UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Properties of ultra fine grain [beta]-CuAlNi strain memory alloys Mukunthan, Kannappar 1987

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1987_A7 M94.pdf [ 7.03MB ]
Metadata
JSON: 831-1.0078496.json
JSON-LD: 831-1.0078496-ld.json
RDF/XML (Pretty): 831-1.0078496-rdf.xml
RDF/JSON: 831-1.0078496-rdf.json
Turtle: 831-1.0078496-turtle.txt
N-Triples: 831-1.0078496-rdf-ntriples.txt
Original Record: 831-1.0078496-source.json
Full Text
831-1.0078496-fulltext.txt
Citation
831-1.0078496.ris

Full Text

PROPERTIES OF ULTRA FINE GRAIN 0-CuAlNi STRAIN MEMORY ALLOYS by KANNAPPAR MUKUNTHAN B.Sc.(Eng.), University of Moratuwa, Sri Lanka, 1983 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF T H E REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES Department of Metals and Materials Engineering We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA August 1987 ® KANNAPPAR MUKUNTHAN, 1987 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of MefcfrU «**»4 Mgtfcevv\U £njio*e^i^ The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date AwjUsC 2-£> A B S T R A C T A method has been developed to produce grain sizes as low as 5M m in 0-CuAlNi alloys and the effect of grain size on mechanical and strain-memory properties was studied. The thermomechanical treatment procedure involved two. sequential warm working and recrystallization steps at 600° C and 800° C respectively on eutectoid alloys. Three different eutectoid alloys, two with M s temperature of around 0 ° C and one with M s = 220° C were used for the present studies. Even at fine grain sizes, the specimens produced were of single 0- phase type without any second phases. Two-stage characteristic stress-strain curves were obtained for most of the specimens in both the strain memory and pseudoelasu'c states. It was found that the ultimate tensile strength and strain to failure increased with decreasing grain size according to a Hall-Petch relationship down to a grain size of 5M m with the exception of one alloy. Fracture "strengths of 1,200 MPa and fracture strains of 10% could be obtained. It was found that the major recovery mode, whether pseudoelastic or strain memory, did not have any significant effect on the total recovery obtained. Recovery properties were not affected significandy by decreasing grain size. Approximately 86% recovery could be obtained for an initial applied strain of 5% at a grain size of around 10M m. Grain refinement increased the fatigue life considerably, possibly due to high ultimate fracture strength and ductile fracture mode. Fatigue life of 275,000 cycles ii could be obtained for an applied stress of 330 MPa and a steady state strain of 0.6%. Most of the fractures are due to intergranular-type brittle fracture. At fine grain sizes, transgranular-type brittle fracture and microvoid coalescence-type ductile fracture dominated the fracture mode. Oxygen segregation at grain boundaries is the possible explanation for the different mechanical properties shown by different alloys in the present work by being a major factor in causing intergranular-type fracture. iii T A B L E O F C O N T E N T S ABSTRACT ii Table of Contents iv List of Tables v List of Figures '. vi Acknowledgement viii 1. INTRODUCTION 1 1.1 GENERAL REVIEW 1 1.2 AIM OF THE PRESENT WORK 5 2. PREPARATION OF ULTRA FINE GRAIN MATERIALS 7 2.1 PREVIOUS WORK ON GRAIN REFINEMENT 7 2.2 ESTABLISHING THE ALLOY COMPOSITION 10 2.2.1 Position of the Eutectoid 10 2.2.2 M s Temperature as a Function of Composition 12 2.2.3 Calculations for the Optimum Alloy Composition 15 2.3 EXPERIMENTS ON GRAIN REFINEMENT 16 2.3.1 Alloy Preparation , 16 2.3.2 Initial Experiments on Grain Refinement 18 2.3.3 The Thermomechanical Treatment for Eutectoid Alloys 21 3. PROPERTIES OF ULTRA FINE GRAIN MATERIALS 27 3.1 GRAIN GROWTH 27 3.2 M s DETERMINATION 31 3.3 TENSILE TESTS 33 3.3.1 Effect of Grain Size on Tensile Properties 34 3.3.2 Effect of Temperature on Tensile Properties 43 3.3.3 Fractography 50 3.4 RECOVERY PROPERTIES 56 3.4.1 Effect of Temperature 56 3.4.2 Effect of Grain Size 59 3.4.3 Effect of Increasing Strain 61 3.5 FATIGUE PROPERTIES 64 3.5.1 Effect of Grain Size on Fatigue Life 64 3.5.2 Fractography '. 70 3.6 EXPLANATION FOR DIFFERENCES IN MECHANICAL PROPERTIES BETWEEN ALLOYS 74 4. CONCLUSIONS 81 REFERENCES 83 iv LIST OF TABLES Chemical Composition and M s of the alloys prepared LIST OF FIGURES 1. Sections of Cu-Al-Ni equilibrium phase diagram. 11 2. Position of eutectoid in Cu-Al-Ni 11 3. Relation between the M s temperature of various Cu-Al-Ni alloys and the Al content in atomic percent 14 4. Schematic representation of the thermomechanical treatment 24 5. Optical micrographs of a) the structure of alloy C after warm working at 600° C showing 7,-precipitates, x800, b) and c) the structures of alloys A and B respectively after recrystallization at 800° C showing grains of 5Mm, xl.200 25 6. Variation of grain size with increasing solution treatment time at 800° C for alloys A, B and C 29 7. Log-Log plot of grain size-vs.-effective true grain growth time at 800° C for alloys A, B and C 30 8. Variation of M s temperature with grain size for alloy B 32 9. Stress-strain curves at varying grain sizes for alloy C, tested at 22° C 36 10. Stress-strain curves at a constant grain size of 50- 60M m for alloys A, B and C, tested at 22° C 37 11. Variation of fracture strength with a) grain size and b) (grain size) --^ for alloys A, B and C, tested at 22° C 39 12. Variation of strain to failure with a) grain size and b) (grain size)""^ for alloys A, B and C, tested at 22° C 42 13. Variation of fracture strength with (grain size)"^ at -195° C for alloys A and B 44 14. Variation of strain to failure with (grain size)"^ at -195°C for alloys A and B .' 45 15. Stress-strain curves for alloy B at varying temperatures .'. 46 16. Variation of transition stress and fracture strength with temperature for alloy B 47 17. SEM fractographs showing tensile fracture surfaces of alloy A at varying grain sizes, fractured at 22° C 52 18. SEM fractograph showing the tensile fracture surface of specimen A (grain size = 10Mm), fractured at -195° C, x500 53 vi 19. SEM fractographs showing tensile fracture surfaces of alloy C at varying grain sizes, fractured at 22° C 54 20. Stress-strain curves of alloy A specimens at varying temperatures showing recovered strain on unloading and heating 57 21. Variation of total recovery and the recovery by pseudoelasticity and strain memory effect with temperature for alloy A 58 22. Variation of percent recovery with grain size at constant applied strain of 2% for alloys A and B 60 23. Variation of percent total recovery and percent pseudoelastic recovery with increasing initial strain for alloy B 62 24. Variation of strain per cycle with cycling in specimens of alloys A, B and C 65 25. Variation of percentage pseudoelastic recovery with number of cycles for the first few cycles for alloy C 66 26. Variation of number of cycles to failure (N) with a) grain size and b) (grain size)"1 / 2 for alloys A, B and C 68 27. Two S-N type plots where the number of cycles to failure is plotted against the maximum applied stress for alloy C 69 28. SEM fractographs showing fatigue fracture surfaces of alloy A at varying grain sizes '. 71 29. SEM fractographs showing fatigue fracture surfaces of alloy C at varying grain sizes 72 30. a) Micrograph and b) O" map of the same region in a specimen of alloy A as obtained on the SIMS 77 31. a) O" and b) Al* trace for a typical SIMS line scan in alloy A 78 vii ACKNOWLEDGEMENT The author wishes to express his sincere gratitude to Dr. L.C. Brown for his advice and encouragement throughout the duration of this investigation. He would also like to thank members of the faculty and fellow graduate students for helpful discussion and other help. The assistance of the technical staff is greatly appreciated. Thanks are also extended to his friends who helped him in so many ways during this work. Financial assistance provided by NSERC (grant number A-2459) and the graduate assistantship awarded by the Department of Metals and Materials Engineering are greatfully acknowledged. viii 1. INTRODUCTION 1.1 GENERAL REVIEW Martensitic transformations are observed in a large number of ferrous and non-ferrous systems. A martensitic transformation is one in which there is no change in composition and the product phase is produced by the coordinated movement of atoms of the parent phase. The transformation results in a shape deformation which gives rise to a tilt on a prepolished surface." The interface between the martensite phase and the parent phase is essentially undistorted and unrotated, and the Miller indices of the habit plane are characteristic of that alloy. The principal directions and planes in both lattices are related by an orientation relationship .^ The transformation may be considered analogous to twinning. Martensitic transformations can be induced by the application of stress as well as by changes in temperature. This interchangeability of 2 temperature and stress as variables can be explained well using the proper thermodynamic and kinetic parameters. There has been a substantial increase in interest in the field of martensitic transformations in recent years. This is possibly due to newly found industrial applications of martensitic transformations in maraging. TRIP and dual-phase steels, and in applications involving the strain-memory effect and in alloys with high damping charactristics. The strain-memory effect (SME) and related phenomena such as pseudoelasticity and the two-way strain-memory effect are closely related to the 1 2 martensitic transformation. Their origin is in the lattice deformation of the martensitic transformation. Accordingly, any martensitic alloy should potentially be able to exhibit these effects. However, the magnitude of the SME differs to a large extent from one 3 alloy to another , and only when the transformation is of the thermoelastic martensitic type is the SME significant A thermoelastic martensitic transformation occurs when martensite forms and grows continuously as the temperature is lowered and shrinks and vanishes continuously as the temperature is raised. Pseudoelastic behaviour is a complete mechanical analogue to the thermoelastic transformation. It occurs at temperatures above M s where the transformation proceeds continuously with increasing applied stress and is reversed continuously when the stress is decreased. The strain-memory effect occurs at temperatures below M s and is realised if a macroscopic deformation is accompanied by a change in the martensite structure which is not reversed by removing the applied stress; in a second step the reverse transformation and a concomitant reversal of the macroscopic deformation are induced by heating above M s . Pseudoelasticity and the strain-memory effect may be associated with a martensitic transformation or reorientation of an existing 4 martensite structure or a combination of both . The two-way strain-memory effect occurs if there is a macroscopic deformation accompanying the thermally induced martensitic transformation and is brought into effect by lowering and raising the temperature through M s . 3 There are two major groups of thermoelastic-type alloys . Both groups are of 0 - phase type at high temperatures and are 3/2 electron compounds. The first group includes those alloys which have a B2 or D0 3 ordered parent phase and 3 transform into a martensite with a periodic stacking structure such as 2H, 3R, 9R and 18R. Ni-Ti, Ni-Ti-X (X = Cu, Fe), Cu-Zn, Cu-Zn-X (X = Al, Si, Sn, Ga), Cu-Al-Ni, Cu-Al-Mn, Cu-Sn, Ni-Al, Au-Cd, Ag-Cd and Au-Cu-Zn are the main examples of this group. The second group is of alloys which have a face-centered tetragonal martensite such as In-Tl, In-Cd, In-Pb, In-Sn, Cu-Mn, Ni-Mn and Fe-Pd. The memory strains and recovery forces are larger with alloys of the first group than those of the second group. Among the alloys of the first group, Ni-Ti, Ni-Ti-X, CuAINi and 5-711 CuZnX are now being used practically for many technological and medical purposes on the basis of their superior recoverability, large range of transformation temperatures and other mechanical properties such as workability and toughness. Some unique applications where strain-memory properties are useful are leakproof couplings for pneumatic and hydraulic lines, thermomechanical and thermostatic control devices, orthodontic dental arch wires, medical implants and heat engines. NiTi alloys are commonly used because of their excellent strain-memory properties with recoverable strains of 8-10% together with good ductility g and fatigue strength . The good mechanical properties are mainly due to their small grain size and low elastic anisotropy. However, they are expensive because of high material cost and the intricacies involved in alloy preparation and fabrication. Hence economic factors 9 limit their use in many applications . The copper-based alloys exhibit excellent strain-memory recoveries of up to 5-7% strain. However, their use for practical purposes is limited due to poor mechanical properties such as limited ductility and low fatigue strength. The inherent brittleness of these alloys is generally considered to be due to 4 high elastic anisotropy and large grain size when in the parent phase condition^. The low material cost and relative ease of alloy preparation and fabrication make these alloys potentially economically attractive and would extend the range of practical applications of 9 these alloys . The relative ease with which the martensitic transformation temperature can be varied by modifying the composition^ and the possibility of using these for high temperature applications^ are additional useful features. A good deal of research work is being carried out with the objective of developing copper-based strain-memory alloys with more useful properties. Improving the mechanical properties and developing ways for these alloys to be used continuously at high temperatures (above 200° C) are two important steps in this direction^. Among the copper-based strain-memory alloys, CuAINi, CuZnAl and CuZnSn are the most useful ones for practical purposes. It has been suggested based on recent 12 investigations that these alloys can be made more useful by giving them a preferred texture or by obtaining finer grain size. There have been only a few investigations which quantify the effect 13.14 of grain size on mechanical properties in strain-memory alloys. Khan and Delaey showed the existence of a Hall-Petch type relationship between grain size and both yield strength and ultimate tensile strength. Similar relationships were shown to exist in CuAINi alloys between grain size and both ultimate tensile strength and strain to failure down to a grain size of 15Mm by Sure and Brown^. They obtained fracture strength values as high as 900 MPa, fracture strain values as high as 7% and recovery values as high as 80% at a grain size of 15M m and concluded that significant improvement in mechanical properties could be obtained by grain refinement without affecting the strain-memory 5 properties to any large extent They also obtained high fatigue strength in their alloys and attributed this to the ductile fracture mode obtained at fine grain sizes. Reasonably similar mechanical properties were also obtained by Duering et al^ in their CuAINi PM alloys with a grain size of 20M m. The crystal structures of martensites forming in Cu-base alloys are 12 well known. In CuAINi alloys , the martensites have long-period stacking order structures with a common basal plane {110}^ . The unit cells of these martensites have been found to be monoclinic. In CuAINi, the matrix to martensite transformation is of the type j3 -> /3', with j3' being 18R type martensite. At high stress levels, martensite to martensite transformations of the type j3 '•* 1' (2H type martensite) also take place. 1.2 AIM OF THE PRESENT WORK The objectives of this work are to produce j3-phase strain-memory alloys with ultra-fine grains and to test them with the aim of studying the effect of grain size on mechanical and strain-memory properties. It has been shown that mechanical properties can be improved significantly by grain refinement without significant deterioration of the strain-memory properties^. There are few studies available for strain-memory alloys which deal with the effect of grain size on mechanical 51L13-18 1518.19 properties' and strain-memory properties ' . Those studies mostly cover the coarser range of the grain size, and the smallest grains that have been studied were of 15/im diameter in the CuAINi system^. The present investigation is directed at reducing the grain size further and seeing the effect of this on the mechanical properties. 6 The CuAINi system was selected because of the possibility of this alloy being used commercially. In addition, this alloy has been used for numerous studies in the past and most of the required information regarding the phases present at 20-23 different temperatures, transformation kinetics , variation in M s temperature values with 15,24-32 changing composition and details of the martensitic transformations and strain 1215 33 34 memory properties ' ' are all readily available. CuAINi has good potential to be used at high temperatures because of its capability to retain the strain-memory effect at temperatures as high as 300° C^ 1 . It is the most brittle of the Cu-based beta alloys; hence if the mechanical properties of this alloy can be improved by grain refinement, the other less brittle beta alloys might show even better mechanical properties on grain refinement 2. PREPARATION OF ULTRA FINE GRAIN MATERIALS 2.1 PREVIOUS WORK ON GRAIN REFINEMENT The following methods have been used extensively in the recent past to obtain effective grain refinement in Cu-based /3-alloys, 17 1) Solution treatment 18.15 2) Two phase structures 3) Alloy additions 3 5 , 3 6 3* 1 1 5 3711 4) Powder Metallurgy techniques ' 38.27 5) Rapid solidification methods 17 Brown used a solution treatment method to obtain various grain sizes in a CuZnSn alloy. The material was cold rolled 15%, recrystallized by heating for various periods in a salt pot at 820° C and subsequently water quenched to give grains in the range of 200 to 1000Mm in diameter. 18 White et al showed that rapid grain growth in CuZnAl alloys could be controlled by the presence of alpha phase as fine widmanstatten plates. They also showed that a small volume fraction of retained alpha at fine grain sizes did not affect the strain memory properties. Sure and Brown15 obtained grain sizes as small as 15ju m in CuAINi alloys by controlled solution treatment although 7 2 precipitates were present in the matrix due to incomplete precipitate dissolution. Detailed investigations on grain refinement by alloy additions to 7 8 CuAINi alloys were carried out by Matsumoto et al and Kamei et al . It was found that Ti, Co, V were most effective for forming equiaxed grains and suppressing grain 30 coarsening of the 0- phase. Sugimoto et al investigated the effect of 0.5 to 3.99 wt% Ti additions to a CuAINi alloy, and demonstrated effective grain refinement and improved hot working ability. Sure and Brown15 used a 0.5% Ti addition to CuAINi alloys and showed that the Ti addition produced a finer cast structure which contributed to better formability. They also showed that addition of Ti gave rise to Ti-rich X-phase spherical particles uniformly distributed in the 0- matrix which were effective in grain growth retardation. They further concluded that the presence of a small fraction of second phase particles (X or 72) had no significant effect on mechanical and strain-memory properties. 37 In CuZnAl alloys , it was found that grain refinement could be obtained using powder metallurgy techniques and the observed superior fatigue properties were attributed to the fine grain structure. Duering et a l 1 1 produced fine grain CuAINi alloys by powder metallurgy techniques and demonstrated good ductility of up to 7% and high fatigue strength. 38 Rapid solidification techniques employed by Oshima et al and 27 Wood for producing fine grain sizes in Cu-based 0- phase alloys have proved very promising and ribbons produced exhibit excellent strain-memory properties with recoverable strains up to 6-7%. The technique can only be applied to thin section sizes. In Cu-base j3-alloys grain sizes in the range of 15-20M m have 11 27 been obtained by powder metallurgy and rapid solidification techniques and also by alloy additions15. Powder metallurgy and rapid solidification techniques involve intricate 9 processing steps where a number of variables have to be controlled carefully to obtain the desired results. In the alloy addition method, a minor modification of conventional alloy preparation techniques is necessary and often a second phase is present in addition to the parent /3- phase. The solution treatment method is the simplest way of obtaining grain refinement and hence was considered as the most suitable method in the present investigation. However, a new thermomechanical treatment procedure had to be developed to obtain ultra fine grains in the order of 5M m in diameter. 10 2.2 ESTABLISHING THE ALLOY COMPOSITION In the present work, the method for producing fine grain alloys involves rolling and subsequent recrystallization steps. The initial experiments performed (described briefly in section 2.3) clearly suggested that the eutectoid composition was the most suitable to work with in obtaining ultra fine grain structure. In addition, it was decided to use an alloy with a martensitic transformation start (Ms) temperature of around 0°C. Alloys with such M s temperatures are particularly useful in practical applications. In addition, mechanical testing of these specimens will be relatively easy since tests can be done at the vicinity of room temperature. Hence it is necessary to find an alloy with the eutectoid composition which also has an M s temperature of approximately 0°C. Such a composition was calculated using existing literature data and some initial experimental results. 2.2.1 Position of the Eutectoid Fig. 1(a) shows the equilibrium phase diagram of the Cu-Al 20 system . The |3- phase has a disordered b.cc. structure at higher temperatures and decomposes eutectoidally into the copper-rich a-solid solution and the hard o 21 12- intermediate phase at 565° C . By quenching from the homogeneous b.cc. region, an ordered b.cc. structure may be produced, and this may transform in turn to martensite. The disorder-order transformation of the |3-phase takes place at about 500°C in this 22 eutectoid binary alloy . Fig. 1(b), (c) and (d) show the equilibrium phase diagrams of the IL ft • uo u 0 / (3 * no 10 20 ALUMINIUM. PER CENT ALUMINIUM. PER CENT. a) 0 wt% Nt b) 3 wr% Ni ALUMINIUM. PER CENT. c) 6 wt% Ni ALUMINIUM, PER CENT d) 10 wt?0 Ni Figure 1. Sections of C u - A l - N i equilibrium phase diagram (after Alexander^). 2 4 6 Ni ( a t o m i c p e r c e n t a g e ) Figure 2. Position of eutectoid in C u - A l - N i (literature data). 12 20 Cu-AI system at three different constant Ni contents . These diagrams clearly suggest that Ni has little effect on the j 3 - phase region when a small percentage of Cu is replaced by Ni. It was also found that Ni in amounts up to 5% had little effect on the -r-r-r. 39 position of the TTT curves . From Fig. 1, it can be concluded that the eutectoid temperature is almost constant at 575+ 15° C for all Ni contents up to 10% in the CuAINi system. It can also be seen that as Ni increases from 0% to 10%, the corresponding Al which is necessary to keep the system as a eutectoid increases from 11.8% to 15.3%. Fig. 2 shows the linear relationship that exists between Ni content and Al content at the eutectoid composition. Using this plot, it is possible to calculate the composition of the eutectoidal CuAINi alloy for any given Ni content which is not more than 10% by weight 2.2.2 M s Temperature as a Function of Composition It is a well known fact that the temperature range over which martensitic transformation occurs depends on the composition of the parent phase. In non-ferrous martensites, the fundamental factor which decides the M s temperature of any alloy composition is the distortion in the basal plane of the martensite crystal structure. The extent of this distortion depends on the type and the degree of long range order, 33 the composition of the alloy and the relative sizes of the constituent atoms . Hence when the M s temperature, of a single system of alloys of different composition are compared, such as CuAINi, the relative atomic sizes and the relative chemical composition are the two important factors affecting the extent of distortion. It is thus possible to 13 have two different alloy compositions which have identical M s temperatures if the 40 distortion produced in either composition is the same. Warlimont and Delaey outlined a linear relationship between the M s temperature of various Cu-Zn-Al alloys and a p-parameter which is a measure of the distortion in the basal plane. They further concluded that the M s temperature is lowered with increasing amounts of distortion. No definite relationship has been reported in the literature relating M s temperature and alloy composition for CuAINi alloys with varying Ni and Al percentages. Such a relationship would be very useful in producing alloys with the desired M s temperature and is particularly important in the present work since both Al and Ni percentages will be varied to make sure that the alloy is of the eutectoid type. In CuAINi, the atomic radius of each atom is: Cu - 1.28A Ni - 1.24A Al - 1.43A Ni and Cu have very similar atomic radius and Ni is completely soluble in Cu. The atomic radius of Al is very different when compared to that of Ni and Cu and hence Al content may be the only factor which decides the extent of distortion. In another approach, the effectiveness of Al to Ni in deciding the basal plane distortion can be considered as 15:4, i.e. their differences in size relative to Cu. Al and Ni seem to have the opposing effect on the basal plane distortion due to the fact that the atomic radius of Cu is higher than that of Ni, but lower than that of Al. All these factors were considered when attempts were made to correlate M s with composition. 14 400-i 300 200 100-0 -100 -200 A O \ O O X O N i (wt %) = 0 • N i ( * t % ) = 1 . 5 - 2 . 5 A Ni (wt %) = 2 . 5 - 3 . 5 O Ni(w t %) = 3 . 5 - 4 . 5 + N i (wt %)=7.7 X N i (wt %) = 3 . 0 - 6 . 5 (present work) A 23 24 25 26 27 28 A l (atomic percentage) 29 Figure 3. Relation between the M s temperature of various Cu-Al-Ni alloys and the Al content in atomic percent. 15 Attempts were made to obtain an empirical relationship by plotting different parameters which are a measure of the basal plane distortion against the M s temperature for various CuAINi alloys using existing literature data and some initial experimental results. In this approach, M s was plotted against parameters such as e/a ratio of alloy, atomic percentage of Al, average atomic radius, [15 x Al(at%) ± 4 x Ni(at%)]. The best correlation was obtained when M s was plotted against atomic percentage of Al (Fig. 3). Hence this reinforces the argument that the effect of Ni on basal plane distortion and hence on M s is negligible when compared to that of Al. The linear relationship obtained here will be very useful in calculating the alloy composition which corresponds to a given M s temperature in the CuAINi system. 2.2.3 Calculations for the Optimum Alloy Composition It is now possible to calculate the optimum values for %Ni and %A1 by using Figs. 2 and 3. These values should satisfy the following conditions, (1) the M s of the alloy is to be 0°C, (2) it is to be of eutectoid composition. Fig. 3 shows that the composition should be 27.6 at% Al to satisfy condition (1). Fig. 2 gives the eutectoid composition of a 27.6 at% Al as 5.5 at% Ni. This will satisfy condition (2). Thus the required alloy will have a composition of 66.9 at% Cu, 27.6 at% Al and 5.5 at% Ni or 79.6 wt% Cu, 14.2 wt% Al and 6.2 wt% Ni. 2.3 EXPERIMENTS ON GRAIN REFINEMENT 16 2.3.1 Alloy Preparation The CuAINi alloys were prepared from 99.9% pure copper, 99.9% pure aluminum and 99.99% pure nickel. The preparation method involved melting weighed quantities of each element in a graphite crucible in a high frequency induction furnace under an argon atmosphere. Considerable time and effort had to be spent in obtaining alloys with the suitable composition because of the difficulty of controlling the Al content at the desired level. Even a change of 1% in Al content in the final alloy shifted the position of the alloy in the . phase diagram from a point which corresponds to eutectoid where the phase boundary temperature is around 575° C to a position where the j3- phase decomposes into two phases at a temperature which is higher than 700° C (Fig. 1). Further an addition of 1% Al in the final composition suppresses the M s temperature by at least 150° C (Fig. 3). Hence Al losses had to be controlled very carefully. A low heating rate was used in the melting process and the superheat of the melt was maintained at a small value before pouring since these could possibly help reducing the Al losses. Initial experiments showed that a 2-3% Al loss occurred in the present casting procedure. When this loss was compensated for, castings with correct composition were obtained relatively easily. The melt was cast into a rectangular copper mold. The castings were then homogenised at 900° C for 24 hours. Stainless steel pouches were used in 17 homogenising to prevent any possible oxidation. Approximately 1kg of alloy was prepared in each batch. Total weight loss during casting was found to be less than 0.3% of the input weight This suggested that the alloy composition was close to that of the charge composition, which was later confirmed by chemical analysis. Chemical analysis of the alloys prepared were carried out by Can Test Ltd. using the plasma spectroscopic technique. The results are shown in Table 1. Table 1. Chemical Composition and M s of the alloys prepared Alloy Cu Al Ni M s Alloy (wt%) (wt%) (wt%) ( ° Q Type X 81.1 13.6 5.32 200 Non-Eutectoid Y 79.8 14.9 5.28 -100 Non-Eutectoid Z 80.7 14.0 5.34 105 Non-Eutectoid A 79.2 14.4 6.32 -15 Eutectoid B 79.7 14.1 6.28 40 Eutectoid C 84.2 12.8 3.04 220 Eutectoid 18 2.3.2 Initial Experiments on Grain Refinement 1) Hot Rolling Throughout the present work, pieces were cut from the cast ingot and were hot rolled at 800° C in the single beta phase region. This was carried out repeatedly as the first processing step until a certain thickness was attained. In the initial stages of rolling when the cast structure was being broken up, the extent of deformation that could be obtained in a single pass without cracking was in the range of 5-10% reduction in thickness. After the cast structure had been broken up and replaced by a recrystallized equiaxed fine grain structure, there was very little tendency to crack along grain boundaries during hot rolling, and up to 25% reduction in thickness could be given at each stage of rolling. In the eutectoid alloys, 40-50% reduction in thickness could be given in a single pass without cracking during the last few passes. It was observed that in a single hot rolling pass, the temperature of the specimens dropped to the two-phase region in alloys X, Y and Z and second phase precipitates formed. Dynamic recrystallization was not observed in any alloys during hot rolling and the material was in a deformed state after a pass. Hence the pieces had to be heated back at 800° C for 120-300s in between every pass so that the material would have a recrystallized 0- phase structure and another pass could be given without any cracking. 19 These hot rolling passes were given until a certain thickness was attained. This thickness varied in the range of 1.2 to 2.5mm according to the final processing step which followed the initial hot rolling passes. At the end of the final processing, the thickness of the specimen was always maintained at 1-1.1mm. 2) Experiments on Hyper-Eutectoid Alloys The alloys X, Y and Z were used for the initial experiments on recrystallization. The temperature at which these alloys undergo 0 -* (|3+72) transformation is in the range of 700 to 750° C, i.e. these are not eutectoid type alloys. Initial experiments were directed at obtaining a fine grain structure by recrystallizing a cold worked material. When two phase (a+72) specimens were cold worked, only 4-6% deformation could be given and no recrystallization was observed, even after 2 hours at 800° C. Such attempts were made following a suggestion of White et 18 al that the presence of a second phase affects the deformation substructure introduced in rolling and also retards rapid grain growth. Different heat treatments were then performed with the objective of obtaining a two-phase material where a small amount of fine precipitate was distributed throughout the structure so that the amount of cold work could be increased. In all cases however, preferential precipitation took place along grain boundaries and no significant increase in cold work could be given and hence no recrystallization was observed. With p'-phase specimens, 10-15% cold work could be given and they could be recrystallized at 800° C. However, in this case the recrystallization was a slow process and in the completely recrystallized state, grains of the order of 25M m could be obtained. No recrystallization was observed in either case when heating was carried out in the two-phase region. Thus static recrystallization was not successful in 2G obtaining a fine grain structure. The possibility of obtaining a dynamically recrystallized fine grain structure was then studied by varying the amount of deformation in a single pass, the 41 rate of deformation and the temperature of rolling as suggested by McQueen and Jonas and it was found out that this alloy does not recrystallize dynamically, i.e. energy can be stored in the structure without simultaneous recrystallization even at a temperature as high as 800°C. In the two-phase region, the amount of high temperature deformation that could be given was limited. In the 0- phase region, specimens could be hot worked 25-30% easily and when this was performed at 850° C and if the deformed structure was retained by immediate water quenching after rolling, then the resulting structure was also of the /3-phase type. When this was recrystall: ;d at 800° C in the /3- phase region, grains of the order of 20-30ym could be obtained in a completely recrystallized structure. These observations led to a successful method to produce fine grains which involved the following steps, 1) rolling the material 25-30% at 750° C and retaining the hot worked structure by water quenching immediately after the pass, 2) heating the specimen at 750° C for a sufficient time. The retained rolled structure had second phase 7 2 precipitates at the grain boundaries and within the grains. This precipitation was due to the temperature - drop taking place during rolling. In this case, complete recrystallization took place in 15-20s and the resulting grain 21 size was 15-20M m. A similar method was used by Sure and Brown^ to obtain grains of the order of 15-40jum in size. The observed fine grain structure may be due to 1) an increase in nucleation rate caused by precipitates as suggested in several previous studies 4243 on recrystallization and 2) a decrease in initial grain growth rate due to dissolving 11 precipitates as suggested by Sure and Brown" .^ 2.3.3 The Thermomechanical Treatment for Eutectoid Alloys Grain sizes of up to 15M m were obtained in the experiments which have been described so far. Since the aim of this work was to reduce the grain size further, new attempts had to be made in this direction. The rate at which a metal recrystallizes is decided by the rate of nucleation and the rate of growth of the nuclei. The growth rate becomes fairly constant 44 for larger strains . In addition, the growth which is accomplished by the migration of 44 high-angle grain boundaries is a fast process in single phase materials. Hence the rate of recrystallization is mostly decided by the rate of nucleation. By increasing the rate of nucleation to a very high value, a large number of nuclei form and grow rapidly and after quick impingement they form a fine grain material. Grain growth is prevented by quenching the specimen after the minimum time necessary for recrystallization. The nucleation rate increases sharply with increasing strain, above a 45 minimum critical strain . In the present work, the maximum possible strain in a single pass (40-50% for eutectoid alloys) was used. The nucleation rate (N ) is related to 44 temperature by, 22 N = A • exp (-Q /RT) where Q n is the activation energy for nucleation over any temperature range within which Q n and A are single valued. Hence with increasing temperature, the nucleation rate also increases rapidly. A temperature as high as 800° C was used in the present work. Temperatures higher than this were not possible since the time involved for recrystallization was reduced to very low values (l-3s) and this would have been difficult to control. It was observed from the initial experiments (section 2.3.2) that recrystallization was achieved only when it was performed in the single-phase region. Hence recrystallization may be obtained at relatively low temperatures if the 0- phase can be obtained at a low temperature. In other words, a higher temperature will be more effective in increasing recrystallization rate in alloys where 0- phase is present at a low temperature than in alloys where |3- phase is present at a higher temperature. Eutectoid alloys are ones where j3-phase is present at a minimum temperature which is around 575° C for the CuAINi alloy system. Hence it was decided to use eutectoid alloys for the following experiments. Proper compositions were calculated as described in section 2.2 and alloys A, B and C were cast. A rolling temperature of 600° C was used in the process. During rolling, the temperature dropped to a lower value and precipitation of 1% took place as shown in the micrograph (Fig. 5(a)). This also may be advantageous in obtaining fine grain structure because of its possible effect on nucleation and initial grain growth. When the material was recrystallized at 800° C after warm work of 23 40-45% given at 600° C, recrystallization took place very rapidly and a fully recrystallized structure of grain size of the order of 10M m was obtained after annealing for 5-7s. Recrystallization took place more slowly at lower temperatures but the nucleation rate was lower and the minimum grain size possible was 20M m after annealing for 15-30s at 700° C and 25- 30Mm after annealing for 60-300s at 600° C. It was observed from micrographs that in most of the specimens, very fine grains of 5M m or less were present along the previous grain boundaries for conditions where recrystallization was not complete. Once recrystallization was complete, these grains had grown and the overall grain size had increased to 10Mm. This clearly suggested that a shortage of nucleation sites which mainly consist of previous grain 44,45 boundarieas in single phase materials was the factor which limited the grain size from being reduced further. Hence it was decided to subject fine grain materials to the above processing step which would give ultra fine grains. In other words, the recrystallization step was done twice in the processing. Thus the final successful thermomechanical treatment process which was able to produce 5M m grain size specimens involved the following steps as indicated schematically in Fig. 4. (1) Several rolling passes were given at 800° C till a thickness of approximately 4.3mm was achieved. Specimens were annealed at 800° C for 120-300s in between passes. (2) The specimens were heated at 600° C for about 15 minutes and a single pass of about 45% deformation was given before they were water quenched. (3) Recrystallization was done at 800° C. When heated for 15s, this yielded grains 800° C 600° C 500-1.000M m 15- 25M m 15s 5-10M m 3-7s 900s WARM ROLLING (=» 45%) WARM ROLLING 60-120s -2.2mm ~ U m m -4.3mm Figure 4. Schematic representation of the thermomechanical treatmenL Figure 5. Optical micrographs of a) the structure of alloy C after warm working at 600° C showing 7precipitates, x800, b) and c^  the structures of alloys A and B respectively after recrystallization at 800 C showing grains of Su m, xl,200. 26 in the range of 15-25/nm. (4) These fine grain specimens were heated at 600° C for 60-120s and subjected to a reduction of approximately 50% before being quenched in water. During this intermediate heating at 600° C, hardly any grain growth but some precipitation was observed. (5) The specimens were recrystallized again at 800° G for 3-7s, which produced grains of the order of 5M m in diameter. At the end of the whole processing the specimen thickness was around 1.1mm. Fig. 5(b) and (c) show the micrographs of 5M m grain size specimens from alloys A and B respectively. The first one is in the pseudoelastic state while the second one is of martensitic type. When recrystallization was performed once, grains of up to 10M m could be outlined. When this was performed twice, grains of 5M m could be obtained. If this is repeated for three times, the grain size may be further reduced slightly or it may reach a limiting value. This was not attempted in the present work since it was expected that the amount of work involved could not be justified by a possible reduction of grain size by a small amount. 3. PROPERTIES OF ULTRA FINE GRAIN MATERIALS 3.1 GRAIN GROWTH As described in section 2.3.3, a final solution treatment time of 3-7s was used to obtain grain sizes in the range of 5-10/ini. Some specimens were solution treated for longer times during which grain growth took place and grains of up to 450M m were obtained and these were used for tests on grain size effects on mechanical properties. For solution treatment times up to 120s, a chloride salt bath at 800° C was used to minimize heating time. A furnace at 800° C was used for longer solution treatment periods. The specimens were water quenched immediately after the solution treatment. The grain size was determined optically by measuring the average grain 46 diameter using the linear intercept procedure . The specimens were etched in a solution containing 2g FeCL,, lOcc HC1 and lOOcc H 2 0. Fig. 6 shows the change in grain diameter with increasing solution treatment time at 800° C for alloys A, B and C. The data would be expected to follow an empirical equation in the form of D = ktn, where D is the average grain diameter, t is the solution treatment time, and k and n are constants which depend on composition 42 and annealing temperature, but are independent of grain size . The parameter n is the 47 grain-growth exponent Hillert shows that n takes a value of 0.5 under ideal conditions and that this is the maximum value for n except for the cases where abnormal grain growth is observed. However, in most alloy systems the value for n is much lower than 0.5 due to inhibitions in grain growth such as second phase particles. The previously mentioned empirical relationship applies best if the initial (pre-growth) grain size is small 27 28 42 compared with the grain size which is being measured during growth . This is the case in the present work because of the ultra-fine nature of the initial grains. A log-log plot for alloys A, B and C is shown in Fig. 7. As can be seen, grain size is plotted against effective true grain-growth time instead of solution treatment time. At first, when the data was plotted as grain size-vs.-solution treatment time, a considerable deviation from a power law was observed up to a grain size of 20jum in alloys A and B and up to 80/xm in alloy C. These two grain size values correspond to a solution treatment time of approximately 15s. Such a deviation from a power law at low grain size values was also observed by Sure and Brown15. They explained their results using the dissolution of 7 2 precipitates. Such an explanation is not relevant in the present work since the alloys are single phase even at the minimum 42 grain size. Cotterill and Mould have given an alternative explanation. They suggest that a power law relationship is obtained only when the initial structure is equiaxed and strain free. In the case of samples which are solution treated in the worked condition, a linear relationship is often found only after annealing has been in progress for a significant period of time and the initial points generally deviate from a straight-line relationship. This arises because the total isothermal annealing time includes a period which is used to convert the initially deformed structure into a strain-free structure by primary recrystallization, before grain growth can begin. In these circumstances, the time which is required to produce a minimum but finite grain size by recrystallization is often different from that which would be required to produce the same grain size by grain growth in a permanently strain-free material with an initial grain size of zero. This effect is predominant at low time values when recrystallization time is comparable to total solution treatment time. 29 500 • A A l l o y A  A H o y B • A l l o y C 100 200 300 400 SOLUTION TREATMENT TIME ( s ) 500 Figure 6. Variation of grain size with increasing solution treatment time at 800° C for alloys A, B and C. 1000 100 ca C/3 < 1 10 100 1000 E F F E C T I V E TRUE GRAIN-GROW TH TIME ( s ) Figure 7. Log-Log plot of grain size-vs.-effective true grain growth time at 800 C for alloys A, B and C. 30 48,49 Beck et al defined the effective true grain-growth time as the actual total annealing time minus the time for completion of primary recrystallization plus the time for imaginary grain growth to the grain size actually produced by primary recrystallization. The same concept is used in the present work. In all the alloys, the time for completion of primary recrystallization was around 3s from the time of placing the specimen in the salt pot A fraction of this time was heating time and the rest of it was the true isothermal recrystallization time. The grain size obtained at the end of this was 5M m for alloys A and B and 10Mm for alloy C. Grain size and solution treatment time were plotted on a log-log scale for times above 15s and a linear relationship was obtained for all the alloys as these times were much longer than the times for primary recrystallization. From these plots the values for n and k were calculated and these values were used to estimate the time for imaginary grain growth from zero to the minimum grain size obtained at the end of recrystallization. The time values calculated were approximately Is for alloys A and B and about 0.25s for alloy C. This time was used in calculating the effective true grain-growth time for alloys A and B, but neglected for alloy C since it is very small. Fig. 7 shows the plot of grain size-vs.-effective true grain-growth time on a log-log scale. As can be seen, a linear relationship was obtained for the complete grain size range for all alloys. In all cases, the value obtained for n was close to 0.5 which is the maximum value for n in most alloy systems. This indicates the rapid grain growth taking place in these alloys. This is to be expected because of the high solution treatment temperature and the single phase nature of these alloys at that temperature. Hence fine grains can only be obtained in the present treatment procedure by quenching the specimens to stop grain growth immediately after recrystallization. 31 3.2 M s DETERMINATION The M s temperature was determined by optical microscopy. In some experiments, a well polished specimen was placed in a shallow copper container which was immersed in a proper bath. In other cases, specimens were mounted in conductive copper and they were placed in the bath. To obtain temperatures which are lower than room temperature, an alcohol bath was used which was slowly cooled by the addition of chilled alcohol at -100°C to obtain a cooling rate as low as 5°C/minute. To obtain temperatures above room temperature, a boiling water bath was used and the temperature was allowed to drop slowly. To obtain temperatures above 90° C, a boiling oil bath was used and the temperature was allowed to drop slowly. An optical microscope with magnification in the range of x8 to x40 was used to observe the structural change. The formation of martensite at M s was revealed by rumpling produced at the polished surface of the specimen. Table 1 (in section 2.3) gives the M s temperatures of the alloys cast. As shown in Fig. 3 (page 14), these results follow the same linear relationship as the other data from the literature when M s temperature was plotted against atomic percentage of Al. Fig. 8 shows the variation of M s temperature with grain size for grains in the range of 15-120M m in alloy B. It can be seen that the observations are 19 consistent with those of Dvorak and Hawbolt on CuZnSn pseudoelastic alloys, where lowering the M s temperature is attributed to increasing grain constraint with decreasing grain size. Sure and Brown15 also obtained similar results with CuAINi alloys. 32 5 0 Figure 8. Variation of M s temperature with grain size for alloy B. 33 3.3 TENSILE TESTS Tensile specimens were prepared by spark machining due to the brittleness of the material in the as-rolled condition. This was done after the final rolling pass of the thermomechanical treatment. Flat tensile test specimens of 20mm gauge length and 5mm width were spark machined from the as-rolled strips of 1mm thickness. The tensile specimens were then solution treated at 800° C for different times as determined in Fig. 6 and water quenched. They were then given a stress relieving treatment at 100-125° C for 300s to remove any quenching stresses. After heat treatment, the tensile specimens were mechanically polished to remove any surface contamination and to obtain a smooth surface. The surfaces and edges were examined metallographically for quenching cracks. Tensile tests were performed on a floor model Instron tensile testing machine at a strain rate of 1.06 x 10"* s"1. For the tests performed at room temperature, conventional type flat tensile test grips were used and a 1.25mm-10% extensometer was clipped on the specimens for accurate measurements of strain. For tests at temperatures different from room temperature, specimens were immersed . in a suitable medium such as hot water, chilled alcohol or liquid nitrogen. Since it was not possible to use an extensometer in these environments, strain measurements were made directly 26 from the crosshead movement with corrections being made for machine compliance . 34 3.3.1 Effect of Grain Size on Tensile Properties Tensile tests to fracture were performed on alloys A, B and C at varying grain sizes. Alloy A was of the pseudoelastic type and alloys B and C were of the martensitic type. Fig. 9(a) and (b) show two typical stress-strain curves obtained for alloy C. As can be seen, there is a difference in the nature of the curve with grain size. The coarse grain material has three portions to the stress-strain curve with an initial region of high slope, then a plateau of lower slope before the final region of high slope leading to fracture. The fine grain material has only the first two portions of the stress-strain curve and fracture occured before the region of high slope. This was explained by Sure and Brown^ using the transition stress between stages 1 and 2 and the slope of stage 2 {ax and da/de respectively in Fig. 9(a)). They showed that both of these parameters increase with decreasing grain size according to a Hall-Petch type relationship. For fine grain materials, the transition stress between stages 2 and 3 becomes very large and fracture intervenes before it is reached. Most of the stress-strain curves obtained in the present work are similar to Fig. 9(b) since the grains are reasonably fine. The general shape of the stress-strain curves obtained for alloys A and B are very similar to those of alloy C. Fig. 10 shows the stress-strain curves of all three alloys at a constant grain size of 50-60M m. This clearly shows that despite the similarities, there are considerable differences in the values of the main parameters which define these stress-strain curves, particularly the stress and strain to fracture (a^ . and e^ .). As far as the values of oF and e f are concerned, the differences between alloys A and B are small although the values for alloy B cases when the same grain size materials are e^ . for alloy C are always much higher. A given in section 3.6. 35 are higher than those of alloy A in most compared. However, the values of and possible explanation for this difference is The transition stress (aj is the stress associated with the transition between the initial elastic and the transformational elastic plus plastic response of the stress-strain curve. For an alloy in the martensitic state, ax is associated with the homogeneous growth of favourably oriented variants. In pseudoelastic alloys, a t is the stress required for the nucleation of stress-induced martensite. The linear portion beyond the transition stress ai is the most useful one for interpretation of martensite deformation behaviour. The strain increase is obtained by homogeneous growth of favourably oriented martensite variants. A lower gradient, da/de, means easy, relatively unrestricted movement of martensite plates over large distances resulting in a large strain increase for a small increment in stress. A' high gradient indicates difficult, restricted martensite plate movement over small distances, giving rise to a small strain increase for the same increment in 15 18 stress. It has been shown previously in CuAINi and in CuZnAl that the parameters Oi and dff/de increase according to a Hall-Petch relationship with decreasing grain size. The Hall-Petch type relationship originates from the barrier effect of grain boundaries to slip propagating from grain to grain and from the build up of stress concentration owing to dislocation pile up. It has been pointed out by Khan and Delaey13 in their work on CuAl martensites that the increase in ot is due not only to increasing grain constraints but also to a decrease in martensite plate thickness with decreasing grain size. For the same reasons, a larger stress is required to overcome the restraining factors inhibiting martensite plate movement which in turn causes dff/de to increase with decreasing grain STRESS (MPa) <JS" 9 o C < CO a o o o o o o o S l a g * 3 J t o -ll II -to Ol O O O " 1 3 (TO STRESS (MPa) o o o o o o o o o o o o o S l o g * 1 o O 00 H > t o to o O 4 U l -l l i i ; to o" o t n n'3 o C L CO CO UJ CC t-co a. 2 CO CO u CC r~ CO 2 « a » S T R A I N (%) CO CO cu e; 400 (-co 200 A 11 or C C S = 10|i m M =220*C 2 « a * S T R A I N (7.) 2 4 a s S T R A I N ( r . ) a) b). c) Figure 10. Stress-strain curves at a constant grain size of 50-60um for alloys A, B and C, tested at 22° C 38 size. In the • present work too, an increase in ax and do/de was also apparent with decreasing grain size as shown in previous studies. 19 Dvorak and Hawbolt suggested that the ratio (grain size/thickness) is the most important parameter in controlling the mechanical properties of polycrystalline alloys. This is because the grain size-to-thickness ratio is a measure of the degree of grain constraint in flat tensile test specimens. Grain constraint can be considered to be a function of the fraction of the total grain surface which is constrained by the presence of adjacent grains. Hence when all the specimens tested are of almost the same thickness (within the range of 1-1.1mm) and when they all have very fine grain structure (the majority of the specimens tested have grain sizes less than 100M m), it is not particularly important to use the parameter(grain size/thickness); instead grain size itself is used in all the plots in the present work. Fig. 11(a) shows the variation of with grain size for alloys A, B and C. The graphs clearly show the major advantage of producing ultra fine grain specimens. At fine grain sizes, specimens are much stronger and as will be seen shortly, they have much higher strains to failure. The strength value of 1,200 MPa obtained in the present work is probably the highest value obtained in the CuAINi system. It is considerably higher than the 800 MPa reported by Duering et a l 1 1 in CuAINi PM alloys of grain size equal to 20M m and the 930 MPa reported by Sure and Brown15 in grain refined CuAINi alloys of grain size equal to 15Mm. Miyazaki et al5^ have proposed that in CuAINi alloys, cracks are formed along grain boundaries as the stress is concentrated there due to the large elastic 39 b) Figure 11. Variation of fracture strength with a) grain size and b) (grain size) - 1 / 2 for alloys A, B and C, tested at 22° C. 40 anisotropy. Cracks usually initiate at three-fold nodes where stress concentration is highest The cracks then propagate along grain boundaries or transgranularly along favourably oriented crystal planes. At large 0-grain sizes, fewer grain boundaries and three-fold nodes are present than at fine grain sizes. However, due to the large elastic anisotropy, high stress concentrations develop at grain boundary nodes in large grains even at small tensile stresses. Thus cracks nucleate at nodes at relatively low tensile stresses. Crack propagation is also relatively unhindered as larger grain size provides larger intergranular and transgranular paths before the crack propagation mode is altered. In fine grain size specimens, higher stress is required for crack nucleation at grain boundary nodes since the stress concentration is less due to the elastic anisotropy. Also, the crack propagation path encounters more restrictions and has to follow an irregular path along the grain boundaries or across the grains, which requires higher propagation energy. These factors contribute to the higher o ^  of fine grain samples. -1/2 Fig. 11(b) shows the variation of with (grain size) . A Hall-Petch type relationship was generally observed in all the alloys. In alloys A and B, the relationship is valid throughout the grain size range. In alloy C, the Hall-Petch relationship seems to be valid down to a grain size of 40Mm. When the. grain size is further reduced, there is an increase in strength but the values are less than predicted by the Hall-Petch relationship. The Hall-Petch equation has been found to express the grain size dependence of stress for both ductile and brittle type fractures .^ While the Hall-Petch equation is a very general relationship, it must be used with some caution especially for ultra fine grain materials. For example, in the present work, if the Hall-Petch equation 41 o is extrapolated to the smallest grain size imaginable (approximately 40A), it would predict strength levels higher than the theoretical cohesive strength. Such an extrapolation is in error because the equations for the stresses in a pile-up on which the Hall-Petch equation is based were derived foT large pile-ups of dislocations. For small pile-ups 52 different equations must be considered . Thus a different mechanism probably occurs in alloy C with very fine grain specimens so that is not as predicted. Another possibility is that at a stress value lower than predicted by a Hall-Petch equation, fracture takes place due to some cumulative effect of the existing cracks or some other defect which becomes critical at these high stress values. 1518 In most previous work ' , when fine grain specimens were produced, a second phase was also invariably present in addition to the j3-phase. In the CuAINi specimens of 15-40M m produced by Sure and Brown15, 72-precipitates were present as a uniform distribution in the 0- phase matrix. They concluded that this distribution did not affect the strength significantly. However, the experiments performed by Kasberg and Mack1^ showed a reduction in ultimate tensile strength and percentage elongation when 7 2 was present in addition to the single phase structure. In the present work, even at the finest grain size, the alloy is of the single /J-phase type without any y 2-precipitates. This is possibly one of the reasons for the high strength values obtained, which are considerably higher than in any previous work. Fig. 12(a) and (b) show the variation of fracture strain e^ . with -1/2 grain size and (grain size) for alloys A, B and C. The plots are similar to the ones which were obtained for a r . The explanation given earlier is appropriate here too. It is particularly interesting to note the validity of the Hall-Petch relationship down to a grain 42 Figure 12. Variation of strain to failure with a) grain size and b) (grain size)"1 /2 for alloys A, B and C, tested at 22° C. 43 size of approximately 40M m in alloy C as in the case of The highest value obtained for e^ . in the present work is 10%. This is considerably higher than the 7% obtained by Sure and Brown15 in their alloy of 15Mm grain size and the 7% by Duering et a l 1 1 in their PM alloy of 20M m grain size. Two possible reasons for this may be the relatiyely finer grain structure and the absence of any second phase. The e ^ values obtained here are almost 16 times higher than the ductility of 0.6% in the cast condition11. For alloy C, the increase in e^ . in a fine grain specimen with a grain size of 10M m is more than three fold compared to a specimen with a grain size of 350M m (3% compared with 10%). A fracture strain of" 10% is particularly attractive in the highly britde CuAINi alloy since this to an extent satisfies the most important purpose of the grain refinement, i.e. making this alloy more ductile to make it more useful for practical applications. 3.3.2 Effect of Temperature on Tensile Properties Alloys A and B were tested at liquid nitrogen temperature for a -1/2 range of grain sizes. Fig. 13 shows the variation of with (grain size) for alloys A and B. Fig. 14 shows the corresponding curve for e j. . The advantage of producing fine grain materials is again quite apparent because of the increase in and e^ . . A Hall-Petch type relationship is valid at liquid nitrogen temperature for the complete grain size range from 200Mm to 5Mm. When the slopes of the Hall-Petch lines corresponding to tests at 22° C and -195°C are compared, it can be seen that the difference between 1/2 the gradients is very small. For example, the gradients are 43 MPa-(Mm) at 22 C 1/2 and 39 MPa-(Mm) at -195 C for alloy A. Thus the gradient is almost constant with changing temperature. The most common mechanism which is used to explain the gradient 44 1000 ~ 800 U 600 -z >^ 4 00 b 2 0 ° -0.0 AT LIOUID NITROGEN A A I lo> A g) A l loy B 0.1 0.2 GRAIN S I Z E 0.3 0.4 -1/2/ 0.5 Figure 13. Variation, of fracture strength with (grain size) - 1 / 2 at -195°C for alloys A and B. 6-a fa. < A A AT LIQUID N I T R O G E N A 1 1 o v A ® A 1 1 o v B 0.0 0.1 0.2 0.3 GRAIN S I Z E ~ , / 2 ( , u m " / 1 ! ) 0.4 0.5 Figure 14. Variation of strain to failure with (grain size) 1 / 2 at -195°C for alloys A and B. 45 is based upon the stress concentration at the' head of the dislocation pile-up which is high enough to create dislocations at the boundary in the new grain. Atomic ledges in grain boundaries are possible sites for this. Such a mechanism is not strongly temperature-dependent The same mechanism probably explains the present results. When intercepts on the stress-axis are compared, an obvious increase can be seen at liquid nitrogen temperature (70 MPa at 22° C and 370 MPa at -195° C for alloy B). This intercept is usually interpreted as the friction stress needed to move unlocked dislocations along the slip planes. This term depends strongly on temperature, strain and alloy (impurity) content Since the friction stress will be high at low temperature, a higher value for the intercept at liquid nitrogen temperature is as' expected. Fig. 15(a) and (b) show the stress-strain curves for alloy B at temperatures -100°C and 90° C at a grain size of lOAim. The M s of this alloy is 40° C. While the general shape of the stress-strain curves are similar, i.e. both having two different regions, the value of ax is considerably higher at 90° C and this reduces the extent of the low gradient second stage before fracture intervenes. Since the second stage is responsible for a considerable portion of the strain to failure, the value foT e ^  is generally higher at low temperatures. This probably explains why a larger e^ . was obtained at liquid nitrogen temperature when compared to that of room temperature (compare Fig. 14 and Fig. 12(b)). Hence in this system, both and seem to take higher values when tests are performed at lower temperatures. This is in contrast to the generally expected behaviour, which is that strength decreases and ductility increases as the temperature is increased. Fig. 16 shows the variation of Oi and a f with temperature for 1 0 0 0 6 0 0 J 3 « 5 8 0 1 2 3 4 STRAIN (%) STRAIN (%) Figure 15. Stress-suain curves for alloy B at varying temperatures. 47 900 T E M P E R A T U R E f C ) Figure 16. Variation of transition stress and fracture strength with temperature for alloy B. 48 alloy B (M s = 40° C) of 10M m grain size. This is very useful in obtaining a clear idea of the variation of tensile behaviour of the alloys with temperature. Although the number of points in the plot are probably not enough to make definite conclusions, ax appears to show a decrease as temperature approaches M s , and below M s the values of a, remains more or less constant The variation of a, with temperature has been fully 53,54 investigated in both single crystal and polycrystalline strain-memory alloys . In CuZn 53 single crystals, Arneodo and Ahlers have shown that c/i decreases linearly with temperature and becomes zero at M s . The variation of ax with temperature can be calculated approximately from a Clausius-Clapyron type equation, da/dT = AH/AeT V 1 o m where AH = heat of transformation. Ae = strain corresponding to complete transformation to martensite. T Q = temperature at which the matrix and martensite phases are in equilibrium at zero stress. V _ = molar volume, m It was shown subsequently by Patel and Cohen"^ that although martensite forms spontaneously on cooling below M s , a finite stress is needed to overcome any friction force resisting the martensite-austenite interface movement Thus the prediction of a zero transition stress, au does not hold even in single crystals. In polycrystalline metals, at is even higher due to increasing grain constraints .^ At temperatures below M s , a, remains more or less constant with temperature .^ In the present work, values of a, are nearly constant at temperatures below M s , confirming the predicted behaviour of d below M s . Fig. 16 appears to show a progressive reduction in a f with 49 temperature although the number of points may not be sufficient to make definite conclusions. This may be explained by considering the effect of martensite growth on the tensile behaviour of the alloy. As pointed out by Miyazaki et al" ,^ j3- CuAINi specimens in the martensitic state have many deformation modes such as twinning or boundary movements between variants. Thus the stress concentrations at grain boundaries caused by high elastic anisotropy are easily relaxed. Consequently a higher stress is required so that stress concentrations at complex grain boundary nodes are high enough to nucleate a crack. Miyazaki et al5^ also showed that martensite plates often impede crack propagation when they are formed at the crack tips and crack propagation along grain boundaries is easily deviated due to relaxation of stress concentration at grain boundaries. Another observation is that the cracking tendency during quenching from 800° C is less when the quenching temperature is lower than M s , since thermal stresses during quenching are relaxed by martensite formation before thermal stresses reach the fracture stress. This may be one of the reasons for the observed slightly higher strength values in the martensitic state (alloy B at room temperature) when compared to the pseudoelastic state (alloy A at room temperature). It can be seen from Fig. 16 that at higher temperatures a higher stress a i is required for nucleation of stress-induced martensite (SIM). At temperatures far above M s , fewer deformation modes are available for relaxation of stress concentrations at grain boundary nodes due to the higher a,. Consequently crack nucleation at grain boundary nodes and crack propagation along grain boundaries is easier, accounting for the lower of the specimens. At temperatures nearer M s , SIM formation is easier due to the lower aY and due to relaxation at high stress concentration regions. Thus higher stresses are required before cracks can nucleate and propagate. This is 50 reflected by a higher for the specimens. It is interesting to note that if both curves in Fig. 16 are extrapolated, ax will be higher than at a temperature of around 100° C. At this temperature, the stress required to nucleate and propagate cracks will be lower than the stress required to form SIM; i.e. tensile fracture will probably occur even before SIM begins to form. 3.3.3 Fractography Fig. 17(a) and (b) show fracture surfaces of alloy A at two different grain sizes. As can be seen, at a grain size of 60Mm the fracture is almost completely due to intergranular brittle fracture. However, at 10M m grain size, the fracture is a combinadon of intergranular and transgranular brittle fracture, the latter being dominant As pointed out in section 3.3.1, the cracks nucleate at grain boundary nodes due to high stress concentration. The grain boundaries provide the easiest crack propagation path and when the grain size is large the grain boundaries . are long and relatively straight Hence the tendency of the crack to deviate from the intergranular path is low and this is reflected in a largely intergranular fracture surface and a low . A t fine grain sizes, the nucleation and initial growth of cracks will probably take place in the same way as in coarse grain size specimens, as is evident by the presence of some areas of intergranular fracture. However, due to fine grain size, the grain boundary orientation changes over small distances and thus the crack propagation path encounters many deviations and can change from an intergranular path to a favourably oriented transgranular path. Higher energy will be required for crack propagation in this case and this is reflected in a higher . However, in alloy A, fracture was primarily brittle in nature over all grain sizes. This clearly explains the observed low values for 51 <7j. and e^ . in this alloy for the complete grain size range. Fig. 18 shows the fracture surface of alloy A with a grain size of 10M m fractured at -195°C. When this is compared to the fracture surface of a similar specimen fractured at 22° C [Fig. 17(b)], it can be seen that the fracture surfaces are somewhat similar. Both specimens show a fracture mode which is a combination of intergranular and transgranular type brittle fracture. A close observation reveals that the contribution of intergranular brittle fracture in the total fracture is higher at -195°C when compared to that of at 22° C. This may be due to increasing brittleness of the grain boundaries with decreasing temperature so that there will be less tendency for the cracks to deviate from the easy path provided by grain boundaries. This observation suggests a lower o ^ at -195° C. However, the current experiments showed results which are contrary to this and this is possibly due to the stress relieving effect of martensite as explained in section 3.3.2. Fig. 19(a) and (b) show two typical tensile fracture surfaces of alloy C at grain sizes of 200Mm and 10M m. At the coarse grain size, fracture is due to a combination of transgranular brittle-type fracture and microvoid coalescence-type ductile fracture. At 10M m fine grain size, the surface is almost completely microvoid coalescence-type fracture. Microvoid coalescence fracture is associated with bulk plastic deformation in the alloys and there is ample evidence in the stress-strain curves of this alloy [Fig. 10(c) for example] to substantiate this. Crack nucleation and propagation is probably due to localised plastic deformation in this alloy. Surprisingly no significant evidence for intergranular-type fracture could be found in this alloy even at relatively coarse grain sizes. The absence of any second phase along grain boundaries in this alloy a) grain size = 60M m, ,\150 53 Figure 18. SEM fractograph showing the tensile fracture surface of specimen A (grain size = 10M m). fractured at -195°C, x500. a) grain size = 200M m, x600 b) grain size = 10M m, x600 Figure 19. SEM fractographs showing tensile fracture surfaces of alloy C at varyin grain sizes, fractured at 22° C. 55 may be one of trie reasons for this. The possible presence of oxide along grain boundaries in alloys A and B is explained in a later section and may be a contributing factor for intergranular-type fracture commonly observed in those alloys. As shown in section 3.3.1, alloy C has high values for and e^ . for the whole grain size range. The values are particularly high for fine grain size specimens. The observed ductile fracture mode clearly explains this. 56 3.4 RECOVERY PROPERTIES Specimens used for recovery tests were similar to those used for tensile tests. Tests were performed on an Instron tensile testing machine at a strain rate of 1.06 x 10"4 s"1. The specimens had not been subjected to any form of straining before being loaded for the recovery tests. Strain measurements were made directly from crosshead movement with corrections being made for machine compliance . Tests to find recovery were performed on alloys A and B, but not on alloy C, because of practical difficulties caused by the unfavourable M s (approximately 220° C). 3.4.1 Effect of Temperature Fig. 20 shows typical stress-strain recovery tests at 20° C and -100°C corresponding to conditions above and below M s [M s = -15°C] . Specimens were loaded to 2-2.5% strain at a suitable temperature and the amount of strain recovered was measured after complete unloading. A higher strain could not be applied because of the low fracture strain of alloy A. Specimens were subsequently heated above M s to obtain the recovery due to heating and this was measured using the crosshead movement which was in the unloading direction to remove the load that developed during heating. The recovery obtained occurs partly during unloading (pseudoelasticity) and partly during heating (strain-memory effect). An obvious difference could be seen in the loading stage of the stress-strain curves [in Fig. 20(a) and (b)] due to the varying temperature as discussed in section 3.3.2. In this case, however, specimens were not strained to fracture. The relative <0 a, CO w a: H co 300-400 A Hoy A CS = l5Mnn M, = - 1 5*C Test Temp. = 20'C 300 200 100-1 / H«aflng 0- / i / STRAIN (%) a) 500-1 400-1 Alloy A GS = I 5LI m M,=-l5*C Teil Temp. = -100*C 2 3 STRAIN (%) b) Figure 20. Stress-strain curves of alloy A specimens at varying temperatures showing recovered strain on unloading and heating. —4 58 1 0 0 >* OS W > o w 02 E-2 W O 02 W o* 8 0 -6 0 -4 0 -2 0 -A A Z3 A A T o t a l R e c o v e r y ^ © 0 S t r a i n M e m o r y R e c o v e r y j»>— © P s e u d o e l a s t i c R e c o v e r y / o ~ 9 © ^ ^ ^ ^ ^ M = -G r a i n S i z e = 1 0 - 2 0 A * m s A p p l i e d S t r a i n = 2 - 2 . 5 J S 15*CO ^ O 1 , - 2 0 0 - 1 0 0 0 T E M P E R A T U R E ( ° C ) 1 0 0 Figure 21. Variation of total recovery and the recovery by pseudoelasticity and strain memory effect with temperature for alloy A. 59 amount of recovery in the unloading stage and in the heating stage changes markedly with temperature. At temperatures above M s [Fig. 20(a)], recovery occurs mainly on unloading due to the pseudoelasticity effect, whilst at temperatures below M s [Fig. 20(b)], the major portion of recovery is on heating due to the strain-memory effect Fig. 21 shows the results of tests for temperatures in the range of -195° C to 90° C. Percentage recovery was used in these plots and this was calculated as the recovered strain relative to the initial loading strain. At temperatures below M s , strain-memory recovery constitutes approximately 75% of the total recovery, whilst at temperatures above M s , pseudoelastic recovery constitutes approximately 90% of the total recovery. Fig. 21 also shows that the total recovery is more or less a constant (within the range of 88-92%) with change in temperature. Thus the major recovery mode, whether pseudoelastic or strain memory, does not have any significant effect on the total 57 recovery obtained. Similar results were obtained in previous work on CuZnSn and on CuAINi 1 5. 3.4.2 Effect of Grain Size Fig. 22 shows the change in percentage recovery (including both pseudoelastic recovery and strain memory recovery) with decreasing grain size in alloys A and B. The initial loading strain used was 2%. A higher strain could not be used because of the low fracture strain of these alloys at coarse grain sizes. Although there is a considerable amount of scatter in the results, it can be seen that there is a continuous reduction in recovery from 96% to 86% as the grain size is reduced from 100M m to 5M m. Despite the reduction, the recovery of 86% at 5M m grain size is quite attractive. 60 100 PH w > o w 90A W CJ 3 8 0 H Applied Strain=2 A A l l o y A • A l l o y B oy /o 70 0 25 50 75 100 GRAIN SIZE^m) 125 Figure 22. Variation of percent recovery with grain size at constant applied strain of 2% for alloys A and B. 61 19 As Dvorak and Hawbolt have pointed out, the amount of non-recoverable strain increases with decreasing grain size, since due to increasing grain constraint more plastic deformation by normal slip takes place along with martensite growth because of the higher stresses involved. A recovery of approximately 90% was obtained at a grain size of 25M m. This is higher than the recovery values of 75-80% obtained for similar grain size 18 values in CuZnAl aloys . This effect is possibly due to the higher ductility of CuZnAl compared with CuAINi. This would result in more plastic deformation during loading, leading to a larger non-recoverable strain, and so lowering the recovery obtained. The present recovery value of 90% is also higher than the value of 85% obtained by Sure and Brown15 in CuAINi under similar testing conditions. This may be possibly due to the single phase nature of the present alloy at fine grain sizes whereas in the other work 72-precipitates were present in addition to the |3- structure. A similar suggestion has 58 also been given by Krishnan et al and they say that any transformation that reduces the volume fraction of parent phase will reduce the absolute magnitude of the recoverable strain. The precipitates interfere with the martensite plate movement and its growth. 3.4.3 Effect of Increasing Strain Tests were performed to study the effect of increasing initial strain on recovery properties of alloy B with specimens of 10-25Mm grain size. Strain values in the range of 0.5% to 5% were used in the different tests. Fig. 23 clearly shows the effect of increasing initial strain on pseudoelastic recovery and total recovery. The total recovery decreases by a small amount from 100% to 86% when the initial strain is 62 © Total Recovery I i i i I i 0 1 2 3 4 5 I N I T I A L A P P L I E D S T R A I N ( % ) Figure 23. Variation of percent total recovery and percent pseudoelastic recovery with increasing initial strain for alloy B. 63 increased from 0.5% to 5%. However, the corresponding decrease in pseudoelastic recovery is large, from 88% to 26%. The total recovery of 86% obtained at 5% applied strain is quite attractive. It is higher than the 80% recovery obtained at 5% applied strain in grain 15 59 refined CuAINi specimens by Sure and Brown . Enami et al have also shown excellent strain-memory recoveries (above 85%) with up to 6% applied strain in grain refined CuZnAl alloys. Alloy B (M s = 40° C) is martensitic at room temperature at which the specimens were tested. At low initial strain, most of the recovery is pseudoelastic. Under these conditions, the stress applied is either not enough to start reorientation or not sufficient to complete reorientation, which is essentially the process of converting martensite of a given variant into a variant of a more favourable orientation. Hence at low initial strains, the strain mainly consists of elastic strain and this is the reason why a high percentage elastic recovery was obtained. At a high applied strain, the irreversible reorientation will be complete and hence the major portion of the strain which is associated with the reorientation process will not be recovered. This explains the large decrease in pseudoelastic recovery with increasing initial strain. The decrease in total recovery with increasing applied strain is probably due to the increasing amount of plastic 19 deformation caused by the grain constraint effect . 64 3.5 FATIGUE PROPERTIES 3.5.1 Effect of Grain Size on Fatigue Life Fatigue tests were carried out on an MTS test machine. Specimens used for fatigue testing were similar to those used for tensile and recovery tests. Specimens with a range of grain sizes were tested in tensile loading under load control. They were cycled between a minimum and a maximum stress. The values of the minimum and maximum stresses were 20 MPa and 250 MPa for alloy A, 20 MPa and 280 MPa for alloy B and 30 MPa and 330 MPa for alloy C. The values of the maximum stress were selected such that the fatigue life of all three alloys was comparable. A separate set of fatigue tests was performed on constant grain size specimens of alloy C with a minimum stress of 30 MPa and with a variable maximum stress. Tests were carried out at a frequency of 5Hz. Strain measurements were taken at regular intervals during the tests at a reduced frequency of 0.05Hz using the stroke measurements of the machine. Fig. 24 shows a typical variation in strain on cycling specimens of 100M m grain size in alloy A, 30M m grain size in alloy B and 250Mm grain size in alloy C under load control conditions. Although the strain measurements may not be very accurate since an extensometer was not used, nevertheless a consistent trend can be observed. A rapid decrease in strain/cycle was observed in the first few cycles and this was followed by a gradual decrease and finally the strain/cycle became more or less a constant and remained so until fracture. This phenomenon on cycling was also observed 17 15 in some previous work, particularly in pseudoelastic 0-CuZnSn and 0 - CuAINi . 65 0.9-1 10 100 1000 10000 N U M B E R O F C Y C L E S ( N ) 100000 Figure 24. Variation of strain per cycle with cycling in specimens of alloys A, B and C. o O 80 o •a D 8 0 z Ed 20 • A l l o y C G r a i n S i z e = 150,um C o n s t a n t A p p l i e d S t r e s s = 4 80 M P i 2 3 4 5 N U M B E R O F C Y C L E S ( N ) Figure 25. Variation of percentage pseudoelastic recovery with number of cycles for the first few cycles for alloy C. 66 17 Brown suggested that the effect was due to the large non-recoverable strain in the first few cycles. It was further suggested15 that with increasing cycles, non-recoverable strain in each cycle progressively decreases and a steady value of strain per cycle is reached. Although such elastic behaviour in pseudoelastic alloy A is as expected, further experiments were performed to show that this was happening in the martensitic state. A specimen of martensitic alloy C was loaded in the Instron machine to a stress of 480 MPa and then unloaded. This was repeated for a few cycles and as can be seen in Fig. 25, the value of pseudoelastic recovery seems to increase with number of cycles and a pseudoelastic recovery of 85% could be obtained after 5 cycles. This is probably because the thermal martensite goes through the non-recoverable reorientation during the first cycle and in the subsequent cycles most of the strain is due to the elastic deformation of the reoriented martensite. In most of the fatigue tests, the strain in the steady state was found to be in the range of 0.50-0.95%. Since this amount of strain is commonly observed in practical loading applications, the fatigue life values obtained here will be useful in assessing the effectiveness of these alloys against fatigue. Fig. 26(a) shows the dependence of fatigue life on grain size for alloys A, B and C. The effect of grain size is very large for alloys A and B, but much smaller for alloy C. Alloys A and B are not very useful for practical applications5 when their grain size is more than 30um since at that grain size a fatigue life of less than 7,000 cycles would be obtained. In alloy C, the fatigue life is 30,000 cycles when the grain size is approximately 250M m and increases significantiy as the grain size is reduced. Hence, alloy C can be used effectively under fatigue conditions without much grain refinement A fatigue life of 275,000 cycles obtained for alloy C at a grain size of 10Mm when tested at a stress of 330 MPa and a strain of 0.6% is very attractive and indicates 67 that this alloy will be useful in cyclic applications. The fatigue life obtained in the current work is the highest value obtained so far in CuAINi alloys and is 5 times more than the 50,000 cycles obtained by Sure and Brown15 in their grain refined CuAINi specimens of 15/xm grain size subjected to similar loading conditions. This increase may be due to the higher fracture strength caused by the finer grain structure and single phase nature of the alloys in the present work as explained in section 3.3.1 . Oshima and Yoshida^ have concluded in their work on CuZnAl polycrystalline alloys that the fatigue life is higher in the martensitic state than in the pseudoelastic state. They explained this by considering the accommodation of local strain by different martensite variants. However in the present work, alloys B and C are martensitic. Alloy C which has a much higher also has a much higher fatigue life. Hence the effect is primarily due to high . The observed slightly higher fatigue life for martensitic alloy B over pseudoelastic alloy A is also most probably due to the same a^. effect (slightly higher for alloy B) rather than martensitic/pseudoelastic effect Fig. 26(b) shows the plots of number of cycles to failure (on a -1/2 log scale) against (grain size) for alloys A, B and C. Despite the considerable amount of scatter in the results, a fairly reasonable Hall-Petch type linear relationship could be observed. This type of relationship was also obtained in the past in some low stacking-fault energy materials and was explained using the grain boundary control of the rate of cracking due to planar slip*'1. Fig. 27 shows two different plots of stress (S = a in this ° max case) against number of cycles to failure (N). As can be seen,- the number of cycles of stress which the alloy can endure before failure increases with decreasing stress. As in 68 1 O O O O O O i a) lOOOOOO 100' i I 0.0 0.1 0.2 0.3 0.4 0.5 GRAIN SIZE" , / J (Mm" / 2 ) -b) Figure 26. Variation of number of cycles to failure (N) with a) grain size and b) (grain size)- for alloys A, B and C. 600 Alloy C Min. Stress=30MPa 200 ' 1 . " T - -7-7-1 M 1 1 i j I I I I I II IT • ; I I I I I I I 100 1000 10000 100000 1000000 C Y C L E S TO F A I L U R E ( N ) Figure 27. Two S-N type plots where the number of cycles to failure is plotted against the maximum applied stress for alloy C. 70 most copper alloys, the S-N curves in the present work slope gradually downwards with increasing number of cycles and no true fatigue limit could be observed. In addition, when both the S-N curves which correspond to 75ym and 400Mm grain sizes are compared, the effect of grain size on fatigue life again becomes obvious. If Fig. 26(a) (cycles to failure vs. grain size) and Fig. 11(a) (fracture stress vs. grain size) are compared, a reasonable amount of similarity can be observed in the plots for all three alloys. This clearly suggests that the fatigue life is controlled by fracture strength of the alloy. This is in accordance with the generally accepted fact that the fatigue life is controlled by the fatigue ratio which is the stress 62 relative to the . Since the applied stress is more or less the same, o ^  becomes the most important parameter and this increases rapidly with decreasing grain size as discussed earlier. When the fatigue mechanism is considered, planar slip causes the grain boundaries to control the rate of cracking^1 and hence grain size is a very important parameter in deciding fatigue life. 3.5.2 Fractography Fig. 28(a) and (b) show typical fatigue fracture surfaces of alloy A at grain sizes of 75um and 5um respectively. For the coarse grain material, fracture is mosdy due to intergranular-type brittle fracture although some transgranular regions are also present For fine grain materials, the fracture mode consists of both intergranular and transgranular-type brittle fracture. These observations are fairly similar to those observed in the fracture surfaces of alloy A after tensile loading. As described previously in section 3.3.3, the observations explain the higher fatigue life for fine grain materials. It 71 a) grain size = 15um, xl50 b) grain size = 5M m, x600 Figure 28. SEM fractographs showing fatigue fracture surfaces of alloy A at varying grain sizes. 72 a) grain size = 250/im, x75 b) grain size = 25M m, x600 Figure 29. SEM fractographs showing fatigue fracture surfaces of alloy C at varying grain sizes. 73 17 was suggested in the past that fatigue crack nucleation begins at three grain intersections which are areas of high stress concentration, and on cycling these cracks slowly propagate along grain boundaries and eventually link up. In the current work too, the observations suggest that the fatigue crack nucleation and initial propagation occur intergranularly and the overload fracture probably occurs transgranularly after the crack has propagated a certain threshold distance. Fig. 29(a) and (b) show fatigue fracture surfaces of alloy C at grain sizes of 250M m and 25M m respectively. At 250M m grain size, fatigue crack propagation has occurred both intergranularly and transgranularly as in specimens of alloy A. However, at fine grain sizes, microvoid coalescence dominates the fracture mode although some intergranular and transgranular regions are also present in most of the specimens. A careful observation of Fig. 29(b) also reveals the presence of fatigue striations in a few areas. In very fine grain specimens, the fracture was completely due to microvoid coalescence-type ductile fracture. In the latter case, the fatigue load applied was around 30% of the fracture strength and this will result in low stress concentration at the grain nodes and is presumably not enough to initiate intergranular cracking. As explained in section 3.3.3, alloy C shows a considerable amount of ductility and the fatigue crack initiation and propagation up to failure must have been due to localised plastic deformation. Thus the very high fatigue life values obtained in alloy C at very fine grain sizes must be due to the high fracture strength and ductile fracture mode of these specimens. 74 3.6 EXPLANATION FOR DIFFERENCES EN MECHANICAL PROPERTIES BETWEEN ALLOYS Alloys A, B and C showed considerable differences in their mechanical properties such as tensile strength and fatigue strength when tested over a grain size range. All three alloys would have been expected to show similar mechanical properties at comparable grain size since they all were single 0- phase type. It was found that the mechanical properties of alloys A and B were fairly similar although alloy B showed slightly higher values compared to alloy A in almost all cases. This difference is probably due to the martensitic state of alloy B at room temperature when compared to the pseudoelastic state of alloy A at the same temperature. As explained in section 3.3.2, stress concentrations at grain boundaries and other possible thermal stresses are easily relaxed due to the several deformation modes available in the martensitic state and hence crack initiation and propagation along grain boundaries become more difficult" .^ In addition, martensite plates often impede crack propagation when they are formed at the crack tips"^. These factors probably contribute to the slightly higher strength values of martensitic alloy B compared with pseudoelastic alloy A. Alloy C was in the martensitic state and showed much higher values for strength than alloys A and B. Such high values for alloy C over alloy B cannot easily be explained since they are both in the martensitic state. The only obvious difference between alloy C and alloy B is in the chemical composition. Hence the difference in mechanical properties is probably due to some factor which is somehow related to chemical composition. The same factor probably explains the difference in properties of alloy C over both the alloys A and B, since alloys A and B have fairly 75 similar chemical compositions. The possibility of having two different types of martensites with different mechanical properties was tested first According to most of the previous work 3412 on CuAINi , 0-phase would transform into only one type of martensite (7 / ) on cooling, although this thermal martensite can be transformed into several deformation martensites (0 , ' , j 3 , ' ' and a, ') upon application of stress. However, according to the 24 study of Garwood and Hull on Cu-12.8Al-7.7Ni (wt%) alloys, 0-phase would transform into two different types of martensites (0 , ' or 1 /) depending upon the prior thermal history of the specimens. 7 ( is of 2H type while 0 , ' is of 18Ri type in Ramsdel 12 notation . The value for the lattice parameter c for 0 1 ' is much higher than that of 1(. However, the values for other lattice parameters a and b are almost same in both 12 cases . X-ray diffraction studies were made on several coarse grain specimens of alloys B and C which were in the martensitic state and these results were compared with the X-ray diffraction traces of 0 , ' and 1( given in the literature^3. The X-ray diffraction traces of these two martensites seem to be fairly similar although there are a few differences. The attempts to classify the martensites formed in alloy B and alloy C as of two different types were not successful. When several specimens were tested, no consistent difference in patterns between the specimens of alloy B and alloy C could be found. Hence using the possibility of having two different martensites to explain the differences in mechanical properties between alloys B and C seems to be not valid. Oxygen segregation at the grain boundaries was considered as the second possibility to explain the differences in mechanical properties in the present alloys. This was based upon recent work by Lee and Wayman" .^ In their tests, they performed 76 Auger electron analysis (AES) on fracture surfaces irnmediately after fracture in both 0- phase dopant-free CuAINi specimens and 0-phase Ti-added CuAINi specimens. Large quantities of oxygen were detected at the grain boundaries but not in the grain interiors in the first case while no segregation was found in the second. They concluded that oxygen segregation must have taken place in these alloys and that was considered responsible for the predominantly intergranular fracture mode observed. In Ti-added alloys, they found Ti-peaks and oxygen-peaks in the same regions and suggested that Ti acts as an oxygen getter. They further concluded that oxygen was no longer segregated at grain boundaries in these alloys, thus changing the fracture mode from intergranular to transgranular. Based on the above suggestion, experiments were carried out on the present alloys, to study the oxygen distribution. A flat surface adjacent to the fracture surface was subjected to Secondary Ion Mass Spectroscopy (SIMS) testing. SIMS with a detection limit of the order of 10"5 % is much more sensitive than AES which has a 64 detection limit of 0.1% , although the oxygen segregation at the grain boundaries may be limited to just 1-2 atom layers and so may be difficult to detect Tests were performed with Ga+ as the primary ion, at 6kV energy and with 80nA target current Experiments were performed on both alloys A and C. Fig. 30(a) and (b) show a micrograph and an Cr map of the same region in a specimen of alloy A. Fig. 31(a) and (b) give the Cr and Al + line traces of a similar region in alloy A. These results appear to suggest that oxygen segregation at grain boundaries has taken place in alloy A. No such results were observed in alloy C. These results further suggest that oxygen segregation is observed mostly along the grain boundaries which were present before the final recrystallization rather than at the grain boundaries of the final fine grain structure. Oxygen diffusion to a) xl .OOO Figure 30. a) Micrograph and b) O - map of the same region in a specimen of alloy A as obtained on the SIMS. 78 79 grain boundaries must have taken place at the solution treatment temperature of 800° C. Since the final recrystallization time is very short (3-5s), there is probably not enough time for the oxygen to diffuse to grain boundaries. However, the treatments given before the final recrystallization involved heating the specimens for much longer times (15-900s). Hence most of the oxygen diffusion must have taken place during this time. This probably explains why oxygen segregation was observed mostly along some previous grain boundaries rather than at the grain boundaries of the final fine grain structure. This is not contrary to the observations of Lee and Wayman^ since they used longer solution treatment times (300s to 21.6ks) in their work so that there was ample time for oxygen diffusion. Fig. 31(b) shows that Ai* peaks were also present in the same grain boundary regions where 0~ peaks were present This suggests that oxygen must have entered the alloy primarily during the casting process in a way associated with Al, although steps were taken to prevent this using an inert atmosphere. Presumably a high amount of oxygen enters the alloy when the Al content is high and this is probably the case in alloy A since it has a relatively high amount of Al (14.4 wt%). Because of the same reason, only a small amount of oxygen must have been present in alloy C which has slightly lower amount of. Al (12.8 wt%). Since the overall oxygen content is low in alloy C, even after possible segregation, the oxygen at grain boundaries is probably so low that it could not be detected. It was clearly shown in sections 3.3.3 and 3.5.2 that in specimens of alloy A, the fracture mode was almost completely intergranular even at fine grain sizes. It was also shown that for specimens of alloy C, even at relatively coarse grain sizes, intergranular type fracture was not commonly observed. As suggested by Lee and Wayman^, it seems likely that oxygen segregation at grain boundaries must have further 80 weakened the already vulnerable grain boundaries caused by high stress concentrations. This must be the cause for the almost complete intergranular fracture mode observed in alloys A and B. In alloy C, either there is no oxygen segregated at grain boundaries or even if there is, segregated oxygen is so low that it could not be detected. It may be reasonable to assume that in such cases even if there is some oxygen present at grain boundaries it may not be very deleterious. This probably explains the lower values obtained for mechanical properties such as tensile strength and fatigue strength in alloys A and B when compared to alloy C. The properties obtained for alloys A and B are also inferior to the properties obtained by Sure and Brown15 with similar alloys which had similar Al content when grain sizes are comparable. The reason for this is probably the use of Ti in their work which possibly acted as an oxygen getter as suggested by Lee and Wayman1^. 4. C O N C L U S I O N S 1) A thermomechanical technique has been developed to produce specimens of ]3-CuAINi with grain sizes as low as Sum. The technique involved two sequential warm working and recrystallization steps in a suitable alloy. Even at fine grain sizes, the specimens produced were of single 0 - phase type without any second phase. 2) Of and increased with decreasing grain size according to a Hall-Petch type relationship down to a grain size of 5Mm. o^.&s high as 1,200 MPa and e^ . as high as 10% could be obtained in a suitable alloy. 3) Grain refinement did not affect the strain-memory properties significantly. Approximately 86% recovery could be obtained for an initial applied strain of 5% at a grain size of around of 10M m. 4) Grain refinement increased the fatigue life considerably possibly due to increased and a ductile fracture mode. A fatigue life of 275,000 cycles could be obtained for an applied stress of 330 MPa and a steady state strain of 0.6%. 5) The major fracture mode changed either from intergranular to transgranular or from transgranular to microvoid coalescence with decreasing grain size. This further explains the better mechanical properties at fine grain sizes. 6) Oxygen segregation at grain boundaries seems to be the factor that caused different 81 82 alloys to show different strength properties by being a major factor in causing intergranular-type brittle fracture. REFERENCES 1. Lieberman, D.S., Phase Transformations, A.S.M. Seminar, Metals Park, Ohio, (1970) 1-58. 2. Warlimont, H., Delaey, L., Krishnan, R.V. and Tas, H., /. Mat. Sci., 9 (1974) 1545-1555. 3. Saburi, T. and Nenno, S., Solid -»Solid Phase Transformations, Ed. by Aaronson, H.I., Laughlin, D.E., Sekerka, R.F. and Wayman, C M . , Metall. Soc. A.I.M.E., Warrendale, Pa., (1982) 1455-1479. 4. Delaey, L., Krishnan, R.V., Tas, H. and Warlimont, H., /. Mat. Sci., 9 (1974) 1524. 5. Wayman, C M . , Journal of Metals, 32 (1980) 129. 6. Schetky, L.McD., Scientific American, 241 (1979) 74. 7. Perkins, J. (Ed.), Shape Memory Effects in Alloys, Plenum Press, New York, 1975. 8. Perkins, J., Metals Forum, 4 (1981) 153. 9. Obrzut, J.J., Iron Age, 111 (1979) 81. 10. Lee, J.S. and Wayman, C M . , Trans. J.I.M., 11 (1986) 584-591. 11. Duerig, T.W., Albrecht, J. and Gessinger, G.H., Journal of Metals, 34 (1982) 14-20. 12. Otsuka, K., Sakamoto, H. and Shimizu, K., Acta. Met., 11 (1979) 585. 13. Khan, A.Q. and Delaey. L., Z. Metalkunde, 60 (1969) 949. 14. Khan, A.Q. and Delaey, L., Scripta. Met., 4 (1970) 981. 15. Sure, G.N. and Brown, L.C., Met. Trans., 15A (1984) 1613. 16. Kasberg, A.H. and Mack, D.T.. Trans. A. I. M. £., 191 (1951) 903. 17. Brown, L.C., Met. Trans., 13A (1982) 25. 18. White, S.M., Cook, J.M. and Stobbs, W.M., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-779. 19. Dvorak, I. and Hawbolt, E.B., Met. Trans., 6A (1975) 95. 20. Alexander, W.O., /. Inst. Met., 63 (1938) 163. 21. Mack, D.J., Metals Technology, 14 (1947) Technical Publication No. 2242. 22. Klier, E.P. and Grymko, S.M., Met. Trans., 185 (1949) 611. 23. Haynes, R., /. Inst. Met., 83 (1954-1955) 105. 83 84 24. Hull, D. and Garwood, R.D., /. Inst. Met., 86 (1957-1958) 485. 25. Rachinger, W.A., / . Aust. Inst. Met., 5 (1960) 114. 26. Oishi, K. and Brown, L.C., Met. Trans., 2 (1971) 1971. 27. Wood, J.V., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-755. 28. Sakamoto, H., Trans. J. I. M., 24 (1983) 665. 29. Yasunaga, M., Funatsu, Y., Kojima, S., Otsuka, K., Suzuki, T., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-604. 30. Sugimoto, K., Kamei, K., Matsumoto, H., Komatsu, S., Akamatsu, K., Sugimoto, T., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-761. 31. Ikai, Y., Murakami, K., Mishima, K., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-785. 32. Miyazaki, S., Kawai, T., Otsuka, K., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-813. 33. Delaey, L., Chandrasekaran, M., Andrade, M. and Van Humbeeck, J., Proc. of an Int'l Conf. on Solid -* Solid Phase Transformations, Ed. by Aaronson, H.I., Laughlin, D.E., Sekerka, R.F. and Wayman, C M . , Metall. Soc. A.I.M.E. Warrendale, Pa., (1981) 1429-1453. 34. Nishiyama, Z., Martensitic Transformations, Academic Press, New York, (1978) 74-134. 35. Matsumoto, H., Tanaka, T., Irizawa, T., Sugimoto, K. and Kamei, K., / . Japan Copper Research Association, 14 (1976) 92. 36. Kamei, K., Matsumoto, H., Sugimoto, K. and Irizawa, T., /. Japan Copper Research Association, 19 (1981) 201. 37. Jassen, J., Willems, F., Verelst, B., Maertens, J. and Delaey, L., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-809. 38. Oshima, R., Tanimoto, M., Oka, T., Fujita, F.E., Hanadate, Y., Hamada, T. and Miyagi, M., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-749. 39. Spencer, C.W. and Mack, D.K., Proc. Symp. Met. Soc. AIME, Ed. by Zackay, V.F. and Aaronson, H.I., Philadelphia, (1960) 549-606. 40. Warlimont, H. and Delaey, L., Prog. Matls. Sci., 18 (1974) 1. 41. McQueen, H.J. and Jonas, J.J., Plastic Deformation of Materials, Ed. by Arsenault, R.J., Academic Press, New York, (1975) 393. 85 42. Cotterill, P. and Mould, P.R., Recrystallization and Grain Growth in Metals, Surrey University Press, (1976) 180-249, 266-325. 43. Doherty, R.D., First Riso International Symposium on Metallurgy and Materials Science, Ed. by Hansen, N., Jones, A.R. and Leffers, T., Roskilde, Denmark, (1980) 57-69. 44. Byrne, J.G., Recovery, Recrystallization and Grain Growth, Macmillan, New York, (1965) 60-92. 45. Cahn, R.W., Proc. Conf. "Recrystallization, Grain Growth and Textures", Ed. by Margolin, H., ASM, Metals Park, Ohio, (1965) 99-128. 46. 1980 Annual Book of ASTM Standards, Part II, Metallography; Nondestructive Testing, American Society for Testing and Materials, Philadelphia, 1980. 47. Hillert, M., Acta. Met., 13 (1965) 227. 48. Beck, P.A., Adv. Phys., 3 (1954) 245. 49. Beck, P.A., Towers, J. and Manley, W.O., Trans. Met. Soc. AIME, 175 (1951) 634. 50. Miyazaki, S., Otsuka, K., Sakamoto, H. and Shimizu, Trans. J. I. M., 4 (1981) 224. 51. Armstrong, R.W., Metall. Trans., 1 (1970) 1169-1176. 52. Cottrell, A.H., The Relation Between Structures and Mechanical Properties of Metals, Vol. II, Her Majesty's Stationary Office, London, (1963) 455. 53. Arneodo, W. and Ahlers, M., Acta. Met., 22 (1974) 1475. 54. Nakamishi, N., Mori, T., Miura, S., Murakami, Y. and Kochi, S., Phil. Mag., 18 (1973) 277. 55. Patel, J.R. and Cohen, M., Acta. Met., 1 (1953) 531. 56. Melton, K.N. and Mercier, O., Acta. Met., 29 (1981) 393. 57. Eissenwasser, J.D. and Brown, L.C., Met. Trans., 3 (1972) 1359. 58. Krishnan, R.V., Delaey, L., Tas, H. and Warlimont, H., / . Mat. Sci., 9 (1974) 1536-1544. 59. Enami, K., Takimoto, V. and Nenno, S., Proc. Conf. ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven, Belgium, (1982) C4-773. 60. Oshima, R. and Yoshida, Y., Proc. Conf ICOMAT-82, Ed. by Delaey, L. and Chandrasekaran, M., Leuven,. Belgium, (1982) C4-803. 61. Thompson, A.W. and Backofen, W.A., Acta. Met., 19 (1971) 597-606. 62. Borik, F., Justusson, W.M.' and Zackay, V.F., Trans. A . S. M., 56 (1948) 327. 86 63. Oishi, K., M.A.Sc. Thesis, The University of British Columbia, Vancouver, Canada, July 1970. 64. Morabito, J.M. and Lewis, R.K., Methods of Surface Analysis, Ed. by Czanderna, A.W., Elsevier Scientific Publishing Co., Amsterdam, 1975. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078496/manifest

Comment

Related Items