak widths. MATERIALS AND METHODS The chemical analyses of the eight heats examined eats No. 30 through 37) are shown in Table L Heat No. 30 has the chemical com~osition of Airsteel X-200, and the remaining seven heats contain various amounts of nickel, chromium, cobalt, and silicon. The heats were air-melted in an induction furnace and cast into 8 by 8 in. ingots that were subsequently heated to 2300 and rolled to 1-in.-thick plate. Oversize 0.505-in.-diam tensile specimens were rough-machined from the plates and austenitized in air at 1700° F for half an hour and oil-quenched. The specimens were then tempered in air at temperatures of 400° to 1000 for tempering times of 1, 4, and 16 hr, and finish-machined to 0.505-in.-diam tensile specimens. Electron micrographs were prepared from Formvar replicas of the midthickness of tensile-specimen shanks that had been electrolytically polished and etched in nital. Electron-diffraction patterns were obtained from carbon-extraction replicas. The replicas were prepared by etching the sample surface in the same manner as for microscopy, evaporating carbon on the etched surface, and re-etching the sample to free the carbon film and the precipitate particles. Electron-probe microanalysis was performed on some of the extraction replicas. To obtain adequate fluorescent X-ray intensities, several thicknesses of replica were exposed to the electron beam so that the results represent an average composition of the extracted particles rather than an analysis of single precipitate particles. The retained austenite contents of the specimens were determined by an empirical X-ray diffraction

Jan 1, 1963

By H. C. Rogers

A phenomenological study of the failure of polycry stalline ductile metals at room temperature was carried out using light and electron microscopy. Tensile fractures as well as sections of partially fractured bars of OFHC copper in particular were examined. The initiation and growth of the central crack in the neck of a tensile specimen occurs by void formation. After the formation of the central crack the f'racture may be completed in either of two ways: by further void formation or by an "allernating slip" mechanism. The first leads to a "cup-cone" failure; the second, to a "double-cup" failure. In the past decade or decade and a half there has been a great deal of emphasis on the solution of the problem of the brittle fracture of metals, particularly those which normally exhibit considerable ductility such as steel. Since the problem of the fracture of metals after large plastic strains has less immediate commercial or defense significance, there has been considerably less effort expended in describing the details of the phenomenology and determining the mechanism of this type of fracture. The present research was undertaken to increase our knowledge in this area. The problem of ductile fracture has not been neglected completely, however. Ludwik1 first found by sectioning a necked but unbroken tensile specimen of aluminum that fracture began with a large internal crack which appeared to have started in the center of the neck. Examination of the fracture indicated that the crack had propagated radially with increasing deformation until a point was reached at which the path of the fracture suddenly left this transverse plane and proceeded at approximately 45 deg to the stress axis until the surface was reached. This gives rise to the commonly observed cup-cone tensile fracture. When MacGregor2 was attempting to demonstrate the linearity of the true stress-true strain curve from necking until fracture, he found that copper was anomalous in that the stress dropped off markedly from the straight line value before fracture occurred. Radiography indicated that in the copper an internal crack was formed long before the final fracture, the stress decreasing during the growth of this crack. One of the most significant advances in the understanding of ductile fracture was the result of work by Parker, Flanigan, and Davis.3 By the use of etch-pit orientations they were able to demonstrate conclusively that the fracture surface at the bottom of the cup, although on a gross scale normal to the tensile axis, did not consist of cleavage facets as had been previously supposed by many investigators. Recently, Forscher4 has shown evidence of porosity near the tensile fracture of hydrogenated zirconium which he attributes to hydride decomposition. The workers at the Titanium Metallurgical Laboratory5 have also shown evidence of porosity in a number of the commonly used metals after heavy deformation. Many metals have relatively low ductility during creep tests at high temperature. The fractures are intercrystalline, resulting from the nucleation and growth of grain boundary voids. The work in this area has been recently reviewed by Davies and Dennison.6 It is possible that some of the observations and conclusions may have a bearing on the present study? especially since at least two studies7,' have been extended down to room temperature and below using magnesium alloys. However, since magnesium does exhibit low-temperature cleavage, these results may not be pertinent to the present one. The use of the electron microscope as an aid to the study of fractures has been extensively exploited by Crussard and coworkers.9 The examination of direct carbon replicas of the fractures of a large number of metals and alloys showed that the bulk of the fracture surface was covered with cup-like indentations of the order of 1 to 2 µ in size. These frequently had a directionality by which Crussard claims to be able to tell the direction of the crack propagation. With this rather disconnected background of information, this investigation was undertaken in the hope of presenting a unified picture of the initiation and propagation of a fracture in a ductile metal. To this end all of the techniques previously used were employed simultaneously so that there might be a good correlation of the data obtained by different techniques. EXPERIMENTAL PROCEDURE The metal which was chosen as the starting material for this investigation was OFHC copper. Of the dozen or so materials considered, it best fulfilled the requirements of commercial availability in large sizes, good ductility, relatively high melting point compared with room temperature and

Jan 1, 1961

By J. R. Long, C. E. Armantrout, A. H. Roberson, T. R. Graham

The preparation and fabrication of copper-manganese-zinc alloys and the evaluation of their engineering properties have for some time been an integral part of a research program of the Federal Bureau of Mines. In this project, the use of electrolytic manganese, along with high purity copper and zinc, has permitted the preparation of alloys which contain a minimum of impurities which are capable of producing detectable effects in the properties studied. The initial work on the physical properties of wrought alpha solid solution alloys and a general outline of the alpha solid solution area for alloys containing up to 40 pct manganese have been presented in previous reports. 1,2,3,4,5 The evidence accumulated in these investigations clearly showed that additions of manganese to the copper-zinc alloys improve the strength properties of the brasses without seriously reducing their characteristic ductility and that considerable quantities of manganese could be added to the copper-zinc alloys without exceeding the mutual solubilities of manganese and zinc in copper. In continuation of these studies, additional data have been obtained in a more extensive investigation of the phase relationships of the copper-manganese-zinc ternary system. The present report delineates the solid phase boundaries of the ternary system from the copper-zinc binary to the manganese-zinc binary for alloys containing up to 50 pct zinc. Summary The alloys used in this investigation were conditioned by working schedules designed to homogenize the structures prior to heat treatments required for the equilibrium phase sludy. The treated alloys were examined by metallographic methods including hardness and X ray data for confirmation. The composite data of the investigation are summarily presented in a series of isothermal sections of the ternary system at 1100, 1200, 1300, 1200, and 1500°F. The salient features of the system as outlined by these data are: (1) the continuous alpha solid solution which extends across the diagram to the gamma-manganese phase of the manganese-zinc system at temperatures of 1200°F and higher, (2) the beta phase of the copper-zinc system similarly forms a continuous solid solution with its counterpart the beta field of the manganese-zinc binary, (3) at 1100°F (593°C) these alpha and beta fields are greatly restricted by complications arising from the allotropic transitions of manganese and the decreasing solubility of manganese in the alpha solid solution. The alpha + beta fields intersect the alpha + alpha manganese field to produce the beta + alpha manganese field and two adjacent three phase fields consisting of alpha + beta + alpha manganese and gamma + beta + alpha manganese. Typical microstructures representative of the various phase areas at each temperature are also presented. Review of Previous Work The copper-manganese-zinc alloys have periodically been the subject of numerous investigatioris of both prac-

Jan 1, 1950

By Vittal S. Bhandary, B. D. Cullity

Swaged iron wire has a cylindrical {001} <110> texture. The texture is also cylindrical after re-crystallization in the absence of a magnetic field, but <111> and <112> components are added to this texture when recrystallization occurs in a field. The mecizanical properties in tension and in torsion are not greatly altered by these changes in texture. AS shown in a previous paper,1 cold-worked wires of the two fcc metals copper and aluminum can be made relatively strong in torsion and weak in tension, or vice versa, by proper control of preferred orientation (texture). The deformation texture can be controlled by selection of the starting texture (texture before deformation), because certain initial orientations are stable during deformation. The present paper reports on similar work performed on bcc iron. In this case it was clear at the outset that there was no hope of controlling the deformation texture, which is one in which <110> directions are aligned parallel to the wire axis. (1t has usually been regarded as a fiber texture, but Leber2 has recently shown that it is a cylindrical texture of the type {001} <110>. In either case, <110> directions are parallel to the wire axis.) There is general agreement on this texture among a large number of investigators, which in itself suggests that the starting texture has no influence on the deformation texture. More direct evidence was produced by Barrett and Levenson,3 who reported that iron single crystals of widely varying initial orientations all had a single <110> texture when cold-worked into wire. Thus <110> is a truly stable end orientation for iron and probably for other bcc metals as well. Under these circumstances attention was directed to the possibility of controlling the recrystallization texture. This texture is normally <110> in iron,4 just like the deformation texture. However, it is conceivable that this texture could be modified by a proper choice of the time, the temperature, and what might loosely be called the "environment" of the recrystallization heat treatment. In the present work the environmental factor studied was a magnetic field. The effect of heating in a magnetic field ("magnetic annealing") on recrystallization texture has been investigated by Smoluchowski and Turner.5 They found that a magnetic field produced certain changes in the recrystallization texture of a cold-rolled Fe-Co alloy. The texture of this material is normally a mixture of three components, and the effect of the field was to increase the amount of one component at the expense of the other two. Smoluchowski and Turner concluded that the effect was due to magnetostriction. With the applied field parallel to the rolling direction, the observed effect was an increase in the amount of the texture component which had <110> parallel to the rolling direction. In the Fe-Co alloy they studied, the magnetostriction is low in the <110> direction and high in the <100> direction. Thus nuclei oriented with <110> parallel to the rolling direction will have less strain energy than those with <100> orientations and will therefore be more likely to grow. In a later paper on the same subject, Sawyer and Smoluchowski6 ascribed the effect to magneto-crystalline anisotropy and made no mention of magnetostriction. In the papers of Smoluchowski et al. the intensity of the magnetic field was not reported but it was presumably large, inasmuch as it was produced by an electromagnet. In the second paper6 it is specifically mentioned that the specimens were magnetically saturated. But if magnetostriction has a selective action on the genesis of stable nuclei during recrystallization, that selectivity must depend only on differences in magneto-strictive strains between different crystal orientations and not on the absolute values of those strains. Thus the saturated state does not necessarily produce the greatest selectivity, because the relative difference in magnetostrictive strains between different crystal directions may be larger for partially magnetized crystals than for fully saturated ones.7 In the present work the specimens were subjected to relatively weak fields (0 to 100 oe) produced by solenoids. MATERIALS AND METHODS Armco ingot iron rod (containing 0.02 pct C and 0.19 pct other impurities) was swaged from 0.25 in. in diam. to 0.05 in., a reduction in area of 96 pct. The mechanical properties in tension and torsion were measured as described previously.&apos; Textures were measured quantitatively with chromium or iron radiation and an X-ray diffractometer,8,1 and

Jan 1, 1962

By Rex B. McLellan, M. L. Rudee, T. Ishibachi

A modelfor dilute quasi-regular interstitial solid solutions is proposed in which the solute atoms can occupy both the octahedral and tetrahedral interstices in the bee solvent lattice. The distribution function, according to which the solute atoms are apportioned between the two kinds of sites, is calculated. The partial thermodynamic functions of the solution are calculated from the distribution function. It is shown that such a solution model can explain the deviation from linearity of the Arrhenius plot of the temperature variation of the diffusivity of carbon through a iron. The published difiusivity data is analyzed to obtain numerical values for the distribution function. Finally it is shown from the numerical values of the distribution function that the possibility of dual-site occupancy will have a negligible effect on the solution thermodynamics for this system. DeSPITE the large amount of experimental data available on the thermodynamics and diffusion kinetics of dilute solid solutions of carbon in a iron, many of the elementary properties of these solutions are not understood due to real or apparent discrepancies in the experimental data. Basically there is a lack of agreement between the solubility and diffusion data obtained by anelastic damping techniques and by bulk chemical thermodynamic methods. SPecifically the heat of solution of cementite in ferrite estimated from thermodynamic measurement1, 2 is about twice as high as that determined from anelastic damping measurements.3 Furthermore the diffusion coefficient D of carbon in a iron in the range -40" to about 150°C (covering about ten orders of magnitude) determined from anelastic damping techniques is such that an Arrhenius plot of In D vs 1/T is linear. However in the range of about 300o to 850°C where mass flow measurements of D have been made (covering about four orders of magnitude), although the data merges with the low-temperature data, the Arrhenius plot departs from linearity. The object of this paper is to present a more refined model of dilute ferrite solutions in which the solute atoms can occupy two kinds of interstitial sites. The experimental data is analyzed in the light of this model and it is believed that thereby some of the anomalies can be resolved. snoek4 explained the room-temperature internal-friction peak (usually referred to as the Snoek peak) in iron as the stress-induced ordering of interstitial carbon and nitrogen atoms occupying the octahedral (0, 0, 1/2) interstices in the bee lattice. From the consideration of a hard-sphere lattice model the (0, 1/4, 1/4) tetrahedral interstices are larger than the octahedrons but their symmetry would not produce the strain dipole necessary to give rise to the Snoek peak. Thus the presence of the Snoek peak is direct evidence that the octahedrons are occupied. Since the publication of Snoek&apos;s explanation, most workers have assumed that all of the carbon and nitrogen in the a solid solution was located in the octahedral sites. This assumption of universal octahedral occupancy has been implicit in the theory of wert5 for interstitial

Jan 1, 1965

By Rex B. McLellan

A simple model for dilute interslitial solid solutions is set up which enables a solubility equation for the equilibrium between the solid solution and the gaseous solute to be deduced. This solubility equation is used to estimate the partial thermody -namic functions of Ag-O and Fe-N dilute interstitial solid solutions from the measured equilibrium between these solutions and the gas phase. It is shown that the partial excess entropy of the solute atom in solution is a vibrational entropy which is due Principally to the oscillation of the interstitial atom and does not arise from electronic or elastic interaction between the solute atoms and solvent matrix. In this work measurements of the equilibrium between dilute solid interstitial solutions and a gaseous phase are used to deduce the important thermody-namic functions of the solid solution with the aid of a theoretical solubility equation formulated from a knowledge of the thermodynamic functions of the gas and a simple model for the solid solution. The quantity of principal interest which is deduced from the equilibrium data is the change in vibrational entropy occurring when a solute atom (u) is placed in an interstial site in the solvent (u) lattice. This paper forms part of a more general study in which the properties of condensed phases are deduced from measurements of the equilibrium between the condensed phase and a gas. Previous investigations have dealt with the equilibrium between krypton and liquid metals,&apos; a pure metal and its vapor,&apos; and dilute substitutional alloys and the vapor,3 and the subject has been popularly reviewed by McLellan.4 In the latter investigation3 the vapor pressures of very dilute Cu-Ag and Cu-Au alloys were measured with a Knudsen cell technique over a large range of temperatures and the analysis of the vapor-pressure data showed that the relative partial excess entropies were large and positive [ds= -Sudk = 30 for Au-Ag alloys and 150 for Cu-Au alloys; these values should be compared to unity, the value assumed for these exponentials in the quasi-chemical theory of solutions]. For many interstitial solutions experimental data is available for the equilibrium between the solution and a gaseous phase, and this can be used to deduce the vibrational entropies of the solute atoms in solution. The general method employed is to calculate the chemical potential of the solute atoms in the gas and equate this to pi, the chemical potential of solute atoms in the solid, which is deduced from a simple model set up for the dilute interstitial solution (in which the solute atoms act as bound oscillators in interstitial sites). This enables a theoretical solubility equation to be set up. In the case of Ag-O and Fe-N solid solutions the solubility c, is found to vary with temperature according to a relation of the type, where PU2 is the pressure of the O2 or Np molecules. Essentially, the theoretical solubility equation enables the parameter A(T), which gives the vibration entropy, and the parameter AH, which gives the energy of solution of a solute atom, to be estimated from the measured variation of solubility with temperature. The solutions treated are dilute enough to enable solute-solute interactions to be neglected so that the partial energy and excess entropy of solution E, and GS are not functions of composition and the configurational entropy is that of a random solution. Large positive values have been found for sxSu this quantity has not been given sufficient consideration in recent work on interstitial solutions and it is felt that the partial thermodynamic quantities of very dilute solutions should be understood before the thermodynamics of concentrated solutions can be fully appreciated. EXPERIMENTAL DATA 1) The Silver-Oxygen System. The measurements of Steacie and Johnson,= using silver foils, indicate that there is a pronounced minimum in the solubility vs temperature curve for O-Ag solutions. Eichenauer and Miiller6 have recently redetermined the solubility of oxygen in silver using a method developed in the investigation of metal-hydrogen systems.7 The solubility determinations were carried out on compact specimens of differing shapes and sizes and the results indicate a much lower solubility than was obtained by Steacie and Johnson with no minimum at 400°C. Subsidiary experiments showed that the erroneous results of Steacie and

Jan 1, 1964

By A. Kelly, S. Sato

Tensile tests have been performed on aluminum single crystals of 99.99 pct purity at 196°K. The resolved shear stress when the stress-strain curve becomes concave to the strain axis depends on orientation, being always higher for crystals with an initial orientation close to the sides of the unit triangle. Observations have been made, with the optical microscope, of the appearance of slip lines on polished surfaces of the crystals. Particular attention has been paid to the first appearance of cross-slip. It is confirmed that marked cross-slip is first seen when the stress-strain curve becomes concave to the strain axis. Measurements are reported of the lengths of the slip lines observed on top and side faces of the crystals at various stages of the deformation. The lengths of the slip lines vary inversely as the flow stress for all crystals, independent of crystal orientation. From measurements of the lengths and separations of the slip traces it is possible to estimate the number of dislocations which are annihilated at each cross-slip site. It is concluded that the annihilation of screw dislocations of opposite sign is the most important cross-slip process occurring. FRIEDEL1 recently pointed out that the stress-strain curve of many pure metals of face-centered-cubic structure can be considered as composed of three stages. Plastic flow commences with a region showing a small rate of hardening which is approximately linear. This has been called "easy glide*. It is followed by a region of more rapid hardening in which the slope of the curve is again approximately constant and this, in turn, is superseded by a final part of the curve which is concave to the strain axis. Following Diehl2 we may designate these regions as stages I, II, and III respectively. The onset of stage 111 occurs at lower stresses at higher temperatures. The appearance of the slip lines formed during these various stages has been described by a number of workers?-5* The object of the present work was to make detailed observations of the lengths and separations of slip lines for a number of crystals of different orientations deformed at the same temperature and strain rate. Although much previous work has been published on the deformation of aluminum single crystals the importance of measurements of these quantities has only recently been realized—they are essential for experimental tests of recent theories of work hardening.&apos; 9 In aluminum, deformed at room temperature, stage 11 of the stress-strain curve is very weakly marked, as the stress for the onset of stage 111 is attained soon after the end of easy glide. In the present experiments it was decided to use a temperature lower than room temperature so that stage II of the curve was better developed. However, at temperatures as low as 90°K slip lines are not measured easily with the optical microscope.5 For this reason 1960k was chosen as a suitable temperature. Even at 1960k however, stage III of the curve commences at glide strains of 6 to 10 pct and hence

Jan 1, 1960

By D. B. Hunter, H. W. Rosenberg

The titanium-rich portion of the Ti-Pd system was investigated from 0 to 75 wt pct Pd by metallo-graphic and X-ray techniques. A 0 eutectoid occurs at 24 wt pct Pd and 1190°F. Two compoutzds are indicated in the region below 75 wt pct Pd, Ti,Pd and Ti2Pd3. The solubility of palladium its a titanium is low, probably less than 1 pct. In 1960 Rudnitskii and Birunl published a complete version of the Ti-Pd phase diagram. However, their work was in disagreement with the earlier literature in that they reported only one inter metallic compound, whereas three had been reported earlier. In view of these discrepancies, it was therefore necessary to redetermine those portions of the diagram of immediate interest. The following account describes our work on the system over the range of 0 to 75 wt pct Pd. MATERIALS AND METHODS Distilled titanium sponge and elemental palladium were used in the formulation of the alloys; the chemistry of these materials is detailed in Table I. The alloys were prepared as 10 to 50 g blended compacts that were melted into buttons by arc melting under gettered argon on a water-cooled copper hearth. Weighing of the ingredients before and after melting showed that negligible weight changes occurred. Therefore, no analyses were undertaken and the compositions of all alloys are nominal. All alloys were fabricated by hot rolling at 1700°F to 0.070-in.-thick sheet. Scale was removed by sandblasting and pickling in a 5 pct HF-35 pct HNO,, balance H20 solution. For metallographic examination, specimens were mounted after heat treatment transverse to the rolling direction, ground on silicon carbide papers of increasing fineness to 600 grit, and then electro-polished using a solution containing 600 ml me-thanol, 60 ml perchloric acid, 360 ml butyl cello-solve, and 2 ml "Solvent X". Unless otherwise specified, etching of alloys containing up to 42 pct Pd was carried out by swabbing with a 12 pct HN03-1 pct HF aqueous etch where a bright etch was required, or by a 1 pct hydrofluoric in saturated oxalic acid solution where contrast between phases was required. The Ti-52.8 Pd alloy was etched with a solution of 25 ml HF, 40 ml glycerine, 35 ml methanol, and 18 g benzalkonium chloride. For X-ray examination, 1/2-in.-square speci- mens of sheet were mounted flat in a standard 1-in. metallographic mount and ground and polished as above. X-ray diffraction was performed using a Norelco type 12045 Diffractometer, employing CuKa radiation with a nickel filter at 40 kv and 20 ma. Specimens were rotated about the sheet normal during exposure. Although this procedure did not remove the effects of sheet texture from the relative intensities, it had the advantage that oxidation or contaminants entering during preparation of powder samples could not confuse the patterns obtained. RESULTS AND DISCUSSION Fig. 1 illustrates the Ti-Pd phase diagram according to Rudnitskii and Birunl with the work of the present authors superimposed. Both interpretations agree that the system is of the 0 -eutectoid type with an extensive 0-phase field, and that the eutectoid temperature is just below 1200°F. There is also agreement that the solubility of palladium in a titanium is restricted. Our work would indicate that the a solubility of palladium is low, probably less than 1 pct. However, whereas Rudnitskii and Birunl place the eutectoid composition at 41 wt pct Pd, this investigation shows it to be at about 24 wt pct Pd. Moreover, this investigation confirms the existence of compounds at Ti2Pd and Ti2Pd3, whereas Rudnitskii and Birun report only a single Berthol-lide phase covering the TiPd to TiPd, range. Laves et a1.&apos; and Wallbaum, whose work was summarized by Maykuth , reported the existence of Ti2Pd3 and TiPd, in addition to Ti2Pd. More recently, Nevitt and Downe~&apos; have reported the structure of

Jan 1, 1965

By E. P. Klier, S. M. Grymko

The transformations in eutectoidal systems have been extensively studied as they occur in steels.&apos; As a consequence of these studies the martensite, bainite and pearlite reactions found for most steels are widely recognized. Further, because of the work of Smith and Lindlief2 and of Greninger and Troiano3 these reactions, except for the bainite reaction, are considered as typical for all eutectoidal systems. The primary alloy studied by Smith and Lindlief2 was a CuAl alloy of near eutectoid composition as was the alloy more recently studied by Mack.4 The transformations in these alloys were, for the most part, studied metallo-graphically, while an X ray analysis of structures was essentially wanting. It has appeared desirable, therefore, to investigate selected CuAl alloys at and slightly away from eutectoid composition. For this purpose three alloys were prepared containing 10.5,11.9 and 13.5 pct A1 respectively. The results of a metallographic and X ray analysis of structures obtained for isothermal transformation of the ß-phase are reported here for the 11.9 and 13.5 pct A1 alloys. Introduction The phase equilibria in the CuAl alloys of immediate concern are indicated in the equilibrium diagram section presented in Fig 1.5 The equilibrium and transition structures which have been reported in this system are as follows: a = copper rich terminal phase (f.c.c.) ß = high temperature eutectoidal phase (b.c.c.) r2 = aluminum-rich intermediate phase (r-brass structure) a&apos; = modified a-phase postulated by Bollenrath and Bungnrdt.26 ß1 = ordered ß (b.c.c.) ß&apos; = martensite structure: A1 < 13 pct (near h.c.p.) ß" = ß&apos; of Smith and Lindlief2 7&apos; = martensite structure: A1 > 13.5 pct (h.c.p.); ß2 of Bollenrath and Bungardt.26 The eutectoid transformation can be suppressed at sufficiently high cooling velocities. The ß-phase is then retained to about 500°C where ordering sets in. The ordering interval is not satisfactorily known but indirectly appears to be non-suppressible. Subsequent to the ordering reaction and consequently at lower temperatures, the ordered ß1 decomposes to a marten-sitic structure. The kinetics of these reactions and in particular those of the martensite reaction have an important bearing on the results obtained in this investigation. Since some confusion exists concerning certain reported aspects of these reactions, they will be considered in some detail. THE MARTENSITE REACTION The martensitic structure is formed on cooling through a critical range6

Jan 1, 1950

By D. K. Deardorff, H. Kato

THE transformation temperature of hafnium from hcp to bcc is 1750° + 20 °C in contrast to previously published values by Duwez and fast2 which are believed inaccurate. The Bureau of Mines determined this value by measuring the temperatures at which a sudden decrease was observed in the electrical resistance of hafnium alloys containing up to 18 at. pct Zr, and then extrapolating the temperatures to 0 pct Zr. Duwez1 previously reported a transformation temperature of 1310°C, obtained by use of a rapid-cooling thermal-analysis technique. Later, Fast2 published a much higher value of 1950° 100°C based upon: a) his own determination of 1690°C for the beginning of the transformation range for hafnium containing 5.7 at. pct Zr, and b) his belief that lattice parameters reported3 for Duwez&apos; hafnium indicated i contained 19 at. pct Zr and that the 1310°C value should be associated with that composition. Russell4 subsequently stated that Duwez and Fast probably were reporting their lattice parameters in different units and reassigned a value of 6.0 at. pct Zr to Duwez&apos; hafnium. If this assumption were true, Duwez&apos; hafnium was of approximately the same purity as Fast&apos;s, but his transformation value was considerably lower. The reason for this divergence is not known. Fast&apos;s value for hafnium containing 5.7 at. pct Zr agrees well with our data. McGeary5 determined a transformation value of 1660°C by metallographic examination of heat-treated specimens, but the zirconium content of his hafnium was not definitely established. The specimens used in this investigation were 1/8-in.-diam rods 6 1/2 in. long, fabricated by hot-swaging bars cut from arc-melted buttons. These buttons were melted from crystal bar hafnium and crystal bar zirconium. The chief impurity in the hafnium

Jan 1, 1960

By M. K. Yen, W. R. Hibbard

Previous studies of plastic deformation of metals have emphasized the important role of bending and constraints during strain under relatively pure stresses.1"5 Some new phenomena such as early conjugate slip6 and polygonization7 are intimately concerned with the relief of bending stresses, the former by slip and the latter by a process analogous to re-crystallization. However, few analyses of the bending deformation of single crystals were found in the literature, and those8,9 are sufficiently previous to the modern interpretations of slip to invite further investigation in this area. The transverse bending mode of testing, using the two-point loading method, was selected because the center portion of the specimen may be considered as under a theoretically pure bending stress with a uniform bending moment. Considerable attention was given to both the lattice distortion and flow characteristics of the crystals during deformation. Experimental Proeedare A specimen under the action of equal and opposite couples at both ends is said to undergo pure bending. The shear and moment diagrams of a specimen under the two-point loading method of transverse bending are shown in Fig 1. The moment over the middle-third of the span is equal to PL/6, and the vertical shear is zero. Fig 2 (A) illustrates the stress distribution in the elastic range on a plane across the middle of the specimen. It can be seen that the normal stress has a maximum value at the extreme surfaces on the tension and compression sides, and decreases proportionately to zero as it approaches the neutral plane. However, when the plastic range is reached, the stress distribution is no longer a straight line relation, but changes in a manner which can be best illustrated as shown in Fig 2(B). The following calculation and stress analysis are mainly in accordance with the works reported by Timoshenkol0 and Kochendörfer.9 From the simple beam formula, the maximum normal stress will occur at the surface of the specimen. For a round specimen with a radius r and for a rectangular specimen with a height 2h and width 2b, the maximum normal stress, Sp, will be given by the following formulas: For round specimen: 2PL Sp = 3pr3 [1] For rectangular specimen: Sn = 8bh2 [2] Since it is reasonable to consider that slip takes place under the same conditions as in uniaxial loading, the resolved shear stress S?, along the operative slip direction will be as follows: For round specimen: 2PL sin x cos ? Sp = 3pr3 sin x cos ? [3] For rectangular specimen: PL S8 = 8bh2 sin x cos ? [4] where X is the angle between the specimen axis and the slip direction and X, the angle between the specimen axis and the major axis of the glide ellipse. It was also reported by Kochenclorfer9 that the critical bending moment, that is, the bending moment exerted on the specimen to initiate slip, should be higher than the value given by the equations stated above. Based

Jan 1, 1950

By S. Isserow

RECENTLY, a description of the wartime work in this laboratory on the U-Si phase diagram was published. This diagram was available earlier in the open literature; as were Zachariasen&apos;s crystal structure identifications for the compounds in the system. The uranium end of the phase diagram is shown in Fig. 1. The most recent work on this system has emphasized the peritectoid E phase, of particular interest for a number of reasons. The early work provided evidence of this phase&apos;s resistance to corrosion in water and of its slight ductility. Further information on these properties has been obtained here. Work elsewhere4 has provided evidence of stability under irradiation. This body-centered-tetragonal phase is relatively isotropic compared to the more usually applied a or ? uranium phases. Unlike the other compounds in this system, it contains no Si-Si covalent bonds.3 In addition, consideration on the basis of Pauling&apos;s concepts leads to the conclusion that in the E phase, uranium has a higher valence than in any of the other phases. Definition of the composition of the ? phase has been of prime concern in the study of this alloy. Evidence is presented below that its silicon content is slightly higher than that of stoichiometric U8Si, the composition frequently assigned this phase on the basis of Zachariasen&apos;s identification. This work on the composition required careful attention to the method of dissolving the alloy for analysis, since apparently low silicon contents could be found under certain conditions, see Appendix. Preparation: Melting, Casting, and Heat Treatment—Almost all the samples were from castings with a diameter of 13/4 or 2 in. and about 1 ft long, weighing 10 to 15 lb. The U-Si charge was melted by heating to about 1650°C in an induction heated vacuum furnace. Beryllia crucibles were used at first, but were replaced by zirconia-washed graphite as the technique of keeping the carbon content low enough (about 500 ppm) was mastered. The alloy was bottom-poured at about 1650°C into graphite molds. A few small buttons, weighing no more than 50 g, were prepared in a small arc melting unit.&apos; Since cast material has a preponderance of uranium and U8Si2, heat treatment is necessary to ob-tain the ? phase by the solid-solid peritectoid reaetion. Once obtained, the ? phase is stable to room temperature, obviating the need for speeial attention to the method of cooling after epsilonization, see Fig. 1. It was found desirable to epsilonize at 800°C for a week. For some purposes, in which residual uranium and/or U8Si2 are tolerable, shorter periods may be adequate. The temperature of 800°C is not critical, since essentially the same results were obtained from 785° to 825°C. Further departure from this epsilonization range is inadvisable.

Jan 1, 1958

By G. H. Kesler, C. E. Dryden, J. H. Oxley

Rates of heat- and mass-transfer from rods to recirculating air were determined within a one-quarter-scale model of a metals deposition bulb. The dependence of local and averaged rates of transfer upon outlet geometry, number of rods, position upon a rod, and airflow rate was established for flow patterns created by radial and tangential air inlets. The results are given a qualitative interpretation in terms of rates and distribution of metal deposition in a prototype deposition bulb. IT is well known that a number of metals can be prepared from volatile compounds containing them by causing dissociation of their compounds to occur at heated surfaces. Examples of such metals include silicon, titanium, zirconium, and aluminum, which can be prepared from their halides. While the ability to prepare high-purity metals by this technique is of importance in itself, the rate of deposition of metal also is important in considering the commercial potential of such a process. If the over-all process of deposition is broken down into separate steps of transport of materials to and from the heated deposition surface and establishment of chemical equilibrium at the surface, as was described in a recent paper,1 then since surface temperatures are usually high and surface equilibrium is attained very rapidly, the rate-controlling step can be considered as either that of transport of reactants to the surface or of transport of products other than deposited metal away from the surface. These two transport steps are related through the reaction stoichiometry. It can be shown1 that the rate and distribution of metal in such a transport-controlled process are determined by the over-all and local mass-transfer coefficients in conjunction with the equilibrium conversions at the hot surfaces. The rate relationship is: w = [KA(y a-y s)] B.E where the bracketed group is the rate of arrival of the reactant at the surface of area A as determined by the mole fraction driving force (y a - y s) and the convective transport constant K. The factor B is the weight fraction of metal in the reactant, and E is the equilibrium fractional conversion to metal. Numerical values of y, and y, are fixed by the vapor composition and the equilibrium compositim of reactant at the surface; and E can be determined by thermodynamic calculations or by experiment. However, values for the convective coefficient K are available generally only for systems of simple geometry, such as cross-flow past a rod or sphere or parallel flow past a flat plate or through a tube. Often it is necessary to work with systems of more complex geometry and in which the vapor flow may not be uniform or exactly parallel or perpendicular to the surface. It is possible, of course, to establish working correlations for these more complex systems by direct experiment; but the experiments usually are costly in both time and money. To circumvent this objection and to aid in arriving at an understanding of the relationships and interactions of the process variables, it is helpful to construct a model of the flow system in which air, water, or other convenient fluids can be substituted for the process vapors and with which mass-transfer measurements or analogous he at-transfer measurements can be made upon a simulated deposition surface. By this means, a large number of measurements can be made in a reasonable time, and the correlation of these results can be applied to the prototype system. Examples of the application of these techniques are to be found in the literature (e.g., Ref. 2). It is the purpose of this paper to illustrate these techniques by description of a model study which was done at Battelle Memorial Institute. The work to be described was of a preliminary nature; the investigation subsequently was extended to develop quantitative relationships among the significant dimensionless transport, geometric, and flow parameters. This more detailed study will be the subject of a future paper.

Jan 1, 1962

By C. A. Siebert, O. S. Duffendack, A. W. Herbenar

IN metallurgical problems involving the study of equilibrium in binary systems, the ,existence of an additional vapor phase, due to the presence of a volatile component in the alloy, has often been neglected. This is well indicated by the modified form of the phase rule, usually accepted in condensed systems in which pressure is not considered as a degree of freedom. In many cases the vapor pressure of the volatile component is exceedingly small and for all purposes can be considered negligible, in which case this form of the phase rule is applicable. However, alloys containing Hg, Cd, Zn, or Mg, tend to show very marked equilibrium vapor pressure, even in the solid state, and therefore present cases where the modified form of the phase rule no longer holds true. A quantitative determination of the vapor pressure of zinc in alpha brasses is not only of practical interest because of the loss of zinc in the processing of these alloys, but also of theoretical interest in that such measurements provide a means of applying the principles of theoretical physics and physical chemistry to problems of applied metallurgy. A number of investigators have reported vapor pressure measurements of zinc in alpha brasses. Hargreaves1 made extensive measurements which were used by Birchenall and Mehl² and Guttman³ for the determination of activity coefficients of zinc in alpha brasses. Hargreaves1 used a differentially heated quartz tube and measured the condensation temperature at one end corresponding to the temperature of the heated brass at the other end. Birchenall and Cheng outlined the condensation method of obtaining vapor pressure of zinc and Cd over some of their silver alloys." The primary object of the present investigation was to increase the range and accuracy of the vapor pressure data for the alpha brasses. Preparation of Alloys: Six alloys were prepared from Bunkerhill zinc and O.F.H.C. copper, both of 99.99 pct purity. The cast ingots were annealed for 24 hr at 550°C, scalped, cold rolled 40 pct and annealed for 50 hr at 550°C. The annealed bars were then surface ground and degasified for 30 min at approximately 550°C in an induction heated vacuum furnace. The surface of the bars was removed before milling the bars into relatively fine chips for chemical analysis and samples for the equilibrium cell. The analyses of the six alloys are given in table I. Experimental Method: The vapor pressure of zinc over the various alloys at a number of temperatures was determined from absorption spectra data. The spark source was produced by means of a high tension discharge between two electrodes of pure zinc. The design of the absorption vessel is shown in fig. 1. The vessel was constructed of transparent quartz which transmits the whole of the visible spectrum and the ultraviolet down to a wave length of 2000 A. The thickness of the vapor space was set at 0.8 mm so as to permit a relatively wide range of measurements without altering the vapor space, and thereby necessitating recalibration of the vessel. The furnace used for heating the absorption vessel consisted of a split wound furnace as shown in fig. 2. The spectrograph used in obtaining the transmitted zinc spectra was a double prism type especially designed for a region of 2500-3500 A. A step diaphragm used in conjunction with a tungsten filament provided the necessary intensity patterns required for calibration of the photographic plates.

Jan 1, 1951

By C. H. Cheng, C. E. Birchenall

The fundamental problem in the thermodynamics of solid solutions is the determinatiorl or calculation of the activities of the components as a function of temperature and composition. Since the theory of metals is not suficiently developed to allow a priori calculation of these quantities, they must be obtained from experiment. C. Wagner1 has reviewed the literature of this subject for the period before 1940. Although several important and extensive studies have been made in the meantime, the number of systems to which any sort of quantitative information can be assigned is vanish-ingly small. These data have become of great potential importance in studies of the structure of solid solutions2 and intermetallic compounds, the nature of the diffusion process,3,4 and perhaps even the mechanism of mechanical deformation.5 For these reasons, it. seems very worthwhile to extend the experimental data in this field. Although there is little use in duplicating Wagner&apos;s review, attention should be directed to the most recently reported investigations. A brief enumeration of the methods of measurement previously employed also should be valuable. The most direct method is the determination of the partial pressure of the components in the vapor phase in equilibrium with the solution. Both equilibrium and kinetic methods have been tried in metal systems, but almost exclusively on liquid phases. Only in the ease of carbon in iron alloys and in the copper-zinc system have extensive measure- ments been made which include solid phases. When one of the components is an element, such as nitrogen or carbon, which forms compounds which are very volatile and stable at normal temperatures and pressures (CH4, Co2, CO NH3), it is frequently possible to equilibrate mixtures containing these (such as CH4-H2, CO2-CO, NH3-H2) independently with the elernents (C, X) and with the metal (Fe, Ni, etc.). Such measurernents have been made for carbon in iron, iron-silicon, and iron-manganese alloys by Smith6 and in iron and iron-nickel alloys by Toensing.7 Differences between carbon activity values at the same temperature and carbon content when carboil is introduced from CH4-H2 rnixtures or CO-CO. mixtures indicate that the effects of hydrogen and oxygen in the system are not negligible, and one can only hope that the true value lies somewhere between these sets of results, probably nearer the CH4-H2 data since hydrogen is less soluble in iron than oxygen. This is essentially an equilibrium method, although flowing gas is employed. A kirletic method has been employed which consists of sweeping an inert gas (generally hydrogen) over the alloy at a series of rates, condensing the metal vapor out,, and analyzing it. The metal content can be extrapolated to zero flow rate (equilibrium). Wejnarth&apos;s8 work on Cd-Mg and Zn-Mg is a good example of this frequently used method. A variant of this consists of equilibrating a known volume of inert gas with the alloy, sweeping it out quickly, and analyzing for metal. In common with the above method, it has the disadvantage that the "inert" gas is generally somewhat soluble in the alloy. Moreover, the former continuously displaces the system from equilihrium and may give low values if solid diffusion is involved. The dew point method applied by Hargreaves9 to the alpha and beta brasses and by Schneider and Stolll0 to A1-Zn avoids the introduction of another component to the system, but is useful only in systems in which one component is much more volatile than the other. The alloy sealed in one end of an evacuated silica tube is heated to the desired temperature. The temperature of the other end is lowered until droplets of the volatile component condense. When the temperature is raised slightly, the droplets will evaporate. By careful adjustment of temperature, the range between evaporation and condensation can be narrowed appreciably. An independent determination of the vapor pressure of the . pure volatile component is necessary to give the partial pressure over the alloy. This is the method employed here to determine the vapor pressures of zinc and cadmium over their silver alloys up to 34 pct in cadmium and 76 pct in zinc. The former involves only the a solid solution, but the latter covers the a, ß, y, and e fields. In recent determinations, Herbenar, Siebert, and Duffenbarkl1 used the

Jan 1, 1950

By R. S. Wagner, W. C. Ellis

A new mechanism of crystal growth involving oapor, liquid, crnd solid phases explains many observations of the effect of implurities in crystal growth from the vapor. The role of the impuuitq is to form a liquid Solution with the crystalline tnalerial to be grown from the vapor. Since the solution is n prefevred site for deposition firorti the uapor, the liquid becorrles supersaturated. Crystal growth occurs by precipitatzon from the supersaturated liquid crt tlie solid-liquid zntevfnce. A crystalline defect, such as a screw dislocation, is not essetztial for VLS (vapor -liquid-solid) growth. The concept of the VLS mechanism is discussed in detail with reference to tire controlled growth of silicon crystals using gold, platinum, palladium, nickel, silver, or copper as an implurity agent. RECENTLY a short communication&apos; described a new concept of crystal growth from the vapor, the VLS mechanism. In this paper we present a detailed description of the process and its application to the growth of silicon crystals and we discuss its relevance to existing concepts of .&apos;whisker" crystal growth. Crystal growth from the vapor is usually explained by a theory proposed by Frank2 and developed in detail by Burton, Cabrera, and Frank.3 In this theory a screw dislocation terminating at the growth surface provides a self-perpetuating step. Accommodation of atoms at the step is energetically favorable, and is possible of much lower supersatu-ration than required for two-dimensional nucleation. Crystals of a unique form resulting from aniso-tropic growth from the vapor are "whisker" or filamentary ones. Such crystals have a lengthwise dimension orders of magnitude larger than those of the cross section. For most filamentary crystals both the fast-growth direction and directions of lateral growth have small Miller indices. The special growth form for a whisker crystal implies that the tip surface of the crystal must be a preferred growth site. sears4 proposed that, according to the Frank theory. a whisker contains a screw dislocation emergent at the growing tip. Such an axial defect provides a preferred growth site and accounts for unidirectional growth. The hypothesis was extended by Price. Vermilyea. and Webb," still implying the presence of a dislocation at the whisker tip. They postulated that impurities arriving at the fast-growing tip face become buried while those arriving on the surface of slow-growing lateral faces accumulate and thereby hinder growth. These considerations led to a whisker morphology. There is increasing evidence that most whisker crystals grown from the vapor are dislocation-free. Webb and his coworkers6 searched for an Eshelby twist7 in zinc? cadmium, iron. copper, silver, and palladium whisker crystals. They found unequivocal evidence for an axial screw dislocation in only one element, palladium. However, not every palladium crystal examined contained a dislocation. Observations with the electron microscope have failed to show dislocations in whisker crystals of zinc, silicon.9 and one morphology of AlN.10 Since many whiskers are completely free of dislocations, an axial dislocation does not appear to be required for whisker growth of many substances. A significant advance in understanding whisker growth has been a recognition of the need for impurities. This requirement has been clearly demonstrated for copper,11 iron,13 and silicon9-1 whiskers. For silicon, detailed studies proved conclusively that certain impurities, for example, nickel or gold, are essential. Another pertinent phenomenon which has received little attention is the presence of a liquid layer or droplets on the surface of some crystals growing from the vapor. Crystals in which this has been observed include p-toluidine,14 MoO3,15 ferrites,16 and silicon carbide.&apos;" The liquid layers or globules were considered to be metastable phases, molecular complexes, or intermediate polymers originating from condensation of the vapor phase. The possibility has been suggested that the halide being reduced is condensed at the tip18 or adsorbed on the surface11 of a growing metal whisker, for example copper. The literature on whiskers discloses illustrations of rounded terminations at the tips. These appear. for example, on crystals of A12O3,19,20 sic,21 and BeO.22 For BeO, Edwards and Happel suggested that during growth of the whisker the rounded termination consisted of molten beryllium enclosed in a solid shell of BeO. A recent paper9 on the growth of silicon whiskers contains many observations pertinent to an understanding of the mechanisnl of whisker growth. These observations are summarized as follows. 1) Silicon whiskers are dislocation-free. 2) Certain impurities are essential for whisker growth. Without such impurities the silicon deposit is in the form of a film or consists of discrete polyhedral crystals.

Jan 1, 1965