Search Documents

By Hsun Hu, R. S. Cline

The effect of rate of deformation on texture formatiotz has been studied with cube-oriented single crystals of 3 pct Si-Fe, compressed 80 pct at two widely different rates. Compression at a low rate (crosshead speed, 0.1 ipm) produces large orienta-tional changes and coarse deformation bands, whereas little change in orientation is observed, particularly at the surface of the crystal, when compressed at a high rate (impact velocity, -500 ips). Thin plates of mechanical twins are produced at the high rate, but their contribution to the deformation texture is practically negligible. The crystal deformed at a high rate has a lower hardness and a more uniform dislocation structure and is morc resistant to recrystallization. There is also a dilference in the recrystallization texture between the slowly and rapidly compressed crystals. These results indicate that plastic flow is less "turbulent" at high strain rates than at low strain ratcs. THE behavior of metals under high-speed deformation has been the subject of wide study in the past decade. Most of these works were concerned with mechanical properties during testing at high rates of loading, the resulting structures, and the theoretical aspects of the observed phenomena. The present status of knowledge and research activities in this field has been documented in recent publications.17' There is, however, very little information available in the literature concerning the effect of high-speed deformation on the texture formation in metals. The possibility of an effect of rate of deformation on texture is suggested by the microstructural differences observed between normally and very rapidly deformed samples. Aside from the strong tendency for twinning (and phase transformation in certain metals) in high-speed deformation, the slip-line patterns and the dislocation configurations show marked differences as a result of large differences in the strain rates. For instance, in mild steel in static and dynamic compression, Campbell and co-worker3,4 observed that coarse slip occurred on several systems in specimens deformed at normal strain rates, whereas specimens deformed rapidly showed only fine slip on fewer systems, the slip-line pattern being very similar to that in specimens deformed at low temperatures and at normal strain rates. The absence of multiple slip was also noticed by Dieter5 in shock-loaded nickel. The dislocation configuration in an iron shock loaded at 70 kbar pressure was shown by Leslie et al.6 to be similar to that produced by light rolling at low temperature—the dislocation lines were mostly straight and uniformly distributed. These observations suggest that the mode of deformation at high strain rates is notably different from that at low strain rates. Consequently, the nature and the extent of lattice reori-entation during deformation should also be different. This investigation was undertaken to test this idea. MATERIAL AND METHOD The remaining part of a single-crystal ingot of a 3 pct Si-Fe alloy,* prepared for an earlier investi- gation,1 was used in the present studies. Specimens 9/16 in. square and 0.365 in. thick were cut from this ingot with (001) planes parallel to the square surfaces, and the [loo] and 1010] directions parallel to the edges of the specimen, within ±1 deg. These machined crystals were etched to remove distorted metal, and annealed at 1300°C for 24 hr in a purified helium atmosphere. The annealed crystals had a hardness of 161 Dph. The crystal orientation was rechecked with X-ray back-reflection techniques. In high-strain rate experiments the crystal was compressed in a Dynapak machine to approximately 80 pct reduction in thickness (from 0.360 to 0.070 in. with an impact velocity of -500 ips). Thus, the duration of loading was approximately 6x 10-4 sec, and the average strain rate was about 103 sec-1. As a result of this rapid, severe deformation, the specimen became a roughly circular thin disc. The original shape of the crystal, however, could still be recognized by a diffusely outlined square area in the center of the disc. For compression at a low speed, the crystal was deformed in a universal testing machine at a rate of crosshead movement of 0.1 ipm. Compared with

Jan 1, 1965

By J. E. Steiner, J. M. Hodge, M. A. Orehoski

Ingots of four steels (1045, 1080, Ni-Mo-V, and Ni-Cr-Mo-V) were cast at pressures varying from about 1 to 760 mm of mercury, so as to obtain a range of hydrogen contents in each steel. The susceptibility of these steels to flaking was evaluated by cooling hot-rolled blooms at a range of cooling rates, varying from water quenching to slag cooling, and studying the occurrence of flaking as a function of cooling rate. The steels exhibited differences in susceptibility to flaking, and a tlzreshold value for hydrogen content for each steel was found, below which the susceptibility was very low and above which the susceptibility increased markedly with increasing hydrogen content. Mechanical property evaluations, metallogvaplzic exantinations, and X-my diffraction determinations o,t retained austenite were made on samples from the blooms. A mechanism of flaking which is consistent with the observed behaviors, the effects of hvdrogen content, and the differences in susceptibility among the steels investigated is postulated, based on the results of mechanical property evaluations and metallographic and X-ray diffraction studies. ThE important role of hydrogen as a factor in the occurrence of "flakes" or "hair-line cracks" in steel has now been established by many investigations. These have included investigations by steel producers and forging manufacturers, to whom the prevention of flaking is a very real and serious problem, and many research studies aimed primarily at establishing the mechanism of this behavior Several investigators3 have reported that steels cast at reduced pressures (absolute pressures of about 1 mm) not only are low in hydrogen content but are essentially insensitive to the formation of flakes. These investigations have shown that a general relationship exists between hydrogen content and susceptibility to flaking. The difficulty of obtaining suitable samples of steels of a given composition with the desired range of hydrogen contents, particularly in the relatively large sections in which flaking most commonly occurs, has been the principal obstacle to quantitative evaluations of these relationships. Steels with intermediate hydrogen contents between 1 ppm, and the 4ppm or higher of air-cast steels are generally not available in commercially produced steels. Laboratory procedures, in which the hydrogen is introduced into solid steel samples by heating in a hydrogen atmosphere under pressure, generally involve a limitation in the size of the sample which can be treated, and thus are not representative of the behaviors in large sections. The availability of a facility at the South Chicago Works of U.S. Steel, in which several ingots could be cast from a single heat of steel at total air pressures ranging from 0.25 to 760 mm of Hg, has made it possible to overcome this obstacle and to obtain ingots of a given steel covering the intermediate range of hydrogen contents desired. These ingots, furthermore, are of a size suitable for the production of product with section sizes large enough to be representative of the products in which flaking most commonly occurs in commercial production. In these experiments, this facility was used to study the relationships between hydrogen content and the susceptibility to flaking of four steels, two plain carbon steels at 0.45 and 0.80 pct C and two low-carbon alloy steels representing a medium and a rather high hardenability level. These flaking studies were supplemented by evaluations of mechanical properties and by metallographic studies on samples from the blooms, in order to obtain pertinent information on the factors other than hydrogen content which influence the susceptibilities of these steels to flaking. Based on the results of these tests, a mechanism for the initiation and propagation of the flaking cracks, which is consistent with the observed behaviors, is postulated. MATERIALS AND EXPERIMENTAL WORK Steelmaking and Conversion. To provide material for the investigation, five 29- by 35-in., hot-topped, big-end-up ingots were cast from each of four 80-ton, basic electric-furnace heats of four different steels (a 1045 steel, a 1080 steel, a Ni-Mo-V steel, and a Ni-Cr-Mo-V steel). A double-slag practice was used on all steels and the two plain carbon steels were Si-A1 deoxidized with an addition of 2 lb of A1 per ton, while the Ni-Mo-V and Ni-Cr-Mo-V steels were deoxidized only with silicon and

Jan 1, 1964

By O. N. Carlson, A. L. Eustice

The tensile properties of iodide vanadium were determined as a function of hydrogen concentration. It was shown that the presence of 10 ppm H is sufficient to cause embrittlement of vanadzum over a limited temperature range. The temperature of the observed ductility mininlum is approximately -l00°C, this being a function of strain rate and hydrogen concentration. The yield stress of hydrogenated vanadium is raised sharply in the brittle temperature range. THE present investigation was initiated to determine the effect of hydrogen on the tensile properties of high-purity vanadium as a function of temperature and the role of hydrogen in the brittle-ductile transition. Loomis and Carlson&apos; reported a change from ductile to brittle behavior in iodide vanadium at -110°C with a return of ductility at temperatures below -140°C. Vanadium wire containing 270 pprn of hydrogen was brittle at room temperature while wire containing 100 ppm was ductile under the same conditions. Roberts and Rogers&apos; reported a ductile-brittle-ductile fracture sequence in vanadium containing 400 to 600 ppm of hydrogen. Their data show that hydrogenated vanadium is ductile at 150"C, brittle at room temperature and ductile again at -196°C. Magnusson and Baldwin found a minimum in ductility in vanadium containing 80 ppm hydrogen from tensile tests performed at several strain rates. Information available on the vanadium-hydrogen equilibrium system4 is somewhat incomplete. The absorption of hydrogen is exothermic in accordance with the decline in solubility with rising temperature. The limit of solid solubility at temperatures below 400°C has not been determined due to the difficulties in obtaining equilibrium at these temperatures. Data on the hydrides of vanadium are rather limited. Roberts6 studied the structure of VD,., by neutron diffraction, and observed the ordering of the deuterium atoms below -65°C into a primitive cubic cell with a lattice constant about twice that of the disordered bcc vanadium solid solution. EXPERIMENTAL PROCEDURE Vanadium of 99.9 pct purity was prepared by the iodide refining process6 for use in this investigation. This vanadium contained the following analyzed impurities; <20 pprn calcium, 150 ppm carbon, 200 ppm chromium, 30 ppm copper, 10 ppm hydrogen, 200 ppm iron, <20 ppm magnesium, <20 ppm nickel, <5 ppm nitrogen, 150 ppm oxygen, <50 ppm silicon, and <20 ppm titanium. specimen Preparation—Tensile specimens were machined from swaged rods of iodide vanadium. The

Jan 1, 1962

By Joseph Lindmayer, Karl M. Busen

If silicon is in contact with silicon oxide, a heterojunction is formed which induces an inversion layer ("channel"). Influences of impurities and structural parameters on the channel are discussed. There seems to be no relation between surface damage in silicon and the channel. Phosphorz~s doping of the silicon oxide has no pronounced effect on the channel. Channels which were well above the range predicted by the heterojunction model were related to the action of H* at the silicon/silicon oxide interface. Channels which were well below tlze range predicted by the heterojunc-tion model were related to a transition region in the oxide adjacent to the interface. In support of the assumed transition region and the H+ are the ex-periments described on the paper. Furthermore, the concept of metastable OH groups in silzca, known from other publications, has been used to explain certain channel characteristics by the formation and the accommodation of hydrogen in the silicon oxide. THE electrical characteristics of a semiconductor device in many cases are influenced by the conditions of its surface. Because these conditions often are changing with temperature, time, ambient, or other parameters, numerous attempts have been undertaken to stabilize them by suitable processes. Several years ago Atallal studied the silicon/silicon dioxide interface and since then increasing attention has been given to the silicon dioxide as a means to stabilize silicon surfaces. This led to the evolution of the planar silicon devices where oxide layers are used to protect places at which the junctions are intersecting the surface. The formation of the oxide layers on silicon can be carried out by a variety of processes: high-temperature thermal oxidation (exposing the silicon surface to oxidizing agents such as 02, HzO, COz), evaporation of SiOz, or anodic oxidation of silicon. Originally the oxides were considered able to "passivate" the silicon surface and it was understood that they 1) protected the surface against any influence from the ambient and 2) stabilized surface conditions which were necessary for proper device performance. It soon was realized that the second quality was not always obtainable: Atalla and Tannenbaum showed that, in the process of forming a layer on doped silicon by thermal oxida- tion, electrically active impurities in the silicon near the interface are redistributed, thus giving rise eventually to the formation of unwanted charge distributions. The degradation of planar junctions during reoxidation of silicon was observed by Barson et a. Surface studies with silicon planar junction structures indicated that the breakdown voltage depended on the kind of procedure which was applied during or after oxidation.*" Yamin and worthinge report on charge storage and dielectric properties of SiOz films at elevated temperatures. Another influential parameter on the silicon surface is discussed by Lindmayer and usen,&apos; namely the appearance of a surface potential on silicon as a consequence of the heterojunction formed by the silicon/silicon oxide interface. In the present paper the influence of impurities and structural parameters in silicon oxide in connection with the heterojunction is a particular object of our investigations. I) EXPERIMENTAL The material used was 10 ohm-cm p-type silicon slices with either damaged surfaces (mechanically lapped or polished) or damage-free surfaces (chemically polished). The majority of the slices were oxidized for 1 hr and 20 min at 1250°C in a steam + oxygen atmosphere generated by passing O2 through HzO at 90°C (wet oxygen). This led to oxide layers of about 9000 A. On some slices oxide layers were grown in pure oxygen (dry oxygen). By a photoresist process a set of 20 by 200 mil windows was then etched out from the oxide layer. The pattern of the windows is shown in Fig. 1. In a subsequent process n-type pockets with surface concentrations between 2 x 10" and 5 x 10" cm"3 were formed by phosphorus diffusion. This allowed ohmic contacts between the silicon surface and tungsten probes which for electrical measurements were placed into two adjoining windows and

Jan 1, 1965

By B. D. Cullity, K. S. Sree Harsha

When a copper or aluminum single crystal is swaged into wire, the resulting deformation texture depends on the original orientation of the crystal. The<100> and <111>orientations me essentially stable, while <110> is unstable. The greater the <100> content of the deformation texture, the stronger the wire is in torsion. the greater the<111>content, the stvonger it is in tenszotz. The preferred orientation (texture) of fcc wires, either after deformation or recrystallization, is usually a double fiber texture in which some grains have <100> parallel to the wire axis and others have <111>. The relative amounts of these two texture components, as reported by different investigators for the same metal, vary considerably. Previous work in this laboratory&apos; has shown that the starting texture of a wire, i.e., the texture which it has before deformation, can have a decided influence on the texture produced by deformation. In particular, it was found that the deformation texture of copper wire is essentially a single <100> texture, if the wire before deformation contains only a <100> component. This is true even when the deformation is carried to more than 98 pct reduction in area. This paper reports on further studies of the role played by the starting texture. Copper and aluminum single crystals of various orientations have been cold swaged into wire, and quantitative measurements of the resulting deformation textures have been made. The tensile and torsional properties of the deformed wires have also been measured, and the relation between these properties has been correlated with the texture of the wire. These measurements were made in order to demonstrate that a cold-worked wire can be made relatively strong in torsion and weak in tension, or vice versa, by proper selection of the texture before deformation. MATERIALS The copper was of the tough-pitch variety, containing, by weight, 99.962 pct Cu, 0.003 pct Fe, 0.025 pct 0, and 0.0021 pct Si. The aluminum contained more than 99.99 pct .&apos;41; the only reported impurities were 0.001 pct Fe, 0.001 pct Si, and 0.003 pct Zn, by weight. Large single crystals of these metals were grown by the Bridgman method in graphite crucibles and a helium atmospliere. Cylindrical specimens of predetermined orientation, about 1.5 in. long and 0.36 in. in diameter, were machined from the as-grown crystals and then etched to 0.25 in. to remove the effects of machining. Their orientations were checked by back-reflection Laue photographs, and they were then swaged to a diameter of 0.050 in. (96 pct reduction in area). 111 order to study the "inside texture" of the deformed wires, they were etched, after swaging, to a diameter of 0.040 in. before the texture measurements were made. TEXTURE MEASUREMENTS The fiber texture which exists in wire or rod can be represented by a curve showing the relation between the pole density I, for some selected crystal-lographic plane, and the angle $ between the pole of that plane and the wire axis (fiber axis). Such a curve will show maxima at particular values of , and these values disclose the texture components which are present. The relative amounts of these components can be determined2&apos;3 from the areas under the maxima on a curve of I sin F vs F. It is seldom necesszlry to measure I over the whole range of F from 0 to 90 deg, since the existence of maxima in the low-F relgion can be inferred from measurements confined to the high-F region. The X-ray measurements were made with a General Electric XRD-5 diffractometer and filtered copper radiation, according to one or the other of the following procedures: 1) A method developed in this laboratory,4 involving diffraction from a single piece of wire. 2) A modification of the Field and Merchant method.5 This method was originally devised for the examination of sheet specimens, but it can easily be adapted to the measurement of fiber texture. Three or four short lengths of wire are held in grooves machined in the flat face of a special lucite specimen holder. The axes of the wires are parallel to the plane defined by the incident and diffracted X-ray beams, and the holder to which the wires are attached can be rotated step-wise about the diffractometer axis for measurements at various angles 9.

Jan 1, 1962

By H. E. McCoy, C. J. McHargue

Single crystals of columbium containing various levels of oxygen, nitrogen, carbon, or hydrogen were deformed by slaw compression and impact loading at -196°C. For the slow deformation rates. 1500 to 1900ppm oxygen, 500 pprn nitrogen, or 150 to 200 ppm carbon completely suppressed mechanical twinning. Hydrogen in concentrations to 1000 ppm had no apparent effect on ease of twinning. Impact loading caused twins to form at all levels of contamination studied, and cleavage mucking on (100) planes occurred in specimens having higher interstitial contents. It is well established that, under some conditions, the bcc metals vanadium, columbium, tantalum, chromium, molybdenum, tungsten, and iron deform by formation of mechanical twins. In general it has been accepted that high strain rates, low temperatures, and large grain sizes are conducive to this kind of deformation.&apos; Since interstitial impurity atoms have a relatively large effect on slip in bcc metals, one might sus- pect a similar effect on twinning. This could occur directly by the interstitials affecting the nucleation or growth of the twins or indirectly from their effect on slip. It has often been argued that the large increase in yield stress caused by interstitial elements enables the effective stress to exceed that necessary to cause twinning. The twinning stress has always been thought to be higher than the stress necessary to cause slip in pure metals.2 simonsens and Tipper and Hall4 found profuse twinning in very pure a iron even at room temperature, whereas Low and Fueste15 and Churchman and tottrell&apos; observed twins in a iron containing carbon but none in highly decarburized specimens. McHargue7 reported mechanical twins to form readily in vanadium which contained 390 ppm interstitigs, while Clough and Pavlovic8 found twins only in the one or two grains adjacent to the fracture surface in vanadium which contained 1500 to 1700 ppm interstitials. In columbium, McHargue9 found high-purity single crystals to twin easily whereas Churchman,10 Adams, Roberts, and mallman,11 and Johnson12 studied less pure material and found far fewer twins. Although the results of Leadbetter and Argent13 showed oxygen to inhibit twinning, a study by sheely14 indicated the opposite to be true. Chromium exhibits this mode of deformation15,16 but there are no data regarding impurity effects. Lawley, Van den Sype, and Maddin17 observed twins in very pure molybdenum single crystals deformed at 77°K but none in less pure crystals. The work by Cahn18 and

Jan 1, 1963

By L. J. Demer, W. W. Walker

AMONG several variables of the indentation hardness test is the effect of loading time on the measured hardness of a material. The most generally recommended duration of loading is 10sec.1-3 Although this short loading duration is undoubtedly satisfactory for harder metals, it is questionable that it is sufficient for very soft metals,4 or for certain plastics.5 The effect of duration of loading on the measured hardness of ionic crystals is largely unknown, although Onitsche and Mitsches found that the indentor (under a very low load) had not yet reached equilibrium after 60 sec in a variety of ionic crystals. To extend further the data of Onitsche and Mitsche, a systematic study was made of the effect of duration of loading for longer periods, up to 50,000 sec, on the 136 deg Dph of synthetic lithium fluoride. For comparison, less extensive tests were made on synthetic magnesium oxide and copper single crystals. The general procedure used was to indent a cleaved and polished (100) face of a lithium fluoride crystal using a Leitz Durimet microhardness tester. In certain cases, the indentor was so oriented as to align the diagonals of the 136 deg diamond pyramid indentation parallel to a (100) edge. In other cases the diagonal of the indentation was aligned at 45 deg to a (100) edge of the crystal. After indenting, the surface tested was etched in the acid etchant recommended by Gilman and Johnston.&apos; This revealed the position of dislocation lines which intersected the surface tested. At these points of emergence were formed the usual sharp, pyramidal etch pits as shown in Fig. 1. The specific procedure used in studying duration of loading was as follows. The surface of the crystal was indented, and the load was left on the indentor for a predetermined time. This duration of loading was varied systematically from 1 to 50,000 sec. After the desired amount of time had elapsed, the indentor was removed from the crystal and the size of the resulting indentation was determined using a micrometer eyepiece. The surface was then etched to reveal the dislocation etch pit pattern around the indentation. Procedures used for lithium fluoride, magnesium oxide, and copper were similar. Preliminary studies revealed that, although an indentation was enlarged by increasing loading time, no corresponding changes occurred in the surface etch pit pattern. Since no evidence of dislocation movement was observed on the surface, it was assumed that the plastic deformation indicated by the indentor sinking into the surface was a result of dislocation movement inside the crystal. This movement was studied bf making a series of surface indentations in a crystal corresponding to various lengths of time, cleaving the crystal vertically through the indentations, etching the freshly

Jan 1, 1964

By T. M. Kegley, J. H. Frye, D. L. McElroy, M. L. Picklesimer, E. E. Stansbury

Measurements of rate of growth, thermodymmic quantities, and partitioning of Mn are reported for high-purity eutectoid Fe-C and Fe-C-Mn steels for the auistenite-pearlite reaction. Evaluztion of the cotnstants of a derived-rate equation with these data indicate that Mn additions cause the decrease observed by increasivg the activation energy and apparent activation entropy or the movenment of complexes at the interface. A diffusion-controlled mechanism is not required. Appreciable partitioning of Mn to cementite occurs entirely behind the austenite-pearlite interface for more than 30°C of subcooling. ThE growth of pearlite from austenite has been extensively studied over a period of years. but the rate-controlling step in the process is not known nor is the mechanisrn of the effect of alloying elements understood. A comprehensive review of the subject has been given by Mehl and Hagel. The present study has been concerned with further considerations of the effects of alloying elements on the rate of growth of pearlite. Analytical relatioriships for the rate of growth of pearlite have been considered by Zener.2 Turnbull,3 Machlin,4 Cahn.5 Frye,6 and Frye et a1.7 The derived expressions in each case involve the free energy of the transformation. the interlamellar spacing. and some rate term of the nature of the diffusion coefficient. mobility or frequency of occurrence of some rate-determining step. The treatments. of course. differ in detail. Further consideration is given in the present paper to the equation of Frye, Stansbury. and McElroy which was derived from modified absolute rate theory without in any way specifying the mechanism of the reaction. This equation is r=c ? T ? F exp -?E*/RT [1] where r = rate of growth in mm/sec c = a constant AT = equilibrium transformation temperature minus actual transformation temperature "C ?F = free energy per mol of austenite minus that of pearlite ?E* = activation energy R= gas constant ?F in Eq. [1] is computed from the following: ?F=?H?T/To - ?T&apos; dT&apos; ?T&apos; ?CdT"/T" [2] To To where AH is the enthalpy per mol of austenite minus that of pearlite at the equilibrium temperature, To, and AC is the difference in specific heat between austenite and pearlite per mol. Subsequently, Darken, Fisher, and Galligan8 have reported that a layer of ferrite separates pearlite from austenite at the growing interface (i,e., the advancing face of Fe3C is in contact with ferrite, not austenite) and the ratio of the thickness of this layer to the spacing of the pearlite has not been found to vary with alloy content or transformation temperature. This observation does not require any alteration in the above equation: however, it may require that a different significance be attached to the factor, AT. The AT term appears in Eq. [I] as a consequence of a factor for the amount of growth per complex transformed at the interface.* It is based *This is not the sole reason for expecting such a relationship. See Ref. 7._______________________________________________________ on a l/2 ?T dependence of the interlamellar spacing. Zener2 has developed a theory which leads to Soa 1/?V where So = the pearlite interlamellar spacing. This relation is as consistent with the data on interlamellar spacing as other relationships, although, as pointed out by Mehl and Hagel,1 the data are not sufficiently precise to consider the relationship as confirmed. The work of Darken, Fisher, and Galligan8 requires that the above be written where f = the thickness of ferrite layer between austenite and pearlite. The present paper is concerned with iron-carbon-manganese eutectoid alloys and reports experimental measurements of the following kinds: 1) the effect of

Jan 1, 1961

By M. Cohen, P. E. Beaubien, D. Caplan

Addition of 1 pct Mn to Fe-26 CY ca/(ses a12 increase in scaling rate at 870° and 1090°C. Whereas only the rhombohedral oxide, formrs on tire manganese-free alloy, with manganese present major amounts of the spinel .Mn0 . Pro-duced through which diffusion is more rapid. At 1090°C the scale formed OH Mn is highly, irregular due to blistering and sintering, This is a second cause of faster oxidation Since much of the blistered oxide does not participate as part of the protective layer. The oxide phases that form at different times depend on the degree to which the surface metal has become depleted, in manganese by preferential oxidation. PREVIOUS work has demonstrated that the oxide scales formed on chromium steels at high temperature consist of both spinel and rhombohedra1 phases and may contain considerably more manganese, especially in the spinel phase, than is present initially in the alloy.&apos; It has further been observed that a binary Fe-26 Cr alloy scaled less rapidly than an equivalent commercial steel and formed no spinel.&apos; However, it is not established if manganese in chromium alloys actually promotes formation of spinel nor if it affects oxidation resistance. It seemed expedient to try to evaluate the specific effect of manganese by oxidizing an Fe-26 Cr alloy containing a small manganese addition under the same experimental conditions as used for the manganese-free alloy.&apos; EXPERIMENTAL Table I shows the composition in weight percent of the Fe-26 Cr-1 Mn alloy and of the Fe-26 Cr alloy used as reference. The alloy with manganese was prepared from the vacuum-melted Fe-26 Cr alloy by remelting in vacuum with the proper manganese addition. The ingot was rolled to strip and specimens 1.1 by 0.035 by 5 cm and 1.1 by 0.5 by 4 cm cut out or machined. As previously described2 the specimens were smoothed by abrading through 600-grit Sic paper, electropolished in acetic-perchloric acid (20 to 1) to remove 10 p of metal, annealed at 1100°C in 20 torr of purified argon, and electropolished to remove a further 10 p of metal. They were then either rinsed in water and methanol and blotted dry or immediately etched for 45 sec in deoxy-genated 4 N HC1 under cathodic polarization at 10 pa per sq cm. Oxidation runs were carried out in a vertical tube furnace in a flow of purified oxygen at 1 atm while the weight increase was recorded continuously on a* automatic balance.3 Alter 98 hr at 870.C or 20 hr at 1090"C the specimens were removed to flowing argon and observed microscopically for evidence of spalling as they cooled. Subsequently they were examined by X-ray diffraction, the oxides analyzed chemically, and metallographic cross settions prepared. RESULTS Figs. 1 and 2 show the oxidation curves of Fe-26 Cr-1 Mn at 870&apos; and 1090C. The Fe-26 Cr reference runs are shown as dashed curves. Experimental details of the six runs are included in Table 11. The first 20 hr of run 1 is repeated in Fig. 2 to show the effect of temperature. At 8&apos;70°C the weight gain for Fe-Cr-Mn is about twice that of Fe-Cr (curve 1 vs curve 2, Fig. 1). Both alloys show regular oxidation curves. On parabolic coordinates (right-hand scale of Fig. 1) the

Jan 1, 1965

By J. W. Spretnak, H. R. Ogden, A. G. Imgram

The effect of working reduction, stress-relief annealing, and recrystallized grain size on the tensile and notch-tensile properties of molybdenum and Mo-0.5 Ti was studied. It was found that increasing the amount of working reduction increased the strength and reduced the tendencies toward brittle behavior. Stress -relief annealing did not alter the strength but lowered the transition temperature. Increasing the recrystallized grain size lowered the strength and raised the transition temperature. The notch -unnotch strength ratio began to decrease with decreasing temperature at temperatures below those at which the notched ductility fell to zero for all of the micro structural conditions investigated. BEING bcc metals, the refractory metals in Groups Va and VIa (V, Cb, Ta, and Cr, Mo, W) undergo a ductile-to-brittle transition and exhibit brittle fractures at low temperatures. Previous research has shown that molybdenum and tungsten not only have higher transition temperatures than columbium and tantalum but show a marked sensitivity of transition temperature to structure as well.&apos; Recrystallized materials become brittle at considerably higher temperatures than is the case with material having a worked or fibered micro-structure. Evidence of structure sensitivity has also been detected in columbium- and tantalum-base alloys containing appreciable amounts of molybdenum or tungsten such as F-48 (Cb-15W-5Mo-1Zr) and Ta-10w.2 This paper describes the results of research conducted to further evaluate the effects of metal. lurgical structure on the low-temperature tensile and notch-tensile properties of molybdenum and Mo-0.5Ti. The effects of rolling reduction, stress-relief annealing temperature, and recrystallized grain size were studied. Considerable progress has been made in explaining the fundamental mechanisms of the flow and fracture characteristics of bcc metals. Several of these concepts are helpful in interpreting the data obtained in this study. Brittle fractures occur when the highly temperature-sensitive yield strength equals, or exceeds, the temperature-insensitive brittle cleavage fracture stress.3 The rapid increase in yield strength with decreasing temperature, characteristic of all bcc metals, is therefore of primary concern. petch4 has described the yield strength, ay, (stress required to initiate plastic flow) of bcc metals by the expression where ai is the frictional resistance to slip, Ky is a constant numerically equal to the slope of a a vs curve, and 2d is the average grain diamefer. The oi term itself is thought to be composed of a temperature-insensitive component dependent on impurity content and second-phase particles and a temperature-sensitive component dependent on the Pierls-Nabarro force.5 This equation has also been demonstrated to hold for the flow stress (stress required to maintain continued plastic flow).&apos; Brittle-crack nucleation is thought to occur by the coalescence of dislocations piled up in slip planes at sessile dislocations or twin and grain boundaries.7,8 These proposed mechanisms require that a large number of dislocations be released into the active slip planes so that sufficiently large stress concentrations are produced at the dislocation pile-ups to enable coalescence into a micro-crack. Therefore, at least a microscopic amount of slip must occur in some grains to facilitate initial crack formation. Cottrell 7 has predicted that brittle fracture, nucleated by slip, should occur when where ty is the shear yield strength, ß = 1 in un- notched xension, u is the shear modulus, and y is the effective surface energy. This equation incorporates the Griffith concept of fracture that a crack will propagate if the decrease in strain energy equals or exceeds the increase in surface energy.

Jan 1, 1964

By A. E. Powers

YEARS of experience and research have shown that molybdenum, tungsten, and vanadium are among the most useful and effective elements in augmenting the high-temperature strength of heat-treatable, ferritic steels. Grün, for example, in conducting an early survey of the strengthening effects of various elements in annealed, low-carbon steel at 750° and 930°F (400° and 500°C), found molybdenum and vanadium to be the most effective.&apos; Holtmann comprehensively studied the strengthening effects of molybdenum and vanadium in quenched and tempered steels at 930°F as a function of microstructure and carbon content.&apos; He found creep strength to be related to the saturation of the austenite at the austenitizing temperature, such that strength is increased by alloying until the y loop is reached—the terminus of complete austenitization. Beyond this degree of alloying, the inability to heat treat will result in lower strength, at least for low and moderate creep temperatures. Tungsten, owing to higher cost than molybdenum, has been little used in ferritic high-temperature steels. As a result, very few investigations have been made of its high-temperature strengthening ability, and reliable data on its effectiveness are not available. Grün rated tungsten far less effective than molybdenum at 930°F (500°C) on a weight-percentage basis.&apos; However, Tammann has found that tungsten is as effective as molybdenum in raising the recrystallization temperature of iron. For this reason Smith, in a review of the subject, has advocated a re-examination of the influence of tungsten on the high-temperature strength of steel.&apos; In any study of the effects of alloying additions on high-temperature properties, recognition must be made of the many variables involved, some of which can be controlled but have been ignored in innumerable investigations of the past, and some of which are difficult or even impossible to control. One cannot, for example, assign a definite strengthening index to any one alloying element, for this will be dependent upon the microstructure (heat treatment and mechanical treatment), testing con- ditions (temperature, time, and stress), and the complete composition of the steel (accompanying alloying elements, impurities, etc.). One ever-appear ing variable, hardness or tensile strength at room temperature, may be eliminated by heat treating all of the alloys to a single hardness level. An attempt may be made to control the variable of primary quenched structure by oil or water quenching the test specimen. The control of such variables may not be possible in cases of widely varying compositions, but at least such deviations should be recognized. Material and Procedure The alloys, made as 30-lb ingots in an induction furnace, all had the same base composition of about 0.18 pct C, 0.85 pct Mn, and 0.48 pct Si. They were grouped as follows: Mo Group—Containing up to 5.2 pct Mo. W Group-Containing up to 6.0 pct W. V Group—Containing up to 3.3 pct V. Mo-V Group—Containing up to 2.7 pct Mo and up to 1.4 pct V. Mo-W Group—Containing up to 2.5 pct Mo and up to 2.5 pct W. The steels, listed in Table I, were forged to ¾ in. sq bars and the material for the high-temperature, rupture-test specimens was further swaged to %-in. round bars.

Jan 1, 1957

By R. F. Mehl, R. W. Parcel

THE rate of the decomposition of austenite in eutectoid steels, forming only pearlite, may be expressed fundamentally in terms of the rate of nucleation and the rate of growth of pearlite nodules. This matter has received much attention in recent years. (The basic references are listed in ref. 1; references of particular import to the present work will be given separately.) Systematic data have been developed for plain carbon eutectoid steels.:! Although the physical factors affecting the rate of nucleation, N, in case of plain carbon steels have presumably been identified, they cannot as yet be evaluated quantitatively.&apos; Several somewhat successful attempts, however, have been made to calculate the rate of growth, G. In the latter case the important factors determining the edgewise growth of pearlite have been assumed to be: 1—the interlamellar spacing; 2—the concentration extremes and concentration contours of carbon in austenite in front of the growing pearlite interface; and 3—the rate of diffusion of carbon in austenite.* The only systematic data on N and G for pearlite in alloy eutectoid steels are those for Co-steels;6 there are no data on alloy eutectoid steels in which the alloying element is known to decrease the rate of formation of pearlite, apart from a few scattering data reported earlier&apos; for steels containing manganese. This paper reports a systematic study of N and G for eutectoid alloy steels containing molybdenum and nickel, respectively. In order to provide information of help in understanding the results, new data are presented on the effect of molybdenum on the interlamellar spacing of pearlite, So; data on the effect of nickel on So have already been published.&apos; The alloys used were in five groups: 1—Pure Fe-C-Mo alloys (Table I) were supplied by the Climax Molybdenum Company of Michigan, Detroit. They were made from Armco ingot iron of standard analysis, Acheson graphite, and ferro-molybdenum, melted in silica-lined crucibles in a high frequency induction furnace. The ferromolyb-denum was added to the melted steel, followed in 5 min by 0.08 pct Al. The ingot as-cast measured 23/4x 31/4x10 in.; it was hot-forged at 1150°C to a bar 1 in. sq.

Jan 1, 1953

By L. F. Porter, G. J. Dienes

The effect of netltron irracliation on mart ensite transformation in the iron-nickel system was studied by means of electrical resistance and magnetic induction measuremertts. Irradiation lowers the M8 temperature of the allay, causes changes in the electrical and magnetic properties of the high-temperature phase which indicate either an accelemtion of phase separation or a clustering of nickel, and enhances strain recovery on specimens partially transformed to martensite prior to irradiation. Possible mechanisms to account for the effects; are discussed. 1 HERE is a good deal of evidence to indicate that neutron irradiation is capable of affecting phase changes in metals and alloys. While the mechanism of the effect often is not well understood, such studies hold promise of improving our understanding not only of irradiation damage but also of the phase changes involved. To date systematic and detailed investigations have been made of the influence of nuclear radiation on order-disorder transformations in the copper-gold, nickel-manganese, and copper-zinc system, and on precipitation reactions in the copper-iron, nickel-beryllium, and copper-beryllium system.&apos; These solid-state reactions are diffusion controlled and the results can be explained largely on the basis of enhanced diffusion due to the presence of irradiation produced vacancies. Reversion of low-temperature eutectoid phases to the high-temperature phase has been observed in the fissionable uranium-molybdenum and uranium-niobium systems.2 Here it is necessary and logical to postulate a displacement spike mechanism to account for the effect. The only case of irradiation induced diffusionless transformation in metals is the allo-tropic transformation of p to a tin reported by Flee-man and Dienes. 3 They found that neutron irradiation markedly decreases the incubation time for transformation, but the mechanisnl is not well understood. Another diffusionless transformation which one might expect to be influenced by neutron irradiation is the martensite transformation. This transformation is known to be sensitive to strain, structural hardening, and the presence of imperfections.6,7 Since the transformation is diffusionless, one also

Jan 1, 1960

By C. H. Samans, G. F. Tisinai, J. K. Stanley

The addition of nitrogen (0.10 to 0.20 pct) to Fe-Cr-Ni alloys of simulated commercial purity results in a real displacement of the u phase boundaries to higher chromium contents. The effect is small for the (Y + s)? boundary, but is pronounced for the (y + a +s)/(y + a) boundary. Although there is an indication of an exceptionally large shift of the n boundaries to higher chromium contents, especially in steels with nitrogen over 0.2 pct, the major portion of this apparent shift results from the fact that carbide and nitride precipitation cause "chromium impoverishment" of the matrices. The effect of combined additions of nitrogen and silicon to the Fe-Cr-Ni phase diagram is demonstrated also. Nitrogen can nullify the effect of about 1 pct Si in shifting the (y + o)/? phase boundary to lower values of chromium at all nickel levels from 8 to 20 pct. NItrogen can nullify this U-forming effect of about 2 pct Si at the 8 pct Ni level, but not at the 20 pct Ni level. The alloys studied were in both the cast and the wrought conditions. There are indications that the u phase forms more slowly in the cast alloys than in the wrought alloys if both are in the completely austenitic state. The presence of 6 ferrite in the cast alloys accelerates the formation of U. Cold working increases the rate of o formation in both cast and wrought alloys. THE major improvement in Fe-Cr-Ni austenitic alloys in recent years has been in the addition or removal of minor alloying elements to facilitate better control of corrosion resistance, sensitization, and heat resistance. One shortcoming of the austenitic Fe-Cr-Ni alloys, which never has been completely circumvented, is their propensity toward u formation. In the AISI-type 310 (25 pct Cr-20 pct Ni) and type 309 (25 pct Cr-12 pct Ni) steels, sufficient amounts of u phase can form, if service or treatment is in a suitable temperature range, to cause severe embrittlement. Also, there is a growing conviction that this phase may be contributory to some unexpected decreases in the corrosion resistance of certain of the 18 pct Cr-8 pct Ni-type steels. The present paper discusses the effect of nitrogen additions on the location of the (r+u)/d and the (y+a+u)/(y+a) phase boundaries in the ternary Fe-Cr-Ni system, for cast and wrought alloys of simulated commercial purity, and in similar alloys containing up to about 2.5 pct Si. The objective is to define compositional limits for alloys which will not be susceptible to u formation when used near 1200°F (650°C). An excellent review of the early studies of the u phase in the Fe-Cr-Ni system has been compiled by Foley.1 Rees, Burns, and Cook2 have determined a high purity phase diagram for the ternary system, whereas Nicholson, Samans, and Shortsleeve3 are- stricted themselves to a portion of the simulated commercial-purity phase diagram. Both groups of investigators show almost an identical position for the commercially significant (y+u)/y phase boundary. Further comparison of the work of the two groups indicates that, below the 8 pct Ni level, the commercial alloys have a decidedly greater propensity toward u formation than the high purity alloys. The two groups of workers agreed that both the AISI-type 310 (25 pct Cr-20 pct Ni) and the type 309 (25 pct Cr-12 pct Ni) steels are well within the (y+~) region and that the 18 pct Cr-8 pct Ni-type alloys straddle the U-forming phase boundaries. Nicholson et al.3 showed, in addition, that these boundaries shift toward lower chromium contents if greater than nominal amounts of silicon or molybdenum are added. The effect of nitrogen on the location of the s phase boundaries in the Fe-Cr-Ni system has not been known with any certainty. In 1942, an approach to this problem was made by Krainer and Leoville-Nowak,&apos; but at that time they apparently were unaware of the slow rate of s formation in strain-free samples and aged their samples for insufficient times, e.g., 100 hr at 650°C (1200°F) and 800°C (1470°F). For this reason, it would be expected that their (y+ u) /y boundary would be shifted toward lower chromium contents (restricted ?-field) when equilibrium conditions were approximated more closely. Procedure for Studying the Alloys The alloys used were prepared in the following way: Heats of 200 lb each were melted in an induction furnace. A 5 lb portion of each heat was poured into a ladle containing an aluminum slug for de-

Jan 1, 1955

By C. E. Birchenall, R. H. Condit

THREE surface activity procedures are in com--L mon use for the determination of diffusion coefficients in solids. In the oldest of these&apos; the activity observed at the original surface is compared before and after diffusion, where all the activity is in a thin, plated layer on the surface before diffusion. Partial absorption of the radiations from active atoms which have diffused below the surface reduces the effective activity after diffusion. The formula employed to calculate the diffusion coefficient is where F, is the ratio of final to initial activity at the surface, measured with identical counting geometry; Z=u3Dt, where D is the diffusion coefficient; t is the diffusion time; and u is the absorption coefficient for the radiation in the solid as defined by Lambert&apos;s law. This absorption law implies that all the rays are parallel to each other and normal to the specimen surface. This assumption of a collimated beam used by Steigman, Shockley, and Nix has been adopted by all other investigators and, although it gives results which are approximately correct under restricted physical conditions, it does not represent many actual circumstances. It is the purpose of this paper to show the importance of the divergent rays and to indicate a procedure for taking them into account. In the residual activity method the activity is determined on a series of surfaces ground on the specimen at progressively greater depths in the diffusion layer parallel to the initial surface. Gruzin2 has de- rived a formula for this procedure, which may be put in the form where I is the ratio of the activity measured at a newly ground surface at distance k below the initial surface to the activity at the initial surface measured after diffusion, and K= k/2p Dt. A third procedure involves the autoradio-graphic determination of activity on a plane ground at a small angle to the initial interface through the penetration zone. Aut oradiography can be particularly sensitive to the oblique rays. Unlike most counters which detect essentially all ß particles or X-rays which enter the active volume as a single event, the photographic emulsion may be thin enough so that oblique rays with longer path length in the emulsion produce more darkening than rays which move normal to the film plane. Consider a plane containing a source of radiation uniformly spread over a circle and a point counter located on the perpendicular through the center of the circle. As the counter moves along this line, the total angle 6, at the counter subtended by the specimen will vary. For a fixed counter distance, let 6 be angle between the counter axis and the line connecting the counter to a point P on the active plane. If an absorber of thickness X is interposed parallel to the source between counter and source, the ray from point P to the counter will pass through an absorber thickness X secant 6. Exponential absorption is assumed to hold for all rays. If r, is the ratio of intensity at the counter with absorber to the intensity without absorber and a = pX, 1 00 The behavior of this integral, evaluated numerically,

Jan 1, 1957

By H. R. Ogden, E. S. Bartlett, A. G. Imgram

Notched and unnotched tensile specimens of wrought and recrystallized, oxygmted and hydro-genated tantalum and columbium were tested over a range of temperatures selected to encompass the ductile-to-brittle transition. Increasing the oxygen content of tantalum and columbium increased their strength, and raised their ductility transition temperature, but did not cause a marked deterioration of notch properties below that of commercially pure material. The adverse effects of oxygen were more pronounced for recrystallized structures than for wrought material. The effects of hydrogen are complicated by strain aging or slow strain rate hydrogen embrittlement at -105°F. However, in general, hydrogen had only a minor effect an strength, but markedly increased the effective transition tem-terature and caused a deterioration of notch properties . BCC metals including the refractory metals in Groups Va and VIa (V, Cb, Ta and Cr, Mo, W) can be expected to undergo a ductile-to-brittle transition and exhibit brittle fractures at low temperatures.&apos; Notch embrittlement is associated with this phenomenon, and can be a serious design problem. Table I compares the tensile transition behavior and notch sensitivity of commercially pure Group Va and VIa refractory metals. The Group VIa refractory metals have considerably higher transition temperatures, and become notch sensitive at higher temperatures, than do the Group Va refractory metals. In addition, the Group VIa refractory metals are sensitive to metallurgical structure. Recrystallized structures generally have higher transition temperatures than wrought material of the same composition. The relatively poor transition behavior of the recrystallized structures of these metals is generally attributed to the segregation of interstitial impurity atoms at grain boundaries. Pronounced sensitivity to structure is not evident with commercially pure Group Va refractory metals, however. Since these metals have higher solubilities for interstitial elements than do the Group VIa metals ,4 it can be expected that higher total interstitial concentrations are necessary to develop a comparable degree of grain boundary segregation and to produce a similar sensitivity of transition temperature to structure. Increasing only the concentration of interstitial atoms in solution would tend to increase the flow stress of a metal in relation to its cleavage fracture stress and thereby raise its transition temperature, but would not be expected to cause a marked sensitivity of transition temperature to structure. A research program has been conducted at the Battelle Memorial Institute to evaluate the effects of interstitial oxygen and hydrogen on the transition behavior and notch sensitivity of tantalum and columbium, and to evaluate the effects of structure on "contaminated" material. EXPERIMENTAL PROCEDURES Notched and unnotched tensile specimens of oxygenated and hydrogenated tantalum and columbium with both wrought and recrystallized structures were tested over a range of temperatures selected to encompass the ductile-to-brittle transition Oxygen levels selected for study in both tantalum and columbium were nominally 500 and 1000 ppm, whereas nominal hydrogen contents were 150 and 500 ppm in tantalum and 150 and 300 ppm in columbium. Oxygen and hydrogen additions were made directly to 6-in. lengths of 1/2-in.-diam electron beam melted wrought bar stock, in a Sieverts Apparatus. Subsequent high-temperature annealing was required for homogenization of the oxygenated material, but was not necessary in the case of hydrogenated stock. The oxygenated and hydrogenated bars were then swaged to 1/4-in.-diam rods. Tensile specimen blanks were cut from the 1/4-in.-diam rods and annealed to achieve stress-re-

Jan 1, 1963

By W. A. Backofen, R. L. Fleischer

AN effect of orientation on the creep behavior of aluminum at 4.2 OK has been observed. Stress relaxation was measured in a hard-type tensile device after stopping the drive. From the spring constant of the device the load-time record could be converted to strain as a function of time. The strain sensitivity was 2-10-=, and the usual creep time of 1 min normally allowed a stress drop of about 25 g per sq mm. Further details of the apparatus and sample preparation appear elsewhere.&apos; It should be noted that neither stress nor strain were constant, so that quantitative comparison with current creep theories will not be made. Shear strain (y) as a function of time (t) was consistent with logarithmic behavior: y = A In t, where A is a constant. In some cases t was made as large as lo3 sec. In the results the quantity A divided by t (the resolved shear stress, which is the driving force for creep) is plotted as a function of strain prior to creep. The quantity A/t is thought to indicate qualitatively the ease with which dislocations can move during creep. In Fig. 1 are drawn tensile stress-strain curves and A/T-strain measurements for single crystals of three orientations: l(a) and l(b) for the maximum resolved shear-stress orientation, l(c) for an orientation 2 deg from [211], and l(d) for a [111] stress axis. The following observations were made: 1) At low strains the A/t values are nearly the same for the three orientations. 2) As strains increase, A/t drops most rapidly for [Ill.] and less rapidly for [211], while for the maximum shear-stress orientation it is constant during easy glide and drops abruptly at the termination of easy glide. 3) In stage 11, A/t values are again approximately the same for the three orientations. These observations are regarded as evidence that 1) the initiation of slip produces obstacles of essentially the same effectiveness for all of these orientations, while 2) the rate at which slip obstacles are produced before stage I1 is fully developed is strongly controlled by the amount of slip on more than one system. This second statement is consis-

Jan 1, 1961

By C. G. Dunn, E. F. Koch, K. T. Aust

Strain-induced grain boundary migration was studied in a series of silicon-iron (3% pct Si) bicrystals of controlled orientations. Each bicrystal was given a 7 pct cold-rolling strain rind annealed at 1100°C. The direction of boundary migration was found to depend on orientation; results for several orientations related by a common [110] axis are given. Analysis of the micrographic and X-ray data is made in terms of the residual strain-energy densities within the crystals (energy per unit volurne) and a driving force on the boundary due to an energy density difference across the boundary. Relative strain energies were obtained for the various orientations investigated. A CHROME-acetic acid electroetching method for revealing dislocation sites in silicon iron was previously applied to clarify the structural features and the energetics involved in strain-induced grain boundary migration.&apos;,&apos; The present paper is a continuation of this study, but is primarily concerned with the influence of crystal orientation. Sperry, as reported by Beck,3 attempted to determine the factors that influence the preferred growth of one grain over another in aluminum due to strain-induced boundary migration. He found that the particular grain of a pair which work hardened more was invaded, although exceptions to this correlation were noted. The index of work hardening was determined from microhardness measurements of each grain before and after cold-rolling. In the present study, orientation itself was taken as a useful variable for studying the behavior of individual grains in a bicrystal and the direction of boundary migration, since the extent of work hardening is orientation dePendent." The effect of strain on boundary migration is evident from a consideration of driving forces. Thus, in the absence of inclusions, the driving force (per unit area of boundary) for growth of grain A into grain B when a bicrystal is strained and annealed may be expressed as follows:2

Jan 1, 1960

By J. D. Livingston, Bruce Chalmers, R. L. Fleischer, R. S. Davis

INTERACTION of dislocations on intersecting slip planes has long been considered a primary cause of work hardening in metals. Easy glide and low work-hardening rates have been observed experimentally only when single-crystal specimens were oriented so that one slip plane was initially favored. When single-crystal specimens have been oriented so that initially two or more slip planes were equally favored, high initial work-hardening rates have been observed. In a polycrsstalline aggregate, glide on five systems is in general necessary for a crystal to deform in conformity with the change in the shape imposed by the other crystals.* Only a small fraction of the crystals of a polycrystal will be oriented so that two

Jan 1, 1958

By Jei Y. Choi, Paul G. Shewmon

The surface self-diffusion coefficient of copper (D,) has been measured between 847° and 1069 "C for six different orientations. These were the(111), (110, (100, and three higher index surfaces. The activation energy for Ds (designated Q s) was found to be about 49 kcal per mol for all six surfaces, and Do about 2 x 104 sq cm per sec. At any temperature Ds varied by no more than a factor of three over these orientations. It is shown that, if the free energy of a surface atom is uniquely determined by its number of nearest neighbors, it follows from the Principle of microscopic reversibility that Qs should have the same value for all surface orientations, and Ds should vary little with orientation. This model also suggests that for clean fee metals Qs ~ 2/3 AH, (heat of vaporization). This is true for copper. ALTHOUGH it has been appreciated for several decades that atoms can diffuse more rapidly on a surface than through the bulk of a crystal, it has only been in the last few years that reliable values of the surface self-diffusion coefficient (Ds) have become available. Tracer studies of Ds had been attempted prior to this period, but when a tracer is placed on a surface, an ever increasing fraction of it is drained off into the lattice. The correction for this loss involves a very difficult, and as yet unperformed calculation. Those who have worked with tracers have not corrected for this loss.1, 2 Thus their results indicate that Ds is greater than the self-diffusion coefficient in the lattice (Dl), but it has not been established that they give quantitative data on Ds. A procedure which avoids the problem of tracer loss is to study the rate of mass-transfer under the effect of surface tension. If the surface asperity being studied is very small, the mass transfer occurs entirely by surface diffusion. The kinetics at which a grain boundary groove forms on an initially plane surface is a well-studied case of this type. The smoothing of a slight scratch in an otherwise flat surface is another procedure that has been studied. If these grooves are up to 20 to 30 µ in width, the dominant mechanism for mass transfer is surface diffusion (at least in the case of metals with low vapor pressures), and the widths can easily be measured with an interference microscope. Of these two, mass-transfer techniques only in the case of grain boundary grooving has a rigorous mathematical treatment been given. This was done by Mullins.3,4 His analysis predicted that in the case of copper in an atmosphere of an inert gas, surface diffusion should be the dominant transport mechanism. This analysis gave an equation for the groove profile and predicted that the width of the groove would increase as (time)1/4. Mullins and Shewmon showed that both of these predictions agreed with experiments.5 Thus the validity of the values of Ds given by this procedure seems to be well established. Gjostein has used copper bicrystals and the grain boundary grooving technique to determine Ds and the activation energy for surface selfdiffusion (9,) in the [001] direction on surfaces ranging between the (100) and (110) planes.= He reported that Qs = 41 kcal per mole and Do = 6.5 x 102 sq cm per sec for all orientations studied. Since the results did not change with the dew-point of the dry hydrogen atmosphere or the type of refractory tube used, he concluded that the surfaces were clean, or at least that the results were not influenced by any impurities chemisorbed from the atmosphere. The work reported here reproduces and extends Gjostein&apos;s study in that D s and Q s were determined for copper over a wider range of orientations. To study the effects of impurities, two purities of copper were used as well as cathodic etching to remove any possible electropolishing film. Gjostein postulated that the diffusing atoms on a surface near a low index plane are the few atoms which are adsorbed on the smooth region between ledges or steps in the surface. A more rigorous derivation of the equation relating Ds to the concentration and jump frequency of these adsorbed atoms is given here. Using this treatment, our empirical observation that Q s and D s are essentially the same for all surface orientations can be shown to follow from the assumption that the free energy of a surface atom is uniquely determined by its number of nearest neighbors. The studies of D s using the scratch technique have been carried out by Blakely and Mukura on nickel,&apos; and by Geguzin and Oveharenko on copper. The latter study using copper gives values of D s roughly

Jan 1, 1962