Search Documents

By Tibor Stefansky, John E. Dorn

MOTT and Nabarro1-5 were first to illustrate the importance of the flexibility of dislocations in accounting for the strengthening that metals undergo due to the presence of internal-strain centers. Such strain-centers might arise from a number of causes such as radiation damage, the presence of semicoherent and coherent precipitates and clusters, as well as individually dispersed interstitial and misfit substitutional solute atoms. Each strain-center induces positive and negative shear stress fields on the slip planes of the surrounding alloy matrix. In order to effect plastic deformation, therefore, the alloy must be subjected to sufficiently high applied shear stresses to push dislocations past all such internal stress fields. If dislocations were inflexible and moved as rigid lines, they would be equally pushed and pulled by internal stress fields. At this extreme the strengthening would be vanishingly small. Dislocations, however, are not rigid; they are flexible and extensible. They wiggle between positive and negative strain-centers, thus increasing their line energy at the expense of their interaction energy with strain-centers. cottrel16 and Friedel7,8 have independently presented lucid reviews of the significance of the Mott-Nabarro dislocation-flexibility concept to the strain-center strengthening of alloys. Their analyses however were only semi-quantitative. There is evidence in the recent literature of renewed interest in a more rigorous analysis of flexibility. Thus, Gleiter9 and Gerald" have shown that the calculated strain-energy interaction between a dislocation and the stress fields of coherent precipitates can be seriously in error if the flexibility of the dislocation is not taken into account. This report will be concerned with a more detailed analysis of the problem based on the following approximations: 1) Although many types of strain-centers exist, the following analysis will be limited to the case of solute-atom strain-centers in fcc substitutional alloys. 2) To facilitate the analysis by avoiding statistical considerations, the strain-Centers are assumed to be in regular arrays on the slip plane. Although the quantitative results are somewhat dependent on the type of array that is assumed, the qualitative trends are similar for various arrays that were studied. 3) An adjustment of the results from those for square arrays to those for the more realistic case of a random distribution of strain-centers is accomplished by adopting the interesting statistical theory first introduced by Kochs,11,12 elaborated on by Dorn, Guyot, and stefansky,13 and analyzed by Foreman and Makin14 in their computerized experiments. 4) For simplicity the dislocation lines are taken to wiggle about their edge orientation and no correction is applied to their line energy for deviations from edge configurations or for the elastic interaction between segments of the now curved dislocation. It will be shown that for extremely low concentrations of atomic strain-centers the strengthening increases with the square root of their concentration. For slightly higher concentrations the strengthening rate becomes less rapid. At a critical concentration a maximum strength is obtained and a decrease in the strengthening takes place as the concentration of solute atoms is further increased. These trends are a direct consequence of dislocation flexibility. They arise from changes in the amplitude of the waviness of the dislocation line at the yield stress with solute atom concentration. Despite the simplifying assumptions that were made, the results agree qualitatively with appropriate experimental data. I) STRESS FIELDS ON SLIP PLANES Internal stresses are introduced in a lattice whenever a host atom of radius ro is replaced by another that has a different atomic radius. According to the classical theory of linear elasticity,15 the stress field about a solute atom at the origin of a spherical coordinate system is given by arr=-4G€(?_0)3 ffflfl ="W =2G€{t) f°rr^ro(l+e) UJ where G is the shear modulus of elasticity and Y, = ro(l + E) is the radius in situ, of the substituted atom. As shown by ~shelb~,'~ this approximation assumes that the bulk modulus of elasticity of the substituted atom is identical with that of the host species. ~abarro'~ suggested that the strain E can be deduced

Jan 1, 1970

By Robin O. Williams

The increase in internal energy as the result of deformation has been measured for lead, aluminum, silver, nickel, iron, and zirconium by using rapid, adiabatic compression. The stored energy increase is roughly Proportional to the strain; the propor-tionality constant increases rapidly with increasing melting point. The fraction of the mechanical energy which is stored increases more slowly, since the strength of the metals also increases with melting point. The values of the stored energy are considered accurate to about 10 pet. The present values appear about 50 pet larger than the more reliable published results where comparisons are possible. It is possible that this difference is due to the high strain rate used in this investigation. Immediately after deformation all these metals release energy at a rate roughly proportional to (time)-'. This release is considered to be associated with dislocation motion but in aluminum (and copper) some additional process seems to be present. This release can represent 20 pet or more of the stored energy. WHEN pure metals are plastically deformed, most of the mechanical energy is converted into heat. The energy remaining within the metal is significant in that it is the energy of the disorder produced and thus detailed knowledge of this energy is a powerful tool in understanding the nature of deformational disorder. While much effort has been expended on this problem, the amount of information available is limited. The situation as of 1958 has been carefully reviewed by Titchener and ever.' The present results have been obtained by a new experimental approach to this problem. The method necessitates high-strain rates which make comparisons with published results less certain, but a high-strain rate is an advantage in that the energy release immediately after deformation can be followed. EXPERIMENTAL PROCEDURE In the experimental method used, the internal (stored) energy is given as the difference between the mechanical work used in deforming the sample and the heat which is released (the first law of thermodynamics). The work is supplied by two identical hammers swinging freely from a fixed height, the available energy being the product of the mass, the gravitational constant, and the distance through which the center of gravity moves. The initial temperature rise of the sample represents the heat produced by the deformation. The sample temperature is determined by a small thermocouple embedded in and supporting the sample. This process can be repeated over and over to produce increased strains. Most samples are run through about five cycles to a total strain of around 0.7. Further details are covered elsewhere. The determination of the heat is dependent upon the sample weight, its specific heat, the rise in temperature, and any gains or losses to the surroundings. One is dependent on published results for specific heals (the values were taken from a collection).3 The values must be accurate to about 0.1 pet in order that they not affect the results. The best values thus do not contribute to the uncertainty, but this is probably not always the case. The accuracy of the temperature rise is limited primarily by knowledge of the thermocouple characteristics over short temperature intervals, but careful calibration eliminates this as an important factor. One can calculate readily the heat flow between the sample and hammers if the time of contact and the sample temperature are known and if there were no oil at the interface. Assuming the maximum temperature difference, upper values for this interchange have been calculated (the time of contact is determined from the decrease in sample length) and it may or may not be significant. However, no correction is colnsidered necessary because of the presence of a thin oil film which has a thermal conductivity much less than the metals. Except for the possible uncertainty in specific heat, the heat is considered to be adequately known. The mechitnical losses which are recognized as important are: 1) friction between the sample and the hammer:;, 2) the kinetic energy in the hammer suspension system which may not be entirely usable, 3) the rebound energy of the hammers, and 4) the vibration of the hammer heads. All these have been covered in detail elsewhere2 and only the more significant points are made here. The friction turns out to be small except for soft materials (lead)

Jan 1, 1962

By E. M. Kalar, I. Iwasaki, J. D. Zetterstrom

Several methods of removing fatty acid coatings from iron ore flotation concentrates were tested both on a Mesabi oxidized iron ore and on a magnetic taconite concentrate, and their effects on the duplex flotation process and on pelletizing were compared. Either acid or alkali alone could not remove the coatings, but a combination of such reagents as lime and quebracho, calcium chloride and pyrophosphate, or sulfuric acid and an organic acid (oxalic, malonic, or citric acid) was quite effective. The use of activated carbon was found to be particularly promising. The removal of fatty acid coatings from iron ores is of much practical interest when a pelletizing process is being considered for a concentrate floated with a fatty acid collector or when a duplex flotation process is being considered for upgrading an oxidized iron ore. A duplex flotation process is a fatty acid flotation of iron minerals followed by an amine flotation of the siliceous gangue from the rougher iron concentrate. Such a process is being used in the Florida phosphate fields.' Although in northern Michigan mills fatty acid flotation concentrates have been pelletized successfully, at other locations fatty acid coatings on iron flotation concentrates have proved so undesirable in the agglomeration operation that other methods of concentration had to be sought.' Since fatty acid coatings cannot be removed as readily with a simple acid or alkali treatment from iron oxide surfaces as from Florida phosphates, a combination of reagents, such as lime and quebracho,3 lime and alkali phosphate,4 or sulfuric acid and oxalic acid,' has been proposed. It was also thought that activated carbon might be effective in removing fatty acid coatings due to its high adsorptive capacity. Activated carbon, with its enormous surface per unit of volume, is well known for its exceptional efficiency in removing organic pollutants from industrial effluents. This article presents the practical aspects of the above methods when they are applied to the duplex flotation process and to the pelletizing of iron flotation concentrates. The article also includes a preliminary investigation of the common parameters involved in the depressant activity of the reagents. EXPERIMENTAL WORK Initially, the experimental work was directed towards the optimization of each method for the removal of hydrophobic coatings from iron oxide surfaces floated with fatty acids. Subsequently, the methods thus optimized were tested in the duplex flotation process with an oxidized iron ore and in the pelletizing process with a magnetic taconite concentrate. Preliminary Flotation Tests: Although the three types of chemical treatment reported in the literature were briefly investigated for the purpose of comparison, major emphasis was placed during the preliminary flotation tests on the use of activated carbon because it was thought to have certain advantages over the chemical methods. LIME AND QUEBRACHO - To obtain a flotation concentrate for optimizing the lime and quebracho method of fatty acid removal, the magnetic taconite concentrate was used. The concentrate had a size consist of 90% passing 325 mesh and analyzed 64.5% iron. After a few preliminary tests the fatty acid flotation procedure with the magnetic taconite concentrate was standardized as follows. A 1200-g sample of the magnetic concentrate was pulped in a Fagergren laboratoy flotation cell to 40% solids, conditioned with 0.5 lb of soda ash per ton for 2 min, and then further conditioned with 1.0 lb of Acintol FA-2 per ton for 2 min. After 5 min of flotation, recovery was nearly complete, averaging 98.6% of the weight and 98.7% of the iron. The cell product, which amounted to 1.4% by weight and analyzed 63.2% iron, consisted mainly of relatively coarse middling particles. Since the repro-ducibility of the flotation concentrates produced by this procedure was vey high, the same procedure was used to prepare fatty acid concentrates for each of the six tests of the lime and quebracho method. To test the removal of the fatty acid from the magnetite surfaces by the lime and quebracho method, a fatty acid froth product was returned to the flotation cell, repulped by 30 sec of agitation, and then con-

Jan 1, 1968

By R. A. Breitrick, J. N. Swinderman, J. J. Cooper, H. L. Bauer Jr.

The Robinson Mining District, near Ely, Nev., is one of the major porphyry copper districts in the world. Within the district is an eight mile long east-trending zone of altered porphyry, metamorphosed sedimentary rocks, and sulphide mineralization surrounded by Paleozoic limestone, dolomite, shale and sandstone, Tertiary volcanics, and Quaternary alluvium. Six major disseminated ore deposits have been exploited in this zone. About 20% of the district production has come from disseminated ore in metamorphosed sedimentary rocks found adjacent to altered quartz-monzonite porphyry. The following brief statements summarize some of the relations between porphyry alteration, metamor-phism, and mineralization: 1) Disseminated copper ore does not occur in unaltered porphyry or in m-metamorphosed sedimentary rocks. 2) Sedimentary rocks have been only weakly metamorphosed where intruded by unaltered porphyry. 3) Intensely metamorphosed sedimentary rocks are found only adjacent to altered porphyry. 4) Many hydrothermal minerals are common to both altered porphyry and adjacent metamorphosed sediments. 5) Disseminated ore is found in altered porphyry and adjacent metamorphosed sedimentary rocks. 6) In general, there is a direct correlation between grade of ore in porphyry and copper content in the adjacent rocks. 7) Chemical changes in the altered porphyry and metamorphosed rocks appear to be related. The foregoing relations support a hypothesis that the porphyry alteration, the metamorphism of the sedimentary rocks, and the ore deposition were all part of a continuous hydrothermal process which probably closely attended the porphyry intrusion. The Robinson Mining District, most often referred to as the Ruth or Ely District, is located in eastern Nev. and is one of the major porphyry copper districts of the world. Production from 1908 through 1962 was nearly 250 million tons of ore averaging a little over one percent copper with a total metal value of nearly one billion dollars. The ore has been produced from six major disseminated-type deposits with the Liberty Pit deposit accounting for 70% of the total district production. Less than one million tons of base metal and gold ores have been produced from the replacement-type deposits of the district. Approximately 20% of the district production has come from underground mining operations; the pit ore has been mined with shovels and removed by train haulage, truck-train haulage combinations, truck haulage, and presently by a combination truck-skip system. Many of the ore deposits formerly worked by underground methods have been and are being exploited by open pit methods. (See Fig. I.) Two co.'s have accounted for virtually the entire production from the district. In 1958 Kennecott Copper Corp. acquired the district assets of Consolidated Coppermines Corp. and is now the sole ore producer in the district. Present operations are confined to the Liberty Pit where 22,500 tons of ore averaging a little less than 0.8% copper is mined per day. Cutoff grade is 0.4% copper, and the stripping ratio is three to one. The ore is hauled by rail 23 miles to the McGill Reduction Plant where it is concentrated to about 20% copper. Molybdenite is recovered as a by-product in the mill. The copper concentrates are smelted in McGill, and the blister copper is shipped for refining to Balt. where gold, silver, platinum and palladium are recovered. The Robinson Mining District has been well known as a porphyry copper district for the past fifty years, but many geologists are unaware of the fact that a large portion of the production has come from disseminated ore found in metamorphosed sedimentary rocks. The scope of this paper describes the occurrences of disseminated ore in the various metamorphosed sedimentary rocks, and presents evidence for the hypothesis that these ores are genetically, spacially, and coevally related to the disseminated ore in the porphyry rocks. Publication of this paper was granted by the management of Kennecott Copper Corp. A. H. James suggested and encouraged the writing of the manuscript. Special mineralogical and spectrographic analyses were performed by the Kennecott Research Center. The deposits of the district have been studied by many geologists since the turn of the century, and

Jan 1, 1964

By M. M. Rao, P. G. Winchell

The growth rates of bainitic plates were measured at 400°C in Fe-Ni-C alloys containing 0.10 atom-fract~on nickel and 0.0012 to 0.0075 atonz-fraction carbon. The growth rates are adequately represented by where xc is nearly the atom fraction carbon of the bulk austenite and PXCy is nearly the carbon atom fractlon in the ferrlte of radiids p In equilibrium with austetzite. The form of the equation is that predicted by a model in which carbon diffusion in austernite controls the gvowth, but the numerical constatnt is two orders of magnitude below that suggested by the model. THE growth of bainitic plates in steel is often assumed to be controlled by the diffusion of carbon away from the advancing plate tip. This hypothesis predicts that the growth rate will increase as the carbon content of the austenite, xCz, is reduced toward the carbon content of the saturated ferrite comprising the plate tip, PxCY The growth rate should vary approximately as (xCg- pxCy)-1. Experimental observation of the growth behavior at low carbon levels should provide a significant test of this model. An alloying element in addition to carbon is required so that low-carbon austenite can be experimentally observed while undergoing bainitic transformation. Nickel was selected. The presence of nickel complicates the interpretation of the data in two ways: First, diffusion of nickel during the transformation would make analysis very difficult. Nickel is assumed immobile during the transformation. Second, nickel affects the solubility of carbon in ferrite and austenite in equilibrium. This effect has been evaluated.' At the completion of our experimental work Goode-now et al.2 published data in essential agreement with the observations to be reported here. Since their discussion is abbreviated and their data are scanty in the region of interest, we believe the present work is of significance. I) THE MODEL OF BAINITIC PLATE GROWTH The rate of lengthening of a plate is assumed to be controlled by the diffusion of carbon from the advancing ferrite-austenite interface into the surrounding austenite. The precipitation of carbides is assumed to be a secondary process. For ease of analysis the carbon-atom ratio,* pxCy, of austenite in equilibrium with ferrite which is convex with minimum curvature radius p, and the carbon-atom ratio, PxCY, of that ferrite in equilibrium with austenite are assumed independent of location on the ferrite-austenite interface. Since these carbon contents vary with the radius of curvature of the ferrite, p, their assumed positional independence must be held as an approximation. The consequences of these assumptions have been developed approximately by zener3 and Hillert,4 and the resulting equation for a platelet has been applied to bainite by Speich and cohen5 and Kaufman, Radcliffe, and Cohen.8 The Zener-Hillert equation* for plates is: The analysis of Hillert is supported by that of Hor-vay and cahn7 which involves no mathematical approximations but does include the assumption that the a/y interface coincides with an isoconcentration line. The solutions of Horvay and Cahn for elliptic paraboloids are replotted in Fig. 1. The shape of the paraboloid is expressed in terms of the ratio of the principal radii of curvature at its tip, A =p1/p2, which is also the ratio of the minor to the major axis of the elliptic cross section. The Zener-Hillert equation for plates is also plotted. The agreement is within a factor of two for (pxyaCr - xyC )/(xyC - PxCaY) between 0.5 and 100. This is the range of interest here and in most other work on bainite. The original form of the Zener-Hillert equation was the form given above with the right-hand side replaced by (pxCya -xCy)/(PxCya). This replacement is not appropriate here. 11) THE EXPERIMENTAL PROCEDURE Alloys were prepared and three kinds of experiments carried out. Continuous-cooling-transformation experiments were carried out on wires by measuring temperature and resistance during continuous cooling. Isothermal-transformation experiments were carried out on wires by measuring electrical resistance as a

Jan 1, 1968

By R. A. Wattenbarger, H. J. Ramey

A systematic study has been made of the application of the real gas pseudo-pressure m(p) to short-time gas well testing. The m(p) function can be used in real gas flow problems to account for the variation of viscosity and z-factor with pressure. A mathematical model was solved numerically to generate solutions of various real gas flow problems. The effects of turbulence, formation damage and wellbore storage were included in the model. The analysis of simulated well tests showed that the interpretation methods normally used for liquid flow are generally accurate when the m(p) is used. For practical rates without turbulence, the solutions are not rate-sensifive, and the flow capacity.. kh and skin effect can be determined accurately from either a buildup or a drawdown test. However, the kh calculated from a drawdown test can be significantly low when turbulence is present. A case was simulated in which this error wa.s 36 percent. Turbulence does not affect the determination of kh from buildup tests. The proper determination of kh from p and pa buildup and drawdown plots is developed by analyzing their relationship to the m(p) method. A simple equation is given that can be used for long-range gas well performance forecasting. This expression is compared with a method pre-sended by Russell et al." Introduction The formation flow capacity and wellbore damage condition can be determined for liquid-producing tests by means of buildup and drawdown tests. These tests make use of short-time pressure transient data rather than stabilized flow tests that are often used for gas well testing. Tracy' presented a method of gas well testing that was based on ideal gas equations and utilized p2 buildup plots. This method was shown to be good for low pressure wells. MatthewsZ suggested plotting p and using an average slope and average gas properties. This method was more successful on high pressure wells. Al-Hussainy, Ramey and Crawfordh howed that variation in gas properties could be simplified by using the real gas pseu do-pressure m(p). Some cases showed that the use of the m(p) function Original manuscript received in Society of Petroleum Engineers office Aug. 8. 1967. Revised manuscript received June 11, 1968. Paper (SPE 1835) was presented at SPE 42nd Annual Fall Meeting held in Houston Tex., Oct. 1-4. 1967, and at SPE 4th Annual Eastern Regional ~eetin$ held in Pittsburgh. Pa., Nov. 2-3. 1967. 0 Copyright 1968 American Institute of Mining, Metallurgical, and Petroleum Engineers, Inc. 'References given at end of paper. provided an accurate method of interpreting buildup and drawdown tests. Because the work of Al-Hussainy et al. included only a few actual cases, there was a need to explore the use of the m(p) method for a greater variety of flow conditions. The occurrence of turbulent flow around the wellbore often is an important factor in gas well testing. Swift and Kiel,' and Carter et a1.' treated turbulence in gas well testing, but did not consider the variation of viscosity and z-factor with pressure. This paper presents the results of an investigation of the application of the m(p) method to buildup and drawdown testing. Vhe investigation included the effects of turbulence, formation damage and wellbore storage. Gas Flow Equations To formulate the mathematical model, many of the assumptions usually used in well testing theory are applied. The system has radial geometry with a closed outer boundary and is composed of a horizontal porous formation that has uniform and isotropic rock properties and uniform thickness. Allowance is made, however, for a radial region of reduced permeability near the wellbore. This region represents formation damage. The geometry of the system is shown in Fig. 1. Darcy's law does not always apply to gas flow. A more general expression is needed for non-Darcy, or "turbulent" REGION OF DAMAGED PERMEABILITY SEALED BOUNDARIES ClRCULAR WELL BORE Fig. 1—Radial flow model. AUGUST, BT7

Jan 1, 1969

By J. D. Verhoeven

The effectiue distriblltion coefficient has been Measured in a series of vertical, normal freezing experirtzents with large current densities passing through the solid-liquid interface. The results indicate that the effect of the current on solute redistribution is due to electvotransport in the liquid. The high currents produced extensive mixing in the liquid which was considerably reduced by solidifying in small lubing. Analysis of the convection indicates that it probably vesulls from a horizontal temperature gradient produced by the Joule heating- of the electvic crirver~t. Due to the strong dependence of this convection upon the specimen radius, it is conclnded that only for specit~ens with radii of less than somewhere between 1 and 2 mm wilt electric field freezing be getlerally effec-tire at enhancing the purification of zone-melting experittierats in loe-melting alloys. Experiments on 0.8-mm lubes illustrate some interesting potential applications of this techniqne as a means of solute control on smcrll specimens. In a number of papers published within the last few years'-4 it has been found that passage of large dc currents through the solid-liquid interface of a freezing alloy produces a change in the solute redistribution accompanying the solidification. It had been predicted theoretically by Hucke et al.s and by Pfann and Wagner that such an effect would occur as the result of electrotransport within the liquid boundary layer at the solid-liquid interface. These authors independently derived the following relation between the effective distribution coefficient and the variables of the process: Where k0 is the equilibrium distribution coefficient, V is the mobility of the solute relative to the solvent, E is the electric field intensity, D is the solute diffusion coefficient, R is the rate of interface motion, and 6 is the boundary layer thickness. Both groups of authors show that this equation predicts the possibility of extending the purification ability of zone-melting experiments by the addition of an electric current. However, there are other effects in addition to electro-transport which could be produced by the high dc current and might, perhaps, invalidate the applicability of Eq. [I]. For example, high electric current might affect the interface reaction or perhaps the equilibrium distribution which was assumed to exist at the interface. It was the purpose of this work to determine the applicability of Eq. [I] by systematically investigating a simple system for which all of the necessary parameters were known. It was of particular interest for speculating on the potential usefulness of this technique to determine the maximum practical field intensities obtainable, and the effect of these high fields upon convective mixing in the liquid. The initial work in this study was published in a previous paper3 which will be referred to as I. In that work, a current density of 2000 amps per sq cm was used with specimen diameters of 5 mm. The results gave a qualitative confirmation of Eq. [I.], and also showed that a large amount of convective mixing was produced in the liquid by the current. The present study is an extension of that work to smaller specimen diameters and higher current densities. Two points were overlooked in I and will be emphasized here: a) the value of ko has not been well-established in the Sn-Bi system and b) the observed convection could be due to thermal convection produced from the Joule heating. EVALUATION OF k0 To evaluate Eq. [ I.] it is necessary to know the value of the equilibrium distribution coefficient, k,. Tn I the value of k, was taken as 0.3 from a published phase diagram.7 According to ansen' the best data on the phase boundaries of the Sn-Bi system are those of Davidson.5 The data of Davidson give a k, value ranging from 0.21 to 0.25 for the compositions studied in this work, or an average value of around 0.23. In a recent publication, Rigaud and Tougas'10 determined ko by means of a solidification technique. For an alloy of approximately the same composition as that used in this work they obtained a ko value of 0.39. Hence, there is considerable uncertainty in the value of ko and, consequently, a series of experiments were made to determine k, more accurately.

Jan 1, 1968

By C. Feng, I. R. Kramer

A study was conducted to determine the influence of the surface on the yield point of fcc metals and high-purity iron. For the high-purity fcc metals, the yield Point produced by restraining a specimen may be eliminated by removing a sufficient amount of the layer. A yield point similar to that in fcc metals was found when iron specimens were pre-strained after a prior deformation. This yield point is not due to an aging effect but is associated with the surface layer. The yield points in age-hardenable aluminum alloy and those connected with "work softening" are not affected by a surface-removal treatment and are, therefore, caused by bulk-type dislocation reactions. The yield point which appears in previously strained fcc metals has been the subject of numerous studies. Several theories have been advanced to explain this "unloading-reloading" yield point. Cottrell and stokes,' from their study on the effect of temperature on plastic properties of high-purity aluminum, considered that it was the result of an avalanche "breakdown" of Lomer-Cottrell sessile dislocations. Haasen and Kelly2 extended this concept of dislocation-dislocation interaction to account for the yield point when unloading and reloading was carried out at the same temperature. On the other hand, strain aging was found to be responsible for the yield point in A1-Mg alloy by Westwood and Broom,3 and in aluminum alloys containing various amounts of alloying elements by Smallman, Williamson, and Ardley.4 In these cases, the yield point is considered to be associated with the locking of dislocations by impurity atoms. Birn-baum and filer5,6 and Takamura and muira7 reported that, when a high-purity copper specimen was prestrained, then held at a lower stress, and finally reloaded at the prestrain temperature, a larger yield point was observed. Such an increase was attributed to the locking of dislocations by point defects. In addition, Bolling8 suggested that both mechanisms, i.e., dislocation-dislocation interaction and strain aging, are responsible for the yield phenomenon, at least in a brass. Thus, there appears to be experimental evidence for both the strain aging and dislocation-dislocation interaction theories to explain the yield point in the fcc metals and alloys. In a previous article,9 one of the authors observed that, if sufficient amounts of metal were electrolti-cally removed after deformation of a high-purity aluminum single crystal, the yield point did not appear upon reloading. Continued pursuit in this area revealed that the effect of the surface on the yield-point phenomenon is not limited to one metal but affects a variety of metals of different lattice structures. Nor is the effect limited to the yield points produced by simple unloading-reloading at constant temperature; the surface exerts an important role on the yield point even when the temperature is changed at some stage during the test. It is the intent of the present article to give additional information regarding the effect of the surface on the yield-point phenomenon and to show the association between the yield point in both fcc and bcc metals and the existence of a heavy concentration of dislocation in the region near the surface of a deformed specimen.10 EXPERIMENTAL Materials. The single crystals used in this study had a nominal cross section of 0.125 by 0.125 in. and a 3-in. gage length and were prepared by a modified Bridgman technique. The initial purity for the metals was 99.997, 99.999, and 99.999 pct for aluminum, copper, and gold, respectively. Bar stock of these same materials was used in the preparation of the polycrystalline specimens of aluminum and copper. The iron specimens were machined from rods of "ferrovac", reported to have a purity of 99.98 pct. In addition, a precipitation-hardening aluminum alloy, containing 1.5 pct Cu, 2.5 pct Mg, and 5.5 pct Zn, was used. The polycrystalline specimens of iron, copper, and the aluminum alloy had a diameter of 0.17 in. and a gage length of 2 in. while the high-purity aluminum specimen had a nominal rec-

Jan 1, 1965

By C. J. Frosch, J. A. May, H. G. White, C. D. Thurmond

Gallium phosphide epitaxial layers were grown from the vapor phase on undoped single-crystal galliurn arsenide substrates in silica tubes by an open-tube wet-hydrogen process. The epitaxial layers were grown over a range of water pressures at three substrate temperatures. Excess donor concentrations were determined by surface barvier capacitance measurelrzents without removing the layers from the substrates. The excess dmlor concentration, ND-NA, is fo~ind to vary approxilnately inversely with the pressure of water added to the hydrogen carrier gas. This is the relationship that would be expected for singly ionized silicon donors on gallium sites in extrinsic galliunz phosphide, with the silicon coming from the SiO generated by the reaction of hydrogen with the silica tube. An increase in the partial pressure of water in the hydrogen stream decreases the SiO pressure. The results indicate that ni, the intrinsic hole and electrmt concentration for gallium phosphide at the three substrate temperatures, is smaller than the concentration estimated from available data for the density of states effective masses and the energy gap. Mass-spectrographic measurements confirm that the dono?, introduced into gallium phosphide is silicon. The equilibrium concentrations of silicon in vapor-flown gallium Phosphide have been estimated from available thernzodynamic information that includes the solubility measurements of silicon in gallium phosphide in equilibrium with a gallium-rich liquid phase. Satisfactory agreement with the measured silicon concentrations is obtained. FROSCH1 has described an open-tube process for growing single-crystal Gap from the vapor phase by a GazO transport mechanism. The method depends upon the reaction of H20 in an H2 carrier gas with a heated source of polycrystalline Gap which provides the necessary vapor species. When the temperature of these vapor species is lowered, super saturation occurs and single-crystal Gap will deposit on a suitable substrate. Unintentionally doped single crystals of Gap grown by the wet H2 process in silica tubes are n type. Evidence is presented to show that the donor introduced is silicon, and that a qua si-equilibrium model accounts for the inverse dependence of the donor concentration on the water partial pressure and predicts the magnitude of the donor concentrations. Ainslie et al. experimentally showed a similar inverse relationship between the carrier density and oxygen pressure for GaAs. Emission-spectrographic analyses showed a decrease in the silicon concentration with increasing oxygen overpressure for GaAs. Cochran and Foster suggested the theoretical possibility of suppressing silicon contamination by using Ga20 generated by the reaction of gallium with water vapor. 1) EXPERIMENTAL The apparatus and procedures are essentially the same as those described by Frosch.' The apparatus consists of a 25-mm-ID SiO2 tube extending through a controlled high-temperature flat zone for the location of the polycrystalline Gap source and a downstream temperature gradient falling at a rate of about 14°C per cm. The latter provides the region of super saturation for the location of the single-crystal substrate. The partial pressure of water in the inflowing hydrogen stream, pA2, O was controlled by mixing me-tered proportions of dry H2 with H2 saturated with H2O vapor at 0°C. The total gas flows were about 200 cu cm per min in all experiments. The Gap sources were prepared by pulverizing boat-grown polycrystalline ingots to pass a 20-mesh sieve. The substrates were cut from an undoped single-crystal boat-grown GaAs ingot purchased from Monsanto. This ingot had a carrier concentration of about 1015 atoms per cu cm, a resistivity of about 5 ohm-cm, and a mobility of about 5000 sq cm per v sec at 25°C. Substrates with dimensions of 1 by 1 by 5 x lo-' cm were employed. The growth faces were chemically polished (111) arsenic faces. Epitaxial layers, at least 7.5 x 10-3 cm thick, were grown,. This required from 1 to 24 hr depending upon the Pii2Q values and the temperatures. In all of the runs, the source temperatures were 50°C higher than the substrate temperatures. Samples were prepared

Jan 1, 1968

By G. Bruggeman, K. G. Kreider

Grain boundary dij]usion coefficienls were measured in tungsten between 1400° and 2200° C and can be expressed by the equation sq cm per sec This activation energy confirms some eavlier estimates made .from tungsten sintering experiments. Grain boundary diffusion was found to occur in sub-bozrndavies having -misorientations of less than 10 deg. The actiuation energy for this subboundavy diffusion is equal to that for dijjusion in incoherent grain boundaries with in the limits of error. This is shown to be consistent with the dislocation model of Low-angle boundaries wheve diffusion occlcvs along- the dislocation 'YPipes" comprising -tile boundary. RECENT investigations of the sintering of tungsten powders all report activation energies which are considerably less than the activation energy for tungsten volume diffusion. Kothari' reports a value of 100 * 5 kcal per mole, Hayden and Brophy' obtained 90 kcal per mole, and Vasilos and smith3 found 110.7 kcal per mole from their sintering studies. Since most determinations of the activation energy for volume diffu-sion4-' fall between 120 and 160 kcal per mole (the true value seems most likely to be nearer 150 kcal per mole), the conclusion is drawn that the mass transport leading to densification during sintering is accomplished by grain boundary diffusion. This interpretation is consistent with various diffusion models of the sintering process. 10-12 Vasilos and Smith calculate diffusion coefficients from their data which fit the equation D * 1.36 x 10* exp(-llO,700/HD However, no direct measurements of tungsten grain boundary diffusion have been made. Furthermore, considerable disagreement exists between the directly measured values of tungsten volume diffusion.'-' In order to corroborate the inferred results of the sintering experiments concerning grain boundary diffusion and to provide accurate diffusion data essential to the analysis of the kinetics of creep, oxidation, precipitation, and so forth, the present work was undertaken to measure self-diffusion in single-crystal and polycrystalline tungsten between 1400" and 2200°C. It is within this temperature range that tungsten sintering is done, the re crystallization of tungsten occurs, and the widest application of tungsten as a high-temperature material will probably be made. EXPERIMENTAL PROCEDURE Radioactive WlE5 was produced by irradiating tungstic acid in a neutron flux of 1.2 x 1012 neutrons per sq cm per sec for 36 hr. A 2-week waiting period was allowed for the decay of w"~ also produced by the irradiation. (w"~ has a half-life of 24 hr.) The half-life of the remaining isotope was determined to be 75 days confirming the presence of w lE5 and the absence of any undesired radionuclide. Specimens 4 in. in diam and $ in. thick were cut from polycrystalline swaged tungsten rods (recrystal-lized) and from Linde single-crystal rods. Chemical analyses of these materials appear in Table I. Actually upon closer examination, the single-crystal specimens were found to consist of several subgrains separated primarily by tilt boundaries in which the misor-ientation ranged from 3 to 10 deg. Thus, it was possible to measure boundary diffusion coefficients in these low-angle subboundaries as well as in the incoherent boundaries of the polycrystalline specimens. The two faces of each specimen were ground flat and parallel within 0.0001 in. The radioactive tungstic acid was dissolved in concentrated ammonium hydroxide, placed on the ground flat of the specimen, and evaporated to dryness. The oxide was then reduced in hydrogen at 1000°C resulting in a layer of wlE5 approximately 1 p thick. The diffusion anneals were performed in vacuum in a tantalum resistance furnace. Time at temperature ranged from 10 hr at 1400°C to 2 hr at 2200°C. The penetration profile was determined by measuring the residual activity after successive removal of surface layers by grinding on metallographic polishing paper. Extreme care was exercised to insure that sections were always taken normal to the diffusion direction; this was verified repeatedly by checking that front and back surfaces of the specimen remained parallel. The activity was measured with an end-window Geiger-Mueller counter. The sides and edges of the specimen were well-shielded to eliminate possible effects due to surface diffusion. The weight of the

Jan 1, 1968

By W. J. Moore, E. M. Swiggard, H. Shenker

Beryllium is the most soluble of the double-acceptor impurities in germanium; the solubility is at least 1 x 10 19 atoms cm3. Photoconductive, optical, and electrical measurements were made on a set of Ge:Be samples grown without other intentionally added imp~ivities. The excited-states spectrum of beryllium in germanium was determined both in zero field and in magnetic fields up to 30 kc. It was found that the absorption 0-f lines in the vicinity of the G line increased markedly with increasing magnetic field. The ioniza-tion energy of neutral beryllium in germanium was determined to be 24.3 * 0.1 mez: using the excited-states spectrum. An infrared detector using a Ge:Be element was found to have a normalized detectivity D* (50 p, 1000 cps, 1 cps) of 9 X ldO cm-~ps"~-w-'. The Ge:Be sambles exhibited low-field breakdown and inibzirity band conduction effects at low temperatures. WHILE the properties of most dopants in germanium have been thoroughly investigated, the properties of beryllium in germanium have had relatively little attention. Much of the work that has been done' has been concerned with the properties of singly ionized beryllium in germanium. The purpose of the present work was to determine the photoconductive and electrical properties of beryllium-doped germanium (Ge:~e) with a small concentration of other electrically active impurities. Ge:Be is of special interest since beryllium is the most soluble of the known double-acceptor impurities in germanium. MATERIALS PREPARATION A set of &:Be crystals with beryllium concentrations between 5 x 1013 and 1x 10'' cmm3 and without other intentionally added impurities was grown in a hydrogen atmosphere by the horizontal zone-refining techniques described by Swiggard and Shenker.2 The segregation coefficient of beryllium in germanium was found to be about 0.07 and its maximum solubility was found to be greater than 1 x 10'' atoms cm-3 which is in agreement with previous results.' Most of the crystals were found to have a small net concentration of shallow acceptors; for example, a crystal having a beryllium concentration of 1.1 X 1016 cm-3 was found to have a net concentration of shallow acceptors of less than 1.5 X 1013 cm-3. Ingots designed to have a beryllium concentration of 1 x 1017 cm-3 or less exhibited a loss of electrically active beryllium. That is, the beryllium concentration was lower than expected and the decrease of the beryllium concentration along the length of the ingot was much larger than expected. It is hypothesized that this effect is due to the oxidation of the beryllium by the residual oxygen in the system. This effect is similar to but not as severe as the effect observed with boron in germani~m.3, 4 OPTICAL ABSORPTION SPECTRUM The excited-states spectrum of neutral beryllium atoms in germanium was studied by measuring the infrared optical absorption spectrum. A typical spectrum is shown in Fig. 1; the labeling of the absorption lines follows that of Fisher and Fan.* The spacings between the observed lines and the D line are nearly equal to the corresponding spacings for the other ac -ceptors in germanium.5-7 By the use of the theoretical result67' that the neutral impurity ionization energy is 2.53 mev larger than the energy of the D line, the ionization energy of neutral beryllium in germanium was determined to be 24.3 * 0.1 mev at 4.2"K. This method yields a more precise value of the impurity ionization energy than the usual Hall effect method.' The absorption spectrum of Ge:Be was also measured as a function of magnetic field using a magnetic field parallel to the incident radiation and using unpolarized radiation. A typical spectrum taken at 30 kG is shown in Fig. 2. An unexpected observation is that the absorption of the lines in the vicinity of the G line (there is a strong possibility that some of these lines may be associated with lines other than the G line) increases with magnetic field. RESISTIVITY AND HALL DATA Measurements of Hall voltage vs temperature were made on several of the samples. The thermal activa-

Jan 1, 1968

By Nicholas J. Grant, A. W. Mullendore, Paul Predecki

The Duwez technique of splat cooling in which a molten droplet of metal is accelerated and made to impact on a cold, highly conducting substrate was investigated.- An apparatus for producing "splat" was constructed employing an explosive powder charge to accelerate the molten metal. Transport of the molten metal just prior to impact with the substrate was studied by means of high-speed photography. The molten particles are small spherical droplets from about 1 to 50 µ diameter. The average cooling rates for aluminum, silver, and a gold alloy splatted on nickel substrates at room temperature were determined experimentally and were found to vary from 18 to 5 x 10' C per sec. The heat-transfer coeficients for pure aluminum and pure silver splats cooled on nickel substrates at room temperature were found to be 2.7 to 6.8 and 13.6 to 54.2 cal per sec sq cm 'C (2 to 5 x 104 and 1 to 4 x 18 Btu per hr sq ft OF), respectively. Solidification rates in pure aluminum and silver splats were calculated. "SPLAT cooling" is a term describing a technique for extremely rapid freezing and cooling of molten metals and alloys to room temperature or below. The technique was developed by Duwez et al.' in 1960, and after refinement2 consisted in transferring a few tenths of a gram of molten metal to near sonic velocity to strike a suitably placed cold copper surface. Upon impact, the metal spread into a thin nonuniform film called a splat, about 10-4 cm thick. The splat particles produced in this manner were thin enough in some areas to be suitable for transmission electron microscopy, without further treatment, and, together with X-ray data, yielded a variety of structures which would be classified as follows: 1) supersaturated solid solutions (increase in solubility limit), 2) metastable crystalline stoichiometric and non-stoichiometric intermediate phases ( in Au-20.5 at. pct Si and in Au-14 at. pct Sb, respectively), 3) amorphous alloys, 4) retained high-temperature phases, 5) alloys having equilibrium phases present, but with unusual, markedly altered structures. The crystalline phases present in splats were usually extremely fine-grained and had low dislocation densities. Present interest in splat cooling centers around a study of the unusual structures produced by the technique, their contribution to alloy theory, and the possibility of the development of new or unusual properties. In addition, the technique is being examined for the production of bulk quantities of new high-strength alloys for low- and high-temperature use. Although a number of alloy systems have already been investigated by the technique, relatively little is known about the mechanism of splat formation and the physical conditions encountered by the metal during solidification and cooling. The purpose of this work was: 1) to determine the velocity, shape, and size of the molten metal droplets just prior to impact with the cold substrate; 2) to try to estimate or measure solidification rates, heat-transfer coefficients, and cooling rates in splats. EXPERIMENTAL TECHNIQUE Splatting Apparatus and Procedure. An apparatus similar to the type originated by Duwez et a1.,2 but employing an explosive charge to accelerate the molten metal, was used. A "Ramset" fastening tool (a gun normally used for driving studs into concrete, steel, and so forth) was mounted vertically above a resistance-heated graphite crucible as shown schematically in Fig. 1. A splat product was produced by melting a few tenths of a gram of metal at the position shown in Fig. 1, and then exploding a powder charge in the breach. The shock wave thus generated traveled down the barrel into the furnace, ejecting the molten metal through the 0.06-in.-diameter hole at the bottom of the crucible. The ejected metal was impacted on a high-conductivity metallic substrate where it formed a splat. High-Speed silhouette Photography. In order to investigate the size and shape of the ejected, molten metal just prior to impact with the cold substrate, high-speed silhouette photographs were taken, and are shown in Figs. 2 and 3. These photographs were obtained with the aid of an Edger-ton, Germeshausen, and Grier microflash unit and a submicrosecond flash drive equipped with a variable time delay. Two types of triggers were

Jan 1, 1965

By B. J. Bayles, J. A. Ford, M. J. Salkind

Unidirecltonally sulidijied AL-A13Ni was found to exliibit excellent microstructural stability to 508°C. Above this temperature the transverse micvostructure coarsened in a manner analogous to "Ostwald Ripening", but the A13Ni whiskers did not spheroidize or foreshorlen. The coarsened whiskers reinforced the matrix as effectirely as the finer whiskevs and no reduction in room-temperature tensile strength mas noled after exposure for 90 hr at 96 pel 0-1 the absollite melling temperature of the eulectic. The slability was auribucted to the presence of low-index (and presurrlably lor-energy) cvystallographic interjaces between phases. The driving force for microstructural coursening was presumed to be the reduction of interfacial energy between the phases. Finer micvoslructures were found to coarsen more rapidly than those which weve imitally coarser. FIBER-reinforced metals have been shown to exhibit good short-time tensile properties at elevated temperatures.1-4 However, many elevated-temperature applications require long-time property retention which dictates the necessity for microstructural stability. Two types of instability are common to fiber composities. The first, chemical instability, is due to chemical reaction between the matrix and the relnforcing phase. Petrasek and weeton5 have shown that chemical reaction between tungsten wire and several copper alloy matrices resulted in degradation of the reinforcing phase and reduction in strength of the composite materials. Kreider and Leverant4 have noted similar behavior for boron-reinforced aluminum above 400°C. The second type of instability arises in systems in which the phases are essentially chemically stable with respect to each other and is characterized by spheroidization and/or agglomeration of the reinforcing phase. parratt6 has reported this type of degradation to occur relatively rapidly at moderate temperatures for nickel and cobalt alloys containing whiskers of silicon nitride, aluminum oxide, and silicon carbide and has identified it as "physicochemical" instability. Although both types of instability are the result of the differences in the chemical potential between the fibrous phase and the matrix, they may be differentiated by the fact that the second type, "physicochemi-call' instability, results in little or no net change in the composition or amount of each phase present. It does, however, result in a morphological change which profoundly affects the strengthening mechanism. Chemical interaction. on the other hand, results in a change in the composition and amount of one or more of the phases present, or the appearance of a new phase. In the latter case, the appearance of the mi-crostructure, that of a fibrous phase aligned in a matrix, may remain essentially unchanged, but the chemical degradation of the fiber results in markedly reduced composite strength. This investigation was initiated in an attempt to evaluate the microstructural and related mechanical property changes of a whisker-reinforced eutectic composite after elevated-temperature exposure. The composite system chosen for this study was the uni-directionally solidified A1-A13Ni eutectic which Lemkey. Hertzberg, and Ford7,8 demonstrated to behave as a whisker-reinforced composite. The microstructure of the unidirectionally solidified material consists of aligned needles of A13Ni in a matrix of aluminum. EXPERIMENTAL PROCEDURE Eutectic ingots having a nominal composition of 6.2 wt pct Ni were produced by melting 99.99+ pct A1 and 99.99 pct Ni in recrystallized alumina crucibles in an argon atmosphere. The analyses of the starting materials are listed in Table I. These master heats were cut up? melted by induction heating, and unidirectionally solidified vertically in 1/2 -in.-ID. 6-in.-long graphite crucibles, in a dynamic argon atmosphere. Therma1 gradients in the liquid were determined by measuring the temperature in the melt as a function of apparatus travel distance, as discussed by Kraft and Albright.9 The rate of solidification was assumed to be equal to the uniform rate at which the crucible assembly was withdrawn from the furnace. Ingots representative of nominal growth rates of 2, 5, and 11 cm per hr were produced and the solidification parameters are summarized in Table 11. Ingots used for met allographic analysis were pre-

Jan 1, 1968

By A. R. Troiano, A. B. Greninger

There is need for an adequate working hypothesis that would describe at least qualitatively the crystallographic mechanism for the transformation from austenite to martensite in steel. A general theory would not only be of great assistance in the correlation of existing data and in planning future experimentation, but would also form the basis for an eventual explanation of the property changes that accompany lattice transformations. The mechanism of martensite formation described in the following pages is the outcome of two new experimental determinations: (1) the accurate evaluation of the lattice relationship between austenite and individual crystals of martensite, and (2) the measurement and analysis of the change in position that a volume of austenite undergoes when it transforms into a crystal of martensite. For this latter study, the stereographic analysis of shear was employed; this method greatly simplifies the solution of lattice-shear problems, and some space is devoted to its description. Lattiice Relationships and Habits PREVIOUS WORK The lattice relationships between austenite and martensite in carbon steels have been studied by Kurdjurnow-and Sachs,' and by Wassermann;2 and in iron-nickel alloys by Nishiyama.3 Wassermann,2 and Mehl and Derge.4 The studies by Kurdjumow and Sachs and by Wassermann were made on single (austenite) crystals of quenched 1.4 pct carbon steel, and the results of these two independent studies agree closely with the postulated lattice relationships: (III)r||(0ll)a [101]r [111]a [1] Both Nishiyama and Wassermann worked with an alloy of iron plus 30 pct nickel (all austenite at room temperature), and "martensitic alpha" was formed by cooling in liquid air. They agree that the lattice relationships for these conditions are: (lll)r||(011)a [112]r[011]a [2] Mehl and Derge worked with a range of iron-nickel alloy compositions centered about 30 pct nickel. They concluded that when the transformation takes place near room temperature or above, relationship [l] holds, but if the same specimen is made to transform at -195°C, relationship [2] obtains;intermediate temperatures evidently produced a combination of these two relationships. Sachs and co-worker~,1,7 as well as Nishiyama,3 have proposed transformation mechanisms for martensite formation.* Both of the mechanisms picture the transformation as initiated by shearing movements along the octahedral [Ill] plane of austenite, followed by suitable expansion and contraction to complete the transformation. These mechanisms have been described and commented upon by Mehl, Barrett, and Smith,8 Mehl and Derge,4 Wassermann,2 and others, and generally accepted by investigators in this field. Greninger and Troiano9 have recently published the results of a detailed study of the shapes and orientation habits of martensite crystals in plain-carbon and nickel steels. They found that, contrary to previous assumptions, martensite plates form not along the octahedral planes but along austenite planes of very high indices; these results were in line with those of previous studies of martensite reactions in nonferrous alloy systems by Greninger and coworkers.6,10 Greninger and Troiano concluded that the transformation theories of Kurdjumow and Sachsl and of Nishiyama3 are untenable. Phragmén11 reached a similar conclusion. EXPERlMENTAL METHODS The object of this particular study was the evaluation of (1) lattice relationship between austenite and martensite, and (2) relationship between the martensite lattice and the martensite plate. The technique used for solving these problems was similar to that used by Greninger6 on martensitic structures in 8 copper-aluminum alloys. Briefly, the method consists of grinding and polishing a specimen on a surface parallel to an individual martensite plate and thus exposing a single martensite crystal for an area of l to 3 mm. The orientation of this exposed martensite crystal is then determined by means of a back-reflection Laue X ray pattern.12 Another pattern is obtained from the matrix crystal, and a stereo-graphic plot of data from these two patterns then provides the solution to the problem. It is necessary to repeat this process on several crystals in order to determine whether or not the solution is unique.

Jan 1, 1950

By K. Mukai, M. Ishihara

This report is a brief discussion of the impurity levels both in primary aluminum and super-purily alnminim in connection with raw materials and proc-essing methods. Particularly, truce amounls of im-purities were analyzed by introducing nittss-speclro-gruphy and nenlron-aclivalian analysis techniques. Vanadium and gallium content in primary aluminum metal corresponded to theor amounls in anode coke. but the nickel content in primary aluminum was hide-pendent oj its amount in the anode cake. Sulfur in the anode colic might result in an increase of the iron content in the molten metal in the reduction cell. The content of sodium and calcium increased with increasing bath ratio. In the holding furnacc sodium content decreased rapidly, and nonmetallic inclusions merit separated front the mollen metal. Three-layer elctrolysis was effectire in eliminating many impuri-ties with the exception of copper, aluminum oxide. aluminum nitride, and some rare-earth elements. THE amount of impurities and nonmetallic inclusions in aluminum metal depends on impurities in the raw materials and various production processes. This paper presents an investigation of the behavior of impurities and nonmetallic inclusions in aluminum during the electrolysis process, the period in the holding furnace, and the process of three-layer electrolysis. Chemical analysis and emission spectrometric analysis have generally been used as methods to determine small amounts of impurities in both primary aluminum and super-purity aluminum; however. mass spectrography and neutron-activation analysis have been recently introduced into this field, and trace amounts of impurities can accurately be determined by these latter methods. EXPERIMENTAL WORK Common impurities were determined by means of absorptiometric. polarographic, flame-photometric. and emission-spectrometric methods. However, in the determination of trace amounts of impurities and non-metallic inclusions, the following methods were applied. In the analyses of trace impurities by mass-spec- trographic analysis, samples were exposed in the vacuum spark of double-focus ing Mattauch-Herzog type mass spectrograph (Japan Electron Optics Laboratory, Ltd.). Ilford Q2 photographic plates were used as ion detectors. Fe was used as the internal standard, and the atomic concentration of each element was calculated by the following expression and converted into the weight concentration: where E, and Ei are the exposures (coulomb) when isotopes of the internal standard and impurity elements (on which the estimate is based) show the same density, X is the percent (atomic) concentration of the standard, IF and Ii are the percent abundances of the isotopes of the standard the impurity elements (on which the estimate is based), and Ms and hfi are the mass of the isotopes of the standard and impurity elements (on which the estimate is based). The coefficient of variation in this method is about 40 pct in the range of 0.1 to 1 ppm. In neutron-activation analysis, irradiations were done in the nuclear reactors TRIGA-I1 (neutron flux: 4.7 x 10 11 n per sq cm per sec; Rikkyo University) or JRR-2 (neutron flux: 2 x 1013 n per sq cm per sec: Japan Atomic Energy Research Institute). After the chemical separation, a NaI (3 by 3 in.) scintillator connected to a 256-channel y-ray spectrometer (R.C.L.) was used for counting and recording various activated impurities. The coefficient of variation in this method is about 20 pct. In the analyses of nonmetallic inclusions, aluminum-oxide determination was carried out according to Fischer's method, 3 as follows. A slice of sample was cut off and the surface was cleaned. It was dissolved in bromine-methanol. The solution was filtered and the residue was washed. The residue was fused with potassium bisulfate flux. Then it was dissolved in weak acid, and oxine solution was added. The aluminum oxinate was extracted with benzene. Finally, aluminum was determined absorptio-metrically. Aluminum nitride determination was carried out as follows. A slice of sample was cut off and the surface was cleaned. It was put in the distillation apparatus and decomposed with sodium hydroxide solution. The solution was distilled. Then ammonia was absorbed and

Jan 1, 1967

By R. F. Mehl, R. F. Bunshah

A fast amplifier technique has been developed for the measurement of the rate of propagation of martensite in an Fe-29.5 pct Ni alloy. The time of formation of one plate of martensite is 3x10 sec and the rate of propagation is 3300 ft per sec approximately. IT has been known for some time that the plate-like structural unit of martensite forms from austenite with great rapidity. Wiester1 and Hane-mann, Hofmann, and Wiester took motion-pictures of the transformation as it occurs in a 1.65 pct C steel; they demonstrated that a single plate formed fully in the time interval between successive frames, viz., 1/20 sec, thus setting an upper limit. Forster and Scheil,3 using an Fe-Ni alloy with 29 pct Ni, recorded the sonic characteristics of the process electrically, upon an oscillograph, setting the upper time limit at 0.002 sec. Forster and Scheil,~ measuring the change in electrical resistance in the same alloy upon a cardiograph, set a limit of 0.02 sec. Forster and Scheil5 later, employing the same alloy, improved their technique, reporting an upper limit of 7.10 sec. In studying signals of such short duration, it is an important question whether the frequency response of the electrical system used is high enough compared to frequency of the pulse measured, or, put differently, whether the system is able to reproduce without distortion the signal arising, in this case, from the formation of a single martensite plate. Forster and Scheil (referring only to their last paper) obtained signals of a frequency of 30 kilocycles (hereinafter kc); this was about the frequency response of the equipment used; thus, if the signal had a frequency higher than 30 kc, it would still appear as a signal of frequency 30 kc. All of these results thus provided upper limits only. Recent developments in electronics have made available equipment with very high frequency response, very high sweep-speeds, high gain, etc. The electrical characteristics of such equipment, used in the present study, are given in Table I. Such equipment offers obvious attraction in the study of the rate of propagation of a martensite structural unit—and perhaps of other structural alterations proceeding at a very high rate. This paper reports an attempt to develop a technique employing such equipment to measure the time of propagation of a martensite structural unit and the variation of this with temperature, with the mode of formation—athermal and isothermal—in both polycrystalline and single-crystal samples; and from such measurements to obtain the rate of propagation. As will be seen, the results obtained are useful theoretically. Materials All data presented here are for an Fe-Ni alloy of the following analysis: 29.5 pct Ni, 0.027 pct C, 0.135 pct Mn, 0.094 pct Si, balance Fe. There were several reasons for choosing this alloy: 1—it is substantially the one used by previous investigators; 2—it exhibits both the athermala and the isothermal' mode of formation of martensite, both studied in detail by Machlin and Cohen; 3—the subzero temperatures of transformation in this alloy are experimentally very convenient; 4—it exhibits the "burst phenomenon";" 5—the change in electrical resistance upon the formation of martensite, a decrease, is great, approximately 50 pct.' The polycrystalline specimens were in the form of wires of 0.025 in. diameter; the single crystals were 1x1/4x1/4 in. Experimental Methods Electrical Apparatus: Fig. 1 is a schematic drawing of the electrical circuit used. The principle used in these measurements is the same as that used by Forster and Scheil." A small direct current, about 1 to 2 amp, is passed through the sample. When a martensite plate forms, the resistance of the sample changes and a high frequency signal is generated. This signal is picked up by the probes attached to the sample, fed into the bank of amplifiers and thence to the vertical deflection plates of a cathode-ray oscilloscope. The signal itself triggers the oscilloscope trace which flashes across the tube face and is photographed by means of a 35 mm movie camera at the end of a light-tight hood. The camera has no shutter. As soon as the signal flashes

Jan 1, 1954

By R. T. O’Brien, J. M. W. Mackenzie

A microelectrophoresis technique has been used to measure the zeta potential of quartz over a range of pH and Ni (11) and Co (Ilj concentrations. Results have been discussed in terms of adsorption of Ni (11) and Co (11) cations and hydroxide precipitates. Possible relationships between the experimental results and the formation of lateritic nickel deposits have been discussed. The adsorption of metallic cations to the surface of silicate minerals is of significance in many industrial and natural processes. In an earlier publication,' the adsorption of Fe (111) to the quartz surface, a phenomenon of importance in flotation, was studied qualitatively using microelectrophoresis techniques. This paper reports the extension of this work to an investigation of the adsorption from aqueous solution of Co (11) and Ni (11) to the quartz surface. The adsorption of these metals was of interest for two reasons; firstly, the importance of such an adsorption process in the formation of nickel deposits by the lateritization of nickeliferous rocks, and secondly, the comparison of electrophoresis data for these metals, whose hydroxides precipitate in the pH range 7-8, with similar data for Fe (111) where the hydroxide first precipitates at pH 2.8. The development of laterites is essentially a near-surface geological process involving vertical and, in part, horizontal redistribution of elements derived from rock-forming minerals. This redistribution of elements leads to the formation of layers within the lateritic profile which are enriched in one or more of these elements and, depending on the nature of the original rocks, continued lateritization may give rise to iron ores, bauxites or nickeliferous laterites. The mechanisms discussed in this paper deal with the last mentioned type of deposit although the considered factors are applicable to laterites in general. Nickeliferous laterites are derived from basic and ultrabasic rocks containing ferro-magnesian silicates. These rocks may be partly or wholly converted to serpentinites. Besides their high Mg, Si and Fe contents, they also contain small amounts of Mn, Al, Cr, Ni, Co, Cu and Zn. The nickel content is usually near to 0.2-0.3%, but for economic use this must be increased during lateritization to over 1% and preferably over 2%. The zone of nickel enrichment is usually at some depth within the lateritic profile. Unpublished work by Mr. O'Brien on nickel-bearing laterites in eastern Australia indicated four visually different types of material which were separated from crushed samples under a stereo-microscope. Although chemical analyses indicated that all four types of fragments contained 2-4% nickel and smaller amounts of cobalt, X-ray diffraction and optical studies revealed only quartz, hematite and chromite as well as, in one of the types, a serpentine mineral. No distinct, nickeliferous minerals were detected. Furthermore, it was possible to extract some 60% of the nickel content of the laterites with cold 0.5N hydrochloric acid. It was therefore postulated that much of the contained nickel was present as amorphous, adsorbed material. In considering this fact, as well as the problem of the actual method of transfer and fixation of elements within the lateritic profile, it was further postulated that the transfer of these elements probably took place partly as solutions and partly as colloids and that the fixation of these was caused by adsorption and flocculation. The existence of nickel in solution in water draining from serpentinites has been confirmed by Smurov.2 Although we have no evidence for the transportation of nickel in the colloidal state, it is well accepted that colloids are important in the formation of laterites. Previous work by Mr. Mackenzie' has already indicated that iron may be adsorbed on quartz under conditions which could be expected in a lateritic profile, but no information was available for nickel or cobalt. Although adsorption possibly also takes place on minerals other than quartz, the latter is significant because silica is an important constituent of the silicate parent rocks. During lateritization, it is released to form free quartz and it is likely that it is initially extremely fine-grained. This would provide abundant opportunities for adsorption and it was de-

Jan 1, 1970

By R. F. Mehl, C. E. Birchenall

SINCE Maxwell1 first considered the self-diffusion process in 1872 its importance in the kinetic theory of matter has been recognized. Until the discovery of isotopes in 1913, a direct measurement of this quantity seemed impossible. The only information that could be gathered indirectly was for gaseous systems for which the kinetic theory was well developed. When methods of direct observation be-came available, investigations on self-diffusion were carried out in condensed phases. In metals these data are presumably of potential importance in the study of recovery, recrystallization, creep, sintering, and related phenomena. Following the pioneer work of Von Hevesy2 and his collaborators, who determined the rate of self-diffusion in lead with the naturally occurring thorium B isotope, several metals have been investigated. Pb, Ag³, Au.6,5, and Cu6,7,8,9 are examples of isotropic crystals existing in only one phase modification. Anisotropy of diffusion has been demonstrated in Bil6 and Znll. This paper is a study of a single metal in two allotropic forms and also of self-diffusion in a body-centered cubic lattice. Despite the fact that the measurements in each of the papers cited on self-diffusion in copper seem to be internally consistent to about 10 to 15 pet, the curves reported by different authors differ by factors of 2 to 4 at the same temperatures. This uncertainty may account, in part, for the failure to establish a successful correlation between the self-diffusion rates and other physical characteristics of the metals. No satisfactory theory of metallic self- diffusion has as yet been proposed to account for all the existing data and to permit estimation in other metals. Experimental Part: Two units of radioactive iron were used in these experiments, each a mixture of Fe53 and Fe59 (12). Fe53 decays by K electron capture and the emission of an X ray with a half life of about 4 years. Fe59 emits two beta spectra, one with a maximum energy of 0.26 Mev, the other 0.46 Mev, and gamma rays of 1.1 and 1.3 Mev, with a half life of about 44 days. They were present in nearly equal concentration initially. The composite half life started at about 50 days and increased as the Fe59 decayed, leaving Fe55 relatively more abundant. After aging a year, the half life was too long to measure significant decay in a month. The absorption properties also changed with time. Because of the importance of the absorption coefficient in the diffusion calculations, it was necessary to determine this quantity frequently over the period during which these experiments were carried out. This correction was unsuspected at the time of publication of notesl3 on this work and accounts for the discrepancy in alpha iron and for part of the discrepancy in gamma iron.* * The data in the present paper supersede those given in the notes entirely. In one note the scale of log D was inadvertently reversed. In addition to the correction for the drift in absorption coefficient, the D values for gamma iron at low temperatures were high owing to insuficient correction for diffusion occurring during heating and cooling through the high temperature part of the alpha range at a rate much slower than that employed in the runs reported here. The high temperature alpha rates are much higher than the low temperature gamma rates and correspond to large equivalent times at these temperatures. These experiments were discarded and new measurements taken. Independent experiments with the same activity units indicated a radioactive contaminant in the iron, but exhaustive attempts to isolate and identify it chemically failed. These experiments? indicate † These experiments will be discussed in a later publication from this laboratory. that the contaminant represented only a trace of little importance in the diffusion results. The active iron was plated from an iron chloride solution. Enough of the solution was dropped on a 1% in. square of filter paper to wet it thoroughly.

Jan 1, 1951

By J. P. Hirth, A. H. Clauer

A simple graphical method is presented for the determination of climb forces on dislocations exerted by uniaxial stresses. In conjunction with standard stereographic projections, the technique is applicable for any crystal system. Climb forces on glide dislocations in bcc crystals are determined as an application. In connection with a study1 of elevated temperature creep of molybdenum single crystals, a detailed analysis of the climb force on a dislocation was necessary. In general, such forces can be calculated for an arbitrary dislocation segment from the Peach-Koehler equation2 (or one of its variants3"6) where Fk are the components of the force per unit length of dislocation line, ?i of the unit sense vector tangent to the line, ojl of the stress tensor, bl of the dislocation Burgers vector, and is the Einstein permutation operator with components ?123 = ?231 = ?321 = —?132 = — ?321 = —?—213 = unity, and all other components zero. It is understood in the suffix notation of Eq. [I.] that repeated indices are to be summed over 1, 2, and 3. For the most general coordinate system, the matrix manipulations required to solve Eq. [I] are fairly complex. However, a simplification can be achieved by specifying a coordinate system based on the dislocation in question. The pertinent system involves cartesian coordinates xi, x;, xi, with attendant unit vectors e'i; ei being parallel to the Burgers vector b, and e'3; being parallel to the climb directions (b x ?), i.e., normal to the slip plane. In this coordnate system b = (bi, 0, 0) and ? = (?'1,?'2, 0). Hence, Eq. [I] yields for the climb force per unit length F3 = ?'2s'11b'1—?'1s'12b'1 [2] The first term on the right side of Eq. [2] is the climb force on the edge component of the dislocation, while the second is the (cross-slip) force normal to the glide plane acting on the screw component. Thus, the problem of determining the climb force reduces to that of resolving the stress components —il and s'12 in the xi coordinate system. Several authors7- l0 have shown, for the glide case, that such stress resolution can be performed most readily by graphical methods. Therefore, we present a similar method for the climb case. Of the suggested methods, we follow that of Hartley and Hirth,9 which can be performed directly with a standard stereographic projection for the crystal system in question. We treat the case of simple tension, or compression, this being the most common creep test in which climb is important. A set of cartesian coordinates (x1, x2, x3) is fixed with x3 parallel to the specimen axis. This set is connected to the x'i set by the factors defined in Table I. A perspective view of the coordinate systems relating them to the glide ellipse, is given in Fig. 1, while a stereographic projection of the coordinate systems is presented in Fig. 2. The transformation matrix connecting the xi and xi coordinates is x'i =aijXj [3] where an a11 a13 aij = a21 a22 a23 a31 a23 [4] s The corresponding transformation for the stresses is i [5] The only nonzero component of okl for the simple tension case is s33,. Hence, using Eqs. [4] and 151, we find that Eq. [2] becomes F3 = b'1s33 sin2 cos ?[cos ? sin a - sin ? cos a] [6] = m1b'1s033 sin a- rn2b's33 cos a where sin a = 5: and cos a = ?i, with a the angle measured from b to ?, i.e., a polar angle in the x1, x2 plane. The same equation with a sign change applies for compression. A graphical plot of the "Schmid" factor m1, relating to the climb force on the edge component, is presented in the stereographic projection of Fig. 3. In this projection the glide plane normal points toward the top of the page and the glide direction points normal to the plane of the page. To use the graph, one

Jan 1, 1970

By G. R. St. Pierre, S. W. Gilby

The equilibrium vapors established over Fe-Cr-Ni alloys containing up to 50 pct Cr at 1600°C were determined by collecting the effusate from thoria Knud-sen cells. From these data, the activities of iron and nickel in the ternary alloys have been calculated. In a separate set of measurements, a mass spectrometer was used to determine the activities of the three components in a series of Fe-Cr-Ni alloys. In binary Ni-Cr melts, the activity coefficient of chromium referred to pure solid chromium as standard state is less than one at low chromium contents but is greater than one for melts containing more than 30 at. pct Cr. For Ni-Cr melts containing less than 10 at. pct Cr, In ?Cr = 0.15-0.92 N2. Ternary activity diagrams for Fe-Cr-Ni alloys are presented. In iron-rich alloys, the interactions between the three components are negligible; however, in nickel-rich alloys, the three principal interactions have significant positive values. THE activities of components in binary Fe-Ni alloys have been measured both in the solid solution1-3 and liquid solution4-6 temperature ranges. The determination for Fe-Ni melts have been made by three different methods: transpiration,4 Knudsen-cell effusion with analysis of an accumulated condensate,' and Knudsen-cell effusion with direct measurements of relative component intensities by mass spectrometric methods.6 The results obtained by the three different methods in three different laboratories are in excellent agreement and considerable confidence is placed in the activity coefficients for Fe-Ni melts. At 1600°C, the activity coefficients in dilute solution are ?°Ni = 0.66 and ?oFe = 0.40 for the pure liquid components as standard states. Measurements made on solid solution Fe-Cr alloys7-10 indicate that the component activities are slightly greater than ideal. The measured activities in iron-rich Fe-Cr liquid alloy11-16 indicate almost ideal behavior for this system. The recommended activity coefficient of chromium in dilute solution in liquid iron at 1600o C, ?oCr, is 1.18 when referred to pure solid chromium as standard state and 1.04 when referred to pure undercooled liquid chromium as standard state." The conversion factor of 1.14 is based on a melting point of 1857°C and an enthalpy of melting of 4047 cal per g at. for chromium.21,37 The activity coefficient of iron in dilute solution in liquid chromium has not been measured. Activity measurements on solid Ni-Cr alloys have been made by several investigators.17-20 During the review of this manuscript, Fruehan12 published the results of a study on liquid Ni-Cr alloys at 1600°C. He used galvanic cells22 with ZrO2(CaO) as the electrolyte and Cr/Cr2O3 for the reference electrode. His results indicate that ?oCr in nickel at 1600°C is about 1.01 when referred to pure solid chromium and about 0.89 when referred to pure undercooled liquid chromium as standard state. The activity coefficient of nickel in liquid chromium is unknown. The compositions of vapors freely evaporating from some molten Fe-Cr-Ni alloys have been measured by Lyubimov, Granovskaya, and Berenstein.23 Their results, combined with known vapor pressures, indicate strong positive deviation for chromium in iron-rich alloys and strong negative deviation for chromium in nickel-rich alloys. EXPERIMENTAL PROCEDURES This research was undertaken to determine the activities of components in ternary Fe-Cr-Ni melts containing up to 50 at. pct Cr. At the initiation of the research, we decided to use a Knudsen-cell method wherein a portion of the effusing vapor is condensed on a cold plate.' The composition of the equilibrium vapor for each alloy composition is determined in this method by holding the system under constant conditions for a period of time sufficient to collect enough condensate on the cold plate for chemical analysis. The activities of components in ternary alloys can be calculated from such vapor composition determinations by Eq. [1].24

Jan 1, 1970