Search Documents

By D. Turnbull, R. E. Cech

EXPERIMENTAL studies of precipitation of iron from solid solution in copper by Gordon and Cohen 1, Bitter and Kaufmann 2 and Smith 3 revealed that the precipitate particles were initially paramagnetic but could be changed to a ferromagnetic

Jan 1, 1955

By D. H. Polonis, J. G. Parr

IN a recent paper on titanium-rich Ti-Fe alloys,' the hardness of quenched powder specimens was given, together with the amounts of martensitic a they contained. The values disagreed in two respects with previous determinations that Worner² made on lump specimens: 1—Some martensitic a' is formed in powder specimens of alloys containing up to 12 atomic pct Fe. On the other hand, Worner reports that, although his results on quenched lumps are not conclusive, alloys containing more than about 4 pct Fe "appear to consist of retained ß'' although they "may not be strictly unaltered solid solutions." 2—According to Worner, quenched lump specimens show a maximum hardness at a composition about 3.5 pct Fe. Earlier -experiments by the authors showed a maximum hardness of martensitic alloys in powder form at 10 pct Fe. Accordingly, more experiments have been conducted in order to confirm these differences and to determine with more accuracy the minimum composition of quenched lump specimens that consist entirely of retained ß. Lump samples?x?x1/20 in. were quenched from 1000°C in helium and, on the whole, showed similar hardness values to the specimens that Worner water-quenched from 950°C. The hardness curve the authors obtained shows a more pronounced peak, but in other respects it is very similar to Worner's curve (Fig. 1). While it was possible to make fairly reliable measurements of the percentage of martensitic a in quenched powders (by Geiger spectrometer), the solid samples allowed only approximate estimates to be made from visual observation of line intensities on film. Retained ? begins to appear in quenched lumps containing 0.2 pct Fe; at 2.2 pct Fe there are approximately equal quantities of ß and a'; in a 4.7 pct Fe alloy only a trace of a' appears. These results are represented as a curve on Fig. 1. The hardness and constitution of powder samples were verified by repeating both types of quenching experiment described in the earlier paper.' Identical results were obtained by the two methods, and they agreed with the figures published previously (Fig. 1). The appearance of the martensitic structure under the microscope is shown in Figs. 2 and 3. The mar-tensite needles are coarser in the lump; otherwise the structures are similar. Three conclusions are made at this juncture: 1— The properties of quenched lumps reported by Worner are confirmed, on the whole. The differences between them and the properties of powders are probably due to an effect of specimen size. Differences of behavior in various sizes of specimen have been observed previously in martensite reactions in Fe-Mn alloys3 and in Cu-A1 alloys. 2—The maximum hardness of both quenched lumps and quenched powders appears at a composition where martensite formation has all but ceased. In this respect, powders and lumps behave similarly. 3—Martensitic a' has essentially the same microscopic appearance both in powders and lumps. Acknowledgment This work is part of a program sponsored by the Defegce Research Board of Canada, Project No. 425.

Jan 1, 1956

By Jack L. Henry, R. D. Holliday

GASEOUS products of an aluminum reduction cell consist mainly of carbon dioxide and carbon monoxide. Sulfur dioxide, hydrogen fluoride, and hydrogen sulfide have also been conc~usively identified. Carbon tetrafluoride has been observed only during the period when the anode is polarized.' This period, known as anode effect, occurs when the alumina content of the molten bath approaches depletion.= Increased resistance at the polarized anode results in a voltage rise from a normal 5 volts to as

Jan 1, 1958

By G. D. Hobson, I. Mrosovsky

Material balance relationships for expansion-type reservoirs above the bubble point have been discussed in recent years by H. N. Hall and M. F. Hawkins, Jr. The former drew attention to the effects of rock compressibility, and the latter to the effects of the compressibility of any interstitial water. Hawkins derived a general relationship which took note of oil, water and rock compressibilities, and then by assuming first, the absence of interstitial water and secondly, no effects due to interstitial water or rock compressibility, obtained relationships symbolically of the same form as those previously derived directly by Hall. Hawkins defined his terms clearly, and noted an approximation, due to the omission of a term involving the product of two compressibilities, made in reaching his final equation. There is, however, possible ambiguity with regard to Hall's written definition of the oil production term. One interpretation would mean that the produced oil and initial oil-in-place, if both required as stock tank barrels, would be obtained by using different formation volume factors in the conversion in order to satisfy the conditions implicit in Hall's formulae. The other would involve an approximation which, if used, was not stated. In both Hall's and Hawkins' studies the effect of the approximation or inconsistency is very slight. Nevertheless, it may be justifiable to give a somewhat more consistent derivation, since the complexity of the final relationships is not thereby greatly increased. Let the initial pore volume be Vi and the volume after a pressure drop of Ap, V. Then V = V, (1 - ?p . cr), where c, is the rock compressibility expressed as change in pore volume/unit pore volume/unit pressure change, and is defined with the initial pressure as base. Suppose that the interstitial water saturation under the initial conditions is Sw. The total volume of interstitial water under those conditions will be ViSw. If the volume of the water after dropping the pressure by ?p is V., this will be given by Vw = ViSw (1 + ?p.c,w), the compressibility of the water being c, which is ex- pressed using the volume at the initial pressure as base. If Boi is the reservoir volume factor of the crude oil for the initial reservoir conditions, and B. the equivalent factor when the pressure has dropped by ?p, with N the initial amount of oil expressed as stock tank barrels. N.Boi = Vi - V,Sw If n is the amount of oil produced (in terms of stock tank barrels) as a result of the pressure drop ?p, the remaining subsurface volume of oil at the lower pressure will be Bw, and Bc can be related to the compressibi:ity of oil and the pressure drop, and if c, is the compressibility factor based on the initial conditions, The relationship derived by Hawkins was case to a standard at the lower pressure, while a term (Ap) had been dropped, because of its smallness arising from its dependence on the products of two compressibility factors. When there is no interstitial water and rock compressibility can be ignored, Eq. 1 reduces to:

Jan 1, 1957

By Murray F. Hawkins

One problem of reservoir engineering is the early estimation of the size of newly discovered reservoirs. Often these reservoirs are the expansion type in which sizeable pressure drops occur incident to the production of small fractions of the oil reserve, above the bubble point, and a material balance equation may be solved to find the reserve. The early form of this equation' considered only the compressibility of the oil. Later Hall2 showed that the compressible nature of the formation rock is a significant factor in production above the bubble point, and he presented a new equation which includes a formation (rock) compressibility term. This note presents a further extension of the material balance equation as applied to volumetric, unsat-urated reservoirs above the bubble point by including a term for the presence and compressible nature of the interstitial water. DERIVATION OF EQUATION During a decline in average reservoir pressure, ?p ?p, — p1 the initial reservoir volume V,, will decrease to some value V, due to the reduction in pore volume as a result of a decline in the hydrostatic pressure within the formation, and v, = V,, [l — cr ?p]......(1) The formation compressibility c, is expressed in terms of change in pore volume per unit pore volume per psi. Simultaneously the connate water within the formation expands from an original volume V,, = S,V,, to a new volume V,, V, = s,v,, [I + c,npl.....(2) The difference [Vp — V,] is the hydrocarbon pore volume at pressure p and may be equated to the volume of the remaining oil, or [N — N,] B, = V,,, [I - C~A;] - S,.Vpi [l + cwnjl.........(3) But the initial oil in place in terms of the initial reservoir volume, connate water, and formation volume factor is Ilividing Equation 3 by Equation 4 eliminates V,, to give,

Jan 1, 1956

By F. E. Werner, B. L. Averbach, Morris Cohen

THAT bainite formed near the M, temperature bears a striking r esemblance to martensite tempered at the same temperature has been shown by the electron microscope.' By means of electron diffraction,' it has been established that carbide and cementite are present in bainite formed at 500°F (260°C); these carbides are also found in martensite tempered at 500°F (260°C).' The investigation reported here is concerned with an X-ray study of the matrix phases in lower bainite and tempered martensite. These phases have turned out to be dissimilar in structure; the matrix of bainite is body-centered-cubic while that of tempered martensite is body-centered-tetragonal. A vacuum-melted Fe-C alloy containing 1.43 pct C was studied. Specimens of 16 in. diam were sealed in evacuated silica tubing and austenitized at 2300°F (1260°C) for 24 hr. One specimen was quenched into a salt bath at 410°+7 °F (210°+4°C), held for 16 hr, and cooled to room temperature. The structure consisted of about 90 to 95 pct bainite, the re: mainder being martensite and retained austenite. A second specimen was quenched from the austen-itizing temperature into iced brine and then into liquid nitrogen. It consisted of about 90 pct martensite and 10 pct retained austenite. The latter specimen was tempered for 10 hr at 410°+2°F (210°+1°C). The specimens were then fractured along prior austenite grain boundaries (grain size about 2 mm diam) by light tapping with a hammer. Single aus-tenite grains, mostly transformed, were etched to about 0.5 mm diam and mounted in a Unicam single crystal goniometer, which allowed both rotation and oscillation of the sample. Lattice parameters were measured by the technique of Kurdjumov and Lyssak. This method takes advantage of the fact that martensite and lower bainite are related to austenite by the Kurdjumov-sachs orientation relationships Thus, the (002) and the (200) (020) reflections can be recorded separately, permitting the c and a parameters to be determined without interference from overlapping reflections. According to these findings, the matrix phase in bainite is body-centered-cubic and, within experimental error, has the same lattice parameter as ferrite (2.866A). On the other hand, martensite, tempered as above, retains some tetragonality, with a c/a ratio of 1.005t0.002. Most workers in the past have assumed that bainite is generated from austenite as a supersaturated phase, but the nature of this product has not been established. The question arises as to whether bainite initially has a tetragonal structure and then tempers to cubic, or if it forms directly as a cubic structure. If it forms with a tetragonal lattice, it might well be expected to temper to the cubic phase at about the same rate as tetragonal martensite. The martensitic specimen used here was given approximately the same tempering exposure, 10 hr at 410°F, as suffered by the greater part of the bainite during the isothermal transformation. About 50 pct bainite was formed in 6 hr at 410°F. On tempering at this temperature, martensite reduces its tetragonality within a few minutes to a value corresponding to 0.30 pct C.' Further decomposition proceeds slowly, and after 10 hr the c/a ratio is still appreciable, i.e., 1.005. Thus, even if the bainite were to form as a tetragonal phase with a tetragonality corresponding to only 0.30 pct C, which might be assumed to coexist with e carbide, it would not be expected to become cubic in this time. It seems very likely, therefore, that bainite forms irom austenite as a body-centered-cubic phase and does not pass through a tetragonal transition. The carbon content of the cubic phase has not been determined, but it could easily be as high as 0.1 pct, within the experimental uncertainty of the lattice-parameter measurements. It has been postulated that retained austenite decomposes on tempering into the same product as martensite tempered at the same temperature. There is now considerable doubt on this point. The isothermal transformation product of both primary and retained austenite at the temperature in question here is bainite," and the present findings show that bainite and tempered martensite do not have the same matrix. Acknowledgments The authors would like to acknowledge the financial support of the Instrumentation Laboratory, Massachusetts Institute of Technology, and the United States Air Force.

Jan 1, 1957

By J. C. Fisher

C. S. Smith has described a method for measuring the relative values of different solid-solid and solid-liquid interfacial tensi0ns. When a polyphase alloy is held at an elevated temperature, individual particles of each phase adjust their shapes and minimize the interfacial energy. Wherever three interfaces meet, the dihedral angles are such that the three interfacial tensions, considered as forces, are in equilibrium. When the angles between three forces in equilibrium at a point are known, it is possible to determine any two in terms of the third. In this way ratios of interfacial tensions have been .obtained. The value of a given interfacial tension ratio was found to be independent of the alloy system from which it was determined. Smith also has employed the method when one of the phases is liquid. When equilibrium is reached with respect to In the course of work on grain growth in high purity aluminum a simple optical method was developed for detecting the presence or absence of preferred orientation in annealed specimens. The specimens were prepared by grinding, polishing and "deep etching." The etching reagent used for high purity aluminum has been described.* The resulting etch pits produced selective

Jan 1, 1949

By H. C. Gatos, Ahmed Azzam

A UTORADIOGRAPHY would appear to be sim-A pler and less time-consuming for diffusion studies than any of the radioactive tracer techniques commonly employed, namely, the sectioning, surface increase, and surface decrease techniques.' Accord- ingly, it was felt worthwhile to investigate the applicability of autoradiography to self-diffusion measurements, and gold was chosen for this purpose. The radioactive isotope Au'!'' was used, as obtained from the Brookhaven National Laboratory. Its radiation consists of p (0.97 mev) and y (0.41 mev) rays, and it has a half-life of 2.69 days. No-Screen X-ray film was found suitable for these experiments: the film has the desired high speed for recording P rays and yet is practically insensitive to y rays. On the other hand, due to the high atomic weight of gold, this metal is a strong absorber for p rays. The

Jan 1, 1953

By G. K. Dumbauld, B. E. Morgan

A satisfactory well-cementing composition must retain its fluidity long enough to be pumped into Place; then it must develop within a reasonable length of time sufficient strength and impermeability to anchor the casing in position and to form an effective seal between the casing and the formations. These properties may be evaluated by pumpability, strength (compressive or tensile), and permeability tests. Laboratory test procedures for the determination of pumpability time and strength development have been outlined previously and are readily available,1 but a satisfactory and convenient laboratory procedure for the measurement of the permeability of set cement has not been developed heretofore. some permeability data have been obtained previously by passing air through dried cement specimens.2 such a procedure is unsatisfactory because the drying of set cement alters it physical and chemical properties and subjects it to conditions dissimilar to well conditions. A more realistic method would be one wherein the cement remains water-saturated durin

Jan 1, 1952

By Rudolph G. Wuerker

THE scarcity of values of tensile strength of rocks has been explained by the lack of successful testing procedures. In the case of mine rock a description is given' of the difficulties encountered in testing a cylindrical specimen, such as a core, by conventional methods. Over a period of years the following method has given definite and reproducible results with the weakest as well as with the strongest rocks. It does not completely supersede the use of cores with special fixtures but is a supplement in all cases where cores cannot be obtained, as from soft rocks, or in cases where it is less expensive to prepare test specimens by cutting them out of the rock instead of drilling cores. Principle and equipment are the same as for the test for tensile strength of hydraulic-cement mortar.' The test specimen, Fig. 1, has the shape of a briquet. While in the original cement mortar test the briquet is cast in a special mold, it is prepared from rocks in different ways, depending on how easily they can be cut and shaped. Soft rocks, which cannot be core-drilled with a carboloy or diamond bit, are simply hand-cut. Only two dimensions need be watched. The first is the 1-in. diam at the narrowest cross-section of the briquet. The other critical measure is the radius of curvature of the waistline, as the roller supports in the grips have a fixed distance. This radius is ground out of the solid by means of a carborundum grinding wheel having a 3/4-in. radius. Medium hard rocks can be core-drilled with a carboloy bit. The resulting core can be used for nondestructive sonic testing first, and after that for any destructive test. By using an EX-bit and by carefully placing the coreholes, preferably by using a tenplate such as shown in Fig. 1, it is possible to obtain from the rock a punched sample from which numerous tensile briquets can be made. The outside radius of the EX-bit differs from the radius of curvature of the briquet by 1/8 in., but this still permits placing and aligning the specimen in the grips. In the case of bedded rocks the core might have bedding planes normal to the plane of the briquets, and rocks can be tested in any arrangement of the bedding planes desired. Hard rocks, limestones, igneous, and metamor-whic rocks can only be diamond-drilled or diamond-cut. Here the method of getting the tension briquets by accurate placing of EX-drill holes is especially economical. The tops of the briquets made from hard rocks cannot be rounded; they are straight cuts made with a diamond cut-off saw and rounded off on a polishing wheel. Results: As long as specimens broke over the waistline the results were considered acceptable. Further statistical treatment of the tests' showed a satisfactory percentage of standard deviation. The tensile strength values obtained by this method do not represent true values because of the stress concentration caused by the curvature of the side of the piece and because of the closeness of the grips. The ratio of maximum to average stress at the plane of failure has been determined to be about 1.75." All tensile strength values listed in Table I are corrected accordingly. To avoid this stress-concentration, if there are a sufficient number of cores, tensile strength can be measured by imbedding the cores in mortar in the two outer briquets in the gangmold.4 Strain-Measurements: The applicability of the briquet specimens for strain observations was tested in the case of sandstone and shale. Two element Rosette SR-4 strain gages were used. Young's modulus and Poisson's ratio, both in tension, were computed and found to be different from those in compression, determined during the same test series and from the same rock, see Table I. References 1 L. Obert, S. L. Windes. and W. I. Duvall: Standardized Tests for Determining the Physical Properties of Mine Rock. U. S. Bur. Mines R.I. 3891 (1946). 2 Test for Tensile Strength of Hydraulic-Cement Mortar, ASTM Standard C 190-44. S F. O. Auderegg, R. Weller, and B. Fried: Tension Specimen Shape and Apparent Strength. Proc. ASTM 11939) 39, pp. 1261-1269. 4 API Code 32.

Jan 1, 1956

By Nicholas J. Grant, H. C. Chang

EMPHASIS on the importance of grain boundary sliding as a mode of deformation at elevated temperatures has been presented elsewhere.' The extent to which boundary sliding occurs under certain creep conditions has been determined for high purity aluminum2,3 and aluminum alloys.' In general, boundary deformation extends quite far into the adjoining grains, resulting in a rather wide band of deformation along the grain boundaries. This has led some investigators to doubt that grain boundary sliding consists of shearing along the boundary surface; it has been contended instead that boundary sliding is a result of concentrated slip in the neighborhood of the deformed grain boundaries."' The purpose of this note is to present evidence that grain boundary sliding takes place by a bulk shear process. gh purity polycrystalline aluminum, wherever and whenever there is grain boundary sliding, boundary migration and concentrated deformation along both sides of the deformed boundary are associated with it. Therefore, it has been rather difficult to obtain unambiguous evidence of boundary sliding. On the other hand, it would be expected in certain alloys of aluminum, when tested under particular conditions. that the severity of the interfering effects of boundary migration and concentrated deformation along the deformed boundary might be reduced. Current research in this laboratory' shows that a 20 pct Zn-A1 solid solution alloy fractures in an intercrystalline manner during creep at 500°F with a small total elongation of not more than about 14 pct, and that it does so with extremely limited grain boundary migration. Accordingly, this alloy was chosen for the present study. The experimental procedure was similar to that used in previous work and has been presented elsewhere.' Two parallel flat surfaces were milled from a round specimen. The gage portion of the specimen was 1 in. x .09 in. x .17 in. The specimens, which were annealed at 1030°F for 24 hr, showed two to four grains across the width of the test bar. The specimens were subjected to creep at 500°F at stresses of 2000 to 5000 psi. The rupture times were from 10 sec to 3 hr. Appreciable boundary sliding occurred in all the tests. A typical example of boundary sliding is shown in Fig. 1. (The tension axis is vertical in the micrographs.) Sliding has taken place along the boundary between grains A and B unaccompanied by important boundary migration. The sharpness of the offset at the triple point can be clearly seen in grain C. As a result of sliding along the boundary between grains A and B, a new boundary surface between grains A and C was created. The offset at the triple point is about 0.07 mm. This boundary sliding is accompanied by the formation of rather sharp kinking bands and, of course, slip on several slip systems in grain C. This type of deformation has been referred to as fold formation' in previous work with high purity aluminum.

Jan 1, 1957

By J. A. Morgan

IN the art of roasting sulfides it is well known that the lower the temperature and the higher the pressure of SO, the larger will be the amount of sulfate present in the product. However, the mechanism of sulfate formation is not generally understood. The following experiment" demonstrates in a direct way that, in the roasting of Cu2S, the sulfate forms from the reaction of Cu2O, formed by the oxidation of Cu2s, with gaseous SO, (or SO3) and 0,. Thus, the sulfate does not form at the sulfide surface. The reaction at the sulfide surface is the same whether or not sulfate is formed. Using an indirect method, Ong, Fassell, and Wadsworth' infer that sulfate forms after oxidation, during the roasting of sphalerite, agreeing with the observations of this paper. The method presented here of demonstrating that sulfate forms after formation of Cu2O is to roast a sample of Cu2S and then to locate and identify the phases formed.' A solid cube of cuprous sulfide, made by melting Fisher reagent grade Cu2S powder and measuring about 1.5 cm on edge, was placed on a Nichrome wire grid suspended over the top of an Alundum combustion boat. The assembly was placed in a tube furnace at 600°C and oxygen gas, flowing at 0.60 liters per min, was passed around the specimen for 24 hr. The furnace tube diameter was 1.5 in. The partially roasted sample was removed from the furnace and impregnated with resin No. BR 0014, obtained from the Palmer Products Co., and the impregnated sample was then mounted in Lucite. A cross section was obtained by grinding and the surface was prepared by the following steps: 1) grinding—60 grit, 3M disk, dry; 2) preliminary polishing —60 through 600 grit, 3M disk, dry; 3) polishing— 8 µ diamond putty, silk cloth; and 4) final polishing —1evigated magnesia in isopropyl alcohol, Selvyt cloth. The color micrograph of the polished surface is shown in Fig. 1. The layers present in the micrograph are given along the side. The comwositions of the various layers were established by Debye-Scherrer patterns. From this evidence the following sequence of reactions which occur in the roasting of Cu2S is inferred: 1) The oxidation of Cu2S occurs with oxygen which has been transported through the oxide-sulfate layer by either gaseous or solid state diffu- sion. The products of this reaction at the Cu2S surface are Cu2O and SO,. 2) The SO, diffuses out through the pores in the Cu2O and finally reaches a position in the Cu2O where the oxygen pressure and the SO, pressure have values such that the thermodynamic conditions for the formation of CuSO4 are fulfilled. Then, the mixture of SO2, SO3, and oxygen reacts with the Cu2O to form CuSO1. 3) The CuSO, decomposes by the evolution of SO,, or SO, and O2, to form the basic sulfate. This is reasonable, since the SO, gradient will decrease toward the oxygen gas phase from the CuSO4 layer. 4) For this same reason the basic sulfate will decompose to form CuO. The above sequence of events is based on the assumption that diffusion through the oxide and sulfate layers, rather than the reaction at the sulfide-oxide interface, is rate-determining. This assumption is supported by the observation that these oxide and sulfate layers combined are almost nonperme-able to the flow of gases through them. Acknowledgments This research was supported by the U. S. Air Force, through the Office of Scientific Research of the Air Research and Development Command. The authors wish to acknowledge the helpful suggestions of Ralph Wells of the Applied Research and Developmeit Laboratory, U. S. Steel Corp. References E. A. Peretti: A New Method for Studying the Mechanism of Roasting Reactions. Discussions Faraday Soc. (1948) 4, p. 174. - J. N. Ong, Jr., M. E. Wadsworth, and W. M. Fassell. Jr.: Kinetic Study of the Oxidation of Sphalerite. Trans. AIME (1956) '206, pp. 257-263; Journal OF Metals (February 1956).

Jan 1, 1957

By H. E. Stauss, J. Hino

INTEREST in silicon has arisen again in the past decade as a result of improvements in crystal rectifiers.' Although the preparation of silicon was first reported by Berzelius in 1880, the early product was of relatively low purity, and only the need for rectifiers in World War II led to the production of a 99.9+ pct pure powder. This material in crystalline form was consolidated into massive silicon for use, and the method developed was to melt it with selected added constituents as "doping" agents. Melting techniques, therefore, are of great importance. There are two basic problems in producing silicon ingots free of doping additions; one is the prevention of spitting and the other is prevention of cracking of the ingot during freezing. The most satisfactory arrangement yet developed for producing massive silicon is to melt and freeze in a cylindrical quartz crucible surrounded by a concentric heating element and concentric radiation shields or insulation. For example, use can be made of a tubular heater with a high frequency generator as the source of power and reflecting shields of alundum cylinders. The spitting of silicon is related to gas evolution, and the gas comes from two primary causes—adsorbed gas and the reaction products of silicon and the crucible. Gas is also released from bubbles contained in the quartz crucible walls. Improved removal of adsorbed gas can be achieved by means of controlled melting and freezing. The seriousness of the problem in vacuo is reduced with an electrically operated mechanical movement of the high frequency power coil. The upper portion of the powder charge is melted first and the high frequency coil lowered until the powder is completely molten. During cooling the high frequency coil is raised slowly. These means also reduce the final nonviolent extrusion of large beads of metal through the ingot top during freezing. Better control of spitting and bead extrusion is obtained when melting is done under helium at. atmospheric pressure instead of in vacuo. The problem of reaction between silicon charge and crucible in practice is confined to the reaction between silicon and quartz. This2 apparently is: Si + SiO2 + 2SiO The part that this reaction plays in spitting has not been isolated for separate study. SiO is a volatile vapor at the melting point; of silicon and is released freely during melting in vacuo, but hardly at all in helium at atmospheric pressure. The cracking of ingots is a major difficulty in melting silicon, and its prevention requires special melting techniques or the addition of "toughening" agents such as aluminum or beryllium.' The cracking of the ingots has been explained as being the result of the expansion that occurs upon freezing; although direct observation of freezing ingots reveals visible cracks on the surface only after a red heat has been reached, suggesting that cracking is the result of differential contraction of silicon and quartz. Silicon wets quartz, and the ingot adheres tightly to the crucible. Therefore as ingot and crucible cool, the two either have to pull apart, or at least one must crack. Surprisingly, in spite of the relative thinness of the quartz and the thickness of the ingot, the ingot and the crucible both crack. Microscopic and X-ray4 studies fail to show any plastic flow other than twinning in the ingots. Slow cooling fails to prevent cracking. Another possible solution to cracking is to weaken the crucible. Use of thin-walled crucibles finally led to success with fused quartz crucibles with a wall thickness of 0.25 to 0.50 mm. With such thin-walled fused quartz crucibles consistently uniform success is secured in producing sound ingots 30 mm in diam from the purest available grade of silicon (99.9+) without the use of any type of addition. Melts are made in the size range of 50 to 100 g. Omission of a deliberately added doping agent is not sufficient to insure pure ingots. The reaction of silicon with crucibles and the resultant solution of impurities in the silicon is well-established." In this laboratory, the presence of Al, Be, and Zr has been found spectroscopically in ingots melted in contact with alumina, beryllia, and zircon. The best crucible materials reported in the literature are MgO and SiO2. Use of MgO in this laboratory has resulted in a heavy deposit of magnesium on the furnace walls, showing that a reduction of the magnesia occurred and the resulting magnesium removed from the melt by volatilization. In the case of quartz, the silica is reduced and SiO liberated to deposit on the equipment walls. There probably is real danger that oxygen is dissolved in the ingot when either magnesia or silica is used as the crucible material. Preliminary analyses by Dean Walter in his vacuum unit in this laboratory6 indicate the presence of oxygen in undoped silicon melted in quartz.

Jan 1, 1953

By J. W. Putman, N. J. Grant, D. S. Bloom

SEVERAL recent determinations of the melting S point of pure chromium have been reported which give values of 1845°C1; 1895°C,² 1930°C,³ 1860°C,' and 1890°C.5 because of this wide spread of values, it appeared desirable to make one additional attempt to obtain a more accurate, reliable figure. bloom and Grantx recently reported a phase transformation in chromium:.--- it was desired to check this, too, in the course of checking the melting point. To accomplish this, a much higher purity chromium was required. Electrolytic chromium containing about 0.5 pct 0 was crushed to a finer size and was then annealed in highly purified, dried hydrogen at 1375°C for 100 hr. This treatment produced chromium with about 0.008 pct 0, 0.002 pct N, and negligible carbon. The balance of the impurities was 0.3 pct Fe, 0.03 pct Si, 0.004 pct S, and less than 0.001 pct Mo. This chromium was melted in a stabilized zirconia crucible under pre-purified argon, using induction heating. Temperature measurements were made with annealed wolfram-molybdenum thermocouples and a Leeds and Northrup Speedomax Recording Potentiometer. The thermocouples were annealed for 1 min in hydrogen at a temperature of about 2400°C. Each leg of the thermocouple was immersed in an ice-water mixture using mechanical connectors to the lead wires. The thermocouple tip was immersed in the molten chromium in a zirconia protection tube which introduces an error of about 3°C as a temperature drop through the walls of the tube. The weight of chromium was about 183 g. The dimensions of the melt in the crucible were about 1.25 in. diam x 2 in. high. The tip of the thermocouple was held about ½ in. from the bottom of the crucible. Two new thermocouples were used with the above set-up giving the results shown in Table I. Since this chromium is purer than that used by Greenaway, Johnstone and McQuillan,' by Carlile et al.,4 and other investigators,²,3 it is believed to be the more accurate. Thermocouple aging effects which are believed responsible for the high value of 1930°C³ were avoided in this work. This value of 1903°C is correct to about ±10°C. The right hand column lists the temperature of the a ? ß transformation on heating and the ß ? a on cooling.3 A sample curve is shown in Fig. 1. The value so determined is 1840°C + 15°C. Fig. 1 gives adequate evidence that there is a transformation in chromium at high temperatures.' References 'H. T. Greenaway, S. T. M. Johnstone, and M. K. McQuillan: High Temperature Thermal Analysis USing the Tungsten-Molybdenum Thermocouple. Journal Inst. Metals (November 1951) 19, p. 109. ²J. W. Putman, R. D. Potter, and N. J. Grant: The Ternary System Chromium-Molybdenum-Iron. Trans. A.S.M. (1951) 43, p. 824. ³D. S. Bloom and N. J. Grant: Chromium-Nickel Phase Diagram. Trans. AIME (1951) 191, p. 1009; Journal of Metals (November 1951). ' S. J. Carlile, J. W. Christian, and W. Hume-Rothery: The Equilibrium Diagram of the System Chromium-Manganese. Journal Inst. Metals (1949) 76, p. 169. 5G. Grube and R. Knabe: Ztsch. Elektrochemie (1936) 42, p. 793.

Jan 1, 1953

By P. Breidt, E. S. Greiner

THIS note reports a determination of the melting point of germanium, and the liquidus of the Ge-Ga alloys; lattice constants of the a phase are given. The alloys were prepared by melting germanium (>20 ohm-cm resistivity) and gallium (99.6 pct) in evacuated silica tubes. The materials (50 to 70 g) were heated at 1000°C, held in the liquid state for about ¾ hr, and then cooled rapidly in the tubes. Points on the liquidus were obtained by thermal analysis which consisted of obtaining time-temperature data during heating and cooling the materials in high purity graphite crucibles and a helium at- mosphere. Temperatures were determined with a Pt—Pt-Rh thermocouple protected by a silica sheath. The couple was calibrated against a National Bureau of Standards calibrated couple and also against the freezing points of copper, silver, aluminum, and the 28.1 wt pct Cu, 71.9 wt pct Ag alloy eutectic. The samples for X-ray analysis were prepared by heating small pieces of the ingots in evacuated glass or quartz tubes at temperatures between 600" and 900°C for periods ranging from 2 hr to 22 days, periods deemed sufficient for equilibrium to be attained, followed by quenching in water. The heat-treated samples were ground to powder for X-ray analysis. The lattice constants of the germanium-rich phase (a) were determined with a baok-reflection focusing camera. The X-ray reflections (331 and 422 Cr Ka, 422 and 333 Cr KB) were corrected

Jan 1, 1956

By L. D. Jaffe

IN previous metallographic work on titanium and its alloys, difficulty has been encountered in distinguishing spheroidal particles of titanium hydride, dispersed in a-titanium, from other phases that may be present, such as p-titanium. This problem is common in examination of commercial unalloyed titanium.' In the course of an investigation of the Ti-H phase diagram, it was noted that titanium hydride, dispersed in a matrix of a-titanium, showed strong optical anisotropy when examined under polarized light. With sensitive tint illumination, particles of the hydride, when the stage is rotated, changed from bright yellow to bright blue in color. Since P-titanium is cubic, it showed no color change on rotation, remaining dark gray or blue-gray. Although a-titanium is hexagonal, it showed only a minor color change, going from pale pinkish-blue to pale purplish-blue. These colors could be modified somewhat by adjustment of the illumination. They were observed in specimens mechanically polished, both without etching and after etching with hydrofluoric acid-nitric acid mixtures in either water or glycerine. The strong color change of the hydride seems to offer a simple method for its metallographic identification. Optical anisotropy of titanium hydride is inconsistent with its accepted face-centered-cubic structure.' Samples of iodide titanium containing 6 to 40 atomic pct deuterium were prepared for diffraction studies. After deuterating and slow cooling' the samples were held 21 % hr at 400 °C, water quenched, held 61 hr at 100°C and quenched, then held 64% hr at 255 C and quenched. Debye ex-posures with MoKa X-rays and 1.111A neutrons were made at room temperature of samples as-deuterated and after the 100" and 255°C treatments. Table I shows the lines observed, in addition to those attributable to a-titanium. These data show that the hydride phase was not cubic, but probably tetragonal. The titanium atoms appeared to form a body-centered-tetragonal lattice complex, two titanium atoms per cell, with approximate dimensions: a = 3.12A, c = 4.18A, and c/a =1.34A. This is equivalent to a face-centered-tetragonal complex, four titanium atoms per cell, with approximate dimensions: a = 4.42A, c = 4.18A, and c/a - 0.946A. The presence of the 2.98 and 1.81A neutron lines indicates that the true Bravais lattice is not identical with this lattice complex, but may be primitive tetragonal with dimensions some multiple of those of the complex. The relative intensities of the hydride and a-titanium X-ray lines as functions of overall composition indicated that the hydride, presumably in equilibrium with the a at 100" to 255"C, contained more than 62 atomic pct H. Acknowledgments The experimental work covered in this report was done at Brookhaven National Laboratory, through the courtesy of D. Gurinsky. The cooperation of J. Sadofsky, 0. Kammerer, A. McReynolds, and A. Cendrowski is gratefully acknowledged.

Jan 1, 1957

By S. C. Carapella, E. A. Peretti

DURING a literature survey of the indium-zinc phase diagram, controversial reports on the composition of the eutectic point were encountered. The value reported in the investigation of Wilson and Perettil and in a later investigation by Valen-tiner' was 4.0 wt pct zinc. The most recent work recorded on this system consisted of an exhaustive survey of alloys by Rhines and Grobe8 in which they indicate a value of 2.8 wt pct zinc for the eutectic composition. To clarify this aspect of the diagram, a series of alloys were prepared by melting the respective constituents in a glass test tube under a covering of mineral oil and slowly cooling the melt in a furnace. The zinc used was Baker's C.P. and had the following analysis: insoluble in H2SO4—0.0l pct; pb—0.001 pct; Fe-—0.001 pct. The indium, which was supplied by the Indium Corporation of America, had a purity of 99.9'7 pet with the following impurities: Cu—0.002 pct; Pb—0.006 pct; Sn—0.01 pct; Zn—0.01 pct. The alloys were prepared for metal-lographic examination using a technique already reported' and a new value of 2.0 ± 0.1 wt pct zinc was observed. An alloy containing 2.00 wt pct zinc was also made by melting the constituents under a vacuum. In both of these methods of alloy preparation the losses can be kept extremely small. Again, the characteristic eutectic structure was obtained, assuring confidence in this value. The accompanying micrographs are presented in support of this point. Fig. la represents an alloy containing 4.00 wt pet zinc, and reveals excess zinc particles, which precludes the value of both Wilson and Peretti, and Valentiner. Fig. lb, gives the structure of an alloy containing 2.2 wt pct zinc, also showing the presence of excess zinc. Fig. lc is a micrograph of a typical structure of the eutectic observed in several sections of the 2.0 wt pct alloy. In the 1.8 pet zinc alloy, fig. id, the definite presence of the primary indium can be noted. From these re-

Jan 1, 1951

By Edward A. Loria

A SERIES of papers on the study of austenite grain A size in cast steels has been published recently.'" These studies show that in the as-cast state, the austenite grains are quite large, for they form as the steel freezes at a very high temperature, but a simple reheating above the critical range produces a new set of austenite grains which, in many instances, is finer than ASTM grain sire No. 8. Furthermore, there is no apparent relationship between as-cast austenite grain size and subsequent grain size produced by heat treatment. The as-cast austenite grain size can be determined easily in plain carbon and low alloy steels which produce a continuous grain boundary f errite envelope in the as-cast structure. In the case of cast steels whose alloy content produces greater hardenability and suppresses the proeutectoid ferrite transforma-

Jan 1, 1952

By R. Bakish, W. D. Robertson

IN the course of a study of stress corrosion cracking of AuCu,, it was necessary to develop new electropolishing and metallographic etching methods. The techniques are generally useful for Cu-Au alloys, including the AuCu composition in which transformation markings are well delineated. For electropolishing, the only satisfactory electrolyte and procedure that has been found1 has the composition given in Table I. The solution must be vigorously stirred while polishing and the temperature maintained in the range of 55" to 62°C. The electrolyte is very unstable and can be used only for 10 to 15 min after which its color changes and a dark-brown precipitate forms. The polishing time is between 1 and 4 min depending on the initial surface preparation. The use of a hood is recommended to avoid the danger of cyanide gas which may form at one of the electrodes even though the solution is alkaline. The optimum current and voltage conditions for vertical anode and platinum cathode are on the plateau shown in Fig. 1, in which voltage is measured across the cell terminals. At voltages below the plateau, etching of grain boundaries occurs and under some conditions, which have not been clearly defined, octahedral etch figures are observed. For general purpose grain-size determination in polycrystalline alloys, a 3 pct solution of iodine in methyl alcohol is used with excellent results. However, when it is necessary to reveal small orientation differences or concentration gradients in an alloy, the iodine etch is not satisfactory. For this purpose, a highly satisfactory etching procedure was developed during this investigation. It involves the use of the vapor given off by freshly prepared aqua regia. The solution employed is 3HC1 to lHNO, which decomposes in accordance with the reaction HNO, + 3HC1 = NOCl + 2H,O + Cl,. The polished specimen surface is held about 1 to 2 cm above an aqua regia solution for 2 to 10 sec depending on the intensity of the contrast required and the surface is not subsequently washed. The results obtained with a coarse grained cast structure are shown in Fig. 2. Subgrain boundaries in a single crystal and concentration gradients within the boundaries are developed by the same treatment with the results shown in Fig. 3. Obviously, any conclusions regarding structure-dependent properties will be modified by the pronounced heterogeneity, which is not clearly evident when more conventional methods are used. Interesting structural details are also revealed by treatment for 4 hr at 850°C in unpurified tank argon. Fig. 4 shows a pattern characteristic of orientation-dependent oxidation.

Jan 1, 1956

By Nicholas J. Grant, Italo S. Servi

SPECIAL polishing and etching techniques were developed during an investigation of the high temperature mechanical properties of high purity aluminum (99.995 pct Al).* Electrolytic polishing was done successfully in the following electrolyte: 5 ml perchloric acid (72 pct), and 95 ml glacial acetic acid. No special precautions need be taken during the preparation of the electrolyte since no appreciable heat of mixing is evolved. Although the electrolyte is not an explosive, it is recommended that Jacquet's safety rules' be followed during the polishing operation. A voltage of 25 to 60 v gives a good polish in about 2 min at room temperature after preparation on 3/0 metallographic paper. Most of the grain boundaries are visible under the microscope after polishing. Since the current is controlled essentially by the resistance of the bath, it is necessary to keep the anode-to-cathode distance as constant as possible. Accordingly, polishing of irregularly shaped pieces will be difficult. A cooling bath and a stirring system are advisable in order to keep the temperature of the bath uniform and constant. Coarse-grained samples can be macroetched in a few seconds by lowering the voltage below 10 v. This is done after electrolytic polishing, without removing the sample from the bath, Fig. 1. The specimen can be repolished in less than 1 min by raising the voltage above 25 v. The "galvanic etching" technique, which has been described elsewhere,' was found to be helpful not only for macroetching, but also for microetching coarse-grained specimens. In fact, the different orientations of adjacent grains provide good contrast at the grain boundaries, when the sample is observed under polarized light. If the specimen is only partially etched, Fig. 2a, the grain boundaries can be located without completely destroying the appearance of the surface of the specimen. The unetched surface appears black under polarized light. Excellent color contrast is obtained by using a "sensitive tint" plate in the optical system of the microscope. In this case, the unetched surface has a magenta color, while the etched part of the sample assumes different colors varying from green to brown, according to the orientation of the grains. Etch pits were produced on high purity aluminum by using the technique suggested by Lacombe and Beaujard." Good results also were obtained by substituting 70 pct nitric acid for the fuming nitric acid recommended. This gave an etching reagent of the following composition: 70 ml nitric acid (70 pct), 50 ml hydrochloric acid (37.5 pct), and 3 ml hydrofluoric acid (48 pct). An example of the results is shown in Fig- 2b. The outline of the etch pits is revealed sharply if the specimen is observed under polarized light, as shown in Fig. 2c. These metallographic techniques are not suitable for the examination of commercially pure (2s) aluminum. References 'P. A. Jacquet: Metal Finishing (Nov. 1949) 47, p. 62. 21. S. Servi: Metal Progress (1950) 58, p. 732. 3 P. Lacombe and L. Beaujard: Journal Inst. of Metals (1947) 74, p. 1._________________________

Jan 1, 1952