Search Documents

By J. E. Sandor, P. F. Schmidt

A general review of the present knowledge of the semiconductor/oxide interface is given, with emphasis on silicon and germanium. The electrical properties of clean surfaces and the changes that occur upon oxidation are described. A brief disczcssion of the methods and mechanisms of oxidation of silicon is given, and the interaction between the semiconductor substrate and the oxide phase is considered. Physical properties of the oxide film, charge storage effects due to the dielectric, and some device applications of oxide films on semiconductors are discussed. THIS paper will be heavily preoccupied with the elemental semiconductors germanium and silicon. Undoubtedly some injustice is done thereby to other semiconductor-oxide systems, but it is not readily possible to cover all aspects of so vast a field in the present review. In considering the semiconductor/oxide interface, we may start out by discussing briefly an idealized clean semiconductor surface, and may then follow the changes brought about by oxidation, as the oxide is allowed to grow progressively thicker. The references given are not a complete list, but are to be taken as guides to the literature. Tamml suggested in 1933, based upon the study of a one-dimensional asymmetrical potential model, that localized surface states may appear at the surface of a crystalline lattice; it was pointed out by Fowler that these states should occur in pairs. shockley3 examined the origin of such surface states on the basis of a more general one-dimensional model and found that surface states may appear in the forbidden region between the highest filled and the lowest vacant band. This happens when the two bands overlap, as, for instance, the "s" and ('p" bands of the diamond structure. In a three-dimensional model these levels would produce a surface band which is half-filled for diamond, and the number of surface states should be approximately equal to the number of surface atoms. This picture of the surface has been refined by more recent theoretical work, like Handler's4 and Koutecky and Tomasek's~ without introducing essential modifications. When these concepts were applied to semiconduc-

Jan 1, 1965

By K. M. Busen, J. Lindmayer

A heterojunction model is suggested for descrihzng the electronic conditions at oxide -semiconductor interfaces. Detailed study of the silicon oxide-silicon interface shows that the heterojunction model explains many experimental facts; however, it is necessary to assume that a transition layer exists at the interface. It will he demonstrated that ionic contrbution way also he present. Extended low-temperature heating stabilizes the oxide to a charge carrier density in the order of 10' cln-: determined by the heterojuncton model. OXIDIZED semiconductor surfaces are widely used, since they offer technological advantages and "passivate" the surfaces of semiconductor devices. However, it has been recognized that there are always n-type inversion layers formed in the silicon oxide-silicon system, which must be regarded as a definite disadvantage. An added complication is the "instability" of these inversion layers;' the inversion layer appears to vary with the processing and they change at temperatures as low as 100" to 200°C This paper will point out that the inversion layer is a result of the dissimilarity of the contacting materials and therefore must always be present. The dissimilarity is the essence of the heterojunction or electronic model and involves directly the structure of the materials involved. In addition, one finds the contribution from ions; this disturbance, however. can be eliminated for the case of silicon oxide-silicon interface. The strength of the inversion layer or channel will be expressed in terms of the density of elementary charges per square centimeter, N*, maintaining the channel. The next section contains the formulation of the heterojunction model by assuming a contact between two homogeneous materials. This calculation should not imply that silicon oxide is homogeneous; it serves only to point out the general properties of the electronic model. Following the electronic model, the practical case of the silicon oxide-silicon interface will be discussed. I) THE HETEROJUNCTION MODEL Contacting an oxide with a semiconductor, a work-function difference must be accommodated. Fig. 1 represents the oxide and the semiconductor before and after contact. When the contact has been made, one finds that qVi + qV, = A@; i.e., the work-function difference is distributed between the oxide and the semiconductor. Normally, the work-function difference appears on the wide-bandgap material, but in the present case the oxide is thin and far from being a single crystal. Accordingly, a significant fraction of the work-function difference must appear on the semiconductor. The work function of silicon2 is about 4.8 ev, but we shall not assign any particular work function for the oxide as it may vary. However, oxides usually have lower work functions and therefore the potentials must build up in the manner shown in Fig. 1. Such an argument predicts that the inversion layer must be n type, as opposed to the clean surface which is p type. Another important factor to be considered is that the oxide is disorderly and therefore contains many trapping levels. As a result, the charges in the oxide arise primarily from the change in trap occupation, while on the semiconductor side the space charge is composed of mobile free electrons. The consequence of this fact is that a change to another thermal equilibrium will require some time, dictated by the traps in the oxide. The conditions can easily be calculated on the semiconductor side. Assuming an intrinsic semiconductor, Poisson's equation is

Jan 1, 1965

By P. Greenfield, P. A. Beck

The phase is a hard and extremely brittle material with a tetragonal crystal structure, containing 30 atoms per unit cell' It occurs in many binary and ternary alloys of the transition elements. The existence of twelve a phases has been reported in binary alloys of metals from the group vanadium, chromium, molybdenum, and tungsten with metals from the group manganese, iron, cobalt, and nickel. It is a characteristic fact that the composition of the o- phases in the various binary systems is not the same. Rather, there is a gradual shift in composition, for instance in the series of chromium a phases, approximately as follows: CrMn,, FeCr, Cr3Co,. In recent years several ideas have been put forward to account for this variability of composition. SullyZ proposed that a was a type of electron compound, characterized by 1.7 electrons per atom in excess of the number required to fill the Pauling atomic orbitals. These excess electrons were supposed to just fill the first Brillouin zone of the u structure. However, later work indicated that the phase is ferromagnetic at low temperatures, so that a- is unlikely to be a full zone compound.~ Another proposal,* which is also based on the structure of the transition elements advanced by Pauling, and in fact leads to essentially the same a compositions as Sully's theory, suggests that the u phase is characterized by a constant number of electron vacancies per atom. The electron vacancy numbers N,. used were originally the same as those in the Pauling theory, but were later modified empirically vn order to bring about better agreement with data for the p phase in certain ternary systems. For alloys, N,, was calculated on the assumption of simple additivity from the electron vacancy numbers of the components, as follows: N,. = 4.66 (Mo + Cr + V) + 3.2 (Mn) + 2.2 (Fe) + 1.71 (Co) + 1.6 (Ni). On this basis a reasonably constant value of N, = 3.4 was obtained for all of the binary u phases and a fairly constant value of N,, = 3.1 for the various p phases. More recently, Bloom and Granta pointed out that a similar correlation with the total number of 3d and 4s electrons per atom is also fairly well fulfilled and concluded that no reference need be made to the Pauling theory. This proposal does not lead to the same binary a compositions as the one described above, so that the relative merits of the two proposals may be determined on the basis of direct comparison with experimental composition data. The purpose of the present paper is to provide such a comparison. Since the experimental data for the composition ranges of several of the known u phases were rather incomplete, it was decided to carry out new experimental determinations for many of the binary systems. Experimental Procedure The alloys were prepared by induction melting in recrystallized alumina crucibles, either in vacuum or in helium atmospheres. Some aluminum was picked up from the crucibles in the alloys of high vanadium content, but this was minimized by keeping the alloys in the molten state for as short a time as possible. The maximum aluminum impurity was 0.13 wt pct by chemical analysis, in an alloy containing over 50 wt pct V. In general, the as-melted compositions agreed accurately with those obtained by chemical analysis. In the case of manganese, however, losses inevitably occurred; consequently all manganese alloys were analyzed. Specimens of all alloys were annealed at temperatures ranging from 1000" to 1300°C in an atmosphere of purified 92 pct helium and 8 pct hydrogen mixture. All manganese alloys were sealed in quartz tubes under vacuum before annealing. The specimens were held at temperature for between four and eight days before being quenched into cold water. Alloys were prepared for microscopic examination by mechanical polishing, usually followed by electrolytic etching. X-ray diffraction specimens were prepared by crushing the homogenized alloys, and by reanneal-ing the obtained powders in evacuated quartz tubes for a few hours at the temperature of the lump anneal. X-ray diffraction patterns were made with unfiltered chromium radiation, using an asymmetrical focusing camera of high dispersion. Using microscopic and X-ray techniques jointly, the accuracy

Jan 1, 1955

By J. B. Clark

In 1933, R. paris1 published a phase diagram for the entire Mg-Ca-Zn system. He reported the existence of a ternary intermetallic phase, Mg5Zn5Ca2, which formed a quasi-binary with the magnesium solid solution. Since then, no extensive work has been reported on the system. Recently however, indications of the existence of a second ternary in-termetallic phase in equilibrium with the magnesium solid solution led to a reinvestigation of the solid phase equilibria involving the magnesium solid solution. The 335°C (635°F) isothermal section was chosen for study. Metallography and powder X-ray diffraction were selected as the best mutually corroborating methods for determining the solid phase constitution. Seventy-six alloys were prepared from singly sublimed (99.95 pct) magnesium, electrolytic (99.95 pct) zinc and USP calcium. The analyses of some of the alloys are shown in Fig. 1. The alloys were melted in a graphite crucible under a chloride flu and chill cast in a split graphite mold. Dif-fusion couples (see Fig. 2) were prepared by casting molten Mg-20 wt pct Zn alloy around a cylinder of Mg-30 wt pct Ca alloy. The alloy samples and the diffusion couples were sealed under argon in Pyrex vials and annealed in a furnace capable of controlling within 1/2°C of 335°C. At the conclusion of the anneal, the alloy samples were quenched in ice water and the diffusion couples were air cooled. Procedures standard for magnesium alloys were used for metallographic examination. Powder X-ray diffraction patterns were taken on a G. E. camera with a 14.32 cm dim using nickel-filtered copper radiation. The redetermined 335°C (635°) section of the Mg-Ca-Zn phase diagram is shown in Fig. 1. TWO ternary intermetallic phases, designated w and

Jan 1, 1962

By R. E. Ogilvie, A. R. Kaufmann, S. H. Gelles

The solid-solubility limits of iron in beryllium were determined between 850o and 1200oC by analysis of differential type multiphase diffusion couples, using an X-ray absorption technique. The maximum value of the solubility limit was found to be 0.92 ± 0.02 at. pct (5.46 wt pet) at the eutectic temperature 1225°C. The solubilities of nickel and beryllium were determined between 900°and 1200°C by the same technique and the maximum solubility was found to be 4.93 + 0.01 at. pct (25.2 wt pet) at the eutectoid temperature, 1065°C. A previously unreported high-temperature phase which decomposes eutectoidally at 1065 °C was found to exist in the beryllium-nickel system at a composition of approximately 8 at. pct Ni (36 wt pet) by diffision-couple analysis. The presence of this phase was confirmed by thermal analysis and metallo-graphic analysis of the structure resulting from the eutectoid decomposition. G. V. Raynor1 has treated the solid solubilities of some of the elements in beryllium on the basis of the "Hume-Rothery" rules2 which have been modified to include ionic size and ionic distortion effects. It was predicted that the solubility of iron and nickel in beryllium should be slightly less than that of copper. The lowering of the solubility, according to Raynor, is due to a more unfavorable relative valency effect and an ionic size effect. Kaufmann and corzine3 have compiled data on the solubilities of elements in beryllium and have discussed them in the light of the Raynor paper. These authors feel that, because the elements having the greatest solubility in beryllium systematically fall in the Group VIII and IB Columns of the periodic table, the electronic structure greatly influences the maximum solid solubility of elements in beryllium. The solubility of iron in beryllium was determined by Teitel and cohen4 as part of the study of the beryllium-iron phase diagram. The determination was carried out by X-ray and thermal analysis and according to the phase diagram presented, the maximum solubility of iron in beryllium is 0.41 at. pct (2.5 wt pct) at 1225oC. However, it is estimated that the uncertainty in the position of the a-beryllium primary solid-solution boundary is about 0.5 at. pct (3wtpct). Losana and Goria3 in studying the beryllium-nickel phase diagram, determined the solid solubility of nickel in beryllium by thermal analysis. They found the maximum solubility to be between 1.65 and 2.65 at. pct (10 to 15 wt pct) at 1240°C. This value decreased rapidly with decreasing temperature. In determining approximate ranges of solubilities for different elements in beryllium, Kaufmann, et al,8 reported a value of between 1.3 and 1.7 at. pct (7.9 to 10.1 wt pct) for the solubility of nickel in beryllium. The value was obtained by metallographic examination of quenched alloys and lattice-parameter measurements. However, the authors also noted a single-phase structure for a 1.7 at. pct Ni alloy (10 wt pct) on cooling from the liquid. This would indicate a higher solubility range than was reported. ~isch,' in his X-ray studies of beryllium-copper, beryllium-nickel, and beryllium-iron intermetallic compounds, reports the disappearance of a second phase (Ni,Be2) in the beryllium primary solid solution at approximately 4 at. pct (20 wt pct). THEORY The analysis of concentration gradients in diffusion couples has proven to be a useful tool in determining phase equilibria.8-14 In this particular study the diffusion couples were chosen to straddle the expected composition range of the phase boundary, then heat treated at a given temperature and the concentration gradient evaluated. The composition of the phase boundary for a given temperature appears at a point of discontinuity of the composition gradient. Examples of typical phase diagrams and the concentration gradients which should be found in such systems are shown in Fig. 1. In the present work, gradients of the form of Fig. l(c) were obtained in diffusion couples made of pure beryllium and two-phase alloys of beryllium with either iron or nickel. The composition at the point where the gradient becomes discontinuous, Cs, corresponds to the solubility limit of either iron or nickel in beryllium. The analysis of the concentration gradients was carried out by an X-ra absorption method developed and applied by Ogilvie and later used by Moll13 and Hilliard.l4 It depends on the fact that the absorption of X-rays by matter is determined by the concentration and type of the various atomic species present. The relationship for the intensity, I, of a monochro-

Jan 1, 1960

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

The solid solubility of holmium in copper, silver, and gold, determined using metallographic techniques, was found to be a maximum of 0.06, 2.4, and 3.2 wt pct, respectively, nt the eutectic temperatures. The temperatures of these eutectic horizontals, determined by differential thermal analysis, were found to be 868", 787", and 778"C, respectively. The first compound on the silver-rich side of the Ag-Ho system was found to be A&Ho2. In a study by Gold et al.,' on the effect of small amounts of soluble impurities on the electrical resistance and thermoelectrical properties of copper, traces of iron were shown to have drastic effects on the thermoelectric power due to the magnetic scattering of the conduction electrons. Since little is known about the state of the iron ion when dissolved in copper, Gold had hoped to repeat these experiments using certain rare earths as the magnetic scattering centers; because of the tightly bound 4f electrons it is not expected that their spectroscopic state would be affected by alloying. In such an experiment the sole role of the trace rare-earth impurity would be to magnetically scatter the copper conduction electrons without producing appreciable distortions in the otherwise pure copper matrix. Since holmium was chosen for such a study because of its high magnetic moment, it was necessary to establish precisely its solid solubility in copper. The study was extended to include the solvents silver and gold, so that an evaluation of the data using classical solubility parameters could be made. No previously reported work on the systems of holmium with copper, silver, or gold was found in the literature. Information on the solubility of other rare earths in these three metals is also very meager. The solubility limit of yttrium in copper was reported as less than 1 wt pct2 and the solubility of praseodymium in gold as 0.35 wt pct,3 while no references were found on solubility studies of rare earths in silver. EXPERIMENTAL Materials. The materials used in this investigation were distilled holmium prepared in this Laboratory, OFHC* copper, electrorefined silver (99.9 *OFHC = oxygen-free high conductivity. pct), and gold with the analyses given in Table I. Sample Preparation. Samples weighing 1 to 2 g were prepared by arc-melting the components together on a water-cooled copper plate under purified argon atmosphere. The samples, turned and remelted at least five times to ensure homogeneity, showed weight changes of less than 0.1 wt pct after melting. Samples containing less than 0.7 wt pct Ho were prepared by diluting 1 to 2 wt pct Ho alloys with the pure solvent, while samples with concentrations greater than 0.7 wt pct Ho were prepared directly from the metals. The samples were wrapped in tantalum foil, sealed in outgassed Vycor tubing under helium, annealed at the desired temperature for at least 24 hr, and then quenched by shattering the Vycor tube in cold water. Metallographic Methods. Standard mechanical polishing methods were used in preparing the silver and gold alloys for examination, while the copper alloys were electropolished using a 40 pct solution

Jan 1, 1963

By Walter Showak

The published phase diagrams for the copper-zirconium system1,2 indicate that the solid solubility of zirconium in copper is not accurately known. Studies made on relatively impure alloys by Pogodin, Shumova, and Kugucheva3 yielded a maximum solubility value of 1 wt pct Zr at the eutectic temperature. Allibone and sykes5 reported that an as-cast 0.2 wt pct Zr alloy is single phase at room temperature. Limited results obtained by Raub and Enge14 indicated that zirconium solubility in copper is less than 0.13 wt pct Zr at 940°C. Tn 1960, on the basis of a microstructural examination of several copper-rich Cu-Zr alloys, saarivirta6 placed the maximum solubility of zirconium in copper at 0.15 wt pct at the eutectic temperature, 980°C. The present note describes results obtained during a re-investigation of the Cu-Zr system for the purpose of accurately establishing the solidus and solvus curves and the eutectic temperature in copper-rich alloys. High-purity copper (99.99 pet) and zirconium (99.64 pet) consolidated reactor grade sponge were used to prepare several alloys, containing up to 2 wt pct Zr, by induction melting in graphite crucibles in vacuum and chill casting into a split steel mold. The melting procedure consisted of first melting the copper charge and then adding the bulk zirconium wrapped in pure copper foil. Chemical analyses were made of both the cold-rolled alloy ingots after homogenization at 925°C for 450 hr in evacuated quartz capsules and of small specimens which had been used in the phase diagram studies. 'The carbon content of each alloy was also determined and was found to be less than 0.001 wt pct in every case. Metallographic and electrical resistivity measurements were used as the primary methods for this investigation, although some thermal analysis studies were also performed. Heat treatments of all the alloys were carried out with the specimens wrapped in molybdenum or alloy foil (of the same composition as the specimen) and sealed in quartz tubes either in vacuum or under a purified helium atmosphere. Following high-temperature anneals for long periods of time, slight depletion of zirconium from the specimen surface was observed. Therefore, careful preparation of samples for microscopic examination and chemical analysis of the samples themselves was necessary to check the composition of samples used in determining phase boundaries. Points on the solvus curve were determined on quenched specimens by both isothermal reaction after solution heat treatment at 950°C for 8 hr and by isothermal annealing of alloy specimens at successively higher temperatures for times ranging from 24 to 500 hr. Liquation studies for positioning of the solidus and eutectic temperatures were performed by holding a pre-

Jan 1, 1962

By F. Weinberg, E. H. McLaren

Dilute binary alloys have been solidified under controlled thermal conditions, and solute distributions, temperatures during freezing and melting, and the position and morphology of the solid-liquid interfaces have been examined. Binary Zn, Sn, and Pb based alloys were investigated with solute additions of AgU°, AU°, Sb'24, TlZo4, snH3, and Zns5. Approximately 150 cc of metal was used in each ingot. Appreciable solute segregation occurred in most of the ingots, depending on the solute concentrations, distribution coefficients. and freezing conditions of the alloys. Measurements were made of the general solute distribution by sectioning layers of material from the ingot circumference and counting the activity of each layer. Local segregation was detected by autoradiographic techniques. Freezing and melting curves for the same alloys were obtained by precision resistance thermometry. These curves are related to the corresponding measured solute distribution curves and those calculated from theoretical expressions. The position and morphology of solid-liquid interfaces during freezing were determined by both decanting and quenching techniques. The results indicate the manner in which freezing and remelting occurred under the conditions investigated. If an isomorphous dilute binary alloy is solidified or melted under equilibrium conditions, the concentration of solute in both solid and liquid would be uniform at all stages in the freezing process. The solute concentrations during freezing could be determined from the phase diagram of the alloy, and the freezing curves would have an alloy slope defined by the solidus and liquidus points. In practice, however, solidification under equilibrium conditions rarely, if ever, occurs, primarily because of low rates of diffusion in the solid and incomplete mixing in the liquid. As a result, the solute distribution and the shape of the freezing and melting curves obtained in a given system depends on a number of factors. These include, in addition to the extent of diffusion and mixing, the initial solute distribution and concentration, the distribution coefficient, the rate of freezing or melting, supercooling of the liquid, the morphology of the solid-liquid interface, and the effect of annealing. The purpose of the present investigation was to determine, experimentally, actual solute distributions and thermal curves for a series of dilute binary alloys, using tracer techniques to determine the solute distributions and precision resistance thermometry for the temperature measurements. In addition, observations were made of the positions and morphologies of the solid-liquid interfaces during freezing and remelting. The distribution of solute in a dilute alloy, assuming complete mixing in the liquid, and negligible diffusion in the solid, is given by1 where Co is the mean solute concentration of the alloy, C, is the solute concentration in the solidg is the fraction of the ingot solidified, and ko is the equilibrium solute distribution coefficient (assumed to be independent of concentration for dilute alloys). For the case of partial mixing, ko can be replaced by an effective distribution coefficient k. If solute transport in the liquid is assumed to be entirely due to diffusion, with negligible diffusion in the solid, then the solute distribution in the solid is given by Tiller et a1.' as where R is the rate of advance of the solid-liquid interface, D is the solute diffusion coefficient in the liquid, and x is the position of the interface from the start of freezing. The temperature at the interface TI is related to the solute concentration of the solid at the interface C, by the expression where Tm is the freezing temperature of the pure solvent and ml is the slope of the liquidus for the binary alloy being considered. Accordingly, by relating TL to the actual temperatures measured in the present experiments, solute distributions can be deduced from the temperature measurements. Using the above expressions, the measured solute distribution of a given alloy may be compared to the theoretical distributions, as well as to the distribution calculated from the observed freezing curve. EXPERIMENTAL The solvent materials used in making up the binary alloys were C. P. New Jersey zinc (99.999 pct), Vulcan Extra Pure Tin (99.999 pet), and Tadanac lead (99.998 pct). The solute elements were of 99.99 pct purity or better, and were activated in the

Jan 1, 1963

By Robert A. Rapp, Jan E. Verfurth

From the measurement of internal-oxidation band widths in Ag-In and Cu-A1 alloys, products for the solubility No and the diffusivity Do of oxygen in silver and apparent NODo products in copper have been obtained over a range of temperature. The NODO products for silver are in agreement with the product of the individually determined No and Do values of Eichenauer and Müller. Although interstitial diffusion of oxygen in silver is expected, the temperature-independent factor Do of the diffusion coefficient does not agree well with the theory of Wert and Zener. For Cu-A1 alloys internal oxidation does not follow the expression derived for ideal kinetics, since the apparent NODo product calculated from internal-oxidation measurements is a function of the alloy content in the specimen. Thus, unequivocal values for No or DO or even the product NODO in copper are not available from the literature. RELIABLE values for the saturation solubility NO and diffusivity DO of oxygen in fcc metals are lacking and difficult to determine, partially because internal-friction measurements are not applicable. Permeability experiments and Sieverts7-type adsorption -desorption methods3 have met with difficulty; however, Eichenauer and Müller4 have recently reported on the solubility and diffusivity of oxygen in silver as determined by the latter technique . Rhines, Johnson, and Anderson,5 and independently Meijering and Druyvesteyn,' introduced the possibility of deriving values for the product NODO in a metal from measurements of the widths of internal-oxidation zones (subscales) formed during the internal oxidation of alloys. In addition to these authors, several other investigators7-9 have made such measurements in various silver and copper alloys at various temperatures, mostly using me- chanically polished specimen surfaces. The purpose of this investigation was to determine over a range of temperature the calculated products NODO in silver and copper from systematic measurements of the widths of internally oxidized zones in Ag-In and Cu-A1 alloys, using specimens with electro-etched surfaces. Consider the internal oxidation of a dilute solid-solution alloy in which the base metal is not oxidized. If diffusion of oxygen is the rate-controlling step, then the depth 5 of the internal-oxidation zone is given by where y is a dimensionless parameter and t the oxidation time. Based on the solution of the diffusion equation, wagnerl0 has shown that under certain conditions several mathematical simplifications may be used in the evaluation of y, so that Eq. [I] may be expressed for the case that y « 1, and y(Do/DB) 1, where N(so)the is the equilibrium solubility (mole fraction) of oxygen in the solvent metal for the given oxidizing atmosphere, is the mole fraction of the solute metal in the bulk alloy, 4 is the diffusivity of the solute in the base metal, and v is the ratio of oxygen atoms to solute atoms in the precipitated oxide BO,. Rearranging Eq. [2] Maak' has derived an expression analogous to Eq. [3] for the case that an external-oxide scale, in addition to an internal-oxidation zone, is formed. For this combined external and internal oxidation for the case that 5/2 where 5 is the distance of the internal-oxidation front from the original external s.urface, and X is the distance of the external oxide/metal interface from the original external interface. The function F is defined by the expression

Jan 1, 1964

By J. F. Enrietto

Internal friction measurements were used to determine the effect of manganese on the solubility and precipitation kinetics of nitrogen. Manganese, in concentrations up to 0.75 pct, has little effect on the solubility at temperatures above 250°C. On the other hand, at Concentrations as low as 0.15 pct, manganese inhibits the formation of iron nitrides, especially Fe4N, even though it may not form a precipitnte itself. The precipitation and solubility of carbides and nitrides have been extensively investigated in the pure Fe-C and Fe-N systems.1-3 In recent years, some effort has been ispent in studying the influence of substitutional alloying elements on the behavior of carbon and nitrogen in ferrite.4 -7 In particular Fast, Dijkstra, and Sladek have investigated the effect of 0.5 pct Mn on the internal friction and hardness during the quench aging of Fe-Mn-N alloys.', ' They found that at low temperatures (below 200°C) the presence of 0.5 pct Mn greatly retarded quench aging. For example, after 66 hr at 200°C very little precipitation had taken place in the iron alloyed with manganese, whereas precipitation was complete after a few minutes in a pure Fe-N alloy. The effect of varying the manganese content and the details of the precipitation process were not mentioned in these papers. Fast' postulated that manganese causes a local lowering of the free energy of the lattice with a resulting segregation of nitrogen atoms to these low energy sites. The segregated nitrogen atoms are bound so tightly to the manganese atoms that they cannot form a precipitate. The internal friction measurements of Dijkstra and Sladek tended to confirm the concept of segregation of nitrogen around manganese atoms, and the increase in free energy on transferring a mole of nitrogen atoms from a segregated to a "normal" lattice site was computed to be - 2800 cal. Dijkstra and Sladek9 distinguished between two types of precipitates: ortho, a nitride of appreciably different manganese content than that of the matrix, and para, a nitride with a manganese content essentially that of the matrix. With each type of precipitate a solubility, again designated ortho or para, can be associated. Since the internal friction maximum in alloys which were aged several hours at 600" C dropped almost to zero, Dijkstra and Sladek9 concluded that the ortho solubility must be very low. The effect of temperature on the ortho and para solubilities has no1: been investigated. There are obviously several gaps in our knowledge concerning the influence of manganese on the behavior of nitrogen in a-iron. It was the purpose of the experiments described in this paper to determine the following: 1) The ortho and para solubilities of nitrogen as a function of temperature. 2) The details of the precipitation process at elevated temperatures. 3) The effect of varying the manganese concentration on the above phenomena. EXPERIMENTAL PROCEDURE Internal friction is conveniently employed in studying the precipitation of nitrides and/or carbides from a -iron because it is one of the few parameters, perhaps the only one, which is not affected by the presence of the precipitate itself. For this reason, internal friction techniques were heavily relied upon in the present experiment. A) Preparat of -. All specimens were prepared from electrolytic iron and electrolytic manganese. Alloys containing 0.15, 0.33, 0.65, and 0.75 wt pct Mn were vacuum melted and cast into 25 lb ingots. After being hot rolled to 3/4 in. bars, the ingots were swaged and drawn to 0.030 in. wires. The wires wen? decarburized and denitrided by annealing at 750° C for 17 hr in flowing hydrogen saturated with warer vapor. To obtain a medium grain size, - 0.1 mm, the wires were then heated to 945oC, allowed to soak for 1 hr, furnace cooled to 750°C, and water quenched. Subsequent internal friction measurements showed that this procedure reduced the nitrogen and carbon concentrations of the alloys to less than 0.001 wt pct. The wires were nitrided by sealing them in pyrex capsules containing anhydrous ammonia and annealing them for 24 hr at 580°C, the nitrogen being retained in solid solution by quenching the capsule into water. Immediately after quenching, the wires were stored in liquid nitrogen to prevent any precipitation of nitrides. By varying the pressure of ammonia in the capsule, it was possible to produce any desired nitrogen concentration. B) Internal Friction. The internal Friction measurements were made on a torsional pendulum of the Ke type,'' a frequency OF 1. or 2 cps being used. For

Jan 1, 1962

By Irving Cadoff, Kurt Komarek, Edward Miller

THE lead-telluride phase diagram as compiled by ansen' has one intermetallic compound, PbTe, with solid-solubility limits extending from 22 to 45 wt pct Te (31.4 to 57.1 at. pct). Pelzel2 redetermined the solubility limits on the lead-rich side of the compound by measuring the relative areas of the two phases in samples annealed in the liquid + compound region, and obtained data that agreed with that of Hansen. Recent studies of the semiconducting properties of PbTe3 however, have indicated that addition of either lead or tellurium to PbTe does not produce samples having carrier concentrations greater than about 10 19 cm-3. Assuming that each impurity atom results in one ionized charge carrier, this carrier concentration corresponds to a solubility range of less than 0.04 at. pct. Fritts and Karrar4 have also observed two phases in samples containing from 0.15 to 3.0 pct excess Pb. To resolve the discrepancy between these data, the solid solubility limits of the compound PbTe were reinvestigated, using several other techniques. Ten 25-g samples with compositions of 1/2, 1, 3, 5, and 10 wt pct excess Pb and Te and two 100-g samples of 15 and 65 wt pct Te were prepared, sealed in evacuated quartz capsules, and melted at 950oC. The 1 pct excess Pb and Te and the 15 and 65 wt pct Te alloys were homogenized for 96 hr at 450°C; the remaining specimens were annealed 50°C below the solidus temperature reported in Hansen. The specimens were polished and etched with dilute HNO,. All the samples were two phase, the second phase appearing at grain boundaries or as spheroidized inclusions in the grains. In the case of the tellurium rich specimens, the second phase could be identified as a eutectic. Debye-Scherrer powder diffraction patterns of the i, 1, 3, 5, and 10 pct excess samples were obtained using CuKa filtered radiation. The films were indexed as the B1 or NaCl structure. The calculated lattice parameter of PbTe was the same for all the specimens, with an average of a0 = 6.460 ± 0.0005A. Lines due to excess lead or tellurium were very faint but could be clearly seen in all the films. The relative areas of the two phases in the 15 and 65 wt pct Te samples were corrected for the differences in density of Pb, Te, and PbTe, and taken as proportional to the relative volumes of the two phases. The corrected relative volumes agreed with a phase diagram having negligible solubility of both lead and tellurium in lead tellurium. The density of the two specimens was obtained to give a third independent determination of the phase limits. The specimens were broken into small pieces to eliminate any internal porosity, and adherent air bubbles were removed by vigorous shaking under water for several hours. The density of lead (11.34 g per cm3) and of tellurium (6.25 g per cm3) was taken from the International Critical Tables and that of stoichiometric PbTe (8.25 g per cm3) calculated from the lattice parameter. For the 15 wt pct Te samples, the density assuming no solubility was calculated to be 9.89 g per cm3; the observed density was 9.86 g per cm8. For the 65 wt pct Te alloy, the density calculated for the case of no solubility was 7.24 g per cm3; the measured density wag 7.22 g per cm3. These results are consistent with those of planimetry and X-ray. SUMMARY Data have been presented which show that the solid solubility of both lead and tellurium in PbTe is

Jan 1, 1961

By E. S. Anolick, Joseph Singer

The magnetic after-effect, in the form of time decay of permeability (l/µ), has been used to obtain independent data on the solubility of' carbon in pure iron. The results differ slightly from the solubilities obtained through internal -friction experiments. The mechanism of the time decay being the magnetic analogue of internal friction, it appears necessary to explain the observed discrepancies, and some suggestions are made. THE purpose of this work was to establish the reliability of the time decay of permeability as a measure of dissolved interstitial in the body-centered-cubic lattice, and further, to establish techniques suitable for such measurements with samples of a form convenient for various other measurements. This particular sample form was the "epstein" strip, i.e., 25 cm by 3 cm by 0.035 cms. While the general conclusions are not limited by this geometry, there are two aspects of the samples that may be relevant: a) texture, and b) grain size, which ranged generally from 1 to 10 mm. Effects due to these two possible parameters were thought to be small and were not investigated. Calibrating the time decay technique and the other experimental procedures by measuring the solubility of carbon in pure iron suggested itself because of the accumulation of data on this subject through the related technique of the internal friction'; The essentials of this report have been published in a somewhat condensed form.5 Since then some additional data have been taken. It is the purpose of this paper to add these data and to provide more complete description of the experimental techniques than was possible then. The time decay of permeability is a form of the magnetic after-effect.6,7 Permeability is measured immediately after demagnetization (µº) and also long after (µ8). The time decay, ?(l/µ), is the difference 1/µ8 - 1/µ0 and has been considered to be proportional to the concentration of a dissolved interstitial6,8,9,12 The relaxation time! 7, is characteristic of a particular diffusion, and affects the permeability as follows: where he being the activation energy for diffusion of the interstitial. Reference is made to the works of snoek6 and Neel7 for a complete discussion. A brief exposition of the principal ideas may be in order here, as follows. On the basis of the simplest model, resistance against domain wall motion increases with time after demagnetization due to the diffusion of interstitials to preferred sites created by the magnetostrictive tetragonolization under the strong magnetic field. Thus permeability, measured with a small field after the abrupt removal of the large field, will continuously fall with the process of diffusion of the interstitials, and the total "time decay" (or "time decrease") should be proportional to the amount of interstitial in solid solution. In the present report, use is made only of the total time decay rather than of the nature of the progress of the decay; the only problem involved in this approach is the need for assurance that only one interstitial is contributing to the effect as used. It is felt that the special use made here of the total time decay warrants the conclusions. Other work has been going on elsewhere on the detailed study of the nature of the time decay.8'9 EXPERIMENTAL A) Materials—12-lb ingots were made in a vacuum furnace from high-purity iron and by a series of hot and warm reductions were brought down to 14-mil strip. This was cut into 25-cm epsteins. The pure iron was decarburized at 750 °C by hydrogen. The time decay measurement for the decarburized iron (see below) indicated a satisfactorily low starting level of interstitials. Grain growth to grain sizes of 1 to 10 mm was produced in the pure iron by a 700" anneal in hydrogen after a critical reduction of about 7 pct as suggested by previous trials. No measurement of texture in the pure iron was made. The grain size was taken to be large enough to be beyond any sharply critical size regarding solubility. B) Infusion of Carbon—Since the permeability measurement required about two dozen epsteins, a special method of carbon infusion had to be developed to ensure homogeneity of infused carbon. This was accomplished by setting the samples edgewise in a

Jan 1, 1961

By K. K. Rao, M. E. Nicholson

IN a recent paper on the solubility of carbon in Ni-Cu alloys,' Nicholson reported that the carbon solubility appeared to be limited by the electronic structure of the alloys and that the solubility was propor- tional to the density of states at the Fermi surface of the "3d" band. In order to determine whether or not this relationship is a general one, the study has been extended to other systems. One of these is the CO-Ni system. This system was chosen for study because 1) it is adjacent to Ni-Cu system in the periodic table, 2) the alloys of cobalt and nickel do not form stable carbides at 1000°C, and 3) the system exhibits no pronounced change (less than 1 pct) in lattice parameter from cobalt to nickel,' see Fig. 1. The alloys were prepared from -200-mesh 99.92 pct Ni furnished by the International Nickel Co. and spectroscopically pure -200-mesh cobalt sponge obtained from Johns on-Matthey and Co. The powders were weighed, mixed, and pressed using a pressure of 40,000 psi. The samples were sintered for 24 hr at 900° C in dry hydrogen and then cold rolled to a thickness of 0.010 in. (Samples containing more than 70 wt pct Co had to be annealed after cold reductions of about 20 pct to avoid cracking.) The alloys were carburized at 1000°C in an alun-durn boat containing Acheson AUC powdered graphite in an atmosphere of carbon monoxide cylinder gas which had been passed through Ascarite to remove carbon dioxide. After carburizing for either 12 or 24 hr the samples were moved to a cool zone of the furnace to cool. Cooling required about 5 min. The samples were filed to remove about 0.005 in. from the surface and then analyzed using a conventional combustion-gravimetric analytic method. The results of this investigation are shown in the lower portion of Fig. 1 and Tabulated in Table I. The nickel percentages for these alloys were determined from the weights of the powders from which the alloys were made. The solubility of carbon in pure cobalt at 1000°C (fcc) is 0.25 wt pct. This is somewhat less than the 0.33 wt pct measured by Boecker.3 From his data, it would appear Boecker's samples were supersaturated before the equilibrating treatment and that a 20-min treatment was not long enough to achieve equilibrium. Since in the present investigation the 12 and 24 treatments produced essentially equal carbon contents, our figure is presumed to represent more nearly the equilibrium value. The solubility of carbon in nickel of 0.19 wt pct agrees quite well with the value of 0.21 wt pct

Jan 1, 1963

By M. E. Nicholson

The solubility of carbon in Ni-Cu alloys was determined at 1000°C. The factors which limit solid solubility are discussed with particular reference to the influence of electronic factors. The results indicate the extent of solid solubility is a function of the density of states at the Fermi surface. THE solid solubility of an interstitial element is primarily limited according to Seybolt and Fullman1 and Speiser and Spretnak2 by (a) the strain energy produced when the solute atom takes its interstitial position, and (b) the partial molal free energies of the solid solution and the coexisting phases. In several binary systems of a transition and a noble metal, where there are no abrupt changes in strain energy or free energies of coexisting phases, the solid solubility of an interstitial third element drops abruptly to a negligible value at a composition corresponding to a complete filling of the "d-band." Such a decrease occurs in the case of hydrogen in the Pd-Ag system at 60 pct Ag.3,4 The solubility of carbon in Ni-Cu alloys seems to vary in a similar manner as indicated by the embrittling effect of carbon on these alloys. These al- loys are embrittled by slow cooling due to the precipitation of graphite. This embrittlement can be removed by heat treatment if the carbon content is less than the solid solubility limit at the heat-treating temperature. If the carbon content of the alloy exceeds the solid solubility at the heat-treating given temperxture, the embrittlement cannot be eliminated by solution heat treatment.' Since pure nickel has a carbon solubility of 0.25 wt pct6 at 1000°C, embrittlement is easily removed from commercial nickel and nickel-rich alloys, which may contain of the order of 0.10 wt pct C. Embrittlement due to carbon cannot be removed from Ni-Cu alloys containing 70 to 80 wt pct Cu by annealing unless: the carbon content is of the order of 0.04 wt pct CL or less."' Thus, it appears that the

Jan 1, 1962

By F. F. Schmidt, H. R. Ogden, E. S. Bartlett

ThE solubility of carbon in tantalum has been reported to be very low at temperatures below 1500 C,1'2 increasing to at least 0.02 wt pctl at the eutectic temperature of 2800°C. The solubility limits at low temperatures were based on metallo-

Jan 1, 1963

By R. P. Smith

Iron-carbon alloys containing a single-phase austenite region and a two-phase azcstenite + cementite region at the desired temperature were prepared by partial carburization of iron or partial decarburization of carburized iron. Each sample was held at a fixed temperature until equilibrium was achieved. Analysis of the austenite for carbon then gave the solubility of cementite in austenite. Solubility measzcrements were made for the temperature range 734° to 1012°C. These values together with other data give 7560 cal for the mean heat of solution of cementite in austenite, and 0.765 pct C for the eutectoid composition of austenite. They also indicate that graphite is the stable phase at all temperatures between the eutectoid and eutectic temperature. Measurements of the solubility of cementite in austenite at 927oand 1011 "C, using a method based on the equilibration of iron with boro-cementites, are reported. The solubility of cementite in austenite (?-iron) has been computed by Darken and Gurry1 from the observed solubility of graphite in austenite, and the thermodynamic properties of pure iron and of iron-carbon alloys. They have shown that the solubility of cementite is greater than that of graphite in the temperature range between the eutectoid and eutectic, and thus that cementite is metastable with respect to graphite and austenite saturated therewith. Since their computation is not in accord with the earlier experimental work of Mehl and wells,' the present direct determination of the solubility of cementite in austenite was undertaken. EXPERIMENTAL PROCEDURE Each iron rod about 1/2 in. O.D. by 4 in. was first carburized at a temperature and with a H2-CH4 gas mixture such that the carbon content would be greater than that corresponding to the solubility of cementite in austenite at the desired (lower) temperature; sufficient carburizing time was allowed that the carbon content became substantially uniform, as illustrated by Curve A, Fig. 1. The sample was then partially decarburized with a hydrogen-water vapor mixture, giving a carbon distribution as schematically illustrated by Curve B. The sample was then held in a helium atmosphere at the temperature at which the solubility was to be determined until the decarburized portion of the sample came to equilibrium with that containing cementite. Curve C of Fig. 1 represents schematically the final carbon distribution; the horizontal portion corresponds to the carbon content of austenite saturated with cementite. To bring the sample to room temperature without change in the carbon distribution and in machinable condition, it was first quenched in a salt pot at 480°C, then in water at room temperature. Successive layers were machined from the center 2 in. of the sample and analyzed for carbon by the usual

Jan 1, 1960

By L. S. Darken, H. A. Wriedt

THE constitution diagram of the iron-copper system derived by Daniloff' exhibits, at the iron-rich end, phase fields similar to those of the iron-carbon diagram. At 1484° 1094°, and 850°C there are, respectively, a 6-y-L peritectic, a y-E-Lperitectic, and a a-y-E eutectoid. The phase designated E is copper-rich, containing 4 pct Fe or less according to the temperature. The composition of copper-saturated ferrite (a-solid solution) has been determined previously only twice as a function of temperature. By electrical conductivity measurements, Buchholtz and Köster2 found 3.4 pct Cu in ferrite at a eutectoid temperature of 810°C. They represented the composition of saturated ferrite by the equation, log (%Cu) = 4.32-4125 (1/T) Gregg and Daniloff3 reported that Köster later modified his opinion, raising the indicated eutectoid temperature to 833 °C and the copper content of eutectoid ferrite to 3.7 pct. The literature contains several indications4-7 that the work of Köster and his co-workers is not reliable. Norton,8 using an X-ray method, found that 1.4 pct Cu dissolved in ferrite at the approximate eutectoid temperature of 850°C and that the solubility decreased to about 0.3 pct at 700°C. Below 700°C his experimental data show almost no change in solubility. These findings, although inconsistent with the work of Ruer and Goerens4 and of Smith and Palmer,7 have been used generally since their publication. In investigations involving the observation of arrests in temperature during continuous heating or cooling the precise location of the eutectoid temperature has been obscured by a large hysteresis. The more reliable work 579 where the results are not effected by this hysteresis indicates a eutectoid temperature between 840° and 850°C. The foregoing discrepancies, along with a few observations in this Laboratory relating to the precipitation of copper from ferrite, prompted us to rein-vestigate the lower temperature portion of the Fe-Cu system. In this investigation the solubility of copper in ferrite was obtained by the chemical analysis of high-purity iron equilibrated with copper-silver alloys. This method eliminated the necessity for retaining by a quench the physical state of the specimen existing at the equilibration temperature. EXPERIMENTAL The reaction of the iron with the copper alloy was

Jan 1, 1961

By E. W. Johnson, M. L. Hill

Equilibrium concentrations of hydrogen in iron were measured at H2 pressures up to 136 atm and temperatures down to 145°C. Residual hydrogen was prominent near 600°C in air-melted but not in vacuum-melted iron. The diffusible hydrogen solubility is abnormally high below 390°C, confirming the hypothesis of hydrogen trapping.' It appears that the traps for hydrogen in iron form suddenly on cooling. The estimated H2 pressure in equilibrium with 5 ppm hydrogen in steel at room temperature is less than 1000 atm. It was shown previously1 that the diffusivity of hydrogen in a-Fe below 200°C is anomalously low, a behavior attributed to the presence in the metal structure of hydrogen "traps." The energy of the trapped hydrogen was computed as 4.8 kcal Per g-atom below that of the interstitially dissolved hydrogen. The strongest confirmation of the trap hypothesis should be obtainable from measurements of the hydrogen solubility at low temperatures. such measurements have been performed and are reported in this paper. The "residual hydrogen" effect1'' must be carefully controlled if valid measurements of the solubility are to be obtained, and such an effort was made in the present work. Previous data for the solubility of hydrogen in iron have been summarized by Geller and Sun.3 Their equation for the most probable solubility at 1 atm in a-Fe above 400°C is

Jan 1, 1962

By R. P. Marshall, L. P. Costas

The solid solubility of lithium in aluminum was determined by two independent techniques, electrical resistivity and microhardness, and the results are in close agreement. The solubility limits X-ray diffraction studies of the intermetallic compound LiAl indicate that its composition is nearly independent of temperature when in equilibrium with the terminal solid solution of aluminum. THE investigation of the aluminum-rich portion of the aluminum-lithium system at the Savannah River Laboratory was prompted by the observation that over-aged alloys consistently differed from the equilibrium states predicted by the latest published diagrams. Chemical analyses of these alloys showed their impurity contents to be quite low; therefore, the results obtained should have very nearly coincided with true equilibrium. A survey of the literature disclosed that previous studies had been made with metallographicig or electrical resistivity methods,4'5 but that agreement between results and technique was poor as evidenced in Fig. 1, which also includes the present results for comparison. The discrepancies have been attributed to impurities,3 and this is undoubtedly true in the early work.1'4'5 The latest Studies,''~ conducted with starting materials of higher purity, employed metallographic analysis, a technique that is quite sensitive to local homogeneity and depends strongly on subjective interpretation. In the present work the problem of defining solubility limits was approached in two ways. The first measured a physical property, electrical resistivity, at the equilibrium temperature. The second, a micro-hardness characterization of a diffusion zone, is a room temperature evaluation of a mechanical property under conditions removed from equilibrium. A close correlation between two such basically different approaches would strongly justify confidence in the results. I) MEASUREMENT OF SOLUBILITY OF LITHIUM IN ALUMINUM BY ELECTRICAL RESISTIVITY Experimental Procedure. Alloys for electrical resistivity measurements were made by melting and

Jan 1, 1962

By Rodney P. Smith

THE solubility of columbium nitride in ? iron has been determined for the temperature range 1191o to 1336°C from the nitrogen content and metallogra-phic examination of a series of Fe-Nb alloys.* (0.0 to 0.92 pct Nb) equilibrated with a purified gas mixture of 1 pct hydrogen in nitrogen. The method is the same as that used by Darken, Smith, and Filer' to determine the solubility of aluminum nitride in y iron. The variation in the furnace temperature during an equilibration period was about ± 2.0°C. Each sample was approximately 1/2 in. wide, 2 in. long, and 1/16 in. thick. A time series at 1200°C using samples of highest columbium content indicated that one day was sufficient for equilibrium; however, to be on the safe side all samples were left in the furnace for either two or three days. Afterequilibration, the samples were brought to room temperature by quenching in water. The experimental results are shown in Fig. 1. To be in accord with the values reported by Wriedt and Gonzalez2 and the temperature coefficient of our results at higher temperatures, the nitrogen content of iron at 1191oC is assumed to be slightly lower than our experimental value. Our data for the single-phase regions are not sufficiently accurate to establish an interaction coefficient for nitrogen in these alloys. However, the change in nitrogen content with change in alloy content for a number of systems (Fe-N-A1;3,1 Fe-N-Ni;372 Fe-N-Si.3949 Fe-N-V;3,5 Fe-N-Cr;3,6 Fe-N-Mo3,7) has practically the same value for the liquid alloy at a temperature near 1600°C as for the solid alloy at temperatures near 1200oC when the alloy content is below a few percent. On this basis the lines through the single-phase regions have been drawn with a slope corresponding to an increase in nitrogen of 0.0013 pct per 0.1 pct Nb.3 The straight lines through the two-phase regions were drawn with a slope corresponding to a one-to-one atom ratio of nitrogen to columbium in the nitride. The X-ray

Jan 1, 1962