Вы находитесь на странице: 1из 11

Mechanics of Materials 36 (2004) 11431153 www.elsevier.

com/locate/mechmat

Absence of yield points in iron on strain reversal after aging, and the Bauschinger overshoot q
Robert A. Elliot 1, Egon Orowan 2, Teruyoshi Udoguchi 3, Ali S. Argon
Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139-4307, USA Received 24 July 2003

Editors abstract As discussed in the Postscript at the end of the paper, the paper emerged in the mid-1990s as Professors Argon and Nabarro were preparing Orowans biographical memoir for the Britain Royal Society and the US National Academy of Science. In addition to their review of the paper, an external review has also been secured. Based on these, the editor is pleased to bring this work to the attention of readers of Mechanics of Materials. (Comments by Editor-in-Chief, Mechanics of Materials). Abstract The yield phenomenon is absent in low carbon steel if the direction of straining after full aging is opposite to that just before aging. That this is not the consequence of macroscopic residual stresses is demonstrated by torsion experiments with thin-walled tubular specimens provided with a shallow circumferential groove, in which yield takes place by widening of a single circumferential Lders band with the elimination of signicant internal stresses. u If the straining is interrupted after partial yielding of the annealed specimen and then stress is re-applied immediately without aging, the consequences are opposite. Yielding continues at the normal level if the direction of straining is not changed; if it is reversed, there is a normal Bauschinger eect in the previously yielded volume, but the Bauschinger stressstrain curve rises without discontinuity considerably above the former yield level. When the Lders band begins u to widen, the stress drops abruptly from the overshooting Bauschinger curve to the normal yield level and remains constant until the end of the yield. Repeated strain aging with straining in the same direction gives yield points rising with the total strain roughly as the strain hardening rises with the strain along the uninterrupted stressstrain curve. After each yielding the rate of strain hardening is roughly the same as it is in uninterrupted straining at the same strain, but the curve is displaced to higher

q *

See Postscript, Section 9. Corresponding author. Tel.: +1-617-2532217; fax: +1-617-2588742. E-mail address: argon@mit.edu (A.S. Argon). 1 1906 N. Prairie Lane, Raymond, MO 64083-9509, USA. 2 Deceased, 1989. 3 Formerly of the Department of Mechanical Engineering, Tokyo University, Bunkyo-ku, Tokyo, Japan, deceased, 1996.

0167-6636/$ - see front matter 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.mechmat.2004.04.001

1144

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

stress levels which rise with the number of aging treatments. The permanent strain-age-hardening may reach 30% or more of the normal yield stress at the same strain. 2004 Elsevier Ltd. All rights reserved.

1. Introduction If an annealed low carbon steel is strained beyond the yield point and then aged (e.g., at 100 or 200 C for 1 h), upper and lower yield points appear again, provided that the straining is continued in the same direction; this is the phenomenon of strain aging. Tipper (1952) observed that the yield phenomenon was completely absent after straining and full aging if the specimen was strained in the direction opposite to that before aging. She extended large rectangular specimens of hot rolled steel plate, machined small compression specimens from them, and compressed these, after aging, in the direction of the preceding extension. She concluded that the absence of both yield points contradicted the explanation of the yield phenomenon as a consequence of locking of dislocations by foreign atoms, and that the Bauschinger eect observed in the compression test could not account for the absence of yield. Polakowski (1951, 1952) found that the yield phenomenon was absent in tensile tests on low and medium carbon steels after cold drawing, temperrolling, surface rollings or torsion followed by aging. He believed that the absence of yield points was due to macroscopic residual stresses. The argument was that yielding would begin in a very thin layer in which the residual stress in the reverse direction was highest, and it would propagate to layers of lower residual stress. Since at any time only an extremely small volume would yield, the stress drop would not be observable. In fact, however, the same argument could also be applied to the case of straining in the same direction as before aging; besides, residual stresses could hide only the upper yield point but not the lower yield point and the yield itself. They would not exceed in iron the order of magnitude of 103 times Youngs modulus, so that a plastic strain of 103 would eectively wipe them out; but in reality yielding

continues usually over a strain range of 13%. A direct refutation of the argument is the fact that the annealed or aged wires show a sharp yield in bending in spite of the non-uniform distribution of stress in the elastic range. In a discussion with Tipper, Polakowski (1953) contended that the absence of the yield phenomenon after reversal could also be explained by the presence of the Bauschinger eect which he attributed to microscopic residual stresses, as it was usual at the time. But the previous argument would also apply to this case; besides, its use implied that yielding would occur at any moment in a number of isolated microscopically small regions, instead of by propagation of Lders bands. X-ray u line broadening in cold worked metals shows that the center of gravity of the residual stress distribution curve is not much above the macroscopic yield stress reached after straining; consequently, the microscopic residual stresses should be also wiped out largely (and replaced by dierent residual stresses) by a plastic strain of the order of 0.1%. That the Bauschinger eect is strong over a strain range of 13% is one of the indications that the microscopic back stress is not its main cause (Orowan, 1959). The present study arose from a few casual experiments carried out in 1959 in order to clarify some properties of strain aging. In subsequent work in 19591960 the absence of the yield phenomenon in reverse straining after aging was observed and the previous work of Tipper and Polakowski was found. Although, for the reason given, macro-residual stresses did not seem likely to provide the explanation, it was attempted to eliminate them by torsion tests on thin-walled tubes. The use of pine-rosin coating showed, however, that this was not achieved because yielding always began with the formation of 24 thin axial Lders bands (Fig. 7) followed by ciru cumferential bands. Since the ends of the gauge section were constrained by the thick end sections

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

1145

of the tube, the axial bands must have caused considerable internal stresses. In further work, this diculty was overcome by the machining of a shallow rounded circumferential groove in the gauge section. Now the rst circumferential Lders band arose along this groove, which acted u by reducing the cross-section rather than causing a stress concentration, and yielding continued by the widening of this band; no axial Lders band was u formed, and therefore no signicant macroscopic internal stress could arise. Since the yield phenomenon was still absent after aging and strain reversal, the eect was an inherent physical characteristic of the yield phenomenon, not a trivial consequence of internal stress. As will become clear from the present experiments, neither the Bauschinger eect as such, nor microscopic residual stresses, can account for the absence of the yield points: a detailed scrutiny of its possible mechanisms leads to the conclusion that it is one of the several properties of the yield phenomenon in iron that resists attempts of explanation on the basis of the theory of dislocation locking by impurity atoms. An alternative interpretation of the yield phenomenon which seems capable of accounting for its known properties will emerge from the present study.

Fig. 1. Thin-walled tubular torsion specimen (Orowan, 1949). Groove G initiates a circumferential Lders band and prevents u axial bands.

2. Experimental details All experiments were carried out in torsion; all but a few with thin-walled tubular specimens of the design shown in Fig. 1 (Orowan, 1949). This kind of specimen reduces non-uniformities of stress and strain distribution to a minimum and permits easy and quick reversal of the stress. The specimens were machined from thick-walled tubes. Both killed and rimming steels were used; analyses are given in Table 1. No. 1 in the Table is the

composition of a seamless cold drawn tube of rimming steel from which the specimens with a wall thickness of 1 mm were made; no. 2 that of a killed steel, obtained in the form of a cold drawn welded tube and used for the thin-walled tubular specimens. Nos. 3 and 4 were cold drawn rods from which the few solid-rod specimens were machined. All specimens were vacuum annealed slightly above the normalizing temperature and furnace cooled. After this the mean grain diameter was between 20 and 40 lm. The dimensions of the tubular specimens from rimming steel are given in Fig. 1. The dimensions of the killed steel specimens diered slightly. The wall thickness of the gauge length, accordingly, was about 1 mm in the rst case and about 0.63 in the second, and the ratio of wall thickness to diameter was between 1/20 and 1/30 respectively. The diameter of the gauge section of the solid-rod specimens was 15.9 mm, and the eective length of the section was slightly over 31.8 mm; the other dimensions were as given in Fig. 1. The torsion-testing machine resembled a lathe. One end of the specimen was gripped in a tailstock, the other in a headstock; both could be moved along the bed, adjusted transversely, and their axes could be aligned before they were bolted to the bed of the machine. The grips in the headstock could be rotated by a large hand-wheel

Table 1 Material no. 1 2 3 4 %C 0.017 0.2 0.17 0.13 %Mn 0.44 0.42 0.55 0.57 %P 0.004 0.005 0.009 0.001 %O 0.003 %Si 0.15

1146

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

through a worm-gear; the stem of the tailstockgrips was supported at two points, one keyed to the tailstock, the other free to rotate in a ball bearing. Between the two points, a set of four wire resistance strain gauges were cemented to the stem for measuring the torque. The torque was measured with a standard resistance-gauge box. The twist was measured by a large graduated circle subdivided into 0.25; it was clamped to the one thick-walled part of the specimen close to the gauge section by means of three sharp-pointed setscrews. A vernier was clamped in the same way to the other thick-walled part of the specimen near the gauge section; its arm was a soft lamella spring which pressed the vernier to the graduated circle. One scale division of the vernier amounted to 1.50 . Later the circle was replaced by a circular arc of about 240 angular range which could be slipped upon the specimen and removed from it while the specimen was gripped in the machine. The grips were provided with sharp teeth; each consisted of two halves pressed together by four bolts. Steel plugs were put into each end of the specimen to resist the pressure of the grips. One end of the grips held the specimen; the other sat loosely on the stem of the tailstock or headstock which was provided with a slightly oval transverse hole for the pin that connected it with the grips. Thus the grips were not rigidly xed to the stem, and bending moments as well as axial forces were avoided. Since the specimen could be put into the grips and removed from the machine without moving the headstock or tailstock, by pulling out the connecting pins and taking out the bolts, the machine could be very carefully aligned by special jigs in the beginning and then left undisturbed. The alignment was so accurate that shear strains up to 40% could be obtained before torsional buckling started. In addition to other advantages of the specimen and of the torsion test, therefore, a strain range could be surveyed far beyond what would have been possible in tensile or compressive tests. In the nal series of the experiments the usual procedure was to give a certain increment of twist by the hand-wheel and then read the torque within 2 min. In this time any transient creep quieted down suciently for the measurement of the torque.

3. Rise of the yield points with strain hardening, and the permanent strain-age hardening Fig. 2 shows two shear-stressshear-strain curves from the earlier series (19591960) of experiments, obtained with tubular specimens of rimming steel, of about 1 mm wall thickness in the gauge section. One is the curve of an uninterrupted torsion test carried beyond the beginning of torsional buckling; the other experiment was interrupted four times and continued in the same direction after aging at 120 C for 1 h. The conspicuous feature of the interrupted curve is that the upper and lower yield points rise, more or less as the strain hardening rises along the uninterrupted curve; yield points and strain hardening are roughly additive. This is signicant because in the theory of the yield phenomenon based on dislocation locking by foreign atoms the upper yield point is interpreted as the stress required for unlocking the dislocations, or at least some dislocation segments; if stress concentrations play a role, the unlocking stress is the applied stress multiplied by a stress concentration factor. In neither case could the level of strain hardening inuence the yield point. When the stress is removed, all dislocations move to positions in which the local residual stress and the dislocation line tension are in equilibrium with the dislocation friction stress (PeierlsNabarro stress), and strain aging would lock them in this position. Since they are unlocked after a displacement of a few nm, long before they could approach the stress peaks

Fig. 2. Shear stressstrain curves of two specimens of rimming steel. One curve with uninterrupted straining; the other was interrupted 4 times and continued in the same direction after aging. Yield points and the level of the curve after yielding rise with increasing strain.

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

1147

that determine the level of strain hardening, the upper yield point could not depend on strain hardening. Thus, unless another possibility has been overlooked, the rise of the yield points with strain hardening cannot be explained on the basis of the theory of dislocation locking. Another feature of the interrupted curve in Fig. 2 is that, after each yield event, the curve rises roughly parallel to the uninterrupted curve, but is at a higher level; with the level dierence increasing with the number of interruptions for aging. The curve may rise 30% or more above the uninterrupted curve at the same strain. In the rst versions of the theory of dislocation locking it was assumed that practically all dislocations were unlocked during yielding; the stress would then return to the basic (uninterrupted) curve after the end of the yield. This is not the case, and so it might be assumed instead that only a few dislocation segments are unlocked; yield would occur by a multiplication avalanche released by these segments while most of the dislocation network would remain locked. Alternatively, the ad hoc assumption could be made that strain aging also precipitates hard particles, in the sense of the theory of the yield phenomenon of Edwards et al. (1943). However, the permanent strain-age hardening eect can be explained without additional assumptions if the theory of dislocation locking is abandoned and a point of view of obstacle-hardening is adopted. Permanent strain-age hardening is bound to occur in warm-rolling; it is likely to account for some of its characteristic eects. The additional hardening should be at least partly responsible for the embrittlement of steel rolled at an unduly low temperature in the nishing passes. Fig. 3 shows the well-known behavior in unloading and reloading of materials not showing a yield phenomenon. The curve is not quite trivial in that it has been obtained with the material used in the experiments of Fig. 2 after nearly complete decarburization, and because the specimen was given the same aging treatment between each unloading and reloading. The rst yield is normal; after strain aging, however, only a trace of a yield point is visible when the re-loading has the same direction as the preceding loading (4th6th incre-

Fig. 3. As Fig. 2, but the specimen was largely decarburized. Strain reversal produces normal Bauschinger eect.

ments of twist). If the straining direction is opposite, there is a large Bauschinger eect. The Bauschinger curve does not join the basic curve but runs at a lower level when it has become nearly parallel to it. Each reversal, therefore, causes a permanent softening; this may be interpreted as a deduction from the total absolute plastic strain as far as the eect on strain hardening is concerned. It is interesting to note that Taylors theory of strain hardening (Taylor, 1934) represented an extreme case of this. The Taylor model should give no Bauschinger eect, but, taken literally, it would lead to the removal by the reverse deformation of all dislocations put into the material by the rst deformation. By the time the plastic strain has been cancelled by reverse deformation, the strain hardening would have disappeared. The actual mechanism of how a part of the plastic strain may be cancelled as a source of hardening by a reverse strain is an integral part of the Bauschinger eect, which is not a subject of discussion here.

4. The strain reversal eect after aging Fig. 4 shows, in addition to another basic (uninterrupted) stressstrain curve reproduced for comparison, an experiment that diers from the one shown in Fig. 2 only in that the strains given after aging vary in sign. Those of the same sign as the initial straining are shown by full curves, those of the opposite sign by dashed curves. Positive and negative stresses and strains are plotted in the same direction for ready comparison. The yield phenomenon is identical with that seen in Fig. 2

1148

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

Fig. 4. As Fig. 2 but the sign of straining was opposite to that of the initial straining where the line is dashed. No yield phenomenon after aging whenever the straining is reversed with respect to the straining just before aging.

whenever the sign of the strain after aging is the same as that immediately before aging; upper and lower yield points are present. If, however, the signs are opposite, there is no trace of any yield phenomenon; a Bauschinger curve of the usual type is observed. Close examination shows that the curve coincides with the elastic line longer than in the case of the decarburized material, but it diverges from it continuously and gradually. Dozens of similar experiments have been carried out with identical results. Only the signs of the straining just before and just after the aging treatment matter; the earlier past of the specimen has no inuence. Naturally, the strain before aging must lose its inuence as it is reduced to very small amounts; what happens in this case is seen in Fig. 5. As in all other gures, open circles indicate aging treatment. After the rst aging and strain reversal, another aging was given after a Bauschinger strain of less than 0.4%; continuation of straining in the same direction has produced a distinct though small yield; without a perceptible upper yield point. In the next straining increment (full lines) the specimen was aged at a higher stress on the Bauschinger curve; the result was the appearance of an upper yield point. Next, aging and reversal were followed by a Bauschinger strain of about 0.6%, aging, and another strain reversal. The strain of 0.6% could destroy the yield points, but the Bauschinger curve branched o the elastic line at a higher stress, with two distinct changes of curvature. After this again, aging and reversal followed, and another aging and reversal after a Bauschinger strain of only about 0.25%. Even this small amount of strain could destroy the yield

Fig. 5. Eect of very small strains before aging and reversal. Rimming steel.

phenomenon, but the curve rose almost without Bauschinger strain and bent over into the continuation of the preceding stressstrain curve of the same sign with a curvature almost as sharp as if there had been no aging between unloading and re-loading. Fig. 6 shows an experiment in which aging was carried out without unloading by placing a heatinsulating box around the specimen in the testing machine and heating it by four powerful light bulbs in the box. There is no obvious dierence between these curves and those obtained with aging after unloading. Figs. 26 show experiments with specimens not provided with the circumferential groove in Fig. 1;

Fig. 6. Aging under load, with and without subsequent stress removal. Rimming steel.

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

1149

they were made in an earlier phase of the study (19591960).

5. The Lders band pattern u With a material that does not display a yield phenomenon the thin-walled torsion specimen of Fig. 1 would give a uniformity of stress and strain sucient for most purposes. With annealed mild steel, however, this cannot be taken for granted. In the elastic region the shear stress is highest in the planes normal to the specimen axis and in the planes containing the axis; Lders bands may be u expected to develop either in the rst kind of planes as circumferential bands, or in the axial planes as strips parallel to the axis. Circumferential bands would not introduce disturbing internal stresses; axial bands, however, would do so because the strips of undeformed material between them are constrained by the thick-walled ends of the specimen. Coating with pine-rosin revealed that yielding began in the tubular specimen always with the appearance of 24 parallel axial bands grouped to one side of the specimen; they are seen in Fig. 7, embedded in an area in which subsequently yield occurred by the widening of a circumferential band.

Although, for the reason given in Section 1, that macroscopic residual stresses should not account for the absence of the yield phenomenon on strain reversal after aging, it is dicult to assess their eects. Finally, a way was found to eliminate the initial axial bands, by machining a shallow circumferential groove in the surface of the gauge section, as indicated in Fig. 1. As mentioned in Section 1, yielding then took place by the formation of a circumferential Lders band along the u groove; the band spread in a regular manner towards the opposite ends of the gauge length. The rosin-covered surface of such a specimen is shown in Fig. 8. The slope of the cracks in the resin shows the direction of twisting; it changes by roughly 90 where the direction was reversed. It is of interest that the areas representing individual increments of twist are separated by thin dark lines even if the twist is not reversed. The cause of this is not clear; however, the Bauschinger overshoot, discussed in Section 7 below indicates that the lower yield point is determined by microscopic internal stresses exerted by the yielded material at the Lders band u boundary upon the neighboring undeformed material, and the rapid relaxation of stress during the two minutes of transient creep between straining and reading the torque shows that a considerable decrease of these internal stresses must occur between two strain increments. The boundary between two consecutive portions of the Lders u

Fig. 7. Lder bands indicated by cracking of pine rosin coating. u Axis of specimen horizontal; entire gauge section (2.5 cm) shown. Yielding begins with the formation of axial L ders u bands (all four bands visible in the gure), and continues with circumferential bands. Specimen without groove.

Fig. 8. L ders bands on specimen with groove G (Fig. 1). No u axial bands.

1150

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

Fig. 9. Lders bands pattern on a solid-rod torsion specimen. u No circumferential band.

band, therefore, has been deformed under conditions dierent from those in the interior of the bands. Fig. 9 shows the Lders band pattern on a sou lid-rod specimen. All bands are axial; no circumferential band has been observed on solid-rod specimens. The yielded volumes seem to be wedges bounded by planes containing the axis of the rod. The stressstrain curves of solid-rod specimens are more irregular than those of tubular specimens, but they show substantially the same features. After the rst upper yield point the stress decreases more gradually towards the lower yield point. After aging within the yield range and stress reversal a minor peak was observed on the Bauschinger curve in one case, followed by continued drop of the stress; the peak, however, was lower than the lowest stress reached before unloading and aging. Since the specimen was of killed steel, the peak might have been a Bauschinger overshoot (see Section 7).

6. Yield behavior after strain aging in the yield range In the earlier series of experiments (19591960) the rst straining extended usually beyond the yield range into the region of strain hardening. In the later series (1967) however, particular attention was given to the phenomena occurring when

yielding was interrupted, so that only a part of the gauge length was covered by the Lders band. u Such experiments were less promising so long as the initial axial bands extending over the whole gauge length could not be avoided. Even after the introduction of the shallow circumferential groove particular care had to be taken to ensure that the boundaries of the circumferential band were circles perpendicular to the axis; obliqueness of the boundaries which must have produced considerable internal stresses arose, for instance, from onesided heating by a 75 W lamp at about 20 cm from the specimen, used for observing the propagation of the Lders bands. However, such disturbances u could be avoided. The interruption of the propagation of Lders u bands followed by aging and by re-loading in the same direction was studied on wires in tension by Sylwestrowicz and Hall (1951) and by Hall (1951). They found that a new upper yield point had to be overcome before the propagation of the band started again. Since the state of stress in their experiments was very uniform, the new upper yield point could be extremely high; after full aging it could be so high that on reloading a new Lders u band started in one of the grips where the shear stress was increased by the transverse pressure of the grips, instead of the propagation of the rst band. The new band propagated at a higher stress in the strain-hardened old band; when the boundary of the old band was reached, propagation continued across its boundary with a sudden drop of stress from the lower yield point of the strainhardened material to that of the virgin material. Naturally, no upper yield point appeared in this process. Since in tensile tests on wires the stress could not be reversed, the behavior of the specimen after aging and strain reversal could not be observed. This could be done, however, in the present torsion tests. It was conrmed that an upper yield point had to be overcome for re-starting the propagation of the Lders band after aging u (Fig. 10) between cycles 1 and 2, but no trace of a yield phenomenon was present when the stress was reversed after aging (Fig. 11). Naturally, after the new yield upper point was reached yielding continued at the same level as before if the direction of

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

1151

in the Lders band. This was recognized by the u absence of any cracking in the rosin coating outside the Lders band during the Bauschinger u deformation, and the absence of any extension of the cracks at the margins of the band.

7. The Bauschinger overshoot To complete the picture, experiments were carried out in the yield range without strain aging. The specimen was unloaded when only a part of the gauge length was covered with a broad circumferential Lders band, and then immediately u re-loaded either in the same or in the opposite direction. Figs. 12 and 13 show the results. When the specimen was re-loaded in the same direction, the stress rose with very little non-elastic deformation and the curve bent into the previous yield with a sharp curvature, as usual (not shown in gure). However, when the specimen was reloaded in the opposite direction, the curve, showing the usual Bauschinger eect, did not bend into the horizontal at the yield level; it went on without any discontinuity until the stress dropped suddenly to the normal level yield point and continued at this level until the end of the yield (Fig. 12, cycle 2). The Bauschinger deformation was conned to the Lders band; the band started to widen when the u

Fig. 10. Upper yield point after aging within the yield range. Subsequent reversal without aging shows Bauschinger overshoot. Killed steel.

Fig. 11. Absence of yield phenomenon after aging followed by reversal within the yield range. Rimming steel.

straining was not changed (between cycles 3 and 4). When the direction was reversed after aging, the usual Bauschinger eect appeared. As would be expected, the entire Bauschinger strain developed

Fig. 12. Reversal without aging in the yield range: Bauschinger overshoot. After reversal, aging followed by another reversal. Killed steel.

1152

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

Fig. 13. Reversal in the yield range followed by another reversal before the stress rose to the lower yield point produced a strong Bauschinger overshoot. Third reversal gave another overshoot (second dashed curve). No strain aging; rimming steel.

stress dropped to the lower yield point (cycles 2 and 3). The simplest explanation of the overshooting of the yield level by the Bauschinger curve is that the lower yield point is determined by the microscopic internal stresses exerted by the yielded grains at the Lders band boundary upon the undeformed u grains just beyond the boundary. After reversal, the Bauschinger strain should almost annihilate the internal stresses of the back-stress type, and so a higher stress must be applied for re-starting the propagation of the band in the opposite direction. Fig. 13 shows an experiment that provides some support for this picture. After the twist and unloading the torque was reversed (dashed line), but it was removed before the lower yield point was reached. Then, the torque was reversed again. The following strong overshoot showed that the preceding reverse loading and the small reverse strain was sucient to produce the desired eect of removing the favorable back stresses at the band border, although the torque did not rise to the yield level; the overshoot, therefore, was caused by the new Bauschinger strain and its eect on the Lders band. u If the Bauschinger overshoot is a consequence of internal stresses exerted by the Lders band u upon the deformed material beyond its immediate boundaries, why is it absent after strain aging? As

mentioned in Section 5, the thin lines between areas of yielding corresponding to consecutive deformations separated by the interval of 2 min probably indicate that the internal stresses relax signicantly at room temperature during this short time. Thus, stress relief during aging at 150 C for 1 h 4 may then remove much of the internal stress at the border. According to X-ray data a much higher temperature would be needed for nearly complete stress relief; but in this case the stresses are relieved also by carbon or nitrogen atoms moving into volumes of hydrostatic tension, according to the principle of Cottrell. Even if the internal stresses opposing the propagation of the Lders band are greatly reu duced, they still do not help the propagation of the band. Their decay alone may be sucient to remove the Bauschinger overshoot, but there may be another eect to be considered. As mentioned, the Bauschinger curve runs along the elastic line longer after aging and reversal than after reversal without aging. This may be the eect of reinforcement by carbon of the relatively weak or widely spaced slipped obstacles in the belt between the position of the dislocations before reversal and their position after the reversed stress has risen nearly to the level of the stress before reversal.

8. Conclusions By machining a shallow and blunt circumferential groove into thin-walled tubular torsion specimens the axial Lders bands which usually u introduce yielding were eliminated and thus macroscopic internal stresses avoided. The yield phenomenon remained completely absent when after full strain aging, the specimen was strained in the direction opposite to that just before aging. Its absence, therefore, is a true physical characteristic of the yield phenomenon in iron. It does not seem compatible with the yield phenomenon model based on the locking of dislocations by impurities.

4 In the later experiments (1967) (Figs. 1013) aging was done at 150 C.

R.A. Elliot et al. / Mechanics of Materials 36 (2004) 11431153

1153

The yield points are present or absent after full strain aging according to the direction of straining, both if the aging treatment is given in the range of strain hardening after completion of yielding, and if it is given after partial yielding when only a part of the annealed specimen is covered by Lders u bands. If the rst yielding is interrupted and then the specimen re-loaded in the opposite direction without aging, the resulting Bauschinger curve overshoots the lower yield point without any discontinuity, and then the stress falls back to the normal yield level suddenly as the widening of the Lders band is resumed. u Experiments on thin-walled tubular torsion specimens conrm that the yield points rise with strain hardening if the straining is interrupted by aging treatments: the yield points and the strain hardening are roughly additive. After each yielding following strain aging the stressstrain curve has roughly the same rate of strain hardening as the uninterrupted curve at the same total strain, but it runs at a raised stress level. The upward displacement of the curve increases with every additional strain-aging treatment. 9. Postscript This research was conducted between 1959 and 1967. It came to the attention of F.R.N. Nabarro and A.S. Argon during the preparation of a biographical memoir of Egon Orowan for The Royal Society of London. The experimental research presented here was accompanied by a second theoretical paper. The two were submitted for publication in 1967 together as a pair. The theoretical paper was declined by the journal, and as a result the entire work remained unpublished. The principal author, Orowan, died on August 2, 1989. F.R.N. Nabarro and A.S. Argon are of the opinion that, even after nearly 40 years the experimental ndings of the authors are still highly

relevant for the understanding of this complex phenomenon. In the long and argumentative accompanying second paper on the theory of the yield phenomenon in iron which had been declined by the journal in its initial submission in 1967, Orowan lists most theoretical models on the yield phenomenon existing at that time. He points out particularly that the experimental observations presented in this paper are not in support of the Cottrell and Bilby (1949) theory and subscribes to the yield phenomena models based on dislocation multiplication such as that of Johnston (1962) but indicates that eective multiplication requires a certain free-run of dislocations that exists in single crystals where the needed multiplication can be accomplished with a minimal upper yield stress. In polycrystalline material where the grain boundaries limit the free-run more prominent upper yield stresses are required to accomplish the needed multiplication.

References
Cottrell, A.H., Bilby, B.A., 1949. Proc. Phys. Soc. A 62, 49. Edwards, C.A., Philips, D.L., Liu, Y.H., 1943. J. Iron Steel Inst. 47, 145167. Hall, E.O., 1951. Proc. Phys. Soc. B B64, 742747, 747753. Johnston, W.G., 1962. J. Appl. Phys. 33, 2716. Orowan, E., 1949. Mechanical testing of solids. In: Principles of Rheological Measurement, Report of General Conference, British Rheologists Club, London, October 1946, Th. Nelson & Sons, London & Edinburgh, pp. 156177. Orowan, E., 1959. Causes and eects of internal stresses. In: Rassweiler, G.M., Grube, W.L. (Eds.), Internal stresses and fatigue of metals, Proc. of a symposium held in Detroit and Warren, Michigan, 1958, Elsevier, Amsterdam, pp. 5980. Polakowski, N.H., 1952. J. Iron Steel Inst. II, 369376. Polakowski, N.H., 1951. Proc. World Metallurg. Congress, Detroit, pp. 553571. Polakowski, N.H., 1953. J. Iron Steel Inst. 1, 283284. Sylwestrowicz, W., Hall, E.O., 1951. Proc. Phys. Soc. B64, 49955502. Taylor, G.I., 1934. Proc. Roy. Soc. A 145, 362387. Tipper, C.F., 1952. J. Iron Steel Inst. 2, 143148.

Вам также может понравиться