Search Documents

By W. C. Ellis, E. S. Greiner, M. E. Fine

C. H. Samans and W. R. Ham (Chicago, Ill., and Dix-field, Maine, respectively)-—For several years we have been studying transitions of this basic type in metals, alloys, glasses, etc. Usually, however, they are not so clearly marked as those which the authors have found, and hence are much more difficult to determine accurately. Since our studies indicate that most of them occur virtually unchanged (as far as temperature is concerned) regardless of the form in which the element appears, we believe that they are a characteristic of the atom. Specifically, we believe that there is an additional rotational degree of freedom possessed by the nucleus which has not been considered heretofore. This nuclear rotation is made up of several components, each related to the several quantum shells of electrons. In chromium there are four of these shells and hence four separate series of characteristic transition temperatures. The lowest temperature at which any transition occurs, based on the present state of our computations, is 125°K, 4° higher than the authors' value of 121°K. A convergence of this series, we believe, shows up at the higher temperature of 2085 °K, surprisingly close to the transformation temperature of 2103°K recently announced by N. J. Grant of M.I.T. for a body-centered to face-centered transformation in chromium. Likewise, our computations indicate a temperature of 311°K for the second transition temperature, reported by the authors as 310 °K. A convergence of this series, we believe, shows up at a higher temperature as the melting point at 2163°K. Although our work on these series must, in a sense, still be regarded as empirical, since we do not understand fully as yet just what the series mean, it is based on a reasonably firm picture. The individual constants, from which the various series are computed for each element, comes directly from the X-ray K absorption limit. Furthermore, the same basic method has accounted for transformation and melting temperatures in about 50 of the chemical elements, which is all we have tried thus far. In many cases the only known transformation is the melting point, but in others the occurrence of transformations or other transitions, equally as well marked as those of the authors, has been pointed out by others. These observations have assisted us greatly. Consequently we were very pleased to see the authors' excellent work in finding these two transitions in chromium. With these confirming data, our picture of this element is clarified considerably, so we expect that at least some of our work can be published in the near future. R. C. Ruder (E. I. du Pont de Nemours & Cu., Wilmington, Del.)—The authors' interpretation of these transitions in terms of 3d to 4s electronic structure transitions is most interesting. It would be interesting to have additional experimental evidence of such transitions from the temperature dependence of the Hall coefficient in the neighborhood of the property changes discussed in this paper. Simple theory15 suggests the Hall coefficient as a measure of the free electron (or s electron) concentration per unit volume. It has been shown that for paramagnetic'" and ferromagnetic1? metals the simple theory is in fact too simple. However, the existence of a discontinuity in the Hall coefficient would provide information which should aid both in our understanding of these transitions and the significance of the Hall coefficient in these metals. It was rather surprising that no significant paramagnetic effects were observed. In this connection the recent work of McGuire and Kriessman18 is cited. They measured the magnetic susceptibility of chromium from 20" to 1460 °C. They also observed no large change in the susceptibility although there might be a change in slope in the vicinity of the 40 °C transition. The existence of these 3d to 4s electronic transitions has been discussed in connection with the paramagnetic susceptibility behavior of nickel and nickel alloys.'"-" Assuming a correspondence principle between classical and quantum mechanical paramagnetic theory and using classical theory to calculate the effective Bohr magneton number from the Curie constant for substances obeying the Curie-Weiss law," it is found that the effective magneton number is a function of temperature. The process of calculation involves the inverse of the differential of the 1/x4 vs. temperature curve so that good and numerous data are necessary to obtain significant results. The data of Fallot23 show a discontinuous increase of about 12 pct in the effective magneton number between 850" and 900 oC, followed by a continuous increase up to the melting point. The data of Sucksmith and Pearce24 show a possible 8 pct increase. The older data of Terry25 and Weiss and Foex26 show a continuous increase. It is possible that small amounts of impurity atoms change these electronic transitions significantly. Fallot's23 data on a nickel alloy with 4.5 atomic pct Fe indicate that the discontinuity occurs around 1300 °C. Systematic investigation of the transition metals for transitions of this nature should provide information which would be very valuable for our understanding of these metals. The absence of antiferromagnetic structures in chromium has been shown by Shull27 using neutron diffraction techniques. M. E. Fine, E. S. Greiner, and W. C. Ellis (authors' reply)—The remarks by Drs. Samans and Ham are certainly very interesting, in particular those pertaining to the close agreement between the theoretically calculated values for transition temperatures in chromium and the experimental values reported by a number of investigators. This is a remarkable achievement and we shall look forward to a more detailed presentation of the method followed in their calculations. We do not believe that the transition in pure chromium near 40 °C remains temperature invariant with alloying, as was reported by Samans and Ham for a number of the substances that they studied. We have not done any work with alloys but base our belief on the results of earlier studies in which less pure chromium was included and considerably lower transition temperatures were observed. The transition tempera-

Jan 1, 1952

By S. E. Bronisz, Dana L. Douglass

a Uranium was deformed by cold rolling, and the effects of this plustic deformation on the microstruc-ture of the metal were observed by the technique of transmission elecbon microscopy. The recrystalli-zation of selected cold-wmked samples was also studied by this method. The samples were prepared for examination by electrolytic thinning in a standard metallographic electrolyte which has not been previously reported as being used in producing thin foils of uranium. The microstructural changes observed in the a uranium were generally similar to those seen by optical microscopy. The changes in the microstructure of a uranium following cold work and recrystallization treatments have been the subjects of several studies employing optical and X-ray techniques.12-19 These studies indicated that twinning was the redominant mode of deformation at room temperature20 and that the regions of high energy associated with the deformation twins were preferred sites for the initiation of recrystallization.15,18, Recently several workers have prepared thin foils of uranium suitable for examination by transmission electron microscopy.'-lo The foils thus prepared have been used to study fission fragment damage; a particular dislocation reaction,' and gas precipitation." At this laboratory the microstructural changes occurring during the cold working and recrystallizing of a uranium samples have been studied by using transmission electron-microscopy techniques. The foils used in this work were electrolytically thinned in a polishing bath not reported previously in the literature as being used for this purpose. EXPERIMENTAL PROCEDURE The uranium used in this investigation was depleted a uranium sheet 4 mils thick. The results of a spectrochemical analysis of this material are shown in Table I. The as-received uranium was cut into strips, 1 in. wide by 5 in. long, and annealed in vacuo at 600°C for 2-1/2 hr. The resulting microstructure consisted of relatively uniform, equiaxed grains having diameters of 10 to 50 p. Cold Working. The cold-worked samples were prepared by rolling several of the annealed uranium strips to various thicknesses. The strips were reduced 3.5 ± 0.5, 6.0 ± 0.6, 10.0 ± 0.5, 19.5 ± 1.0, and 28.5 ± 2.0 pct in thickness. It was not possible to achieve these reductions with a single pass through the rolling mill, so the strips were turned end for end during the rolling procedure. The cold-rolled strips were stored in dry ice until samples from them could be thinned and examined (within 24 hr after rolling). Annealing. The samples used in studying the re-crystallization behavior of a uranium were obtained by annealing sections of the strips which had been reduced 10.0 and 28.5 pct. These samples were sealed in evacuated Pyrex capsules and were annealed in a tube furnace at temperatures from 200" to 600°C and for times from 1/2 to 2 hr, see Table l. Thinning: The samples from the rolling and heat treatments were prepared for examination in the electron microscope by electropolishing in a bath composed of 25 g chromic acid, 133 ml glacial acetic acid, and 25 ml water. This electrolyte has been used and recommended for preparing specimens for optical metallography It was necessary to mask the edges of the sample with a stop-off paint in order to prevent preferential attack in these areas. The paint used was made by dissolving polystyrene in benzene. The electrolytic cell consisted of a cylindrical stainless-steel cathode placed in a beaker containing the electrolyte, with the sample as the anode positioned at the cell axis. The temperature of the electrolyte was held below 15°C by cooling with liquid

Jan 1, 1963

By R. S. Goodrich, G. S. Ansell

The microstructure of three Al-Al2O3 SAP-Type alloys (containing 2.0, 3.0, and 5.7 wt pct alumina, respectively) was studied utilizing transmission electron microscopy. These alloys were fabricated from aluminum powders by powder-metallurgical techniques, cold-rolled into thin strips, and recrys -tallized. Thin foils suitable for an electron-transmission study were then prepared from the strip material by electropolishing. The microstructure of these alloys is shown to consist of small (0.1 mean diameter) plate -shaped alumina particles finely distributed in an aluminum matrix. In the higher-oxide content alloys the platelets are somewhat larger and more irregular in shape. The distribution of oxide particles is not uniform and there is an increasing tendency toward clumping SINCE the discovery by Irmann1 that aluminum alloys produced by compacting, hot pressing, and extruding fine aluminum powders possess remarkable high-temperature mechanical properties, the Al-A12O3 SAP-Type alloys have been utilized extensively as model dispersion-strengthened alloys in both experimental and theoretical invest igations.2-8 The use of the SAP-Type alloys as model dispersion-strengthened alloys implicitly assumes that their microstructure has been completely delineated. This is, however, not the case. with increasing oxide content. In all the alloys studied, the mean center-to-center spacing of particles was quite small (0.2 u) due to the very fine oxide debris present. Selected-area diffraction of the specimen identified the oxide as either y Al2O3 or y' Al2o3. The re crystallized MD-5100 alloy exhibited a mean grain diameter of only 0.7u, while the grain size of the other alloys was several hundred microns. Severe pinning of the grain boundaries by oxide particles was rarely observed, and the grain boundary triple points in the small-grained MD-5100 alloy were often particle -free. Diffraction contrasts visible around the particles are attributed to elastic strain fields set up by differential thermal contraction during cooling from the recrystallization anneal. Early investigation demonstrated that optical metallography did not possess sufficient resolution to observe the microstructure of the SAP-Type alloys.' Later, extensive use was made of replication techniques to study the microstructure by means of electron microscopy.7-9 These studies indicated that the SAP-Type alloys consist of a dispertion of irregular Al2O3 platelets, approximately 150A thick 9 and a micron or so on edge,9 distributed in a matrix of commercial-purity aluminum. These oxide platelets tend to be oriented in the aluminum matrix, with their largest dimension parallel to the extrusion direction. 8,9 A study employing polarized-light techniques1' indicated that the grain size of a series of as-extruded SAP-Type alloys was several microns in diameter, while the grain size after re-crystallization was several millimeters in diameter. A more recent study 12 utilizing surface-replication techniques examined the particle morphology,

Jan 1, 1964

By J. Nutting, J. W. Cahn

WITH the development of thin film techniques for the direct examination of metals in the electron microscope some new problems in quantitative metallography have become apparent. In order to obtain certain contrast effects in the electron optical image when examining thin films from alloys containing dispersed phases, it is essential to use films of a thickness at least two or three times greater than that of the mean particle diameter of the dispersed phase. From micrographs of such specimens it may be desired to obtain the volume fraction of the dispersed phase and the number of particles of the phase in a unit volume of the matrix. Under these conditions the particles sectioned by

Jan 1, 1960

By R. W. Bartlett

The rates of oxidation of tungsten have been determined at temperatures between 1320" and 3170°C and oxygen pressures to 1 amn using a surface -recession measurement technique. Above approximately 2000°C and 10-6 atm the rate is independent of temperature and can be calculated from gas collision theory assuming a constant reaction probability, e, of 0.06. Oxygen molecules react at surface sites where oxygen atoms have previously chemisorbed. This provides a direct pressure dependence at low pressures but at high pressures tungsten oxide molecule s form an adjacent gas boundary layer which lowers the PO2 at the tungsten surface. A correction for this effect using free-convection theory fits the rate data over the entire oxygen-pressure range from 10-8 to 1 atrn as well as data using O2-A mixtures. Below 10-6 atrn and above 2000°C, e decreases with increasing temperature because of desorption of oxygen atoms. Below 2000°C the rate decreases with decreasing temperature at all oxygen pressures following an apparent activation energy of 42 kcal per mole and depending on (Po2)n with n varying between 0.55 and 0.80. MOST of the previous tungsten oxidation studies have employed gravimetric methods and have been limited to temperatures below 1000°C where the weight loss associated with evaporation of tungsten oxides is negligible compared with the weight gain from oxidation.' At higher temperatures, oxygen-consumption rates have been determined from pressure measurements, usually at constant flow rates, by Langmuir,2 Eisinger,3 Becker, Becker, and Brandes,4 and Anderson.5 The sensitivity of this method decreases with increasing pressure and, with the exception of Langmuir's work, these investigations were confined to pressures below 10-6 atm. Above approximately 1300°C, depending on the oxygen pressure, the rate of oxide evaporation is greater than the oxide-formation rate and the recession of the tungsten surface can be measured optically without interference from an oxide layer. This was first done by Perkins and crooks6 who heated tungsten rods in air pressures from 1 to 40 torr at temperatures between 1300" and 3000°C. The present investigation of the oxidation kinetics of tungsten at high temperatures emphasizes oxygen pressures from 10-6 to 1 atm. This is the range of interest for earth atmosphere re-entry applications of tungsten for which little data were previously available. APPARATUS The apparatus is a modification of the type used by Perkins and crooks.' Ground tungsten seal rods, 6 in. long by 0.125 in. diam, were mounted vertically between two water-cooled electrodes, one fixed and the other having free vertical travel. The movable counter-weighted electrode is prevented from undergoing horizontal displacement by three sets of runners mounted at 120-deg intervals. Electrical contact is made by means of a water-cooled mercury pool. A 24-in. vacuum bell jar having a volume of approximately 267 liters was used as the reaction chamber with the sample holder mounted in the middle of the chamber. Power was supplied from an 800-amp dc variable power supply. Temperature readings were made by means of a Latronics automatic two-color recording pyrometer. With this instrument, corrections for emissivity are not necessary provided the spectral emissivi-ties at two closely spaced wavelengths are equal. Supporting measurements were made with a micro-optical pyrometer corrected for emissivity of bare tungsten and window absorptivity. The micro-optical pyrometer was calibrated against a National Bureau of Standards calibrated tungsten lamp and both pyrometers were periodically checked against the melting points of tungsten and molybdenum using the oxidation apparatus. Above 10-6 atm, pressures were measured with an Alphatron gage calibrated against a McCleod gage. At 10-6 atm, a hot-filament ionization gage was employed. A magnified image of the self-illuminated tungsten rod was formed using a 360-mm objective lens mounted outside the bell jar. When the experiment exceeded 1 hr, the image was focused on a ground-glass plate about 10 ft from the tungsten rod at about X8 and the recession of the thickness of this image was monitored with a Gaertner cathe-tometer. When faster rates were encountered, a 35-mm time-lapse cinecamera with a telephoto lens and bellows extension was substituted for the ground-glass plate and cathetometer. Diameter recession rates were determined from the photograph image projected on the screen of an analytical film reader. EXPERIMENTAL PROCEDURE After installing the rod in the apparatus and cleaning it with acetone, the system was evacuated to 5 1 x 10-5 torr. Before oxygen was introduced,

Jan 1, 1964

By J. L. Ratliff, R. I. Jaffee, H. R. Ogden, D. J. Maykuth

An experimental program was carried out to improve the low-temperature ductjlity of tungsten through the combined use of dispersed oxides for grain-size control and Groups VII and VIII metal additions for matrix softening. Separate binary -alloy studies showed that increasing amounts of finely dispersed thoria or zirconia and critical amounts of rhenium, osmium, and iridium were effective in significantly reducing the ductile-to -brittle bend transition temperature of tungsten as well as in increasing its re crystallization temper - LIKE most of the high-melting point bcc metals, tungsten undergoes a marked transition from ductile to brittle behavior with decreasing temperatures. The material factors which influence the transition temperature include grain size, metal purity, and the presence of dispersed phases. ature and decreasing its recrystallized grain size. The beneficial effects shown by each of these additions on transition temperature and grain size in binary combinations appeared to be additive in tevnary combinations. Thus, combined additions of 2 to 8 vol pct thoria with 5 pct Re, 0.87 pct Os, or 0.30 pct Ir were effective in lowering the transition temperature of tungsten pom 2000 to 75"C, increasing its re crystallization temperature from 1600 to 1800°C, and decreasing its recrystallized gain size from 1175 to 6800 grains per sq mm. The mechanism by which these factors are operable in tungsten is not yet clearly understood. Nevertheless, it has become increasingly apparent that the interrelationship of grain size and matrix composition governs to a large extent the degree of low-temperature ductility which can be achieved. Though it is not possible to accurately assess the relative importance of these variables, the available evidence suggests that, at low temperatures, grain-size effects predominate over purity effects. This was indicated in recent studies by Allen, Maykuth, and Jaffee,' involving single crystals containing 20 to 40 ppm 0 and 1 to 30 ppm C. Thus, while each

Jan 1, 1964

By J. T. Norton, Pekka Rautala

The phases and equilibria in the W-Co-C system have been studied by X-ray diffraction methods, metallographic technique, and thermal analysis. In addition to the 7 phase, two double carbides, called 8 and have been revealed. The compositions correspond to CO3 W6C2 and Co3-W10C4. The reactions leading to these phases have been explained and tentative diagrams of stable and metastable equilibria proposed. The basic reactions in sintering cobalt cemented tungsten carbides are discussed. THE W-Co-C system is of fundamental importance in practical carbide manufacturing as well as for the understanding of the sintering mechanism. Surprisingly little is known about this system, for the probable reason that all the important alloys are of two-phase structure and that the diagram Co-WC has been treated as a quasi-binary. This obviously is incorrect, because WC decomposes before melting. One ternary phase, 7, of composition Co3W3C has long been known. It was first studied by Adelskold, Sundelin, and Westgren,1 although the isomorphous iron-tungsten carbide was known earlier. There have been in the literature several reports of two 7 phases.' " Also the 7 phase has been considered unstable by Takeda4 and Westgren.1 The Co-WC diagram has been studied by Wyman and Kelley5 and a quasi-binary diagram has been published by Sandford and Trent.2 Takeda has published a tentative Co-W-C diagram, considering both metastable and stable equilibria. However, the lines of two-fold saturation are shown only partially and it seems impossible to complete the diagram without violating the phase theory. Therefore it seemed desirable to examine the system in more detail. Experimental Procedure The alloys used in the present investigation were made of powders of tungsten, tungsten monocarbide, cobalt, and carbon and were of the grade used in manufacture of commercial cemented carbides. The powders were ground and mixed in small stainless steel ball mills, using balls of the same material. Benzene was used as a dispersing agent. The mixing period was 1 hr, since this was shown to give suffi- ciently good mixing of the powders without too great a contamination from the mill. After ball milling, the specimens were pressed in cylindrical or rectangular dies. No paraffin or other lubricant was used and the small compacts had sufficient green strength to be handled without difficulty. Several sintering furnaces were employed. The most satisfactory arrangement was a vacuum furnace based on the Arsem principle which employed a graphite helix as the resistance heating element. Specimens were placed on graphite stands and there was generally a slight carburization or decarburiza-tion of the specimen surface, depending on the carbon content of the alloy. The evaporation of the cobalt at a sintering temperature of 1400°C was not significant, but became severe at 1500°C and higher. The sintering time was 1 hr at 2000°C and 2 to 4 hr at lower temperatures. It was not possible to quench the specimens, but the cooling rates were rather fast, greater than 300°C in the first minute. In the system under investigation, the reactions are sluggish, and it is believed that the high temperature structures are satisfactorily retained. The principal method of investigating the sintered specimens was X-ray diffraction by the Norelco recording spectrometer. Approximate determinations of the phase boundaries were made by the disappearing phase method. Ternary Phases To study the phase formation in W-Co-C system, a series of specimens was sintered at 1400°C. In this experiment two ternary phases, called here ? and k were formed in addition to the well-known 7 phase. The 7 phase, which has been completely described by Westgren: showed a range of homogeneity from 7 to 20 pct C and from 38 to 48 pct Co. At 1400°C the 7 phase was found to be in equilibrium with monotungsten carbide, 8, tungsten, 8, #?, and liquid. The boundaries toward #? and liquid were difficult to determine and appeared to be very temperature sensitive. The other boundaries are believed to be well fixed. The homogeneity range of 7, as measured in this work, is considerably smaller than the one

Jan 1, 1953

By J. C. Sarace, S. M. Sze, C. R. Crowell

Thin films of tungsten 077 n-type germanium, silicon, and gallium arsenide were obtained by reacting tungsten hexafluoride with the semiconductor surface in an argom atmosplrere at temperatures between 325° and 400° C. Capacity-voltage, current-iloltage, and photoelectric measurements were used to investigate the characteristics of the tungsten -semiconductor diodes thus Produced. The junctions are shown to he very close to ideal Schottky barlp/ers with barrier heights measured with respect to the Fermi energy of 0.18, 0.65, and 0.78 1.1 jar W-Ge, W-Si, and W-GaAs, respectively. The electrical properties of the W-Si interface show no deterioration when heated to 1000°C in dry forming gas for 5 min. A theoretical value of the Richardson constant, A, appropriate to the semiconductor-hand structure has been used in evaluating the current-voltage characteristics. ThE W-Si surface-barrier diode was initially proposed for investigation because the eutectic temperature with silicon (1400°C) is much higher than that in the Si-Au system 1370°C).1 This would permit more flexibility in heat treatment and possibly provide greater reliability at elevated temperatures. The lower work function of tungsten (4.54 ev)2 compared with that of gold (4.78 ev)3 also suggested that a lower barrier height would be obtained with tungsten and hence a lower forward bias and lower minority carrier injection ratio for a given current density. The investigation was extended to include the characterization of W-Ge md W-GaAs surface-barrier diodes. The tungsten films have been produced by reacting WF6 with germanium, silicon, and GaAs surfaces in an argon atmosphere at temperatures from 300° to 500°C.4 This process is a very satisfactory alternative to the relatively difficult process of evaporating tungsten films in vacuo. To ensure an adequate electrical characterization of the tungsten-semiconductor interface, three types of barrier-height measurements have been performed. The mutually consistent results obtained lead to the conclusion that the tungsten-semiconductor junctions are indeed of the Schottky type. EXPERIMENTAL PROCEDURE The apparatus used for producing tungsten films is shown schematically in Fig. 1. It consists of an argon carrier gas line to which metered amounts of tungsten hexafluoride can be rapidly added. The mixture passes through a heated reaction tube containing the semiconductor slices and is exhausted to a hood. The argon is purified by passage through a 6-in. column of titanium turnings maintained at 800°C. The tungsten hexafluoride dispensing arrangement was designed by V. C. Garbarini and W. R. Bracht.4 A measured amount of liquid tungsten hexafluoride is injected into the argon stream and vaporizes. The mixture passes through a sodium fluoride absorption cell to remove traces of hydrogen fluoride, then into the nickel reaction tube. The tube is 12 in. long with an inside diameter of 1/2 in. The center section is heated by a furnace of the self-supporting wire-filament type. It was chosen for its rapid heatup and cooling. The wall thickness is 10 mils except for a 3-in. hot zone which is 30 mils thick to reduce thermal gradients along the length. The samples to be coated are placed on a sapphire plate and centered in this section. The tungsten is deposited with the following sequential steps: the loaded reaction tube is flushed with argon at a rate of 500 cu cm per min and heated to 370°C. Then 0.45 g of liquid tungsten hexafluoride is injected into the argon stream. The samples are held at this temperature for 2 min. The tube and samples are then cooled to room temperature and the samples removed. Tungsten films were grown on (111) faces of silicon and germanium polished with Linde A abrasive and lightly etched with HF-HNO3. The GaAs surfaces were (100) faces chemically polished with a H2SO4-H2O2 solution. The films have typical sheet resistances of 0.2, 8, and 15 52/0 when grown on germanium, silicon, and GaAs, respectively. After the tungsten deposition, ohmic contacts were alloyed on the back surfaces of the wafers. Ohmic contacts to the germanium and silicon were obtained by alloying Au-Sb at 370°C, ohmic con-

Jan 1, 1965

By C. E. Morris, S. R. Pollack

Current -vo1tage -temperature characteristics were studied for Al-A12O3-Al, Au structures. The oxide film was grown by gaseous oxidation in an 0, glow discharge. The electron-transfer mechanism was identified as tunneling and the barrier heights and thickness were determined. It was found that the barrier height at the parent aluminum-oxide interface is 1.5 ev and the built-in vloltage with an aluminum counter electrode is 0.35 v. The growth mechanism of the oxide is also discussed and it is suggested that Mott&apos;s theory is not applicable in describing the growth. Discrepancies were observed in the detailed comparison of tunneling theory, based on a trapezoidal model, with the data, and several possible explanations are discussed. THEORETICAL analyses of electron tunneling from one nonsuperconducting metal to another through an extremely thin insulating film have been presented.1-6 These calculations relate the current density, J, to the voltage, V, between the two metals, and to the temperature, T. The results of these calculations indicate that the magnitude of J and the detailed shape of the J-V characteristic will in general depend most strongly upon such properties of the insulating film as the thickness, the dielectric constant, and the structure of the electron energy barrier profile. The experimental results of many7-11 workers have corroborated the main features of these calcu- lations and have in fact been used to determine properties of the thin insulating film. It has been demonstrated, for example, that the tunneling current formulation of simmons4 is adequate to describe the observed J-V characteristic over approximately 10 orders of magnitude of current. Furthermore, from these observations and analyses, insight into the nature of the A1-A12O3 (thermally grown) interface region has been obtained.&apos;&apos; The temperature dependence of the tunnel current has also been studied and, although a self-consistent set of tunneling parameters can be chosen to describe these data at low voltages (i.e., <0.5 v), the tunneling parameters are different from those obtained from higher voltage (i.e., of the order of 1 to 5 v) measurements." We have measured the effect of temperature and voltage upon the tunnel current at higher applied voltages and find that the theory is inadequate to describe all of the effects reported here. A discussion of the origin of this inadequacy is presented following a description of the samples used for these experiments, and the experimental results. SAMPLE PREPARATION Metal-insulator-metal structures were fabricated in the following way. A 99.999 pct A1 charge was used to deposit, in vacuo, thirty films, 2000Å thick, onto a 1 by 2 in. glass substrate (7059 Corning glass). Prior to deposition the substrate was glow-discharged for final cleanup and then baked at 200°C for 2 hr at a pressure of 5 x 10-7 torr. The pressure during deposition of the metal films was 2 X 10-6 torr and the deposition rate was 50 to 100Å per

Jan 1, 1965

By H. K. Birnbaum

The termination of a twin lamella in the interior of a crystal requires accommodation of the strains at the tip of the twin which result from coherency of the twin and matrix. In zinc and magnesium crystals,

Jan 1, 1960

By R. E. Reed-Hill

The shear continuity conditions under which one mechanical twin may cross another are considered. Twin intersections usually involve various types of slip deformation in addition to twinning. Because the shear of the crossing twin may be partly transmitted through the crossed twin by slip mechanisms, continuity of the twinning shear vectors is not required at a twin intersection. It is concluded that analyses based on twin intersections should not he used as an independent method for deducing twinning elements in newly observed twinning modes. WHEN one mechanical twin passes through another, a twin intersection is said to occur. In the simplest form of intersection, Fig. 1, the crossing of one twin by another is accomplished with the aid of a second-order twin induced, in the twin that is crossed, by the crossing twin. The subject of twin intersections is significant to the field of deformation studies because it has been postulated by cahn1 that twin intersections should occur only under the following continuity conditions: 1) the traces of the primary and second-order twin must be parallel to each other in the plane of the crossed twin; 2) the direction 771 and magnitude, S, of the twinning shear must be identical in the crossing and second-order twins; also, the sense of shear must be the same. If these postulates are valid they constitute a valuable tool for determining twinning elements of newly observed twinning modes in metals. In his classic paper in 1953 on the plastic deformation of a uranium, Cahn1 used these continuity conditions to deduce twinning elements for several uranium modes. However, in the same year, pratt2 showed that both of Cahn&apos;s continuity equations were violated for (1012) twin intersections in zinc. Later, Kiho and Maruyama3, 4 showed that Cahn&apos;s second postulate was violated in {301) and (101) twinning in tin and (101) twinning in bismuth. In spite of this documentation that Cahn&apos;s second postulate is often violated, several recent papers5&apos;6 have been published which have attempted to use the condition of continuity of twinning shear vectors for the evaluation of twinning elements. The primary purpose of this paper is to call attention to the possible ambiguity inherent in applications of Cahn&apos;s postulate, and to present additional information concerning the nature of twin intersections.

Jan 1, 1964

By Ursula E. Wolff

Il.lechanica1 twins have been observed in brigsten single crystals of a variety of axial orientations defornzed at room temperature in tension or bending. The twins formed near the final fracture and were up to 3 p thick. The twinning plane was always of the type (113). Twinning at room ternperature occurred only after considerable prior slip. Twins usually preceded fracture and nucleated cracks. The crack surfaces often alternated between a (001) plane and the twin interfaces. Secondary twins could also nucleate from cracks. In the absence of twinning tensile samples failed by shear; twinned samples failed by cleavage. On the assumption that observed sudden load drops in the stress-strain curves are associated with twinning, a critical resolved stress for twinning of 60, 000 to 65, 000 psi was calculated. MeCHANICAL twinning has been observed in iron and iron alloys at low temperature by a number of investigators.&apos;-7 Similarly, for high-purity tungsten, twinning has been reported for deformation temperatures below 211°K.8-10 In W-Re and Mo-Re alloys, 20 pct Re was determined as the lower limit for twin formation at room temperature." Since the twins are generally very narrow, they are usually identified only by etching techniques. The expected (112) habit plane has always been confirmed by angular analysis of the surface traces. Recently, however, a complete identification by X-ray analysis has been performed on twins in a Mo-35 pct Re alloy.12 During an investigation of the deformation charac- teristics of tungsten single crystals, twins were revealed in tensile and bend samples deformed at room temperature. This paper deals with the conditions for such twinning and the relation of twinning to fracture. EXPERIMENTAL PROCEDURE Tungsten single crystals, 1/8 in. in diam, of a variety of axial orientations were grown from un-doped material by four-pass electron-beam zone melting. A representative analysis is given in the following table: From these crqstals, tensile samples with 1 in. gage lengths were made by surface grinding, leaving shoulders for gripping. The diameter after grinding was 0.082 in. The distorted surface layer was then removed by ele1:trolytic polishing which reduced the diameter of the gage length to 0.080 in. Tensile tests were carried out at room temperature in an Instron machine under two different sets of conditions: a) Rigid mounting; strain rate 0.02 in. per in. per min. b) Gimbal mounting; strain rate 0.002 in. per in. per min up to a total strain of approximately 0.2 pct, then 0.01 in. per in. per min to fracture. Stress axis and growth direction coincided in all samples and are subsequently called orientation.

Jan 1, 1962

By J. W. Taylor, E. G. Zukas

THE results of shock-loading studies on copper were reported several years ago by smith. In his experiments, Smith found that there was a correlation between the shock direction and the orientation of twins which the shock produced. By using free-surface-type shots in which the free surface of the specimen was not normal to the shock direction, he showed conclusively that twinning of copper in impulsive loading was caused by the rarefaction wave. Similar tests with iron, however, showed that it twinned during compression. More recent studies of iron2 and on Fe-Si single crystals3 have shown that twinning can be suppressed in these materials by shock loading at pressures high enough so that the velocity of the shock wave exceeds the elastic-wave velocity. From these observations, it appeared that twinning was caused by the shear stresses present in the zone undergoing adjustment from one-dimensional to three-dimensional loading. At high pressures, such a zone becomes so thin that there is insufficient time for the relatively ordered motion of atoms involved in twinning to occur. Twinning in copper, therefore, cannot be produced by a single compressive shock wave because resolved shear stress large enough to create twins exists only for a very short time (0.01 psec or less) before the pressure increases suddenly and loading becomes essentially hydrostatic. (A typical pressure vs time record for a copper shot, illustrating the short rise time of the compression wave, is shown in Fig. 1.) If, on the other hand, a second shock wave can be created in a precompressed sample, then the new elastic limit is several kilobars,4 and a resolved shear stress sufficient to cause twinning should be maintained for an adequate time period (0.1 psec or more). This experimental situation can, in fact, be realized. The experimental arrangement used to do so is shown in Fig. 2. By utilizing the pressure transformation in Fe-V alloys, it was possible to shock-load the copper in compression while the copper was already precompressed. In the first test, an 8 wt pct V-Fe sample was used as a driver for loading the copper, and the copper was exposed to the following series of compressive shocks: 1) an elastic wave of about 8 kbars, which is usually ignored in shock studies; 2) the pressure transformation wave for the driver of 200 kbars; and 3) the driving shock wave of 270 kbars pressure. Metallurgical determination of the angular twin distribution, shown in Fig. 3, with respect to the shock direction and the shock rarefaction direction (which was 23 deg from the shock direction) showed that twinning took place

Jan 1, 1965

By Carl J. McHargue

Mechanical twins were produced in electron-beam melted columbium by high-speed impact at room temperature and by slow or fast compression at -196°C. The composition plane of the twins was { 112} and the shear direction was <111>. Notches in the twin bands often corresponded to traces of {110) of the matrix and appeared to be untwinned regions. Markings within the twin bands were interpreted as resulting from {110} slip in the twins. THERE has been much work in recent years concerning plastic deformation by glide, and the dislocation theory relating to glide has reached a relatively high degree of development. On the other hand, there have been fewer studies of deformation by mechanical twinning, and understanding of this process is far from satisfactory. This method of deformation is of interest for at least two reasons. First, it provides a mechanism in addition to glide for the relief of stresses, and, in the bcc and hexagonal close-packed metals may result in significant amounts of plastic flow. Secondly, there is the possibility that twins may act as barriers for dislocation movement, resulting in pile-ups which could nucleate cracks. As might be expected, the bulk of the literature on mechanical twinning in the bcc metals is concerned with iron. A good summary of the work done prior to 1954 is contained in the book by all.&apos; Recently the refractory bcc metals have become increasingly important. Limited studies have shown that tantalum,2,3 molybdenum,4,5 vanadium,6,7 tungsten,&apos; and columbium9-11 deform by mechanical twinning under some conditions. Alloys of molybdenum with rhenium and tungsten with rhenium show extensive deformation by twinning at room temperature.I2-l4 Most of these studies have dealt primarily with mechanical properties at low temperatures or have shown the existence of twins, and there is only a small amount of information concerning the conditions under which they form. The subject of the present paper is the formation of twins under stress in columbium with a consideration of their morphology. EXPERIMENTAL PROCEDURE The material used for these studies was taken from an ingot of columbium which had been melted twice by the electron-be am-method. The analysis of the ingot was (in ppm): B < 1, C = 10, Fe < 100, The cast ingot contained very large grains, and it was possible to obtain single-crystal prisms which measured from ¼ to 3/4 in. on a side. A few experiments were conducted on polycrystalline plate which was prepared by rolling material from the same ingot at room temperature and annealing at 1000 in a dynamic vacuum of 10-6 mm Hg. This gave a plate in which the grains had an average diameter of 3 mm. After the specimens were cut from the ingot, the six faces were metallographically polished and elec-tropolished to remove all traces of cold work. Most of the observations were made on the surfaces of the deformed specirllens without further treatment. Occasionally, etching after deformation was desirable. In these cases, an etchant of the composition 50 parts H2O, 5 parts HNO3 25 parts HF, and 10 parts H2SO4 was found to delineate the twins very well. Unless considerable care was taken to ensure the removal of all disturbed metal left by the mechanical polishing, etching failed to reveal many of the features discussed in this; paper. The specimen&apos;s were deformed either by impact or slow compression at 77°K (liquid-nitrogen coolant), 198°K (dry ice and acetone coolant), and 298°K. The impact load was delivered by a hammer except in one case where the load was delivered by a bullet. Slow compression was carried out on a hydraulic testing machine equipped with a chamber to hold the coolant. EXPERIMENTAL RESULTS It has been generally believed that the conditions favoring the formation of deformation twins are large grains, low temperature, and impact loading. In fact, Barrett and Bakish2 found twins in tantalum only after impact deformation at 77°K, and Adams, Roberts, and Smallman10 observed twins in columbium only at 20 For these reasons, the initial experiments of this study used impact loading. Hammer blows caused many bands resembling twins in single crystals a.t 77" but not at 198°K. Only a few slip lines were observed on any of the single-crystal specimens of this study—essentially all the deformation occurred by twinning. The appearance of the twins on the as-deformed surface is shown in Fig. 1. Although both Figs. 1(a) and l(b) are photomicrographs of twins taken at the same magnification and from the same crystal, they are startlingly different in appearance. Fig. 1(a) was taken from the crystal face approximately perpendicular to the shear direction, whereas Fig. 1(b) was taken from

Jan 1, 1962

By D. O. Hobson, J. O. Stiegler, C. J. McHargue

The effects of alloy composition, deformation temperature, heal treatment, ad inlerstilial contamination on the occurrence of deformation twins were studied. The twinning transition temperature varied with composition and was a maximuni of 140°C for the 40 pcl V composition. Tire remperature range of the transition was very narrow. Alloys of 20 to 70 pet V could he cold-rolled and exhibited profuse twinning. The alloys which did not twin at room temperature (10, 80, and 90 pel V) could plot be cold-rolled. Oxygen additions systematically lowered the transition temperatures for all vanadium contents. The transition temperature for a 30 pet V alloy increased with annealing temperature, apparently due to changes in dislocation density and configuration. Twinning did not nucleate fracture in any specimen in the 20 to 60 pet V composition range. THERE is great current interest in the refractory metals and alloys as structural materials. An important factor in considering the ductile-brittle behavior of such materials is the role of mechanical twinning as a deformation mechanism. Mechanical twins have been observed after deformation at low temperatures or at high loading rates in the pure metals iron,1 vanadium,2,3 columbium,4,5 tantalum,6 Chromium,7 molybdenum,8 and tungsten.9 Binary alloys in which mechanical twins have been observed include: iron with silicon,10 phosphorous," or aluminum;12 and chromium, molybdenum, or tungsten with rhenium.13-17 The increase in low-temperature ductility obtained by alloying the Group VI-A elements with rhenium is very important with regard to fabricability, and there has been considerable effort to understand the so-called &apos;(rhenium effect". Factors suggested to explain this effect include:&apos;&apos; 1) increase in solubility of interstitial impurities with alloying (electronic effect): 2) changes in the nature of grain boundary phases; 3) changes in interfacial energies of grain boundaries and stacking faults: and 4) promotion of twinning. Although an outstanding characteristic of these alloys is the profuse twinning that accompanies low-temperature deformation, most studies of the rhenium effect have been concerned with the other factors. During recent studies of refractory-metal alloys, it was noted that solid solutions of columbium and vanadium exhibited high hardness and strength at room temperature and elevated temperature and, surprisingly, also showed considerable ductility at room temperature and below. These alloys deformed by twinning at room temperature when subjected to slow loading rates. It appeared that the low-temperature ductility resulted at least in part from the relaxation of stresses by twinning in a system where slip is difficult. The reason for enhanced twinning behavior in some bee solid-solution alloys is unknown. There have been suggestions18,19 that, as in fee alloys, the

Jan 1, 1965

By E. Hornbogen

Discontinuous precipitation in a iron can occur by at least two different mechanisms. These mechanisms are compared, using observations made on an Fe-22 at. pct Zn alloy and an Fe-19.5 at. pct Mo alloy. In the Fe-Zn alloy, precipitation began on grain boundaries. These boundaries moved into the supersaturated solid solution, providing nucleation sites and an easy path for diffusion. The precipitated phase pew as lamellae, perpendicular to the PRECIPITATION reactions are called discontinuous if the nucleation of particles does not occur on individual sites in the supersaturated matrix.&apos; In discontinuous precipitation, nucleation and growth occur on a reaction front that moves into the supersaturated solution and leaves a two-phase aggregate behind. The term cellular precipitation2 has also been used, but the term autocatalytic precipi- moving reaction front. The orientation of the depleted iron differed from that of the supersaturated iron. In the Fe-Mo alloy the growing precipitate particles apparently created dislocations in the growth front which served as nucleation sites. The precipitate grew on (110) planes of the iron, and growth occurred through volume diffision. The depleted iron retained the orientation of the supersaturated solid solution. tation seems to describe the nature of such reactions best. The basic effect of such a precipitation mechanism in a solid is that it is able to move or create sites that can catalyze nucleation. Such sites can be grain boundaries and dislocations. A mechanism of autocatalytic precipitation must, therefore, include some rational process by which a grain boundary may move into the supersaturated matrix or by which dislocations may be generated in front of the reaction. Discontinuous precipitation is always in competition with individual nucleation and growth of par-

Jan 1, 1963

By Robert E. Green

Formulae are given for calculating the modes of wave propagation in a single-crystdl specimen possessing a given crystallographic orientation. Such calculations lead to determination of the orientation factor appearing in the Granato-Lucke theory for the attenuation of these waves due to dislocation damping. A detailed treatment is given for the case of an aluminum crystal possessing the orientation having a maximum Schmid fac -tor of 0.5. FOR a number of years the attenuation of ultrasonic waves propagating in metals has been used to investigate various phenomena causing internal friction.Most of the experiments were performed on polycrystalline test specimens and the waves were assumed to be propagating in an isotopic medium. However Alers 6 in 1955 and more recently Hikata et at. Chiao and Gordon,8 and Swanson and Green 10 have used measurements of attenuation of ultrasonic waves to study internal-friction changes in single crystals. These authors were concerned with the energy losses experienced by the ultrasonic waves due to dislocation motion induced by plastic deformation of the crystals. Alers measured the attenuation of 7 mc per sec ultrasonic pulses propagating in zinc single crystals before, during, and after plastic deformation. Simultaneously he measured the time-dependent plastic strain at constant stress in order to study the correlation between this strain and attenuation. The crystals were oriented in such a way that the applied stress provided a direct shear on the slip plane, thus causing the deformation to proceed by slip on this plane. At the same time the high-frequency sound pulses were sent through the crystal perpendicular to the slip plane. He found that the attenuation of transverse waves, the stress vectors of which lie in the slip plane, was very sensitive to the deformation. The polarization of the particle displacements, associated with these waves, relative to the applied static shear stress, and hence the direction of slip, had no obvious effect on the results. The attenuation of longitudinal waves was affected by the deformation only when the slip plane was not perpendicular to the propagation direction. This indicated that the attenuation was caused by dislocations introduced by the deformation and moving only in the slip plane. Hikata et al. made simultaneous measurements of attenuation, velocity changes, and stress-strain relations on aluminum single crystals deformed in tension. Both longitudinal and shear waves were used, at frequencies of 10 and 13 mc per sec. They placed emphasis on the choice of propagation modes and polarization of the ultrasonic waves, such that interactions between these waves and dislocations on selected slip systems could be investigated. They also selected specimens possessing specific crys-tallographic orientations, which allowed investigation as a function of the number of equally favored slip systems under the applied stress. They found that, for symmetrical orientations, the change of attenuation decreases with the increasing number of equally favored slip systems, i.e., in the order

Jan 1, 1964

By Robert E. Green, Robert A. Swanson

Ultrasonic attenuation and stress were measured simultaneously as a function of strain for aluminum single crystals tested in compression. The propagation mode and polarization of the ultrasonic waves were chosen with reference to the theoretical analysis presented in Part I. Good experimental agreement was found with this theory and with earlier experimental work in tensile tests. Support is given to the contention that ultrasonic -attenuation measurements are much more sensitive to the early stages of plastic deformation than are conventional stress -strain measurements. Dislocation meclzanisms are suggested which are compatible with both the present and the earlier experimental observations. A brief historical introduction of previous work concerned with the study of dislocation motion by ultrasonic-attenuation measurements has been given in Part I. The purpose of the present experimental investigation was to make use of the theoretical considerations from Part I in order to gain further knowledge concerning dislocation motions during plastic deformation. Emphasis was placed on the choice of propagation modes and polarization of the waves, such that interactions between these waves and dislocation movement on a specific slip system could be investigated. EXPERIMENTAL PROCEDURE Since the primary concern of this investigation was dislocation damping, other sources of damping were eliminated as much as possible. The test specimens were 0.5-in.-diam single crystals of 99.99+ pct pure aluminum grown from the melt by a modified Bridgman technique. Ultrasonic-attenuation measurements were made at a frequency of 10 megacycles per sec using a Style 56A001 Ultrasonic Attenuation Comparator manufactured by Sperry Products Inc. after a design by Chick, An- derson, and Truell.1 At this frequency the particle vibrations are essentially adiabatic and thermo-elastic damping should be absent. This frequency was selected since it was one of the frequencies used by Hikata et al.2 n conjunction with tensile tests on aluminum crystals, thus allowing direct comparison with this work. The use of quartz-crystal transducers insures that the amplitude of vibration is not sufficient to cause the break away of the dislocation loops from weak pinning points. The quartz-crystal transducers used all possessed a resonance frequency of 10 megacycles per sec and were thin circular discs 0.375 in. in diameter. X-cut crystals were used to generate the longitudinal waves and AC-cut crystals to generate the transverse waves. Aluminum was chosen as the basic material because it is a typical fcc metal whose deformation behavior is well-known and because the most recent work at the time the present research was initiated was that by Hikata et al. who used aluminum. The present tests were run in compression to see if such tests could be satisfactorily run in compression without being severely masked by grip effects.. A second reason was to compare the results in compression with those of the tension experiments of Hikata et at. Finally, shorter specimens could be run in compression with less over-all damping than in tension. All of the aluminum-crystal test specimens used were oriented for plastic deformation by single slip; i.e., they possessed the orientation having a maximum Schmid factor of 0.5. Because of the well-known time dependence 3 of plastic strain at constant stress, the test specimens were subjected to a quasi-static loading. Previous investigations4-&apos; have indicated that attenuation after plastic deformation is also time-dependent, so any continuous loading of the specimen would show that changes in attenuation were also a function of strain rate. By loading the specimen to a given stress level and holding it constant until the recovery process was essentially complete, the time dependence was eliminated. The quartz transducer was permanently bonded to the aluminum crystal using Eastman 910 cement. Excess cement was removed from around the quartz with DMF (dimethyl formamide). Eastman 910 was found to be far superior for bonding, especially with regard to the shear-wave transducer, since it formed a solid bond which readily supports shear-wave propagation. The solid bond formed by the Eastman 910 cement also insured that the quartz

Jan 1, 1964

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 C. S. Smith, Chih-Chung Wang

D. Turnbull—In the opinion of the writer the most interesting result described in this paper is that the distribution of tin with respect to solidification temperature has several fairly well-defined maxima. It appears that each peak may be associated with a catalytic impurity of some rather specific composition. Recently we have obtained some additional evidence in favor of this viewpoint in experiments on the supercooling of lead droplets coated with different surface films. The distribution of lead droplets, ostensibly coated with lead sulphate, with respect to solidification temperature exhibited two distinct and reproducible maxima, one at 20" supercooling and the other at 47". On the other hand, when the droplets were ostensibly coated with halide films, no solidification was perceptible until the aggregate was supercooled more than 55". Upon further cooling a single well-defined peak centering at 60" supercooling was observed. Since the particle size distribution was the same in these two sets of experiments it appears that the most reasonable interpretation of the results is that the sulphate surface film or its reaction products with lead are much more effective in catalyzing the nucleation of lead crystals than are lead halide films. C. S. Smith (authors&apos; reply)—Dr. Turnbull&apos;s experiments with lead drops coated with various surface films are most interesting. However, before drawing conclusions from the cooling curves described in the paper as to the existence of a few nucleus types effective at a few different temperatures, one must consider the state of subdivision of the liquid metal. In the A1-Sn alloy used, the tin was not in uniform droplets but tended to collect in three distinct shapes—in long interconnecting prisms along grain edges, in somewhat elongated interdendritic branching pools. and in isolated droplets. A single nucleus will solidify only the metal in contact with it, and the heat evolved will be proportional to the connected volume. A uniform distribution of nuclei of random effectiveness in such an alloy will cause peaks on the specific heat curve on cooling, which correspond to the volume distribution of tin and tell nothing about the nucleus type. In any real alloy both effects will be combined, and it seems impossible at present to distinguish between them. M. B. Bever—The thesis of this interesting paper is supported by a combination of positive and negative evidence. The authors have shown that under certain conditions a liquid phase may undercool, if it is dispersed in a thoroughly discontinuous manner throughout a solid matrix. They also found that undercooling

Jan 1, 1951