Search Documents

By K. J. Brondyke, P. D. Hess

Lack of correlation between densities of aluminum alloy samples solidified under reduced pressure (vacuum gas test) and hydrogen content of the metal is explained on the basis of inclusions serving as nuclei for hydrogen bubble formation. As a result of these findings, the vacuum gas test is suggested as a practical tool to indicate metal cleanliness and the potential effects of hydrogen. Visual obseroation of the sample during solidification under vacuum provides more useful information than that obtained either from density meas -urements or sectioning of the solidified sample. A rapid, practical method of detecting and measuring gas in molten-aluminum alloys has been sought for many years. The development of the vacuum gas test appeared to solve this problem. However, the vacuum gas test is subject to certain limitations, and the results must be interpreted mindful of these limitations, as described herein. Since the only gas which is appreciably soluble in aluminum is hydrogen, the terms gas and hydrogen will be used interchangeably in this paper. In the early days of the aluminum industry, when techniques were not as refined as they are today, it was customary to pour small "pancakes" of alumi- num on a metal plate. These were allowed to solidify under atmospheric pressure. Excessive gas in the metal was evidenced by the appearance of small bubbles on the surface of the "pancake".' It is obvious that such a method would only show very high gas levels far above any which could be tolerated by today's standards of quality required for most applications. Various attempts were made to develop methods which would be more sensitive. Among these were measurements of density of aluminum alloys solidified under controlled conditions, radiography of cast slabs, and examination of fractures. The method which showed the greatest promise and appeared most practical was the solidification of a sample of the molten aluminum alloy under vacuum. The samples so solidified are evaluated either by observation for bubble emission during solidification, by sectioning the solidified sample and examining for porosity, or by determining the density of the solidified sample. This test is known internationally as the Straube-Pfeiffer test or the reduced-pressure solidification test, but in America it is commonly referred to as the vacuum gas test2-' and will be so designated in this paper. The determination of density appeared an attractive and convenient means of providing a numerical rating for the samples. Attempts were made in some instances to correlate densities to hydrogen contents. Unfortunately, the visible loss of gas during solidification of samples with high gas contents precluded a correlation of vacuum gas test densities with the observed gas evolution. This led to Alcoa's use of two different pressures; 50 mm of

Jan 1, 1964

By D. H. Gurinsky, D. G. Schweitzer

The empirical equilibrium constantsd the heat of reaction for the reduction have been determined from 300° to 500°C. The mechanisms of the oxidation of uranium and magnesium from dilute bismuth solutions have been investigated. The kinetics of some of the reactions were affected by air adsorbed on the uranium oxides. THE Liquid Metal Fuel Reactor studied at Brook-haven National Laboratory uses a solution of uranium in bismuth as the fuel. Corrosion and stability problems require that the fuel solution contain corrosion inhibitors and deoxidants. Of the additives tested, magnesium has been found to be the best deoxidant and zirconium the best corrosion inhibitor in bismuth. Some experiments1 indicate that magnesium may play a secondary role in increasing the effectiveness of zirconium in inhibiting corrosion. The work reported here was intended to determine the relative concentrations of Mg and U required to prevent oxidation of the uranium fuel in the event of an air leak. In addition, an attempt was made to identify some of the mechanisms of the oxidation and reduction reactions occurring in the liquid bismuth solutions. EXPERIMENTAL Apparatus and Procedure—The equipment shown in Fig. 1 was designed so that the reaction kinetics could be studied as functions of temperature, melt composition, pressure, and flow rate. At the start of a run, the 10-liter bell jar (F) is removed by opening the bottom Dresser fitting (L), and the bismuth and additives are placed in a pyrex crystallizing dish (I). The bell jar is replaced, the system is evacuated to 10-5 to 10-6 mm Hg and then the heater (J) is turned on. When the selected temperature is reached, the oxidizing gas is admitted to the system to the desired pressure as read on the manometer (D). Gas is introduced through the slide tube (Q) which has a pyrex frit (R) sealed to its bottom. Stirring is accomplished by immersing the end of the tube in the melt. To maintain a constant pressure under flow the input and evacuation rates must be equalized. To obtain this condition stopcock (C) is closed and flowmeters (B) are matched. The reaction is "stopped" by closing the inlet flowmeter and quickly evacuating the system. To obtain a liquid metal sample, the system is evacuated with the sampler positioned near the surface of the liquid. The sampler, Fig. 2, is then immersed until its thermocouple reading matches thermocouple (G) Fig. 1. The system is then pressurized with helium which forces the solution through the frit. Twenty to 30 min elapsed between the time the system was evacuated and sampled. During evacuation and pressurization of the system, the temperature variations did not exceed ± 5°C. While bubbling occurred the temperature remained constant to within 1/2 oC. The volume of the equipment was determined by adding a known volume of air at atmospheric pressure to the evacuated system and noting the resulting pressure.

Jan 1, 1962

By Milton E. Wadsworth, R. Ted Wimber

Cobalt sulfate solutions containing ammonium acetate and chloroplatinic acid were reduced by hydrogen in a pyrex-glass lined autoclave in the temperature range of 170o to 232°C and hydrogen partial pressure range of 115 to 830 psia. The reduction rate was directly proportional to the hydrogen partial pressure and surface area of the pyrex glass and was independent of the quantity of chloroplatinic acid added initially. Experiments involving the variation of the relative concentration of ammonium acetate indicated that the reducible cobalt complex was probably the diacetate complex of cobalt, Co(AC)4H20, or a new mononcetate complex Co Ac, which was in solubility equilibrium with a pink precipitate of CO(AC)-4HzO. THE reaction in which a metal is dissolved by an acid to produce gaseous hydrogen and a salt solution was discovered early in the history of chemistry. In 1859 Beketoff found experimentally that this reaction could be reversed; i.e., a salt solution could be reduced by gaseous hydrogen to produce a metal and an acid. A review of work done on this phenomenon since that time may be found elsewhere., The hydrogen reduction of a cobalt salt solution is facilitated by complexing the cobalt ion. An ammonia complex of cobalt has been reduced commercially in the recovery of cobalt metal. A new reducible complex of cobalt was discovered5 when it was found that a co-baltous sulfate solution containing ammonium acetate could be reduced by hydrogen at temperatures in the region of 200°C. When a cobalt sulfate-ammonium acetate solution was heated to a temperature below its normal boiling point, a violet color became apparent, indicating complex formation. The nature of this complex was investigated5 by the addition of NH4Ac to a CoSO4 solution maintained at 85o. During the first additions of NH, Ac, the pH of the solution remained fairly constant at about 5.85. However, as the ratio of NH,Ac to CoSO, approached two, the pH rose and then leveled off at about 6.05. The absorption spectra of a Co(Ac), solution and a CoSO,, NH Ac solution were obtained at 85°C and were compared and found to be the same. These findings suggested that the diacetate complex of cobalt, Co(Ac),.4H20, was formed at 85°C. When a cobalt sulfate-ammonium acetate solution was heated to a temperature above about 165o, a finely divided pink precipitate appeared. The X-ray diffraction pattern of this precipitate indicated that it was Co(Ac), - 4H,O. In addition, it was discovered that when chloroplatinic acid, H,PtCl;, was added initially to the cobalt sulfate-ammonium acetate solution, a faster reduction was obtained. The roles of the solution complex, pink precipitate and chloroplatinic acid in the reduction process were then investigated. APPARATUS The experimental work was carried out in a two-liter stainless-steel autoclave. Adetailed description of the autoclave and the auxiliary equipment used in maintaining constant temperature and pressure may be found elsewhere.= Because the stainless steel was corroded, and also because it acted as a hydrogena-tion catalyst, an all-glass liner was fabricated such that the solution came only into contact with flame-polished pyrex glass. EXPERIMENTAL PROCEDURE The solutions used in the experimental work were prepared by dissolving reagent grade chemicals in distilled water. Although variation of the brand of ammonium acetate appeared to have no effect on the experimental results, CoSO, 7H O from the J. T. Baker Chemical Co. of Phillipsburg, N. J.,was found to give faster reductions than that prepared by Allied Chemical and Dye Corp., N. Y. The former was used throughout the course of the experimental work and was weighed up at the outset of each experiment. The ammonium acetate was dissolved to form a 6M stock solution, which was stored under refrigeration. A 10 pct solution of chloroplatinic acid (J. T. Baker Chemical Co.) was diluted to a 1.15 x 1Q2 M stock solution, which was standardized by precipitation of K,PtCl, as outlined by Scott. The appropriate volume of the chloroplatinic acid, H,PtCl,, solution was pipetted into the clean, dry glass liner. The cobalt sulfate-ammonium acetate solution, which had previously been saturated with

Jan 1, 1962

By P. Marier, T. R. Ingraham

When anhydrous cupric sulfate is heated in a stream of nonreactive gas, cupric oxysulfate is formed. When this reaction is complete, the cupric oxysulfate then decomposes to cupric oxide, which is the normal end product of reaction. The kinetics of each of these reactions has been studied using pellets prepared from finely divided cupric sulfate and from finely divided cupric oxysulfate. In each case, the reactant-product interface within the pellet is well-defined, and by normalizing the decrease in inter facial area with the weight fraction of the pellet decomposed it has been shown that the interface migrates into the pellet at a uniform rate at constant temperature. The activation energy estirnated for the decomposition of cupric sulfate is 57 ± 7 kcal per mole and that for cupric oxysulfate is 67 ± 8 kcal per mole. The rate of interfacial reaction is flow-sensitive, increasing with increasing flow of a sweep gas in the range from 50 to 2000 cu cm per min. The rate of decomposition is retarded by sulphur trioxide in the sweep gas. The relationship between sulfur trioxide Partial pressure and rate is consistent with the Langmuir Adsorption Isotherm governing the retention of sulfur trioxide in the interfacial layer. MOST of the work reported on the thermal decomposition of cupric sulfate and of cupric oxysulfate is related to the thermodynamics of the individual decompositions.'-4 No information is available on the rates of the individual reactions, on their activation energies, or on the effects of gas flow or of partial pressure of sulfur trioxide on the reaction rates. Some dubious conclusions based on relative reaction rates have been drawn from studies on the oxidation of sulfides. For example, on the basis of the products observed in a quenched system, it has been suggested5" that, when copper sulfide is roasted, the cupric oxide which is formed arises from the decomposition of either cupric sulfate or cupric oxysulfate. This is very difficult to reconcile with the thermodynamics of the C-S-O System,3 which support the probability that both cuprous oxide and cupric oxide precede cupric oxysulfate and cupric sulfate in the sequence of products formed during the oxidation of cuprous sulfide. In this paper the individual sulfate decomposition reactions will be examined and suggestions will be advanced for the mechanism of reaction. This will be based on the effects on reaction rate of changes in interfacial area, flow rate, and back-pressure of sulfur trioxide. EXPERIMENTAL Materials. The copper source material used in all experiments was Fisher Certified Reagent Grade anhydrous cupric sulfate, for which the following analysis was supplied by the manufacturer: chloride, 0.002 pct; alkalies and earths, 0.2 pct; insoluble, 0.002 pct; iron, 0.020 pct. Cupric oxysulfate was prepared by roasting cupric sulfate for 48 hr, with frequent rabbling, in an open tray in a muffle furnace maintained at 725°C. At this temperature, cupric sulfate develops about 0.16 atm pressure of combined sulfur trioxide, sulfur dioxide, and oxygen. Cupric oxysulfate develops only about 0.05 atm pressure, and is stable until

Jan 1, 1965

By M. J. Spendlove, H. W. St. Clair

The problem frequently arises, particularly in refining metals or smelting scrap metals, of separating metals in the metallie state. Many metals may be separated by taking advantage of their difference in vapor pressure. Such separations can be made at atmospheric pressure, but the separations are much more selective and can be carried out at considerably lower temperatures if the distillation is done at pressures of a few millimeters or less in an evacuated enclosure. Until recently, this has not been considered feasible as a metallurgical operation, but the recent improvemcnts that have been made in vacuum technology have broadened the applicability of vacuum processes and have prompted re-examination of low-pressurc distillation of metals as a practicable process. The distillation of zinc from lead is one separation that has already been reduced to practice.l This paper is the first of a series of studies being made on separation of nonferrous metals by distillation at low pressures. Although these experiments were confined to the separation of zinc from aluminum, the significance of the results is by no means confined to these two metals. The purpose has been to investigate a metallurgical technique rather than merely to devise a means of separating specific metals. The experimental work on distillation of zinc from zine-aluminum alloys at reduced pressure grew out of earlier work on distillation at atmospheric pressure.2 The earlier work indicated that it would not be practicable to decrease the zinc in the alloy much below 10 pct owing to the high temperature required. Therefore attention was turned to distillation ah low pressures, at which lower temperatures are required. After preliminary tests were made in a small, evacuated tube furnace, a larger furnace having a capacity of 100 to 150 Ib of metal per charge was constructed. Distillation tests were first made on pure zinc and then on aluminum-zinc alloys of various composition. Particular attention was given to the limit to which zinc could be reduced in the residual metal. Data were also taken on the rate of evaporation, and heat balances were made for both the crucible and the condenser. Distillation Furnace The vacuum-distillation unit is illustrated schematically in Fig 1. The major components are the induction furnace, the condenser, the vacuum system, and the power-conversion unit. Power is supplied to the induction furnace from a 50-kw 3000-cycle motor-driven alternator. The pressure in the furnace is reduced by a vacuum pump having a nominal pumping speed of 10 liters per sec. When in operation, the metal vapors travel upward from the furnace to the water-cooled condenser where they are collected in amounts of 50 to 100 lb. The condenser is removed with aid of an electric hoist. When the system is under vacuum, the condenser is made self-sealing by a rubber gasket between the smooth-faced, water-cooled flanges at the top of the furnace and the bottom of the condenser. The pressure of the atmosphere is more than sufficient to insure sealing. At the conclusion of an experiment, the residual metal can be removed from the furnace by removing the condenser and tilting the furnace with the electric hoist. The metal was cast into the molds carried on a mold truck. A photograph of the furnace and auxiliary equipment is shown in Fig 2. The details of the vacuum furnace are illustrated in Fig 3. The furnace proper is made vacuum-tight with rubber gaskets placed at each end of a fused quartz cylinder. A clamping plate at the bottom and a ring at the top are made to squeeze the rubber between the metal and the end of the quartz tube. A large graphite crucible placed inside the quartz cylinder is thermally insulated and physically supported by refractory insulating bricks. A thermocouple in a quartz protection tube is located at the bottom of the crucible: the leads pass through a rubber seal in the bottom plate. The supporting structure for the furnace is an angle iron frame with transite board sides. The condenser is made in the form of a water jacketed cylinder with an opening to the vacuum line at the top. The bottom has a projecting skirt inside the machined flange to provide additional cooling for the rubber gasket. Condenser sleeves are made in the form of two semicylindrical pieces of sheet metal that fit snugly inside the cooling jacket. The split sleeve facilitates removal of the condensate. Measurement of Temperatare and Pressure The metal temperature was measured by a platinum-platinilm rhodium thermocouple inserted in a well extending up into the bottom of the graphite crucible. During rapid evaporation there is a wide difference in temperature between the surface and the main body of metal in the crucible because of the large amount of heat that must be conducted to the surface to supply the heat of evaporation. The heat of

Jan 1, 1950

By A. Steiner, E. Miller, K. L. Komarek

Equations have been derived to calculate the chemical potentials of the components of liquid binary alloys from liquidus and enthalpy data. The equations are applicable to systems with intermetal-lic compounds of limited solid solubility. The liquidus curve of the Mg-Sn system was accurately redetermined, and the melting point of the compound Mg2Sn was found to be 770.5° * 0.3°C. The thermo-dynamic properties of the liquid alloys were calcu -lated from the liquidus data. The activity of tin shows both positive and negative deviations from Raoult&apos;s law and the relative integral entropy is positive. THE phase boundaries in an equilibrium phase diagram are functions of the thermodynamic properties of the coexisting phases. Specifically, the shape of the liquidus curve near an intermediate compound is determined by the properties of the compound and the liquid phase. Hauffe and wagnerl have derived an equation for the chemical potential differences (ui) of the liquid alloys near the congruent melting point when the heat of fusion of the compound and the liquidus curve are known. In their derivation the temperature dependence of the ui value was neglected, and liquid of stoichio-metric composition was chosen as the reference state. To obtain chemical potentials over the ntire composition range of the phase diagram with the pure components as the reference state, equations have been derived which incorporate the temperature dependence of the chemical potentials. Such a thermodynamic analysis can be carried out when the liquidus curve and the enthalpy values are known. The equations have been applied to the redetermined liquidus curve of the Mg-Sn system to obtain the thermodynamic properties of liquid Mg-Sn alloys. The Mg-Sn phase diagram has been studied by various investigators and the results have been compiled and critically assessed by Hansen.2 Preliminary thermal analyses showed that the basic features of the diagram were correct. The system has one congruent melting compound, Mg2Sn, two eutectics, and some solid solubility of tin in magnesium. However, the published liquidus temperatures show considerable scatter. In order to perform an accurate thermodynamic analysis, the liquidus curve between the magnesium-rich and tin-rich eutectic compositions was therefore redetermined by careful thermal analysis using highest-purity starting materials. EXPERIMENTAL PROCEDURE The magnesium metal (Dominion Magnesium Ltd, Toronto, Canada) had a purity of 99.99+ pct with the following impurities (in ppm): 20 Al, 30 Zn, 10 Si, <1 Ni, <1 Cu, <10 Fe; tin (Cominco Products, Spokane, Wash.) contained 99.999+ pct Sn. The graphite crucibles were machined from graphite rods (United Carbon Products Corp., Bay City, Mich.) which had an average total ash content of less than 30 ppm. All the graphite parts were baked out in vacuum at 950" to 1000°C for a minimum of 24 hr to remove volatile organic materials. The graphite crucibles (3 in. long, 1-5/8 in. OD, 1-3/8 in. ID) were tightly closed with a threaded cover (1/2 in. thick) which had a central thermocouple well (2-1/4 in. long, 5/16 in. OD, 3/16 in. ID). Cover and thermocouple well were machined from one piece of graphite. A thin sheath of quartz was inserted in the well to protect the thermocouple

Jan 1, 1964

By H. W. Mossman

Three aspects of magnetite smelting are discussed. The first is the working out of equilibrium conditions for eliminating sulfur. The second is the influence of magnetite solubility on the difficulty of tapping the reverb matte. The third is an approximation of the equilibrium conditions in the reverb gases which govern whether magnetite is mode or reduced in the reverb slag by these gases and by any iron sulfide in the slag. MAGNETITE has had a varied history in the Hurley Smelter since its start in 1939. Magnetite determinations on the smelter products are made regularly only on the monthly composite samples. Variations on the monthly averages are shown in Table I. Magnetite which drops from the slag and matte in the reverb has some slight bottom buildup which comes and goes, but no substantial accumulation from this source has been found at the end of a normal nine months&apos; furnace campaign. However, there has been some low grade magnetite bearing material mixed with considerable A1,0,, which has slid down from the bottom of the sloping flue between the reverberatory furnace and the waste heat boilers. This accretion has required drilling and blasting near the skimming end of the furnace. The magnetite has interfered with tapping at times. When the smelter was first started, tapping trouble from magnetite was extremely severe. Increasing the reverberatory furnace temperature by putting in an air preheater and a Dutch oven has helped greatly, although there still is occasional tapping trouble. When the present series of physical chemistry articles on copper smelting started coming out in 1950, they were read with interest, but no immediate application was seen for them. Results of some laboratory work in 1952 aroused a much stronger interest in this physical chemistry. A series of melts was made on some converter slags, which had magnetite in very large grain sizes, with the object of reducing the grain sizes in the slag, as it was known that it was easier to handle in the reverb in that condition. Anything done in the tests greatly reduced the grain sizes—even in the controls, where nothing was done except melt the slag and cast it. There was more magnetite in the slags after the tests than before, and with wide variations. There were no obvious reasons lor much of what happened in these tests. Much of the base material published in English in this field was made available for study. Recalculations were made on many of the type problems, and part of the data was reduced to local temperatures and compositions. Explanations were found for what happened in the 1952 series of tests on converter slags, and the same principles turned out to be a description of much of what magnetite does in the reverb. This article is to present the results of that study, from the viewpoint of applying the technical material in definite numerical form to the operating conditions in both the converters and the reverberatory furnaces at the Hurley smelter. Table I. Magnetite Variations on Monthly Averages, 1939 to 1955 Pet Magnetite Lowest Highest Average Converter slag 13.6 43.3 25.4 Roverb slaa 2.7 20.9 8.7 Matte 28 15.9 98 In general it was found that magnetite is made or reduced in both the converters and the rever-beratory furnace, depending on variations of temperature, matte composition, and reverb gas composition occurring in ordinary plant operation. Within reasonable limits, the field conditions for formation or reduction can be predicted, and probably can be set up and maintained. Converter conditions affecting magnetite formation can be put into numerical values better than for the reverb from purely technical calculations. The converter can be operated so as to keep the magnetite in the slag down to between 12 and 14 pct and still give satisfactory life for the converter brick. This depends upon having converter flux available which will make a slag with a good separation without raising the temperature too high. In the Hurley reverb and others with similar conditions, it is likely that a compromise of conditions will give a reasonably good control of combustion and still keep the magnetite from building up on the bottom. This discussion consists of three main parts. The first is the working out of the equilibrium conditions in the converter for determining in which direction the reaction 3 Fe,,O, (s) + FeS (1) F? 10 FeO (1) + SO, will go under actual converter operating conditions. The second deals with the influence of the solubility of magnetite in the slag and matte in the reverb on the difficulty of tapping matte. The third is an approximation of the equilibrium conditions in the

Jan 1, 1957

By Robert D. Pehlke, Arden L. Bement

A mass transfer coefficient for the removal of hydrogen from liquid aluminum by inert flush degassing has been determined experimentally at 700°C. A mathematical model has been derived, assuming transport control, which adequately describes the experimental data. The removal of hydrogen is shown to increase with decreasing bubble size, and with increasing flow of flushing gas. The presence of hydrogen in liquid metals in amounts which exceed the solid solubility at one atmosphere hydrogen pressure is highly undesirable, causing unsound ingots or castings, or resulting in difficulties during further fabrication. The removal of hydrogen from liquid metals can be accomplished by several methods, the principal two being vacuum treating and inert flush degassing. The design of a process involving either of these techniques is based on the mass transfer coefficient for hydrogen between the liquid metal and a hydrogen-dilute gas phase. In an effort to extend our knowledge in this area, an investigation of the rate of removal of hydrogen from liquid metals by inert gas flushing was undertaken. In view of the difficulties encountered with gas porosity in the casting of aluminum and aluminum alloys, and because of certain experimental advantages, liquid aluminum was selected for the study. PREVIOUS INVESTIGATIONS The removal of gases from liquid metals, particularly steel, has received considerable attention in recent years, The first definitive work of a theoretical nature was presented by Geller1 who considered the case where equilibrium exists between purging gases and metal bath during flushing, and also deviations from this ideal case. Hulme2 has surveyed a number of experimental studies and discussed some of the practical limitations of removing dissolved gases from molten metals. More recently a number of studies have been reported on degassing of steel melts by flushing with an inert gas.3&apos;4 Considerable work has also been reported on aqueous and organic liquid-gas systems, and an excellent summary prepared by Hovis.5 The treatment of liquid aluminum by a flushing gas to remove dissolved hydrogen has been known for several decades.6 A number of gases have been used for this purpose, including nitrogen, argon, and chlorine, or mixtures of these gases. Tikkanen and Erkko7 have studied the relative density changes of aluminum castings poured from melts degassed with chlorine, nitrogen, and mixtures of the two gases. Their

Jan 1, 1962

By Carl Wagner

AT elevated temperatures. most metals react with oxygen, sulphur, or halogen rather rapidly, although a coherent layer of the reaction product is formed and separates the two reactants from each other. The reaction proceeds, since one of the reactants or both dissolve and diffuse across the reaction product. In many cases the rate is diffusion-controlled as is indicated by the validity of the parabolic rate law of Tammann&apos; and Pilling and Bed-worth. wott and Gurney&apos; have suggested that diffusion processes also play an important part in the case of reduction reactions. This is confirmed by observations reported below. In general, however, the rate of reduction reactions is determined not exclusively by diffusion processes. Reduction of Silver Sulphide Above 179°C silver reacts with liquid sulphur very rapidly. Since the volume of silver sulphide is much greater than that of the equivalent amount of silver, a coherent layer of silver sulphide is formed in accordance with a rule established by Pilling and Bedworth.&apos; According to Wagner&apos;,&apos;&apos; silver ions and electrons migrate from the Ag-Ag,S interface across the silver sulphide layer to the Ag,S-S interface where the following chemical reaction takes place: 2 Ag +2 e + S Ag,S If silver sulphide is reduced by means of hydrogen, the equivalent volume of the reaction product, silver, is less than that of the initial substance. Accordingly, no coherent layer of silver is formed. Instead, fine filaments of silver grow from the silver sulphide surface into the ambient gas. This phenomenon has been investigated by many authors, especially by Kohlschiitter," who has also reviewed previous literature. Kohlschiitter has interpreted the observed phenomena by migration of "silver" dissolved in silver sulphide. According to Wagner,&apos; silver ions and electrons are the migrating particles. The sequence of consecutive steps involved in the reduction of silver sulphide is shown in Fig. 1. Every- where at the silver sulphide surface, hydrogen reacts with sulphur so that excess silver ions and electrons are formed, which migrate at random until they are captured by a nucleus of metallic silver. If such a nucleus lies on the surface and material is furnished only from the base, the original nucleus is pushed outward and thus a filament grows. Removal of sulphur and formation of metallic silver may also take place at different points when silver sulphide is reduced by copper. The reaction is usually formulated as: AgL,S + 2 Cu = Cu.S + 2 Ag but the actual reaction products are silver and a solid solution of the two sulphides. Primarily, excess silver and electrons dissolved in silver sulphide are formed. After a certain degree of supersaturation has been reached, nuclei of metallic silver may form. If, however, a silver foil is in contact with the silver-sulphide phase, as is shown in Fig. 2, silver ions and electrons will migrate to the Ag,S-Ag interface where they are captured. To confirm this scheme, a composite sample, with the geometry shown in Fig. 2, was made by pressing silver sulphide tablets 0.6 cm in diam and 0.4 cm high together with a copper cylinder and a silver foil in crucibles of iron, which does not react appreciably with silver sulphide at 400°C. After heating for 3 to 7 hr at 400°C. the specimens were cut normal to the phase boundaries. It was found that the thickness of the silver foil had increased from 0.01 cm as initial value to a maximum value of 0.06 cm. In the interior of the silver sulphide tablets only traces of metallic silver were observed. These observations show clearly that silver ions and electrons may migrate over distances of several millimeters before they form metallic silver. Reduction of Cuprous Sulphide by Iron The behavior of cuprous sulphide is very similar to that of silver sulphide, since the lattices of the high temperature modifications are identical, but in Cu.S there is a variable deficit rather than an excess of metal. To investigate the displacement of copper by iron, cuprous sulphide powder covered by a copper foil was pressed in iron crucibles and heated up to 800 °C for 3 to 7 hr. The thickness of the copper foil increased from an initial value of 0.005 cm to a nlaximum value of 0.029 cm. In the interior of the

Jan 1, 1953

By C. G. Bieber, R. F. Decker

The effects of minor constituents on the malleability of nickel alloys are described. These effects are related to the atomic diameter, valence, and position on the Periodic Table. The basic methods for removal or neutralization of deleterious elements are then reviewed. Finally, practical methods for melting nickel alloys are described in detail and related to minor constituent control. L HE first malleable nickel was produced by Fleit-man in about 1870 by the addition of manganese and magnesium to the molten metal to counteract the harmful effects of sulfur. Merica and Waltenbergl explained the mechanism by which these elements accomplished the desired result. They showed that sulfur in concentrations in excess of about 0.005 pct produced films of a brittle low melting eutectic of nickel-nickel sulfide which completely surrounded the grains. When manganese was added to molten nickel it reacted with the sulfur to form manganese sulfide which precipitated as widely scattered globules in the grain boundaries. The further addition of small amounts of magnesium converted the sulfur to a very stable sulfide of magnesium which was relatively insoluble in both the liauid and solid states and therefore occurred as finely divided particles within the grains where it had relatively little detrimental effect upon the malleability of nickel or its alloys. The practice of treating nickel with manganese and magnesium came to be called "deoxidation" and was employed without appreciable modification for the commercial production of wrought nickel and high nickel alloys (in this paper—50 pct or more nickel) until about 1930 even though the results were never entirely satisfactory. At that time a more extensive study was made of the effects of minor constituents on malleability and metallurgical quality. The results, small portions of which have been published,2-4 are presented here with other pertinent data. EFFECTS OF MINOR CONSTITUENTS UPON NICKEL Systematic additions were made of 0.5 pct or less of each of most of the elements of the periodic ta- ble. It was established in this study that some minor constituents other than manganese and magnesium were extremely influential. The malleability, as evaluated by the reduction of area in tension tests, depended roughly on temperature as shown in Fig. 1. Four temperature zones are of interest; the cold malleable zone (up to about 1000° F), the red short zone from roughly 1000° to 1500&apos; F, the hot malleable zone around 2000° F and the incipient melting zone. Most of the malleability problems are associated with the red short zone and the discussion will concern this mainly. The reduction of area in normal tensile tests on high-purity nickel (see curve A of Fig. 1) appeared to have no

Jan 1, 1962

By W. Hodge, A. F. Haskins, R. M. Evans

Refractory boron compounds are shown to resist corrosion by molten zinc. Coatings were made from ferroboron and manganese boron by several methods: welding, hard facing, and pack diffusion; and techniques of coatings are very important. Mechanical failure of the diffusion coatings can be partially eliminated by applying them to type 416 chromium steels rather than carbon steels. Welded coatings made with a tungsten arc are better than those made by other welding methods. Sintered compacts of mixtures of iron and chromium borides developed strengths of about 30,000 psi. They resisted corrosion of zinc at 600°C and oxidation at higher temperatures. MOLTEN zinc rapidly attacks common metals of construction. The rate of attack is greatly amplified at temperatures over 500°C. However, there are many applications where it is necessary, or at least advantageous, to use metal construction, such as for galvanizing pots and zinc smelter equipment. Condensers and agitator linings used in continuous zinc smelting processes are eroded and corroded by violently agitated molten zinc. Pumps, tubing, and other parts to permit transfer of molten zinc would be desirable in many zinc handling operations. The development of metallic materials which are resistant to molten zinc also might permit its use as a heat-transfer medium in modern high efficiency power plants. In reviewing previous work on zinc-resistant metals, it was evident that most of the available data refer to corrosion at temperatures just above the melting point. Tungsten and high Mo-Fe alloys containing more than 80 pct Mo resist corrosion by molten zinc.&apos;, &apos; Imhoff states that because chromium is not attacked by zinc, chromium plating improves galvanizing pot life." Nitrided cast iron has been used in die-casting machines&apos; where the molten alloy is 96 pct Zn and 4 pct Al. Other metals that have been suggested as of possible value for zinc resistance are columbium, tantalum, and titanium.V he corrosion of steel by zinc has been discussed by many authors,z, "2 None of these materials resisted corrosion by molten zinc at the temperatures used in this investigation, i.e., 600" to 700°C. I. Exploratory and Resistant-Coating Tests To make a study of zinc-resistant metals, exploratory tests were conducted to sort out roughly those materials which seemed most worthy of further development. This preliminary study included both solid metals and surface coatings. Exploratory Corrosion Tests in Molten Zinc—In initiating a study of this problem, numerous samples of metals and alloys were evaluated by dynamic corrosion tests made at 440 °C. Each metal was tested in its most readily available form, which, because of shape and size, was often unsuited for standard corrosion testing. The equipment used for dynamic corrosion testing of all materials, except in a few initial experiments, is shown in Fig. 1. Samples were tied to tungsten rods with tungsten wire, immersed in the zinc, and revolved at 12 rpm about the circumference of a 4 in. diam circle. Speed of travel was about 12.5 fpm. The molten zinc was protected from oxidation by an atmosphere of carbon monoxide, or, at a later period in the program, by a 9:1 mixture of nitrogen and hydrogen. When the effect of both molten metal and atmosphere was desired, the zinc level was lowered. Special high grade zinc was used in all of the corrosion tests. Corrosion losses for materials tested are shown in Tables I and 11. Whenever the materials were available in suitable form, the losses are reported in mils per year. It was recognized that weight losses on irregularly shaped specimens did not show the corrosion rate but the important consideration—poor resistance to corrosion—could be determined readily.

Jan 1, 1956

By Herbert H. Kellogg

Graphs representing the standard free-energy of formation as a function of temperature for 21 fluorides are presented, along with estimated values for the standard free-energy of formation of 20 atedadditional fluorides. A few of the many possible uses of these data in metallurgical calculations are discussed, including the fluorination of oxides, sulphides, metallurgicaland chlorides, and the reduction of metal fluorides. RECENT papers which have presented free-energy data for oxides, sulphides, chlorides, carbonates, sulphates, and silicates in the form of free energy vs. temperature plots,1-5 have found important uses in teaching and metallurgical development work. This method of presentation of thermo-dynamic data serves to make available in readily usable form thermodynamic constants which are scattered throughout the literature, and to show by graphical means the stability relations which exist between the compounds of a given type. It is certainly desirable to have these free-energy-temperature diagrams for as many types of compounds as are of present or potential use in metallurgical processes. Thus, even though there are but few uses of metallic fluorides in extractive metallurgy today, the potential uses of fluorine processes are of great importance and serve as the principal justification of this paper. Recent developments in the production of fluoro-carbons have resulted in vastly improved methods for the production and handling of fluorine and corrosive fluorides, with consequent reduction in costs.".&apos; The extractive metallurgist should no longer overlook possible fluorine processes when development of a new process for a relatively high unit-price metal is being considered. In a previous paperv he author outlined in some detail the importance and use of free-energy data in metallurgical calculations and also discussed the properties of a standard free-energy vs. temperature diagram. The same method of presentation is adopted in this paper, and the reader is referred to the earlier work for explanations of the use of free-energy data. Thermodynamic Data This paper presents standard free-energy vs. temperature diagrams, Figs. 1 and 2, for those fluorides for which the major thermodynamic constants are known. In the calculations of the standard free-energy of formation (AGO), values of the enthalpy of formation at 298°K (?Ho,) were taken from "Selected Values of Thermodynamic Constants"8 most of the values of standard entropy at 298°K (S°298) were taken from "Selected Values"," but in a few cases the values given by Kelley9 ere preferred; data for the Phase changes came from "selected Values" and several of Kellley&apos;s publics- tions;0-12 and most of the specific-heat data were taken from Kelley.l0 The appendix gives a detailed list of the sources of the data. As explained in an earlier paper,6 he accuracy of some of the standard free-energy relationships is not high. An error of +1 kcal at 1000°C is expected in most of the curves of Figs. 1 and 2. The error may be as high as +3 kcal in some of the curves. Rather than attempt to estimate the reliability of each curve, the source of the data used is reported in the appendix, and the reader may compare these data with newer values as they become available. Figs. 1 and 2 fail to show values for the fluorides of many important metals. In most cases this is because the entropy of the metallic fluoride is unknown. For preliminary calculations it is often useful to have available the best estimate of an unknown datum, rather than no information at all. With this in mind, Table I lists estimated standard free-energy values for a number of important metallic fluorides not included in Figs. 1 and 2. The values reported in Table I were, calculated by Brewer et al.13 and are based on estimated entropy values. The estimated uncertainty of the values for each fluoride were calculated from the uncertainty values given by Brewer1&apos; for ?H°, and the entropy function.

Jan 1, 1952

By J. E. Powell, F. H. Spedding

WHILE rare earths are reported to be widely distributed in nature and are not really rare," in practice, there are only a few minerals which are sufficiently rich in rare earths to serve as practical sources. Perhaps the best known of these is monazite which is a phosphate mineral containing rare earths and thorium. This mineral occurs as a dense brown sand in gravel beds and is particularly rich in the light rare earths of the cerium subgroup. This mineral is processed commercially for its thorium, cerium, and lanthanum content, and, consequently, furnishes rich concentrates from which neodymium, praseodymium, samarium, europium, and gadolinium may be obtained. Unfortunately, monazite is rather lean in rare earths heavier than gadolinium. A second mineral which is rich in the light rare earths is bastnasite, a fluoro-carbonate. Extensive deposits of this ore have been discovered in the western United States and have received considerable newspaper publicity in recent years. While bastnasite is very rich with respect to cerium, lanthanum, and neodymium, it contains even less heavy rare earths than does monazite. One of the better sources of heavy rare earths of the yttrium subgroup is gadolinite, a black silicate rock from which the rare-earth content can be extracted readily by acid leaching. It is obtained chiefly from Norway at the present time, although there are known deposits in the United States. Other sources of heavy rare earths include fergu-sonite, euxenite, and samarskite which are refractory tantalo-columbate ores. These minerals require caustic fusion or reduction to carbides with carbon before the rare-earth content can be extracted. All of the minerals which are rich in the heavy rare earths contain yttrium as a major constituent. After the rare earths have been extracted as a group from an ore by chemical means, it is generally convenient to precipitate them from acid media with oxalic acid in order to eliminate certain non-rare-earth impurities such as iron, beryllium, etc., which are usually present. The oxalate can then be readily ignited to R2O3. The oxide can be dissolved in acid and is the starting point for subsequent separation into the pure components. Perhaps the principal reason why the rare earths have not been studied as extensively as other elements of the periodic table, whose natural abundances are comparable, is that they are extremely difficult to separate from each other by the usual chemical means. Prior to 1945, the separation of one trivalent rare earth from another was a laborious process. All separations were based on repeated fractionation processes, i.e., fractional precipitation, fractional decomposition, fractional crystallization, etc. These processes were repeated from a few hundred to many thousands of times in order to obtain individual rare-earth salts of reasonable purity. Of course, mention should be made that, in the few cases where a rare earth could be oxidized or reduced to a valence state other than three, more conventional chemical means could be utilized to separate the oxidized or reduced ion from the other normally trivalent rare earths. The ionic states which deserve special mention are CeIV, SmII, Eu11, and Yb11. When it is possible to remove an element of the series efficiently, due to an optional valence state, its immediate neighbors also become easier to isolate. For example, binary mixtures of lanthanum and cerium, and praseodymium and cerium can be obtained by a relatively small number of fractional operations. The tetravalent state of cerium then allows the complete resolution of the binary mixtures by ordinary chemical means. Although the tetravalent state of cerium has been known for a long time, the divalent states of samarium, europium, and ytterbium were not used extensively in separations prior to 1930 because they are relatively unstable in aqueous media.&apos;-" No attempt will be made to give a comprehensive review of the extensive literature dealing with the separation of rare earths. Rather, this paper will be confined to a review of those methods which have been developed at Iowa State College during recent years, and which have proved extraordinarily successful for the isolation of highly pure rare earths in quantity. It was obvious that, if pure rare earths were to become generally available, methods would have to be developed wherein the thousands of fractional operations made necessary by the similarity of rare-earth properties could be performed automatically. The development of chromatographic techniques and ion-exchange resins appeared to offer a mechanism by which this objective could be accomplished. A number of early attempts were made to separate rare earths by these means; for example, Russell and Pearce12 passed a mixture of rare earths through a cation-exchange column and reported

Jan 1, 1955

By A. L. Labbe, J. J. Donoso

Hard-won experience in the operation of smelting plants has pointed the way to the most efficient design and economical operation of a baghouse system for recovery of metallurgical fumes. This paper treats of the principles governing bag filtration and the factors most directly affecting bag life. Methods for temperature conditioning of smoke previous to filtering are described and the evolution of bag shaking mechanisms is brought up to date. THE history of the bag filter process, as applied to the recovery of metallurgical fumes, goes back to 1876, when the Lone Elms Works of Joplin, Missouri, erected a two-section baghouse for filtering fumes from Scotch Hearth lead smelting furnaces. By modern standards this installation was crude, but the basic operating principles remain essentially the same today. Great progress, however, has been made in the mechanization of the process by the use of devices for mechanically and automatically cleaning bags of adhering fumes, in the use of instruments for recording and controlling the variables affecting baghouse operation, and in the conditioning of the smoke previous to filtering. The principles governing bag filtration of fumes were investigated in 1929 by A. J. Seitz of American Smelting and Refining Co., and he arrived at the following conclusions: 1. Pressure on Bags at Constant Volume: The rate at which the pressure builds up in a bag with constant volume is directly proportional to the time, with the same kind of smoke and dust burden (fig. 1). 2. Pressure on Bags for Change in Volume: With equivalent smoke and dust burden, the bag pressure increases with the square of the increase in volume. The rapid increase in pressure is due to (1) increase in volume, (2) increase in amount of dust on the bag, and (3) more intense packing of the dust at higher velocity (fig. 2). 3. Effect of Intensive Filtering on Pressure: Since the pressure builds up directly as the square of the volume, with the same smoke and dust burden, it is necessary to shake the bags in that increased ratio, and vigorously, to keep the pores of the bag open and pressure down. This increases the wear on the bags and the dust losses. 4. Effect of Shaking on Dust Losses: The losses are in direct proportion to the frequency of shaking because the loss takes place immediately after shaking, before the bag seals itself. Tests show that with hand shaking the last 20 pct of the removable dust requires seven times more shaking per unit weight of dust than the first 80 pct. 5. Effect of Fineness of Dust on Losses: The fineness of the dust particle increases the loss in two ways: (1) When the bag is clean a greater percentage of it will pass through before the bag seals itself; (2) fine material penetrates farther into the bag pores, building up pressure faster and necessitating more shaking, thereby increasing the loss. The physical principle of operation of the bag filter is simply the separation of solids from a gaseous carrier through a porous medium, and is analogous to the filtration of solids from liquid carriers. Basically, the mechanics of the process are as follows: Fume-laden gases drawn from the operating fur-

Jan 1, 1951

By Alex Boozenny

The results of potential measurements between 1) a titaniurn electrode in NaCl-TiCl, melts and a graphite-cizlorine reference electrode and 2) a titanium electrode in NaC1-Nu "melts and a graphite-chlorine reference electrode show that the following relationship expresses the equilibrium relationship between the TiCl, and the dissolved metallic sodium concentrations, expressed in mole fraction units, in dilute NaCl welts at 850°C. The results of potential measurenments between a graphite electrode in NaCl-TiCl, melts that were saturated with TiCl, and a graphite-chlorine reference electrode, when combined with the standard free energy data given in literature, gave the following expression for the equilibrium relationship between the TiCl, and TiCl, concentrations, expressed in mole fraction units, in dilute NaCl melts at 850°C. These relationships are for melts in equilibrium with pure titanium. In impure systems, less sodiurn and more TiCI, can be copresent with a given concentration of TiCl,. AFTER 10 years of existence, the titanium metal industry still relies on the original Kroll and the sodium reduction processes for the only important means of converting titanium-bearing ores to titanium metal via production of titanium tetrachloride as an intermediate. Through years of modification and development these two processes have evolved to a point where any further major economic advances in the field of titanium metal production must be achieved through implementation of processes which are distinct departures from simple magnesium or sodium reduction of titanium tetrachloride. Concurrently with the refinement of the Kroll and sodium reduction processes, industries and government have sponsored research and development in the field of producing titanium by electrolysis of oxides,&apos; carbides,&apos; and halides3&apos;4 in fused salt media. However, the incomplete understanding of the physical chemistry of fused salt systems has in many cases hindered the development of promising new titanium production processes to the point where they will have a clear advantage over the processes presently employed. Fortunately the volume of information on fused salt physical and electrochemistry is increasing at an accelerating pace. The first approaches tbward investigating the electrochemical behavior of fused salt systems have been the measurements of equilibrium potentials of metals immersed in melts containing their halides.&apos; Some work on the kinetics of fused salt electrochemical processes has been reported.&apos;09" Some descriptions of the structural composition of fused salt systems are also available. The work described herein is a continuation of the study of metal electrode potentials in fused salts. Specifically, this contribution to the growing store of fused salt chemistry information presents a description of the equilibrium relationships between the concentrations of titanium dichloride, titanium trichloride, and metallic sodium in fused sodium chloride. The investigation of the titanium dichloride-titanium trichloride equilibrium relationships is not completely new. Kellog and Krye and Melgren and opie15 have presented the results of chemical analysis of salts containing titanium chlorides. Flengas and lngrahaml&apos; have published the results of an investigation of these equilibrium relationships obtained by measurement of electrode potentials in a fused NaCl-KC1 mixture. The investigation described herein differed from that of Flengas and Ingraham principally in the use of straight NaCl as the solvent in place of the more costly and more difficult to purify mixture. This investigation was conducted in three parts. First, the equilibrium potentials of titanium in sodium chloride containing known concentrations of titanium dichloride were measured. A graphite-chlorine electrode was employed as reference. Second, the equilibrium potentials of titanium immersed in sodium chloride containing known concentrations of elemental sodium were measured. Third, the equilibrium potentials of graphite immersed in titanium tetrachloride-saturated sodium chloride containing known concentrations of titanium trichloride were measured. The results of these three related investigations are presented in the above order. The results are combined with the thermodynamic properties of titanium chlorides and sodium chloride, obtained from literature, to give the titanium dichloride-titanium trichloride-elemental sodium equilibrium concentration relationships. In the test, whenever a melt contains titanium chlorides in predominantly divalent form, the melt

Jan 1, 1962

By J. R. Stone, G. D. Storm

Design features and operating methods of ASARCO&apos;s new unit for the continuous melting and casting of tough pitch copper at Perth Amboy are described. Preliminary studies made for determinitzg economic feasibility and design details are also presented. Cathodes are preheated with oil and melted in an electric furnace at 40-ton per hr rate, from whence the molten copper is fed simultaneously to the wire bar casting wheel and the continuous casting of cakes and billets. In the early 1950&apos;s the copper refining furnaces at ASARCO&apos;s Perth Amboy plant reached a point where major expenditures would have to be made for replacement of the furnaces and the auxiliary equipment. The problem was whether to continue with the standard reverberatory furnaces or to consider continuous melting in electric furnaces. Reverberatory furnaces have been used for years in the copper industry and for some 50 years at the Perth Amboy plant. Reverberatory melting practice, however, imposes certain restrictions. First, labor costs are high, usually requiring three shifts per day. Secondly, they operate on a batch basis, the conventional operation being a 24-hr cycle including

Jan 1, 1961

By J. W. Byrkit

Design features and operating methods at the new Ajo smelter are described in detail. Successful operation of a novel method of handling and charging wet concentrates to a deep bath type reverberatory furnace contribute to the daily production of 200 tons of anodes with good results from the standpoint of both metallurgy and economy. THE New Cornelia Branch of the Phelps Dodge Corp. is located at Ajo, Ariz. Large scale mining operations were started at Ajo in 1917, when a 5000 ton leaching plant was put in service to treat the copper carbonate ore that overlaid the sulphide ore-body. In 1924 a 5000 ton flotation plant was built to treat the sulphide ore and the leaching operation was abandoned a few years later. Changes in practice and additions to plant facilities have resulted in a gradual increase in the milling rate which now approaches 30,000 tons daily. Prior to the erection of the Ajo smelter, flotation concentrate was shipped by rail to Douglas, Ariz, where it was treated in the Phelps Dodge smelter. Construction of a one reverberatory furnace smelter to produce an average of about 200 tons per day of copper anodes at Ajo was started early in 1949. Initial heating of the reverberatory furnace was begun on June 21, 1950. Charging the furnace was started on july 8, and the first anodes were cast on July 14. Neither the anode furnace nor the reverberatory furnace has cooled since the initial firing. Fig. 1 is a photograph of the new smelter. Built primarily to eliminate the &apos;300 mile haul of concentrate to the Douglas plant, the new smelter was designed to treat the Ajo concentrate, with a minimum of capital investment. A novel, and relatively simple, method of concentrate handling and furnace charging made unnecessary the installation and operation of expensive storage facilities and reclaiming equipment, and obviated the necessity of holding in storage large quantities of copper-bearing materials. (Patents are pending in the United States and abroad on the Ajo process.) In the selection and arrangement of equipment, consideration was given to the full utilization of all smelting facilities, reducing the amount of idle equipment to a minimum. The layout of the smelter is shown in Fig. 2. The important problem of internal transportation was minimized by arranging the reverberatory furnace, converters, and anode furnace in a compact group and providing storage bins inside the smelter building, within reach of overhead cranes, for fluxing materials and other supplies needed in the daily operation. This can be seen in the sectional view of the smelter given in Fig. 3. Incorporated in the design were many innovations in smelting equipment intended to facilitate operations and effect economies in maintenance expense and manpower requirements. Table I gives operating data of the entire plant. Metallurgy The metallurgical practice at Ajo is based essentially upon the use of a single, deep bath reverberatory furnace for smelting wet concentrate, without concentrate storage facilities. The large reservoir of slag and matte maintained in the furnace serves to equalize, to some extent, the day to day variations in the nature and grade of concentrate and permits

Jan 1, 1954

By A. W. Sommer, H. H. Kellogg

It is shown that SO3-O2 mixtures react with sphalerite at an appreciable rate ill the temperature range of 361° to 527°C to fornz ZnSO4. The rate of reaction follows a parabolle lax. Oxygen, or O2-SO2 mixtures, have a negligible effect on sphalcrite in the same temperatuee range. PRACTICAL roasting of sphalerite is usually performed at 800°C or higher. The calcine is composed of zinc oxide unless the roaster atmosphere contains relatively large amounts of SO2 and the temperature is held to 800oC or lower, in which case zinc sul-fate may also form. Ong, Wadsworth, and assell&apos; have made a careful kinetic study of sphalerite oxidation under these "normal conditions" and have concluded that the rate of the reaction is controlled by the decomposition of an activated complex adsorbed on the sphalerite surface. They found the rate of oxidation to be small at 700°C. Extrapolation of their data to 400°C would indicate an almost negligible rate at this temperature. In our work sphalerite was reacted with air, mixtures of SO2, O2, and N2, and mixtures of SO3, O2, and N2 in the temperature range 361o to 527OC. Negligible rates of oxidation were found, except for the gas mixtures containing SOs. With this latter gas, the oxidation of finely divided sphalerite was fairly rapid and zinc sulfate was the product. Based on the limited evidence available, the rate of sphalerite oxidation by SO3 is postulated to be controlled by gaseous diffusion through pores or cracks in the zinc-sulfate coating. This evidence for the direct reaction of sphalerite and SO, at low temperature may prove of importance to the understanding of zinc-sulfate formation in the dust-collecting equipment usually associated with zinc roasters. The roaster gas carries off fine dust, much of which may be unreacted sphalerite. The temperature in dust-collecting equipment drops from the roaster temperature (about 800°C) through the range of temperature we studied, to ambient temperature. Such dusts are known to contain far larger amounts of zinc sulfate than the primary calcine. It has been assumed in the past that this sulfate is formed by reaction of ZnO dust with SO3 (or SO, plus 0,) in the partly cooled gases. Our work shows that an alternative possibility exists—the direct reaction of SO3 with ZnS dust. EXPERIMENTAL Reaction rates for sphalerite at 350° to 600°C are relatively small so that it was necessary to use finely divided material with a relatively large surface area in order to obtain measureable amounts of reaction. In the preliminary experiments, pure mineral sphalerite, ground to pass a 325-mesh sieve, and chemically precipitated ZnS (in the form of an impalpable powder) were used. In the quantitative rate experiments the ground mineral sphalerite was processed in an Infrasizer to obtain a product with particle size between 9 and 18 µ. The analysis of this material was 66.7 pct Zn, 0.09 pct Fe, and 32.7 pct S (theoretical sphalerite is 67.1 pct Zn, 32.9 pct S). The sized powder was carefully mixed and divided into l-g samples. The apparatus used for the quantitative rate measurements is shown in Fig. 1. For the preliminary experiments the following change was made: The external catalyst furnace (B, C)* was not used. *Letters refer to Fig._______1 .___- Rather, when catalysis of the SO2 + O2 reaction was desired, the glass beads in basket M were replaced by vanadium-oxide pellets. Procedure—A l-g sample of ZnS was placed on a shallow stainless-steel tray, I, and spread evenly to a depth of about 1mm. The tray was placed in the reaction assembly and the assembly inserted in the cold furnace. The furnace tube was flushed with dry argon and then heated to the desired reaction temperature. The temperature was controlled to ±0.5oC. Gas mixtures (O2 + N2 or O2 + SO2) were prepared from tank gases by means of a mixing device identical to that employed by Darken and Gurry.2 The total flow rate of gas was 205 ml per min to point A of Fig. 1. The accuracy of the individual and

Jan 1, 1960

By H. Sigurdson, C. H. Moore

Extensive studies have been carried out on electric furnace and blast furnace slags obtained in the winning of iron from its ores. These slags normally consist of elements of the gangue minerals present in the ores, as well as the added flux materials. In consequence, melts of CaO, MgO, Al2O3 and SiO2 can be considered as representing typical slag compositions. When a slag of this composition cools, it usually crystallizes according to predictions possible from an equilibrium diagram of these constituents, providing the melt is not undercooled to form glass. The melt is either viscous or fluid, depending upon the ratio of binary cations to silica, and crystallizes easily or forms a glass for the same reasons. If the melt is not overheated so that carbides of the metal components of the slag are formed and if the composition of the slag is so adjusted that it has a high fluidity, liquid equilibrium is attained and the slag can be held in a liquid state for extended periods of time. Upon tapping, the slag crystallizes into minerals, the type and proportion of which are determined by the melt composition. Since equilibrium is attained, the holding period is not critical. In melts containing a large increment of titanium, however, the normal slag procedures are not applicable. Titanium, as one of the atomic transition elements, is, at elevated temperatures, capable of being reduced to form metalloid compounds much more readily than the refractory oxides present in normal slags. In consequence, an oxide melt containing titanium never reaches equilibrium in a reducing environment, but continues to shift its composition until cooled. If melts of this nature are cooled and samples submitted to metal-lographic and X ray analysis the course of reaction and crystallization in this type of slag can be determined. Preparation of Slag The slags investigated fell into the system CaO-MgO-TiO2-Al2O3-SiO2 and were produced from ilmenite ores reduced by carbon in an electric furnace. Since the equilibrium series1 and the laboratory smelting of ilmenite2 are described in two of the accompanying papers, detailed description of the smelting procedure is not required here. However, certain essentials must be mentioned. Two types of melts were used to produce slags studied in this investigation. The first series of smelts made to determine proper flux addition were produced in a 4 lb Ajax induction furnace. The charge, consisting of ore with the proper flux addition, was heated in a graphite crucible until fluid, held fluid for a sufficient time period to obtain 1-5 pct FeO content, and poured. Because of the small size of the charge only the final sample of these melts could be examined. In the melts made in the 50 lb arc furnace, however, grab samples taken at 10 min. intervals between time of initial melting and final pouring were available for examination. These samples allowed a much clearer picture of the course of reaction and crystallization. amounts of ferrous oxide and reduced titanium compounds is opaque to transmitted light. Therefore, all petro-graphic studies had to be made on polished slag sections. A representative sample of slag was cut or broken, mounted in a thermosetting plastic, ground flat using 400 grit silicon carbide, the coarse scratches removed with 600 grit silicon carbide and polished on billiard cloth using levigated alumina. Rouge was avoided because of the entrainment of the red particles in pores in the slag, causing a possible confusion with some of the mineral phases. In order to prevent sample projection above the plastic surface red bakelite was used to hold the sample, and backed up with clear lucite. In this manner sample labels could be permanently retained in the mounting. The polished samples were examined on a Bausch and Lomb metallograph at magnifications of 250 X, 500 X, 1000 X and 1800 X. The instrument was equipped for examination of specimens under bright field illumination and with crossed nicols. A magenta tint plate to aid in color tone differentiation was also used. Petrology of Slags In order to determine the composition and mineral relatinos of a previously unreported system petrologically, it is essential that the starting composition, reaction temperature and final composition be known. The chemical composition of the ilmenite ore used in these smelts is given in Table 1, and the complete analysis of a typical high titanium, low iron slag is given in Table 2. In the winning of TiO2 from ilmenite by a smelting process it is necessary to produce a slag which will melt at an economically feasible temperature, remain molten as the iron is removed by reduction, be fluid enough to be readily removed from the furnace, contain a high percentage of TiO2 and a low percentage of reduced titanium com-

Jan 1, 1950

By Arnulf Muan, D. P. Masse

PHASE relations in the system CoO-SiO2 have been determined as a basis for further investigations of thermodynamic properties of olivine solid solutions involving Co2SiO4 as a component. Previous data on the system CoO-SiO2 are incomplete or uncertain. Biltz and Lemke1 determined the melting point of cobalt orthosilicate as 1345°C , but Asanti and Kohlmeyer2 have later found a considerably higher temperature, 1420°C. The latter authors also studied melting relations of one mixture on the CO side and two mixtures on the SiO2 side of the orthosilicate composition and sketched a tentative phase diagram for a limited composition range of the system. However, no phase identification was carried out, and the authors left unanswered the question of the possible existence of a stable meta-silicate phase. Greig3 showed that a 90 wt pct SiO2-10 wt pct Co mixture at 1725°C consists of two immiscible liquid phases. The quenching technique was used in the present investigation. Mixtures of high-purity cobalt oxide ("Fisher Certified") and dehydrated silicic acid ("Baker Analyzed") were equilibrated in air atmosphere. The samples were then quenched to room temperature and the phases present were identified by microscopic and X-ray examination. The samples were kept in small envelopes made from thin (0.0004 in.) platinum foil. Thermodynamic data which have become available recently for Pt-Co alloys4 were used to check that the losses of COO from the oxide samples by alloying of cobalt with platinum in the present investigation were too small to change significantly the compositions of the oxide material during the equilibration. A vertical tube furnace with an 80 pct Pt 20 pct Rh resistance winding was used in runs up to 1510°C. Temperatures in these runs were measured with a Pt vs 90 pct Pt 10 pct Rh thermocouple calibrated against the melting point of diopside (CaMgSi2O6, 1391.5?C). The given temperatures as measured by this technique are estimated to be accurate to *5°C. Quench runs at temperatures above 1650°C were made with a modified Roberts and Morey5 strip furnace, using strip resistors composed of a 60 pct Pt-40 pct Rh alloy. Temperatures were measured with an optical pyrometer which was calibrated against the liquidus temperature of a mixture composed of 10 wt pct CaO, 90 wt pct SiO2 (1707°C). The temperatures measured with this technique have an estimated accuracy of +15°C. The results are shown graphically in Fig. 1. Only one intermediate phase, the orthosilicate Co2SiO4 with olivine-type structure, is stable in addition to the end members cobalt oxide and silica. The melting point of the orthosilicate was found to be 1415" ± 5°C. The metastilicate CoSiO3 is not stable at the temperatures of the present investigation. The two eutectic points in the system CoO-SiO2 where COO plus olivine, and olivine plus silica, coexist with liquid were found to be 72 wt pct COO and 1407°C and 63 wt pct COO and 1381°C, respectively. The melting point of cobalt oxide in air was taken as 1745?C, based on the previous data of Aukrust and Muan.6 Within limits of error (+ 5?C), identical eutectic temperatures to those determined in air were found when two representative mixtures were equilibrated at 1 atm O2 pressure and in an atmosphere of 84 vol pct CO2, 16 vol pct H2. This suggests that changes in oxidation state of cobalt in the condensed phases are not significant as far as their effect on the phase relations are concerned. This inference was substantiated by failure to detect7 any "excess oxygen" (i.e., oxygen in excess of that contained in Co2SiO4) in a sample of orthosilicate composition equilibrated in air 10°C above the liquidus temperature and subsequently quenched to room temperature. This work was carried out as part of a research program sponsored by the U.S. Atomic Energy Commission under Contract No. AT(30-1)-2781.

Jan 1, 1965