Search Documents

By J. W. Downey, M. V. Nevitt

The phase boundaries for the 950" isothermal sections in the ternary systems Zr-Co-0 and `Zr-Ni-0 have been determined for the composition range from 50 to 100 at. pct Zr. The two systems show very similar phase relations, having no extensive solid solution phase fields. Each contains a ternary phase. These phases are apparently isostructural, but their structure has not been determined. Some aspects of the phase relations are discussed in terms of the alloying behavior of transition metals. THE work described in this paper is the outgrowth of a recent study of the occurrence of phases having the Ti2Ni-type structure (structure-type E9,) in certain ternary systems involving Ti, Zr, or Hf with another transition metal and O.", In the Zr-CO-0 and Zr-Ni-0 systems no Ti2Ni-type phases were found to occur. However, there are several interesting aspects of the phase relations in the two systems which have significance from the point of view of the alloying behavior of transition metals. The results of this investigation may also have some importance in studies of the oxidation of Zr-Co and Zr-Ni alloys. In both the Zr-Co-0 and Zr-Ni-0 systems only the 950" isothermal sections were investigated and, as a further restriction, the study was limited to the composition range from 50 to 100 at. pct Zr. A tentative Zr-Co binary diagram has been published by Larsen, Williams, and Pehin. In the composition range pertinent to the present work they report a eutectic at 980°C and 75.9 at. pct Zr, the products of which are the terminal solid solution based on /3 Zr and the compound Zr2Co, and a eutectic at 1080" and 64.8 at. pct Zr whose products are Zr,Co and ZrCo. The solid solution based on /3 Zr is shown to decompose eutectoidally at 826°C into a Zr and Zr2Co. The limits of solubility of Co in a and /3 Zr have not been established. The structure of Zr2Co is not identified in the publication just cited. Dwight has reported that ZrCo has the CsC1-type strcture. The Zr-rich portion of the Zr-Ni diagram has been determined by Hayes, Roberson, and Paasche." The phase relations are very similar to those of the ZrCo system. A eutectic reaction whose products are /3 Zr containing 2.9 at. pct Ni and Zr,Ni occurs at 961°c, and a eutectic between Zr,Ni and ZrNi is found at 985". The solid solution based on /3 Zr decomposes eutectoidally at 808°C. The solubility of Ni in a Zr is not known accurately but is believed to be very small. Smith, Kirkpatrick, Bailey, and Williams7 have found that Zr2Ni has a tetragonal structure of the A1,Cu-type and that ZrNi is orthorhombic. Domagala and McPherson8 have published a constitution diagram for the system Zr-ZrO,. At 950" their diagram indicates that the solid solution of 0 in 0 Zr is stable from 0 to 0.5 at. pct while the phase field of 0 in a Zr extends from 6 to 29 at. pct. These solubility limits were adopted in the present study and no binary Zr-0 alloys were made. No previous data on the phase diagrams of the ternary systems are known to exist. EXPERIMENTAL PROCEDURE The experimental details involved in the preparation of alloys in this laboratory by arc melting have been described in several previous papers"3 and they will not be repeated here. Information concerning the purity of the metals used is given in Table I. Oxygen was added in the form of reagent grade ZrO,. All of the cast specimens in both alloy systems were annealed in air-atmosphere tube furnaces at 950 3' for 72 hr and water quenched. The specimens were protected from oxidation by wrapping them in Mo foil and sealing them in quartz tubes that had been evacuated at room temperature to a pressure of 1 x 10B mm of Hg. The phase boundaries were determined by metallography, and identification of the phases was accomplished primarily by X-ray diffraction methods which employed a powder camera having a diameter of 114.6 mm. The diffraction techniques which are in use in this laboratory have been previously described.' An etchant that proved satisfactory for most of the alloys consisted of 5 pct by vol of AgNO

Jan 1, 1962

By C. Wagner

When an alloy solidifies and the composition of the solid differs from the composition of the liquid, atoms of the alloying elements rejected at the solid-liquid interface have to diffuse toward the bulk liquid. Diffusion may be supported by convection. Theoretical calculations have been made for a liquid without convection, for a liquid involving natural convection, and for solidification of a liquid alloy at the surface of a rotating disk as an example of forced convection. WHEN an alloy solidifies, the equilibrium composition of the originating solid phase differs in general from the composition of the liquid phase. Thus, segregation may take place, i.e., the composition of the solid phase formed at different times differs from the bulk composition. The degree of segregation depends on several factors such as the partition ratio of the alloying elements between the liquid and the solid phase, the rate of solidification, and convection. To clarify, a theoretical analysis has been worked out. The following presuppositions are made: 1—Calculations are confined to the solidification of binary alloys in which the concentration of component A (solvent) is much greater than that of component B (solute). 2—The equilibrium concentration of the alloying element B is assumed to be lower in the solid phase than in the coexisting liquid phase. 3—At the solid-liquid interface, partition equilibrium is assumed. 4—The dependence of the partition ratio of the solute between the liquid and the solid phase on concentration and temperature is disregarded since only small concentrations of the solute are considered. 5—The concentration of the alloying element B in the solid phase is supposed to be lower than the saturation concentration for the formation of a second solid phase. 6—diffusion in the solid phase is disregarded, since diffusion rates in the solid state are comparatively low. When the partition ratio of the solute between the liquid and the solid phase is greater than unity, a part of the solute is rejected at the solid-liquid inter- face and has to diffuse toward the bulk liquid. Thus there is necessarily a concentration difference in the liquid between the solid-liquid interface and the bulk liquid as has been pointed out by Hayes and Chipman,' Rutter and Chalmers,' Tiller, Jackson, Rutter, and Chalmers, and others (see Fig. 1). In view of the concentration gradient in the liquid phase, the ratio of the concentration of the solute in the solid phase to that in the bulk liquid differs from the equilibrium ratio, although at the solid-liquid interface, partition equilibrium is assumed. Under certain limiting conditions, the solid may even have virtually the composition of the bulk liquid as is shown below. General Equations Let u be the linear crystallization velocity in cm per sec, D the diffusion coefficient of component B in the liquid phase in cm2 sec-', c',, and c", the concentrations of component B in the liquid and the solid phase, respectively, in mol per cm", and s the distance from the solid-liquid interface toward the bulk liquid at a given time t. The partition ratio is assumed to be equal to a partition constant K. Thus. The difference in the specific volume between the liquid and the solid phase may be disregarded. In view of the discontinuity of the concentration of the solute at the interface according to Eq. 1, u(c', — c",), , , = tic', (S = 0)(1 — K) gram-atoms B per unit area per unit time are rejected at the interface and have to diffuse toward the bulk liquid. From Fick's first law it follows that uc',,(s = 0) (1 ~K) == - D(dc',,/?>s). _-„. [2] The concentration profile in the liquid near the interface depends on the interplay of diffusion and convection as is discussed in the following sections. A schematic concentration profile is shown in Fig. 1. The "effective thickness of the diffusion boundary

Jan 1, 1955

By D. R. de Fontaine

By computing the values of the resolved shear stress for a great many orientations of the tensile axis on all 12 (111) <110> slip systems, Taylor and Elam&apos; were able to map out a stereogram of slip systems, given here, Fig. 1, in Diehl&apos;s notation (see, for example Hauser and Jackson&apos;). The purpose of the present note is to derive such a stereogram by mathematical reasoning and to give a simple method of defining the system with highest resolved shear stress for any direction of the tensile axis, thereby eliminating tedious computation of the Schmid factor. Let the tensile axis, of direction cosines [xyz], make the angles 0 and relative to the slip plane normal [HKL] and the slip direction [hkl] respectively. Fig. 2 shows a simple cubic cell whose edges define a Cartesian coordinate system. Let A be a unit vector of origin 0 supported by the tensile axis. Its end point [xyz] then lies on a sphere of unit radius, on the positive hemisphere of which the planes z = 0, x = y, y = z limit the so called primitive triangle of a standard [loo] stereographic projection. For any point of this triangle, the following hold: 0«2 «y *sx « 1 [l] Now, for given [xpz], the most highly stressed system will be the one for which the following expression of the Schmid factor has the largest possible value: (HKL) belonging to the form (111) and [hkl] to the form <110>, the denominator on the right hand side of Eq. [2] is equal to 1.fi.a. The unknowns H, ... h, . . . can now be determined by correctly distributing +1 and -1 to H, K, L, and +1, -1,O to h,k, 1 giving the numerator of Eq. [2] its greatest positive value (the angles 0 and cpare chosen acute) subject to the condition: Hh+Kk+Ll = 0 [3] since the slip direction must lie in the slip plane. Condition [3] can be satisfied only by having one of its terms vanish and the other two equal to +1 and -1. Taking the case of the tensile axis lying in the primitive triangle, x2y, z and therefore Hh will be chosen equal to +1 with H = h = +1 in order to retain a maximum of + signs. There remain two alternatives: either 1 = 0 or k = 0. 1) take 2 = 0 Then Kk = -1 and the numerator of [2] becomes: Retaining a maximum of + signs gives L = k = +1 and becomes: 2) take k = 0 Then L1 = -1and, likewise, the numerator of [2] becomes: xz - z2 + xy + yz [5] But the difference [5] - [4] can be written: (y - z)(x+y +z) [6] positive or zero by virtue of [I]. If [6] is positive (y > z), then the largest possible value for the numerator of [2] for any xyz inside the primitive triangle is given by [5], i.e., for: That is to say, the active slip system corresponding to an orientat:~on of the tensile axis inside the primitive triangle is:

Jan 1, 1962

By P. A. Flinn

ALTHOUGH many physical properties of superlat-tices have been studied intensively, relatively little attention has been paid to their mechanical properties until recently. Even for the well-known transformation in 0 brass, the effect of the transformation on plastic behavior has only lately been extensively investigated.&apos;&apos;&apos; A partial analysis of dislocation structure in superlattices has been given by cottrel13 and his theory agrees well with the experimental results on CU3AU.4,5 His analysis, however, is implicitly limited to low-temperature effects, since it does not include the behavior of dislocations at temperatures high enough for diffusion, and hence climb, to occur. The work of Herman and Brown2 on the creep of 0 brass indicates that important effects occur under such conditions, and consequently, the theory should be extended to include them. Such high-temperature effects cannot be observed in Cu3Au, since it disorders below the temperature at which diffusion occurs at a reasonable rate. There are materials of the same structure, however, such as Ni3A1, which retain their order up to, or close to, the melting point. GEOMETRY OF THE COMMON SUPERLATTICES The two best-known ordered structures are the 0 brass type, B2, shown in Fig. 1, and the Cu3Au type, Ll2, shown in Fig. 2. In the B2 structure the (000) and (1/2, 1/2, 1/2) sites are different in that they are occupied by different types of atoms. The symmetry of the structure is then lowered from a body-centered cubic, A2, in which the sites are equivalent, with atoms located randomly in the sites, to simple cubic. This lowering of symmetry has the consequence that two types of regions may occur within a crystal: regions where A atoms are in the (0, 0, 0) sites, and regions where they are in (1/2, 1/2, 1/2) sites. The boundary between two such regions will contain bonds between like atoms, and be a surface of higher energy. These regions are known as antiphase domains and the boundary as antiphase boundary. In the B2 structure, boundary formed during the ordering process should disappear as a result of domain growth, since any domain growing to the boundary of two others will join one and engulf the other.&apos; In any well annealed crystal, only a single domain should exist, with antiphase boundary present only in connection with dislocations, as discussed below. The L1, structure is related to the A1 (face-centered cubic) in much the same way as the B2 is to the A2. The face-center sites are occupied by A atoms, and the corner sites by B atoms. The symmetry is again lowered to simple cubic, and antiphase domains can exist. In this case, however, there are four initially equivalent sites in the unit cell: (0,0,0), (0, 1/2, 1/2), (1/2,0, 1/2) and (1/2, 1/2, 0). In the ordered structure any one may be occupied by a B atom and the other three by A atoms. Thus four types of antiphase domains can occur in this structure. This is sufficient to permit a metastable "foam structure" to exist within a crystal.&apos;&apos;7 Antiphase boundaries along three different planes are shown in Figs. 3, 4, and 5.* *The presence of a layer of disordered material between antiphase domains, as suggested by Jones and Sykes8 does not seem likely. Such a layer would be a region of unnecessarily high energy which could readily be reduced by the formation of the type of boundary discussed here.__________________________________________________________ The notation (111) [110] means a boundary lying along a (111) plane between antiphase regions related by a 1/2 [I10] translation. It may be seen in Figs. 3 and 4 that an antiphase boundary usually results in incorrect nearest neighbors (shown connected by braces). However, as pointed out by Wilson, a boundary of the (001) [I101 type can exist

Jan 1, 1961

By David Turnbull

IN isothermal recrystallization processes, new crystals generally grow into the matrix until they impinge upon other new crystals or an external surface, at constant linear rates G. Before impingement the perceptible course of growth can be described by the equation: 1 = G(t-7) C1I where, G = dl/dt, 1 is a crystal dimension measured in a constant direction, t is the time, and 7, the nucleation period, is a positive intercept on the time axis. Fig. 1 is a schematic representation of I as a function of time for a recrystallizing grain. G is dependent upon temperature, driving energy (strain or surface energy), relative grain and boundary orientations, but is generally independent of time. The frequency of nucleation, fi, (time" volume") can be defined by the equation: N = 1/fV [2] where ? is the mean nucleation period and V is the volume of the specimen that has not recrystallized. The kinetics of primary and secondary recrystallization generally can be described satisfactorily in terms of the parameters N and G only.&apos;-" After recrystallization is complete the average grain size 7 increases with time by "normal grain growth;" didt, the average rate of grain growth, is strongly time dependent and has not yet been precisely related to G for the motion of the individual grain boundaries constituting the system. It has been suggested4* " that the elementary act in grain boundary migration is closely related to the elementary act in grain boundary self-diffusion. Although the distance of atom movement in the two processes may be somewhat different, there is reason to expect that the activated states may be very similar, so that the free energy of activation for grain boundary migration should be of the same order of magnitude as for grain boundary self-diffusion. Therefore, it is desirable to develop a satisfactory basis for comparing data on self-diffusion and grain boundary migration and to make such comparisons where possible. Theory The formal theory of grain boundary migration rates is analogous to the theory for the rate of growth of crystals into supercooled liquids reviewed elsewhere 6-8. Boreliuss has shown that the latter theory describes, within the theoretical uncertainty, the growth of selenium crystals into supercooled liquid selenium. Motto and more recently Smolu-chowski" have derived expressions for the rate of boundary migration in recrystallization. The treatment to be presented is similar to Mott&apos;s excepting that the formalism of the absolute reaction rate theory will be used. The atomic mobility, M, in grain boundary migration is defined by: G = -M6p/6x where p is the chemical potential per atom and x is the coordinate measured in the direction of grain boundary movement. Let AF be the free energy difference per gram atom on the two sides of the boundary and k the thickness of the boundary. For RT>>AF the potential gradient across the boundary (6p/6x) is essentially linear and it follows that: SF/8x = - aF/N\ [4] where N is Avogadro&apos;s number. According to the Nernst-Einstein equation, M is related to a diffusion coefficient, Do, for matter transport in grain boundary migration by the equation: M = Da/kT [5] Substituting eqs 4 and 5 into eq 3 gives the basic relation between Do and G: G = DoaF/\RT [6] Do values may be calculated from experimental values of G from eq 6 and directly compared with the coefficient of self-diffusion within the crystal, DL, or the grain boundary self-diffusion coefficient D,. However, a more convenient, though equivalent, basis for comparing atomic mobility in grain boundary migration and self-diffusion is through the constants of the absolute reaction rate theory. According to this theory diffusion coefficients may be written:" D = k2(kT/h) exp [-AF,/RT] 171 aFa, the free energy of activation, is related to the measured energy of activation, Q, by the equation: AFA = Q - T aSx - RT [8] where aSa is the entropy of activation. Substituting eqs 8 and 7 into eq 6 gives: G = ek(kT/h) (aF/RT) exp [(AS,,)C/R] exp C-Qc/RTI C91 where the subscript G refers to boundary migration. The relationship between the driving free energy and the free energy of activation in boundary migration is indicated schematically in Fig. 2. Experience indicates that the variation of G with temperature can be described by: G= Go exp [- Qc/RT] [10] where Go and Qc are generally temperature independent over wide ranges of temperature. Comparison of eq 9 with eq 10 gives:

Jan 1, 1952

By E. S. Machlin

The alternate processes by which solute atoms can limit the migration of grain boundaries have been considered. At the lowest solute concentrations the controlling process is "mechanical breakaway" in which the solute atoms affect the driving force and not the mobility. In the next higher range of solute concentration, solute atoms are bound to the boundary and diffuse by boundary diffusion along cusps in the boundary developed at the solute atoms. In this region the boundary velocity is controlled by the component in the direction of boundary migration of the rate at which the bound solute atoms can diffuse along the boundary. At still higher concentrations, it is predicted that boundaries break away from the binding solute atoms by thermal activation. The specific behavior Predicted is consistent with the observations of Aust and Rutter. The most striking agreement between prediction and experiment is that obtained for a range of solute concentration in which the boundary velocity is inversely proportional to the cube power of the solute concentration. MaNY models have been devised for the atomic steps in the motion of a grain boundary. McLean1 has listed some of these models and Aust and utter&apos; have evaluated these and other models. If there are

Jan 1, 1962

By J. Weertman

An explanation is advanced for the recent results of Barrett and Sherby on the high-temperature creep of fee metals. Their measurements indicate that metals with a low stacking fault energy creep at a much slower rate than metals with a high stacking fault energy. The explanation is based on the assumption that the rate of core diffusion is much smaller in partial dislocations than in perfect dislocations. An experiment is suggested to test this assumption. The explanation also requires that the jog density is small on widely split dislocations. SHERBY and Barrett1,2 recently established an important result in the field of high-temperature steady-state creep. They have shown that fee metals whose dislocations split into partial dislocations separated by a wide stacking-fault ribbon creep at a much slower rate than fee metals whose dislocations are not split appreciably. The creep rates in the two cases may differ by a factor as high as 6000. (For the purpose of comparison, the test temperature of each metal was chosen so that its rate of self-diffusion was equal to a standard value. Similarly the stress levels were such that the ratio of the stress to an appropriate elastic modulus was the same for each metal. In all cases the stresses were sufficiently low that the steady-state creep rate was proportional to the stress raised to a power.) It is the purpose of the present paper to attempt an explanation of this interesting and significant result of Sherby and Barrett. The activation energies of high-temperature creep of the metals tested by Sherby and Barrett (aluminum, nickel, copper, and silver) were equal to their self-diffusion activation energies. Hence it appears reasonable to assume that this type of creep is controlled by a dislocation-climb process, which in turn is controlled by the diffusion of vacancies to or from dislocation lines. (It will be assumed that the diffusion of interstitial atoms is unimportant in the climb process. The analysis could be repeated for the case in which interstitial atoms are important.) In prior papers374 I have treated dislocation climb by using the assumptions that core diffusion is extremely fast down a dislocation and that a dislocation always maintains an appropriate vacancy concentration in its immediate vicinity. In this

Jan 1, 1965

By A. G. Metcalfe

Observations at temperature are used to investigate the phase changes in alloys containing more than 50 pct Co and above 1000°C. The nonsuppressible transformations in cobalt above 1120°C and in the intermetallic compound Co2Cr3 are studied. The liquidus and solidus surfaces are determined. Insufficient data is obtained to complete the constitution diagram. A REVIEW of some of the earlier work on cobalt&apos; and on the binary systems with chromium2-4 and with molybdenum,5 showed that one likely source of differences might be attributed to the presence of phases at high temperatures which could not be retained by quenching. Consequently, the decision was made that only methods of examination at temperature should be used in this work. The correctness of this decision was confirmed during the course of the work by the investigation of the martensitic nature of the transformation in cobalt- and by the decomposition of the high temperature form of the phase Co2Cr3 on quenching.&apos; The use of high temperature methods, including dilatometry and thermal etching led to the postula-tion of the existence of a high temperature form of cobalt with a probable hexagonal close-packed structure, transforming from cubic cobalt over a range of temperature from 1119° to 1145°C.8 One of the important pieces of evidence for this transformation was that it was observed in alloys with chromium and molybdenum which were nonmagnetic at this temperature so that confusion with magnetostriction effects was impossible. These results were not reported in this earlier paper, so that the data given here supply further evidence for the existence of a second allotropic transformation in cobalt. High temperature observations are necessarily much slower in producing results than the examination of quenched specimens at room temperature by conventional means such as X-ray diffraction or microexamination. The investigation has only covered a limited part of the field, as a result, extending from the liquidus down to about 1050°C and from the cobalt corner up to 50 pct alloying elements. The differences between the various reported diagrams for the constitution of the Co-Mo and Co-Cr systems have already been mentioned. The best available diagram in the first system is due to Sykes and Graff&apos; which is shown in Fig. 1. Fig. 2 shows the latest Co-Cr diagram due to Elsea, Westerman, and Manning.&apos; A comparison reveals that these are incompatible because of the disagreement with regard to the number of allotropes of cobalt. Elsea, Westerman and Manning took their liquidus and solidus curves from Wever and Haschimoto.2 The temperature horizontal in the two-phase field beneath the eutectic was placed 35" to 39°C higher than in earlier work2,4 with a new interpretation of the nature of this transformation. This interpretation also required a peritectic horizontal at about 1470°C, which has been confirmed here for an alloy with 53 pct Cr. No investigation had been made of the ternary system until after this work had been terminated, when a section at 1200°C was published." In this work a ternary compound was found by the X-ray diffraction examination of quenched alloys. This compound contained less than 50 pct Co, and was thus outside the field investigated here. However, some reactions were found dilatometrically which could not be fitted on the diagram and are now believed

Jan 1, 1954

By Atmaram H. Soni

A statistical study of machinability led the writer to examine existing data in regard to a thermal conductivity-mechanical properties relationship. Various functional relationships were proposed and the multiple correlation determined by statistical methods discussed in detail in the 1iterature.4 The relationship having the greatest engineering and statistical significance is discussed below. The data for cast and wrought aluminum alloys&apos;&apos;2 have been compiled and analyzed separately. The results of these analyses are presented in Table I. The empirical equations for estimating the thermal conductivity (cal sec-&apos; cm-I °C) are: for cast aluminum alloys K = 0.584 -0.00984/ -0.00263 Hb [1] and for wrought aluminum alloys K = 0.604 -0.00424l- 0.00170 Hb [2] where 1 and Hb are the percentage elongation and Brinell hardness number. Results of a similar analysis pertaining to stainless steel3 are also given in Table I. This empirical equation for estimating thermal conductivity (Btu per sq ft per hr per °F per in.) is K = 216.19- 1.716l - 0.0997 Hb [3] The relationship between electrical resistivity and mechanical properties was also examined for aluminum alloys using the same statistical method. These results are listed in Table II. While the situation of exact multiple correlation

Jan 1, 1965

By O. D. Gonzalez, R. A. Oriani

A thermo-osmosis technique has been used to measure the heat of transport, Q* , of hydrogen and of deuterium dissolved in a iron and in nickel, and of hydrogen in Feo.6Nio.4 in the tempevature range 400° to 600°C. For all these systems, Q* is negative and has a large temperature coefficient; an isotope effect can be established for the solutes only in nickel. The magnitude of Q* is considerably larger than the activation energy, E , for migration in the case of these isotopes in a iron, so that all variants of the Wirtz model must be rejected. The phenomenological definition of Q* is developed to show that Q* is related to the mechanisms by which the activation energy is dissipated back into the lattice, and that Q*/E is a function of the ratios of the mean free paths of electrons and of phonons to the distance of jump of the diffusing atom. A correlation is shown to exist between the sign of Q* in thermal diffusion and that of the effective charge in electro migration, and may be understood as due to the large ratio of the mean free path of electrons to that of phonons. THE study of nonisothermal diffusion in the solid state is of interest because certain aspects of the mechanism of diffusion are made manifest which remain concealed though still present in the isothermal case. The thermodynamics of irreversible processes characterizes nonisothermal diffusion, known as thermal diffusion or as the Ludwig-Soret effect, by a quantity, Q*, called the heat of transport, which essentially measures the sign and the magnitude of the steady-state concentration gradient produced by the imposed temperature gradient. The problem then is to understand the quantity Q* at the atomistic level. There does not exist a theory for the atomistic interpretation of the heat of transport. There are, however, two classes of simple models which seem to afford some physical insight into Q*. Wirtz&apos;s model1, z and its subsequent generalizations partition the activation energy, E, for the migration of the atom spatially about the region of the migrating atom, so that Q* is the difference between the portion of the activation energy centered about the original site of the jumping atom and that portion about the arrival site. Clearly, the magnitude of Q* cannot on this model be larger than the activation energy for migration. The model of Oriani3 also considers the spatial distribution of E not only before but also after the atomic jump, and Q* is related to the variation of the spatial distribution of E that is produced by the jump. The simplest kind of system with which to compare the consequences of these ideas is the interstitial solid solution, since one may easily choose a temperature at which only the interstitial component moves, the solvent lattice serving as a completely satisfactory frame of reference. Thus, one avoids complications arising when both species in a binary system move. In addition, one avoids the necessity of knowing the energy for the formation of a vacancy, something which is needed for the analysis of thermal-diffusion data on metals which diffuse by the vacancy-exchange mechanism. However, the number of well-measured interstitial systems is extremely small, and, in particular, there are almost no data in the published literature for the temperature dependence of the heat of transport. If one works with solutes which are gases in the pure state at ordinary temperatures, one may use the technique of thermo-osmosis4 which permits thermal diffusion to be measured over a small temperature difference so that the temperature dependence of Q* can be more easily determined. The choice of dissolved hydrogen in iron and nickel was based on these considerations as well as on the desire to look for an isotope effect associated with localized

Jan 1, 1965

By G. D. Cody, D. S. Beers, B. Abeles

The thermal diffusivity (thermal conductivity divided by specific heat) of Armco iron has been measured over the temperature range 30º to 1025ºC. The results are in good agreement with the thermal diffusivity computed from published values of thermal conductivity and specific heat, except near the Curie point and at the a to y phase transition, where the measured diffusivity undergoes rabid variations. The differences observed are ascribed to a rapid variation in the electronic thermal conductivity near the Curie point, and a decrease in the lattice thermal conductivity, at the a to ? phase transition. A major difficulty in measuring the thermal conductivity of materials at high temperatures is the determination and elimination of radiation losses. It has been shown, however, that in measuring thermal diffusivity D (thermal conductivity/specific heat) radiation losses can be taken into account exactly.&apos; In order to test the performance of a newly designed apparatus for measuring thermal diffusivity of solids,&apos; we measured Armco iron to an accuracy of 2 pct in the temperature range 30" to 1025°C. This material was selected since both the thermal conductivity,3&apos;4 k, and specific heat,5 C, are known accurately in this temperature range and thus the measured values of D could be compared with D computed from the pub-

Jan 1, 1962

By R. M. Treco

THE thermal expansion of pure beryllium was first investigated by Hidnert and Sweeneyl in 1925 on a single cast specimen stated to be of 98.9 pct purity. A study of the coefficients of expansion by X-ray methods has recently been made by P. Gordon&apos; on powder obtained from the Brush Beryllium Co. assaying 97 wt pct beryllium with about 2 wt pct oxygen as an impurity. The present work pertains to about the highest purity metal obtainable from the Brush Beryllium Co. This metal was in the form of lumps which were remelted and cast in a high-vacuum induction furnace. Expansion data in this paper deal with beryllium extruded into bars of rectangular cross-section 4x1/2 in. and the anisotropic expansivity of a large cast single crystal. Analysis of the cast metal after extrusion indicated a beryllium assay of 99.28 pct. Impurities were 0.170 pct Fe, 0.140 pct Al, 0.014 pct Mg, 0.086 pct Si, 0.020 pct Cu, 0.020 pct Mn, 0.007 pct Ni, 0.007 pct Ca, 0.080 pct C, 0.179 pct 0, (as BeO). Part I—Extruded Metal Thermal expansion measurements were obtained with a fused quartz-tube differential type dilatom-eter3 on specimens cut from the extruded sections in longitudinal and transverse directions. Specimens were machined to a diameter of 0.250 in. and a length of 3.000 in. One sample from each direction was vacuum annealed for 1 hr at 800°C. The dilatometer tube was placed in a vertical tube furnace having a uniform temperature zone which was twice the length of the sample. Average time for a complete thermal cycle to 500°C was 18 hr. Temperatures were observed with two chromel-alumel thermocouples attached to the sample about 1 M in. apart. The accuracy of temperature measure-ment was ± 0.25°C. An argon gas atmosphere sur-rounded the specimen to prevent oxidation of the beryllium. Samples were identified as shown in table I. Results - The original plan included measuring expansion up to 1000°C, but the first two runs on sample No. 5 showed that a permanent contraction occurred from

Jan 1, 1951

By D. E. Furman

The thermal coefficients of linear expansion for several stainless steels have been determined over the temperature range from —300° to 1000°F. The steels studied include types 301, 304, 316, 347, 310 and 330. This wide selection of compositions allows an insight into the effects of austenite stability and alloy content on the expansion characteristics. THE recent increased use of materials in low temperature applications has created a field for which the properties of austenitic iron, chromium, nickel alloys are particularly suited. Tensile and impact values of stainless steels at subzero temperatures have been made available in the technical literature but, as yet, little information has been published with respect to expansion characteristics. In order to supply this need, a study was made of several of the more common compositions. Those selected were types 301, 304, 316, 347, 310 and 330 stainless steel. Both the mean and instantaneous coefficients of linear expansion were determined over the temperature range from —300°F to 1000°F. Procedure The materials used were commercial grades with the exception of type 301 which was made in a 10# high frequency furnace. The compositions are listed in table I. All steels were given an anneal at 1950°F for 30 min and water quenched prior to machining the expansion specimens. Expansion readings were taken on specimens 0.250 in. in diam by 4.000 in. long with a fused quartz tube expansion apparatus similar to that described by Hidnert and Souder.1 Temperature measurements were made with a chromel-alurnel thermocouple calibrated for use at subzero temperatures. A bath of propane was used for temperatures from —300° to —100°F and one of alcohol from —100°F to room temperature. These baths were contained in large mouthed Dewar flasks and cooled to the desired temperature levels by passing liquid nitrogen through glass cooling coils placed in the flasks. The baths were stirred constantly for temperature equalization and were heated by immersing metal rods which were replaced frequently enough to give a heating rate of approximately 150°F per hr. A resistance wound furnace was used for heating from room temperature to 1000°F and throughout this range rates of about 450°F per hr were maintained. The furnace was designed so that the temperature gradient over the specimen was within 5°F. In most instances, the expansion readings were taken during heating from —300° to +300°F at

Jan 1, 1951

By Alan T. Gorton, T. L. Joseph, Gust Bitsianes

High-temperature X-ray diffraction techniques were used to determine thermal expansion coefficients of iron and its oxides. Lattice parameters of a and iron, wiistite, magnetite, hematite, and goethite were determined and plotted as a function of temperature. From these results, the true and mean linear thermal expansion coefficients were calculated. In the case of noncubic crystals, the expansion coefficients were determined along each of the crystallographic axes. The effect of the magnetic transformation on lattice parameter was measured for a iron and magnetite at their respective curie temperatures. COEFFICIENTS of expansion for iron and its oxides were determined from the temperature dependence of their lattice parameters. From these data, the true linear coefficients were derived by means of the formula: at =[(l/ao )(dao/dt)] [1] where at is the true linear coefficient of expansion, a, is the lattice parameter in angstrom units, and all variables are referred to the temperature, t, in OC. Thus the true expansion coefficient represents the instantaneous change in lattice parameter on a unit basis and at a particular temperature. A second expansion coefficient is also often used. It is termed the mean linear coefficient and is defined in the following way: am = [ao(t) - ao(O)]/ao(0) l [21 This coefficient represents the average change in lattice parameter on a unit basis but at a specified parameter length, ao at O°C, and over a specified temperature range,1 to 0°C. Generally the true co-

Jan 1, 1965

By D. N. Williams

THE present study was undertaken with two specific objectives: measurement of the thermal-expansion coefficient of ß titanium and examination of the effect of solid-solution alloying on the thermal expansion coefficient. Four titanium alloy niaterials were available in the form of annealed 5/8-in. diam rod. Three of these, Ti-8V, Ti-20V, and Ti-6A1-4V, were prepared in the laboratory from 140 Bhn sponge. The

Jan 1, 1962

By B. A. Kulp, R. R. Reeber

Lattice parameters for the wurtzite form of&apos; CdS mere measured by powder X-ray diffraction techniques over the temperature range 26° to 1000 K&apos;. A negative thermal -expansion coefficient was determined parallel and perpendicular to the c axis below 13.5°K. Near 808°K, the curves of the a, parameter, the differential thermal expansion, and the resistivitv vs absolute temperature exhibited a reversible discontinuity. j\. direct determination of the lattice parameters of CdS over an extended temperature range is of particular use in interpretation of the effects of high pressure on the band structure of semiconductor materials. For example, Langer&apos; has studied the effects of pressure on the absorption edge of CdS from 77°K to room temperature. In order to properly interpret his results it is necessary to know the change in lattice constants as a function of temperature for the wurtzite, zincblende, and rocksalt2 modifications of this material. Thermal-expansion measurements for CdS, wurtzite phase, have previously been determined over a limited temperature range.s74&apos;5 The zincblende phases, of other compounds with similar bonding, have been investigated by Novikova et a! .&apos;&apos; Previous thermal-expansion measurements for metals with hexagonal symmetry have for the most part used bulk sample techniques for low-temperature result. Most X-ray measurements at low temperature have been on materials with cubic symmetry.&apos;jQ The use of X-ray powder-diffraction techniques offers a convenient and accurate method of measurement over a wide temperature range. The X-ray method, besides measuring unit cell size, also provides information about change in structure type and the nature of some of the stacking defects present. I) EXPERIMENTAL PROCEDURE The thermal expansion was measured by X-ray powder-diffraction techniques. Lattice parameters for the range 300" to 1000°K were determined in a 19-cm Unicam camera. From 26" to 300°K a 34-cm back-reflection camera with the Van Arkel" film arrangement was used. The sample was bonded to a copper cold finger below a helium, hydrogen, nitrogen, oxygen, dry ice, or ice brine bath in a metal cryistat. che inner bath was surrounded by another one for radiation shielding. Nitrogen was used in this outer bath for the very low-temperature runs while dry ice and acetone was used for the 195°K point. CuKa X-rays, filtered with a nickel foil, entered the cryostat through a 0.001-in.-thick aluminum window. During a typical 9-hr exposure the sample was continuously rotated through a 30-deg arc. Heat losses from the sample holder limited the minimum temperature attained to 26°K with helium in the inner dewar. Temperature calibration between 25" and 125°K was determined with an Allen Bradley 300-ohm carbon resistor. The resistor was attached to the back of the copper finger with a thin layer of silicone grease. Liquid He, HZ, N2, and O2 baths were used to calibrate the resistor. Temperature variation is estimated to be *2"K with a slightly higher uncertainty at room temperature and 264°K. The thermal expansion of high-purity germanium from 373" to 900°K was measured as a calibration for the Unicam camera. This value was compared with the known value1&apos; for the thermal expansion. A discontinuity in the thermal-expansion coefficient of the CdS a, parameter was observed. Similar discontinuities in the resistivity and differential thermal analysis at the same absolute temperature indicate that the temperature error is *5°K in the temperature range above 500°K. Temperature errors in the Unicam camera below 500°K are appreciably higher due to large gradients in the resistance furnace. II) SAMPLE PREPARATION The CdS was Eagle Picher ultrapure grade. The crystallite size range of the as-received powder

Jan 1, 1965

By R. J. Wasilewski

Axial expansion has been determined by X-ray diffraction up to 600° to 760°C in a titanium and four high-oxygen alloys. Expansion data cannot be fitted to the usual quadratic expression and anomalies observed are believed indicative of the changes in the bonding configuration of the valence electrons. Furthermore, there is significant disagreement between data obtained on poly crystalline and single crystal powder specimens. The major disagreement occurs within the temperature range (-470°C) where both mechanical and electrical properties also show markedly anomalous behavior. X. HIS work forms part of an investigation of physical properties of pure titanium and its osolid solutiono alloys with oxygen and nitrogen. In view of the anomalies observed in both the electrical resistivity&apos; and the susceptibility behavior it was decided accurately to determine the thermal expansion behavior of the pure and alloyed a titanium. PREVIOUS WORK The axial expansion coefficients reported for pure titanium are listed below. As can be seen, there is a marked discrepancy between the axial expansion and the experimentally determined coefficients on poly crystal line material. Thus, if we take the most recently reported axial expansivities3,4 we obtain perature interval of 25o to 700°C. The experimental value reported for polycrystalline metal is, however, of the order of 9.7 x 10o6 per OC. EXPERIMENTAL I) Pure Titanium. High purity titanium powder of -325 mesh was used throughout most of this work. The only significant impurities present were the interstitials, oxygen and nitrogen, which analyzed respectively 650 and 80 ppm. The purity of the ma-

Jan 1, 1962

By Nicholas J. Grant, Noboru Komatsu

Metallographic and X-ray studies were made of oxide dispersion strengthened Cu-12 vol pet SiO2 and Cu-3.5 vol pet Al2O3 alloys following time exposures at temperatures approaching the melting. point of the matrix metal. Changes in hardness, oxide particle size and crystal structure were studied to determine the time-temperature stability of these alloys. An interpretation of the mechanism of oxide agglomeration is proposed. One of the most striking features of dispersion strengthened metal-metal oxide alloys is their stability at elevated temperatures, both with respect to recrystallization and maintenance of mechanical properties after exposure at high temperatures. A small amount of work has been done on the re-crystallization phenomenon in dispersion hardened metal-metal oxide alloys&apos;l and several theoretical treatments have been advanced regarding the stability of these alloys by Lenel and Ansell,8 and by Grant and his co-workers.9, 10 This paper reports on metallographic and X-ray studies following various heat treatments of Cu-AI2D3 and Cu-SiO2 alloys at elevated temperatures and the effect of subsequent cold work on the recrystallizalion temperature. An interpretation of the recrystallization behavior of this type of alloy, based on the decomposition and agglomeration of oxides, is proposed. EXPERIMENTAL PROCEDURE Cast ingot bars of copper-1.59 pet Si and Cu-0.77 pet Al, 7/8 in in diam, were annealed in argon for 80 hr at 1000CC for homogenization. The homogenized rods were machined into fine chips and ball milled. Milled particles in the range minus 20 to plus 60 mesh were separated for internal oxidation. The internal oxidation was accomplished by the same method used by Preston and Grant,9 except that the oxidalion temperature for the Cu-Si alloy was 750°C, and for the Cu-A1 alloy was 800°C. These powders were then hydrogen-reduced at 450°C, compacted in a copper container, evacuated at room temperature to 0.04 µ, and sealed for extrusion. They were extruded at 1400°F (760°C) with a ram speed of 55 in. per min for an extrusion ratio of 15.6 to 1 to give a rod 0.38 in. diam by about 3 ft long.

Jan 1, 1962

By Y. H. Liu, R. Kiessling

The chromium, iron, and tungsten borides have been treated with ammonia at different temperatures. They are attacked, forming metal nitride and boron nitride, and the results are summarized in the tables. In the tungsten-nitrogen system a new phase has been observed, closely related to the known ß phase. THE present investigations were begun to determine whether any ternary phases exist in the transition metal-boron-nitrogen systems for the metals chromium, iron, and tungsten. Attempts to prepare such phases were made by using the borides of the metals as starting materials and ammonia as the nitriding agent. Although no new ternary phases have been found to exist, the results may be of some interest with regard to the technical use of the borides. General Procedure A steady stream of dry ammonia was passed through a silica reaction tube, the central part of which was kept at the reaction temperature while its ends were water-cooled. Weighed amounts of the borides were placed in a silica or porcelain boat and heated at different temperatures for a period of 10 to 24 hr. The boat was then rapidly removed from the reaction zone to the quenching zone, the water-cooled end of the tube, by means of a simple magnetic device. The nitrogen content of the products was determined by the increase in weight of the specimens, and the phase analysis was carried out by X-ray methods. For this analysis a Guinier-type camera, constructed at this Institute, was used with a bent, ground quartz crystal monochromator. The general reaction, which occurred in all the experiments, may be summarized as: metal boride + nitrogen -* boron nitride + metal nitride For the higher temperatures pure metal was formed instead of metal nitride. It is sometimes difficult to Discussi detect boron nitride among the reaction products by X-ray methods. Prolonged exposures of some specimens showed, however, that the strongest reflections of boron nitride were present. Additional evidence for its presence was given by the increase in weight of the specimens after nitriding. This increase was always greater than that corresponding to nitride formation only and was often in close agreement with the formula given above. Finally, a white powder, which was shown by X-ray methods to be boron nitride, was left if the reaction products after nitriding Fe,B at 448°C and FeB at 768°C were dissolved in diluted sulphuric acid. Ch rom iu m - Boron - N itrogen Five phases have been found to exist in the Cr-B system,&apos; , and all were used as starting materials. Investigations on the Cr-N system have shown the existence of two nitrides with ideal composition Cr,N and CrN.".&apos; * The results are summarized in Table I. All the borides are attacked and decomposed into chromium nitride and boron nitride, and the kind of nitride formed depends only on the temperature and not on the boride used as starting material. The lowest reaction temperature, however, is different for the different borides. Thus the 6 phase begins to react at about 735 "C, the r phase at about 800°, the S phase at about 900°, the 7, phase

Jan 1, 1952

By W. C. Ellis, M. E. Fine

WHEN certain binary Fe-Ni alloys are worked cold and then stabilized by a stress-relief anneal, their Young&apos;s moduli are nearly invariant over a substantial temperature range determined by composition and work-anneal history.&apos; In the present investigation addition of molybdenum to such alloys (besides increasing the yield strength) was found to diminish the sensitivity of the thermal coefficient of Young&apos;s modulus to composition changes. Such alloys are of interest for applications requiring 1 a metal with controlled, low temperature coefficient of Young&apos;s modulus and substantial magnetic permeability.&apos; Young&apos;s modulus ordinarily decreases with rising temperature. In ferromagnetic alloys a change in modulus on heating due to loss of ferromagnetism modifies the .temperature dependence. Far below the Curie temperature ferromagnetism alters the modulus in two ways: a—Addition of the energy of magnetization, a function of interatomic distance, changes the relation between interatomic energy and distance.&apos; This lowers the modulus in the Fe-Ni and Fe-Ni-Mo alloys of this investigation. b—An applied stress changes the ferromagnetic domain arrangement so that linear magnetostriction contributes to the strain and further lowers the modulUs.~ Since in the alloys of this investigation both effects lower the modulus, loss of ferromagnetism on heating changes the slope of the modulus-temperature curve in a positive direction. With increasing temperature the slope of the modulus-temperature curve, due to progressive loss of ferromagnetism, becomes less negative, goes through zero (the modulus is minimum), becomes positive, and resumes its normal negative value above the Curie temperature.&apos; The increase in modulus and decrease in curvature in the modulus-temperature curve occurring on work hardening have been attributed to a reduction in effect (b).&apos;. Vn the absence of ferromagnetism work hardening through introduction of residual strains would be expected to decrease the modulus.&apos; Alloy Preparation—Test Methods Four series of alloys having nominal molybdenum contents of 5, 7, 9, and 10 pct and containing 47 to 58 pct Fe (the analyzed compositions are given in Table I) were melted in air and cast as bars weighing from 2 to 6 lb. The charge consisted of electrolytic nickel, Armco iron, molybdenum (99+ pct) or ferromolybdenum (62 pct Mo), manganese, and aluminum as a deoxidizer. The bars were swaged cold to the desired diameters with intermediate anneals. Alloys in this composition range are reported" to be face-centered cubic and ferromagnetic at room temperature. The Curie temperatures" decrease with molybdenum. For example, keeping the Ni/Fe ratio at unity and increasing the molybdenum from 0 to 17 pct decreases the Curie tempera- ture from approximately 500°C to room temperature. (The 17 pct Mo alloy requires quenching to retain a single phase.~) Young&apos;s modulus at a series of temperatures from —50" to 100°C was determined by a dynamic method previously described.&apos;, " In this method Young&apos;s modulus is calculated from the resonant frequency (in this case approximately 50,000 cycles per second) of a rod sample (approximately 2 in. long) vibrating longitudinally in the first mode. The densities of the samples at room temperature, Table I, were determined by the standard method of weighing in air and reweighing in redistilled bromobenzene. The linear coefficients of expansion, Table I, needed to calculate the sample length and density at the temperature of measurement, were obtained by observing the change in length (22" to 100°C) of an 8 in. gage distance with a two microscope cathetometer. The values of density and coefficient of expansion in the few samples checked are independent of treatment to the accuracy reported. This was assumed to be general for all the samples. The modulus values are accurate to &0.05x101&apos; dynes per sq cm. However, values of thermal change in modulus are accurate to 20.005~ 10" dynes per sq cm. Young&apos;s Modulus Values at 25°C: The modulus values of O&apos;, 5, and 10 pct Mo alloys at 25 °C are plotted in Fig. 1 as functions of nickel. After working cold, the samples were either recrystallized at 950" to 1000°C or given a low temperature stabilizing anneal at 400° C. The 400°C anneal does not reduce the hardness nor cause recrystallization. Annealing at 400 °C does

Jan 1, 1952