)V1531_J RETURNING MATERIALS: P1ace in book drop to unuuas remove this checkout from .—:—. your record. FINES um be charged if book is returned after the date stamped below. PHASE EQUILIBRIA IN A T1-6WT%Al-2NT%Nb-IWT%Ta- INT%M0 ALLOY: MORPHOLOGY, CRYSTAL STRUCTURE AND SOLUTE PARTITIONING By Daniel Oraegbuna Nnamdi Obikwelu A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY IN METALLURGICAL ENGINEERING Department of Mechanics, Metallurgy and Materials Science 1982 ABSTRACT PHASE EQUILIBRIA IN A T1-6NT%Al-2WT%Nb-lNT%Ta- lNT%Mo ALLOY: MORPHOLOGY, CRYSTAL STRUCTURE AND SOLUTE PARTITIONING by Daniel Oraegbuna Nnamdi Obikwelu The objective of this investigation was to establish the crystallographic and morphological aspects of various phases in a Ti-6wt%Al-2wt%Nb-lwt%Ta-lwt%Mo alloy (henceforth called Ti—62ll) as a function of thermal history. Various equilibrium and metastable phases can exist in such an alloy depending on the isothenmal holding time and temperature in the two;phase or single phase region. The crystal structure and morphology can also be modified by the cooling rate and/or quenching. The high-temperature equilibrium B-phase can either be retained on rapid quenching or it can transfonm to two possible martensitic phases, a' and a"; where the a' is a hcp unit cell and the a" is orthorhombic. Whether the 3-phase is retained or the a' or thecfl'martensites form, depends on the solute supersaturation of the equilibrium B-phase. In fact the solute supersaturation of the B-phase suppresses the martensitic start temperature, Ms, and martensitic final temperature, Mf and if the Ms and/or Mf are below room-temperature then retained 3 is obtained. As-received Ti-Gle alloy samples were equilibriated at 102500 (l00% B-phase) in an argon atmosphere, down quenched to 900°C, 800°C, 700°C, 600°C and 500°C, and then isothermally annealed and quenched to room temperature. Optical microscopy, scanning electron microscopy, wavelength dispersive x-ray analysis and x-ray diffraction studies were conducted to correlate structure, morphology and alloy chemistry. Based on these results it was shown that quenching after 900°C, 800°C and 700°C, isothermal annealing produced an a + 8 structure. For both 600°C and 500°C, the structures obtained were (a + a") and (a + a“) respectively. The B-phase in all of these cases was, of course, retained B and the a phase could be either equilibrium a or martensitic a'. Based on the x-ray diffraction of the as-quenched alloys a tentative partial phase diagram for Ti-Gle was suggested. Peak splitting was observed only on the x-ray pattern of the samples in which a" was identified. a2, Ti3Al phase was not detected in this alloy in the present study. A The solute partitioning study showed that the Al-rich regions were Ta-poor and vice versa. Dedicated to my mother, Ezelagbo Anuka ii ACKNOWLEDGMENTS The author wishes to express his immense gratitude to Professor Kali Mukherjee and Professor Masaharu Kato for their prodigious guidance and encouragement in our numerous meetings for the successful completion of this academic endeavour. Appreciation is expressed to Professor Robert Summitt and Professor K. Baker, the Director of Electron-Optical Research Center, for serving in my guidance committee and allowing me to use the electron-optical facilities for this research. A special expression of gratitude is due to Professor David Sikarskie, the Chairman of the Department of Metallurgy, Mechanics and Materials Science, for creating an exceptionally peaceful atmosphere, resources, and facilities for the completion of this research. The author's gratitude also goes to the following people without whose help and moral support the success of this research would have been a mirage: nurwife, Victoria, Professor 0. J. Montgomery, Professor Nicholas Altiero, Mr. Leo Szafranski, the technician, Mrs. Diane Sorensen, the typist, and Mr. V. Shull, the Electron Microprobe Specialist. Finally the author wishes to profoundly thank the U. S. Office of Naval Research under Contract No. N00014 82-K-0268 for providing the fund for this research. TABLE OF CONTENTS Chapter 3222. LIST OF TABLES .................. vii . LIST OF FIGURES ........... . ...... ix INTRODUCTION ................... l 2 REVIEW OF PREVIOUS STUDIES ............ 4 2.1 Titanium Alloys--An Introduction ......... 4 2.2 Phases and Phase Transformations in Titanium Alloys ...................... 5 2.3 Morphology of Equilibrium Phases in Titanium Alloys ...................... 13 2.4 Titanium Alloy Martensites ............ 18 2.4.l Hexagonal (a') Martensite . . . . . . . . . . . . 19 2.4.2 Orthorhombic (a") Martensite ........... 20 2.4.3 Crystallography of a' and a" Martensites ..... 23 2.4.4 Other Titanium Martensites ............ 28 2.4.4.l Face Centered Tetragonal Martensite ....... 28 2.4.4.2 Stress-induced Products ............. 29 2.5 Omega Phase ................... 31 2.6 Interface Phases ................. 34 2.7 Decomposition of Supersaturated Phases ...... 35 2.7 l Alpha-phase ................... 35 2.7.2 Beta-phase .................... 38 2.8 Phenomenological Crystallographic Theory of Martensitic Transformation ............ 39 2.8.l Mathematical Formulation-Large Deformation Theory 44 2.8.2 Strain Energy Minimization Criterion-Small Deformation Theory. . . , ............ 47 iv O —l (A) (a) h w (A) (A) w (A) w m N (A) (A) O O I O O O O C O 0 EXPERIMENTAL DESIGN AND PROCEDURE ............ 54 Heating Treating of Ti 6211 ............... 54 Selection of Test Conditions .............. 54 Procedure ........................ 54 Analysis of the Heat Treated Samples .......... 55 Optical Microscopy ................... 55 X-ray Diffraction .................... 55 Electron Microprobe Analyzer .............. 57 Specimen Preparation .................. 57 Optical Microscopy ................... 57 X—ray Diffraction. . .......... ' ........ 57 Electron Microprobe Analyzer .............. 57 Instrumental and Experimental Limitations ........ 58 Heat Treating ...................... 58 ARL Electron Microprobe Analyzer ............ 58 EXPERIMENTAL RESULTS .................. 60 Brief Description of the Results ............ 60 X-ray Diffraction .................... 60 Optical Microscopy ................... 75 Scanning Electron Microscopy .............. 81 Electron Microprobe Analyzer .............. 84 DISCUSSION OF EXPERIMENTAL RESULTS ........... 87 SUMMARY AND CONCLUSION ................. 99 BIBLIOGRAPHY ...................... 103 APPENDICES ....................... 109 Appendix II III IV VI VII VIII IX XI XII Table A-1. Structural analysis of Ti 6211 annealed at 5000C for 48 hours in argon atmosphere and quenched in water at room temperature .............. Table A-2. Structural analysis of Ti 6211 annealed at 600°C for 48 hours in argon atmosphere and quenched in water at room temperature ............... Table A-3. Structural analysis of Ti 6211 annealed at 7000C for 48 hours in argon atmosphere and quenched in water at room temperature ............... Table A-4. Structural analysis of Ti 6211 annealed at 800 C for 48 hours in argon atmosphere and quenched in water at room temperature ............... Table A-5. Structural analysis of Ti 6211 annealed at 900°C for 48 hours in argon atmosphere and quenched in water at room temperature ............... Table A-6. Structural analysis of Ti 6211 annealed for 30 minutes at 10700C in argon atmosphere and quenched in water at room temperature .............. Table A-7. Structural analysis of the as-received Ti 6211 plate prior to heat treatment ......... Accurate lattice parameter calculation for cubic structures with correction in sample displacement . . . Accurate lattice parameter calculation of the hexagonal close packed phase. . . . . . . . . .......... Calculation of volume fractions of a', B using the direct comparison method ................ Extinction rule for orthorhombic structure ....... Figure A-3. Atomic scattering factor of titanium versus sine e/x ........................ vi 109 110 111 112 113 _114 115 116 120 121 123 130 Table NOQNOSUI-DMN _a o u—l -—l 12 13 14 15 16 LIST OF TABLES Phase transformation in titanium alloys .......... Omega phase ........................ Two types of a-phase ................... a/B interface phase .................... Formation and decomposition of orthorhombic a" martensite. Possible reflections of HCP phases (a, a') in Ti 6211. . . Possible reflections of orthorhombic (6") phase in Ti 6211 Possible reflection of the BCC (B-phase) in Ti 6211. . . . Possible reflections of the BCC (B-phase) in Ti 6211,8000C 26 angle for samples annealed at 500°C.for 48 hours in argon atmosphere and quenched in water at room temperature Experimental 26 angles for samples annealed at 600°C for 48 hours inargon atmosphere and quenched in water at room temperature ........................ Experimental 26 angles for samples annealed at 700°C for 48 hours in argon atmosphere and quenched in water at room temperature ........................ Experimental 26 angles for samples annealed at 800°C for 48 hours in argon atmosphere and quenched in water at room temperature ........................ Experimental 26 angles for samples annealed at 900°C for 48 hours in argon atmosphere and quenched in water . . . . Experimental 26 angles for samples annealed 30 minutes at 10700C in argon atmosphere and quenched in water at rOOm temperature ........................ 26 values of the as-received Ti 6211 plate prior to heat treatment ......................... vii Page 14 15 16 16 66 67 68 69 7O 71 72 17 18 Calculated diffusion data for aluminium in beta-phase for Ti-4.55 w/o AL ...................... g. . 91 Calculated diffusion data for aluminium in the alpha-phase for Ti-4.l w/o AL alloy . . ................. 92 viii Figure 01-th O5 10 11 12 13 14 LIST OF FIGURES Partial phase diagram of the alpha-stabilized system. . . . Partial phase diagram of the beta isomorphous system. . . . Partial phase diagram of the beta eutectoid system ..... Constitutional diagram of titanium-aluminium system . . . . Constitutional diagram of titanium-niobium (columbium) system ........................... Constitutional diagram of titanium-tantalum system ..... Ti-Ta phase diagram (National Bureau of Standards) ..... Illustration of mechanism proposed by Burgers for transformation of body-centered cubic lattice to hexagonal close-packed lattice in zirconium ........ Lattice correspondence of martensites showing the four unit cells of the b.c.c. crystal and its face-centered tetragonal equivalent ................... Crystal structure change from B.C.C. to alpha prime and alpha double prime .................... Lattice correspondence and distortion proposed by Bain for the FCC BCC (BCT) martensitic transformation in ferrous alloys ........................... Schematic representation of inhomogeneous lattice-invariant shear by (a) internal twinning and (b) internal slip within martensite plates ..................... Sphere-ellipsoid representation of homogeneous deformation in which matching planes of no distortion exist (cross-hatched) Calculated habit plane normals for titanium alloy martensites ................ ' ........ ix 24 26 27 41 42 51 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 3O Calculated habit plane normals for Ti-Mo alloy martensites ......................... Heat treating equipment ................... Typical X- -ray diffraction pattern for Ti 6211 annealed at 500°C, 600°C, 700°C, 800°C and 900° for 48 hours and quenched in water ...................... Optical micrograph of the as-received Ti 6211 shbwing fine alpha + beta colony microstructure ............. Optical micrograph of Ti 6211 annealed 30 minutes at 900°C and quenched in water at room temperature .......... Typical alpha-colony and basketweave microstructure for Ti 6211 ........................... Optical micrograph of Ti 6211 aged for 48 hours in argon atmosphere at 500°C, 600°C, 700°C, 800°C and quenched in water ............................ Scanning electron image of Ti 6211 annealed for 30 minutes at 900°C and quenched in water at room temperature ..... Scanning electron image of Ti 6211 showing the alpha blocks in the trijunction ..................... Qualitative solute partitioning of aluminium in the sample annealed for 90 minutes at 500°C and quenched in water . . . Solute partitioning of aluminium and tantalum in Ti 6211 annealed 20 hours at 500°C and quenched in water, showing mirror image partitioning tendency of AL and Ta ....... Solute partitioning of aluminium and tantalum in Ti 6211 annealed 20 hours at 600°C and quenched in water showing mirror image partitioning of AL and Ta ........... Beta-transformation at 500°C ................ Ti-AL phase diagram showing effect of 0.50 a/o Mo on Ti-Al binary system ........................ Schematic illustration of the microstructures for samples annealed for short times at 800°C and 900°C and quenched in water at room temperature ................ Schematic a and a + 8 phase boundaries in Ti 6211 alloy based on quenched structures ................ X 53 56 73 75 76 77 81 82 83 84 85 86 89 94 98 101 CHAPTER 1 INTRODUCTION Ti-6Al-2Nb-lTa-1Mo (Ti 6211) has been developed for possible marine applications requiring thick sections. The microstructure of a near alpha or a.+ B.alloys such as the Ti 6211 has a great influence on various mechanical properties such as strength, ductility and fracture toughness. a + 6 alloys generally (1) have good low cycle fatigue properties because of fine-grained mixture of a + 8 structures with relatively high proportion of primary alpha. Fracture toughness can be improved by raising the solution temperature to give a micro- structure with coarser widmanstatten alpha transformed from beta. Ductility may be achieved if beta can be retained at the same time having dispersed fine martensitic phases. The beta-Ti alloys have been widely used for their good high strength/high toughness potential, but their major setback is that they exhibit low aged tensile ductility especially in the transverse direction of rolling (1). Hence beta-Ti alloys have restricted use especially in forgings or sheet rolling. Detailed study of alternative alloys like near 3- or a + 3 alloys is warranted. A host of microstructural features can be produced in Ti 6211 by varying hotworking conditions, heat treatment and variations in quenching or cooling rates. Apart from the microstructural variations produced by a deliberate and controlled processing and heat treatment parameters, some unControlled and generally unavoidable microstructural variations result during welding, especially in thick sections. Although some of the microstructural components in this class of alloys are produced by a diffusional decomposition, there are several diffusionless and marten- sitic phases. Furthermore, a martensitic phase or phases can occur in these alloys, under an applied stress. Such a stress-induced martensitic transformation can occur if a metastable beta-phase exists in the alloy either due to a rapid cooling rate or a supersaturation 0f the beta-phase with beta-stabilizing elements at the solution treatment temperature. Depending on the heat treatment in the a + 3 field, metastable beta-phase can be retained (2) in Ti 6211 alloy. A possibility exists that such a metastable beta-phase will transform martensitically. under a service load. James and Moon (3) have discussed the effects of stress-induced martensite on the mechanical properties of three Ti alloys. These authors (3) have attributed the observed low yield strengths, low values of elastic modulus and high internal friction to the stress induced martensitic transformation of the metastable beta-phase. Recently (2) two martensites, alpha prime (hexagonal) and alpha double prime (orthorhombic) were reported for Ti 6211 alloy. However, no detailed study of the extent, formation and stability of these phases and solute partitioning at various times and temperatures has been done. A unified investigation of the aforementioned aspects of the as-quenched and aged Ti 6211 is accomplished by the present study. In addition accurate lattice parameters of these phases are calculated and the strain energy minimization criterion formulated by Mura et al (4) and the lattice correspondence and crystallography proposed by Mukherjee and Kato (S) are described for possible application to Ti 6211. Although the stabilizing tendencies of aluminum, niobium, tantalum and nolybdenum are known, the partitioning tendencies of these solute elements in the as-quenched structure of Ti 6211*and other titanium alloys are not known. Solute partitioning characterization of Ti 6211 at various times and temperatures would help evaluate the effectiveness of the component elements in Ti 6211. This work also clarifies the solute partitioning tendencies in the various titanium as-quenched microstructures at various times and temperatures. This dissertation is presented in six chapters. Chapter 1 describes the rationale for the present study. Chapter 2 reviews the existing literature relevant to the present study. Chapter 3 describes the experimental procedure and techniques employed in the present study. In Chapter 4 the results of the experiments performed are presented. In Chapter 5 the results presented in Chapter 4 are discussed in detail with existing theories. Chapter 6 summarizes the findings of the present study and concludes with recommendation for further research. *Chemical Composition (w/o): A1=5.8, Nb=2.06, Ta=0.96, Mo=0.06, Fe=0.04, C=0.03, Ti (balance). CHAPTER 2 REVIEW OF PREVIOUS STUDIES There is a wealth of information concerning phase transformations in titanium alloys available in the literature in recent years. Some of the information are complementary, some contradictory and others are not on a specific titanium alloy. In this review an attempt is made to bring these pieces of information under the same umbrella. .2.l Titanium Alloys--An Introduction Titanium alloys are classified into the following structural types, depending on the alloying elements, alpha or near alpha, alpha-beta; beta or near beta titanium alloys (6). (i) The Alpha System: Figure 1 represents a typical binary constitution diagram of alloying elements that stabilize the alpha phase. Alpha stabilizing elements are Al, Ge, Ga, C, 0 and N. The first three elements are of the substitutional type and the others are of the interstitial type. In the alpha-beta coexistence region, these elements are more soluble in alpha than in beta. Adding such elements to titanium stabilizes the alpha phase to higher temperatures. (ii) The Beta System: Alloys containing beta-stabilizers are of two types. Those which are completely miscible with the beta phase such as V, Mo, Ta, Nb are a o Internet-“5c Phase Temperature J, Alloy Content Fig. 1. Partial Phase Diagram of the Al- pha-Stabilized mstem._ The alpha-stabilizing elements are AI, Ge, Ga, C, O, and N. beta isomorphous, Figure 2. Those beta stabilizers that are not miscible with the beta phase transform the beta phase in an eutectoid reaction into beta eutectoid, Figure 3. The beta eutectoid elements which de- compose the beta phase into alpha and an intermetallic phase are Mn, Fe, Cr, Co, Ni, Cu, and Si. Unlike the beta and alpha stabilizers there are neutral alloying elements, namely Zr and Sn which are extensively soluble in both alpha and beta phases. They act as strengthening agents. Alloy Ti 6211 is composed of 5.8 w/o aluminum, 2.06 w/o niobium, 0.96 w/o tantalum and 0.6 w/o molybdenum. From the alloy element class- ification alunfirumi is an alpha stabilizer; niobium, tantalum and molybdenum are beta stabilizers. Constitutional diagrams of Ti-Al, Ti-Ta, Ti-Nb and Ti-Mo systems are shown in Figures 4-7 (6, 95, 96). 2.2 Phases and Phase Transformation in Titanium Alloys Phase transformations and resulting phases in titanium alloys can be classified into eight groups after Murakami (8) namely, 1. Decomposition of beta-phase during quenching producing alpha prime, alpha double prime, athermal omega, retained beta. 2. Isothermal transformation of metastable beta phase producing isothermal omega, alpha double prime, beta prime, alpha, 02, 82. 3. Decomposition of retained beta phase producing isothermal omega, beta prime, widmanstaten alpha, raft alpha, a2. 82. stress-induced martensite, isothermal (thermoelastic) martensite. 4. Decomposition of martensite producing alpha, compound, beta. S . 3 K 3 1‘1 \\"" \ " ‘ 1 \ \ M. All. v v 1 L mm Fig. ‘ 2. Partial PhaeeDiagramoitheBeta bomowhoue Syetem. Alloy!!! elemeuteotthehetaieomor- MWWV,MO,T},MCD. mm Fig. 8. Partial Phase Diagram of the Beta Eutectoid System. Alloying element: of the beta eutectoid typeueMmFe,Cr,Co,Ni,Cu,and a. Figure 4. IEIGNT PEI CENT ALUMINUM so 4.0 so 70 00 90 1000 1 11 1 1 1720' 900 1700 ‘ .— “:‘17\ 2' » N L :: aoo— -1 16 x ' 700— 0.15 . 1500 :2 mo‘ I? 665 _‘ }¢93\ “a 'r’wau I “00 .’ I \\‘ o iVlt%TiZ -\ \ 1 \ ‘ 1340' ‘ 1300 : \ C t’ (z «.9 / a \ z.‘zoo ZA {Tl IL, 3 / I c 7 fl 31106 A & T // E \ U .. I a 1000 ! 1.: i W 900 fi 1 eez‘ 'I' ’ 36.5 49.: H '°° «24.5) l(35.1.) w I I II I I :1 I l J 700 , , | ‘ I 1' “5° 660° I | l 1' I m I ll i i i | 1 l u (we :00 i l l J o 20 0 so so so 70 so 90 100 TI ATOIIC PERCENT ALuumuu at Constitutional diagram of titanium-aluminum system. Cimmifim thoNb a o :0 x n 50 to 70 a M m .0 T T I I T T 1 V 7 * I / 4/9 I” “’2” . I I”’r / 3:00 ’I’ / /” / / 1 ,4 ’1‘“? «3500 ,w’ M L” / 1 4’ 1‘ ['0’ I, I .4 ’A/ , s / 3 .m t - & k p . i rwa E «:m" moo «raw \\ 4 a“ an! “"h.‘ «7300 WI 1 ‘~~m w ""0 20 an p re :0 so 70 a) 30 ’ cmumm LIP/nor (ob Figure 5. Constitutional diagram of titanium- niobium (columbium) system. 10 mm» thJe 0100000 «0 50 60 I0 00 90 100 (533 T1 -71-1 T T l ___l I 51004:.--- . _ ‘1 ‘ L. I, - "’1” 1 m z’ / ~ W P 4? /:/// l' ‘r m I}, // . " “7' - 0000 L ,2/ V l— /’ .. I I 2706‘ r I .- ’/VA 4 W t e. .- I A a . \ I g , Z' / g : fl 1 k E“ p .. m, - J 750:; _ J )- 'iNW 7.299 ’ 2mm '\ 4 ”H! 3" ~ 1000 ’99 l \ 0 10 80 .10 I0 50 80 70 80 .90 100' mm: etJJa m 1 P \ \ '\ - M0 '\ \A k\ _ _______. 1 ‘\. S . \'\_ 5 70 ——-oc+p \ E \.\,\\b\ .‘E‘ l \\ ' _‘J 500 I l 20 M 0 ’0 with 8.7.1: Figure 6. Constitutional diagram of titanium- tantalum system. .. 0 . lcmpcl ulurc L 11 Weight Percent Tantalum Figure 7. v v v vvv vv '7 v v v . vv vvv v w 22‘. 30 #6 SC ‘ “ 1’ at U. I Atomic Percent 'z'cn:c!um Ti-Ta Phase Diagram vvvvvvvv vv vrvvv 8. 12 Alpha-phase formation through omega and beta prime, omega + alpha: beta + alpha (Type 1 alpha and 2 alpha). Eutectoid decomposition, pearlitic constituent or massive product. Precipitation from alpha-solid solution producing compound, (I . 2 Formation of interface phase, the alpha/beta interface phase. A brief explanation of each phase shown above is given below: 1. Beta phase: The high temperature allotrope of titanium with a body centered cubic crystal structure (BCC). Alpha: The low temperature allotrope of titanium with a hexagonal close packed crystal structure. Alpha prime: A hexagonal martensite. It is a super- saturated, non-equilibrium alpha phase formed by a diffusionless transformation of the beta phase. Alpha double prime: An orthorhombic martensite. It is a supersaturated, non-equilibrium orthorhombic phase formed by a diffusionless transformation of the beta phase in certain titanium alloys. Omega phase: A hexagonal titanium phase that forms either from quenching from B-phase field below the omega start temperature or isothermal aging of e-titanium alloys. Athermal omega: The type of omega that forms during rapid quenching from the e-field to temperatures below the omega start temperature. It forms as uniform dispersions of 20-40A° precipitates. 13 7. Isothermal omega: The type of omega that forms during isothermal aging of e-alloys or retained beta. It is generally either ellipsoidal or cuboidal in morphology. 8. Isothermal (thermoelastic) Martensite: The type of martensite in which there is a balance between the strain energy generated in the parent phase and the chemical free energy made available by the lower free-energy state of the martensite phase. 9. Beta prime: Decomposition product of metastable beta. It is a solute lean bcc zone. . ' 10. a2: Ordered phase based on Ti3Al in Ti-Al alloys with more than 10 atom a/o aluminum. ll. 82: Ordered bcc phase formed by the decomposition of retained beta. Its formula is TizMoAl in Ti-Mo-Al alloys. 12. Interface phase: A beta decomposition product that forms along the alpha/beta interfaces of Widmanstatten structures. Phase transformations and morphology in titanium alloys are summarized on Table 1-5 after Murakami (8). 2.3 Morphology of Equilibrium Phases in Titanium Alloys The nucleation and growth of alpha phase after cooling from the beta region result in a plate-like morphology of alpha which is often described as Widmanstatten structure. In alloys which have a relatively low con- centration of alpha phase, these plates often form in colonies or stacks. The formation of the colony and basketweave microstructures depends on the cooling rate from the beta or a + 3 regions and the alloy chemistry. In the basketweave alpha plates with or without interleaved beta, platelets occur TABLE 1 PHASE TRANSFORMATION 14 IN TITANIUM ALLOYS , TRANSFORMATION PROCESS DECOMPOSITION 0F 8 - (i) Ti Martensite: 6—-°a'. n" PHASE DURING QUE NCH- ING (ii) Athermal u - Phase : B-—0 “an? 0 A DECOMPOSITION OF B - PHASE 1N ISOTNERHAL TRANSFORMATION 8 (6 I o) -_. n ' “rich‘ ulean 8 ' 6‘rich‘ 0 8loan Type (i) Isothermal a -Phase : Br --*’ ui * B DECOMPOSITION OF (ii) Phase Separation : 6‘ --~ 8' t B III RETAINED S - (iii) a - Phase Formation : 8 --~ a t B PHASE I Hidmanscitten a4,_§§£tio ) (iv) Compound Precipitation : Bt-*-Odered 0; t B 8g“ Odered 32 T 8 (v) Stress - Induced Martensite (vi) Isothermal ( Thermoelastic ) Martensite I (i) Hexagonal o' ( B - isomorphous )——~e t O DECOMPOSITION Hexagonal o' ( B o eutectoid ) ——-¢ 0 t CW W or MARTENSITE (ii) orthorhombic a” ( H§.>>rooa teIIp )—- a o a Orthorhombic a“ ( e eirooe temp ) : eF-» B qPHASE FORMATION (i) B t u——' n t B V THROUGH INTERMEDI- ATE PHASE (ii) 8 6 B'—* a t B ' EUTECTOlO (i) Active Eutectoid: Pearlitic Constituent v1 ~ (4Cellular Reaction ) DECOMPOSITION (ii) Sluggish Eutectoid: Massive Product ( Cellular Reaction )7 PRECIPITATION FROM (i) Compound Precipitation Vll n - SOLID SOLUTION (11) 02 Phase Precipitation VIII FORNAT ION OF INTERFACE L— L PHASE n / 6 Interface Phase Formation during Slow Cooling 15 l . I TABLE 0 OMEGA PHASE I SPACE GROUP. ogd‘ P3ml (06h- P6/nm,when u=ll2 ) f ATOM POSITION. ( 0.0.0 ) : t( 1/3. 2/3. u ) : u'O.SZOS0.003 AND LATTICE a= 4.60725kX. c=2.82123kx.c/a = 0.613 ( Ti-SZCr alloy ) CONSTANT “WWW" (112O)U//(110)8 . (0001)...“111) e RELATION Athermal w Isothermal u n MORPHOLOGY 3 very fine particle ellipsoidal (low misfit. long axes] (11081 cuboidal ( high misfit. cube face/l (lOO)8) FORMATION displacement controlled Diffusional controlled reaction nucleation followed by “EC”A"‘5” displacement fluctuatio RELATION BE- TWEEN 0 AND solute lean u - Type 1 __. Type 2 w -PHASE 16 TABLE III THO TYPES OF 0 - PHASE TYPE I 0 TYPE 2 o' ( Obeying Burgers relation Not obeying Burgers relation 22:31:30" - (llO)B// (OOOI)m (101210011) twin orientation - to Burgers orientation 0 t (111) 810120)“ I FORMATION Initially formed during aging Transformed from Type l T—fl after longer holding 3 Monolithic plate (ternary alloy) Colony of fine particles MORPVOLOGY Monolithic needle (binary alloy) ._.Iu,_ - __.. L§2:2:?§:ATlON Nucleation and growth more likely than mechanical twinning TABLE IV a / “ lNTERFACE PHASE - l MORPHOLOGY CRYSTAL STRUCTURE ORIENTATION RELATION -.S~_---__H._.__....__“m..--.e-- --__--”_n...__ I Icc I a = 436 pm ) (OOOl)n//(lll)v ; (ll-20>n/(TIOM ' HONOLITHIC or (Single Crystal) I“ I d = 432 pm. (DOWN/(”ll IteAllOTB; STRIATED LAYER (Polycrystalline) c/a = I.I3 ) cIIéo).-.// Ito/(Tim.- hexagonal _ “DION/(1010),»: (COMM/(1213);, hexagonal . ( Burgers related. Often twinned on ‘; (TOIIIQ ' TABLE V FORMATION AND DECOMPOSITION 'OF ORTHOROHMBIC a" 17 MARTENSITE TRANSFORMATION PROCESS FORMATION i) ii) iii) Decomposition of Metastable 8 during Quenching B --+ a") (8) Decomposition of Retained e by Intermediate ( Bainitic ) Transformation during Isothermal Aging _1— 3' A Stress-Induced Transformation from Retained B B ——» a" + Twinned B i DECOMPOSITION i) Decomposition of o" by both Spinodal Decomposition and Reverse Martensitic Transformation during Cooling and Aging. 0” ____. a" + a" . _. a” + 8 lean rIch lean L_2 I “T1T- I Decomposition of a" in the Normal Way a + B a“ --+ B 18 in colonies in a Widmanstatten structure. In colonies there are the alpha platelets having nearly identical orientations. In addition to the above-mentioned microstructures, alpha mucleates on the prior beta grain boundary forming the grain boundary alpha (Gba). Such alpha grows rapidly along the boundary and frequently leads to a continuous layer of grain boundary alpha. In addition to the alpha morphologies which result from nucleation and growth or from tempering of martensite, hotworking or thermomechanical processing at temperatures in the two phase alpha + beta region has an important effect on alpha-phase morphology. Thermomechanical processing results in in-situ deformation of the alpha phase which is present at the working temperature and this deformed alpha subsequently recrystal- lizes. The recrystallization reaction leads to the formation of equiaxed alpha particles. 2.4 Titanium Alloy Martensites Generally in quenched alloys the martensitic structure typically exhibits a plate-like morphology and in some alpha alloySIIlath or packet martensite can be formed. In the plate-like martensite the individual plates each tend to have a different crystallographic orientation. This is in contrast to the colony microstructure formed by nucleation and growth in which all plates in a colony have the same crystallographic oriéntation. The martensitic structure also decomposes during subsequent aging to form a relatively inhomogeneous distribution of beta-phase precipitates in a matrix of alpha. 19 2.4.1 Hexagonal (Alpha Prime) Martensite The most prevalent athermal Ti martensite is the hexagonal alpha prime martensite. Two morphologies of hexagonal martensite have been observed, namely massive martensite or packet or lath martensites (9,10) and acicular (10,11,12) martensite. The massive martensite was found to consist of colonies which were resolvable optically (10). The interiors of these colonies exhibited surface relief effects; thin foil electron microscopy showed that these surface reliefs resulted from parallel martensite plates contained within the colonies, all of which belonged to the same variant of the orienta- tion relation (10). The boundaries bewteen the plates have been shown to consist of walls of dislocations, whereas the interior of the plates contained a relatively dense tangle of dislocations (13). The conmon difficulty of indirect methods employed by various workers was that an orientation relation between the martensitic product and the parent beta-phase must be assumed. Assuming the Burger relation (14) (110) beta // (0001) alpha, beta // alpha the following habit planes have been obtained: iodide Ti, {8.9.12} beta (9), Ti-Cu, {lOTllalpha (13), Zr Nb {334} beta (14). These results indicate that for the massive martensites habit plane may vary with alloying elements. With increasing solute (decreasing Ms temperature) the massive martensite colony size was found to decrease (l3) and at a solute con- centration which was shown to vary between alloy systems the colonies totally degenerated to individual plates (12), each being a different variant of the orientation relation. This alpha prime martensite was shown to be frequently internnally twinned on (1011) e'. 20 In sufficiently concentrated alloys an incompletely transformed structure could be produced permitting verification of the orientation relation as the characteristic Burgers relation, ie. (llO)beta // (001) alpha prime, [llllbeta // [11201alpha prime (16.17.18). Habit plane .determination on the acicular martensite have been performed and these martensites have. for the most part, the {334lbeta habit plane (12.18.19) which had been predicted by the Bowles and Mackenzie crystallographic theory of martensite. Several workers have also reported a {344lbeta (12.18) habit plane and the possible origin of this had been discussed by Hammond and Kelly (20). The kinetics of alpha prime decomposition is rapid compared with those of the super saturated alpha phase. The dislocations in the alpha prime martensites act as heterogeneous nucleation sites. In many systems the equilibrium precipitate in the alpha prime decomposition such as the beta-phase is heterogeneously nucleated. hence curtailing or precluding the occurrence of uniformly nucleated metastable transition precipitates. However in several beta-eutectoid systems solute-rich coherent zones have been observed during alpha prime decomposition. Such zones have been reported in Ti-Si alloys containing more than 2wt%Si (21) and Ti-Cu alloys containing about 1.5w/oCu (22). In Ti-Cu. disc zones nucleated between dislocations in the mar- tensite substructure during aging at low temperatures. At high temperatures heterongenous nucleation of the equilibrium phase at dislocations and interfaces dominated the decomposition process. 2.4.2 Orthorhombic (Alpha Double Prime) Martensite This martensite occurs in binary alloys like Ti-Mo. Ti-Nb, Ti-W. Ti-Re and not in others like Ti-V. In alloys such as Ti-V which do not 21 formalpha double prime. addition of aluminum was shown to stabilize the orthorhombic phase over relatively wide composition ranges (23). Alpha double prime was recently observed in the commercial a + B alloy Ti-6Al-2W -4Zr-6Mo and its occurrence in this alloy was shown to be accompanied by a reduction 'hi tensile ductility (24). The lattice parameters of orthorhombic martensite were strongly composition dependent (13,23). The decomposition of alpha double prime has been discussed by several workers (25-27) but they disagreed on the early stages of alpha double prime decomposition. Thin alpha prime platelets formed parallel to (100), (010) and (110) alpha double prime planes during the early stages of tempering of alpha double prime in Ti-W alloys (25). Bywater and Christian (26) examined the tempering of alpha double prime in Ti-Ta alloys whereas Young et al (27) examined the tempering process in Ti- 6AL-25n-4Zr-6Mo. Both investigators suggested that the initial preci- pitates are beta-phase. Williams and Hickman (25) in addition presented X-ray diffraction data to support their observation and their data showed that the alpha-phase precipitation led to a simultaneous increase in the orthorhombic distortion of the alpha double prime matrix. Reexamina- tion of the results of Young and the others shows that the time-temperature combination they have used led to a fairly advanced stage of decomposition. The Ti-W system studied by Williams and Hickman is a eutectoid system while the Ti-Ta and Ti-6A1-2Wn-4Zr-6Mo are isomorphous systems. Thus the composition changes required to precipitate beta-phase in the latter 22 system should be much smaller than the Ti-W system where the beta-phase is essentially pure tungsten. In view of this. it was suggested that the a" +'a" +-o decomposition might be peculiar to the Ti-W system (28). However both investigators agreed that at the advanced stages of the decomposition the structure was a + 8. Thus the apparent disagree- ment may result from the change in the decomposition processes from a"'+ a" + ae (one is orthorhombic martensite enriched in beta-stabilizers) during the very early stages of tempering to a" +.a + 8 during the later stages or at higher tempering temperatures. Finally in binary Ti-W alloys the a" + a mixture transformed to a + B by a cellular reaction (25). This reaction was not observed in Ti-6Al-ZSn-4Zr-6Mo and this is consistent with earlier direct nucleation of beta-phase in this quinary alloy, compared to the binary Ti-W alloy where most convincing evidence of a" +~a" + “e has been obtained. - ,The end result of alpha double prime decomposition during tempering is generally agreed to be a + 8. Two distinct modes of formation of this mixture are the cellular reaction and the heterogenous nucleation of beta-phase similar t6 that seen in the case of alpha double prime. Davies et al (29) recently proposed that the transition in martensite: crystallography from hexagonal alpha prime to orthorhombic alpha double prime occurred with increasing solute content and hence decreasing Ms, and that at least one alpha double prime point of inflection muSt exist in the alpha double prime free energy. They proposed a form of the free energy/composition relationships for alpha prime. alpha double prime and beta phases at a certain temperature within the a + B field, and also the corresponding phase diagram showing the chemical Spinodal of alpha double prime and Ms of the beta phase. 23 2.4.3 Crystallography of o' and a" Martensites The Burger's mechanism for the beta (BCC) + alpha prime (hcp) (14) Fig. 8 transformation in Titanium alloys is still tenable. Also the formation of the alpha double prime (the orthorhombic phase) and alpha prime can be explained in terms of the beta-stabilizing element composition and simple atomic shuffling. Oka and Taniguichi (30) found that with increasing Mo content a beta-stabilizer, the lattice parameter of the stress induced orthorhombic phase approached that of the beta-phase. Sasano et al (31) confirmed that with increasing beta-stabilizing agent the crystal structure of Ititanium martensites changes from hcp to orthorhombic. T. W. Duerig et al (32) and Sasano et al. tried to explain beta + alpha prime or alpha double prime by simple atomic movements. Duerig et al (32) reported the following atom positions of the orthorhombic alpha double prime martensite in a Ti-lOV-2Fe-3AL alloy: (0,0,0); (a/2, b/2. 0), (0,2b/3, c/2); (a/2, b/6, c/2). The lattice parameters of the orthorhombic cell is reported to be: a = 3.01Ao b = 4.83A0 c = 4.62A° In addition to alpha prime and alpha double prime, f.c.o. (face centered orthorhombic) martensite was obtained in Ti-12V alloy bulk crystal (33). It has been pointed out that all the titanium martensites (alpha prime. alpha double prime. f.c.o.) can be generated from a face centered tetragonal equivalent of the four unit cells of the b.c.c. beta phase (Figure 9). 24 ..$: cope—e .5358? E 35»: vuxuoaiomopo pocomoxm; 3 3.5qu 953 vocoucooiéoa o8 5555322» .8“. mamas-a E 3833 .5228... mo 85950.3: .w 853“. < :31 «no: 12.1 <93. 4 will 5.8 a 2... n3 _. 5 LN-a \ dubs-w LNI‘U :2. \\\\ . 1.7.; «as-aseo as 25 Mukherjee and Kato (5) unified these physical interpretations of beta-+ alpha prime or alpha double prime and proposed that for the alpha prime, the Bain strain is brought about by Burger's mechanism where the operation of the Ill2} B shear systems followed by a small volume increase results in the (0001)“. plane from the (Oll)B plane. In addition to the above Bain strain a aB[OlT]/6 shuffling on every other (Oll)B plane is necessary to create the h.c.p. structure. The formation of alpha double prime martensite can be explained in the same way. The difference lies in the fact that after the operation of the {112} B shears, the (Oll)B plane does not completely change into the hexagonal close packed plane but stops midway between the (Oll)B and (0001)“. leaving the angles 61 and 62 Figure 10 slightly different from 120 degrees. Thus the alpha double prime martensite can be con- sidered a "distorted" alpha prime martensite. In the case of f.c.o. martensite no shuffling is necessary and the lattice correspondence was very similar to that in Au-Cd martensite (33). From the Bain strains calculated in Ti-12.6V by Mukherjee and Kato (S) for the f.c.o. martensite it is conclusive that f.c.o. martensite unlike alpha prime and alpha double prime cannot be considered a "distorted" alpha prime martensite. Sasano et al (31) and other investigators have demonstrated that orthorhombic phase x-ray line derives from the hcp line with increasing Mo or beta-stabilizing element. for example in Ti-Mo-3 w/o Al, for O to 5 w/o M0 the hcp lOTO line is intact. but for 6 w/o to 12 w/o No the hcp lOTO line split into {110} and {020} orthorhombic lines. the {lOTl} hcp lines split into {111} and {021} orthorhombic lines and the separation 26 x (Tom, I x 2(001), Il / \yIOIOI, Figure 9. Lattice correspondence of martensites showing the four unit cells of the b.c.c. crystal and its face-centered tetragonal equivalent. 27 100030 (OlTlp ——~> (01 (Comic, planes TIIOOIO. IOlOlau (01113 planes \ 0 Is: and 3rd layer oloms e 2nd layer oloms (b) (0000. planes Figure 10. Crystal structure change from B.C.C. to alpha prime and alpha double prime, after Mukherjee and Kato (S). 28 between these twin orthorhombic lines increased with increasing Mo content. The {020} and {lll} lines of the orthorhombic approached gradually the {002} of orthorhombic or {110} of bcc respectively as No content became higher. Alpha double prime was difficult to retain in thin foil form, however it was found to be internally twinned in a Ti-8.5 Mo-O.SSi on (111) planes (25). 2.4.4 Other Titanium Martensites 2.4.4.1 Face centered tetragonal martensite— In addition to the well known alpha prime and alpha double prime titanium martensites. several workers have observed face-centered tetragonal martensite with c/a ratio nearly unity in Ti-Fe (35). Ti-Cr (16). complex ternary titanium alloy (35) and quartenary titanium alloys (34,24). This martensite was reported on the basis of thin foil electron microscopy evidence; recent evidence obtained by Williams and Rhodes shows that this martensite is orthorhombic in its bulk form (24 ). For example, if a Ti-l4V—6Al alloy was quenched from 900°C. X-ray diffraction evidence obtained from bulk samples showed that this alloy contained orthorhombic martensite. whereas electron microscopy examination of thin foils prepared from the same sample used for X-ray diffraction showed evidence of face-centered cubic martensite. Furthermore. light microscopy of the thin edge of the electron microscopy specimens showed a marked rumpling characteristic of Shears resultant from mechanical instability along the thin edges of the foils (24). In other alloys which have bulk orthorhombic martensite, thin foil results showed that the martensite reverted to the bcc, beta-phase along the edge of the foil during thinning. Thus the balance of the evidence at present indicates that the face-centered cubic and/or face-centered tetragonal martensites 29 reported by several workers on the basis of thin foil electron microscopy evidence, are in fact thin foil version of a bulk orthorhombic marten- site. Oka et al (33) referred to earlier again reported a face-centered orthorhombic martensite on Ti-ll.6wt%V. alloy. 2.4.4.2 Stress-induced products . There is conflicting evidence regarding the nature of martensite produced in metastable bcc Ti alloy during stressing. In some alloys. clear evidence for mechanical twinning (36.37) has been presented while evidence for a stress induced martensite has been obtained in other alloys (24,37). It appears that the con- centration and type of solute has an important influence on whether the 'stress-induced product in a given alloys is twinning or martensite. Electron microscopy of Ti-ll.5Mo-4.5$n-62r (B-III); Ti-l4Mo-3Al (24.36) showed the occurrence of bcc twins in these alloys. Also for some other alloys the fact that {122} twinning (37,38) and {322} twinning (36.39) could occur during stressing of metastable bcc alloys was verified. There is good evidence to Show that other alloys do form stress-induced martensite and in these cases the martensite product has been shown to have orthorhombic structure. Koul and Breedis (40) reported a contrary evidence of a hexagonal stress induced product in Ti-V and Ti-Mo alloys. However if one considers the relatively orthor- hombic distortion which can occur in some alloys like Ti-lOat%Al+V (41) this difference can be resolved thus; the separation between the (110) alpha double prime and (020) alpha double prime and the (111) alpha double prime and (021) alpha double prime is sufficiently small that high re- solution X-ray diffraction is required to differentiate between a hexagonal and an orthorhombic product. 30 Additionally. in the present case; the straining results in some line broadening which makes resolution difficult. Williams (41) studied the above subject involving 25-50 different alloy compositions including those in which hexagonal streSs induced martensites have been reported and was unable to reproduce the hexagonal stress-induced martensite but observed the orthorhombic alpha double prime phase. The factors which control the occurrence of twinning or martensite formation during stressing are not clear but it appears that Al addition to Ti-Mo and Ti-V alloys promotes stress-induced alpha double prime. Blackburn and Feeney (36) analyzed the occurrence of the {112}B and {332}B twin modes and found that alloys in which the parent bcc beta- phase contained athermal omega-phase exhibited {332} twinning, whereas more concentrated alloys characteristically twinned on {112}B (38). In a number of beta-alloys. Md lies above room temperature. As a result, stress-induced reactions have been reported in the beta- stabilized Ti-Mo (42) and Ti-V (40) systems as well as in other complex systems. The structure of the stress-induced martensitic products in beta-Ti alloys was reported as FCC (35), HCP (43). and recently as orthorhombic (44.45). T. W. Duerig et a1 (32) studied the stress- induced transfonnation in Ti-lOV-2Fe-3Al. They conclusively found that the transformed microstructure contained orthorhombic martensite (alpha double prime) and the martensite plates were accompanied by mechanical twinning. The orientation relationship between this martensite and beta-matrix was shown to be (HOIB // (0min. [1T1]8 // [1101an 31 Oarkfield imaging (32) of alpha double prime spots showed the martensitic plates to be "mottled" or "spotty." Oka and Taniguchi (30) in their study of the crystallography of stress-induced products of some Titanium alloys again found SIM (stress- induced orthorhombic martensite) and a {332} twin of the beta-matrix in Ti-9 ~l6%Mo deformed_by compression or cold rolling. SIM formed in Ti-9 ”11%Mo had their habit planes close to {443} plane. ' 2.5 Omega Phase In alloys where the martensite reaction is suppressed by sufficient concentration of beta-stablishing elements which depress the martensite finish temperature beneath room temperature, beta-phase can be retained in metastable form by quenching. In such alloys the beta-phase was shown by Silock et a1 (46) and Bagarjatskii et al (47) to partially decompose during quenching into a metastable precipitate called omega-phase. Athermal omega was considered for sometime to form by a martensitic transformation largely because its formation could not be suppressed by quenching. Blackburn and Williams (37) using Electron microscopy showed that athermal omega occurred as extremely small (20-40A°) particles with a very high density. Such a product is not compatible with existing theories of martensitic transformation since the key features such as habit plane cannot be assigned to such small particles. It thus appears that athermal omega is not a martensitic product at least in the classical sense. A new class of transformation called displacement controlled transformation, was proposed by de Fontaine (48) who used omega as an example of such a transformation. 32 Paton. de Fontaine and Williams (56) used cold storage transmission electron microscopy to show that the omega-phase could be reversibly formed on cooling beneath room temperature. Such reversibility is consistent with a displacement controlled transformation which occurs without change in composition. The displacement controlled reaction argument can also account for a related phenomenon, namely the complex distribution of diffuse scattering which is observed in beta-alloys. The streaking has been ascribed to random or uncorrelated displacements of the atoms so that the omega-phase lattice is not generated but the periodicity of the bcc lattice is reduced. The athermal omega-phase was shown to be trigonal in heavily beta- stabilized alloys but became hexagonal in leaner alloys (49). The formation mechanisms of athermal omega has been suggested to be displacement-controlled reaction while that of isothermal omega was diffusion controlled nucleation followed by displacement fluctuation (SO-54). Recently Duerig et al (55) suggested that a composition alloy invariant growth mechanism does not differ from "classical" diffusion controlled growth. that as the alloy content becomes leaner or the beta- structure less stable. the compositionally invariant growth mechanisms begin to dominate and a certain "unusual" growth phenomenon can be observed. which distinguishes the beta + omega reaction from most others. They observed a morphology transition from ellipsoids to cuboids in Ti-lOV- 2Fe-3Al at a constant size. Hence they proposed that the morphology change could result from a rapid, chemically invariant omega particle growth, followed by a slow composition equilibration. There is evidence that athermal omega-phase reverts during reheating to the isothermal aging temperature to much lower density isothermally formed particles. 33 This suggests that each athermal omega-particle does not act as a pre-existing nucleus for isothermal omega-phase. Also the incubation period which preceded the isothermal omega-phase formation as shown in (54) suggests that the athermal omega-phase was reverted and the iso- thermal omega-phase formed by nucleation and growth. The similarity between bcc beta-phase and the hexagonal omega-phase (48.56) can account for the athermal 6 :Im transformation. It would also result in a low barrier to nucleation and subsequent growth which would lead to rapid precipation kinetics of isothermal omega-phase. Both athermal and isothermal forms of omega-phase are coherent. this also reflects the similarity between beta and omega. Isothermal omega phase is produced by isothermal aging of alloys that are relatively dilute as well as alloys richer in beta-stabilizing alloy additions. Isothermal omega-phase was shown to have a cuboidal or ellipsoidal morphology depending on the precipate/matrix misfit (57,58). It was observed that the isothermal ellipsoidal omega has low precipate/ matrix misfit, its long axis are parallel to and cuboidal omega has high misfit the cube face is parallel to (lOO)B. This misfit arises in the isothermal case because the omega-phase is solute lean and therefore leads to enrichment of the beta-matrix. The effect of ternary additions of Zr, Sn, Al and oxygen on the occurrence and stability of isothermal omega-phase was studied (59). The elements reduced the occurrence and/or stability of omega-phase. Both Sn and Zr increased the stability of beta-phase when present together with other beta-stabilizers like Mo or V and thus these elements reduced volume fraction of omega-phase. Al promoted early separate 34 nucleation of the alpha-phase and the occurrence of this competitive reaction product also reduced the time of stability of omega-phase by consuming it via diffusional growth of the alpha—phase. 2.6 Interface Phase A reaction product which has an apparent close relationship to the type 1 and type 2 alpha-phase formation is the interface phase. This is a reaction product which forms along the alpha/beta interfaces of Widmanstatten structures in alloys such as Ti-6Al-4V. The interface phase has a somewhat variable morphology (60). It forms in extensive quantities during continuous cooling from temperatures above 950°C. Attempts to produce and control this product by isothermal reaction studies were unsuccessful and it appeared that it only formed athermally (61,62). The reasons for this are nOt clear at present. In addition, the nature of the interface phase is still subject to discussion. Rhodes and Williams (60) Suggested that it is a heavily twinned and dislocated region of the Widmanstatten alpha plates. The interpretations are tied to each of these investigators' suggested mechanism of interface phase formation and each has some difficulties associated with it. For example, Rhodes and Williams (60) suggested that the interface phase is a nucleation and growth product. but if this is the case. the apparent athermal character of the reaction is difficult to explain. Margolin et al (63) suggested that the interface phase forms as a result of mechanical twinning which arise because of stresses at the interface which are created by thermal expansion mismatch of the alpha and beta phases. If this is true. then one might expect the interface phase to be very sensitive to cooling rate. This does not appear to be 35 the case as shown in (61). Secondly, if the interface phase is simply localized mechanical twinning. then each twin should have accomodation dislocations in the matrix ahead of its terminus; this generally is not the case. Finally. Rhodes and Williams (60) reported a monolithic face- centered cubic layer of the interface phase which formed at intermediate cooling rates. It is possible that this is a transition reactiOn product which subsequently decomposes into the striated. hexagonal interface phase. Recently Chenu et al (64) found that in Ti685 the alpha/beta interface phase was of two forms, the monolithic (ML) and a striated (SL) layer and was dependent on homogenisation temperature in the B0 or a + B fields and the cooling rate. The layers have f.c.c. crystallographic structure and in twin relationship with the ML twin axis. The formation of the interface and its width wasinfluenced by the Mo content of the beta phase and the cooling rate. The ML and SL layers have identical composition. Two different orientation relation- ships were found between ML, beta and alpha phases in the same sample 1. {llO}B // {OOOlla // {lll}ML <111>B // <1210>a // <110>ML 2 (110)B // (OOOl)a // (OOl)ML <111>B // <1210>a // <110>ML Banerjee and Arunachalam (65) in their study of two Titanium alloys IMI 685 and BT9 maintained that the f.c.c. interface phase does not form as part of the transformation sequence beta-rfxc.c + alpha, it rather had precipitated on the alpha side of the alpha/beta interface at a fairly late stage of the beta-ralpha transformation. They concluded that beta is retained whenever the beta + alpha transformation occurs 36 by nucleation and growth and the interface phase is observed at all alpha lath/retained beta interfaces. but not when the beta + alpha transformation is martensitic. the f.c.c. interface with lattice para- meter of 404A0 exists in two crystallographic orientation relationship to alpha, namely (lTO) // (1100)“; [001] // [000110 and (111) // (0001)“; IlTO] // [1120]a. type I and type II f.c.c. respectively. They suggested that the f.c.c. phase is a hydride of titanium because there is no f.c.c. based phase in any binary systems of their alloy concen- tration levels except for gamma-hydride. the lattice parameter of the f.c.c. phase is in close agreement with that of gamma-hydride, and alpha are identical to those observed between hydride and alpha. 2.7 Decomposition of Sppersaturated Phases There are two principal modes of decomposition of the super- saturated alpha phase: precipitation of a second phase and ordering of alloys whose compOsition is in the 25at%Al range. The principal pro- ducts of the precipitation reations are the ordered o2 phaSe. intermetallic compounds. such as TiZCu or Ti55i3; the hydride phase, Tin; and compounds which are based on the rare earth elements such as Y or Er. 2.7.1 o2 Formation The alpha-phase which contains more than lOat.%Al is supersaturated and can decompose during aging at 500°C. This decomposition of supersaturated Ti-Al alloys containing 10at%Al has been observed by many workers (66.67) and there is general agreement that the decomposition product is an ordered phase based on Ti3Al. This product is usually designated the o2 phase and has a 0019 structure. 37 Only relatively recently it was shown that the a2 phase formed as uniformly distributed, coherent precipitates (67). Further. the exact position and extent of the a + o2 phase field is still subject to some question. The morphology of coherent o2 during early stages of formation is equiaxed whereas at later stages (larger sizes) it is ellipsodial with the long axis lying along the 'c' axis of the matrix and precipitate. This morphology can be accounted for by considering the precipitate misfit, which is 0.35% parallel to [0001] and 0.83% perpendicular to [0001]. It has been shown that the precipitate aspect ratio and particle size limit for coherency can be altered by the addition of ternary solutes such as Sn and Ga (68). Of these. Sn causes the o2 misfit to become more isotopic whereas Ga causes it to become more directional and larger along the 'a' axis. As a result. the a2 phase is nearly equiaxed in Ti-Al-Sn alloys and forms a long. slender. semi-coherent rods in Ti-Al-Ga alloys. Another variable which influences the formation of oz in Ti-Al alloys is the amount of oxygen present as an impurity. Paris and Williams recentlyinvestigated the role of oxygen in (.2 formation (69) and found that a 2000 wt ppm increase in oxygen content in a Ti-7.3wt%A1 alloy led to an increase of 100°C in the coherent a/o + o2 solvus position. The reasons for this are not clear at present but several factors pertain. First oxygen expands the alpha-Ti lattice which increases the misfit of the a2, at least when the unconstrained lattice parameters of the "pure" alpha and o2 phases are considered. This would be expected to suppress the coherent solvus rather than raise it. Gehlen (70) has suggested that the structure of the o2 phase is distorted and he has used x-ray diffraction measurements and intensity calculations re ex On to su tel be‘ (Hi We uni bcc Nb. 38 to support this. Assuming that the distortion described by Gehlen really exists, some of the octahedral holes in the ca lattice will be expanded. This tends to increase solubility of oxygen in the a2 phase. On this basis, Paris and Williams (69) suggested that oxygen partitions to the a2 phase and stabilizes it to higher temperatures. 2.7.2 Beta-Phase In alloys which contains between 7 and 25at%Al the addition of sufficient quantities of beta-stabilizing elements still permits re- ,tention of the bcc allotropic form during quenching. This Al rich beta—phase exhibits several decomposition modes which are somewhat different from those of the lower Al beta-phase. In a Ti-16at%Al-7at%Mo alloy the metastable beta-phase decomposed to form an ordered bcc precipitate (71) designated B2 (72). The 32 phase forms as a fine, uniform dispersion of ordered, coherent precipitates in a disordered bcc matrix. By contrast to the Ti-Mo-Al alloy described above, in Ti- Nb-Al and Ti-Nb-W-Al alloys containing 25at%Al the beta-phase trans- formed to lOOvol% B2 during quenching (73-75). This beta-phase contains fairly large thermal antiphase domains; this suggests that the order- disorder temperature for formation of the B2 (CSCl) structure is reasonably high so that growth of the ordered domains occurs until they impinge leaving a fully transformed structure. Although the order-disorder temperatures have not been experimentally determined in these alloys. the following explanation for the differences between Ti-Mo-Al and Ti-Nb-Al is suggested. The effect of Mo on the suppression of the beta + alpha transus temperature is much greater than that of Nb. Thus the 39 beta-phaseis stable to lower temperatures in the Ti-Mo-Al alloy. As a result. beta-phase instability will occur at a lower temperature where diffusion is slower in the Ti-Mo-Al alloy. This may limit the extent of ordering kinetically. In the alloy compositions in which the ordered bcc phase is observed, it becomes unclear which types of atoms occupy the A and 8 sites in the AB supperlattice. Even if it is assumed that 1 Ti occupies the A sites and all other solutes occupy the B sites. there typically are insufficient 8 atoms to achieve stoichiometry. For example. the ordered bcc phaseINasobserved in Ti-25at%Al-lOat%Nb and in Al-25at%Al— %at%Nb-lat%W yet in these alloys the total Al + Nb or Al + Nb + W amounts to less than 50at%. Therefore. the B2 lattice must be stable over a wide composition range. Selected area electron diffraction patterns obtained from aS-quenched samples showed extensive streaking along (110)82. Bouchard and Thomas (76) reported similar effects in (Cu-Mn)3 A1 alloys where a 82 lattice decomposed into a L21 phase and a 003 phase. It is possible that a similar decomposition reaction is occurring in Ti-Al-Nb alloys. Further work to examine this point is warranted. The equilibrium structure of the Ti-Al-Nb system is o2 (0019) + 6(Bl) and thus is different from that of the (Cu-Mn)3 Al alloys which is Cu3AL (003) + CuMnAl (L21). Thus if the 82 +-L2] reaction does occur in the Ti base system. it must be a metastable state relative to 02 ‘1’ B, 2.8 The Phenomenological Crystallographic Theorypof Martensitic Transformation The phenomenological crystallographic theory of displacive martensitic transformations was developed simultaneously by Wechsler. Read and Liberman (77) and Bowles and MacKenzie (78). In 1924 Bain (79) suggested 40 that the bcc structure of iron can be produced from the fcc by a contraction of about 17% in the direction of one of the austenite cube axes and an expansion of about 12% in the plane perpendicular thereto. Figure 11. This Bain Model and many other theories summarized by Cohen (80) were not able to account for all of the important crystallographic features of martensitic transformation in a self-consistent general manner, that is. if the crystal structure change was correctly described. the habit plane and/or orientation relationship usually was not. The Wechsler, Lieberman and Read theory (W-R-L) asserts that the crystallographic features of martensite transformation can be completely explained in tenns of three basic deformations: l. A Bain distortion, which forms the product lattice from the parent lattice, but which in general does not yield an undistorted plane that can be associated with the habit plane of the deformation. 2. A shear deformation which maintains the lattice symmetry (does not change the crystal structure) and, in combination with the Bain distortion. produces an undistorted plane. In most cases. this undis- torted plane possesses a different orientation in space in the parent and product lattices. 3. A rotation of the transformed lattice. so that the undistorted plane has the same orientation in space in both the parent and product crystals, Figures 11. 12, 13. No attempt wasmade in W-L-R theory to give physical significance to the order of the steps listed above and the entire theory is best viewed as an analytical explanation of how one lattice can be formed from the other. z‘bExPWS‘O“ «e \ ' Figure 11. 41 Lattice correspondence and distortion proposed by Bain (79) for the FCO+BCC (BCT) martensitic transformation in ferrous alloys. The correspondence cell (heavy lines) in the parent phase becomes a unit cell of martensite after a homogeneous lattice deformation. The principal dis- tortions along x', y', and z' are indicated. 42 Figure 12. Schematic representation of inhomogeneous lattice-invariant shear by (a) internal twinning and (b) internal slip within martensite plates after (81). 43 Figure 13. Sphere-ellipsoid representation of homogeneous deformation in which matching planes of no distortion exist (cross- hatched) after (81). 44 On the other hand Bowles and MacKenzie (78) B-M theory. almost identical but mathematically somewhat different to the W-L-R theory. has adjustable parameters in the form of a uniform dilation in the interface (habit plane). The surface dislocation approach of Bullough , and Bilby (82) is mostly closely related to the matching method of Frank (83) and Bilby and Frank (84). For excellent summaries and critical reviews of the theoretical and experimental work on martensitic phase transformation the reader is referred to Bilby and Christain (85) and Cohen (80). 2.8.1 Mathematical Formulation of the Martensitic Transformation-- _Large Deformation Theory (BIT The basic equation of B-M (78) and W-L-R (77) theories can also be presented in matrix form for completeness thus: E 8 R S B ...................... I ....... (l) where B is the Bain distortion, S is a lattice-invariant shear (mathe- matically following the Bain distortion). R is the rigid body rotation and E is the resulting shape deformation. B. S, R. and E are all (3 x 3 matrices. If the rotation R returns the plane left undistorted by S B to its original position, then E = R S B.is an invariant plane strain (IPS). Although equation (1) Shows the inhomogenous Shear S following the Bain distortion, the same mathematical result is obtained by "allowing" the shear to occur in the parent phase prior to the Bain distortion. By the latter reasoning the basic equation becomes E = R B S ............................ (2) where 5, like S represents a simple shear. In the formulation by Wechsler, Lieberman and Read (77) RB is the lattice deformation part of the observed shape change. while S is the 45 fine-scale inhomogeneous part. Although the operation of S may not be apparent from the macroscopic homogeneity of the shape deformation, there is now ample evidence from electron microscopy that such lattice- invariant defOrmation do occur on a fine scale. The equation E . R B S can be rewritten as proposed by Bowles and MacKenzie (78) - E S"1 . R B = lattice deformation ............... (3) 5-1 corresponds to a shear deformation of the same magnitude and on the same plane as S. but in the opposite direction. It can be regarded as the invisible portion of the lattice deformation which is cancelled by S. and is sometimes called the complementary shear. With reference to the lattice correspondence in Ti 6211 in the beta martensite transformation asff+ bu. aBIE-T'ao'Ij- the Bain distortions n1 = aa'/a8 n2 . ba./aa[2 n3 = aa.[37a8]2— Thus the Bain distortion for BCC + HCP can be written as C T. ./a O O 1 a 8 3 e o bayani O .......... (4) L0 0 aa.[3/a8 s This B-matrix does not have an undistorted plane and so cannot conform to the invariant-plane condition. 46 The IPS shape deformation (E) can be expressed as (78) E 8 I + mdp' ................. . ........ (5) where I is the unit (3 X 3) matrix, p' (p transpose) is (l X-3) row matrix designating the normal to the invariant plane. d is a (3 X 1) column matrix representing the direction of the shape displacement. and m_is the magnitude of the displacement. By means of B-M theory (78) the invariant line strain can be calculated if the Bain distortion is known (from the correspondence and lattice parameters) and if the inhomogeneous shear mode) that is the plane and direction) is presumed. S'1 in equation (3) being a simple shear is also an IPS and the 1 product ES- 3 L 2 R8 is an invariant line strain, defined by the, intersection of the two planes which are invariant to E and,S"1 respectively. The crystallographic concepts and their mathematical formulation presented have both successes and discrepancies. A departure from the basic IPS criterion has been introduced in the form of a “dilatation parameter" (78) to improve the agreement between the predicted and measured habit-plane orientations. This empirical adjustment relaxes the requirement that the habit should be undistorted. and so the shape deformation is not a true IPS. There would be no way to physically justify the existence of such arbitrarily determined parameters. Other attempts to account for such difficulties have included the assumption of two or more inhomogeneous Shear systems (86,87). However. it was pointed out by Kato (88) that the introduction of three independent IPS's in the phenomenological theory makes any arbitrary plane as a habit plane. 47 Thus, the introduction of such new adjustable parameters is both dangerous and meaningless. 2.8.2 Phenomenological Analysis Based on the Strain Energy_ MifiimizatTOn CFTterion--Small Deformation Theory Mura et al (4) and Kato et al (89) have shown that there is a one-to-one correspondence between the phenomenological crystallographic theory of martensitic transformations (78) and the strain energy minimization approach based on the small deformation theory (4, 89). According to this analysis. if the transformation strain eTU (Bain strain plus lattice invariant strain) in X1 - X2 - x3 Cartesian coordinate system satisfies the condition. T=T=T= 6 ll 5 22 5 12 the strain energy becomes minimum (zero) and the condition in equation 6 O . . . . . . . . . . . . . . . . . Q . . (5) with the X3 - axis as an invariant plane (habit plane) normal. is equivalent to the invariant plane condition in the pehnomenological theory. The advantage of using the small deformation theory is that the calculation becomes much simpler and even if the three principal components of the Bain strain are different. such as the case in the Titanium martensites. the final solutions can be obtained in simple analytical forms. Mukherjee and Kato (5) applied this strain energy theory to analyze the crystallography of the alpha prime. alpha double prime and face-centered orthorhombic martensites. The Bain Strain in the t-system for the three martensites can be written in the form of elij 3 0 £2 0 .................. (7) O 0 c L 3. t the Bain distortion,ii= c + l. 48 They found that e1 < O (contraction). e2. e3. > 0 (expansion for the orthorhombic (alpha double prime) and the hexagonal (alpha prime) martensites whereas e1 > O and e2. e3 < O for the f.c.o. martensites. If equation 7 is viewed based on the x-y-z system in the parent beta phase (beta system) shown in Figure 9 the following equation results c.' 0 O T clij = o (62 + e3)/2 (c3 - c2)/2 ......... (8) .0 (c3 - 521/2 (82 ‘1' 69/24 (Variant 1) It has been reported that the lattice invariant shear for each martensite occurs dominantly by twinning and the twinning planes are {1011} alpha prime, (20) {Ill} alpha double prime (29) and {Ill} f.c.o. (33). All these twinning planes correspond to the (110) beta plane of the parent beta-phase Figure 9. Thus by using the beta system. the twinned variant should have the Bain Strain of . V52 I 831/2 0 (32 ‘ e3V2 Ezij 8 0 ET 0 eeeeeeeeee (9) From equations 6. 7 the total transformation strain , eTij can be written as cTij = f e‘ij + (l - f) czij, (O < f < l) ........... (10) where f is the volume fraction of the variant 1 which is to be determined. If the unit vector 5, normal to the habit plane is written in the beta system as K = [sin 6 cos a. sin 6 sin o. cose] ............. (11) By rotating the coordinate system, a new X1 - X2 - X3 system can be found in which the total transformation strain (90) satisfies the 49 invariant plane condition of equation 6. In this new system '6 becomes parallel to the X3-axis. Equation 6 gives three simultaneous equations with f. 6 and o as the three unknown parameters. Solving these equations these results are obtained tan a 8- . . . ........... (12) 2f81- (1 "' f) (62 + 53) tan a = $1" 9 {2(2f ' 1’61 T (I ‘ 2f) (52 T 53)} ..... (13) (£2 - c3) {(1 - f) tana + f} 2 _ 1 1 A - 4AB 1' ' 2 A = {3.1 (.2 -.:3)2 + 2(c1-cz) (52 + 5:3) (a1 - c3)}c/4 . . . . (14) B = ' e18283 The authors Mukherjee and Kato (5) showed that solutions corresponding to the positive and negative signs in equation 14 are crystallographically equivalent. Also. it should be noted that there are two independent (crystallographically different) solutions for the orientation of the habit plane. depending on the positive and negative values of tan a in equation 12. Using the Bain strains the authors calculated the habit plane normal. The volume fraction of the variant 1 were calculated from equations 12 through 14 for each of the three martensites. This present analysis was compared for the case of alpha prime martensite in Ti-Mn with that of Knowles and Smith (91) who based their analysis on the phenomenological theory and essentially the same orientation of the habit plane normals are predicted. one of the solutions belonging to the class A (o+. m+) and the other to the class A (o-. 6+) in B-M notation (92) Figure 14. From this comparison the authors (5) concluded that the small-deformation theory which does not involve any 50 numerical calulation can be used in lieu of the phenomenological theory to discuss the crystallography of the titanium alloy martensites. Furthermore it has been found that in Ti-Mn alloys (20.91) there are two different types of alpha prime, one with the {334}B and the other with the {344}8 habit planes. None of these two solutions for the alpha prime accurately explains the above habit plane indices. Hamond and Kelly (20) were thus forced to use an adjustable "dilatation" parameter, a. in the phenomenological theory to bring the irrational habit plane normals to the observed indices. ’Knowles and Smith (91) analyzed the zig-zag parent-martensite interface in the {344}B martensite on the basis of the Burgers vector minimization criterion of the interface disloca- - tions. According to their calculation the total magnitude of the Burgers vector averaged over the entire interface becomes zero. in other words no long range strain field is created due to the transformation and the invariant plane condition in the macroscopic habit plane is still satisfied in this model. Thus, the orientation of the calculated habit plane on the average remains the same as that predicted from the phenomenological analysis shown in Figure 14. The reason adduced by Mukherjee and Kato (5) for the discrepancy between the calculated orientations of the habit plane and the observed {334}B or {344}B is that Hammond and Kelly (20) as well as Knowles and Smith (91) used the lattice parameter of the beta phase at room tempera- ture. Since beta + alpha prime martensitic transformation takes place at a higher temperature the lattice parameter should be compensated by taking into account the thermal dilatation. Davies et al (29) on the other hand have systematically obtained the lattice parameters of the beta phase due to the change in the molybdenum content. According to their 51 OIO parallel lo / (0001100 ”0 [OOIIQN f (CHDII Io°-°-\ Twinning " plane _ Class Ala-«n+1 00' I00 100 . - A /1 —- I 3I3 0? 3 3 a Class A‘afuit) 434 334 o f.c.o. I; a’ by Knowles 8 Smilh 010 Figure 14. Calculated habit plane normals for titanium alloy martensites. Mukherjee and Kato (S). 52 Observation, the alpha prime martensite formed in the alloy with as much as 4% molybdenum and the alpha double prime martensite formed in Ti-6% molybdenum and Ti-8% molybdenum. Similar tendency from hexagonal to orthorhombic structure change associated with increasing beta- stabilizing elements was reported in other titaniun alloys (31). Mukherjee and Kato further calculated the Bain strains for Ti-Mo from the lattice parameters of the beta and the martensitic phases and applied the strain energy minimization criterion and found as shown in Figure 15 that the habit plane normals lie close to the [4 3 4]B and [4 3 3]B poles and that with increasing beta stabilizing agent in this case molybdenum there is continuous change in the orientation of the habit plane as shown in Figure 15. This finding is a result of the continuous change in Crystal structure from alpha prime martensite to alpha double prime martensite with increasing beta-stabilizing element (31). 53 ' \./ \ - l\ 10' PO pure T1 (at) O Ti-Z'leMo. -4'/oMo (a') \ A T1‘6./0M0 (all) _. TI-BT/0M010") Figure 15. Calculated habit plane normals for Ti-Mo alloy martensites. Mukherjee and Kato (S). IOI CHAPTER 3 EXPERIMENTAL DESIGN AND PROCEDURE 3.1 Heat Treatipg of Ti 6211 3.1.1 Selection of Test Conditions Preliminary heat treatment showed that,with the heat treating equipment,30-35 minutes was adequate for solution treatment at the upper furnace. Longer time intervals were chosen to ensure equilibrium at the various isothermal annealing temperatures. An argon atmosphere and pure titanium getters were used to minimize adsorption of interstial elements like hydrogen and oxygen in the alloy. 3.1.2 Procedure The as-received sample plate was cut into small rectangular blocks (5 x 2 x 2)cm3 and polished in 120 grade abrasive to remove blade marks and flatten the surface. More polishing was done through 600 grit before the sample was hot-worked. The slight hotworking operation was done at about 1090°C and the samples were aircooled. The hotworked metal was profusely polished to remove any alpha case and oxidized layer. The clean samples of final 3 were subsequently polished to 600 grit and dimension (3 x l x l)cm then heat treated in a two-zone coaxial furnace assembly circulated with argon gas. 54 55 The furnace assembly was designed. as shown in Figure 16. to avoid contamination and ensure very rapid quenching rate. The sample was solution treated in the upper furnace for about 35 minutes at around 10300C and step-quenched to the lower furnace which was separately controlled. The equilibrium temperature on the sample was measured by means of a potentiometer and a digital millivoltmeter which was used to monitor temperature. 3.2 Analysis of the Heat Treated Samples 3.2.1 Optical Microspppy_ This technique was used to study the variation of microstructure at various times and temperatures. An Olympus Table Microscope was used to have a quick glance on the samples but photomicrographs were taken with a Heerbrugg Wild Microscope Model M20 with attached 35 mm camera . 3.2.2 X-ray_piffraction X-ray diffraction were performed on all the heat treated samples to identify the phases and index the planes. CUKa radiation was used at 20 KV and 15 ma. 3.2.3 Electron Micrpprobe Analyzer This technique was uSed to determine the elemental distribution of the various microstructures in the heat treated samples. Working con- ditions were 20 xv and lOnA. Beam diameter was : lOOOAo. AlKa. TaLa. NbLa and MoLo wavelengths were used. Thermocouple to tempera tu re V11 potentiometer 56 —1$4_ Argon in 61 - l Figure 16. ———> Heat treating equipment. Sample wires to variac To variac Upper furnace Quartz tube Ceramic tube To controller Lower furnace Sample wires Argon out 57 3.3 Specimen Preparation 3.3.1 Optical Microscopy_ Each heat treated sample was ground flat and any trace of oxide or coloration removed with the water-cooled Buehler Belt Surfacer. Rough polishing was done on 240. 320. 400, 600 emery papers adequately cooled to avoid deformation. Final polishing was done on the polishing wheel with microcloth using 0.05 micron alumina. The sample was etched for 40 seconds in 95 cc H20. 3.5 cc HNO3 and 1.5 cc HF. (Kroll's Etchant) and washed in water and then methanol and dried. 3.3.2 X-rayppiffraction The size and Shape of the sample necessitated the design of a special glass sample carrier into which the sample was placed with the samples' surface on the same level with surface of the glass sample carrier. Each sample prepared for the Optical Microscopy was repolished on a rotating wheel with cloth using 0.05 micron alumina and then etched slightly (15 seconds) in 95 cc H20. 3.5 cc HNO3 and 1.5 cc HF. (Kroll's Etchant). 3.3.3 Electron Microprobe Analyzer The samples for the Probe analysis were prepared as for the x-ray diffraction. 3.4 Instrumental and Experimental Limitations (Situational) 3.4.1 Heat Treating ' Titanium alloys are reactive at high temperatures to elements like oxygen and hydrogen. Hence as soon as the sample was introduced into the furnance. the furnace was flushed with argon gas and a continuous slow flow of argon gas was maintained with the aid of a flowmeter. This 58 flow was normally interrupted during the quenching Operation. At this time of quenching some air must have unavoidably rushed into the furnace. However, it was hoped that this exposure was much shorter compared with the annealing time. Quenching in water has its disadvantage. There was spontaneous nucleate boiling and decoloration of the sample at the instant of quenching. This decoloration was due to the spontaneous reaction be- tween the hot metal and water. This decoloration was. however. removed during grinding. 3.4.2 ARL Electron Microprobe Analyzer The best attainable beam diameter of the microprobe was lOOOAo. Good beam diameter was. however. produced with high voltage and low cur'- rent; hence, in this project 20 KV and lOnA were maintained constant in all the microanalysis. This 20 KV was just good enough to give reason- able counts. To reduce statistical error in the count rates. each spot count was for 1 minute. More than five count rates were recorded for each spot and the mean count rate was taken. The detector in the equipment saturates between 10,000 - 15.000 counts per second. Quantitative microprobe analysis was not done in this project. Solute partitioning tendencies were studied qualitatively in the form of concentration profiles and concentration of various solutes in the various microstructures was measured in count rates. It is assumed that these raw data will reflect the concentration of each solute in the microstructures. 59 The current of lOnA was maintained constant, but it sometimes varied slightly. On the standard carrier, pure elements (standards) were embedded with other elements; for example. aluminum was embedded with cobalt. The possibility of interference of cobalt atoms in the emitted A1 radiation may not be ruled out. CHAPTER 4 EXPERIMENTAL RESULTS After the solution treatment and annealing of Ti 6211 at various temperatures. x-ray diffraction technique was used to identify the stable phases at each temperature after 48 hours annealing time. Detailed analysis of the x-ray results was conducted. The detailed analyses are shown in the Appendix while the summaries are presented in Tables 10 to 16. Based on the sample annealed at 700°C for 48 hours and quenched in water. average values of the lattice parameters for the HCP phase are calculated to be 2.938A° a a C (1 "Similarly the lattice parameters for the BCC and orthorhombic phases 4.670A° were calculated from the samples annealed for 48 hours at 800°C and 500°C respectively and quenched in water. These lattice parameters are for the beta phase as = 3.225A° and for the orthorhombic phase 0 a .. 8 3.04OA a o b .. =4.944A a o c .. =4.637A a 60 61 Based on these lattice parameters a complete Set of allowed reflections for the hcp and BCC phases were calculated. This was calculated using CuKa radiation. The results of these calculations are summarized in Table 6 to 9 with increasing 26 angles. A typical X-ray diffraction pattern for Ti 6211 annealed at 500°C, 600°C, 700°C. 800°C and 900°C for 48 hours in argon atmosphere is . shown in Figure 17. It is not possible to distinguish between the co and o' in the x-ray diffraction work alone since they have the same crystal structure and very nearly the Same lattice parameters. Thus all the hcp reflections in this study are designated alpha. Optical microscopy of the aS-received and the heat treated samples were performed to show morphological features: Figure 18 shows the optical micrograph of the as-received sample. The type fine a + B colonies are observable. Figure 19 shows the optical micrograph of beta-solution treated. held for 30 minutes at various temperature and quenched in water at room temperature. The grain interior alpha and the packets of lath martensite are observable. More comments on this microstructure are included in the discussion. Typical alpha-colony and basketweave microstructures in Ti 6211 obtained in samples annealed at 600°C, 500°C and 400°C and quenched in water are shown in Figure 20 A.B.C, respectively. Figure 21 A.B.C and 0 shows the samples annealed at 500°C, 600°C, 700°C and 800°C for 48 hours and quenched in water. The coarsening of alpha plates is observable. At higher temperatures this coarsening produces more equiaxed structure. 62 Table 6: Possible reflections of HCP phases (a, 0') in Ti 6211. Data taken from the sample annealed for 48 hours in argon atmosphere at 700°C and quenched in water at room temperature; with CuKa radiation. a . 2.9378A c =- 4.66973 HCP 26 (in deg.) (CuKa)* hkil dhkl A II I2 1010 2.544 35.279 35.249 35.340 0002 2.334 38.573 38.540 38.640 1011 2.234 40.373 40.339 40.443 1012 1.720 53.257 53.210 53.352 1120 1.468 63.356 63.298 63.473 1013 1.327 71.966 70.966 71.170 2020 1.272 . 74.611 74.539 74.756 1122 1.243 76.662 76.587 76.812 2021 1.227 77.849 77.772 - 78.002 0004 1.167 82.691 82.607 82.858 2022 1.117 87.287 87.196 87.468 1014 1.061 93.203 93.103 93.404 2023 0.985 103.009 102.890 102.248 2130 0.961 106.682 106.555 106.937 2131 0.941 110.020 109.889 110.292 1124 0.913 115.211 115.061 115.510 *X. = Weighted average of the wavelengths of an unresolved K doublet of Cu radiation. A] = Wavelength of CuKo1 radiation 12 = Wavelength of CuKo2 radiation. 63 Table 7: Possible reflections of orthorhombic (a") phase in Ti 6211. Data taken from the sample annealed for 48 hours in argon atmosphere at 500°C and quenched in water at room temper- ature, using CUKa radiation. a = 3.0403 Atom positions (0,0,0). (1/2.1/2,0), ‘ b a 4.9443 (l/2,l/6,l/2). (0.2/3.1/2). c . 4.6373 26 (in deg.) (Culgl) hkl dhkl A A1 32 110 2.5896 34.647 34.617 34.706 020 2.472 36.342 36.311 36.405 002 2.318 38.850 38.817 38.917 111 2.261 39.871 39.836 39.940 021 2.181 41.399 41.363 41.471 112 1.727 53.024 52.977 53.119 022 1.691 54.245 54.196 54.342 200 1.520 60.952 60.897 60.046 130 1.449 64.286 64.226 64.405 113 1.327 71.034 70.966 71.170 220 1.295 73.068 72.998 73.209 023 1.310 72.099 72.030 72.237 202 1.271 74.680 74.607 74.825 221 1.247 76.372 76.297 76.521 040 1.236 77.176 77.101 77.328 004 1.159 83.389 83.303 83.558 222 1.130 86.036 85.947 86.213 114 1.058 93.547 93.446 93.749 024 1.049 94.598 94.496 94.804 310 0.99269 101.899 101.783 102.133 223 0.99256 101.918 101.801 102.152 311 0.9707 105.157 105.033 105.405 240 0.9589 107.020 106.892 107.277 330 0.8632 126.529 126.341 126.907 313 0.8352 134.747 134.520 135.205 64 Table 8: Possible reflection of the BCC (B-phase) in Ti 6211. Calcu- lation based on various lattice parameters (0.02 A steps) using cuka radiation. a. a0 = 3.21 A 26 (in deg.) (CuKu) hkl dnk1 A AT AZ 110 2.2698 39.710 39.675 39.778 200 1.6050 57.413 57.361 57.516 211 1.3104 72.074 72.005 72.212 220 1.1349 85.575 85.487 85.751 310 1.0151 98.831 98.721 99.053 222 0.9266 112.606 112.464 112.891 321 0.8579 127.952 127.758 127.342 a0 = 3.23 . 110 2.2839 39.454 39.420 39.522 200 1.6150 57.024 56.973 57.127 211 1.3186 71.556 71.488 71.693 220 1.1419 84.926 84.840 95.100 310 1.0214 98.010 97.900 98.228 222 0.9324 111.545 111.405 111.824 321 0.8632 126.529 126.341 126.907 30 = 3.25 110 2.2981 39.200 39.167 39.268 200 1.6250 56.642 56.591 56.741 211 1.3268 71.046 70.979 71.182 220 1.1490 84.280 84.194 84.45 310 1.0277 97.205 97.097 97.42 222 0.9382 110.510 110.374 110.78 321 0.8685 125.157 124.974 125.52 do = 3.27 110 2.3112 38.969 38.936 39.036 200 1.6350 56.264 56.214 56.366 211 1.3349 70.551 70.484 70.685 220 1.1561 83.645 83.560 83.815 310 1.0340 96.416 96.310 96.628 222 0.9439 109.516 109.385 109.788 321 0.8739 123.807 123.630 124.164 Table 9: Possible reflections of the BCC (B-phase) in Ti 6211. 65 Data taken from the sample annealed 48 hours in argon atmosphere of 800°C and quenched in water at room temperature OuKa radiation was used. a0 = 3.2253 20 (in deg.) (CuKa) hkl dhkl X Al I2 110 2.280- 39.525 39.490 39.593 200 1.612 57.140 57.089 57.244 211 1.316 71.719 71.651 71.856 220 1.140 85.101 85.014 85.276 310 1.019 -98.321 98.211 98.540 222 0.930 111.981 111.840 112.262 321 0.861 127.113 126.923 127.496 66 Table 10: 26 angle for samples annealed at 500°C for 48 hours in argon atmosphere and quenched in water at room temperature. Data taken at 22°C with CuKa radiation.* Peak# 26 (deg) (hkl) or (hkil) Phase Identification l 35.38 (1010) ' o 2 28.51 (0002) a 3 39.87 (111) . a" 4 40.21 (1011) o 5 53.11 (1012) a 6 63.30 (1120) a 7 71.95 (1013). (113) o, a" 8 74.56 (2020). (202) a. a" 9 76.45 (221). (1122) a",a 10 77.57 (2021) a 11 87.0 (2022) a 12 102.70 (2023) - a 13 111.95 (2131)o- o *For detailed analysis seed Table A-1 in the Appendix. 67 Table 11: Experimental 26 angles for samples annealed at 600°C for 48 hours in argon atmosphere and quenched in water at room temperature. Data taken at 22°C with CuKo radiation.* Peak# 20 (deg) (hkl) or (hkil) Phase Identification l 35.31 (1010) I a 2 38.60 (0002)a a 3 39.73 (lll)a", o" 4 40.28 (1011) a 5 53.11 (112)a", (1012)a a", a 6 63.23 (1120) o 7 71.90 (lOT3)a. (ll3)a" o, a“. . 8 76.53 (1122). (221) a, a" 9 77.73 (2021) a 10 82.21 (0004) a 11 85.50 (222)." o", 12 106.71 (2130) ' o 13 111.90 (2131) o 14 115.06 (1124) a *For detailed analysis see Table A-2 in the Appendix. 68 Table 12: Experimental 26 angles for samples annealed at 700°C for 48 hours in argon atmosphere and quenched in water at room temperature. Data taken at 22°C with CuKa radiation.* Peak# 26 (deg) (hkl) or (hkil) Phase Identification l 35.31 (1010) a 2 38.60 (0002) a 3 39.53 (110) B 4 40.34 (1011) a 5 53.24 (1012) a 6 63.30 (1120) o 7 70.72 (1013) a 8 71.76 (211) B 9 74.64 (2020) a 10 76.61 (1122) a 11 77.69 (2021) a 12 82.45 (0004) a 13 87.0 (2022) o 14 111.10 (2131)o. (222)B a, B *For detailed analysis see Table A-3 in the Appendix. 69 Table 13: Experimental 29 angles for samples annealed at 800°C for 48 hours in argon atmosphere and quenched in water at room temperature. Data taken at 22°C with CuKa radiation.* Peak# 261(deg) (hkl) or (hkil) Phase Identification 1 35.38 (1010) a 2 38.58 (0002) a 3 39.52 (110) 8 4 40.48 (1011) a 5 53.20 (1012) a 6 57.12 (200) 8 7 63.50 (1120) a 8 71.0 (1013). a 9 71.64 (211) 8 10 74.64 (2020) a 11 76.45 (1122) a 12 77.77 (2021) a 13 82.21 (0004) . a 14 84.53 (220) 8 15 86.85 (2022) a 16 102.78 (2023) a 17 106.19 (2130) a 18 111.60 (2131)a, (222)s a, B *For detailed analysis see Table A-4 in the Appendix. 70 Table 14: Experimental 26 angles for samples annealed at 900°C for 48 hours in argon atmosphere and quenched in water. Data taken at 22°C with CuKa radiation.* Peak# 2e)(deg) (hkl) or (hkil) Phase Identification 1 35.38 (1010) a 2 38.24 (0002) a 3 39.51 (110) 8 4 40.14 (1011) a 5 52.89 (1012) a 6 62.96. (1120) a 7 71.60 (211) 8 8 74.64 (2020) a 9 76.46 (1122) a 10 81.41 (0004) a 11 85.00 (220) e 12 106.25 (2130) a 13 111.00 (2131)6, (222)8 a, 8 *For detailed analysis see Table A-5 in the Appendix. 71 Table 15: Experimental 26 angles for samples annealed 30 minutes at 1070°C in argon atmosphere and quenched in water at room temperature. Data taken at 22°C with CuKa radiation.* Peak# 26 (deg) (hkl) or (hkil) Phase Identification 1 35.31 (1010) a' 2 38.44 (0002) a' 3 40.21 (1011) 6' 4 53.15 (1012) 6' 5 62.97 (1120) a' 6 71.20 (1013) 6' 7 74.16 (2020) a‘ 8 76.50 (1122) 6' 9 77.25 (2021) a' 10 81.57 (0004) a' 11 87.19 (2022) a' 12 92.24 (1014) a' 13 102.25 (2023) a' 14 110.00 (2131) 6' *For detailed analysis see Table A-6 in the Appendix. 72 Table 16: 26 values of the ias- received Ti 6211 plate prior to heat treatment. Data taken at 22°C with CuKa radiation.* Peak#. ' 26 (deg) (hkl) or (hkil) Phase Identification 1 35.30 (1010) a 2 38.55 (0002) a 3 39.60 (110) B 4 40.35 (1011) a 5 53.15 (1012) a 6 63.30 (1120) a 7 71.70 (211) 8 8 74.60 (2020) a 9 76.52 (1122) a 10 77.60 (2021) a 11 82.20 (0004) a 12 102.70 (2023) a 13 106.02 (2130) a 14 110.00 (2131) a 15 114.50 (1124) 01. *For detailed analysis see Table A-7 in the Appendix. 73 Figure 17. Typical X-ray diffraction pattern for Ti 6211 annealed at 500°C, 600°C, 700°C, 800°C and 900°C for 48 hours and quenched in water. 74 +28 7 75 Figure 18. Optical micrograph of the as-received Ti 6211 showing fine alpha + beta colony microstructure. Etchant is Kroll's solution, magnification: 200x 76 Figure 19. Opticalomicrograph of Ti 6211 annealed 30 minutes at 107000, and 900 C and quenched in water at room temperature. Etchant is Kroll's solution, magnification: 200x. 77 Figure 20: Typical alpha-colony and basketweave microstructure for Ti 6211. Etchant Kroll's solution. A: Bas etweave microstructure for sample annealed at 400 C and quenched in water at room temperature. Alpha-colony microstructure for sample annealed at 500°C and quenched in water at room temperature. Alpha-colony microstructure for sample annealed at 600°C and quenched in water at room temperature. Mag. for al1 = 150x 78 79 Figure 21. Optical Micrograpg of Ti06211 aggd foro 48 hours in argon atmosphere at 500 C, 600 C, 700 8000 C and quenched in water. Etchant: Krolls Solution (1. SHF, 3. 5HN03, 95H20). Mag. for all I 200x 80 81 Scanning Electron Microscopy was also conducted to depict some features in the microstructure. Figure 22 shows the scanning electron image of the beta solution treated and subsequent 30 minutes annealing at 900°C and quenching in water. The martensite plates occurring as alpha prime plates, the retained beta and the alpha colony are observed. Figure 22 also shows the heterogeneously nucleated grain boundary alpha. Figure 23 shows alpha blocks in the trijunction. From the Electron Microprobe Analysis further characterization of Ti 6211 is conducted. Figure 24 shows the solute partitioning of aluminum 'hi the sample annealed 90 minutes at 500°C and quenched in ' water. This profile is non-regular because of some inhomogeneities in the heat treated sample. This non-regularity is still observed at long annealing times as shown in Figure 25. Figures 25 and 26 depict the mirror image partitioning of aluminum and tantalum. The aluminun partitions to the alpha-plate while Tantalum partitions to the beta-phase. The alpha-plate here has light colour while the beta-phase has the dark colour. 82 / 8.85 25K 4844. Figure 22. Scanning Electron Image of Ti 6211 annealed for 30 minutes at 900°C and quenched in water at room temperature. Etchant is Kroll's solution. Mag: 750x 83 1 . 8883 zsxm Figure 23. Scanning Electron Image of Ti 6211 showing the alpha blocks in the trijunction. Etchant is Kroll's solution. Mag: 500x 84 17397 AL 16408 Counts per 30 seconds 1135 Ta 1008 Figure 24. Qualitative solute partitioning of aluminium in the sample annealed for 90 minutes at 500°C and quenched in water. 85 20937 AL 18275 Counts per 30 seconds 1219 Ta 1173 Figure 25: Solute partitioning of aluminium and tantalum in Ti 6211 annealed 20 hours at 500°C and quenched in water, showing mirror imagepartitioningtendency of Al and Ta. 86 A1 17902 Counts per 30 seconds Ta Figure 26. Solute partitioning of aluminium and tantalum in Ti 6211 annealed 20 hours at 6000C and quenched in water showing mirror image partitioning of A1 and Ta. CHAPTER 5 DISCUSSION OF EXPERIMENTAL RESULTS The objective of this research as stated in the abstract was achieved by solution-treating and annealing the samples in two coaxial furnaces circulated with argon gas, such that the sample was inside the furnace throughout the whole operation. Rapid quenching after annealing was ensured by burning a high resistance wire holding the sample and the sample dropping instantaneously into water at room temperature. . From subsequent analysis it was found that the phases that can be formed in Ti 6211 are no, a', a", and 8. no. a' and the 8(retained) were identified in the samples annealed at 900°C, 800°C and 700°C and quenched in water, with the beta decreasing in volume fraction as the temperature of annealing was lowered. For the samples annealed at 600°C and 500°C and quenched in water the phases identified were no and a". It has been shown (31) that when alpha double prime martensite formed the splitting of alpha-peaks occurred with increasing beta- stabilizing molybdenum content in Ti-Al. In our present alloy splitting was only observed at 5000 and 600°C where alpha double prime was formed. However a systematic 8 peak shift at around 39.510, 28 angle at 900°C to 39.530 28 angles at 700°C was observed in the present study. This peak was identified as beta-phase at 700°C, 800°C and 900°C and as alpha double prime at 500°C and 600°C. The shifting was accompanied by very small 87 .88 change in the beta-lattice parameter, which decreased as the annealing temperature was decreased. At 500°C the alpha double prime peak almost merged with the (1011)a line, Figure 17. It must be noted that when a sample was annealed at 102500 and then down-quenched to a lower temperature such as 600°C or 500°C the sample passed through the two phase a + 8 region in a relatively short time. For example experimental studies indicated that the time required for the sample to come to an equilibrium at 500°C was about 10 minutes. Thus an equilibrium was not attained during the passage of the sample through the a + a phase field. The sample at 500°, 600°C and 700°C started with a non-equilibrium amounts of alpha and beta phases. If 500°C were below the alpha-solvus line, then conversion of the retained beta to the equilibrium alpha at this temperature would occur by diffusional decomposi- tion of the beta-quenched-in at 500°C. The rate of diffusion at 500°C, however, is quite slow. The situation can be illustrated diagrammatically in Figure 27. Further this decomposition involved diffusion of A1 out of the beta-phase to nucleate the Al-rich equilibrium alpha-phase. Such nucleating regions must also reject the beta-stabilizing elements. The regions that have a very high supersaturation of the beta stabilizers upon quenching will be trapped as retained beta between the the equilibrium alpha-laths. However, depending on the degree of this supersaturation the quenching could transform these regions to alpha prime or alpha double prime martensites. It has been shown (93) that a transition from alpha prime to alpha double prime martensite in titanium alloy occured asa function of increasing beta-stabilizing solutes. 89 B I I 0+.8 '1 a - A - 2 [ac + untransformed B] Isoghermal annealing at 50058 ' [ac + 80500 + supersat. B] - I Quenching [ac + 0050 + B or a" l T‘ Figure 27. B-transformation at 500°C. 8 = Equilibrium a that nucleated from the B-phase during passage through 8+8 region. “0500 a Equilibrium a at 500°C. 90 From the present investigation since we observed alpha double prime martensite at 500°C and 600°C it can be concluded that such a mechanism . was operating in this system. In order to obtain a qualitative idea regarding the nature of the diffusion, we calculated the diffusion coefficients of aluminum in Ti-Al alloys in the beta phase. Using the data reported by K. Ouchi et a1 (94) Table 17 and 18were constructed for various annealing temperatures and 48 hours annealing time. Based on the data in Tables 17 and 18 it is clear that diffusion ’ rates of aluminum in the beta-phase were much slower compared with the diffusion rates of’ aluminum in the alpha-phase. The proposed model of localized supersaturation of the beta-phase which transformed to alpha double prime on subsequent quenching could be justified from this difference in the diffusion rates. Additionally the low diffusion rate of aluminum in the alpha-phase, at low temperatures, would account for some inhomogeneities in the alpha-phase eSpecially at short annealing times when the ac was not fully homogenized. These inhomogeneities, due to the low diffusion rates, gave rise to the non-regularity (as seen in Figure 24) of the wavelength dispersive x-ray solute partitioning profiles, especially at short times of annealing. At long times such non-regularity was still observed (see Figure 25) but the occurrence dependedcnithe size of the alpha-plates and beta-films. The alpha-plate was about two microns and the beta-region was less than one micron depending, however, on the annealing temperature. 91 Table 17. Calculated diffusion data for aluminium in beta-phase for Ti-4.55w/0A1. Q Temperature,K 00x10"6 kJ/mol. 773 2.90 180 873 2.90 180 973 2.90 180 1073 2.90 180 1173 2.90 180 0(m2/sec) in B-phase 2.0 x 10"18 -17 6.0 x 10"16 5.0 x 10‘15 5.0 x.10 28.0 x 10"15 Average diffusion distance i = (0t);5 in microns 48 hours annealing time “ 0.6 ” 3 ” 10 “ 29 ” 69 92 Table 18. Calculated diffusion data for aluminium in the alpha-phase for Ti-4.1w/oA1 alloy. Average diffusion distance 0 D mz/sec i = (Dt)15 in microns Temp K 00 x 10'13 K0 in a-phase 48 hours annealing time 773 8.70 70.7 1.0 x10'17 ~ 1.5 873 8.70 70.7 5.0 x30"7 ~ 3 973 8.70 70.7 1.0 x 10'16 ~ 5 1073 8.70 70.7 3.0 x10"16 ~ 7 16 1173 8.70 70.7 6.0 x 10- ~10 93 The solute partitioning tendencies of tantalum a beta-stabilizer, and aluminum an alpha stabilizer further confirmed the model of localized supersaturation in the beta-phase and the occurrence of inhomogeneities in the alpha-phase. The overall observation, however,was that tantalum partitioned to the inter alpha regions which were the beta-regions at elevated temperatures. Aluminum on the other hand partitioned to the alpha plates both on the grain boundary when the grain boundary alpha formed and in the grain itself. The partitioning tendencies of tantalum and aluminum were mirror images to each other, see Figures 25,26. This trend was clearly observable at low temperatures and long annealing times. The occurrence of supersaturated regions was further confirmed by occasional high peaks and deep troughs in the solute partitioning profiles. Figures 25 and 26 show the partitioning of tantalum and aluminum at 500°C and 600°C after 20 hours annealing and quenching. The aluminum rich regions corresponded to the alpha-plates and the aluminum poor regions which were tantalum rich regions corresponded to the inter-alpha regions which were either beta-phase or alpha double prime depending on the annealing temperature. This stems from the obvious fact that tantalum and other beta stabilizers dissolve preferrentially in the beta-phase and aluminum and other alpha-stabilizers dissolve preferrentially in the alpha-phase. Considering the binary phase diagrams Figure 4, 5, 6, and 7 after Hansen et al (95,96) it is observed that the beta stabilizers, namely Nb, Ta and Mo lower the a/ a + 8 solvus line. The overall effect of these elements to Ti-Al system is speculative. Recently Shull et al (97) showed the phase diagram for Ti-Mo-Al alloy system, Figure 28, in which the effect of 0.5 a/o Mo on Ti-Al 94 ' i l 11004 I B 1 i A .. O 3, 1 1 . CD 900 1° h. 2 1 g l ' . a: a l° ‘3' l S l/ 1'- 700‘ 11. (22 "‘ l l - o - 500 1/1162111 L 1 10 20 30 Atomic 96 Al T199.5M°o.5 Figure 28. Ti-Al phase diagram showing effect of 0.50 a/o Mo on Ti-Al binary system. Thin background lines indicate the diagram for Ti-Al system, after Shull et al (97). 95 binary system is the slight lowering of both the 8/8 + 8 and a + 8 /8 phase boundaries. It is likely that the d/a + 8 and a + 8/8 phase boundaries are further lowered with accompanying widening of the a + 8 phase field. There is also the possibility that at lower annealing temperatures, Ti 6211 may have 62 an order 0019, T13Al phase (97). This possibility was examined based on the x-ray diffraction results. The position of fundamental diffraction peaks of az-phase was almost identical to those of the alpha-lines (98). In addition supper- lattice peaks from the ordered DD19 structure should appear.. But in the present investigation all x-ray diffraction lines were accounted for as a, 8 or a" and there were no extra lines which could be identified as the supperlattice reflections. Thus either the volume percent of 02 phase in this alloy was below the limit of detection by x-ray diffraction or the occurrence of 02 could be ruled out for the annealing times and/0r temperatures of the present study. It has been pointed out by Margolin et a1 (99) that with an increasing amount of beta-stabilizer in the beta-phase the independent nucleation of separate plates was enhanced and the colony microstructure was replaced by the basketweave or the widmanstatten microstructure. Typical alpha colony microstructure at 600°C and the basketweave at 500°C for Ti 6211 are shown in Figure 21. These microstructures further confirm that the beta stabilizers in the equilibrium beta increased as annealing temperature was lowered. Further morphological information was obtained from the scanning electron microscopy. Figure 22 shows the electron image of a sample aged 96’ for 30 minutes at 900°C and quenched in water. The martensitic plates occurring as alpha prime plates, the retained beta and the alpha colony are observable. The heterogeneous nucleation of alpha on the prior beta-grain boundary is shown in the scanning electron micrograph, Figure 23. This alphaeplate nucleated and grew along the grain boundary to form a continuous grain boundary alpha. 1 From the electron microprobe analysis it appeared that the aluminum content of the grain boundary alpha was slightly smaller than the Aluminium content of alpha-colony. The pattern of the various alpha colonies was noteworthy. The alpha colony terminating at the 'grain boundary alpha showing that the two alpha microstructures must have nucleated_differently. At high annealing temperatures, such as 900°C or 800°C, a hetero- geneous nucleation of equilibrium alpha phase occured at the prior beta grain boundaries. As these grain boundary alpha grew there was a depletion of A1 (and thus a supersaturation of beta-stabilizers) adjacent to the grain boundaries. The grain interior, however, did not suffer such an Al depletion. Thus simultaneously or concurrently equilibrium alpha-colonies formed at the grain interior. The remaining matrix, after the isothermal heat-treatment, consistedcfl’beta-phase. Among the retained beta regions, the volume adjacent to a grain boundary was the most supersaturated beta because of the higher diffusion rate of solutes near the boundary. Upon quenching, the beta-phase transformed martensitically giving rise to a "packet" type martensitic structure. The retained beta near the grain boundaries,however, was supersaturated with beta-stabilizers to such a degree that upon cooling or quenching 97 to room-temperature, it was retained. The x-ray data confirmed the existence ' of the beta-phase in alloys quenched from 800°C and 900°C. The scanning electron and optical micrographs, Figure 22 and Figure 23, show the nature of equilibrium alpha colonies, grain interior, packet martensites, and the structureless zone (enriched beta-phase) between the grain boundary alpha and the martensitic packets. Schematically this situation is explained in Figure 29. Further it is appropriate to make few remarks about the phenomeno- logical theories as well as the strain energy minimization criterion mentioned in this research. Both tacitly assume the uniform distribu- tion of transformation strain and lattice invariant shear. It is true that using these phenomenological analysis one can obtain the situation where the elastic strain energy is zero. If this theory is accepted the supercooling phenomenon which characterizes the martensitic transformation should not be observed because there would be no factors opposing the transformation. Thus in order to understand the energetics of martensitic trans- formation it is essential to take into account the actual periodic distribution of transformation strains and lattice invariant shear as has been done by Mura et a1 (4) and Kato et a1 (89). Graln BOundany a Reta1NEd B ".=i'I-. (ll/8% /)iéz’9/ 1 (ll/ll) j 1:," /{/ ‘ a -j[qiaggggaiiiilligaat’/ :;;\ o 7 ;, I o : \ , fi 2‘ \\\\’ / Figure 29. Schematic illustration of the microstructures for samples annealed for short times at 800 C and 900°C and quenched in water at room temperature. Plate martensue . CHAPTER 6 SUMMARY AND CONCLUSION In this research fundamental principles were used to characterize the phase transformations in this complex quartenary Ti alloy and the following conclusions are made. 1. The x-ray diffraction experiments showed that do, a', a", and 8 phases could be obtained in Ti 6211 after isothermal and quenching heat treatments. Specifically at higher temperatures 700°C, 800°C, and 900°C do, a' and retained 8 were stable, at lower temperatures 00. 6“ phases were stable. The existence of 62 (113Al) in this system in the prescribed heat treatments was not confirmed because of the absence of the extra superlattice peaks from 62, 0019 structure. This absence might result from very small volume fraction of dz in this alloy. The splitting of a-peaks to satelite orthorhombic peaks is not observed in this system, at high temperatures where a" was not obtained on quenching. At 5000 and 600°C where a" was obtained, splitting occurred. 2. The lattice parameters of the prevalent phases in Ti 6211 are found as follows: 8-phase: a8 = 3.2253 99 100 a-phase: a = 2.9383 0 o c = 4.670A a Orthorhombic: o a n = 3.040A a - o b u - 4.944A “ o c u = 4.637A a Slight variation of these parameters at various temperatures was observed. 3. The Electron Microprobe Analysis showed qualitatively that aluminum has strong partitioning in the alpha regions and the grain boundary alpha (Gba). Tantalum exhibited weak partitioning to beta, alpha prime and alpha double prime regions. This was expected as alpha prime and alpha double prime came from the high temperature beta phase on quenching. Aluminum and tantalum partitioning occurred in mirror . image to each other especially at low temperatures. 4. Based on the phases identified in quenched alloys and the solute diffusion model proposed in this study, tentative a and a + 8 phase boundaries are schematically shown in Figure 30. Our proposed phase boundaries are in qualitative agreement with those proposed by Shull et al (97) for a ternary Ti-Mo-AL alloy. We believe, however, that due to the additional B-stabilizers, namely Ta and Nb, 900°C and 800°C are also within the cx+8 phase field. Further our results as well as Shull's et al results, indicate that the phase equilibria proposed by Williams (2) wherein 500°C to 900°C are all in two phase Ql+f3 region is incorrect. 1100 1025 1000 700 Tent) 600 500 Figure 30. Schematic o and 8+8 phase boundaries in Ti 6211 31on 101 I / I L , I ______;P___ ..