obvifSIvj RETURNING MATERIALS: P1ace in book drop to LIBRARJES remove this checkout from Jun—nun. your record. FINES will be charged if book is returned after the date stamped below. GRAIN BOUNDARY MIGRATION AND DYNAMIC RECRYSTALLIZATION 0F NICKEL DURING HIGH TEMPERATURE LOW CYCLE FATIGUE By Shuhrong Chen A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Metallurgy, Mechanics and Materials Science 1987 ABSTRACT GRAIN BOUNDARY MIGRATION AND DYNAMIC RECRYSTALLIZATION 0F NICKEL DURING HIGH TEMPERATURE LOW CYCLE FATIGUE By Shuhrong Chen Microstructural developments of Ni polycrystals during high temperature low cycle fatigue were studied. Extensive grain boundary migration and dynamic recrystallization were observed. Dynamic recrystallization can be initiated at a drastically lower stress value during cyclic deformation at large cumulative strains than in monotonic tests. A total strain amplitude as small as 0.375% was found sufficient to set off dynamic recrystallization in Ni polycrystals during low cycle fatigue at high temperature, and there is circum- stantial evidence that dynamic recrystallization is the major softening mechanism. The microstructural investi- gations suggest that recrystallization twinning and grain boundary bulging may be responsible for the nucleation.of dynamic recrystallization during low cycle fatigue. ACKNOWLEDGEHENTS First and foremost, I would like to thank my advisor, Professor G. Gottstein, for his patient guidance and beneficial discussions in making this project a success. Also, I would like to express my deepest gratitude to my mother and my parents-in-law for their boundless love and full support. The considerateness, understanding and encouragement coming from my wife, Bihling, made everything go smoothly. I would also like to thank my friends in.the MMM department, for teaching me experimental techniques and conveying their experience. Finally, the support of the U.S. Department of Energy, Office of Basic Science, under grant number DE-FGOZ-SSER 45205 is gratefully acknowledged. ii TABLE OF CONTENTS Page LIST OF TABLES iv LIST OF FIGURES v I INTRODUCTION 1 II LITERATURE SURVEY 2.1 Grain Boundary Migration During LCF 3 2.2 DRX In Monotonic Tests 4 2.3 DRX During LCF 9 III EXPERIMENTAL PROCEDURE 3.1 Testing Materials 11 3.2 Specimens Preparation 11 3.2.1 Mechanical Testing Specimens 11 3.2.2 Optical Microscopy Specimens 14 3.2.3 Electron Microscopy Specimens 14 3.3 Mechanical Testing 15 3.4 The Microhardness Test 18 IV EXPERIMENTAL RESULTS 4.1 Mechanical Behavior 19 4.2 Microstructural Development 25 4.3 Evidence Of DRX Under LCF 34 4.4 Dislocation Structures 43 4.5 Result Of A Cu Single Crystal 47 V DISCUSSIONS 50 VI CONCLUSIONS 55 VII REFERENCES 57 iii LIST OF TABLES Table Page 1. Composition of Ni studied 11 2. Composition of Al studied 11 3. Jet polishing conditions for Ni and A1 15 4. Hardness distribution in Fig. 22 41 iv Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 10. ll. 12. LIST OF FIGURES Figure Captions True resolved shear stress vs. true resolved shear strain for <111> Ni single crystal [6]. Flow curves of a plain 0.25% C steel in the austenite 0 state (fee) at 1100 C (0.76 Tm), illustrating the strong influence of strain rate [27]. Geometry and dimensions of testing samples. Gauge sectional geometry and orientation of the Cu single crystal. Arrangement of the high temperature mechanical testing setup. True stress amplitude vs. cumulative strain of Ni with 1% total strain amplitude. True stress amplitude vs. cumulative strain of Ni with 0.5% total strain amplitude. True stress amplitude vs. cumulative strain of Ni with 0.375% total strain amplitude. True stress amplitude vs. cumulative strain of A1 with 0.5% total strain amplitude. True stress amplitude vs. cumulative strain of a Cu single crystal with 1% total strain amplitude. o The microstructure of Ni after annealing at 800 C for 5 hours. The microstructure of Ni after 200 cycles with 0.5% 0 total strain amplitude at 600 C. Page 13 l3 17 20 21 22 23 24 26 26 Figure Captions Page Fig. 13. The microstructures of Ni in a plane parallel to the Fig. Fig. Fig. Fig. Fig. 14. 15. 16. 17. 18. loading direction after cyclic deformation with 0.5% o 0 strain amplitude at 600 C (a) Annealed at 800 C for 5 hours; (b) after 10 complete cycles at 600°C, slightly polished and etched again to reveal the new grain boundary positions; (c) right after 10 complete cycles, positions A, 3,0 and D indicate inhomogeneous strain distribution on the surface; (d) etched to reveal the positions of new and original grain boundaries simultaneously; (e) right after 40 cycles; (f) the positions of new and original grain boundaries after etching again; (g) slightly polished and etched again, only new grain boundaries can be seen; (h) right after 160 cycles; (i) after etching; (3) after polishing and etching. 28 Detailed structure of a surface marking after 10 cycles. 32 o The microstructure of A1 (a) after annealing at 310 C for 5 hours; (b) after 720 cycles with 0.5% total strain amplitude at 200°C. The two figures are not taken from' the same area. 32 Serrations of a grain boundary in Ni after 200 cycles with 0.5% total strain amplitude. 35 A protrusion and a new grain at a grain boundary in Ni after 200 cycles with 0.5% total strain amplitude. 35 A new grain on the grain boundary in Ni after 510 cycles with 0.5% total strain amplitude. 36 vi Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. Figure Captions Page A new grain formed at a twin boundary in Ni after 200 cycles with 0.5% total strain amplitude. 36 A new grain formed near a twin in Ni after 300 cycles with 1% total strain amplitude. 37 A new grain formed in the interior of a grain at the end of a twin after 510 cycles with 0.5% total strain amplitude. 37 The microstructure of a Ni specimen after 510 cycles with 0.5% total amplitude. The whole cross sectional area perpendicular to the loading direction is revealed. 38 The microstructure in Ni after 300 cycles with 1% total strain amplitude. 40 The microstructure in Ni after 200 cycles with 1% total strain amplitude followed by a tension test until complete DRX occurred. 40 True stress-strain curves of Ni. Curve 1 and 2 represent the first and the 200th cycles, and curve 3 is the initial part of the hardening curve in a subsequent monotonic tension test. 42 The dislocation structure inside a Ni grain after 300 cycles with 1% total strain amplitude. 44 The dislocation structure around a grain boundary after 300 cycles with 1% total strain amplitude. 44 The dislocation structure within a Ni grain after monotonic tension test to a strain 6 - 0.15. 45 The dislocation structure in A1 after 720 cycles with 0.5% total strain amplitude. 45 vii Fig. Fig. Fig. Fig. Fig. 30. 31. 32. 33. 34. Figure Captions The dislocation structure in Ni around a grain boundary after 510 cycles with 0.5% total strain amplitude. The dislocation structure in Ni of a large area around a grain boundary after 300 cycles with 1% total strain amplitude. Surface markings on a Cu single crystal after 420 cycles with 1% total strain amplitude. (Due to the cylindrical specimen shape only a small strip is in focus) Qualitative decomposition of the cyclic flow stress. DRX in Pb-2% Sn alloy after 50 fatigue cycles at a strain ,3 amplitude of i0.75% and frequency 5.5x10 Hz. Recrystallization (a) at grain boundaries and (b) adjacent to a high density of slip bands [39]. viii Page 46 48 49 52 S4 I INTRODUCTION When a material is subjected to or even deformed at high temperatures, the microstructure and consequently the properties may change and cause failure. One well known high temperature phenomenon is the recrystallization of the material during deformation, also referred to as dynamic recrystallization (DRX). DRX has been reported in a variety of fee metals and alloys in which the degree of dynamic recovery is restricted, such as Cu [1,2], CuAl [3], CuZn [4]; Ag [5]; Ni [6,7], Ni-Fe [7], Ni superalloys [8]; stainless steel [9], etc.. These experimental reports were based on the tests conducted in monotonic loading conditions, namely in tension, compression, or torsion. In a recent paper, DRX was found at the grain boundaries during low cycle fatigue (LCF) in bicrystals of Ni with special grain boundaries [10] . But no systematical research has been reported yet on DRX during LCF. Most parts under high temperature service conditions, however, are subjected to cyclic loading conditions due to vibration or temperature fluctuations. A major difference between cyclic and monotonic loading is that the strain amplitude under cyclic conditions is commonly small compared to the critical strain to set off DRX in monotonic tests. However, the critical conditions for DRX to occur are determined by the microstructure, namely by the development of a cell-or subgrain structure[2,6,11]. In recent investigations on LiF [12] , stainless steel [13,14,15] and Cu [16] it was found that also during LCF at elevated temperatures a microstructure develops that strongly resembles the dislocation structure, which occurs in high temperature monotonic tests. Under appropriate conditions this microstructure may collapse and set off DRX. Extensive grain boundary migration was observed during LCF. For monotonic tests at lower strain rates, the nucleation of DRX may occur by bulging of an existing grain boundary [7] . Also DRX caused by the bulging of original grain boundaries was reported on polycrystalline dilute Cu alloys [17]. Therefore, the grain boundary migration during LCF may set off DRX at a lower stress than the critical recrystall- ization stress for monotonic tests. In the present study the problem was addressed by investigating the microstructural development during LCF. II LITERATURE SURVEY 2.1 Grain Boundary Migration During LCF It is clear from the data available to date that the movement of grain boundaries is an important microstructural process in high temperature fatigue. Quantitative studies have been reported on Pb [18,19] and A1 [20,21]. Metallographic observations show that there is a one-to-one correspondence between the markings from grain boundary migration and the number and pattern of cyclic loading. Apparently, the grain boundaries respond to the imposed stresses by migrating in a cyclic manner and it is assumed that migration occurs such as to move the boundaries into a diamond shape grain configuration at 450 to the stress axis. The average distance of grain boundary migration ii for polycrystalline A1 with strain amplitude less than ~ 3 0.4% was found to obey the empirical relationship - -o.ss 0.5 o.ee m - Af N Ac exp(-62.0/RT) where f : frequency N : number of loading cycles Ac : strain amplitude R : gas constant in kJ/mol.k T : absolute temperature 1 0.35 A : == 3.2x10 pms Also, the occurrence of cyclic migration leads to the disappearance of grains and thus is a mechanism of grain coarsening. A change of the grain morphology due to grain boundary migration during cyclic deformation at 650°C at a low strain rate (é- 4x10J/s ) was also reported on OFl-iC Cu [22]. But this change in grain shape was found to be strain rate dependent and at high strain rates , no such change in grain morphology occurred. That impurities or precipitates slow down or inhibit grain boundary migration explains why type 304 stainless steel does not exhibit any change in grain structure at 760°C at the lower strain rate. A relation between grain boundary migration and nucleation of DRX was reported by Lim and Raj [10]. In their work on Ni bicrystals, they found that low to moderate extent of boundary migration was beneficial to the nucleation of fresh grains during LCF. They also found that grain boundary cavitation and migration were inversely related, i.e. , in regions where migration had occurred there was little cavitation and vise versa . 2.2 DRX In Monotonic Tests DRX has been investigated extensively in monotonic loading. As is known from previous investigations, the onset of DRX in single crystals is indicated by a sharp drop in the flow stress (Fig. l) [5,23] . It has already been shown previously [24], the shear stress (rR) rather than shear strain (1R) is the critical quantity governing the initiation of CHNHUM mefim U0>H0mmh 05H“ .Ho_ Hmumsuo mascam Hz AHHHV you .m> mmmHUm Hdflfim U$>H0m0H 03H. 22um mwumasado .m> ennuwanas mmouum mane .o .mwm _ X _ 225m .323 OONF 600 O _ 4] _ ngm \\ j I L O / f IIK — _ . cm can now 0 30>0 “.0 .02 [new] ssauis anal 21 .oeSuHHQEe :Hmuum kuou am.o £uw3 Hz mo :Hmuum o>HumHSESQ .m> oesuwaaem mmouum same .5 .wwm fix; :2um :5550 So one o _ \\ J O 4. [saw] ssauis anal O (D O cum omN m._o>o .10 .oz 22 .oUSDHHQEQ Guouum Hmuou wmmm.o nu“: Hz mo :Hmuum o>wumasedo .m> ennuwamee mmouum osuH .m .mwm § _ 225m .5230 8: one o .I O 0 I l I 0 [saw] ssauis anal / O (O O _ 0mm owe m40>0 "—0 .02 23 .oesuHHQEm :Hmuuw Heuou mm.o news H< mo :Hmuum o>wumH3830 .m> oesuuaaEd mmouum oSHH .o .wwm _o\o _ 225m .5550 \\\\\\I\\\J com one 0 // k m40>0 “.0 .02 0N5 con 0 or O [can] ssaui anal + Op 24 NO. OF CYCLE o 250 500 35 . fiF‘X \\ + CD TRUE STRESS [MPa] F X/ \. _——‘ 35 l O 600 1200 CUMUL. STRAIN [°/.] Fig. 10. True stress amplitude vs. cumulative strain of a Cu single crystal with 1% total strain amplitude. 25 4.2 Microstructural Development Substantial microstructural changes were found when comparing the microstructures before (Fig. 11) and after the test (Fig. 12). These changes are due to extensive grain boundary migration. In Fig. 11 most grain boundaries showed smooth curves, nevertheless, serrated segments and sharp turns were found in Fig. 12, which is the microstructure after 200 cycles deformation with 0.5% strain amplitude. In order to investigate the phenomenon more systematically, one specimen was machined flat on one side surface such that the microstructures could be examined after different defamation stages. The test was interrupted after 10 complete cycles , 40 cycles. and finally 160 cycles. A series of micrographs was taken after each specific stage. They are shown in Figs. 13 (a)-(j). All these micrographs were taken from the same area which was parallel to the loading direction. Fig. 13(a) is the microstructure of the annealed state. The average grain size as measured by the linear intercept method is about 300 pm. There are many small grains embedded between large ones in this annealed state. Fig. 13(b) is the microstructure after 10 complete cycles. Position A, B and D show extensive grain boundary migration. One grain boundary disappeared and one twin boundary formed at position C. About 25 pm thickness was polished away in the sequence from Fig. 13(a) to Fig. 13(b). Comparing these two microstructures. it was found the small grains tend to shrink. Fig. 13(c) is the microstructure taken immediately after 10 cycles. The strain distribution is inhomogeneous in that some areas show much stronger grain boundary migration. The grain boundary arrangement is complex because of the superposition of grain boundary positions of all previous cycles due to grain boundary 26 a Fig. 11. The microstructure of Ni after annealing at 800 C for 5 hours. Fig. 12. The microstructure of Ni after 200 cycles with 0.5% 0 total strain amplitude at 600 C. 27 sliding perpendicular to the surface. Therefore. the sample was etched again slightly such that the boundary position before and after deformation could be seen simultaneously. Fig. 13(d) is the result after etching. The grain boundary at position B in Fig. 13(a) reveals a series of approximately parallel markings in front of this grain boundary. The new grain boundary position can be seen in Fig. 13(d) . Those markings delineated the positions of the boundaries at the end of each stress cycles. A one-to-one correspondence between the number of grain boundary markings and the total number of whale cycles imposed on the specimen was established. A more detailed structure of surface markings is shown in Fig. 14. The same procedure was followed for the investigation at higher numbers of cycles. Figs. 13(a)-(g) are the microstructures after 40 cycles. They show the same phenomenon, but the migration rate has decreased compared to the first 10 cycles. The positions indicated by arrows make it is obvious that the grain boundaries tend to migrate such as to align under 450 with respect to the stress axis. This will facilitate the grain boundary sliding and is believed to also may cause cavity fomatian. Figs. 13(h)-(j) are the micrographs after 160 cycles. Although compared to Figs. 13(a)-(g) the sample had deformed another 120 cycles . Fig. 13(j) doesn't show much difference compared to the microstructure in Fig. 13(g), except most a grain boundaries now are aligned under 45 to the stress axis. The average grain size of the sample after 160 cycles (Fig.13(j)) is approximately 400 pm . Grain boundary migration was also observed in Al. The micrographs in Fig. 15(b) were taken after 720 cycles defamation. Actually, it is the whole cross sectional area perpendicular to the loading direction. 28 Fig. 13. The microstructures of Ni in a plane parallel to the loading direction after cyclic deformation with 0.5% o a strain amplitude at 600 C (a) Annealed at 800 C for 5 0 hours; (b) after 10 complete cycles at 600 C. slightly polished and etched again to reveal the new grain boundary positions; 29 (G) Q5mm Fig. 13. (c) right after 10 complete cycles. positions A. .C and D indicate inhomogeneous strain distribution on the surface; (d) etched to reveal the positions of new and original grain boundaries simultaneously; 30 Fig. 13. (e) right after 40 cycles; (f) the positions of new and original grain boundaries after etching again; (g) slightly polished and etched again. only new grain boundaries can be seen; 31 Fig. 13. (h) right after 160 cycles; (1) after etching; (j) after polishing and etching. 32 a Fig. 15. The microstructure of A1 (a) after annealing at 310 C for 5 hours; 33 (b) Fig. 15. (b) after 720 cycles with 0.5% total strain amplitude at 200 C. The two figures are not taken from the same area. 34 The grain boundary migration can be seen by comparison of Figs. 15(a) and (b). Fig. 15(b) shows curved grain boundaries instead of perfectly straight ones shown in Fig. 15(a), which is the microstructure of nude famed specimen . 4.3 Evidence Of DRX Under LCF The occurrence of DRX in polycrystals is difficult to evidence since the sample remains polycrystalline and the newly famed grains are rapidly deformed by the concurrent deformation. It is particularly difficult in LCF because of the small driving forces and thus small boundary migration rate. The occurrence of DRX was therefore indirectly confirmed by metallographic investigations. After each test, the sample was cut perpendicular to the loading direction. and examined the microstucture in details under higher magnification. Special attention was paid to observe the grain boundaries. Several micrographs are shown in Figs. 16-21. The grain boundaries show serrations (Fig. 16) , sharp protrusions (Fig. 17) and even new grains (Figs. 17-21). These phenomena are typical for the nucleation of recrystallization at grain boundaries [26,37] . Figs. 17.19-21 show the details near a twin. It occurs as if the twin was created during deformation. This is important, since recrystallization twinning was found to be the major nucleation mechanism of DRX in monotonic tests. This seems to indicate that these processes are nucleation phenomena of DRX. Fig. 22 is a micrograph of the whole cross sectional area perpendicular to the loading direction after 510 cycles with a total strain amplitude of 0.5%. The microstructure contains areas with many small grains. While Fig. 35 Fig. 16. Serrations of a grain boundary in Ni after 200 cycles with 0.5% total strain amplitude. 17. A protrusion and a new grain at a grain boundary in Ni after 200 cycles with 0.5% total strain amplitude. 36 Fig. 18. A new grain on the grain boundary in Ni after 510 cycles with 0.5% total strain amplitude. Fig. 19. A new grain formed at a twin boundary in Ni after 200 cycles with 0.5% total strain amplitude. 37 Fig. 20. A new grain formed near a twin in Ni after 300 cycles with 1% total strain amplitude. Fig. 21. A new grain formed in the interior of a grain at the end of a twin after 510 cycles with 0.5% total strain amplitude. 38 Fig. 22. The microstructure of a Ni specimen after 510 cycles with 0.5% total amplitude. The whole cross sectional area perpendicular to the loading direction is revealed. 39 the migration processes which were observed in the beginning of defamation, generally resulted in a coarsening of the microstructure, this micrograph. in contrast indicates a refinement of the microstructure. Fig. 23 shows the microstructure after 300 cycles with 1% strain amplitude. Compared to Fig. 22, the grain boundaries in this figure appear more snaky than in Fig. 22, also the nucleation phenomenon is more pronounced. Fig. 24 is the microstructure after 200 cycles cyclic deformation with 1% strain amplitude followed by a tension test until DRX occurred. This microstructure evidenced both boundary migration and DRX. but only DRX can be seen in monotonic tests. A parallel study was conducted on Al. Due to the high stacking fault energy Al is not able to recrystallize during deformation [23,38]. Fig. 15(b) shows the microstructure of an Al polycrystal after 720 cycles defamation with a total strain amplitude of 0.5%. No small grains were detected in this micrograph. Obviously without DRX the microstructure has a tendency for unifom coarsening. The microstructure in Ni however shows local grain refinement. This further substantiates the conclusion that Ni polycrystals undergo DRX during LCF at high temperature. Due to the lack of rapid quench equipment, the cooling rate of the sample in air was relatively slow. This might raise the problem that the microstructure after cooling to room temperature was not the same as the microstructure at high temperature before cooling. More to the critical point. the question is whether the observed recrystallization phenomena could be due to static recrystallization. This was tested by means of the flow stress which is very sensitive to recrystallization. A Ni sample was first cyclically deformed for 200 cycles with a total 40 Fig. 23. The microstructure in N1 after 300 cycles with 1% total strain amplitude. Fig. 24. The microstructure in Ni after 200 cycles with 1% total strain amplitude followed by a tension test until complete DRX occurred. 41 strain amplitude of 1% at 600°C and then kept under zero load at same temperature for 5 hours in the testing machine. and finally tested in tension until DRX occurred. The result is shown in Fig. 25. The true stress-strain curve of the first and the last (200th) cycles during cyclic deformation are indicated by l and 2 respectively in this plot. Curve 3 is the initial part of the hardening curve in subsequent monotonic tension test. The yield stress in curve 3 was measured as 40 MPa compared to 42 MPa which appeared in the last cycle under cyclic deformation. The small stress drop during annealing is typical for static recovery. This test substantiates that the recrystallization characteristics observed in the microstructure of fatigued specimens are not due to static recrystallization after the test but have to be attributed to DRX. In addition microhardness tests were conducted to measure the hardness distribution of the grains shown in Fig. 22. The results are listed in Table 4. Table 4 Hardness distribution in Fig. 22 Grain size [pm] Average L [mm] Average DPH (a) >200 0.075 t 0.001 33.0 i 0.9 (b) <100 0.079 t 0.001 29.7 i 0.8 The average DPH of small grains is slightly but consistently smaller than that of large grains. This result is compatible with the 42 .umou Camacou owcouoaoa usozvomnsm o cw o>udo maucoouon on» ma anon Howuwcw ecu mg m o>hao was .moa0ho Spoom one use umufim onu udomoumau N one H o>uso .Hz mo mo>uao :Houum-mmouum asuh .mm .wwm TX; Z_