Search Documents

By W. Evans, C. E. Ells

The agglomeration of hydrogen in pure aluminum and A1-Mg alloys has been studied through use of hydrogen introduced into the metal by cyclotron proton irradiation. Both the growth and dispersal of the agglomerates have been followed by metal-lographic techniques. During the cyclotron irradiation, agglomeration occurs at temperatures i 100°C for hydrogen concentrations as low as 1 ppm. On heating at elevated temperatures, 300°C to 600°C, intragranular agglomerates first coarsen and then disperse. The agglomerates formed at grain boundaries are much larger than the intragranular agglomerates; with hydrogen concentra- tions > 10 PPm, heating for 1 hr at 500°C results in extensive grain-boundary cracking. The hydrogen agglomerates have been shown to inhibit grain growth, but no inhibition of recrystallization has been observed. The ability of hydrogen to enter aluminum at a free surface, combined with its high-diffusion rate at elevated temperatures, accounts both for growth and dispersal of the agglomerates. Irradiations at elevated temperatures give results consistent with the explanation proposed for agglomerate behavior on post-irradiation heating. 1 HE formation of hydrogen-filled bubbles, or blisters. in aluminum and its alloys has been studied for over' thirty years. The work up to 1952 has been reviewed by Kostron' and since then Ransley and

Jan 1, 1963

By G. T. Murray

The aging behavior of a Cu-Ti (3.2 at. pct Ti) alloy has been followed by electrical resistivity, hardness, and metallographic changes. The resistivity data indicate two principal processes, the first causing a sharp decrease in resistivity, and the second, which occurs over a long period of time, a much smaller rate of decrease. Lamellar nodules, Widmanstätten platelets, and homogeneous continuous precipitation was observed. The Widmanstätten platelets grow at the expense of the latter. When this study was initiated, little information was available on the aging behavior of copper-base titanium alloys. Several articles'-3 that appeared in the early 1930's showed that these alloys could be strengthened by precipitation, but a detailed study was not made. Within the past year, however, two other more comprehensive studies4'5 have been reported in the literature, one consisting primarily of an investigation of mechanical properties, the other

Jan 1, 1961

By J. M. Dupouy, R. A. Rawe, M. B. Bever

The aging characteristics of a titanium alloy containing 13 pct V, I1 pct Cr, and 4 pct A1 have been investigated by hardness measurements, X-ray diffraction, and metallography. The P phase decomposes into 0 + a, and 0 + a, + TiCr, above about 1050°F; below this temperature a transition phase, believed to be w, precipitates first, followed by a, and TiCr,. The observed hardening is attributed to these reactions. Cold working of solution-treated specimens accelerates hardening and raises the hardness level attained. A TTT diagram is proposed. NEW titanium-base alloys, designated as "all-beta," have recently been developed. These alloys contain large amounts of P-stabilizing elements such as molybdenum, chromium, or vanadium. When cooled from the solution temperature, they consist only of P; in particular, the martensitic a,' phase does not form.

Jan 1, 1961

By H. C. Rogers

It has been shown previously'- 3 that when a mild steel or iron is charged with hydrogen, the normally observed yield point is eliminated or considerably

Jan 1, 1960

By A. R. Kaufman, P. Gordon

THE large-scale manufacture and use of uranium in conjunction with the atomic energy development during the war led to a need for knowing the equilibrium diagrams of uranium with various other metals. The alloys of uranium with aluminum and with iron were among the first that were studied for this purpose. Much of the work was done in the manner of a survey since it was necessary to determine the important features of the phase diagrams rather than to obtain completely detailed results. For this reason no attempt was made to locate exactly such features as liquidus lines, solid solubility limits and compound and eutectic compositions. It is believed, however, that there are no great inaccuracies in the diagrams reported here and that no important points were overlooked. Earlier work on alloys of uranium with aluminum and iron is almost nonexistent and in each case is considered to have been of little value by M. Hansen. It is interesting to note, however, that the compound UAl3 was first reported in 1902 by L. Guillet. Experimental Procedure Materials: The uranium used for making most of the aluminum alloys was obtained from material prepared by Metal Hydrides, Inc., and melted in vacuum at M.I.T. Because of the lack of an adequate supply of metal, it was necessary to use material from several different batches and hence it is impossible to state an exact analysis. However, a few results indicated about 0.01 to 0.03 pct iron and probably about 0.03 to 0.05 pct carbon. Toward the end of the work a small amount of metal was obtained from Brown University and from Iowa State College and this was used in determining the transformations in pure metal and in some dilute alloys. All of the aluminum used in this investigation came from a supply of high purity metal (greater than 99.9 pct aluminum) which was obtained through the courtesy of Dr. Mehl of the Carnegie Institute of Technology. The uranium used for making the alloys with iron was produced by the bomb reduction process which has been described by J. Chipman.² This metal, after vacuum melting, contained about 99.9 pct uranium by weight with carbon and iron as the chief impurities. Electrolytic iron containing about 0.012 pct carbon, 0.020 pct nickel and 0.009 pct copper was used in making the melts. Melting Equipment: The melting of the alloys went through a stage of development which culminated in the use of beryllia or beryllia-lined crucibles and in the use of high frequency heating in either vacuum or in an atmosphere of argon. It was found that uranium at a few hundred degrees above its melting point would pick up aluminum and to a lesser extent silicon from an alundum thimble while aluminum above 1000°C would extract silicon from the impure alundum. No reaction of either metal with beryllia was noted at temperatures as high as 1800°C. The distillation of aluminum from the aluminum-rich alloys was quite troublesome at temperatures above about 1300°C and was only partially avoided by the use of argon instead of vacuum. Melting in resistance furnaces,

Jan 1, 1951

By A. R. Kaufman, P. Gordon, C. H. Schramm

AS a part of the general program on alloys of uranium carried out at the Massachusetts Institute of Technology under contract W-7405-eng-175 for the Manhattan Project during the recent war, it was considered desirable to investigate the boundary binary phase diagrams of the ternary alloy system uranium-tungsten-tantalum. Previous work on these alloys was practically nonexistent. A brief statement to the effect that tantalum and tungsten form a continuous series of solid solutions was found in a general paper on tantalum by W. V. Bolton1 but no information was available on the uranium-tungsten and uranium-tantalum systems. The research to be described in this paper has served to delineate the main features of the uranium-tungsten, uranium-tantalum phase diagrams and to check Bolton's work on the tungsten-tantalum diagram. Experimental Details Metals: The tungsten and tantalum used for preparing the alloys were 99.96 and 99.65 pct pure, respectively. The chief impurities in the metals were about 0.02 pct molybdenum in the tungsten, and 0.2 pct columbium, 0.05 pct tungsten, 0.02 pct barium and 0.01 pct silicon in the tantalum. The uranium used was 99.9 pct pure, with the main contaminants being iron and carbon. (An appreciable amount of magnesium was present in the uranium as received, but this distilled off during melting and so is not reported as an impurity.) The composition of specimens indicated in the subsequent figures and tables of this paper are the results of chemical analyses on the cast alloys except for the tantalum-tungsten alloys. For the latter, chemical analysis was not very reliable as a result of the difficulty of separating tungsten and tantalum chemically. Consequently, the nominal

Jan 1, 1951

By Irving Cadoff, Jack Medoff

THE linear thermal expansions of oriented single crystals of zinc were measured in the range from 20" to 416°C using a Leitz HTV optical lever differential dilatometer. The single crystals, supplied by Dr. J. J. Gilman, General Electric Research Laboratory, Schenectady, N.Y., were prepared from 99.999 pct-purity zinc, and were in the form of rods approximately 2 in. long by 0.15 in. in diameter having the following orientations: x = 1.5, 12, 20, 30, 37.5, 49, 60, 73, and 82 deg where x is the angle between the specimen axis and the (0001) plane of the crystal. From the family of expansion vs temperature curves, the instantaneous expansion coefficients were calculated at 100°C intervals, and Fig. 1 shows a plot of these derived coefficients for each orientation. Coefficients were obtained from the raw data using the chord-slope method. The curves in Fig. 1 indicate a positive temperature coefficient for low x, a negative temperature coefficient for high X, and an expansion coefficient invariant with temperature, amounting to about 43 x 106, for a crystal orientation of approximately 52 deg. From the available data, it is possible to arrive at the values for the axial thermal-expansion coef- ficients a, and a, by extrapolating the curve of the expansion coefficient vs orientation to zero and 90 deg orientations. More precise values for a, and a, are obtained if the instantaneous coefficients are plotted against the square of the cosine of the orientation angle, as shown in Fig. 2. The extrapolated terminal points of the best straight line through the data points of such a plot define the axial coefficients at that temperature. The theoretical justification for the linearity of this curve comes from inspection of a rearranged form of the general anisotropic equation for expansion: ax=a, + (a, -a,)cosZx [1I in which a, is the expansion coefficient in the "c" or [0001] direction and a, the expansion coefficient in the "a" or [ll50] direction. From Fig. 2, the axial coefficients at 100°C, for example, were found to be a, = 13.9 x 10'6 and a, = 62.4 x 10"6 per "C. In the same way, the coefficients were determined for 20", 200°, 300°, and 400°C. This data is summarized in Fig. 3. It is of interest to observe the common intersection of the curves in Fig. 3, which seems to bear out the earlier observation, from Fig. 1, of a

Jan 1, 1964

By Paul Gordon, A. S. Iyer

Apreliminary investigation has been carried out on the annealing behavior of explosively deformed copper as compared to that of conventionally deformed copper, using hardness, stored energy, and the sharpening of X-ray spots on Debye rings as the measures of annealing. Some metallographic observations were also made. Two purities of copper were studied. The first was tough pitch copper obtained as a small piece from the guard ring of a sample explosively deformed to approximately 500 kbars.* The second consisted of mentedand deformed these samples quite badly so that only a few small fragments were available for the annealing experiments. As a result, little in the way of quantitatively significant data was gleaned from the measurements. However, several important qualitative differences between the annealing of explosively deformed and conventionally deformed copper have been observed. It is these effects which are described in this report. Fig. 1 presents the results of hardness and stored energy measurements during the annealing of the high-purity copper after explosive and after conventional deformation. The data for the conventionally deformed copper were taken from the work of Gordon;' in this copper it is evident that most of the stored energy is released concommittantly with the hardness drop. As shown in orddon's' investigation, these effect:: are definitely associated with recrystal-lization. In the case of the explosively deformed copper, on the other hand, a large fraction of the stored energy is released before any appreciable hardness change takes place. At 160°C about 40 pct, and at 170°C about. 80 pct, of the stored energy appears to have been released prior to the onset of softening, i.e., undoubtedly prior to recrystallization. In view of the facts that the hardness and energy measurements were made on separate samples and that there is reason to suspect that the intensity of the explosive deformation from sample to sample may not have been uniform, a check run was carried out in which both energq release and hardness changes were measured cnn the same sample. The results of this run, shown in Fig. 2, confirm the conclusion that much of the stored energy is released before the hardness begins to change appreciably. In an effort to obtain some idea of the activation energy associated with the early release of energy, a single callorimetric sample was annealed in two steps, first at 149.5"C and then at 159.5"C, Fig. 3.

Jan 1, 1962

By R. O. Scattergood, P. Beardmore, M. B. Bever

THE stored energy and microhardness of a Au-Ag alloy deformed by explosive loading were measured previously as functions of shock pressure' and compared with corresponding data for drawn wires.' The difference of the two processes has been explored further by investigating the annealing behavior of explosively loaded specimens and drawn wires. Sheet specimens, 0.042 in. thick, of a Au-Ag alloy containing 82.6 wt pct Au were explosively loaded in stacks of four.3 Wires prepared from the same melt of alloy were drawn through dies.' The annealing kinetics were determined by microhardness measurements on separate specimens for each annealing temperature and time. Several values of the stored energy were measured by tin-solution calorimetry Experimental details are given elsewhere. Wires were drawn until the microhardness or stored energy equalled the corresponding values of the explosively loaded specimens as nearly as possible. Wire drawn to a true strain 7 of 0.09 had approximately the same hardness and stored energy as specimens explosively loaded to a shock pressure of 70 kbar. The microhardness of specimens

Jan 1, 1963

By A. E. Deruyttere, J. Van der Planken, M. Laurent, Van den Bergen

FahRENHORST and schmid1 observed that zinc single crystals work hardened less rapidly when strained in liquid air (- 185°C)than in a bath at -82°C, whereas at higher temperatures the rate of work hardening decreased with increasing temperature, as would be expected. Cadmium and magnesium showed no such decrease at -185°C. At that time the observed anomaly for zinc at liquid air temperature probably seemed insignificant. However it later was confirmed2 by experiments in liquid nitrogen (-196°C) and in the mixture solid carbon dioxide-acetone (-77"C) and therefore appeared to be real. seeger3 then suggested that the anomaly might be due to the different wetting of the zinc crystals by the cooling media. Recently a new confirmation was obtained,4 however the cooling media again were different: the rate of strain hardening was found to be lower in liquid nitrogen and liquid oxygen than in pentane cooled to various temperatures down to -126°C. Ekperiments will now be reported in which the medium surrounding the specimen at the different temperatures was the same, namely gaseous air. The single crystals were made from "Overcor 99.99" zinc. They were grown in a vertical traveling furnace on seeds of nearly the same orientations. The diameter of the crystals was 5 mm and the length between shoulders 140 mm. The brass shoulders were stuck to the crystals by means of araldite. A thin steel wire was fixed to each shoulder and served to fasten the specimen to a Houns-field tensile testing machine. The crystals with the shoulders and part of the wires were surrounded by two concentric cylindrical containers. The coolant. either liquid air or ground solid carbon dioxide, was poured into the outer container only. During cooling the containers rested on a support which was removed when the straining was started. In order to check the temperature which the crystals would attain inside the inner container, a hollow bar of zinc in which a thermocouple was fitted was first mounted in the apparatus as a substitute for a crystal. It appeared that a constant temperature, respectively - 183° and -72°C, was attained inside the bar, if a soaking time of the order of 1 hr was allowed and if the outer container was constantly kept filled. Accordingly, this procedure was adopted for every tensile test. To take account of the large scatter, fourteen tensile tests were performed with liquid air and seven with solid carbon dioxide. The orientation of the crystals and the results are presented in the table, in which the symbols have the following meaning: 1 : temperature x : initial angle between basal plane and specimen axis : initial angle between slip direction and specimen axis a : shear strain at fracture 0: average rate of strain hardening = —-——- 100 t0 and Te representing respectively the critical shear stress and the shear stress at fracture. Despite the scatter of the a- and 8— values which is of the same order of magnitude as in the work reported by Seeger and Trzuble,4 it is evident that the strain hardening rate is significantly lower at -183°C thanat-72°C. The table also contains the average values calculated from Seeger and Trauble's results at the corresponding temperatures: z.e., four results at -183oC and two at-70°, -71°C. These authors calculated 9 in a slightly different way, but this is unimportant. The agreement between the a-values shows that both sets of experiments are comparable despite the differences in experimental techniques. The rates of strain hardening also agree and it may therefore be concluded that the anomaly is not

Jan 1, 1962

By A. H. Daane, B. J. Beaudry

The uranium-antimony system has been investigated by metal-lographic, therma1, and X-ray methods. There are four intermediate phases present: USbz and U,Sb, which undergo peritectic decomposition at 1355°and 1695OC, respectively, and USb and UPSbs which melt congruently at 1850"and 1800°C, respectively. There is a eutectic between U4Sbs and USb which melts at 1770°C. USb,, UsSb4, and USb are line compounds while U4Sbs has a solubility range of approximately 1.5 at. pct at 1770°C. Single crystal studies of the cmpo,und U.,SbS show it to have a hexagonal unit cell with a, = 9.268A and c, = 6.201b;. The solubility of uranium in liquid antimony increases from 0.1 at. pct at 650°C to 1.5 at. pct at 900°C. The solubility of antimony in a, P, and y uranium is approximately 0.02, 0.1, and 0.05 at. pct respectively. IN nuclear reactors, uranium has been used mainly in the form of solid-fuel elements of the pure-metal or uranium-rich alloys. However, in attempts to overcome some of the difficulties inherent in solid-fueled reactors, reactor concepts hive been proposed in which the fuel would be a liquid metal, such as a solution of uranium in another metal. The liquid-metal fueled reactor (LMFR) designed by the Brook-haven National Laboratory of the U.S. Atomic Energy Commission is an example. Fuel for this reactor is a solution of approximately 800 ppm enriched uranium in bismuth.' To design such liquid-fueled reactors, a background of information on the solubility of uranium in various metals is obviously necessary. The present study was undertaken to pro- vide information for such purposes, and in addition, to study the uranium-antimony system as a whole. A preliminary literature search showed that some work had been done previously on uranium-antimony alloys, but no attempts had been made to make a complete study of this system. The solubility of uranium in antimony was studied by Cammack and Bridger' up to 950°C and by Hayes and Gordon up to 900'-C. Their data are plotted in Fig. 1. Ferro, in an X-ray structure study of some uranium-antimony compounds, found USg to be tetragonal with a, = 4.2724 and c, = 8.471A, U3S4 to be b.c.c. with aO = 9.095A and USb to be f.c.c. with a,= 6.191A. Colani reported the compound UsSba; however, work by Ferro and by this Laboratory showed that this compound does not exist. foote' reported several intermediate phases in the uranium-antimony system, but did not identify them. Greninger' re-ported finding an intermediate phase in a 5 at. pct Sb

Jan 1, 1960

By G. Koves

The behavior of case-hardened AISI C-1213 free -machining steel under complex impact-fatigue -wear load conditions was investigated. The inherently poor dynamic properties of the steel are mainly affected by the stress distribution in the case, component shape, and tempering. Cold-riveted components showed unexpectedly high dynamic strength compared to basic parts. A compromise beatment—which provides good dynamic properties, reasonable wear resistance, and fair riveting characteristics—zuas developed. A number of mechanical components in data-processing machines are made of free-machining carbon steel. Although, selected primarily due to manufacturing requirements, components made of these grades frequently undergo complex heat treatment and must sustain impact-fatigue loads. A typical free-machining steel is the AISI C-1213 grade, which is a resulfurized and rephosphorized open-hearth steel. While the undesirable effects of sulfur and phosphorus on the mechanical properties are well-known, data on dynamic properties are scarce.1"3 Rapid rotating or oscillating motion while carry- ing a high cyclic load requires a combination of wear resistance with high endurance limit and impact toughness. The inherently poor dynamic properties of the free-machining steel are further degraded by the machining, which introduces local stress-concentration areas. While the heat treatment (usually case hardening) improves some mechanical properties, it necessarily impairs others. Another problem arises when components, such as studs, must be fastened by riveting. The high crack-sensitivity of the free-machining steel, especially in the presence of a hardened case, makes this operation difficult. In view of the preceding considerations, it is of interest to investigate the behavior of a typical component under complex impact-fatigue-wear loading conditions in order to explore the effects of material, machining, and heat treatment, to obtain specific data on the dynamic characteristics of free-machining steel, and to arrive at a process which would adequately utilize the inherent qualities of this material. EXPERIMENTAL PROCEDURE To simulate actual service conditions, a typical component was selected, based on frequency of application, complexity of load, and processing difficulties. The dimensions and shape of the component—a stud used in many electromechanical document-handling machines—are shown in Fig. 1.

Jan 1, 1964

By N. F. Foster

The use of piezoelectric semiconducting materials for the fahrication of ultrasonic transducers has raised the upper fundamental frequency limit for transducers from about 200 mc to above 10 kmc, Several types of semiconductor transducer configurations have been proposed and examined experimentally. Depletion-layer transducers have been made and operated at frequencies below 1 kmc hut they appear to he better suited for operation at considerably higher frequencies around 10 knzc. Diffusion-layer transducers have been made and operated in the range 50 to 1000 mc and have exhihited considerably improved conversion losses and bandwidths compared with other transducers available in the frequency range. A variety of novel structures are possible using these new types of transducers, some of which could lead to substantial advances both in research and in the fabrication of ultrasonic delay lines and memory elements. THE field of ultrasonics today spans some five decades of the frequency spectrum from the audible to the kmc range and finds applications ranging from fundamental research into the atomic structure of matter to the degreasing of machine parts. To cover this wide range it has been necessary to develop a great many types of ultrasonic trandsucers utilizing a variety of materials and physical phenomena. The first ultrasonic transducers were made during World War I for a submarine-detection system. They consisted of plates of single-crystal quartz bonded between metal plates and were one of the first applications of the phenomenon of piezoelectricity discovered nearly 30 years previously. Despite the many advances which have been made in transducer technology, piezoelectric transducers using single-crystal quartz are still used almost exclusively for application above about 30 mc. Transducers at these frequencies are normally made in the form of thin quartz plates which are then bonded onto the medium in which the ultrasonic wave is to be propagated. The crystallographic orientation and geometry of these plates determines the type of wave generated and the way in which the efficiency of the electromechanical conversion varies with frequency. Analysis of the design parameters shows1 that the frequency of maximum efficiency (the fundamental or resonant frequency) is that frequency at which the transducer is approximately half the wavelength of the ultrasonic wave in the material. As the frequency increases, the ultrasonic wavelength, and therefore the optimum transducer thickness, decreases and in practice the fabrication of quartz transducers with fundamental freauencies above 100 mc becomes extremely difficult due to the extreme thinness required. For frequencies above 100 mc quartz transducers can be used in overtone-mode operation, and for very high frequencies (500 to 24,000 me) a technique using a quartz rod inserted in a high-Q microwave cavity4 is available. These techniques, however, exhibit comparatively low fractional bandwidths and rather poor conversion efficiencies. The present trend in ultrasonics is towards high frequencies and greater bandwidth to allow the use of shorter pulses which can give increased discrimination in radar-system delay lines and greater storage capacity in computer applications. The increasing demand for more efficient transducers with greater bandwidth

Jan 1, 1964

By J. W. Cunningham, W. A. Tiller, D. H. Lane

The effect of ultrasonic vibrations on ingot solidification has been considered both theoretically and experimentally. The theoretical section elucidates the mechanisms by which the ultrasonic vibrations might refine the ingot structure. The experimental section describes a very efficient means of introducing the ultrasonic vibrations to the freezing interface. The application of this technique to the consumable-electrode arc melting process shows that the columnar mode of freezing may be effectively suppressed yielding an ingot structure consisting predominantly of equiaxed grains. The results support the contention that the ultrasonic vibrations increase the nucleation frequency in the layer of liquid adjacent to the freezing interface. CERTAIN structural characteristics of as-cast ingots are undesirable since they create considerable difficulty for subsequent processing and often produce defects in the finished product. These structural defects are illustrated in Fig. 1(a), and are: 1) planes of weakness where columnar grains growing from the different mold faces meet, 2) strength anisotropy due to the preferred orientation of the columnar grains and 3) inhomogeneity due to segregation at dendrite boundaries within the columnar grains and at the columnar grain boundaries. The type of ingot structure that eliminates 1) and 2) and diminishes 3) is the fine-grained equiaxed structure illustrated in Fig. 1(c). Many attempts have been made to produce the desirable ingot structure illustrated in Fig. 1(c) by inoculating, casting with varying superheats, vibrating or stirring, or by irradiation with ultrasonic vibrations. These attempts have met with various degrees of success, the ingots generally having the structure illustrated in Fig. 1(b). All four methods have been the subject of numerous investigations,1"4 the latter producing the most erratic behavior and inconsistent success. The method of ultrasonic irradiation during ingot solidification has shown considerable promise on small-scale ingots (diam 2 in.) but has not yet been successfully adapted to large-scale operation. The present paper considers the application of ultrasonic radiation to ingot solidification both theoretically and experimentally. The purpose of the paper is to elucidate the mechanism by which ultrasonic energy refines the ingot structure and to describe a method for producing the desirable ingot structure in both laboratory scale experiments and in ingots of commercial size. THEORY Conventional Casting—Before discussing the possible effects of ultrasonic radiation upon ingot structure let us first briefly review the reasons for the breakdown of columnar crystallization to equiaxed crystallization during ingot solidification. The following is a slightly oversimplified description. In the conventional method of casting, the alloy having a certain degree of superheat, T, is poured to fill a cold mold which is at room temperature, Tc,. Due to conduction of heat through the mold wall the outer rim of liquid metal cools rapidly to the liquidus temperature and begins to supercool. At some degree of supercooling, nuclei of solid form at random in this supercooled layer at the mold surface. The grains in this chill layer begin to grow inwards producing a solute build-up, thus lowering the freezing temperature at the interface as illustrated in Fig. 2. As growth proceeds, the temperature gradient in the solid conducts away not only the latent heat of fusion evolved during growth but it also conducts away some of the superheat of the melt so that the temperature

Jan 1, 1961

By W. A. Tiller, D. H. Lane

A simple zone melting technique for investigating the effect of ultrasonic irradiation upon ingot solidification is described. The effect of i) ultrasonic power level, ii) freezing velocity, iii) constant or variable frequency, iv) direction of irradiation, and v) transducer-cozlpling bar joint upon the grain size of type 316 stainless steel have been investigated. The results of this study support the postulate that the ultrasonic vibrations increase the nucleation probability in the layer of liquid adjacent to the freezing interface. In the first paper of this series,l the possible mechanisms whereby ingot structure may be altered by irradiating a solidifying melt with ultrasonic vibrations have been discussed. A new technique for efficiently introducing the ultrasonic energy into the solid-liquid interface region was presented and applied to ingot preparation by the consumable-electrode arc-melting technique. The experimental results support the postulate that the ultrasonic vibrations may refine the ingot structure by increasing the nucleation probability in the zone of liquid adjacent to the solid-liquid interface. The present paper describes the application of the same technique to a relatively simple laboratory-scale zone-melting apparatus. This apparatus is found to be well suited to the detailed study of ultrasonic technology in this area and allows one to determine, in particular, the minimum ultrasonic power level needed to produce a certain grain size in a particular material. The results of this study support the nucleation mechanism for grain refinement, but do not provide any deeper insight into the specific nucleation mechanism responsible for the effect. EXPERIMENT The apparatus used in these experiments is shown in Fig. 1 and illustrated schematically in Fig. 2. The material to be studied, type 316 stainless steel, was in the form of a 3/4-in. sq bar 30 in. long. This bar was placed in a square open-topped mold with open ends made from high-purity graphite coated with a flame-sprayed Al2O3, skin. Surrounding the crucible was a molybdenum susceptor with 3/8-in. diam holes spaced at 1-in. intervals along its length to serve as sight ports. A tapered transducer horn was brazed to one end of the specimen. The entire assemblage was enclosed in a quartz tube, the transducer itself projecting out one end and resting in a cooling tank. A vacuum seal between the coupling bar and the end plate enclosed one end. A similar plate and seal enclosed the other end. The system is gas tight and can be operated under vacuum or an inert atmosphere. For these experiments, flowing argon at 15 in. of Hg pressure was used. The heat source for melting the stainless steel specimen was a 2-in. long 3-kc induction coil placed around the quartz tube; this source heated the susceptor, inductively melting about a 3-in. zone of the sample by radiation. To move the molten zone at a specified rate, the induction coil was moved down the tube at this rate with the aid of a simple pulling device. Thus, the sample could be irradiated at varying ultrasonic frequencies and power levels and could be solidified at varying rates. The power input to the transducer was measured using a "John Fluke Model 120 VAW-Meter." 1 The power delivered to the interface region is probably only about 50 pct of that measured on the VAW-meter. In a typical run, when the ultrasonic power was turned on, the interface of the molten zone closest

Jan 1, 1961

By W. D. Robertson, D. A. Koss, A. J. Goldman

THE significance of the hcp (E) structure, which appears when Fe-Cr-Ni alloys (stainless steels) are transformed martensitically, has been the subject of considerable study and speculation.'-7 It has been proposed that the hcp structure is an intermediate phase in the transformation: fcc (austenite) — hcp (E) -bcc (0, martensite). However, recent studies by transmission electron microscopy6,7 have indicated that the hcp structure may be a consequence of the large shear strain (10 deg) associated with the fcc to bcc transformation. That is, the transformation to a bcc structure is accompanied by extensive dissociation of dislocations in the parent austenite, which has a low stacking-fault energy.'-'' X-ray diffraction analysis of the transformed alloy reveals a pattern of lines corresponding to an hcp structure, together with the parent austenite and the bcc product. The fact that the hcp structure is observed only in association with the bcc product in these alloys1'4'6'7 lends support to the conclusion that the hcp structure is a secondary product of the transformation. But, the evidence is circumstantial and more definite resolution of the two possibilities mains to be obtained. If the hcp structure is an intermediate product of the transformation, as in 1) above, then two distinct Ms temperatures should be observed, Ms and To evaluate this possibility, three different techniques that are particularly sensitive to the structural changes accompanying such transformations have been used in an attempt to detect two transformation temperatures in an Fe-15.1 Cr-11.7 Ni alloy: 1) high-sensitivity differential thermal analysis, 2) the temperature dependence of resonant frequency (elastic moduli), and 3) internal friction of cylindrical specimens vibrating in either longitudinal or torsional modes. To obtain the necessary sensitivity, a differential thermal analysis technique was developed for the purpose, Fig. 1. The differential temperature between a nickel standard and the specimen was amplified to produce a sensitivity of 0.02OC (0.7 v) in temperature difference. The specimen and the standard were surrounded by a copper block and the entire assembly was immersed in a solution of methyl and ethyl alcohol, freezing point, - 140OC, which provided the necessary thermal stability to reduce temperature fluctuations to negligible proportions. A cooling rate of about 1.5OC per min was obtained by bubbling liquid nitrogen through the liquid bath. In order to calibrate the experiment with respect to the evolution of heat in the transformation from fcc to hcp structures, differential cooling curves were obtained for an Fe-19.8 Mn alloy, which transforms martensitically to E.'"-' a Ms(e) temperature of 145°C was clearly defined by a temperature

Jan 1, 1964

By R. W. Powers, M. V. Doyle

ThE solution of a diatomic gas such as 0, or N2 in a metal usually follows Sieverts' law; i. e., Here C is the solute concentration at equilibrium and P, the gas pressure. The proportionality constant Ks is also the equilibrium constant for the reaction 1/2 02 (at pressure, P) - 0 (in solution at [21 concentration., C). to use oxygen as an illustrative gas. Sieverts' law is therefore interpreted to indicate that diatomic gases go into metallic solution atomically and that the solute "gas" atoms interact negligibly with each other. Of course, this interpretation is strictly valid only over the temperature interval in which the reaction indicated in [Z] can be investigated. At temperatures several hundreds of degrees below this interval, the possibility of interactions between solute atoms is conceivable. Indeed, from internal-friction measurements some evidence for the existence of groupings of oxygen atoms in tantalum solid solutions was presented a few years ago.1,2 Over a range of solute concentration, the experimental oxygen internal-friction peak in tantalum, measured as a function of temperature at constant frequency of vibration, can be described as the sum of two elemental peaks, each corresponding to a separate relaxation process. At 0.7 cps, the height of the elemental peak found at 137°C is proportional to the oxygen content, whereas the height of the one at 162°C varies as the square of the oxygen concentration. The 137°C peak was presumed to arise from the stress-induced ordering of oxygen atoms among various classes of octahedral sites in the manner proposed by Snoek.3 The 162°C peak was attributed to a somewhat similar stress-induced ordering process, in which each oxygen atom was interacting with a neighboring oxygen atom. Similarly, the experimental peak due to nitrogen in tantalum can also be described as the sum of two peaks. The height of one varies as the nitrogen concentration and the other as the square of this quantity. In fact, some evidence of solute interactions exist for all oxygen and nitrogen solid solutions of Group V transition metals. However, the interaction energy was not obtained previously for any alloy. The determination of this quantity, as a part of a more intensive study of the interaction between oxygen atoms in solid solution in tantalum, is reported in this paper. PRINCIPLE OF THE EXPERIMENT Consider an equilibrium in a solid solution of tantalum between oxygen atoms which are bound together pairwise and those solute oxygen atoms which are not so associated. 20 (Ta) - O2 (Ta) [3]

Jan 1, 1960

By T. Yoshioka, Paul A. Beck

SILICON, germanium, and tin occur with both the white tine-type structure and the diamond cubic structure. In the latter form these elements are semiconductors; in the former they are metallic. The metallic form of germanium and silicon is obtained only under high pressure.' For all three elements the transition from the semiconductor to the metallic state is accompanied by a decrease in the atomic volume of somewhat more than 20 pctily2 Table I. At such a large compression the widening of the s and p bands is sufficient to make the energy gap of 1.1 ev, or less, of the semiconductor disappear, giving rise to the metallic state.*

Jan 1, 1965

By P. S. Rudman

The allotropic volume changes of the ,metallic elements are reviezoed with the conclusion that in general atomic volume is conserved to better than 1 pct in such transformations. A table of the atomic volumes of the metallic elements is given and its use in considering the size factor in alloys is discussed. The importance of the concept of atomic-size factor, due to Hume-Rothery and coworkers,' is firmly established. Most attempts at quantitative formulations of this concept have employed the atomic radius (half some interatomic distance) as the measure of size and innumerable tables of such radii exist. This is probably, however, an unfortunate choice of parameter. Firstly, discussion in terms of interatomic distances requires a knowledge of the crystal structure. While this is known, except for a few cases, for the elemental structures, there are still many undetermined alloy structures. Secondly, except for the simpler structures, face-centered cubic (Al),* body-centered cubic (A2),* and hexagonal-close-packed (A3).* with ideal axial ratio, there is no unique in- teratomic distance, an m y decision must be made as to which distance, or which average of the distances, should be used. However, the most important point is that for metals the dominant bonding parameter can generally be expected to be atomic volume and not interatomic distance. This is perhaps most explicitly stated in the Wigner-Seitz cellular method.' In this approximation, cellular polyhedra of atomic volume v are replaced by spheres of radius Y,: 4/3 ?rs3 = v. That is, there is no reference to the interatomic distances. If atomic volume is in fact the dominant parameter, then in allotropic transformations we might expect to find atomic volume approximately conserved. For example, in the A1 ? A2 transformation we would thus have

Jan 1, 1965

By M. G. Benz, J. F. Elliott

The austenite solidus of the iron-carbon system has been determined using a series of diffusion couples, each of which consisted of a specimen of austenite held in contact with a melt saturated with austenite. After the equilibrium distribution of carbon had been established by diffusion at a specified temperature, i.e. the austenite specimen had become austenite of the solidus conzposition, the diffusion couple was cooled, sectioned and anazlyzed for carbon. The solidus was found to be a straight line: A revised iron-carbon temperature-composition diagram is presented, COMPOSITION has a marked effect on the temperature at which austenite begins to melt and the austenite solidus of the iron-carbon system describing this effect has been the subject of many investigations.1-10 The significant results of these investigations are presented in Fig. 1. The experimental methods and purity of the materials used in them are summarized in Table I. In plotting the data in Fig. 1 no attempt has been made to convert these results to the International Temperature Scale of 1948," except for the data of Adcock,5 as this conversion would do little to reduce the uncertainty that exists as to the position and shape of the solidus. A point of major concern in evaluating these data is that the alloys studied, except those used by Adcock, were not binary alloys of iron and carbon, Table I. Also, it would appear that several of the methods did not permit equilibrium to be established through the system being studied. All that can be said for the results of these investigations is that they indicate only approximately the location of the solidus with the uncertainty as to its location at 1 pct C being approximately 100°C. The current investigation was undertaken to provide reliable data by which the austenite solidus could be established. It was hoped that information on the liquidus also could be developed at the same time, but experimental limitations prevented this as austenite segregated from the liquid on cooling. After a careful study of possible experimental methods and extensive laboratory tests, it was decided that the use of a series of austenite-liquid diffusion couples would provide the most reliable results. This paper describes the method and its results and also includes a complete iron-carbon temperature-composition diagram based on what are considered to be the best available data. EXPERTMENTAL METHOD The diffusion couple used for this investigation consisted of a small cylindrical pellet of austenite held in contact, at a specified temperature, with a melt saturated with austenite. The composition of the cylindrical austenite pellet was chosen to be approximately 0.1 wt pct C less than the estimated

Jan 1, 1962