Search Documents

By D. H. Davis, I. Fatt

Oil bearing rocks, usually found at depths of 2,000 to 10,000 it, are elastically deformed by overburden pressure. Although the change in porosity with pressure in a number of example rock.: has been shown to be small, it was thought that overburden pressure might have a significant effect on permeability. This note presents results of an investigation to determine the magnitude of the change in permeabilty with pressure. In routine core analysis permeabilities are measured on rock samples which are not under overburden pressure. If permeablity measured in this way differs from permeabilities measured under overburden pressure, a systematic error is introduced into well productivity calculations. The experiments described in this report were made to determine the magnitude of the change and thus to give an estimate of the error introduced into the calculations. The error introduced by the neglect of the change in permeability when overburden pressure is removed may account for part of the difference between well productivity calculated from core analysis and the actual well productivity. The results obtained in this investigation have only quali-tative significance because of the difficulty in reproducing in the laboratory the stress conditions on the rock buried in the earth. The assumption which is usually made is that the rocks in the earth are under a uniform pressure equal to the weight per unit area of the vertical overburden column from the rock in question to the surface of the earth less the pressure of the liquid in the pores of the rock. For an overburden consisting of sandstones and shales (of assumed average specific gravity of 2.3) and for a liquid pressure equal to the pressure in a salt water column (of specific gravity 1.0) reaching from the rock to the surface, the resulting net pressure on the rock is approximately 0.56 psi per ft of depth. EXPERIMENTAL METHOD Clean, dry sandstone core plugs one in. in diameter and about three in. long were mounted in a copper foil jacket as shown in Fig. 3 or molded in a Lucite jacket as shown in Fig. 4. All cores except those labeled "2" in Fig. 1 and "8" in Fig. 2 were measured in the copper foil jacket. Cores "2" and "8" were measured in the Lucite jacket. The jacketed core was placed in a high pressure hydraulic bomb in which the liquid pressure outside the jacket could be raised to 15,000 psi. Flow lines from the core were brought out of the bomb through special fittings in the bomh head and connected to a laboratory type gas permeameter. Measurements were made with nitrogen gas; the maximum mean gas pressure in the core was 16.1 psia. RESULTS The specific permeability of sandstone decreases with in. crease in overburden pressure. Most of the decrease takes place over the range of zero to 3,000 psi overburden pressure. If the net pressure on the rock is assumed to be 0.56 psi per ft of depth as discussed above, 3,000 psi of overburden pressure is equivalent to about 5,000 ft of overburden. At 3,000 psi overburden pressure, the permeability of the eight sandstone core samples tested ranged from 59 to 89 per cent of the permeability without overburden pressure. Figs. 1 and 2 show the permeability reduction as a function of overburden pressure. The permeability of the eight cores without overburden pressure ranged from 4.35 md to 632 md; these permeabilities are listed in Figs. 1 and 2. REFERENCES 1. Carpenter and Spencer: "Measurements of Compressibility of Consolidated Oil Bearing Sands," U. S. Bureau of Mines, R.I. 3450, (1940). * * *

Jan 1, 1952

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

N recent reports, Sully' and Beck and coworkers' I have advanced hypotheses concerning the formation of the phase. Both of these hypotheses are based on Pauling's theories of the electronic configuration of the elements of the first transition group. Sully considered that the number of electrons which can be absorbed in filling electron "holes" is the factor which determines whether or not the structure can form. Beck suggested that the presence of a certain concentration of electron "holes" is the controlling value. It can be shown that a similar criterion can be developed without making reference to Pauling's theories. Table I shows the elements of the first transition group and the incidence of the a structure in the binary systems of these elements according to the data presently available. Also appended is the number of electrons considered to exist in the 3d-4s levels of these elements. If the atomic percentages for the phase are listed as shown in Table 11, and if the number of electrons in the 3d-4s levels per atom is calculated for the a boundary values, the results are as shown in the last column. It can be seen that the numerical values are all in the vicinity of 7 electrons per atom. This is true also for the ternary Cr-Mo-Ni IT phase, which occurs in a system in which none of the binary compositions are known to develop the phase. In the Fe-Mo binary, the high temperature a phase, which occurs at a composition of 50 atomic pct of each element, also yields a ,;1 e of 7 electrons per atom. From inspection of Tables I and 11, it develops that any series of sequential numbers which is assigned to the elements under consideration will result in the numbers pertaining to a compositions grouping themselves around one specific number. It does not follow, however, that this random number necessarily has physical significance. Such a point has already been discussed by Hume-Rothery." One interesting point develops from this study, however, and that is the position of manganese with reference to the other elements and to the phase. In the system used here, the formation of phase is associated approximately with the value of 7 electrons per atom, which is also the number of electrons per atom that manganese has. Manganese, it will be recalled, exists in three different crystal forms, none of which are the usual, simpler metallic forms; on the other hand, elements just to the left of manganese (in Table I) generally have lower coordination-number structures (body-centered cubic) than do elements to the right of manganese (face-centered cubic). The contribution to a complex crystal structure by the interalloying of elements from opposite sides of manganese is significant in that the a phase is formed in binary systems including one element from each side of manganese, with manganese capable of participating with both sides. It is indeed an interesting coincidence that manganese, with 7 electrons per atom, solidifies in complex crystal forms, and that when alloys of the transition elements are made which have seven 3d-4s electrons per atom, they, too, tend to assume a complex crystal form. From the above, it might be deduced that the study of the a phase might be advanced by a critical examination of the element manganese and how its structure is affected by small additions of elements which change the 3d-4s electrons per atom number. The role of atom size would also be of interest in formation. References ' A. H. Sully: The Sigma Phase in Binary Alloys of the Transition Elements. Journal Inst. Metals (1951) 80, Part 4, p. 173. 2 S. Rideout, W. D. Manly, E. L. Kamen, B. S. Lement, and P. A. Beck: Intermediate Phases in Ternary Alloy Systems of Transition Elements. Trans. AIME (1951) 191, p. 872; Journal of Metals (October 1951). "W. Hume-Rothery, H. M. Irving, and R. J. P. Williams: The Valencies of the Transition Elements in the Metallic State. Proc. Royal Soc. (September 1951) 208A. p. 431.

Jan 1, 1954

By C. G. Dunn, M. Sharp

HE orientations of 137 crystals of copper pro-T duced by secondary recrystallization in a cube texture matrix have recently been obtained. Although the crystals were not prepared for the purpose of determining a secondary recrystallization texture, the orientations obtained proved to be rather interesting, as they confirmed, in part, published results and also disclosed additional preferred orientations in the texture. The investigation of Dahl and Pawlek,&apos; Bowles and Boas,&apos; and Kronberg and Wilson3 showed respectively that the orientations of secondary grains growing in a cube texture matrix were related to those of the primary grains by a rotation of approximately 30" about [001], the rolling direction, by 30" about a <111> direction, and by either 22" or 38" about a <111> direction. Kronberg and Wilson also obtained a component of the texture similar to that of Dahl and Pawlek except the angle of rotation about [001] was 19". When the cube texture matrix was deformed a small amount, Kronberg and Wilson also obtained large grains on subsequent annealing corresponding to a rotation of 15" about the normal to the sheet, but these grains should not be classified as secondary recrystallization grains since they were produced subsequent to mechanical straining. The particular texture obtained, of course, depends on the composition of the copper and the processing schedule employed, since these factors markedly affect the primary recrystallization texture and the sharpness of any cube texture produced. In the present investigation, pieces of OFHC copper in the form of M in. round bars were cold forged to a thickness of about 0.40 in., annealed 1 hr at 350°C in an atmosphere of hydrogen, and cold rolled crosswise to a thickness of 0.200 in. These pieces were given a penultimate anneal of 1 hr at 350°C in hydrogen and a 90 pct cold-rolling reduction lengthwise to a thickness of 0.020 in. Indentations generally were put in one end of the as-cold-rolled strips to promote earlier formation of secondary grains.

Jan 1, 1953

By Gordon W. Hodgson, Edward Tipman

INT&apos;RODUCTION An appreciable number of the oil fields in Western Canada are accumulations of heavy black oils in more or less unconsolidated sandstones. When the crude oils are produced by conventional means, they are frequently contaminated with both sand and oilfield water. The water is usually present as emulsified water, and the sand and other solid materials are suspended in the emulsions. The presence of the solid material in the crude oils is troublesome and, in general, little is known about the behavior of suspended solid particles in emulsions. The Stokes equation describes the behavior of suspended particles in homogeneous fluids in terms of particle size, density differential, and fluid viscosity. In the case of an emulsion, particle size and density differential introduce no difficulties, but considerable doubt centers around the viscosity term. It has been reported1 that the viscosity of an emulsion is related to the viscosity of the continuous phase by an expression of the type: n = a, (1 + a C) where n is the viscosity of the emulsion, n,, is the viscosity of the continuous phase, a is a constant, C&apos; is the volume concentration of the disperse phase. The value of 17, the coefficient of the concentration term, has been determined for various systems, and has been found to vary from a value of 1 to 4.75. The value of 2.5 has been found to be applicable to a large number of systems, and this is the value that is widely used in the equation which then becomes the Einstein equation for the viscosity of emuls~ons. It is of considerable significance that neither the particle size nor the viscosity of the disperse phase enters the expression; the expression for the viscosity of the enlulsion is based solely upon the viscosity of the continuous phase and the concentration of the disperse phase. For crude oil emulsions the Simha equation and the data presented by Fukushima and lchimurai show a non-linear dependence of the viscosity ratio on the water concentration. While it is undoubtedly reasonable to use these data for predicting the viscosity of a specific emulsion, and while it is equally reasonable to expect the calculated viscosity to predict the settling rate of a solid spherical particle in the emulsion by means of Stokes equation, there seems to be no direct evidence that this calculation is indeed sound for describing the settling rates of solid particles in emulsions of water in crude oils. PROCEDURE A direct approach to the problem was thus indicated. Laboratory equipment was designed and built for the direct determination of the rate-of-fall characteristics of solid particles in emulsions of water in oil. Although the apparatus incorporated the basic design of conventional falling-ball viscometers, the study was concerned directly with the rate-of-fall characteristics of particles in emulsions, rather than with the viscosity of the emulsions as such. The crude oil used in this study was a very heavy black crude oil—the McMurray oil obtained from oil-soaked sands outcropping on the Athabasca River in northeastern Alberta. It was produced from the sands by strip mining followed by hot-water separation of the oil in the separation plant of the Alberta Government at Bitumount&apos;. The primary plant product was a wet crude oil containing about 5 per cent solids and 35 per cent water. The final plant product was only slightly different from the native oil, having lost only about 4 per cent of its bulk as light ends during dehydration distillations. In the present investigation two kerosene dilutions of the McMurray oil were used in addition to the dry plant product. The McMurray oil contained 1.92 per cent solids. Analysis of the solids by X-ray diffraction showed the presence of kaolinite, pyrite, and illite as major constituents. with lesser amounts of quartz. Particle size of the pyrite and quartz was less than 10

Jan 1, 1957

By K. S. Milliken

NORMAN W. Lord&apos; has shown that the resultant impurity distribution after n zone passes is given by in the ingot section OSa<N—n, where a is the displacement in zone lengths, N is the ingot length in zone lengths, C,, is the initial uniform impurity concentration, k is the segregation constant = Ceoll C,,,,,,,, at equilibrium, and The foregoing expression is formidable in respect to computational labor. However, the triple sum can be reduced as follows is better if computing for only a single value of n. For k<l, it can be shown that this last equation is equivalent to The infinite series is useful only when k is small enough to force rapid convergence. In such cases, the absence of subtraction from unity has the definite advantage of preserving accuracy. Acknowledgments It is a pleasure to acknowledge stimulating correspondence with Dr. Howard Reiss of Bell Telephone Laboratories Inc. This note is published by permis-sion of the Deputy Minister, Dept. of Mines and Technical Surveys, Ottawa, Ont., Canada.

Jan 1, 1956

By N. L. Ananthanarayanan, J. F. Libsch

A LITERATURE survey1-&apos; of sintering metal pow-A ders and powder compacts indicates that studies have so far been confined primarily to the latter stages of sintering in relatively coarse powders. Presumably this is due to the difficult techniques associated with studies of the early stages of sintering. Recent experimental and theoretical investigationsL-" on the preparation and the magnetic properties of ultrafine ferromagnetic powders suggest the possi- bility of following the sintering process from a very much earlier stage than has previously been considered. It is now well-known that ferromagnetic powder particles which are below a critical size behave as single domains and exhibit permanent magnet qualities characterized by a very high coercive force. When the same particles are fully sintered, the coercive force decreases to very low values corresponding to those of the solid metal. Thus the coercive force appears as a sensitive parameter which may be measured to study the early stages of sintering. To determine the feasibility of this technique, experi-

Jan 1, 1954

By N. J. Grant, I. S. Servi, A. Chaudhuri

IN a recent paper by Servi and Grant,&apos; it was illustrated that the slip band spacing and stress based on creep test data were related by the equation: d =1/ where d is the spacing in millimeters and v is the stress in psi, based on a more complete equation proposed by Orowan. It was shown further,&apos; that for a. given grain size, the slip band spacing became larger with decreasing stress (and was independent of temperature), until such time that the slip band spacing becomes greater than about one-half the grain diam- eter. When this occurred, the mechanism of deformation in creep of 2s aluminum at high temperature changed from slip to grain boundary deformation (sliding). Additional metallographic evidence is now available which confirms this in a most interesting fashion.

Jan 1, 1954

By C. W. Chen

MARTENSITE transformations in ß Cu-Al alloys have been studied by Greninger1 and other investigators. According to Greninger, the parent phase ß1 an ordered body-centered-cubic structure obtained from ß phase by suppressing the eutectoid decomposition, transforms into an ordered hexagonal-close-packed phase in composition containing 12.9 to 14.7 pct Al. The M, temperature decreases with increasing aluminum content; for the alloy containing 14.5 pct Al, for example, the ß1??&apos; transformation occurs below room temperature. More recently, Kurjumow2 studied the transformation in ß Cu-Al alloys with the addition of nickel. His report stimulated new interest in the subject due to the observation of completely reversible transformation without hysteresis in the transformation temperature ranging from 10° to —10°C. In the present paper some characteristics are described of the transformation of Cu-Al-Ni alloys that were partly studied by Kurjumow. Experimental Procedure High purity copper (99.999 pct) and aluminum (99.99 pct) and electrolytic nickel were used in the preparation, by the Bridgman technique, of single crystal specimens which contained aluminum and nickel of 14.5 and 0.5 to 3.0 pct, respectively. Polished surfaces were prepared mechanically. Specimens were then chemically etched to remove distorted material, homogenized at 1000°C for several hours, and quenched drastically to room temperature in a 10 pct NaOH bath to produce the parent phase ß1. The transformation was studied under a microscope and, in some cases, recorded by means of motion pictures. A device similar to that designed by Greninger and Mooradian3 was used to cool and reheat the specimens. Results and Discussion When the specimens were cooled below room temperature, the ß1 to ?&apos; transformation began at 10°C with the appearance of ?&apos; crystals In relief, Fig. la. As the specimen temperature dropped further, the transformation continued, either by the growth of the ?&apos; crystals, with the ß1 — ?&apos; interface moving into the ß1 phase, Fig. 1c and 1d, or by the formation of new ?&apos; crystals, Fig. 1b. As a consequence of the former process, banded structure is observed as a common feature of the low temperature phase. According to the theory of the formation of martensite by Wechsler, Lieberman, and Read,&apos; the bands of ?&apos; phase are probably twin-related, as is the case in the diffusionless phase change of In-T1 alloys,5 but this was not revealed by X-ray tech- niques. New ?&apos; crystals, in needle form, often emerged suddenly across the ?&apos; bands during the transformation. These acicular crystals then grew, both in length and in width, see Fig. 2a through 2d. The transformation on cooling is completed at about -35°C. Upon heating, the reverse transformation started at —10° C, in a manner nearly opposite to the transformation on cooling, and completed at 35°C. There was no noticeable change in the transformation temperature when the nickel content was varied within the limits previously mentioned. Through control of the specimen temperature, the transformation can be started, stopped, or reversed at will. This phenomenon has frequently been observed in the martensite transformation of many nonferrous alloy systems. Other systems are Au-Cd6 and In-Tl.5 ow-- ever, in the latter systems, the transformation is accomplished by single interface motion if the specimen composition is homogeneous and the temperature gradient in the specimen is uniform and sharp, whereas in the Cu-Al-Ni specimens, only multiple interface transformation is observed. The speed of the interface motion appears to be a functionof the rate of temperature change and the temperature gradient across the specimen length. In one case, in which the temperature increased at the rate of 10°C per min and there was no temperature gradient along the specimen axis, the speed of the disappearance of a ?&apos; plate was determined, by the study of the motion pictures made, to be 26 µ per sec. Quench markings were observed on the polished surfaces of specimens. The markings were grouped into one or more sets of different orientations, and were parallel in each set. The ?&apos; plates formed in subsequent transformation were parallel to the markings, indicating that the ?&apos; plates and the quench markings had the same geometric relation-ship to the ß1 matrix. The quench markings on two intersecting surfaces of a specimen were therefore used in the determination of the habit plane of transformation, by the trace method suggested by Barrett.&apos; Results obtained from five sets of markings in three specimens indicate that the habit plane is an irrational plane about 2" from one of the {221} planes. This is very close to the habit plane (3" from 221 planes) of ß Cu-Al alloys containing more than 13.0 pct Al.1 The martensite transformation of Cu-Al-Ni alloys is reproducible. No sluggishness was found between consecutive transformation cycles, although a slight difference in the distribution pattern of the ?&apos; plates was observed, compare Figs. Id and 2d. The transformation can be strain-induced. This characteristic has been tested by a simple method. When a specimen was elastically strained slowly in a vise, ?&apos; plates were gradually produced in the same fashion as during transformation on cooling, This test was done at room temperature, and thus above the M,

Jan 1, 1958

By R. V. Riley

IN the metallographic examination of the alloys of iron and silicon, several investigators1-&apos; have reported on the occurrence of an unusual type of etched structure, particularly in the alloys of higher silicon content. This, the so-called "barley shell" structure, has been observed with iron-silicon alloys over a wide range of silicon content, and under certain circumstances it can be seen upon the micro-sections of pure iron. The barley shell structure may be demonstrated most easily with iron-silicon alloys in the range 8 to 20 pct Si after etching in a hydrofluoric acid base etchant. Two etchants which produce the structure are: Corson&apos;s reagent, which is 2 parts hydrofluoric acid, 1 part nitric acid, and 2 parts glycerol; and picric acid reagent, which is 1 part hydrofluoric acid and 1 part saturated solution of picric acid in alcohol. Both when freshly prepared and used for etching commercial silicon-iron alloys containing 14 to 16 pct Si, invariably produce the barley shell structure shown in Fig. la. It is now well known that the barley shell structure is a pseudomorph and the true structure of an alloy of this composition consists of a single phase of silico-ferrite together with graphite as shown in Fig. lb. Lower silicon alloys and even pure electrolytic iron may also be etched in the above reagents to give the barley shell provided the etchant has been either: 1—Stored in contact with glass or silica, or 2—used previously for etching an iron containing silicon. Pure iron etched in perfectly pure hydrofluoric acid base etchants free from silicon contamination gives a normal structure free from the barley shells. Therefore, although the hydrofluoric acid in the reagent primarily is responsible for the barley shells, silicon, or its compounds, also must be present, either in the etching reagent or in the specimen. It was of considerable interest to the author to read the paper by Wachtell,? in which the barley shell etch was deliberately used for the identification of the boundaries of high and low silicon areas in diffusion experiments made upon laminate bars of compacted and sintered pure iron and ferrosili-con powders. This method of utilizing the barley shell etch is novel and in the hands of the investigator gave a useful differentiation between the high and low silicon areas in those specimens. For the reasons already given, however, the technique is not to be regarded as of universal application. Misleading results might be obtained, for example, if two or more specimens were etched in the same etchant. Fresh pure reagents only must be used each time. Provided this is done and the investigators are well aware of the limitations of the method, then the barley shell pseudomorph can be made of limited use in distinguishing between higher and lower silicon areas in a microspecimen of a ferrous material. The greatest of care must be exercised in the interpretation of the results which may be regarded as qualitative only and may, in fact, not be specific to silicon alloys.

Jan 1, 1952

By O. Schaaber

THE existence of isothermal martensite without the presence of the athermal form thus far has been established only in a 6 pct Mn, 0.6 pct C steel&apos; and a 23 pct Ni, 3.5 pct Mn-Fe alloy.&apos; " Theoretical calculations, moreover, predict this phenomenon to be observable only within narrow composition limits.&apos; In all other steels, it is generally considered that martensite inevitably forms upon cooling below the M, temperature. Even the possibility of isothermal martensite transformation has been questioned for the common steels, although Averbach and Cohen" showed that martensite formation does not stop when the cooling is arrested and that as much as 5 pct retained austenite may transform isothermally to martensite at room temperature. This note presents some observations made on a number of low alloy steels by dilatometric and inductive measurement of isothermal austenite transformation at various temperatures close to the Ms. In the dilatometer studies reported previously, austenitized specimens were immersed in a salt bath maintained at proper temperature and immediately inserted into the dilatometera8&apos; already at thermal equilibrium with the bath. The dilation was read on a dial gage divided into 0.001 mm intervals. Induction measurements were made with a separate excitation source and a test coil, sealed together in a salt-tight case. The increase in the induced voltage caused by the transformation products was recorded at a paper speed of 4 mm per sec. The steels studied are listed in Table I. Specimens prepared from steels 1 and 2 were in the form of hollow cylinders 60 mm long, 10 mm ID, 14 to 16 mm OD, while specimens of steels 3 and 4 were in the form of notched-bar impact test specimens, DVM type. Specimens of steel 5 were in the form of strip, 7x0.3 mm. In the bainitic range, curves representing the amount of transformation vs log time have a single step appearance, as is shown by the curve of isothermal transformation at 320°C in Fig. 1.Y ransformation proceeds in accordance with the Austin-Rickett equationo P -----------= k t" 100-P where P is the percentage of transformed austenite, t is time, and k and n are constants. Near the M, temperature, however, complex curves having two steps are observed. As shown in a previous papera for steel 3 isothermally transformed at 245"C, such complex curves can be separated into two separate transformations or reactions each following the Austin-Rickett equation. The values of n in the equation for a number of such two-step transformations are listed in Table 11 for steel 1. At transformation temperatures of 250" and 235"C, evidence of a second transformation is very

Jan 1, 1956

By P. R. Hines

Diphenyl guanidine is used as an accelerator in vulcanizing rubber. Other rubber accelerators are also flotation collectors, e.g., dithiocarbamate, thiazole, and the xanthates. Urea and its derivatives are good flotation collectors,&apos; and guanidine is the nitrogen analog of urea. These two characteristics suggested testing diphenyl guanidine as a flotation collector. Diphenyl guanidine has been tried in flotation work previously, but references give no details.&apos;.&apos; Bunker Hill Co. of Kellogg, Idaho, supplied the ore sample used in these tests. The sample was typical of the ore milled in 1949 and contained 4.52 pct Pb and 1.08 pct Zn. A flotation test made by the author with potassium ethyl xanthate as a collector, the same flotation reagents, and the same grind employed in regular Bunker Hill mill practice was the standard for comparing results with the compounds and derivatives of guanidine. The compounds and derivatives of guanidine are only fair collectors of galena. However, if potassium ethyl xanthate were used in the galena float, its presence later in the sphalerite float would make it impossible to introduce another type of collector and be certain which collector contributed the result, so the compound or derivative of guanidine under test was used in both the galena and sphalerite floats in Table I. On both the galena and the sphalerite in the Bunker Hill ore, the depressing action of sodium cyanide and zinc sulfate is much greater with collectors of the guanidine type than with the xanthates. If the depressing agents are left out when a guanidine type of collector is used, but a pH of 8.6 is maintained with either lime or sodium carbonate, the depressing action is approximately the same as in the standard xanthate test. Consequently the depressing agents, sodium cyanide and zinc, were not used in the galena floats of tests 1226, 1228, and 1286, Table I, so the galena floats would be on a comparable basis with the standard xanthate test. As shown by the tests in Table I, the loss in the tail when diphenyl guanidine is used is 8.3 to 9.3 pct, as compared to 17.6 pct with potassium ethyl xanthate, or a recovery of 2.3 to 2.0 lb more zinc. Concentrate grade is 13.7 pct with potassium ethyl xanthate and increases to 22.9 and 31.8 pct with diphenyl guanidine, showing a much higher selectivity. Unless the Barrett&apos;s No. 4 coal tar creosote is considered to be a collector, no collector other than diphenyl guanidine is present in the tests, Nos. 1226, 1228, and 1286. Tests 1218, 1253, and 1263 in Table I1 were run to check the effect of the depressing agents used in flotation of the galena (sodium cyanide and zinc sulfate) upon subsequent flotation of sphalerite by diphenyl guanidine. Urea was substituted for potassium ethyl xanthate in tests 1253 and 1263 because it is a good collector for galena and a poor one for sphalerite, and the effect of the depressing agents is more marked with urea than with potassium ethyl xanthate. Test 1205 checks the effect of urea alone without any depressing agents when used for the flotation of galena. Table I1 shows that diphenyl guanidine as a collector for sphalerite is compatible with other flotation reagents used in floating galena. When the depressing agents sodium cyanide and zinc sulfate are used in the galena flotation step, the amount of lime in the sphalerite flotation must be increased (compare 1253 and 1263). Table I11 gives some of the compounds of guanidine which were tested and compared. Some of the compounds and derivatives of guanidine* are solu- * Diphenyl guanidine is in commercial production and is available on the market. ble in water and others are not. Those tested were fed dry. When an organic collector is promising, its compounds and derivatives should be tested and their comparative collecting power recorded. Eventually, in this way, it may be possible to discover some of the essential chemical characteristics of a good collector. For example, it is interesting to compare the diphenyl and dibutyl derivatives in Table IV with those of urea in Table 111, Example 3, U. S. Patent 2,664,198. All are good collectors. Table I shows some of the effects of lime on recovery of sphalerite by diphenyl guanidine. Table V

Jan 1, 1960

By E. P. Klier, Volker Weiss, George Sachs

IT has been shown that the isothermal transformation of austenite in a special steel can be modified by isothermal holding in the high subcritical transformation range.&apos; Since this treatment potentially leads to increased hardenability, limited testing has been completed to determine if the stabilization effects observed with this treatment can be used to advantage in the heat-treating of 4340 steel. As an index of the effects of the several austenitizing treatments on the hardenability of the steel, Jominy-type hardenability tests were completed for specimens 2.5 in. in diameter and 8 in. long. Since the steel studied is to be used at strength levels where notch sensitivity is of major concern, notch sensitivity after selected austenitizing treatments has been evaluated by notch tensile tests for several specimen sizes. The steel composition and the austenitizing treatments are given in Table I. The test specimens are presented in Fig. 1. The hardenability data are given in Fig. 2 and the notch tensile properties are presented in Fig. 3. The hardenability of the steel investigated when determined by use of the standard Jominy test specimen2 as high and remained essentially unchanged for the different austenitizing treatments. For the larger test specimen, as has been shown by other investigations," the characteristic hardenability indexes are changed and it becomes possible to isolate austenitizing treatments which promote maximum hardenability. The stepped austenitizing treatment promotes higher hardenability as is indicated in Fig. 2. The evaluation of notch sensitivity for very high strength steels is best accomplished by means of the notch tensile test. The experimental conditions of the test, however, modify the notch strengths measured, and in the present experimentation an effort was made to achieve a limiting notch condition by fixing the notch root radius at 0.001 in. while the specimen diameter was allowed to increase from 0.3 to 1.1 in.* For this root radius the notch strength, 0.3-in.-diam specimen, and this trend differs from that for the larger specimens. At the highest tempering temperatures the 1.1-ip.-diam specimens yield comparable results for the austenitizing treatments compared. However, as the tempering temperature is progressively lowered below 800°F, em-

Jan 1, 1958

By C. G. Dunn, M. Sharp

IN twinned crystals of the face-centered cubic metals the lattice of one twin is a mirror image of the other in a common twin boundary. When several twins appear within large grain in a sheet specimen, the twin one boundaries form a set of lines at the surface of the specimen which coincide with (111) planes of the large grain. Furthermore, for twins of the same orientation, these lines are parallel. Generally, the presence of identically oriented regions with straight parallel boundaries coinciding with a (111) plane of the surrounding crystal is strong evidence for identifying the island regions as twins of the parent crystal. However, Fig. 1, which shows the macrostructure of a large grain of copper with island regions that satisfy these conditions. is not an illustration of (111) twins. Since the reverse side of the specimen has much the same appearance, it was thought at first that these regions, which appear dark in the macrograph, actually were twins. According to X-ray data, however, these regions are second-order twins of the large crystal. With regard to their formation, these second-order twins formed by secondary recrystallization in a cube texture matrix. Growth occurred in the direction of the arrow (see Fig. 1) as the specimen moved slowly into a gradient temperature furnace as described previously.&apos; Nucleation of the second-order twins occurred, therefore, on the ends facing opposite the arrow. If the origin of the second-order twins were due to repeated twinning, some first-order twin structure should be visible on these ends. This proved to be the case, as very small twins were readily found with the aid of a microscope, and probably could have been seen, in some instances, under ideal lighting conditions without aid of a microscope. Fig. 2 shows a cross-section view taken perpendicular to both the surface and the (111) trace of the parent crystal (visible as a straight boundary in Fig. 1) at the beginning point of growth of a second-order twin and where one first-order twin was relatively thick. In the micrograph, A is the large parent grain; B is the first-order twin of A; and C, which is a first-order twin of B; is a second-order twin of A. Between A and B and between B and C the major straight portions are traces of common (111) twin boundaries. The straight portion of boundary between A and C, however, is not a common crystallographic plane to the two lattices; it is a (111) plane of A and a (115) plane of C. Without considering the mechanism of twinning itself, the origin of the second-order twins may be accounted for in terms of repeated twinning and special growth characteristics. After each nucleation, a selective growth process can be thought of as favoring growth of the first-order twin in local spots only and favoring growth of the second-order twin to an extent comparable with that of the parent grain over relatively large areas in a way similar to that described for twinning in aluminum.&apos; It has already been pointed out that the boundary between the large grain (A) and the second-order twin (C), which is responsible for the straight boundary portions in Fig. 1, involves a (111) plane of A and a (115) plane of C. The same combination of planes is not only possible in first-order twins, but actually appears quite frequently.3 Their prevalence in first-order twins and their presence here in second-order twins, together with the necessary occurrence of a large number of common lattice sites at the boundary, is an indication that this combination produces an "energy cusp"&apos; boundary. (Energy cusp boundaries have been described by Shockley and Read.") The configuration of atoms near a {Ill), (115) boundary in first-order twins is of course different from the configuration near the same type of boundary in second-order twins. References 1 M. Sharp and C. G. Dunn: Secondary Recrystallization Texture in Copper. Journal of Metals (January 1952) Trans. AIME, p. 42. 2W. G. Burgers and W. May: Stimulated Crystals and Twinning in Recrystallized Aluminum. Recueil des travaux chimiques des Pays-Bas (1945) 64, p. 5. aD. Whitwham, M. Mouflard, and P. Lacombe: Discussion of W. C. Ellis and R. G. Treuting, "Atomic Relationships in the Cubic Twinned State." Trans. AIME (1951) 191, p. 1070; Journal of Metals (October 1951). 4 W. Shockley and W. T. Read: Dislocation Models of Crystal Grain Boundaries. Physical Review (1950) 78, p. 275.

Jan 1, 1953

By J. Gordon Parr, D. H. Polonis

IN a previous study of hypereutectoid Ti-Ni alloys&apos; a substructure was observed in quenched powder specimens when the constitution was 100 pet retained ß body-centered-cubic phase. The alloy powders were quenched from 1000°C by a blast of helium gas and precautions already described were observed throughout the work, to minimize contamination. Ogden et al.2 have previously observed subboundaries in the ß phase of a Ti-Mn-N alloy; but this structure was attributed to nitrogen-rich a precipitation. The nitrogen content of the Ti-Ni alloys does not exceed 0.02 pet and the quenching rate was sufficiently rapid to suppress precipitation reactions. Since X-ray diffraction results have confirmed the absence of any second phase which might account for this phenomenon, it seems unlikely that this substructure is explicable in terms of nitrogen content. Fig. 1 shows the substructure in adjacent grains of 100 pet ß phase in a quenched 6 atomic pet Ni alloy. Fig. 2 shows adjacent ß grains in a similarly treated alloy; one grain contains marked intracrys-talline boundaries and the other is completely free from any substructure. The suberystals appear to have fairly regular shapes, suggesting a regular orientation relationship among them. Tempering the retained ß structure at temperatures below the eutectoid eliminates the substructure at an early stage of ß decomposition. At the present time it is possible only to guess at an explanation for this subboundary phenomenon, since very little work has been reported on substructures resulting from phase changes. It is believed that the present phenomenon is due to poly-gonization resulting from unrelieved stresses set up

Jan 1, 1957

By W. M. Reilly, Martin E. True, O&apos

A technique, equipment, and a compound have been developed and field tested for sealing tubing joint thread leaks without removing the tubing from the well. This eliminates the necessity of killing the well with possible damage to the producing formation, which could result if it were exposed to drilling mud. A thread sealing compound, developed and tested in the laboratory, is introduced into the tubing and squeezed into the joint to eliminate the leak. A high percentage of successful field applications have eflected an appreciable cost reduction in repairing leaking tubing strings in East Texas and in the Louisiana Gulf Coast area. INTRODUCTION In producing oil and gas wells the industry has been plagued with a never ending problem of tubing connection leakage. This situation is becoming increasingly more significant as more gas wells are placed on production and as higher pressures are encountered at greater depths of current drilling. Substantial progress has been made in minimizing thread leaks by careful handling and cleaning of joints and by the use of improved sealing compounds in combination with controlled makeup. But even with the use of the best known methods and sealing materials during initial makeup, occasional leaks through tubing joint threads appear inevitable. In some instances, tubing joints in high-pressure gas wells have been found to leak gas at relatively less pressure than that at which the tubing had been tested satisfactorily using liquid. Many expedients have been employed to prevent the physical replacement of tubing when leaks have developed. Perhaps the most commonly used is the sodium silicate treatment in which the solution is lubricated into the tubing to form a crystalline plug in the leak. Generally, any benefit realized is of a temporary nature. In addition to field studies on handling and makeup of tubing, extensive laboratory work has been undertaken on the development and testing of thread dopes.&apos; Also, substantial effort has been devoted to testing tubing joints using high-pressure gas as a pressure medium.&apos; The results of this work have emphasized that tubing joints are susceptible to occasional leaks, especially when subjected to relatively high differential pressures. To cope with the problem of leaking connections, an extensive investigation was undertaken to devise means of sealing leaks without removing the tubing from the well and without reducing the inside diameter of the tubing. The purpose of this paper is to present a technique, together with the associated equipment and compound, which have been developed for effecting a seal of leaking tubing in place. DISCUSSION Work was initiated simultaneously on two ventures: (1) to develop a satisfactory leak sealing compound and (2) to develop tools necessary to find the leak and place the compound. Leak Sealing Compound In the development of a sealing compound, tests were conducted on those compounds commercially available which include: (1) conventional thread lubricants, (2) thermo-setting resins and (3) other materials which held promise of effecting a seal. Although none proved completely satisfactory in their available form, two commercial compounds proved nearly equally effective in sealing leaks during laboratory tests after proportioned quantities of graded silica had been added. Field usage has indicated that the preferred compound is one prepared of an aluminum-stearate base thinned to the proper constituency and carrying a mixture of fine mesh graphite and graded silica. The compound developed by Humble Oil & Refining Co. is designated as Tubing Thread Leak Sealant 800. Placement Tools and Technique In developing tools to detect joint leaks and to place the sealing compound, work was done initially on wireline equipment. Preliminary designs were completed of a tool somewhat on the order of a pressure bomb to test individual joints with gas at high pressure to determine if the joint was leaking. An entirely different tool equipped with opposed packers was intended for placement of the sealing compound at high pressures in each leaking joint. The application of these tools

Jan 1, 1958

By D. L. Martin

THAT small amounts of impurities have a harm-J- ful effect on the malleability of metals and alloys is well-known. One common type of em-brittlement involves the formation of a small quantity of eutectic at the grain boundaries, and is typified by sulfur embrittlement of nickel. Another example of eutectic embrittlement—sulfur embrittlement of cobalt—is described in this report. In the preparation of specimens for a study of Co-Pt alloys, it was observed that the hot swagabil-ity of the alloys appeared to be influenced by the grade of cobalt used as melting stock. The cause of the hot-shortness was not traced to any particular impurity, although sulfur was suspected.&apos; The damaging action of sulfur on the malleability of cobalt has not been generally recognized,&apos;; in spite of the fact that the Co-S phase diagram8 eveals the same basic conditions—low solubility and eutectic formation—responsible for sulfur embrittlement of iron and nickel.7, 8 Apparently sulfur embrittlement has not been a serious problem in the fabrication of cobalt alloys. One related factor is likely the limited commercial experience with high cobalt alloys, since the predominant use of cobalt has been in alloys containing less than 75 pet Co. A series of Co-S alloys was prepared and studied to establish the effect of sulfur on the hot working characteristics of cobalt. Specimens, 5/8 in. diam by 4 in. long, were prepared by vacuum melting in magnesium oxide crucibles and casting in a copper mold. Two grades of low sulfur electrolytic cobalt were used as the basic material for the study, one from Johnson, Matthey and Co. and the other from the International Nickel Co. Each contained about 99.5 pet Co and had a nominal sulfur content under 0.005 pet. The sulfur was added initially as cobalt sulfide powder. Later, some low sulfur alloys were prepared by diluting the high sulfur alloys with cobalt. Trouble was experienced in controlling the sulfur content. Apparently much of the cobalt sulfide powder was lost during melting, as most of the samples contained a lower sulfur content than was added. The alloys were evaluated for hot brittleness by heating to 1000°C in a hydrogen furnace and swaging in air. The results of these tests are summarized in Table I. All the alloys containing less than 0.008 pet S were successfully hot swaged to 1/4 in. diam. The bars containing more than 0.015 pet S could not be swaged because of the formation of intercrystal-line cracks similar to those illustrated in Fig. 1. Examination of the microstructure of the cast bars revealed cobalt sulfide eutectic at the grain boundaries, Fig. 2. The sulfide was not always continuous at grain boundaries in bars that were hot embrittled but, in general, the particles did form a network that could account for the brittleness above the eutectic temperature, 880°C. There is no question about the brittle nature of a specimen exhibiting a continuous sulfide network such as is shown in Fig. 2b, though a sulfide distribution similar to that in Fig. 2a is marginal. This particular bar cracked when hot-swaged. However, other bars with lower sulfur content but with some sulfide phase at grain boundaries were successfully worked. The reduction of area of a hot tensile test specimen was also found to be a sensitive indicator of sulfur embrittlement. The variation of reduction of area with sulfur content is shown in Fig. 3 for specimens fractured at 900°C in an evacuated capsule. The ductility of cobalt is greatly reduced when the sulfur content exceeds 0.005 pet. Similar results were obtained for alloys tested at 800°C, below the eutectic temperature. The upper sulfur embrittlement limit of 0.015 pet, Table I, agrees with the observation of Kalmus8 hat

Jan 1, 1957

By J. G. Richardson

The performance of laboratory water floods is compared with the flooding behavior calculated by the Buckley-Leverett techniques from steady-state relative permeability-saturation relations. Both the steady-state and unsteady-state experiments were conducted on the same sand-packed column. Relative permeabilities were calculated by the Welge technique from the unsteady-state waterflooding displacement experiments and compared with those measured by steady-state techniques. An evaluation of the various calculation procedures was made. It was demonstrated that steady-state relative permeabilities could be used to calczcJate waterflood performance. INTRODUCTION To calculate the waterflooding behavior of a given reservoir, relative permeability-saturation relations of the reservoir rock are required. These relative permeabilities are commonly measured on core samples by steady-state experiments such as the Penn State technique.&apos; The relative permeabilities obtained in steady-state experiments are calculated from oil and gas flow rates and pressure drops measured under conditions of equilibrium distribution of oil and water in the pore spaces with slow approach to equilibrium. On the other hand, in water floods the water saturation at a given point changes rapidly with time as the flood front passes. Extensive research2 has shown that the steady-state relative permeabilities to oil and water are independent of the rate of flow and the fluid viscosities. Research has also demonstrated that the waterflooding performance of homogeneous sands uniformly contacted by water is independent of rate when capillary pressure gradients, which are important in short laboratory columns, are made negligible. However, there is little experimental data in the literature comparing relative permeabilities calculated from steady-state flow experiments with those calculated from flooding experiments. In this paper, the production data of two water floods and the relative permeabilities calculated from a series of steady-state experiments on an unconsolidated sand column are presented along with the comparisons calculated by the Buckley-Leverett and by the Welge techniques. The purpose of this study was to determine if steady-state relative permeabilities could be used to predict the waterflooding performance. This study also permitted an evaluation of the Buckley-Leverett and Welge calculation procedures and allowed an insight into the nature of microscopic displacement of oil by water. EXPERIMENTAL APPARATUS AND PROCEDURE The apparatus used in this investigation was a Lucite column 1 ft long and 11/2 in. in diameter. A drawing of the apparatus is shown in Fig. 1. Six pressure taps were evenly spaced along the top of the column to measure the oil pressures. Three pressure taps fitted with small porous disks were spaced along the side of the column to measure water pressure. Water was admitted to the inflow face of the sand through a porous disk while oil was admitted through a hole in the porous disk and distributed to the face of the sand by grooves machined in the porous disk. Retainer screens of 325-mesh stainless steel were cemented over the inlet and outlet ports and the pressure taps to prevent loss of sand. The column was packed with a clean dry sand all of which was finer than 200 mesh. Next, the sand was saturated with a solution of 150,000 ppm sodium chloride solution. The column was weighed dry and water-saturated to obtain data for the calculation of the pore volume. A 151-cp oil was then injected into

Jan 1, 1958

By Pol Duwez

IN the course of an investigation of the V-CO system, two intermediate phases were found. One of these phases corresponds approximately to the stoichiometric composition VCo and is isomorphous with the sigma phase in the Fe-Cr system.&apos; The second phase has the composition V3Co; its crystal structure is described in the present note. The alloys were prepared by mixing the two metals in the powder form, pressing a small disk weighing about 5 g at 80,000 psi, and arc melting this disk on a water-cooled copper plate in an atmosphere of pure helium. The details of this technique have been described.&apos; The vanadium powder was obtained from Westinghouse Electric Corp., Bloomfield, N. J. This powder is probably of very high purity, since when it is properly sintered or melted in the above-mentioned arc furnace, ductile specimens are obtained. The cobalt powder, from Charles Hardy, Inc., New York, contained 0.5 pct Ni, 0.1 pct Cr, and traces of Si and Fe. After melting, the V,Co samples were sealed in evacuated quartz tubes and homogenized for ten days at 800°C. Powder diffraction patterns were obtained with a 14.32 cm diam camera, using Ka copper radiation. The patterns were readily indexed on the basis of a primitive cubic lattice with a parameter equal to 4.675A. The density, determined by the immersion method, was 6.71 g per cu cm; hence the number of molecules per unit cell is approximately 1.95; i.e., 2. At this point, the possibility that the structure might be that of beta tungstena became apparent. The beta tungsten structure is described as follows: Space group 03,, — Pm3n 2 Co in (a) : 000; ?4lhYZ (hhl) reflection present only if 1 = 2n. Assuming this structure to be the correct one, intensities were computed by means of the usual eauation: 1 + cos220 I oc p F sin 0 cos 6 where F is the structure factor, 0 the Bragg angle, and p the multiplicity factor. The observed and calculated values of sin 0 and the intensities are given in Table I. The agreement between the observed and the calculated sin 0 is good and there are no flagrant discrepancies between the calculated intensities and those estimated visually. The (hhl) reflections for which 1 is odd are not observed, as required by the space group. In addition, the (410), (430), and (531) reflections are missing as expected, because of the special (a) and (c) positions in0%. However, six reflections—(llo), (220), (310), (411), (422), and (510)—which have very weak computed intensities were not observed. For these reflections, the structure factor is proportional to the difference between the scattering factors of the two atoms in the structure. Since the scattering factors of vanadium and cobalt are not very different, these reflections are weak. However, by using Ka chromium radiation, whose wavelength is just above the absorption edge of vanadium, the effective scattering factor of vanadium may be decreased by one or two units; consequently the difference between the cobalt and vanadium scattering factors is increased. It was, indeed, found that in a powder pattern taken with chromium Ka radiation, the three reflections (110), (220), and (310) were actually present. The three other reflections (411), (422), and (510), with spacings smaller than half the wavelength of chromium Ka, were obviously not obtainable with chromium radiation. All the experimental results appear to confirm the beta tungsten structure for V,Co. In this structure, each cobalt atom is surrounded by twelve vanadium atoms at 2.61A; each vanadium atom is surrounded by two vanadium atoms at 2.34A, four cobalt atoms at 2.61 A, and eight vanadium atoms at 2.86A. Acknowledgment This work was done at the Jet Propulsion Laboratory, California Institute of Technology, under contract number W-04-200-ORD-455 with the Army Ordnance Department, Washington, D. C. The author wishes to thank this agency for the permission to publish the results of this investigation. References &apos;P. Duwez and S. R. Baen: X-Ray Study of the Sigma Phase in Various Alloy Systems. Symposium on the Nature, Occurrence, and Effect of Sigma Phase. ASTM Special Tech. Pub. No. 110, pp. 48-54. Philadelphia, 1951. 2 C. H. Schramm, P. Gordon, and A. R. Kaufmann: The Alloy Systems Uranium-Tungsten, Uranium-Tantalum, and Tungsten-Tantalum. Trans. AIME (1950) 188, pp. 195-204; Journal of Metals (January 1950). 3 M. C. Neuburger: The Crystal Structure and Lattice Constants of Alpha and Beta Tungsten. Ztsch. fiir Krist. (1933) 85, pp. 232-237.

Jan 1, 1952

By Robert G. Nisle

In conducting a pressure build-up test on an oil well, it is often necessary to shut-in the well for a short time prior to initiation of the test. The effect of such a short tern shut-in on the resulting pressure build-up curve is derived. The derivation is based on the Kelvin point-source solution and is a straight forward extension of the fundarnental theory. The magnitude of the effect is evaluatecl and illustrated by an example. It is shown that if the production subsequent to the short term shut-in is at least 10 times the duration of the shrct-in, the effect will he less than 10 per cent. INTRODUCTION It often happens that before a pressure build-up test can be made an additional valve must be mounted on the wellhead, or on arrival at the well, it might be learned that the well had been shut-in several hours the previous day because of a shortage of tankage. In such cases the question is How will such an interruption of production affect the results of a pressure buildup test This work explains how to take account of such a short term shut-in in order that the test can proceed without further delay. The solution presented here makes use of the Kelvin point-source solution used by Horne1 in his treatment of the pressure build-up in wells. Use is also made of the super-position theorem. MATHEMATICAL DERIVATION The well is assumed to have been in production at a constant rate for a relatively long time (several days or more). It is then shut-in for a short time, after which it is again placed in production at the previous constant rate. It continues to produce at this rate until shut-in for a pressure build-up test. The Kelvin point-source solution is expressed by Eq. 1. The sequence of production rates is listed in Table I. TABLE 1 — SEQUENCE OF PRODUCTION RATES Time interval Production Rate At time AT the pressure will be the resultant of the sequence of production rates shown in Table 1. An expression for the super-posed effects may be derived from this table. (Effect of + q for T + ? T) + (Effect of - q for T + ?T-t,) + (Effect of + q for T + ? T - t,) + (Effect of - q for ? T) = Resultant........(2) The desired result is obtained by substituting the effect defined by Eq. 1 for the terms in Eq. 2. The result is Eq. 3. When the logarithmic approximation is made for the exponential integrals and the terms collected the result is Eq. 4. Let T — t2 = T = flowing time following short term shut-in t2 — t1 = 6 = duration of short term shut-in Then T - t1 = r + 6 Substitution of these quantities in Eq. 4 gives p(rw ?T) = P,, -When d<<t - AT then d—&apos;-, T— approaches 1. r + AT In this case the effect of the short term shut-in may be ignored. In order to get some idea of the numerical magnitudes involved take the worst possible case, when AT = 0, that is at shut-in. If the error is made less than some maximum acceptable value at this time, it will be less than the maximum for all subsequent times. d + r Let the difference between the value of y-— and 1 be less than 0.1. Then —6+r — 1 < 0.1 S < O.l T Therefore, if the flowing time following a short term shut-in is at least 10 times the duration of the short term shut-in, the error introduced by ignoring the short term shut-in will not exceed 10 per cent of T+AT A T

Jan 1, 1957

By A. R. Troiano, W. J. Barnett

The existence of a grain size stabilization effect on martensitic type reactions as shown by Scheill in an iron-29 pct nickel alloy and by Troiano and Tokich2 in cobalt prompted a review of other martensitic reactions to determine if this stabilization effect is characteristic of all martensite type reactions. Preliminary work has shown this effect to be present both in the gamma —+ epsilon transformation in an iron-20 pct manganese alloy and the martensite transformation of SAE 4140 and 3 pct Cr-O.40 pct steels. No appreciable effect was found in an SAE 2340 steel. The M, temperature for the gamma —+ epsilon transformation was lowered from 380°F to about room temperature by varying the annealing treatment following cold work as shown in Table I. The effect of austenitic grain size on the M, of three steels is shown in Table 2. M, was deter- mined by the conventional quench-temper method of Greninger and Troiano. In a single crystal the change in external shape concomitant with the homogeneous shear mechanism of the martensite transformation can take place unhindered. In a polycrystalline specimen a restraint of these changes must occur at the grain boundaries. Therefore, the smaller the grain size the greater the shear strain energy necessary for the formation of a martensite crystal. The degree of lowering of M, due to the reduction of grain size is much greater when the change in crystal structure involves only a small energy change, such as face-centered cubic —+ close packed hexagonal (that is, CO and Fe-20 pct manganese alloy). The maximum lowering of M, observed in the steels investigated was 50°F with a change in the ASTM austenitic grain size of 1-2 to 8-9. In comparison, the crystal structure change face-centered cubic —+ body-centered tetragonal involves a considerable energy change. There is little doubt that with much finer grain sizes the increased grain boundary restraint would greatly enhance this stabilization effect. The implications of this phenomenon are multifold. Undoubtedly. under certain conditions it offers a method of stabilizing high temperature phases for room temperature

Jan 1, 1949