Search Documents

By R. F. Mehl, C. S. Barrett, A. H. Geisler

THE studies of the mechanism of precipitation and of the resulting property changes in aluminum-silver alloys1-3 have presented some new concepts of the aging reaction-concepts that may be fundamental in establishing a complete quantitative theory of age-hardening. With both aluminum-copper and aluminum-silver alloys, thin platelike aggregates form parallel to the Widmanstätten planes of precipitation.1,4 These platelets grow in lateral dimensions and in thickness with aging time until they are thick enough to produce an X-ray diffraction line pattern. Their structure is that of a transition lattice which is coherent with the matrix, since the atomic spacings in the matrix and the precipitate are the same on each side of their common interface.2,4 The equilibrium precipitate forms later from and at the expense of the transition lattice. Since age-hardening occurs during the formation and growth of the transition lattice, the major contributor to the hardening no doubt is the stress condition in the matrix resulting from the matrix holding the precipitate in the coherent condition. The development of these stresses also appears to be responsible for an increase in the electrical resistance superimposed on the normal decrease resulting from a change in the concentration of the matrix during aging.3 The two consecutive reactions constituting the aging process are accompanied by characteristic microstructural alterations. The initial precipitation process results in the formation of a Widmanstätten figure, while the later transformation of the transition lattice to the equilibrium lattice is accompanied by a coalescence, in the case of aluminum-copper alloys,5 and by the appearance of light areas at the grain boundaries with aluminum-silver alloys showing a structural rearrangement of matrix and precipitate.3 This latter reaction is similar microscopically to the so-called "discontinuous precipitation." Possibly discontinuous precipitation is a process that often if not always follows true precipitation of a coherent lattice and during which the transformation of the coherent transition lattice to the equilibrium lattice takes place accompanied by matrix recrystallization.3 EXPERIMENTAL PROCEDURE The alloys were made using 99.97 per cent Al or distilled magnesium, with

Jan 1, 1943

By V. H. Gottschalk

SEVERAL years ago a group in the metallurgical division of the U. S. Bureau of Mines began a study of the application of new developments in physics to metallurgical problems'. Among these developments were the extensions of what may be termed by analogy the "sound spectrum." Electromagnetic waves, from X-rays to radio waves, have been quite thoroughly explored for metallurgical use, but the sound spectrum, extending from low-frequency vibrations involved in certain types of metal failure by fatigue through the range of audible sound to inaudible vibrations having frequencies of many kilocycles, has had only meager consideration from metallurgists. Because of the general interest in the process recently demonstrated by St. Clair3 for flocculating smoke by means of sound waves, this article is written to point out possible applications of sound and supersonic (ultrasonic) waves in the metallurgical industry.

Jan 1, 1937

By Gerhard Bockstiegel

Kuczynski's formula has been derived for the case of nonspherical particles. TWO formulae of Kuczynski's type have been derived, one describing the increase in tensile strength, the other describing the progress of shrinkage of a powder compact. It has been strength,shown that the exponents of all three formulae each contain two magnitudes of different physical characters, viz, the geometrical factor a and the kinetic factor ß. The interrelationships between the three exponents are stated. SOME years ago Kuczynski1 experimentally showed that the radius, x, of the area of contact between very small spherical metal particles and a metallic block is related to the time of sintering, t, by the following equation x = constant tk [11 where k has the value 1/5 or 1/7. Assuming that the metal particles were perfect spheres and the metallic block was perfectly flat, he derived the foregoing equation from theoretical considerations of the process of material transport in metals, and he showed that exponent k is different for different mechanisms of transport, e.g., k = 1/2 for viscous flow (according to Frenkel2), k = 1/3 for evaporation and condensation, k = 1/5 for volume diffusion, and k = 1/7 for surface diffusion. From this Kuczynski concluded that the mechanism of transport was either volume diffusion or surface diffusion, depending on whether the value of k, as found in his experiments, was 1/5 or 1/7. Subsequently. Cabrera8 corrected Kuczynski's calculations with regard to surface diffusion, showing that the theoretical value of exponent k is 1/5 for both volume and surface diffusion. He supposed that the different experimental values of k were due to slight differences in the shape of the metal particles. An exponential relationship similar to the aforementioned was found by Okamura, Masuda, and Kikuta,4 Masuda and Kikuta, and Takasaki8 when studying the rate of shrinkage on powder compacts during sintering. The authors measured the shrinkage by means of the fraction w = Vp — V./Vp — V,,,, where V,, is the volume of the green compact, V, is the volume of the sintered compact, and V,,, is the volume of the compact in its densest state. This fraction, w, they found, is related to the time of sintering, t, by the equation w == constant tm. [21 Further, Bockstiegel, Masing, and Zapf7 observed that the tensile strength, s, of sintering iron powder compacts can also be related to the time of sintering, t, by an equation of the foregoing type, i.e., s = constant tn. [3] For exponent n the values 0.28 (S=2/7) and 0.35 ( 2/5) were obtained, and the authors pointed out that there might exist a simple interrelation between exponent n as found in their experiments and exponent k in Kuczynski's equation. The authors supposed that 2k = n, since the strength of adhesion between a metal sphere and a block (as in Kuczyn-ski's experiments) must approximately be proportional to their contact area, p. x2. Theoretical Considerations This paper is an attempt to correlate the fundamental experiments of Kuczynski's type with the results obtained with powder compacts as represented by Egs. 2 and 3. In particular, the paper is to show how the rate of sintering is influenced by the geometry of the sintering particles and by the type of material transport. As the geometry of particles conglomerating in a powder compact is very complex, some simplifying assumption has to be made, of course, in order to adapt the problem to mathematical treatment. In the following paragraph a suitable simplification is introduced, and Kuczynski's formula is derived for the case of nonsphcrical particles. Relation Between Area of Contact and Sintering Time—As the face of contact between two particles in a sintering powder compact is not necessarily a circle (as in the case of spheres sintering to a block), Kuczynski's formula is modified as follows: Let the perimeter of the face of contact be described by means of polar coordinates R, 4, as shown in Fig. la, so the area of contact, f, is determined by f= 1/2 . S112p[R(Æ) ]2 dÆ [4] Then, let the two particles be intersected by a plane perpendicular to area f. The intersection is shown in Fig. lb. According to the nomenclature in this figure, the distance, h, between the surfaces of the two particles is a function of T and Æ: h = h(r,Æ). For the particular case of spherical particles, as in Kuczynski's theory, this function becomes: h = constant r2. It shall be assumed here that in the close neighborhood of their

Jan 1, 1957

By R. Taggart, D. H. Polonis, W. C. Gallaugher

In the Ti-Cu system, the a' phase can be produced over a wide range of alloy composition witJwut the retention of measurable amounts of the ß or ? phases. This paper reports on the decomposition of this hexagonal martensite phase front the standpoint of solute corzcentratiorz, tempering temperature, and coherency strain condition. The mechanism of precipitation during tempering comprises localized precipitation which is followed by discontinuous precipitation if sufficient coherency strains are present. The localized precipitation process in hypoeutectoid alloys is described by the generalized Johnson-hiehl equation and an activation energy of 46,000 cal per mole. In hypereutectoid alloys the corresponcEing activation energy is 51,000 cal per mole. The rate-controlling process has been proposed as the difision of copper along a' platelet interfnces. SEVERAL investigations have been reported concerning the kinetics of decomposition of the mar-tensitic a' phase in titanium binary alloys.' In the Ti-Mo system,' two modes of a' decomposition have been observed in the presence of /3 phase; one of these involves the precipitation of fine a from a', and the other involves diffusion across the a'-ß interface. In studies of the Ti-Cr system, Rostoker2 did not detect the formation of TiCr2 during tempering. In Ti-Ni alloys,3 the intermediate phase Ti2Ni has been found to precipitate directly from a' along the interfaces between plates. With the exception of the study of Ti-Ni alloys the previous investigations of tempering phenomena in substitutional martensites are mainly qualitative and do not present a detailed description of the precipitation processes. The following limitations restrict the correlation of the microstruc-tural processes and reaction kinetics during tempering in most binary-alloy systems. 1) A mixture of at least two phases characterizes the constitution of quenched alloys. 2) Difficulties have been encountered in obtaining uniform structures throughout quenched samples. 3) The reaction products, and in particular their morphology. have not been clearly resolved. 4) The martensitic a' phase forms over only a limited composition range for most titanium-base binary alloys. Gomez and polonis4 showed that the Ti-Cu system provides an excellent basis for investigating the tempering of a' over a range of composition without the complication of retained P or the transition w phase. In the present work the effects of solute concentration, tempering temperature, and coherency strain conditions are considered with reference to the over-all precipitation process in quenched alloys of the Ti-Cu system. Both microstructural observations and kinetic data are correlated to define the rate-controlling processes which govern the observed localized and discontinuous precipitation reactions. An earlier paper5 was devoted to a discussion of the modes of heterogeneous nucleation of Ti2Cu from the martensitic a' structure. The present paper will therefore emphasize the progress of the tempering reaction beyond the initial stages. EXPERIMENTAL METHODS The Ti-Cu alloys for this study were prepared by conventional arc-melting procedures. Chemical analysis revealed that the alloy buttons contained approximately 0.02 wt pct 0, 0.04 wt pct N, and within 0.1 wt pct of the intended copper concentration. The progress of the tempering reaction was followed by means of electrical-resistance measurements utilizing a Kelvin double bridge, microhard-ness readings with a 400-g load, and X-ray dif-fractometry patterns to reveal line-broadening effects. Metallographic specimens examined at magnifications greater than X1000 were shadowed with germanium to reveal fine structural details. Direct carbon replicas were prepared for the electron-microscopy studies. EXPERIMENTAL RESULTS Property Changes. The changes of microhardness that accompany the precipitation of Ti2Cu from a' are shown in Figs. 1 and 2. As the solute concentration increases the peak hardness, for a given tempering temperature, increases. In alloys of given composition the time to reach the peak hardness agreed with the time to attain maximum X-ray diffraction peak breadth, as shown in Fig. 3. The maximum hardness and the maximum line breadth increased with lower tempering temperatures, and the time to reach the maximum also increased. The a' peaks broadened during the initial stages of tem-

Jan 1, 1965

By Donald J. Knight

HEAT Treatment at 1500' F of a mild steel containing 0.1 pct C, in an atmosphere which is oxidizing to both carbon and iron, results in the progressive oxidation of the metal surface with little or no change in the carbon content of the underlying metal. Further heat treatment of the scaled metal in either a neutral atmosphere or in a vacuum may or may not result in the decarburization of the metal depending upon the physical nature of the scale. It is apparent that if a continuous layer of oxide is present on the steel surface, the reaction OScale + Cmetal Cogas cannot progress since the SO formed gas would be unable to leave the oxide/metal interface. Thus, in order for the reaction to progress, the oxide must be cracked or porous. Vacuum heat treatment at 1500" F of samples with a thick continuous oxide scale showed no decarburization of the steel. However, when cracks were introduced mechanically into the scale prior to the vacuum heat treatment notable decarburization of the metal resulted and a structure as shown by Fig. 1 obtained, i.e., reduction of the crack walls to ferrite. With reference to the diagram of Fig. 2, it would appear that a mechanism of the following type takes place. Carbon atoms arriving at an oxide/metal/void interface, such as 'A', Fig. 2(a), can combine with oxygen atoms from the scale and leave the reaction site as Cogas thereby reducing the oxide to metal. On a greatly magnified scale, Fig. 2(b),this will appear as a ferrite projection along the crack wall. Carbon atoms entering this projection can react with oxygen atoms at point 'C', which is a new interface, and so on leading to a progressive growth of the ferrite along the crack wall and to the surface, Figs. 2 (c) and 1. It would appear that the only means by which thickening of the ferrite film can take place is by a diffusion of oxygen through the ferrite film to react with carbon at the newly formed ferrite/ void interface. Since the diffusion rate of carbon is reportedly much greater than that of oxygen, and since there is a low probability factor of oxygen and carbon atoms arriving simultaneously at the same site on the ferritehoid interface, there will be a tendency for more rapid lengthening than thickening of the ferrite layer, as is observed.

Jan 1, 1962

By John M. Kelly

New Mexico produced 39,751,868 bbl. of oil in 1941 and ranked seventh among the oil-producing states. Its 1941 production established an annual record, exceeding the 1940 record year by 854,498 bbl. The average daily production for the year was 108,909 bbl. The average daily non-marginal well allowable for New Mexico on Jan. I, 1941, was 36 bbl.; at the close of the year it was 43 bbl. The average daily pipe-linc runs were approximately 3 per cent less than the allocations of the New Mexico Oil Conservation Commission, which followed closely the recommendation of the United States Bureau of Mines in setting the daily outlet demand. The production in Lea County was 34,333,448 bbl.; Eddy County, 5,009,354 bbl.; and northwestern New Mexico (San Juan and McKinley Counties), 406,306 barrels. The total number of completions in the state was 402, of which 301 were oil wells, 19 gas wells, and 82 dry holes. Drilling during 1941 was 68 per cent of 1940, and the ratio of dry holes to producers, I to 4, was the greatest since 1935. Of these dry holes 43 were drilled in proven fields and 39 were wildcats. One new oil field and one new gas field were discovered as a result of the 1941 wildcatting program. However, numerous extensions were made to the existing fields, which, although they were called extensions to proration areas, were in reality field discoveries, being on separate geologic structures. Southeastern New Mexico Lea County The Monument field was the largest producing field in Lea County, with 6,973,808 bbl. It was closely followed by the Eunice field, with 6,644,181 bbl.; Vacuum field, with 4,800,608 bbl. and Hobbs, with 3,624,159 bbl. Hobbs field, however, still has the largest total recovery, this being 97,407,929 bbl. at the end of 1941. The new oil-field discovery in Lea County during 1941 was the Salt Lake field, T. 20 S., R. 33 E. The discovery well was Leonard-Welch State No. I, NW. NW. of sec. 18, T. 20 S., R. 33 E. This well was drilled to a total depth of 3103 ft. and was completed for an initial production of 250 bbl. per day pumping. The oil is 26" gravity and the producing horizon is in the San Andres limestone of Permian age. The discovery well was completed on June 22, 1941, and since that date four more producers have been completed, with no failures. The area is about 10 miles north-west of the Lynch pool, to which the oil from this field is being trucked because there is no pipe-line connection. The greatest drilling activity during 1947 was in the Maljamar field of western Lea County, with 62 completions, 59 being oil wells and 3 dry holes. The limits of this pool are now being defined on the north and south but to date have not been defined either to the east or west. The total number of 40-acre oil-producing units in the field at the close of the year was 158. On April I, 1941, a pressure-maintenance project was installed at the southern end

Jan 1, 1942

By John M. Kelly

New Mexico produced 39,751,868 bbl. of oil in 1941 and ranked seventh among the oil-producing states. Its 1941 production established an annual record, exceeding the 1940 record year by 854,498 bbl. The average daily production for the year was 108,909 bbl. The average daily non-marginal well allowable for New Mexico on Jan. I, 1941, was 36 bbl.; at the close of the year it was 43 bbl. The average daily pipe-linc runs were approximately 3 per cent less than the allocations of the New Mexico Oil Conservation Commission, which followed closely the recommendation of the United States Bureau of Mines in setting the daily outlet demand. The production in Lea County was 34,333,448 bbl.; Eddy County, 5,009,354 bbl.; and northwestern New Mexico (San Juan and McKinley Counties), 406,306 barrels. The total number of completions in the state was 402, of which 301 were oil wells, 19 gas wells, and 82 dry holes. Drilling during 1941 was 68 per cent of 1940, and the ratio of dry holes to producers, I to 4, was the greatest since 1935. Of these dry holes 43 were drilled in proven fields and 39 were wildcats. One new oil field and one new gas field were discovered as a result of the 1941 wildcatting program. However, numerous extensions were made to the existing fields, which, although they were called extensions to proration areas, were in reality field discoveries, being on separate geologic structures. Southeastern New Mexico Lea County The Monument field was the largest producing field in Lea County, with 6,973,808 bbl. It was closely followed by the Eunice field, with 6,644,181 bbl.; Vacuum field, with 4,800,608 bbl. and Hobbs, with 3,624,159 bbl. Hobbs field, however, still has the largest total recovery, this being 97,407,929 bbl. at the end of 1941. The new oil-field discovery in Lea County during 1941 was the Salt Lake field, T. 20 S., R. 33 E. The discovery well was Leonard-Welch State No. I, NW. NW. of sec. 18, T. 20 S., R. 33 E. This well was drilled to a total depth of 3103 ft. and was completed for an initial production of 250 bbl. per day pumping. The oil is 26" gravity and the producing horizon is in the San Andres limestone of Permian age. The discovery well was completed on June 22, 1941, and since that date four more producers have been completed, with no failures. The area is about 10 miles north-west of the Lynch pool, to which the oil from this field is being trucked because there is no pipe-line connection. The greatest drilling activity during 1947 was in the Maljamar field of western Lea County, with 62 completions, 59 being oil wells and 3 dry holes. The limits of this pool are now being defined on the north and south but to date have not been defined either to the east or west. The total number of 40-acre oil-producing units in the field at the close of the year was 158. On April I, 1941, a pressure-maintenance project was installed at the southern end

Jan 1, 1942

By J. H. Wernick, S. Geller

A number of new AB, compounds, in which A is a rare earth or yttrium atom and B is Al, Mn, Fe, Co, or Ni, having the cubic MgCu, structure (Laves phase) are reported. In most of the compounds, the interatomic distances are normal; the lanthanide contraction is observed. However, in the compounds CeB2, B = Fe, Co, Ni, Ce appears to have a valence substantially greater than three; in CeAl2, Ce appears to be trivalent. These results are interpreted as vesulting from an energetically more favorable situation for the 4f electron of Ce to be lost to the 3d band of Fe, Co, nnd Ni. Evidence for the existence of homogeneitj! ranges in these compounds is presented. THE purpose of this paper is to report on the crys-tallographic investigations of a large number of new compounds, Table I, with the MgCuz structure (cubic Laves phase). It is a part of a continuing study of the magnetic properties and structures of binary systems involving the rare earths. In a previous paper' some of the magnetic properties of such systems were reported briefly. We have also reported2 a number of compounds of the CusCa type. The compounds were prepared by induction melting in Argon atmosphere of mixtures of stoichiomet-ric amounts of the constituent elements.* For most of the compounds fused alumina or silica crucibles were used as the containing vessel; graphite crucibles were used to prepare CeA12, NdAl2 and TbA12. X-ray powder photographs were taken with CrKa radiation and the use of Straumanis-type Norelco cameras of 114.6 mm diam. The compounds AB2 belong to space group Fd3m-Ol with 8A in a): (000; 01/2 '/2;3) + 000-; b % XA and 16B in dl: (000; oVa %; 3) + 5/87; % %; 7e % Vs\£>. Because the intensities of the reflections on the powder photographs of the different compounds are similar, relative intensities were computed only for DyMnz for CrKa radiation, Table 11. In the calculation of intensities, the Thomas and umeda3 atomic scattering factors, corrected for dispersion,4 were used. The agreement between observed and calculated intensities leaves little doubt that DyMnz has the MgCuz structure. Crystallographic data for the ABz compounds are given in Table JII. Twelve of these compounds have been reported by other authors. A comparison of the published results with our work is given in Table W.* ers of elements of lower purity, to peritectic reactions in some cases, and to the presence of homogeneity ranges in these alloys, so that melting losses would still result in single phase alloys. The possibility of existence of homogeneity ranges in these compounds was investigated systematically only for GdMnz. At room temperature, the solid solution range is between

Jan 1, 1961

Jan 1, 1943

Jan 1, 1943

By F. von Gemmingen, E. J. Fasiska, L. Zwell

EARLIER studies of the dependence of creep on structure1 and of the dependence of creep rupture on creep behavior2 did not include detailed discussion of the phases present in the specimens under examination. It is of interest to determine the nature and amounts of such phases present since it is well known that these are dependent on composition, heat treatment and creep conditions.3 This investigation has therefore been undertaken to identify these constituents and to determine their relative amounts. This report presents the results of this investigation and describes an improved "extraction replica" technique developed in the course of the work. The composition of the Type 316 austenitic stainless steels studied in this investigation is given in Table I. The specimens were solution treated at 2000°F for 1/2 hr and water quenched. Those specimens subjected to prestrain at room temperature Methods for extracting precipitate particles from materials for electron and X-ray diffraction analysis have been described by Fisher,4 Smith and Nutting: and Booker and coworkers.6'7 Electrolytic extraction methods have been described by Andrews and Hughes.8 The first two methods have been used at this Laboratory, but have not proved satisfactory for many problems requiring X-ray diffraction analysis. It has been difficult to remove the replicas or to avoid tearing them during removal. These difficulties are overcome by backing the replica with a water-soluble plastic such as Dupont 51-05 Elvanol (polyvinyl alcohol). The backing increases the thickness of the replica so that it can be mechanically stripped from unloosened particles without folding or tearing. This procedure eliminates the stripping of wet replicas, which usually have a low tear strength. By stripping replicas after drying, it is possible to extract partially loosened particles which otherwise would remain imbedded in the metal. The modified method is as follows: after the specimen is polished, it is etched for 60 min in modified glyceregia consisting of 30 ml glycerol, 7 ml nitric acid, and 4 ml of hydrochloric acid. The sample is then washed, air dried, and coated with a 20 pct Baker and Adamson 1611 collodion (85 pct amyl acetate by volume) film. After the collodion film is dry (about 15 min), the sample is again etched in the modified glyceregia for 50 min, washed with isopropyl alcohol (to prevent dissolving the collodion) and air dried. A very thick coat of polyvinyl alcohol (about 50 pct solution "Elvanol Dupont 51-05" in water) is poured over the sample and is permitted to dry for about 24 hr. The composite film containing the precipitates is then stripped from the specimen and floated, polyvinyl alcohol side down, in water (at about 50°C) until the polyvinyl alcohol dissolves. This usually takes about 15 min. Should any of the polyvinyl alcohol remain on the replica, it does not give a crystalline diffraction pattern and therefore does not hinder identification of the precipitates. The extraction replica, floating in water, is picked up on a glass fiber and mounted in the Debye-Scherrer camera for X-ray diffraction analysis. Concentration of the etchant, and also etching time, had to be increased for several of the samples. Fig. 1 shows two extraction replicas taken from different areas of spec. M74-37; X-ray analysis showed a strong diffraction pattern of M23C6 and a medium one of u phase. A

Jan 1, 1962

Jan 1, 1924

Jan 1, 1924

By Franklin J. Hill, Theodore D. Tiemann

Sized fractions of Wisconsin Gogebic taconite were reduced with hydrogen over the temperature range from 600° to 1000°C. In general, the degree and rate of reduction increase with temperature. Particle size has no observable effect except in the smaller fractions at 900°C and above, where reduction is impeded, possibly due to the siliceous nature of the ore. Gaseous diffusion appears to be significant as a rate controlling factor at certain stages. THE depletion of high grade iron ore reserves in the United States has established the need to utilize all available iron ore deposits, both in long range planning and from the viewpoint of national emergency. This has led to extensive research and development in this field. The beneficiation of Wisconsin Gogebic taconite ore has been previously studied by investigators at the University of Wiscconsin, 1 the United States Bureau of Mines,, Battelle Memorial Institute, 3 and others. Methods investigated have included concentration by flotation processes, magnetic and gravity separation, magnetic roasting and various combinations of them. The foregoing studies have been concerned with the production of an enriched ore suitable for blast furnace use. If the ore could first be reduced to metallic iron and then beneficiated, the product could be used directly in the open hearth or electric steel-making furnace. The by-passing of the blast furnace should enhance the economic feasibility of the beneficiating process. The work reported here,4 was undertaken to establish data on the hydrogen reduction of Wisconsin taconite that would eventually lead to the development of an economic direct reduction process for this and perhaps other low grade siliceous ores. The volume of work on direct reduction can be judged by the fact that 240 direct reduction processes were patented in the United States alone up to 1951, 5 and numerous additional processes have since been developed. Some of the factors contributing to direct reduction development are: 1) The increasing capital cost of the blast furnace and its accessory equipment. 2) The decreasing availability of high grade domestic ore. 3) The high cost of steel scrap for use in the open hearth and electric furnaces. In general, direct reduction processes can be divided into two types; those using a solid reductant (coal, peat, lignite) usually carried out in a kiln, and those using a gaseous reductant (H2, CO) in a shaft furnace or a fluidized bed retort. As there is no low cost coal readily available in northern Wisconsin but natural gas could be quite easily obtained, it is believed a gaseous reductant process would offer the greater feasibility for use on Wisconsin taconite. It is hoped that this preliminary work will beof interest to other investigators in this increasingly important field. TACONITE ORE The ore used in the investigation was obtained originally from the U. S. Bureau of Mines, and consisted of a representative composite made up from trench samples of the Norrie, Pabst, Plymouth, Pence, and Yale members of the Wisconsin Gogebic range iron formation. Chemically, the ore contains approximately 52 pct SiO, and 30 pct Fe. X-ray diffraction and other studies indicate the principal minerals to be quartz (SiO2), hematite (Fe2O3), and goethite (Fe2O3. H2O). Minor amounts of iron containing silicates, siderite (FecO3), and magnetite (FeO-Fe2O3) are present. The iron minerals and quartz are so finely disseminated that the ore requires grinding to less than 200 mesh (74 µ) for essential liberation.l,2 The iron content of each of the several size fractions studied is shown in Table I. EXPERIMENTAL METHOD The method used throughout the investigation was based on that developed by the U. S. Bureau of Mines in Minneaopolis6,7 in which the course of reduction is followed by the loss in weight of the ore sample. More specifically, a 100-g sample of the size frac-

Jan 1, 1962

T. L. Joseph (University of Minnesota, Minneapolis, Minn.)—Mr. Killian is to be commended for his inquiry as to why a decrease of 15.3 pct in coke consumption was accompanied by a decrease of only 1.9 pct in the CO/CO, ratio of the top gas. Although I encouraged him in his efforts to find a more significant and meaningful carbon-gas ratio, the method, calculations, and conclusions originated entirely with the author. Whether we agree entirely with the assumptions underlying his corrected ratio, I think we can all agree that a better ratio is needed, and that he deserves credit for recognizing the need and for doing something about it. To be most useful, the CO/CO, ratio for the top gas should reflect the way in which the carbon is gasified in the furnace, particularly the degree of oxidation and the extent of heat release. The effect of CO, from the decomposition of the limestone and the effect of the reaction of CO and water vapor in the upper part of the furnace on the quality of top gas should be isolated from the effect of more basic reactions such as ore reduction and solution loss. The reduction of sufficient iron oxide to produce a ton of pig iron and the accompanying changes in the proportions of the carbon gases can be expressed on a mol basis as follows: 16.5FeA + 16.5CO - 33FeO + 16.5CO, [1] 33FeO + 33.0CO = 33Fe + 33CO. [2] The CO, from eq 1 normally escapes from the furnace because it does not react with coke at the temperature of its release. However, as much as 50 pct of the CO, from eq 2 may react with coke because it is released at temperatures above 1000°C: 16.5CO. + 16.5CO2 = 33CO [3] Thus we see that the gasification of 16.5 mols or about 200 lb of C in solution loss will regenerate as much CO as is required for the last stage of reduction. The pronounced effect of reaction 3 on the final proportions of CO and CO, in the gas stream is obvious. It also follows that reaction 3 materially alters the heat release but it cannot be entirely eliminated in Lake ore practice. A previous study indicates that 100 lb of solution loss may be necessary to maintain the reducing power of the gas. A reduction of 100 lb from the normal level of 200 lb would mean that over 1,-000,000 additional Btu are released in the process. The 100 lb of additional carbon which will be burned at the tuyeres will release 437,000 Btu and the absorption of 583,000 Btu by eq 3 would be prevented. One of the chief objectives of sized ore burdens and more uniform distribution of the gas stream is to minimize solution loss by completing reduction higher in the furnace and in turn to release the maximum amount of heat from the carbon charged. It was odd, indeed, to find that the use of sized ore and sinter in the careful tests reported by Dobscha had changed the CO/CO2 ratio very little. The author has stressed the close correlation between his corrected CO/CO, ratio and production. A correlation of fuel consumption and the CO/CO, might be more logical. Fuel consumption on the prepared ore burden decreased 15.3 pct whereas the corrected CO/CO2 ratio decreased 19.0 pct. Although this agreement between the reduction in fuel and the reduction in corrected CO/CO2 ratio is not as close as the correlation with tonnage, it indicates the corrected ratio is superior to the normal ratio which decreased less than 4 pct. Some adjustment should be made for the reaction: CO + H2O = CO2 + H2 —700Btu [41 because there is evidence to indicate that about one half of the hydrogen is formed in the upper 21 ft of

Jan 1, 1953

By Horace R. Graham

CHEMICAL nitrogen and the "nitrates" of commercial significance are derived mainly from three basic sources: (1) the natural deposits in the form of nitrate-bearing earth and clay, which, being largely water-soluble, can exist only in the most arid portions of the earth; (2) coal, which yields not only nitrogen but also the additional by-product hydrogen required for the synthesis of ammonia and its compounds; (3) the atmospheric air, which contains 75.5 pct by weight of pure nitrogen gas. NATURAL NITRATE DEPOSITS Occurrence Although natural nitrate deposits consisting principally of the salts of sodium and potassium have been reported in Egypt, South Africa, Mexico, Argentina, Colombia, Peru and the USA, the only known natural nitrate resources of commercial importance are the vast deposits found in the Atacama, Tarapaca and Antofagasta regions of northern Chile, lying in a narrow strip 10 to 50 miles wide and roughly 450 miles long, between latitudes [19O] and [26OS]. In addition to yielding nitrates of sodium and potassium, these deposits are the source of the major portion of the world's supply of iodine. The nitrate-bearing ores known in Chile as caliche are variable in composition and may contain from 5 to 30 pct each of nitrate, chloride and sulphate. While the deposits are present chiefly in the form of the sodium salt, potassium, magnesium and calcium also occur in percent- ages from 0.1 to 5; usually as double salts with sulphate, which are but slightly soluble in nitrate leach solutions. These double salts are bloedite(Na2S04.MgS04.4H,0), glauberite (Na2S04.CaS04), polyhalite (K2SO4.- MgS04.2CaS04.2H20). Some deposits contain up to30 pct of their ni-

Jan 1, 1949

By R. B. Sosman, J. C. Hostetter

It is well known that ferric oxide, Fe2O3, is paramagnetic, while magnetite, Fe3o4, is classed among the highly ferromagnetic substances. But magnetic data on oxides intermediate in composition between Fe2o3 and Fe3O4 have been almost completely lacking. Fe2O3 and Fe3O4 form a solid-solution series, according to present evidence.' In this series the properties change continuously from Fe2O3 toward Fe3O4, as the percentage of FeO increases from zero toward 31.03, which is the percentage in magnetite. It is possible that there is a break in the series near Fe3O4, but it has not yet been possible to establish the existence of such a break experimentally. The occurrence of solid solution in this series is shown by the dissociation pressure, or oxygen pressure in equilibrium with the oxides, which falls continuously over the range from Fe2O3 to Fe3O4. The oxygen-pressure curves at 1100" and 1200" are shown in Fig. 1. The existence of solid solution is also demonstrated by the continuous change in optical properties from Fe2O3 over to a composition containing about 18 per cent. FeO, at which point the opacity of the oxide becomes so great that it is impossible to obtain measurements at higher percentages of ferrous iron. Natural oxides of iron intermediate in composition between Fe2O3 and Fe3O4 are much more common than is generally supposed. If a mineral oxide is not strongly attracted by a small hand magnet and if it gives a red streak, it is usually labeled "hematite." If it is strongly attracted by the hand magnet it is usually labeled "magnetite" without further tests. As we shall show later, oxides containing from 1 up to 31 per cent. FeO can be thus erroneously lumped together as "magnetite." It is a fact, nevertheless, that the great bulk of natural oxides of iron lie fairly near either Fe2O3 or Fe3O4 in composition. The reason will be

Jan 1, 1918

By M. G. Benz

The superconducting performmce of diffusion-processed Nb3Sn is influenced by its micro structure. High isotropic transverse current density may be achieved in this material by a process which forms a precipitate of ZrO, within the Nb3Sn. FOR an ideal type-I1 superconductor, little or no transport current can be carried in the mixed state; i.e., little or no transport current can be carried above the lower critical field H,,, where the field penetrates abruptly in the form of current vortices or fluxoids, even though full transition to the normal state does not occur until the upper critical field H,,.' Fortunately, nonideal type-I1 superconductors can be readily obtained and these carry large transport currents up to the upper critical field H. Both theoretical and experimental investigations have attributed this current-carrying capability for nonideal type-I1 superconductors to pinning of the fluxoid lattice by heterogeneities in the microstructure of the superconducting material. These heterogeneities may take the form of dislocations or dislocation clusters,2"5 grain boundaries: structural imperfections introduced by phase transformations; radiation damage,8"10 or precipitates.11"15 Nb3Sn formed by diffusion processing is a type-I1 superconductor. Heterogeneities are needed for high superconducting critical currents above H,,. This paper will cover: a) what the microstructure of diffusion-processed NbSn looks like; b) what changes in the microstructure take place when the system is doped with precipitates, and c) how these changes in microstructure influence the superconducting critical currents. EXPERIMENTAL Preparation of Samples. Diffusion processing was used to form the Nb3Sn. The procedure used was as follows: a) coat the surface of a niobium tape with tin; b) heat-treat this tape at a temperature above 930°C to form a layer of Nb3Sn at the Sn-Nb interface. Such a layer of NbsSn is shown in Fig. 1 The thickness of the NbsSn layer formed was controlled by the time and temperature of the heat treatment. The same general procedure was used for preparation of both undoped samples and samples doped with a precipitate. An additional step was included in the preparation of the doped samples which consisted of internal oxidation of zirconium to form ZrOn. The details of the doping process will be reported in a later paper. Sample Testing. The Nb3Sn tape samples were soldered to a copper or brass shunt. Current and voltage leads were then attached to the sample in the usual four-probe resistance measurement configuration. The sample was cooled to 42°K. In some cases it was cooled in the presence of a high magnetic field and in other cases with the field turned off. The results were the same for both cases. The samples were oriented in a configuration with field transverse to current but could be rotated such that the angle between the field vector and the wide side of the tape sample could be changed. Measurements up to 100 kG were done in a superconducting solenoid and measurements above 100 kG in a water-cooled copper magnet at the MIT National Magnet Laboratory. Once the test field was reached, the current in the sample was increased until voltage was detected across the sample. The critical current was taken as the current at which voltage was first detected in excess of background noise. In most cases this was 1 to 2 x 10~6 v for a— in.-wide sample carrying several hundred amperes with a in. separation between voltage leads and with a 10 "-ohm shunt resistance. RESULTS AND DISCUSSION Microstructure. Examination of the microstructure of the undoped Nb3Sn shows rather large-diameter (1 to 2 columnar grains growing outward from the niobium surface toward the tin surface. As the layer is made thicker by longer diffusion times, these grains grow longer. Few new grains are started. Transmission electron microscopy shows little or no second-phase material within the bulk of the Nb3Sn layer. The microstructure of a diffusion-processed NbsSn layer changes quite drastically when the system is doped so as to form a precipitate within the NbsSn layer. Instead of large-diameter columnar grains of NbaSn forming, smaller-diameter (0.5 to 1 ) equiaxed grains of Nb3Sn decorated with the precipitate form. Fig. 2 shows a transmission electron micrograph of a Nb3Sn layer doped with zirconium oxide. This layer has been etched so that one may look between the grains

Jan 1, 1969