Search Documents

By S. Senderoff, D. E. Couch

A PROCESS for the electrolytic preparation of molybdenum from molten salts has been described previously. This previous work centered on electrowinning and electroplating characteristics of the process. In this paper the properties of the process as a refining medium are discussed. In producing metals from ores, it frequently has proven most economical to obtain a comparatively impure metal by smelting, and to purify the product by electro refining. Direct electrowinning from molten salts is rather difficult when a porous diaphragm is required in the process, as in the case of molybdenum, since satisfactory refractory materials for such a diaphragm in molten salts have not yet been developed. In electrorefining, however, a soluble anode is used. By suitable control of anode and cathode current densities, the lower-valent molybdenum compound used in the electrolyte is not oxidized, and a diaphragm is not needed, thus avoiding difficulties which are encountered with other electrowinning processes. Electrorefining may result from two different processes in the cell: Dissolution of the molybdenum and the less noble impurities with a separation of the more noble impurities as an anode sludge; the electrodeposition of molybdenum and the more noble metals at the cathode, leaving the less noble metals in the electrolyte. The amount of separation which occurs in each process is determined by the potential of the operating electrode and the relative concentration of the various species at (or in) that electrode. Since molybdenum has a comparatively noble position in the EMF series in molten alkali halides,2 most refining would be expected to occur at the cathode. In this work only the cathodic process was studied. The objective was to determine to what extent various metallic ions, if they dissolve from the anode or are otherwise introduced, were concentrated in the electrolyte. The necessary conditions for the separation of each and the limiting concentration which may be tolerated in the electrolyte for effective refining were studied. The deposits were chipped from the cathode and analyzed spectrochemically for metallic impurities. Besides determining the major contaminant, a general qualitative analysis was obtained to be sure that other metallic ions were kept at a very low level during each series of runs and to be sure that the baths were not being contaminated accidentally. Samples of the bath, for analysis, were taken at the temperature of operation so there was no chance of selective crystallization during cooling. In this way the ratio of molybdenum in the bath to the other metal could be determined readily. These data permitted a check on the anode efficiency and any volatilization that took place during the electrolysis. Experimental Apparatus—The apparatus as shown in Fig. 1 was used to determine the effect of the various foreign ions upon electrodeposited molybdenum. The entry chamber A allowed the cathodes to be removed without cooling the bath and without allowing oxygen or moisture to enter." The cathodes were made Prom 0.25-in. tungsten rods and were rotated during the electrolysis. The balloon in the lower arm B was inflated with helium after the cathode was drawn up into the upper section for removal. A new cathode was then placed into the entrance chamber A after which the chamber was flushed

Jan 1, 1959

By E. R. Macgregor, J. Halpern

The kinetics of the reduction of cupric salts in aqueous solution by molecular hydrogen to metallic copper are described. The rate of reduction appears to be homogeneously determined and shows a marked inverse dependence on the hydrogen-ion concentration. Precipitation of copper is more rapid and complete in sulfate than in per-chlorate solution. A mechanism is proposed which accounts for this behavior. IN recent years a number of important hydro-metallurgical processes have been developed in which metallic salts in aqueous solution are reduced by gaseous hydrogen, usually at elevated temperatures and pressures, and thereby precipitated either in metallic form (e.g. Ni, Co, Cu) or as an oxide or salt of lower valence (e.g. VO, and VA).1 ; In connection with applications such as these, most of which involve heterogeneously catalyzed reactions, recent demonstrations"-" that salts of certain metals, e.g., Cu", Ag' , Hg" , and Hg,+ can activate molecular hydrogen homogeneously and thus catalyze hydrogenation reactions in aqueous solution, are of considerable interest. For example, cupric salts catalyze the otherwise slow homogeneous reduction of dichromate and ceric salts by hydrogen, i.e. Cr,O; + 3HL. + 8H1 ? 2Cr'- -- 7H,O [1] 2Ce ' ! H.. ? 2Ce" | 2H . [2] In perchloric-acid solution, where Cu" is not appreciably complexed, the kinetics have been found" " to be of the form -d [H,] _- k,[HJ [CU"]' dt -LCu+' l+ (k ,/k,) [H] [3] where, K = 1x10"' exp [—26,600/RT] 1, mole-' min-' ... [4] and (k ,/k,) - 0.25 at 110°C. This kinetic behavior suggests that the reaction proceeds by a mechanism involving the following sequence of steps kt Cu" +H1CuH' 4-HA [51 k., CUH + CU" ? + 2Cu f H' [GI 6Cu'+ Cr,07 + 14H'+ 2Cr- (¦ 7H,O + 6Cu" ['i] fast The catalytic activities of the undissociated cupric-sulphate and cupric-acetate complexes are somewhat greater (by factors of 6.5 and 120, respectively) than that of the simple hydrated Cu" ion.'" These considerations suggest that, in contrast to the other hydrometallurgical hydrogenation reactions, referred to previously, reduction of cupric salts in aqueous solutions by molecular hydrogen, should proceed homogeneously, and in the absence of an extraneous catalyst. This has already been demonstrated" for acetate-buffered solutions where the product of the reaction is a precipitate of cuprous oxide (Cu,O). In the present paper, an investigation of the reduction of cupric salts in perchloric-acid and sulphuric-acid solutions, leading to the precipitation of metallic copper, is described. Some results for the sulphate system, which is of considerable hydrometallurgical interest, have been reported previously by Ipatieff",' " and by Schaufel-berger' but these do not give a quantitative account, of the kinetics and mechanism of the reaction. Experimental The reduction experiments were conducted in the 1-gal, titanium-lined autoclave described in connection with earlier related investigations," " and using a similar procedure. The solutions were made up by dissolving known amounts of CP grade Cu(ClO,), and HC10, (or CuSO, and HISO,) in dis-tilled water. While in the autoclave, the solution was stirred and maintained at a constant temperature (±loc) and under a constant partial pressure of hydrogen gas. It was ascertained that the agitation (provided by two 3-in., four-bladed stirrers, rotating at 900 rpm) was adequate so that at no time was the rate of reaction limited by solution of hydrogen. During each experiment, samples of the solution were withdrawn periodically, allowed to cool, and the concentration of copper remaining in solution was then determined electrolytically. Some Cu' is formed during the reaction, but undergoes disproportionation when the solution is cooled to room temperature. For purposes of kinetic calcula-

Jan 1, 1959

By R. H. Atkinson, R. N. Rhoda

An improved technique has been developed for the deoxidation of palladium with aluminum which is especially suitable for use in making small castings with an induction-melting and casting machine. The effect of various crucible linings on the efficiency of the residual deoxidizer and accompanying oxide film in protecting palladium against oxidation during remelting also has been examined; the preferred crucible lining consists of zirconia cement. THE objective of this work was to develop a simple method of deoxidizing induction-melted palladium which would be suitable for use by manufacturers of jewelry and other small castings. In a common method of deoxidizing torch-melted palladium a piece of aluminum foil is dropped on the surface of the molten metal. If a film is present, which is usually the case, it must first be removed by oxidizing the molten metal thoroughly because it would interfere with the entry of the aluminum into the melt. Removal of the film is probably facilitated by a mild fluxing action of the silica crucibles which are used for torch-melting. However, removal of an alumina film from molten palladium in the crucibles normally used for induction melting is a much slower process and it became necessary to modify the de-oxidation procedure. In the modified procedure the charge is simply melted and the desired amount of aluminum, crimped to one end of a tungsten wire, is plunged through the film into the palladium and the wire immediately withdrawn. The trace of tungsten which dissolves along with the aluminum can be avoided, if considered objectionable, by using a palladium wire. In view of the persistence of the alumina film under the conditions of induction-melting it also was decided to ascertain the efficiency of this film in protecting palladium against oxidation during repeated remelting and casting. Melting Equipment and Materials The equipment used for melting and casting consisted of an Ecco GC:6, 6 to 8-kva high-frequency converter and an Ecco M34B centrifugal casting machine. This unit is being increasingly used for making jewelry and dental castings and it also has been found useful in making exploratory melts of a variety of nickel alloys, for which purpose it is well suited on account of its speed of melting and economy of material. A description of the machine and its operation will be published elsewhere.' A common palladium jewelry alloy, 95.5 palladi-um-4.5 ruthenium (referred to as "palladium" in the trade), was selected for the deoxidation tests. It was available as melting stock which had been melted in air to free it from oxidizable impurities, and also as clean scrap from previous melts which had been deoxidized with aluminum. The latter was thought to be typical of clean "palladium" scrap to be found in a manufacturing jeweler's establishment and it is also similar to one brand of "casting palladium." By the same token, the oxidized melting stock corresponds to "casting palladium" after the latter has been melted and thoroughly oxidized: Pure aluminum wire (99.9 pet), 0.032 in. diam; 1 in. = 0.036 g. Scrap tungsten wire, 0.014 to 0.020 in. diam. Argon, commercial water-pumped; 99.9 pet pure Experimental Procedure Samples of about 50 to 75 g of the palladium alloy were melted in a crucible lined with a proprietary refractory alumina cement. Some charges were melted in air, others in argon. The oxidized melting stock was melted and solidified a number of times in order to remove as much oxygen as possible before deoxidizing, but this treatment was unnecessary in the case of scrap. In either case the metal was deoxidized with aluminum (0.01 to 0.05 pet) by the

Jan 1, 1959

By R. G. Hudson, H. W. Paxton

The activities of iron and chromium in their binary system have been measured by the Knudsen orifice technique. The system is close to ideal at 1200- 1250 C with slight positive deviations. The effect of this on the liquidus and solidus is discussed and these are calculated for two limiting assumptions. The vapor pressure of pure iron, pure chromium, and one alloy is compared with that obtained by the Langmuir free-evaporation technique. The results agree, indicating a vaporization coefficient close to 1. HERE are no reported data on the thermody-A namic activity of iron and chromium in the solid portions of the phase diagram. It is generally assumed'.' that the liquid solutions of iron and chromium obey Raoult's law. Chen and Chipman" have inferred that the liquid solution obeys Raoult's law. The minimum' in the solidus-liquidus curves for the iron-chromium system shows that there are deviations from ideality in either the solid or liquid or both. Because of the importance of developing reliable techniques for obtaining activity data in alloy systems at high temperatures and of the data so obtained, an investigation of the activity of iron and chromium in the binary at 1200-1250 C was undertaken using the Knudsen cell. Experimental The apparatus and experimental procedures used in this investigation were described in a previous paper." The various materials used for the construction of the Knudsen cell are listed in Table I. The metal used for the determination of the vapor pressure of chromium was 99.9 pet pure, obtained from A. D. Mackay, Inc. The iron was obtained as electrolytic grade from Fisher Scientific Company. The Fe-Cr alloys used are shown in Table 11. The largest error and controlling factor in the accuracy in the experiments reported here are introduced because of temperature uncertainty, which is estimated to be ±3 deg C. This introduces an uncertainty of 26 pet in the measured pressure. Experimental Results and Calculations The vapor pressures of pure iron and pure chromium were calculated from the measurements by the Knudsen equation3 on the assumption that only Cr or Fe atoms were present in the gas phase. For metal vapors, at the high temperatures and low pressures involved in these measurements this should be a reasonable assumption. Whitman" corrections for the cells were made. The data and results of pressure calculations are presented in Table I, and are plotted in Fig. 1. In addition the heat of vaporization at the absolute zero, ?E°°, has been calculated for chromium. These calculations are summarized in Table 111. The data used in these calculations are as follows kelley13 (F° - E°°/T ) Speiser, et al H (F° - E°°/T ) 010 where F" — standard free energy at temperature T It was necessary to calculate the total pressure above the alloys from the total weight loss W of the cell during a run, because of the fact that all attempts to collect and analyze the chromium and iron vapor effusing from the cell in the apparatus available failed. The failure resulted from the re-

Jan 1, 1959

By H. R. Peiffer

Recently artin has indicated that the recovery of resistivity at 145°C following elongation of molybdenum at room temperature was the result of the annihilation of vacancies. The activation energy for this process was, according to Martin, 1.26 ev. This energy seems to be a reasonable one for the activation energy for the diffusion of vacancies in molybdenum. However, impurity atoms such as carbon, nitrogen, and oxygen4 have an activation energy for diffusion in molybdenum of the same order of magnitude (i.e., 1.45, 1.15, and 2.09 ev, respectively). Cottrell and Churchman have already correlated the return of impurity atoms to dislocations in cold-worked iron with the return of the upper and lower yield point. According to cottrel16 the diffusion of carbon atoms to dislocations results in the return of the yield point. As the carbon atoms become trapped by dislocations, Bhatia has shown that a decrease in the electrical resistivity is to be expected. Cottrell and Churchman were able to correlate experimentally the decrease in electrical resistivity with the return of the yield point. Since one would expect strain aging to occur in molybdenum, an experiment was performed to determine whether the resistivity recovery at 150°C for molybdenum cold-worked at room temperature can be correlated with the return of an upper and lower yield point. For this experiment 3/16-in. tensile bars of Climax molybdenum, 99.961 pct pure, were used. The specimens were annealed at 1000° C for 1 hr in an argon atmosphere and then were elongated up to 10 pct on a Baldwin-Lima-Hamilton FGT

Jan 1, 1959

By J. B. Newkirk, W. G. Martin

Oxidized cobalt powder is known to have a magnetic hysteresis loop which is asymmetric with respect to the magnetization axis. The experiment described herein shows that the orientation relationship between the basal plane of hexagonal cobalt and the oxide which forms upon it at 400°C is {111}ux//{100.1 }co <110>ux//<11.O> This orientation relationship allows a magnetic interaction between the antiferromagnetic oxide and the ferromagnetic substrate which could account for the offset hysteresis loop. Oxidized nickel powder has a symmetrical hysteresis loop and so is apparently not influenced by any magnetic interaction between metal and oxide. The orientation of the oxide (cubic) was found to be identical with that of the nickel substrate when the oxide forms on a polished surface parallel with {111} Ni. IN 1956 W. H. Meiklejohn and C. P. Bean discovered that fine particles of cobalt which had been prepared in a certain way have a magnetic hysteresis loop which in a strong field is asymmetric relative to the magnetization axis.&apos; The cobalt powder exhibited this unusual magnetic property only after it had been oxidized in air or oxygen; it lost the shifted hysteresis loop when the oxide was reduced in hydrogen. X-ray and neutron diffraction&apos; experiments showed the presence of cobaltous oxide (COO) and hexagonal cobalt in samples exhibiting the biased hysteresis loop and specifically showed no indication of any other compound with the exception of mercuric oxide.* The magnetic hysteresis loop becomes symmetrical above the Nee1 temperature (paramagnetic state) of COO. Therefore, it was concluded that the anomalous magnetic behavior is associated with the influence of cobaltous oxide upon the metallic cobalt. It has been proposed that crystallographic coherency may exist between the cobalt and a plane of some antiferromagnetic material which has unbalanced spin distribution and sufficient magnetic anisotropy to hold its spin in the direction which existed when the specimen was cooled. COO has these properties. Therefore, Roth&apos; has suggested that cobaltous oxide may form with a {111} plane parallel and coherent with the basal plane of the hexagonal cobalt metal and proposed that, as consequence of the antiferromagnetic interaction between COO and the underlying cobalt, the magnetization direction in oxidized fine Co particles may be ro-lated from the easy c-direction.2 Such a relationship, he proposed, would explain the observed magnetic effect in oxidized cobalt powder. The main purpose of this study was to determine the orientation relationship, if any, between the basal plane of cobalt and the oxide which forms upon it. Attempts to produce a film of COO which was strong enough to be handled were not successful. However we did succeed in making a film of CoCo2O, which was strong enough to be removed from the cobalt substrate and mounted on an electron-microscope grid. Because of the close structural similarity of COO and CoCo2O, we believe the orientation relation found for CoCo2O, on cobalt probably also holds for COO on cobalt. The epitaxial relationship of NiO and Ni also was investigated. To date no shifted hysteresis loop has been observed with nickel powder. However, the similarity of atomic array in cobalt and nickel leads to the prediction that a {111} of NiO may be parallel with a (111} of nickel. Experimental Method Cobalt-Cobalt Oxide—A coarse-grained specimen of a (hexagonal) cobalt was prepared by allowing a large crystal of fee cobalt to transform slowly at 400°C. The crystal was then cut to expose a surface parallel with the basal plane. A back reflection Laue X-ray photograph showed that the 00.1 plane was within 2" of the cut surface. The surface was mechanically polished and then electro polished in 85 pet orthophosphoric acid after which the crystal was held for 30 min at 400°C in air. During this heat-treatment the surface darkened slightly due to the oxide film which formed. The film was not thick enough to give an X-ray diffraction pattern, even by a glancing-angle technique. Glancing-angle electron diffraction was not possible either, owing to the interference of the electron beam with the high magnetic fields which exist at the 00.1 surface of the cobalt. However, it was possible to make an electron-diffraction photograph of the oxide film by stripping it from the cobalt substrate using the method described later. By maintaining reference marks carefully, it was possible to preserve the orientation relationship between the stripped film and the substrate on which it was formed. The oxidized surface of the crystal was first covered with a 1 pet solution of collodion (cellulose nitrate) in amyl acetate. When the film was dry, small rectangles were scored on the surface with a needle point. The specimen was then repolished

Jan 1, 1959

By C. Feng, C. Elbaum

Knoop hardness measurements were carried out on large grains of iodide titanium containing different amounts of oxygen. For each oxygen content the hardness is recorded ainingas a function of the crystal orientation and of the angle 4 between the indenter diagonal and the projection of the C (hexad) axis in the surface. The results can be expressed by the following relation; hardness = n-k cos 2ø, where n and k are constants for a given oxygen content and crystal surface orientation. A NISOTROPY of hardness is a well-known phe-nomenon and has been reported for many metals.1-3 At the present time, however, such information is not available in the literature for titanium. Consequently, a systematic study of the simultaneous effects of oxygen content and crystallographic orientation on the hardness of titanium was undertaken in this laboratory. Evaluations of oxygen content in titanium are frequently carried out through hardness tests. Although such procedures may be justified in small grain specimens, the results of the present investigation indicate that the anisotropy of hardness should be taken into account in the case of coarsegrained specimens as well as in the case of pronounced preferred orientation. Materials and Experimental Procedures Iodide titanium, 99.99 pet pure and containing of the order of 0.002 wt pet O, was used throughout this investigation. The metal was arc melted in an argon atmosphere and cold rolled into strips 1/16 in. thick. These strips were subsequently recrystal-lized and strain annealed in vacuum (10" mm Hg or better) at 850°C in order to produce crystals varying in diameter from 10 to 25 mm. Oxygen was introduced into selected large crystals by heating at 700°C in an atmosphere of pure oxygen under an observed pressure of approximately 50 mm Hg. Each sample was subsequently annealed in vacuum at 850°C for 100 hr, in order to homogenize the oxygen content throughout the specimen. It was found that longer annealing was not necessary, for no change of surface hardness in a given direction was detected after prolonged annealing. The oxygen content was determined by the difference in weight of the specimens before and after the oxidation-annealing cycle. The accuracy of oxygen determination is within ±0.1 mg in samples weighing 0.5 to 1 g. A vacuum fusion analysis of oxygen was performed on several samples and gave results in good agreement with the ones obtained by the weight change. It should be emphasized that expected weight losses through evaporation of the metal in vacuum at 850°C&apos; are many orders of magnitude lower than the precision of these measurements. Following the annealing treatment, both oxidized and unoxidized crystals were carefully polished. All detectable traces of surface cold working were removed by repeated polishing and etching. Laue back reflection X-ray patterns were taken for each crystal for the purpose of determining the orientation and ascertaining the absence of substructures detectable by this method. Hardness measurements, using a 1 kg load, were subsequently carried out on a Tukon hardness tester equipped with a Knoop indenter. The Knoop indenter was used throughout this work because indentations obtained by means of a Vickers indenter are considerably distorted due to the anisotropy of the elastic constants in titanium. Such distortions are much less pronounced in the case of Knoop indentations, although for some orientations a cer-

Jan 1, 1959

By C. S. Barrett, D. B. Masson

Martensitic transformations have been found in AgCd both upon cold-working at room temperature and cooling to lower temperatures. The crystal structures of the transformation products were found to be close-packed hexagonal in the first case and or-thorhombic in the latter. The transformation characteristics also were studied. Similar treatment of AuZn alloys failed to yield any transformation. THE literature on martensitic transformations of body-centered cubic metals and alloys indicates widespread instability of this structure at low temperatures, a tendency which merits further study. The body-centered cubic ß-phase of CuZn, for example, when it contained less than 42 wt pct Zn, was found by Greninger and Mooradian&apos; and others&apos; to transform on cooling to subzero temperatures to a phase that was tentatively considered to be tetragonal. Further work by Massalski and Barrett" demonstrated that cold work at low temperatures causes a partial transformation of ß CuZn of higher Zn content. The structures of the strain-induced transformation products were found to be different from those induced by simple cooling. Another well-known martensitic transformation among body-centered cubic phases is that of ß AuCd, first studied with X-rays by Olander,&apos; then very thoroughly by Chang and Read.5,6 Here again transformation was found to occur both as a result of cooling and of strain. The alkali metals Li and Na, as well as certain Li-Mg alloys, also undergo transformation from their normal body-centered cubic structure, both on cooling and on cold-working. The phase AgCd, the subject of the present report, should be expected to behave quite similarly to CuZn for a number of reasons. It is the chemical analogue of CuZn, silver being directly below copper and cadmium directly below zinc in the periodic table. The size factors (percent difference in size) of the atoms, possibly a consideration in the stability of the ß-phases, are +3.1 in AgCd and + 4.2 in CuZn. Likewise the relative electronegativities of the two pairs of metals are about the same, CuZn possessing an electronegativity difference of 1.0 and AgCd one of 0.8. These values are from a table of electronegativities given by Darken and Gurry." The phase diagrams of the two systems around 50 atomic percent are similar, both possessing a disordered body-centered cubic phase, ß, at high temperatures and an ordered bcc phase, ß, at room temperature; however, the ß and ß&apos; phases of AgCd are separated by a hexagonal phase, [, stable in the range 220-440oC, which has no analogue in CuZn. There has been confusion in the literature as to the structures of the ß&apos; and [-phases. The phase diagram in the ASM handbook&apos;" has the structures of the two phases interchanged; however, Hansen" reports the correct relative positions and Muldawer" verified that ß&apos;, the low-temperature phase, is body-centered cubic. Work by the present authors also

Jan 1, 1959

By A. Keil

TO measure the absolute conductivity of metallic specimens it is generally necessary to have exact geometrical shapes and good electrical contact with the test-piece. To overcome limitations of the standard methods, an electronic conductivity-meter* Trade name, Sigmatest; model Dr. F. Forster; Reutlingen, Ger-many. has been developed, which works on the principle of inducing high frequency eddy-currents in the specimens and measuring the reaction of these curients on a measuring coil.&apos;,&apos; To attach the instrument to the test specimen, a flat area of about 10 mm diam is all that is necessary. The reading is taken off a scale calibrated in pet of International Annealed Copper Standard. The induced eddy-currents are propagated in a circular motion parallel to the surface. Thickness of the layer included in a measurement depends on the frequency used. With a frequency of 60 kc per sec, the penetration of the currents was 0.8 mm at the upper limit of the scale, and 2 mm at the lower limit. The method is limited to nonferrous metals and other applications have been described elsewhere." The object of the present investigation was to detect the formation and growth of surface layers of differing conductivity, and to determine the depth to which these changes could be observed. A well known phenomenon which illustrates the problem is the internal oxidation of certain silver alloys containing base metal solutes, such as copper or cadmium, when they are annealed in an oxidizing atmosphere.4, 5 Until now, the only methods available for following these phenomenon have been metallo-graphic examination and hardness tests.0, 7 Changes in mechanical properties of copper alloys by internal oxidation were recently discussed in a more general way by Martin and Smith." When measuring these conductivity changes it must be remembered that the only absolute values of conductivity are those of 1) the nontreated homogeneous specimen, and 2) the final values obtained with a surface layer of depth comparable to the penetration of the eddy-currents. All intermediate values refer to the mixed conductivities of the transformed surface layer and the untransformed alloy, and are thus readings of relative value only. Experimental Results Specimens were annealed for 1/2 hr at 800°C in hydrogen, and the internal oxidation was carried out by thermal treatment in air at 800°C. Fig. 1 shows the results of the first series of experiments. With internal oxidation of a solid solution of 8 pet Cd in silver, the conductivity rises continuously until the surface layer formed has become thick enough to be the deciding factor (about 400µ). Precipitates of CdO are mainly in the form of points, Fig. 2, and, as a result, the electric contact between the silver crystals is not distorted by them. Brine11 tests were run with a 1.25 mm ball and a 7.812 kg load. The specimen annealed in hydrogen and oxidized to a depth of 400µ had a hardness of 40 Bhn,

Jan 1, 1959

By H. W. Weart, D. J. Mack

SINCE the first recognition of eutectic solidification as the simultaneous formation of 2 solids from one liquid,1 many complex structures thus produced have been observed. Despite many attempts to ascertain it, however, the detailed mechanism by which these structures develop remains obscure. In an attempt to clarify the origin of eutectic structures, the study described in this paper was undertaken. The broad objective of the study was to investigate the effect of different solidification conditions on the structure of single eutectic grains. It was expected that the eutectic structure would thereby be isolated for study, but preliminary experiments revealed another structure within each grain. When a literature surveye revealed only 1 previous mention of the subgrain structure,&apos; it was decided to determine the origin of the structure and to study its response to various growth conditions. Before presenting the details of the investigation, it is apropos to distinguish carefully among the 3 structures produced during eutectic solidification. This is especially desirable in view of the apparent confusion existing in the literature concerning the structures in a eutectic specimen. The confusion is due largely to the existence of the subgrain structure noted above which is easily mistaken for the grain structure. In the following paragraphs, an effort is made to distinguish the 3 structures by briefly describing their origin during solidification, even though the results are thereby anticipated slightly. The 3 structures produced during eutectic solidification are the grain structure, the colony structure, and the eutectic structure, each one contained by the one preceding it. The grain structure is analogous to the polycrystalline structure of a metal or inle-phase alloy in that each grain grows from a single nucleus. A somewhat broader interpretation is needed here, however, because of the presence of 2 phases in each grain. Since in most eutec-tics particles of 1 phase are dispersed in a matrix of the other, a grain may be taken to be the region in which the matrix phase is monocrystalline.*

Jan 1, 1959

By F. Weinberg

It has been suggested, for some time, that the behavior of metals under high-temperature creep conditions is strongly influenced by the behavior of the grain boundaries present in the material. This has led to a number of investigations1-6 in which attempts were made to measure the properties of the grain boundaries themselves, both to determine their contribution to the creep process and to obtain further knowledge of their intrinsic structure and properties. The present investigation was undertaken to extend the observed data under more closely controlled experimental conditions and to attempt to clarify sonie of the discrepancies in the literature. A brief report of the preliminary results has been published.7 EXPERIMENTAL PROCEDURE For the present investigation, controlled orientation specimens were used having plane boundaries which terminated at the specimen surface. Trierys-tal specimens were used instead of the bicrystal type of Puttick and King,&apos; in order to simplify the gripping technique required to apply a shear stress along the boundary. The aluminum used was obtained from the Aluminum Francaise and was of 99.994 pct purity, containing 0.002 pct each of Cu, Fe, and Si. The aluminum tricrystals were grown horizontally from the melt, in graphite boats, and in an argon atmosphere, following the technique developed by Chalmers.&apos; The orientation of the tricrystals was controlled by using seed crystals separated by mica inserts. Thermal conditions were adjusted to produce tricrystals having boundaries which were macroscopically straight, parallel to the growth direction, and perpendicular to the top surface. Following growth, the tricrystals were carefully cut with a jeweller&apos;s saw to the test specimen shapes shown in Fig. l(a) and l(b). In Fig. 1(a), the boundaries subjected to shear stresses are indicated AB and CD: Points A and C are the ends of the mica inserts separating the seed crystals; points B and D are at lateral cuts made in the specimen. It was found that the slots left by the mica inserts at A and C had to be enlarged slightly, by cutting, to ensure that the boundary terminated freely at these points. A shear stress was applied to the boundarks by gripping the upper two legs in file-faced Inconel grips and suspending a weight from the lower part of the specimen held in similar grips, as shown schematically in the diagram. The test specimens shown in Fig. 1(b) were cut from long

Jan 1, 1959

By L. W. Berger, R. I. Jaffee, D. N. Williams

The effects of hydrogen on the properties of binary Ti-A1 alloys containing 2.5, 5, and 7 pct aluminum and ternary alloys of the same compositions with added interstitial elements oxygen, nihrogen, and carbon were studied. Aluminum increases the apparent hydrogen solubility of the a-phase, resulting in a marked increase in resistance to impact embrittlement. In the Ti-5Al and Ti-7Al alloys, for example, the addition of 180 and 305 ppm hydrogen, respectively, did not cause embrittlement, while unalloyed titanium was embrittled by as little as 75 ppm. The role of aluminum in increasing hydrogen solubility of the a-phase probably accounts for the excellent hydrogen tolerance of a-ß alloys containing considerable amounts of aluminum. There is some evidence that aluminum retards the rate of hydride precipitation. If so, these alloys may be susceptible to slow-strain embrittlement. UNALLOYED a-titanium is extremely sensitive to small amounts of added hydrogen in that there is a marked decrease in notch properties. For example, as little as 100 ppm hydrogen is sufficient to lower the room-temperature notch-impact strength from over 100 to less than 20 in.-lb.&apos; This loss of impact Strength has been associated with the Presence of hydride phase in the microstructure. Hydrogen in the amounts encountered in commercial practice

Jan 1, 1959

By L. W. Berger, and R. I. Jaffee, D. N. Williams

The effects of hydrogen on the properties of binary Ti-A1 alloys containing 2.5, 5, and 7 pct aluminum and ternary alloys of the same compositions with added interstitial elements oxygen, nihrogen, and carbon were stzdied. Aluminum increases the apparent hydrogen solubility of the a-phase, resulting in a marked increase in resistance to impact embrittlement. In the Ti-5Al and Ti-7Al alloys, for example, the addition of 180 and 305 ppm hydrogen, respectively, did not cause embrittlement, while unalloyed titanium was embrittled by as little as 75 ppm. The role of aluminum in increasing hydrogen solubility of the a-phase probably accounts for the excellent hydrogen tolerance of a-ß alloys containing considerable amounts of aluminum. There is some evidence that aluminum retards the rate of hydride precipitation. If so, these alloys may be susceptible to slow-strain embrittlement. UNALLOYED a-titanium is extremely sensitive to small amounts of added hydrogen in that there is a marked decrease in notch properties. For example, as little as 100 ppm hydrogen is sufficient to lower the room-temperature notch-impact strength from over 100 to less than 20 in.-lb.&apos; This loss of impact Strength has been associated with the Presence of hydride phase in the microstructure. Hydrogen in the amounts encountered in commercial practice

Jan 1, 1959

By R. N. Zitter

PRESENT work was done in connection with a program for obtaining a low energy gap semiconductor suitable for photodetection. The inter-metallic compound InSb, which is now being in-

Jan 1, 1959

By C. F. Frey, P. J. Ahern, J. F. Wallace

THE high affinity of molten titanium for oxygen and nitrogen has resulted in considerable difficulty in developing a satisfactory melting procedure. It has been found necessary to perform melting operations either in a vacuum or inert atmos-phere, and failure to obtain a satisfactory crucible material has necessitated the use of special melting techniques. These techniques have included cold-mold arc-melting utilizing both consumable and nonconsumable electrodes, skull-melting, drip-melting, and levitation melting.&apos; It has been recognized generally that the melting of both titanium sponge and titanium scrap would be expedited by the use of induction-melting procedures but the lack of a crucible capable of containing molten titanium without objectionable contamination thus far has prevented the use of induction-melting procedures. An extensive investigation: which included the common refractory materials was not successful in developing a satisfactory crucible for induction-melting. Some success in induction-melting has been attained with the use of a graphite crucible material, but in all instances sufficient carbon contamination was encountered to cause an appreciable reduction in the optimum mechanical properties of the titanium or titanium alloys. These investigations"" report an increase in the carbon content from approximately 0.3 to 1.0 pct. A serious reduction in the tensile and forming properties has been observed as a result of this carbon contamination. The development of a satisfactory induction-melting procedure for titanium without undue contamination undoubtedly would have considerable economic significance. Such a process would be particularly useful in melting the large quantities of low-cost scrap now available that can only be reprocessed at considerable expense and difficulty by the more elaborate techniques referred to previously. In addition, the process could provide a satisfactory method of melting and alloying of bri-quetted sponge for forging ingots.

Jan 1, 1959

By D. Turnbull, H. N. Treaftis

Precipitation of tin from Pb-Sn alloys (lead-rich) occurs by the nucleation and growth of hemispherical cells which consist of tin lomelloe interspersed in the depleted solid solution. Nucleation and growth of these cells and the tin interlamellar spacing at various compositions and temperatures have been investigated. The interlamellar spacing increases approximately as the reciprocal of the logarithm of the supersaturation ratio. For constant initial tin concentration the concentration of cell nuclei increases with decreasing temperature to a limiting temperature-independent value. The kinetic results support the view that the cell growth rote is governed by the diffusion of tin along the cell boundary. In the main, the results confirm the authors&apos; earlier deductions, drawn from resistometric and calorimetric data, about the precipitation mechonism. IT has been shown1, 2 that tin (ß) precipitates from lead-rich (a) PbSn solutions by the nucleation and growth of cells (cellular or discontinuous precipitation). These cells consist of tin lamellae, with a spacing 1, interspersed in a depleted a solution. Reorientation of a also accompanies cell growth. The kinetics of cellular precipitation of tin from lead have been measured by resistometric2, 3 and ca1orimetric1, 3 techniques. The time-transformation isotherms (t equals time; x, volume fraction of specimen transformed) of the initial rapid precipitation, which drains about 60 pet of the tin from solution, have been described by a kinetic law having the form x = 1 —exp (—bt n) [I] where n usually has the value 3.0. It was shown" that this law and the other kinetic facts can be explained satisfactorily if it is assumed that the cells grow at a constant rate and that all the cells nucleate in a time negligible in comparison with the period of cell growth before impingement. Thus, for hemispherical cells b = (2/3)pNG3 [21 where G is the rate of cell growth in cm per sec, and N is the number of cell nuclei per unit volume. The rapid rate of cell growth at low temperatures can be accounted for"&apos; if it is assumed that the dissolved tin drains to the lamellae edges by diffusion along the sweeping cell boundary rather than by diffusion through the a crystal (volume diffusion). For cell growth governed by cell boundary diffusion it was shown, approximately G=2(X° - X° / X° ) DBA / l2 where A equals the thickness of the cell boundary; D R, the coefficient of diffusion of tin atoms within the boundary; and X° and X, are the atom fractions of tin in the supersaturated and saturated solutions, respectively. Zener8 had already shown that, where

Jan 1, 1959

By G. W. Preckshot, J. W. Gorman

MOLTEN metals offer an excellent medium for the study of molecular diffusion and interphase transfer. In the absence of intermetallic compound formation, solutions of molten metals are solutions of elements. Therefore, complicating factors such as molecular structure, dissociation, association, polarization, and so on, are generally absent. Further, metals remain in the liquid state over a wide range of temperature. Thus, the effect of temperature on transport processes can be studied in more detail. Although some data are available for the diffusion of metals in the molten state at high temperatures,"""".&apos;!&apos; the most useful and accurate data are those for the diffusion of various elements in mercury at room temperature. The effective radius of diffusion for the various elements in mercury may be calculated from the Stokes-Einstein relation.&apos;-These calculated radii are compared with the atomic and ionic radii calculated from X-ray data on crystals." The effective radii of diffusion of the alkali and alkali-earth metals correspond to those of the neutral atom; those of cadmium, silver, lead, zinc, thallium, bismuth, and tin correspond to ionic radii; and those of copper and gold indicate larger radii due to association with mercury or more likely, faulty diffusion data. Independent evidence"&apos; supports the notion that alkali metals are uncharged in amalgams. The concept of metals moving free of their conductance electrons in the liquid state has been discussed by Eyring, et al,&apos;" in connection with studies on the viscosity of pure molten metals. Considering the approximate nature of the tabulated atomic and ionic radii and the variance in the diffusion measurements, the application of the Stokes-Einstein equation to the diffusion of metals in mercury gives gratifying results. It is noted that a number of these metals have article radii eaual to or less than those of the solvent mercury.".&apos;*,"&apos; Soba1,&apos;" sing the inaccurate Roberts-Austin20 data, indicated that the effective radius of diffusion might well be a function of temperature. Additionally in studies of mass transfer between phases, many conflicting opinions have been expressed concerning the assumption of equilibrium at the interface." ,","," The results of the present work on the diffusion of solid copper in molten lead throws light on the tempera- ture variation of the diffusion radius as well as the extent of the interfacial resistance. The theoretical background and experimental procedure are presented here in detail. Results on other systems will be presented in subsequent papers. Nomenclature The following nomenclature is used in the paper: A—cross-sectional area of diffusion column, sq cm C, C,—concentration of solution and saturated solution respectively, g per cc D,D&apos;—-diffusivity, sq cm per sec e—Naperian base h—ratio, k/D k—mass-transfer coefficient, cm per sec L—diffusion column length, cm n—series index q—flux, g per sq cm-sec r—radius, cm or A R—-gas constant t—time, sec T—temperature, K V—volume, cc w—weight of copper-lead alloy on copper cylinder WT—-total copper transferred to lead diffusion, column, g Z—distance from top of diffusion column, cm 2&apos;—distance from copper-lead interface, cm (2&apos; = L-2) u,,—roots of diffusion equation Mathematical Apparatus The model shown in Fig. 1 applies to the diffusion of solid copper into molten lead. Here Fick&apos;s first law may be written where all terms are defined in the nomenclature. It is assumed that the volume of the lead column is constant. This is justified since experimental densities compared favorably with calculated densities based on perfect interstitial mixing. Thus there is no counter diffusion of lead. It is clear that 2 in Equation [I] should be measured relative to the interface which moves relative to the diffusion cell but not to the liquid phase. Hence, the frame of reference is the plane of the copper-lead interface. The diffusivity calculated at constant volume, D", is that defined by Hartly and Crank" and is designated herein as D. No Interfacial Resistance—The two situations of the presence and absence of interfacial resistance will be treated. The simplest is the latter which re-

Jan 1, 1959

By G. C. Fryburg, H. M. Murphy

ThE rates of oxidation of metals are usually obtained by heating the metal specimens in furnaces. Such a procedure is satisfactory for most metals. However, there are several metals that oxidize according to the reaction.

Jan 1, 1959

By J. J. Becker

Changes in magnetic properties with particle size are used to study the precipitation process in a Cu-Co alloy. In particular, the effect of a field during aging in producing anisotropy is shown to occur entirely during the growth of the particles, after precipitation is complete. HEAT-treatment in a magnetic field is done commercially to improve the magnetic properties of a number of materials. These can be divided into solid-solution alloys and precipitation alloys. The Permalloys are representative of the first type, and magnetic annealing greatly increases their maximum permeability. On the other hand, in precipitation alloys such Alnico 5 the important result is an increase in maximum energy product. The effect of the field in all cases is to bring about an additional magnetic anisotropy, uniaxial in nature, which makes the hysteresis loop more upright when measured in the direction in which the field was applied. The structural origin of the anisotropy brought about by magnetic annealing is not well understood. In the solid solution alloys, one possible explanation is based on ordering.&apos; Another possibility is the orientation of like-atom pairs.&apos; The precipitation systems known to respond to magnetic heat treatment are Alnico 2, in which the effect was first discovered," Alnico 5, cobalt ferrite (oxide),&apos; and Cu-Co alloys.3 n these materials, the field affects the precipitation process itself in some way. Although there has been much work on Alnico 5. its nature is still incompletely understood. According to Geisler,&apos; the precipitation reaction is of the

Jan 1, 1959

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

HEAT-to-heat variations in properties of an alloy of constant nominal chemical composition have been a perplexing problem to the metallurgist. These heat-to-heat differences have been especially baffling in high-temperature metallurgy. In order to establish fundamental reasons for these heat-to-heat variations, the National Advisory Committee for Aeronautics has sponsored and financially assisted an extended research program on heat-resistant alloys at the University of Michigan. As part of the program to find the contribution of processing variables, Frey, Freeman, and white&apos; studied effects of heat treating on nickel-base alloys and Ewing and reeman&apos; investigated effects of hot-working practice on N155 alloy. To extend the program to the study of effects of melting practice on high-temperature properties, a 55 Ni-20 Cr-15 Co-4 Mo-3 Ti-3 pct A1 composition was selected for experimentation. The objective of this program was to establish the melting-practice variables which influenced the hot-working characteristics and creep-rupture properties and, at the same time, establish the metallurgical mechanisms for the effects. During the course of the investigation of practices using vacuum melting, it became increasingly evident through chemical analyses supplied by industrial concerns that a major variable was melt reaction with the crucible. This paper is devoted to the results of this phase of the melting-practice study. EXPERIMENTAL PROCEDURES Experimental 10-lb heats were induction melted in the University of Michigan vacuum-melting unit with the melting practices outlined in Table I. Aim analysis for the basic alloy, in wt pct, for all heats was as follows: Cr Co Mo Ti Al 20.0 15.0 4.0 3.1 3.1 Mn Si C Ni 0.12 0.12 0.08-0.15 Bal. Small amounts of boron and zirconium were added to certain heats as listed in Table I. These additions

Jan 1, 1959