Search Documents

By Walter R. Hibbard, Harold Margolin

The effect of ten ternary additions on the aging of a Cu-Ag alloy was measured by X ray, microstructure and hardness. A supersaturated copper-rich transition structure was found. The effect of the ternary element was related to differences between transition and precipitate lattice parameters and analogies with solution and strain hardening. IN addition to the prerequisite of diminishing solid solubility with falling temperature, the theory of coherency hardening1 proposes that the principal source of hardening during precipitation is strain caused by coherency between the matrix and precipitate. For a given system greater coherency hardening would result from lengthening the period of coherent growth and/or increasing the number of nuclei, other things remaining equal. However, the hardness change during aging is a function of a number of phenomena in addition to coherency hardening, such as dispersion hardening, loss of coherency, solute depletion, matrix recovery or recrystal-lization, coalescence, and their relative time sequences. The present investigation was undertaken to study the effect of ternary additions on the age-hardening of Cu-Ag alloy. For this purpose a series of alloys were made containing as ternary constituents Mg, Al, Si, P, Mn, Ni, Zn, As, In, and Sb. These elements were added to a master 5 wt pct Ag in Cu alloy in such amounts as to obtain alloys listed in Table II containing the following approximate atomic percentages: 96 pct Cu, 3 pct Ag, 1 pct ternary addition. A 97 atomic pct Cu, 3 atomic pct Ag alloy was used as a comparison standard. There have been numerous investigations of age-hardening in Cu-Ag alloys2-lo but few of ternary Cu-Ag alloys. Pfister and Weist" found that 1 wt pct Ni reduced the solubility of Ag in Cu above 600 °C and that the rate of precipitation was slower than in polycrystalline Cu-Ag alloys of similar silver content. More recently Hodge, Jaffee, Dunleavy, and Ogden12 investigated the effect of ternary additions of Mg, Li, P, Cd, Zn, Zr and Be on strength properties of Cu-Ag alloys. Mg, Cd, and P were found to increase strength in the aged and cold-worked condition. Experimental Procedure Charges weighing 7.5 lb were melted in a closed carbon crucible under a reducing atmosphere* of nitrogen and hydrogen. A Durville-type casting procedure with a hot-top arrangement was used to obtain a sound casting, 2x2x43/4 in. Prior to rolling and annealing of the ingots, a trial solution anneal of rolled sections of the hot-top portions indicated that at 780°C melting was avoided except for the Cu-Ag-Sb alloy. This alloy could not be produced as a single phase within the limits of temperature control. i 3" to 5oC, since melting occurred at 64O °C, and below this temperature precipitation was

Jan 1, 1952

By G. D. Gemmell

Effects of solid-sdutidn alloying with titanium, molybdenum, and tungsten at concentrations up to 10 pct on the strength of pure columbium at elevated temperatures (mainly 2000°F) have been investigated by means of creep-rupture, hot-tensile, and hot-hardness testing and recrystallization studies. ThIS paper presents some results from the initial part of a study of the effects of alloying on the high-temperature strength properties of columbium. The elevated-temperature strength characteristics of pure columbium were first studied. Binary alloys of columbium with titanium, molybdenum, and tungsten containing 1, 5, and 10 pct of the solute elements were chosen on the basis of their high melting points, their known solid solubility in columbium, and results of separate oxidation-resistance studies of columbium alloys. EXPERIMENTAL PROCEDURE All the materials used were of highest purity. A typical analysis of the columbium is shown in Table I. Compositions of the alloys are also given in Table I. To evaluate the strength properties of pure co-lumbium and the alloys at elevated temperatures, creep-rupture, hot-tensile, and hot-hardness testing was done on recrystallized material. To prepare the pure metal and the alloys for these tests, 1- 1b buttons were made in a water-cooled copper crucible, inert atmosphere, tungsten arc furnace. Tapered ingots, 3 in. long, were made from these buttons by a skull-casting technique and these were machined to 1/2 in. diam, swaged to 50 pct reduction (pure columbium at room temperature, alloys at lowest temperature possible without cracking, generally in range 1100" to 1800°F) and heated in vacuum (lo-' mm Hg) for 16 hr at 2550°F for re-crystallization and homogenization. The resulting grain diameter was 0.1 to 0.3 mm. The 10 pct Mo alloy casting was not successfully swaged at temperatures up to 1800°F without cracking and was not included in the test program. Creep-rupture testing was done at constant load in vacuum on cylindrical specimens of 0.16 in. diam, using Instron capsules. The pressure at test temperatures was less than 10-1 mm Hg. Hardneas readings taken on the test specimen before and after testing were used as a check on possible gaseous contamination. Hot tensile tests in vacuum on cylindrical specimens of 0.125 in. diam were also made with Instron equipment at a strain rate of 0.067 min.-l. Hot-hardness measurements were taken with a unit constructed at the Du Pont Experimental Station. The test piece, supported on an alumina anvil, and the molybdenum indenter shaft with sapphire tip is heated in a platinum-wound furnace. The load applied to the pyramid indenter is measured by strain gages attached to a proving ring located outside the hot zone. In the measurements reported here the load was I kg. The tester operates in a vacuum of about 10"5 mm Hg and is capable of temperatures up to 1400°C.

Jan 1, 1960

By B. A. Rogers, K. O. Stamm

BECAUSE it is the basis of the economically important operation of hardening steel by quenching, the A3 transformation in iron has been the subject of numerous investigations. Although the contributions to this subject are interesting, no general review will be given here. Readers who care to follow the matter further are referred to the comprehensive summary by Cleaves and Thompson,1 which covers the literature to about 1935. Their references to 53 papers on the subject suggest that further investigation is hardly justifiable. However, considerable interest has developed recently concerning the reversibility of the transformation. From thermodynamic considerations, the transition ought to occur at the same point on heating or cooling, if the change of temperature is sufficiently slow. Yet all of the earlier investigators found a difference of several degrees. Burgess and Crowe,2 who used thermal methods, obtained average temperatures of 912° for Ac3 and 900°C. for Ar3 on electrolytic iron. Their data indicate a considerable range of temperature over which the change occurs, both on heating and on cooling. Terry3 measured the susceptibility in the critical range and estimated from the steepest parts of his curves that Ac3 came at 9180 and Ar3 at 903°C. Ishiwara4 was one of the first investigators to report a shorter interval between the temperature of the change in the two directions. He measured the change of susceptibility of kaolin-coated samples of steel, in a hydrogen atmosphere, at rates of heating and cooling so low that about 3 or 4 hr. were required for the transformation to complete itself. His results show some variability between samples, but indicate that the difference between Ac3 and Ar3 is in the vicinity of 6°C. Roberts and Davey5 measured the space lattice in pure iron in the vicinity of 900°°C. and concluded that a change occurs reversibly between 907° ± 3° and 9100 ± 3°C., probably between 907° and 910°C. Wells, Ackley, and Mehl6 made dilatometric determinations of the Ac3 and Ar3 points in samples of hydrogen-treated carbonyl iron maintained under a vacuum. They used three rates of heating and cooling: 2°, '1/2°, and 1/8°C. per min. and found the spread between the two points decreased with low rates. At the slowest rate, the Ac3 transition occurred at 909° to 909.5°C. and the Ar3 change at 907°C. in three of the samples believed to contain the lowest percentages of alien elements. They concluded that if the rate of temperature change and the content of alien elements were low enough, the A3 transformation would be completely reversible at 909.5° ± I°C. The question of reversibility is an interesting one and when some iron with especially small percentages of alien elements became available the opportunity to conduct an investigation was welcomed. This iron was prepared by the Division of

Jan 1, 1941

By J. W. Halley, G. L. Plimpton

Because of the two distinct stages in the solidification of rimmed steel, segregation in the rimmed ingot is more complex than that in the killed or semikilled ingot. In the earlier stage, chemical reactions, resulting in evolution of gas and vigorous stirring of the metal, effect a degree of negative segregation unique to rimmed steel. The composition of this zone constitutes the principal quality advantage of this type of steel. In the later stage of solidification, after the evolution of gas has ceased, the segregation follows the general pattern of a more completely deoxidized steel. However, the composition of the liquid remaining after the first solidification stage varies markedly from that of the original melt, some constituents having been partially removed as reaction products while the concentration of others has been intensified by rejection from the metal solidifying in the first zone. The aim, from the standpoint of segregation, in rimmed steel of high quality is to effect a maximum negative segregation in the rim zone without excessive concentration of undesirable elements at the center of the ingot. To achieve this aim, the composition of the melt may be varied and the furnace and pouring practice may be adjusted within certain limits. It is the purpose of this study to discuss segregation in rimmed steel from the standpoint of the mechanisms that produce it and to survey the phases of open-hearth practice that influence those mechanisms. Mechanisms Producing Segregation The mechanisms that determine the degree of segregation in the rimmed ingot are closely interrelated, making it difficult to select the primary factors and to determine a satisfactory order for presenting them. For instance, the question of sulphur segregation is relatively simple, since sulphur is not directly involved in chemical reactions in the ingot; but its concentration in the rim zone is dependent not only on its original analysis in the melt, its segregation coefficient, and the freezing rate, but also on the rate of its removal from the impure film on the solidifying surface. This washing action involves all the factors (such as freezing rate, the carbon-oxygen product, pressure, etc.) that determine the rate of gas evolution and hence the efficiency of film removal. Sulphur distribution between the rim and core is affected by the volumetric ratio between the two zones, and its segregation in the core is a function not only of the freezing rate of the core but also of all the foregoing factors that determine its concentration in the metal when solidification of the core starts. To simplify the presentation, it is proposed to treat the mechanisms affecting segregation independently in this section, reserving a discussion of their over-all effects for the section in which actual segregation data are given. The factors affecting segregation will be treated in the following order: (I) differential solidification, (2) removal of constituents as

Jan 1, 1945

By Nelson Severinghaus

Mineral fillers have been defined as inert materials that are included in a composition for some useful purpose. Because this definition and the scope of this review are not specific-indeed, exceptions abound-accurate total statistics are unavailable. One source estimates the total production in the United States in 1979 at 11.3 Mt (12.5 million st) valued at $534 million, averaging $49.19 per ton. Included are a broad group of minerals-asbestos, barite, bentonite, ground limestone and dolomite, clays, diatomite, feldspar, micas, silica, talc. Additional references will be made to synthetic mineral-derived fillers such as aluminum hydrate, precipitated calcium carbonate, and pyrogenic and precipitated silicas and silicates. A section on related uses at the end of this chapter covers absorbents and filters. Mineral fillers can be included in compounds to accomplish a variety of final results. Hardness, brittleness, impact strength, compressive strength, softening point, fire resistance, surface texture, electrical conductivity, and a host of other physical characteristics may be modified by the inclusion of fillers. These effects are the result of properties of the fillers, including chemical activity, hardness, particle size, shape and distribution, surface structure, color, density, and refractive index. To facilitate an overview of the mineral filler industry, this chapter will review the general usage of fillers, the inherent characteristics that intensify their utilization, and similarly their manufactured characteristics. Major using industries will then be examined in detail. Specific information on deposits and processing for industrial minerals will be found in the appropriate commodity chapter. All of these same minerals have additional significant uses based on their chemical composition or other characteristics. References will be found in other use chapters and summarized in commodity chapters. The chart in Table 1, updated from the 4th edition of Industrial Minerals and Rocks, illustrates the major mineral fillers (both natural and manufactured) and their more important end uses. Use of Mineral Fillers Archaeologists have found that more than 5000 years ago, bitumen used for mortar and waterproofing contained mineral contaminants that acted as fillers. Since that time man has added mineral products to compounds to improve certain characteristics. Perhaps the most common objective of filler usage is reduction of cost of the end product. Certainly the inclusion of a filler that costs 2¢ per 0.45 kg (lb) will reduce the cost of a plastic product whose resin costs 50¢ per 0.45 kg (lb). But the decision to use a specific filler is not that simple. It may "soak up" binder, making a compound more viscous and perhaps preventing its flow into a mold as an injection molding compound, but similarly will give good anti-sag characteristics to putty, caulking compound, or undercoating. Fillers may be incorporated to modify: 1) Cost 2) Physical characteristics 3) Flow characteristics 4) Fire resistance 5) Density 6) Heat conductivity 7) Color or brightness and opacity 8) Hardness, brittleness, impact strength 9) Deformity, viscosity, softening point 10) Processing problems 11) Electrical conductivity 12) Surface texture 13) Thermal expansion characteristics 14) Abrasion resistance

Jan 1, 1983

By Cyril Wells, Robert F. Mehl

The practical importance of a knowledge of the rates of diffusion of carbon and of alloying elements in steel has often been pointed out.l,13 This importance lies chiefly, though by no means only, in the study of carbufizing and in the study of reactions involved in the eutectoid inversion in steel. The present paper is one of a series in which the rates of diffusion of carbon and of alloying elements in ?-Fe and their mutual effects are measured with precision. The present paper reports the results of measurements of diffusion coefficients D of manganese in ?-Fe over wide ranges of temperature, and concentration, and of the effect of carbon and impurities and of grain size upon these diffusion coefficients.‡ Data on the rate of diffusion of manganese in ?-Fe have been reported by Fry2 and by Paschke and Hauttmann.3 Claussen4 calculated the diffusion coefficient D from Fry's data as 3 X 10-10 sq, cm. per set. at 960º, which will be shown to be incorrect by many orders of magnitude. Paschke and Hauttmann gave a rough value of 9.6 x 10-8 Sq. cm. per set. at 14000. Neither of these workers considered the variation of D with concentration and with temperature, and neither studied the effect of carbon. Method and Materials Method.—The method employed is identical with that used earlier.5 Owing to the volatility of manganese and the consequent loss of manganese from the outer surfaces of the welded diffusion bars, these surfaces were discarded. The diffusion anneals were performed in argon, except One (test 18) in which hydrogen was employed for purposes noted below. The temperatures reported are accurate to about 3" and constant within 1°C; the diffusion times are reported to the nearest half hour. The specimens were furnace-coo1ed after the diffusion ation, at about 2° per min. Layers, usually 0.004 in. thick, parallel to the welded interface, were machined from the specimens after the diffusion anneal and analyzed for manganese to 0.I Per cent Mn; the distances of these layers from the interface was measured to ±0.0002 in. The concentration-penetration curves obtained were analyzed by both the Grube and the Matano solutions.8 The D values calculated by these two methods do not differ appreci000. The Grube solution was used except in tests in which the effect of concentration upon D was greater than so per cent; in such cases the Matano method, which explicitly provides for the variation of D with concentration,' is the more reliable. In general, the number of layers machined from the zones of variable composition were large enough to permit the use of the average layer composition as that of the center of the layer. In 3 of the 3' tests, however, this is not justified, and the D values

Jan 1, 1941

By J. D. Hanawalt, C. E. Nelson, J. A. Peloubet

THE subject of the salt-water corrosion resistance of magnesium alloys is somewhat like that of the pitting of stainless steels, in that it involves a relatively small percentage of the applications but receives a much greater proportion of the publicity. Actually magnesium alloys are in use in many exterior applications as well as interior, with no trouble from corrosion. Magnesium and its alloys have a striking stability against atmospheric attack, being far superior to iron. Specimens exposed unprotected to atmospheric weathering for 10 years show only a superficial surface-film formation. The basic nature of magnesium and of its important alloys is that they are also resistant to the attack of salt water even in an unprotected state, giving substantially no loss in weight after one year of alternate immersion in 3 per cent NaCl solution (Fig. I). This basic behavior is often masked because of the extreme sensitivity of these alloys to certain impurities and combinations of impurities. It is known, and the subject of increasing emphasis, that higher purity leads generally to greater stability for many materials, nonmetallic as well as metallic. However, a review of the literature on the corrosion factors in magnesium alloys shows that almost all researchers worked with base metal in which metallic impurities varied in uncontrolled amounts. The major factor, that of metallic impurities, probably greatly overshadowed the effects of minor ones such as flux inclusions, nonmetallic impurities, porosity, metallographic state, which these researchers presumed to study. In the present investigation the first observation of interest is that whereas magnesium of commercial purity (99.9 per cent pure) has a corrosion rate in alternate immersion in 3 per cent NaCl solution at room temperature of anywhere from 5 to 100 mg. per sq. cm. per day, the corrosion rate of high-purity magnesium is 0.15 ± 0.05 mg. per sq. cm. per day. The principle followed in the experiments was to add to such base magnesium other elements, singly and in combination, in controlled and known amounts and to observe the corrosion behavior of the resulting alloy. These experiments show the existence of what may be called "tolerance limits" of magnesium and some of its alloys for certain impurities. The extremely small percentages of impurities involved and the sharpness of the corrosion discontinuities appear to be quite unpredictable. In order to carry on this work, it was necessary to develop an analysis technique suitable to the problem. In some cases it was necessary quantitatively to analyze impurities to less than 0.001 per cent. The experimental data to be presented are based on observation of about 5000 alloy specimens and on more than 50,000 quantitative analyses. While the primary purpose of this paper is to present the experimentally obtained data on corrosion rates as a function of composition, and particularly to demon-

Jan 1, 1941

By F. Keller, R. H. Brown

BECAUSE of the extensive use of Alclad 24s alloy sheet in aircraft construction, there is much interest in the metallurgical changes caused by heat-treatment of this product.1,2 One of these changes is the result of the diffusion of soluble elements from the core into the coating during treatment. This paper deals with the thermal treatments that may influence diffusion and with the practical effects of diffusion on the serviceability of the product. As is well known, Alclad 24s-T sheet has a core of high-strength alloy combined with surface layers of relatively pure aluminum. The microstructure of a cross section of this product is shown by Fig. I. hot only does the relatively high-purity coating have a high resistance to corrosion, and hence serves as a mechanical barrier between the core metal and the corroding environment, but in addition this coating I exerts electrochemical protection over areas where the core may be exposed, such as cut edges. In addition, the coating affords electrochemical protection to projecting heads of *417S-T, 17S-T and 24S-T rivets, or attached members made from these alloys. Under some conditions of heat-treatment, however, diffusion of copper into the coating will alter the electrochemical relationship between the coating and the core to the extent that only partial protection will be obtained under some corrosive conditions. The reason for the electrochemical protection lies in the fact that the coating layers of pure aluminum are anodic to the alloy core. The addition of copper to aluminum produces alloys in which the potential is lower than that of the pure metal. Moreover, as the amount of copper in solid solution in z4S alloy is increased, the potential of the alloy becomes more cathodic. When heat-treated at 920°F. and quenched in cold water, 24s-T is about 0.16 volt cathodic to pure aluminum. However, copper will diffuse from the core into the coating, and if it reaches the surface the potential will be altered in a cathodic direction. Thus the potential dif-erence between the core and coating may be less than 0.16 volt, depending upon the thickness of the coating, and the time and temperature of heat-treatment. Of the methods available for determining the extent of diffusion of copper from the core into the coating of Alclad 24S-T, two are generally used; namely, the metallographic and the solution potential method.3 METALLOGRAPHIC METHODS Metallographic methods will reveal the progress of diffusion of copper into the coating. For instance, the diffusion zone in the coating intermediate between the coating surface and the core is shown in Fig. I. This photomicrograph is of a polished and etched cross section of 0.040-in. Alclad 24S-T sheet. Diffusion proceeds into the coating more rapidly along the grain boundaries than through the body of the grains. consequently, the diffusion zone is not of uniform width

Jan 1, 1944

By Cyril Wells, Robert F. Mehl

The practical importance of a knowledge of the rates of diffusion of carbon and of alloying elements in steel has often been pointed out.l,13 This importance lies chiefly, though by no means only, in the study of carbufizing and in the study of reactions involved in the eutectoid inversion in steel. The present paper is one of a series in which the rates of diffusion of carbon and of alloying elements in ?-Fe and their mutual effects are measured with precision. The present paper reports the results of measurements of diffusion coefficients D of manganese in ?-Fe over wide ranges of temperature, and concentration, and of the effect of carbon and impurities and of grain size upon these diffusion coefficients.‡ Data on the rate of diffusion of manganese in ?-Fe have been reported by Fry2 and by Paschke and Hauttmann.3 Claussen4 calculated the diffusion coefficient D from Fry's data as 3 X 10-10 sq, cm. per set. at 960º, which will be shown to be incorrect by many orders of magnitude. Paschke and Hauttmann gave a rough value of 9.6 x 10-8 Sq. cm. per set. at 14000. Neither of these workers considered the variation of D with concentration and with temperature, and neither studied the effect of carbon. Method and Materials Method.—The method employed is identical with that used earlier.5 Owing to the volatility of manganese and the consequent loss of manganese from the outer surfaces of the welded diffusion bars, these surfaces were discarded. The diffusion anneals were performed in argon, except One (test 18) in which hydrogen was employed for purposes noted below. The temperatures reported are accurate to about 3" and constant within 1°C; the diffusion times are reported to the nearest half hour. The specimens were furnace-coo1ed after the diffusion ation, at about 2° per min. Layers, usually 0.004 in. thick, parallel to the welded interface, were machined from the specimens after the diffusion anneal and analyzed for manganese to 0.I Per cent Mn; the distances of these layers from the interface was measured to ±0.0002 in. The concentration-penetration curves obtained were analyzed by both the Grube and the Matano solutions.8 The D values calculated by these two methods do not differ appreci000. The Grube solution was used except in tests in which the effect of concentration upon D was greater than so per cent; in such cases the Matano method, which explicitly provides for the variation of D with concentration,' is the more reliable. In general, the number of layers machined from the zones of variable composition were large enough to permit the use of the average layer composition as that of the center of the layer. In 3 of the 3' tests, however, this is not justified, and the D values

Jan 1, 1941

By T. J. Woods

Tars paper deals with the origin, definition and prevention of scabs an semifinished rolled-steel product. Mold coatings, which are considered essential in scab prevention, were found to be effective only in relation to the ingot surface of semikilled and killed steels and not necessary for the rolled product. That is to say, the scabs on ingots caused by having been cast in uncoated molds are eliminated in the soaking and rolling operations, therefore are not present on the rolled product. This fact, which is contrary to the average opinion, was established after extensive and thorough experiments had proved the ordinary conception of the origin of scabs to be incorrect. This subject may be of little or no concern to steel plants producing small ingots for the manufacture of small billets and bars, except as it may be of interest in showing why their surface troubles with scabs do not exist on the product of heats with normal pouring practice. But to the producer of large ingots to be rolled into slabs for large plates used in shipbuilding, it is a very important subject. DEFINITION OF A SCAB The generally accepted opinion is that the individual scabs present on rolled products (as shown by Fig. I) are considered to have been individual splashes on the ingot surface. This opinion was found to be incorrect. Experiments showed that there were no individual scabs on the product that had appeared as individuals on the ingot surface. The single scabs seen are either pieces of a shell that is present near the bottom of all top-cast ingots, or a veneer that may be present on any part of an ingot improperly poured, as when the stream strikes the mold wall. This definition is based on a detailed study of scabs and their characteristics in which the following have been found true: Location.- On the product of heats poured without trouble, scabs, if present, are always on the product that corresponds to the bottom end of the ingot, and usually occur in clusters, This is true whether molds are coated or not. Shape.-There is no regularity as to shape, size or thickness. It is sometimes possible to separate a few pieces and place them together, like pieces of a jigsaw puzzle, indicating that at one time they had been united as one piece. If individual splashes were the origin of scabs, the foregoing would not be true. Instead, the splashes, which occur throughout the pouring of the ingots, would be present uniformly over the entire product and there would be considerable similarity in the shape and thickness of the scabs. EXPERIMENTAL WORK Our experimental work was conducted on 200-ton heats of semikilled plate steel tapped into an oval ladle 15 ft. high. All ingots were top-cast direct into 24 by 54 by 82-in. molds, straight-sided, set on fat stools. Two separate studies were made about one year apart, each study extending

Jan 1, 1943

By H. E. Cline, D. F. Stein

A fine lamellar structure with interlamellar spacings ranging from 0.I to 3.7 was Woduced by high-speed directional solidification and used to study the interaction of dislocations with an array of bimetallic bound-am.es. The tensile flow properties were measured at elevated temperatures and the results interpreted in terms of an image force model. As the width of the plates decreases the strength increases according to a Hall-Petch relation, suggesting pile -ups. Considerable strengthening was also observed at elevated temperatures. In recent years composites containing a strong but brittle phase and a ductile but weak phase have been studied extensively in the search for a high-strength material with adequate ductility.' However, the strength of these composites has been limited to a volumetric average of the strengths of the component phases. In this paper we will show that a composite made by directional solidification of a Cu-Ag eutectic has strength properties that substantially exceed the strength of either of the metals in the composite, and an explanation for this increased strength will be presented. The Cu-Ag eutectic is a near ideal material to study for several reasons. The eutectic structure is made up of alternate plates of silver-rich and copper-rich phases with the same crystal structure.2'3 In addition, the mechanical and elastic properties of silver4 and copper5 are known over a wide range of temperatures and the eutectic microstructure is stable to nearly the melting point.6'7 Therefore, the properties of the eu-tectic are readily comparable to various theories involving the mechanical properties of copper and silver. The experiments consisted of producing directionally solidified Ag-Cu eutectic material, determining the structure and performing tensile tests at different temperatures and strain rates. EXPERIMENTAL A) Solidification and structure. A cylindrical ingot 2.2 cm in diam and 15 cm long was prepared from 99.999 pct pure materials and cast in argon. The ingot was placed in an alumina crucible and was direc-tionally solidified at 7 x 10"4 cm per sec in a Bridg-man apparatus in argon. Metallographic examination at both ends and along the length of the ingot showed a uniform structure that consisted of fine alternate plates of silver- and copper-rich phases. The plates were aligned parallel to the growth direction, as shown in the longitudinal section, Fig. 1, and were in small randomly oriented domains of parallel plates as viewed in a transverse section, Fig. 2. Four smaller ingots 0.175 in. in diam were solidified at speeds of 2Xlo-', 4.4 xlo-', 4.4 x 10-3, and4.4x10-4 cm per sec to produce a range of lamellar spacings. Replica electron micrographs were taken to resolve the individual plates as shown in Fig. 3. The copper-and silver-rich plates in the larger ingot were about

Jan 1, 1970

By B. H. Cody

THE Morenci concentrator of the Phelps Dodge Corporation is a part of the project that was completed after five years had been spent in development of the open-pit mine, equipment and process testing in the former No. 6 concentrator, and construction of the reduction works. The reduction works, of which the concentrator is a part, is about a mile and a quarter southeast of Morenci, on the highway that leads up Morenci Canyon. The concentrator has a nominal capacity for the treatment of 45,000 tons of ore daily. The original plant, which has a rated capacity of 25,000 tons of ore daily and is now in its fifth year of operation, began production on Jan. 30, 1942. The addition, to increase the capacity 8o pct, began operation on Dec. 1, 1943. The equipment of the original plant has been described in MINING AND METALLURGY (May 1942). No changes in equipment were recommended for installation in the concentrator extension and only one major change, forced by a shortage of maintenance labor, has since been made. Metallurgical treatment has been altered in some ways as experiments progressed and as labor and economic conditions dictated. No complete description of the operation of the concentrator has been published, so this paper will record such operating and equipment changes as have been necessary, the operating methods that have been developed and the metallurgical results that have been obtained. Descriptions are given in the order of the processes with a flowsheet and an equipment list of each process for convenient reference. The questions asked by engineer visitors have guided the detail in which operation and maintenance are described. During the period Jan. 30, 1942 to Sept. 1, 1946, ore milled has totaled 43,595,253 dry tons, from which 1,643,885 tons of concentrate have been made, containing an indicated production of 783,164,809 lb of copper. The principal results obtained in operation of the concentrator during the years 1942 to 1945, inclusive, and to Sept. 1, 1946 are shown in Table 1. A graphic general flowsheet of the plant is given in Fig I. ORE The Morenci ore that is now being treated is a medium-hard monzonite porphyry in which the principal sulphide minerals are chalcocite and pyrite. A minor amount of covellite is present. Chalcopyrite and bornite have been observed under the microscope. The chalcocite occurs mainly as a coating on pyrite. Some of the coatings are so thick that the pyrite presents the appearance of being an inclusion in chalcocite. Some are less than one micron thick and are barely visible at 500 diameters magnification of a polished cross section. Some chalcocite is present as a dissemination in the gangue. Various oxidized copper minerals are present as a result of oxidation in place but they represent a minor proportion of the copper minerals of the ore body. Oxidation

Jan 1, 1947

By C. F. Clausen

"Cement" means binding agent or glue. It is derived from the Latin word "Caementum," the name of a limestone, chips of which were used in mortar more than 2,000 years ago in Italy. During the middle ages "cement" or "cement" generally denoted mortar. Today it means any binding material. The Thomas Register lists nearly 200 classifications for "cement" including automobile cement, household cement, rubber cement, etc. The material most commonly associated with this word, however, is portland cement which, when mixed with water and allowed to hydrate, is a ceramic binding material used to cement aggregates (crushed stone, gravel and sand) together in "concrete," the most universally used building material of today. It is estimated that more tons of concrete are used annually than all other building materials put together including steel, lumber, gypsum, bricks, tiles, aluminum and glass. Its use in foundations, building construction, sanitary work, dams, railroads, and highways, makes it difficult to conceive of modem civilization without concrete. A survey made in October 1955 showed the following uses of portland cement: [Percent Highways20 Nonresidential buildings20 Residential buildings15 Military construction10 Sewer and water works8 Public utilities10 Other (conservation, farm, and maintenance and repair)17 100] As of today, the percentage for highways is probably even higher. Ceramic binding materials have unquestionably been used in construction work for thousands of years. The mortar between the blocks of the great pyramids was calcined gypsum (plaster of paris). The use of calcined and slaked limestone goes back earlier than the Roman Empire. Gypsum, however, is relatively soft, and common lime mortar is not "hydraulic," i.e., it does not harden under water. The Greeks and Romans discovered later that an addition of a finely divided active siliceous material (Santorin earth or Pozzolana) would react with slaked lime to form a stronger, and hydraulic, cement. The Appian Way in Italy, Colosseum, and the great Aqueducts built by the Romans were made with this type cement, and some of these structures are still in existence today. However, not until 1756 did some serious research result in the development of better cements. In that year the English engineer, John Smeaton produced a stronger cement by burning a limestone containing considerable proportions of clayey impurities. Other investigations and improvements were made by Bryan Higgins, Bergmann, Joseph Parker, L. J. Vicat, J. F. John, James Frost, and I. C. Johnson.2 Traditionally, but perhaps unfairly to others, the invention of the prototype of today's portland cement is credited to the English mason Joseph Aspdin who in 1824 obtained a patent and gave the product its name, because in color the mortar produced from it resembled a natural building stone obtained from the British Isle of Portland. The contributions of Johnson and Aspdin were primarily the important step of preparing a synthetic mixture of lime and clay without depending on raw materials found in nature which might contain such elements in desirable proportions. However, even these early

Jan 1, 1960

By Margaret G. Scroger, Luke L. Y. Chang, Bert Phillips

Using the sealed-system technique, isothermal sections at 1000°, 1400°, and 1700°C for the system Co-W-O hare been determined. The equilibrium data were obtained by microscopic and X-ray diffraction examination of samples winch had been equilibrated in pellet form fit sealed capsules and subsequcelly quenched. The only ternary phase in the system is CoWO, which mells congruentlly at approximately 1280°C. At 1000°C, this ternary phase exists in equilihrinm with all the allows and oxide compounds in the hounding binary systems. At 1400°C, liquid forms in thin oxygen-rick region near the CoWO4 composition and only one phase assemblage in the system (W + WO2 + H'5CV7) contains no In order for tungsten to be used to its full potential as a refractory metal. it is necessary that coatings be developed which will preclude oxidation of the metal. It is true that coatings now in use do protect tungsten for short periods at temperatures near 1700 °C. but no coating has been developed which protects adequately at or above this temperature where the use of tungsten is most required. This situation exists despite the great effort in recent years to develop such coatings. and it has become evident that success in this area can be achieved only after a more fundamental knowledge of the thermochemical and therrnophysical propertics of tungsten oxidation and tungsten -metal oxidation is obtained. In our 1aboratories. a project liclrriel. At 1700°C, tungsten is the ouly solid phase which is stctble, and a large range of compositions melt to two liquids, one metallic and one oxidie. The sealed-system method of etjuilibra lion prevents the determination of the partial pressure of oxygen and the total gas pressure orer the condensed phases at equilibrium. This data is not available and can be obtained only by some other method. We can generalize on these unknowns and indicate that both partial pressure of oxygen and total pressure of the gas phase must increase as compositions change from the W-Co binary toward the oxygen apex and as temperatures increase. has been initiated which is directed at providing some of this fundamental information through a study of metal-tungsten-oxygen reactions and equilibria and the properties of any intermediate phases which result. This paper is based on a part of this program in which the system Co-W-O was studied between 1000° and 1700°C. I) RELEVANT INFORMATION FROM THE LITERATURE A) The Notable Lack of High-Temperature Work on Metal-Tungsten-Oxygen Systems. It should be pointed out that there has been no published report (of which we are aware) which deals with the study of a metal-metal-oxygen system at 1700°C. Also, there has been no report on a complete metal-luugsten-oxgen system at any temperature. As a matter of fact. before this program started. The

Jan 1, 1965

By P. R. Sperry, M. L. Holzworth, P. A. Beck

The basic work of Z. Jeffries 1,2,3 has long ago established the main features of grain growth in the presence of a dispersed second phase. Working with sintered specimens of initially fine grained tungsten, to which various amounts of thoria had been added, Jeffries found that grain growth was inhibited, that is, practically prevented, up to a certain annealing temperature. When this temperature, which increased with the thoria content, was exceeded, extremely large grains developed abruptly from the fine grained matrix. such a a temperature" was later found by Gross-mann~3,14 in certain types of steel, while ~~i~ showed15 that in other steels grain growth was gradual, without inhibition and abrupt coarsening. Later work demonstrated the connection between coarsening and aluminum additions. The viewpoint that the direct cause of inhibition and coarsening in aluminum killed steels is a fine dispersion of aluminum oxide21 appears to be held quite generally, although some doubts have been voiced even very recently, 20 In steels with Ti additions the titanium carbide phase is considered responsible for the inhibition effects found.20 The coarsening temperature increases with the titanium content. That coarsening may occur, as a result of, certain heat treatments, even in low carboil rimmed sheet steel, was shown by Samuels. He also detected coarsening in a steel ingot of similar composition, after giving it the same heat treatment as that used by him for the sheet material. In this instance the identity of the inhibiting phase has not been determined. However, Tangerding found23 that even the few small carbide particles, which occur in carbonyl iron, have a very marked inhibitive effect. As a result of an oxidizing anneal at 850°C, large grains begin to form at the surface of the carbonyl iron specimen, and gradually grow inward as the subsurface oxidation of the carbide particles progresses and eliminates the obstruction. The carbide particles Can be removed also by annealing in hydrogen, with similar results. But if the oxidizing anneal of 10 hr at 850°C in air is followed between in hydrogen for 75 hr at the Same temperature, grain growth is more hibition usually the whole specimen is transformed into a single crystal (approx. 1.5 cm X 5 cm). The reversed procedure, that is, hydrogen annealing followed by oxidizing annealing leads to smaller grains. Tangerding arrived at the logical conclusion that there must have been in his specimens some additional obstruction to grain growth hibition from a minor impurity, other than carbon. It seems likely that this obstruction was caused by an oxide formed during the oxidizing anneal and gradually reduced rtain the subsequent hydrogen-annealing. Other instances of grain growth inhibition of varying severity, associated with a dispersed second phase, have been described by several authors. Notable examples: lead with 0.06 pet Cu or Ni,24 cartridge brass with 0.12 pct Cr,25 or with up to 0.15

Jan 1, 1949

By C. S. Smith

The partial persistence of grain size and grain shape on melting and resolidifying crystalline substances, as well as the general effects of pre-solidifi-cation and of superheating on nuclea-tion rate, have been attributed to the presence in the liquid of crystalline impurities bearing some structural relation to the principal solid and hence capable of serving as nuclei for its crystallization. That such lattice matching does sometimes occur is well established, but it is surely improbable that a compound possessing both high-temperature stability and the requisite structural similarity should be naturally available in almost every system. It is now suggested that a mechanism for producing appropriate nuclei exists in the crystallization of all but absolutely pure substances. Impurities can he of many kinds. Some will give rise to compounds that are less soluble in the solid crystal than in the liquid. A . crystal forming from a melt or solution containing these will be supersaturated in some degree, and the foreign atoms will commence to segregate during or after solidification. They do not necessarily form pure, unstrained, crystals of the precipitating compound with its normal composition and structure. Generally—at least in the metallurgical examples that have been most studied —the first recognizable nuclei have lattice coherency with the surrounding matrix and have practically the same crystal structure and lattice spacing as the material in which they form, differ- ing only in composition. An extremely thin oxide layer formed under appropriate conditions on or in a solid metal will tend to be coherent with the metal lattice, and the first few atom layers of the oxide in contact with the metal will differ in both composition and structure from massive crystalliue oxide. It will resemble the metal more closely than the oxide. If it did not subsequently change on melting, such an oxide film would constitute an almost perfect "template" or two-dimensional matrix to hold a few atoms of the primary metal in the exact array needed to form a nucleus for solidification. Any precipitating material will adopt the correct spacing over a few atom diameters, and with appropriate substitution of other atoms of the correct sizes a precipitate can match the parent lattice over large areas of cohererlce without excessive strain. In metals, the componentsof a host of oxides, silicides, borides, nitrides, carbides, sulphides, and other compounds are always present in minute amounts. Some of these will be eliminated because of extreme insolubility in the liquid, while others will remain soluble in the solid; occasionally, however, particles will be formed in the solid by coherent precipitation—not of pure substances, but of whatever assortment of atoms best satisfies the joint requirements of availability, affinity, and appropriate average atom size. Of these, many will disappear on melting, but a few will persist unmelted as little rafts of complex composition, maintaining in the liquid a surface of the exact geometry needed to nucleate the solid on subsequent cooling. Their number and size distribution will determine the resulting grain size; the dimensions of the largest will determine the degree of undercooling that can occur before solidification commences. A similar process may operate in the case of phase transformations in solids. The coherent precipitation of supersaturated minor constituents in a low-temperature phase should facilitate the nucleation of this phase on subsequent cooling after transforming into another phase at higher temperatures. At first sight this hypothesis may seem little different from the older ones. It should therefore be emphasized that it does not depend on chance to provide the right compound to form a nucleus, but postulates that every crystal, containing minute amourlts of any of a wide range of impurities, automatically engenders particles having the correct surface structure to serve as nuclei for subsequent solidification. It could operate in any system, metallic, organic, or inorganic. The suggested mechanism is closely analogous to the production of antibodies from antigens in living organisms—indeed, it was a description of the matrix theory of this process that suggested to the writer its crystalline analogue.

Jan 1, 1950

By E. T. Turkdogan, L. E. Leake

A number of measurements have been made on the vapor pressure of pure iron at 1600°C. Experiments were carried out by the transportation method in which a sample of iron is exposed in a furnace to a stream of pure argon, metered at a predetermined flow rate for a specified period of time. The vapor pressure was obtained from the weight loss of the sample. A detailed diagram of the apparatus used for the experiments is given in Fig. 1. It consisted of a horizontal molybdenum-wire-wound furnace fitted with a recrystallized alumina tube 1 1/8 in. internal diam. The boats, 1 by 3/16 by 1/8 in., were fired in argon at 1600°C to constant weight before use. After this treatment, subsequent heating of the boats to 1600°C gave little or no further change in weight, e.g. ±0.05 mg in 24 hr. The furnace and reaction tube were separately fed with argon from two independent flowmeters. These were connected to a common supply source via an argon purification furnace containing metallic titanium sponge at 700°C, and an anhydrone drying-tower. A rotary displacement-type meter was included in the gas train to provide a direct means of measuring the total volume of gas passed through the reaction tube. The material used was vacuum melted high-purity iron in the form of wire 2mm in diam. The following impurities were present in the iron: 0.008 pct C, 0.002 pct Si, 0.005 pct Mn, 0.008 pct S, 0.001 pct P, 0.01 pct Ni, 0.001 pct Cr, 0.006 pct Cu, and 0.001 pct Al. Samples consisting of three small pieces of this wire were laid end to end in a recrystallized alumina-combustion boat, and were separated by small fragments of alumina. The wire samples were then melted in argon to form spherical beads 2 to 3 mm in diam. After weighing, the boat with its contents was placed on the boat carrier and introduced into the cool end of the reaction tube. The entire furnace system was then flushed with argon for 30 min. At the end of this period, the flow of argon to both furnace and reaction tubes was adjusted to the desired level; then the boat was slid rapidly into the hot-zone of the reaction tube. After a known volume of gas had passed through the reaction tube, the boat was withdrawn quickly to the cool end of the furnace. When quite cool, the boat was removed from the fur-mace and placed in a desiccator for subsequent re-weighing. The semi-micro balance used in these experiments had a sensitivity of 0.005 mg. The above procedure was repeated at several flow rates within the range of 25 to 300 ml per min. The flow of argon in the annular space between the two tubes was at a rate of 300 ml per min in all experiments. During all the experiments the furnace temperature was maintained at 1600° ± 5°C by means of a temperature controller. The temperature of the reaction zone was measured with a Tinsley high precision disappearing filament pyrometer which was initially calibrated against a Pt-13 pct Rh/Pt thermocouple; the errors due to temperature measurements were within * 5°C. RESULTS Assuming that the iron vapor is in the monatomic state, the vapor pressure can be calculated from the weight loss of the sample and the volume of the carrier gas. The results obtained are plotted in Fig. 2 against the rate of flow of argon through the chamber containing liquid iron beads. At low flow rates, the thermal diffusion of iron gives an apparent high vapor pressure, but at high gas flow rates, the carrier gas is not saturated with the vapor, and therefore, low vapor-pressure values are obtained. This is now a well-known phenomenon. There is a distinct inflexion of the curve at about 120 to 160 ml per min rate of argon flow and the vapor pressure of iron within this

Jan 1, 1961

By J. E. Drury, W. M. Byers, E. J. Lynch

Measurements made with starch ~~~~s indicate that the Presence of starch serves to depress the streaming potential of a clay mud system. Measurements with pure starch slurries show that the addition of NaCl or CaO reduces the streaming potential significantly. The addition of NaOH, however, causes only a minor reduction in potential even when there are large changes in mud resistivity. INTRODUCTION In their excellent study of the streaming potential, Hill and Anderson' proposed that the presence of conductive solids within the filter medium will reduce the generated potential. This theory was demonstrated with measurements on shaly sandstone cores having various "b" factors; the cores with the higher negative b factors showed markedly lower potentials. (The factor b was originally defined by Hill and Milburn2 in an expression relating the resistivity of a shaly sandstone to the resistivity of the interstitial water: log F, = log F,, + b (log 100R,)'. 'he factor b, therefore, is a measure of the contribution of the conductive solids to the total conductivity of the sample. It has been found experimentally to be directly related to the amount of chemically exchangeable ion per unit pore volume.) In the logical extension of this hypothesis to the streaming potential created by filtration through mud cakes, they suggested that the presence of nonconductive starch in a drilling mud would have the opposite effect; i.e., it would cause the streaming POtential to be increased. The high streaming potentials that they observed with lime muds were attributed to this cause. A cross-plot of some of their results which are relevant to this discussion is given in Fig. 1. The lime-mud curves and untreated-mud curve are taken from curves given in their paper. The gyp-mud line was estimated from their statement that the measured streaming potential for gyp muds in the range of 1.5 to 2.0 ohm-m was approximately 18 mv/1,000 psi. These gyp-mud potentials were the lowest that they reported for any of the bentonitic-clay muds, and were the only ones which were less than the untreated-mud potentials. This very low streaming potential of gyp muds has also been observed in field studies.' By way of comparison, a U. of Texas publication4 gives the approxi-ate compositions for lime and gyp muds shown in Table 1. These materials are added to a bentonitic-clay mud base. It can be seen that both of these muds contain significant amounts of calcium salts and that the gyp mud has the higher concentration of starch. It would appear from this and from the data shown in Fig. 1 that starch may well be the cause of a streaming-potential reduction, rather than a potential increase. EXPERIMENTAL A series of streaming-potential experiments has been performed using a cell of a design similar to that reported by Wyllie.5 Calomel electrodes were used in both the mud slurry and the filtrate. Pressure in the cell was generated with high-pressure nitrogen. The primary objective of the investigation was to determine the effect of starch on the potential. Five different slurries were made which ranged in composition from untreated ben-tonitic mud (Baroid Aquagel) to a pure starch suspension (Baroid Im-permex). The intermediate slurries were prepared by mixing together a 9-lb/bbl clay mud with a 12-lb/bbl starch suspension in the ratios 1:3, 1:1 and 3:1. When starch was used alone, it was necessary to use an 18-Ib/bbl mixture, since leaner mixtures would not form a suitable filter cake. The clay mud used in the experiments was aged for three months before use. In addition to this series of measurements on the starch-clay mud mixtures, other experiments were performed to determine the effect on the potential of the absolute starch and Aquagel concentration, and the effect of adding various chemicals. RESULTS The effect of starch on the streaming potential is shown clcarly in Fig 2. Increasing percentages of starch resulted in n continually decreasing

By G. B. Thomas

Several methods of analyzing pressure build-up data in wells have been presented by various authors. This paper reviews the theory and method of D. R. Horner and presents example calculations performed on data obtained by testing several different types of wells. These calculations include, (1) graphical estimation of final static pressure, (2) determination of the productive capacity of the pay away from the well bore and. (3) the degree to which the formation adjacent to the well bore has been damaged by completion or other causes. The methods of testing and precautions which should be taken to assure the best data possible are discussed. Limitations and reliability of calculated results are also treated. INTRODUCTION Pressure testing of wells is generally limited to the determination of producing and static mean formation pressures. The so-called "static" pressure determination.. along with PVT, electric log and production data, enable the reservoir engineer to determine, within reasonable limits, the drive mechanism of the reservoir and in some cases. the amount of edge water encroachment. Producing pressure tests enable calculation of productivity indices and allow the engineer to plan the systematic production of a pool for optimum conservation of suhsurface energy. The radial flow formula advanced by Muskat' has been based on the assumption of incompressible radial fluid flow. It has been known that reservoir fluids do not behave in an ideally incompressible manner. For example, incompressible flow theory indicates a simple logarithmic relationship between the difference of the instantaneous and static well pressures when plotted against time. 'The latter stages of this type of plot of pressure build-up data generally show a marked devia-, tion from the earlier straight line trend. which deviation may be shown to be due to the compressible flow of fluids toward the well bore. Shut-in times of 24, 48. 72. or at most 96 hours are currently in wide use for determining so-called "static" reservoir pressure. Due to the continuation of compressible flow of fluids into the well bore long after this arbitrarily taken shut-in time, the determination of static pressures has almost invariably resulted in lower than equilibrium values. Materials balance calculations made early in the life of a reservoir often result in a calculated reserve which later observations prove to be too low. Failure to obtain reliable "static" reservoir pressures within the prescribed 24 or 48-hour build-up period has undoubtedly been a major factor in obtaining these low estimates. Comparison of theoretical and actual productivity performance has indicated that formation damage and not being able to attain true static pressures have been partially responsible for the observed discrepancies. Research into the theory of compressible flow behavior has resulted in methods of applying these theories to the testing of wells. Application of the developed theories to pressure build-up performance indicates that in many cases the following can be estimated. 1. The static reservoir pressure. 2. The in-place effective formation permeability away from the well bore. 3. The degree or extent by which the formation has been damaged adjacent to the well bore, either through completion methods or subsequent damage due to fluid entry. REVIEW OF THEORY Horner2 has shown that the pressure build-up within a "point source" well is approximated by the following formula: Equation (1) is the approximate "point source" solution to the radial compressible flow equation advanced by Muskat. The solution assumes the following: 1. A point source well is producing at constant rate from the center of an infinite reservoir with a constant pressure at its external boundary. 2. The fluid flowing is present in one phase only. 3. The compressibility and absolute viscosity of the fluid remain essentially constant over the range of temperature and pressure variation encountered. 4. The well i.; shut-in at the sand face arid there is 110 after production into the well bore. 5. The formation permeability is homogeneous in the direction of flow. From Equation (1) it can be seen that if an ideal well were shut-in while producing from a reservoir under the conditions assumed, the pressure build-up would be a logarithmic func-

Jan 1, 1953

By T. R. A. Gokcen

THE possibilities of separating and purifying metals by high vacuum distillation were examined by Kroll.1 He suggested vacuum treatment for the removal of zinc from the lead produced after Parkes desilverizing. The St. Joseph Lead Co. developed the first commercial vacuum dezincing process at their Herculaneum refinery, as described by Isbell.² The Broken Hill Associated Smelters at Port Pirie, Australia, has, after several years of pilot plant operations and fundamental investigations, developed a continuous process, which will be described briefly in a forthcoming publication." As the full-scale continuous vacuum dezincing plant at Port Pirie is still experimental, publication of full practical details of the plant will be deferred until the unit is operating as a normal part of the continuous refinery. This paper deals only with theoretical aspects of vacuum distillation processes, with particular reference to vacuum dezincing. The method of mathematical analysis is of general interest as it may be applicable to other metallurgical separations which have been investigated recently.4-6 Evaporation Processes At about atmospheric pressure, or higher, most liquids possess a boiling point—a temperature at which any heat put into the liquid is absorbed only as latent heat, not as specific heat. If a steady heat input is supplied, the liquid's temperature rises to this value, then remains constant while bubbles of vapor form beneath the surface. The rate of evaporation is determined solely by the rate of heat transfer to the liquid; the temperature of boiling is determined by the partial pressures of the volatile constituents in the liquid, and the total pressure above the surface. If the rate of heat transfer to the liquid is increased, the temperature remains constant, and the rate of boiling increases. When evaporating metals under vacuum, however, the partial pressures concerned are generally so small that boiling does not occur, because at even a fraction of a millimeter below the surface the hydrostatic pressure is usually too great to permit the formation of a pocket of vapor. In addition, the high thermal conductivity of metals tends to prevent the local superheating which is necessary for bubble formation.' Although this effect is doubtless also exerted when boiling metals at higher pressures, the magnitude will be less because the degree of superheat required to form bubbles is very much less at the higher temperatures involved. Under vacuum, therefore, evaporation of volatile constituents takes place only from the exposed surface, and the rate of evaporation depends upon the surface area, the surface concentration of volatile constituents, the surface temperature, and the partial pressures of volatile constituents immediately above the surface. If the heat input is raised above a certain level, the effect is not to increase the evaporation rate at constant temperature, but to raise the temperature of the liquid until at some higher level an increased rate of evaporation (and thus of latent heat absorption) again balances the heat input rate.? Many substances (including metals) have a very large intrinsic evaporation rate at quite low temperatures—far below their normal boiling points. However, at atmospheric pressure, the large numbers of atoms evaporated are almost completely deflected back into the liquid (or solid) surface by air molecules. Thus the back condensation rate is practically equal to the gross evaporation rate, and the net evaporation rate is practically zero. It can therefore be seen that an overall distillation rate depends not only upon the intrinsic evaporation rate, but also upon the ability of the volatile atoms to move away from the evaporating surface. This movement is facilitated by the provision of a condensing surface close by. Vacuum Distillation: The function of the vacuum above the evaporating surface is to remove foreign molecules, so that the chances of deflection of an evaporated atom back into its source are reduced. When the residual gas pressure is reduced so far that the evaporated atoms have a high probability of reaching a nearby condensing surface without suffering collision with a foreign molecule, the state of affairs is termed "molecular distillation." This process is practiced commercially today for the purification of numerous organic chemical products of high unit value, but not, to the writer's knowledge, for any metallurgical separation. When the degree of vacuum produced in a still is not sufficient to promote molecular distillation, then the evaporated molecules must diffuse through the

Jan 1, 1954