Search Documents

By Richard H. Siller, Rex B. McLellan

A model for interstitial solid solutions has been considered in which a repulsive Potential exists between interstitial atoms in the solvent lattice. It has been shown that this model is consistent with the thermodynamic Properties of certain interstitial solid solutions. The model is then applied to the calculation of the variation of the diffusivity of carbon in au-tenite with composition. Although the energy levels at the saddle point, and thus the absolute values of dif-fusivity, cannot be calculated, it is shown that by taking the repulsive interactions between the diffusing atoms into account the functional form of the dependence of diffusivity on solute concentration can be derived. ThE composition dependence of the diffusivity, D, of carbon in austenite at a given temperature is well-known.'-3 smith3 tried to reconcile the observed composition dependence of D at 1000°C with the previously observed4 composition dependence of carbon activity using equations for D derived by arken5 and Fisher, Hollomon, and Turnbull,6 but was unsuccessful. In this paper we shall modify the approach of Fisher et al. by considering the "blocking" model7 for interstitial solid solutions, based on concept due to Speiser and Spretnak,' in which a repulsive potential exists between interstitial solute atoms in the solvent lattice, causing a certain number of interstitial sites to be "blocked" from occupancy by further solute atoms. We shall thus be able to describe the intimate relationship between the composition dependence of carbon diffusivity, a kinetic phenomenon, and the composition dependence of carbon activity in austenite. THEORY Consider for convenience one-dimensional interstitial diffusion in an fcc lattice in the [loo] direction. Fig. 1. Assume that solute atoms occupy octahedral sites only.7,9'10 These octahedral sites are located on {lOO} planes a/2Å apart. A solute atom, A, occupying an octahedral site on a {100 } plane denoted by 2, has twelve nearest-neighbor octahedral sites: four on plane 1, four on plane 2, and four on plane 3. Diffusion occurs when there is a net flux of solute atoms, through an imaginary plane shown with dashed lines, from plane 2 to plane 3. Fisher et al6. use the concept of activated complexes, as given in the theory of absolute reaction rates." They show that the number of solute atoms per cubic centimeter per second which pc — Fig. 1-Diagram showing the arrangement of octahedral sites (x) in the fcc lattice. pass in the direction of diffusion from activated to new interstitial position is: ?where C is the concentration of the solute in atoms per cubic centimeter, l/m + is the fraction of new interstitial positions corresponding to motion in the forward diffusion direction (transmission coefficient), ?F* is the free-energy difference between the activated complex and the reactants when each is in its standard state at the temperature of the reaction (independent of temperature and composition in the case of carbon in austenite), y is the activity coefficient of the solute species, and, finally, ym is the activity coefficient of the activated complex (assumed constant). The flux of atoms moving in the forward direction through an area of 1 sq cm perpendicular to the direction of diffusion is therefore: m+ Ym l ' atoms per second where is the distance between planes. The blocking model states that a nearest-neighbor octahedral site of a given solute atom has a probability, f, of being inaccessible to other solute atoms. The number of available sites for diffusion on a given plane will, therefore, depend on f, which is constant, and Nu, the number of solute atoms on that plane. The effect is to have more available sites in the diffusion direction, thereby increasing the net flux of solute atoms. This is reflected in the transmission coefficient, l/m. To calculate the transmission coefficient for solute

Jan 1, 1970

By G. W. Toop

In a recent publication,31 Darken has derived an equation to describe thermodynamic behavior in ternary metallic solutions with compositions near pure component 1: Eq. [I] is understood to be a general equation, consistent with the ternary Gibbs-Duhem equation, which is proposed as a logical formalism in view of observed behavior in some binary metallic solutions31'32 rich In Eq. and am are constants obtainable from the 1-2 and 1-3 binaries. The parameter 023 may be determined from measurements on the ternary The purpose of this discussion is to compare Eq. [I) or [3] with other ternary formulations involving binary terms, within the regions of validity of Eqs. [I], [2], and [3]. First, the ternary point is considered to be at the intersection of two composition paths, one with constant N1 and another with constant N2/N3 as shown in Fig. 10. The terms in Eq. [3] may be related to the binary points shown by multiplying the first term on the right-hand side of Eq. [3] by N2 (1) /l -N1, the second term by the third term by and the fourth term by each multiplier being unity: It should be noted here that no suggestion has been made that either the binary or the ternary systems form regular solutions. However, Eq. [5] and hence Eq. [3] shows that the contribution made to the ternary excess free energy from the 1-2 and 1-3 binaries has the same form as for a regular ternary system.1 Before considering the term a23 for the above case, a second geometrical relationship is examined with the ternary point at the intersect ion of three composition paths with constant N1/N2, N2/N3, and Nl/N3, respectively, as shown in Fig. 11. For this geometry, ternary values of NZ and N3 in terms of the binary values shown may be substituted into Eq. [I]. Then by making use of identities of the form Eq. [6] contains terms that are similar to those in a ternary regular solution formulation34 based on the geometry shown in Fig. 11. Although at finite concentrations, unless the 1-2 and 1-3 binaries form

Jan 1, 1968

By W. F. Flanagan, K. R. Evans

The strong temperature dependence of the flow stress of bcc metals has prompted considerable interest and controversy concerning the rate-controlling influence of the Peierls' mechanism upon deformation in metals.18"21 Dorn and Rajnak" recently offered a criterion for Peierls' deformation utilizing a line-energy model of a dislocation to estimate the saddle-point activation energy, for the nucleation of a pair of kinks as a function of applied stress and the height and shape of the peierls9- barrier. It was found that Un/2Uk, where 2Uk is the excess energy of a kink pair, was a unique function of where is the effective stress and 7p is the Peierls' stress as defined in Fig. 20. The relationship found by Dorn and Rajnak between is given by the solid line of Fig. 21. may be expressed in terms of the temperature of test, T, and the athermal temperature, T,, defined in Fig. 20. Considering the strain-rate, 9, of deformation for a thermally activated process to be given by where fo is nearly a constant, it is seen that and the theoretical relationship of Fig. 21 may be examined with readily obtainable experimental parameters. Dorn and Coworkers analyzed a large amount of experimental data in this manner for materials sus-

Jan 1, 1967

By J. V. Gluck

In the present paper," as in earlier publications, V1'2"1 the authors present experimentally obtained relations for the free energy of solution of oxygen in various metals as a function of temperature. The relations are linear and given in slope-intercept form. The authors give values for the "probable error" of the slope and intercept, citing Birge's method17 for obtaining these quantities. Birge discussed these methods as early as the period 1929-32.25726 In Ref. 25 the probable error is defined as "the numerical value which the true error is as likely to exceed as not", i.e., 50 pct probability. Ref. 26 is even more detailed and includes a lengthy discussion of the applicability of the Gaussian distribution to experimental data. However, the use of the "probable error" has come under some question. For example, Ezekiel'~ states that "the use of the probable error tends to make the results appear rather more accurate than the standard error". Moroney28 stated (in 1951) that "the term 'probable error' was used in former times to denote 0.67450 (50 pct confidence using the normal distribution)....the term, a poor one, is now very much obsolescent. It has been said that 'it is neither an error nor probable' ". Bennett and Franklin in 1954" in discussing statistical confidence levels mentioned that "the customary use of the 5 and 1 pct levels is not completely a matter of chance; these levels probably represent a balance between the desire to be correct and the desire to be specific. Also in most people's minds the borderline between something that could possibly happen and something that is not likely to happen is somewhere in this region. Certainly 5 or 1 pct confidence limits are to be preferred to limits such as the 'probable error' which has only a 50/50 chance of including the correct value." It may readily be seen that the 95 pct confidence level, which is given by 1.960 (assuming large sample size), thus corresponds to approximately three times the "probable error". Indeed, limits of one standard deviation correspond only to about 68 pct probability. Even if we triple the tolerances given by the authors in the equations of their Table 111, the deviations do not become excessive and would not detract from their excellent results. It could then be said with a high degree of confidence (95%) that the precision of the slope and intercept values is about *3-5 percent of their absolute levels rather than the 1-2 percent implied in Table 111. This is then compatible with an earlier suggestion of Gokcen~' of a probable accuracy of 5 pct in dissolved oxygen measurements by the technique employed. Alternatively, the authors could have

Jan 1, 1967

By John Chipman

The concentration of a solute in a dilute ),zetallic solution may be measured by any of several parame- ters including weight percent, atom fraction, atom ratio, and lattice ratio. The ratio of filled to unfilled interstitial sites is useful for interstitial solutes. A variable 2 proportional to this ratio is used as a measuve of concentration. For component 2 irz a bitzary solution z2 = n2/Ym - nz/b) where b is the numberber of interstitial sites per lattice atom. For a t~lul-ticortzporzent solution this becomes zz = n2/(nl + Cvjnj) in which Vj = - l/b for an interstial solute and +1 for a substitulional solute. In the infinitely dilute solution the activity of an interstitial solute 2 is proportional lo z2. At finile concentration the departure from this limiting law is expressed us an activity coefficient, his coefficient is a function of concentra1io)z expressed as tevactiolz coeffcient 8; is analogous to the jark~iliar e£ bul is found to be independent of concentvation in certain solutions for which data are available. It is found that the same equations may be used to express the activity of a nonmetallic solute, sulfur, in liquid solutions of iron containing other solutes, both metallic and nonmetallic. For a nonmetallic solute or for one which strongly increases the actiuity of sulfur, it is convenient to assign arbitvarily a value vj = — 1. When this is done the derivative is found to be constant in each of the ternary solutions studied. The activity coefficient of sulfur in a complex liquid iron solution may be expressed as where nk is a second-order cross product determined in the quaternary solution Fe-S-j-k. The equation is used to calculate tlze activity of sulfur i)z three sevetl- component solutions. IN thermodynamic calculations concerning dilute solutions it is unnecessary to invoke laws and relations which extend across the concentration range to include concentrated solutions. In most binary metallic systems, as arkeen&apos; has recently pointed out, there exist two terminal composition regions of relatively simple behavior, connected by a central region of much greater complexity. When the solute is a nonmetal there is only one such region and in many systems the concentration range is extremely limited. It is the purpose of this paper to suggest a method for the calculation of activities in such a terminal region in which one or more solutes are dissolved in a single solvent of predominantly high concentration. HENRY&apos;S LAW In the usual textbook statement of Henry&apos;s law, concentration is stated in mole fraction. This has the advantage that it makes Henry&apos;s law thermodynamically consistent with Raoult&apos;s law. Since all measures of concentration at infinite dilution are related by simple proportion it follows that mole fraction, molality, atom ratio, weight percent, or any other unit of concentration can be used with the appropriate constant. At finite concentrations, however, calculations based on the law depend upon the unit employed. Deviations from Henry&apos;s law at finite concentrations depend upon the composition variable employed. They are evaluated in terms of activity and interaction coefficients2 which have become familiar features of metallurgical thermodynamics. It is the purpose of this paper to propose a measure of concentration for metallic solutions containing interstitial or nonmetallic solutes by means of which the calculation of activities in complex solutions may be simplified. The discussion will be restricted to free-energy interaction coefficients3 typified by Wagner&apos;s c|a BINARY SOLUTIONS The several measures of concentration which are to be considered are shown in line a of Table I. The corresponding activity coefficients are in line b and the deviation coefficients, sometimes called self-interaction coefficients, are in line c. Henry&apos;s law simply states that the activity coefficient approaches a constant value at infinite dilution. By adoptihg the infinitely dilute solution as the reference state and defining the "Henrian" activity as equal to the concentration in this state, the activity coefficient is always unity at infinite dilution. This convention is far sim~ler and more useful in dilute solution than emploiment of the &apos;Raoultian" activities and it will be used in the following discussion. The several definitions and equations of Table I will be referred to by means of their coordinates in the table. Early observations of deviations from Henry&apos;s law in metallic solutions were shown graphically4 rather than analytically. For the case of sulfur in liquid iron5 the slope of a plot of logfs vs (%S) was constant in the range 0 to 4.8 pct S, indicating constancy of eh2&apos; in Ic. He was proposed by wagnerz and has been widely adopted. The a function of IIIc recently employed by ~arkenl was designed specifically for dilute solutions. Darken has shown that the value of a12 remains essentially constant for many binary solutions within a substantial range of compositions. The atom ratio is directly proportional to the molalitv.<, a conventional measure of concentration. IVb and C served as the basis for smith&apos;s6 classic studies of

Jan 1, 1968

By Klaus Schwerdtfeger

In order to test the performance of ZrO, (CaO) and Tho,(Yz03) electrolytes in the electrochemical determination of oxygen activities in liquid metals at steel-rrzakitng temperatures , electromotive force measurements have been made with the following cells: The oxygen-- activities are calculated from the electmrnotive forces for the investigated systems. The standard free energy of- formation of MnO and the activity of manganese in Fe-Mn melts are derived. Indications are that the Zr 0z(Ca0) electrolyte exhibits little electronic conductivity at oxygen pressures as low as 10&apos;a atnz at 1600°C; however, the partial electronic conductivity may become noticeable at lower oxygen activities. The Tho,(Yz03) electrolyte is found to be an ionic conductor at the oxygen pressure as low as that for the Mn + MnO equilibrium, but exhibits partial electronic conductivity at oxygen pressures higher than 10"1 atm (1500°C). fi. better control of oxidation and deoxidation reactions in steelmaking may be facilitated if the oxygen activity in the steel is known during various stages of the process. In the present work oxygen activities have been measured in liquid iron, manganese, and iron alloys with galvanic cells using ZrOz(CaO) and s solid electrolytes with the object of establishing the range of oxygen pressure in which these materials can be used as electrolytes at steelmaking temperatures. Calcia-stabilized zirconia and Tho2-Y203 solid solutions are oxygen ion conductors. Such electrolytes were first used by Kiukkola and Wagner1 and Peters and ~ann&apos; in galvanic cells for the determination of thermodynamicquantities of oxide systems in the temperature range 700" to 1200°C. At these moderate temperatures calcia-stabilized zirconia has been found to exhibit predominantly ionic conductivity at oxygen pressures as low as those for the Cr + Crz03 equilibrium,394 but at lower oxygen pressures elec- tronic contribution to conductivity becomes significants4 The Th02-Y203 solid solutions are ionic conductors at oxygen pressures lower than those applicable to zirconia base electrolytes.5&apos;6 At high oxygen pressures Thoz base electrolytes exhibit significant p-type semiconduction. At the time the present experiments were carried out several papers1&apos;-* were published on the electrochemical determination of oxygen activities in liquid iron using the ZrOz(CaO) electrolyte. The results of different authors do not agree well. For instance, according to Fischer and ~ckermann" stabilized zirconia can be used to measure oxygen contents in liquid iron larger than 0.02 wt pct, but according to Baker and west1&apos; ZrOz(CaO) cannot be used for the determination of oxygen content in liquid iron because of high electronic contribution to the conductivity. Fig. 1 is a schematic diagram of the apparatus used. About 500 g of liquid metal is contained in a crucible into which a stabilized zirconia or Th02(Y203) tube is immersed. A platinum contact wire is pressed (with a spring mechanism) against the inside bottom wall of the electrolyte tube. The tube is flushed continuously with a stream of air, CO-CO, (or Hz-C02) (at 0.96 atm and at a rate of about 300 cu cm per min), via an alumina tube inserted inside the electrolyte tube. This alumina tube also contains the Pt-Pt/10 pct Rh thermocouple. A capillary alumina tube shielding the other contact wire is immersed in the melt. The molybdenum contact wires were used mostly; in some cases platinum wire was employed. Molybdenum has the advantage that it dissolves more slowly in the melt than platinum. The whole cell was contained in the alumina reaction tube of a resistance furnace with molybdenum winding. The reaction tube was closed gas tight at both ends and flushed continuously with argon. Because of the poor thermal shock resistance of the electrolyte tubes the cell had to be assembled in the cold furnace. The electrolyte tube and alumina sheath

Jan 1, 1968

By F. S. Pettit

The oxidation of Ni-3 to 25 wt pd Al alloys has been studied in 0. 1 atm of oxygen at temperatures between 900° and 1300°C. These alloys have been found to oxidize by three different mechanisms which depend on the temperature of- oxidation and the alloy composition. Two of the three mechanisms do not permit a continuous layer of Al,0, to be formed on the alloy surface and the oxidation rates are greater than that for pure nickel. The third mechanism results in the formation of an external A1203 scale and lke oxidation rates are about three orders of magnitude smaller than those for pure nickel. The minimum amount of aluminum required for the formation of external scales of A L,0, has been determined. NICKEL-base alloys are currently the main source of materials for use at elevated temperatures in gas turbine engines. These alloys are usually coated to obtain oxidation resistance. Coatings on nickel-base alloys are frequently formed by reaction of the alloy with aluminum whereby alloyed nickel aluminides are formed. The alloyed nickel aluminides provide protection to the nickel base alloy because external scales of A1203 (alumina) are formed during oxidation and mass transport through A120, is slow in comparison to mass transport through most other oxides. In view of the protective properties of A1203, it is important to know how much aluminum is required in these alloys in order to form external scales of AlzO,. The present paper is concerned with the oxidation kinetics and the oxidation mechanisms of Ni-A1 alloys and the minimum amount of aluminum required in these alloys for the formation of external scales of Alz03. THEORETICAL CONSIDERATIONS When a Ni-A1 alloy is heated in oxygen at elevated temperatures, the following reactions can take place on the surface of the alloy where the oxide phases are assumed to be virtually pure: These oxide phases are in the form of nuclei scattered over the surface of the alloy and, in view of their rapid formation, they need not be in equilibrium with the alloy. As the oxidation process continues, equilibrium between the alloy surface and the oxide phases is approached and the stability of the oxide nuclei is determined by the composition of the alloy at the alloy/oxide interface because of the following reactions: 3NiA1204 + 2A1 (alloy) = 4AlzO3 + 3Ni (alloy) [41 4Ni0 + 2Al (alloy) = NiA120, + 3Ni (alloy) [ 5 1 Application of the mass-action law to Eqs. [4J and [5J yields the following equilibrium conditions for these reactions: where aA1 and aNi are the activities of aluminum and nickel, are the standard free energies of formation of NiO, A1203, and NA1204, respectively, R is the gas constant, and T is the absolute temperature. If the composition of the alloy at the alloy/oxide interface is such that (akl/ahi) is greater than the equilibrium values defined by Eqs. [6] and [7], then Reactions (41 and [5] will go to the right as written. Conversely, if the alloy composition is such that the activity ratio (aLl,/aki) at the alloy/oxide interface is less than the equilibrium values, then Reactions [4] and [5] will proceed to the left. The equilibrium activity ratios in Eqs. [6] and 171 can be calculated since values for the standard free energies of formation of the oxide phases are available. Standard free energies of formation for NiO and A1203 have been tabulated by Elliott and ~leiser.&apos; The standard free energy of formation for NiA1204 can be obtained from the data of Tretjakow and Schmalzried.&apos; The results of these calculations are tabulated in Table I. Table I shows that the following inequality is valid over the temperature interval 900" to 1300°C: (Reaction [5]) (Reaction [4]) « 1and therefore the aluminum activity for these compositions can be taken as equal to the square root of the activity ratios (i.e., aNi = 1). If equilibrium is estab-

Jan 1, 1968

By E. P. Sadowski, L. P. Trudeau, C. R. Cupp

The apparent high radiation resistance of two varieties of maraging steel is described and an indication is given of some phenomena that require further study. Two aspects were included in this work: a) effects of neutron irradiation on the room-ternperature tensile properties of 18 Ni (250) maraging steel; b) effects of neutron irradiation on the room-temperature tensile properties and on the impact resistance of 12 pct Ni-5pct Cr-3 pct Mo maraging steel. Subsize tensile and impact test specimens were used. The 18 pct Ni steel was irradiated to a neutron fluence of 3.3 x 10 n per sq cm (>1 mev). The 12-5-3 variety was irradiated to two neutron fluence levels: 1) 1.5 x 10 per sq CTTL (>I mev) 2) 3.1 x 10&apos;9nper sq cm (>1 mev) In all respects both varieties showed unusually high resistance to change in strength and ductility after neutron irradiation. There was no noteuorthy over-aging in fully aged specimens. The aged 18 pct Ni variety showed a 4 pct drop in yield strength upon irradiation with little or no decrease in ductility. The aged 12-5-3 variety showed a 5 pct increase in yield strength on irradiation with a very slight decrease in ductility. The impact transition of the 12-5-3 variety only increased 35 to 45 Co upon irradiation, and the shear energy fracture value for the subsized specimens decreased only half as much as expected frorn experience with conventional steels. THE development of new reactor concepts and nuclear technology now depends to a large extent on the availability of new alloys capable of operating in stringent reactor environments. One of the most important new types of metallurgical material that has been developed in recent years is the maraging steels. This type of alloy is notable for its remarkable retention of toughness at high strength levels. In the course of a research program which incorporated ambient temperature radiation effects studies to determine the mechanisms of strengthening of 18 pct nickel maraging steel,&apos; it became evident that the resistance of the alloy to neutron-induced changes in mechanical properties should be studied. This paper is to present evidence for the apparent high radiation resistance of maraging steel. The experimental program was in two parts: a) Effect of neutron irradiation on the room-temperature tensile properties of 18 Ni (250) maraging steel. b) Effect of neutron irradiation on the room-temperature tensile properties and on the impact resistance of 12 pct Ni-5 pct Cr-3 pct Mo maraging steel, a cobalt-free alloy. EXPERIMENTAL PROCEDURE 18 Ni (250) Maraging Steel. Tensile test specimens were machined to the dimensions shown in Fig. 1 from $-in.-diam hot-rolled bar of the composition shown in Table I. All specimens were austenitized for 1 hr at 1500°F (815°C) in argon and water-quenched. Half of the specimens were then mar aged for 3 hr at 900° F (482°C) in argon. The specimens were irradiated in a fast neutron transformer rod at approximately 120°F (50°C) to a neutron fluence of 3.3 x 10 10 n per sq cm (>1 mev). The irradiation was performed in the NRX reactor of Atomic Energy of Canada, Ltd., at Chalk River, Ontario, Canada. In all cases, the thermal neutron flux was three times as high as the reported (> 1 mev) fluxes. The room-temperature tensile tests were performed with an Instron tensile testing machine operating at a crosshead speed of 0.05 in. per min. The loading frame of the machine, mounted in a hot cell, was remotely controlled. The specimens were tested either

Jan 1, 1968

By G. S. Cole

The temperature has been measured ahead of stationary solid-liquid interfaces under conditions approximating luzidirectional heat flow and therefore unidirectional solidification. Natural convection flow patterns may be deduced from the temperature distributions , temperature fluctuations, and shape of the interface. Fluid flow increases with the height and the rate of heat transfev through the interface and this is further manifested by a deviation of the interface from aflat vertical plane. The influence of fluid flow on solute inhomogeneity during alloy crystal growth can be inferred from the observed temperature distributions. A buoyancy force exists in the liquid ahead of a vertical solid-liquid interface, caused by a difference in density between cold fluid near the interface and warmer fluid in the bulk liquid. When the viscous and inertia forces in the melt exceed this buoyancy force, a flow of fluid takes place, termed natural or free thermal convection. Natural convection in purely liquid systems has been extensively studied for many years. 1"u On the other hand, fluid flow during horizontal crystal growth has oniy recently been the subject of experimental investigation."-25 The requirements for horizontal crystal growth differ from other heat flow systems. The small aspect ratio (ratio of height of cold wall to length of fluid) has never been considered. The uniform furnace gradient which supplies heat radially (and not necessarily symmetrically) differs from previous boundary conditions of uniform heat flux at the cold and hot ends. And most important of all an isothermal s/l interface is present which can adjust its shape and position to conform to heat and fluid flow. All of these boundary conditions involve complexities which cannot be readily solved analytically. Preliminary observation has demonstrated the penera1 shape of the natural convective flow pattern in transparent media.20&apos;24&apos;25 The flow is circulatory, directed toward the interface at the surface of the liquid, down and away from the interface at the bottom, and then up at the hot end of the melt. During crystal growth such a flow may interact with the solute boundary layer at the s/l interface to affect solute incorporation.M&apos;1B|28&apos;i!T Evidence has also been presented recently to show that thermal convective flow will affect the structure of ca~tin~s.~~-~~ The rate of heat transfer (conduction plus convection) in a given fluid system is a function of the temperature difference between the hot end of the liquid and the s/l interface and the height of the interface. At the lowest values of these variables* all heat is *It has been shown&apos; that adverse temperature gradients as low as O.OOS°C oer cm are sufficient to cause convection. transferred by conduction. When the temperature difference or interface height are increased, laminar fluid flow commences and heat transport takes place by laminar convection as well as by conduction; turbulent heat transfer takes place at higher values of these variables. In the transition region between laminar and turbulent flow, boundary layer separation takes place; fluctuations in temperature are also noted and increase in amplitude and frequency as turbulence becomes dominant. In this paper, fluid flow patterns in the melt ahead of a stationary interface are deduced from observations of temperature distribution and fluctuations, heat flow rate, and interface shape. The fluid flow ahead of advancing interfaces and the effect of such flow on solute incorporation may be inferred from these measurements on stationary interfaces. Observations during enforced fluid motion will also be considered. EXPERIMENTAL PROCEDURE The metal was contained in a lava boat 10 cm in length, 1+ cm in width, and 2 cm in height, as shown in Fig. 1. A water-cooled, molybdenum block heat-sink is at one end of the boat; surrounding this assembly is a slotted stainless-steel tube noninductively wound with a nichrome heating element. Temperature was measured by 38-gage Chrome1 vs Alumel thermocouples sheathed in 0.05-cm-diam graphite-coated stainless-steel tubing, which were moved longitudinally and vertically through the melt by means of a two-dimensional motorized micrometer stage. In some experiments the thermocouple junction was exposed, but in the majority of experiments the junction was welded to the sheath tip; no significant difference

Jan 1, 1968

By G. R. Purdy, J. A. Malcolm

The isothermal dendritic precipitation of y in p CuZnSn is compared with the isothermal formation of Widmanstatten a plates in p CuZn. Although both precipitates are evidently bounded by semicoherent interfaces, the dendritic shape of y evolves through smooth "sinuoidal" shape perturbations, while a-p interfaces tend to be strictly planar, or rnicroscopically faceted. An estimate of the average inter-facial free energy of cube-oriented y in p through the Mullins-Sekerka analysis indicates that it is low (of the order of 50 ergs per sq cm). Interface mication is apparently controlled by volunle dij&apos;usion, and the observed anisotropy in growth probably results from anisotropy of inter facial free energy. Studies of the grain boundary precipitation of y in p indicate that the precipitate can grow in both $ grains, but becomes unstable with respect to the grain with which it shares a mobile semicoherent interface. Facets observed at the tips of Kurdjumov-Sacks-oriented Wid-manstatten a plates correspond to variants of an irrational hahit, near (231)0. Comparison with Hu and Smith&apos;s determination of relative interfacial free energies of a-(3 boundaries suggests that this system possesses an extremely deep o cup, which results in an ej&apos;ective low-o zone. The possible effects of such a cup on the stability of a-p interfaces are discussed, a needles, grown during rapid continuous cooling, possess microscopically smooth interfaces, and are characterized by a high degree of&apos; damage in the region of the boundaries. These boundaries, induced to move on tempering at 200"C, become faceted as equilibrium is approached, which seems to indicate that boundary structure and mobility in this system are dependent on the local degvee of departure from equilibrium MULLINS and sekerkal have elucidated the phase boundary stabilizing effect of surface tension. a factor not taken into account in previous treatments of morphological development. Their analysis applies to the isothermal growth of a pure spherical precipitate (planar in the limit of infinite radius) in a slightly supersaturated solution, under conditions of inter-facial local equilibrium and isotropy of interfacial free energy. More recently, several other factors which might tend to stabilize planar interfaces have received attention. These are: i) interface diffusion;4"8 and ii) slow interfacial reaction. cahn7 included a slight anisotropy in interfacial free energy, and showed that it has no stabilizing effect, but forces shape perturbations to reflect the anisotropy. shewmon4 noted that dendritic morphologies are rarely observed in solid-solid systems, and, in particular, postulated that slow interfacial reaction occurs at all boundaries between proeutectoid ferrite and austenite, except those which propagate parallel to {11 l)y planes (for Kurdjumov-Sachs [K-S] related phases). Previously, Aaronson~ had proposed interfacial reaction at the broad faces of Widmanstatten plates, and expected that the lateral motion of high-mobility incoherent ledge edges would produce planar precipitates bounded by law-mobility semi-coherent interfaces. The problem of interfacial reaction during the dif-fusional formation of precipitates has often been ignored, although it is apparent that a small component of chemical reaction must exist at any moving interface." Further, all solutions of the diffusion equations for boundary conditions corresponding to isothermal growth predict infinite growth rate at time zero, so that elements of interfacial reaction would be most easily detected at early times. However, reported deviations from diffusion control are comparatively rare," and the local-equilibrium diffusion-control model appears adequate for the description of many isothermal growth systems.&apos;2"9 Cahn14 has treated the problem of migration of an interface which is able to reach positional energy minima, and concluded that such an interface, if subjected to a high local ther-modynamic "driving force", may migrate normal to itself without the benefit of ledges. As local equi-

Jan 1, 1968

By Emerson F. Heald

A systematic study was made of new and old data on chemical activities in Fe-Pt alloys at elevated ternperatuves. Experimental results may be expressed in terms of the excess free energy using Least-squares analysis of the data gave the following values for the constants: for the temperature range 1130° to 1350°C, and tentatively to 155O°C, B = -3.326564 and C = 0.221051; for the temperature range 650" to 850°C, B = -2.555690, C = 1.762735, and D =0.097196. In the study of iron-containing silicate systems, it is sometimes desirable to have a direct experimental measure of the activity of iron in the system. The well-known solubility of iron in platinum, often a headache in experimental work on iron compounds under reducing conditions, can be used to advantage in this respect. If the activity of iron in an Fe-Pt alloy in equilibrium with the silicate is known as a function of the composition of the alloy, chemical analysis of the alloy will give a knowledge of the activity of iron in all of the phases in the system. The present study was undertaken in order to elucidate the characteristics of Fe-Pt alloys as iron activity indicators. This work is intended to tie together some previous work, which may be summarized as follows. Larson and Chipman1 determined the activity of iron in Fe-Pt alloys at 1550°C by equilibrating platinum metal with calcium oxide-iron oxide-silica melts of known iron activity. Compositions of the resulting alloys were determined by chemical analysis. A similar study was carried out by Taylor and ~uan,&apos; who worked at 1300°C. They brought the Fe-Pt alloys into equilibrium with iron oxide under conditions of known partial pressure of oxygen, and thus, from the work of Darken and ~urr~,~ conditions of known iron activity. Compositions were determined indirectly, by following the change in weight of the sample. Sundaresen et el* used the electromotive force of cells in which the alloy formed one electrode in order to measure the activity of iron in the alloy at 650" and 850°C. These temperatures were chosen to be above and below the first-order phase transition which takes place upon the ordering of Fe3Pt and the second-order transition which occurs upon the ordering of FePt3. EXPERIMENTAL 1) High Temperatures. The starting materials used were thin platinum foil, about 0.002 mm thick, and Fisher Certified reagent ferric oxide, Fez03, which had been heated for 24 hr at 1000°C. An intimate mixture of 80-mesh Fez03 and platinum platelets was placed in a thin platinum foil envelope. The latter was suspended from thin platinum wires in the hot zone of a vertical-tube, platinum-wound furnace of the type described by Muan and ~sborn.~ A capillary gas mixer similar to that used by Darken and Gurry3 was used to prepare a precisely known mixture of carbon dioxide and hydrogen, which was allowed to flow upward through the furnace tube. The partial pressure of oxygen in contact with the sample was thereby fixed at a value which was calculated from the charts prepared by porter? Temperatures were measured with a Pt-10 pct Rh-in-platinum thermocouple, which was calibrated using the melting points of gold (1062 .@C) and diopside, CaMgSizOB (1391.5"C). Temperature control was maintained to within i3"C with a Geophysical Laboratory proportional controller, using the furnace resistance as the sensing element. Samples were quenched by passing a small current through the platinum suspension wires, allowing the sample to drop into a bath of dibutyl phthalate at the bottom of the furnace tube. Prior to chemical analysis the samples were washed with acetone and dried. 2) Chemical Analysis. It proved possible, in almost all cases, to separate the Pt-Fe platelets physically from particles of iron oxide. The platelets were dissolved in a small volume of aqua regia, evaporated to dryness, and redissolved to 0.1 M HC1. In order to determine iron in the platinum alloy potentiometrically, it is necessary first to remove the platinum. A 10-cm column of Amberlite IR-120 cation exchange resin in the hydrogen form provided separation quickly and quantitatively: The mixture of iron and platinum in 0.1 M HC1 was added to the top of the column, and washed with about 100 ml of 0.1 M HC1. Under these conditions, the iron, principally in the form of cations such as FeC1" and FeCl;, is held quantitatively in the uppermost centimeter of the column. The platinum, in the form of anions such as PtC&- , is washed through without being adsorbed. After a qualitative test with stannous chloride indicated all of the platinum was removed, the iron was

Jan 1, 1968

By K. Mukherjee

A quasi-thermodynamic calculation is presented for estirt~ating vacancy Parameters in fcc metals. For fcc metals the equilibrium vacancy concentration at the melting point can be approximated by a constant mole fraction. The Gibbs free energy of formation of the vacancies can then be expressed as a simple function of the absolute melting temperature alone. From this description the formation energy Ef, the vibrational entropy Sv, and the formation volume of a vacancy can be calculated in terms of known thermodynamic parameters. The calculated values are in good agree-rnent with available experimental results. IT appears that many of the physical parameters associated with lattice defects can be empirically related to thermodynamic melting parameters.&apos;-7 Besides the implicit correlations between melting phenomenon and point defects, the melting temperature itself at atmospheric pressure can be used as a reference temperature for the purpose of comparison of vacancy concentrations in a given crystal class. The concentration at the melting point then corresponds to the maximum vacancy super saturation in the crystalline state at a fixed pressure. For pure fcc metals such as Au, Cu, Al, Ag, Pb, and Pt the mole fraction of vacancies at the melting point8 is of the order of 10*"3 to 10"1. The equation expressing the equilibrium vacancy concentration at the melting point contains three parameters: the formation energy, the vibrational entropy, and the absolute melting temperature, see Eq. [2]. Since these three parameters cover a wide range of values8 in the above-mentioned metals, it can be assumed that in other fcc metals, for which no reliable data are available, the mole fraction of vacancies at the melting point most likely lies in the range 10~3 to lO"4. This is then the basis of the calculations described in this article. Within this range the mixing entropy can be linearized and the vibrational entropy, the formation energy, and the vacancy formation volume can be calculated in terms of known thermodynamic constants. 1} LINEARIZATION OF MIXING ENTROPY The Helmholtz free energy of a crystal containing n vacancies at the melting point can be writteng as where Ef, T,, k, and S, are the formation energy of a vacancy, the absolute melting temperature, Boltz-mann&apos;s constant, and the vacancy formation entropy, respectively. After applying Stirling&apos;s approximation in Eq. [l] and minimizing with respect to n where (n/N), is the equilibrium vacancy concentration in mole fraction at the melting temperature at 1 atm pressure. If instead of setting (6~/6n) = 0 immediately after Stirling&apos;s approximation in Eq. [I], the configurational partition function is simplified, then some useful and significant results can be obtained by utilizing the experimental inforrnation After applying Stirling&apos;s approximation and rearranging is plotted as a function of (n/N) in Fig. 1. The points, not all of which are shown, are fitted by a straight line substituting from Eq. [5] into Eq. [I] and minimizing with respect to n combining Eqs. [6] and [2] It must be emphasized that Eq. [7] is only a crude approximation based on an oversimplified model of

Jan 1, 1967

By C. Y. Cheng

THE purpose of this note is to show that the dislocation mechanism controlling the creep of Fe-4 pet Si alloy1 may equally account for the behavior of a-Fe2 over the same temperature range. A recent study of the creep behavior of a-Fe has been made by Ishida et al.2 Their results were in agreement with the mechanism of nonconservative motion of jogged screw dislocations over the entire range of temperatures investigated, from 480° to 1045°K. However, the observed creep rate was much higher than that which could be accounted for by the above model unless the authors assumed that pipe diffusion became significant below 775°K. More recently, Cheng et al.1 inves-gated the creep of Fe-4 pet Si alloy over about the same temperature range and found that the results could be correlated fairly well with a dislocation climb mechanism as represented by Eq. [l]. Proceeding as shown previously,1 if creep is controlled by a single climb mechanism, at intermediate stresses, the shear creep rate, y, can be represented by where n and K are material constants {K is independent of the temperature but dependent on the structure, that is, K decreases during the primary stage and reaches a steady-state value over the secondary stage of creep), t is the applied shear stress, G is the shear modulus, gc is the free energy of activation for creep (usually equal to that for self-diffusion, gd), and RT has the usual meaning. By solving for y, the apparent activation energy for creep, Q which can be determined experimentally, is given by where hc is the enthalpy of activation for creep which usually is also equal to that for self-diffusion, h^. By using Eq. [2], the data in Fig. 2 of Ref. 2 can be re-plotted as shown in Fig. 1. By comparing the two figures, it is seen that the slight increase in the apparent activation for creep over Region II, as noted in Fig. 2 of Ref. 2, is due to the influence of temperature on the shear modulus rather than of the stress on the

Jan 1, 1969

By Richard J. Fruehan

ThE thermodynamic activities of nickel and chromium in liquid Ni-Cr alloys have not been previously reported. There have been several investigations on the solid alloys.1-3 In the present investigation the activity of chromium in liquid Ni-Cr alloys at 1600° C was determined by the use of the following galvanic cell, The use of galvanic cells employing solid oxide electrolytes for determining the activity of metals in solid alloys was first demonstrated by Rapp and Maak4 on Cu-Ni alloys between 700° and 1000°C. Recently Schwerdtfeger5 tested the performance of the ZrO2(CaO) electrolyte at 1600°C and found it satisfactory for oxygen pressures as low as 10"12 atm. He also demonstrated the use of solid oxide electrolytes for determining the activities in liquid metal systems in measuring the activity of manganese in Fe-Mn alloys at 1500°C. Extensive investigations have been carried out in this laboratory, exploring the oxygen partial pressure lower limit at which ZrO2(CaO) electrolyte can be used satisfactorily in oxygen galvanic cells at 1600°C. In a paper to be published at a later date,9 it is demonstrated conclusively that ZrO2(CaO) exhibits insignificant electronic conductivity at oxygen pressures down to about 3 x 10"13 atm which is lower than the oxygen partial pressure in equilibrium with Cr-Cr2O3 at 1600°C. Therefore, the activity of chromium can be related to the measured reversible electromotive force of the cell in the present investigation since no electronic short circulating will occur. Fig. 1 is a schematic diagram of the cell used. The Cr-Cr2O3 electrode was made of equal weights of chromium and Cr2O3 powders which were mixed and packed into a recrystallized alumina crucible. A Pt-wire lead was embedded into the Cr-Cr2O3 compact to insure good contact. A ZrO2(CaO) crucible, supplied by the Zirconium Corp. of America, contained the Ni-Cr melt equilibrated with solid Cr2O3. The ZrO2(CaO) crucible was placed on the Cr-Cr2O3 electrode and additional Cr-Cr2O3 mixture was packed around it. Contact with the Ni-Cr alloy was made with a Cr2O3 tube, which is a weak electronic conductor. A platinum wire lead was attached to the Cr2O3 tube. The galvanic cell was lowered slowly into the hot zone of the furnace tube in order to avoid too large a thermal shock which could break the ZrO2 (CaO) crucible. The furnace was con- tinuously flushed with purified argon during the experiment. The temperature was measured with a Pt-Pt/10 pet Rh thermocouple and maintained at 1600° ± 5°C. The electromotive force was measured with a recording potentiometer. The furnace was turned off when measurements were being made in order to avoid induction effects and the reversibility of the cell was demonstrated in the usual manner. The electromotive force was measured several times over a period of approximately 1 hr. During that period the electromotive force remained constant within ± 3 pet and did not continuously drift in any one direction. After the experiment the alloys were analyzed for chromium in the usual manner. The measured electromotive forces of the cell at 1600°C, (2912°F), are listed in Table I. The activity of chromium, with solid pure chromium as the reference state, (a), was calculated using Eq. [l], , n F E I, n lna RT- U] where E is the measured electromotive force, F is the Faraday constant, and n is equal to 3, the number of equivalents per mole, (nF/RT = 18.59 volts&apos;1 at 1600°C). The resulting activities are listed in Table I. The activity of chromium relative to pure liquid chromium was also calculated. The free energy of fusion of chromium at 1600° C was calculated using

Jan 1, 1969

By J. H. Swisher

The solubility of graphite in CaO-S2O2-Al,O3 slags was measured by equilibrating slag samples with graphite crucibles and CO gas. Carbon contents as high as 2 ut pct were obtained in CaO-saturated, CaO-A1,O3 slags, and 1.3 wt pct in slags of the composition CaO.Si0,. Although the observed conditions for Sic formation were in agreement with those predicted from thermodynamic data, CaC, was found to form at a lower temperature than predicted frotn thermodynamic data. From measurements of the equilibrium carbon content as a function of CO Partial pressure, it was found that carbide ions dissolve in CaO-A12O3 melts with a valence of minus two. The carbon content increased with CaO concentration in Ca0-Al,O3 melts and increased with SiO, content along the CaO&apos;AlO3-CaOSi0 join in the ternary system. When solid CaC2 was added to CaO-A12O3 and CaO-SiO2-A12O3 slags, it was found that one of the oxides in the slag was reduced by the carbide (Al2O3 in the forrner and SiOz in the latter). In electric furnace steelmaking, a double-slag practice is frequently used to meet alloy specifications. Initially a flush slag, which is oxidizing in nature, is used to remove phosphorus and carbon from the steel bath. Later in the refining period, the flush slag is replaced by a highly reducing carbidic slag. When calcium carbide is formed in or added to a finishing slag, the slag is effective as a desulfurizing agent and also permits alloying elements such as chromium, vanadium, and tungsten to be added to the slag in the form of oxides. The oxides are readily reduced by calcium carbide, thereby minimizing the use of expensive ferroalloys. More work has been done on the thermodynamics of silicon carbide in slags than on calcium carbide. Baird and alor&apos; and Kay and alor&apos; determined the free energy of formation of Sic by measuring the partial pressure of CO in equilibrium with solid silica, silicon carbide, and graphite. Using a similar technique, they determined SiOz activities in CaO-SiOz and Ca0-Si0,-A1203 slags. Rein and chipman3 also determined the free energy of formation of Sic using slag-metal equilibrium measurements. A literature survey has uncovered only one experimental study of the behavior of CaC, in slag systems. Shanahan and cooke4 report the results of some preliminary experiments on the solubility and stability of CaC, in a CaO-A1,03 and a Ca0-Si0-A1,03 slag at a temperature of about 1500". The carbon solubility as CaC, in a slag containing 50 pct CaO and 50 pct A1203 was reported to be 0.6 pct. They also review earlier work on the binary CaO-CaC, system. A eutectic exists in this system, but various investigators disagree on the eutectic temperature and composition. eal has given an explanation for carbide furnace erruptions in terms of the thermodynamic properties of CaC,; his analysis is not based on experimental data, but on compiled data for the free energies of formation of CaC, and CO.&apos; , These data for steel-making temperatures are all extrapolated from the results of low-temperature measurements. In the experiments described in this paper, slag samples were equilibrated with graphite crucibles and with mixtures of CO and argon or with CO gas at 1 atm total pressure for measurement of the carbon solubility. Most of the work was done on Ca0-A1203 binary slags, although in some experiments CaO-SiO, and Ca0-Si0,-A1,03 slags were used. EXPERIMENTAL Slag samples of the desired composition for the solubility measurements were obtained by blending pre-fused master slags. The master slags were prepared by fusing mixtures of reagent-grade CaC03 with either A1,03 or Si0, in a graphite crucible. The master slags were crushed, then decarburized in air in a muffle furnace at 1200O C. A schematic diagram of the apparatus is shown in Fig. 1. The source of carbon for the solubility meas-

Jan 1, 1969

By David V. Ragone, James A. Craig, Richard E. Balzhiser

The Gibbs free energies of formation of UC2 and UC were measured by equilibrating two-phase mixtures of UC2 + C and UC, + UC with liquid bismuth. The measured equilibrium concentrations of uranium in the bismuth combined with available activity coefficient information yielded uranium activities in these two-phase regions. Measurements on UC2 were performed over the range 1015 to 1160 K and on UC over the range 1115° to 1165°K. The values obtained for UC2 are in excellent agveement with values determined using combustton calorimetry and low-temperature heat capacity measurements. The UC values are in fair agreement with the calorimetric values, differing by about 10 pct from the calorimetric determination. MUCH interest has been shown in the uranium carbides in recent years because of their potential as nuclear fuels. As nuclear reactor operating temperatures have been raised to increase thermal efficiency, new fuels have been required. The uranium carbides are of interest because of their high melting points, >2000°C, high thermal conductivity, and resistance to irradiation damage. The manufacture and use of uranium carbides are aided by a knowledge of their stability and reactivity under various conditions. In this regard the thermo-chemical properties of the carbides, i.e., heat capacity, heat of formation, free energy of formation, and so forth, are particularly important. A great amount of work has recently been done on the thermodynamics of the U-C system. Unfortunately, however, there has not been good agreement among the various investigations. This study of the free energies of formation of uranium monocarbide and uranium di-carbide in the temperature region 1100°-1200° K was undertaken to help clarify the situation. It should be noted that in the temperature range of this investigation, 1100°- 1200° K, the thermodynam-ically stable phase in equilibrium with graphite is U2C3, not UC2, Fig. 1. However, U2C3 is reported to be stable only under extremely low oxygen and nitrogen pressure. Although UC2 is metastable below 150O°C, it exists under most experimental conditions. It should also be noted that "UC," never exists as UC2 but with compositions near UC1.9. However, "UC2" will be used to designate the dicarbide phase except when the stoi-chiometry is necessary for clarity. METHOD The approach taken in this study of the free energies of formation of UC and UC2 was to determine the thermo- dynamic activities of uranium in the two-phase regions UC, + C and UC + UC, by equilibrating the two-phase mixtures with the liquid bismuth. When a mixture of solid UC2 + C is brought to equilibrium with liquid bismuth, the reaction that occurs is: Neither carbon nor UC, could be detected as dissolved material in the bismuth. The standard free energy for this reaction may be written: In the two-phase region UC, + C, Fig. 1, the thermo-dynamic activities of UC2 and carbon are unity. The activity of uranium in this region was determined by measuring the concentration of uranium in a liquid bismuth melt in equilibrium with the two-phase mixture. This concentration was then used in conjunction with the activity coefficient data of uranium in bismuth reported by Tien et al.&apos; to determine the activity of uranium in the two-phase region, UC, + C. All of the quantities in Eq. [2] were then known, and the AG; (uc,) was evaluated. UC + UC2. For the studies involving UC + UC,, the reaction taking place when the two-phase mixture is brought to equilibrium with liquid bismuth is: The standard free-energy change for this reaction is given by:

Jan 1, 1969

By J. D. Verhoeven

In the past 15 years a considerable amount of experimental and theoretical work has been done concerning the onset of convection in liquids as a result of interm1 density gradients. This work, which has been doue in many different fields, is reviewed here and extended slightly to give a rrlore quantitative understanding to the probletrz of conzection in liquid metal dlffusion experinletzts. In liquid metal systems the capillary reservoir technique is currently used, almost exclusively, to measure diffusion coefficients. In this technique it is necessary that the liquid be stagnant in order to avoid mixing by means of convection currents. Convective mixing may result from: 1) convection produced as a result of the initial immersion of the capillary; 2) convection produced in the region of the capillary mouth as the result of the stirring frequency used to avoid solute buildup in the reservoir near the capillary mouth; 3) convection produced during solidification as a result of the volume change; and 4) convection produced as a result of local density differences within the liquid in the capillary. The first three types of convection have been discussed elsewhere1-a and are only mentioned for completeness here. This work is concerned only with the fourth type of convection. Local density differences will arise within the liquid as a result of either a temperature gradient or a concentration gradient. It is usually, but not always, recognized by those employing the capillary reservoir technique that the top of the capillary should be kept slightly hotter than the bottom and that the light element should be made to migrate downward in order to avoid convection. In the past 15 years a considerable amount of work, both theoretical and experimental, has been done in a number of different fields which bear on this problem. This work is reviewed here and extended slightly in an effort to give a more quantitative understanding of the convective motion produced in vertical capillaries by local density differences. The Stokes-Navier equations for an incompressible fluid of constant viscosity in a gravitational field may be written as: %L + (v?)v = - ?£ + Wv - g£ [1] where F is the velocity, t the time, P the pressure, p the density, v the kinematic viscosity, g the gravitational acceleration, and k a unit vector in the vertical direction. A successful diffusion experiment requires the liquid to be motionless, and under this condition Eq. [I] becomes: where a is the thermal expansion coefficient [a =-(l/po)(dp/d)], a&apos; is a solute expansion coefficient [a&apos; = -(l/po)(dp/d)], and the solute is taken as that component which makes a&apos; a positive number. Combining with Eq. [3] the following restriction is obtained: Since there is no fixed relation between VT and VC in a binary diffusion experiment, Eq. [5] shows that the condition of fluid motionlessness requires both the temperature gradient and the concentration gradient to be vertically directed. Given this condition of a density gradient in the vertical direction only, it is obvious that, as this vertical density gradient increases from negative to positive values, the motionless liquid will eventually become unstable and convective movement will begin. The classical treatment of this type of instability problem was given by aleih&apos; in 1916 for the case of a thin fluid film of infinite horizontal extent; and a very comprehensive text has recently been written on the subject by handrasekhar.&apos; It is found that convective motion does not begin until a dimensionless number involving the density gradient exceeds a certain critical value. This dimensionless number is generally referred to as the Rayleigh number, R, and it is equal to the product of the Prandtl and Grashof numbers. For the sake of clarity a distinction will be made between two types of free convection produced by internal density gradients. In the first case a density gradient is present in the vertical direction only, and, since the convection begins only after a critical gradient is attained, this case will be called threshold convection. In the second case a horizontal density gradient is present and in this case a finite convection velocity is produced by a finite density gradient so that it will be termed thresholdless convection. Some experimentalists have performed diffusion experiments using capillaries which were placed in a horizontal or inclined position in order to avoid convection. These positions do put the small capillary dimension in the vertical direction and, consequently, they would be less prone to threshold convection than the vertical position. However, if the diffusion process produced a density variation, as it usually does, it would not be theoretically possible to avoid thresh-

Jan 1, 1969

By H. E. Cline, J. L. Walter

Grain boundary sliding and migration were studied in pure aluminum bicrystal and polycrystal samples with two-dimensional grain structure. Scratches, 50 P apart, were used for measurement of sliding and migration distanceso. Samples were deformed at constant rate at 315C and events recorded continuously on wrotion picture film. Electron micrograPhs of boundary-scratch intersections were obtained. Yield and flow stress values were measured. The sequence of sliding and migration events for a three-grain junction is described in detail. Sliding depended only on the resolved shear stress imparted to the boundary. Sliding was accowmodated by formation of shear zones in grains opposite triple points and adjacent to curved boundaries. These shear zones provided the driving force for grain boundary migration. Migration caused rumpling of the boundaries, decreasing the sliding rate. Sliding and migration generally began at the same time, occurred simultaneously and ended at the same time. In the bicrystal, sliding and migration rates were proportional. Initial sliding rules of 5 X joe cm per sec. were measured for the polycrystal and bicrystal samples. These sliding rates agree wilh the internal friction experirnents of K;. The observations seem consistent with a viscous boundary sliding nzechanism. GRAIN boundary sliding is the translation of one grain relative to its neighbor by a shear motion along their common boundary. Sliding is thought to be an important mode of deformation at elevated temperatures and at low strain rates such as prevail in creep,&apos; and perhaps in the area of superplastic behavior.2"4 Although much work has been done to investigate grain boundary sliding, the effort has not led to the identification of a mehanism. KG showed that grain boundaries in aluminum exhibit a viscous nature under very small displacements of internal friction measrements. Various dislocation mechanisms have been proposed but are without conclusive experimental support. Attempts to relate sliding to 6&apos;s viscous boundaries have been unsuccessful in that measured rates of sliding are always several orders of magnitude lower than KG&apos;S results would predict.= In bi crystals7and polycrystalsR of aluminum tested under constant load, the grain boundary sliding was found to be proportional to the total creep elongation which indicated that sliding might be controlled by deformation of the grains. Shear zones were observed to extend beyond grain boundaries at triple points to accommodate the sliding.8 Surface observations brought forth the opinion that sliding and migration occurred alternately, in sequence.&apos; Measurements of sliding at the surface have been criticized because they might not be representative of the interior of the sample. Generally speaking, it seemed that much of the previous work and knowledge was based on observations made at relatively low magnification and examination of samples after deformation had been accomplished. Thus, it was the purpose of the present study to continuously record, at high magnification, the events occurring during the deformation of pure aluminum. Samples with two-dimensional grain structures were used to simplify interpretation of the results. The sliding and migration of small areas of many samples were continuously recorded by time-lapse motion pictures. Replicas of the surface were used to provide high-resolution electron micrographs. These observations, coupled with tmsile strength data, provide sufficient information to arrive at an understanding of the phenomenon. EXPERIMENTAL PROCEDURE An ingot of 99.999 pct A1 was rolled to sheet, 0.127-cm thick. Tensile specimens, with a gage length of 0.85 cm, were machined from the sheet. Bicrystal tensile specimens, of the same dimensions, were spark cut from a large bicrystal ingot. The grain boundary was oriented at 45 deg to the tensile axis. The surfaces of the tensile samples were ground flat on fine metallographic paper and were then electropolished in a solution of 75 parts absolute alcohol and 25 parts of perchloric acid. The solution was cooled in an ice-water bath. Using a weighted sewing needle suspended from a small pivot on a precision milling machine, a grid of fine scratches, 50 p apart, was scribed on one surface of the sample. The polycrystalline samples were then annealed in hydrogen for 15 min at 350" to 400°C to produce a two-dimensional grain structure of about 0.2-cm average grain diameter which would not undergo further growth at the test temperature, 315OC. Examination of both surfaces of the samples showed that the grain boundaries were perpendicular to the surface of the polycrystal and bicrystal samples. A hot-stage tensile machine was constructed for use with an optical microscope as shown in Fig. 1. The specimen is shown mounted in the grips. The grips ride in V-ways so that the sample can be mounted without damage. The rear grip is free to slide so that when the sample expands during heating it is not put under a compressive stress. When the grips and samples are at temperature, the rear grip is locked in place by two set-screws. The other grip is connected to a synchronous drive motor which, through a worm gear and a fine-threaded rod, deforms the

Jan 1, 1969

By J. Brett, L. Seigle, L. Castleman, T. Montelbano

Impurity-induced low-temperature recrystallization of cold-worked tungsten was inuestigated with emphasis on the influence of nickel on the reaction. Palladium, nickel, aluminum, manganese, platinum, and iron greatly lower the recrystallization temperature of doped tungsten, which zs normally very high, but the recrystallization temperature of electron-be am zone-refined tungsten wzre is slightly raised by conlacl with nickel. Recrystallization can be induced at low temperature by the presence of solid nickel on the surface of doped tungsten wire, but apparently not by exposure to nickel vapors alone. Approximately 200 ppm of Ni dijjused into 10-mil wires at 1200 from a deposit of nickel on the surface produced total recrystallization, whereas more than 600 ppm of Ni could be absorbed frotn a vapor source without altering the fibrous structure of cold-worked tungsten. Once initiated, nickel-induced recrystallization required a continued source of&apos; nickel for propagation of the recrystallization front. The solubility of nickel in fibrous 10-mil W wire was approximalely 500 ppm at 1150&apos; C, and the activation energy for penetration of the recrystallization front was 52 kcal per mole. In many applications the usefulness of tungsten depends on critical control of its structure. Cold-worked tungsten with the fibrous structure developed by suitable thermo-mechanical treatment has a low, but technologically significant, ductility. It has long been known1 that traces of nickel, and perhaps other metals, are profoundly deleterious in doped tungsten, because they induce recrystallization at low temperature, which produces a brittle, equiaxed grain structure. This effect appears to be an exception to the general observation that recrystallization is impeded and the recrystallization temperature raised by the presence of impurities.2"9 Previous studies7-&apos;&apos; of the annealing and recrystallization of tungsten wire have divided the phenomenon into prior recovery stages, primary recrystallization and secondary recrystallization. The present investigation is concerned principally with primary recrystallization which is defined here as the replacement of the fibrous structure of deformed tungsten by equiaxed grains. The objective of this study was to explore the nickel-induced recrystallization reaction in tungsten and attempt to elucidate its mechanism. As well, an effort to define which other elements give rise to low-temperature-induced recrystallization was carried out. EXPERIMENTAL PROCEDURE The procedure adopted for these experiments was essentially to bring nickel and other elements into diffusive contact with cold-worked tungsten wires. The process of recrystallization was followed as a function of time and temperature by light and electron microscopic observations. First the influence of nickel on the recrystallization temperature of arc-melted, zone-refined, and variously doped tungsten wire was determined by electroplating a deposit of nickel on the surface of the wire and annealing at a variety of temperatures for 3 hr. The chemical analyses of the tungsten wires used in this investigation are given in Table I. The surface of the tungsten wire was etched with Murakami&apos;s slution&apos; and approximately 0.005 in, of Ni was deposited from a Watts-type low pH bath14 for the conditions of these experiments. Variations in plating thickness from about 0.001 to 0.005 in. had no discernible influence on the resulting structures. The wires were then annealed in an atmosphere of dry hydrogen to establish the recrystallization temperature. Concurrently, un-plated specimens were annealed to establish the recrystallization characteristics of nickel-free wire. The criterion of recrystallization was that the fibrous structure be completely replaced by equiaxed grains after a 3-hr treatment of temperature. This provided more reproducible results than use of the first recrystallized grain or a fixed proportion of re-crystallized structure as the critical observation. The structures encountered in longitudinal and transverse sections were examined by both light and electron microscopy at magnifications up to X32,000 using parlo-dion-carbon replicas shadowed with platinum for the latter method. Second, the influence of a variety of metals on the recrystallization temperature of 0.010 in. D alumina-silica doped tungsten wire, AW136-64, was determined. The elements were applied by electroplating whenever possible. Alternatively, they were vapor-plated on the tungsten wire and a greater thickness built up by coating with a dispersion of metal powder in nitrocellulose lacquer. Elements not amenable to either of these procedures were merely slurry coated on the tungsten. The recrystallization temperature was determined as above. Third, the nickel-induced recrystallization process in doped wire was studied more closely by electroplating 8 mils of Ni on 65-mil alumina-silica doped tungsten wire, AW153-NS10, and exposing the wire to temperatures of llOO°, 1200°, or 1300°C for various times in a hydrogen atmosphere. A circular recrystallization front, &apos;Onsisting of equiaxed grains, developed at the periphery Of the coated tungsten wire, and the advance of this front into the fibrous interior was studied. These experiments employed relatively coarse 0.065 in. D wire because the 0.010 in. D wire recrystallized too quickly to permit observation of the pene-

Jan 1, 1969

By Colin Baker, G. C. Smith

Investigation of the initiation and propagation of ductile failure in OFHC copper was undertaken to determine the role of nonmetallic inclusions. The effect of inclusion initiated voids on the formation of the internal cavity and the final shear separation was studied by metallographic eranzination of strained test pieces. A strain anneal technique was used to enlarge the voids under uniaxial stress conditions to elinzinate triaxial stress effects. Measurements of void size us stress and strain were made to show the point at which void im&apos;tiation begins and becomes an important factor in the deformation process. The work of separation of copper-cuprous oxide was determined to attempt to correlate the breakdown of the matrix inclusion interface with void initiation and propagation. The zloid shape and position relative to the tensile axis suggested an interface breakdown mechanisnz of initiation. Evidence is presented that shows a basic similarity between the central cavity propagation and the 45-deg shear portions of the failure. DUCTILE fracture has been studied by a number of workers1-lo and attention drawn to the importance of hard second phase particles in the initiation of the failure. Holes formed at the matrix-particle interface can elongate by plastic deformation and then subsequently expand sideways to link up and produce a major crack. This is usually observed first in the center of the macroscopically necked region of a test-piece where the hydrostatic stresses are at a maximum. As the crack spreads sideways towards the free surface of the specimen, well defined shear zones develop from the crack tip and the final separation is along a direction at approximately 45 deg to the stress axis. This shear failure may also be associated with voids formed adjacent to second phase particles. In this way a cup and cone type fracture is produced. The stage at which separation takes place between particles and the surrounding matrix has not been clearly identified. In addition, although researchers have dealt with anisotropy of tensile behavior" as a result of material fabrication variables, not much is known about the microstructural features of aniso-tropic behavior. In the present work evidence on these points is presented in relation to the behavior of copper containing second phase particles of cuprous oxide. I. MATERIALS AND PROCEDURES EMPLOYED The material used was +-in. diam or 2-in. sq cold-drawn OFHC copper bar which contained 0.6 pct by volume of cuprous oxide inclusions. These ranged in COLIN BAKER, Junior Member AIME, formerly at -mnF of Metallurgy, University of Cambridge, Cambridge, England, is presently Research Scientist Reynolds Metals Co., Richrnand, Va. G. C. SMITH, Member AIME, is Senior Lecturer, Department of Metallurgy, University of Cambridge. Manuscript submitted June 20, 1967. IMD size from approximately 1 to 6 p in length and 1 to 4 p in width. The shape was generally slightly ovoid. Tensile tests were made on specimens having a gage length of 2.5 cm and diameter of 0.643 cm. Metallographic examination was carried out by sectioning deformed and fractured specimens; in addition fracture surfaces were examined optically and with a scanning electron microscope. Some measurements of the work of separation between copper and cuprous oxide were made, using a sessile drop technique which was a modification of that used by Kingery and umenick." The best metallographic results were obtained by using a vibratory polisher, which minimized smearing of the surface. 11. RESULTS A) Initial Experiments. Specimens from the +-in. diam rod were annealed for 2 hr at 650°C in uacuo, at which temperature complete recrystallization occurred without any change in the form of the inclusion. They were then fractured at temperatures from -190" to 600°C. Cup and cone fractures were obtained at all temperatures from -196" to 400°C. With increase in temperature there was, however, a continuous increase in the extent of the central transverse area and a corresponding decrease in the shear portion of the fracture. Above 400°C, the fractures became intergranular. Sections of specimens tested below 400°C revealed extensive small voids which were always associated with inclusions. However, the voids only reached dimensions greater than the inclusion size in the region of the macroscopic neck, where they were many times longer. Lateral expansion was found only near the fracture surface of the test pieces. As observed by Puttick, the voids were either (a) triangular holes initiated in the direction of the tensile axis and elon-

Jan 1, 1969