Search Documents

By C. van der Poel, R. L. Chuoke P. van Meurs

When an initially planar interface between two im-ttitcihle liquids is displaced at constant rate, U, nor-mat to the front, instability will occur for all rates greater than a critical rate. U, given by provided the Fourier decomposition of the spatial pertmrbcrtiott or deformation of the moving displacetment front contains modes with wavelengths. A, greater than a critical wavelength, given by tit there expressions, p, p and k, with the subscripts I and 2 distinguishing the two 1iquids under consideration, are density, viscosity and effective pemeability coefficients respectively; g is the absolute value of the acceleration due to gravity; and cos (zz') is the direction cosit7e between the vertical carlesian coordinate z' (positive upward) and the z coordinate norma1 to the initially plane macroscopic interface taken positive in the direction from Liquid I to Liquid 2. U is the average volumetric velocity (injected volume of liquid pet. unit time per unit of area normal to Z) and is positive for Liquid I displacing Liquid 2, negative for the reversc displacement and r2 is an effective inter-fucial tension (for displacements between parrallel plates, t0= s the ordinary bulk interfacial tension). further, there is a wavelength of maximum instability given by For natural perturbations, this wavelength will preclotninate and characterize the form of the instability as a quasi-sinusoidal deformation, i.e., viscous fitzgers of peak-to-peak separatiotr, A,,. Comparison of experitient with theory rei'eals tltat these considerations are recsonably valid for displacements in hoth parallel plate chantzels and nnconsoli-clutt'd glass ptnvder packs. These results are. tij particular interest for interpretation of data from laboratory experiments with regard to fielil applica/iotz.s. It is quite possible for A,, although st7zall relative to the large lateral dimension of the reservoir, to rxceed the greatr.st lateral extent of the lahoratory sample or model. If it does, frontal displace- ment occurs in the laboratory experiments, and oil INTRODUCTION Thc puposes of this paper are to present theoretical and experimental evidence for occurrence of macroscopic instabilities in displacement of one viscous fluid by another immiscible with it through a uniform porous medium and to compare available experimental data with some predictions of a theory of instability developed by the first author. The instabilities are referred to as macroscopic in the sense that spatially quasi-sinusoidal, growing fingers of the displacing liquid are formed, the width and peak-to-peak separation (wavelength) of which is large relative to a characteristic length of the particular permeable medium such as grain size. Visual models of two kinds have been used to obtain observations: displacement of oil by water-glycerine solutions through the flow channel formed by closely spaced parallel plates and displacement of oil by water with and without initial interstitial water through unconsolidated glass powder packs, employing the technique of matching indices of refraction. In all cases we have observed macroscopic instabilities or fingers under conditions predicted by the theory to be favorable for their occurrence. The phenomenon discussed here is not the production of streamers due to gross inhomogeneities such as permeability stratification of the porous medium. It is our object to show, on the contrary, that a quantitative prediction of finger spacing is possible in a porous medium known to be macroscopically homogeneous and isotropic throughout. The importance of the phenomenon in its influence on the configuration of oil and water with respect to oil production behavior was noted earlier by Engel-berts and Klinkenberg' who coined the term "viscous fingering". THEORY Necessary and Sufficient Conditions ior Instability anc Initial Kinematics There are several levels of increasing complexity in the theoretical description of instabiIity of fluid displacements in permeable media. Of these, the simplest description, ;tdapted to low permeability systems, is selected for presentation.' More inclusivc tlescriptions are reserved for separate publication.

By V. Kondic, W. M. Rumball

HUNT and Jackson' have recently suggested that binary eutectics might be classified according to the entropy of fusion of the component phases. Such a classification was in fact proposed2 some time ago by the authors and the purpose of this communication is to present some supporting evidence. As pointed out by Hunt and Jackson' typical metals of low entropy of fusion, ?S, tend to form normal eutectics (using Scheil's terminology3). Where one phase is metallic (low ?S) and the other nonmetallic (high AS) then anomalous structures are formed. These anomalous structures may either be regular (e.g., Bi-Sn)or irregular (e.g., A1-Si). Finally Hunt and Jackson suggest that where both phases are of high ?S regular faceted structures are formed. In the proposed classification2 the ratio of the entropies of fusion is taken as the quantitative parameter, i.e., ?Sa/?SB = E, where e > 1.0. An extensive literature survey has been carried out2,4 for reliable photomicrographs of eutectic structures. The structures, numbering 151, have been classified into two classes, viz., normal (i.e., lamellar and rodlike) and anomalous. The latter group has been arbitrarily subdivided into regular and irregular anomalous eutectics. Details and sources of the structures are given elsewhere.2,4 In the absence of better thermodynamic data the value of E has been determined for each of the systems making the following assumptions and approximations: l) The entropy of fusion of solid solutions may be calculated additively;5 ii) Inter metallic compounds are treated similarly but with the addition of the following term for fully ordered structures.5 -4.573(Ni lOg10 Nl + N2 log10 N2) where Nl and N2 are the atomic fractions. It might be argued that the entropies of fusion are meaningless as the eutectic phases form at some temperature (T,°K) below their true melting point (Ta or Tb). As AS, = Lo/To if La the latent heat of fusion of a, is assumed independent of temperature, the actual entropy change AS& is given by La/Te. The ratio of entropies of fusion then becomes Fig. 1 shows E' plotted against cumulative totals for normal and anomalous systems from which it is seen that normal eutectics predominate at low values of E'; e.g., some 90 pct of normal systems lie below E' = 1.5 with 65 pct of anomalous systems above this value. Fig. 1 also shows that the regular anomalous structures occur mostly at low values of c' possibly suggesting growth more akin to normal than anomalous growth. The choice of c as a classification parameter was based on nucleation concepts, it being observed that metals widely separated in Sundquist and Mondolfo's8 nucleation series were anomalous and those close together normal. The basis of these classification views is really outside the intended scope of this communication but it is hoped to explain them fully in the near future.7 Having arrived at this entropy of fusion classification by nucleation considerations it is interesting to note that Hunt and Jackson's ideas are based on growth considerations. Elsewhere one of the authors4 has shown that an even better classification of the metallographic structures reviewed is obtained from purely conventional heterogeneous nucleation concepts. It has been found that systems having a value of the parameter +b less than 0.28 are almost certainly normal and above this anomalous: V, is the gram atomic volume. This parameter is nevertheless closely related to c and c', these being an approximation of p. The parameter is also related to the other classification3,8,9 which have met with any success in the past. All these classifications are to be reviewed shortly by the authorsU7 A significant thing about the classifications based on

Jan 1, 1968

By E. Miller, N. Chou, K. Komarek

The solid solubilities of the compounds PbS, PbSe, and PbTe have been determined by equilibrating single crystals with two phase alloys. The solubilities were determined by measuring the Hall coefficient of the equilibrated samples. The observed maximum deviations from stoichiometry on the lead rich and the chalcogen-rich side of the phase fields were: at. pct excess Pb at pct excess S at 850°C; PbSe--4 x 10'" at. pct excess Pb at at. pct excess at 800°C; and PbTe-2.8 x 10 4 at. pct excess Pb at 780°C, 10"1 at. pct excess Te at 720°C. The previously developed theory of retrograde solubility has been expanded to include intrinsic disorder in the compounds, and the experimental results are in agreement with the theory. THE solid solubility limits of intermetallic semiconducting compounds are of considerable interest, since many of these compounds have potential use as thermoelements in thermoelectric devices, as infrared detectors, or in semiconductor devices. The solidus curves of a significant number of these compounds have been reported to be retrograde with the maximum solid solubility occurring between the eu-tectic and the maximum melting temperatures. In a previous paper,' equations were derived which permit calculation of the maximum solid solubility of an AB compound exhibiting retrograde solubility. Derivation of these equations was based on the assumption that the compound is a nondegenerate semiconductor at the temperature under consideration, that the defects were unassociated, and that the vibrational contribution to the entropy of the compound could be evaluated by the Einstein approximation. It was shown that at the temperature of the maximum solid solubility The entropy term ASS is composed of the entropy of formation of the solid stoichiometric compound (AS$)*, and the entropy due to deviations from stoichiometric composition which is the sum of a con-figurational term (aS*), a vibrational term (AS,), and an electronic term (AS,). In the previous paper the positional entropy was evaluated considering the con- N. CHOU, Member AIME, formerly Graduate Student, New York University, N. Y., is with IBM, Yorktown Heights, N. Y. K. L. KOMAREK, Member AIME, formerly with New York University, is Professor, Department of Inorganic Chemistry, University of Vienna, Vienna, Austria. E. MI LLER, Member AIME, formerly with New York University, is Associate Professor of Mechanical Engineering, California State College at Long Beach, Calif. The paper is based on a thesis submitted by N. CHOU to the Graduate Division, School of Engineering and Science, New York University, in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Materials Science. tributions from defects arising from deviations from stoichiometric composition. The derivation is generalized in the appendix to include intrinsic disorder. The electronic entropy term is also recalculated in the appendix. Of the various semiconducting compounds, the solidus curves of Pbs,' and ~b~e,~ have been most extensively studied. The solidus of PbSe has also been determined, but the temperature of maximum solubility has not been accurately established.5129 The present investigation was undertaken to compare the experimental solid solubility data for these systems with the values predicted theoretically. Since the temperature and composition of the maximum solubility of PbSe has not been accurately determined, experimental emphasis was placed on determining the solidus curves for PbSe in the region of maximum solubility, although experiments were also performed on PbS and PbTe at several temperatures. X-ray diffraction studies were also made on carefully prepared powder samples of PbSe to determine the variation of lattice parameter with composition. EXPERIMENTAL PROCEDURE The procedure of Brebrick and Allgaier was employed for determining the solidus data. Monocrystal-

Jan 1, 1970

By F. R. Scauzillo

Equilibrium data which should be useful in the design and/or evaluation of glycol dehydration units were prepared from an analysis of various published data and the correlation of these data by the use of the thermo-dynamic equilibrium ratio. The equilibrium ratios of water are used to solve the glycol absorber problem; such solutions are necessary to define the number of trays and the glycol circulation rate needed to meet drying requirements. Activity coefficients were obtained which relate directly to the equilibrium ratios of water in the water-TEG-natural gas system. These activity coefficients have been used to calculate the equilibrium dew points for aqueous TEG concentrations of 60 to 99.9 weight per cent for the temperature range of 40° to 120° F. They also provide a means of calculating equilibrium ratios for water in the water-TEG-natural gas system; this applies to any desired TEG concentration and to the temperature range from 40° to 120°F. INTRODUCTION During the last several years, the drying of natural gas with aqueous triethylene-glycol (TEG) solutions has become very prominent. Most users of TEG as a drying agent have been satisfied with the performance of TEG solutions at the conditions used; however, there always has been some discussion of the drying ability of TEG solutions at conditions not commonly encountered such as temperatures below 50° or 60°F, but more particularly temperatures above 100°F, and pressures above 1,000 psia. Today, when higher wellhead temperatures such as 120°F are more commonly encountered many are skeptical of TEG's ability to dry sufficiently well to provide dew points around 32°F, which is generally the maximum tolerable when attempting to dry a gas to contract specifications of 7 Ib/MMscf. These views probably have evolved to some extent from the days when suppliers would not guarantee dew-point depressions in excess of 65° to 75 °OF. Also, the feelings about TEG drying may have arisen from the lack of information about the equilibrium relation of water in the drying operations. Porter and Reid' have reported equilibrium data for a 95 per cent TEG solution, while Townsend2 reported equilibrium data for 95, 98 and 100 per cent TEG solutions. Townsend also presented a calculational method which obtains activity coefficients in the liquid phase and, subsequently, the equilibrium water content of a saturated gas over glycol from published3 atmospheric dew points. Wise, Puck and Failey' presented activity data for the water-TEG system at atmospheric pressure. In the past, the equilibrium ratios published for water in a 95 per cent by weight aqueous TEG solution have been used indiscriminately by many for all concentrations of TEG encountered in gas drying operations. Since essentially all such gas operations require the use of more concentrated TEG solutions, a study was undertaken to correlate the existing equilibrium data for aqueous TEG and gas systems and to provide some means of calculating the equilibrium ratio of water in the natural gas-water-TEG system where TEG concentrations were other than 95 per cent. The data presented herein consider only the equilibrium drying ability of the TEG solutions and do not consider the effect of temperature and pressure on the tray eficiency of contactors. In other words, this paper is concerned primarily with the development of the equilibrium relationship between water in the dried natural gas and the water in the lean TEG entering the top tray of the absorber. BASIC EQUILIBRIUM RELATIONS A relationship which relates the K value of water in the natural gas-water-TEG system to the vapor pressure of water at the system temperature, the total pressure of the system and the activities of the liquid and gas phases has been evaluated. It is well known that, for any component of a mixture at equilibrium between a gas phase and a liquid phase,

By P. R. Sahm, T. V. Pruss, D. A. Gandolfo

Cb 3 Sn powders were hot-pressed to estahlish densi-fication conditions and assess the possible use of such compacts in multiphase structures with ceramic or magnetic inclusions. Superconducting transition temperatures, Tc, were measured as a function of powder particle size, pressing temperature, and second-phase admixtures. The highest transition temperatures were found in coarse powder (5 to 500 u) compacts and were only weakly depended on pressing temperatures up to 1200°C. Fine powders (below 5 p), however, both single-phase CbsSn and Cb3 Sn-composites, showed a stronger dependence on hot-pressing temperatures. While Al2 O3 phase admixtures did not affect T, much more, a Pronounced T, degradation was observed upon the addition of ferrimagnetic Fe2 Mno. 5 Zno.5 O4 powders. Likewise, critical current densities, Jc, at 4.2° K were lowered appreciably in the ferrite doped samples, whereas the neutral A1203 inclusiow gave slightly improved Jc values. MULTIPHASE structures in electronically active materials have recently gained in interest as pointed out by Galasso in a review article.' In superconductors the most widely reported group are eutectic alloys which have been mostly investigated with respect to the proximity effect.2"4 In addition, "artificial eu-tectics", for example in the form of mercury-vycor glass sponge5 or fine niobium wires imbedded in a copper matrix,= have been prepared to attain higher critical fields and critical current densities. The idea was to enhance flux pinning phenomena by second-phase inclusions or ho1es. 7 Flux pinning should also be possible by the introduction of insulator inclusions into a superconducting matrix. Ceramic-metal combinations are most easily produced by hot pressing (= pressure sintering), as it permits an almost unlimited selection of material combinations. Cb 3 Sn has been prepared by vapor phase growths or in the form of thin films. Large compacts must be sintered,1° however, and normally present difficulties. At temperatures below 1350°C insufficient densities are obtained and at higher temperatures the loss of tin due to dissociation upsets the stoichiometry. Hot pressing is a natural approach to circumvent the problems, since one can obtain satisfactory densities despite low sintering temperatures. This work was mostly concerned with establishing conditions to hot press Cb3 Sn and Cb3 Sn multiphase structures (= composites), because no data are presently available, and to evaluate these with respect to their transition temperatures. However, a few measurements on critical current densities were also carried out, in particular to assess the value of ferrimagnetic inclusions. I) EXPERIMENTAL PROCEDURE The experimental procedure involved the preparation of feed material and powders, hot pressing, and evaluation of the resulting ingots. As no attempts were made to synthesize Cb3Sn from the elements during hot pressing, feed material had to be produced. For this purpose, a stoichiometric Cb-Sn powder mixture was compacted into a 3-cm-diam ingot (80,000 psi) and then inserted into a quartz tube and evacuated. Rf heating was employed. Above the tin melting point (230°C) liquid tin was forced out of the ingot, but was pulled back by capillary action after approximately 0.5 hr of continuous pumping at 700°C. Thus, it appears that evacuation at this temperature caused desorption of impurity gas creating wetting surfaces which, later on, favored alloying at relatively low temperatures. After a presintering period of 17 hr at 800°C, the quartz tube was tipped off and sealed into another such larger-diameter tube which, in turn, was connected to a pump system. This configuration permitted a temperature treatment at 1400°C, carried out for 65 hr, without collapsing of the glass walls or unfavorable chemical interaction of Cb3Sn with the quartz. The resulting material was

Jan 1, 1969

By John P. Hager, Igor A. Wilkomirsky

The thermodynamic properties of the Pb-Ag system have been determined between 775° and 950°C by the cell: Electrotransport measurements on silica-saturated PbO-SiO2 melts established that the conduction was entirely ionic and that lead is present only in the divalent state. The activities of lead and silver exhibit Positive deviations from ideal behavior. In dilute solutions the activities of lead and silver follow Henry's law. The values obtained for yhb and yig at 1200°K are 1.49 and 1.69, respectively. Determination of a portion of the liquidus curve agrees well with the established phase diagram. provides an accurate means for the thermodynamic study of liquid alloy systems. However, Wagner and wernerl have shown that cells of type I are limited to alloy systems for which the reaction at the alloy-electrolyte interface will be extremely small. The traditional electrolyte for cells of type I has been a low melting point mixture of group IA chlorides, to which a small amount of the compound ACIZa has been added. However, the choice of a chloride electrolyte may not always be practical, in that the Gibbs free energies of formation of the compounds ACIZ, and BCIZb may not be sufficiently different to minimize Reaction [I]. On the other hand, the Gibbs free energies of formation of the oxides of metals A and B may be sufficiently different to minimize Reaction [I]. In this case, the use of molten oxide electrolyte would be clearly advantageous. In the present study, the thermodynamic properties of the Pb-Ag system have been investigated using the cell: Previous galvanic cell studies on the Pb-Ag system2'= are not reliable due to the displacement reaction between the alloys and the chloride electrolyte. Wagner and ~erner,' using Laitinen and Liu's' value at 450°C for the standard electrode potential of Reaction 111, have estimated the percent error in the activity of lead bPb) when using a chloride electrolyte. At 1000°C the error for xpb = 0.10 is 12 pct. Taking the Gibbs free energies of formation for PbClz and Ag cI,' and the equations of Wagner and Werner,' the error in apb at 1000°C and Xpb =0.10 was calculated to be 140 pct. However, using the Gibbs free energies of formation for Pb O' and an estimated value for Ag 20, the error in apb when using a molten oxide electrolyte at 1000°C was estimated to be less than 1 pct for The virtual reaction of cell I1 for the passage of n Faradays of positive electricity from left to right in the cell is: where 5 is the Faraday equivalent (23,063 cal per v equivalent), E is the electromotive force of the cell in volts, and n is the number of Faradays of electricity required for the transfer of 1 g-atom of lead through the cell. The value of n should be equal to 2 since lead is generally divalent in oxide melts. However, the presence of either tetravalent lead ions or electronic conductivity through the molten oxide electrolyte will result in a value of n greater than 2. Bockris, Kitchener, and Davies' have determined the amount of ionic conductance in a liquid PbO-SiO2 mixture (xPbO= 0.50) at 920°C by measurement of the volumes of CO and C& evolved at a carbon anode during electrolysis. From their data a value of 2.13 * 0.16 was calculated for n. Although their measurements indicate the possible presence of multivalent lead ions and/or electronic conductivity, their measurements were not at silica saturation and no definite conclusions can be drawn. However, as a result of the electrotransport measurements described in the next section the ;due of n can be taken as equal to 2.

Jan 1, 1969

By C. S. Smith, Chih-Chung Wang

TURNBULL and his collaborators1,2 have developed the theory of homogeneous nucleation as applied, inter alia, to solidification of liquid metals. Vonnegut³ and Turnbull4 have shown that if a liquid metal is subdivided into small droplets a vast majority of them will undercool very considerably before solidification, generally to as low as about 0.8 of the freezing temperature on the absolute scale. The nuclei effective at small degrees of supercooling in bulk metal seem to be internal or surface heterogeneities, relatively small in number. If the metal is subdivided, those droplets that happen to contain such nuclei will solidify at a temperature not greatly below the true freezing point, but only a small part of the whole volume will be affected and the majority of the drops will under-cool to the much lower temperature at which homo- geneous nucleation occurs as a result of fluctuations. It occurred to one of the authors that an appropriate subdivision to give effective freedom from random nuclei is produced during the solidification of many alloys that contain a minor amount of a phase of low melting point, and that one might then expect marked undercooling of the distributed phase. Experimental: The alloys selected for initial study were copper with minor amounts of lead and bismuth, and aluminum with tin. In all these alloys, the major component freezes at a temperature not much below that of the pure metal, and there is little further change in constitution on cooling until the lower melting point constituent freezes in an almost pure state. Cooling curves were taken using the controlled heat flow method permitting approximate specific heats to be obtained. Chromel-alumel thermocouples were used, with the standard emf tables, since extreme precision was not needed. The crucible (3/4 in. id, 5/16 in. wall) was made of B & W K-20 insulating brick. It held about 10 cc of the alloy being investigated. The cooling rate was 2.5" to 2.9°C per min under a controlled temperature difference of 20°. Approximate specific heats were computed from the inverse rate curves together with data from a blank run and from a standard run with a copper cylinder of known heat capacity. The alloys for investigation were made from high-purity metals (99.99+pct) and cast into graphite molds. The castings were machined to fit the crucible and to provide a hole for the inner thermocouple. Cooling curves were taken after heating to a temperature about 50" above the melting point of the minor, lower melting-point, constituent. Results The lead phase in an alloy of copper with 5 pct lead did not undercool more than 3" below the melting point of lead (327°C) either as cast or after annealing to produce a new dispersion of the liquid phase. A copper-zinc-lead alloy with 23.75 pct zinc and 5 pct lead undercooled more but showed no thermal effect below 319°C. A cast alloy of copper with 5 pct bismuth undercooled to 249°C (M.P. bismuth 27l °C), but once solidification started it was completed at the same temperature. This was anticipated, since the bismuth forms a nearly continuous phase between the grains of copper, and a solid crystal nucleated anywhere would rapidly* con- sume the entire network of liquid, unless the physical continuity of the liquid were broken through volume changes, inadequate fluidity, or gas evolution. Similar arguments apply in the case of copper-lead alloys, where the lead-rich liquid forms a network along grain edges, though not grain faces. In both cases there would be a few isolated particles

Jan 1, 1951

By N. J. Grant, B. C. Giessen, U. Wolff

With the aid of the splat-cooling technique of rapid quenching, metastable alloy phases based on antimony ad bismuth have been prepared. At room temperature, simple cubic phases were found in the Sb + Au, 56 + Yd, and Sb + Ni systems. Another phase was retained in the Au-Bi system at -190°C; in this systent, a microcrystalline phase and a rhombohedrally distorted phase were also observed. The crystal chemistry of these phases is discussed in terms of their valence electron concentration range. A remarkable five-step Phase sequence was found upon heating and equilibration of certain metastable Au-Bi alloys. USING the splat- cooling technique of rapid quenching from the liquid,' a large number of metastable intermediate alloy phases have been prepared in recent years. Thus, Hume-Rothery phases of the hexagonal type based on the valence electron concentration have been produced in systems where they are missing in equilibrium, e.g., Ag-Ge 1 or AU-Sb. 3 In a comprehensive study of metastable alloys based on tin, simple hexagonal metastable y phases were found, raising the total of stable and metastable y phases from 3 to 13. 4 In the In-Bi system, five new metastable intermediate phases were retained.' In most cases, the new metastable phases have some crystal chemical characteristics in common: a) They have rather simple, elementlike structures with few atoms per unit cell, e.g., y phase,4'5 or the metastable primitive cubic phases in the Ag-Te and and Au-Te systems.= b) Their structures are often related to those of the terminal phases, or to elements that are straddled in their position in the periodic table by the constituent elements; this is the case for metastable ß (In-Bi) with the structure of white ß Sn,' or metastable a (Cd-sn). 4 c) They seem to represent the crystal structures appropriate to certain average group number (AGN) ranges in which no or few single-phase structures exist in equilibrium. These average group number ranges can probably be equated in the case of many elements of the Ba(Zn)-Bs(As) groups with the averaged valence electron concentration. In the present work, the splat-cooling technique has been applied to the alloys of antimony and bismuth. In equilibrium, neither element forms intermediate alloy phases with other B elements from groups Bl(Cu) to B4(Si) that contain more than 66.7 at. pct of Sb or Bi, respectively, e.g., AUSb 2. 7 Thus, antimony- or bismuth-rich phases, which might be closely related to antimony or bismuth, and which would reflect struc- tural changes occurring on alloying with small amounts of addition elements, do not exist. Also, the terminal solubility of antimony and bismuth for all B metals is very small. It was therefore of interest to investigate the influence of enforced, nonequilibrium solid solution of addition elements in antimony and bismuth on their crystal structures. It was found that additions of group B 2-B4 metals act distinctly differently than additions of group T 10 and B1 metals. The latter set of additives will be treated here. EXPERIMENTAL TECHNIQUES Alloy preparation and splat-quenching followed closely the procedures described in Refs. 1, 4, and 5; they consisted of the preparation of master alloys from high-purity starting materials (99.99+ pct) and blast-atomizing of 20-mg quantities onto copper sheets. The resulting approximately 1 to 5 u thick foils were investigated by X-ray diffraction. The cooling rates obtained are of the order of 107 to 10 8 C per set,'" causing solidification to proceed at a rat; of approximately 100 cm per sec.' Powder patterns of the resulting fine-crystalline materials were taken with filtered CuK, radiation. For antimony alloys the substrates were held at room temperature and for bismuth alloys at -190°C both during splat foil deposition and X-ray investigation. Annealing treatments at several

Jan 1, 1969

By W. G. Pfann

Under certain conditions, a molten zone can be made to move through a solid by impressing a stationary temperature gradient across the solid. This phenomenon can be utilized in fabricating semiconductive devices, growing single crystals, joining, boring fine holes in solids, measuring diffusivities in liquids, small scale alloying, and purification. Fundamentals and exemplary applications are outlined. ANEW aspect of the travel of a molten zone through a solid is considered in this paper. Whereas, in previously described zone melting techniques,'-' zones were usually caused to move by changing the position or temperature of a heat source, in the present technique a zone is made to move by impressing a temperature gradient across it. For example, a thin layer of molten A1-Si alloy, sandwiched between two silicon slabs having a temperature gradient normal to the layer, will travel through the hotter slab. One feature of temperature gradient zone melting, as this technique will be called here, is that zones of unusually small dimensions can be maintained. This leads to such diverse uses as have been listed in the abstract. In this paper fundamentals of the movement of a molten zone in a temperature gradient, and some practical applications thereof, are outlined. Assumptions The movement of a molten zone through a solid body of solvent substance in which a temperature gradient exists will be discussed. The length of the zone will be its dimension in the direction of motion. The direction of the temperature gradient will be toward the region of higher temperature. To be considered is a binary or higher order solute-solvent system in which the concentration of solute in the molten zone is sufficient to produce a lowering of the freezing temperature of the solvent which is at least of the order of the temperature range impressed on the charge. While we discuss primarily molten zones in a solid matrix, the principles are applicable to solid or vaporous zones in a solid matrix, a requirement being that the diffusivity in the zone be substantially greater than in the matrix. Diffusion of solute into the surrounding solid will be assumed negligible. In general, its presence will not basically alter the conclusions to be drawn. For illustration, a system represented by the partial constitutional diagram of Fig. la will be discussed, in which k, the distribution coefficient defined in the figure, is constant and less than unity. However, the technique is applicable to any solute- solvent system in which one component lowers the melting point of another component. Movement of a Molten Zone in a Temperature Gradient Consider, in the system AB of Fig. 1, where B is designated the solute and A the solvent, the effect of sandwiching a thin layer of solid B between blocks of solid A and placing the whole in a temperature gradient such that the temperature of the layer is above the lowest melting temperature of the system. Surrounded by an excess of A, the layer will dissolve A, becoming molten, and expand in length." As so- lution of A continues, at both interfaces, the mean solute concentration in the zone will move to the right in Fig. la, until at temperature T, the liquid at the cooler interface reaches the liquidus concentration C,. Solution of A thereupon ceases because the liquid T1 is saturated with A. The liquid at the hotter end of the zone, being at temperature T2, is not saturated with A at concentration C1. Hence solution of A continues there and concentration C2 is approached. A concentration gradient therefore is set up in the zone, causing A to diffuse toward the cool end and B toward the hot end. As a result, the liquid at the cool interface becomes supersaturated and a layer of crystalline A containing concentration kc, of B in solid solution freezes. Since a source of A

Jan 1, 1956

By P. E. Arnold, G. S. Ansell

RELAXATION in metals has been studied in detail by many workers in recent years.1-5 These studies have shown that there is an energy-loss peak observed in a metal placed in mechanical resonance at low frequencies which may be correlated with various metallurgical processes. These experiments have been largely used to study single-phase materials. In order to study the effect of a finely dispersed second phase in a metal matrix upon this phenomenon, the relaxation behavior of an aluminum-aluminum oxide SAP-Type alloy in both the fine-grained as-extruded and coarse-grained recrystallized conditions was investigated by the use of internal-friction measurements. The alloys studied, designated MD 2100, consist of a matrix of commercial-purity aluminum containing a finely dispersed second phase of aluminum oxide. The aluminum oxide is present in this alloy in the form of fine platelets approximately 130A units thick and several microns on edge. The mean free path betwen dispersed platelets is approximately 0.5 p. The grain size of this alloy is extremely stable at temperatures as high as the melting temperature of the aluminum matrix. In the as-extruded condition the grain diameter is several microns, while in the recrystallized condition it is several millimeters in diameter. The damping characteristics of the alloy samples were determined using a torsional-pendulum device. The internal-friction constant was determined for this alloy in both the fine-grained as-extruded and coarse-grained recrystallized conditions over the temperature range from 343" to 923°K at frequencies of 0.9, 1.3, and 1.8 cycles per sec. An internal-friction peak was observed in testing the fine-grained alloys which was not observed for the coarse-grained alloys. This peak, which might be associated with a type of grain-boundary relaxation, was observed at each of the three different test frequencies for the fine-grained sample. The temperature at which the relaxation peak occurred for each of the three different frequencies is shown in Table I. From these data an activation energy of relaxation for each pair of frequencies was determined by comparing the temperature shift of the relaxation peak with frequency between each of the test frequencies, assuming that an Arrhenius-type rate equation is applicable. The activation energy of relaxation determined from these data is also shown in Table I. The mean value of the activation energy was about 6 kcal per mole. The activation energy for the relaxation process determined in this investigation indicates that the relaxation associated with the grain boundaries in this aluminum-aluminum oxide SAP-Type alloy is quite a different process than grain boundary relaxation in single-phase aluminum alloys. In these latter alloys, the activation energy for relaxation has been found to be the activation energy for self-diffusion, approximately 36 kcal per mole1,2 Where the activation energy is the same as the activation energy for self diffusion, it has been proposed that the relaxation phenomena is dependent upon vacancy formation and motion. It is apparent therefore, that, in this aluminum-aluminum oxide SAP-Type alloy, the relaxation must be due to another type of atom transport. It has been shown previously in studies of the high-temperature, low-stress, steady-state creep behavior of as-extruded aluminum SAP-Type alloys, that the activation energy for steady-state creep is one that may be characterized by the nucleation of dislocations from the grain boundaries. It is proposed that the relaxation behavior observed here is due to a similar effect. In this case, the internal-friction peak is a resonance behavior due to the time-dependent nucleation of dislocations from grain boundaries in the sample. A process of this type would have an activation energy which is undoubtedly stress-dependent and of the form Q = Q, + (dQ/du)u

Jan 1, 1962

By Leonard R. Weisberg

The general properties of diffusion in GaAs are reviewed. A total of .fourteen atoms have been studied to date, and activation energies for eleven reported are (in ev): Ga (5.6), As (lo), Zn (2.49), Cd (2.43), Sn (2.5). Mn (2.75). S (4.0). Se (4.2). Ag (0.33), Cu (0.52). and Li (1.0). The dijfusion of the first eight atoms, as measured by Goldstein using radiotracer techniques, established the principle of sub lattice diffusion for both host and impurity atoms, since atoms substituting for gallium have lower activation energies than those substituting for arsenic. The diffusion constant of interstitial copper has been determined for the first time by Hall and Racette using heavily doped p-type samples in which interstitial copper predominates. The anomalous behavior of zinc, which has an abrupt diffusion front and a strong dependence on zinc concentration, has recently been quantitatively explained in terms of an interstitial-substi-tutional equilibrium-diffusion mechanism. Lithium and silver diffuse mainly interstitially, while the diffusion of phosphorus is complicated by the continual formation ofGap. Junction-migration measurements indicate diiferent results than radiotracer techniques, and this is one problem that still remains, together with discrepancies in cadmium-diffusion results, and an unexplained dependence of impurity diffusion on arsenic pressure. BESIDES the usual insights into the behavior of atoms in solids, studies of diffusion in semiconductors have additional importance for device applications. In fact, the amount of diffusion studies performed on a semiconductor might serve as a simple barometer of its technological importance. Thus, the information accumulated for GaAs is not yet quite comparable to that for germanium or silicon, but considerable knowledge is available. The diffusion of at least fourteen atoms has been investigated for GaAs, and the diffusion constants have been determined for eleven of these. Interesting results were found for GaAs that elucidated problems not previously solved for germanium or silicon. For example, the interstitial-diffusion constant for the very fast diffusant, copper, was first measured in GaAs. Also, solutions for interstitial-substitu-tional equilibrium diffusion in extrinsic material were provided first for GaAs, although they are ap- plicable to germanium, silicon, and other semiconductors. Studies in GaAs firmly established the concept of sublattice diffusion, which should have application in many different compound systems. The purpose of this paper is to review briefly the general state of knowledge concerning diffusion in GaAs. First some general aspects of diffusion will be covered, and next attention will be given to the results mentioned above. Diffusion in GaAs has been measured both by radiotracer and junction-penetration techniques, and some discussion will be given of differences in the two sets of results. The comprehensive work of oldsteinl-' has served as the primary reference for the diffusion of well-behaved substitutional atoms in GaAs. Since a review has recently been presented of many of these reults, details of the diffusion data will not be repeated. Neither will any discussion be given of the experimental techniques of diffusion measurements, except to note here, in passing, the simplicity and effectiveness of the precision lapping device employed by Goldstein.' I) EXPERIMENTAL RESULTS AND DISCUSSION A) Summary of Experimental Results. Diffusion constants D = Do exp(E/kT) have been determined for eleven atoms in GaAs: a,' AS,' n,'" Cd,'" n,' n,8 s, e,' Ag,O CU," and i." These results are summarized in Table I. In addition, diffusion was investigated for phosphorus,4'12 germanium,I3 and ilicon,' but Do and E were not determined. Radiotracer techniques were used for most of the studies except for lithium, where a combination of chemical and electrical techniques was employed, and germanium and silicon, where junction-migration techniques were applied. The diffusion of zinc, manganese, sulfur, and tin was also measured using junction migration.13'14 Goldstein demon-

Jan 1, 1964

By C. S. Matthews, P. Hazebroek, F. Brons

A method has been developed for calculating the average pressure in a bounded reservoir. The reservoir is first divided into the individual drainage volumes of each well, by using the criterion that at steady state each individual drainage volume is proportional to a well's production rate. The average pressure in each drainage volume is then calculated by a method developed in the report. By volumetrically averaging these individual drainage volume pressures, the average pressure in the entire reservoir is obtained. To calculate the average pressure in each drainage volume, a correction is applied to the ordinary extrapolated pressure, i.e., the pressure obtained by extrapolating to infinite time the linear portion of the graph of closed-in pressure versus log[?t/(t + ?t)], where At is the closed-in time and t the production time. The correction, which is a function of the production time, is presented in graphical form for different shapes of the drainage area (horizontal cross section of the drainage volume). INTRODUCTION It is important to be able to find the volumetric average pressure in a reservoir so that the size of the reservoir may be determined from material balance calculations. It is also desirable to be able to find the approximate distribution of pressure within a reservoir for detection of fluid movement. 'The purpose of this paper is to present a method for calculating both the average reservoir pressure and the approximate distribution of pressure within a bounded reservoir — that is, a reservoir with no water drive. In reservoirs where the pressure builds up rapidly after wells are shut in, the determination of average pressure generally poses little problem, for one often need only average the final buildup pressures. It is when pressure buildup is slow that difficulties arise. For practical and economical reasons, the time allowable for closing in wells is limited. If at the maximum allowable closed-in time the pressure has not reached a con- stant value (and this is more often the case than is generally realized), calculation of average pressure presents difficulties. One method which has been used in the past for obtaining an "average" pressure involves the extrapolation to infinite closed-in time of curves of pressure as a function of closed-in time. Contour maps of these extrapolated pressures are used in arriving at an average pressure. In some cases wells are shut in for arbitrary Lengths of time, and contours of these arbitrary shut-in pressures are used for calculation. All these methods are somewhat arbitrary and suffer from lack of fundamental theoretical basis. For example, no precise meaning can be given either to the extrapolated or to the arbitrary shut-in pressure. Furthermore, there seems to be no reason why either set of pressures should be contourable. It was because of the vagueness of these old methods and their lack of a sound theoretical basis that, the present method for determining average reservoir pressure was developed. SYMBOLS A = drainage area of a well, cm' A, = drainage area of well j, cm2 A , = total drainage area of all wells in a reservoir, cm2 c = coefficient of compressibility of fluid at reservoir temperature, atm-1 f = hydrocarbin-filled porosity, fraction h = net thickness of formation, cm k = permeability of formation, darcy p = pressure at a well, atm p0 = initial reservoir pressure, atm p* = pressure obtained by straight-line extrapolation of linear portion of plot of p versus log [At/(t+ At)] p = volumetric average pressure inside drainage volume of a well q = volume flow rate at a well, cc/sec a: prevailing reservoir conditions q, = total volumetric flow rate from a reservoir, cc/sec at prevailing reservoir conditions rw = radius of well bore, cm T = kt/fµcA t = corrected time of production of a well, seconds, defined by

Jan 1, 1955

By J. N. Ong

The oxidation of tantalum is assumed to occur by four simultaneous first-order chain reactions; solution of oxygen in the metal, nucleation and growth of a suboxide phase at the metal surface and two phase boundary processes which give rise to two different modifications of Ta2O5. A rate equation is then formulated which accounts for all principal details of oxidation behavior, unifies and correlatss the results of the principal investigations, and expresses the rate as a function of pressure, temperature, and time. For purposes of practical application, the expressions: may be used below and above 700T respectively to determine the regression rate of tantalum in cm per hr, where T is to be expressed in "K and P in atmospheres pressure of oxygm. The equations will reproduce within a factor of two all reported experimental rates in the temperature range 475" to 1400°C and pressure range 0.000026 to 40.8 atm 0. 1 HE oxidation kinetics of tantalum have been investigated thoroughly in the temperature range 50" to 1400 and pressure range 1.32 x l0-' to 40.8 atm pressure of 0,and as a result, the principal features of the oxidation of tantalum are known and have been described in considerable detail. The exposition of the characteristics of the behavior of tantalum in oxygen has been largely descriptive, with considerable attention being devoted to elucidating the structure, composition, and morphology of the resulting oxide products. The weight change vs time results have usually been reported graphically. With the exception of the linear rate below 700 C and the process of diffusion of interstitial oxygen into the metal,6'9 no effort has been made to characterize the reaction rates in functional form. It is the purpose of this paper to demonstrate that when the fundamental principles of chemical kinetics are rigorously followed, the results of the principal investigations will be correlated and unified and the oxidation rale may be expressed as an analytic function of pressure, temperature, and time. In addition to accounting for the temperature and pressure dependency of the oxidation rate, the following minor features will be accounted for as a result of the basic analysis: the initial period of nonline-arity in rate at low temperatures, which decreases with increasing temperature;3'5'7 the rate vs temperature cusps obtained at constant pressure"5 the concentration gradient of oxygen interstitially dissolved in the metal.5'7 The analysis will also be consistent with the overall chemistry of reaction and with all visual, microscopic, and X-ray observations. DEVELOPMENT OF RATE EQUATION As a fundamental basis for the treatment of metal-oxygen reaction kinetics, it will be regarded that the rate of reaction is determined by the concentration and not the activity or chemical potential. It is further assumed that all metal-oxygen reactions are first order complex chain reactions" whose rate is dependent on the concentration of an intermediate species which is a complex function of the oxygen concentration, or oxygen pressure, only. The basic rate equation is therefore: is the weight of oxygen consumed by the reaction, A is that portion of the area of the metal over which the reaction is occurring (cm2), t the time (sec), kf the rate constant for forward reaction, C the intermediate species concentration and f '([Oz]) and f (Po,) are complex functions of oxygen concentration and Oxygen pressure respectively. Reverse reactions may usually be neglected in metal-oxygen systems since a large negative free energy generally accompanies these reactions. Eq. [I] takes different forms when the rate of reaction is controlled by diffusion into or through a phase or at a phase boundary. For diffusion into the metal phase, the rate is given by1' Di is the diffusion coefficient of species diffusing into the metal phase (cm2/sec), and Q is the corre-

Jan 1, 1962

By J. C. Fisher

Nucleation theory is applied to martensite nucleation in substitutional iron alloys. Composition fluctuations are neglected, and a steady rate of nucleation is predicted for any composition and temperature. The maximum rate of nucleation (as a function of temperature) is shown to be measurable only for on extremely narrow composition range, being too great or too small outside this range. A number of experimental observations, including completely isothermal transformation in some alloys and composition-dependence of M. temperature in others, are compared with the calculated behavior. IN a previous report,¹ nucleation theory was applied to the formation of isothermal martensite in an Fe-Ni-Mn alloy. The experimental observations of Cech and Hollomon² were accounted for quantitatively. It now is of interest to examine the predictions of nucleation theory with respect to martensite transformation in substitutional iron alloys of other compositions. Since Fe-Ni alloys have received considerable attention, both as to their transformation kinetics and their free energies, they will be treated in detail, although the results apply with equal validity to other substitutional alloys. In substitutional alloys, where composition fluctuations can be neglected in volumes of critical nuclear size, the rate of nucleation of martensite in subcooled austenite is¹ n - Nv exp (—W*/kT) [I] where the free energy of formation of the critical size nucleus is W* = 8192 p (?² s³) [2] In the expressions for n and W*, the symbols have the following meanings: N is the number of atoms per unit volume; v, the atomic vibration frequency; 8, a function of the elastic constants of austenite and the shear angle of martensite; , the austenite/mar-tensite inter facial free energy; and ?fc is the free energy change per unit volume accompanying the transformation. Jones and Pumphrey² have determined the austenite to martensite free energy change as a function of nickel content and temperature for Fe-Ni alloys. Their expression is ?F = 2500 C + 1.25 (1 —C)?FFc cal per mol [3] where C is the atom fraction of nickel, and ?Ffe is the free energy change per unit volume for pure iron as tabulated by Zener4 (or, equivalently, Fisher"). From Eqs. 1 to 3, it is possible to calculate isothermal nucleation rates as a function of temperature and composition for Fe-Ni alloys. The only unknown parameter is (?² s³). The best value of this parameter for Fe-Ni alloys, determined from M. temperatures in a manner to be described shortly, is (?² s³)= 9.92 (10)21 cgs units [4] Taking this value of (?² s³), C-curves for isothermal nucleation were calculated for Fe-Ni alloys of several compositions, and are plotted in Fig. 1. It is evident from Fig. 1 that completely isothermal nucleation of martensite in Fe-Ni alloys can be observed experimentally for only a very narrow range of compositions. Cech and Hollomon find that a nucleation rate of about 10' nuclei per cc per sec is as fast as they can follow dilatometrically, and a rate of about 10-1 nucleus per cc per sec is about the slowest that can be measured conveniently. If the nucleation rate is to lie between the limits 10-1 10 the nickel content of an Fe-Ni alloy must fall in the atomic fraction range 0.299 5 C 5 0.302. Nickel contents lower than 0.299 correspond to nucleation rates too great to follow, and those higher than 0.302 to nucleation rates too slow to measure in a reasonable time.* In view of the narrow corn- position range in which successful isothermal measurements can be made, it is not surprising that only one substitutional alloy has been described' that comes close to being in the proper range.

Jan 1, 1954

By B. H. Bergstrom, J. J. Gilvarry

Previously the authors showed that the Gilvarry equation correctly describes the distribution of fragment size in single fracture, provided the exoclastic particles showing original surface of the specimen are removed from the distribution. In the present work this conclusion has been strengthened in two respects. First, it has been shown that the existence of two peaks in the differential distribution for the endoclastic particles, corresponding to the effect of surface and volume flaws in the interior of the specimen, can be demonstrated from the average of data for a large number of specimens. Second, it has been shown by means of a Coulter counter that data obtained from specimens fractured in gelatin display the behavior at fragment sizes approaching 1 p which is demanded by the limiting form of the Gilvarry equation in this case. In previous work, a rigorous derivation of the proper distribution function for fragment size in single fracture of a brittle solid has been given by Gilvarry,' based on a closely defined physical model and deduced strictly by the laws of probability. Experimental work has been reported by Gilvarry and Berg-str~m,~ confirming the theory in its broad outlines. Subsequently, Gilvarry and Bergstrom extended the theoretical discussion to include comminution and presented experimental data bearing on this point. They have described further experimental results for the case of single fracture recently. The experimental evidence regarding single fracture was rather convincing. However, it was based on only five specimens for which fragment sizes were determined by sieve analyses and three for which measurements in the finer sizes were made on a Coulter counter. The purpose of the present paper is to extend the experimental results as obtained by means of both sieve analyses and the Coulter technique. In the former case, sufficient specimens are included so that the average data should vield ac- curately the detailed behavior of the distribution function for the coarser fragment sizes. In the latter case, measurements for the finer sizes are reported which correspond to single fracture more closely than prior results and fill a lacuna left in the experimental verification. In addition, an extension of the theory given recently bGilvarry for the case of fracture of a two-dimensional solid will be outlined, and experimental results obtained by him will be discussed.6 THREE-DIMENSIONAL CASE Gilvarry' has shown that the Schuhmann size modulus is a measure, in a sense specified later, of the number of activated edge flaws in a specimen undergoing single fracture. Single fracture is defined as that resulting from a single instantaneous application of energy, such as the blow of a hammer whose motion is halted when fracture begins. Plural fracture consists of a short sequence of single fractures, as in a jaw or gyratory crusher; multiple fracture obtains when many applications of energy are made, as in a ball mill. The Griffith flaws associated with fracture of a specimen are flaws occurring within the volume of the body, on the new surfaces created within its interior and on the new edges (intersections of two surfaces) within the body. Fracture proceeds by activation of flaws in the volume of the specimen, in the fracture surfaces throughout its interior, and in the edges produced by intersection of fracture surfaces. If all the flaws and all groups of flaws are distributed at random, independently of other flaws or groups thereof, it has been shown by Gilvarry that the distribution function for fragment size has the form of a Poisson distribution. The fraction y, by weight or volume, passing a sieve of mesh size x is given by where k is an average spacing of activated edge flaws, j2 is an average amount of surface containing one activated surface flaw and i3 is an average volume corresponding to an activated volume flaw. The averages in question are referred to the mean dimension x of a fragment. On the basis of the theoretical

Jan 1, 1962

By P. G. Carpenter, R. G. Nisle, F. W. Bubb

It is the purpose of this paper to describe an electrical computer which has been constructed to solve the equations for vapor-liquid equilibrium in multi-component systems. The instrument consists of seven component-computing units each with proper indicating means and power supplies. Each unit is a resistance network with a voltage matching servomechanism, and each provides an output voltage proportional to the mol-fractions for vapor and liquid phases. These voltages are summed and matched with a reference voltage to provide the solutions. Any reasonable number of such units may be put together to make a computer. The theory and operation of the computer is discussed. A number of applications and examples of computer results are given. The computer yields the over-all vapor or liquid fraction to a probable error of 0.002. An interpolation method is described which reduces the probable error to 0.0002. INTRODUCTION The process known as Flash Vaporization may be described as follows: A mixture of known composition of relatively volatile components is allowed to come to thermodynamic equilibrium at some given temperature and pressure by any path whatsoever. In general, a vapor and a liquid phase will be present at equilbrium. The problem is to determine the fractions of the mixture in the vapor and liquid phases, and to determine the mol fractions of the various components in each phase. The relations that exist between these various quantities at equilibrium are well known and will be given later. These equations are difficult to solve, yielding only to trial-and-error methods. The virtues of the analog computer are its speed and the automatic character of its calculations. The device is operated by turning a crank which varies the value of v, the total vapor fraction. This actuates a number of servomechanisms which perform automatically all computations. This computer facilitates the solution of such problems as (to mention a few] : Analysis of separator operation, studies of changes in composition of reservoir fluids with pressure decline, and analysis of natural gasoline plant operation. Examples of some of these problems are given in detail later. STATEMENT OF THE PROBLEM The essential equations are xm = 1+(km-l)v . . (1) Kmzm ym =------------------------. . (2) 1 + (Km-l)v ; n n S S xm= --------------------=1 1 + (Km-l)v m = 1 m=1 ......(3) n n S = S Km Zm =1 S ym 1 + (Km-1)v m=1 m = l ......(4) where ym = Vm/V = Mol fraction of m-th component in vapor phase. xm = Lm/L = Mol fraction of m-th component in liquid phase. Zm = Fm/F = Mol fraction of m-th component in mixture. KM = ym/xm = Equilibrium constant for m-th component at the given temperature and pressure. v = V/F = Mol fraction of vapor in the mixture. F = Total mols of a mixture of n components. V - Total mols of vapor in the mixture. L = Total mols of liquid in the mixture. Fm = Total mols of m-th component. Lm = Mols of m-th component in liquid phase. Vm = Mols of m-th component in vapor phase. Assuming the total mols of the mixture F and its composition (all the Fm) to be known — either from a quantitative analysis or from the amounts put together to make the mixture — the zm = Fm/F may be regarded as known quantities. From the given temperature and pressure of the mixture the K., are known (principally from experimental data). The primary problem is then: having given all the zm and Km to calculate v and all the fractions xm and ym. Having calculated these ratios, it is a simple matter to calculate V. L. Vm, Lm from their defining equations. SOLUTION OF THE PROBLEM Basic Computer Unit The computer consists of n units, each of which provides voltages xEo and YE, proportional to x and y as in

Jan 1, 1950

By W. S. Hyler, W. L. Bruckart

The powder metallurgy and arc-cast types of wrought molybdenum stock were studied in rotating beam fatigue. Endurance ratios of unnotched specimens after 5x10 cycles were found to be 0.74 and 0.81, respectively. Fatigue notch factors (K,) of 1.36 and 2.05, respectively, were obtained on V-notched specimens having a theoretical stress-concentration factor Kt = 3.1. MOLYBDENUM and its alloys are rapidly gaining importance as engineering materials of construction for high temperature applications. AS a wrought material, molybdenum and its alloys are outstanding above 1800°F. It is necessary, however, with molybdenum to provide elevated temperature oxidation protection. This is currently being studied by a number of investigators. Thus far, there have been no published data on the fatigue behavior of molybdenum either at room temperature or at elevated temperature. The knowledge of fatigue behavior, particularly notched fatigue behavior, becomes increasingly important as the engineering importance of a material increases, especially from the design standpoint. The work described here was directed toward determination of the notched and unnotched rotating beam fatigue characteristics of molybdenum at room temperature. Also, it was desired to learn whether any difference exists between wrought molybdenum of the powder metallurgy and the are-cast types. Fabrication History of Bar Stock For this work, bar stock was procured from Climax Molybdenum Co. (are-cast type) and from Westing-house Electric Corp. (powder metallurgy type). The processing histories of these materials are listed in Table I. The first lot of are-cast molybdenum was deoxidized by aluminum additions. The second lot was deoxidized by carbon." Both lots of the powder Microstructure A microscopic examination of these materials, as received, revealed the powder metallurgy molybdenum to be free of inclusions and to be finely fibered. On the other hand, the are-cast C molybdenum possessed numerous carbide inclusions, and the arc-cast A appeared to have coarse grained fibers. These are shown in Fig. 1. Owing to the large apparent difference in fineness of fiber between the are-cast A and the powder product and are-cast C, a check was made of the recrystallized grain sizes to determine whether these differences persisted. A series of recrystallization anneals revealed mixed grain sizes in both the are-cast A and the powder product. Variations in grain diameter of as much as 50 times were found after full recrystallization at 1200°C for 30 min. There was no distinction between the two. Arc-cast C was not checked. Room Temperature Mechanical Properties Prior to subjecting the materials to fatigue tests, room temperature mechanical properties were determined. These data, together with the tensil: strength of the notched bars, data,togetherare listed in Table tensil: There were no significant differences in the tensile strengths of the two types of molybdenum. However, the powder metallurgy material was generally more ductile and yielded at a somewhat lower stress than the arc-cast material. Tensile strengths of the notched specimens indicated a slightly greater notch strengthening effect in the are-cast C than in the powder product, lot 2. Room Temperature Fatigue Properties Sample Preparation: Specimens were prepared for use in R. R. Moore rotating beam fatigue machines. These specimens had a continuous radius test section with a minimum diameter of 0.260 in. The specimens were prepared by grinding to shape and polishing to produce a smooth unworked surface. Polishing consisted of rotating a disk in the continuous radius section on which were attached, successively, emery belts of 240, 400, and 600 grit. The specimens were slowly rotated during the operation. The belt speed was high, and rotation was such that polishing scratches were longitudinal. Unnotched Fatigue Properties: Unnotched fatigue data were obtained from a number of specimens of each of the four lots of molybdenum. The results of the tests are presented in Fig. 2. From this figure,

Jan 1, 1956

By M. T. Hepworth, T. P. Papazoglou

Hydrogen diffusion in hcp and bcc titanium in the temperature range 610° to 900°C was measured by reacting cylindrical specimens of high-purity titanium at constant temperature with small constant pressures of hydrogen of the order of a few Torr. In the range 0 to 6.7 at. pct H in hcp titanium, the diffusion coefficient D in sq cm per sec is given by: where T is in °K and R in cal per g-atom OK. Only three measurements of diffusivity in bcc titanium were made: at 900' : the diffusivity in ß titanium was found to be 1.8 x 10-4 sq crn per sec. McQUILLAN1 has studied the constitution of the Ti-H system and reported the concentration of hydrogen as a function of pressure of hydrogen and temperature. He found hydrogen absorption was reversible and rapid at temperatures above 500°C in both crystallographic modifications, the a (hcp) phase and the ß (bcc) phase, with the latter phase stabilized by hydrogen additions. Wasilewski and Kehl 2 measured the diffusivity of hydrogen in hcp and bcc Ti-H alloys. For bcc alloys, they reacted cylindrical specimens of titanium with hydrogen gas at 1 atm pressure. They measured the rate of absorption of hydrogen isothermally on specimens ranging from zero hydrogen content when above the transformation temperature to hydrogen contents sufficient to stabilize the bcc phase when below the transformation temperature. By integrating Fick's second law for a cylindrical geometry, assuming a diffusion coefficient independent of composition, they calculated diffusivities as a function of temperature. In the range 650" to 1000°C their results in bcc titanium yielded the expression: where D is in sq cm per sec, T in OK, and R the gas constant in cal per g-atom-OK. Their studies on the hcp a phase were conducted by exposing long cylindrical titanium specimens to hydrogen at a sufficiently high pressure to establish a surface layer of fl phase 80 to 120 u thick for the duration of an isothermal experiment. Concentration profiles were then determined by hydrogen analysis from chips obtained upon machining quenched specimens. The diffusivity of hydrogen was then calculated yielding the expression: The present investigation was undertaken principally to redetermine the diffusivity data on a titanium by a direct absorption from the gas phase at sufficiently low hydrogen pressures to avoid the presence of a surface bcc phase and also to check the diffusivity data at low hydrogen concentrations in the bcc stmcture. EXPERIMENTAL PROCEDURE High-purity electrolytic sponge titanium (99.9 pct) of BHN 91 was arc-melted, forged, rolled, and machined into cylinders of final hardness, upon annealing, of BHN 92. Cylindrical specimens of two sets of dimensions were used: 1.524 by 2.931 cm and 1.905 by 3.663 cm diam and length, respectively. The experimental apparatus used for annealing specimens and for admitting known amounts of hydrogen at constant pressure is shown in Fig. 1. In a typical experiment, a specimen was lightly abraded with fine-grain silicon carbide, washed with water and detergent, and then thoroughly degreased with reagent-grade tri-chlorethylene and acetone. The specimen was suspended by a Chrome1 wire in the center of a quartz furnace tube, without contact with the tube, and heated to 925°C (as measured by a thermocouple adjacent to the specimen, outside the tube) for a period of at least 4 hr under a vacuum of from 10-6 to -7 Torr as measured by an ionization gage. The furnace temperature was then adjusted to the temperature chosen for the particular diffusion study and the specimen allowed to equilibrate thermally. Hydrogen was then ad-

Jan 1, 1969

By A. G. Guy

In a special type of electrodiffusion experiment the entire specimen is at constant temperature. The vacancies are then created at one end of the specimen and flow to the other end where they are destroyed. The energy required to operate the corresponding sources and sinks results in a super saturation of vacancies at one end of the specimen and an under -saturation at the other end. This gradient in vacancy concentration produces a chemical-diffusion flux, J(D), of vacancies in addition to the flux, J(E), caused by the electric current. Using the experimental data of Ch. Herzig,5 it is shown that a super saturation of 1 pct can cause J(D) to be about 25 pct of J(E), and thus account for the anomalous behavior of short specimens. THERE are advantages in a phenomenological treatment as contrasted to an atomistic one. For example, there are many reactions involving the element copper that are quantitatively and usefully understood from a phenomenological viewpoint, even though electron theory is not yet able to predict whether copper should be fcc or cph.' Similarly, the lack of a complete atomic theory of vacancies need not prevent the development of a treatment based on thermodynamics and chemical kinetics. In the following discussion, models will be used for the vacancies and for the mechanisms by which they move or are created or destroyed, but the phenomenological treatments are not dependent on the details of these models. The important product of a phenomenological analysis is an equation that describes the relation among the measurable quantities determining the behavior of the system. Based on such an equation, experiments can be performed to study these quantities in detail. Eventually atomic theory will supply an adequate theoretical base for the phenomena in question. A chemical species whose chemical potential was invariant during reactions in binary and ternary systems of varying composition would indeed be an anomaly. Yet this is the role that has frequently been assigned to the vacancies in pure metals, in dilute solutions, and even in ordinary alloys. The weight of available evidence supports the view that supersatu-rations of vacancies are on the order of 1 pct in usual diffusion processes.' Although it may seem strange that such a small deviation from equilibrium could significantly affect the thermodynamic properties of the vacancies, it is well known that the driving force for an ordinary chemical reaction is the departure of the components from their equilibrium state. Similarly, the reaction of vacancies that results in the formation of pores in various types of diffusion3 can reasonably be attributed to the supersaturation of vacancies. In the present paper, it will be shown that small degrees of supersaturation can significantly affect the kinetics of the diffusion process itself. ISOTHERMAL ELECTRODIFFUSION Two different experimental techniques are convenient for studying the behavior of vacancies in a pure metal in the presence of an electric field. Gilder and Lazarus4 studied self-electrodiffusion in a gold specimen in which only the central portion was at high temperatures. In this case the significant vacancy sources operate inside the specimen in the presence of a temperature gradient. In the laboratory of Th. Heumann a method has been devised for studying self-electrodiffusion at constant temperature.5 In principle, there is then no effect of thermodiffusion. It will now be shown why in this case the significant ATungsten disks \^^ ^^ ^Gold disks^^ ^^"T>.... "TTZA ? Up Iy'7 markers ///// ////// initial plane p I y/y/y of joining \I 7//// V7//A_________I_________V//// -I 0 +I Distance from center of specimen (a) -— Direction of flow of electrons —Direction of flw of vacancies %yy} Matano I z~irkendall /%W, vy/A inlerfacej "interface ///// -} HfiK- +/ 0 Distance from center of specimen (6) Fig. 1—-Schematic illustration of the essential features of isothermal self-electrodiffusion. (a) Initial condition of specimen. (b) Condition of specimen after electrodiffusion.

Jan 1, 1970

By B. Trafton, H. L. Hartman

An earlier paper1 introduced a rapid solution for mine ventilation network problems, employing the digital computer. In this paper, refinements and additions to the computer program which expand its scope of application and a solution to a ventilation network having a coal-mine configuration are presented. The revised program-written in FORTRAN for the IBM 7074 computer in Penn State's Computation Center-is applicable to any mine ventilation network employing natural splitting and having its fan located externally. It accomplishes the following: 1) Distributes an assumed total quantity of air flow among all airways of the network, performing iterations until an acceptable level of error is attained and the flow in each airway is known. 2) Computes the pressure drop (head loss) in all airways and the overall network and mine head losses. 3) Determines the operating point for the mine and a given fan, adjusting all air flows and head losses to the actual values. The basis of analysis used is the Hardy Cross method of successive approximation for solution of any fluid network, explained in detail in the earlier article. In brief, corrections to assumed air flows are computed for all the loops of a network; following distribution to the airways of each loop, the process is repeated until acceptable accuracy is attained. When a hand or desk-calculator solution is used, there is advantage to close approximation of the actual flows in the original assumption of air-flow distribution in the network. Since publication of the previous paper, in which assumption of initial quantities was also necessary in the computer solution, the program has been modified to make the first distribution itself, assigning the total quantity to one specified path in a split and then proceeding to correct the flows. The number of iterations required for a high-accuracy solution is little affected, at least in networks of few loops, 10 to 15 being sufficient to reduce the error to 1 pct or less. The principal limitation of the original program was that it contained no provision for handling a fan. Further, it did not determine the overall head loss in the network and was not designed for multiple parallel openings, such as occur in coal mines. The revised program was written to overcome these disadvantages and also to print out the adjusted flows for all airways, corrected for the system operating point. At the present time, the program can handle only an external fan. The procedure employed in solving a ventilation network is now as follows: 1) Input data consisting of airway characteristics and loop designation along with assumed direction of flow, for each branch of the network and any trunk airways, are fed into the computer. 2) Employing the resistance relationship, where R is resistance, K is friction factor, P is perimeter, L is length, A is area, and n is number of airways (where multiple intakes and returns are employed); values are computed for all airways. 3) An arbitrary, continuous route of flow through the circuit—from beginning to end—is selected and the path identified by airway numbers and read in. 4) Based on a total assumed quantity of flow Q (usually 100,000 cfm.) the computer employs the procedure described in the previous article in distributing quantities of flow among all the airways until a satisfactory level of accuracy has been attained (in the current program, the machine is turned off when the cumulative corrective error in cfm, is reduced to five times the number of airways). 5) Airway head losses H are calculated by the relation H = R Q2 and cumulated along the original designated flow route to find the overall head loss for the network. Trunk airways are then included to obtain the mine head.

Jan 1, 1964