INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps.
Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9” black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.
Bell & Howell Information and Leaming 300 North Zeeb Road, Ann Artxjr, Ml 48106-1346 USA 800-521-0600 UMI
FUSION BOUNDARY MICROSTRUCTURE EVOLUTION IN ALUMINUM
ALLOYS
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
B y
Anastasios Dimitrios Kostrivas. Dipl. Eng.. M.S.
*****
The Ohio State University
2000
Dissertation Committee: Approved by
Dr. John C. Lippold. Adviser
Dr. Michael J. Mills Adviser Dr. William A. Baeslack III Welding Engineering Program UMI Number; 9994890
UMI*
UMI Microform 9994890 Copyright 2001 by Bell & Howell Information and Leaming Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 ABSTRACT
Aluminum alloys c-xliibit a variety ot'mierostruciurcs within the fusion zone adjacent to the fusion boundary. Under conventional weld solidification conditions, epitaxial nucléation occurs off grains in the heat affected zone (H.A.Z) and solidification
proceeds along preferred growth directions. In some aluminum alloys, such as those containing Li and Zr. a non-dendritic equiaxed grain zone (EQZ) has been observed along the fusion boundary that does not appear to nucleate epitaxially from the HAZ
substrate. The EQZ has been the subject of considerable study because of its
susceptibility to cracking during initial fabrication and repair. The motivation of this
investigation was to develop a technique that would allow the nature and evolution of the
fusion boundary to be studied under controlled thermal conditions and to use this
technique to study fusion boundary' microstructure evolution in a number of alloy
systems.
.A melting technique was developed to simulate the fusion boundary of aluminum
alloys using the Gleeble" thermal simulator. Using a steel sleeve to contain the
aluminum, samples were heated to incremental temperatures above the solidus
temperature of a number of alloys. In alloy 2195. a 4wi%Cu-l\vt%Li alloy, an EQZ
could be formed by heating in the temperature range from approximately 630 to 640 °C. At temperatures above 640 °C. soliditication occurred by the normal epitaxial nucléation and growth mechanism. Fusion boundary behavior was also studied in alloys
5454-H34. 6061-T6. and 2219-T8. Details of the technique and its effectiveness for performing controlled melting experiments at incremental temperatures above the solidus are described.
.Additionally, experimental alloy compositions were produced by making bead on plate welds using an alloy 5454-H32 base metal and 5025 or 5087 filler metals. These tiller metals contain zirconium and scandium additions, respectively, and were expected to influence nucléation and growth behavior. Both as-welded and welded/heat treated
(540°C and 300°C) substrates were tested by melting simulation, resulting in dendritic and EQZ structures depending on composition and substrate condition.
Orientation imaging microscopy (OIM'^') was employed to study the ciy stallographic character of the microstaictures produced and to verity the mechanism responsible for EQZ formation. OIM^^' proved that grains within the EQZ have random orientation. Specific non-dendritic equiaxed grains, located next to the fusion boundary,
were separated from the adjacent grains in the partially melted zone by high angle
boundaries. In all other cases, where the simulated microstructures were dendritic in
nature, it was shown that epitaxy was the dominant mode of nucléation. In this case
showed a clear orientation relationship between dendritic solidification and the
PMZ. The lack of any preferred crystallographic orientation relationship in the EQZ
supports a theor.- proposed by Lippold et al that the EQZ is the result of heterogeneous
nucléation within the weld unmixed zone.
Ill EDS analysis of the 2195 on STEM revealed particles with ternary composition consisted of Zr. Cu and Al and a tetragonal type crystallographic lattice. In general, for
Zr and Sc bearing intermetallics. it was difficult to detect them using selected area diffraction due to their submicron size and similarity of their unit cell structure (LD lattice) with that of the aluminum matrix (fee). Microdiffraction line scans on EQZ grains in the alloy 2195 showed very good agreement between the measured Cu composition within the interior of the non-dendritic grains and the corresponding value the Scheil equation predicts for the first solid to form upon solidification for a binar\- .A.l-Cu alloy with identical Cu composition.
It is also important to note that a threshold value of 0.05wt% Zr was found below which no EQZ was formed irrespective of the substrate condition examined. This value can be compared against the 0.14 wt%. which is the Zr content of the alloy 2195 that exhibits EQZ formation.
In the context of the alloys, compositions and substrate conditions examined a mechanistic model for EQZ zone formation is proposed. This model can be helpful in adjusting base metal compositions and/or substrate conditions to control fusion boundary microstructure.
IV DEDICATION
To my parents
Ariiiiiipri and KaAi.ippor] for their unconditional love and support
and
to my brothers
Etppvri. ©oôcopÛKTi and navayicbiri
who I missed so much ACKNOWLEDGMENTS
I wish to thank my adviser. John Lippold. for his intellectual support, encouragement and enthusiasm that made this dissertation possible, and tor his patience in correcting both my stylistic and scientific errors. You afforded me the opportunity to learn so much. Your friendship and support has meant a lot to me.
1 would like to express my appreciation to Michael Mills, for co-advising this work. Gratitude and appreciation is also expressed to Perena Gouma for providing invaluable help with TEM work. I am grateful to William Baeslack for serving on my committee and reviewing my manuscript.
1 need to express my sincere thanks to the members of the Welding and Joining
Metallurgy Group for their friendship and cooperative atmosphere and Arturo Sanabria for his assistance in producing quality welds.
I would also like to thank Alcan International and Lincoln Electric for donating the base and filler metals used in this investigation.
Special thanks go to Oak Ridge National Labs for granting me the opportunity to visit their facilities and perform OIM analysis and to Ed fCenik for his assistance.
I am indebted to Gregory Rohrer at Carnegie Mellon University for making possible for me to use their OIM equipment and to TexSEM Labs for providing the OIM analvsis software.
VI VITA
July 20, 1970 ...... Bom - Kalamata. Messinia. Greece
1993 ...... Diploma Engineer. Metallurgical Engineering. National Technical University of Athens
1996 ...... VI.Sc.in Welding and Joining Technology. Gran field University. UK
1996-present ...... Graduate Research .Associate. The Ohio State Universitv
PUBLICATIONS
Research Publications
1. Kostrivas and J.C. Lippold; ".A Method for Studying Weld Fusion Boundary Vlicrostructure Evolution in .Aluminum AWoys".WeUiing Journal, v.79 (1), January 2000. pp. l-8s.
2. Kostrivas and J.C. Lippold: " Weldability of Li-Bearing Aluminum Alloys - Review Paper". International Materials Review, v.44 (6). December 1999. pp.217-237.
3. Kostrivas and J.C. Lippold: "The Circular Patch Test: A Review". Welding Technology {\0). 1998. pp. 3- 23. (published in Greek).
4. Kostrivas and J.C. Lippold: "Metallurgical Characteristics of Al-Cu-Li .Alloy .AF/C489 Varestraint Weldability Test Samples". Final Report Submitted to U.S. A ir Force. August 15. 1996.
5. Kostrivas and J. Norrish: "Orbital Welding of Mild Steel Pipelines with Twin Wire GMAW Process". M.Sc Thesis. Craniield University. Cranfield. UK. June 1996.
VII 6. A. Kostrivas and G. Papadimitriou: "Welding of 5XXX series Aluminum Alloys with GTAW Process". Eng. Dipl. Thesis. NTUA. .Athens. Greece. October 1993.
FIELDS OF STUDY
Major Field: Welding Engineering
Mil TABLE OF CONTENTS
Page
A bstract...... ü
D edication ...... v
Acknowledgements ...... vi
V ita...... vii
List ot'Tables ...... xii
List o f Figures ...... xiv
Nomenclature ...... xxv
Chapters
1. Introduction ...... 1
2. Literature Review...... 4
2.1 The Need for Welding Li-Bearing .Aluminum Alloys ...... 4 2.2 Classification and Physical Metallurgy of Al-Li-X Alloys ...... 5 2.2.1 Classification ...... 7
2.2.2 Physical Metallurgy ...... 8
2.3 Principles of Solidification ...... 20
2.3.1 Driving Force for Solidification and Homogeneous Nucléation ...... 20 2.3.2 Heterogeneous Nucléation ...... 23 2.4 Microstructure Evolution During Welding ...... 27 2.4.1 Fusion Zone M icrostructure ...... 27 2.4.2 Scheil Partitioning ...... 32
ix 2.4.3 Heat Affected Zone Microstructure ...... 35 2.5 Weld Cracking Susceptibility' ...... 36 2.5.1 Weld Metai Solidification Cracking ...... 36 2.5.2 HAZ Liquation Cracking ...... 49 2.6 Previous Investigation and Hypotheses ...... 52 2.7 Equiaxed Grain Zone Formation and Associated Fusion Boundary Cracking in .Al-Li-X Alloys ...... 54 2.8 Summary of Literature ...... 57
3. Objectives...... 60
4. Experimental .Approach-Procedures ...... 61
4.1 M aterials...... 61 4.2 Welding Procedures ...... 62 4.3 Pickling ...... 64 4.4 Heat Treatments...... 64 4.5 Melting Simulation ...... 66 4.5.1 Gleeble” 1500 Characteristics...... 67 4.5.2 Nil Strength Temperature Measurement ...... 69 4.5.3 Controlled Melting Experiments ...... 70 4.6 Characterization Techniques ...... 73 4.6.1 Metallographic Preparation ...... 73 4.6.2 Base Metal Dilution Measurement ...... 74 4.6.3 Microstructure Characterization Techniques ...... 75 4.6.4 (Scanning)-Transmission Electron Microscopy ...... 76 4.6.5 Orientation Imaging Microscopy ...... 77
5. Results...... 82
5.1 Microstructures Produced with Controlled Melting on Gleeble® 83 5.1.1 Microstructures in Commercial .Aluminum Alloys - .As Received Substrate...... 84 5.1.1.1 Fusion Boundary Microstructures in Heat-Treated .Aluminum Alloy 2195 ...... 95 5.1.2 Al-Mg-Zr System ...... 102 5.1.2.1 .As-Welded Substrate Condition ...... 102 5.1.2.2 Welded and Heat -Treated Substrate Conditions (540°C for4Hrs) ...... 106 5.1.2.3 Welded and Heat-Treated Substrate Conditions (300°C for 5 hrs) ...... 108 5.1.3 Al-Mg-Sc System ...... 110 5.1.3.1 .As-Welded Substrate Condition ...... I l l 5.1.3.2 Welded and Heat -Treated Substrate Conditions (540°C for 4 hrs) ...... 114 5.1.3.3 Welded and Heat -Treated Substrate Conditions 118 (300°C for 5 hrs) ...... 5.1.4 Cross Welds ...... 121 5.1.5 Summary on Microstructures ...... 123 5.2 Element Distribution Profiles and Composition Analysis ...... 124 5.2.1 Compositional Effect o f the Steel Sleeve/Crucible ...... 142 5.2.2 Summary on Composition .Analysis ...... 143 5.3 .Analytical Electron Microscopy Results ...... 144 5.3.1 .Analysis of Electron Backscattered Diffraction Patterns (O IM ™ )...... 144 5.3.2 TEM and High Resolution I'EM .Analysis ...... 161 5.3.3 Summary of .Analytical Electron Microscopy Results ...... 171
6. Discussion ...... 172
6.1 Development Characteristics of the Melting Simulation Technique ... 172 6.1.1 Nil Strength Temperature ...... 172 6.1.2 Fusion Zone Simulation ...... 173 6.2 Simulation Aided Study of Fusion Boundary Microstructure 176 Evolution ...... 6.2.1 Fusion Boundary Microstructure Evolution in Commercial .Aluminum .Alloys ...... 177 6.2.2 Fusion Boundary Microstructure Evolution in .Aluminum Alloy 2195 ...... !...... 179 6.2.2.1 Heterogeneous Nucléation Mechanism and Thermal Effects on EQZ Fomiation ...... 179 6.2.2.2 Substrate Effects - OIM™ .Analysis ...... 186 6.2.3 Fusion Boundary Microstructure Evolution in Al-Mg-Zr alloys - Substrate Effects ...... 188 6.2.4 Fusion Boundary Microstructure Evolution in Al-Mg-Sc .Alloys-Substrate Effects ...... 189 6.2.5 Heterogeneous Nucléation - Nucleus Size Effect ...... 193 6.3 EQZ versus Dendritic Microstructure Formation - Composition Effects.... 196 6.4 Precipitation Analysis Using STEM ...... 197 6.5 Simulated versus Weld Fusion Boundary Microstmctures ...... 198 6.6 Practical Implications ...... 200
7. Conclusions ...... 208
8. Future Work...... 211
List of References ...... 213
XI LIST OF TABLES
Tabic Pasc
2.1 Cr\ ogenic properties of the high strength alloy 2090 ...... 5
2.2 Nominal compositions (vvt %) and typical mechanical properties of commercial .41-Li alloys ...... 9
2.3 Precipitation in Li-bearing aluminum alloys ...... 12
2.4 Solidification temperature range and fraction eutectic at maximum
cracking for binary alloy systems ...... 41
2.5 Solidification data for commercial .Al-Li-X alloys based on binar>' phase diagrams ...... 42
2.6 The solidification cracking temperature range (SCTR) determined using the Varestraint test ...... 44
2.7 Spot Varestraint test results ...... 51
4.1 Composition (%vvt.) o f some commercial aluminum alloys (base and filler metals) ...... 61
4.2 Welding parameters and resulting BMD range for welds used in the current investigation ...... 63
4.3 Summary of the different materials and heat treatment schemes used in the current investigation ...... 65
5.1 Nil strength temperature (NST) and nominal melting temperature for various commercial aluminum- alloys ...... 82
XU 5.2 NST values determined for the experimental compositions produced from welds made using filler metals 5087 and 5025 and 5454-H32 substrate...... 83
5.3 Effect of thermal conditions on EQZ formation and epitaxial nucléation for various commercial aluminum alloys melted on Gleeble ...... 94
5.4 EDS spot-analysis results produced in selected fusion zone microstructures ...... 141
6.1 Summary of the combined effects composition and substrate exert on the EQZ formation ...... 198
Xlll LIST OF FIGURES
Figure Page
2.1 Elastic modulus of Al-Mg-Li alloys versus lithium content ...... 6
2.2 Yield strength and YS/Density ratio for various Al alloys ...... 7
2.3 Ternary and quatemar\' aluminum alloy systems containing lithium, showing commercial alloy designations and primary strengthening precipitates ...... 8
2.4 Binary phase diagram for the Al-Li system ...... 10
2.5 The lower part of the Al-Li binary system ...... 11
2.6 Binar\ phase diagram for the .\1-Zr system ...... 14
2.7 Binary phase diagram for the .A.1-SC system ...... 19
2.8 Free energy difference vs. temperature for solid and liquid phases ...... 21
2.9 Free energy change vs. sphere radius r for homogeneous nucléation 23
2.10 Schematic showing heterogeneous nucléation of spherical cap on flat substrate surface...... 25
2.11 Free energy change vs. spherical cluster radius r for both homogeneous and heterogeneous nucléation ...... 27
2.12 Constitutional supercooling during alloy solidification ...... 29
2.13 Mode of solidification versus amount of constitutional supercooling 30
2.14 Representative fusion zone microstructure of autogenous welds ...... 31
XIV 2.15 Representative fusion zone microstructure of VP PA welds in 2195 using 2319 filler metal ...... 32 2.16 Copper concentration profile estimated using the Scheil equation for an Al-2wt% Cu binary- alloy ...... 34
2.17 Electon microprobe analysis of Cu across primary solidification boundaries in an .Al-2vvt%Cu CTA weld ...... 34
2.18 Weld metal cracking susceptibility as a function of composition in a simple binary alloy system containing a eutectic reaction ...... 37
2.19 Cracking susceptibility versus normalized composition (Co/Cma\) for the .A.1-CU. .Al-Mg. .Al-Li and .Al-Si binary system s ...... 39
2.20 Cracking susceptibility versus normalized composition (Co/Cma\) for the .Al-Cu. Al-Mg and .Al-Li binary system s ...... 40
2.21 Cracking susceptibility versus normalized composition (C,/Cma\) for the .*\1-Cu and Al-Li binarv^ svstems ...... 40 n in Representative weld solidification cracking in Varestraint samples tested at 3% strain...... 45
2.23 Cracking susceptibility of autogenous and filler metal welds on 01441- T8 aluminum allov sheets usina the Houldcroft test ...... 47
2.24 Total crack length contour map ...... 48
2.25 Schematic illustration of suggested Heterogeneous Nucléation Mechanism for the formation of the Equiaxed Grain Zone ...... 53
2.26 Fusion boundary region in Varestraint sample of alloy 2195 ...... 55
2.27 Fusion boundary region in weld of alloy 2195 using 2319 filler metal 55
2.28 Fusion boundary region in Varestraint sample of alloy 8090 ...... 56
2.29 Solidification crack along EQZ close to fusion boundary. Alloy AF/C489-Longitundinal Varestraint test ...... 56
4.1 Schematic showing weldment location from which simulation samples removed and a cross weld deposited across a previous pass ...... 63
XV 4.2 View of the atmosphere tank and gripping assembly o fth e Gleeble® 1500 ...... 68
4.3 Schematic showing sample configuration for Gleeble® thermal simulation and the associated axial thermal gradient ...... 68
4.4 An axial slice of cylindrical sample heated by the Gleeble®. Tliermocouple location and isothermal planes are shown ...... 70
4.5 Experimental configuration for melting aluminum samples on the Gleeble® ...... 72
4.6 Typical thermal cycles applied (programmed) during controlled melting of aluminum samples ...... 72
4.7 Illustration showing the BMD calculation ...... 75
4.8 Schematic showing specimen arrangement and pattern generation due to backscattered electron diffraction in a bulk sample ...... 78
4.9 Illustration depicting the various steps followed to collect and analyze EBSP signal generated on the surface of a bulk sample ...... 79
4.10 An indexed FCC electron backscatter pattern ...... 80
5.1 .Alloy 2195-T8. Base metal microstructure ...... 85
5.2 Weld fusion zone simulation of alloy 2195-T8. Microstructure corresponds to specimen heated to 625 °C ...... 85
5.3 Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. A) 630 °C. B) 660 °C ...... 86
5.4 Simulated fusion boundary region in alloy 6061-T6 produced using the Gleeble®. A) 645 °C. B) 620 87
5.5 Simulated fusion boundary region in alloy 2219-T8 produced using the Gleeble®. A) 655 °C. B) 587 °C ...... Z ...... 88
5.6 Simulated fusion boundary region in alloy 5454-H34 produced using tlie Gleeble®. A) 623 °C.B) 610 °C ...... 7...... 89
5.7 Hardness traverse across the fusion boundary for alloy 2219-T8. Specimen was heated at peak temperature of 640°C ...... 91
XVI 5.8 Hardness traverse across the fusion boundary for alloy 6061-T6. Specimen was heated at peak temperature of 645°C ...... 91
5.9 Hardness traverse across the fusion boundary for alloy 5454-H34. Specimen was heated at peak temperature of 623°C ...... 92
5.10 Hardness traverse across the fusion boundary for alloy 2195-T8. Specimens were heated at peak temperatures of 630°C and 640°C respectively ...... 92
5.11 Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen heated to 640°C peak temperature ...... 93
5.12 Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 600° for 15 min prior to melting at 630°C ...... 96
5.13 .A.lloy 2195. Heat-Treated/Quenched substrate condition (600°C. 4h) ...... 98
5.14 Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 600° for 4 hrs prior to melting at 632°C ...... 99
5.15 Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 625°C for 4 hrs prior to melting at 640°C ...... 100
5.16 Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 625°C for 4 hrs prior to melting at 640°C ...... 101
5.17 .A.S welded microstructure produced using cold wire GTAW and 5454- H32 base metal with 5087 filler metal alloy at 50% BMD ...... 103
5.18 Simulated microstructure produced on the Gleeble® using 5454-H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 590°C peak temperature ...... 103
5.19 Simulated fusion boundary region produced on the Gleeble® using 5454-H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 605°C peak temperature ...... 104
xvn 5.20 Simulated microstructure produced on the Gleeble® using 5454-H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen was heated to 625°C peak temperature ...... 104
5.21 Simulated fusion boundary region produced on the Gleeble® using 5454-H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen was heated to 631°C peak temperature ...... 105
5.22 Simulated fusion boundary region produced on the Gleeble® using 5454-H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen was heated to 660°C peak temperature ...... 105
5.23 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen was heated to 605 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after w elding ...... 106
5.24 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen was heated to 616 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding ...... 107
5.25 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen was heated to 628 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding ...... 107
5.26 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen heated to 660 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after w elding ...... 108
5.27 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen was heated to 632°C. Prior to simulation the sample was aged at 300°C for 5h after welding ...... 109
5.28 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen was heated to 650 °C. Prior to simulation the sample was solution heat treated at 300°C for 5h after welding ...... 110
5.29 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Specimen heated to 580 Ill
XVIII 5.30 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Specimen heated to 600 °C ...... 113
5.31 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Specimen heated to 605 °C ...... 112
5.32 Weld fusion zone simulation produced on the Gleeble® using 5454- H32.G025 at 50% BMD in the as welded substrate condition. Specimen heated to 015 “C ...... 113
5.33 Weld fusion zone simulation produced on the Gleeble® using 5454- H32.G087 at 50% BMD. Specimen heated to 633 °C ...... 113
5.34 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Specimen heated to 644 °C ...... 114
5.35 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Specimen was heated to 582 T ...... 115
5.36 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Specimen was heated to 6 1 0 °C ...... r ...... 115
5.37 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Specimen was heated to628°C ...... T...... 116
5.38 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Specimen was heated to633°C ...... 117
5.39 Higher magnification of the EQZ shown in Figure 5.39 ...... 117
XIX 5.40 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Specimen was heated to 6 5 5 ° C ...... r ...... 118
5.41 Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Specimens received a heat treatment cycle at 300° for 5 hrs prior to melting and then were specimen heated (A) to 629° C and (B) 636° C T...... 119
5.42 Hardness traverses across the fusion boundary for two Sc-containing alloys ...... 120
5.43 Cross weld fusion zone microstructure produced on 5454-H32/5087 at 50% BMD in was in as weld/heat treated substrate condition (540°C-4h) ...... 121
5.44 Cross weld fusion zone microstructure produced on 5454-H32/5025 at 50% BMD in was in as welded substrate condition ...... 122
5.45 Cross weld fusion zone microstructure produced on 5454-H32/5025 at 50% BMD in was in as welded substrate condition ...... 122
5.46 BSE photomicrograph ofthe simulated fusion boundary microstmcture in the alloy 2195-T8. Specimen was heated to 632°C on Gleeble® ...... 125
5.47 BSE photomicrograph ofthe simulated EQZ in the alloy 2195-T8. Specimen was heated to 632°C on the Gleeble® ...... 125
5.48 X-Ray maps of the simulated EQZ in the alloy 2195-T8. Specimen was heated to 632°C on the Gleeble® ...... 126
5.49 BSE photomicrographs of the simulated EQZ in the alloy 2195-T8. Specimen was heated to 635°C on the Gleeble® ...... 127
5.50 BSE photomicrographs of the simulated dendritic structure next to the fusion boundary in alloy 2195-T8. Specimen was heated to 650°C on the Gleeble® ...... 128
5.51 BSE photomicrographs of the simulated dendritic structure in the alloy 2195-T8. Specimen was heated to 650°C on the Gleeble® ...... 128
XX 5.52 BSE photomicrograph (top) and X-Ray maps (bottom) of the simulated fusion zone in the alloy 2195-T8. Specimen was heated to 650°C on the G leeble© ...... 129
5.53 BSE photomicrographs of actual weld fusion boundary microstructure in alloy 2195-T8...r...... 130
5.54 BSE photomicrographs of actual weld fusion zone microstructure in alloy 2195-T8...7 ...... 130
5.55 BSE photomicrographs and X-ray line scan of weld/solution heat-treated alloy 2195-T8 (600°C-4hrs) ' 132
5.56 BSE photomicrographs ofthe simulated fusion boundary microstructure in weld/heat treated alloy 2195-T8 (600°C-4hrs). Specimen was heated to 632'^C on the Gleeble® ...... 133
5.57 BSE photomicrograph and X-ray maps of weld/solution heat-treated alloy 2195-T8 (625°C-4hrs).....’ ...... 134
5.58 BSE photomicrographs ofthe simulated to fusion boundary microstructure in weld/heat treated alloy 2195-T8 (625°C-4hrs). Specimen was heated to 640‘^C on the Gleeble® ...... 135
5.59 BSE photomicrograph and X-ray line scan in the simulated fusion boundar\- microstructure. Specimen was alloy 2195-T8 in weld/heat treated condition (625°C-4hrs) and it was heated to 640°C on the G leeble® ...... 136
5.60 BSE photomicrographs of the simulated to fusion boundary microstructure in weld^heat treated alloy 5454/5025 (540°C-4hrs). Specimen was heated to 655°C on the Gleeble® ...... 138
5.61 BSE photomicrograph and X-Ray maps of the simulated fusion zone weld/heat treated alloy 5454/5025 (540°C-4hrs). Specimen was heated to 632°C on the Gleeble® ...... 139
5.62 BSE photomicrographs of the simulated to fusion boundary microstructure in welcEheat treated alloy 5454/5087 (540°C-4hrs). Specimen was heated to 640°C on the Gleeble® ...... 140
XXI 5.63 BSE photomicrographs of the simulated to fusion boundary microstructure in weld/heat treated alloy 5454/5025 (540°C-4hrs). Specimen was heated to 632°C on the Gleeble® and shows the fusion zone/steel sleeve interface ...... 142
5.64 SE photomicrograph showing fusion boundary' microstructure in autogenous GT.A. weld in alloy 2195-T8. As-electropolished surface ...... 146
5.65 Image quality grain map of fusion boundary microstructure in autogenous GT.A weld in alloy 2195-T8 ...... 147
5.66 Inverse pole figure plotted the entire grain population shown in Figure 5.65 ...... 148
5.67 Two SE micrographs of the simulated fusion boundaiy microstructure in weld/heat treated alloy 2195-T8 (600°C-4hrs). illustrating the distortion introduced in the image by tilting the specimen for EBSP ...... 149
5.68 Inverse pole figure plotted the entire grain population shown in Figure 5.67 ...... 150
5.69 Discreet pole figures showing individual orientations by color of small non-dendritic equiaxed grains located adjacent to big HAZ grain shown in Figures 5.67-B and 5.68 ...... 151
5.70 Image quality grain map of fusion boundary microstructure in 5454/5025 GT.A weld and inverse pole figure plotted the entire grain population ...... 153
5.71 .As electropolished surface and image quality map of fusion boundary microstructure in Gleeble® sample. Specimen was in as welded 5454/5087 substrate condition and was heated to 630 °C on the Gleeble® ...... 154
5.72 Inverse pole figure plotted the entire grain population shown in Figure 5.71 ...... 155
5.73 .As -electropolished surface and image quality map of fusion boundary microstructure in Gleeble® sample. Specimen was 5454/5087 in weld/heat treated substrate condition (540°C-4h) and was heated to 629°C on the Gleeble® ...... 157
X.XII 5.74 Inverse pole figure plotted the entire grain population shown in Figure 5.73....1 ...... 158
5.75 As-electropolished surface and image quality map of fusion boundary microstructure in Gleeble® sample. Specimen was 5454/5025 in weld/heat treated substrate condition (540°C-4h) and was heated to 629°C on the Gleeble® ...... 159
5.76 Inverse pole figure plotted the entire grain population shown in Figure 5.75 ...... 160
5.77 TEM Bright-field image illustrating the EQZ microstructure in 161 autogenous GT.A weld of the alloy 2195-T8 ......
5.78 TEM Bright-held images illustrating the EQZ microstructure and second phase decorating a non-dendritic grain ...... 162
5.79 TEM dark-field image illustrating a Zr-bearing precipitate in the EQZ microstructure in autogenous GT.A weld of the alloy 2195-T8 along with EDS map and S.AD images from matrix and precipitate ...... 163
5.80 Copper profile in non-dendritic equiaxed grains of an autogenous 2195-T8 GTA weld ...... 7 ...... 7 ...... 164
5.81 Precipitates with needle-like morphology were detected in the EQZ grain interior an autogenous 2195-T8 GT.A weld ...... 165
5.82 Fligh resolution TEM image showing precipitation within the equiaxed non-dendritic grain an autogenous 2195-T8 GTA weld ...... 166
5.83 TEM images showing a general view of the microstructure in the EQZ grain developed in an Al-Mg-Sc weld/solution heat treated sample. melted on the Gleeble® at 632°C ...... 167
5.84 TEM images showing the same structure as in Figure 5.83. SAD image taken along after long exposure time ...... 168
5.85 TEM bright field images showing secondary phase along EQZ grain boundaries. Sample was Al-Mg-Sc alloy in the weld/heat treated condition prior to melting on the Gleeble® at 632°C ...... 169
XXIII 5.86 High resolution image taken from the interior of an EQZ grain. Sample was Al-Mg-Sc alloy in the weld/heat treated condition prior to melting on the Gleeble® at 632°C ...... 170
6.1 Schematic showing the heating cycles operating on the Gleeble® as sinusoidal function of time ...... 175
6.2 Schematic highlighting fusion boundary microstructure evolution in simulation specimens of alloys 2219-T8. 5454-H34 and 6061-T6. when gross melting occurs upon heating on Gleeble® ...... 178
6.3 Lattice misfit between aluminum matrix a and a - .A.l3(Li\Zrt.\) precipitate versus Li content in a ' ...... 180
6.4 Schematic illustrating fusion boundary microstructure evolution in simulation specimens in alloy 2195-T8. upon heating on the Gleeble® within the temperature range between 630°C and 640°C ...... 182
6.5 Schematic highlighting fusion boundary microstructure evolution in simulation specimens of alloys 2195-T8. upon heating on the Gleeble® at peak temperatures exceeding 640°C ...... 182
6.6 Plot showing velocity V (cms"') versus temperature gradient G (K cm'') for .Al-3\vt%Cu and N=1000 cm''’ for four different nucléation undercoolings ...... 183
6.7 Plot of velocity V (cms'') versus temperature gradient G (K cm'') for Al- 3wt%Cu and ATn=0.75 K showing the effect of varying number N (cm''’) of nucléation sites ...... 184
6.8 Plot of velocity V (cms'') versus temperature gradient G (K cm'') for Al- 3wt%Cu for four different alloy compositions C (wt.%) at N = 1000 cm'"* and ATn=0.75 K ...... 184
6.9 Schematic view of columnar and equiaxed growth ...... 190
6.10 Schematic illustrating atom packing in a diffuse interface along with solid growth rate as a function of interface undercolling ...... 193
6.11 Plot of the strength of both grain and grain boundary as function of temperature ...... 201
6.12 Schematic illustrating the stress pattern within the EQZ and columnar grain regions during the Longitudinal Varestraint test ...... 203
XXIV NOMECLATURE
A VC...... Automatic Voltage Control 13'v I...... I3asc Metal BMD...... Base Metal Dilution BSE...... Backscattered Electrons CTE...... Coefficient of Thermal Expansion EBSP...... Electron Backscattered Diffraction Patterns EBVV...... Electron Beam Welding EDS...... Energy Dispersive Spectroscopy EQZ...... Equiaxed Non-Dendritic Grain Zone FB...... Fusion Boundary FCC...... Face-Centered Cubic FEG...... Field Emission Gun FM...... Filler Metal FSW...... Friction Stir Welding FW ...... Friction Welding GMAW ...... Gas Metal Arc Welding GTAW ...... Gas Tungsten Arc Welding HAZ...... Heat-Affected Zone IPF...... Inverse Pole Figure LBW ...... Laser Beam Welding MCD...... Maximum Crack Distance NST...... Nil Strength Temperature QIM™ ...... Orientation Imaging Microscopy PAW ...... Plasma Arc Welding PFZ...... Precipitation Free Zone PMZ...... Partially Melted Zone SCTR...... Solidification Cracking Temperature Range SDV...... Standard Deviation SE...... Secondary Electrons SEM ...... Scanning Electron Microscopy SGB...... Solidification Grain Boundary SSGB...... Solidification Subgrain Boundary STEM ...... Scanning Transmission Electron Microscopy TCL...... Total Crack Length TEM ...... Transmission Electron Microscopy VPPAW ...... Variable Polarity Plasma Arc Welding
XXV CHAPTER 1
INTRODUCTION
A number of Al-Li-X alloys have been developed during the last 20 years and much work has already addressed the vveldability of those alloys. Since they have very good mechanical properties coupled with low density they are potential candidates to substitute for other alloys in the aerospace industry. Highly critical applications such as cry ogenic tankage require production of defect free welds. As with other aluminum alloys, there are a number of weldability issues associated with these alloys, including resistance to defect formation during fabrication, mechanical property degradation, and ser\ ice performance. .As a result, weldability problems common in welds of Al-Li-X alloys such as weld solidification and heat affected zone (HAZ) liquation cracking should be studied in depth in order to control their formation.
The commercial Al-Li-X alloys are welded using a variety of processes, including arc welding (GTAW. P.AW. GMAW). high energy density welding (EBW and LBW). and solid-state welding (FW and FSW). The strength of welds in these alloys varies widely, depending on the welding process, filler metal selection, and postweld heat treatment. In general, these alloys have low joint efficiency (ratio of weld to base metal strength) in the as-welded condition and require postweld aging to achieve efficiencies substantially above 50%. Weld porosity has been a particular problem with these alloys in part due to the hygroscopic nature of Li-containing aluminum oxides. However, this problem can be controlled if proper surface preparation and cleaning procedures are used.
The -A.l-Li-X alloys tend to be more susceptible to weld solidification cracking than comparable alloys without Li additions. Basic weld solidification theory is used to explain this increase in susceptibility. Some of these alloys exhibit an unusual fusion boundaiy' cracking phenomenon that is associated with an equiaxed grain zone (EQZ) that forms in alloys containing Li and,/or Zr during fabrication and repair welding.
Despite these problems and the potential for substantial economic and human loss resulting from failures of highly critical structures such as the space shuttle hydrogen and oxygen tanks, the mechanism of formation of the EQZ is not yet fully understood and cannot be effectively controlled.
The current investigation is a fundamental study of weld fusion boundary microstructure evolution in aluminum alloys in an effort to understand EQZ formation and fusion boundary nucléation and growth phenomena. Weld fusion boundary microstructures were simulated on the Gleeble®. which has proven to be a very useful tool that consistently reproduced fiision zone microstructures. Among other commercial aluminum alloys, experimental Mg-bearing alloys with Zr- and Sc- additions were studied along with the widely used Cu- and Li-containing alloy 2195-T8. The objective here is to understand the interrelation among composition, base metal substrate, and temperature as they relate to nucléation and growth phenomena at the fusion botmdarv. Such an understanding will provide significant insight as to the operable mechanism for the EQZ formation and the parameters affecting the overall microstructure evolution next to the weld fusion boundarv. CHAPTER 2
LITERATURE REVIEW
2.1 The Need for Welding Li-Bearing Aluminum Alloys
Despite their wide use. stemming from a high strength over density ratio and their enhanced corrosion properties, aluminum - lithium alloys initially were not designed with weldability in mind with the exception of the Weldalite 049 alloys (refs. 1-2). since mechanical fastening was the primary joining process in aerospace applications.However, riveting is a tedious operation, unhealthy for the operator (ref.3) and very slow due to large volume of bolted and riveted joints. On the other hand, welded structure imposes weight savings (elimination of rivets, sealants, rubbers), less use of labor in manufacture, and reliability in performance due to uniform distribution of stresses in joint elements as compared to riveted or bolted joints (ref.4). For these alloys to have an extended use in a wider range of applications weldability should be studied and potentially more weldable alloy variants should be developed. Weldability incorporates issues such as mechanical property degradation in both the weld region and heat affected zone (HAZ). weld metal
porosity, solidification cracking susceptibility and liquation cracking susceptibility along
with resistance to corrosion coupled with alloy composition and heat treatment. 2.2 Classification and Physical Metallurgy of Al-Li-X Alloys
Lithium-bearing aluminum alloys constitute a group of high performance wrought aluminum alloys intended for use principally in aircraft and aerospace structures.
Reduced density and increased elastic modulus coupled with very good cryogenic toughness and ductility (Table 2.1) make Li-bearing aluminum alloys an attractive alternative relative to more conventional precipitation hardenable aluminum alloys (e.g.
2024,2014.2219.6061 and 7075) (refs. 1. 5-12).
Test Tem p. UTS, ksi (iVIpa) YS, ksi (M pa) Elongation. % K ic k siV m
(M Pa \fin ) "c FL LT LLTL LT L-T T-L 25 77 81.9 81.9 77.6 77.6 11.0 5.5 30.9 22.7 (565) (565) (535) (535) (34) (25) -1% -320.8 103.7 100.8 87.0 90.6 13.5 5.5 47.3 30.9 (715) (695) (600) (625) (52) (34) -269 -452 119 118 89.2 102 17.5 6.5 59.1 35.5 (820) (818) (615) (705) (65) (39)
Table 2.1. Cryogenic Properties of the High Strength .-\lloy 2090 (ref. 12).
It is known that every 1% wt. Li in aluminum decreases density (refs.l. 8-9) by
about 3% and increases elastic modulus (E) by about 6%. Figure 2.1 shows that effect
for Al-Mg-Li alloys. Pickens in his review paper (ref.8) states that for an Al-3%wt. Li alloy, structural weight savings would be in the range between 10% and 16%. Figure 2.2 illustrates the yield strength/density ratio for some commercial Al-Li-X alloys as well as the same ratio for some conventional aerospace Al alloys.
Such a weight reduction could be cost effective even when compared to composite materials considering that they have high production costs (ref. 10).
Composites are being used extensively in certain areas of high performance aircraft offering potential weight savings of about 25% compared to that achievable by conventional aluminum alloy products (ref.l 1 ). However. Li-bearing aluminum alloys appear to be competitive alternatives.
11.8
80
© a £
3 © 11.0 3 75
10.6
1 2 3
Wt. % LI
Figure 2.1. Elastic modulus of Al-Mg-Li alloys versus lithium content (ref. 8). ZL 8 I
0 U 2 0 -r6 H0*)O.r6 2020 2(WO-r8 2091. n 2095.T » 2 I‘>5-T8 2024-T3 22I4-T6 "U'S-TA
Figure 2.2. Yield strength and YS/Density ratio tor various A\ alloys (rets. 9-8. 10.12).
Classification
Despite the difficulties in production of Li-bearing aluminum alloys due to the high reactivity of Li. initial alloys were produced as early as in 1920‘s in Germany
(ref.9). Further improvements in alloy processing techniques enabled the production of a range of Li-bearing aluminum alloys (ref. 11). Today's commercial or near - commercial alloys (ref.7) can be divided into four major groups namely. Al-Mg-Li and Al-Cu-Li ternary alloys, and Al-Cu-Li-Mg and Al-Li-Cu-Mg quaternary alloys. The commercial alloys that fall within each of these groups are shown in Figure 2.3. The nominal compositions of the specific alloys are listed in Table 2.2. Minor additions of other elements such as Zr and Mn and/or Sc are used to modify microstructure and as a result the mechanical properties of Al-Li-X alloys but these effects will be discussed later.
-Aluminum - Lithium -Allovs
.Al-Cu-Li-Mii Al-Cu-Li -Al-Li-Cu-Mg
01420 2020 01421 01450 01460
Medium Copper Medium Copper High Copper Low Copper High Magnesium Low Magnesium Low Magnesium High Magnesium Medium Lithium Hiah Lithium Low Lithium Hiah Lithium T i.0 -.0 -
2091 2090 2094 8090 01430 2095 8091 2195 01440
Figure 2.3. Ternary and quaternary aluminum alloy systems containing lithium, showing commercial alloy designations and primary strengthening precipitates (refs. 7. 13-15)-
2-2-2 Physical Metallurgy
Bimirv Al-Li Allovs. The Al-Li phase diagram (ref-16) is shown in Figure 2.4. The
limited solid solubility of Li in Al (= 4%wt- at 600 C. ô'-AljLi phase (Figure 2.5) is the primary strengthening precipitate and is considered to be coherent with the matrix (ref. 17) phase since its lattice parameter is close to that of aluminum and hence results in low misfit strain and coherency with the matrix. The maximum sol vus temperature for the phase 6' is 300-350 ”C at approximately 14%wt. Li. Yield U ltim ate .Alloy Li Cu Mg Z r O th e r S tren g th S tren g th (ksi) (ksi) 01420-T6 2.0 - 5.3 0.10 0.5 Mn 41 67 01421-T6 1.9 - 5.0 0.08 0.17 Sc 52 70 01430 1.7 1.6 2.7 0.11 - 54 67 01440 2.3 1.5 0.9 0.15 56 70 01441 1.9 2.0 0.9 0.09 0.05 Be. O.llFe 72 01450 1.9 3.15 - 0.10 0.08 Ti 71 84 01460 2.1 3.1 - 0.09 0.075 Sc 67 75 8090-T8 2.5 1.0 1.0 0.10 - 70 77 8091-T8 2.6 1.9 0.9 0.12 0.1 Fe.0.1 Si 78 84 2020-T6 1.2 4.4 -- 0.5 Mn. 0.2 Cd 77 82 2090-T8 2.3 2.7 0.2 0.12 - 75 81 2091-T8 2.0 2.2 1.5 0.10 - 64 70 Weldalite 1.3 5.4 0.4 0.04 Ag 101 104 049-T8 2094 1.3 4.7 0.4 0.14 0.4 Ag 2095-T8 1.0 4.0 0.4 0.14 0.4 Ag 84 90 2195-TS 1.0 4.0 0.4 0.12 0.4 Ag 86 90 Table 2.2. Nominal compositions (wt %) and typical mechanical properties of commercial Al-Li alloys (refs. 1.7-8.10.18-20). Atomic Percenlaqe Lithium 10 20 30 40 50 60 70 80 85 90 95 718 * 7 0 0 600 ' 9 .9 2 0 .4 OOOF 521' 5 0 0 34 (Al) : 3 0 0 500F 1805' 200 3O0F (U)----- 00 lOOF 20 30 40 50 60 70 80 90 L.A.W. Weight Percentage Lithium Figure 2.4. Binarv- phase diagram for the .A,l-Li system (ref. 17). Upon aging following a solution heat treatment. S' precipitates homogeneously in the form of spheres (ref.9). The precipitation sequence can be described as: Ctsohii solution ------► a + S' + S ►a -r S where 6 is the equilibrium phase (AlLi). 10 Overaging results in formation of the Ô - phase. The transformation 6' —> ô is likely to occur by 6' dissolution and subsequent 6 precipitate formation. Although 6' precipitates result in high strength they are considered to be detrimental to ductility and fracture toughness since O' particles can be sheared by dislocations (refs. 20-21). 400 a + 5 Su Q. 200 0 8 16 24 atomic Li (%) Figure 2.5. The lower part of the Al-Li binary system (ref. 22). All the Li-bearing alloys are strengthened to some degree by precipitation. In some alloys, multiple precipitates may be present and/or complex precipitation sequences 11 used to optimize strength, ductility, toughness, and fatigue resistance. For easy reference, these precipitates are listed in Table 2.3. Precipitate Stoichiometry Crystal Structure Alloys Ô .AlLi B32 (cubic) 01420. 01421. 01430. 01440.01450,01460. 2090. 5" .-\l-,Li L12 (ordered fee) 2uvl. 2uv4. 2uv5. 2195. 8090.8091 01450.01460.2020.2090. 0' .ALCu tetragonal 2094.2095.2195 0 b ct(C I6 ) 01450.01460.2020.2090. I T, ■ALCuLi i hexauona! 2094. 2095. 2195. 8090 T: •ALCuLi; icosahedral. 2020. 2091 group = M3 5 T„ Aii.CusLi: j t'cclCn 2020 01430.01440. 2090.2091. S' .ALCuMg j orthorombic 2094.2095.2195. 8090. i 8091 ! S .ALMgLi Cubic 01420.01421 a ' .AL(Zr,_...LiJ 01420.01421. 01430. 01440.01450.01460.8090. LI2 (ordered fee) 8091.2090. 2091. 2094. P- A!,Zr 2095.2195 Psc AL(Zr,.Sci„) 01421.01460 Table 2.3. Precipitation in Li-bearing Aluminum Alloys (refs. 17.23-30). 12 Al-M s-Li (- Zr) Allovs . Alloy 01420 is the product of many years of Russian development (ref.8) and was employed in the construction of the welded supersonic aircraft MIG - 29. It has a nominal composition Al-5Mg-2Li-0.1Zr. Additions of Mg in the Al-Li binary system result in lower solid solubility of Li in aluminum (ref.31 ). The reduced solubility of Li in the matrix yields a higher volume fraction of 6' which contributes to higher levels of strength. Since alloy 01420 does not contain Cu it possesses low density (ref.4) (2.47 g/cm"). In addition Mg does not form precipitates (with the exception of .ALMgLi upon overaging) due to its high solid solubility limit. However, its contribution to the strength of the alloy is via solid solution. The precipitation reaction can be described as; ctsoiiii solution a - J - o '^ a-i-Ô -( ALMgLi) .ALMgLi is a product of extended aging and adversely affects alloy strength, ductility and toughness. Formation of .ALMgLi phase within the matrix also promotes precipitation in the precipitation free zones (PFZ's) which are narrow regions along high angle boundaries associated with poor ductility and fracture properties (ref.20). PFZ's are regions were precipitate nucléation is difficult due to vacancy and/or solute atom depletion since they are lost to the nearby sinks (ref.32). Vacancy-solute binding energies, degree of initial supersaturation and subsequent thermo-mechanical treatments determine the extent of the formation of PFZ's. The effect of those PFZ's on alloy properties such as stress corrosion cracking (SCC) will be discussed later. 13 Zr Additions. Zirconium has low solubility in the aluminum and. as a result, a fine dispersion of metastable P‘ [.A.tyZr] and /or a [Aty(Zr. Li)] spherically-shaped particles is formed (ref.22). These precipitates inliibit recrystallization and promote grain refinement. It is stated that zirconium intermetallics are considerably finer than manganese intermetallics and are distributed within the matrix with greater density' providing more effective strengthening (refs. 15.33). The P‘ precipitates also serve as heterogeneous nucléation sites for 6' [.^tyLi] while p‘ particles also resist dislocation cutting due to a high antiphase boundary energy. PFZ's are strengthened due to p' precipitation (refs. 7.20). The .Al-Zr phase diagram (ref.34) is shown in Figure 2.6. Atomic Percenloqe Zirconium 0.2 0.3 0.4 0.5 7 0 0 0.11 660 .7 * I200F 0.28 (Al) 6 0 0 0.2 0.4 0.6 0.8 1.0 1.2 I. 2.0 L.A.W. Weight Percentage Zirconium Figure 2.6. Binary phase diagram for the Al-Zr system (ref. 34) 14 Strensthenin^ Phases in Al-Cu-Li Allovs. The precipitation reactions are quite complex in this system. The precipitation sequence can be described as tbilovvs: 0" — ► 0 (AliCu) Cisolii] solution ^ GP Zones — ► 0 ' T, (AhCuLi) Copper reduces the solubility of Li in aluminum while Li modifies the structure of GP zones and also affects precipitation of the 0 " and 0 ‘ phases (refs. 17.20). The primary strengthening phases present seem to be a function of the Cu content and are brietlv described for the following allov svstems. 20% and Weldalite 049 Allovs. .Alloy 2090 and the Weldalite family alloys have a higher copper/magnesium ratio. Strengthening precipitates are 0*. T| and 6' while P' particles act as nucléation sites for 0" in addition to their grain refining role. Precipitation of TI [.ALCuLi] phase can be affected by P' dispersoids since they cause retention of the well-defined substructure and enhance heterogeneous nucléation (refs.7.17.20). Ti is nucleated at subgrain boundaries and on dislocations in the form of platelets (refs.2.17). Precipitation of Tt(increase in its volume fraction in the matrix) occurs at expense of 0 ' since both phases compete for the available Cu. Increasing volume fraction of Tp platelet precipitates results in high alloy strength (ref.21) and hence, both 2090 and 15 Weldalite alloys possess higher yield strength than other alloys (see Table 2.2). However, increased Cu content results in higher alloy density. Alloy 2020. The first commercial alloy based on the .A.l-Cu-Li system was produced by .Alcoa in the late '50s and designated as 2020 by the Aluminum Association. Alloy 2020 possesses high strength, low density (Figure 2.2). high elastic modulus and very good corrosion properties, which made it attractive for use in high-performance military structures such as the Navy's A-5A and RA-5C Vigilante (ref.35) airplanes. However, its low ductility and poor fracture toughness led to termination (ref. 10) of its production in the late '60s. The primary strengthening phases in 2020 are the partially coherent 0 ' (.ATCu). TilAhCuLi) and Tg (Al| more prominent (refs.35-36) than either T| or Tg. Cadmium suppresses 0 " precipitation, aids nucléation of 0 ‘ and inhibits grain and precipitate growth by segregating to the 0 ', matrix interface (refs.37-38). Manganese also inhibits grain growth by forming dispersoids (ref.36) which are present in 2020 as 0.5 - 1 pm AEoCuiMn]. Low ductility is suggested to relate to both formation of PFZs due to heterogeneous precipitation of equilibrium phases along grain boundaries and metastable coherent 6' (ALLi) and partially coherent T, precipitates which form upon aging. For aging conditions that do not produce PFZs cracking occurs within the grains due to shearing of coherent and partially coherent precipitates by moving dislocations. When PFZs are present, cracking occurs due to plastic deformation in these soft regions 16 (refs.39-40) and cracks propagate intergranularly (or just adjacent to the grain boundary in the PFZ). Cadmium segregation is likely to enhance crack propagation along grain boundaries by lowering the surface energy associated with fracture (ref. 36). 2091 and S090 Allovs. These alloys have a low copper/magnesium ratio and consequently 0 ' is not a significant strengthening precipitate. The S' phase (ATCuMg) is favored and strength is lower compared to that of the Weldalite alloys. The S' promotes cross slip, improves ductility and toughness but reduces fatigue resistance. Equilibrium 6- phase does not seem to form in alloys with a higher magnesium content, and is generally not observed in these alloys. Other strengthening phases are 6' and T|. The latter competes with S' for available Cu atoms and heterogeneous nucléation sites (ref.20). Straining (or stretching) of wrought alloys prior to aging provides new heterogeneous nucléation sites and therefore eliminates Ti and S' PFZs. Scandiurn Atkliiions. Small additions of scandium (Sc) together with Mg seem to increase the strength of Al-Cu-Li alloys (ref.41). aged in the temperature range from 20 "C to 180 "C. Increase in strength is due to grain and subgrain refinement and a higher density of (3 sc -Al 3(Zr.Sc)x dispersoids. The Al-Sc phase diagram (ref.42) is shown in Figure 2.7. It is important to note that due to lower diffusion coefficients of both Sc and Zr. the distribution of the dispersoids is primarily dependent on the distribution of these elements in the matrix of the cast alloy (ref.33). Scandium also promotes refinement of 17 6' and T , precipitates which results in good alloy ductility in addition to enhanced strength properties (ref.41). Formation of layered particles was also observed with the addition of Sc. In the case of .A.l-Mg-Li-Zr alloys. 5‘ was found (ref.43) to precipitate and form a shell around the (3 sc particle. Aging in the temperature range between 140 "C and 170 caused thickening of this shell and increased the strength of the alloy. Alloying of the .Al-Mg-Li system with Sc has been reported to promote an improvement of alloy resistance to overaging in the HAZ as compared to 01420 Soviet alloy (ref. 44). This means that Sc-bearing alloys are relatively insensitive to prolonged heating at 300 -350 "C and their hardness remains essentially unchanged. Fridlyander et al (ref.43) suggest that .Al-Mg-Li-Zr-Sc alloys could be considered as a replacement for both 2219 and 01420 commercial weldable alloys. In alloys of the system Al-Mg-Li. scandium decreases the stability of the solid solution in the region of .AfMg: and .AlizMgn phases. Thereby, segregation of those phases decreases the Mg/Li ratio in the solid solution and decreases its stability in the region of the .ALMgLi phase at 290 °C which causes an overall increase in the strength potential of the alloy (ref.31 ). However, fracture toughness (Kic) of the alloy through the thickness (short transverse direction) seems to decrease. Weight Percent Aluminum 30 30 40 so 60 70 80 90 ICO 1600 t 1200 - o 1000 - 3 .Q 8 0 0 - ua a, c 6 0 0 - 6- 4 0 0 - (aS c) (Al) :o o - 10 40 SO 60 7( 90 too Sc Atomic Percent Aluminum ■0 940 (Al) ScAl, 4- (Al) 9 2 0 - 900 99 7 999 100 Atomic Percent Aluminum Al Figure 2.7. Binary phase diagram for the Al-Sc system (ref. 42). 19 2.3 Principles of Solidification Solidification is a transformation from a non-crystallographic to a crv stallographic state of a metal or alloy, which takes place ahead of an advancing solid- liquid interface (ref.45). Two important mechanisms constitute solidification viz.. nucléation and growth of the crystallographic state out of the melt. These phenomena are basic in the field of metallurgy and they have been the subject of considerable research in the past decades in order to promote advancements of such technological aspirations as ingot casting, single crystal growth, rapidly solidified metals and alloys, fusion welding etc. .A brief introduction to the solidification principles is presented in the following sections. 2.3.1 Drivina Force for Solidification and Homogeneous Nucléation Reducing the temperature of a pure metal by an amount AT below its melting point Tm results in a difference AG ( = G l- Gs) in the free energy between the liquid and solid states with the solid state having lower free energy thus, being thermodynamically stable (Figure 2.8). More specifically the change in the volume free energy between the two phases is given by dC, (eq. 1) where Lv is the latent heat per unit volume (ref. 45). 20 AG Temperature m Figure 2.8. Free energy difference vs. temperature for solid and liquid phases (ref. 45). However, lowering the temperature of the melt below the equilibrium melting point does not necessarily imply that spontaneous formation of the solid phase will take place. There is a free energy barrier that needs to overcome before transformation to the solid state occurs as described below. Assuming that metal atoms form a solid cluster instantly within a pure liquid metal already undercooled by an amount AT below its melting point, then a change in the overall free energy of the system will take place AG- -VsAGv + A s lY s l (eq. 2) 21 where Vs is the volume of the solid cluster. Asl is the solid/liquid interfacial area. ysL is the solid liquid interfacial free energy and AGv is given by the equation 1. Since the excess interfacial free energy becomes minimal for a cluster of spherical shape the equation 2 can be re-written as 4 AG^ = /T7-'AG,. + (eq. 3) where r is the radius of the solid sphere and ysi. is assumed isotropic. .A plot of the change in free energy versus cluster size is shown in Figure 2.9. It can be seen that there is a critical cluster radius r* for which the free energy of the system becomes maximum. For clusters with size r (nuclei), the free energy of the system will be further reduced. Solid spheres with size r=r* are in an unstable equilibrium with the liquid atoms. It is important to observe the dominant effect of the interfacial energy for solids with sizes r * _ V n y Sl.^m 7^ AG,, and (eq. 5) 3{AG| 31:. ( A n - It is interesting to note that AG* decreases faster when compared to r* with increasing undercooling AT. AG interfacial 2 energy ar AG AG, Volume free energy ar AT Figure 2.9. Free energy change vs. sphere radius r for homogeneous nucléation (refs.45- 46) 2.3.2 Heterogeneous Nucléation Homogeneous nucléation requires significant amount of liquid undercooling (up to 250°C) but in most practical cases the liquid to solid transformation takes place at much lower undercoolings (1-20°C) starting on sites such as impurity particles within the liquid, mould walls etc. via a mechanism knowm as heterogeneous nucléation (refs.45- 47). This type of nucléation is facilitated by a smaller nucléation energy barrier as compared to that one for the case of homogeneous nucléation and is the result of surface energy modifications assuming the solid nucleus forms in contact with heterogeneous sites as mentioned above. Figure 2.10 shows a solid embrvo in contact with such a heterogeneous site (eg. flat mould wall). The energy of the system is minimized if the shape of that solid embryo is a spherical cap assuming yst. is isotropic. From the balance of the interfacial tensions in the plane of the mould it is 7ML=7s.\r 7sl(cos0) (eq. 6) where 0 is the wetting angle. The formation of the solid cap results in overall change of the free energy of the system given as ^Ghct = -V sAGv •A.sl7sl‘^Asm7sm-Asm7ml (eq. 7) where V’s is the volume of the solid cap. .A.sl and Asm are the areas of the solid-liquid and solid-mould interfaces. 7 sl. 7sm and 7ml are the free energies of the solid-liquid, solid- mold and mould -liquid interfaces. 24 Liquid Substrate Figure 2.10. Schematic showing heterogeneous nucléation of spherical cap on tlat substrate surface (ref. 45.47). Creation of the solid-liquid and solid - mould interfaces have a positive contribution in the above equation while disruption of the mould - liquid interface has a negative one. Formation of the solid from liquid phase also has a negative effect since the solid state is thermodynamically stable at temperatures below Tm. .Appropriate mathematic formulation of the equations 6 and 7 results in : ^Arcr'-y,, |-S((9) (eq. 8) where (2-i-C0S^)(l-C0S^)' S{6) = (eq. 9) and r is the cap radius. Figure 2.11 shows the change in the free energy of the system during both heterogeneous and homogeneous nucléation. It can be seen that the AG*het is lower than the AG*horn- .Actually it can be derived from the equation 8 that AÜ:, = = ACLS(S) (eq. 10) j AO|‘ and r* = (eq. 11). AG,. It is interesting to note that the activation energy barrier is the same as that for homogeneous nucléation reduced by the factor S(0). Many parameters seem to control the value of this factor (ref. 47). Based on classical heterogeneous nucléation theory the general characteristics of a good grain retlner include high surface energy between particle and melt ('/m l ) and low surface energy between the nucleating agent and the growing solid C / s m ) which results in small contact angle between the nucleating agent and the growing solid (eq. 9). The value '/sm decreases as lattice disregistry between particle and solid decreases. Increased chemical affinity between heterogeneous particle and solid also results in low ’/sm quantity. Additionally an effective nucleating agent possesses maximum surface area, optimum surface character and stability in the molten metal (ref 47). It should be noted, although many researchers have established (refs.47-50 ) that the undercooling required for solidification increases with lattice mismatch this is not 26 always the case. It is suggested that surface nature, surface area or other factors not predicted by the classical nucléation theory may potentially have an overriding effect in determining the end point, viz.. the grain size of a cast (ref. 47). AG AG AG AG. Figure 2.11. Free energy change vs. spherical cluster radius r for both homogeneous and heterogeneous nucléation.( refs. 45-46). 2.4 Microstructure Evolution During Welding 2.4.1 Fusion Zone Microstructure A typical fusion weld contains a number of distinct regions that can broadly be classified as a fusion zone, where melting and resolidification occur, and heat affected zone (HAZ). where the surrounding base metal microstructure is altered by the heat of welding. Fusion zone microstructure in aluminum alloys is primarily dictated by the solidification process, rather than by post-solidification transformations. During welding. 27 a localized region of the base material is melted and solidified. The solidification morphology, as well as the solute distribution, depends on the interrelationship of temperature gradient, solidification growth rate, and diffusion (refs.45-47 ). As a result, different welding processes affect the weld microstructure and subsequent properties differently. For example, high energy density processes like EBW produce refined weld metal microstructures while arc welding processes such as UI AW result m coarser microstructures. The constitutional supercooling (ref. 46) theoiy has been developed to explain morphological stability along an advancing solidification front. For a binary alloy system and a given composition Co. a variation in liquidas temperature is developed ahead of the advancing solid-liquid interface, attributable to solute build up (partition coefficient k < 1 ). as shown in Figure 2.12. If the actual thermal gradient Gaa. at the solid-liquid interface is greater than that for the predicted liquidas temperatures (G) then a planar front is maintained. When Gad During welding the aforementioned subgrains nucleate directly from randomly oriented grains in the HAZ. which form the substrate at the fusion boundary. The driving force for nucléation is low since the thermodvnamic barrier for solidification is almost 28 Liquidus Solidus Region of (Cs)‘ Cq (Cl)‘ Concentration Constitutional Supercooling Distance (x) * ^ACT. > _ _ miCo(\-k) k = E l kDi L Interface Distance (x) R ~ D , ~ Cl Figure 2.12. Constitutional supercooling during alloy solidification. (A) Phase diagram. (B) solute-enriched layer. (C) unstable (supercooled) interface. The criterion for morphological stability for a moving solidification front with velocity R is also shown (refs. 51-52). eliminated and only growth of new grains occurs. Each of these subgrains solidifies along a crystallographic direction (<100> type for cubic structures) dictated by the orientation of the parent grain in the HAZ. described as epitaxial growth. Subgrains with their solidification growth direction parallel to that of the maximum temperature gradient, grow faster than others less favorably oriented and. this results in formation of solidification grain boimdaries at regions where these subgrains impinge (competitive growth). 29 Planar Cellular Growth Growth !i Equiaxed Columnar i-i 1 Gimn Dendritic Dendritic 4tOO> , (' . ^ 1 unowoooied Figure 2.13. Mode of solidification versus amount of constitutional supercooling (ref.51). Strengthening of the fusion zone by aging following welding is limited since much of the solute needed to form precipitates is tied up in a eutectic constituent, which forms at the end of solidification. Figure 2.14 shows the fusion zone microstructure of autogenous GTA welds in alloys 2090. 8090. 2195. while Figure 2.15 shows the microstructure of a VPPA weld in alloy 2195 using 2319 filler metal. As will be described in a subsequent section, non equilibrium weld solidification conditions promotes the formation of some eutectic in the fusion zone during the final stages of solidification in all the Al-Li-X alloys. As a result, the predicted solidification temperature range is the difference between the equilibrium 30 liquidus temperature and the eutectic temperature. • " 4L' 7 ^ J-X 1 .* / /I a Figure 2.14. Representative fusion zone microstructure of autogenous welds. A) alloy 2090. B) alloy 8090. C) alloy 2195. 31 Figure 2.15. Representative fusion zone microstructure of VPPA welds in 2195 using 2319 filler metal. 2.4.2 Scheil Partitioning Solidification of fusion welds is accompanied by segregation of alloying and impurity elements at intercellular and interdendritic regions under conditions of rapid cooling. Solute partitioning along solidification subgrain boundaries (SSGB) is modeled using the Scheil's approach (refs.45-47) (case II solidification) for non-equilibrium solidification, which is also know as non-equilibrium lever rule. The following assumptions are made (ref.52): i l 1. Uniform liquid composition or complete mixing in the liquid state. 2. Negligible solid-state diffusion. 3. Local equilibrium at the solid/liquid interface is dictated by the phase diagram (partition coefficient k is considered constant). The solute concentration (Cs) in the solid as function of the fraction solidified (fs) is given by C,=^C„(1(eq.l2) Based on this approach solidification will progress with the liquid becoming continuously enriched in solute until it reaches the eutectic composition. .At this point the remaining melt will solidify as eutectic along the SSGB (invariant point of solidification). Given the equation 12. the eutectic fraction (IL) for a binary system can be estimated using the following expression C — A = ( 7 T-)^-' (eq.l3) 0 where C e is the eutectic composition dictated by the equilibrium phase diagram. It is clear that the mechanism described above always leaves the solidified structure with a constituent decorating the SSGB (Figures 2.16-2.17). which has lower melting temperature compared to the solidus of the bulk metal. This is very important with regards to cracking propensity of the alloy during welding, as will be described later. z,nm 0.2 0.4 0.6 0.8 Normalized distance (z/L*) Figure 2.16. Copper concentration profile estimated using the Scheil equation for an Al- 2\vt% Cu binary alloy. Superimposed are the predicted composition for the first solid to form (k„Cu) and the nominal alloy composition Co (ref.52). GTAW (Ai-2wt% Cu) 7.5 2.5 40 Position, urn Figure 2.17. Electron microprobe analysis of Cu across primary solidification boundaries in an Al-2\vt%Cu GTA weld (ref.52). 34 2.4.3 Heat Affected Zone Microstructure Microstructure within the HAZ is primarily dictated by solid-state reactions that occur in regions heated to elevated temperature as heat flows away from the fusion zone. In the Li-bearing aluminum alloys, most HAZ microstructural changes are associated with an alteration of the strengthening precipitates. Other changes, such as grain growth, may also occur, but their effect on HAZ properties is relatively minor. Coarsening of precipitates is anticipated to be limited to locations experiencing relatively low peak temperatures whereas precipitate dissolution seems to take place at regions experiencing higher peak temperatures. It is difficult to produce a quantitative analysis of the heat affected zone of alloys possessing a variety of strengthening precipitates since one phase may dissolve while another coarsens. Both dissolution and coarsening are primarily diffusion-controlled processes. The stability of the various precipitates described earlier, at elevated temperatures depends on the ability to minimize the free energy of the system viz.. the minimization of chemical free energy which is the driving force for dissolution and the reduction of the interfacial energy which takes place with precipitate coarsening (ref.53). Dissolution of precipitates close to the fusion boundary provides solute in the matrix and results in increased strength of the alloy due to solid solution strengthening but the overall increase of strength in these regions stems from reprecipitation during cooling from elevated temperature. Regions where partial dissolution exists have lower strength due to lower fraction of precipitation whereas regions experiencing lower temperatures possess lower strength due to the precipitate coarsening effect (ref. 21.54). 33 2.5 Weld Cracking Susceptibility 2.5.1 Weld Metal Solidification Cracking Weld solidification cracking is an inherent problem in many structural aluminum alloys resulting from a combination of a large solidification temperature range and significant contraction stresses that arise due to a large coefficient of thermal expansion (GTE). The interrelationship among alloy composition, solidification temperature range, and the amount and nature of eutectic constituent that forms during the final stages of solidification controls cracking susceptibility in aluminum alloys. Weld solidification cracking usually occurs along solidification grain boundaries. Partitioning of alloy and impurity elements during solidification promotes the formation of low melting liquid films along these boundaries which may separate, i.e. crack, if sufficient stress is imparted across the boundary' (ref.55). The effect of composition on weld metal cracking susceptibility can be explained using the simple eutectic phase diagram in Figure 2.18. ,A.t low solute contents the solidification temperature range is small and only a small amount of eutectic forms during the final stages of solidification resulting in a crack-resistant structure. .As the solute content approaches the solid solubility limit both solidification temperature range and amount of eutectic liquid at grain boundaries increase. High cracking susceptibility results from this combined effect. Under non equilibrium conditions, such as during welding this maximum response occurs at compositions below the maximum solid solubility (refs.56-57). As solute content increases above maximum solid solubility, the solidification temperature range again narrows and more liquid of eutectic composition is present during the final stages of 36 solidification. The increased eutectic product will tend to "heal" any cracks which form. Thus, cracking resistance in aluminum alloys can generally be achieved with either "solute lean" or “solute rich" compositions. Non-Equilibrium Solidus Liquid Equilibrium .Non-Equilibrium oc N, V Composition Figure 2.18. Weld metal cracking susceptibility as a function of composition in a simple binary alloy system containing a eutectic reaction. Both equilibrium and non-equilibrium solidification conditions are considered. 37 This is illustrated in Figure 2.19 for the Al-Cu. Al-Li. and Al-Mg systems based on solidification cracking data generated using these binary systems (refs.57-58). Rather than the actual composition, a value normalized by the maximum solid solubility is used so that all the alloys can be plotted relative to the theoretical peak in cracking susceptibility under equilibrium conditions (Figure 2.18). Based on the curves in Figures 2.19. 2.20 and 2.21 the peak in cracking susceptibility occurs at 2.5% Li. 3% Cu. 3% Mg. and 0.8% Si. Note that these concentrations are very close to the nominal composition of many of the commercial alloys listed in Table 2.2. Thus, it is not surprising that many of the Al-Li-X alloys show- some level of susceptibility to weld solidification cracking. It should be noted that most .Al-Li-X base metals were designed to optimize strength and the compositions that promote good mechanical properties may not be optimum for weld cracking resistance. 38 Al-Cu Al-Ll Al-Si S 12 w « im u 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 Normalized Composition Figure 2.19. Cracking susceptibility versus normalized composition (Co/Cmax) for the Al- Cu. .41-Me. Al-Li and Al-Si binary systems. The information from Figure 2.17 can further be used to estimate the weld solidification temperature range and fraction eutectic that results in a maximum in cracking susceptibility. This is done by taking the composition at the point of maximum 39 01420 01440 S090 • S£ 02 0.4 0.6 0.3 I 12 1.4 1.6 1.3 22 Normalized Composition Figure 2.20. Cracking susceptibility- versus normalized composition (Co/Cmax) for the .Al-Cu. .Al-Mg and .Al-Li binary- systems. The location of various allovs is shown. 2090 e 2195 V C. 3 SC c w 2 u 0 0.4 0.6 O j 1 1.4 1.6 1.3 2 Normalized Composition Figure 2.21. Cracking susceptibility versus normalized composition (Co/Cmix) for the Al-Cu and Al-Li binary systems. The location of various alloys is shown. 40 cracking susceptibility in Figure 2.19 and determining the temperature differential between the liquidus at this composition and the eutectic temperature, using the appropriate binary phase diagram. The Scheil Equation (refs.46-47) is used to estimate the fraction eutectic liquid for the composition of maximum cracking, again based on the binary phase diagrams. These data are provided in Table 2.4. .\iIoy System Composition of Max. Solidification Temp. Fraction Eutectic Cracking (wt%) Range (°C) (vol %) Al-Li 2.5 55 9.9 .Al-Cu 3.0 100 5.5 .Al-Mg 3.0 205 1.4 Al-Si 0.8 90 4.2 Table 2.4. Solidification temperature range and fraction eutectic at maximum cracking for binan.' allov svstems. Estimates of solidification temperature range and fraction eutectic can also be determined for the commercial alloys using binary systems. These data are provided in Table 2.5. It is recognized that these serve only as estimates, since there will be interactive effects among elements. Unfortunately, ternary systems are not available for Al-Li-X alloys to make these estimates more accurate. The data in Tables 2.4 and 2.5. and Figures 2.19-2.21 can be used to provide some estimate of weld solidification cracking susceptibility by comparing the solidification 41 temperature range and determining how close the nominal composition is to the peak in cracking susceptibility in Table 2.4. For the Ai-Cu-Li alloys, the addition of Cu appears to expand the solidification range in all cases. Alloys such as 2090 and 2091. whose Cu and Li content are very close to the Cu and Li peaks in Figure 2.21 (3.0 and 2.5 wt %. respectively) would be expected to be very susceptible to cracking. In contrast, an .A.I-C 11 alloy such as 2219. which has a Cu content of 6.5 wt %. Is relatively resistant to weld solidification cracking. This results from both the narrowing of the weld solidification temperature range and the large amount of eutectic that forms at the end of solidification. Predicted Weld Solit iification Fraction Eutectic Calculated from .\lloy Temperature Ran ge ("C) Scheil Equation (ref.47) Al-Li Al-Cu Al-Mg .Al-Li .Al-Cu Al-Mg 01420 48 - 178 6.8 - 3.6 8090 45 109 - 9.9 1.5 - 2020 53 97 - 2.9 8.8 - 2090 46 103 - 8.6 4.9 - 2091 48 105 - 6.8 3.8 - 2094 52 96 - 3.3 9.6 - 2095 54 98 - 1 -> 7.9 - 2195 54 98 - 2.2 7.9 - 2014 - 97 - - 9.1 - 2219 - 91 - - 13.6 - Table 2.5. Solidification Data for Commercial Al-Li-X Alloys based on Binary Phase Diagrams The fraction eutectic can also serve as a good measure of cracking susceptibility. From Table 2.4. it can be seen that cracking is a maximum in the binary systems when 42 the amount of eutectic is between 4 and 10% (ignoring the Al-Mg system). At levels below this, insufficient liquid of eutectic composition is present at the end of solidification to completely wet the grain boundaries. At higher levels, sufficient liquid is present to heal cracks that may form. This explains the characteristic of eutectic systems, whereby cracking susceptibility first increases to a maximum and then decreases with the addition of solute, for the Ai-Lu-Li system, the reduction of Li below 2.5% and increase in Cu above 3.0% should reduce cracking susceptibility. This was the approach taken with the development of the Weldalite^" family of alloys in order to improve their weldability (refs.56.59). The negative effect of Cu on cryogenic fracture toughness, however, resulted in the reduction of Cu levels in these alloys to around 4.0%. In spite of its high solidification temperature range (Table 2.5). alloy 01420 is reported (refs. 1.6.8) to show good weldability. From Figure 2.19 and Table 2.4 it can be seen that the binaiy .41-Mg system has a peak in cracking susceptibility at approximately 3.0 \vt% Mg (ref. 16). The high Mg of this alloy. 5.3 wt%. places alloy 01420 away from the cracking susceptible regime. Figure 2.22 shows representative weld metal solidification cracks in autogenous welds in alloys 8090 and 2195. This cracking was produced using the Varestraint test as described in the following paragraphs. As is typical, weld solidification cracks follow solidification grain boundaries since these are the last regions to solidify. Note that these boundaries exhibit a continuous network of eutectic constituent. This indicates that liquid films of eutectic composition existed along these boundaries at the end of solidification. There is no evidence of "backfilling" in either of these alloys, suggesting 43 that the amount of liquid was sufficient to wet the boundary, but insufficient to heal any cracks that might form. This is consistent with the data in Table 2.5 that suggests fraction eutectic on the order of 5vol. %. Weld solidification cracking susceptibility data from the literature have been generated using different types of tests such as Varestraint and Houldcroft tests. Since these tests are fundamentally different and neither are standardized, it is difficult to compare results published by different investigators. Recently. Lippold and Lin (ref.55) have proposed a technique using the Varestraint test that measures the actual solidification cracking temperature range (SCTR) in the fusion zone. This is an alloy- specific parameter determined by measuring the length of the longest solidification crack in a test sample where the cooling rate and solidification velocity are known. Data gathered for various .Al-Cu and .Al-Cu-Li alloys are presented in Table 2.6. Longitudinal Varestraint Transverse Varestraint .Alloy MCD (mm) SCTR(°C) MCD (mm) SCTR (°C) 2014 6.3 185 2219 3.9 155 2.04 125 2090 7.2 190 2094 14.5(1) 225 2195 13.3 (1) 220 4.07 182 ( 1 ) contribution from fusion boundary cracking Table 2.6. The Solidification Cracking Temperature Range (SCTR) Determined using the Varestraint Test (ref. 55). 44 * / X • v / : ' 50 pm Figure 2.22. Representative weld solidification cracking in Varestraint samples tested at 3% strain. 400 X. A) alloy 8090. B) alloy 2195. The arrow in B highlights the tip of the crack. Note that the SCTR is greater than the predicted solidification temperature range shown in Table 2.5 for both the Al-Cu and Al-Li systems. This suggests that the interactive effects of Cu and Li in conjunction with other alloying elements or impurities (Mg. Si. Mn. Fe. etc.) tend to extend the solidification temperature range to lower 45 temperatures. Increased resistance to weld solidification cracking has been reported in both Weldalite™ 049 and 2090 alloys in studies using filler metals (ref.56) with high copper content, such as 2319. Figure 2.15 shows the weld metal microstructure of the alloy 2195 welded using 2319 filler metal. The higher volume fraction eutectic in this structure relative to lower Cu alloys (Figure 2.14) is apparent. .A.dditions of Mg result in lower melting point Mg-rich eutectics which extend the "mushy" solidification range and change the wetting characteristics of the interdendritic liquid, thus promoting solidification cracking (ref.56). Small amounts of Ti refine solidification grain sizes and thus the distribution of eutectic liquid at grain boundaries is modified resulting in lower cracking susceptibility. High levels of Ti are reported to result in the formation of brittle intermetallics at the grain boundaries thus, promoting higher cracking susceptibility in the solid state (ref. 56). Alloy 01420 has been reported to exhibit "good weldability" (refs. 1.6.8) when using filler materials boosted in Mg and with small Ti additions. Low-copper, high- magnesium 8090 alloys have also been reported (refs. 1.7.60) to exhibit good resistance to cracking using when using Al-Mg based (5000-series) filler materials. Employing the Houldcroft test. Reddy et al (ref. 19) conducted weldability studies for the alloy 01441 using a range of filler alloys i.e. 01441. 4043. 2319 and 5356. It was found that weld cracking susceptibility was reduced in the order: autogenous. 01441. 4043. 2319. 5356 (Figure 2.23). Alloys 4043 and 2319 have the smallest solidification temperature range (61°C and 46 °C respectively), which is significantly lower than that 46 for the alloy 01441 (106 °C). It is also known that fillers 4043. 2319 and 5356 are eutectiferous alloys. The fact that alloy 5356 results in better overall weldability irrespective of its higher solidification temperature range, is attributed to the fact that the eutectic liquid is not continuously distributed around the grain boundaries thus, allowing extensive network of solid-solid bridges to form during the last stages of solidification. 140 120 100 - 80 - en o Filler Alloys Figure 2.23. Cracking susceptibility of autogenous and filler metal welds on 01441-T8 aluminum alloy sheets using the Houldcroft test (ref. 19). The weldability of some commercial Al-Li-X alloys that contain Cu and Mg can be assessed using a map of total crack length (TCL) versus Cu and Mg content (ref.6I). as shown in Figure 2.24. This map does not reflect the influence of Li. but shows how the 47 combined effects of Mg and Cu influence cracking susceptibility. It is interesting to note that the peak in cracking susceptibility is very close to the 1.4% Mg and 3.0% Cu values from Table 2.4 for maximum cracking in binary systems. High levels of either element reduce susceptibility, as shown alloys 2219 or 5083. .Alloys containing substantial amounts of both elements such as 2024 (4.4 Cu. 1.5Mg) and 7075 (1.6Cu. 2.5Mg) are known to have poor weldability as reflected by their high values of I'CL. Based on Figure 2.24. the order of increasing resistance to cracking of some of Al-Li-X alloys is 2091. 8090. 2090 and Weldalite'"^' 049. Contour lines: total crack length (TCL). in. 4 3 I 7071 E 3 2 !024 1 7005 • 2091 • 80906061 I 2014. 0 2 3 4 5 Copper, welght% Figure 2.24. Total crack length contour map. Li-bearing aluminum alloys 8090. 2090. 2091 and Weldalite 049 are superimposed in the above diagram (refs.61- 62). 48 2.5.2 HAZ Liquation Cracking Heat affected zone liquation cracking by definition is a type of high temperature weld cracking which takes place in the HAZ adjacent to the fiision boundary and is associated with the formation of liquid films at grain boundaries. This region is defined as the partially melted zone (PMZ) of the HAZ. Although metallurgically this type of cracking is associated with the presence of liquid, no theories exist that quantify the potential for HAZ liquation cracking based only on the amount and/or the characteristics of the liquid evolved. Sufficient tensile stress is also required in order for cracking to occur {crack initiation) and such stresses do not generally develop until the weld pool begins to cool. Since the PMZ is a region adjacent to the fusion zone where melting ranges from 0% to 100% evolution of liquid in the HAZ is. by definition, confined to the PMZ. The peak temperatures experienced by PMZ fall between the liquidas (T l) and effective solidus of the BM. The effective solidus is always below the equilibrium solidus due to segregation, which is invariably present in commercial aluminum alloys (ref.63). In general, the paucity of data currently available in the literature precludes a straightforward assessment of HAZ liquation cracking in Al-Li-X alloys. In addition, those data have been generated using a variety of tests such as the Gleeble® hot ductility test or the Varestraint test. Results from different types of tests cannot be directly compared since each test uses different criteria in assessing weldability. Using the Gleeble® hot ductility test Zacharia et al (ref.64) studied the HAZ liquation cracking of alloys 2090 and 2091. They concluded that both alloys do not 49 exhibit a strong tendency for HAZ liquation cracking, since the temperature range in which the material recovers its ductility upon cooling from the peak HAZ temperature is extremely small. The same conclusions were drawn by Yunjia et al (ref.65) who studied the weldability of both 2090 and 2024 alloys using the Varestraint test. Alloy 2090 was judged to be less susceptible to HAZ liquation cracking compared to alloy 2024. In another study Yunjia et al (ref.66) studied the weldability of both 8090 and 2024 alloys using the longitudinal Varestraint test. They again concluded that alloy 8090 was less prone to HAZ liquation cracking compared to 2024 alloy. Studies were performed by R.V. llyushenko et al (ref.44) to simulate the response of the HAZ of alloy 01420 at different thermal cycles using isothermal soaking of specimens in a molten tin bath. Specimens previously had been heated to different temperatures. It was deduced that by heating in the range from 550-580 "C (T l= 635 "C for 01420 alloy) both dissolution of the strengthening phases and intensive melting of primary intermetal lie compounds took place. Hardness of those specimens was measured to be minimal and it couldn't be restored even after artiftcial aging (at 160 for 16 hours). .Although the tendency of liquid formation was traced in the previous work, strain and/or stain rate effects were not examined. Consequently, the overall HAZ liquation cracking susceptibility of alloy 01420 was not clearly determined. The weldability of several variants of Weldalite049 alloy with different levels of Cu content was studied by Kramer et al (ref.59) using the spot Varestraint test. Alloys 2090. 2219 and 2014 were also examined. Employing the total crack length (TCL) in the HAZ as a weldabilitv criterion, it was shown that little or no cracking occurred in the 50 HAZ of alloys 2090 and Weldalite^^’ 049. However, a significant amount of HAZ cracking was found in alloys 2219 and 2014 (see Table 2.7). Liquation of low melting- point eutectics on grain boundaries accounted for the formation of cracks in the HAZ of those allovs. Alloy TCL in HAZ per Sample Weldalite* ^' 049 Variants AVGSTDV 6.1 wi% Cu 0.0 0.0 5.2 wt% Cu 0.5 0.9 4.8 wt% Cu 0.0 0.0 2090 0.0 0.0 2219 8.2 0.7 2014 9.0 2.8 *Total Crack Length in Weld Heat Affectec Zone Table 2.7. Spot Varestraint Test Results (ref. 59). It should be pointed out that the HAZ liquation cracking susceptibility of Al-Li-X alloys is also influenced by the cracking susceptibility of the fusion zone. If weld solidification cracking susceptibility is high, then cracking will tend to be concentrated in the fusion zone even if the HAZ is susceptible. In cases where filler metals are used improve weld solidification cracking resistance (such as when 2319 is used with 2000- series alloys), it is possible that HAZ liquation cracking could be a problem in some alloys. It is understood that when the weld cracking resistance of the weld metal is improved by filler metal selection, the susceptibility of the HAZ may then be realized 51 since liquated HAZ grain boundaries may then become the weakest link in the microstructure. 2.6 Previous Investigation and Hypotheses Currently two theories have been proposed to explain EQZ formation (refs.65. 67-68). The first hypothesis (ref. 67) is based on the premise that the EQZ is located in the PMZ and forms by a recrystallization mechanism. It is proposed that during the weld thermal cycle the HAZ immediately adjacent to the fusion boundary undergoes recry stallization and grain boundary liquation within the PMZ/HAZ. The second hypothesis proposed by Lippold et al (ref.68) and Yunjia (ref. 65) involves a solidification mechanism requiring both nucléation and growth. It is based on heterogeneous nucléation within a molten layer near the fusion boundary, perhaps a stagnant liquid region defined by the unmixed zone. Solid nuclei (precipitates or dispersoids) that are able to survive the thermal conditions (Figure 2.25) in this region serve as heterogeneous nucléation sites for growth of the equiaxed grains. Yunjia et al proposed that dispersoids such as Al]Ti. AljZr or Alj(Lix. Zri.x) are possible nuclei. Such nucleating particles exist in the parent metals following homogenization and solution heat treatment operations (refs. 68-69). Guttierez et al (ref. 70) showed that the EQZ can be eliminated by altering one of the following. 1) weld pool stirring. 2) composition, and 3) nature of the base metal substrate. As noted earlier, vigorous weld pool fluid flow can effectively stir the EQZ away. In a simple laboratory experiment, an electromagnetic coil around a OTA torch 52 was used to rapidly spin the molten pool during welding. Fusion zones produced in this fashion exhibited no EQZ. while those produced under similar conditions but without the coil showed a distinct EQZ. Guttierez also clearly showed the effect of composition on EQZ formation. In special alloys made with varying Li and Zr. alloys with less than 0.5 wt% Li and 0.04 wt% Zr did not exhibit an EQZ. Unfortunately, the individual (uncoupled) effects of Li and Zr could not be studied. Locus of Peak Temperatures ^ T^-StO'C L Unaffected True Composite Base Metal HAZ region Weld Pool vilTurbuience ^ AI3Zr & A13(U.Zr) Particles Equiaxed Grain Zone Figure 2.25. Schematic illustration of suggested Heterogeneous Nucléation Mechanism for the formation of the Equiaxed Grain Zone (ref.70). 33 [n a study of weld substrate effects, it was found that no EQZ formed when the substrate was as-welded (or cast). This occurs because the nucleating particles are assumed to dissolve or melt, and do not reform during weld cooling. When these welds are solutionized and then rewelded, the EQZ reappears, presumably because the nucleating particles have reformed during the solution heat treatment. 2.7 Equiaxed Grain Zone Formation and .Associated Fusion Boundary Cracking in Al-Li-X .Alloys The equiaxed grain zone ( EQZ) has been observed along the fusion boundary in Al-Li-X alloy welds, (refs 65.67-69) as mentioned above. This region consists of fine equiaxed grains and is located adjacent to the fusion boundary within the fusion zone. The grain boundaries are typically decorated with a eutectic constituent (ref.67). Careful microscopic examination of the EQZ has revealed (refs.67-68) that the fine grains do not have dendritic morphology (Figures 2.26 and 2.27). Not all Al-Li-X alloys exhibit an EQZ. as shown in the photomicrograph of the fusion boundar\- of .Alloy 8090 (Figure 2.28). The EQZ has been associated with severe cracking during fabrication and repair of structures. This cracking has been reproduced in .Al-Li-X alloys using the Varestraint test (ref.71 ) (Figure 2.29). Cracking is most often not isolated in the EQZ. but may propagate from the HAZ and/or weld metal to fusion boundary and continue along the EQZ (refs.67-68). Studies by Lippold and Lin (ref.55) have shown that vigorous stirring of the weld pool eliminates the EQZ and prevents this form of cracking. 54 ■ ;.» Fusion Zone HAZ ^ 100 |im S Figure 2.26. Fusion boundar\- region in Varestraint sample of alloy 2195. Arrows indicate the location of fusion boundary. HAZ T EOZ ® > 100 jjin § Figure 2.27. Fusion boundary region in weld of alloy 2195 using 2319 filler metal. Arrows indicate the location of fusion boundary. The fusion zone is on the right. 33 s f e e f f i s ÿ : a c . t s .: ; 100 pm Figure 2.28. Fusion boundary region in Varestraint sample of alloy 8090. Arrows indicate the location of fusion boundary. Fusion Zone > . 100 pm Figure 2.29. Solidification crack along EQZ close to fusion boundary. Alloy AF/C489- Longitundinal Varestraint test (ref.71). 56 2.8. Summary of Literature Li-bearing aluminum alloys can be welded using a variety of processes including GTAW. PAW. LBW. EBW, RW and FW. The use of these alloys is not restricted to certain welding processes and process selection is equivalent to other structural aluminum alloys. A variety of standard (Li-free) tiller metals can be used with these alloys. Li-bearing tiller metals are not generally available, or recommended, because of increased potential for porosity formation. .Al-Cu-Li and .A.l-Mg-Li alloys are more susceptible to weld solidification cracking than comparable commercial alloys that do not contain lithium. By evaluating the individual effects of Li. Mg. and Cu using appropriate binary alloy systems, many of the .Al-Li-X alloys were found to contain levels of alloying addition that increased cracking susceptibility. This is primarily the result of the amount and distribution of eutectic liquid in the weld microstructure, which allows grain boundaries to be wet by liquid films, but is insufficient to provide any crack "healing" effects. Where filler metals are employed, cracking susceptibility can be reduced by the selection of appropriate filler metal compositions. In the Al-Cu-Li system, the use of filler alloys 2319. 4043. and 4047 significantly increase cracking resistance. For the Al-Mg-Li alloys, filler metals with higher Mg content than the base metal appear to improve weldability. Improvement in the inherent weld cracking susceptibility of the Li-bearing base metals will require further alloy development and consideration of the micro structural factors that control cracking susceptibility. Tltese factors include control of the eutectic content of the weld metal through alloy additions and the use of grain 57 refinement by the addition of elements such as titanium, zirconium, and scandium. The presence of an EQZ along the fusion boundary presents another, and more intractable, problem. Because the EQZ most likely forms from the unmixed zone region of the fusion zone, elimination of the EQZ by filler metal additions is not possible. In Al-Li-X alloys containing zirconium, the EQZ has been shown to be highly susceptible to cracking during both fabrication and repair. This aspect of weld cracking susceptibility warrants additional investigation. .Adjusting base metal compositions can infiuence weldability. Increasing copper content above 5 wt% or Mg above 3 wt% will improve weld cracking resistance by promoting more eutectic liquid healing during solidification. However, such an approach may adversely affect other material properties, such as toughness, ductility, strength, or fatigue crack resistance. "Tweaking" of other elements may provide some improvement in weldability. for instance regarding EQZ formation, but these changes may be counter to optimizing mechanical properties. Weld metal strength, although always inferior to that of the parent metal, can also be improved by implementing post weld heat treatment and aging. But again, this is not always feasible especially for field fabrication and large scale assemblies such as cryogenic tankage. Compositional modifications of Al-Li-X for structural applications, such as cryogenic tankage, must consider not only weldability issues such as weld solidification cracking. EQZ cracking. HAZ delamination, etc.. but the implications that such modifications may have on alloy performance. Despite considerable development of Al- Li-X alloys over the past 30 years, the relationship between weld cracking susceptibility 58 and composition is not fully understood and weldability problems persist with these alloy systems. In the absence of any systematic, coordinated study of this relationship, development of truly "weldable" Li-bearing aluminum alloys will be difficult. Improvements in the welded/aged strength of these alloys will greatly expand their engineering usefulness by offering welded aluminum structures that greatly exceed the properties of currently available structural alloys. 59 CHAPTER 3 OBJECTIVES In order to understand the cracking phenomena within the non-dendritic EQZ it is necessity to understand the mechanism by which it forms and establish the parameters that are critical to its presence in aluminum alloy weldments. The objectives of this investigation were 1. To develop a simulation technique in order to facilitate the study of weld fusion boundary structures in aluminum alloys 2. To determine the evolution of the fusion boundary structures of Li-bearing and other commercially available aluminum alloys including those containing Li. Zr and Sc 3. To confirm the heterogeneous nucléation mechanism for formation of the non- dendrtitic equiaxed grain in Zr- and Sc bearing aluminum alloys 4. To create a mechanistic model to predict fusion boundary microstructure evolution in aluminum alloys based on alloy composition and substrate condition 60 CHAPTER 4 EXPERIMENTAL APPROACH PROCEDURES 4.1 Materials Commercially available aluminum alloys 2219-18. 6061-16. 5454-H34 and 5454- H32 were investigated along with the .Al-Cu-Li alloy 2195-T8. Cylindrical specimens used for Gleeble^ simulation were 102 mm (4 in.) long and 5mm (0.196 in.) in diameter. Zr-bearing alloy 5087 and Sc-bearing alloy 5025 were obtained as filler metals in bare wire form with nominal diameter 1.2 mm (3/64 in.). The composition of these alloys is listed in Table 4.1. .Alloy Cu Mg Li M n Si C r Zn Z r O th er 5454 - H34 0.057 2.61 — 0.63 0.11 0.072 <0.05 <0.05 Fe 0.23 5454 - H32 <0.05 2.97 — 0.77 0.090 0.079 <0.05 <0.05 Fe 0.28 6061 -T6 0.35 0.93 — 0.15 0.64 0.21 0.17 <0.05 Fe 0.29 2 2 1 9 -T 8 6.3 0.02 — 0.24 <0.2 — <0.1 0.12 V 0.08 2 1 9 5 -T 8 4.0 0.36 1.0 <0.1 0.02 — 0.01 0.14 Ag 0.28 5087* <0.05 4.8 — 0.63 <0.4 — <0.25 0.154 FeO.14 5025* <0.05 5.0 — 0.12 <0.25 <0.12 <0.1 — Sc 0.2 (*) Filler metal Table 4.1. Composition (%wt.) of some commercial aluminum alloys (base and filler metals). 61 4.2 Welding Procedures Experimental alloy compositions were produced by conducting gas tungsten arc welds (GTAW) using 5025 or 5087 filler and 5454-H32 base metals. All welds were produced using a Jetline Engineering sidebeam with carriage, cold wire feed and automatic voltage control (AVC). which enables maintaining a constant arc length. The welds were deposited as a single pass on coupons with dimensions 20.3 cm (8 in.) long by 6.3 cm (2.5 in) wide and 6 mm (0.24 in.) thick. Samples for simulation were removed from these coupons in the manner shown in Figure 4.1 A. A 3.18 mm ( 1/8 in.) diameter. 2% thoriated tungsten electrode with 60° included angle and He shielding gas was used for all welding. Current, voltage and travel speed were held constant for all the welds produced with filler metal additions, at 160 .Amps. 12.5 Volts and 1.7 mm/sec (4 ipm) respectively. Wire feed rate was also held constant at 8.5 cm/sec (200 ipm) for the above mentioned welds. The resulting base metal dilution (BMD) was kept within the range from 45 to 55 %. However, autogenous welds on alloy 2195 were made by adjusting both current and travel speed in order to achieve uniform penetration and avoid excessive melting. Cross welds were produced by overlaying an autogenous weld pass perpendicular to the direction of the previously deposited pass with filler wire (Figure 4.1B). The current and travel speed were again customized in order to achieve constant penetration and sound weld metal characteristics. The welding parameters along with the type of welds produced are given in the table 4.2. 6 2 (A) Cylindrical Specimen. As Cast. Solution Heat Treated - Aged Conditions Weld Metal (B) GTA Cross Weld Figure 4.1. Schematic showing .A) weldment location from which simulation samples removed. B) a cross weld deposited across a previous pass. M aterials C u rren t Voltage Travel Speed W ire Feed BMD BM/FM Rate cm/sec range (Amps) (Volts) mm/sec (ipm) (ipm) (%) 5454H32/5087 160 1.7 (4) 8.5 (200) 45-55 5454H32/5025 5454H32/5087 C ross W eld 15012.5 3.4 (8 ) 5454H32/5025 N/A 100 C ross W eld 2195-Autog. 145 4.2 (10) 2219-Autog. 155 Table 4.2. Welding parameters and resulting BMD range for welds used in the current investigation. 63 4.3 Pickling AH coupons were subjected a surface cleaning treatment three to four hours prior to welding. The aim was to remove most of the tenacious aluminum oxide formed on the surface of the aluminum samples, which is associated with porosity formation in the weld metal. Chemical pickling (ref. 72) was the method employed in order to remove sufficient thickness of material from the specimen surface and it is described as follows: 1. .A 5% wi. aqueous NaOH solution was heated to approximately 70°C. 2. Coupons were placed in the hot solution for approximately one minute. 3. Coupons were then removed form the basic bath and rinsed with cold water. 4. Samples were placed then in a bath of concentrated HNO 3 for approximately 30 seconds to remove the smut that formed during the first step. 5. Samples were removed from the acidic bath and rinsed for approximately one minute in a cold water bath. 6 . Subsequently the samples were air dried and kept sealed in plastic bags until welding. This procedure gave samples a dull and smooth finish with minimal oxide layer on their surface. 4.4 Heat Treatments Heat-treating was conducted in a laboratory furnace. The furnace features an alumina tube that is heated from the outside by heating coils. The power of the coils is 64 regulated electronically by a control unit connected to a thermocouple located midspan outside the tube. For better control of the specimen temperature, another thermocouple was inserted midspan inside the tube and its temperature was continuously monitored. The specimen is loaded in an alumina boat inside the tube. Argon was allowed to flow at a minimal rate from one end of the tube in order to avoid air entering in it. Cylindrical samples were used to evaluate the effects of an as-cast/solution heat treated substrate on the Gleeble^ and they were removed from the weld zone (Figure 4. l.A) of weldments made with base metal 5454-H32 and either 5087 or 5025 filler metal. Welded coupons of the same base metal alloy and filler metal combinations were also heat treated to evaluate the effects of substrate conditions during actual welding practice. The following table 4.3 summarizes the heat treatment schedules used in this studv. Material 5454-H32/5087 5454-H32/5025 2195-T8 BM/FM (45-55 % BMD) (45-55 % BMD) Temperature 540 300 540 300 600 625 (°C) Time 4 5 4 5 4 4 (hrs) Table 4.3. Summary of the different materials and heat treatment schemes used in the current investigation. 65 The specimens were quenched immediately after removed from the furnace and were used in the solution annealed/aged condition. 4.5 Melting Simulation •A. number of models for weld solidification have been developed (refs. 47.73-74). but few experimental techniques to verify these models have evolved. Direct observation of weld solidification has proven difficult due to both the scale of the solidification subgrains (on the order of microns) and the high temperatures associated with the solidification process. Simulation of metallurgical and welding processes has been proven a practical and cost effective means for studying new materials and optimizing both process- and material-related parameters. The Gleeble“ thermo-mechanical simulator has been extensively used to study metallurgical phenomena related to the heat affected zone (HAZ). e.g. hot ductility behavior, precipitation kinetics, stress relief cracking, strain age cracking, constitutional liquation etc. Simulation of metallurgical behavior in the fusion zone has proven to be much more of a challenge using the Gleeble® due to problems associated with containing the molten metal. This can be a particular issue with aluminum, since relatively high 66 currents are needed to heat the samples and maintaining thermocouple contact with the molten metal can be difficult (refs. 75-76). These problems can literally result in the sample "exploding" due to loss of contairunent or thermocouple contact. In this investigation, the use of a simple steel sleeve fitted around the aluminum sample was found to effectively contain the molten aluminum while allowing the temperature of the molten region to be monitored. 4.5.1 Gleeble^ 1500 Characteristics The Gleeble® 1500 thermo-mechanical simulator was used to perform the controlled melting experiments. The gripping (i.e. jaw) assembly within a vacuum, or atmosphere, tank is shown in Figure 4.2. The Gleeble' allows the simulation of complex thermal and mechanical cycles in small samples, while allowing continuous thermal monitoring and feedback control. The specimen is held between two sets of water-cooled grips and heated by its own resistance (I'R heating) by the passage of 60 hertz alternating electric current. An axial thermal gradient is created since the heat generated is being continuously extracted by conduction through the massive copper grips (Figures 4.2 and 4.3). 67 Figure 4.2. View of the atmosphere tank and gripping assembly of the Gleeble® 1500. THERMOCOUPLE STEEL SLEEVE WIRES ilail COPPER JAWS M- FREE SPAN -W Figure 4.3. Schematic showing sample configuration for Gleeble® thermal simulation and the associated axial thermal gradient. 68 Despite the small amount of heat radiation and/or convection from the specimen surface to the surrounding atmosphere, the radial temperature gradient is nearly zero due to uniform heating of the specimen. In addition, metallographic evaluation of isothermal sections revealed no change in microstructure in the melt zone, adjacent to the steel containment sleeve. Based on the small sleeve size (6.35 mm OD) and wall thickness ( 1.1 mm). It IS assumed that the sleeve reaches melt zone temperature. .A.s a result, planes perpendicular to specimen axis are isothermal (Figure 4.4). Temperature monitoring is achieved continuously via a thermocouple attached mid-span on the specimen corresponding to the isothermal plane of maximum temperature. The output of the thermocouple is compared with the instantaneous value of the programmed temperature. The polarity and the magnitude of the error signal are used to determine the amount of the current flowing through the specimen viz.; the specimen heating and cooling rates. The thermal profile generated by a given peak temperature. Jaw separation, specimen size, geometry and. material property is determined by a complex interrelation between the conduction, convection and radiation mechanisms in establishing a thermal balance. 4.5.2 Nil Stremith Temperature Measurement Initially, the nil strength temperature (NST) was determined for each alloy. The NST provides a good estimate of the fusion boundary temperature and was the starting point used for subsequent fusion boundary simulations. The test was performed by heating a sample a rate of 150 °C/sec under a small tensile load (approximately 10 69 pounds). The temperature was increased until specimens failed. The temperature at failure was recorded as the NST. Three NST tests for each alloy were pertbrmed and an average NST determined. Thermocouple Wires Peak Isothermal Plane Figure 4.4. An axial slice of cylindrical sample heated by the Gleeble '. Thermocouple location and isothermal planes are showm. 4.5.3 Controlled Melting Experiments Controlled melting samples were contained within a steel sleeve 2.54cm (1 in.) long during heating in order to hold the molten aluminum in place. Other sleeve dimensions are given in the previous section. Clearance between the 5mm diameter sample and ID of the sleeve was approximately 0.1 mm (0.004 in). A small hole 3 mm (0.12 in) deep by 0.5 mm (0.02 in.) in diameter was drilled midspan in the aluminum specimens and a chromel-alumel thermocouple (0.25 mm diam.) was inserted through a hole machined in the steel tube. Thermocouple wires were mechanically attached to the 70 specimen by inserting them into the small hole drilled in it and mashing aluminum material around them. A ceramic dual-hole tube was slid around the thermocouple wires for even better stability and electrical insulation (Figure 4.5). The specimen assembly located in the atmosphere chamber, was tightly secured in the water-cooled copper jaws (Figure 4.5). Prior to testing, the chamber was evacuated and back- purged with argon to prevent specimens from oxidizing. After repeating the evacuation and argon purging procedure twice, melting simulation tests were performed in the protective atmosphere at nearly atmospheric pressure. .Argon was flowing continuously during test at minimal rate as a means to prevent air from entering into the chamber. Samples were heated to the NST at a rate of 100 ‘^C/sec. Above NST. heating to peak temperature was conducted at a rate of 7 to 1 'C/sec to control the melt zone. Specimens were held at peak temperature for 1 sec and were control cooled at a programmed rate of 100 °C/sec while the melt zone temperature was continuously monitored using the control thermocouple. Some deviations in the actual cooling rate from the program cooling rate were observed, mainly attributable to release of latent heat from the melt. Tests were performed at temperature increments of approximately 10°C above NST (Figures 4.3 and 4.6). .A variety of different heating and cooling rates is possible to be programmed for each specimen configuration. However, the free span i.e. the distance between the surfaces of the two copper jaw sets (Figure 4.3) remained constant at 3 cm (1.2in). 71 CER.\.MIC HOLDER THERMOCOUPLE WIRES I cm I STEEL SLEEVE COPPER JAWS Figure 4.5. Experimental configuration for melting aluminum samples on the Gleeble' To higher peak temperatures V dOfl OS 500 < 400 Cm 300 0 2 4 6 S 10 U 14 16 18 20 22 24 26 28 30 32 34 36 38 40 TIME (sec) Figure 4.6. Typical thermal cycles applied (programmed) during controlled melting of aluminum samples. 72 4.6 Characterization Techniques During welding there is always a transition from the base metal microstructure to that of the weld zone. This change occurs along a narrow region across the weld fusion boundary and the resulting microstructure significantly affects the properties of the weldments. It is important then to characterize the structure within this region in adequate detail and study its composition, grain orientation etc. as a function of base metal dilution, composition and substrate condition. In order to achieve that, a series of characterization techniques were employed. 4.6.1 Metallographic Preparation .\ significant number of cross sections were produced from both simulated samples and actual weldments. The simulation samples were cut axially using a standard wafer cutting saw. For all other cuts, a standard water cooled cut off saw was used. Numerous metallographic samples were prepared by mounting them in cold epoxy- having an average cure time of 8 hours. Following mounting the samples were ground sequentially starting from 600 grit silicon carbide paper and continuing to 800 and 1 2 0 0 . The next polishing stage continued with 5 pm and 1 pm alumina slurry. The last stage of polishing was completed using 0.03 pm colloidal silica dispersion on a vibratory polisher for 40 minutes. Electrolytic etching was found to give the best results under examination in polarized light. For that reason an aqueous solution 5 volume percent of fluoboric acid 73 (Barker's etchant) was the electrolyte. This etching was performed in a cell with 20 volts DC and a 0.2 Amps/sec current density for up to one minute depending on the alloy composition. The edges of the simulation samples that carried the steel sleeve around them were covered using masking tape prior to etching. In this way differential etching between steel and aluminum allov was avoided. 4.6.2 Base Metal Dilution Measurement Two sections per weldment were used to estimate the base metal dilution. The sectioned samples were polished successively starting from 600 grit silicon paper, continuing to 800 and finishing with 1200. Macroetching was performed by dipping the samples in aqueous NaOH solution 5%wt. heated up to 70° C. until the fusion boundar}' became visible. Images from the etched samples were taken using a digital system and stored directly into a computer. The electronic images were analyzed using the digital analysis software Scion Image Beta 4.01 (ref. 77). The analysis was based on measuring the weld nugget area of the cross section area and the BMDs were calculated using equation 12 (Figure 4.7). 74 %BMD =*100 (eq. 1 2 ) Fiüure 4.7. Illustration showing the BMD calculation. 4.6.3 Microstructure Characterization Techniques Optical microscopy was carried out at magnifications up to 400x in order to reveal the nature of the resulting microstructures across the fusion boundary. Illumination with polarized light was shown to result in good grain contrast and reveal easily both the dendritic and non-dendritic character of the microstructures along with the eutectic constituents decorating the grain boundaries. Scanning electron microscopy (SEM) was performed on as-polished specimen surfaces with either backscatter (BSE) or secondar}' electron (SE) detectors using a Phillips XL-30 microscope featuring a field emission electron gun (PEG). Typical operating conditions were 10 mm working distance and 10 -15 kV accelerating voltage with an average probe size of 3 }im. 75 BSE was employed to reveal the composition contrast within the microstructures. Energy dispersive spectroscopy (EDS) was employed to study the microstructure composition using 10 -12 kV accelerating voltage. Spot analysis, line scans and X-ray maps were the modes carried out in order to obtain optimum compositional profiles from the various microstructure constituents. 4.6.4 Transmission Electron Microscopv Thin foils were prepared from transverse sections of selected samples for additional analytical work. The sections were removed using a low speed cut off saw to an average thickness of 300 - 400 pm. They were mechanically thinned using 800 and 1 2 0 0 grit silicon carbide papers until their thickness was reduced to 2 0 0 pm. Subsequently these samples were lightly etched using a warm 5% aqueous NaOH solution in order to highlight the fusion boundar}'. Discs 3 mm in diameter were punched from the etched sections along the fusion boundar}'. Thinning continued using 1200 grit SiC paper until average thickness became 100 pm. Initially, electropolishing was selected to be the final thinning process using a twin jet equipment and electrolyte made of 25 % HNOj solution in methanol. Electropolishing was performed at -25°C and 15 volts was the applied potential. However, it was found this technique resulted in either severe attack of the grain boundary constituents or perforation of the sample away from the fusion boundary within the fusion zone. TEM foils were then produced by switching the final thinning process to ion milling. Discs had been already dimpled on both sides using 3 pm diamond paste. 76 Subsequently samples were set into a Gatan dual stage ion mill at 6 kV. approximately 0.5mA per side at 10° angle tor 8-12 hours, until perforation was achieved. .A.nalytical work continued on a Phillips CM-200 microscope featuring a LaBô cathode and 200 kV accelerating voltage. High resolution imaging and precipitate mapping was carried out on a Phillips CM-300 microscope featuring a high brightness tleld emission electron gun (PEG) for high coherence and small probes (5.4) operating at 300 kV accelerating voltage. Bright and dark field conditions were used to highlight microstructural characteristics. 4.6.5 Orientation Imaging Microscopv This is a new SEM based technique whereby the spatial distribution of crystallographic features in a microstructure can be easily mapped (refs.78-79). For example, cri stallographie orientations of individual grains and the distribution of such orientations along with the distribution of the grain boundary misorientations can be easily represented using this technique and linked directly with the location of the corresponding microstructural features. As a result pole tlgures generated using the orientation imaging microscopy (OIM™) technique consist of projections of poles belonging to discreet grains. This is contrasted to pole figures produced using X-ray reflections which are the result of simultaneous diffraction of many grains while the pole 77 figures in this case describe the distribution of the normals of a particular set of planes and not the distribution of individual grain orientations. The electron backscattered diffraction (EBSD) patterns (Kikuchi bands) from bulk specimens (Figure 4.8) convey all the information that is unique to a certain ciy stallographic orientation and crystal symmetry. These bands are analyzed (Figure 4.9) and cry stallite lattice orientation is measured (ref. 8 U) trom their geometry which carries information on the spatial distribution of lattice orientations, including the grain boundary- character and distribution (refs.78-83). Diffracting Electron beam plane Tilted specimen (70°) Kikuchi Bands Phosphor Screen Figure 4.8. Schematic showing specimen arrangement and pattern generation due to backscattered electron diffraction in a bulk sample (ref. 82). 78 Electron beam Phosphor screen Specimen Image enchancement Monitor display of indexed pattern Semi - automatic : x i : pattern indexing Figure 4.9. Illustration depicting the various steps Followed to collect and analyze EBSP signal generated on the surface of a bulk sample (ref.83). Data are collected rapidly on a point-by-point basis (refs.78-79.83) over a selected area of the sample. The points are arranged in a regular grid and either the electron beam of the SEM is programmed to step at each point in turn or the beam is held stationary and the specimen stage is programmed to traverse beneath it. A t each step, the coordinates of the point, the crystal phase and the orientation are stored. These data can 79 be plotted and interrogated in various ways afterwards in order to determine the nature and character of different microstructural features of interest (refs.78-9.83-85). An example is shown in Figure 4.10 where boundaries between 5 and 15"^ are plotted with thin lines while boundaries greater than 15" are plotted with bolder lines icj./fiH.n CI-0H57 'ft nn -w;snn Figure 4.10. An indexed FCC electron backscatter pattern (ref. 8 6 ) In comparison to TEM. which is one technique that combines spatially specific diffraction information with tine microstructural detail. OIM^^ is characterized by easier specimen preparation and ability to directly relate the area of view to the whole 80 specimen. Specimen preparation for OIM™ usually involves a final step of light electrolytic polishing to remove the thin deformed layer formed on the specimen surface due to previous mechanical polishing that may obscure the low intensity EBS diffraction patterns (EBSP). However TEM is the recommended technique when high spatial and angular precision (less than 1°) is needed (ref.83). Specimen preparation for OIM in this study included a final step of electropolishing after the mechanical polishing steps described in section 4.6.1. A Struers Lectropol-5 polishing unit equipped with automatic tracking of voltage, current and electrolyte flow rate and temperature was employed. Electrolyte was 25% HNO 3 in methanol held at a constant temperature of 20°C. The applied voltage was 37 volts and electrolyte flow rate was set such as to always maintain contact between solution and specimen surface without any liquid overflow. 81 CHAPTER 5 RESULTS Melting simulation was performed on the Gleeble' with the intent of accurately reproducing the weld fusion boundary microstructures present in welds of aluminum alloys including those alloys forming an equiaxed non-dendritic zone next to the fusion boundary. Such a simulation technique facilitates both studying the nature of such microstructures and understanding their evolution from molten metal and/or parent grains located within the partially melted zone. Development of melting simulation was performed by systematically heating aluminum samples on the Gleeble® at incremental temperatures above the nil strength temperature. The NST values measured for each alloy tested are presented in tables 5.1 and 5.2. AMoy NST (°C) Solidus Liquidus Temperature (°C) Temperature (°C) 2219-T8 540+3 543 643 5454-H34 587+8 602 646 6Ü61-T6 591 ±3 582 652 2195-T8 623+6 540' 640 *DTA Results Table 5.1. Nil strength temperature (NST) and nominal melting temperature (ref.87) for various commercial aluminum- alloys. 82 Materials Temper Condition BMD NST (°C) BM/FM (%) A s W e ld e d 590.3+8.6 5454H32/5087 SHI* at 54U"C tor4h 50 597.5+10.6 A s W e ld e d 582.7+4.5 5454H32/5025 SHI* at 341)'-'C for 4h 3 8 0 Solution annealed and quenched Table 5.2. NST values determined for the experimental compositions produced from welds made usine filler metals 5087 and 5025 and 5454-H32 substrate. 5.1 Microstructures Produced with Controlled Melting on Gleeble® Melting simulation produced a range of microstructures as a function of alloy composition, substrate condition and peak temperature. Optical. SEM. TEM and STEM photomicrographs along with microhardness data from selected samples highlighting regions close to peak temperature are shown in this section. Although simulated bulk fusion zone and HAZ microstructures will be described, emphasis will be placed on the composition and nature of microstructures developed next to the fusion boundary. A non- dendritic equiaxed zone (EQZ) present in certain alloys will warrant special focus. Comparison will be made also between weld and simulation microstructures from alloys having similar composition and temper in order to reveal any differences ancL'or similarities between them. 83 5.1.1 Microstructures in Commercial Aluminum Alloys - As Received Substrate Optical microscopy has revealed that melting/solidification tests on the Gleeble® can reproduce the various regions surrounding the fusion boundary in alloy 2195. In all commercial alloys, except 2195. the melt zone microstructure was dendritic. The wrought base metal microstructure of alloy 2195 is shown in Figure 5.1. In samples heated to the NST ('625°C). gross melting is not observed and the microstructure is similar to the partially melted zone (PMZ) observed adjacent to the fusion boundary. This region, shown in Figure 5.2. exhibited a crack due to the shrinkage strains within the sample during cooling. When heated into the temperature range from approximately 630°C to 640"C. non-dendritic equiaxed growth occurs, as shown in Figure 5.3 A. When heated above 640 "C. standard epitaxial nucléation and growth occurs off the substrate and dendritic microstructures are observed (Figure 5.3 B). Some equiaxed grains maybe observed at the solid-liquid interface. When 6061 is heated above its NST (-591*^0. only epitaxial dendritic growth occurs (Figure 5.4). There appears to be no tendency for non-dendritic EQZ formation in this alloy. Similar behavior was observed for alloys 2219 and 5454. irrespective of the peak temperature applied (Figures 5.5 and 5.6). 84 100 |im Fiuure 5.1 Allov 2 195-T8. Base metal microstructure. Figure 5.2 Weld fusion zone simulation of alloy 2195-T8. Microstructure corresponds to specimen heated to 625 °C. Heating Rate =100'^C/sec, Cooling Rate = dO'^C/sec. Arrows highlight location of solidification crack. 85 Heating Rate -HAZ 100"C/sec. Cooling Rate 65"^C/sec Heating Rate lOO"C/sec Cooling Rate 30"C/sec I-A FUSION ZONE 100 Lim Figure 5.3. Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. A) 630 °C. B) 660 °C. Arrows indicate location of fusion boundar}'. 86 Heating Rate iOO"C/sec ft Cooling Rate 4o'*C/bce FUSION ZONE 100 urn Heating Rate lOO"C/sec Cooling Rate 90“C/sec 100 gm Figure 5.4. Simulated fusion boundary region in alloy 6061-T6 produced using the Gleeble®. A) 645 °C. B) 620 °C. Arrows in (A) indicate location of fusion boundary. 87 Heating Rate 3FUSI0N ZONE 100“C/sec. Cooling Rate 30‘’C/sec Heating Rate IOO"C/sec. Cooling Rate 85"C/sec Figure 5.5. Simulated fusion boundary" region in alloy 2219-T8 produced using the Gleeble®. A) 655 °C. B) 587 °C. .Arrows in (A) indicate location of fusion boundary. 88 Heating Rate lOOV/sec FUSION ZONE f-' 3 Cooling Rate 87‘’C/sec Heating Rate lOOV/sec Cooling Rate 87°C/sec Figure 5.6. Simulated fusion boundary region in alloy 5454-H34 produced using the Gleeble®. A) 623 °C. B) 610 °C. Arrows in (A) indicate location of fusion boundary. 89 Hardness traverses made from the base metal across the fusion boundary for alloys 2219-T8 and 6061-T6 show typical HAZ degradation (softening) due to precipitation dissolution and/or precipitate coarsening (Figures 5.7 and 5.8). The extensive softening observed in the HAZ of alloy 5454-H34 can be attributed to recovery, reciystallization. and grain growth (Figure 5.9). .Alloy 2195-T8 also exhibited a softening reaction in the HAZ due to reversion and overaging. However, a sharp increase in hardness was observed (Figure 5.10) within the EQZ. This hardness increase disappeared in samples heated at higher peak temperatures having a dendritic structure upon solidification (Figures 5.10 and 5.11). This is consistent with hardness traverses reported across the fusion boundary of actual welds in this alloy (Refs. 21. 88). It is proposed that the higher hardness in the EQZ is related to both the finer grain size and the higher proportion of eutectic constituents relative to the epitaxial, dendritic structure. 90 120 IIU Fusion Zone Heat Afrected Zone I u n > S‘l ^ ftU 50 40 30 II : 4 6 Distance from Fusion Boundary (mm) Figure 5.7. Hardness traverse across the fusion boundary for alloy 2219-T8. Specimen was heated at peak temperature of 640°C. 110 Ç Fusion Zone llcat AfFected Zone # 'HI 30 -4 *2 0 2 4 Distance from Fusion boundar) (mm) Figure 5.8. Hardness traverse across the fusion boundary for alloy 606I-T6. Specimen was heated at peak temperature of 645°C. 91 Fusion llcat .VITcctcii Zone Base ■5 50 2 4 6 H 10 i: 14 Distance from Fusion Boundary (mm) Figure 5.9. Hardness traverse across the fusion boundary for alloy 5454-H34. Specimen was heated at peak temperature of 623°C. Heat AfTected /.one ...... ♦ - TceaRs 440 "C ♦ ♦ To*Jk= 430 "C ♦ ♦ u -0 -I 0 I : J 4 5 6 7 S 9 Distance from Fusion Boundars- (mm) Figure 5.10. Hardness traverse across the fusion boundary for alloy 2195-T8. Specimens were heated at peak temperatures of 630°C and 640°C respectively. 92 FUSIO N ZO n F 1 0 0 u m Figure 5.11. Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen heated to 640°C peak temperature. Heating Rate - lOO‘^C/sec. Cooling Rate = 40‘^C/sec. Arrows indicate location of fusion hoLindarv. Table 5.3 provides a summary of the temperature ranges at which either EQZ forms or epitaxial nucléation takes place when the various commercial alloys are melted on Gleeble". NST and liquidus temperatures are also listed for each alloy in order to facilitate the understanding of this thermal effect on the nature of the resulting microstructures. It is important to note that an EQZ formed only in the aluminum alloy 2195-T8 within a relatively narrow temperature range. 93 Temperature for EQZ Liquidus (ref.91) Alloy NST Bulk Melting and Temperature Temperature (°C) Epitaxial Nucléation CC) (°C) (°C) 2195-T8 623 640 625-640 640^*' 2219-T8 540 590-600 Do not form 643 6061-T6 591 620-630 Do not form 652 5454-H34 587 610-620 Do not form 646 (a) DTA Results Table 5.3. Effect of thermal conditions on EQZ formation and epitaxial nucléation for various commercial aluminum allovs melted on Gleeble^. TEM and OIM techniques will help us understand the mechanism by which the non-dendritic equiaxed zone is produced, as it will be described later. This investigation has demonstrated that controlled melting of commercial aluminum alloys can be accomplished routinely and that this method is very useful for studying fusion boundary behavior. This tool is further used to evaluate the combined effect that alloy composition and substrate condition have on the EQZ formation. Previous investigations (refs. 68-69) have associated the EQZ in welds of the alloy 2195 with the presence of elements such as Zr and Li. The effect of Zr will be further explored along with the effect of Sc which is an element added recently to some aluminum alloys in order to improve their strength over density ratio (refs.3.41). Experimental compositions based on Al-Mg-X system were used in this study and produced in the manner described in the section 4.3. 94 The effect of substrate condition is investigated by performing a series of controlled melting experiments on specimens in both as welded (i.e. as-cast) and heat treated (solution annealed or aged) and quenched conditions. Specimens were prepared as described in sections 4.2 and 4.4. 5.1.1.1 Fusion Boundary Microstructures in Heat-Treated Aluminum Allov 2195 .A.S section 4.5 describes, alloy 2195-T8 was solution heat treated at high temperatures for various times. Initially, samples being in the heat treated/quenched conditions (600°C. 15 mins) were tested on the Gleeble®. at 630°C since this temperature was shown earlier to favor EQZ formation in the fusion zone. Figure 5.12-A provides a general view of the type of microstructure produced after testing on Gleeble®. Three general zones can be distinguished viz. unaffected-heat treated base metal, grain growth- heat affected zone and EQZ. Figure 5.12 B illustrates the microstructure produced across the simulated fusion boundaiy. It is interesting to note an apparent region of plane front solidification and then interfacial breakdown present in the HAZ grains. 95 Heat Treated- Grain Growth EQZ Unaffected BM Zone - HAZ 300 |o.m ■pMZ X - • Figure 5.12. Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 600° for 15 m in prior to melting at 630°C. Arrows indicate location of fusion boundary. 96 Alloy 2195-T8 also received two other cycles of heat treatment. Samples held for 4 hours at 600°C and 625 °C respectively, were quenched and then immediately tested on the Gleeble®. Specimens heat-treated to 600°C were melted at 632°C while those heat- treated to 625° C were tested at 640° C on the Gleeble®. The heat treatments resulted in microstructures consisting of huge, precipitate-free grains. A significant amount of second phase (eutectic constituent) was found to decorate the grain boundaries. It is apparent (Figure 5.13) that the typical pancake grain structure present in the alloy 2195- T8. was totally wiped out after the alloy received the above mentioned treatment cycles. In both cases, microstructure within the simulated fusion zone was composed uniformly of non-dendritic equiaxed grains. The large grains located in the PMZ were shown to be decorated by a "neckless" structure, which was made of small non-dendritic grains similar in appearance with those observed within the adjacent EQZ (Figures 5.14 - 5.16). This occurs due to local grain boundary melting and resolidification as equiaxed grains. 97 Figure 5.13. Alloy 2195. Heat-Treated/Quenched substrate condition (600°C. 4h). 98 • ^ à Figure 5.14. Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 600° for 4 hrs prior to melting at 632°C. (A) Large grains in PMZ decorated by a neckless of small non-dendritic equiaxed grains. (B) Arrows indicate location of fusion boundary. 99 ...... PMZ %. m Ê m 25 (jun I Figure 5.15. Simulated fusion boundary region in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 625°C for 4 hrs prior to melting at 640°C. (A). (B) - Large grains in PMZ decorated by a neckless of small non dendritic equiaxed grains 100 . .f c c v 100 p.m Figure 5.16. Simulated fusion boundarv' région in alloy 2195-T8 produced using the Gleeble®. Specimen received a heat treatment cycle at 625°C for 4 hrs prior to melting at 640°C. Arrows indicate location of fusion boundary. The effect of substrate condition is investigated by performing a series of controlled melting experiments on specimens in both as welded (i.e. as-cast) and heat treated (solution annealed or aged) and quenched conditions. Specimens were prepared as described in sections 4.2 and 4.4. 101 5.1.2 AI-Mg-Zr System These samples were prepared using the GTAW process with the combination 5454BM/5087FM. The composition in this alloy system (weld metal) is estimated on the basis that 50% BMD was achieved approximately during the bead on plate welding. Hence, the weld metal contains approximately 3.9 wt% Mg. 0.7 wt% Mn and 0.07 wt% Zr. 5.1.2.1 .As-Welded Substrate Condition Optical microscopy was performed in order to evaluate the nature of the microstructures produced from melting simulation on the Gleeble'®. A typical as welded microstructure of an Al-Mg-Zr alloy produced with 50% BMD is shown in Figure 5.17. This structure is comprised of columnar grains. Specimens heated to a peak temperature close to NST (590°C) exhibited partial grain melting and growth (Figure 5.18). Samples heated at higher peak temperatures exhibited a dendritic structure upon solidification. The types of microstructures developed in this case are shown in Figures 5.19 - 5.22. 102 ^ 100 lim^ Figure 5.17. .\s welded microstructure produced using cold wire GT.A.W and 5454-H32 base metal with 5087 filler metal allov at 50% BMD. M m Figure 5.18. Simulated microstructure produced on the Gleeble® using 5454-H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 590°C (NST) peak temperature. 103 FUSION ZONE Figure 5.19. Simulated fusion boundary region produced on the Gleeble using 5454- H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 605°C peak temperature. Arrows indicate location of fusion boundary. * --w • Figure 5.20. Simulated microstructure produced on the Gleeble® using 5454-FI32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 625°C peak temperature. Arrows indicate location of fusion boundary. 104 L| FUSION ZONE ' HÂZ ' Figure 5.21. Simulated fusion boundary region produced on the Gleeble® using 5454- H32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 63 rC peak temperature. .A.rro\vs indicate location of fusion boundary. Figure 5.22. Simulated fusion boundary region produced on the Gleeble® using 5454- FI32 base metal and 5087 filler metal alloy at 50% BMD. Specimen heated to 660°C peak temperature. Arrows indicate location of fusion boimdary 105 5.1.2.2 Welded and Heat -Treated Substrate Conditions (54Q°C for 4Hrs) Specimens were prepared as sections 4.2 and 4.4 describe. Optical microscopy was used to reveal the type of microstructures produced when these samples were melted on the Gleeble®. The resulting microstructures were similar in nature with those presented in the previous section 5.1.2.1. Partially melted grains were observed when specimens were heated to a peak temperature close to NST (597.5°C). The fusion boundary microstructure consisted of dendrites produced during the cooling stage of the applied thermal cycle when the peak temperature exceeded the NST. Figures 5.23 -5.26 illustrate the observed microstructures. Figure 5.23. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Microstructure corresponds to specimen heated to 605 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Arrows highlight location of liquation crack. 106 w m m m Figure 5.24. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Microstructure corresponds to specimen heated to 616 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. .Arrows highlight location of liquation crack. d ^ 1 4.f ' IjV; FUSION ZONE V U. PMZ 50 p m Figure 5.25. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Microstructure corresponds to specimen heated to 628 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Arrows indicate location of fusion boundary. 107 w m FUSION ZONE Tc ' ; . HAZ Figure 5.26. Weld fusion zone simulation produced on the Gleeble using 5454- H32/5087 at 50% BMD. Microstructure corresponds to specimen heated to 660 °C. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Arrows indicate location of fusion boundary. 5.1.2.3 Welded and Heat -Treated Substrate Conditions (300°C for 5 hrs) Figures 5.27-5.28 are characteristic examples of the microstructures observed when specimens in the weld/heat treated condition (300 °C. 5hrs) were melted on the Gleeble^. In all cases, when samples were heated at peak temperatures beyond the NST. 108 the resulting solidified microstructure consisted of columnar dendrites. This type of heat treatment was selected in order to assess the effect of aging of the substrate have on the nature of the solidifying microstructures. Dendritic structures were formed in all simulation samples after gross melting occurred (Figures 5.27 - 5.28) FUSION ZONE 100 um Figure 5.27. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Microstructure corresponds to specimen heated to 632°C. Prior to simulation the sample was aged at 300°C for 5h after welding. Arrows indicate location of fusion boundary. 109 FUSION ZONE Figure 5.28. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5087 at 50% BMD. Microstructure corresponds to specimen heated to 650 °C. Prior to simulation the sample was solution heat treated at 300°C for 5h after welding. Arrows indicate location of fusion boundary. 5.1.3 .Al-Mg-Sc System These samples were prepared using the GTAW process with the combination 5454BM/5025FM as it was described in the section 4.2. The Composition in this alloy system (weld metal) is estimated on the basis that BMD is approximately 50%. Hence, this weld metal is about 4.0 wt% Mg. 0.5 wt% Mn and 0.1 wt% Sc. 110 5.1.3.1 As-Welded Substrate Condition Figures 5.29- 5.35 show the microstructures produced after samples were melted on the Gleeble^. Optical metallography was employed to characterize the resulting microstructures. Specimens heated to NST exhibited partially melted grains while heating at higher peak temperatures resulted in formation of dendritic structures upon solidification. ' P M Z > 1 . ^ 2 f- , 1 0 0 p m Figure 5.29. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Microstructure corresponds to specimen heated to 580 °C (NST=583°C). Arrows highlight location of a wide liquation crack. I l l ÎL#&- • ' • '. ■\y . r - ; V • Figure 5.30. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Microstructure corresponds to specimen heated to 600 °C (NST=583°C). .Arrows highlight location of liquation crack. K i S i » • PM Z Figure 5.31. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Microstructure corresponds to specimen heated to 605 °C (NST=583°C). Arrows highlight location of a wide liquation crack. 112 Figure 5.32. Weld fusion zone simulation produced on the Gleeble using 5454- H32/5025 at 50% BMD in the as welded substrate condition. Microstructure corresponds to specimen heated to 615 °C (NST=583°C). .A. wide liquation crack can be seen in the upper right comer. ■■jenL '•* FUSION ZONE Figure 5.33. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Microstructure corresponds to specimen heated to 633 °C (NST= 583°C). Arrows indicate location of fusion boundary. 13 m : I " # FUSION ZONE /7 . : Figure 5.34. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Microstructure corresponds to specimen heated to 644 °C (NST= 583°C). .A.rro\vs indicate location of fusion boundary. 5.1.3.2 Welded and Heat -Treated Substrate Conditions (540°C for 4 hrs) Heating samples of this substrate condition at peak temperatures close to its NST (580°C) and up to 620°C. it was found that gross melting did not occur and the resulting microstructure consisted of partially melted grains. Increasing the peak temperature in the range approximately between 628°C and 636°C. it was shown that the entire simulated fusion structure consisted of non-dendritic grains. This grain structure looks similar to the non-dendritic equiaxed zone produced when alloy 2198-T8 was tested on Gleeble® at a comparable temperature range (630°C-640°C). Figures 5.35 -5.40 show the microstructure produced when samples were tested on Gleeble®. 114 Figure 5.35. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Microstructure corresponds to specimen heated to 582 °C (NST=580°C). Figure 5.36. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Microstructure corresponds to specimen heated to 610 °C (NST=580°C). Arrows highlight location of liquation crack. 115 .... . 100 g^~, hSjh I 1 HAZ Figure 5.37. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Microstructure corresponds to specimen heated to 628°C (NST=58G°C). Arrows indicate location of fusion boundary. 116 L ' E Q Z V - ^ 50 |im Figure 5.38. Weld fusion zone simulation produced on the Gleeble® using 5454- H32/5025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Microstructure corresponds to specimen heated to 633°C (NST=580°C). .A.rrows indicate location Figure 5.39. Higher magnification of the EQZ shown in Figure 5.39 117 50 (iin Figure 5.40. Weld fusion zone simulation produced on the Gleeble using 5454- H32,G025 at 50% BMD. Prior to simulation the sample was solution heat treated at 540°C for 4h after welding. Microstructure corresponds to specimen heated to 655°C (NST=580°C). .A.rrovvs indicate location of fusion boundarv. 5.1.3.3 Welded and Heat -Treated Substrate Conditions (300°C for 5 hrs) Selected samples in the welded/heat treated substrate condition (300 °C for 5hr) were tested on the Gleeble^ at temperatures around 630° C. which have been proven to produce EQZ formation in other alloy compositions and/or heats as it was shown in previous sections. Figure 5.41 show microstructures produced after samples were heated to 629°C and 636° C respectively. It is apparent that in both cases the resulting microstructures are fully dendritic and no evidence of non - dendritic equiaxed grains was found. 18 y ï m m m s r ^ ^ FUSION ZONE h r ' W.’ ' V» : • - \ .. > * * c 1. , FUSION ZONE Figure 5.41. Weld fusion zone simulation produced on the Gleeble using 5454- H32/5025 at 50% BMD. Specimens received a heat treatment cycle at 300° for 5 hrs prior to melting. Microstructure corresponds to specimen heated (.A.) to 629° C and (B) 636° C. Arrows indicate location of fusion boundary. 119 Hardness traverses (Figure 5.42) were taken across the simulated fusion boundary microstructures from two samples, which were subjected to low (300°C-5hr) and high temperature (540°C-4h) heat treatments respectively. Both samples were heated to nearly the same temperature on the Gleeble®. Plotted data show that EQZ had higher hardness, while the sample that produced a dendritic structure exhibited low hardness in the fusion /one. Il ib ulbu inlciesting tu nutc that bulh bUinplcb have nearly identical hardness values in the HAZ. 90 EQZ 35 EFFECT 30 ♦ ♦♦♦ ♦♦ ♦ '5 ♦♦ # • • • • • ♦ ♦ ♦ ♦ ♦ * ♦ S£ • • S 65 • • • « # #g ** ** 4*, •••• • •• FISION e # ZONE 55 • • • 50 45 40 -4 -3 -2 A 0 1 Distance (mm) ♦ WcIdcdy'SMT at 540=C4h-Tp=629=C • W elded/SHT at 300°C. 5h-Tp=628°C Figure 5.42. Hardness traverses across the fusion boundary' for two Sc-containing alloys. Both samples had identical composition but different substrate condition and they were heated to almost the same peak temperature on the Gleeble®. 120 5.1.4 Cross Welds The simulation results stated above were confirmed by producing cross welds. An autogenous GTAW was performed perpendicular to the direction of a previously deposited weld bead. Samples for metallurgical evaluation were removed in the manner described in paragraph 4.3. Micrographs below show the nature of the fusion zone microstructures produced in each case. Heat affected zones in figures 5.43 and 5.44 correspond to prior weld fusion zone structures comprised of twins (refs.89-90) also known as "leather" structure. Cross Weld Substrate ‘S FUSION ZONE v s ■ HAZ Figure 5.43. Cross weld fusion zone microstructure produced on 5454-H32/5087 at 50% BMD in was in as weld/heat treated substrate condition (540°C-4h). Arrows highlight location of fusion boundary. 121 Cross Weld , Substrate ■-'èf- . ’ FUSION ZONE 100 um i Figure 5.44. Cross weld fusion zone microstructure produced on 5454-H32/5025 at 50% BMD in was in as welded substrate condition. Arrows highlight location of fusion boundarv. PASS OVERLAY PASS Figure 5.45. Cross weld fusion zone microstructure produced on 5454-H32/5025 at 50% BMD in was in as welded substrate condition. Arrows highlight location of fusion boundarv. 122 5.1.5 Summary on Microstructures A variety of fusion boundary microstructures were produced using the Gleeble® simulation technique described above to melt aluminum samples. Using this technique it was possible to discriminate between the substrate effects and composition on the resulting fusion zone microstructures. Alloy 2195-T8 in either as received or heat- treated conditions at 600°C and 625°C respectively was found to affect formation of the EQZ upon melting simulation tests. The same result was observed whenever Sc-bearing alloy 5454/5025 was employed in the weld/heat treated condition (540°C). Using the same alloy either in the as-welded or weld/aged (300°C) substrate conditions did not promote formation of EQZ in the fusion zone at any peak temperatures. Zr-bearing alloy 5454/5087 was found to consistently result in dendritic fusion zone microstructures in either as welded or weld/heat treated conditions. In addition, cross welding experiments showed good agreement between simulation and actual fusion welding microstructures. It is interesting to note that an EQZ was produced in the alloy 2195 when simulation samples were tested on the Gleeble^ in the temperature range from 625°C to 640°C (Table 5.3). Increasing the peak temperature above 640°C resulted in formation of dendritic structure in the fusion zone upon solidification. In contrast, melting simulation of the Al-Mg-Sc specimens (weld/heat treated substrate condition) at peak temperatures exceeding the 625°C always resulted in EQZ formation. 123 5.2 Element Distribution Profiles and Composition Analysis Energy dispersive spectrometry (EDS) in SEM was performed in order to obtain data regarding the element distribution within the microstructures produced. Line scans. X-ray maps and X-ray spot analyses along with the use of a backscattered electron detector (BSE) were the techniques employed to collect numerous composition data and element distributiun piufiles. Allhough qualitative in nature, data from X-ray spot analysis can be used for semi-quantitative purposes. Emphasis is placed on the non-dendritic equiaxed structures produced in certain commercial alloys and experimental compositions. Data from selected dendritic structures are also included for comparison with the non-dendritic equiaxed structures. Figures 5.46 and 5.47 are BSE images revealing that the non-dendritic equiaxed grains are decorated by a second phase containing atoms with higher atomic number when compared to the aluminum matrix. EDS analysis shows that this phase is rich in copper and depleted in aluminum (Figure 5.48). .Analysis at higher magnification shows the EQZ grains are decorated with either lamellar or divorced eutectic (Figure 5.49). This type of eutectic is known to form during the last stages of solidification in aluminum alloys (refs.47. 91). It is interesting to note that the as-solidified dendrites generated by heating samples of the alloy 2195-T8 at 655°C on the Gleeble'®. exhibit a similar behavior with that one described above. Columnar dendrites are decorated by a copper rich eutectic phase as BSE images (Figures 5.50 and 5.51) and EDS analysis indicate (Figure 5.52). For comparison purposes EQZ and dendritic structures from an actual weldment are also shown in Figures 5.53 and 5.54. 124 Figure 5.46. BSE photomicrograph of the simulated fusion boundary microstructure in the alloy 2195-T8. Specimen was heated to 632°C on the Gleeble®. White areas correspond to copper rich phases while the darker regions are aluminum rich matrix. Arrows indicate location of fusion boundary. 43 m y \c c V bpol Ma Figure 5.47. BSE photomicrograph of the simulated EQZ in the alloy 2I95-T8. Specimen was heated to 632°C on the Gleeble®. Notice white phase decorating the non-dendritic (dark) grains. 125 % , A '--' r - i . j . v a i 0 um LazaeNa BSE Image Cu L y ; - 'r - A . 1» um :### A1 K Figure 5.48. X-Ray maps of the simulated EQZ in the alloy 2195-T8. Specimen was heated to 632°C on the Gleeble®. Notice the copper segregation at grain boundaries. 126 V Acc V Spot Magn Dot WO 10 0 kV 3 0 400ÜX B S t 10 ^ Figure 5.49. BSE photomicrographs of the simulated EQZ in the alloy 2195-T8. Specimen was heated to 635°C on the Gleeble®. Notice both the divorced and lamellar eutectic decorating the non-dendritic (dark) grains. 127 FUSION ZONE Figure 5.50. BSE photomicrographs of the simulated dendritic structure next to the fusion boundar)' in alloy 2195-T8. Specimen was heated to 650°C on the Gleeble®. Notice the white phase decorating the dark grains. Arrows indicate location of fusion boundary. FUSION ZONE Figure 5.51. BSE photomicrographs of the simulated dendritic structure in the alloy 2195-T8. Specimen was heated to 650°C on the Gleeble®. Notice the white phase decorating the dendritic (dark) grains. 128 FUSION ZONE 20 um Figure 5.52. BSE photomicrograph (top) and X-Ray maps (bottom) of the simulated fusion zone in the alloy 2195-T8. Specimen was heated to 650°C on the Gleeble®. Notice the copper segregation at grain and subgrain boundaries. 129 Figure 5.53. BSE photomicrographs of actual weld fusion boundary- microstructure in alloy 2195-T8. Notice again the white phase decorating the dark grains in HAZ and EQZ. .Arrows indicate location of fusion boundary. m FUSION ZONE YY»:- Y-- V. ■; Acc.V Spot Maqn Det WD Lxp ------1 20 pm lOOkVSO IQOOx BSE 9 9 I COLUMNAR OENRir-ALLOY 2195 Figure 5.54. BSE photomicrographs of actual weld fusion zone microstructure in alloy 2195-T8. Notice again the white phase decorating the dark dendrites. 130 After solution heat-treating alloy 2195-T8 at 600°C or 625°C for 4 hours a new microstructure was produced as was described in the section 5.1.1.1 (Figure 5.13). The second phase that surrounds the big grains in this new structure are also found to be rich in Cu (Figure 5.55). The cellular character (Figure 5.55-B) observed at higher magnifications in this phase, indicates that partial melting has taken place around the copper rich grain boundary during the last stage of heat treatment. Table 5.3 includes semi-quantitative data representing the composition of the different microstructure constituents mentioned above. The EQZ grains composing the "neckless" structure shown in Figures 5.14-5.16. were found to be decorated by a eutectic phase rich in copper, as analysis with both BSE and EDS detectors reveals (Figures 5.56 - 5.59). In particular, an X-ray line scan taken across two non-dendritic grains shows the characteristic Cu depletion in the grain interior as contrasted with the Cu enriched grain boundary phase (Figure 5.59- B). Lamellar or divorced type of eutectic structures can be observed along the grain boundaries. .A. similar effect was found in the EQZ produced in the Al-Mg-Sc system. Figure 5.60 shows both EQZ and HAZ grains (prior dendrites) decorated by a white phase indicating that it contains atoms with higher atomic number compared to the aluminum matrix. It is interesting to note that the amount of second phase surrounding the prior dendrites in HAZ (Figure 5.60-.A) is significantly smaller compared to that one surrounding the EQZ grains. This has to do with the fact the substrate was in the weld/solution-annealed condition prior to melting on the Gleeble®. It is 131 1 .Arc V S|»>l n-r Wl) \ XJ. ------1 ;’()0|itii i>»i)kV ;h> 10(1* ir.j U)H o .11 hit t.iun; tor A h A Acc: V spill M.kjii n.ti Wl) I *p ------1 s pm 1h0kv:tl) SOdO* MSI 1UK 0 ?l*)h M ll.it t.UO'C 1 h (lit to o u m Figure 5.55. BSE photomicrographs (A) - (B) and X-ray line scan (C) of weld/solution heat-treated alloy 2195-T8 (600°C-4hrs). The white phase decorating the grains exhibits an enrichment in copper. Figure B shows the cellular structure of the eutectic (white) phase. 132 Acc V Spot Magn Del WD ------1 200 pm 15 0RV30 I50x BSE 9 7 GL_2195 SHTat 600‘C -1 h MAZ Figure 5.56. BSE photomicrographs of the simulated fusion boundary microstructure in weld/heat treated alloy 2195-T8 (600°C-4hrs). Specimen was heated to 632°C on the Gleeble®. Notice small non-dendritic equiaxed gains decorating the large dark grains in HAZ. Arrows in (A) indicate location of fusion boundary. 133 Acc / tipot Magn J e t Wl) I------1 to o pm 12 0 kV 3 0 25ÜX B l F 9 A 2l95_SHTat 625'C_'I ti 40 um 40 um Figure 5.57. BSE photomicrograph (A) and X-ray maps (B) - (C) of weld/solution heat-treated alloy 2195-T8 (625°C-4hrs). The white phase decorating the large grains is rich in copper. 134 Acc V Spot Maqn Dot wD **l?0kV30 ?00x RSr 100 CA 2105 r,flTnt6?5-C 4 ti Figure 5.58. BSE photomicrographs of the simulated to fusion boundary microstructure in weld/heat treated alloy 2195-T8 (625°C-4hrs). Specimen was heated to 640°C on the Gleeble®. Notice small non-dendritic equiaxed gains decorating the large dark grains in HAZ. Arrows in (B) indicate location of fusion boundary. 135 Figure 5.59. BSE photomicrograph (A) and X-ray line scan (B) in the simulated fusion boundary microstructure. Specimen was alloy 2195-T8 in weld/heat treated condition (625°C-4hrs) and it was heated to 640°C on the Gleeble®. The white phase decorating the large grains is rich in copper. 136 reasonable to assume that a significant portion of the second phase at grain boundaries was dissolved during the heat treatment process. EDS mapping shows that the white phase, which decorates the EQZ grains, contains significant amount of iron as well as some Mn. .A,t the same time the grain boundaries are shown to be rich in Mg atoms while the grain interior is relatively depleted. .A.luminum depleted along with Mg enriched areas better define the grain boundaries on the EDS map since they are not clearly identified using the BSE detector only (Figure 5.61). Table 5.3 shows semi-quantitative data regarding the composition of the phases mentioned above. Data were generated using EDS-spot analysis. .A. similar phenomenon was observed in the columnar structure produced in the .Al-Mg-Zr system. HAZ grains (prior dendrites) are decorated by a smaller volume fraction of eutectic when they compared with the as solidified columnar structure (Figure 5.62). Solution heat treatment immediately after welding may account for the small eutectic fraction in the HAZ as it was explained previously. The eutectic phase (white) contains heavier atoms than the aluminum matrix (dark) does. 137 Figure 5.60. BSE photomicrographs of the simulated to fusion boundary microstructure in weld/heat treated alloy 5454/5025 (540°C-4hrs). Specimen was heated to 655°C on the Gleeble®. Notice small non-dendritic equiaxed gains decorated by white eutectic phase. Arrows in (A) indicate location of fusion boundary. 138 A ^-V Spor Magn Dot ,WD Exp 15.0kY3.0 BSE M 6.2 5G/Sc/SHT^40.*ï5üh-Tp-632'C ■ WÊ WÊ 10 u ni f Ê Ê M gK Figure 5.61. BSE photomicrograph (top) and X-Ray maps (bottom) of the simulated fusion zone weld/heat treated alloy 5454/5025 (540°C-4hrs). Specimen was heated to 632°C on the Gleeble®. Notice the Mg and Fe segregation at grain at grain boundaries. 139 FUSION ZONE FUSION ZONE Figure 5.62. BSE photomicrographs of the simulated to fusion boundary microstructure in weld/heat treated alloy 5454/5087 (540°C-4hrs). Specimen was heated to 640°C on the Gleeble®. Notice dendritic gains decorated by white eutectic phase. Arrows in (A) indicate location of fusion boundary. 140 Alloy T yp e - Composition (Selective Elements) Substrate Condition wt.% (at.%) Grain Location Grain Interior Grain Boundary 2195-T8-Gleeble Cu (not detected) Cu 49 (29) EQZ Tp= 632 HAZ (pancake) Cu 3.5 (1.4) Cu 35 (19) 2195-18-Gleeble Fusion Zone Cu (not detected! Cu 36 (20! Tp= 650 'C (Dendrite) EQZ Cu 1.8 (0.8) Cu 50 (30) 2195-T8 .As welded Fusion Zone -- Cu 54. (33) (Dendrite) BM5454/FM5025 EQZ Mg 4 (5) Mg 3.2-8 (4-9.4) weld/heat treated Fe 14-23 (7-13) 540°C-4h-Gleeble Mg 2-5 (2.5-5) Tp=629°C HAZ Mg 3 (3.4) Fe 12-18 (6-10) Mn 4-8.6 (2- 5) BM5454/FM5025 M g 2 (3) weld/heat treated Fe 23 (12.5) 540°C-4h-Gleeble EQZ Mg 3 (3.3) Tp=655°C BM5454/FM5025 As welded /Cross weld HAZ Mg 5 (5.5) Fe 21 (12) Mn 8 (4.7) Fusion Zone Mg 4 (4.5) Fe 22 (12.6) (Dendrite) Mn 7 (4) Table 5.4. EDS spot-anaiysis results produced in selected fusion zone microstructures. 141 5.2.1 Compositional Effect of the Steel Sleeve/Crucible There was not found evidence of interaction between steel crucible and molten aluminum phase. Besides, aluminum specimens were tested in all cases at temperatures below 660°C. which is the theoretical melting point of pure aluminum. This temperature is well below the melting temperature of steel. Figure 5.63 clearly shows that there is not any iron diffusion in the simulated aluminum fusion zone. However, it is possible a small amount of iron diffusion not to be detected bv BSE. Figure 5.63. BSE photomicrographs of the simulated to fusion boundary microstructure in weld/heat treated alloy 5454/5025 (540°C-4hrs). Specimen was heated to 632°C on the Gleeble®. Notice the steep composition gradient between the white Fe-bearing sleeve and the darker aluminum phase. Arrows indicate location of Al/Fe interface. 142 5.2.2 Summary on Composition Analysis BSE images showed that dendritic structures in the fusion zone are decorated by a second phase, eutectic in nature. In the case of alloy 2195-T8 the eutectic phase surrounding the dendrites was found to be rich in copper. Non-dendritic equiaxed grains were also decorated by a phase rich in copper, although this was not continuous around the grain. Li cannot be detected using EDS since the X-rays emitted from its atoms are ver>' soft. Mg-bearing alloys exhibited a similar trend. Both dendrites and equiaxed grains in these alloys, when observ ed, were decorated by a discontinuous phase, rich in Fe. EDS maps have revealed Mg solute enrichment along grain boundaries in the EQZ. This was shown to be continuous around the equiaxed grain boundaries. In addition. EQZ grain interior in both 2195 and Sc-bearing alloys was depleted from solute. No significant differences in the composition of grains located in the fusion zone and HAZ were detected by EDS spot analysis. 143 5.3 Analytical Electron Microscopy Results Advanced characterization techniques were employed in order to identify crystailographic characteristics like grain orientations (microtexture) or precipitate type and structure within the microstructures generated. TEM and High Resolution STEM analyses were performed on selected samples. 5.3.1 .Analysis of Electron Backscattered Diffraction Patterns (QIM*^^') In an effort to identify potential orientation relationships among the grains located across the fusion boundary, samples were evaluated using and included both EQZ and dendritic microstructures. The step size used is shown on each picture and was in the order of 1 pm. Some variations occurred in an effort to achieve optimal resolution within a reasonable data collection time. Data are presented as image quality maps with grain boundaries superimposed as lines with different thicknesses. Low angle boundaries with up to 5° missorientation are highlighted with thin lines while high angle boundaries with misorientations greater than 15° are highlighted with bolder lines. Inverse pole figures (IFF) were created for each area mapped, revealing the distribution of certain crystailographic directions per grain [<111>. <100> and <101>) relative to a pre-selected direction in the bulk sample. In this way individual grain orientations can be easily viewed along with their spatial specificity. In some instances, pole figures were also created using a small number of selected grains since this is the most direct and simple means of displaying the statistics of the orientations of individual grains. 144 Figure 5.64 is an illustration of the as-electropolished surface of a cross section taken from a 2195-T8 GTA weld. Both the pancake grain structure in the HAZ and the EQZ formed next to it are shown in these pictures. Figure 5.65 exhibits the results obtained using the OlAH" from the region shown in Figure 5.64. High angle boundaries have been superimposed and are shown with a light (yellow color). In the present investigation boundaries having more that 15° misorientations will be highlighted with light (yellow) color while boundaries with smaller misorientation angles will be shown as dark (green or red) lines. Figure 5.66 shows the [001] inverse pole figure map illustrating the relative orientation of HAZ and EQZ grains along with the unit triangle of the inverse pole figure . which is the key used to read the map. It is interesting to note the random orientations among the EQZ grains, which are decorated by high angle boundaries. HAZ grains are covered with a similar color indicating similar orientations among them. Figure 5.67 shows the as-electropolished surface of an axial section removed from a simulation sample. This sample was solution heat treated at 600°C for 4 hours and then was heated on the Gleeble' to 632°C. producing the microstructure type shown in Figure 5.14. O I\r^' results regarding the grain orientations are shown in Figure 5.68. which is the IFF map corresponding the microstructure illustrated in Figure 5.67. It is apparent from this image that non-dendritic grains are randomly orientated again and decorated by high angle boundaries. This is clearly seen in the pole figures shown in Figure 5.69. With reference to the pole location of the huge black HAZ grain (colored blue) on the crystailographic projection, the small equiaxed grains are rotated randomly. 145 Figure 5.64. SE photomicrograph showing fusion boundary' microstructure in autogenous GTA weld in alloy 2195-T8. Note EQZ zone and typical pancake HAZ grain structure. As-electropolished surface. Arrows indicate location of fusion boundary 146 20.00 pm = 40 ste p s 20 CO pm = 40 steps Image Quality Rotation Angle Boundaries 30.10 - 152.90 2.0 5.0; 5.0 15.0; 15.0 180.0; Figure 5.65. Image quality grain map of fusion boundary microstructure in autogenous GTA weld in alloy 2195-18. Note the high angle boundaries decorating the entire number of the non-dendritic grains in EQZ. Arrows indicate location of fusion boundary. 147 20.00 pm = 40 steps Figure 5.66. Inverse pole figure plotted the entire grain population shown in Figure 5.65. Results show random orientations ofEQZ grains and similar orientation of grains within HAZ. Arrows indicate location of fusion boundary. 148 Figure 5.67. Two SE micrographs of the simulated fusion boundary microstructure in weld/heat treated alloy 2195-T8 (600°C-4hrs). illustrating the distortion introduced in the image by tilting the specimen for EBSP. A) specimen tilted 70° and B) flat specimen. Sample was heated to 632°C on Gleeble® As-electropolished surface. Arrows indicate location of fusion boundary. 149 EQZ HAZ Boundary levels 5 O' 15 O' 60 00 ym = Ô0 steps lPFMap[001| [001] 001 101 Figure 5.68. Inverse pole figure plotted the entire grain population shown in Figure 5.67-B. Results show random orientations of EQZ grains. 150 Bourcar/ 5 O' ’5 C 60 CO k.m s M 4tec9 100 110 =D =D Figure 5.69. Discreet pole figures showing individual orientations by color of small non- dendritic equiaxed grains located adjacent to big HAZ grain shown in Figures 5.67-B and 5.68. 151 Figure 5.70 presents both the image quality and the inverse pole maps for a cross section taken from a 5454/5025 GTA weld. It is interesting to note that the fusion boundary on the IFF image is crossing the long dendrites, which start from the HAZ and continue towards the fusion zone. This is indicative of epitaxial growth. In this case, since fusion zone dendrites and adjacent HAZ grains have identical orientations, they appear having the same color on the IFF image. As a result they will be shown as one continuous long grain. The same effect is shown in Figure 5.71 for a simulation specimen of the combination 5454/5087 (as welded substrate). The sample was in the as welded substrate condition before heating to 630°C on the Gleeble^. Figure 5.71-.A shows the as- electropolished surface while Figure 5.71-B illustrates the image quality map based on the 0 1 \r^' analysis. Again, the fusion boundary crosses the long dendrites. These dendrites start off the HAZ and are colored with a single color on the IFF image (Figure 5.72). Epitaxy can explain this phenomenon, since both HAZ grains and adjacent dendrites have identical crystallographic orientations. Figures 5.73 - 5.74 also illustrate the effect of epitaxial growth in a simulation sample of the combination 5454/5087 in the weld/heat treated substrate condition (540'^C-4h). As explained above epitaxy accounts for the fact that fusion boundary crosses the long single colored grains on the IFF image. 152 4 FUSION ZONE Boundary levels. 5 O' 5 O' 15 0 4$ CO urn - 45 steps Continuous lO 0 0 24 7 (B) [001] 111 HAZ 001 101 FUSION ZONE 45 X ym = 45 steps iP^ Map [001] Figure 5.70. Image quality grain map (A) of fusion boundary microstructure in 5454/5025 GTA weld and inverse pole figure plotted the entire grain population (B). Note elongated grains grown epitaxialy from HAZ towards fusion zone. Dot line highlights fusion boundary. 153 (A) FUSION ZONE m FUSION ZONE Ü Boundary levels: S 0* l S G* 60 X urn = 60 steps Continuous >Q 3 9 49 2 Figure 5.71. As-electropolished surface (A) and image quality map (B) of fusion boundary microstructure in Gleeble® sample. Specimen was in as welded 5454/5087 substrate condition and was heated to 630 °C on the Gleeble®. 154 FUSION ZONE HAZ 60 00 tifn = 60 steps IPF Map [001]* [001] Figure 5.72. Inverse pole figure plotted the entire grain population shovvn in Figure 5.71-B. 155 Finally, a situation similar to that one observed in the EQZ of the alloy 2195. is shown in Figures 5.75 through 5.76 for an Al-Mg-Sc simulation sample. The specimen was in the weld/solution heat-treated substrate condition before melting on the Gleeble® at 629°C. EQZ grains in the fusion zone appear to be randomly oriented and separated with high angle boundaries. Epitaxy is not shown since the HAZ grains have different orientation from the adjacent non-dendritic grains as the IPF image reveals in Figure 5.761. 156 • \^ - ' '-*' «i* "-I* *-0 ' *. ,-w* T * - . '- * » « . *• *•■.•■.»•••-..--' ^ .1 ^ • • - ' f ~ ; - . • ■.■: ' T ^ . '-I*.-i '.fj'" .'V" HAZ . ^ . -1 & cc V Spot Magn Det WD Exp 100 |im 20 0 kV 5 0 200x S E 16 8 0 (B) _ FUSION ZONE gcurdarv 'cvel« à 0" î5 0‘ 1C5 0 jm : îî e » Ccntnuous G ^ 5 '36 2 Figure 5.73. As -eiectropolished surface (A) and Image quality map (B) of fusion boundary microstructure in Gleeble® sample. Specimen was 5454/5087 in weld/heat treated substrate condition (540°C-4h) and was heated to 629°C on the Gleeble®. 157 f m Ê t FUSION ZONE 105 0 pm : 70 steps IPFMap(10O] [100] Figure 5.74. Inverse pole figure plotted the entire grain population shown in Figure 5.73-B. Note elongated grains grown epitaxialy from FIAZ towards fusion zone. Dot line highlights fusion boundary. 158 V-r^- - HAZ EQZ W i g » . : . cc V Spot Maqn Det WD Exp 20 0 kV 5.0 250k SE 17 I 0 W EQZ i m Boundary levels: 5.0° 15.0° 60 00 pm = 60 steps Continuous IQ 18.0.. 179.6 Figure 5.75. As-electropolished surface (A) and image quality map (B) of fusion boundary microstructure in Gleeble® sample. Specimen was 5454/5025 in weld/heat treated substrate condition (540°C-4h) and was heated to 629°C on the Gleeble®. 159 -# 60.00 pm = 60 steps IPF Map [001] [001] Figure 5.76. Inverse pole figure plotted the entire grain population shown in Figure 5.75. Results show random orientations of EQZ grains and similar to the situation shown in figures 5.66 and 5.68. Arrows indicate location of fusion boundary. 160 5.3.2 TEM and High Resolution TEM Analysis Transmission electron microscopy was performed on foils removed from a location close to weld fusion boundary of a 2195-T8 autogenous GTA weld. Figures 5.77 - 5.78 are TEM bright field images and show the non-dendritic grain structure in the EQZ. It can be seen that grain boundaries are decorated with a second phase, which in some cases has been preferentially attacked during eloctropolishing (Figure 5.78). Figure 5.77. TEM Bright-field image illustrating the EQZ microstructure in autogenous GTA weld of the allov 2195-T8. 161 « Figure 5.78. TEM Bright-Held images illustrating the EQZ microstructure in autogenous GTA weld of the alloy 2195-T8. Second phase decorating the non- dendritic grain is also shown (right). Scanning - transmission electron microscopy was performed at very high magnifications. The aim was to detect Zr-bearing intermetallic particles within the aluminum matrix. Figure 5.79-A shows such a particle in a grain located in the EQZ. EDS maps reveal that it contains Zr and Cu. TEM bright field and SAD images from this particle are shown in Figure 5.79-B. An EDS line profile shown in Figure 5.80 reveals Cu solute enrichment at grain boundary separating two non-dendritic equiaxed grains and solute depletion in the grain interior. The Cu composition of the first solid to form as predicted by the Scheil equation for an Al-4wt%Cu binary alloy is also superimposed. 162 50 nm Figure 5.79. TEM dark-field (A) image illustrating a Zr-bearing precipitate in the EQZ microstructure in autogenous GTA weld of the alloy 2I95-T8. (A) EDS map. (B) Bright field and <110>a SAD images from matrix (left) and precipitate (right). 163 10 kCo” 0.68 4 6 8 10 12 14 16 fbsi(ion(imn) 100 % 90 5 10 15 20 Position (un^ Figure 5.80. Copper profile in non-dendritic equiaxed grains of an autogenous 2195-T8 GTA weld. Superimposed is the copper composition for the first solid to form predicted by Scheil equation for an Al-4.0wt% Cu alloy. 164 Needle like precipitates were detected occasionally within the grain interior using both conventional and high resolution TEM analysis. Figure 5.81 is a high resolution TEM image showing such precipitates in the aluminum matrix. Figure 5.81. Precipitates with needle-like morphology were detected in the EQZ grain interior of an autogenous 2195-T8 GTA weld. 165 In an effort to detect AlgZr particles in the EQZ grain interior, precipitation such as the one shovvn in Figure 5.82 was also encountered. Unfortunately, both precipitates showed in Figures 5.81 and 5.82 had very small size (<10 nm) and as a result no EDS analysis could be performed on them in order to obtain information regarding their composition. Figure 5.82. High resolution TEM image showing precipitation within the equiaxed non- dendritic grain of an autogenous 2I95-T8 GTA weld, B= <110>a. 166 Non-dendritic equiaxed grains produced when melting simulation performed on the Al-Mg-Sc system were free from precipitates compared to those in the 2195-T8 weldment. The following figures illustrate the microstructure within an EQZ grain, which was developed when a sample in the weld/heat treated condition (540°C - 4h) was melted on the Gleeble'® at 632° C. Figure 5.83-A shows the precipitation free grain as can be also confirmed by the SAD image next to it (Figure 5.83-B). Figure 5.83. General view of the microstructure in the EQZ grain developed in an Al- Mg-Sc weld/solution heat treated sample melted on the Gleeble® at 632°C. (A) Bright field image. (B) SAD image along the <110>a zone axis. 167 The same effect is shovvn in Figure 5.84. After long exposure times no evidence of extra reflections was found on the SAD image indicating precipitation. 100 nm Figure 5.84. General view of the l ucrostructure in the EQZ developed in an Al-Mg-Sc weld/solution heat treated sample melted on the Gleeble® at 632°C. (A) Bright field image, (B) SAD image taken along the <110>a zone axis after long exposure time. 168 Figure 5.85 illustrates the second phase shovvTi to decorate some of the grain boundaries of the EQZ grains. It was previously shown that this phase was detected to be rich in iron. ... 200 nm Figure 5.85. TEM bright field images showing secondary phase along EQZ grain boundaries. Sample was Al-Mg-Sc alloy in the weld/heat treated condition prior to melting on the Gleeble® at 632°C. 169 Absence of any detectable precipitates was also found when high resolution TEM analysis was performed as Figure 5.86 indicates. Figure 5.86. High resolution image taken from the interior of an EQZ grain. Sample was Al-Mg-Sc alloy in the weld/heat treated condition prior to melting on the Gleeble® at 632°C. 170 5.3.3 Summary of Analytical Electron Microscopy Results It is interesting to note that OIM™ analysis shows there is random distribution in crystal lographic orientations among EQZ grains and those located in HAZ. High angle boundaries separate the non-dendritic equiaxed grains from the HAZ grains located next to them. Whenever the fusion zone is comprised of dendrites, it was shown that epitaxy is the dominant nucléation mode. Dendrites located next to fusion boundary appear to have the same ciy'stallographic orientation with the neighboring HAZ grains. Consequently, both the fusion zone dendrite and its neighbor HAZ grain are highlighted with the same color on the IPF image. .Although Sc- and Zr - bearing precipitates were not easily detected using TEM analysis, it was shown in one instance that such a particle has ternary composition and complex crystal structure. Cu is shown to be part of its composition along with Zr and potentially Li atoms. Sc-containing precipitates were not detected in the EQZ formed in the .Al-Mg-Sc system even when atomic resolution imaging was employed for this purpose. X-ray microdiffraction line scans showed that the interior of non-dendritic equiaxed grains was depleted in solute, while some solute enrichment was evident at grain boundaries. High amounts of Cu were detected in the grain boundaries decorating the EQZ grains in the alloy 2195 while non-dendritic equiaxed grains in the Al-Mg-Sc system were decorated by boundaries rich in Mg and Fe as it was shown in the section 5.2. This is consistent with the heterogeneous nucléation theory, which predicts that the EQZ grains are produced through a solidification process. 171 CHAPTER 6 DISCUSSION 6.1 Development Characteristics of the Melting Simulation Technique Discussion will begin with some fundamentals of the simulation melting technique since this is the tool used to produce the majority of the microstructures in the current investigation. In addition, it is important to understand the physical characteristics of the technique along with the limitations imposed in order to help establish the basis for understanding the metallurgical behavior of the systems examined. 6.1.1 Nil Strength Temperature The nil strength temperature is close to the "effective" solidus of a material, since considerable liquid is normally present in the microstructure at this temperature. However, some of the NST values measured in this study are below the nominal bulk solidus temperature reported in Table 5.1. Metallographic studies of aluminum alloys (Refs. 92-94) and in other systems (ref. 63) suggest that the NST closely approximates the temperature at the fusion boundary in an actual fusion weld. Consequently, these temperatures were used as the starting point for the controlled melting experiments. 172 6.1.2 Fusion Zone Simulation Melting aluminum alloys on the Gleeble* has been accomplished in the past using various techniques but presents considerable experimental challenge (refs. 75-76). Previous work using quartz containment tubes with an axial slot machined has proven troublesome since the tubes often cracked during testing. The use of steel tubes in this methodology avoids that problem and provides bener containment if the relative specimen and tube sizes are properly controlled. Since steel has a much higher melting temperature than aluminum, there is no degradation of the sleeve material and no metallurgical interaction has been observed between the molten aluminum and steel (Figure 5.63). .Another advantage is that aluminum has a higher coefficient of thermal expansion than steel over the temperature range from room temperature to 600 °C. allowing the aluminum to expand into the sleeve and provide a tight seal at the ends. Despite these advantages, some problems attributable to melt zone instability were encountered. Initially, it was difficult to contain the molten aluminum in the steel tube, and some "leaking" occurred through the small hole on the sleeve used for thermocouple access. This problem was exacerbated at higher heating rates and prolonged times at peak temperature. Experiments conducted in vacuum aggravated the melt zone instability, possibly due to degassing effects. •Aluminum has low resistance (high conductivity) and thus large currents are required to achieve melting. This in turn, results in strong electromagnetic pinch forces. It has been proposed (ref. 75) that as soon as the molten zone is formed, the pinch forces produce necking of the molten region. Once the necking takes place, the reduced cross 173 sectional area promotes higher current density and. in turn, necking is augmented. In addition, the thermocouple response is slower when suspended in the molten zone than when the thermocouple wires are welded directly to the specimen (ref. 76). Hence, a local instantaneous overheating and melt zone instability leads to removal of a significant mass of molten aluminum out of the crucible. It was found that using slower heating ralcb (Figure 4.6) at the temperature range above NST along with reduced dwell time at peak temperature, minimized that effect and resulted in a less vigorous melting process. Reducing the number of heat cycles after the specimen temperature exceeds NST also helps stabilize the aluminum melting process by reducing the liquid stirring. At temperatures above the NST the liquid state becomes dominant within a narrow temperature range thus, enabling a lower amount of current to flow in order to keep temperature constant. By deleting one. two. or three heating cycles (Figure 6 .1 ) still each operating cycle is at power line frequency. Hence, the approximate value can be decreased by 29%. 43% and 50% respectively, in each case. This feature allows better control on the temperature of the melt since low heating frequency results in both smaller heating current and longer time over which a weak magnetic field occurs, which finally helps preventing molten metal from overheating and removing itself from the steel crucible. The outcome is a robust melting technique accompanied by a high level of success. 174 Cycles O ff I Figure 6 .1. Schematic showing the heating cycles operating on the Gleeble® as sinusoidal function of time. How we can achieve better control over the cooling rate following melting? During melting simulation tests on the Gleeble® it has been observed that the actual cooling rate was slower than the programmed one. This effect was intensified when the peak temperature was significantly higher than the NST. This can be understood if we 175 consider that gross melting occurs if the alloy is heated at temperatures well above the NST. Upon cooling, the latent heat released from the solidified liquid results in slower cooling rates at temperatures close to NST. Fast heat extraction rate is required so that the extra heat released to be efficiently transferred away from the fusion zone enabling the maintenance of the programmed cooling rate. This can be achieved by reducing the free span between the copper jaws, which is expected to produce steeper thermal gradient in the fusion zone; a necessary thermodynamic condition for achieving high cooling rates. External means of quenching can also help in accomplishing the programmed cooling rate. 6.2 Simulation Aided Study of Fusion Boundary Microstructure Evolution .Alloys were selected in this study, to provide a wide spectrum of compositions. The aim was to include commercial alloys such as Mg-containing alloys 6061 and 5454 or the Cu-containing alloy 2219. which are known to produce dendritic microstructures upon solidification in actual welding practice, and compare their metallurgical behavior with that of other less studied alloy systems that form equiaxed structures. Such alloys are those bearing Sc- and/or Zr- additions and are known for their importance in cryogenic and aerospace applications but are plagued by weldability problems during fusion and repair welding. The simulation technique developed on the Gleeble® provides us with a tool that was proven to be very efficient in studying alloy composition, substrate condition and 176 thermal effects in a consistent and systematic way. These are some of the parameters that seem to play important role in the evolution of the fusion boundary microstructures, as it will be discussed below. The good agreement between simulation and actual weld microstructures makes the melting simulation technique even more promising for studies of complex metallurgical problems relating to weld fusion zone microstructure. 6.2.1 Fusion Boundary Microstructure Evolution in Commercial Aluminum Alloys Solidification phenomena and microstructure evolution in multicomponent aluminum alloys are complex since there is always an interaction among the alloying elements. However, alloys such as 5454. 6061 and 2219 are routinely employed in a wide range of commercial applications while minor element additions in them do not exceed 1 wt% altogether (Table 4.1). Hence, solute redistribution in these systems can be treated in the same way it is done for binary alloys. Negligible diffusion in the solid and the other assumptions used in deriving the Scheil equation permit a simple treatment of the problem. \^Tten these alloys are heated on the Gleeble® at temperatures close to NST (effective solidus) grain growth occurs on-heating in the solid state. Increasing the peak temperature a few degrees higher, then some grain boundary liquation takes place. By bringing the peak temperatures well above the NST gross melting is promoted. Solute redistribution affects constitutional supercooling phenomena during the cooling portion of the thermal cvcle. as it was described in the section 2.4.1. This leads to solidification 177 interface breakdown and formation of dendrites. It is interesting to note that dendrites located next to simulated fusion boundaries grow out from HAZ grains towards the inner part of the fusion zone. This is the result of epitaxial nucléation and competitive growth phenomena commonly observed in fusion welds in aluminum alloys. Figure 6.2 below shows schematically the evolution of microstructure next to the fusion boundary in Gleeble samples of commercial aluminum alloys 5454-H34. 6U61-I 6 and 2 2 19-1 S. It should be noted that hardness traverses made from the base metal across the fusion boundary for alloys 2219-T8 and 6061-T6 and 5454-H34 (Figures 5.7-5.9). show HAZ degradation (softening) that is typical for these alloys (ref. 95). HAZ Fusion Zone | HAZ | Base Metal «m s D endrites Figure 6.2. Schematic highlighting fusion boundary microstructure evolution in simulation specimens of alloys 2219-T8. 5454-H34 and 6061-T6. when gross melting occurs upon heating on Gleeble®. 178 6.2.2 Fusion Boundary' Microstructure Evolution in Aluminum Alloy 2195 When simulation tested on the Gleeble® at peak temperatures close to NST. alloy 2195-T8 exhibited recrystallization in the rolled pancake grains, while heating at peak temperatures slightly exceeding the NST resulted in some grain boundary liquation around them. Heating at higher peak temperatures resulted in formation of an equiaxed non-dendritic grain zone in this alloy, which cannot be explained by simple application of constitutional supercooling principles, as it was done above for the other commercial aluminum alloys. Previous investigation on this alloy has proposed that EQZ is the result of a heterogeneous nucléation phenomenon, which takes place immediately adjacent to the fusion boundary (ref.70). Thermally stable intermetallics such as .A.l]Zr and/or .-\.l;(Li\Zri.\) that already exist in the matrix are considered the nucléation sites upon which the a-aluminum phase grows. It is suggested that during welding these particles survive the tlow and thermal conditions within a stagnant thin liquid metal layer next to the fusion boundary (unmixed zone). 6.2.2.1 Heterogeneous Nucléation Mechanism and Thermal Effects on EQZ Formation The ability of Zr-containing intermetallics to act as nucléation sites has been verified in a number of studies (refs.47. 96-97). It is known that when Zr- is added to Mg alloys during casting an equiaxed grain structure is produced with a spherical morphology. Recently C.E Cross et al (ref. 98) have observed formation of a non- dendritic equiaxed zone in autogenous GTA welds of the Zr-bearing alloy 7108. Based 179 on the EQZ geometry and constitutional supercooling hypothesis of Burden and Hunt, he was able to estimate the presence of high undercooling next to the fusion boundary, which was gradually reduced with increasing the distance from the fusion boundary. On the other hand, analysis by F.W Gayle et al (ref. 30) suggests that AHfLixZri.x) intermetallics have LG ordered crystal structure with decreasing particle/matrix misfit as Li amount increases in the precipitate (Figure 6.3). This means that lattice disregistry between Zr intermetallics and a-aluminum matrix is significantly reduced leading to a reduced thermodynamic barrier for heterogeneous nucléation of the aluminum phase on 0.8 0.6 Lattice Misfit 0.4 0.2 - 0.2 0.5 Figure 6.3. Lattice misfit between aluminum matrix a and a'- ALfLLZri.x) precipitate versus Li content in a' (ref. 30). 180 the intermetallic crystal structure. Therefore small undercoolings of the order of magnitude of 1°C can suffice to activate most heterogeneous nucléation sites. The Gleeble ® based melting simulation technique enabled study of the thermal effects on the EQZ formation and shed light on the heterogeneous mechanism from the perspective of substrate condition. Fusion zones produced by melting 2195-T8 samples between 630°C and 640°C consisted uniformly of equiaxed non-dendritic grains. However, increasing the peak temperatures above 640°C resulted in elimination of the EQZ trom the fusion zone. A few non-dendritic equiaxed grains were observed positioned randomly along the simulated fusion boundary The structure was fully dendritic with some dendrites originating from the fusion boundary. A few of the dendrites were equiaxed. It can be assumed based on the previous analysis, that heterogeneous nucléation took place upon cooling the microstructure from peak temperatures between 630°C and 640°C. This solidification phenomenon should be spontaneous resulting on a significantly high number of grains nucleating and growing simultaneously. Hence, the growing grains impinge each other very soon before they develop any substantial subcell structure. Formation of dendritic structure upon cooling from peak temperatures exceeding 640°C can be explained by considering the intense stirring of the molten pool. Since in this case higher temperature levels require increased amount of electric current, the result is an intense magnetic field stirring vigorously the melt zone. This may enhance the dissolution of nucleating particles due to increased diffusion rates. It is also important to note that AT.Zr intermetallics have a DO 23 ordered crystal structure at temperatures 181 exceeding 600°C (refs.30. 99-100). This type of structure is tetragonal and is expected to increase lattice disregistry between a-aluminum matrix and A^lLi^Zri.x). Therefore, both the nucléation efficiency and the number of highly efficient heterogeneous sites should be decreased as peak temperature increases. This means that a small number of grains will eventually nucleate heterogeneously upon cooling. These grains will have more space and hence more time to grow and develop subcell structure, before they impinge another neighboring grain. Depending on the dispersion of the nucléation sites within the fusion zone it will be possible now for the epitaxially grown columnar dendrites to grow a significant distance away from the fusion boundary', before they meet an equiaxed dendrite. Figures 6.4 and 6.5 illustrate the nature of fusion zone microstructure developed upon simulation melting of aluminum alloy 2195-T8. 182 Wâ Eoz Figure 6.4. Schematic illustrating fusion boundary microstructure evolution in simulation specimens in alloy 2195-T8. upon heating on the Gleeble® within the temperature range between 630°C and 640°C. HAZ Fusion Zone HAZ D en d rites Figure 6.5. Schematic highlighting fusion boimdary microstructure evolution in simulatioti specimens of alloys 2195-T8. upon heating on the Gleeble® at peak temperatures exceeding 640°C. 183 It is interesting to note how the increased efficiency of heterogeneous nucléation sites (lower undercooling) widens the range of parameters over which a fully equiaxed grain zone is formed according to the analysis made by J.D Hunt (ref. 101). In his model he obtains a simple expression used to predict when fully equiaxed structures should occur during directional solidification. According to Figure 6 .6 . for increased nucléation potency (lower AT\j) the critical growth velocity over which the columnar to equiaxed transition occurs is decreased. Figures 6.7 - 6 .8 illustrate the effect of the number of nucléation sites and alloy composition on the above range of parameters (growth velocity V and thermal gradient G). E quia.\ed - 2.0 C o lu m n ar O o AT.,=0.75 K -3 1.0 -0.5 0.0 0.5 1.0 LOG (G) Figure 6 .6 . Plot showing velocity V (cms"') versus temperature gradient G (K cm'') for Al-3wt%Cu and N=1000 cm'"* for four different nucléation undercoolings. Solid lines correspond to fully equiaxed transition (ref. 101). 184 E q u iax ed > N=10 N=1Ü- - 2.0 -2.5 C o lu m n ar 1.0 -0.5 0.0 0.5 1.0 5 LOG (G) Figure 6.7. Plot of velocity V (cms‘‘ ) versus temperature gradient G (K cm'') for Al- 3\vt%Cu and ATnj=0.75 K showing the effect of varying number N (cm'') of nucléation sites. Solid lines correspond to fully equiaxed transition (ref. 101). 1.5 E quiax ed - 2.0 O 2.5 C o lu m n ar -0.5 0.0 0.5 1.01.0 1.5 LOG(G) Figure 6 .8 . Plot of velocity V (cms'‘) versus temperature gradient G (K. cm'') for Al- 3\vt%Cu for four different alloy compositions C (wt.%) at N = 1000 cm'^ and ATn=0.75 K. Solid lines correspond to fully equiaxed transition (ref. 1 0 1 ). 185 The analysis made by Hunt can be used to qualitatively predict the type of microstructures formed upon solidification. Although thermal gradient and growth velocity parameters were not measured in this investigation it is interesting to note that the increased nucléation efficiency of Zr - and Sc-bearing intermetallics in concert with their distribution numbers in the molten aluminum will favor formation of equiaxed structure. The absence uf the dendritic character frutn this equia.\ed structure will be discussed in a later section. 6.2.2.] Substrate Effects - OIM™ Analysis Zr-bearing intermetallics preexist in the aluminum matrix. The small amount of Zr solid solubility in the aluminum matrix results in an alloy easily supersaturated upon fast cooling from high temperatures. Thus, upon subsequent heat treatment, hot rolling etc. Zr-containing aluminum alloys form Zr-bearing precipitates within the aluminum matrix. These precipitates have been observed in many Zr-bearing alloys. F.G.Gayle (ref. 30) considers complex intermetallics of the AbCLi^Zri.O type form each time Li is added in the alloy composition as it was mentioned earlier. He reports Li additions increase the volume fraction of the above type of precipitates since part of the Zr atoms in the ALZr structure are replaced by Li atoms. Heat treating the alloy 2195-T8 at temperatures as high as 600°C or 625°C affects the alloy microstructure in various ways. Cu-containing phases in the aluminum matrix like 0' or T| are expected to dissolve at this high temperature (refs. 21. 8 8 ). It can be seen the as quenched structure consists of huge grains, which are free from any surface 186 pits that usually form upon etching on copper containing precipitates. However, Zr- bearing precipitates not only are expected to survive that high temperature level but also an increase in their size or volume fraction should be anticipated since there is more Li available due to dissolution of Li-bearing precipitates such as Ô' or T|. At the same time, significant Cu enrichment seems to take place at grain boundaries. This is expected to decrease the local melting temperature at grain boundaries compared to that of bulk grains. Heating this structure at temperatures slightly above 630°C on the Gleeble® leads to the formation of significant amount of liquid surrounding the HAZ grains. Upon solidification this liquid results in formation of small equiaxed non-dendritic grains decorating the larger grains (Figures 5.14-5.16. 5.56. 5.58). These grains are thought to nucleate heterogeneously on Zr intermetallics already distributed in the aluminum structure. It is very interesting to see that equiaxed grains located next to the large HAZ grains are randomly oriented with regards to the crystallographic orientation of the big grains (Figures 5.68-5.69). The same phenomenon was observed in the EQZ that was formed in a 2195-T8 autogenous GTA weld. HAZ grains were separated from non- dendritic equiaxed grains by high angle boundaries while the EQZ grains exhibited a random crystallographic orientation among them (Figures 5.65-5.66). This result can be easily understood if we assume that the nucléation sites are floating in the liquid aluminum and maintain a random orientation. 187 On the other hand pancake grains in the weld HAZ are colored with a similar purplish color on the IPF image (Figure 5.66). indicative of their similar crystallographic orientation. Although some degree of recrystallization is taking place eg. in some broken pancake grains, this result can be easily understood since the HAZ grains are rolled and a texture is expected to remain in the alloy structure after it is subjected to short heat treating cycle during welding. 6.2.3 Fusion Boundary Microstructure Evolution in .Al-Mg-Zr allovs - Substrate Effects .Alloys of this system were simulation melted in both as welded and weld/solution heat treated substrate conditions. The aim was to identify the effects that substrate condition exerts on the nature of the simulated fusion boundarv' microstructure and compare them with the behavior already observed in alloy 2195. Heating this system in the as welded substrate condition at peak temperatures well beyond the NST results in bulk melting. The fusion boundary microstructure consisted of columnar dendrites grown from HAZ grains towards the inner part of the fusion zone. No evidence of EQZ was observed in the solidified fusion zone of this alloy. This result is expected since Zr-containing intermetallics are not predicted to form in the as welded condition. Zr. which has relatively small diffusivity in the aluminum matrix, is anticipated to remain in solid solution after the fast cooling process during welding (ref. 102). Besides, tests on the Gleeble® were made within few days after welding completion while samples were kept at room temperature. The fact that long columnar grains were observed in the simulated fusion boundary structure is also in agreement with 188 the analysis made by Hunt (ref. 101). Figure 6.5 predicts that the solidified structure will consist predominantly of columnar dendrites in the absence of any nucleating particles. Surprisingly, when specimens in the weld/solution heat treated conditions were melted on the Gleeble® the resulting fusion boundary' microstructure consisted again of columnar dendritic grains. This phenomenon was not expected since the heat treatment cycle was thought to promote formation of AhZr precipitates in the as welded structure. However, formation of columnar dendritic grains epitaxially grown from the HAZ substrate (Figures 5.71-5.72) implies the absence of such nucléation sites. .Although structural analysis was not conducted to detect the amount and distribution of the Zr- containing intermetallics in the substrate, previous work with alloy 2195 has shown reduction in the EQZ size as Zr and Li contents were reduced and below a certain Zr composition this zone was completely eliminated (ref. 70). The above is a notable result referring to the composition effect on EQZ formation, as it will be discussed in a later section. 6.2.4 Fusion Boundarv Microstructure Evolution in .Al-Mg-Sc Alloys-Substrate Effects Alloys of this system were simulation melted in both the as-welded and weld/solution heat treated substrate conditions. The aim was to identify the effect substrate condition exerts on the nature of the simulated fusion boundary microstructure and compare them with the behavior already observed in alloy 2195. Alloys tested had almost identical composition. Sc was included in this study since it is used in many Russian aluminum (refs.3. 43) alloys to improve their strength over density ratio. 189 Heating samples in the as welded substrate condition at peak temperatures close to NST resulted in grain growth or grain boundary melting phenomena. Heating samples at higher peak temperatures resulted in gross melting in the alloy. The as solidified structure consisted of dendritic grains. As was shown, epitaxy was confirmed for the dendrites located immediately next to HAZ grains (Figures 5.73-5.74). Absence of nucléation sites can account for the dendritic nature of simulated fusion boundary microstructure in this system. Similarly to Zr. Sc has low diffusivity in the aluminum matrix. Since Sc has a high solute distribution coefficient in aluminum (k s 0.7) (Figure 2.7) it is expected that the majority of Sc will remain in solid solution form in the as welded structure, after the fast cooling process during weld metal solidification. However, the presence of equiaxed dendrites (Figure 5.34) in the simulated fusion zone may imply existence of a small number of nucléation particles in the liquid metal. Hunt's model predicts that for a small number of effective nucléation sites a mixed microstructure (Figure 6.9) may form upon solidification consisting of both columnar grains and equiaxed dendrites (Figures 6 .6 -6 .7). Melting samples in the weld/heat treated condition at temperatures well above NST. resulted in gross alloy melting. The simulated fusion zone upon solidification consisted of non-dendritic equiaxed grains similar in nature with those observed in alloy 2195 (Figures 5.3A. 5.37-5.39). This phenomenon was observed even after melting the alloy at temperatures as high as 655°C (Figure 5.40). After solidification the entire fusion zone was uniformly consisted of non-dendritic equiaxed grains. 190 9 8 Figure 6.9. Schematic view of columnar and equiaxed growth (ref. 101) Solid solubility of Sc is very low eg. 0.l3wt.% at 640°C and 0.03wt% at room temperature (ref. 103). A^Sc is the most important intermetallic compound that forms in dilute Sc-bearing alloy. This phase has been confirmed to have a cubic Lli ordered structure at all temperatures with a lattice parameter a=0.4105 nm (ref. 28). The lattice parameter of A1 is 0.404 nm. which is only 1 .6% less than that of AlgSc. Its low misfit with the aluminum matrix on all exposed crystal planes makes it a very efficient nucleating agent. It has been reported that small equiaxed non-dendritic grains are 191 developed upon solidification in alloy Al-0.7Sc (ref. 104). Norman reports grains containing big L h AI 3SC particles (ref. 104). He performed EBSP analysis and found that the missorientation between some particles and matrix was less that 1 ° although there were some .AbSc intermetallics having 50° missorientation between them and the matrix. .As a result, he concludes that not all the intermetallics act as nucleating agents. Using analytical microscopy was not able to detect such intermetallics within the equiaxed grains produced in this study. This can be attributed to their small size (10-40nm) which makes them difficult to detect even with atomic resolution imaging. However. EDS mapping showed moderate enrichment in Mg taking place along the grain boundaries as well as some iron rich precipitates (Figure 5.61). Norman also reports second phase formation (divorced eutectic) around equiaxed grains produced in an Al-4.5Cu-0.8Sc alloy, although Cu was not as much segregated as it was observed in the case of a binary .Al-4.5Cu alloy. The refinement in the grain size compared to large dendritic grains can also be understood since the equiaxed grain has small size in the order of the wavelength of the solid/ liquid interface instability. This is shown characteristically for the alloy 2195 - T 8 in Figure 5.12-B. Under these conditions the solid/liquid interface cannot move far enough during the growth of the small grains for solute levels at the interface to built up sufficiently and destabilize the solid/liquid interface before grain impingement occurs. In the current study non-epitaxy was confirmed using the OIM™ technique. Small equiaxed grains appear to have random orientations while are separated by high angle boundaries from the adjacent HAZ grains (Figures 5.75-5.76). Again it is 192 interesting to note that since AljSc intermetallics have stable LI? structure even at high temperatures this does not compromise their nucléation efficiency as it occurs with A^Zr intermetallics. 6.2.5 Heterogeneous Nucléation - Nucleus Size Effect .-\s was mentioned in the section 2.3.2, the critical nucleus radius r* during heterogeneous nucléation remains the same as that for homogeneous nucléation (Figure 2.9). However, only a spherical cap with radius r* is required to nucleate on the substrate in this case. It is interesting to note that this cap reduces in volume as the wetting angle 0 is decreased (Figure 2.10) until it disappears when 0 = 0°. .*\t this point there is no barrier for nucléation as equations 8 and 9 predict and only growth takes place. This means that liquid atoms approach the substrate-liquid interface and form the solid, which is grown directly from the existing substrate since formation of a solid cap is not necessary. As it was mentioned above Norman et al has observed epitaxy between Sc-intermetallics and surrounding a-aluminum matrix (ref. 104). Thus, prerequisite for the size of the nucleating substrate does not exist any more as it happens when a spherical cap forms (0 > 0°). Figure 6.10 sketches how the solid grows continuously from the substrate surface resulting in a diffuse solid-liquid interface as expected for metals (refs. 45. 47). It is interesting to note that the growth rate of the solid in this case is linear function of the undercooling ahead of the solidification front. This can serve as another likely possibility 193 to explain the non-dendritic nature of the grains formed in the EQZ. The non-dendritic grain character implies that a small undercooling was developed ahead of the solidification front, which suffices to initiate nucléation but prevents dendrites to form. Hence, the morphology of the solidification front remains planar and the grains grow spheroidally until they impinge each other. LIQUID SOLID LIQUID UNOERCOOLING.AT^ Figure 6 .10. Schematic illustrating atom packing (a) in a diffuse interface (b) along with solid growth rate as a function of interface undercooling (c) (ref.47) 194 It is interesting to note, that the heterogeneous nucléation rate (I) depends on the size of the substrate as the following equation shows (ref. 45.47). / = ^10"exp[-AGLS(5)] (eq.l4) n w here n/ is the number of the surface atoms of the substrate per unit volume of liquid and n is the number of atoms in the liquid per unit volume. The expressions for AG'hom and S(0) are presented in the sections 2.3.1 and 2.3.2. For 9 = 0°. it is / = —10" (eq.l5) It can be seen that even though the size of the heterogeneous nuclei can be small, a minimum surface is required area for the solid to form upon. Besides, a small cap will be always required to nucleate since the case 0 = 0° is mostly theoretical. In practice a small wetting angle will be present attributable to lattice disregistry and surface characteristics arising from the degree of the chemical affinity between solid and particle etc (ref. 47). Zr and Sc containing precipitates were found in Mg-bearing aluminum alloys with small additions of Zr and Sc (ref. 105). The diameter of these spherical precipitates was estimated to be between 50 and 90 nm. 195 6.3. EQZ versus Dendritic Microstructure Formation - Composition Effects In the previous discussion only the substrate condition was taken into account to explain fusion boundary microstructure evolution. However, equally alloy composition seems to play an important role in formation of equiaxed non dendritic versus dendritic structures. It has been already mentioned that Li additions in alloy 2195 suggest increase m the volume fraction of the efficient nucleating particles ALiLi^Zri.x) within the aluminum matrix. It was also shown in this study that EQZ was not formed in the Al- Mg-Zr system even after the substrate was in heat treated condition. Based on the dilution levels, the estimated amount of Zr content in this system should be 0.07wt%. However, alloy 2195 that forms EQZ contains approximately 0.14wt.% Zr while Sc in the Al-Mg-Sc system is expected to be on the order of 0.1 wt.%. These values suggest that a cut off value exists in the Zr content, below of which an EQZ does not form, attributable to either absence of nucléation sites or a very small volume fraction of such precipitates in the system. Gutierrez (ref. 70) in a similar study with experimental compositions showed that an EQZ was not present in an Al-2.5Li-2Mg-0.08Zr alloy even after it was heat treated. Another significant observation is that although Sc content is expected to be around 0.1 wt.%. it is still capable of producing an EQZ in the weld/heat treated substrate condition. Since the values 0.07wt.% and 0.1 wt% are very close, this may also signify the higher potency of the Sc-containing alloys to form EQZ. The stable ordered LI: structure of the AlgSc intermetallics in contrast to the two different structures LG and DO:: that ALjZr particles exhibit, may account for this effect. It has already been 196 pointed out that the tetragonal DO 23 structure offers smaller number of planes with low intermetallic/matrix mismatch as it compares with the cubic LN cell. Finally another effect that should be taken into account is the heat treatment temperature and time. It was found that although EQZ was formed in the Al-Mg-Sc system when it was heat treated at 540°C for 4 hours, a dendritic structure was produced when the same type of alloy was heat treated at 300°C for 5 hours. .A.pparently kinetics of the intermetallic formation and coarsening are different at these two different temperature levels, also affected by the relative stability of other precipitates which may dissolve or coarsen. In general, diffusivity is expected to decrease at the lower temperature level. .Although more experiments including prolonged heat treatment times at lower temperature levels were not incorporated in this study, time seems to be an important parameter for the reasons mentioned above. Besides, both temperature and time may also affect the structure of potential heterogeneous nuclei as some of them appear to have more than one structure. 6.4. Precipitation Analysis Using STEM An effort was made to identify intermetallics in the EQZ structures produced. However, analytical work even at atomic resolution level was not able to detect such precipitates (Figures 5.81. 5.84. 5.86). The reason can be the very small size of these particles (1 0 -4 0 nm) along with the fact that only a very small number of them are expected to be surrounded by the relatively massive EQZ grain. This makes it extremely 197 difficult to detect them even with the fine electron beam size allowed on STEM since the field on the scope is limited only at the very thin edge of the foil rendering the majority of the grain non transparent. Examination even at the very thin region is not without problems since transparency is often intermittent due to local thickness variations. As a result, grain boundary precipitates (Figure 5.78. 5.85) or other relatively big particles like the needle like cross section phase (5.81) and the big spherical Zr- containing precipitate in the alloy 2195-T8 (Figure 5.79) were easily detected. 6.5. Simulated versus Weld Fusion Boundary Microstructures The cross welds produced showed very good correlation between microstructures generated using the simulation melting technique on the Gleeble^ and those developed in actual welds (Figures 5.43-5.45). In spite of the fact that fluid flow patterns can be quite different in the two situations, the good correlation was anticipated since the melting technique simulates the temperature conditions that occur within a narrow region next to the weld fusion boundary. On the other hand, fluid flow conditions in weld pool seem to influence the distance over which the EQZ extends within the weld pool. Gutierrez (ref.70) suggests the observed random change in the thickness of the EQZ produced in 2195-T8 GTA welds, is mainly the outcome of different fluid flow patterns prevailing along the weld fusion boundary. He proposes Al]Zr particles are swept by the liquid flow toward the inner part of the weld pool until they melt or dissolve. Based on the previous discussion 198 it should be also added that these particles may change structure and lose their nucléation potency well before they melt or dissolve in hotter regions of the weld pool. As a result the width of the EQZ should be significant lower especially if the local thermal gradient happens to be steep. Table 6.1 below summarizes briefly the effects both composition and substrate condition exert on the formation of EQZ and how these findings correlate with actual welding as it was discussed above. .Mloy Substrate Zr Content Sc Content EQZ C o n d itio n (w t% ) (\vt% ) Gleeble Sample W eld m en t 5454/5025 ,\W N ot formed N ot form ed =50“ o BMD AW-SHT at 300°C - 5h = 0.1 N ot formed .wwsn r at 540'C - 4h Formed Form ed 5454/5087 AW Not formed N ot form ed s50°b BMD AW*SHT at 300=C - 5h = 0.07 Not formed — AW-SHT at 540“C - 4h --- N ot formed N ot form ed 2195-T8 As-received Formed Form ed SUT at 600°C - 4h 0.14 Formed Form ed Table 6.1. Summary of the combined effects composition and substrate exert on the EQZ formation. 199 6.6 Practical Implications Implications related to the presence of the EQZ as well as weld cracking susceptibility associated with the EQZ formation are addressed below. Some of these points have already been discussed above. This section provides a summary of both the observations associated with EQZ and their explanations. WTiv is EQZ crack sensitive ? It was already described in section 2.7 that the EQZ in welds of the .A.l-Cu-Li-Zr alloys is associated with extensive cracking (Figure 2.29). Weldability tests have already shown an extensive cracking formation within the EQZ zone. Vigorous stirring of the weld pool eliminates the EQZ and prevents this form of cracking (ref.55). .Although experimental data regarding the cracking propensity of the EQZ were not collected in the current investigation, an attempt is made to explain this phenomenon. It has been observed that cracking in the EQZ is intergranular (Figure 2.29). It is known that for transgranular fracture the slip planes are weaker than the grain boundaries while for intergranular fracture the grain boundary is the weaker link in the structure. Since the amount of grain boundary area decreases with increasing grain size, fine grained structures such as the EQZ will have lower overall strength compared to that of the coarser dendritic ones at elevated temperatures. This can be graphically seen in Figure 6.11, which illustrates the strength of grains and grain boundaries as function of temperature (ref. 106). 200 Grain boundary Transgronular fracture Intergranular fracture ECT Temperature Figure 6.11. Plot of the strength of both grain and grain boundary as function of temperature (ref. 106). ECT is the temperature at which the strength of the grain equals the strength of the grain boundary. In addition, it should be also noted that during the last stages of solidification both the EQZ grains and the dendrites are covered by significant amount of eutectic liquid. This reduces the ability of the grains to accommodate the applied strain during welding through mechanisms such as grain boundary migration or grain boundary sliding. As a result the ductility in the smaller EQZ grains is exhausted sooner than that in the dendritic region. Since the liquated grain boundaries around them cannot accommodate more strain, rupture will occur preferentially along them (intergranular). 201 The orientation of the grain boundaries relative to the applied stress pattern (Sappi.) may intensify the cracking susceptibility. Comparison of the orientations of the grain boundaries between EQZ and dendritic structures during the Longitudinal Varestraint test shows that the majority of the dendrite grain boundaries should be inclined in respect to the applied load (Figure 6.12). Columnar grains are slightly bent in order to accommodate the direction of the maximum thermal gradient, which is continuously changing during welding (refs.51-52). This means that the resolved stress (Sp) perpendicular to the liquated (dendritic) grain boundar, should be small. In contrast, a significant number of liquated grain boundaries within the fine-grained EQZ are oriented so that the resolved stress acting perpendicular to them has the same magnitude as the applied one (Sappi ) (see Figure 6.12). Hence, significant cracking will occur in the EQZ due to the inability of these grain boundaries to accommodate the applied load perpendicular to them. 202 Columnar Dendrites Figure 6 .12. Schematic illustrating the stress pattern within the EQZ and columnar grain regions during the Longitudinal Varestraint test. Why is the thickness of the EQZ in actual welds non-uniform? EQZ zone is not of uniform thickness in actual welds. It has been observed that EQZ is absent at the top while its thickness becomes maximum at the bottom part of the weld fusion zone. This can be understood if we consider that EQZ zone is formed within a stagnant liquid metal 203 layer next to the ftision boundary (unmixed zone). Within this liquid layer thermally stable Zr- or Sc- bearing intermetallics survive the thermal conditions and non-dendritic equiaxed grains are produced via a heterogeneous nucléation mechanism during weld solidification. The fluid flow patterns affect the extent of the unmixed zone and as a result the thickness of the EQZ. Intermetallic particles (nucleating agents) are swept by the liquid flow toward the inner part of the weld pool where they melt, dissolve (ref. 70) or become ineffective nucleating agents. Fluid flow is stronger closer to the top part the weld pool and this results in a small thickness of the EQZ a there but it is diminishing at the bottom of the weld where the EQZ has higher thickness. Another interesting observation is the banding shown in Figure 5.45. The second EQZ band is separated by the first by dendritic structure. This can be understood if we consider the solute enrichment (k Where is Li going in the structure? Is there anv wav to detect it? As was mentioned earlier Li cannot be detected using an EDS detector since its very small atomic number results in emission of soft X-rays when high energy electrons interact with the Li atoms. However. Li can be detected using Electron Energy Loss Spectrometry (EELS) on TEM. 204 It has already been mentioned that studies of the Ai-Li-Zr system show that Li atoms replace Zr atoms in the AbZr structure. This results in production of high volume fraction of ALlLi^Zri-x) particles since now more Zr is available. Introduction of Li atoms in the L1 2 ordered AljZr structure reduces the misfit between the intermetallic and the tcc aluminum structure. hmaliy it is interesting to note that the partition coetTicient predicted by the binary Al-Li phase diagram (k=0.4). indicates that a significant amount of Li will be incorporated in the eutectic phase which surrounds both the equiaxed and columnar grains in the fusion zone. In fact the synergistic effect of the other alloying elements may enhance Li partitioning during solidification. The low melting (180.54 °C) and boiling (1347.0 °C) points of Li also indicate that a small amount of its initial composition will evaporate under the high temperatures prevailing in the weld pool during welding. \&liv does an EQZ not form from an as-cast substrate? During welding of dilute Zr- and Sc- bearing aluminum alloys, the as solidified structure is not anticipated to contain AljZr or AI 3SC precipitates. These are considered responsible for the formation of the EQZ via a heterogeneous nucléation mechanism. In the case of Sc the partition coefficient (k Sc rich solidification fronts will result in EQZ bands as it was explained above. 205 In the case of Zr its partition coefficient (k>l) (ref. 70) predicts that during solidification the grain boundaries will be depleted of Zr which is kept in solid solution form in the interior of the dendrites. \iVTiv does solution annealing of weld metal restore EQZ? .A.s was stated above both Zr and Sc are in solid solution within the as-welded structure with some of the Sc taking part in the eutectic composition along the grain boundaries. Upon solution annealing there is enough solute available to form the equilibrium AfZr or .AfSc precipitates. These later effect the EQZ zone formation as it has been already described. Alloys provided in the solution annealed and subsequently stretched and artificially aged conditions are expected to have high amounts of these precipitates, since their nucléation will be facilitated due to the high amount of crystal defects such as dislocations, introduced to the alloy during stretching. Why is an EQZ not formed easily in the 2219-T8 welds'? The alloy 2219-T8 has a nominal composition 0.12 \\ 1 .% Zr which is slightly lower than that of the alloy 2195-T8 (0.14 wt.% Zr). .Although both alloys are rich in Cu and are provided in the same temper (T 8 ). alloy 2219 does not contain Li. During weld metal simulation, an EQZ was not observed in the samples examined. This can be understood if we consider the effect of the Li. Its addition to the alloy 2195-T8 increases the volume fraction of the Al 3(LLZri.x) precipitates which besides their big number have also better nucléation potency compared to that of the smaller in number AljZr precipitates. As a result EQZ in the alloy 2195-T8 206 will be much more easily formed compared to that expected in the alloy 2219-T8. In the latter EQZ formation will be much more dependant on the variations of the Zr composition within different batches. Compositions richer in Zr will effect formation of the EQZ during fusion welding. Why does EQZ hardness increase relative to that of the dendritic structure? This phenomenon is not clearly understood. The smaller size of the EQZ grains compared to that of the dendrites can partially explain this phenomenon based on the Hall-Petch effect (ref. 106). However, the smaller size of the EQZ grains does not seem sufficient to explain this observation. Therefore another mechanism should be also responsible for the higher hardness within the EQZ. Since the Scheil equation applies to a relative wide range of cooling rates it will be then valid if we use it to assess the compositions of both dendrites and EQZ grains. By doing so it is expected that the smaller non-dendritic EQZ grains will have different (average) composition in solute such as Cu. compared to that of the bigger dendrites. On the other hand this may affect the composition and quantity of the eutectic constituents surrounding the two types of grains. It is interesting to mention that microsegregation phenomena are strongly affected by the flow patterns of the liquid within the interdendritic region and the lack of such branching on the EQZ grains is expected to affect segregation phenomena in the EQZ region as well. Differences in the solute composition between the two grain types (dendrites versus EQZ) may result in hardness variations due to natural aging effects. 207 CHAPTER 7 CONCLUSIONS 7.1 Melting Simulation Technique H. 1. A melting simulation technique was developed on the Gleeble and is characterized by robustness and reproducibility. 2. .A steel sleeve was used as a crucible during melting simulation and no interaction was observed between the steel and the molten aluminum. 3. Melting simulation was proven very eftlcient in evaluating the role of alloy additions such as Sc. Li or Zr play in the fusion boundary microstructure evolution of aluminum alloys. 4. This technique makes it possible to study the effect that thermal conditions have on the fusion boundary microstructure evolution in aluminum alloys. 5. Simulation generated fusion boundary microstructures have been accurately correlated with those present in actual welds making the melting simulation technique a very promising tool for other studies relating to complex fusion zone metallurgical phenomena. 208 7.2 EQZ Formation in Zr- and Sc- Bearing Alloys 1. The effect of thermal conditions on the EQZ formation has been identified for the Zr- and Li-bearing alloy 2195. EQZ was found to form when samples were simulation melted on the Gleeble'® in the temperature range between 630°C and 640°C. 2. Dendritic structures were developed next to the fusion boundary when alloy 2195- T 8 was simulation melted on the Gleeble'® at temperatures exceeding 640°C. 3. .An EQZ similar to that one developed in the alloy 2195 was formed in the Al- Mg-Sc system even when specimens were simulation melted at temperatures as high as 655°C on the Gleeble®. 4. OIM analysis reveals random orientation among non-dendritic equiaxed grains and adjacent grains in HAZ in both Al-Mg-Sc system and alloy 2195. 5. Thermally stable particles of the type AULi^Zr,.^) and AlgSc with cubic structure (LI2 type ordering) and reduced partiele/matrix disregistry are the potential heterogeneous nucléation sites for the EQZ grains. 6 . A change in the Al^Zr precipitate structure from cubic LI? to tetragonal DO 23 type taking place at temperatures above 600°C. may account for dendritic structure formation in alloy 2195. 7. Fusion boundary microstructure is primarily dictated by epitaxial growth when an EQZ is absent. 209 7.3. Composition and Substrate Condition Effects 1. Non-dendritic EQZ was not formed in welds along the fusion boundary using 5454-H32 substrate and Sc or Zr-bearing filler metals. 2. Zr or Sc-containing as-cast substrates were not found to promote EQZ production in the alloys tested. 3. Weld/ heat-treated substrate conditions of the Ai-Mg-Zr system did not promote EQZ production. 4. Weld/ heat-treated substrate conditions of Sc- containing alloys is associated with EQZ production. 5. Low temperature aging of either Zr or Sc-containing alloys did not promote EQZ formation. 6 . Irrespective of substrate temper, lowering Zr and potentially Sc contents seem to be efficient in eliminating EQZ formation. 210 CHAPTER 8 FUTURE WORK The current investigation makes visible the salient effects both alloy composition and substrate condition have on the nature of the fusion boundary microstructures generated. .A. systematic study is recommended for the future, which will include different substrate heat treated and aged conditions. Such a study is anticipated to reveal the nature and potentially the structures of the intermetallic precipitation since different temperature levels of aging or heat treatment seem to affect drastically the EQZ formation in the Al-Mg-Sc system. Results from the precipitate structure analysis before simulation melting can be compared with those produced from the simulated fusion boundary' microstructures. In addition, interesting information can be gained regarding the effect of alloy composition has on the nature of the microstructures evolved. This investigation can be conducted with the use of simple binary or ternary alloys such as the Al-Li. .Al-Zr. Al-Sc or Al-Li-Zr. Al-Cu-Zr and Al-Mg-Sc systems. Using these simple alloy systems as a starting point, it will make the analytical work easier as compared with the alloy 2195 or other multi-component experimental compositions, since the number of the potential precipitates in the matrix will be 21! significantly reduced. Thus, precipitate detection and associated structural analysis will be accomplished with less degree of complexity. Since the structure of the intermetallics formed in the substrate seems to be crucial affecting their potency to act as heterogeneous nucléation agents, other elements such as Ti should be tried in the alloy compositions. These elements can completely replace Zr and or Sc additions or complement them in the alloy chemistry. Emphasis needs to be placed in elements that produce precipitates with structures that deviate from the fee aluminum structure. .A.nother important aspect that needs to be addressed is the nature of the second phases that form around the non-dendritic grains. Hardness traverses in this investigation show an increased hardness level within EQZs as it compares with that of as solidified dendritic structures. The nature of these "eutectic" phases in EQZs needs to be studied in conjunction with solidification modes. It is important to understand the potential link such second phases have with cracking phenomena observed in EQZ during fabrication and repair welding. Varestraint weldability test could be used as a means to assess this effect. 212 LIST OF REFERENCES 1. J.R.Pickens: ’'Review: Recent Developments in the Weldability of Lithium- Containing Aluminum Alloys". Journal o f Materials Science- 25(7). 1990. pp.3035-3047. 2. J.R. Pickens. L.S. ICramer. T.J. Langan and F.H Heubaum: "The Effect of Zn Nucléation in Al-Cu-Li-Ag-Mg Alloy. Weldalite™ 049 (X2094)". Paper presented at Sixth International Aluminum-Lithium Conference- Eds. M.Peters and P.-J. Winkler. Vol.l. 1991. Garmisch-Partenkirchen. Germany, pp.357-362. j . l.N.Fridlyander. A.G. Bratukhin, and V.G. Davydov: "Aluminum-Lithium Alloys for Welding .Aviation Structures". Russian Metallurgy, No.3. 1992. pp. 107-109. 4. .-\.G. Bratukhin: "Metallurgical Principles of Making Permanent Joints in High Strength Stainless Steels and Lower-Density .Aluminium-Lithium Composition .Aluminium Alloys for High Quality and Reliability and Long Service Life Aircraft". Welding Research Abroad Vol.41(3). 1995. pp.29-31. P.S. Fielding and G.J. Wolf: ".Aluminum-Lithium for aerospace". Advanced Materials and Processes- October 1996. 6 . i.e. Lippold and J.K Watson: "The Weldability of Aluminum-Lithium Alloys: A Review . proc. o f North American Welding Research Seminar, Edison Welding Institute. 1987. 7. C.E.Cross and W.T.Tack: "Selection and Weldability of Aluminum-Lithium Alloys ~ a s m Handbook-Welding, Brazing and S o l d e r i n g " - 1995. pp.549-552. 8 . J.R.Pickens : "Review-The Weldability of Lithium Containing Aluminum alloys' Journal o f Materials Science, 20. 1985. pp.4247-4258. 9. C.E Cross: "Weldability of Aluminum-Lithium Alloys: An Investigation of Hot Tearing Mechanisms". Ph.D. Thesis. Colorado School of Mines. August 1986. 213 10. T.S.Srivatsan and T.S. Sudarshan: "Welding of Lightweight Aluminum-Lithium Alloys". Welding Journal ^0. Jul. 1991. pp. 173s-185s. 11. J.T. Staley: ".Aluminum-Lithium Alloys: Their Properties and Concerns". U ght Meial A^e- Vol.48 (11). 1990. pp.31-32. 12. S.J.Cieslak et al: ".ALCO.A .Alithalite Alloy 2090 Technical information". Paper presented at the l~th International SAMPE Conference- October 22-24. 198o. Kiamesha Lake. New York. 13. P.M.Elkin. N.A.Naryshkina. V.N. Anad'ev and A.V. Rudoi: "Structure and Properties of .Aluminum-Lithium Alloy 1430". Russian Metallurgy', NO.3. 1992. pp. 110-115. 14. .A.M.Dritz and T.V Krymova: "Extrusions of 1450 Alloy for Modem Transport Aircrafts Structures . Paper presented at Sixth International Aluminum-Lithium Conference- Eds. M.Peters and P.-J. Winkler. Vol.2. 1991. Garmisch-Partenkirchen. Germany, pp. 1233-1238. 15. l.N.Fridlyander: "Structural Aluminum-Lithium .Alloys", v/c/a/ Science and Heat Treatment- Vol.32 (4). 1990. pp.235-245. 16. L..A.Willey. Metals Handbook. Vol.8. 8th Edition. 1973. p.261. .ASM Metals Park. Ohio 44073. USA. 17. H.M. Flower and P.J. Gregson: "Solid State Phase Transformations in Aluminum .Alloys Containing Lithium". \{aterials Science and Technology’- Vol.3. February 1987. pp.81-92. 18. J.R. Pickens, F.H. Heubaum. T.J. Langan. and L.S. Kramer : "Al-(4.5-6.3)Cu- 1.3Li-0.4Ag-0.4Mg-0.14Zr Alloy Weldalite™ 049". Paper presented at Fifth International Aluminum-Lithium Conference- Eds.. T.hTSanders Jr.. E.A. Starke. Jr.. Vol.III. 1989. Williamsburg. Virginia, pp. 1397-1414. 19. G.M. Reddy. A.A. Gokhale. and K. Prasad Rao : " Effect of Filler Metal Composition on Weldability of Al-Li alloy 1441". Science and technology of Welding and Joining. Vol.3(3). 1998. pp. 151-158. 20. J.C. Lippold: "A Review of the Physical Metallurgy and Welding Characteristics of Aluminum-Lithium Alloys". EWI. Project No 3064. August. 1987. 214 21. W.A.Baeslack III and K.H. Hou: "Characterization of the Heat-Affected Zone in Gas Tungsten Arc Welded Aluminum alloy 2195-T8 ” Journal o f Materials Science Letters 15- 1996. pp. 239-244. 22. D.B. Williams: "Microstructural Characteristics of Al-Li Alloys". Paper presented at First International Aluminum-Lithium Conference' Eds. T.H Sanders. Jr. and E.A. Starke. Jr.. 1981. Stone Mountain. Georgia, pp. 89-100. Metallurgical Societv/AIME. Warrendale. Pa. 15086. 23. H.K. Hardy and J.M. Silcock: " Ihe Phase Sections at 500 '"C and 350 °C of Aluminum-Rich Aluminum-Copper-Lithium Alloys" Journal o f the Institute o f Metals- Vol.84. 1955-56. pp. 423-428. 24. V. Vladmilovic and G. Thomas : "Atomic Resolution Imaging in Al-Li-Cu Alloy". Journal de Physique- Colloque C3. Suppl.9. Vol. 48. September 1987. pp. C3-385- 396. 25. .A.K. Gupta. P.Gaunt and M.C. Chaturv^edi : "The Crystallography and M orphology of the S'-Phase Precipitate in an Al(CuMg) Alloy". Philosophical Magazine A- Vol.55(3). 1987. pp.375-387. 26. D.W. Levinson and D.J. McPherson : "Phase Relations in Magnesium - Lithium Aluminum Alloys". Transactions o f the ASM- Vol.48. 1956. pp.689-705. 27. V.l.Elagin. V.V. Zakharov and T.D. Rostova : "Nonferrous Metals and Alloys - Scandium-.Alloyed Aluminum Alloys". ^letal Science and Heat Treatment- Vol.34 (1-2). 1992. PP. 37-45. 28. N. Blake and M.A. Hopkins : "Constitution and Age Hardening of Al-Sc Alloys". Journal o f Materials Science- 20. 1985, pp.2861-2867. 29. F.W.Gayle and J.B. Vandersande : "Composite precipitates in Al-Li-Zr Alloy". Scripta Metallurgica- Vol. 18. 1984. pp.47j-478. 30. F.W. Gayle and J.B. Vandersande : ".Al 3(Li.Zr) or ( f Phase in Al-Li-Zr System". Paper presented at Third International Aluminum-Lithium Conference- Eds. C. Baker. P.J. Gregson. S.J. Harris. C.J. Peel. 1986. Oxford University. UK. pp.509- 515. 215 31. G.L.Shneider and A.M. Drits: “Hardenability of Aluminum-Lithium Alloys”, \fe ta l Science and Heat Treatment, (9-10). 1995. pp. j7j-j77. 32. C.Kumai. J. Kusinski. G. Thomas and T.M. Devine: "Influence of Aging at 200 C on the Corrosion resistance of Al-Li and Al-Li-Cu Alloys". Corrosion, April 1989. pp. 294-302. 33. l.N Fridlyander. V.S. Sandler, and T. Nikoloskya: "Dispersoids in Deformed Aluminum Alloys Containing Lithium . \fetal Science and Heat Treatment' Vol.32(6). 1990. pp. 466-468. 34. Metals Handbook. Vol.S. 8 th Edition. 1973. ASM. 35. .A.Gysler. R. Crooks and E..A.. Starke. Jr: "A Comparison of Microstructure and Tensile Properties of P/M and 1/M Al-Li-X Alloys". Paper presented at First International Aluminum-Lithium Conference- Eds. T.H.Sanders. Jr. and E.A. Starke. Jr. TMS-AIME. Warrendale. PA. 1981. pp.263-291. 36. E.A. Starke. Jr and F.S.Lin: "The Influence of Grain Structure on the Ductility of the Al-Cu-Li-Mn-Cd .Alloy 2020 . Metallurgical Transactions A- Vol.l j. December 1982. pp.2259-2269. 37. J.D.Boyd and R.B. Nicholson: "The Coarsening Behavior of Q"and q'Precipitates in Two .Al-Cu Alloys". 4eta Metallurgica- Vol.19^ December 1971. pp.1379-1391. 38. J.M. Silcock. and T.J. Heal. H.K. Hardy: "The Structural Aging Characteristics of Ternary .Aluminium-Copper Alloys with Cadmium. Indium, or Tin". Journal o f the Institute o f Metals, Vol.84. 1955-56. pp.23-31 39. F.S.Lin. S.B. Chakrabortty. and E.A. Starke. Jr.: "Microstructure-Property Relationships of Two Al-3Li-2Cu-0.2Zr-XCd Alloys". Metallurgical Transactions .4 . Vol.l3A. 1982. pp. 401-410. 40. T.H. Sanders. Jr. and E.A.Starke. Jr.: "The Effect of Slip Distribution on the Monotonie and Cyclic Ductility of Al-Li Binary Alloys", jeta Metallurgica- Vol.30(5). 1982. pp.927-939. 216 41. L.I. Kaygorodova, A.M. Dritz. T.V. Krymova and l.N. Fridlyander: "The Effect of Small Sc and Mg Addition on Al-Li-Cu-Zr Alloy Structure and Mechanical Properties . Paper presented at Sixth International Aluminum-Lithium Conference- Eds. M.Peters and P.-J. Winkler. Vol.l. 1991. Garmisch-Partenkirchen. Germany, pp.363-367. 42. K.A. Gschneidner Jr. and F.W. Calderwood: "The Al-Sc (Aluminum-Scandium) System". Bulletin o f Alloy Phase Diagrams- Vol. 10 (1). 1989. pp.34-36. 43. l.N. Fridlyander. S.F. Danilov. E.N. Malysheva. T.A. Gorokhova and N.N Kirkina: "Structure and Properties of Al-Li Alloys Alloyed with Scandium”. Paper presented at Sixth International Aluminum-Lithium Conference- Eds. M.Peters and P.-J. Winkler. Vol.2. 1991. Garmisch-Partenkirchen. Germany, pp.381-386. 44. R.V. llyushenko. A.V. Lozovskaya. I.E. Sklabinskaya and E.O. Paton: "Peculiarities of Weakening of Alloy 1420 (Al-Mg-Li System) in Fusion Welding". Paper presented at Sixth International Aluminum-Lithium Conference- Eds. M.Peters and P.-J. Winkler. Vol.2. 1991. Garmisch-Partenkirchen. Germany, pp. 1251-1256. 45. D A. Porter and K.E. Easterling. 1992. "Phase Transformations in Metals and Allovs "- Ed.. Chapman and Hall. London. 46. W.Kurz and D.J. Fisher. 1992. "Fundamentals o f Solidification - 3'^* Ed.. Trans Tech Publications. 47. M. C.Flemings. Solidification Processing McGraw-Hill Inc. New York. NY. 48. M.E. Gligsman and W.J. Childs: “Nucléation Catalysis in Supercooled Liquid Tin", Acta Metallurgica- Vol. 10. October 1962. pp.925-933. 49. Bruce L. Bramfitt: "The Effect of Carbide and Nitride Additions on the Heterogeneous Nucléation Behavior of Liquid Iron”. Metallurgical Transactions. Vol. l.M y 1970. pp. 1987-95. 50. J.A. Reynolds and C.R. Tottle: "The Nucléation of Cast Metals at the Mould Face”. Journal o f the Institute o f Metals-^ Vol. 80. pp. 1951-52. 217 51. S. Kou. 1987. p. 168. New York. NY. John Wiley & Son. 52. O. Grong. 1994. "Metallurgical Modelling o f Welding ”• London. UK. The Institute of Materials. 53. R.P. Martukanitz and P.R. Howell: "Relationships Involving Process. Microstructure, and Properties of Welderrmts of Al-Cu and Al-Cu-Li Alloys". Manuscript submitted for publishing in Conference Proceedings o f World Trends in Welding Research by ASM International. 1996. 54. S.D.Dumolt: ".An Investigation of the Microstructural Changes in the Heat Affected Zone of .Age Hardenable Aluminum Alloys Using Transmission Electron Microscopy". Ph.D. Dissertation. Camegie-Mellon University. Pittsburg PA. July 1983. 55. J.C. Lippold and W. Lin: "Weldability of Commercial Al-Cu-Li .Alloys". Paper (Resented at Fifth International Conference on .Aluminum .Alloys. Lds. J.H. Driver. B. Dubost. F. Durand. R. Fougeres. P. Guyot. P. Sainford. and M. Suery 1-5 July 1996. Grenoble. France, pp. 1685-1690. Transtec Publications Ltd. Zeurich-Uetikon. Switzerland. 56. L.S. Kramer. C.E. Cross and J.R. Pickens: "The Effect of Composition on the Weldability of Al-Cu-Li-.Ag-Mg Alloys in the High Cu:Li Regime". Paper presented at Sixth International .Aluminum-Lithium Conference- Lds. M.Peters and P.-J. Winkler. Vol.2. 1991. Garmisch-Partenkirchen. Germany, pp. 1197-1202. 57. C.E. Cross. D.L. Olson. G.R Edwards and J.F. Capes: "Weldability of Aluminum- Lithium Alloys". Aluminum-Lithium Alloys II. Eds. E.A. Starke. Jr.. T.H. Sanders. Jr.. TMS/AIME. .April 1983. pp.675-682. 58. W.I.Pumphrey and J.V. Lyons : "Cracking During the Casting and Welding of the More Common Binary Aluminum Alloys". Journal o f Institute o f Metals- Vol.75. 1948. pp.439-455. 59. L.S. Kramer. F.H. Heubaum and J.R. Pickens: "The Weldability of High-Strength Al-Cu-Li Alloys . Paper presented at Fifth International Aluminum-Lithium Conference- Eds.. T.H. Sanders Jr.. E.A. Starke. Jr.. Vol.III. 1989. Williamsburg. Virginia, pp.1415-1424. 218 60. M.F.Gittos: "Gas-Shielded Arc Welding of the Al-Li Alloy 8090'\ Paper presented at the International Conference o f Advanced Aluminium and Magnesium Alloys' Amsterdam. The Netherlands, pp. 181-188. 1990. ASM International. Materials Park. Ohio 44073. USA. 61. W.I.Pumphrey and D.C.Moore: "Cracking During and After Solidification in Some Aluminium-Copper-Magnesium Alloys of High Purity". Journal o f Institute o f Metals' Vol.75. 1948. pp.425-438. 62. C.t.Cross. W. 1. lack. L.W. Loechei and L.S. ICramer : ".A.iuminum Weldability and Hot Tearing Theory " Proc. o f the Materials Weldability Symposium' Detroit. ML October 8-12. 1990. pp.275-282. ASM International. Materials Park. Ohio 44073. USA. 63. W. Lin : ".A Methodology of Quantifying Heat-Affected Zone Liquation Cracking Susceptibility". Ph.D. Thesis. The Ohio State University. 1991. 64. T. Zacharia. S..A. David. J.M. Vitek and R.P. Martukanitz: "Weldability and Microstructural Characterization of Al-Li Alloys". Paper presented at Fifth International Aluminum-Lithium Conference' Eds.. T.H. Sanders Jr.. E.A. Starke. Jr.. Vol.111. 1989. Williamsburg. Virginia, pp. 1387-1396. 65. H. Yunjia. G. Dalu and S. Zhixiong: "Solidification Mode and Cracking Behavior ot Aluminum-Lithium Alloy Welds . Paner presented at Sixth International Aluminum-Lithium Conference' Eds. M.Peters and P.-J. Winkler. Vol.2.1991. Garmisch-Partenkirchen. Germany, pp. 1215-1220. 6 6 . H. Yunjia. G. Dalu and S. Zhixiong: "Microstructure and Crack Susceptibility of .Aluminum-Lithium Alloy 8090 Weldment . Paper presented at Sixth International Aluminum-Lithium Conference' Eds. M.Peters and P.-J. Winkler. Vol.2. 1991. Garmisch-Partenkirchen. Germany, pp. 1203-1208. 67. S.R. Shah. J.E Wittig and G.T Hahn: "Microstructural Analysis of a High Strength Al-Cu-Li (Weldalite TM 049) Alloy Weld. International Trends in Welding Science and Technology, ASM International. 1993. pp. 281-285 6 8 . A.E.Gutierrez and J. C. Lippold and W. Lin: " Nondendritic Equiaxed Zone Formation in Aluminum - Lithium Welds . Paper presented at Fifth International Conference on Aluminum Alloys, Eds. J.H. Driver. B. Dubost. F. Durand. R. Fougeres. P. Guyot. P. Sainford. and M. Suery 1996. Grenoble. France, pp. 1691- 1696. 219 69. A.E.Gutierrez and J.C Lippold: "A Proposed Mechanism for Equiaxed Grain Formation along the Fusion Boundary in .A.luminum-Copper-Lithium Alloys”. Welding Journah Vol.77(j). 1998. pp.l2j-s. 70. A.E.Gutierrez: "A Study of Non-Dendritic Equiaxed Grain Formation Along the Weld Fusion Boundary of Aluminum-Lithium Alloys". Master Thesis. The Ohio State University. Columbus. 1995. 71. A.Kostrivas and J.C. Lippold: "Metallurgical Characteristics of Al-Cu-Li Alloy .AF/C489 Varestraint Weldability Test Samples". Final Report Submitted to U.S. Air Force. August 15. 1996. 72. "Metals and their weldability". Welding Handbook' Vol.4. 7“’ Ed.. .American Welding Society. Miami. Florida, 1987. p.333. 73. S.A David... and J.M. Vitek: "Correlation Between Solidification Parameters and Weld Microstructures", international Materials Reviews- Vol.34(5). 1989. pp.213- 245. 74. S. Kou. and Y. Le: "Nucléation Mechanisms and Grain Refining of Weld Metal". Welding Journal- 65(12). 1986. pp.305-13s. 75. C.C. Chen and W.A Baeslack III: "Weld Fusion Zone Simulation in Aluminum .Alloy 2090 . Paper presented at The International Symposium o f Physical Simulation o f Welding, Hot Forming and Continuous Casting. Bowker. Ottawa. Canada. 1988. pp.lV55- IV64. 76. R. Batra. G.A. Wilber. H.F. Breit and W.J. Childs: "Programmed in Situ Melting. Freezing, and Tensile Testing for Laboratory Study of High Temperature Properties of As-Cast Metals . Journal o f Testing and Evaluation' Yol.j (1). 1975. pp.68-74. 77. Scion Image Beta 4.01 Win. Public Domain Software. 2000. National Institute of Health. 78. D.J. Dingley and V. Randle : "Review: Microtexture Determination by Electron Back-Scatter Diffraction". Journal o f Materials Science' Vol. 27(17). 1992. pp. 4545-4566. 220 79. D.J.Dingley and D.P. Field: "Electron Backscatter Diffraction and Orientation Imaging Microscopy". Materials Science and Technology^ Vol.63(I).1997, pp. 69- 78. 80. N.C. Krieger Lassen: "Automatic Localisation of Electron Backscattering Pattern Bands from Hough Transform". Materials Science and Technology- Vol. 12. 1996. pp. 837-843. 81. H.S. Ubhi and A. W. Bowen: "Analysis and Representation of Electron Backscattered Diffraction Texture Data - Examples from Heat Treated Al-Li Alloy Sheet . Materials Science and Technology- Vol. 12. 1997. pp. 880-884. 82. D. van der Wal and D.J. Dingley: "An Introduction to EBSP". Electron Optics Bulletin- V.134. Jan. 1996. pp. 19-25. 83. V. Randle. 1992: "Microtextrure Determination and Its Applications "- Institute of Materials. London. UK. 84. V. Randle and M. Caul: "Representation of Electron Backscatter Diffraction Data". Materials Science and technology- Vol. 12. October 1996. pp.845-50. 85. V. Randle and .A. Day: "Use of Rodrigues-Frank Space for Representation of Micro texture and Grain Boundary Parameters". Materials Science and Technology- V ol.9(12). December 1993. pp. 1069-78. 8 6 . T.W.Nelson: "A Fundamental Investigation of Boundaries and Structures in Dissimilar Metal Welds . phQ Dissertation. The Ohio State University- 1998. 87. Metals Handbook. Vol.2. 9th Edition. 1979. pp.86.109. 115.261. ASM Metals Park. Ohio 44073. USA. 8 8 . R.P. Martukanitz. K.G. Lysher. and P.R. Howell. "Laser beam welding of aluminum- lithium structures" Conference Publication 3287- 1994. pp. 237- 244. 89. J.C. Lippold. M.J. Cola. T. Gruber: "The Nature and Significance of "Ghost" Indications in Aluminum Weldments". Research Report- EWI, March 1991. 221 90. S. Henry. T. Minghetti. M. Rappaz: "Dendrite Growth Morphologies in Aluminium Alloys", .{eta Materialia^ Vol.46 (18). 1998. pp. 6431-43. 91. G..A.. Chadwick. 1972. Solidification Processing- London. UK. Butterworth&Co (publishers) Ltd. 92. J.P. Boyle. D.A. Mannas and D.W. Walsh: "The effects of grain size on hot cracking of welded 2014 aluminum alloy rolled ring forgings". Paper presented at Fifth International Symposium on Physical Simulation. Delft. Netherlands. April 22-24. 1992. .p p .1 6 5 -173. 93. J.P. Balaguer. D.W. Walsh and E.F. Nippes: "Hot ductility response of Al-Mg and .Al-Mg-Li alloys". Welding Journal- 68(7). 1989. pp.253 - 261-s. 94. J.C. Lippold. W. Lin. W..A. Baeslack 111. D. Xia and A. Szabo. "Weld cracking susceptibility of aluminum-lithium alloys". Paper presented at Aluminum-Lithium .Alloys for Aerospace Applications Workshop- NASA Conference Publication 3287. 1994. pp.288-289. ^5. ASM Handbook-Welding, Brazing and Soldering"- Vol. 6 . June 1995. pp.528-536. 96. T.Z. Kattamis: "Experimental Observations on the Growth Stability of Spherical Crystals During Freezing". Journal o f Crystal Growth- Vol.18. 1973. pp.45-53. 97. T.Z. Kattamis. U.T. Holmberg and M.C. Flemings: "Influence of Coarsening on Dendrite Arm Spacing and Grain Size of Magnesium-Zinc Alloy". Journal of the Institute of Metals. Vol.55. 1967. pp.343-347. 98. C.E. Cross. -. Grong and M. Mousavi: "A Model for Equiaxed Grain Formation Along the Weld Metal Fusion Line". Scripta Materialia- Vol.40 (10). 1999. pp. 1139-1144. 99. E.Sahin and H. Jones: "Extended Solid Solubility. Grain Refinement and Age- Hardening in AI—1 to 13 wt.-% Zr Rapidly Quenched from the Melt". Rapidly Quenched Metals III- Vol. 1. 1978pp. 138-146. 100. S. Rystad and N. Ryum: "A Metallographical Investigation of the Precipitation and Recrystallization Process in an Al-Zr Alloy". Aluminium- Vol. 53 (3). 1977. pp. 193- 195. 101. J.D. Hunt: "Steady State Columnar to Equiaxed Growth of Dendrites and Eutectic”, Materials Science ami Engineering- 1984. pp.75-8 j 102. J. Caputi: " The Effects of Zirconium Kinetics of 5 * (Al^Li) in Binary Al-Li Alloys". Master Thesis. Northwestern University. 1986. 103. M.E. Drits. E.S. Kadaner. T.V. Dobatkina. and N.l Turkina: "Interaction of Scandium With Aluminum in the .Aluminum-Rich Portion of the Al-Sc System". Iz\\ Akail. MnikSSSR Mel - (^)- 1973. pp. 213-217 (Russian). 104. A.F. Norman. P.B. Prangnell and R.S. McEvven: "The Solidification Behaviour of Dilute Aluminium-Scandium Alloys". Materialia- Vol.46 (16). 1998. pp.5715- 5732. 105. M.C.Carroll. M.J.Mills. O.S. Daehn. P.l. Gouma. M.F. Savage, and B.R. Dundar: "Effects of Minor Sc and Zr Additions on Commercial Al-Mg-Mn Alloys". Paper Presented at TMS Conference on Automotive Alloys. San Diego- 1999 (to be published). 106. G. E. Dieter Jr.. 1961. "Mechanical Metallurgy"- McGraw-Hill Inc. New York. NY. 223