“'7 >__ I -—.—-—-—-—- a llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllil 3 1293 00561 4825 LIBRARY Michigan State University Th‘" is to certify that the thesis entitled DYLMC RECRYSIALLIZATION OF BORQJ IDPED NiBAl POLYCRYSI‘AL DURING HIGH TIE/[PERATlJ'tfi CCMPRESSION TEST presented by Hang-Pei Yung has been accepted towards fulfillment of the requirements for u- S degree in Jetallmgx ' ,3 61;“ 4’63 ’fx {fa—x. (/ Maj ryéfessor Drofessor G: Gottstein Date 11' “2%- ”7‘7 February-24-l989 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution MSU LlBRARlES .—:—. RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. DYNAMIC RECRYSTALLIZKTION 0P BORON DOPED N13A1 POLYCRYSTAL DURING HIGH TEMPERATURE COMPRESSION TEST BY HANG - PEI YUNG A THESIS Sub-itted to Michigan State University in partial fulfillment of the require-eats for the degree of MASTER OF SCIENCE Department of Metallurgy, Mechanics and Materials Science 1989 sci-2.237 ABSTRACT DYNAMIC RECRYSTALLIZATION OF BORON DOPED Ni3A1 POLYCRYSTAL DURING HIGH TEMPERATURE COMPRESSION TEST BY HANG-PEI YUNG The features of dynamic recrystallization ( DRX ) of NiaAl and pure Ni during compression at temperature between 0.5 - 0.8 of the melting temperature ( Tm ) and strain rate between 2 x 10-3 8-1 and 2 x 10"5 8-1 were studied. From the metallographic observation and stress - strain behavior, it was found that DRX actually occured during high temperature deformation. For the given deformation condition only a single maximum of the flow curve was observed for Ni Al, while an oscillation flow behavior was found for Ni. The 3 microstructure of Ni3Al always underwent grain refinement, while the microstructure of Ni underwent grain coarsening during DRX. The observations are in line with the prediction that a transition from sigle peak to multiple peak flow behavior only occurs for Do / ZDS . 1. For Ni an activation analysis of OR yielded a strss exponent, n s 6.8. and an activation energy, 0 a 57.79 kcal / mole. An activation analysis for Ni3Al was not feasible since n and 0 depend on temperature. ACKNOWLEDGEHENTS I am grateful to my advisor, Professor G.Gottstein, for his kindly instruction and necessity support for experiment. Also, I would like to thank my colleague, S.R.Chen, for his technical assistance and discussion and others who exchanged view on the DRX experiment. Finally, I would like to appreciate my wife for her mental support and fortitude during the past two years. 111 TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES 1. INTRODUCTION 2. LITERATURE SURVEY 2.1 Effect of Strain Rate on Grain Size of DRX 2.2 Effect of Temperature on the DRX Grain Size 2.3 Microstructure Change Corresponds to the DRX 3. EXPERIMENTAL PROCEDURES 3.1 Machine 3.2 Specimen Preparation 3.3 Mechanical Test 3.4 Microstructure Observation 3.5 Protection 4. RESULTS 4.1 The True Stress - True Strain Curve 4.2 Metallographic Examination 5. DISCUSSION 5.1 Evidence of the DRX 5.2 The Grain Size of the DRX 6. CONCLUSIONS 7. REFERENCES iv Page IV QNUN 10 10 13 13 16 17 17 SS 62 65 66 LIST OF TABLES Table 1. The a 8 a under various test condition for Ni A1 with R S 3 D . 170 um. 2. The IR & as under various test condition for Ni3Al with D0 = 9 um. 3. The “R & as under various test condition for pure Ni with D0 = 136 um. 4. The US under various test condition for Ni3A1 with DO 3 170 Ill]. 5. The Us under various test condition for Ni3Al with 6. The Ds under various test condition for pure Ni with Do a 136 um. Page 22 22 3O 30 49 49 LIST OF FIGURES Figure Captions Page Fig. 1. (a) Effect of strain rate on the flow curves. (b) Effect of temperature on the flow curves [3]. 3 Fig. 2. A microstructure mechanism map for distinguishing between the occurrence of two types of DRXIIB]. 4 Fig. 3. Predicted stress - strain curves for DRX. (a) A steady state curve for the condition when cc < ex. (b) a cyclic flow curve when the ac > ex [3]. 6 Fig. 4. The metallograph of Ni3A1 annealed at 800°C for 4 hours. 11 Fig. 5. The metallograph of prestrained NiBAl ( 60 X ) annealed at 800°C for 1 hour. 12 Fig. 6. The metallograph of pure Ni annealed at 1000°C for 30 min. 14 Fig. 7. The metallograph of pure Ni when waited for the thermal equilibrium at 1000°C for 1 hour. 15 Fig. 8. The flow curves of Ni3Al ( D0 = 170 um ) which were compressed at 650°C with various strain rates. 18 vi Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 10. 11. 12. 13. 14. 15. 16. The flow curves of N13A1 ( Do u 170 um ) which were compressed at 750°C with various strain rates. The flow curves of N13A1 ( Do . 170 um ) which were compressed at 850°C with various strain rates. The flow curves of Ni3A1 ( Do - 170 um ) which were compressed at 1000°C with various strain rates. The flow curves of Ni3Al ( Do 170 um ) which were 38-1 at various compressed with strain rate 2 x 10' tempera tures o 170 um ) which were 4 The flow curves of Ni3Al ( Do compressed with strain rate 2 x 10' S"1 at various temperatures. The flow curves of Ni3Al ( Do 170 um ) which were 5 1 compressed with strain rate 2 x 10- S- at various temperatures. The flow curves of Ni3A1 ( Do - 9 um ) which were compressed at 800°C with various strain rates. The flow curves of Ni3Al ( Do - 9 um ) which were compressed with strain rate 2 x 10.45"1 at various temperatures. vii 19 20 21 23 24 25 26 27 Fig. Fig. Fig. Fig. Fig. Fig. Fig. 17. 18. 19. 20. 21. 22. 23. The flow curves of pure Ni ( Do a 136 um ) which were 4 compressed with strain rate 2 x 10- S-1 at various temperatures. The flow curves of pure Ni ( Do . 136 Hm ) which were compressed at 850°C with various strain rates. The metallograph of Ni3A1 ( D0 = 170 Um) after compression test at 650°C with 2 x 10.38.1 shows the necklace structure along the existing grain boundary. The metallograph of Ni3Al ( Do a 170 um ), compressed -4 -1 at 650°C with 2 x 10 S , shows the necklace structure along the existing grain boundary. The metallograph of Ni3Al ( Do a 170 um ), compressed -5 —1 at 650°C with 2 x 10 S , shows the necklace along the existing grain boundary. The metallograph of Ni3A1 ( Do . 170 um ), compressed -3 -1 at 750°C with 2 x 10 S , shows the necklace along the existing grain boundary. The metallograph of Ni3Al ( D0 = 170 um ), compressed -4 -1 at 750°C with 2 x 10 S , shows the necklace along the existing grain boundary. viii 28 29 32 33 34 35 36 Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 24. 25. 26. 27. 28. 29. 30. 31. 32. The metallograph of Ni3Al ( Do . 3 at 850°C with 2 x 10' 3", shows the necklace. The metallograph of Ni3A1 ( Do at 850°C with 2 x 10’45‘1. D II The metallograph of Ni3Al ( at 850°C with 2 x 10'55‘1. The metallograph of Ni3Al ( D a at 1000°C with 2 x 10'35'1. The metallograph of Ni Al ( D . 3 0 at 1000°C with 2 x 10‘45'1. The metallograph of Ni3Al ( D a at 1000°C with 2 x 10'55’1. The metallograph of Ni A1 ( 3 0 at 600°C with 2 x 10‘43'1. U l The metallograph of Ni D II 3A1 ( o at 750°C with 2 x 10’45'1. C II The metallograph of Ni A1 ( 3 0 at 800°C with 2 x 10‘35‘1. ix 170 um the grain 170 um 170 um 170 um 170 Hm 170 um )9 ). )9 )i )9 ). compressed growth within compressed compressed compressed compressed compressed 9 um ), compressed 9 pm ), compressed 9 um ), compressed 37 38 39 40 41 42 43 44 45 Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 9 um ), compressed The metallograph of N13A1 ( Do at 800°C with 2 x 10'45'1. The metallograph of Ni3A1 ( Do at 800°C with 7 x 10’55’1. 9 um ), compressed The metallograph of Ni 9 pm ), compressed 3A1 ( D0 at 850°C with 2 x 10"s'1. The metallograph of pure Ni, compressed at 650°C with 2 x 10‘45'1. The metallograph of pure Ni, compressed at 850°C with 2 x 10'35'1. The metallograph of pure Ni, compressed at 850°C with 2 x 10‘45‘1. The metallograph of pure Ni, compressed at 850°C with 2 x 10‘55'1. The metallograph of pure Ni, compressed at 1000°C with 2 x 10’45'1. The relationship of strain rate and 0R for Ni3A1 with DO 8 170 um. The relationship of the temperature and OR for Ni3A1 46 47 48 50 51 52 53 54 56 Fig. Fig. Fig. Fig. Fig. Fig. 43. 44. 45. 46. 47. 48. with D0 = 170 um. The relationship With DO I: 9 um. The relationship Vith D0 = 9 um. The relationship The relationship The relationship the relationship of of of of of of the strain rate and 0R for Ni3Al the temperature and OR for Ni3Al the strain rate and o R for pure Ni. the temperature and on for pure Ni. and final grain size for Ni Al. ° 3 S ‘3 xi and final grain size for pure Ni. 57 58 59 60 61 63 64 1. INTRODUCTION The dynamic recrystallization (DRX) of the material with Ll2 structure such as N13A1 [1] and Zr3Al [2] has been reported to explore the fracture mode by using tension test. In order to observe the kinetic of the grain growth under DRX process of the NiBAl, we chose various compression test conditions and prepared two different initial grain sizes (Do). Controlling 2 value and Do’ we supposed that the transition of DRX flow curve for Ni3Al from continuous to periodic should be taken place because lower Z value and smaller Do tend to the occurrence of multipeak. Finally, we found that the behavior of the DRX can be related to the recrystallization stress ( OR ) and the final grain size (Ds) has relationship to the steady-state flow stress (as). The DRX behavior of pure Ni was compared with NisAl because Ni3Al has a large amount of Ni and the similarity of the melting temperature (Tm), i.e. 1395°C, 1450°C, also to inspect the Do is independence of DS . 2 LITERATURE SURVEY Materials would soften due to the recrystallization while they are subjected to deform at high temperature, above half of the melting temperature, which called dynamic recrystallization (DRX). Since 1939, the topic of DRX has been mentioned in the creep test of Pb [3]. Up to now, several reports in this field have been stated in various deformation tests, such as torsion test [4,5,6,7,8,9], tension test [1,2,10,11], rolling test [12,13,14], and compression test [15,16,17,18]. According to these authors, the most remarkable phenomenon when material is deformed at elevated temperature with given strain rate (é) is the flow stress-strain (o-s) curve with a characteristic shape, either periodic (multipeak) or continuous (single peak) (Fig. 1) [3]. This dynamic—softening process is thermal-activated, giving a dependence of a on é and temperature which can be expressed in terms of the Zener-Hollonom parameter, Z, (Fig. 2) [18] Z a é exp (0/RT) where, Q is activation energy. The style of the flow curve can be changed from periodically to continuously as the 2 value is increased. On the other words, when the Z is decreased the flow curve shows from single peak to multipeak. Therefore, when the temperature is decreased or the t is increased, the single peak will occur and the multipeak takes place with increasing temperature or decreasing é . Correspond to the initial grain size (Do)and steday-state recrystalized grain size (05), when the oscillation flow curve appears, the grain coarsening occurs (Do << DS), and when the continuous flow ._m_ mo>m=o node on“ :o ouaumuoaeou mo uoouum Any .mo>u=o aoau may no menu :«muum no uuommm Amy .H .wfim £02» an; mo mo v0 mo 0 a _ . 1 coBmocano 1 ON CV 05 Dub 00 mum UOOQN I k l 00 |.oo_ 73.0. .6... w .063 0 $00.0 ON. 358135 80.1.] (aw/NW) AWN AVN D. o. mAV a _ d _ ‘ co .mcoh _ 60.0 800.0 {H1 58.0 885 , a 6.0 a 08.0 _ 30.0 v. .o 98 l 0000: I k 363 othnwd .ON cm 00 00. 3 (aw/mm ssaiis am] Stable dynamic grain size figs/2.4444? ___//////2 / High 2 High 0' N N I V l / ’ H ' // ____\_-a I / Multiple peak. and grain coarsening region : / 3w l I \ Low 0 Low 2 l | 0 00 0| 2 Initial grain size Fig. 2. A microstructure mechanism map for distinguishing between the occurrence of two types of DRX[18]. curve shows means the specimen is undergoing grain refinement (Do >> DS). 2.1 Effect of Strain Rate on Grain Size of DRX When the material undergoes DRX, the dislocations substructure develops in the initial stage of deformation which means the new cell structure is smaller and has more tangled cell walls [18]. At higher é , a finer tangled cellular structure is developed throughout all the grain, which does not leave grain boundary segment long enough to bulge. With increasing strain, some tangles build up to high misorientations, the nuclei create within each grain. The density is higher near the grain boundary because the higher strain. By the time, steady-state flow is attained, new small equiaxed grains have replaced the original ones. According to the Fig. 3a [3], before recrystallization (RX) is complete, the region which first recrystallized reaches the critical strain for a second nucleation. Under this condition, the dislocation densities at the center of the recrystallized grains have increased with further straining the material, which tend to grow another nuclei. Because nucleation or necklace structure occurs at the existing boundaries, the dynamic grain size attains the final stable dynamic grain size(DS ). Therefore, continuous deformation results in work hardening to take place within the dynamic recrystallized grains and reducing the driving force for growth. Hence, the final stable grain size is refined, and leaves a single peak stress - strain curve. ._m_ xw A ow was can: o>u=o 36am omauzu a Aav .xu v on can: cemuficaoa ecu mow o>u=u oumum atmoum < Amy .xmn mow mo>a=u :«muum I mmouum coaamvoum .m .uwm 3. . E Tuwl'Tluwl'wlluwi'levl'Jl CVJ TCVI‘TIUWIUVIICV Cereal! C Ia \ J a L NomH [No DU 1 can.» 13%.». I. wdmsiwo. 100 WW. 3.... mowu . J III-1.8 w o.m. J o..w ssaJIs ssaJts .538”. «.mm .253. comm Conversely, at lower strain rate, the grain undergoes coarsening due to nucleation production by bulging of the pre-existing grain boundary. In this case, the impingement of boundary results in the termination of a new dynamic grain growth. From the observation of stress-strain curve (Fig. 3b), the nucleation of all new grains occur at when a > ex [scz critical strain for the initiation of RX, ep : peak strain for RX, ex : strain for the completion RX]. Therefore, the growth is restricted to the region ex until the material reaches another RX can be initiated. Hence, one RX cycle is virtually complete before the next cycle starts and results in a multipeak stress-strain flow curve. However, normally higher strain rate gives rise to higher flow stress, or vice versa [4,10]. Thus, when the material is subjected to deform with lower strain rate at high temperature, the RX stress, a , is lower than at high strain rate under the same test condition and shows periodic a - a curve [4]. 2.2 Effect of Temperature on the DRX Grain Size First of all the test temperature for RX should be higher than 1/2 Tm at least. As has been mentioned previousely, temperature has something to do with DRX because of the 2 value. At given é , the 2 value will be increased as temperature is decreased which results in a single peak stress-strain curve. Because a higher density of nuclei is given rise to the increasing in subgrain density as the 2 value increases [18]. Inversely, increasing temperature at given strain rate will increase the RX process [13] and decrease the Z value which results in a multipeak behavior and the appearance of the grain coarsening. Also, akthe decreases as the temperature increases which can be described in terms of the empirical relation [11], "R = A él/m exp (U/mKT) where, A, m, U are constants. Therefore, RX is periodic at lower stress when temperature is increased, and becomes continuous with increasing stress when temperature is decreased [4]. 2.3 Microstructure Change Corresponds to the DRX The effects of strain rate and temperature on the DRX have been discussed a while ago. Both effects have mentioned one important feature, that is microstructure change, either grain coarsening or refinement, under different test conditions. Also, the shape of the DRX stress-strain curve is dominated by the final grain size. In case grain coarsening (DO << Ds)takes place, the curve shows periodically and grain refinement (Do >> DS) pans up a continuous curve. From the map (Fig. 2), however, changing the Do seems that the shape of the flow curve can be altered. Thus, it is to be expected that the rate of DRX process decreases, i.e the transition from periodic to continuous behavior, with increasing Do [15]. Nevertheless, it is found that the in terms of domination element, D ,is normally related to the a S S ’ -n as =OO+AD where, do, A, n are constants. Formularily, a is only dependent of D S but Do [8]. Changing D0 will not affect the shape of the flow curve. S However, it increases the probability of the transition. It is clear that decreasing the D0 will increase the rate of the RX and decrease the flow stress of RX. It is not necessary, however, that DS should be larger than D0 to produce an oscillation curve. Sakai et a1. [3,18] cited the criterion for the transition, DO IZDS a 1. Under this example, the oscillation flow curve occurred when we found that Ds < DO < ZDS , it is considered that the grain coarsening take place due to the small final grains resulting from the growth of the RX nuclei. 3. EXPERIMENTAL PROCEDURES The material, Ni Al, with composition Ni-76,Al-24 and B-0.24 a/o, 3 was supplied by Oak Ridge National Lab. The pure Ni was ordered from Materials Research Corporation with Ag, Al, Cu, Fe, K, Li, Mg, Mo, Pb, Si, Sn, Ti, V, Zr less than 10 ppm, 0-14ppm and N-lppm. 3.1 Machine The expriment machine is MTS 810 with 458.20 microconsol which was linked to Znith AT personal computer to remote control the MTS unit and to record the data. The vacuum furnace was set up on the HTS with Centorr S60 - 26350 temperature control unit. The thermocouple was placed at the edge of the cental ceramic rod. 3.2 Specimen Preparation As-casted Ni3Al was machined to 5 mm height and 3 mm diameter cylinder by using electric discharge machine (EDM). To keep a normal stress on the surface when applied axisymetric compression load, the sample surfaces were grinded to very flat with flatness no larger than 1/1000 and very perpendicular to the cylindric surface, i.e. aligned by the EDM. All of the specimens were annealed at 800°C for 4 hours in the vacuum furnace at a pressure of approximately 10'5 Torr, and allow us to get a grain size with 170 um (Fig. 4). The small grain size sample, with 9 um (Fig. 5), was obtained from the above material with large 10 11 .muso: a you oooow um voHaoaaa H< n 2 no £83232. 2; .e a: .uso: H man Queen um vonoccm A N Co v H saw: Deane um vmmmouaeoo mesa same: A as oh“ u ea c Hanaz ae.aw>aae some use .m .maa 7% 225m may: om ow ow ON 0 m I N 1 79-35 n or. m :PeENIsIm 1-9-0me u a as 0.03 n a s coo oomfi [Ban] SSHHIS anal l9 .mouma :wmuum m=Owum> nu“: Owens um commouaeoo saws games A a: aka u on v Hamaz me wwaaaa same was .a .uaa E 225m mSE 00_ on ow 0v om . m I {myoaxm I a I m .m.-2xm n a ll Poms n s com coca [Edi/ll] ssaals anal 20 .moumu :Amuum msofium> nu“; caomm um commauaaou saws :eAsa A 5: 0AA u on v AamAz as asaaae aaAa was .oA .mAa E 225m may; Avm DAV omw Av 00A ow Tmtoflxm H .8 rmtofixm H .6 Tmabfixm H w. 000mm | Est—INC") l mxvm mzvm [9cm] 883818 anal 21 0A.: .moumm awmuum mzoAum> nu“: voocofi um commouasoo saws seAaa A a: oAA u on v AemAz we assess aeAa was .AA .mAa #3 225m may: on cm 0% ON 0 rmabfixm H .w l _ I _ . _ . _ . O azum m N A E oaooofi H F t _ i A . _ . 006 [Edit] SSEIHLS anal Table 1. The a Note: The unit for a Note: The unit for & a under various test condition for Ni3Al with R 8 D0 = 170 um ' in 9 -3 -1 I —4 -1 -5 —1 ' I (at | 2x10 S | 2x10 S 2x10 5 r°c s 650 1035 757 694 (1096) (584) (348) 750 833 703 ——- (660) (422) 850 593 462 319 (295) (195) (144) 1000 408 219 114 (258) (159) (110) & a is MPa. R S Table 2. The a & a under various test condition for Ni A1 with R s 3 Do - 9 um. 2x10'3s'1 2::10"'s'1 7x10'5s'1 --- 837 --- (781) 750 --- 536 --- (485) 800 479 330 298 (421) (293) (256) 850 -—- 228 --- (194) °a 8 aS is MPa. .mouaumaoasou msoAum> um Atmmqu x N mama :Ammum AAA: commomasoo one: :aAsa A E: 0AA n on v am=o Seam one .NA .wAm The 225m may: 2: cm ow 0A4 ON 0 erooA H 1 ocean H 080mm H Uaomw H A rmtoAxm . cow E-IE-‘F I l I 01 rd (32 03 ST 00 [98141] 883818 8081 A . A . A w A . OONH 24 .mousumuoaeau msoAmm> um commouasoo one: :oAsa A E: as“ u on v H< E 229% may: m- o” x N mama :Aauam AAA: Az we assess aeAm was .mA .mAa A- a- m 00A ow ow ow om n? m ,/\ 0.82 n A I e 0.0mm n a. I m A Doomkt H F In N 0.08 n A IA amtoAxm I. 6 azum 000A [8cm] $83818 3081. 25 .mousumaoneou msoAum> an m oA x N ouau :Amuum zuAa 6A- m- commounsoo one: :aAga A E: 0AA a o v Hmmu Sofia one .aA .mAm GOA ow E 225m. m3: ow ow ON 0 Uaomo H A. TmtoAxm A In. I erccA H A. I m Uacmw H L. I N A a I 00% A w A p A a cow [9cm] $83818 8081. 26 .moumm :Amuum msoAum> 59A: oaooo Am commouasou was: eeAaa A aa a I ea 3 AamAz we awaaae aeAA was .mA .uAa E 225m NEE 2: cm cm as cm 0 . a I A I J . _ I O Tmthxb H w I m 7m.-oAxm H w I m ;-ma-oAxm H w I A Uaoow H A. m I m I com A L A s A .r A w A . com [Edi/II] 883818 3081 27 .mousuauoasou msoAua> an m oA x N mama aAauum :uAa NI at commouasou who: sows: A a: m u on v au=u mean one .eA .wAA 00A ow ANA 225m may: ow ow ON 0 0.0mm 1 Poem 0.03. 0.08 .I. 79.0? meter—«E 'UJ *4 CQ 0? 5T m - ooh . I _ I _ _ oooA [edit] $88818 8081 28 .mousummoasmu msoAum> an m OH x N mama :Ammum suAa NI «I commomnsoo one: noAna A s: oma u on v A: mama mo mo>uao seam one .NA .wAA E 225m may; 00 0A4 om § I a \1 a! w o! - UoOOOH ........ ermw I I I 6.0% .-m 702m I a om GOA [9am] 883818 510211. 29 .mwumu :Amuum msoAum> :uAa veomm um vommmuasou mama :oAna A a: oMA u on A A2 muse mo mo>uso scam one .mA .NAA :2 23%. ASE. ow ow ON 0 . _ _ o :mtoAxm :mtoAxm ....... M .tmtoAxm .II a 0.0mm I A [8cm] $83818 3081 30 Table 3. The a 8 a under various test condition for pure Ni with R 8 DO 2 136 um. °k to? e -3 -1 -4 -1 ' -5 -1 5 2x10 S 2x10 8 2x10 S T°C 650 --- 75 _-- (85) 850 44 28 22 (36) (25) (23) 1000 --- 20 --- (20) Note: The unit for OR & aS is MPa. Table 4. The D under various test condition for NiBAl with 5 Do a 170 um. 05 e I 2::10‘35'1 2x10"‘s’1 | 2::10'55’1 T°C 650 1.7 3 3.5 750 2 2.5 --- 850 3 4 6 1000 8 14 30 Note: The unit for D is um. 31 structure ( Fig. 19 - 23 ) with very tiny nuclei occurs along the original grain boundary while it is subjected to compress at 650°C and 750°C. Until 850°C, the small grains ( Fig. 24 — 26 )almost grew up within the old grains and replaced them. This phenomenon is remarkable when the strain rate lowered to 2 x 10'55-1( Fig. 26 ). At 1000°C, we can observed that grain growth ( Fig. 27 - 29 ) occurs very strikingly and becomes larger ( Fig. 29 ). Considered the flow curve and compared with the initial grain size ( D0 = 170 um ) in Table 4, all of the DS are much smaller than the DO , give rise to a single peak flow curve. As narrated before, the smaller of Do , the higher of chance of the transition from single peak to multipeak. We prepared a much smaller D0 which was prestrained to 60 X at room temperature and annealed at 800°C for 1 hour to inspect whether the transition would take place or not. A series of metallographs ( Fig. 30 - 35 ) were taken, we can discover that the D is smaller than Do . The largest one, Ds - 4.4 pm S ( Table 5 ), is interesting. Applying the transition criterion, Do / 2DS = 1, we found the appearance of the single peak behavior. Therefore, all of the curves show the samples undergo the continuous RX process. As for the pure Ni with Do - 136 um , carefully compared with each Ds ( Table 6 ), it is out of question that occur the oscillation RX process ( Fig. 36 - 40 ). 32 .Aumvazon :Aamw mafiumfixo any N:OHa omsuosuum oomeuo: on» maozm :oAmmmuaeou mouua As: oNH n 0A 4 N saAs u.oma as saws mAz me aaaauaAAsawe was .aA .uAa A-am- a A A< 33 .Auavazon :Amuw mafiumAxo on» macaw wmsuuzuum oumaxuoa one macaw . Im IoA x N :AAS oaomo as o A me commounsoa .A E: 0AA n n v H< “2 we :nmuuoHHmums 6:9 .oN .mAm .Aumvazoa :Aauw maAamAxo ecu macaw ouaaxuo: as» macaw . Im ION x N :AAS o.ons as ewmawaaeae .A ea oAA I on A A