Search Documents

By Arnt Solbu, Jomar Thonstad

The reaction between CO, and aluminum in cryolite-alumina melts in contact with aluminum has been studied by passing CO2 over the melt. In unstirred melts a homogeneous reaction between dissolved metal and dissolved CO2 was observed. In stirred melts in which convection was induced by bubbling argon through the melt, the dissolved metal apparently reacted mainly with gaseous CO2. The rate of formation of CO increased slightly with increasing depth of the melt, and it did not depend on whether CO2 was passed over or bubbled through the melt. The rate of formation of CO increased with increasing area of the metal/melt interface and with the application of anodic current to the metal. It is concluded that the dissolution of metal into the melt is the rate-determining reaction. THE current efficiency in aluminum electrolysis is determined by the rate of the recombination reaction between the anode gas and the metal: 2A1 + 3CO2—A12O3 + 3CO [1] as originally stated by Pearson and waddington.1 The occurrence of this reaction in cryolite-alumina melts in contact with aluminum was first verified experimentally by Schadinger.2 Thonstad3 has shown that the reaction may proceed further to give free carbon: 2A1 + 3CO— A12O3 + 3C [2] Normally only a few percent of the CO formed undergoes such reduction. The mechanism of these reactions has not yet been clarified. Aluminum, as well as CO,, is soluble in the melt. The solubility of aluminum in cryolite-alumina melts at around 1000°C corresponds to 75 x 10- 6 mole A1 per cu cm,4 while that of CO2 is only 3 x 10-6 mole CO, per cu cm.5 Taking into account the stoichiometry of Reaction [I], the ratio between dissolved aluminum and dissolved CO2 available for the reaction in a saturated melt is about 40. Therefore, as will be shown in the following, the reaction probably mainly occurs between gaseous COa and dissolved aluminum. The dissolved aluminum presumably consists of subvalent ions of aluminum and sodium.4'6 Since the interpretation of the present results is not dependent upon the nature of this solution, the dissolved metal will be designated solely as Al+ in the following. The reaction can then be divided into four steps: A) dissolution of metal, e.g., 2A1 + Al3 — 3A1+ [3] B) diffusion of dissolved metal through a boundary layer; C) transport of dissolved metal through the bulk of the melt; D) Reaction [1]. If dissolved CO, takes part in the reaction, three additional steps embodying the dissolution and transport of CO2 must be added. schadinger2 observed, when bubbling CO2 through the melt, that the rate of formation of CO (in the following designated rfco) did not depend on the distance from the metal surface. The results also indicate that the rate of bubbling did not affect the rfco. When passing CO, over the melt, Revazyan7 found that the loss of metal did not depend on the depth of the melt above the metal or on the flow rate of CO2, and concluded that Step A is rate-determining. In an unstirred melt, however, Gjerstad and welch8 found that the rfCo decreased with increasing depth of the melt, indicating that step C was rate-determining. It thus appears that the rate control of the process depends on the experimental conditions, particularly on the convection. In the present measurements the reaction has been studied in unstirred as well as in stirred melts. EXPERIMENTAL AND RESULTS The experiments were carried out at 1000°C in a Kanthal furnace with a 10-cm uniform temperature zone (±0.l°C). The melts were made up of "super purity" aluminum (99.998 pct), hand-picked natural cryolite, and reagent-grade alumina. In experiments where alumina crucibles were used, the alumina content in the melt was close to saturation (13.5 wt pct9); otherwise it was 4 wt pct. Pure Co2 (99.85 pct) was passed over the melt, and the exit gas was analyzed for CO2 and CO by the conventional absorption method.3 From the weighed amount of CO (as CO2) the rfco was calculated as the number of moles of CO formed per min per sq cm of the surface area of the melt. The amount of carbon formed by Reaction [2] was not determined. As already indicated the rfco is much higher than the rfC, by Reaction [2]. Since the rfC probably is proportional to the rfco, the measured rfco should then the proportional to, but slightly lower than, the total rate of Reactions [I] and 121. In general the scatter of results obtained in duplicate measurements was ±5 to 10 pct, while within a given run a precision of ±3 to 5 pct was obtained. The various crucible assemblies that were used will be described below. Measurements in Unstirred Melts. When carrying out aluminum electrolysis in small alumina crucibles. Tuset10 observed that after solidification the lower part of the electrolyte was gray and contained free metal, while the upper part near the anode was white and contained no metal. One may test for the presence of free metal by treating with dilute hydrochlorid acid.

Jan 1, 1969

By Louis A. Rosales, Kanji Ono

The anomalous behavior of hydrogen in iron is examined and a new model is proposed It is postulated that hydrogen single and di-interstitials are in equilibrium with each other. When the binding enthalpy of the di-interstitial, B, is assumed to be about 10 kcal per mole, the predicted temperature dependence of the solubility agrees reasonably well with the reported data. Effective diffusiuity is calculated assuming the di-interstitials to be immobile. Since the number of mobile single interstitials is greatly reduced at lower temperatures, the effective diffusivity is decreased. A good agreement in the temperature dependence of diffusivity between theory and experiment is provided with B - 10 kcal per mole. The predicted diffusivity is also a function of total concentration of hydrogen and this appears to be Partly responsible for the large scatter in the reported data. HYDROGEN is known to form an interstitial solid solution with iron.''' At elevated temperatures, it has been well-established that hydrogen enters iron in the dissociated form and occupies the interstices, and that diffusion of hydrogen in iron is via the interstitial mechanism. The concentration, c, permeability, P, and diffusivity, D, of hydrogen in iron have been separately determined and are consistent with each other under the steady-state condition for temperatures above approximately 400°C; that is, the relationship P = c-D holds. At lower temperatures, however, anomalous behavior of hydrogen in iron has been observed by many investigators. The results of experimental studies4-'' of the diffusion of hydrogen in iron are summarized in Fig. 1, in which the values of D are plotted against the reciprocal temperature. Although there is an increasing divergence in observed values of D with decreasing temperature, the values of D at temperatures below approximately 200°C are definitely small as compared with the extrapolation from high-temperature data. The lines shown bracket the region containing the most probable data. The observed values of c6," are plotted against the reciprocal temperature in Fig. 2, where the dashed lines represent Geller and sun's'' most probable isobars at 1 and 100 atm. At high external hydrogen pressure and low temperatures, the values of c observed are higher than those expected from high temperature data. It is generally accepted that certain trapping mechanism is responsible for the anomalies, but the nature of the trapping mechanism has not been clarified. Various types of hydrogen traps were suggested in the literature, such as "rifts" in the lattice dislocations, voids, microcracks, and so forth. However, no quantitative account of the observed behavior has been given in terms of these trapping mechanisms. Formation of methane in hydrogen-charged steel was found by ~odgurski." This appears to explain the existence of the so-called "residual" hydrogen,4 but not the observed anomalies of hydrogen in iron. McNabb and Foster' proposed a mathematical analysis in terms of a trapping mechanism. However, the physical significance of the traps was not clarified. The necessity of evaluating five parameters in their theory made comparison with experiment difficult. In the present paper, a model is proposed in which the anomalous behavior of hydrogen is attributed to the formation of hydrogen di-interstitials at lower temperatures. In this model, it is postulated that single and di-interstitials are in equilibrium with each other. (The hydrogen di-interstitial is to be distinguished from a hydrogen molecule which may exist in voids.) It can then be shown that most of the anomalies can be explained by assuming that the di-interstitials are virtually immobile and that the binding enthalpy of the di-interstitial, B, is approximately 10 kcal per mole. Predictions of the theory agree with the available experimental data reasonably well. THEORY The basic postulate of the present analysis is that single and di-interstitials are in equilibrium with each

Jan 1, 1969

By R. L. Klueh, W. W. Mullins

During the course of a study on the periodic precipitation (Liesegang phenomenon) of water vapor bubbles in solid silver, some interesting observations were made concerning the formation and growth of these bubbles. Under certain conditions it was found that the bubbles which formed in the single crystals led to the formation of large blisters ad cracks, and when poly crystalline specimens were used grain boundary cracking was observed. These two phenomena were examined in some detail, and it was found that an important step in the formation of the blisters and grain boundary cracks was the impingement of neighboring bubbles. DURING the study of Liesegang precipitation in solid silver,' some interesting observations were made concerning the formation and growth of water vapor bubbles. The bubbles were formed by annealing oxygen-saturated silver in hydrogen. Because of the similarity of this phenomenon to the hydrogen embrittlement of tough pitch copper,' it is known in the literature" as the hydrogen embrittlement of silver. In some respects it appears that the hydrogen embrittlement of silver is similar to the processes of hydrogen attack of steel5 and blister formation.' Hydrogen attack results from the high-temperature reaction of hydrogen with dissolved carbon or carbides in the steel to form methane (usually at grain boundaries). Blister formation occurs in steel and other materials as a result of hydrogen produced by corrosion, pickling, and so forth. Because of these similarities, some speculations were made concerning these other processes. EXPERIMENTAL PROCEDURE Water vapor bubbles were formed in 99.999+, 99.99+, and 99.9 pct Ag by annealing in a partial pressure of oxygen (pq) until saturated and then in a partial pressure of hydrogen (pH,). The hydrogen diffuses into the silver and reacts with the dissolved oxygen to form water vapor which precipitates in the form of bubbles. The specimens were then sectioned, silver-plated, mounted in cold-setting plastic, ground and polished through a preliminary polishing, and finally chemically polished.' The bubbles were visible without etching.

Jan 1, 1969

Jan 1, 1969

By John W. Cahn

The spinodal has long been regarded as a limit beyond which a homogeneous phase could no longer be metastable. But only recently has it become apparent that a phase beyond the spinodal would decompose by simple diffusional clustering mechanism quite different from the nucleation and growth mechanisms encountered for metastable phases. The theory of spinodal decomposition, which is based on the diffusion equation modified by thermodynamic requirements , is phenomenological, and each parameter can be measured by independent thermodynamic or diffusion experiments. The details of the theory have been carefully verified in a number of simple systems. It appears that the spinodal mechanism is commonly encountered in clustering reactions in solids and glasses. HISTORICAL INTRODUCTION The understanding of the earliest stages of precipitation in alloys has been the subject of three prior Institute of Metals Lectures. In 1932 Paul ~erica' reviewed the puzzling evidence that hardening of duralumin is completed before precipitation of CuAlz takes place and suggested that hardening occurs when the copper atoms "assemble in random groups on lattice positions of the duralumin, not of the CuAlz lattice". He called these random groups knots. This modification of his dispersion hardening theory seemed to fit all the facts, but Merica worried that the required diffusion "is just the reverse of what we deem normal; although quite spontaneous, it proceeds uphill, if you please". The reality of Merica's knots was quickly confirmed by Guinier's' and preston's3 X-ray work and these knots are today known as G.P. zones in their honor. Theoretical justification for uphill diffusion was also quickly provided by Becker and Dehlinger~ who noted that the diffusion coefficient could be negative in certain regions of phase diagrams. The boundaries of such regions could be described by a 90-year-old thermodynamic concept first described by ~ibbs' in 1877 as a limit of metastability for fluid phases, which by 1890 had received the name spinodal (see Appendix A). Spinodal in fluids is as old as nucleation theory. Both appear for the first time in the same paper by Gibbs who called the spinodal the limit of metastability. For the first 50 years, nucleation theory was forgotten and the spinodal was extensively treated in every good thermodynamics textbook.81Q The idea was simple and logical and it was accepted as part of a rigorous thermodynamic development of metastable phases, and for a while it was extensively investigated theoretically.

Jan 1, 1969

By V. Pasupathi, R. E. Hummel, J. W. Koger

Specimens of P1 brass were plastically deformed at room temperature to various degrees of deformation and subsequently cooled in order to transform them to low-temperature martensite. Deformation shifts Ms. A, , and the temperature of minimum resistivity to lower temperatures, and also decreases the temperature coefficient of electrical resistivity. These properties change rapidly up to about 15 pct reduction but vary very little with higher deformation. The possible relationships between martensite formed by deformation and the M, temperature of low-temperature martensite are discussed. Evidence is given that deformation martensite delays the formation of low-temperature martensite. BETA' brass undergoes at least two different types of martensitic transformations. One of these transformations (B1- B2) was first observed by Kaminski and ~urdjumov' and occurs when 81 brass with a zinc content between 38 and 42 wt pct (quenched from the single-phase region) is cooled below room temperature. Jollev and Hull' determined the structure of 0" from X-ray and electron-diffraction data as ortho-rhombic. Kunze came to the conclusion that the super-lattice cell of 0" is one-sided face-centered triclinic (pseudomonoclinic). The second martensitic transformation (B1-A1) occurs when the specimens are deformed at or somewhat above room temperature. This type of martensite will be called deformation martensite. Horn-bogen, Segmuller, and Wassermann4 determined the structure of deformation martensite to be bct. (An intermediate phase, az, occurs before the final phase appears.) At deformations higher than 70 pct, a, transforms into a.4 A critical temperature Md exists above which no transformation occurs during deformation and is estimated to be around 400°C in P1 brass.5 This martensite has elastic properties.6 When the sample is stressed, martensitic plates appear; when the stress is released, the plates disappear. The present paper studies the effect of deformation martensite on the formation of low-temperature martensite. The experiments involved samples of 8, brass which were plastically deformed by various amounts and were subsequently cooled below the transformation temperature. EXPERIMENTAL PROCEDURE The 13 brass investigated was made from 99.999 pct pure copper and 99.9999 pct pure zinc and contained 38.8 wt pct Zn. The specimens, consisting of foils 0.1 mm in thickness, were heat-treated at 8'70°C for 15 min in an argon atmosphere and then quenched into ice water. They were then deformed by cold rolling and subsequently cooled at a rate of 1°C per min. The martensitic transformation that occurred during cooling was followed by electrical resistivity measurements. The resistance measurement technique and its accuracy have been described in a previous paper. Because the transformation 81 —-8" occurs below room temperature, the samples were placed in a cryo-stat which contained isopentane as a cooling medium. The isopentane was cooled by liquid nitrogen pumped under pressure through a 15-ft coil of copper tubing which was immersed in the isopentane. The nitrogen flow was regulated by a temperature controller using two thermistors in the cooling medium. The cryogenic liquid could be heated with an immersion heater. The useful temperature range with this device was from +25° to approximately -155~C. EXPERIMENTAL RESULTS Resistivity Measurements. The following abbreviations are used in this paper to label the characteristic temperatures during the martensitic transformation. M, is the starting point of the martensitic transformation and is defined as that temperature where the resistivity vs temperature curve on cooling first deviates from a straight line. Mf is the temperature at which the martensitic transformation is completed. On reheating, the transformation from martensite to the parent phase starts at a temperature A, and ceases at a temperature Af. Fig. 1 presents five different resistivity vs temperature curves corresponding to the transformation of brass from Dl to 8" after different degrees of reduction in thickness. The following observations can be made from these curves. 1) With increasing degree of deformation the Ms temperature is shifted to lower temperatures. This shift ranges up to 35°C compared to the undeformed state. This is also indicated in Fig. 2, where AM, (the shift of Ms, compared to the undeformed state) is plotted vs the degree of deformation. AM, increases rapidly until a reduction of about 15 pct is reached. With higher deformations, no additional increase in AM, was found. 2) With increasing degree of deformation the temperature of minimum resistivity (M) is also shifted to lower temperatures. The shift, attains a maximum of about 61°C compared to the undeformed state. In Fig. 3, AM is plotted as a function of deformation. It can be seen that, as in 1 above, AM increases rapidly and no further shift of M occurs for deformations greater than 15 pct. 3) The temperature coefficient of resistivity, is given by the slopes (dp/dT) of the linear portions of

Jan 1, 1969

By Jerry L. Straalsund, D. Bruce Masson

A dew-point technique was used to determine the thermodynamic activity of zinc at 430°C in a series of e phase Ag-Zn alloys. The composition of the alloys ranged from 72 to 88 at. pct Zn. This range included the composition at which Massalski and King3 found a reversal in the composition variation of the crystallo-graphic c/a ratio, which they attributed to an overlap of the Fermi surface across the (002) faces of the Brillouin zone. The data is presented in a graph in which RT In yz,, where yz, is the activity coefficient of zinc, is shown as a function of atomic percent zinc. This curve has an unmistakable change in slope at the same composition that Massalski and King observed the beginning of the reversal in the c/a ratio. This change in slope of the thermodynamic data is also attributed to a Brillouin zone overlap. Equations are presented to demonstrate that the thermodynamic activity can be related to the density of states of the conduction electrons, and that the observed phenomena are consistent with this model. It is also demonstrated that the contribution of the density of states can be related to the excess stability, a phenomenological parameter recently shown by Darkeen" to be significant in the interpretation of thermodynamic data of metallic phases. The data seem to indicate that zone overlap has caused a spinodal point, and the resulting misci-bility gap, in the phase diagram. THE problem of developing an adequate thermodynamic model of solid solutions has proven to be difficult, and is still only partially solved. The main approach has been to develop a statistical model, such as that of an ideal solution, regular solution, and so forth, to which can be attached corrections for electronic, vibrational, magnetic, ordering, or other contributions. Such corrective terms are usually derived on an ad hoc basis, and it is difficult to predict in advance what their relative importance will be. This problem has been discussed in general terms by Oriani and Alcock,' who have reviewed several thermodynamic models and a few empirical correlations. The measurements described in the present paper were made to demonstrate in a special case the importance of one such corrective term, the contribu- tion of the energy of the conduction electrons of an alloy. It was our premise that the contribution of the energy of the conduction electrons to the thermodynamic activity of the alloy components could be detected; further, that such an effect would be observed at alloy compositions where other phenomena, also ascribed to the energy and density of states of the conduction electrons, are observed. The idea of the importance of the conduction electrons is hardly new. Hume-Rothery and his adherents have developed the well-known theory of alloy phases in which the sequence of phase fields in binary equilibrium diagrams, especially those involving the noble metals with the IIB, IIIB, and IVB subgroups, can be correlated by replacing the composition variable with the ratio of conduction electrons to atoms, e/a. Jones and others have developed a physical explanation for this correlation, in which they consider the solubility limits of phase fields to be restricted by an intersection between the Fermi surface and a Brillouin zone. The general features of the model are also quite well-known—presumably zone intersection causes the density of states to decrease at critical alloy compositions. The attendant increase in energy of conduction electrons in the original crystal structure allows an alternate structure to become more stable as the concentration of polyvalent solute is increased. In spite of the wide acceptance of these ideas on phase stability, there is only indirect* evidence, such as the variations in lattice parameter recorded extensively by Massalskizy3 and others, that Brillouin zone interactions occur. There are few experimental measurements, other than the correlations of the phase sequence, that substantiate the premise that the energy of conduction electrons affects the solubility limits of alloy phases. Much thermodynamic data of alloys has been found to be consistent with the theory; yet there is a lack of detailed data at compositions where zone intersection and overlap are thought to occur. One would expect that the energy of the conduction electrons would make a measurable contribution to the thermodynamic properties of alloys at compositions near zone intersection and overlap if the theory of Hume-Rothery and Jones is correct. This conclusion cannot be avoided, because the phase boundaries are determined by the requirement that the chemical potential of the components be equal in both phases at equilibrium. An electronic effect large enough to alter the stability of a phase should also affect the thermodynamic activity by a measurable amount.

Jan 1, 1969

By Amitava Gangulee, Michael B. Bever

The effects of changes in short-range order on some thermodynamic, electrical, and mechanical properties of the silver-rich Ag-Mg solid solutions have been investigated. The heats of formation at 273°K of several alloys after two different thermal treatments were measured by liquid metal solution calorimetry. Their electrical resistivities were measured at temperatures ranging from 4° to 773°k. Tensile tests and microhard-ness measurements were carried out at room temperature. An effective short-range order parameter and an effective interaction energy are defined. The measured changes in the properties are interpreted in terms of these quantities. The short-range order parameter is evaluated separately from calorimetric and X-ray data; the results are in fair agreement. Increasing short-range order lowers the resistivity of- the silver-rich Ag-Mg alloys but does not measurably affect the temperature dependence of the resistivity. Short-range order increases the yield stress of these alloys but does not affect the ultimate tensile stress. Changes in the effective short-range order parameter independently obtained from measured changes in the heat of formation, resistivity, and yield stress are in fair agreement. THE thermodynamic, electrical, and mechanical properties of solid solutions have been investigated as functions of composition in many alloy systems, but the effects of configurational variables, such as short-range or long-range order, on these properties have received much less attention. The silver-rich terminal solid solutions in the system Ag-Mg are well-suited for the investigation of the effects of order. The composition of these solid solutions may be varied over a range extending to about 27 at. pct Mg.1 Appreciable short-range ordering is likely to take place in these solid solutions; the degree of short-range order can be varied by suitable thermal treatments. In the alloy Ag3Mg long-range ordering is possible.2 This paper is primarily concerned with the effects of short-range order on some thermodynamic, electrical, and mechanical properties of the silver-rich Ag-Mg solid solutions.-The long-range ordering transition is the subject of a concurrent paper.3 1) EXPERIMENTAL PROCEDURES 1.1) Preparation of Specimens. Ag-Mg alloys containing up to 26.4 at. pct Mg were prepared by melting 99.99 pct pure Ag (obtained from Baker Chemical Co.) and 99.99 pct pure Mg (obtained from Eastman Chemi- cal Co. or Johnson Mathey & Co.) in graphite crucibles under molten potassium chloride. The solidified ingots were homogenized at 823°K for 10 days in evacuated Vycor capsules. A surface layer 6 in. thick was then removed by machining. The ingots were swaged into rods (2.5 mm diam) which were drawn into wires (1.0 mm diam). The wire specimens were annealed at 773°K for 24 hr in vacuum and either quenched into iced brine or cooled to room temperature over a period of 15 days. The average grain diameter (obtained from the linear intercepts) was about 0.1 mm. 1.2) Calorimetry. The heats of formation of the alloys were measured in a tin solution calorimeter as the difference in the heat effects of alternate additions of samples of an alloy and the mechanical mixture of its component elements. The additions were made from 273°K to the bath at about 623°K. The heat effect on addition of the alloys was approximately 5.5 kcal per g-atom. The procedure and the method of calculation have been described.4 1.3) X-Ray Diffraction. Short-range order parameters of two alloys were calculated from diffuse scattering intensities,' obtained with briquettes made from powdered specimens. A GE-XRD 5 diffractometer with filtered copper radiation was used. The absolute intensities were based on calibrations with paraffin. 1.4) Resistivity Measurements. The electrical resistivities were measured by a potentiometric method in which the potential drop across the specimen was compared with that across a standard resistance. For measurements below room temperature, the specimens were immersed in liquid helium (4°K), liquid nitrogen (78°K), dry ice and trichloroethylene (195°K), or ice and water (273°K). Measurements above room temperature were made on specimens held in a furnace under helium or in vacuum. The resistivity measurements were reproducible to ±0.5 pct or less. 1.5) Mechanical Tests. Tensile tests were carried out at room temperature with a Tinius-Olsen XY elec-tromatic Universal testing machine. Wire specimens (nominal gage length 1 in.) were used. The strain was measured with an extensometer, which had a sensitivity of 4 x 10-5; the strain rate was 10-3 min-1. Microhard-ness measurements were made with a square-pyramid indenter and a 100-g load.11 2) RESULTS AND DISCUSSION All measurements were made on two parallel sets of specimens, one quenched and the other slowly cooled from 773°K. The reported values are the averages of at least three measurements. In the range of magnesium concentration substantially below that of the composition Ag3Mg, quenched alloys have less short-range order than slowly cooled alloys. As the magnesium concentration approaches that of Ag3Mg, slow cooling develops long-range order. Quenching sup-

Jan 1, 1969

By Amitava Gangulee, Michael B. Bever

The order-disorder transition in Ag-Mg alloys in the range 17 to 26 at. pct Mg was investigated and some thermodynamic, electrical and mechanical properties of ordered Ag-Mg alloys were measured. A modification of the phase diagram is proposed on the basis of measured transition temperatures. The stability of the ordered structure is analyzed in terms of the quasichemical theory. Kinetic aspects of the order-disorder transition were also investigated. The energies of ordering at 273°K of Ag-Mg alloys were measured by liquid metal solution calorimetry. The electrical resistivities and the tensile Properties of the ordered alloys were measured. The exothermic energy of ordering increased with magnesium concentration up to the composition Ag3Mg; it is discussed in terms of the quasichemical theory. Pronounced order hardening was observed and can be explained by several strengthening mechanisms, which are particularly effective because of a change in crystal structure associated with the ordering transition. IN silver-rich solid solutions containing approximately 17 to 26 at. pct Mg, long- range ordering may occur during appropriate therma1 treatments. Clare-brough and Nicholas1 first detected such long-range order in the solid solution of composition Ag3Mg and suggested that the unit cell of the superlattice contained 256 atoms. x-ray2 and electron diffraction3 investigations confirmed the occurrence of long-range order, but indicated a shifted Ll2-type structure. On the basis of recent X-ray measurements the crystal structure of ordered Ag3Mg has been confirmed as type D023.4 The present paper is concerned with the order-disorder transition in silver-rich Ag-Mg solid solutions and some thermodynamic, electrical, and mechanical properties of the ordered alloys. The effects of short-range order on the properties of silver-rich solid solutions containing up to 26 at. pct Mg are discussed in a concurrent paper.5 1) EXPERIMENTAL PROCEDURES 1.1) Preparation of Specimens. Specimens of Ag-Mg alloys containing from 18 to 26 at. pct Mg were prepared in the form of 1.0-mm-diam wires as described in Section 1.1 of Ref. 5. Disordered specimens were prepared by annealing at 773°K for 1 hr and quenching into iced brine. Ordered specimens were prepared by annealing at 773°K for 1 hr and slowly cooling to room temperature over a period of 15 days. The specimens were stored at 78°K. 1.2) Resistivity Measurements. Electrical resistivi- ties were measured by a potentiometric method.' Equilibrium values were obtained at several temperatures. The kinetics of ordering were investigated by following the time-dependent changes of the resistivity of initially disordered specimens during annealing. The specimens were enclosed in copper capsules and immersed in salt pots; the heating up was accelerated by injecting preheated helium into the capsules. 1.3) Calorimetry. The energies of ordering of the alloys were measured in a tin solution calorimeter as the difference between the heat effects of additions of completely ordered and disordered specimens of the same composition. The procedure and the method of calculation have been described.6 Magnesium was used for thermal compensation in most calorimetric runs in order to improve the accuracy.677 1.4) Mechanical Tests. Tensile tests were carried out with wire specimens at room temperature as described in Section 1.5 of Ref. 5. Microhardness measurements were also made.? 2) RESULTS AND DISCUSSION The characteristics of the order-disorder transition of silver-rich Ag-Mg solid solutions will be discussed first. Some thermodynamic, electrical, and mechanical properties of the ordered alloys will then be considered and compared with the corresponding properties of dis-disordered alloys. 2.1) The Order-Disorder Transition. 2.1.1) Thermodynamic and Structural Aspects. The transition temperatures Tt were determined from discontinuities in the slope of the equilibrium resistivity vs temperature curves. Normalized curves for three compositions are shown in Fig. 1. The transition temperature increases with the magnesium concentration and reaches a maximum at the stoichiometric composition Ag3Mg. The transition temperature of the alloy Ag3Mg was measured as 665° ± 2°K and compares with published values of 660°,1 663°,8 and 66°k.3 In the slopes of the resistivity vs temperature curves of alloys containing 22.2 and 22.5 at. pct Mg, discontinuities were observed at two temperatures. Such upper and lower discontinuities indicate a two-phase field. A modified form of the published phase diagram9 is shown in Fig. 2. A two-phase field was found only on the low-magnesium side of the composition Ag3Mg. On the high-magnesium side, the existence of a two-phase field could not be established because of insufficient resolution, but such a field must be present. This part of the phase diagram can be made complete by a eutec-toid-type reaction a = (a' + ß). The existence of a boundary between the two-phase fields (a + ß) and (a' + ß) is also in accord with published lattice parameters.3 The crystal structure of ordered Ag3Mg (type Do23)

Jan 1, 1969

By M. Yokoi, K. Yoshida, T. Ishihara

The solubility of aluminum in cryolite melts has been determined by measuring the weight loss of aluminum submerged in melts contained in closed silicon nitride crucibles. The solubility in pure cryolite has also been checked with radioactive isotopes. The solubility decreases with increasing A1203 concentration, decreasing temperature, and decreasing cryolite ratio. The change of aluminum solubility in pure cryolite with the activity controlled by using Al-Ag and A2-Cu alloys supported the assumption that aluminum would dissolve in the form of monovalent ions according to the reaction 2AI + A1' —3 A1'. DURING the electrolytic production of aluminum, the electrodeposited metal redissolves in the bath in some form and is subsequently oxidized back to the former ionic state. This dissolution is believed to cause a decrease in current efficiency. Reported data on the solubility of aluminum in cryolite melts are scattered. The mechanism of the dissolution has not yet been determined, although certain tentative interpretations have been proposed. Many previous studies were performed under different test conditions. Studies were performed in open Systems and closed systems,'o and some were apparently not given sufficient time to Saturate." The solubility can be determined either analytically or gravimetrically. The analytical determination involves measurement of hydrogen gas evolved when frozen cryolite bath saturated with aluminum is treated with sodium hydroxide or hydrochloric acid. The gravimetric determination, adopted by many investigators,'-37'3>14 involves measurement of weight loss of the metal after equilibration with a known amount of salt. It appears that the former technique yields much smaller values of solubility than the latter technique. This large difference in results strongly suggests that the dissolved metal in the cryolite melt is no longer in the metallic state. In the present study, the solubility was determined by the gravimetric technique at various activities of aluminum using A1-Ag and Al-Cu alloys. An attempt was also made to determine the ionic (or atomic) state of the dissolved metal from the solubility data. 1) EXPERIMENTAL Apparatus. Preliminary tests demonstrated that silicon nitride was suitable as a container material since it is compatible to both the fluoride melts and molten aluminum. Crucibles were made from silicon powder nitrided with Nz gas at 1600°C. The good compatibility between crucible and melts is due to the near perfect nitrogenation of silicon. Only small amounts of silicon were found in the aluminum metal after the experiments. The crucible dimensions were 90 mm inside height, 38 mm ID, 105 mm outside height, and 50 mm OD. Close-fitting lids made of silicon nitride were also prepared. The crucible was placed in a close-fitting graphite crucible heated by a vertical Kanthal furnace, Fig. l. The lid and crucible are fastened tightly together by the condensation of vaporized bath in the narrow gap between them. The temperature was measured in the wall of the graphite crucible and controlled within * 5°C. The temperature difference between the measuring point and the interior of the melts was initially determined. Materials. The materials used were: Greenland cryolite (impurities CaF,, 0.5 pct; AlzO3, 0.2 pct; high-purity aluminum metal (purity 99.99 pct); A1203 and AlF, (provided by Nippon Light Metal Co., Ltd); reagent-grade NaF and CaF2; and 99.9+ pct Ag and Cu. Powdered reagents were carefully dried before use. A1-Ag and A1-Cu alloys were prepared before the experiments; their composition was checked by analysis

Jan 1, 1969

By M. J. Pool, P. J. Spencer

Using liquid tin as the solvent metal, the heats of solution of silver and gold in dilute Ag-Au-Sn alloys have been determined at 723°K by solution calorimetry. From the values obtained, the various enthalpy interaction coefficients in the liquid solutions have been calculated. The self-interaction coefficients Au and Ag values of 2413 and -6322 cal per g-atom in binary Au-Sn and Ag-Sn alloys, respectively. For an alloy containing 0.04 mole fraction Ag, Au is equal to 2443 cal per g-atom, and in an alloy containing 0.04 mole fraction Au, Ag has a value of —8883 cal per g-atom. Ag Au has been determined as —2650 cal per g-atom. ADDITION of a second solute to a dilute binary alloy may cause striking changes in the thermodynamic characteristics of the alloy. As yet, such influences cannot be reliably predicted and it is only by accumulation and critical assessment of data that definite patterns can be established. Solute interactions in dilute solutions were first treated mathematically by Wagner,1 who derived expressions for the Gibbs energy interaction coefficients. For an alloy consisting of. two dilute solutes, i and j, in a solvent s, these coefficients are given by: where yi is the activity coefficient of component i and x represents the mole fraction of solute or solvent. Other interaction coefficients have since been established in terms of the various partial molar thermodynamic properties of an alloy; the definitions of these are summarized in a recent paper by Lupis and Elliott.2 Mathematical and experimental studies of such coefficients have received much attention recently, providing a very useful theoretical background to an understanding of the nature of solute interactions taking place in dilute solutions. One of the definitions to arise from such work has been the extension of Wagner's treatment by Lupis and Elliott4 to include an enthalpy interaction coefficient, defined as. where ?Hi is the relative partial molar enthalpy of component i. ?Hi is related to the relative partial molar enthalpy at infinite dilution, ?Hio, and to the enthalpy interaction coefficients by the expression: Values of ?Hi can often be obtained with a high degree of accuracy by calorimetry, and, if the composition of the alloy is carefully controlled by making calculated solute additions throughout a series of measurements, application of Eq. [2] will enable reliable enthalpy interaction coefficients to be determined. Partial heats of solution must be investigated as a function of solute concentrations down to xi = 0, and, since ? values are obtained from the slope of mi plotted against x, great care must be taken to obtain precise data. The data for binary systems which are presently available often require extrapolation of ?Hi values from compositions as great as xi = 0.05 or even 0.1 to obtain limiting heats of solution. Reliable enthalpy interaction coefficients are thus impossible to obtain from published data for most of the alloy systems already investigated. The aim of the present work was to determine values of Au, Ag, Ag, and Au in the limit xsn -1 for the relatively simple Ag-Au-Sn system and from these results to gain some insight into the possible influence

Jan 1, 1969

By V. Koump, R. H. Tien

An approximate mathematical solution is found for the problem of unidimensional solidification of a slab of a binary alloy with time-dependent surface temperature. For the model considered, the solid skin does not form until the liquidus front has moved a certain distance. This distance depends on the rate of cooling and can be as large as half the thickness of the slab when the cooling rate is small. Assuming that the THE process of unidirectional solidification of binary alloys with constant surface temperature was analyzed by the present authors in a recent paper.' In the previous analyses, the surface temperature was considered to be lowered instantaneously, at the beginning of solidification, from its initial value to some surface temperature decreases linearly with time and that the solidus and the liquidus lines are straight, calculations are made of the temperature and concentration distributions, movements of the liquidus and the solidus fronts, as well as changes of temperature, solid fraction, and solute concentration of the liquid at the center of the slab. arbitrary temperature, lower than the solidus temperature. Although the solution of the problem for the case of constant surface temperature has some practical applications, instantaneous changes of temperature are physically impossible. Present theoretical study of the unidirectional solidification of a binary alloy is based on a more realistic model, i.e., in this case, the surface temperature of the slab decreases from its initial value continuously with solidification time. The model of solidification within the solid-liquid region is the same as the one described in the

Jan 1, 1969

By P. Bolsaitis, L. Skolnick

Vapor pressures of silver over ternary alloys of six different copper to gold ratios have been measured by means of the Knudsen effusion technique. The data obtained in the temperature range of 1330° to 1375° have been used to estimate the integral excess free energies of the ternary alloys and the activities of copper and gold. The results compare favorably to the predictions based on the electron cell model of alloys recently formulated by the authors. THE orifice effusion method for measuring vapor pressures of materials in condensed phases was first introduced by knudsen, and has since been used extensively for the determination of thermodynamic properties of condensed phases. The basis of the measurement is the effusion, through a small orifice, of vapor in equilibrium with a condensed phase. Assuming free molecular flow into high vacuum, the vapor pressure, P (in mm Hg), of the effusing species is calculated from the following equation:' where w (g) is the weight of effusing species lost during a run, t (sec) the time length of the run, A, (sq cm) the area of the effusion orifice, T ("K) the temperature at which the run is conducted, M the molecular weight of the effusing species, and k the Clausing factor, a constant that corrects for deviations from free molecular flow caused by the finite thickness of the orifice walls.17 The determination of partial and integral thermodynamic variables from vapor pressure data of ternary alloys is possible only if the vapor pressures of all but one component are negligible or can be esti-

Jan 1, 1969

By Richard W. Greene, Joel S. Hirschhorn

MATERIALS containing very large amounts of porosity, in the order of 50 to 90 vol pct, are extremely difficult to evaluate according to conventional mechanical behavior tests. This is usually a consequence of their low strength and brittleness, and the difficulty of preparing standard test specimens. A simple fatigue test has been developed and used successfully on porous nickel sheet materials requiring no specimen preparation; it could also be used for configurations other than sheet or plate. Samples of the sheet. 2 in. sq and about 0.025 in. thick, are placed in a vessel that is evacuated to about 50 f Hg by a mechanical pump. Distilled and deairated water is allowed to enter the vessel thereby filling the porosity (virtually all the porosity in these nickel ma-

Jan 1, 1970

By L. A. Tarshis, H. E. Cline

A stability criterion for a lamellar eutectic interface was derived previously1 assuming that the wavelength, A,. of the perturbation which grows most rapidly is much larger than the lamellar spacing. A. This assumption was required to permit averaging of both the interface temperature and composition over the complicated geometry of a lamellae. If this assumption is invalid, as has been suggested.2,3 then the problem becomes too complicated to solve completely. The first of these analyses1 did not calculate the wavelength. A,. of the initially unstable perturbation. whereas the more recent analysis.' which led to a contradiction of this assumption. did not include one of the key factors for determining this wavelength. To rigorously determine the wavelength, the effect of the curvature of the perturbed interface (not considered by either of the above treatments) must be included. By omitting this term, one would be led to an incorrect conclusion regarding the magnitude of A,. but to an accurate criterion for stability as evidenced by the excellent agreement with experimental observations.' In the case of single-phase alloys. the critical wavelength. A,. for instability can be calculated from the Mullins and Sekerka4 (M.S.) analysis which includes the effects of capillarity. In the following discussion, the M.S. analysis is applied to lamellar eutectics and the critical wavelengths calculated for reasonable growth conditions in Pb-Sn eutectic alloys. In order to apply the M.S. analysis to eutectics. one may substitute i) the effective liquidus slope.1 m*. for the single-phase liquidus slope and ii) the ratio of the bulk alloy composition. C,. to the average interface composition in the liquid (Ci = CE where CE is the equilibrium eutectic composition) for the distribution coefficient to give:

Jan 1, 1970

By E. A. Metzbower

STEREOGRAPHIC projections of hcp metals depend upon the c/a ratio. Hu1 has recently published standard projections for titanium (c/a = 1.5873) of the (000l), the basal plane, the (1010), the most prominent slip plane, and the (1012), the most prominent twinning plane. Caplan et a1.2 and Hul have also published tables of angles between poles of planes, between directions, and between directions and poles of planes. From these tables standard projections and double stereograms can be constructed. However, by using the on-line CALCOMP plotter with a CDC 3800 computer, direct plots of standard stereographic projections of titanium (c/a = 1.587) have been made.

Jan 1, 1970

By Leonard B. Griffiths

In a recent communication, Smith and Wood described some results of 800°C compression/creep tests on WC-12 wt pct Co alloys in which plastic flow in the WC particles was believed to have occurred. The deformation observed in the WC particles was found to be a function of the rate of loading. The authors suggested that their data constituted support for the Ansell6-Lenel theory of yielding in dispersion-strengthened alloys. I do not agree with this interpretation since the Ansell-Lenel theory does not apply to the type of alloy in question. In this alloy, the dispersoid is present to the extent of about 80 vol pct. Even where there is no continuous skeleton, which apparently is the case in this instance, the simple fact is that nearly all of the load is carried elastically by the stronger material (wc).7 The constraint imposed upon the matrix, by virtue of the differences in Young's modulii and Poissons ratios, severely limits the stress level to which it may be subjected (assuming that the interfaces remain intact). If one allows for the constraint and then performs the same computation as in Smith and Wood, one finds that the stress on a WC particle is about 4 x 105 psi. This is about 21/2 times lower than the predicted fracture stress for WC. I would completely agree with the final paragraph of Smith and Wood, namely, that loading rate is of considerable significance. However. a more likely interpretation of their data is, I submit, that in the absence of stress relaxation (by secondary slip or climb) those WC-Co interfaces subjected to a normal tensile stress begin to fail. Once this occurs, the stress distribution will change rapidly and "premature" fracture of the WC then becomes quite probable. Tne serrated edges shown in Fig. 1 may very well be fine, interface fissures.

Jan 1, 1970

By E. Ryba, D. J. Michel

SAMPLES of REZn2 (RE = rare earth) intermetallic compounds prepared by a variety of techniques have been shown to exhibit an extensively twinned micro-structure.1-4 We have recently conducted compression tests at room temperature on specimens of HoZn2 prepared by a powder metallurgy technique.' The microstructures of specimens before and after testing are shown in Fig. 1. The "before" specimen, Fig. l(a), shows the basic twin structure of this material. Fig. l(b) shows the microstructure of the "after" specimen just below the fracture surface and reveals considerable cross-twinning as well as additional

Jan 1, 1970

By C. S. Barrett

WhEN a collimated beam of protons or other energetic particles is scattered by a single crystal, the emerging particles produce patterns that contain much crystallographic information. Calculations of the trajectories of individual particles can predict with reasonable accuracy the complete pattern (which is related reciprocally to channeling, now studied in detail6) but the task is a lengthy one which has been completed for very few cases. A many-beam dynamic diffraction calculation of the patterns would be even more difficult and in some cases may be invalid. Because of these complexities, simple methods that will predict to a rough approximation the relative intensities of the lines in films of the back-scattering patterns of a given crystal are wanted. I have found a method which, although it is not a valid computation of intensities or of film darkening by either dynamic or kinematic diffraction principles or by trajectory calculations, nevertheless is found to be useful in practice. It has been tested on several crystal types,7 and is here tested on additional ones. The method consists in first calculating the structure factor F(hk1) for the (hkl) plane, of spacing d, by the familiar equation: where the summation extends over the atoms in the unit cell and fj is the atom scattering factor of the jth atom (using tables computed for electron diffraction, preferably those computed without exchange). Then the F values are squared and summed over all the orders, n, for which F2 has an appreciable value, giving a quantity: Finally, I, is expressed as a percentage of Ic for the

Jan 1, 1970

By A. U. Seybolt, R. E. Honnemon

THE purpose of this communication is to present activity data of nickel in the Ni-A1 binary system that was obtained for oxidation studies in this system. This data has been calculated from previous aluminum activity data at 1000°C1,2 by two types of Gibbs-Duhem integrations. These results should be of particular current interest to oxidation studies of aluminide coated nickel-base superalloys3 as well as other problems. The aluminum activity of Steiner and Komarek1 is plotted in Fig. 1 as a function of composition in Ni-A1 alloys at 1000°C. The dashed parts of the aluminum curves represent our approximate extrapolations from the directly measured regions. In the terminal solution regions (a and L phases) standard parabolic solution functions of the following form are assumed: where i is the component subscript, Ki is a constant at a fixed temperature, and xi is the atom fraction. The resultant nickel activity curve at 1000° C from the Gibbs-Duhem integration4 of the aluminum data is shown in Fig. 1. The solubility ranges of the phases present in the Ni-A1 binary at 1000°C are also denoted at the bottom of Fig. 1. Corresponding values of the nickel activity, aNi and yNi, are listed in Table I at useful compositional intervals. including phase boundary values. The aNi and yNi data listed in Table I are subject to the uncertainties present in the experimental aluminum activity data1 and to deviations from parabolic solution behavior in the terminal solutions. The errors introduced into aNi and yNi due to the parabolic solution assumption are expected to be relatively small (<20 pct of the values given) in all the phases except the liquid phase. Some deviation from parabolic solution behavior in the liquid phase is indicated

Jan 1, 1970