Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Fatigue crack propagation in Al-Li alloy 8090 : environmental effects Chen, Xiande 1995

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

Item Metadata


831-ubc_1995-982667.pdf [ 5.85MB ]
JSON: 831-1.0078590.json
JSON-LD: 831-1.0078590-ld.json
RDF/XML (Pretty): 831-1.0078590-rdf.xml
RDF/JSON: 831-1.0078590-rdf.json
Turtle: 831-1.0078590-turtle.txt
N-Triples: 831-1.0078590-rdf-ntriples.txt
Original Record: 831-1.0078590-source.json
Full Text

Full Text

FATIGUE CRACK PROPAGATION IN Al-Li ALLOY 8090- ENVIRONMENTAL EFFECTSbyXIANDE CHENB.Sc., Zhengzhou University, China, 1982M.Sc., Research Institute of Solid State Physics, Academia Sinica, China, 1985A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTSFOR THE DEGREE OF DOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Department of Metals and Materials Engineering)We accept this thesis as confirmingto the required standardTHE UNIVERSITY OF BRITISH C UMBIADecember 1994© Xiande Chen, 1994In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives, It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)_________Department of t,tij MSe.raJs jn•eJrmnThe University of British ColumbiaVancouver, CanadaDate fr4t 24 “??cDE-6 (2/88)11AbstractFatigue crack propagation behaviour in Al-Li alloy 8090 plate were studied in four maintesting environments: a relatively inert environment - desiccated air; and three freely corrodingaqueous environments consisting of distilled water, 0.6 M NaCl and 1 M A1C13. It was foundthat the major role of the aqueous environments in the T-L orientation crack growth behaviourof the Al-Li alloy plate was to promote S-L splitting (delamination) at grain boundaries, witha subsequent effect on the stress state at the crack tip. The splitting preceded the main crackadvance, and helped to keep the crack in the same macroscopic crack plane and restricted thefracture surface roughness, and consequently reduced crack closure effects. In contrast, theabsence of the splitting effects in desiccated air, combined with the planar deformation featuresand the strong crystallographic texture of the Al-Li alloy plate, led to out-of-plane cracking andridge formation in the mid-thickness of the specimen. This resulted in severe crack growthretardation and a crack growth rate plateau at zK values above -3 MPa.m’. Analyses of theobservations led to the conclusion that the S-L splitting phenomenon is associated with bothlocalized anodic dissolution processes and hydrogen embrittlement effects. The effect ofchanging the loading frequency was not obvious, except in the dry air environment wheredecreasing the frequency from —80 to -‘0.5 Hz led to the disappearance of the crack growth rateplateau. The fatigue cracking resistance of the Al-Li alloy plate is superior in the dry airenvironment and slightly better or equivalent in the aqueous environments when compared toa conventional crack-tolerant Al-alloy plate, 2024-T35 1.The effects of a re-aging treatment on fatigue cracking in the S-L orientation of the Al-Liplate were also studied. The presence of aqueous environments also accelerated crackpropagation in the S-L orientation, and crack propagation is faster in the S-L orientation thanin the T-L orientation. The re-aging treatment was effective in increasing the short-transverse111fracture toughness and also improved the intergranular corrosion resistance of the material.However, the re-aging treatment did not improve the corrosion fatigue crack propagationresistance.ivTable of ContentsAbstract iiTable of Contents ivList of Tables ixList of Figures xList of Symbols and Abbreviations xvAcknowledgements xix1 Introduction 12 Literature Review 32.1 The corrosion fatigue failure 32.1.1 Crack initiation 32.1.2 Crack propagation 52.1.3 Types of corrosion fatigue crack growth 102.1.4 Models for corrosion fatigue Superposition model Competition model Interaction model 132.1.5 Mechanisms for corrosion fatigue Oxidation-related mechanisms Hydrogen assisted cracking mechanisms Surface-related mechanisms 152.1.6 Crack closure effects Plasticity induced closure Oxide induced closure 17V2.1.6.3 Roughness induced closure. Viscous fluid induced closure Phase transformation induced closure 192.1.7 Rate controlling process in corrosion fatigue 192.2 Corrosion Fatigue of Li-Containing Aluminium Alloys 202.2.1 Brief historical development of Li-containing aluminium alloys 202.2.2 Mechanical properties and fracture toughness of Li-containingaluminium alloys 212.2.3 Corrosion and stress corrosion cracking of Al-Li alloys 242.2.4 Fatigue and corrosion fatigue of Al-Li alloys 293 Objective 354 Experimental 374.1 Materials 374.1.1 Chemical compositions 374.1.2 Mechanical Properties 374.1.3 Grain structures 384.1.4 The re-aging treatment 394.2 Specimen design and preparation 414.2.1 Electrochemical test specimens 414.2.2 Fatigue specimens 444.2.2.1 Single edge notched specimen 444.2.2.2 Compact tension (CT) specimen 454.2.2.3 Double cantilever beam (DCB) specimen 484.2.3 Preparation of specimens for transmission electron microscopy 484.3 Test environment selection 49vi4.4 Test setup and procedures.514.4.1 Potentiodynamic polarization tests 514.4.2 Corrosion fatigue crack propagation tests 534.4.2.1 High frequency 534. Precracking 534. Fatigue tests 544.4.2.2 Low frequency 554. Precracking 574. Fatigue tests 574.4.2.3 Intermediate frequency 594. Precracking 594. Hardness and fracture toughness tests 604. Fatigue tests 604.4.3 Fractographic study 614.4.4 Crack profile study 614.4.5 Microstructural and compositional study 634.4.5.1 TEM 634.4.5.2 STEM + EDX 634.4.5.3 EELS 635 Results 645.1 Electrochemical behaviour 645.1.1 Al-Li alloy 645.1.2 Al-Cu alloy 73vii5.2 Crack growth behaviour 755.2.1 Al-Li alloy 755.2.1.1 T-L crack plane orientation, 80 Hz loading frequency 755.2.1.2 T-L crack plane orientation, 0.5 Hz loading frequency 785.2.1.3 S-L crack plane orientation, 30 Hz loading frequency 845.2.2 Al-Cu alloy 865.2.2.1 T-L crack plane orientation, 80 Hz loading frequency 865.2.2.2 T-L crack plane orientation, 0.5 Hz loading frequency 865.3 Fractography 935.3.1 Al-Li alloy 935.3.1.1 T-L crack plane orientation, 80Hz loading frequency 935.3.1.2 T-L crack plane orientation, 0.5 Hz loading frequency1085.3.1.3 S-L crack plane orientation, 30Hz loading frequency 1085.3.2A1-Cualloy 1125.3.2.1 T-L crack plane orientation, 80 Hz loading frequency 1125.3.2.2 T-L crack plane orientation, 0.5 Hz loading frequency 1135.4 Macroscopic fracture surface appearance, roughness and crack profile 1175.4.1 Al-Li alloy 1175.4.1.1 T-L crack plane orientation, 80Hz loading frequency 1175.4.1.2 T-L crack plane orientation, 0.5 Hz loading frequency 1225.4.1.3 S-L crack plane orientation, 30Hz loading frequency 1225.4.2 Al-Cu alloy 1235.5 Microstructure and compositional distribution 126vu6 Discussion.1326.1 Corrosion fatigue crack propagation in the Al-Li alloy in the T-L orientationand comparison with the Al-Cu alloy 1326.1.1 Role of splitting 1326.1.2 Role of environment 1466.1.3 Effects of loading frequency 1526.1.4 Effects of microstructure and deformation behaviour 1546.2 Corrosion fatigue crack growth in the Al-Li alloy in the S-L orientation andeffects of the re-aging treatment 1576.2.1 Possible changes in the material caused by the re-aging treatment 1576.2.2 Effects of re-aging on the S-L crack growth rates 1606.2.3 Effects of environments on the S-L crack growth rates 1617 Summary and Conclusions 1648 References 1669 Appendix 1819.1 Appendix I. Increased stiffness of Al-Li alloys 1819.2 Appendix II. Identification of grain boundary precipitate by selected areaelectron diffraction 1829.3 Appendix III. EELS analyses 1869.4 Appendix IV. Threshold stress state 1899.5 Suggestions for further work 190ixList of TablesTable 4.1 Chemical Compositions (Nominal wt.%) 37Table 4.2 Physical and Mechanical Properties 38Table 5.1 Values of Microhardness (H) and Short Transverse (S-L) OrientationFracture Toughness (K1) 84Table 5.2 Threshold Cyclic Stress Intensities (AKth) of S-L Fatigue Cracking in theTwo Tempers of the Al-Li Alloy in Different Environments 85Table 6.1 Boundary Conditions, ZS.Kb, above Which Plane Stress ConditionsBecome Important in the Presence of NB Splits 139Table 6.2 The Operative [111 } < iTO> Slip Systems and Schmid Factors UnderUniaxial Loading ([1111 Direction) 142Table Al Comparison of Calculated Spacing and Angle Values of a Possible Phase184xList of FiguresFigure 2.1 The three modes of loading [32] 5Figure 2.2 Schematic variation of fatigue crack growth rate as function of stressintensity rang [36] 7Figure 2.3 Schematic diagram illustrating the model for fatigue crack propagationbased on restricted slip reversal at the crack tip [41] 9Figure 2.4 The plastic blunting process of fatigue crack propagation in the Stage IImode [43] 10Figure 2.5 Three types of corrosion fatigue crack growth behaviour [44] 11Figure 2.6 Schematic illustration of the principal mechanisms of fatigue crackclosure [56] 16Figure 2.7 Terminology used to describe the various orientations for crackextension in an isotropic material containing specific planes ofweakness in one direction [1071 24Figure 4.1 Grain structures of (a) Al-Li 8090-T8771 and (b) Al-Cu 2024-T351 40Figure 4.2 Crack plane orientation identification code for plate [1881 42Figure 4.3 Geometry of the electrochemical test specimen 43Figure 4.4 Geometry of the single edge notched (SEN) specimen 46Figure 4.5 Geometry of the compact tension (CT) specimen 47Figure 4.6 Geometry of the double cantilever beam (DCB) specimen 50Figure 4.7 Schematic experimental set-up of the potentiodynamic polarization test.52Figure 4.8 Schematic environmental cell set-up for the high frequency fatigue test.56Figure 4.9 Schematic environmental cell set-up for the low frequency fatigue test.58Figure 4.10 Schematic environmental cell set-up for the intermediate frequencyfatigue test 62Figure 5.1 Potentiodynamic polarization behavior of Al-Li alloy 8090-T8771 in 0.6M NaCl and 1 M AIC13 66xiFigure 5.2 Appearance of polished surfaces (T orientation) of Al-Li 8090 afterfreely corroding in 1 M A1C13 for —20 hours 68Figure 5.3 Appearance of polished surfaces (T orientation) of Al-Li 8090 afterfreely corroding in 1 M A1C13 for —170 hours 69Figure 5.4 Sectioning parallel to the L plane showing depths of intergranularcorrosion after corroding in 1 M Aid3 for —170 hours 70Figure 5.5 SEM + EDX analysis of the intergranularly corroded slots in the re-agedAl-Li alloy, showing uncorroded Cu-containing particle 71Figure 5.6 SEM micrographs showing location of Cu-rich particle containingintergranular surface of the re-aged Al-Li alloy before and aftercorroding in 1 M A1C13 for —12 hours 72Figure 5.7 Potentiodynamic polarization behavior of Al-Cu alloy 2024-T351 in 0.6M NaC1 and 1 M A1C13 74Figure 5.8 Effect of environment and LS.K on the growth of fatigue cracks in Al-Lialloy 8090-T8771 at —80 Hz 76Figure 5.9 Consecutive fatigue crack growth behaviour in Al-Li alloy 8090-T8771when the testing environment is changed from distilled water todesiccated air 79Figure 5.10 Fatigue crack growth behaviour in Al-Li alloy 8090-T8771 at —80 Hzin desiccated air, starting from AK 5 MPa.m’ 80Figure 5.11 Fatigue crack growth behaviour in Al-Li alloy 8090-T8771 at —80 Hzin desiccated air after hydrogen precharge 81Figure 5.12 Fatigue crack growth in Al-Li alloy 8090-T8771 at 80Hz in 1 M HC1solution under controlled potential of -2.1 VSCB. In this potential andpH condition, Al is in the immunity region and the rate of hydrogenevolution is high 82Figure 5.13 Effect of environment and AK on the growth of fatigue cracks in Al-Lialloy 8090-T877 1 at —0.5 Hz 83Figure 5.14 Fatigue crack growth in the S-L orientation of the Al-Li alloy plate indesiccated air at —30 Hz 88Figure 5.15 Fatigue crack growth in the S-L orientation of the Al-Li alloy plate in 1M AIC13 at —30 Hz 89Figure 5.16 Fatigue crack growth in the S-L orientation of the Al-Li alloy plate indistilled water at —30 Hz 90xliFigure 5.17 Effect of environment and AK on the growth of fatigue cracks in Al-Cualloy 2024-T351 at —80 Hz 91Figure 5.18 Effect of environment and AK on the growth of fatigue cracks in Al-Cualloy 2024-T351 at —0.5 Hz 92Figure 5.19 (a) SEM fractograph showing transgranular fracture surface of Al-Lialloy specimen fatigued at —80 Hz in distilled water near AKth.Macroscopic direction of crack propagation is from the right to the leftof the page. (b) Schematic diagram of the transgranular crack front inthe primary crack plane showing tongues extending into uncrackedmaterial, with uncracked ligaments between the tongues 94Figure 5.20 SEM fractograph of Al-Li alloy specimen fatigued at —80 Hz in 1 MAlCl3 near AKth, showing transgranular cracking on the primary crackplane and S-L splitting of grain boundaries normal to the primaryplane. Macroscopic direction of crack propagation is from the right tothe left of the page 96Figure 5.21 Effect of environment and AK on the number of splits (NB) across thethickness of the T-L orientation Al-Li 8090-T8771 specimen fatiguedat —80 Hz 99Figure 5.22 SEM micrograph of Al-Li alloy 8090-T8771 specimen precrackedsurface exposed to 1 M AlC13 during —80 Hz test. Area showslocalized dissolution of grain boundaries 100Figure 5.23 Optical micrograph of sectioned and etched Al-Li alloy 8090-T8771specimen fatigued at —80 Hz in 0.6 M NaC1. Sectioned normal to theprimary crack plane in Stage II at AK = 5 MPa.m”2 101Figure 5.24 Optical micrograph of sectioned and etched Al-Li alloy 8090-T8771specimen fatigued at —80 Hz in 0.6 M NaC1. Test terminated in StageII and sectioned normal to the primary crack plane very close to themain crack front 102Figure 5.25 SEM fractograph of Stage 11 cracking (AK = 6 MPa.m”2)of Al-Li8090-T8771 in 0.6 M NaCl at —80 Hz, showing river patterns runningaway from the S-L split. Note very faint striations normal to the riverlines. Macroscopic cracking direction is from the right to the left ofthe page 103Figure 5.26 SEM fractograph showing change of fatigue fracture surface of Al-Li8090-T8771 produced at —80 Hz in distilled water and desiccated airconsecutively. Macroscopic cracking direction is from the right to theleft of the page 104Figure 5.27 Change of the number of splits (NB) across the thickness of the T-Lorientation Al-Li 8090-T877 1 specimen with AK, when fatigued at —80Hz under continuous hydrogen charging at -2.1 VSCE 105xliiFigure 5.28 SEM fractographs of near AKth cracking of Al-Li alloy 8090-T877 1 indistilled water at —80 Hz. The opposing fracture surfaces indicating a“peak to valley” match of small details. Macroscopic crackingdirection is from the top to the bottom of the page 106Figure 5.29 SEM fractographs of matching fatigue fracture surfaces of Al-Li alloy8090-T8771 produced by Stage II cracking (AK = 6 MPa.m) indistilled water at —80 Hz. Macroscopic cracking direction is from thetop to the bottom of the page 107Figure 5.30 SEM fractograph showing transgranular fracture surface of Al-Li8090-T8771 specimen fatigued at —0.5 Hz in desiccated air. AK= —6MPa.m’, location near the mid-thickness of the specimen.Macroscopic direction of crack propagation is from the right to the leftof the page. Micrograph (b) shows the central region of Micrograph(a) in higher magnification 109Figure 5.31 Effect of environment and AK on the number of splits (NB) across thethickness of the T-L orientation Al-Li 8090-T8771 specimen fatiguedat —0.5 Hz 110Figure 5.32 SEM fractographs showing transition from fatigue precracking tomonotonic overload fracture in the S-L orientation fracture toughnesstest specimens of Al-Li alloy 8090. Macroscopic cracking direction isfrom the right to the left of the page 111Figure 5.33 SEM micrographs showing fatigue fracture surfaces of Al-Li alloy8090 in the S-L orientation in the re-aged temper near IXKth.Macroscopic cracking direction is from the right to the left of the page.114Figure 5.34 SEM fractographs of near AKth cracking of Al-Cu alloy 2024-T35 1 inthe T-L orientation at —80 Hz 115Figure 5.35 SEM fractograph of Stage II cracking (AK = 8 MPa.m1”2)in Al-Cualloy 2024-T35 1 in the T-L orientation in distilled water at —80 Hz 116Figure 5.36 Macroscopic appearance of opposing fracture surfaces of Al-Li alloy8090-T8771 tested in the T-L orientation at —80 Hz. Scale showscrack length 118Figure 5.37 (a) Schematic diagram of ridge formation in mid-thickness ofspecimens fatigued in the T-L orientation in desiccated air at —80 Hz.Arrow shows direction of macroscopic crack propagation. (b) Theangle of the central ridge with respect to the original macroscopiccrack plane 120Figure 5.38 Optical micrographs of sectioned and etched Al-Li 8090-T8771specimen fatigued at —80 Hz in the T-L orientation in desiccated air.Sectioned normal to the primary crack plane 121xivFigure 5.39 Macroscopic appearance of opposing fracture surfaces of Al-Cu alloy2024-T351 fatigued in the T-L orientation in 0.6 M NaC1 at —80 Hz.Scale shows crack length 124Figure 5.40 Optical micrographs showing crack front profiles of specimensfatigued in the T-L orientation in 0.6 M NaC1 at —80 Hz. Sectionednormal to the primary crack plane at AK = 5 MPa.mla 125Figure 5.41 TEM micrograph showing ö’ phase in Al-Li alloy 8090. Centered darkfield image 127Figure 5.42 TEM micrograph showing precipitates containing grain boundaries inAl-Li alloy 8090-T8771. Bright field image 128Figure 5.43 STEM + EDX analyses showing Cu content profiles across a grainboundary containing precipitates 129Figure 6.1 Schematic diagram showing the triaxial stress state near the tip of acrack 133Figure 6.2 Schematic diagram showing effect of S-L splitting on linkage ofuncracked ligaments near the crack front. Arrow shows direction ofmacroscopic crack propagation 136Figure 6.3 Deformation patterns [32] 137Figure 6.4 (a) Schematic diagram showing the relation of the initial T-Lmacroscopic crack plane with the texture in the mid thickness of theAl-Li 8090-T8771 plate. (b) Standard (111) projection for aface-centred cubic crystal 143Figure 6.5 Relative orientations of the operative slip systems in the overall(111) [1211 cracking 144Figure 6.6 Schematic theoretical polarization diagram showing possible cathodicprotection on the other part of the grains by the active dissolution of thegrain boundary regions. I <I due to enhanced corrosion at the grainboundaries (1GB) 156Figure Al Selected area diffraction pattern obtained from a grain boundaryprecipitate 183Figure A2 Selected area diffraction pattern obtained from a grain boundaryprecipitate, showing five-fold symmetry 185Figure A3 Schematic diagram of an electron energy loss spectrum 187xvList of Symbols and Abbreviationsa crack lengthb beam thicknessB specimen thicknessC scaling constantCLI, CMg atomic concentration of Cu, Li and MgC hydrogen concentration at crack tip surfacehydrogen concentration at a distance of x ahead of crack tipd plane spacingda/dN crack propagation rateD diffusion constantE elastic modulusE00ff free corrosion potentialE open circuit potentialpitting potentialplasmon energy lossh beam heightH specimen heightVickers hardnessi current densityI currentK crack tip stress intensityK1 opening mode stress intensityK1 fracture toughnessxvithreshold stress intensity for stress corrosion crackingK, K maximum stress intensity, minimum stress intensityK10 stress intensity for crack closureAK stress intensity rangeAKb boundary cyclic stress intensityAKff effective stress intensity rangeiXKth threshold stress intensity range1 support distance of beamL camera lengthm scaling exponent in Paris lawn scaling exponent in Forman equationN load cycle numberNB split numberP load1rnax, p maximum load, minimum loadr crack tip plastic zone sizeAr crack tip cyclic plastic zone sizeR load ratio, = P.jPmax K/KmaxR1,R2,R3 distance between diffraction spotsS beam stiffnesst time variabletB specimen thickness required for plane strain conditionsW specimen widthW., W beam weightx distancexviiC elongation( angle between the loading direction and the normal of the slip planeangle between the loading direction and the slip directionelectron wavelengthV Poisson’s ratiop densityaT, aL principal tensile stresses in S, T, L directionsyield strengthultimate tensile strengthe angle coordinate in front of a crack tipCF corrosion fatigueCOD crack opening displacementCT compact tensionCTOD crack tip opening displacementDC direct currentDCB double cantilever beamEDX energy dispersive X-ray analysisEELS electron energy loss spectroscopyFCC face centred cubicLT long transverseM molarityPFZ precipitate-free zoneRH relative humidityRSR restricted slip reversibilityS, T, L directions with respect to rolling plane: S - short transverse, T -transverse, L - longitudinalxviiiSCC stress corrosion crackingSCE saturated calomel electrodeSEM scanning electron microscopySEN single edge notchedSHE standard hydrogen electrodeSSRT slow strain rate techniqueST short transverseSTEM scanning transmission electron microscopyTEM transmission electron microscopyVSCB potential with respect to saturated calomel electrodepotential with respect to standard hydrogen electrodeWDX wavelength dispersive X-ray analysisf3’ phase A13ZrS phase A1Li8’ phase A13LiS’ phase A12CuMgT1 phase A12CuLiT2 phase Al6CuLi3e’ phase AI2CuxixAcknowledgementsI would like to express my sincere gratitude to Professor Desmond Tromans for hisguidance and encouragement throughout this project. My thanks are also extended to otherfaculty and staff members, and fellow graduate students for their help, particularly to Mr.Ru-hong Sun for his assistance in the initial stages of the project.Special thanks go to my wife for her patience, understanding and support during thecourse of this work.The author is grateful for the financial support provided via Research Assistantship bythe Canadian Department of Defence and the Natural Sciences and Engineering Council ofCanada, and via Teaching Assistantship by the Department of Metals and MaterialsEngineering.Introduction 11 IntroductionAged Li-containing aluminium alloys are candidate materials for airframe applicationsbecause their reduced density and increased elastic modulus combine to give a class of alloysthat have a higher specific stiffness than the more widely used, and more conventional, highstrength 2000 and 7000 series Al-alloys [1-3]. The early problems of poor ductility andtoughness that were associated with Li-addition have now been largely overcome [2-4]. Thishas increased the opportunity for their use in high stiffness, crack tolerant situations. Theproblem ofcrack tolerance is a critical consideration in airframe structures where the fluctuatingloading conditions can lead to time-dependent sub-critical crack propagation (fatigue).Furthermore, it is well established that conventional Al-alloys are susceptible to increasedfatigue crack propagation rates in the presence of water vapor, water, and saline solutions [5],leading to the phenomenon of corrosion fatigue (CF). Consequently, it is essential to study theCF behaviour of the new Al-Li alloys, and compare such behaviour with that of the moreconventionalAl-alloys, in order to confirm that the new alloys are a suitable replacement materialfor airframe structures.Extensive studies have been conducted on the fatigue and corrosion fatigue behaviour ofLi-containing aluminium alloys and comparisons have been made to the more conventionalAl-alloys. In particular, these studies have found that the commercial Al-Li alloys have betterresistance to fatigue propagation in air environment than the conventional Al-alloys when testedin specific plate orientations such as L-T [6]. This improved resistance is attributed mainly tothe “extrinsic toughening” effects of crack deflection, crack branching and roughness inducedcrack closure [6-1 1], which are associated with the planar slip characteristics inherent in Al-Lialloys. Several studies have been conducted to determine the effects of the environment onIntroduction 2fatigue crack propagation in Al-Li alloys, especially the presence of water and saline solutions,and deleterious effects have been explained by differentmechanisms based on anodic dissolution[12], hydrogen embrittlement [13-16], change of crack closure level [15,17-19] and slipreversibility effects [15,20]. However, defmitive studies on the mechanism ofenvironment-related fatigue cracking behaviour are still relatively scarce.The present research is concerned with the corrosion fatigue crack propagation behaviourofan Al-Li alloy, 8090-T8771, particularly in the near threshold low cracking rate region, whereenvironmental effects are more likely and the cracking behaviour plays an important role indetermining the overall service lifetime of a structure or component. Corrosion experimentsare conducted in controlled environments, with different loading frequencies, crack orientationsand heat treatments. Comparisons are also made to the corrosion fatigue crack propagationbehaviour of a conventional crack tolerant Al-Cu alloy, 2024-T351.Literature Review 32 Literature ReviewSince the recognition of the influence of different environments on fatigue ofmetals aboutseventy years ago [21,22], corrosion fatigue has been extensively studied because it is frequentlycited as the cause ofengineering failures in a broad range ofmaterial-environment systems [23].Although a generalized framework describing corrosion fatigue has emerged [241, a thoroughunderstanding of this complex problem is still far from complete. No single theory has reachedthe point of fully predicting or even explaining all the problems encountered in load bearingstructures. New challenges for higher efficiency and reliability continue to arise, and newmaterials, environments and loading situations have been continuously extending the domainof corrosion fatigue research.2.1 The corrosion fatigue failureThe corrosion fatigue process leading to failure is usually divided into two steps or stages:(i) crack initiation and (ii) crack propagation. However, the transition between the two stagesis not sharply defined.2.1.1 Crack initiationThe fatigue crack initiation stage under cyclic (repeated) loading conditions is generallydefmed as the period required for the formation of microcracks of sufficient length that theybegin to affect the local slip behaviour. Crack initiation typically occurs on (i) external surfaces(or subsurface interfaces) at stress concentrating defects and flaws or (ii) smooth surfaces whereLiterature Review 4slip reversal has produced persistent slip bands containing intrusions/extrusions which then actas local stress concentrators, or flaws [251. The presence of an environment may accelerate thisinitiation process.Several mechanisms have been proposed to explain the effects of environments on crackinitiation, depending on the nature of the environment:Gaseous environments:1. Obstruction of reversible slip [26].2. Inhibition of crack re-welding in slip bands [27].3. Reduction of surface energy at highly stressed locations [28].4. Formation of bulk oxide [291.Aqueous environments [30]:1. Pitting, where nucleation of cracks occurs at pits formed through corrosive attack.2. Preferential dissolution, where highly deformed material acts as a local anode with thesurrounding undefonned material acting as a local cathode. The anodically dissolvedregions then serve as crack nucleation sites.3. Rupture of protective oxide films, where localized corrosion occurs on the exposedmetal at the sites of deformation-induced rupture of the protective oxide. Fatigue thenstarts at the regions of localized attack.4. Surface energy reduction, where adsorption of environmental species promotes crackformation by reducing the surface energy.Literature Review 52.1.2 Crack propagationCrack propagation refers to the sub-critical, time-dependent crack growth that occurs afterinitiation. Based on the realistic assumption that cracks and flaws are always present in astructure due to its processing history before service [31], the crack propagation stage isconsidered to be the governing factor on the total service life of an engineering structure. Inrecent years, there have been many studies where precracked fracture mechanics specimenswere used to investigate fatigue crack propagation behaviour. These studies avoided theproblems associated with using smooth specimens, where both initiation and propagation stagesare present and a clear separation between them is difficult.For a solid containing a crack, three modes of loading can be classified, as schematicallyshown in Figure 2.1 [32],mode]1Iteoring modeFigure 2.1 The three modes of loading [32].Normal stresses give rise to the “opening mode” or Mode I loading; the displacements of thecrack surfaces are perpendicular to the plane of the crack. In-plane shear results in “slidingmodeflmode s’iding modeLiterature Review 6mode” or Mode II loading; the displacements of the crack surfaces are in the plane of the crackand perpendicular to the leading edge of the crack. The “tearing mode” or Mode Ill loading iscaused by out-of-plane shear; the crack surface displacements are in the plane of the crack andparallel to the leading edge of the crack. Superposition of the three modes describes the generalcase of loading. Mode I loading is technically the most important one. In experimental studies,this is also the most commonly applied loading condition.Fatigue crack propagation behaviour is further refined to include short crack and longcrack behaviour; as studies have revealed differences between the two situations [33,34].Generally, cracks defined as microstructurally short, mechanically short, physically short andchemically short cracks [35] all propagate faster than long cracks under the same nominalmechanical driving force, and can even propagate below the apparent crack propagationthreshold for long cracks. To a certain extent, short fatigue crack propagation can be consideredas covering the transition region between crack initiation and (long) crack propagation.For long cracks, the fatigue behaviour can be well described with the application of afracture mechanics factor (stress intensity, K). The curve in Figure 2.2 shows the typical variationof fatigue crack propagation rate per cycle, da/dN, with the alternating stress intensity rangeiSK(= Km - K) [36]. Crack growth behaviour is usually divided into three regimes or stagesin terms of their different sensitivities to various microstructural, mechanical and environmentalvariables. In detail, the three stages of crack propagation for long cracks have been defined asfollows:Stage I, near threshold propagation: Crack growth rate is very low, usually <1Omm/cycle,and can be even less than the lattice parameter, indicating crack growth is not occurring uniformlyover the entire crack front or propagating on every cycle. The threshold AK, is the stressLiterature ReviewUEEzI,0wI—0(9UU7intensity range value below which no crack growth occurs or can be detected within a specifiednumber of load cycles or time. Some researchers [37,38] suggested that the crack growth ratecorresponding to the defmed AKth should be lower than iO minlcycle. The value of AKth andthe crack growth rates are strongly sensitive to microstructure, mean stress and environment[39]. From the engineering point of view, Stage I crack growth is important to the high cycle,low amplitude resistance of the materials.Regime ANon-continuummechanismslarge influence of(I) microstructure(ii) mean Stress(iii) environmentPRIMARY MECHANISMSRegime BI Continuum mechanismI (striation growth)I little influence ofI (I) mlcrostructLreI (II) thickness101OC lmrnlminBm I Regime C 1nIhi I Static mod&mechanismsI (cleavage. intergranular.I / and fibrous)— 1/Iarge influence o1 large influence of:4(iif) certain combinations (i) microstructureI— I lattice spacing / of environment, mean (ii) mean stressper cycle / i stressand frequency (iii) thickness — 11m1Y— I / little influence of:I i (iv) environmentIl A IIl ithreshold K0 I I4, i__________LOG KFigure 2.2 Schematic variation of fatigue crack growth rate as function of stressintensity range [36].Literature Review 8Stage II, often called the Paris power law region: Crack growth rates are usually between10.6 and i0 mm/cycle. In this region, the relationship between the crack growth rate per cycle,da/dN, and AK follows the empirical Paris power law:daldN=CAKTM (2.1)where C and m are scaling constants [40]. Although there is no strict theoretical foundation forthe Paris law, it is the best description for fatigue crack growth behaviour and has been widelyaccepted. In this region, the crack growth rates are generally insensitive to microstructure andmean stress relative to the effect of AK. While certain combinations of environment, meanstress and frequency may have a large influence on the crack growth.Stage ifi: As K approaches K1 for unstable cracking, the crack growth rate is muchfaster than that predicted by the Paris law and da/dN is usually> 1 O. It is better described bythe Forman equation [32]:da CAK22dN(1—R)(K—K) (.Equation (2.2) is based on a modification of the Paris law and takes into account the increasingvalue of the maximum stress intensity factor, where R (=K11JK) is the load ratio and K1 isthe fracture toughness value for unstable cracking. Crack growth in this stage is influenced bymicrostructure and mean stress, but is not significantly affected by the test environment.The Stage I fatigue cracking normally produces a crystallographically dependent fracturesurface without well defined striations. It is generally accepted that the crack advances througha process of restricted slip reversals on relatively few slip systems. The model depicting theevent at a propagating crack tip is shown schematically in Figure 2.3 [4 1,42], which assumesLiterature Review 9alternative slip systems operating at the crack tip. The amount of slip reversal is affected by (i)work hardening in the forward slip and (ii) the degree of oxidation of the bare slip step createdduring the forward slip.(c)Figure 2.3 Schematic diagram illustrating the model for fatigue crack propagationbased on restricted slip reversal at the crack tip [41].The Stage II fatigue crack normally proceeds by transgranular fracture with ductilestriation formation on the fracture surface. The crack growth rate da/dN corresponds closelyto the striation spacing [43]. It is widely accepted that the crack advances through a process ofcrack tip blunting and re-sharpening. The model for crack advance and striation is shown inFigure 2.4, which assumes multiple slip systems operating simultaneously at the crack tip.(a) (e)(b) (f)(d)(g)ç_/ \ø’(h)Literature Review 10(a.)Figure 2.4 The plastic blunting process of fatigue crack propagation in the StageII mode [43].In the Stage ifi fatigue cracking, static fracture modes, such as cleavage, intergranularand fibrous fracture, occur in addition to striation growth.2.1.3 Types of corrosion fatigue crack growthIn the presence of an environment, the general shape of the da/dN - AK curve describedin Figure 2.2 may be changed. Three general types of changes are shown schematically inFigure 2.5 [44]. They illustrate the corrosion fatigue crack growth resulting from differentmechanical-environmental interactions.=(d)=4(e)Literature Review iiKic OrKc K KiccrKc “Kis K1c0(Kcog Kmox log K,, logTypeA TypeB TypeCFigure 2.5 Three types of corrosion fatigue crack growth behavior [44].Type A is termed “true corrosion fatigue”. The combined action of cyclic loading andcorrosion enhances crack growth under all conditions of loading, except when AKis very highand mechanical crack growth becomes so rapid that the effect of chemical corrosion isovershadowed.Type B is termed as “stress corrosion fatigue”. Environmental contributions to cycliccrack growth only occur at AK levels with K> where is the threshold for stresscorrosion cracking (SCC). The waveform of loading has a strong effect on cracking rate inaddition to cyclic frequency, stress intensity level, and stress ratio.Type C is the combination of A and B types.Literature Review 122.1.4 Models for corrosion fatigueBased on the types of crack growth, three major models have been proposed for corrosionfatigue. Superposition model [45,46]Initially, the superposition model assumed that the rate ofcrack growthin any environmentcan be predicted by summing the crack growth rate for pure mechanical fatigue and the rate forstress corrosion cracking in the same environment [45], analogous to the type B situation. Later,a third term was added to include the synergistic interaction between mechanical fatigue andenvironmental actions [46]. Therefore the crack growth rate in any aggressive environment isof the form:(daldN)e = (daldN)r + (da/dN) + (da/dN) (2.3)where (dafdN)r, (da/dN)scc and (daIdN) are the cracking rates for pure mechanical fatigue,SCC and mechanical fatigue - environment interactions, respectively. Competition model [47]The competition model is based on the assumption that the fatigue crack growth rateinvolves competition among various processes. It suggests that the processes of fatigue andstress corrosion cracking are mutually competitive and a crack will grow at the fastest availablerate pertinent to the prevailing stress intensity factor.Literature Review Interaction model [48]The interaction model was developed to allow interactions between mechanical fatigueand stress corrosion cracking to occur, particularly in alloy systems which exhibit a high degreeof susceptibility to stress corrosion cracking. In this model, the effects of predominantlymechanical and predominantly electrochemical crack propagation processes are added, but eachis modified to account for the presence of the other. Owing to interactions, one process maybe inhibited or enhanced by the action of the other.2.1.5 Mechanisms for corrosion fatigueThe main mechanisms by which the environment may affect the crack growth can be putinto three categories as oxidation-related mechanisms, hydrogen assisted cracking mechanisms,and surface-related mechanisms. Oxidation-related mechanismsThe main mechanisms include:1. Anodic or strain enhanced dissolution [49,50]. Preferential anodic dissolution of thestrained material at the crack tip increases the extent ofcrack growth during each cycle.The anodic dissolution can also help by dissolving the cyclically strain hardenedmaterial or other barriers to slip processes (e.g. a hard precipitate) at the crack tip andthereby sustaining the deformation related cracking process.2. Film rupture and dissolution [51]. The protective film is ruptured at the crack tip bythe cyclic strain from the alternating stress. The bare surface exposed acts as anodeand the rest of the film covered metal serves as cathode. This results in localizedLiterature Review 14corrosion at the crack tip and crack growth.3. Restricted slip reversibility [41,42]. The bare metal formed in the forward slip part ofa fatigue cycle is covered with oxide patches. These oxide patches restrict slipreversibility on the same slip plane during the reverse part of the cycle, in addition tothe work hardening effects on the slip system. The amount of restricted slip is equalto the crack growth per cycle. Hydrogen assisted cracking mechanismsIn these mechanisms, hydrogen (either generated by corrosion processes during fatigueor pre-existing in the metal) accumulates in the region ahead of the crack tip and contributes tohydrogen assisted cracking via one or a combination of the following processes [35,52,53]:1. Pressure buildup. Molecular hydrogen precipitates at internal interfaces or defects andgenerates a pressure that is in equilibrium with the local activity of the atomic hydrogendissolved in the metal lattice. The pressure augments the applied tensile stress andassists crack propagation.2. Decohesion. Presence of hydrogen at the crack tip region causes a reduction ofmetal-metal bond strength and allows mechanical separation to occur more easily.3. Hydride formation. Hydrides are well known to exhibit brittle behaviour. The formationand cracking of brittle metal hydrides will facilitate advance of the fatigue crack. Surface-related mechanismsThe mechanisms include:1. Surface energy reduction. Reactions at the crack tip that produce adsorbed species mayLiterature Review 15lower the surface energy and increase the ease of crack growth [49,50,54], because ofthe lower energy required to create new surfaces.2. Adsorption-induced localized slip. Chemisorption of species facilitates the injectionof dislocations from crack tips and thereby promotes the coalescence of cracks withvoids ahead of cracks [25,55].Processes 1 and 2 constitute another form of hydrogen embrittlement when hydrogen isthe adsorbing species.2.1.6 Crack closure effectsIt has been found that a fatigue crack can be in the closed state, with opposing crack tipsurfaces in contact, even during part of the tensile load cycle. Therefore, based on the conceptthat the load is effective in propagating the crack only when the crack is open, this leads to thedefinition of the effective stress intensity range,effKmax(do (2.4)where K10 is the stress intensity value at which the crack begins to close. Many crack closuremechanisms have been identified, the principal ones are shown schematically in Figure 2.6 [56].The phenomenon of crack closure is now widely acknowledged to exert a strong influence onnear threshold fatigue crack growth. Closure ideas have helped to explain, at least in a qualitativemanner, many crack propagation effects, especially concerning the near threshold region.However, the crack tip sharpening effect and changes to the plastic zone during the compressivepart of the loading cycle, or after the crack is closed, is not well accounted for in thesemechanisms.Literature Review 16(a)(b)(c)Plasticity -InducedClosureOxide-InducedClosur•Roughness-InducedClosureCd)tronsfcwmed zone4 4 4 4 4444a.Ce) 4I, tt+t -+Viscous Fluid-InducedClosurePhase Transformation-inducedClosureFigure 2.6 Schematic illustration of the principal mechanisms of fatigue crackclosure [56]. Plasticity induced closurePlasticity induced closure arises from the fact that the propagating crack generates a plasticzone (enclave) in the crack tip region that is contained in an elastically strained matrix. Due toelastic constraint from the uncracked material, the crack tip region is placed under compressivestress on the closing portion of the load cycle. The resulting effect is to press the sides of thecrack together near the crack tip before the gross applied load reaches zero. This compressiveLiterature Review 17stress effectively imposes a constant stress on top of the applied cyclic stress, shifting the meanstress of the cycle to a lower value. Therefore, if the applied load cycle is at a sufficiently lowR ratio the residual stress will bring part of the load cycle into the compressive region, creatingclosure effect, even when R is positive (R = KLfl/KX = Ptnji/Pinax).This fundamental source of closure was first conjectured and later establishedexperimentally by Elber [57]. After Elber’s pioneering work, other sources of crack closurehave also been identified. Oxide induced closureWhen oxide deposits or corrosion products build up on the fracture surfaces to the extentthat their thickness is of the same order of magnitude as the calculated crack tip openingdisplacement, crack closure may result.Ritchie, Suresh and coworkers [58,59] reported the first quantitative experimental studyof this kind of closure, in which they noticed that oxide deposits on the fracture surfaces of asteel tended to be more pronounced near threshold and at low R values, causing black bands ona fracture surface which delineate the position of the crack front under near threshold loading. Roughness induced closureRoughness induced closure arises from the scenario that the relativemotion of the opposingfracture surfaces at the crack tip may contain some Mode II and ifi displacements, even thoughfatigue may be occurring under nominally Mode I displacements. In particular it has beenestablished that significant Mode II stresses operate at the crack tip even under far field ModeLiterature Review 18I loading. When any such relative motion occurs, the two opposing faces of the crack will notfit together perfectly on the closing part of the cycle. Opposing asperities will come into contactat some point and, since these can transmit stresses across the crack faces, some degree ofclosurewill occur.This kind of closure was first suggested by Purushothaman and Tien [60], and Walkerand Beevers [61]. Viscous fluid induced closure [56]Fluid induced closure relies on the fact that a viscous fluid is capable of transmitting stress.Therefore a crack filled with, say, oil, may demonstrate some closure effects at all stress levels,depending on the loading frequency. This mechanism is confined to fluids which are viscousenough to exert an effect but also capable of flowing into the narrow openings afforded by nearthreshold cracks. Phase transformation induced closure [56]In materials which undergo stress or strain induced phase transformations under cyclicloading, a further mechanism of fatigue crack closure can result. This phase transformationinduced closure arises also from the constraint effect of the surrounding elastic material on thetransformed region, analogous to the plasticity induced closure mechanism. When the phaseformed is more voluminous than the original phase, a compressive residual stress results, leadingto crack closure.Literature Review 192.1.7 Rate controlling process in corrosion fatigueCorrosion fatigue involves many processes, any of which can act independently orsynergistically to control the cracking rate. A list of such processes includes [35,49-63]:1. Transport of reactants to the crack tip region.2. Rupture of a protective film by straining.3. Dissolution of freshly exposed metal.4. Passivation of freshly exposed metal.5. Production of hydrogen atoms.6. Adsorption of deleterious species (hydrogen atoms).7. Diffusion of hydrogen atoms into the matrix and ahead of the crack tip.8. Promotion offracture by the deleterious species (e.g. surface energy reduction, hydrogenembrittlement).9. Build up of corrosion products which influence crack closure and the effective stressintensity range.10. Transport of corrosion products from the crack tip region.2.2 Corrosion Fatigue of Li-Containing Aluminium AlloysCorrosion fatigue is related to both the mechanical and corrosion properties of the materialand the interaction between them. Therefore, the mechanical properties, fracture toughness,corrosion and stress corrosion cracking behaviour of Li-containing alloys will also be includedin the review. Adequate knowledge of these properties is necessary for the understanding ofthe fatigue and corrosion fatigue behaviour of Li-containing aluminium alloys.Literature Review 202.2.1 Brief historical development of Li-containing aluminium alloysThe development of Al-based alloys containing lithium began in Germany in the 1920’s[64] and was primarily concerned with additions of small amounts of lithium to age-hardeningalloys to increase their strength. In the 1950’s, metallurgists at ALCOA recognized that lithiumincreases the elastic modulus of aluminium and developed the high strength Al-Cu-Li alloy2020 [65]. In the mid-1960’s Al-Mg-Li alloy 1420 was introduced in the former Soviet Union[661. However, there were only limited applications because these early generationLi-containing alloys exhibited major problems relating to poor ductility and low fracturetoughness.Major development of Al-Li alloys began in the 1970’s [2,3,67]. The stimulus was theconsistent pursuit of materials with high strength/weight ratio, which is a critical requirementof the airplane and aerospace industries. Moreover, to meet demands for increased fuelefficiency, one of the most feasible methods is to replace conventional construction materialswith new materials which have similar structural characteristics but lower specific weight.Unlike other alloying elements in aluminium alloys, lithium increases the elastic modulus andstrength while reducing the density of the alloys. Each weight percent lithium added to analuminium alloy reduces the density approximately 3% and increases the elastic modulusapproximately 6% for lithium additions up to 4% (which is the maximum solid solubility of Liin Al at 610 °C) [1,3]. Therefore the specific stiffness can be increased significantly (seeAppendix I). As such, the development of Li-containing alloys makes aluminium alloys moreresistant to replacement by composite materials, with the added advantage that Li-containingaluminium alloys can be produced and processed by existing equipment with only slightmodifications [1,2,4,68].Literature Review 21Alloy development has led to the introduction of two of the most promising Li-containingalloys, Al-Li-Cu-Zr alloy 2090 by ALCOA and Al-Li-Cu-Mg-Zr alloy 8090 by ALCAN in themid-1980’s. Generally, the 2090 alloys are planned primarily as a replacement for theconventional high strength 7000 series Al-alloy, and the 8090 alloys are intended to substitutefor conventional 2000 series Al-alloys in medium strength and damage tolerant applications[67,69,70].Several conferences have been held in the past decade [71-76] that relate specifically tothe development, characterization, and properties of Li-containing Al-alloys.2.2.2 Mechanical properties and fracture toughness of Li-containingaluminium alloysThe major strengthening phase in Li-containing aluminium alloys is the metastable 6’(A13Li) phase which is coherent with the Al matrix, although a study by Noble et al. [77] foundthat the main increase in modulus of Li-containing aluminium alloys comes from lithium insolid solution, and to a lesser degree from the 6’ phase. The major strengthening increment inall Al-Li based alloys comes from the formation of this phase upon age-hardening. In addition,Li and other alloying elements also contribute to strength by solid solution strengthening andother precipitation hardening processes (e.g., formation of T1 (A12CuLi), 0’ (AI2Cu) and I’(AI3Zr) in Al-Li-Cu-Zr alloys, andT1(AI2CuLi), S (A12CuMg) and f3’(AlZr)inAl-Li-Cu-Mg-Zralloys) [1,3,78,79].The fatal problem with the early generation Al-Li alloys was poor ductility and toughness,which was the cause for the withdrawal ofthe early 2020 Al-Li alloy from commercialproductionLiterature Review 22in the late 1960’s. Extensive studies attributed the poor ductility and toughness to several causes.These include grain boundary segregation of impurity alkaline elements (e.g. sodium andpotassium) introduced with the lithium addition [3,80-84], strain localization at grain boundariesassociated with the shearing of W phase by dislocations and resultant planar slip [1,3,85,861,strain localization favoured by the highly textured structure [85], formation of a precipitate-freezone (PFZ) at grain boundaries and large grain boundary precipitates [3,86-92]. Considerableeffort has been made to combat these problems by (i) increasing the purity of the alloying Lior adding impurity bonding elements, like bismuth, silicon or iron [93-95], (ii) addition ofdispersoid forming elements like Mn and Zr to promote more homogeneous deformation [79],(iii) addition of solid solution and precipitate hardening elements (Mg and Cu) [79,96], (iv)thermomechanical processing [87], (v) mechanical alloying and powder metallurgy [97,98].While slip planarity is generally considered to have a harmful effect on fracture toughness,studies by Suresh et al. [92,99,100] showed that it can also lead to a concomitant beneficialgeometric effect on toughness by causing crack bifurcation. Crack bifurcation occurs in theunder-aged Al-Li-Cu-Zr alloys when the amount of grain boundary precipitates is insufficientto cause intergranular fracture, leading to enhancement in crack growth resistance.For the newly developed Al-Li alloys like 2090 and 8090, the goal of improving ductilityand toughness without significant loss in other mechanical properties has been reached withgreat success. Moreover, excellent cryogenic properties have been found with Al-Li alloys, asshown by increasing strength, ductility and toughness with decreasing temperature[81-83,101-110]. The increased fracture toughness values of Al-Li alloy plates in the L-T andT-L orientation at low temperature have been attributed primarily to the effects of crack-dividerdelamination toughening [107,108]. Modem Al-Li alloy plates usually have the pan-cake shapedLiterature Review 23grain structures, analogous to laminates. These structures lead to anisotropic mechanicalproperties. For cracks in laminate material, three distinct orientations exist, as shown in Figui2.7 [107]. According to the delamination toughening mechanism, the enhanced splitting(delamination) along the weak interfaces not only consumes energy but also changes the stressstate at the crack tip from plane strain to plane stress, resulting in increased toughness. On theother hand, some studies attributed the increased toughness at low temperature to thesolidification of embrittling liquid impurity phases of alkaline metals [82] and the greaterhomogeneity of plastic deformation with increased strain hardening [103-105].(L—T,T—L)______arresr____ ____ ____(T—s,L--S)Crackdelamination(S—L,S—l’)Figure 2.7 Terminology used to describe the various orientations for crackextension in an anisotropic material containing specific planes of weakness in onedirection [107].Literature Review 24Associated with the anisotropic grain structure of Al-Li alloy plates, fracture toughnessvalues are low when the crack plane is parallel to the plane of the pancake-shaped grain (i.e. inthe S-L or S-T orientation). In order to improve the low fracture toughness of Al-Li alloy8090-T8771 in these orientations, a re-aging heat treatment has been proposed [111-114]. Thisre-aging treatment has been shown to increase the S-L fracture toughness by 2 to 3 times withonly slight sacrifice in strength (-.7%). However, different mechanisms [112-114] have beenreported to explain the effects of the heat treatment. These include grain boundary Lidesegregation by Lynch [112] and reduced slip planarity by Blankenship et al. [113] and Slaviketal. [114].2.2.3 Corrosion and stress corrosion cracking of Al-Li alloysIt is expected that Li-containing aluminium alloys will be more susceptible to corrosionthan other aluminium alloys, because of the reactive nature of lithium. Studies have found thatcorrosion is generally associated with the precipitates in the material [115]. In addition to thewell known deleterious effect on corrosion by the S (A1Li) phase, which is anodic to thealuminium matrix [116,117], corrosion has also been attributed to other precipitates, like T1(AI2CuLi) and S’(A12CuMg) [118-120] andT2(AI6CuLi3)[121], to the formation ofPFZs [122]associated with precipitation, and to the increasing size distribution of 5’ precipitates with aging[123]. These precipitates cause the formation of galvanic cells, leading to corrosion. Whethera heat treatment improves or degrades the corrosion behaviour depends on its effect on theformation of the precipitates and PFZS and the resultant driving force for the corrosion process.There are some “indirect studies suggesting that the Li content in the Al alloys is beneficialfor their corrosion resistance. The result obtained by Gui et al. [124] from a corrosion study ofLiterature Review 256061 Al-alloy in aqueous solutions with and without Li ions suggested that Li in solution hasa beneficial effect on corrosion resistance. Instead of increasing corrosion, they found that Liin an alkaline solution decreases the corrosion rate by incorporating Li in the formation of aprotective film that results in anodic passivation behaviour. When Li is not present in thealkaline solution, passivation behaviour is not observed. The result of Fernandes and Ferreira[125], based on electrochemical polarization measurement of pure Al in mixed solutions ofNa2CO3and Li2CO3,also showed that the presence of Li in solution improves the passivationof aluminium above a critical concentration of carbonates. The presence of chlorides above acertain level in these solutions is deleterious to the passivating films, causing film breakdownand pitting. Generally, the corrosion resistance of Al-Li alloys is equivalent to or better thanthe conventional alloys that they are intended to replace.Similar to the conventional high strength aluminium alloys [126,1271, the stress corrosioncracking susceptibility of Li-containing alloys is of great concern. Studies of SCC behaviourof Al-Li alloys in sodium chloride test solutions have shown that SCC susceptibility changeswith material orientation, heat treatment, electrochemical potential, loading and environmentexposure conditions [128-145].A study by Braun et at. [128] on peak-aged 8090 alloy showed that it is susceptible to SCCin the short transverse (ST) direction but not in the long transverse (LT) direction. Lumsden etal. [129] also found much higher SCC severity in the ST than in the longitudinal direction intheir study on 8090 alloy and that SCC behaviour is insensitive to aging treatment. A study byDorward et al. [130] showed that the under-aged 2090 alloys are susceptible to SCC in both LTand ST directions, but the near peak-aged alloy is immune to SCC.Literature Review 26Although it is established that over-aging improves SCC resistance for conventional highstrength aluminium alloys [126,127], many studies on Al-Li alloys [130-140] have shown thatheat treatment produces different effects. Studies by Christodoulou et al. [131] on binary Al-Lialloys and Lane et al. [132] on Al-Li-Cu-Mg alloys showed that the peak-aged tempers are mostsusceptible to SCC and over-aging improves SCC resistance, similar to conventional Al alloys.Gray [133] showed that under-aged tempers of Al-Li-Cu-Mg alloys exhibit greater SCCsusceptibility than either peak-aged or overaged tempers. Rinker et al. [134] reported that theSCC resistance of Al-Li alloy 2020 is excellent in both peak-aged and under-aged tempers.Meletis [135] reported that the peak-aged temper of an Al-2.9Cu-2.2Li-0.l2Zr alloy is mostresistant to SCC, over-aging induces susceptibility to SCC owing to the development of strainat the grain boundaries resulting from preferential T1 precipitation. Vasudévan et al. [136] andBalasubramaniam et al. [137,138], reported a decreasing SCC resistance with increasing agingtime. A recent study by Hu et a!. [121] showed that over-aging increases SCC resistance ofAl-Li alloy 8090 with simultaneous significant reduction in tensile strength, while aretrogression and re-aging process is ab1e to improve the optimum combination of tensilestrength and the SCC resistance, i.e., achieving the high SCC resistance of the overaged temperwithout losing the high ultimate tensile strength and yield strength of the peak-aged temper.Lumsden et a!. [129] studied the effect of electrochemical potential on SCC behaviour of8090 alloy in 3.5% NaC1. They found that susceptibility to SCC depends on electrochemicalpotential. There is a critical potential below which SCC does not occur. The critical potentialis near the pitting potential (only about 70 mV below E1J. Strain-to-failure decreases rapidlyas the potential approaches the pitting potential.Literature Review 27While most workers [128-138,141-145] have studied SCC of Al-Li alloys using a varietyof testing methods (e.g. alternate immersion, drip feed or constant immersion together withconstant strain, constant load, slow strain rate technique (SSRT) on smooth specimens and boltloaded precracked double cantilever beam (DCB) specimens), Hoiroyd et at. [146] and Craiget at. [147,148] have drawn attention to the influence of the environment exposure conditionson SCC of Al-Li alloys. They found that SCC does not occur in 3.5% NaC1 under constantimmersion conditions with either smooth or precracked specimens under constant load (orstrain), but occurs under alternate immersion orpre-exposure testing conditions. They proposedthat the local chemistry at the crack tip, or within other limited geometries, in Li-containingalloys is alkaline after the specimen is removed from the bulk solution. This alkaline chemistryis opposite to the acidic local chemistry that forms inside long cracks under total immersion inAl-Li alloys (as proven by artificial crevice corrosion tests [119,147,148]) and in conventionalAl-alloys [149]. They explained that under alternate immersion, drip feed or pre-exposuretesting condition, a critical balance between activity and passivation necessary for SCC candevelop, due to the absorption of CO2 from the atmosphere. They supported their explanationwith further test results that showed exposure of the test solutions to CO2 determines whetherSCC occurs, and that SCC occurs inLi2CO3ILiHCO buffered NaCl solution under constantimmersion condition. The need for CO2 in the SCC was attributed by Moran et at. [150,151]and Buchheit et at. [152] to an increase in carbonate concentrations which eventually leads tothe passivation of blunted fissures by precipitation ofLi2CO3. These studies also attributed thelocal alkaline chemistry to the reduction ofprotons (H) that occurred within the crack or creviceupon removal of the bulk solution. However, several researchers [141,142,145] have reportedresults of SCC susceptibilities obtained with DCB specimens under constant immersionconditions, which are contrary to the above assertion on environment exposure conditions.Literature Review 28The proposed mechanisms of SCC of Al-Li alloys are similar to those for conventionalaluminium alloys. Both anodic dissolution and hydrogen embrittlement have been proposed.Anodic dissolution mechanisms are based on the potential difference between grain boundaryprecipitates (such as 6, T1 and T2), PFZs and the interior of grains (and possibly the solutedepleted zone) [121,130-134,140,144,150-153]. Hydrogen embrittlement mechanisms arebased on pre-exposure embrittlement [146], the discontinuous nature of crack growth,hydrogen-generating corrosion reactions behind the crack tip [131,141], hydride formation[137,138], and facilitated hydrogen entry by T1 phase [154,155]. Hydrogen has been shownto dramatically decrease tensile properties of 2090 alloy [156,157]. The recent work of Lee etat. [145] on an Al-Li-Zr alloy proposed a transition in the SCC mechanism from anodicdissolution in stage I to hydrogen embrittlement in stage II by considering the temperaturedependence of SCC susceptibility based on the anodic dissolution and hydrogen embrittlementmodels.It is probable that anodic dissolution and hydrogen embrittlement both play a role in SCCof Al-Li alloys. In comparison with the conventional Al-alloys, the SCC resistance of Al-Lialloys is graded as equivalent or superior.2.2.4 Fatigue and corrosion fatigue of Al-Li alloysLow resistance to fatigue crack initiation was observed for the early generation Al-Lialloys such as 2020. This was attributed to strain localization caused by the shearing of thecoherent 6’ phase by dislocations and resultant planar slip [158,159]. In similar manners to thefracture toughness, fatigue crack initiation resistance can be improved by dispersing the planarslip and increasing the strain hardening ability, e.g., by adding dispersoid forming elements andLiterature Review 29producing semi-coherent/incoherent precipitates [90] or by mechanical alloying to incorporatefmely distributed oxide and carbide particles [98]. The recently developed Al-Li alloys, like2090 and 8090, have equivalent or better fatigue crack initiation resistance than the conventionalAl-alloys [6].The fatigue crack growth resistance of Al-Li alloys is generally superior to theconventional aluminium alloys [6]. Coyne et al. [158] attributed the improved resistance oftwo Al-Li-Mn and Al-Cu-Li alloys to sub-critical crack growth at low cyclic stress intensities(about 4 to 10 MPa.m)to the increase in elastic modulus by addition of Li, which leads todecreased crack opening displacement (COD) and hence decreased fatigue crack growth rate[160].Besides changes in elastic modulus, Bretz et al. [159] and Vasudévan etal. [161] attributedthe much slower crack propagation rates of alloy 2020 relative to those of alloy 7075 to thecoarse, recrystallized grain structure and highly non-linear and deflected crack proffles resultingfrom planar slip of the Al-Li alloy. The cracking rates in alloy 2020 are an order of magnitudelower at intermediate AK levels and up to 2 orders of magnitude lower in the near-thresholdregion than the cracking rates in alloy 7075. Non-linear and deflected crack surfaces cansignificantly reduce the effective crack propagation driving force, AK [162-164].In studies with three high purity Al-alloys containing different amount of lithium andcopper, Vasudévan et al. [7] and Petit et al. [11] found that increasing the Li/Cu ratio inaluminium alloys leads to an improved resistance to cyclic crack growth at room temperaturein moist air and vacuum environments. They explained that differences in elastic moduli of thematerials are not the cause of their different resistances to cyclic crack growth, as considerabledifferences still exist between the materials when propagation rate da/dNis plotted against AKIE.Literature Review 30The beneficial fatigue properties appear to arise from the crystallographic crack growthmechanism and a crack deflection process induced by the 6’ precipitates in the alloys of higherLi content.Harris et a!. [8] studied the effect of grain size and shape on mechanical properties andfatigue resistance with 8090 alloys. In addition to the reduced fatigue crack growth rate of 8090at AKvalues close to threshold and up to 20 MPa.mtt2,relative to the 2000 series Al-alloys, theyfound that for the Li-containing alloy the fatigue crack growth rate is slower when the grainsize is large and when subgrain boundaries contain the precipitated S (A12CuMg) phase. Theimprovement in fatigue properties with increasing grain size is largely due to the highly facetedcrack path (on { 111) slip planes), the resulting surface roughness causing early crack closureand hence reducing the effective AK range at the crack tip [162]. High slip reversibility inLi-containing alloys also contributes to the improved fatigue properties. The improvement offatigue crack propagation resistance with increase in grain size was also indicated by Ruch eta!. [165] in their study on fatigue propagation in mechanically alloyed Al-Li-Mg alloys. Theyelucidated that increased grain size allows an increase in slip length, resulting in more reversibleslip, and that sharp texture increases the effective grain size by making slip transfer across grainboundaries easier, because of small misorientation angles between neighboring grains.Rao et at. [9,10,166-168] studied fatigue crack propagation of long and short cracks indifferent Li-containing aluminium alloys. They found that fatigue crack growth is highlyanisotropic for long cracks [9,166]. In a commercial 2090 alloy, L-T, T-L and T-S orientationsshow the best crack growth resistance, and S-L, S-T and L-t-45° show the worst [9]. Differentgrain orientations cause different magnitudes of crack tip shielding (reduction in local crackdriving force) arising from crack deflection and roughness induced crack closure. Short crackLiterature Review 31propagated much faster than long cracks at equivalent nominal AK, and propagated below thelong crack threshold stress intensity [10,167,168]. The major reason for this behaviour wasattributed to the lack of crack deflection and roughness induced closure for short cracks. Theabsence of an intrinsic threshold, together with crack growth below the closure-corrected AKeffthreshold for long cracks, was explained by factors such as enhanced cyclic plastic strain at thetip of small cracks and the differing statistical sampling effect of large and small cracks withmicrostructural features [33,34].Studies by Yoder eta!. [169] related the extraordinary fatigue fracture surface roughnessand crack closure level of the Al-Li alloy 2090-T8E41, which are responsible for its uncommonlygood fatigue crack growth resistance, to the microstructure of the material. They concludedthat the slip band facets and the unusual height of asperities in the fatigue fracture surface ofthe Al-Li alloy plate are the consequence of (i) the propensity for a planar slip mode, and (ii)the alloy’s unusually intense crystallographic texture ({ 110) <112> and { 123) <634>).Oxide induced closure was also investigated in some of the above studies [7,9,98], butwas found to have only a relatively minor effect when compared with the crack tip openingdisplacement (CTOD).Al-Li alloys exhibit strong tensile overload retardation effects on crack growth[11,159,170,171], but they are also very sensitive to periodic compression overload cycles[9,13,172]. In the latter case, the compressive loads act to reduce closure effects by crushingthe asperities, which decreases roughness induced closure stresses, causing accelerated crackpropagation rates and crack growth at the threshold.Literature Review 32Besides the increased toughness with decreasing temperature for Al-Li alloys[81-83,101-1101, fatigue resistance was also found to increase with decreasing temperature[104,1731 and attributed to deeper and larger delamination that occurred on the fracture surface[173]. Also, it has been noticed that factors which improve resistance to fatigue crackpropagation tend to have a detrimental influence on fatigue initiation [17], such as grain size,and planarity of slip.While structure related effects play an important role in fatigue resistance ofLi-containingaluminium alloys, the influence of environment is another key factor that needs to be includedin corrosion fatigue of Al-Li alloys. Vasudévan et al. [161] found with an Al-Li-Cu alloy thatonly a small difference in fatigue crack growth rates can be detected between moist air (90%relative humidity (RH)) and dehumidified helium (<3 ppm moisture). It was suggested thatdue to the highly reactive nature of lithium with moisture, a moisture content of even less than3 ppm can cause embrittlement. Many studies conducted in air and vacuum or dry argon haveshown that the crack propagation rate can be accelerated and the threshold value decreasedwhen tested in air [11-17,20]. The presence of water vapor, liquid water and sodium chloridesolution can further accelerate fatigue crack propagation rate and decrease the threshold[13,15-19,142,174], although sometimes the opposite effects have been observed [17,18]. Withsmooth specimens, environmental effects lead to reduced fatigue lifetime and fatigue limit[175-1801.Many mechanisms have been proposed to explain the influence of environment on fatiguecracking resistance. These are primarily anodic dissolution [12,176,178,180], hydrogenembrittlement [13-16,176,179,180], environmental effects on crack closure level[15,17-19,179] and on slip reversibility [15,20], or a combination of any of them. For example,Literature Review 33anodic dissolution is assumed to accelerate crack initiation and crack propagation in a corrosivesolution through pitting [178,180], intergranular corrosion [12,176], and slip-enhanceddissolution [12,180]. On the other hand, the studies of Piascik et at. [16] suggest that a hydrogenembrittlement mechanism may be operative. They reported the following results as supportiveto the hydrogen embrittlement mechanism: (i) the fatigue crack growth is accelerated by thepresence of only ppm levels of water vapour where condensation is not possible; (ii) thedependence of da/dN on water pressure agrees with the predictions of an impeded molecularflow model; (iii) the fatigue crack growth is not affected by the presence of film-forming 02,and is the same as those in purified helium and vacuum; (iv) the crack paths in the ppm levelwater vapour and the aqueous NaCl environments have identical brittle morphologies andprocess zone volume dependence [181]. Pao et a!. [19] concluded that the presence of anaggressive environment may enhance or retard fatigue crack growth depending upon whetherthe crack closure level is reduced or elevated during corrosion fatigue, based on their study of2090 and 7075 in air and 3.5% NaC1 solution. Fatigue crack growth rates of alloy 2090 in 3.5% NaC1 were increased approximately six fold relative to those in air, due to the conjoint actionof salt water corrosion and mechanical fretting which largely removed the fracture surfacetortuousity and reduced roughness induced closure. Tests on alloy 7075 showed that in the lowAK region, corrosion product induced crack closure in salt water and led to lower fatigue crackgrowth rates than those in air. Peters et a!. [17,18] also attributed the higher threshold value ofAl-Li alloys in aqueous NaCl, relative to air, to corrosion product induced crack closure. Studiesby Jata et at. [20] and Shin et a!. [15] found that although crack path tortuousity, and hencecrack deflection and fracture surface roughness, were increased in vacuum relative to those inair, the crack closure level was lower in vacuum. They proposed that in vacuum the slipreversibility increases due to the lack of a significant surface oxide. Thus although fractographyLiterature Review 34ofvacuum tested samples indicated extensive faceting, it did not result in increased crack closurecaused by the mode II displacements behind the crack tip. The extent of slip reversibility playsan important role in the fatigue crack growth rates of planar slip material. The lower fatiguecrack growth rate in vacuum is due to increased slip reversibility rather than an increase in theroughness induced crack closure.Overall, the corrosion fatigue resistance of Al-Li alloys is superior or equivalent to theconventional aluminium alloys. The addition of reactive lithium element does not appear tocause severe environmental sensitivity problems, while the increase ofelastic modulus, strengthand sup reversibility, and especially the “extrinsic toughening” effects from crack deflection,roughness and branching, all contribute to improve the fatigue crack growth resistance. Ideally,we would like to have new materials with both improved fatigue crack propagation resistanceand increased fracture toughness. Therefore, when comparing to the materials to be substituted,the new materials will not only exhibit lower crack growth rates at comparable AK but also cansustain longer critical crack lengths before unstable fracture occurs (i.e. when K,, in the fatiguecycle reaches K1). Consequently, much longer service lifetime will be obtained for theengineering structures or components made from the new materials.Objective 353 ObjectiveThe main objectives of the present study are as follows:1. To evaluate the corrosion fatigue crack propagation behaviour of the Al-Li alloy 8090in the T8771 temper in the T-L orientation, covering a broad range of fatigue crackgrowth rates from the near threshold (Stage I) to the Stage II region, and to investigatethe effects of controlled environments and cyclic load frequency on the fatiguebehaviour.2. To compare the fatigue crack propagation behaviour of the Al-Li alloy in the T-Lorientation with that of the conventional aluminium alloy 2024 in the T35 1 temper.3. To study the corrosion fatigue crack propagation behaviour of the Al-Li alloy in theshort-transverse orientation (S-L) and to determine the effects ofa toughness-enhancingre-aging treatment on the fatigue and corrosion fatigue crack growth resistance in theS-L orientation.4. To gain further understanding of the mechanisms related to the generally reportedsuperior fatigue crack growth resistance of Li-containing aluminium alloys, byassessing the influence of alloy composition, microstructure, deformation behaviour,environmental sensitivity and electrochemical corrosion behaviour on fatigue crackingresistance.Objective 365. To clarify the relative importance of anodic oxidation processes, hydrogen evolutionand surface-related processes in the mechanism of corrosion fatigue crack growth, andthe possible roles they play in affecting corrosion fatigue crack propagation in Al-Lialloys through effects on crack closure level and/or slip reversibility.Experimental 374 Experimental4.1 Materials4.1.1 Chemical compositionsThe materials used for the study were commercial Al-Li 8090 alloy plate of 12.7 mmthickness, and commercial Al-Cu 2024 alloy plate also of 12.7 mm thickness. The nominalchemical compositions [2,67,182] of the alloys are given in Table 4.1.Table 4.1 Chemical Compositions (Nominal wt.%)Si Fe Cu Mn Mg Cr Zn Li Zr Ti Al8090 0.20 0.30 1.0- 0.10 0.6- 0.10 0.25 2.2- 0.04- 0.10 balance1.6 1.3 2.7 0.162024 0.50 0.50 3.8- 0.30- 1.2- 0.10 0.25 0.15 balance4.9 0.9 Mechanical PropertiesThe Al-Li alloy was received in the T877 1 temper, which involves a solution treatmentat 545 °C that is followed by a cold water quench, a 7% stretching operation, and a final agingtreatment at 170 °C for 32 b [112]. The Al-Cu alloy was received in the T351 temper, whichExperimental 38generally consists of a solution treatment at 495 °C, a 1.5-3 % stretching and natural aging[182,183]. The actual and nominal mechanical properties [2,184,185] of the alloys in theas-received conditions are listed in Table 4.2.Table 4.2 Physical and Mechanical Propertiesp E e K1(g/cm3) (GMPa) (MPa) (MPa) (%) (MPa.m’)T-L S-L8090-T8771 * 540 480 9 28 148090-T8771 2.55 77 460-515 380-450 4-6 13-30 16minimum typical2024-T351 * 490 395 18 30 302024-T351 1185] 2.78 73 470 325 19 32 26* Data supplied by Morrison [184].4.1.3 Grain structuresThe grain structures of both materials were examined in three orthogonal planes, withtheir normals in directions L, T and S, where L is the longitudinal or rolling direction, T is thetransverse direction and S is the short-transverse or through-thickness direction. Subsequently,Experimental 39planes normal to the L, T and S directions will be designated as the L-plane, T-plane and S-planerespectively. The specimens were ground with SiC paper down to 600 grit and subsequentlypolished with diamond suspensions to a 1 im finish. The grain structures were revealed byimmersion etching in Keller’s etchant for 10-20 seconds, followed by washing in a stream ofwarm water and ethanol, and drying in a stream of air. The Keller’s etchant is composed of thefollowing reagents [186]:1.0 ml HF, 1.5 ml HC1, 2.5 ml HNO3 and 95 ml H20.Both plates exhibited anisotropic grain structures that are characteristic of the commercialmaterials, as shown in the composite optical micrographs in Figure 4.1. The grain dimensionswere not very uniform in the Al-Li alloy, even when measured and compared within each ofthe 5, T, L directions, and indicated a partially recrystallized structure. The grain thickness inthe S-direction varied by a factor of —5 with a mean of —7 x iO mm. Grain diameters in therolling plane (S-plane) were larger, but even less uniform, varying by a factor of —19 with amean of 2 x 102 nim in the T direction, and varying by a factor of —20 with a mean of —7 x 10.2mm in the L direction. The grain dimensions were relatively uniform in each direction in theAl-Cu alloy, witha mean of—3 x 1(12 mm, —7 x 1(12 mm and —1.3 x 10.1 mmin the 5, Tand Ldirections respectively.4.1.4 The re-aging treatmentA re-aging treatment was applied to the Al-Li alloy to investigate its effects on the S-Lorientation corrosionfatigue crackpropagationresistance. The denotation ofcrack plane followsthe conventions as defined in ASTM E399-78a [1881 and is shown here in Figure 4.2. It hasbeen shown that a re-aging treatment at 230 °C for 10 minutes can largely improve the fractureS(a)(b)Figure4.1Grainstructuresof(a)Al-Li8090-T8771and(b)Al-Cu2024-T351.LI-_,_.-:I,__.g%“k:‘.•:T200pmExperimental 41toughness in the short-transverse orientations with only minor decrease in strength [112]. Inthe present study, the re-aging treatment was conducted in a salt bath of —50% NaNO3+ —50%KNO3 [187] with coupons near the fmal size for the S-L orientation double cantilever beam(DCB) specimens. The salt bath was heated in an electric furnace at 230 °C. Each coupon wasput in the salt bath for 10 minutes and then quenched in ambient temperature tap water (—20°C). The DCB specimens were made from these coupons after the heat treatment.4.2 Specimen design and preparation4.2.1 Electrochemical test specimensSmall square samples of—15 mm x 15 mm x 12.7mm were cut from plates of both alloysto make the specimens for electrochemical tests. Each metal section was mounted in epoxyresin to leave only one surface exposed. The exposed surface was metallographically polishedto a 1 jim fmish and cleaned with ethanol. The exposed edges between the epoxy and alloywere masked with cellulose acetate lacquer to prevent crevice corrosion effects during theelectrochemical tests. A copper wire was electrically connected to the unexposed rear face ofeach specimen and passed through a glass tube that was sealed into the epoxy mount, as shownschematically in Figure 4.3. The polished and lacquer masked specimens were stored in adesiccator before use.IFigure4.2Crackplaneorientationidentificationcodeforplate[188].EpoxyResinH\,1GlassTube /CopperWire/Figure4.3Geometryoftheelectrochemicaltestspecimen.Metal/()Experimental 444.2.2 Fatigue specimensThree types of specimens were used in the present fatigue study. The choice of specimengeometry was determined by the comprehensive consideration of intended crack orientations,environmental exposure feasibility, as well as the loading capability and applicability of theavailable fatigue machines. Single edge notched specimenSingle edge notched (SEN) specimens with a T-L orientation cracking plane and direction,as defined by ASTM E399-78a [188] and shown in Figure 4.4 were used for high frequencyfatigue testing at —80 Hz on an Instron electromechanical resonant machine (Model 1603). Eachspecimen had a chevron-shaped machined starter notch, —13 mm in length at the centre of thespecimen and —18 mm at the surfaces. The chevron-shaped starter notch was used in order toobtain a good starting crack front profile for the ensuing fatigue test after the fatigue precrackingpreparation procedures. The two side surfaces of each specimen were polished to ametallographic finish and fiducial lines were scribed on the surface to assist measurement ofcrack growth with a travelling microscope and the location of fractographic features. The stressintensity for opening mode loading for the SEN specimens were calculated from the followingpublished K-calibration equations [41,189]:(4.1)KP(ait)”2f(a/W) (42WBExperimental 45f(aIW) = (tanQJ112 {0.752 ÷ 2.202(aIW) ± 0.37(1 — sinQ)3}(4.4)where a is the crack length, P is the applied load, W is the specimen width and B is the specimenthickness. Compact tension (CT) specimenCompact tension specimens with the T-L crack plane orientation were used for lowfrequency fatigue testing at —0.5 Hz on a motor driven, microswitch activated, modifiedHounsfield tensile machine. The geometry of the specimen is shown in Figure 4.5. In a similarmanner to the SEN specimens, a chevron-shaped starternotch, side surface polishing and scribedfiducial line were used on the CT specimens. The K-calibration for this specimen geometryunder Mode I loading [190] is given by:K = P (2+ o) (0.086 + 4.46x — 13.32cz2+ 14.72cz3— 5.6c4) (4.5)BIW(1 _)3aa(4.6)where a is the crack length, W is the specimen width, B is the specimen thickness and P is theload.TB=12.7mmW=55mmFigure4.4Geometryofthesingleedgenotched(SEN)specimen.06.35mmSTB=12.7mmLFigure4.5Geometryofthecompacttension(CT)specimen.Experimental 484.2.2.3 Double canifiever beam (DCB) specimenDouble cantilever beam specimens were used for studying the corrosion fatigue crackpropagation resistance of the Al-Li alloy in the S-L crack plane orientation, particularly theeffects of the re-aging treatment. The same specimen geometry was also used to determine theeffects of the re-aging treatment on the fracture toughness. The geometry of the specimen isshown in Figure 4.6. A chevron-shaped starter notch, -6.4 mm at the centre and —13 mm longat the surfaces, was introduced along the mid-plane of the plate to ensure a good starting crackfront profile after precracking. Similar side surface polishing and fiducial line scribing wereused as the SEN and CT specimens, concerned with crack length measurement and location offractographic features. The K-calibration [1911 for this specimen geometry is given below:K= Pa 13.45+2.415— (4.7)BH312\ a)where P is the load, a is the crack length, His the half specimen height and B us the specimenthickness.4.2.3 Preparation of specimens for transmission electron microscopyElectron transparent specimens were prepared from the Al-Li alloy plate (in both theas-received and the re-aged tempers) by the following steps:1. Thin strips of alloy, —1 - 2 mm thick, with the surface normal to the T-direction weresectioned from the plate with a fluid-cooled saw.2. Small disks of 3 mm diameter were cut out from the strips using a spark machine.3. The disks were then thinned to -400 tm by mechanically polishing on 600 grit sandpaper using a special holder.Experimental 494. The disks were then mechanically dimpled to remove about 30 im of materials on bothsides using a dimpling machine (Dimpler Model D500).5. The fmal thinning was done by Ar ion beam sputtering with an ion beam thinner equippedwith liquid nitrogen cooling specimen stage (Maxmill Model 703), operating at 6 KVand 40 A beam intensities.4.3 Test environment selectionFour environments were chosen for the corrosion fatigue studies. They were:1. Desiccated air.2. Distilled water.3. Sodium chloride solution (0.6 M).4. Aluminium chloride solution (1 M).Desiccated air was used to eliminate any aqueous environmental effects and to allowstudies to be made solely on the effects ofmechanical factors on the crackpropagation behaviour.This reproducible inert environment also served as the baseline condition for determiningwhether other environments significantly affected the corrosion fatigue behaviour. Distifiedwater was used to study the effect of this reproducible contaminant-free environment relativeto that of dry air. Sodium chloride solution was used to study the effects ofhalide-anion-containing corrosive media on the crack propagation behaviour. Aluminiumchloride simulates possible local chemistry conditions that form inside occluded (restricted)geometries such as pits, cracks and crevices. It is known that a local acidic chemistry, enrichedin chloride ions forms inside restricted geometries for both the conventional Al-alloys andLi-containing Al-alloys when exposed to neutral bulk chloride solution [119,147,148,149].SFigure4.6Geometryof thedoublecantileverbeam(DCB) specimen.4LT B=12.7mm2H=12.7mmW=49mmaw 55mmCExperimental 51All the aqueous solutions were prepared from regent grade chemicals and distilled water.4.4 Test setup and procedures4.4.1 Potentiodynamic polarization testsPotentiodynamic polarization tests were conducted in the aqueous chloride environments,using a conventional three electrode system with one working electrode, a platinum counterelectrode and a saturated calomel reference electrode (SCE). The working and counter electrodeswere placed in a single compartment electrochemical cell containing —1 liter of test solution,and the reference electrode was interfaced to the test solution via a salt bridge and Luggincapillary that terminated —2 mm from the working electrode surface. The potential wascontrolled by means of a microprocessor controlled potentiostat (PAR Model 350A) withautomated recording of potential (E) and current density (i). The data were then transferred toa personal computer and the polarization curve, presented as log(i) versus E, was constructedusing software, Lotus 123 and Lotus Freelance Graphics. Each test was conducted by firstimposing a potential on the working electrode that was —250 mV cathodic with respect to itsopen circuit potential (E) and then scanning the potential in the anodic direction at 1 mV/s.All solutions were deaerated by purging with nitrogen before and during the polarization tests.A schematic set-up of the experiment is shown in Figure 4.7.[MicroprocessorControlledIPotentiostatrElectrochemicalCeJNitrogenGasCylinderIFigure4.7Schematicexperimentalset-upofthepotentiodynanhicpolarizationtest.NitrogenGasInlet[PersonalComputerWE:WorkingElectode(Specimen)CE:CounterElectrode(Pt)RE:ReferenceElectrode(SCE)Experimental 534.4.2 Corrosion fatigue crack propagation tests4.4.2.1 High frequencyHigh frequency fatigue tests with the SEN specimens were conducted on the Instronfatigue machine. The loading frequency of the fatigue machine depends on the stiffness of thetesting specimen and the load level, with the frequency decreasing as the crack length increases.For the materials and specimen geometries used in the present study, the frequency change wasless than 5%. The load cycled sinusoidally between maximum (Pm) and minimum (P) loads,and crack growth was monitored on the metallographically polished external surface with aninternally illuminated travelling microscope having a micrometer stage resolution of 10 pm.The fatigue precracking and subsequent fatigue tests followed different loading sequences, asdetailed below. PrecrackingFatigue precracking was conducted in laboratory air at -22 °C using a load sheddingtechnique combined with an increasing R ratio (P1jP). This involved commencing theprecracking at an initial cyclic stress intensity (AK) of —5 MPa. m”2 and an R ratio of 0.1. TheAK was gradually decreased by the incremental lowering ofP, to reduce the maximum stressintensity (K), while slowly increasing theR ratio to 0.5 by raisingP, to increase the minimumstress intensity (K). Upon reaching R = 0.5, incremental decreases (—10%) of AK werecontinued at constant R until no crack growth increment could be detected after 1 x 106 cycles(equivalent to 1 x 108 mm/cycle). At this point, the crack front was essentially straight andthe overall length ofcrack (a) was —22 mm. The above loading procedures were used to minimizeExperimental 54crack closure effects due to load shedding while approaching the threshold without causinglarge increases in crack length. The fatigue precracking was normally fmished within a cracklength increase of —5mm. Fatigue testsDuring subsequent fatigue cracking studies at —22°C in the test environments, a constantR ratio of 0.5 was used throughout the test in order to minimize crack closure effects, and theinitial AK was chosen to be —10% lower than the final value used for precracking. The AK wasthen raised or lowered in small increments ( 10%) until a crack propagation rate close to 1 x10.8 mm/cycle (the lower limit of crack rate detectability) was detected. Thereafter, the cyclicloads, P and P, were held constant, and the crack was allowed to propagate tinder risingAK conditions while recording the crack growth behaviour. Crack length (a) was monitored asa function of the number of load cycles (N) and the average crack growth rate per cycle, doldN,was obtained by the secant method described in ASTM E 647-78T [190].The environments used during corrosion fatigue testing were contained in a transparentacrylic (Plexiglas) cell that was mounted around the mid-section of the specimen and sealed tothe specimen by silicone rubber cement, as shown schematically in Figure 4.8. For tests in theaqueous environments, the top of the cell was left open and the external polished surface of thespecimen was coated with a thin layer of transparent nitrocellulose lacquer to minimize surfacecorrosion and facilitate crack growth monitoring. For tests in desiccated air, the precrackedspecimen was first placed in a vacuum chamber (—1 x i(i torr) for over 60 hours to eliminateany possible reversible hydrogen embrittle effects caused by water moisture in the ambient airduring precracking [192]. It was then placed in a test cell containing freshly baked silica gelExperimental 55desiccant, and the top of the cell was closed with a rubber sheet.The electrochemical potentials of the specimens were not controlled during the fatiguetests in the electrolyte solutions. However, the freely corroding potentials (E) were measuredwith respect to the SCE electrode by means of a salt bridge assembly, similar to that used in thepolarization tests, and a high impedance voltmeter.High frequency loading tests are useful for fundamental studies of fatigue in aqueousenvironments, because the more rapid fluid flow and mass transfer effects inside the crackincrease the probability of maintaining the crack solution chemistry close to that of the bulksolution [41], minimizing complications arising from changes in local solution chemistry thatnormally occur in static cracks and crevices. Low frequencyIt takes longer time at lower frequency to initiate and grow a fatigue crack to a certainlength. Therefore, in order to speed up the specimen precracking processes, the CT specimensused for the low frequency tests were precracked at a different, relatively high frequency. Inthe ensuing low frequency tests, higher starting AKwere used for similar concern, i.e. to shortenthe experimental time period.HighImpedanceVoltmeterAWEPEnvironmental CellFigure4.8Schematicenvironmentalcellsetupfor thehighfrequencyfatiguetest.(JiC’Experimental 574. PrecrackingPrecracking of the CT specimens was conducted in laboratory air at -22 °C on a Sontagelectrical synchronous fatigue machine (Model SF-i-U) which has a fixed loading frequencyof 30 Hz and sinusoidal load waveform. Similar load shedding (AK decreasing) technique tothe high frequency precracking was used except that an R ratio of 0.5 was chosen right fromthe beginning. Precracking was proceeded until the AK value was lowered below the plannedstarting AK value for subsequent low frequency fatigue tests (the starting AK was higher thanthat used on the high frequency tests). Fatigue testsThe low frequency fatigue tests with the precracked CT specimens were conducted onthe modified Hounsfield tensile machine fitted with a microswitch activated DC reversing motor.The frequency was controlled to -.0.5 Hz by adjusting the motor speed and the load waveformwas thangular. The load was applied horizontally and for testing in the aqueous environments,the uncracked portion and some of the cracked portion of the specimen was immersed in thesolution, as shown schematically in Figure 4.9. For tests in desiccated air, the specimen wassealed in a plastic bag with freshly baked silica gel desiccant. All tests were conducted at roomtemperature (-22 °C). Crack growth was also monitored by a travelling microscope, similar tothat used in the high frequency tests.The starting AK values for the low frequency tests were relative high to compensate thelonger time involved. Therefore the real threshold region of low cracking rate ( iO mm/cycleas suggested by Bucci [37] and Taylor [38]) was never reached.Environmental CellHighImpedanceVoltmeterFigure4.9Schematicenvironmental cellset upfor thelowfrequencyfatiguetest.00Experimental 594.4.2.3 Intermediate frequencyThe relatively low stiffness of the DCB specimens and the relatively high working loadlevel of the Instron fatigue machine available prevented the fatigue tests from being conductedat the high frequency with the Instron fatigue machine. Therefore, the short-transverse (S-L)orientation fatigue crack. propagation behaviour of the Al-Li alloy plate was studied at afrequency of 30 Hz with the Sontag fatigue machine. Studies on the effects of the re-agingtreatment on the S-L orientation fatigue crack growth behaviour were also conducted with thesame machine. To overcome the restriction by the limited size of the DCB specimens, the loadwas applied through two threaded eye-bolts which were screwed into the specimen (see Figure4.10). PrecrackingPrecracking of the DCB specimens was done in laboratory air at -.22°C with load shedding(decreasing zK) technique on the Sontag fatigue machine at 30 Hz. An R ratio of 0.5 was usedfrom the start and precracking was terminated at the point that no crack extension could beobserved by the travelling microscope used to monitor the crack progress after iO cycles(equivalent to 1 x i(i mnilcycle cracking rate). Precracking was normally finished with acrack extension of about 4 mm beyond the tip of the -.13 mm long chevron notch seen at thespecimen surface. Hardness and fracture toughness testsHardness and fracture toughness tests were conducted for materials in the as-receivedstate and after the re-aging treatment, to ensure the accomplishment of a proper re-agingExperimental 60treatment, which has been reported to be able to significantly improve (double or even triple)the toughness in the short-transverse orientations with only slight sacrifice in yield strength(—.7%) [112]. Microhardness (H, Vickers hardness) tests were conducted across the platethickness. The S-L orientation fracture toughness values were measured using some of thefatigue precracked DCB specimens. The specimen was tested on a Hounsfield Tensile Machineand the load continuously increased until fracture occurred. The load at this point and the cracklength were used to calculate the toughness value via the stress intensity fomiula for the DCBspecimen, equation (4.7). Fatigue testsCorrosion fatigue tests were conducted on the Sontag machine at 30 Hz at roomtemperature (—22 °C). An R ratio of 0.5 was used throughout the tests. Testing environmentswere contained within a Plexiglas cell placed around the specimen, and Plexiglas sheaths, rubbergaskets and acetate cellulose lacquer were used to prevent the loading bolts from contacting thetesting solutions, as shownschematically in Figure 4.10. Other experimental and data processingprocedures were similar to those described for the high frequency tests.At the end of each fatigue test, the specimen was rinsed with ethanol, dried in a streamof air and broken open on a Hounsfield Tensile Machine by tensile overload. This overloadwas recorded and used to calculate the apparent fracture toughness to re-confirm that its valuewas within the normal limits, and in the case of the re-aged material, to re-confirm that a properre-aging treatment had been done.Experimental 614.4.3 Fractographic studyAfter fatigue testing, the fracture surfaces were prepared for fractographic study. Thefracture surfaces were normally cleaned with denatured ethanol ultrasonically. When necessary,the fracture surfaces were cleaned with an inhibited acid solution containing 70 ml phosphoricacid, 32 mg chromic acid and 130 ml distilled water [193]. Crack surface fractography wasexamined by standard scanning electron microscopy (SEM), using secondary electron imagingwith 20 KV and 5 KV acceleration voltages. The 5 KV excitation was chosen because it wasfound sometimes that small details on the surface were more easily visible under low excitationconditions. Two scanning electron microscopes, a Hitachi S-570 equipped with energydispersive X-ray (EDX) spectrometer and wavelength dispersive X-ray (WDX) spectrometer,and a Hitachi S-2300 were used in the studies.4.4.4 Crack profile studyMetallographic sectioning procedures and conventional optical microscopy were used tostudy the crack front profiles in the plane normal to the direction ofcrackpropagation. Sectioningwas made at several AK values on specimens tested in different environments and conditionsin order to study the changes of crack front proffle. The sectioning was done by cutting with afine jewelry saw, grinding with SiC paper down to 600 grit and then polishing to 1 Im finish.The polished surface was etched with Keller’s etchant to reveal the grain structure.HighImpedanceVoltmeterWEEnvironmentalCell‘PFigure4.10Schematicenvironmentalcellsetupfortheintermediatefrequencyfatiguetest.Experimental 634.4.5 Microstructural and compositional study4.4.5.1 TEMTransmission electron microscopy was conducted to characterize the microstructures ofthe Al-Li alloy in the as-received temper and after the re-aging treatment. Bright field imaging,centred dark field imaging and selected area diffraction were used to observe and identifystructures and phases in the material. The transmission electron microscope, Hitachi H-800,was operated at 200 Ky. STEM + EDXScanning transmission electron microscopy (STEM), combined with EDX analysis, wasperformed in order to detect composition profile across grain boundaries in the Al-Li alloy andthe effect of the re-aging treatment. Quantitative analyses at points along a line across a grainboundary and directly on a precipitate, or between two precipitates, were conducted. EELSLithium in the Al-Li alloy is a light element which is not detectable by the commonanalytical X-ray techniques like EDX analysis and WDX analysis. Therefore, electron energyloss spectroscopy (EELS) was used, in order to study the compositional distribution of Li inthe material. Attempts were made to find the Li excitation edge and the shift of the plasmonpeak energy which can be related to the local composition [194,1951.Results 645 Results5.1 Electrochemical behaviour5.1.1 Al-Li alloyThe potentiodynamic polarization behaviour of the as-received Al-Li alloy (T877 1) in thechloride solutions is shown in Figure 5.1. The ahoy exhibited different polarization behaviourin the two chloride solutions, particularly withrespectto the open circuitpotential (E), passivityand the pitting potential (E). The polarization curve obtained in 0.6 M NaC1 shows annear -1.025 V and a series of very low anodic current densities ( 3 x 1(18 A/mm2), as thepotential is raised —350 mV above E, that is consistent with the presence of a passive film.Near -0.705 V, there is a pronounced increase in current that is characteristic of filmbreakdown and the onset of pitting, suggesting that the pitting potential (E1J is close to -0.705V. The curve obtained in 1 M AIC13 shows a higher near -0.875 V, a narrow poorlydefined region of passivity with a higher current density of —5 x i(1 A/mm2,and a much lowernear -0.785 V. The differences in polarization behaviour are readily attributable to thelarge difference in chloride ion concentration between the two solutions, and the acidic natureof the 1 M A1C13 (pH = 2.0) relative to the near neutral 0.6 M NaCl (pH = 6.0).Thermodynamic equilibria in aqueous solutions at one atmospheric pressure and 25 °C[196] show that the potential, EH2, below which it is possible to reduce H ions to hydrogen isgiven with respect to the standard hydrogen electrode (SHE) by Equation (5.1),= —0.059pH, V (5.1)Results 65Consequently, since the conversion from the SHE to the SCE scale is obtained via the relationship= -0.242 + V, Equation (5.2) gives E on the SCE scale,EH2 = —0.242 — O.O59pH, VSCE (5.2)Therefore, with respect to Figure 5.1, hydrogen can be evolved on the alloy surface at potentials-0.360 VSCE in 1 M A1C13,and -0.596 V in 0.6 M NaC1.Electrochemical studies were conducted with specimens of different exposed surfaceorientations (S and T). Changing the orientation did not produce any obvious effect on thepolarization behaviour, even though the material has an anisotropic grain structure so thatspecimens of different orientations expose different portions of grain and grain boundaryregions. The orientation-independence of the polarization behaviour of the highly texturedmaterial is consistent with the study of Yasuda et a!. [200] on Al-Cu single crystals, where itwas shown that the presence of alloyed Cu reduces the dependence of EPil on surface orientationso thatEbecomes invariant atCu level above—i wt.%. The re-aged alloy showed no differencein electrochemical polarization behaviour either.The appearances of metallographically polished T-surfaces of both the as-received andre-aged materials, after immersion in 1 M Aid3 under freely corroding conditions at roomtemperature (—22 °C) for —20 hours, are shown in Figure 5.2. It is seen that the materials wereattacked preferentially along lines which delineate the grain boundaries. However, no significantdifference can be seen in the intergranular attack between the two tempers. Figure 5.3 showsthe same surfaces after continuing the inimersion in 1 M Aid3 solution to 170 hours. It isobvious that the material is not as badly corroded in the re-aged temper as in the as-receivedtemper. When sectioning was made perpendicular to these surfaces, as shown in Figure 5.4, itrevealed that the intergranular attack depth was much larger for material in the as-received(1.mV/s,—22°C)C.,————%S-0.4-0.5-0.6-0.7-0.81-0.9 -1—1,1-1.2-1.3-1.4ioio810-6io4102i(A/mm2)-Figure5.1PotentiodynamicpolarizationbehaviorofAl-Li alloy8090-T8771in0.6MNaC1and1 MA1C13.O.6MNaC11MA1C13I--I-_---IIIIResults 67temper than in the re-aged temper. These differences indicate that the short time re-agingtreatment was able to cause changes in the material which affect its corrosion behaviour.Analyses of the sectioned surfaces by SEM + EDX suggested that the corrosion may be relatedto the local Cu content distribution, because Cu-rich particles were unattacked while materialsaround them were eaten away, as shown, for example, in Figure 5.5. Further studies wereconducted with precipitates-containing intergranular fracture surface obtained by monotonicloading, in order to locate the sites where corrosion starts preferentially. The results are- illustrated in Figure 5.6, where a comparison is made between the same location before andafter immersion (corrosion) in 1 M A1C13 for —12 hours. It is evident that corrosion occurredpreferentially at sites where there were precipitates and/or where precipitates had been pulledout during the previous mechanical fracture. One Cu-rich precipitate (identified by EDX)remained while the others disappeared after the immersion. The disappearance of the particlesmay be due to either the corrosion of the particles themselves or the corrosion of the materialsaround them that the particles fall out subsequently. With regard to the cavity growth at SiteA in Figure 5.6 and consistent with Figure 5.5, the corrosion of the materials around the particlesis considered more likely. The sites where corrosion starts preferentially are likely to beassociated with Cu-depleted regions in the matrix around Cu-rich precipitates, based on theeffects of local Cu concentration on localized corrosion of aluminium alloys [200,201].Figure5.2Appearanceofpolishedsurfaces(Torientation)of Al-Li8090after freelycorrodingin1MAIC13for -20hours.(a)(b)(a)As-received;(b)re-aged.00(a)(b)Figure5.3Appearanceofpolishedsurfaces(Torientation)ofAl-Li 8090afterfreelycorrodingin1MA1C13for —170hours.(a)As-received;(b)re-aged.(a)(b)Figure5.4SectioningparalleltotheLplaneshowingdepthsof intergranularcorrosionafter corrodingin1MAIC13for170hours.‘‘P’’’ti[injHJ4OO/.Lm(a)As-received;(b)re-aged.Figure5.5SEM+EDXanalysisoftheintergranularlycorrodedslotsinthere-agedAl-Li alloy,showinguncorrodedCu-containingparticle.(a)Linescanposition;(b)CuX-raylinescan.(a)(b)Figure5.6SEMmicrographsshowinglocationofCu-richparticlecontainingintergranular surfaceof there-agedAl-Lialloybeforeandaftercorrodingin1MA1C13for-l2hours.(a)Beforethecorrosion;(b)afterthecorrosion.(a)(b)Results 735.1.2 Al-Cu alloyThe potentiodynamic polarization behaviour of the Al-Cu alloy in the chloride solutionswas similar to the Al-Li alloy. Representative polarization curves for the Al-Cu alloy are shownin Figure 5.7, where the test in the acidic 1 M A1C13shows a higher E(), a lower EPk and a narrowpoorly defined region of passivity with a higher current density than in the neutral 0.6 M NaC1.Also, in 1 MA1C13,E1,, of the Al-Cu alloy is —-0.725 VSCB, Eft is —-0.670 VSCB and the passivationcurrent density is 10.6 A/mm2 Corresponding values in 0.6 M NaCl are E = —-0.960 V,= —-0.650 V and the passivation current is 3 x 108 Jfl2.The Al-Cu alloy exhibited slightly higher pitting potential than the Al-Li alloy in bothchloride solutions, suggesting a slightly better resistance to pitting corrosion.-0.4I-1.3(1mV/s,—22°C)1MA1C1310.1010-8io610iO2i(A/mm2)-0.5-0.6-0,7-0.8-0.9 —1—1.1-1.2—a———SSS SS S4-1.4O.6MNaC1IIII.IIIFigure5.7PotentiodynamicpolarizationbehaviorofAl-Cualloy2024-T351 in0.6MNaCland1 MA1C13.Results 755.2 Crack growth behaviour5.2.1 Al-Li alloy5.2.1.1 T-L crack plane orientation, 80 Hz loading frequencyThe corrosion fatigue crack propagation behaviour of the as-received Al-Li alloy in thechloride solutions, distilled water and desiccated air at about 80 Hz loading frequency is shownin Figure 5.8. Crack growth rates (da/dN) increased with rising AK in all the aqueousenvironments, with the A1C13 solution producing the highest rates and causing the mostdeleterious effects in the very low AK region. The threshold of cyclic stress intensity (AKth)was reproducibly close to 1.05 MPa.m1in 1 M Aid3,whereas AKth was higher and close to1.7 MPa.m1t2 in 0.6 M NaCl and distilled water. Between a AKof 2.2 and 4.0 MPa.m”2,crackgrowth rates in both chloride solution were similar and slightly higher than those in distilledwater. Above a AK of —4.0 MPa.m1,the fatigue curves obtained in the aqueous environmentsconverged and crack growth kinetics became independent of solution composition. In thisregion, the aqueous fatigue curves obeyed a power law relationship, da/dN (AK)’1 with n =—4, that is typical of Stage II fatigue behaviour.Changing the environment to desiccated air caused a remarkable and reproducibledifference in crack growth behaviour. In particular, the crack growth rates became essentiallyindependent of AK above -3 MPa.m1’,resulting in a crack rate plateau where da/dN was —2 x10.6 mm/cycle. This plateau extended into the region of Stage II cracking. Consequently, atAK —9 MPa.m”2 (where AK is based on crack length measurements made at the specimen1C.) I+1III11111 10LLK(MPa.m”2)IIII11111 100(R=0.5,—22°C,—80Hz,T-LOrientation) 4QD11 1o-10-61 0-10-81+1 MA1C13,EcorrD0,6MNaC1,Ecorr+DistilledWater°DesiccatedAirFigure5.8EffectofenvironmentandAKonthegrowthof fatiguecracksinAl-Lialloy8090-T8771at40Hz.Results 77surface), crack growth rates in desiccated air were —100 times lower than those in the aqueousenvironments However, the near-threshold behaviour was indistinguishable from that in 0.6M NaCl and distilled water, where AKth was —1.7 MPa.m”2The average E potentials of specimens fatigued in 0.6 M NaCl and 1.0 M A1C13 were—-0.730 Vand —-0.780 V, respectively. Comparison with Figure 5.1 shows that these areclose to or coincident with the E1 potentials in each solution, indicating the possibility oflocalized electrochemical dissolution processes during fatigue. In addition, comparison of eachwith Equation (5.2) confirms that hydrogen could have been evolving simultaneously onthe alloy surface.Some modified fatigue tests were conducted to study the effect of changing the testingenvironment from aqueous to the dry air condition on the crack growth behaviour of the sametest specimen. Figure 5.9 shows the result of a fatigue test conducted in two differentenvironments consecutively. The test was first conducted in distilled water until the AKreachedthe value of 5 MPa.m’. Then the distilled water environment was removed and the specimenwas put in a vacuum chamber for 72 hours to remove remaining water and any possible reversiblehydrogen embrittlement effects [192]. The fatigue test was then continued in desiccated airenvironment. It is clear that the cracking rate dropped an order of magnitude abruptly from thevalue in distilled water to that in desiccated air, and was followed up by a crack rate plateau.Another fatigue test in desiccated air was started from a AK of 5 MPa.m1and exhibited thecrack rate plateau right from the onset of cracking, which continued to an apparent AK valueof 50 MPa.m’ at the finish of the test (Figure 5.10). The corresponding apparent Km valueIt will be shown later in Section that at AK> 3 MPa.m’’2,the crack front in desiccated air is non-linearand that the AK values in this environment are apparent values and not true values.Results 78at the termination of the test was -.100 MPa.m”2and is well above the fracture toughness (K1)value reported in Table 4.2. Additional high frequency tests were conducted with a hydrogenprecharged specimen, and a specimen subjected to continuous hydrogen charging. The fatiguecrack propagation behaviour of the hydrogen precharged specimen in desiccated air is shownin Figure 5.11, where the crack growth rate plateau still persists. The fatigue crack propagationbehaviour of the specimen under constant hydrogen charging condition is shown in Figure 5.12.With continuous on-site hydrogen production at the crack tip, the crack propagation curve issimilar to those obtained in the aqueous testing environments, where a crack rate plateau isabsent. The test conditions for Figure 5.12 were such that the specimen was controlled at apotential of -2.1 V in 1 M HC1 solution which was circulated around the specimen with apump. The potential was chosen to place the alloy in the corrosion immunity region for Al(EM+++,A, = -2.02 V for [A1i = 106 [196]) so that hydrogen effects were dominant. Thecirculated acidic solution was chosen in order to maintain the pH at the metal surface and avoidsevere corrosion of the specimen surface caused by local alkalization. Under highly hydrogencharging conditions the reduction ofhydrogen ions can increase the local surface pH and tarnishthe specimen surface, making crack monitoring with the travelling microscope very difficult.1 0-(R=O.5,-p22°C,8OHz,T-LOrientation)10-a+—10-s+10-6+10-v+DistilledWater+108++0DesiccatedAir9IIIII11111IIIIIIILl1110100LK(MPa.m”2)Figure5.9ConsecutivefatiguecrackgrowthbehaviourinAl-Li alloy8090-T8771whenthetestingenvironmentischangedfromdistilledwatertodesiccatedair.10-s(R=O.5,--22°C,8OHz,T-LOrientation),.1O-C.) 110-61O8IIIII11111IIIIIIII110100zK(MPa.m”2)Figure5.10Fatiguecrackgrowthbehaviour inAl-Li alloy8090-T8771at-80Hzindesiccatedair, startingfromAK=C5MPa.m(Note: AKvaluesareapparentvalues.)10-s(R=O.5,-22°C,-80Hz,T-LOrientation)1 0-)ir-5,—lIV I 106•0p.10-8 9IIII11111III1111110110100LK(MPa.m”2)Figure5.11FatiguecrackgrowthbehaviourinAl-Lialloy8090-T8771at--80Hzindesiccatedairafter hydrogenprecharge.(Note:zSXvaluesareapparentvalues.)1 0-(R=O.5,-22°C,-8OHz,T-LOrientation)V1o-V Vi0-Wv110-610-v10-8 ic1—9IIItillIII11111110100zK(MPa.m”2)Figure5.12FatiguecrackgrowthinAl-Li alloy8090-T8771at—80Hzin1 MHC1solutionundercontrolledpotentialof-2.1 VSCB.InthispotentialandpHcondition,Alis intheimmunityregionandtherateof hydrogenevolutionishigh.Results 835.2.1.2 T-L crack plane orientation, 0.5 Hz loading frequencyThe corrosion fatigue crackpropagation behaviour in the chloride solutions, distilled waterand desiccated air at a loading frequency of 0.5 Hz is shown in Figure 5.13. The cracking ratesin 1 M Aid3and 0.6 M NaCl solutions showed similarbehaviour in the range ofAKinvestigated.The cracking rates in distilled water and desiccated air were not largely different from eachother, however they were lower than those in the chloride solutions. Also, the crack growthrate curves in distilled water and desiccated air show an apparent cyclic stress intensity thresholdof about 5.5 MPa.m1,which is much higher than the threshold values observed during the highfrequency tests. The averageE0potentials monitored during fatigue at -0.5 Hz in 0.6 M NaC1and 1 M A1C13were —-0.810 V and —-0.780 V respectively. S-L crack plane orientation, 30 Hz loading frequencyThe microhardness and fracture toughness values of Al-Li 8090-T877 1 before and afterthe re-aging treatment are listed in Table 5.1. The re-aging treatment doubled the fracturetoughness with only--.8% decrease in hardness, consistent with reported data in the literature[112]. The data confirm that the re-aging treatment is an effective way to improve the S-Lorientation fracture toughness with only minor sacrifice in strength (hardness).The fatigue crack propagation behaviour of the Al-Li alloy in both the as-received andre-aged states when tested in a desiccated air environment is shown in Figure 5.14. It is obviousthat fatigue crack propagation is faster and the threshold stress intensity (IXKth) for crackpropagation is lower in the re-aged temper in desiccated air. Figure 5.15 shows the crack growthbehaviour in 1 M A1C13 solution under free corrosion condition, where it is seen that there is no10-s,APDQi_*+ø•D4;f;$:10-51n-6\—,LU÷1MAid3,Ecorrio-00.6MNaC1,Ecorr+DistilledWater10-8“DesiccatedAir(R=0.5,22°C,-0.5Hz,T-LOrientation)IIIIIIliiiIIIIIliii110100K(MPa.m1”2)Figure5.13EffectofenvironmentandAKonthegrowthoffatiguecracksinAl-Li alloy8090-T877 1at-0.5Hz.Results 85difference between the as-received and re-aged tempers. Figure 5.16 shows the crackingbehaviour in distilled water and again there is no observable difference between the two tempers.However, when compared with the crack propagation behaviour in desiccated air, it is seen thatthe Aid3environment accelerated crack growth of both tempers by both increasing the crackingrate and decreasing the threshold value AK11, whereas the distilled water only increased thecracking rates of both tempers at AK —2 MPa.m1. Below AK value of —2 MPa.m” thecracking rates were decreased in distilled water and showed a higher AK than in desiccatedair. The AKth values of the two tempers in the different environments are summarized in Table5.2. The cracking in 0.6 M NaCl solution under free corrosion condition was not well defmedand the data were very scattered because of problems arising from crack retardation effects.The average EC(,ff potentials during fatigue at 30 Hz in 0.6 M NaCl and 1 M A1C13were —-0.730VSCE and —-0.780 V respectively.Table 5.1 Values of Microhardness (H) and Short Transverse (S-L) Orientation FractureToughness (K1)As-received (T8771) Re-aged (230 °C, 10 minutes)H K1 (MPa.m”) H K1 (MPa.m”2)159 —14.7 146 —29.8Results 86Table 5.2 Threshold Cyclic Stress Intensities (AKth) for S-L Fatigue Cracking in the TwoTempers of the Al-Li Alloy in Different EnvironmentsEnvironment As-received Re-agedAKth (MPa.mla) AKth (MPa.m1t2)Desiccated Air—1.6 —1.3Distilled Water —1.8 —1.81MA1C13—1.0 —1.0When compared to the T-L orientation fatigue crack growth data at —80 Hz, crackpropagation resistance in the S-L orientation is lower by showing faster cracking rates and lowerAKth values.5.2.2 Al-Cu alloy5.2.2.1 T-L crack plane orientation, 80 Hz loading frequencyThe T-L orientation crack propagation behaviour of the as-received Al-Cu alloy in 1 MA1C13,0.6 M NaC1, distilled water and desiccated air environments at the testing frequenciesof -80 Hz is shown in Figure 5.17. There was little difference between the cracking behaviourin 0.6 M NaC1, distilled water and desiccated air. While in 1 M A1C13, the crack propagationthreshold AKth was reduced to —1 MPa.mhtZ, with respect to the value of —2 MPa.m’ in the otherenvironments. In the relatively high AK region (i.e AK> —4 MPa.m1),the cracking rates in 1Results 87M A1C13converged with those in the other environments. The average Eff potentials monitoredduring fatigue at -.80 Hz in 0.6 M NaC1 and 1 M A1C13 were -0.640 and -0.700 Vrespectively.The crack growth rate curves of the Al-Cu alloy are very similar to those of the Al-Lialloy in the respective aqueous environment. T-L crack plane orientation, 0.5 Hz loading frequencyThe crack growth behaviour of Al-Cu alloy in the same four environments at 0.5 Hzloading frequency is shown in Figure 5.18. Crack growth rates were similar in 0.6 M NaCl,distilled water and desiccated air environments over the whole range of AKinvestigated. Crackpropagation was accelerated in 1 M A1C13 in the lower end of the uK range investigated (i.e.near MPa.m1),and merged with cracking rates in the other environments at the higher AKend (i.e. above —8 MPa.mtt2) For tests in distilled water and desiccated air, the curve showsan apparent threshold cyclic stress intensity of --5.5 MPa.m’. When compared to the highfrequency results, the major difference in behaviour is the decreased crack growth per cycle atthe low frequency in the low end of the AK range investigated. The average potentialsduring fatigue at -05 Hz in 0.6 M NaC1 and 1 M AICI3 were -0.700 and -0.7 10 Vrespectively.(R=0.5,22°C,-30Hz,S-LOrientation)C.) I1IIII11111 10z\K(MPa.m112)IIIII1111 100106108•As-received•As-receivedoRe-agedoRe-agedFigure5.14FatiguecrackgrowthintheS-Lorientationof theAl-Lialloyplateindesiccatedairat—30Hz.00 001 0-(R=0.5,—22°C,—30Hz,S-LOrientation)10-aA10h1 I106AAs-received10-8Re-agedIIIIIIIIIIIIIIIIIII110100L\K(MPa.m”2)Figure5.15FatiguecrackgrowthintheS-Lorientationof theAl-Lialloyplatein1MA1C13at—30Hz.10-(R=0.5,—22°C,—30Hz,S-LOrientation)io-41‘1)ir-5 lvI.E10.60.iO•As-received10-8cRe-agedIIIIIIIIIIIIIIIlIlt110100zK(MPa.m”2)Figure5.16FatiguecrackgrowthintheS-Lorientationof theAl-Lialloyplateindistilledwaterat—30Hz1 0-‘PC.) I 10-8(R=O.5,—22°C,—80Hz,T-LOrientation)‘DesiccatedAir1IIIIIIII 10IIIIIliii 100z\K(MPa.mL’2)1 0-10-s10”61 0-14+ +0 01 MA1C13,EcorrD0.6MNaC1,Ecorr+DistilledWater1Figure5.17Effectof environmentandzS.Konthegrowthof fatiguecracksinAl-Cualloy2024-T351 at—80Hz.1C10-i‘4>1- Q__1004C.)1MA1C13,Ecorrd10-600.6MNaC1,EcorrDistilledWater10<>DesiccatedAir10-8(R—O.5,-p22°C,O.5Hz,T-LOrientation)IIIIIIliiiIIIII1111110100K(MPa.m’’2)FIgure5.18EffectofenvironmentandAKonthegrowthoffatiguecracksinAl-Cualloy2024-T351 at -O.5Hz.Results5.3 Fractography5.3.1 Al-Li alloy5.3.1.1 T-L crack plane orientation, 80 Hz loading frequencyTransgranular cracking was the predominant mode of crack propagation in both the nearthreshold Stage I region and the linear Stage II region (or Paris law region) for the Al-Li alloywhen tested in all four environments at the high loading frequency (—80 Hz). With the exceptionof the Aid3solution, the threshold fractography was very similar in the different environments,consistent with the similarity in their crack growth behaviour near AKffi. A representativefractograph, typical of the desiccated air, distilled water and NaC1 environments, is shown fordistilled water in Figure 5. 19a. Numerous fme scale tear and shear ridges (river lines) are evidentwhich are inclined to the macroscopic direction of propagation. Therefore, assuming that theseridges are normal to the local direction of the propagating crack front, the fractography indicatedthat the crack front was not straight on the microscopic scale; instead, it was composed of aseries of tongues that linked together by the widening of the tongues in the wake of thepropagating crack tip. A schematic diagram illustrating the tongue-like nature of the propagatingcrack front is shown in Figure 5.19b.The near threshold fractography obtained in the Aid3 was different from the otherenvironments. A distinguishing feature was the large degree of secondary cracking (splitting)normal to the primary crack plane (i.e. the splits were consistent with S-L orientation cracking).The splitting was microstructure dependent and occurred along grain boundaries, whereas theprimary crack plane (T-L orientation cracking) followed a transgranular path. Figure 5.20 showsUC.,uncrackedcrackedetr’(a)(b)Figure5.19(a)SEMfractographshowingtransgranular fracturesurfaceofAl-Li alloyspecimenfatiguedat-.80HzindistilledwaternearAKth.Macroscopicdirectionofcrackpropagationisfromtherighttotheleft ofthepage.(b)Schematicdiagramofthetransgranular crackfrontintheprimarycrackplaneshowingtongues extendingintouncrackedmaterial,withuncrackedligamentsbetweenthetongues.\0Results 95the characteristic splitting (delamination) features observed nearAKth (i.e. nearAK= 1 MPa.m112)Increasing the AK caused significant change in the degree of S-L splitting in allenvironments. Splitting was quantified by measuring the number (NB) of splits across the widthof the specimen, using linear intercept analyses conducted in the SEM. In the cases of distilledwater and 0.6 M NaC1, the number of splits was found to increase from near zero at AKth andreached a peak at AK values of —5 to -.6 MPa.m’t2 This was followed by a decreasing numberof splits as AK continued to increase. Tests in 1 M Aid3 showed that the number of splitsdecreased from a maximum at AK to a minimum as AK increased to —3 MPa.m1”2.Thereafter,the splitting behaviour imitated that in the other aqueous environment by exhibiting a peak at—5 MPa.m’. Splitting effects in desiccated air were very small relative to the aqueousenvironments and were virtually independent of AK. Details of the splitting analyses arepresented in Figure 5.21.The large number of splits at AKth in the A1C13 solution appeared to be related primarilyto localized stress-independent dissolution of the grain boundaries, because the precrackedfracture surface, which was not fatigued in A1C13,also exhibited the same degree of splitting.An example of this phenomenon is shown in Figure 5.22. The precracked fracture surface ofspecimens tested in the other environments did not show the splits. Consequently, it followsthat the decreasing degree of splitting as AK increased to —3 MPa.m was due primarily to thefact that the propagating crack front spent increasingly smaller amounts of time in contact withthe solution, thereby causing increasingly smaller amounts of grain boundary dissolution.Localized dissolution processes are consistent with the observation that EC(,ff during fatigue wasclose to E1 and that the anodic dissolution current at was much higher in the 1 M Aid3solution than in NaC1 (see Figure 5.1). The result of severe intergranular corrosion in the A1C13Figure5.20SEMfractographofAl-Li alloyspecimenfatiguedat‘-80Hzin1MA1C13nearIL\Kth,showingtransgranularcrackingontheprimarycrackplaneandS-Lsplittingofgrainboundaries normaltotheprimaryplane.Macroscopicdirectionofcrackpropagationisfromtherighttotheleftofthepage.C.,C.’Results 98local crack front. The average spacing of these striations is -1 x i0 mm corresponding verywell with an average crack propagation rate of —1 x 1O mm/cycle for the testing conditions(see Figure 5.8) and providing further support for the interpretation of Figure 5.25.Figure 5.26 shows the region of fracture surface when the test environment is changedfrom distilled water to desiccated air. The secondary cracks (splits) produced in distilled waterdisappeared in the desiccated air environment, and the fracture surface changed to contain largefacets. Fracture surfaces from the fatigue test conducted under constant hydrogen charging at-2.1 also contained splits, as shown by the results of the SEM analyses in Figure 5.27. Ina similar manner to behaviour in distilled water and 0.6 M NaC1 environments, the number ofsplits increased from about zero at AKth and peaked at intermediate AK values of —7 MPa.m”2.In addition, detailed studies of the near-threshold fractography were conducted byexamining stereo pairs of opposing fracture surfaces. In this manner, it was found that thefractographic features on the opposing surfaces interlocked (i.e. matched peak to valley). Anexample is shown in Figure 5.28 for a test conducted in distilled water. The production of thisfractography is easily explainable by the restricted slip reversibility (RSR) model [41,42] forfatigue cracking, where alternative slip systems are operating in the near threshold region.However, using the same matching surface techniques, it is less certain whether the fractographicstriations produced in Stage II are matching on opposing fracture surfaces in a peak to peak orpeak to valley mode, as shown in Figure 5.29. From the crack tip blunting and re-sharpeningmodel [43] for Stage II fatigue crack growth, it would be expected that the match on opposingfracture surfaces should be peak to peak and valley to valley (i.e. mirror images).Cl)I. I C C C z300 280260__240220200180160140120100 80 60 40 20 0zK(MPa.m”2)Figure5.21EffectofenvironmentandAKonthenumber ofsplits(N8)acrossthethicknessof theT-LorientationAl-Li8090-T8771specimenfatiguedat—80Hz.1357911Results 97immersion tests supports the above conclusion. Furthermore, consistency of reasoning leadsto the conclusion that the increasing degree of splitting as AKrises to values of —5 to —6 MPa.m,observed in all the aqueous environments, was due to a synergism between the cyclic load andthe environment.Optical microscopy studies of sectioned specimens showed that the depth of penetrationof the splits below the primary crack plane often reached —200 urn or more in Stage II. This isseen in Figure 5.23 for a specimen fatigued at —80 Hz in 0.6 M NaCl and then sectioned throughthe primary crack plane at the position corresponding to AK =5 MPa.m1.The grain structurewas revealed by Keller’s etchant and shows the intergranular path followed by the splits. Sometests were terminated in Stage II after reaching a AK condition corresponding to the formationof the peak number of splits. In these cases, each specimen was sectioned behind the leadingedge of the main crack and carefully polished towards the crack front until the first tongue-likepenetrations of the crack could be detected. The result is shown in Figure 5.24 for a test in theNaC1 solution and clearly shows a section through two tongues, one on the left and one on theright side of the micrograph. A vertical split is present that is not connected to either tongue,confirming that the split extends ahead of the crack front. Consequently, it is possible that suchsplits can assist linkage of uncracked ligaments between the crack front tongues by acting asinitiation sites for microcracks which subsequently propagate away from the split along theprimary crack plane. Detailed SEM fractography observations supported this possibility, asshown in Figure 5.25 for AK =6 MPa.m112,where the orientation of river lines clearly indicatesthat local microcracks initiated at a split and propagated away from the split in a direction thatwas inclined to the overall direction ofmacroscopic crack propagation. Furthermore, very faintstriations are detectable normal to the river lines and they indicate successive positions of theFigure5.22SEMmicrographof Al-Li alloy8090-T8771specimenprecrackedsurfaceexposedto1 MA1C13during-80Hztest.Areashowslocalizeddissolutionofgrainboundaries.Figure5.23OpticalmicrographofsectionedandetchedAl-Lialloy8090-T8771specimenfatiguedat-.80Hzin0.6MNaC1.SectionednormaltotheprimarycrackplaneinStageIIatAK=5MPa.m1,,..1•.1i!I•1)I;II.100pm\-Figure5.24OpticalmicrographofsectionedandetchedAl-Lialloy8090-T8771specimenfatiguedat-80Hzin0.6MNaC1.TestterminatedinStageIIandsectionednormaltotheprimarycrackplaneveryclosetothemaincrackfront.Figure5.25SEMfractographofStageIIcracking(tK=6MPa.m”2)ofAl-Li8090-T8771in0.6MNaC1at—80Hz,showingriverpatternsrunningawayfromtheS-Lsplit.Noteveryfaint striationsnormaltotheriverlines.Macroscopiccrackingdirectionisfromtherighttothe leftofthepage.DesiccatedAirFigure5.26SEMfractographshowingchangeoffatiguefracturesurfaceofAl-Li 8090-T8771producedat—80Hzindistilledwateranddesiccatedairconsecutively.Macroscopiccrackingdirectionisfromtherighttotheleft of thepage.IDistilledWater300280Ci’)200 I: 140 120 80 60 Z 40 20 0/K(MPa.m”2)Figure5.27Changeof thenumberof splits(NB)acrossthethicknessoftheT-LorientationAl-Li 8090-T8771specimenwithK,whenfatiguedat—80Hzundercontinuoushydrogenchargingat-2.1V.1357911Figure5.28SEMfractographsofnearAKhcrackingofAl-Lialloy8090-T8771indistilledwaterat-80Hz.Theopposingfracturesurfacesindicatinga“peaktovalley”matchofsmalldetails.Macroscopiccrackingdirectionisfromthetoptothebottomofthepage.Figure5.29SEMfractographsofmatchingfatiguefracturesurfacesofAl-Lialloy8090-T8771producedbyStageIIcracking(AK=6MPa.m’)indistilledwaterat-80Hz.Macroscopiccrackingdirectionisfromthetoptothebottomof thepage.Results 1085.3.1.2 T-L crack plane orientation, 0.5 Hz loading frequencyChanging the loading frequency of the fatigue tests to —0.5 Hz did not cause any obviouschange in the fractography obtained. In the AK range tested, the fractography was stillpredominantly transgranular, as shown in Figure 5.30 for fracture surfaces produced indesiccated air. Similar split number analyses to those at the high loading frequency wereconducted, the result is shown in Figure S-L crack plane orientation, 30 Hz loading frequencyFractographic studies of surfaces obtained from the fracture toughness tests and fatiguetests in the S-L orientation showed that the crack path followed the “pan cake” grain boundaries,i.e. the cracking was intergranular. Figure 5.32 shows the fracture surfaces from the K1 testsof the material in both the as-received and re-aged tempers, with the transition between thefatigue precracked region and the monotonic loading cracked region located near the centre ofthe respective micrograph. Careful SEM examination close to the transition region of the re-agedspecimen showed that the monotonically loaded fracture surface exhibited continuous fine scaledimples, characteristic of a ductile microvoid coalescence process, whereas the adjacentprecracked surface was relatively featureless with no dimples. In addition, trenches wereobserved lying normal to the primary crack in the transition region. For the as-received material,the fracture surface showed discontinuous patches of dimple features on the monotonicallyloaded side of the transition region. These were interspersed with patches of featureless surface.Also no trenches were found in the transition region. At locations more distant from the transitionregion, the monotonically overloaded fracture surface of both the as-received and re-agedtempers showed a patchy fractographic mixture of dimpled and smooth features.Figure5.30SEMfractographshowingtransgranular fracturesurfaceofAl-Li 8090-T8771specimenfatiguedat-.0.5Hzindesiccatedair.AK=—6MPa.m’,locationnearthemid-thicknessof thespecimen.Macroscopicdirectionofcrackpropagationisfromtherighttotheleftof thepage.Micrograph(b)showsthecentralregionofMicrograph(a)inhighermagnification.(a)(b)300280260240CI2220C/)200180III_____EK(MPa.m”2)Figure5.31EffectofenvironmentandAKonthenumberofsplits(NB)acrossthethicknessoftheT-LorientationAl-Li8090-T8771specimenfatiguedat—0.5Hz.Aid3,EcorrCNaC1,Ecorr÷DistilledWaterDesiccatedAirMonotonicOverload(a)(b)Figure5.32SEMfractographsshowingtransitionfromfatigueprecrackingtomonotonicoverloadfractureintheS-Lorientationfracturetoughnesstest specimensofAl-Li alloy8090.Macroscopiccrackingdirectionisfromtherighttotheleftofthepage.(a)As-received;(b)re-aged.IPrecrackMonotonicOverloadIlecrackResults 112The S-L fatigue crack fractographies of the as-received and re-aged Al-Li alloys werevery similar to each other when obtained in the same environment. The fractography wasgenerally flat and featureless, and followed an intergranular path along the S-L orientation grainboundaries. Detailed differences in fractography between different environments could berelated to environmental effects on the crack propagation rates. For example, Figure 5.33compares the near threshold fatigue fracture surfaces of the re-aged material tested in distilledwater and desiccated air. Dark features indicative of rubbing or fretting, and corrosion productsare visible on the fracture surface obtained in distilled water, while a clean intergranular fracturesurface was obtained in desiccated air. These differences are consistent with the higher thresholdvalue in distilled water than in desiccated air, as displayed in the da/dN - AK curves (Figure5.14 and Figure 5.16), and can be attributed to corrosion product induced crack closure[38,58,59].5.3.2 Al-Cu alloy5.3.2.1 T-L crack plane orientation, 80 Hz loading frequencyThe T-L fractography of the Al-Cu alloy tested in the different environments was alsopredominantly transgranular. Fractography in the near threshold cracking region showed asimilar irregular crack front to the Al-Li alloys, with numerous tongues extending into theuncracked material and uncracked ligaments between the tongues. Figure 5.34a shows amicrograph obtained from a test in distilled water, and is typical of the other environmentsexceptAid3. In the AIC13environment, the fracture surface was more corroded (Figure 5.34b).As the crack grew into Stage II with increasing AK, well defined ductile striations began toResults 113appear on the fatigue crack surface in all environments. A characteristic striated fractographyis shown in Figure 5.35, a test conducted in distilled water. Similar striations were observed inthe other environment. T-L crack plane orientation, 0.5 Hz loading frequencyDecreasing the loading frequency of the fatigue test from —80 Hz to —0.5 Hz did not causeany obvious change in fractography. When compared to the Al-Li alloy, no secondary cracking(splitting) effects were observed in the Al-Cu alloy. This is the major difference in fractographybetween the Al-Li and Al-Cu alloys.7’Figure5.33SEMmicrographsshowingfatiguefracturesurfacesofAl-Lialloy8090intheS-Lorientationinthere-agedtempernearAKth.Macroscopiccrackingdirectionisfromtherighttotheleftofthepage.(a)Indistilledwater;(b)indesiccatedair.(a)(b)Figure5.34SEMfractographsofnearL\KthcrackingofAl-Cualloy2024-T351intheT-Lorientationat-80Hz.(a)Indistilledwater;(b)in1MAid3.C.,(a)(b)CiT7Figure5.35SEMfractographofStageIIcracking(AK=8MPa.m112)inAl-Cualloy2024-T351intheT-Lorientationindistilledwaterat—80Hz.Results 1175.4 Macroscopic fracture surface appearance, roughness and crack profile5.4.1 Al-Li alloy5.4.1.1 T-L crack plane orientation, 80 Hz loading frequencyFor the Al-Li alloy tested at -8O Hz, the T-L fatigue fracture surfaces produced in theaqueous environments were relatively flat on a macroscopic scale and remained in the initialcrack plane, with no particular tendency for the crack to lead at the centre or edge ofthe specimen.A typical example is shown in Figure 5.36a for two opposing fracture surfaces of a specimentested in the A1C13 solution. The fatigue crack was allowed to propagate until the overall cracklength (a) was —48 mm, after which the test was terminated and the specimen opened byincreasing the load monotonically until unstable crack propagation occurred upon reaching K1.Fatigue surfaces obtained in desiccated air exhibited a characteristically differentmacroscopic appearance from other tests, as shown for opposing fracture surfaces in Figure5.36b. Initially, the crack remained relatively flat and stayed in the original crack plane untilAK reached -3 MPa.m1,corresponding to a 31 mm in Figure 5.36b. Beyond this point, thecrack started to deviate (climb out of the cracking plane) near the centre of the specimen. Asthe crack continued to propagate, the deviation increased in height and width in the centralregion while the outer sections of the crack front remained in the original macroscopic crackingplane. The result was the production of a centrally ridged fracture surface as shown in Figure5.37a, and the angle of the ridge with respect to the original macroscopic crack plane is _600(Figure 5.37b). With reference to the crack length scale in Figure 5.36b and the AK values inFigure 5.8, the ridge was 2 mm in height at a —34 mm and AK -.4.6 MPa.m’, and 4 mm in——l[flhIIIII“innim11111111IIflW”WIN!!miImfrm,rI!I1H!II!!nI!IrOmmlO20304050Ommi0203.040510€(a)(b)Figure5.36MacroscopicappearanceofopposingfracturesurfacesofAl-Li alloy8090-T8771testedintheT-Lorientationat—80Hz.Scaleshowscracklength.(a)In1MA1C13;(b)indesiccatedair.00Results 119height at a —40 nmi and 8.4 MPa.mtt2 Furthermore, the formation of the ridge caused the crackfront to trail in the centre of the specimen so that when the crack length at the surface was —46mm it was only —40 mm at the centre. This trailing effect is clearly seen from the shape of thecrack front in Figure 5.36b, where the specimen was fatigued until the surface length of thecrack was —46 mm before terminating the test and opening the specimen. The trailing effectwas confirmed by additional tests where the fatigue crack was allowed to reach a length of —35.5mm at the surface, corresponding to an apparent AKnear 5.6 MPa.m, after which the unopenedspecimen was sectioned and polished on a plane passing through a = 34.6 mm, correspondingto an apparent AK near 5.0 MPa.m”2 No crack was visible in the central —46% of the sectionwidth, but was present in each outer —27% (Figure 5.38), confirming that the crack was leadingat the edges of the specimen and trailing at the centre. Furthermore, continued grinding andpolishing of the specimen back to the plane where the apparent AK was —4 MPa.m”2revealeda crack that was continuous across the thickness and ridged in the mid-thickness region (Figure5.38), supporting the proposal that the trailing effect was directly associated with out-of-planecracking. All of these observations show that at AK> 3 MPa.m’ the reported AK values areapparent values only, due to the non-linearity of the crack front developed in this desiccated airenvironment.f’o.-ov(b)Figure5.37(a)Schematicdiagramofridgeformationinmid-thickness ofspecimens fatiguedintheT-Lorientationindesiccatedair at—80Hz.Arrowshowsdirectionofmacroscopiccrackpropagation.(b)Theangleof thecentralridgewithrespect totheoriginalmacroscopiccrackplane.(a)t) C(a)(b)Figure5.38Opticalmicrographsof sectionedandetchedAl-Li 8090-T8771 specimenfatiguedat-8OHzinthe T-Lorientationindesiccatedair.Sectionednormaltotheprimarycrackplane.(a)K=5MPa.m”2;(b)LK=4MPa.m”Results 122Another significant observation arising from the study of surface roughness is that the AKvalue at which the central ridge began to form in desiccated air also corresponded very closelywith the value of AK at which the crack rate plateau commenced in this environment (Figure5.8). Fatigue test in desiccated air that commenced from a AK value of 5 MPa.m”2is consistentwith this observation because they exhibited a cracking rate plateau (Figure 5.10) and roughridged fracture surface immediately. Ridge formation and an associated crack rate plateau werealso observed in the hydrogen precharged specimen tested in desiccated air, whereas thecontinuously hydrogen charged specimen showed no ridge and no crack rate plateau (c.f. Figure5.11 and Figure 5.12). T-L crack plane orientation, 0.5 Hz loading frequencyThe macroscopic fractography produced at 0.5 Hz loading frequency in the aqueousenvironments was similar to that of the -.80 Hz tests. The fracture surface produced in desiccatedair at 0.5 Hz was much rougher than those in the other environments at the same loadingfrequency. However, the same large scale ridge as that in the centre of the specimen tested at-.80 Hz in desiccated air was not observed for the 0.5 Hz test. This is also consistent with thefact that no cracking rate plateau was observed for the 0.5 Hz test in desiccated air. This absenceof the large ridge and the related crack growth rate plateau may be associated with the longerexposure times and lower crack tip deformation rates at the lower loading frequency, therebyallowing a greater degree of interaction between the environmental and mechanical effectsduring each load cycle.Results 1235.4.1.3 S-L crack plane orientation, 30 Hz loading frequencyThe S-L orientation fatigue cracking fracture surfaces were very flat on a macroscopicscale, and no obvious difference in surface roughness was observed among fatigue fracturesurfaces produced in the different aqueous and dry environments. Also no obvious differencein surface roughness was observed between the as-received and the re-aged tempers.5.4.2 Al-Cu alloyFor the Al-Cu alloy tested at -80 and -0.5 Hz in the T-L crack plane orientation, themacroscopic appearance of fatigue fracture surfaces produced in the different environmentswere much flatter than those of the Al-Li alloy produced in the aqueous environments. Figure5.39 shows the macroscopic fracture surfaces of a specimen tested in 0.6 M NaC1 at —80 Hzthat are typical of the other environments. Specimens sectioned at different AK values showedvery linear crack front profiles as compared to the crack front proffles of the Al-Li alloy. Acomparison of the sectioned alloys at AK =5 MPa.mla is shown in Figure 5.40 for tests in 0.6M NaCl.Unlike the Al-Li alloy, fatigue tests on the Al-Cu alloy in desiccated air did not producea ridged fracture surface and associated crack rate plateau at —80 Hz. Similar to the Al-Li alloy,no ridge and no crack rate plateau were observed at -0.5 Hz.Figure5.39Macroscopicappearanceof opposingfracturesurfaces ofAl-Cualloy2024-T351fatiguedintheT-Lorientationin0.6MNaC1at-80Hz.Scaleshows cracklength.(a)(b)Figure5.40Opticalmicrographsshowingcrackfrontprofilesof specimensfatiguedintheT-Lorientationin0.6MNaC1at—80Hz.SectionednormaltotheprimarycrackplaneatAK=5MPa.m1.(a)Al-Cualloy2024-T351;(b)Al-Li alloy8090-T8771.IJ400pm—Results 1265.5 Microstructure and compositional distributionFigure 5.41a shows the finely dispersed, precipitation hardening W phase in the Al-Lialloy in the as-received temper (T8771). The appearance is the same as that reported byBlankenship et al. [113] for 8090-T877 1. For the Al-Li alloy after the re-ageing treatment,similar micrographs were obtained (Figure 5.41b). Figure 5.42 shows the grain boundaries withprecipitates on them. From TEM studies with both bright field and centred dark field imagingtechniques, no difference was noticed regarding either the size or the distribution of thestrengthening phases between the two tempers ofmaterials. However, small changes in chemicalcomposition between the two tempers were detected by STEM + EDX analyses. For example,Figure 5.43 compares changes in the Cu concentration of precipitates and matrix in the grainboundary region of the as-received and re-aged material. It is apparent that the grain boundaryprecipitates have become enriched in Cu in the re-aged alloy. Electron diffraction analyseswere conducted to identify the Cu-containing precipitates in the boundary of the as-receivedand re-aged alloy. Analyses of the limited patterns obtained, and comparison with reporteddata in the literature [197-199], led to the conclusion that the precipitates in the boundary wereclose to the T2 (A1CuLi3)phase, as shown in detail in Appendix II.The EELS analyses, conducted by searching the excitation edges and the plasmon peakenergy shifts to detect Li distribution, did not produce useful results in the present study. Theplasmon peak energy shifts measured with the Al-Li specimens relative to a pure Al (99.9995%)indicated a change in the correct direction. However, no reproducible and sensible compositionalprofile could be drawn out from the multiple measurements conducted, due to the limitedattainable resolution of the equipment and the accuracy of the energy shift measurements fromthe energy loss spectra (see Appendix ifi for detail).(a)(b)Figure5.41TEMmicrographshowing6’phaseinAl-Lialloy8090.Centereddarkfieldimage.(a)As-received;(b)re-aged.Figure5.42TEMmicrographshowingprecipitates containinggrainboundaries inAl-Li alloy8090-T8771.Brightfieldimage.(a)As-received;(b)re-aged.(a)(b)Figure5.43STEM+EDXanalysesshowingCucontentprofilesacrossagrainboundarycontainingprecipitates.(a)Grainboundarycontainingprecipitatesinthere-agedAl-Liahoy.Thecontaminationcyclesmarkthelocationsofcompositionanalyses;(b)Cucontentprofilesthroughprecipitates(averagefrom4tests);(c)Cucontentprofilesbetweenornearprecipitates(averagefrom4tests).(a)20 I18 16 14 12 10 8 6 4 2 0 -900-700-500-300-1000100300500700900DistancefromtheGrainBoundary(nm)(b)CL) L)10 9 8 7 6 5 4 3 2 1 0-900-700-500-300-1000100300500700900DistancefromtheGrainBoundary(nm)(c).1Discussion 1326 Discussion6.1 Corrosion fatigue crack propagation in the Al-Li alloy in the T-Lorientation and comparison with the Al-Cu alloy6.1.1 Role of splittingIt is clear from the experimental studies that the S-L splitting phenomenon plays animportant role in the T-L orientation fatigue cracking behaviour of the Al-Li alloy at the highfrequency (—80 Hz) and that the splitting tendency is influenced by environmental effects. Anytensile stress-dependent effects causing S-L splitting must arise from through-thickness stresses.It is well established that the stress state near the tip of a crack is significantly different fromregions remote from the crack tip [32]. With respect to the T, L and S directions, three principaltensile stresses GT, 0L’ and a5 may be present due to constraints imposed by the less stressedmaterial outside the crack tip region, as shown in Figure 6.1. The value of a drops to zerounder plane stress conditions (relatively thin specimens) and has a value equal to V(CTT + aL)under plane strain conditions (relatively thick specimens), where V is Poisson’s ratio [32].Consequently, the through-thickness stress component contributing to the S-L splitting is dueto a. The significant effect of the environment on splitting is apparent from a comparisonbetween the different test conditions. For example, significant splitting was observed in all theaqueous environments and the lowest AKth was observed in the aqueous solution ofA1C13,whichpromoted the largest degree of splitting in the low t\Kregion. In contrast, out-of-plane cracking,the greatest macroscopic surface roughness, and a cracking rate plateau at AK values above —3MPa.m112were observed only in the non-aqueous environment (desiccated air), where negligibleamounts of splitting occurred.SITFigure6.1Schematicdiagramshowingthetriaxialstressstatenearthetipof acrack.Discussion 134The fractography indicated that splitting influences crack propagation by assisting thelinkage of uncracked ligaments between crack front tongues, as shown schematically in Figure6.2. Assuming that linkage occurs by slip dominated shear processes, then the implication isthat additional local slip systems become operative during the linkage process. These additionalsystems are different variants of { 111 }<1 10> slip in the face centred cubic (FCC) matrix andoperate simultaneously with other variants that control slip reversibility advance [41,421 of thepropagating crack front tongues. The situation may be envisaged as a change in the stress statein the local crack tip region, whereby a smaller degree of crack tip plasticity and lower numberof slip systems are present under plane strain conditions, whereas a larger degree of crack tipplasticity and a larger number of slip systems are present under plane stress conditions. Forexample, with respect to Figure 6.1 and shown in Figure 6.3 [32], under plane strain conditionsthe principal shear stress planes at the crack tip are confined to planes that are parallel to theS-direction and inclined 45° with respect to the L-direction, with the maximum of the shearstress being (aT -aL)!2. Only slip plane variants oriented closely to these planes will be operativeand the orientation of these planes are not conducive to shear linkage of uncracked ligaments.In contrast, under plane stress conditions, other principal shear stress planes become dominant.These are parallel to the L-direction and inclined 45° to the S direction, with the maximum ofthe shear stress beinga.1/2 (which is higher than in the plane strain condition). Slip plane variantsoriented closely to these planes now become operative, and the orientation of these planesfavours the shear linkage of the uncracked ligaments. Consequently, splitting may exert itsinfluence on crack propagation behaviour by changing the stress state in the crack tip regionfrom one that is predominantly plane strain to one that includes plane stress. An equivalentway of envisaging these changes in stress state is to consider that splitting causes a change inthe cyclic loading mode near the crack tip from a combination of Model I (opening) and ModeDiscussion 135II (in-plane shear) to a more complex combination of Mode 1,11 and ifi (anti-plane shear). Thisis consistent with the results of a recent study by Kamat and Prassad [202] on an Al-Li 8090-T3alloy plate that fatigue crack propagation resistance is lower under mixed Mode I and Mode ifiloading as compared to that under pure Mode I loading.The concept ofa change in local stress state due to S-L splitting may be tested by comparingthe required number of splits to produce such a change against the number of splits observedduring fatigue. The generally accepted thickness criterion [188] for plane strain testingconditions at the tip of a crack is given by Equation (6.1),t82.5J (6.1)where tB is the minimum specimen thickness above which plane strain conditions are dominantand below which plane stress conditions become increasingly important, and a is the tensileyield stress. Consequently, if there are NB splits along the crack front, the average thickness ofthe material between each free surface, or split, will be decreased from the initial specimenthickness of B to B/(NB + 1). Also, under cyclic loading conditions, the value of K governingthe minimum thickness criterion will beK, as given by equation (4.1). Therefore, substitutingtB =BI(NB + 1) and K= iXKI(1 - R) into Equation (6.1), and rearranging, leads to aNB controlledboundary condition for the cyclic stress intensity (zIKb) above which plane stress conditionsstart to become an important consideration in the crack tip region. The boundary condition isgiven by Equation (6.2),B 1/2IXIc_[(N+l)(25)] [(l—R)a] (6.2)cracked Uuncracked(a)(b)Figure6.2Schematicdiagramshowingeffectof S-Lsplittingonlinkageof uncrackedligamentsnearthecrackfront.Arrowshowsdirectionofmacroscopiccrackpropagation.(a)Beforelinkage;(b)afterlinkage.IT(a)L(b)Figure6.3Deformationpatterns[32].(a)45°sheardeformationinplanestress;(b)hingetypedeformationinplanestrain.ISDiscussion 138Calculated values of AKb for different NB values, using the relevant (480 MPa) in Table4.2, are listed in Table 6.1. Comparisons of the data in Table 6.1 with the splitting analyses inFigure 5.21, and the crack propagation rates in Figure 5.8, clearly show that the observed NB inthe threshold region leads to plane stress conditions (AKffi AK) in the AId3 solution and toplane strain conditions (AK AK,) in all the other environments. Thus, the lower AKth of 1.05MPa.m’ in the AlCl3 solution, relative to —1.7 MPa.m1in the other environments, may beattributed primarily to differences in the crack tip stress state caused by the presence of S-Lsplitting, leading to the conclusion that AK is near--i MPa.m1t2under plane stress conditionsand closer to —1.7 MPa.m1under plane strain conditions. These two situations are analyzedin more detail in Appendix IV. In addition, comparison of Table 6.1 with Figure 5.8 and 5.21confirms that fatigue in the A1C13 solution occurred under conditions of AK> AKb over thewhole range of AK values investigated, suggesting that the crack tip was always subjected tosome degree of plane stress. Similar comparisons and analyses of the splitting data in Figure5.21, fatigue data in Figure 5.8 and the boundary stress intensity data in Table 6.1 show thatplane stress crack tip conditions become important at AK above —3 MPa.m112 in all the otheraqueous test environments, and indicates that the great similarity in Stage II behaviour betweenthese environments and the A1C13 solution was due primarily to plane stress state conditions atthe crack tip.In contrast, for tests in desiccated air, there was a general absence of splitting and NB —0 over a wide AK range. This resulted in plane strain dominated condition, AK < AKb, up to AKvalues at least as high as 7 Consequently, examination of Figure 5.8 shows that thecrack rate plateau, which commenced near a AK of —3 MPa.m”2,developed and grew underplane strain crack tip conditions. These conditions should affect the number of operative slipsystems at the crack tip and contribute to significant out-of-plane cracking and ridge formationDiscussion 139Table 6.1 Boundary Conditions, AKb, above Which Plane Stress Conditions BecomeImportant in the Presence of NB SplitsNB K, (MPa.m’) AKb (MPa.mhIZ)300 1.97 0.99275 2.06 1.03250 2.16 1.08200 2.41 1.21150 2.78 1.39100 3.40 1.7075 3.92 1.9650 4.79 2.3925 6.71 3.3610 10.32 5.165 13.97 6.982 19.75 9.881 24.19 12.100 34.21 17.10Discussion 140in the mid-thickness region of the specimen. The resulting crack morphology inevitably leadsto the retardation of cracking by significantly reducing the effective cyclic stress intensity (AKeff)at the crack tip to a value significantly below that calculated from Equations (4.1) to (4.4).Several processes may decrease including crack deflection effects, lower stresscomponents acting on the inclined surfaces of the ridged crack plane, and significant roughnessinduced closure effects caused by the macroscopically large ridge. Another probablecontributing factor to ridge formation is the tendency for the commercial Li-containing alloysto exhibit marked differences in crystallographic texture in the mid-thickness of plate products[169]. The study of Yoder eta!. [169] showed that in highly textured 2090 Al-Li alloy material({110) <112> and {123) <634>), large slip band ({11l} type) shear facets can form due toplanar slip across many closely oriented grains (i.e. the effective grain size is larger than theactual grain size), resulting in unusual height of asperities on the fatigue fracture surface andthus roughness induced crack closure. Tromans and Mogambedze [2031 observed a dominant{ 110) <112> texture in the mid-thickness region of the same Al-Li 8090-T8771 plate used inthe present study. This { 110) <112> texture in the 8090-T8771 plate may lead to the ridgeformation on the fatigue fracture surface produced in desiccated air at -.80 Hz. Detailed analysisis given below.Because of the pronounced [1OT} <T2T> rolling texture in the mid-section of the Al-Li8090-T877 1 plate, the T-L orientation crack plane in this region exhibits a texture that is { 111 }<12T>, as shown in Figure 6.4a where the macroscopic crack plane is parallel to (ill) and themacroscopic cracking direction is parallel to [T2T]. The [ill } < 110 > slip systems are expectedto be operative in Al-Li alloys [169]. The texture in the 8090-T8771 plate is such that one slipplane, (ill), is parallel to the crack plane, and three others, (1 11), (111) and (iii), areDiscussion 141symmetrically oriented with respect to the load axis, [111], as shown in Figure 6.4b. Theoperative slip systems and their corresponding Schmid factors under uniaxial ([1111 direction)loading conditions are list in Table 6.2.Therefore, considering an applied stress parallel to the T-direction (i.e. normal to the (111)plane), together with the Schmid factors listed in Table 6.2, the following combination of fourslip systems will be required for a slip-dominated (slip reversibility) cracking process to producea macroscopically flat fracture plane (i.e. parallel to (111)) with a <12T> direction of crackpropagation:(1T1)[i10] +(1T1)[O1i] = (1T1)[121](iii) [121] + (ii 1) [TOT] + (liT) [10ii = (111) [T21]The respective orientations of the slip planes and slip directions are shown schematically inFigure 6.5.Macroscopic flat fracture surfaces were obtained during fatigue in the aqueousenvironments, and may be atthbuted to the operation of the above four slip systems. However,under plane strain conditions (i.e. in desiccated air with no splitting phenomena), the hinge typedeformation occurs (Figure 6.3b), favouring slip on (hi), rather than on (iii) and (liT) planes.Hence, the crack will tend to move out of the T-plane, onto (111) in the mid-thickness regionof the plate. The planar slip characteristics of Al-Li alloys, coupled with the out-of-planecracking, wifi form large slip band facets on (111) planes, leading to ridge formation in themid-thicknessregion, as observed (see Figures 5.36 and 5.37). Consistent with this, the measuredangle between the climbing ridge and the initial crack plane is —60° (Figure 5.37b), whichcompares favourably with the 70.5° angle between (111) and (iii) planes.Discussion 142Table 6.2 The Operative { 111 } < 110> Slip Systems and Schmid Factors Under UniaxialLoading (([111] Direction)Slip Plane Slip Direction p (Degree) (Degree) Schmid Factoror cosqcos?..(111) [O1T] 0 90 0[101] 0 90 0[Tb] 0 90 0(liT) [Tb] 70.53 90 0[0111 70.53 35.26 0.27[TOT] 70.53 35.26 0.27(111) [Toll 70.53 90 0[011] 70.53 35.26 0.27[1101 70.53 35.26 0.27(Til) [011] 70.53 90 0[TOT] 70.53 35.26 0.272[110] 70.53 35.26 0.272= angle between the loading direction and the normal of the slip plane= angle between the loading direction and the slip directionSLCrackingDirection4(a)[10ii1121233oh1s101.113•.023203-012115102•I5..155132.1S3013.‘17‘103.I5•313312•123—001213.•l222.113•.•Ti?112II•115l5103013-•3321315••,i5•1351232I—212••102113012•221•313203’112135.023133JL2221..101.315on•—331•533.-5i5213.•.123•155-•353•3312iI312.515’.313•55o‘132•Tai-32•153.231-.55132!’513.302‘21222.133551110531•311•201••.31232.•021•131.351110shi•513•.I2I151-320’210ITo.230—71I•30!311•32122123!•131•Th310’7II.331••151•.iii.130351•100531..55!.35!•oo120321-—‘711310’210.lIp..•-—171.—31511•711’-320230-.171•151311•—•511•531‘551•35J151•13i30!.31i•321’•331.231•13i.031121211201-.-02!--513•-i2ii—221-•121-•153.513‘533353•153.312302.••31232’••o3132.212122133——..5l5•313335••155ioiOil(b)213-123112Figure6.4(a)SchematicdiagramshowingtherelationoftheinitialT-LmacroscopiccrackplanewiththetextureinthemidthicknessoftheAl-Li8090-T8771plate.(b)Standard(111)projectionforaface-centredcubiccrystal.I[121]Direction4(111)Plane01][110][011]Figure6.5Relativeorientationsoftheoperativeslipsystemsintheoverall (111)[T2T]cracking.Discussion 145In the desiccated air environment at —80 Hz load frequency, the fatigue crack growth rateplateau and the corresponding ridge formation on the fracture surface commence at a AK of -3MPa.m1.At this AK value, the monotonic crack tip plastic zone size in the (iT 1) plane underplane strain conditions in the mid-thickness region of the plate, can be calculated from thefollowing formula [32],r(O) = K, [sin20 + (1 — 2v)(1 + cos e)] (6.3)4ita 2where, in the present case, 0 is the angle between the (1T1)and the T-plane. Substituting 0 =70.5°, Poisson’s ratio v = 1/3, Km, =6 MPa.m”2and a = 480 MPa from Table 4.2, Equation(6.3) gives, Ty (70.5°) 1.8 x 10.2 mm. This value is very close to the T-direction grain size ofthe Al-Li alloy plate (-.2 x 102mm). The similarity between these values may indicate that thecentral ridge starts to form when the extent of plastic deformation reaches the point such thatthe planar slip on (iT 1) planes is extending fully across one grain to the neighbouring grain. Inthe situation of textured material with closely oriented grains, the planar slip band can cut throughthe grain boundary and extend into the next grain without being diverted. As the extent ofplastic deformation at the crack tip increases with increasing AK, due to crack growth, the planarslip band can cut through many closely oriented grains and create a large asperity on the fracturesurface, resulting in the large ridge formation in the mid-thickness region of the specimen, asobserved in the present study.The effects of texture on fracture surface roughness are largely reduced in the aqueousenvironments when the splitting effects are present, due to the change in stress state ahead ofthe crack tip from plane strain to plane stress. The 45° shear deformation under plane stressconditions (Figure 6.3a) is more uniformly distributed among (Ti 1), (iii) and (iT 1) planes. InDiscussion 146this situation, the above slip systems are likely to operate coordinately to produce overallmacroscopic cracking on (11 1)i.e. the T-plane. Therefore, the presence of splitting effects willdecrease roughness on the fatigue fracture surface.It is worth noting that, Rao et al. {107,1 08] have reported that splitting effects, which theydescribed as crack divider delamination, may occur at the onset of unstable (critical) crackpropagation in some commercial Al-Li alloys and increase the fracture toughness (K1). Theyinterpreted the improved toughness as being due to a change in stress state from plane strain toplane stress and a consequent increase in energy required for fracture. The present studyconsiders stable (subcritical) crack propagation under fatigue conditions where splitting hasbeen obtained at stress intensities well below K1. The splitting phenomenon, and developmentof a plane stress state will not lead to improved resistance to fatigue cracking when the fatiguecrack propagation process is slip controlled, because plane stress conditions enhance the degreeof plastic deformation and facilitate crack tip linkage.6.1.2 Role of environmentIt seems reasonably clear from the observations of the T-L orientation fatigue crackingbehaviour of the Al-Li alloy at -80 Hz and the related fractography that the large degree ofsplitting in the Aid3solution at zKth, and as AK increased to -‘3 MPa.m, was associated withsignificant stress-independent dissolution of grain boundaries lying in the rolling plane (i.e. theS-plane containing the T and L directions). Such localized dissolution is almost certainly dueto chemical heterogeneities in the grain boundary region, which cause the boundary to be anodicrelative to the grain, and this is enhanced by the acidic pH of the test solution. Direct evidencefor the above assertion is provided by the immersion test in the Aid3 solution in which theDiscussion 147initial corrosion delineated the grain boundaries and extended corrosion period led to deepintergranular corrosion. The intergranular penetration rate on the fracture surface of the fatiguespecimen may be relatively faster than that on the immersion test specimen, because thedislocation movements inside the cyclic crack tip plastic zone during the previous fatiguefracture process may have further enriched the grain boundaries near the fatigue fracture surfacewith solute or impurity elements.Indirect evidence for solute segregation effects at the S-plane boundaries is provided bythe low K1 value of —14.7 MPa.m”2obtained in the S-L orientation (short transverse fracturetoughness), which is listed in Table 4.2. Lynch [1121 has studied the toughness of 8090 Al-Lialloy plate in this orientation and his work strongly indicates that low S-L toughness values of—15 MPa.m’ are associated with solute segregation effects at the grain boundary. He wasunable to positively identify the segregant, but concluded that lithium segregation effects wereconsistent with his experimental results. The general absence of splitting phenomenon at AKthin the NaCI solution and distilled water (see Figure 5.21) is attributable to the fact that theseneutral pH environments are sufficiently less aggressive than the A1C13 solution that localdissolution rates at the grain boundaries are too small relative to the plane strain thresholdcracking rates to produce any significant grain boundary penetration.The increase in degree of splitting as AKrises from —3 to—S MPa.m’ in the A1C13solution,and as AK rises from AKIh to —5 or 6 MPa.m112 in the other aqueous environments (see Figure5.21), appears to be due to a synergism between the cyclic load and the environment that resultsin grain boundary cracking (splitting) ahead of the main crack front. The lower AKth value inthe AJC13 solution and the higher cracking rates in A1C13 and distilled water, as shown in theS-L orientation fatigue cracking behaviour of the Al-Li alloy, are consistent with the splittingDiscussion 148results. While this may possibly be due to preferential dissolution of the grain boundaries arisingfrom acombination of film rupture and grain boundary segregation effects, the role ofdissolutionmay be less important than that of hydrogen embrittlement, because theE0potentials observedduring fatigue testing were fully consistent with thermodynamic conditions that allowedhydrogen evolution. Also, dissolution and hydrogen evolution need not be confined solely tothe grain boundaries during each load cycle. Local slip processes at the crack tip should enhancecorrosion along the whole crack front by rupturing corrosion films to expose newly generatedmetal surface to the environment, although the more actively corroding grain boundary regionsmay put the other part of the grains under a certain degree of cathodic protection. Some of thehydrogen generated during the corrosion process will be absorbed into the metal lattice asmonatomic hydrogen, which is mobile at room temperature, and will permeate slowly throughthe metal. For example, Gest and Troiano [204] have clearly demonstrated room temperaturepermeation ofhydrogen through thin stressed membranes ofa high strength 7075 Al-alloy underfreely corroding conditions in aqueous NaCl. Consequently, it may be the absorbed hydrogenthat is promoting the grain boundary splitting. Therefore, it is important to estimate the diffusiondistance, x, of monatomic hydrogen into the crack tip region during each load cycle and tocompare this distance with the average crack growth increment per cycle (da/dIV).If the assumption is made that the concentration of dissolved hydrogen at the crack tipsurface is held at a reasonably constant value, C, by local corrosion processes, mathematicalsolutions to the mass transport problem in one dimension [205] show that the concentration ofhydrogen, C,, at a distance x ahead of the crack tip after time t is given by Equation (6.3):C = C[l erf(2(D)1] (6.4)Discussion 149where D is the diffusion constant for monatomic hydrogen in the metal lattice. Also, errorfunction tables [2051 show that C, = 0.5 C whenx (Dt)1’2 (6.5)Thus, Equation (6.5) provides a reasonable estimate of the distance at which a significantconcentration of hydrogen, relative to C6, is present in the metal lattice.For the present purposes, t is the period of one load cycle at 80 Hz, 1.25 x io second,and Ricker and Duquette [206] concluded that values of D for hydrogen in aluminium, basedon the published literature, should lie between 1 x 10.12 and 1 x 10m2Is. In a later paper, Chenand Duquette [179] considered 1 x i’m2/s to be an appropriate value of D for Al-Li alloys.Consequently, using the latter value, Equation (6.5) shows thatx is 3.54 x i0mm. Examinationof Figure 5.21 shows that a peak in the number of splits occurs between cyclic stress intensitiesof 5 and 6 MPa.m when tested in the aqueous environments, corresponding to crackpropagation rates between 2 x iO and 6 x iO mm/cycle in Figure 5.8. The good agreementbetween these cracking rates and the magnitude of the computed value of x strongly suggeststhat the diffusion of hydrogen during each load cycle is sufficiently rapid to influence grainboundary splitting up to AK values of 5 to 6 MPa.m’. At higher cracking rates, the overalladvance of the crack front per cycle will become larger than x and the degree of splitting shoulddecline, consistent with the decreasing amounts of splitting that were observed at AK> 6MPa.m”2 in Figure 5.21. It is possible that other transport processes may influence thedistribution of hydrogen in the cyclic plastic zone, such as dislocation transport and grainboundary diffusion processes. Also, the amount of hydrogen absorption at the crack tip, andthe actual concentrations C and C, will be dependent on such factors as the local pH andhydrogen overvoltage [207], and the area fraction of surface that is covered by corrosion films.Discussion 150These, in turn, will affect the prevalence of splitting in different solutions. Nevertheless, withinthese limitations, the simple calculation based on Equation (6.5) is considered to be sufficientlyadequate to demonstrate that splitting effects are reasonably consistent with the presence ofdissolved hydrogen.The experimental results from thehigh frequency fatigue test under continuous hydrogencharging provided more support to the above argument. From the change of split number withAK (Figure 5.27), it is evident that splitting is caused by the synergistic interaction betweenhydrogen embrittlement and the lateral stress arising from through-thickness constraints. Atlow AK values near the threshold, the acidic solution (pH = 1) did not cause a large number ofsplits to form by anodic dissolution, because the dissolution process was sufficiently suppressedby the cathodic polarized potential (-2.1 V). While at a relatively higher AK, a large numberof splits were observed at the splitting peak due to the interaction of hydrogen and the cyclicload. This peak was reached at a higher AK value relative to the other test conditions, and isprobably associated with the higher hydrogen production rate occurring under the hydrogencharging conditions.Considering the test conditions of all the high frequency tests on the Al-Li alloy, it appearsthat a continuous hydrogen source is required at the crack tip to produce the splitting effect andto maintain the crack in its initial macroscopic cracking plane. If this hydrogen source isremoved, the splitting process will stop. This was shown by the combined test conductedconsecutively in distilled water and desiccated air, where the splits produced at the crack tip inwater disappeared when the test was continued in desiccated air and cracking did not continuein the previous macroscopic plane at the previous growth rate (see Figure 5.26 and Figure 5.9).Discussion 151Itis concluded that the hydrogen precharged specimen contained insufficient absorbed hydrogento have any significant effect, because subsequent fatigue of the specimen in desiccated airremained free from the splitting phenomenon.The precise mechanism by which hydrogen promotes splitting is not established, but itmay be related to the low S-L toughness caused by the possible segregation of Li at the grainboundaries [112]. Electronic interactions between Li and the monatomic hydrogen in the metallattice could reduce the cohesive energy between the atoms at the boundary so that thethrough-thickness stresses produced by lateral restraint in the crack tip region during fatigueloading are sufficient to cause boundary decohesion (splitting). Strong interactions betweenthe Li segregant and the dissolved hydrogen are expected because of the stability of the hydride,LiFT [208]. Balasubramaniam et al. [137,138] showed by transmission electron diffractiontechniques that a brittle hydride, LiAIH4can form from grain boundary precipitates of 6 (AILi)phase in hydrogen charged Al-Li alloys, and they suggested that the hydride plays a major rolein stress corrosion cracking of these alloys.It is possible that impurities (mainly Na and K) contained in the Li alloying addition mayalso play a role. There has always been a strong argument that the 3 to 10 ppm Na and Kimpurities contained in commercial purity Al-Li alloys could be responsible for their lowershort-transverse fracture toughness [81,82,209].6.1.3 Effects of loading frequencyLowering the loading frequency may lead to the following changes:1. Decreased strain rate at the crack tip;2. Increased time for environmental exposure during each load cycle;Discussion 1523. Increased time for material recovery and increased probabilities for other thermallyactivated processes (e.g., cross slip) during each load cycle;4. Change of local environment chemistry due to relatively poor mixing with thesurrounding bulk solution.Interactions among the changes 1 to 4 may lead to different consequences depending onwhich factor is dominant. For example, the similarity in crack growth behaviour between theAl-Li alloy fatigued in 0.6 M NaC1 and that tested in 1 M AiCl3 at —0.5 Hz indicates that theircrack tip chemistries may be similar. This suggests that the crack tip solution in the neutralNaCl moves towards an acidic A1C13 composition due to the hydrolysis of dissolved Al duringthe corrosion fatigue process,A13+nH2OAl(OH)3+nH (6.6)The higher apparent threshold values at lower frequency may be associated with crack closureeffects due to increased amounts of corrosion product at the crack tip and/or the increased timefor the material at the crack tip to recover from deformation damage during each load cycle.However, the thresholds observed at —0.5 Hz are possibly overestimated because time constraintsdid not allow fatigue to be studied at the same low crack growth rates per cycle as those conductedat —80 Hz.Although the fatigue fracture surface of Al-Li 8090-Th771 produced at 0.5 Hz indesiccated air was much rougher than those produced at 0.5 Hz in the aqueous environments,it did not show an obvious large central ridge as those observed on fatigue fracture surfaces ofthe as-received Al-Li alloy tested at 80 Hz in desiccated air. Consistently, the crack growthcurve for the 0.5 Hz test in desiccated air did not show a crack rate plateau. Study of the fatiguefracture surfaces of the Al-Li alloy produced at 0.5 Hz in desiccated air revealed that the ridgeDiscussion 153still tended to form but the scales were much smaller as shown in Figure 5.30, and the faces ofthe small ridge correlated very well with the operative slip planes drawn in Figure 6.5. Theridge was terminated at small scales when slip deformation on the three slip planes ((111), (1 11)and (111)) intercepted and combined. A convincing explanation on why the ridge formation at0.5 Hz loading frequency did not develop to the same large scales as that at —80 Hz loadingfrequency has not be reached,but it may be related to fact that the split number is slightly higherat the low frequency than at the high frequency (see Figure 5.31 and Figure 5.21). Althoughthe increased number of splits at 0.5 Hz is not yet high enough to change the crack tip stressstate from plane strain to plane stress (see Figure 5.31 and Table 6.1), it is closer to plane stressconditions which produce a flatter fracture surface in this Al-Li plate, as discussed in Section6.1.1 regarding the effects of the presence of splits on crack tip stress state and fracture surfaceroughness. The increased number of splits at 0.5 Hz may be attributed to the longer time forinteraction between the remaining moisture in the environment and the lateral stress during eachcycle. Other factor, such as increased probability for cross slip at lower frequency, may alsoplay a role in restricting the scales of the ridge. Therefore, the fatigue crack growth curve at0.5 Hz in desiccated air did not show a crack rate plateau as that at 80 Hz which was caused bythe severe roughness induced crack closure from the large ridge formation.6.1.4 Effects of microstructure and deformation behaviourThe Al-Cu alloy exhibited different high frequency fatigue characteristics from those inthe Al-Li alloy. The difference were more significant for tests conducted in desiccated air,where a crack rate plateau and ridge formation were not observed in the Al-Cu alloy. DifferencesDiscussion 154were less evident in the aqueous environments, but still detectable, as shown by the flatter crackfront proffles in the Al-Cu alloy relative to the Al-Li alloy (c.f. Figure 5.40) and the absence ofsplitting effects.The behaviour of the Al-Li alloy is closely related to the planar slip characteristicsconferred by the coherent 6’ (AI3Li) phase, the anisotropic grain structure, and the stronglytextured microstructure, as explained in the previous section (Section 6.1.1). In contrast, theAl-Cu has a more equi-axed grain structure, is less strongly textured and is not expected toexhibit strong planar deformation characteristics [11,13]. These factors probably contribute tothe similarity in behaviour of the Al-Cu alloy in the desiccated air, distilled water and NaC1environments.The lower AKth of the Al-Cu alloy in the A1C13 solution, relative to AKth in the otherenvironments, is likely associated with the higher corrosion and accompanying hydrogengeneration rates at the crack tip in this solution. Such processes are enhanced at the crack tipdue to the generation of freshly exposed metal by the cyclic deformation, and more time isavailable per cycle for environmental effects to exert an effect at low crack growth rates. Asthe K value increases, the mechanical effects on fatigue cracking become more and moredominant so that the environmental effects are overshadowed, and the crack rates converge withthose in the other environments.The above explanation for the low AKth of the Al-Cu alloy in the AId3 in the absence ofsplitting effects does not contradict the explanation proposed for the low AK, of the Al-Li alloyin A1C13,where a change of stress state was attributed to splitting effects. In the Al-Li alloy,the sensitive grain boundary region dissolves preferentially and the other part of the fresh metalexposed by cyclic slip deformation is under a certain degree of cathodic protection caused byDiscussion 155the higher dissolution rate of the grain boundary regions, as illustrated schematically in Figure6.6. Therefore, the direct contribution from anodic dissolution to transgranular crack tip advancein the Al-Li alloy is not as large as that in the Al-Cu alloy.Fatigue crackpropagation behaviour ofthe Al-Li and Al-Cu alloys at the low test frequencyis more similar, with the Al-Li alloy showing a slightly higher environmental sensitivity.In summary, when compared to the Al-Cu alloy, the Al-Li alloy showed very superiorT-L orientation fatigue crack propagation resistance in the dry air environment, and slightlybetter or equivalent fatigue crack propagation resistance in the aqueous environments at thesame AK values. (Note that the Al-Cu alloy has a slightly higher K1 and is able to sustain alonger critical crack length before unstable fracture.).0Figure6.6Schematictheoreticalpolarizationdiagramshowingpossiblecathodicprotectionontheotherpartofthegrainsbytheactivedissolutionof thegrainboundaryregions.I,<4duetoenhancedcorrosionatthegrainboundaries(JOB)=CorrosionCurrentonTransgranular RegionswithoutProtectionfromBoundaryRegions‘p=CorrosionCurrentonTransgranular RegionswithProtectionfromBoundaryRegions‘GB=CorrosionCurrentonBoundaryRegionsAnodicEox=ReversiblePotentialofOxidant(e.g.HIons)En=ReversiblePotentialofGrainBoundaryRegionsEG=ReversiblePotentialof TransgranularRegionsOxidationandReductionCurrent:———GrainBoundaryRegionsTransgranular RegionsCombinedEG EGB\IIIICathodicI 1IpL1JOBCurrentDiscussion 1576.2 Corrosion fatigue crack growth in the Al-Li alloy in the S-L orientationand effects of the re-aging treatment6.2.1 Possible changes in the material caused by the re-aging treatmentIt is clear from the fracture toughness and hardness tests that the short time re-agingtreatment can largely increase the short transverse toughness of the material with only a slightdecrease in hardness, confirming that the re-aging treatment is an effective way to improve theshort-transverse fracture toughness.Two mechanisms have been proposed to explain the effects of the re-aging treatment onthe short-transverse fracture toughness value in the literature:1. Transient decrease of Li segregation at the grain boundaries, by Lynch [112].2. Reduced slip planarity due to a combination of W (Al3Li) phase dissolution and S’(Al2CuMg) growth, by Blankenship eta!. [1131 and Slavik et at. [114].In the study of short transverse fracture toughness of 8090 Al-Li plates, Lynch [112]proposed that Li segregation along the grain boundaries is responsible for the low fracturetoughness of the T877 1 heat treatment. He rationalized that the short time re-aging treatmentincreases the fracture toughness because it allows the Li segregant to diffuse from the grainboundaries to nearby precipitates at the grain boundaries before bulk diffusion from the matrixreplenishes the segregant at the grain boundaries. Re-embrittlement occurs when the grainboundaries are re-segregated with Li due to the diffusion ofLi from the matrix. However, Lynchprovided no direct evidence to support his proposal. In his fractographic study, Lynch noticedthat fracture surfaces of the as-received material consisted of a mixture of well defined dimplesDiscussion 158and relatively smooth facets, while that of the 230°C, 5 minutes re-aged material was completelycovered with well defined dimples. Our SEM studies of the transition region between the fatigueprecracked and monotonic fractured surfaces showed similar features to those in Lynch’sobservations. In addition, lines of trenches along part of the transition region were observed inthe re-aged material in the present study, which were not reported in Lynch’s article [112].These trenches may cause blunting of the crack and provide an additional contribution to theincreased fracture toughness.In studies by Blankenship and Starke [113] and Slavik et al. [114], the increased fracturetoughness obtained by the re-aging treatment was attributed to a combination of 6’ (A13Li) phasedissolution and S’ (A12CuMg) phase growth. They proposed that these microstructural changeslead to a transition from a coarse planar deformation mode to more homogeneous deformationand results in the increased fracture toughness of the re-aged material. They pointed out thattheir TEM study did not reveal any differences in 8’ distribution between T877 1 and the re-agingtempers, although they could identify an increase of S’ phase in the re-aging temper. Thedecreased 6’ volume fraction in the re-aging temper was documented by Guinier X-ray analysis[113] and small angle neutron scattering (SANS) [210].Consistently, the TEM observation in the present study showed no discernible differencesin 6’ phase distribution between the two tempers (Figure 5.39). Attempts to measure Licompositional profiles by EELS, using Li absorption edge measurements and plasmon peakenergy shifts, were unsuccessful, as explained in Appendix ifi.Intergranular segregation effects were suggested by the immersion etching tests in 1 MAid3 solution, where a detectable improvement in intergranular corrosion resistance wasobserved after the re-aging treatment. The SEM + EDX and STEM + EDX analyses indicatedDiscussion 159a possible relationship between corrosion resistance and the effects of re-aging on the Cudistribution. It is known that alloyed Cu in solid solution ennobles Al-alloys against corrosion[200]. The improved intergranular corrosion resistance of the re-aged material may be relatedto a local increase of Cu in solid solution. The re-aging temperature (230 °C) is higher than theoriginal aging temperature (170 °C), and the increased solubility at higher temperature may putmore Cu back into solution by dissolving some small precipitates during the re-aging treatment.The obvious increase of Cu concentration in the grain boundary precipitates after re-aging(Figure 5.43b) indicates that a diffusion controlled re-distribution of Cu is achievable duringthe re-aging treatment. This possibility is supported by the fact that the Cu content that wasdetected near the grain boundary region was slightly higher in the re-aged material (Figure5.43c).It is known that localized corrosion starts at sites with the lowest Cu concentration (Cudepleted zone) in binary Al-Cu alloys [201]. The grain boundary regions in the as-receivedmaterial were only slightly more active than those in the re-aged material, and the differencewas not easily discernible in the early stage of corrosion (Figure 5.2). However, as corrosionproceeds, Cu enrichment will occur on the side surfaces of the corrosion sites, allowing thesesurfaces to act as cathodes in the local corrosion cells. The Cu enrichment will accelerate thelocal cathodic evolution of hydrogen because the hydrogen overpotential on Cu is lower thanthat on Al [154], resulting in accelerated dissolution of the local anodic sites at the tip of thepropagating corrosion slots. Therefore, deeper corrosion slots and larger differences betweenthe intergranular corrosion rates of the as-received and re-aged tempers were observed in thelater stages of the intergranular corrosion tests (Figure 5.3 and Figure 5.4).Discussion 160Difference in intergranular corrosion behaviour may also be accounted for via changesin Li segregation at grain boundaries, as proposed by Lynch [112]. A decreased concentrationof the reactive Li element at the grain boundaries should lead to reduced sensitivity to corrosion.Unfortunately, it was not possible to confirm the presence or absence of Li segregation in thepresent work.Moreover, during corrosion the W phase is likely to be anodic to the Al matrix in thecorrosion process [123], like the (AILi) phase. The decreased volume fraction of W phase inthe re-aged material, as proposed in the studies by Blankenship et al. [113] and Slavik et at.[114], may also be reflected in the less generally corroded appearance of the re-aged material.Although, the present study detected no significant changes in the volume fraction of W phase.The low K1 of the as-received material and its improvement is similar to classical temperembrittlement phenomena in steels that are caused by segregation of tramp elements. In Al-Lialloys, tramp Na and K elements are introduced with Li. There have always been argumentsthat these tramp elements could be responsible for the lower short-transverse fracture toughnessof commercial purity Al-Li alloys [8 1,82,209].6.2.2 Effects of re-aging on the S-L crack growth ratesIt is interesting to see that the fatigue crack propagation resistance of the re-aged materialin desiccated air is inferior to that of the as-received material, showing a lower threshold stressintensity value for cracking and faster crack growth rates.Considering the possible changes in the material caused by the re-aging treatment, thedecreased fatigue crack propagation resistance in the dry air environment caused by the re-agingDiscussion 161treatment may be due to one or more of the following reasons:1. Decrease of strength (hardness): A decreased yield strength produces a larger crack tipopening displacement, leading to faster cracking rate and a lower AKth value [160].2. Reduced planarity of slip: The superior fatigue crack propagation resistance ofLi-containing alloys is attributed mainly to the “extrinsic toughening” effects from crackdeflection, crack branching and roughness-induced crack closure [6-1 1]. These effectsstem from the planar slip characteristics of Li-containing alloys. Therefore, reducedplanarity of slip [113,114] can lead to reduced fatigue crack propagation resistance.3. Replenishment of the embrittling element(s): According to Lynch’s theory [1121, there-aging treatment caused a transient decrease ofLi segregation at the grain boundaries.However, it may be argued that during the fatigue process, the slip movement at thecrack tip and the associated dislocation motion can facilitate the re-segregation ofembrittling species from the matrix to the grain boundaries. Grain boundaries insidethe cyclic crack tip process zone can be replenished with the embrittling element(s) bydislocation transportation during fatigue. Therefore, similar fatigue resistance couldresult for both the re-aged and as-received materials, if other related material properties(e.g. yield strength) were not changed by the heat treatment.6.2.3 Effects of environments on the S-L crack growth ratesInterestingly, no obvious difference in fatigue crack propagation resistance was observedbetween the as-received and re-aged materials when tested in each aqueous environment. Thesimilar fatigue crack propagation resistance between the two tempers of materials in the aqueousDiscussion 162environments may be attributed to the combination of the inferior “mechanical fatigue”resistance with the superior intergranular corrosion resistance (accompanied by lower hydrogenproduction during corrosion) of the re-aged material relative to the as-received material.When comparing the fatigue crack growth in both tempers of materials in differentenvironments, large differences were observed. The higher threshold (LKth) in distilled waterthan in desiccated air is attributed to corrosion product induced crack closure[15,17-19,58,59,179]. The SEM observation showed the blackening marks ofrubbing or frettingand a thicker layer of corrosion product on the fatigue fracture surface produced in distilledwater near iXKth (Figure 5.33a), while no such marks were observed on fracture surface indesiccated air (Figure 5.33b). However, at relatively higher b.K, the closure effect was largelyovercome [7,9,98], and the embrittling effect of water dominated, resulting in faster crackingrates in distilled water. In the ALC13 solution, no crack closure resulted, because the corrosionproduct is less stable and is soluble in the acidic environment. Moreover, the higher corrosionand hydrogen production rates, and their interaction with cyclic deformation at the crack tip,combined together to produce a lower threshold (AKth) and higher cracking rates in thisenvironment.The S-L fatigue cracking fractography was predominantly intergranular for both tempers(Figure 5.33). The easier cracking along the grain boundaries in distilled water and AIC13solution is consistent with the observation that fatigue crack propagation in the T-L orientationwas accompanied by lateral splitting along the S-L orientation in these environments.Furthermore, the lower S-L fatigue resistance in Aid3 also corresponds to the higher splittingtendency (see Figure 5.21 and Figure 5.31).Discussion 163Considering the faster crackpropagation rates and the lower threshold (liKth) ofthe re-agedmaterial in desiccated air, the beneficial effects of the re-aging treatment on improved fracturetoughness may be largely compromised in regard to the overall lifetime of a structure orcomponent. Although a higher fracture toughness value means that unstable cracking will bedelayed to a longer critical crack length, the lower threshold (zKth) and higher cracking ratewill reduce the time period (lifetime) required to reach this critical length. Near threshold fatiguecrack growth behaviour is important in detennining service lifetime, because this periodoccupies the largest fraction of the lifetime under normal situations.In summary, a re-aging treatment that has been shown to be effective in increasing theshort-transverse fracture toughness does not improve the corrosion fatigue crack propagationresistance. This means that the re-aging treatment is less promising for commercial applicationswhere fluctuating loads prevail.Summary and Conclusions 1647 Summary and Conclusions(1) Regarding the T-L orientation crack propagation behaviour, the following conclusionsemerge:(i) The T-L orientation fatigue crack propagation resistance of Al-Li 8090-T8771 issuperior to Al-Cu 2024-T351 in a desiccated (dry) air environment, and slightly better than orequivalent to Al-Cu 2024-T351 in the aqueous environments.(ii) The superior crack growth resistance of the Al-Li alloy in desiccated air comes mainlyfrom the crack deviation and roughness induced crack closure.(iii) The presence of corrosive environments can largely accelerate the fatigue crackpropagation rate. For Al-Li 8090-T8771, the primary effect of aqueous environments is topromote S-L splitting (delamination) along grain boundaries normal to the initial crack planein the crack tip region. Splitting does not occur in the Al-Cu alloy.(iv) The splitting occurs ahead of the main crack front and affects the main crack growthby changing the local stress state, keeping the main fatigue crack in the original crack plane andrestricting the fracture surface roughness induced crack closure.(v) In the absence of an aqueous environment, S-L splitting is avoided. A large degreeof crack deviation and fracture surface roughness result, leading to severe crack growthretardation. The central ridge formation on the fatigue fracture surface is related to the strongtexture in the mid-thickness of the Al-Li 8090 plate.Summary and Conclusions 165(vi) Splitting arises from both localized anodic dissolution and hydrogen embrittlementprocesses. The occurrence of significant splitting at IXKth is due to anodic dissolution, whereasthe increased splitting that occurs with rising AKis attributed mainly to hydrogen embrittlement.(2) Regarding the S-L orientation fatigue crack propagationbehaviour ofAl-Li alloy 8090,the following conclusions are reached:(i) The fatigue crack propagation resistance is lower in the S-L orientation than in the T-Lorientation. The easier intergranular cracking along the S-L orientation is consistent with theobserved S-L splitting phenomenon in the aqueous environments during the T-L crackorientation tests.(ii) The re-aging treatment, effective in increasing the short-transverse fracture toughness,does not improve the corrosion fatigue crack propagation resistance. This makes the re-agingtreatment less promising for commercial applications, especially in situations where fluctuatingloads prevail.(iii) The re-aging treatment improves the intergranular corrosion resistance of the material.(iv) Dislocation transport processes during fatigue may facilitate re-segregation of theembrittling species from the matrix to grain boundaries inside the cyclic crack tip process zone.The similar crack growth behaviour between the two tempers of materials in each environmentis due to the combination of the inferior “mechanical fatigue” resistance with the superiorintergranular corrosion resistance (accompanied by lower hydrogen production duringcorrosion) of the re-aged material relative to the as-received material.References 1668 References1. Polmear, I.J., LightAlloys, 2nd edition, London, 1989, Pp. 105-109.2. James, R.S., “Aluminium-Lithium Alloys”, in Metals Handbook, 10th edition, Vol. 2, 1990,pp. 178-199.3. Starke, Jr., E.A., T.H. Sanders, Jr. and I.G. Palmer, “New Approaches to Ahoy Developmentin the Al-Li System”, Journal ofMetals, Vol. 33, No. 8, 1981, pp. 24-33.4. Webster, D. and C.G. Bennett, “Tough(er) Aluminium-Lithium Alloys”, Adv. Mater. &Processes, 10/89, pp. 49-54.5. Aliaga, D. and E. Buillon, “Corrosion Fatigue Behaviour of Some Aluminium Alloys”, inCorrosion Fatigue, AGARD Conference Proceedings No. 316, NATO, 1981, pp. ffll-ffl12.6. Rao, K.T.V. and R.O. Ritchie, “Fatigue of Aluminium-Lithium Alloys”, IntL MaterialsReviews, Vol. 37, No. 4, 1992, pp. 153-185.7. Vasudevan, A. K. and S. Suresh, “Lithium-Containing Aluminium Alloys: Cyclic Fracture”,Metall. Trans. A, Vol. 16A, 1985, pp. 475-477.8. Harris, S.J., B. Noble and K. Dinsdale, “Fatigue Crack propagation in Al-Li-Mg-Cu-Zr (8090)Alloys”, in Aluminium Technology’86, T. Sheppard, Ed., The Institute of Metals, 1986, pp.45 1-458.9. Rao, K.T.V., W. Yu and R.O. Ritchie, “Fatigue Crack Propagation in Aluminium-LithiumAlloys 2090: Part I. Long Crack Behaviour”, Metall. Trans. A, Vol. 19A, 1988, Pp. 549-561.10. Rao, K.T.V. and R.O. Ritchie, “Mechanical Properties ofAl-Li Alloys, Part 2: Fatigue CrackPropagation”, Mater. Sci. & Technol., Vol. 5, 1989, pp. 896-907.11. Petit, I., S. Suresh, A.K. Vasudevan and R.C. Malcolm, “Constant Amplitude andPost-Overload Fatigue Crack Growth in Al-Li Alloys”, see ref. 73, pp. 257-262.12. Magnin, T., P. Rieux, C. Lespinasse and C. Bathias, “Fatigue Crack Initiation andPropagation Properties of Al-Li-Cu Alloys in Air and Aqueous Corrosive Solution”, seeref. 74, pp. 8 17-822.13. Ohrloff, N., A. Gysler and G. Lutjering, “Fatigue Crack Propagation Behaviour of 2091T8and 2024 T3 under Constant and Variable Amplitude Loading”, see ref. 73, pp. 801-807.14. Tintilier, R., H.S. Yang, N. Ranganathan and J. Petit, “Near Threshold Fatigue Crack Growthin 8090 Lithium Containing Al Alloy”, see ref. 74, pp. 777-784.References 16715. Shin, K.S. and S.S. Kim, “Environmental Effects on Fatigue Crack Propagation of A 2090Al-Li Alloy”, inHydrogenEffects onMaterialBehaviour, N.R. Moody and A.W. Thompson,Eds., TMS-AIME, Warrendale, PA, 1990, pp. 919-928.16. Piascik, R.S. and R.P. Gangloff, “Environmental Fatigue of an Al-Li-Cu Alloy: Part I.Intrinsic Crack Propagation Kinetics in Hydrogenous Environments”, Metall. Trans. A, Vol.22A, 1991, pp. 241 5-2428.17. Peters, M., K. Welpmann, W. Zink and T.H. Sanders, Jr., “Fatigue Behaviour ofAl-Li-Cu-Mg Alloy”, see ref. 73, pp. 239-246.18. Peters, M., V. Bachmann and K. Welpmann, “Fatigue Crack Propagation Behaviour of theAl-Li Alloy 8090 Compared to 2024”, see ref. 74, pp. 785-79 1.19. Pao, P.S., M.A. Imam, L.A. Cooley and G.R. Yoder, “Comparison of Corrosion-FatigueCracking of Al-Li Alloy AA2090-T8E41 and Alloy AA7075-T651 in Salt Water”,Corrosion, Vol. 45, No. 7, 1989, pp. 530-535.20. Jata, K.V. and E.A. Starke, Jr., “Fatigue Crack Growth and Fracture Toughness Behaviourof an Al-Li-Cu Alloy”, Metall. Trans. A, Vol. 17A, 1986, pp. 1011-1026.21. Haigh, B.P., .1. Inst. Metals, Vol. 18, 1917, pp. 55.22. Gouch, H.J. and D.G. Sopwith, “Behaviour of Single Crystal of Al under AlternatingTorsional Stresses While Immersed in Slow Stream of Tap-Water”, J. Inst. Metals, Vol. 49,1932, pp. 93.23. Crooker, T.W. and B.N. Leis, Eds, Corrosion Fatigue, ASTMSTP 801, 1983.24. Sudarshan, T.S. and M.R. Louthan, Jr., “Gaseous EnvironmentEffects on Fatigue Behaviourof Metals”, Intl. Mater. Reviews, Vol. 32, No. 3, 1987, pp. 121-151.25. Lynch, S.P., “Mechanisms of Fatigue and Environmentally Assisted Fatigue”, in FatigueMechanisms, ASTM STP 675, J.T. Fong, Ed, 1979, pp. 174-213.26. Thompson, N, N.J. Wadsworth and N. Louat, “The Origin of Fatigue Fracture in Copper”,Philos. Mag., Vol. 1, 1956, pp. 113-126.27. Duquette, D.J., “Mechanisms of Crack Initiation and Propagation in Corrosion Fatigue”, inMechanisms ofEnvironment Sensitive Cracking ofMaterials, P.R. Swann et al., Eds., TheMetals Society, London, 1977, pp. 305-321.28. Latanision, R.M. and A.R.C. Westwood, “Surface- and Environment- Sensitive MechanicalBehaviour”,Adv. Corros. Sci. Technol., Vol. 1, 1970, pp. 5 1-145.29. Shen, H., S.E. Podloseck and I.R. Kramer, “Effect of Vacuum on the Fatigue Life ofAluminium”, Acta metall., Vol. 14, 1966, pp. 341-346.References 16830. Laird, C. and D.J. Duquette, “Mechanisms of Fatigue Crack Nucleation”, in CorrosionFatigue, O.F. Devereux etaL, Eds., NACE, Houston, TX, 1972, pp. 88-117.31. Chen, C. Q.G. Cai and R.Z. Wang et al., Eds, Engineering Fracture Mechanics, Vol. I,National Defence Industry Publishing Company, China, 1978.32. Broek, D., Elementary Engineering Fracture Mechanics, 4th edition, Kiuwer AcademicPublishers, Dordrecht, Netherlands, 1986.33. Suresh, S. and R.O. Ritchie, “Propagation of Short Fatigue Cracks”, Intl. Metals Reviews,Vol. 29, No. 6, 1984, pp. 445-476.34. Miller, K.J. and E.R. de Los Rios, Eds., The Behaviour of Short Fatigue Cracks, EGF1,MEP, Bury St. Edmunds, UK, 1986.35. Suresh, S., Fatigue ofMaterials, Cambridge University Press, 1991.36. Ritchie, R.O., “Near-Threshold Fatigue-Crack Propagation in Steels”,Intl. Metals Reviews,No. 5&6, 1979, pp. 205-230.37. Bucci, R.J., “Development of a Proposed ASTM Standard Test Method for Near-ThresholdFatigue Crack Growth Rate Measurement”, in Fatigue Crack Growth Measurement andData Analysis, ASTM STP 738, S.J. Hudak, Jr. and R.J. Bucci, Eds., 1981, pp. 5-28.38. Taylor, D. Fatigue Thresholds, Butterworths, London, 1989.39. Davidson, D. and S. Suresh, Eds, Fatigue Crack Growth Threshold Concepts, TMS-AIIvIE,Warrendale, PA, 1984.40. Parris, P.C. and F. Erdogan, “A Critical Analysis of Crack Propagation Laws”, J. Basic Eng.(Trans. ASME. D), Vol. 85, 1963, pp. 528-534.41. Fong, C. and D. Tromans, “High Frequency Stage I Corrosion Fatigue of Austenitic StainlessSteel (316L)”, Metall. Trans. A, Vol. 19A, 1988,.pp. 2753-2764.42. Fong, C and D. Tromans, “Stage I Corrosion Fatigue Crack Crystallography in AusteniticStainless Steel (316L), Metall. Trans. A, Vol. l9A, 1988, pp. 2765-2773.43. Laird, C., “The Influence of Metallurgical Structure on the Mechanism of Fatigue CrackPropagation”, in Fatigue Crack Propagation, ASTMSTP 415, 1967, pp. 13 1-168.44. McEvily, A.J. and R.P. Wei, “Fracture Mechanics and Corrosion Fatigue”, in CorrosionFatigue: Chemistry, Mechanics andMicrostructure, O.F. Devereux, A.J. McEvily and R.W.Staehle, Eds., NACE-2, 1972, pp. 381-395.45. Wei, R.P. and J.D. Landes, “Correlation between Sustained-Load and Fatigue Crack Growthin High Strength Steels”, in Mater. Research and Standards, Vol.9, No.7, 1969, pp. 25-2g.References 16946. Wei, R.P., ‘On Understanding Environment-Enhanced Fatigue Crack Growth--AFundamental Approach”, in Fatigue Mechanism, AS7M STP 675, J.T. Fong, Ed., 1979, pp.816-840.47. Austen, I.M. and P. Mclnture, “Corrosion-Fatigue of High Stingth Steel in Low PressureHydrogen Gas”, Metal Science, Vol. 13, 1979, pp. 420-428.48. Rhodes, D., J.K. Musuva and J.C. Radon, “The Significance of Stress Corrosion Crackingin Corrosion Fatigue Crack Growth Studies”, Eng. Fract. Mech., Vol. 15, 1981, pp. 407-419.49. Duquette, D.J., “Mechanisms ofCorrosion Fatigue ofAluminium Alloys” ,AGARD-CP-316,1981, pp. 11-111.50. Duquette, D.J., “Corrosion Fatigue”, in Corrosion Mechanisms, F. Mansfeld, ed., MarcelDekker Inc., New York, 1987, pp. 367-395.51. Ford, F.P., “Current Understanding of the Mechanism of Stress Corrosion and CorrosionFatigue”, in Environment-sensitive Fracture: Evaluation and Comparison ofTestMethods,ASTM STP 821, S.W. Dean, E.N. Pugh and G.M. Ugiansky, Eds., 1984, pp. 32-5 1.52. Kim, Y.H. and S.D. Manning, “A Superposition Model for Corrosion Fatigue CrackPropagation in Aluminium Alloys”, in Fracture Mechanics: 14th Symposium--Volume I:Theoiy andAnalysis, ASTMSTP 791, J.C. Lewis and G. Sines, Eds., 1983, pp. 446-462.53. Trethewey, K.R. and 3. Chamberlain, Corrosion, Longman Scientific & Technical, 1988,pp. 184. /54. Duquette, D.J. and M. Gell, “The Effect ofEnvironment on the Mechanism of Stage I FatigueFracture”, Metall. Trans., Vol. 2, 1971, pp. 1325-1331.55. Lynch, S.P., “Environmentally Assisted Cracking: Overview of Evidence for anAdsorption-Induced Localized-Slip Process”, Acta metall., Vol. 36, 1988, pp. 2639-266 1.56. Suresh, S and R.O. Ritchie, “Near-Threshold Fatigue Crack Propagation: A Perspective onthe Role of Crack Closure”, in Fatigue Crack Growth Threshold Concepts, TMS-AIIvIE,Warrendale, PA, 1984, pp. 227-26 1.57. Elber, W., “Fatigue CrackClosure UnderCyclic Tension”, Engineering FractureMechanics,Vol. 2, 1970, pp. 37-45.58. Suresh, S., J. Toplosky and R.O. Ritchie, “Environmentally affected Near-Threshold FatigueCrack Growth in Steels”,in Fracture Mechanics: Fourteenth Symposium - Volume I: TheoryandAnalysis, ASTM STh 791, J.C. Lewis and G. Sines, Eds., 1983, pp. 329-347.59. Suresh, S. and R.O. Ritchie, “On the Influence ofenvironment on the Load Ratio Dependenceof Fatigue Threshold in Pressure Vessel Steel”, Engineering Fracture Mechanics, Vol. 18,1983, pp. 785-800.References 17060. Purushothaman, S. and J.K. Tien, in Strength ofMetals and Alloys, ICSMA5 Conf. Proc.,Vol. 2, P. Haasen et a!., Eds., Pergamon Press, New York, 1979, PP. 1267.61. Walker, N. and C.J. Beevers, Fat. Eng. Mat. Struct., Vol. 1, 1979, pp. 135.62. Jaske, C.E., J.H. Payer and V.S. Balint, Corrosion Fatigue ofMetals inMarineEnvironments,Columbus, Ohio, 1981, pp. 23.63. Wei, R.P. and M. Gao, “Hydrogen Embrittlement and Environmentally Assisted CrackGrowth”, in Hydrogen:.Effects on Material Behaviour, N.R. Moody and A.W. Thompson,Eds., The minerals, Metals & Materials Society, 1990, pp. 789-8 16.64. Realeaux, 0., “Scieron Alloys”, J. Inst, Met., Vol. 33, 1925, pp. 346.65. Spuliler, E.H., “Alcoa Alloy X-2020”, Alcoa Green Lett., 1958, pp. 156-158.66. Fridlyander, I.N., V.F. Shamray and N.y. Shiryayera, “Phase Composition and MechanicalProperties ofAluminium Alloys Containing Magnesium and Lithium” ,Russian Metallurgy,Vol. 2, 1965, pp. 83-90.67. Rioja, R.J. and R.H. Graham, “Al-Li Alloys Find Their Niche”, Adv. Mater. & Processes,Vol. 142, 6/92, pp. 23-26.68. Grimes, R., A.J. Cornish, W.S. Miller and M.A. Reynolds, “Aluminium-Lithium BasedAlloys for Aerospace Applications”, Metals and Materials, 6/85, pp. 357-363.69. Technology Forcast’86, Metal Progress, 1/86, pp. 75-79.70. Rao, K.T.V. and R.O. Ritchie, “Mechanical Properties of Al-Li Alloys Part 1: FractureToughness and Microstructure”, Material Science and Technology, Vol. 5, 1989, pp.882-895.71. Sanders, Jr., T.H., and E.A. Starke, Jr., Eds., Aluminium-LithiumAlloys, Proc. of the 1st IntlAluminium-Lithium Conference, Stone Mountain, Georgia, May 1980, TMS-AIME,Warrendale, PA, 1981.72. Starke, Jr., E.A., T.H. Sanders, Jr., Eds., Aluminium-LithiumAlloys II, Proc. of the 2nd Intl.Aluminium-Lithium Conference, Monterey, California, April 1983, TMS-AIME,Warrendale, PA, 1984.73. Baker, C., P.J. Gregson, S.J. Harris, and C.J. Peel, Eds.,Aluminium-LithiumAlloys Ill, Proc.of the 3rd Intl. Aluminium-Lithium Conference, Oxford, England, July 1985, The Instituteof Metals, London, England, 1986.74. Champier, G., B. Dubost, D. Miannay and L. Sabetay, Eds., 4th InternationalAluminium-Lithium Conference, 10-12 June 1987, Paris, France, Journal de Physique,Colloque C3, Supplement au No. 9, Vol. 48, 1987.References 17175. Sanders, Jr., T.H. and E.A. Starke, Jr., Eds., Aluminium-Lithium Alloys, Vol. 1-ifi,Proceedings ofthe 5th Intl Aluminium-Lithium Conference, Williamsburg, Virginia, March27-3 1, 1989, Materials and Component Eng. Pub. Ltd., UK, 1989.76. Peters, M. and P.3. Winkler, Eds., Aluminium-Lithium, Vol. I-il, Proceedings of the 6th IntlAluminium-Lithium Conference, Garmishch - Partenicircher, Germany, DGM, 1992.77. Noble, B., S.J. Harris, K. Dinsdale, “The Elastic Modulus of Aluminium-Lithium AlloyS’,J. Mater. Sci., Vol. 17, 1982, pp. 461-468.78. Gregson, P.J., H.M. Flower, C.J. Peel and B. Evans, “The Microstructure and Toughnessof Aluminium-Lithium Alloys”, in The Metallurgy of Light Alloys, Spring ResidentialConference, No. 20, March 1983, The Institute of Metallurgists, England, 1983, PP. 57-64.79. Flower, H.M. and P.J. Gregson, “Critical Assessment Solid State Phase Transformations inAluminium Alloys Containing Lithium”, Mater. Sci. & Technol., Vol. 3, 1987, pp. 8 1-90.80. Vasudevan, A.K., A.C. Miller, M.M. Kersker, “Contribution of Na-Segregation to FractureBehaviour of Al-ll.4 At.% Li Alloys”, see ref. 72, pp. 18 1-199.81. Webster, D., “Aluminium-Lithium Alloys”, Metal Progress, 4/84, pp. 33-37.82. Webster, D., “The Effect of Low Melting Point Impurities on the Properties ofAluminium-Lithium Alloys”, Metall. Trans. A, Vol. 18A, 1987, Pp. 218 1-2193.83. Kobayashi, T., M. Ninomi and K. Degawa, “Temperature Dependence of Impact Toughnessin Al-U-Cu-Mg-Zr Alloys”, Mater. Sci. & Technol., Vol. 5, 1989, pp. 1013-1019.84. Wert, J.A. and J.B. Lumsden, “Intergranular Fracture in an Al-Li-Cu-Mg-Zr Alloy”, ScriptaMetall., Vol. 19, 1985, pp. 205-209.85. Gregson, P.3. and H.M. Flower, “Microstructural Control of Toughness inAluminium-Lithium Alloys”,Acta metall., Vol. 33, No. 3, 1985, pp. 527-537.86. Sanders, Jr., T.H., E.A. Ludwiczak and R.R. Sawtell, “The Fracture Behaviour ofRecrystallized Al-2.8%Li-0.3%Mn Sheet”, Mater. Sci. & Eng., Vol. 43, 1980, pp. 247-260.87. Harris, S.J., B. Noble and K. Dinsdale, “Effect of Composition and Heat Treatment onStrength and Fracture Characteristics of Al-Li-Mg Alloys”, see ref. 72, pp. 219-233.88. Vasudevan, A.K., E.A. Ludwiczak ,S.F. Baumann, R.D. Doherty and M.M. Kersker,“Fracture Behaviour in Al-U Alloys: Role of Grain Boundary 6”, Mater. Sci. & Eng., Vol.72, 1985, pp. L25-L30.89. Miller, W.S., M.P. Thomas, D.J. Liord and D. Creber, “Deformation and Fracture in Al-LiBase Alloys”, see ref. 73, pp. 584-594.90. Tosten, M.H., A.K. Vasudevan and P.R. Howell, “Grain Boundary Precipitation in Al-Li-Cualloys”, see ref. 73, pp. 490-495.References 17291. Dodd, B. and H. Kobayashi, “Dynamic Fracture of Binary Al-Li Alloy in Torsion”, Mater.Sci. & Technol., Vol. 6, 1990, PP. 447-452.92. Suresh, S., A.K. Vasudevan, M.H. Tosten and P.R. Howell, “Microscopic and MacroscopicAspects of Fracture in Lithium-Containing Aluminium alloys”, Acta metall., Vol. 35, No.1, 1987, pp. 25-46.93. Vaynbiat, Y.M., B.A. Kopeliovich and Y.G. Gol’der, “Subboundary Brittleness of AlloyAl-Mg-Li Caused by Horophile Impurity of Sodium”, Physics ofMetals andMetallography,Vol. 42, No. 5, 1976, pp. 105-112.94. Talbot, D.E.J and C.E. Ransley, “The Addition ofBismuth to Aluminium-Magnesium Alloysto Prevent Embrittlement by Sodium”, Metall. Trans. A, Vol. 8A, 1977, pp. 1149-1154.95. Zheng, Z.Q., M.G. Liu, X.Z. Liu and D.F. Yin, “Microstructure and Tensile Properties ofthe Bismuth-Containing 2091 Al-Li Alloy”, Scripta Metall., Vol. 30, 1994, pp. 3 1-36.96. Harris, SI, B. Noble and K. Dinsdale, “The Mechanical Behaviour of Aluminium AlloysContaining Lithium”, in The Metallurgy of Light Alloys, Spring Residential Conference,No. 20, March 1983, The Institute of Metallurgists, England, 1983, pp. 48-56.97. Sankaran, K.K., J.E. O’Neal and S.M.L. Sastry, “Effects of Second-Phase Dispersoids onDeformation Behaviour ofAl-Li Alloys”, Metall. Trans. A, Vol. 14A, 1983, pp. 2174-2178.98. Chen, R.T. and E.A. Starke, Jr., “Microstructure and Mechanical properties of MechanicallyAlloyed, Ingot Metallurgy and Powder Metallurgy Al-Li-Cu-Mg Alloys”, Mater. Sci. &Eng., Vol. 67, 1984, pp. 229-245.99. Suresh, S., A.K. Vasudevan, “Influence of Composition and Aging Treatment on FractureToughness of Lithium-Containing Aluminium Alloys”, see ref. 73, pp. 595-601.100. Vasudevan, A.K. and S. Suresh, “Microstructural Effects on Quasi-Static FractureMechanisms in Al-Li Alloys: the Role of Crack Geometry”, Mater. Sci. & Eng., Vol. 72,1985, pp. 37-49.101. Webster, D., “Temperature Dependence of Toughness in Various Aluminium-LithiumAlloys”, see ref. 73, Pp. 602-609.102. Dorward, R.C., “Cryogenic Toughness of Al-Cu-Li Alloy AA2090”, Scripta Metall., Vol.20, 1986, pp. 1379-1383.103. Glazer, J., S.L. Verzasconi, R.R. Sawtell and J.W. Morris, Jr., “Mechanical Behaviour ofAluminium-Lithium Alloys atCryogenic Temperatures”, Metall. Trans. A, Vol. 1 8A, 1987,pp. 1695-1701.104. Glazer, J., S.L. Verzasconi, E.N.C. Dalder, W. Yu, R.A. Emigh, R.O. Ritchie and J.W.Morris, Jr., “Cryogenic Mechanical Properties of Al-Li-Cu-Zr Alloy 2090”, Adv. Cryo.Eng. Mater., Vol. 32, 1986, pp. 397-404.References 173105. Jata, K.V. and E.A. Starke, Jr., “Fracture Toughness of Al-Li-X Alloys at Ambient andCryogenic Temperatures”, Scripta Metal!., Vol. 22, 1988, PP. 1553-1556.106. Rao, K.T.V., H.F. Hayashigatani, W. Yu and R.O. Ritchie, “On the Fracture Toughnessof Aluminium-Lithium Alloy 2090-T8E41 at Ambient and Cryogenic Temperatures”,Scripta Metal!., Vol. 22, 1988, Pp. 93-98.107. Rao, K.T.V., W. Yu and R.O. Ritchie, “Cryogenic Toughness of CommercialAluminium-Lithium Alloys: Role of Delamination Toughening”, Metal!. Trans. A, Vol.20A, March 1989, pp. 485-497.108. Rao, K.T.V. and R.O. Ritchie, “Fracture-Toughness Behaviour of 2090-T83Aluminium-Lithium Alloy Sheet at Ambient and Cryogenic Temperatures”, ScriptaMetal!., Vol. 23, 1989, pp. 1129-1 134.109. Rao, K.T.V. and R.O. Ritchie, “Mechanisms Influencing the CryogenicFracture-Toughness Behaviour of Aluminium-Lithium Alloys”, Acta metal!. Mater., Vol.38, No. 11, 1990, pp. 2309-2326.110. Chu, D. and J.W. Morris, Jr., “The Effect of Through-Thickness Anisotropy on theCryogenic Mechanical Properties ofan Al-Cu-Li-ZrAlloy (Vintage ifi 2090-T8 1)”, Metall.Trans. A, Vol. 22A, 1991, pp. 1789-1799.111. Peel, C.J. and D.S. McDarmaid, “The Present Status of the Development and Applicationof Aluminu-Lithium Alloys 8090 and 8091”, Aerospace, Vol. 16, May 1989, pp. 18-23.112. Lynch, S.P., “Fracture of 8090 Al-Li Plate I. Short Transverse Fracture Toughness”, Mater.Sci. & Eng., Vol. A136, 1991, pp. 25-43.113. Blankenship, Jr., C.P. and E.A. Starke, Jr., “Mechanical Behaviour of Double-AgedAA8090”, Metall. Trans. A, Vol. 24A, 1993, PP. 833-841.114. Slavik, D.C., C.P. Blankenship, Jr., E.A. Starke, Jr. and R.P. Gangloff, “Intrinsic FatigueCrack Growth Rates for Al-Li-Cu-Mg Alloys in Vacuum”, Metal!. Trans. A, Vol. 24A,1993, pp. 1807-18 17.115. Ambat, R. and E.S. Dwarakadasa, “The Influence of pH on the Corrosion of MediumStrength Aerospace Alloys 8090,2091 and 2014”, Corrosion Science, Vol. 33, No.5,1992,pp. 681-690.116. Niskanen, P., T.H. Sanders, Jr., J.G. Pinker and M. Marek, “Corrosion of AluminiumAlloys Containing Lithium”, Corrosion Science, Vol. 22, No. 44, 1982, pp. 283-304.117. Colvin, E.L., G.L. Cahen, Jr., G.E. Stoner andE.A. Starke, “Effect ofGermanium Additionson the Corrosion Behaviour of an AILi Alloy”, Corrosion, Vol. 42, No. 7, 1986, pp.416-421.References 174118. Buchheit, R.G., J.P. Moran and G.E. Stoner, “Electrochemical Behaviour of the T1(A12CuLi) Intennetallic Compound and Its Role in Localized Corrosion of A1-2% Li-3%Cu Alloys”, Corrosion, Vol. 50, No. 2, 1994, pp. 120-130.119. Buchheit, Jr., R.G., J.P. Moran and G.E. Stoner, “Localized Corrosion Behaviour of Alloy2090 - The Role of Microstructural Heterogeneity”, Corrosion, Vol. 46, No. 8, 1990, pp.610-617.120. Moran, J.P., E.A. Starke, Jr., G.E. Stoner and G.L. Cahen, Jr., “ The Influence ofComposition and Microstructure on the Corrosion Behaviour of Two A1LiX Alloys”,Corrosion, Vol. 43, No. 6, 1987, pp. 374-381.121. Hu, Z.Q., Y. Zhang, Y.L. Liu and Z.Y. Zhu, “Corrosion Behaviour of 8090 Al-Li Alloy”,Corrosion, Vol. 49, No. 6, 1993, pp. 491-498.122. Kumai, C., J. Kusinski, G. Thomas and T.M. Devine, “Influence of Aging at 200 Con theCorrosion Resistance of Al-Li and Al-Li-Cu Alloys”, Corrosion, Vol. 45, No.4, 1989, pp.294-302.123. Baumert, B.A. and R.E. Ricker, “Effect of Heat Treatment on Corrosion Resistance ofAl-Mg-Li Alloy”, see ref. 73, pp. 282-286.124. Gui, 3. and T.M. Devine, “Influence of Lithium on the Corrosion of Aluminium”, ScriptaMetall., Vol. 21, 1987, pp. 853-857.125. Fernandes, J.C.S. and M.G.S. Ferreira, “Effect of Carbonate and Lithium Ions on theCorrosion Performance of Pure Aluminium”, ElectrochimicaActa, Vol. 37, No. 14, 1992,pp. 2659-266 1.126. Speidel, M.O., “Current Understanding of Stress Corrosion Crack Growth in AluminiumAlloys”, in The Theory of Stress Corrosion Cracking in Alloys, J.C. Scully, Ed., NATOScientific Affair Division, Brussels, 1971, pp. 289-341.127. Speidel, M.O., “Stress Corrosion Cracking of Aluminium Alloys”, Metall. Trans. A, Vol.6A, 1975, pp. 63 1-651.128. Braun, R. and H. Buhl, “Corrosion Behaviour of Al-Li-Cu-Mg Alloy 8090-T651”, see ref.74, pp. 843-849.129. Lumsden, J.B. and A.T. Allen, “The Stress Corrosion Cracking Behaviour of AILi Alloy8090”, Corrosion, Vol. 44, No. 8, 1988, pp. 527-532.130. Dorward, R.C. and K.R. Hasse, “Stress Corrosion Characteristics of A1CuLi AA2090Alloy”, Corrosion, Vol. 44, No. 12, 1988, pp. 932-941.131. Christodoulou, L., L. Struble and J.R. Pickens, “Stress Corrosion Cracking in Al-Li BinaryAlloys”, see ref. 72, pp. 561-579.References 175132. Lane, P.L., J.A. Gray and C.J.E. Smith, “Comparison of Corrosion Behaviour ofLithium-Containing Aluminium Alloys and Conventional Aerospace Alloys”, see ref.73, pp. 273-281.133. Gray, A., “Factors Influencing the Environmental Behaviour of Aluminium-LithiumAlloys”, see ref. 74, pp. 89 1-904.134. Rinker, J.G., M. Marek and T.H. Sanders, Jr., “Microstructure, Toughness and StressCorrosion Cracking Behaviour of Aluminium Alloy 2020”, Mater. Sci. & Eng., Vol. 64,1984, pp. 203-221.135. Meletis, E.I., “Stress Corrosion Cracking Properties of 2090 Al-Li Alloy”, in CorrosionCracking, V.S. Goel, Ed., American Society for Metals, 1985, pp. 3 15-326.136. Vasudevan, A.K., P.R. Ziman, S.C. Jha and T.H. Sanders, Jr., “Stress Corrosion Resistanceof Al-Cu-Li-Zr Alloys”, see ref. 73, pp. 303-309.137. Balasubramaniam, R., D.J. Duquette and K. Rajan, “On Stress Corrosion Cracking inAluminium-Lithium Alloys”, Acta metal!, mater., Vol. 39, No. 11, 1991, pp. 2597-2605.138. Balasubramaniam, R., D.J. Duquette and K. Rajan, “Hydride Formation in an Al-Li-CuAlloy”, Acta metall. mater., Vol. 39, No. 11, 1991, pp. 2607-2613139. Dorward, R.C., “On the Mechanical Properties and Stress Corrosion Resistance of TernaryAl-Cu-Li and Quaternary Al-Li-Cu-Mg Alloys”, Mater. Sci. & Eng., Vol. 84, 1986, pp.89-95.140. Sheppard, T., N.C. Parson, “Corrosion Resistance ofAl-Li Alloys” ,Mater. Sci. & Technol.,Vol. 3, 1987, pp. 345-352.141. Pizzo, P.P., R.P. Galvin and H.G. Nelson, “Utilizing Various Test Methods to Study theStress Corrosion Behaviour of Al-Li-Cu Alloys”, in Environment-Sensitive Fracture:Evaluation and Comparison ofTest Methods, ASTMSTP 821, S.W. Dean, E.N. Pugh andG.M. Ugiansky, Eds., 1984, pp. 173-201.142. Pizzo, P.P., R.P. Galvin and H.G. Nelson, “Stress-Corrosion Behaviour ofAluminium-Lithium Alloys in Aqueous Salt Environments”, see ref. 72, pp. 627-655.143. Ahmad, M., “Correlation between Aging Heat Treatments, Microstructure and StressCorrosion Properties of Al-Li-Cu-Mg Alloys”, see ref. 73, pp. 871-879.144. Vasudevan, A.K., J. Liu and R.E. Ricker, “Mechanism of Stress Corrosion Crack GrowthResistance of Al-Li-Cu Alloys: Role of Grain Boundary Precipitates”, in EnvironmentalDegradation ofEngineering Materials III, M.R. Louthan, Jr., R.P. McNitt and R.D. Sisson,Jr., Eds., Pennsylvania State University, 1987, pp. 321-327.145. Lee, S., S. Pyun and Y. Chun, “A Critical Evaluation of the Stress-Corrosion CrackingMechanism in High-Strength Aluminium Alloys”, Metal!. Trans. A, Vol. 22A, 1991, pp.2407-2414.References 176146. Holroyd, N.J.H., A. Gray, G.M. Scamans and R. Hermann, “Environment-SensitiveFracture of Al-Li-Cu-Mg Alloys”, see ref. 73, pp. 3 10-320.147. Craig, J.G., R.C. Newman, M.R. Jarrett and N.J.H. Hoiroyd, “Stress-Corrosion Crackingand Pre-Exposure Effects in Al-Li-Cu-Mg Alloy 8090”, in EnvironmentalDegradation ofEngineering Materials III, M.R. Louthan, Jr., R.P. McNitt and R.D. Sisson, Jr., Eds.,Pennsylvania State University, 1987, pp. 3 13-320.148. Craig, J.G., R.C. Newman, M.R. Jarrett and N.J.H. Hoiroyd, “Local Chemistry ofStress-Corrosion Cracking in Al-Li-Cu-Mg Alloys”, see ref. 73, pp. 825-833.149. Gangloff, R.P., Ed., Embrittlement by the Localized Crack Environment, TMS-AIME,Warrendale, PA, 1984.150. Moran, J.P., and G.E. Stoner, “Solution Chemistry Effects on the Stress Corrosion CrackingBehaviour of Alloy 2090 (Al-Li-Cu) and Alloy 2024 (Al-Cu-Mg)”, see ref. 75, pp.1187-1196.151. Moran, J.P., R.G. Buchheit and G.E. Stoner, “Mechanisms ofSCC ofAlloy 2090 (Al-Li-Cu)- A Comparison of Interpretations from Static and Slow-Strian-Rate Techniques”, inParkins Symposium on Fundamental Aspects of Stress Corrosion Cracking, S.M.Bruemmer, et al., Eds., The Mineral, Metals & Materials Society, 1992, pp. 159-170.152. Buchheit, Jr., R.G., J.P. Moran, F.D. Wail and G.E. Stoner, “Rapid Anodic DissolutionBased SCC of an Al-Li-Cu Alloy by Isolated Pit Solutions”, ibid, pp. 141-158.153. Buis, A. and J. Schijve, “Stress Corrosion Cracking Behaviour of AlLi 2090-T83 inArtificial Seawater”, Corrosion, Vol. 48, No. 11, 1992, pp. 898-909.154. Meltis, E.I., “Microstructural Effects on the Environment-Assisted Fracture Mechanismof Al-Li Alloys”, in Parkins Symposium on Fundamental Aspects of Stress CorrosionCracking, S.M. Bruemmer, et a!., Eds., The Mineral, Metals & Materials Society, 1992,pp. 353-370.155. Meltis, E.I. and W. Huang, “The Role of T1 Phase in the Pre-Exposure and HydrogenEmbrittlement of Al-Li-Cu Alloys”, Mater. Sci. & Eng., Vol. A148, 1991, pp. 197-209.156. Kim, S.S., E.W. Lee and K.S. Shin, “Effect of Cathodic Hydrogen Charging on TensileProperties of 2090 Al-Li Alloy”, Scripta Metall., Vol. 22, 1988, pp. 1831-1834.157. Shin, K.S., N.J. Austin and S.S. Kim, “Effect of Hydrogen on Mechanical Behaviour of a2090 Al-Li Alloy”, in Hydrogen Effects on Material Behaviour, N.R. Moody and A.W.Thompson, Eds., The Metals & Materials Society, 1990, pp. 1023-1031.158. Coyne, Jr., E.J., T.H. Sanders, Jr., and E.A. Starke, Jr., “The Effect of Microstructure andMoisture on the Low Cycle Fatigue and Fatigue Crack Propagation of Two Al-Li-XAlloys”, see ref. 71, pp. 293-305.References 177159. Bretz, P.E., L.N. Mueller and A.K. Vasudevan, “Fatigue Properties of 2020-T651Aluminium Alloy”, see ref. 72, PP. 543-559.160. Donahue, R.J., H. McI. Clark, P. Atanmo, R. Kumble and A.J. McEvily, “Crack OpeningDisplacement and the Rate of Fatigue Crack Growth”, Intl. J. Frac. Mech., Vol. 8, 1972,pp. 209-219.161. Vasudevan, A.K., P.E. Bretz and A.C. Miller, “Fatigue Crack Growth Behaviour ofAluminium Alloy 2020 (Al-Cu-Li-Mn-Cd)”, Mater. Sd. & Eng., Vol. 64, 1984, pp.113-122.162. Suresh, S. and R.O. Ritchie, “A Geomethc Model for Fatigue Crack Closure Induced byFracture Surface Roughness”, Metall. Trans. A, Vol. 13A, 1982, pp. 1627-163 1.163. Suresh, S., “Crack Deflection: Implications for the Growth of Long and Short FatigueCracks”, Metall. Trans. A, Vol. 14A, 1983, pp. 2375-2385.164. Suresh, S, “Fatigue Crack Deflection and Fracture Surface Contact: MicromechanicalModels”, Metall. Trans. A, Vol. 16A, 1985, pp. 249-260.165. Ruch, W. and E.A. Starke, Jr., “Fatigue Crack Propagation in Mechanically AlloyedAl-Li-Mg Alloys”, see ref. 73, pp. 121-130.166. Rao, K.T.V. and R.O. Ritchie, “Fatigue Crack Propagation and Cryogenic FractureToughness in PowderMetallurgy Aluminium-Lithium Alloys”,Metal!. Trans. A, Vol. 22A,1991, pp. 191-201.167. Rao, K.T.V., W. Yu and R.O. Ritchie, “Fatigue Crack Propagation in Aluminium-LithiumAlloy 2090: Partil. Small CrackBehaviour” , Metall. Trans. A, Vol.1 9A, 1988,pp.5639.168. Rao, K.T.V., W. Yu and R.O. Ritchie, “On the Growth of Small Fatigue Cracks inAluminium-Lithium Alloy 2090”, Scripta Metall., Vol. 20, 1986, pp. 1459-1464.169. Yoder, G.R., P.S. Pao, M.A. Imam and L.A. Cooley, “Unusual Fracture Mode in the Fatigueof an Al-Li Alloy”, in Advances in Fracture Research, K. Salama, et al., Eds., ICF7,Pergamon Press, 1989, pp. 919-927.170. Lespinasse, C and C. Bathias, “Fatigue Crack Growth of 8090 Alloy under Overloading”,see ref. 73, pp. 793-799.171. Rao, K.T.V. and R.O. Ritchie, “Mechanisms for the Retardation of Fatigue CracksFollowing Single Tensile overloads: Behaviour in Aluminium-Lithium Alloys”, Actametall., Vol. 36, 1988, pp. 2849-2862.172. Yu, W. and R.O. Ritchie, “Fatigue Crack Propagation in 2090 Aluminium-Lithium Alloy:Effect of Compression Overload Cycles”, J. Eng. Mater. & Technol., Vol. 109, 1987, pp.81-85.References 178173. Xu, Y.B., L. Wang, Y. Zhang, Z.G. Wang and Q.Z. Hu, “Fatigue Fracture Behaviour ofan Aluminium-Lithium Alloy 8090-T6 at Ambient and Cryogenic Temperature”, Metal!.Trans. A, Vol. 22A, 1991, PP. 723-729.174. Haddleton, F.L., S. Murphy and T.J. Griffm, “Fatigue and Corrosion Fatigue of 8090Al-Li-Cu-Mg Alloy”, see ref. 74, pp. 809-8 15.175. Srivatson, T.S., E.I. Meletis, C.P. Dervenis, E.J. Coyne, Jr. “Environmental Effects on TheFatigue Behaviour of Lithium-Containing Aluminium Alloys”, in EnvironmentalDegradation ofEngineeringMaterials III, M.R. Louthan, Jr., R.P. McNitt and R.D. Sisson,Jr., Eds., Pennsylvania State University, 1987, pp. 543-552.176. Magnin, T. and M. Rebiere, “The Effect of Hydrogen During Stress Corrosion Crackingand Corrosion Fatigue ofAl-Li-Cu Alloys in 3.5 % NaCl Solution”, see ref. 74, pp. 835-841.177. Dervenis, C.P., E.I. Meletis and R.F. Hochman, “Corrosion Fatigue in Al-Li Alloy 2090”,Mater. Sci. & Eng. A, Vol. 102, 1988, pp. 15 1-160.178. Rebiere, M. and T. Magnin, “Corrosion Fatigue Mechanisms of an 8090 Al-Li-Cu Alloys”,Mater. Sci. & Eng., Vol. A128, 1990, pp. 99-106.179. Chen, G.S. and D.J. Duquette, “The effect of Aging on the Hydrogen-Assisted FatigueCracking of a Precipitation-Hardened Al-Li-Zr Alloy”, Metal!. Trans. A, Vol. 23A, 1992,pp. 1551-1562.180. Chen, G.S. and D.J. Duquette, “Corrosion Fatigue of a Precipitation-Hardened Al-Li-ZrAlloy in a 0.5 M Sodium Chloride Solution”, Metal!. Trans. A, Vol. 23A, 1992, pp.1563-1571.181. Piascik, R.S. and R.P. Gangloff, “Environmental Fatigue of an Al-Li-Cu Alloy: Part II.Microscopic Hydrogen Cracking Processes”, Metall. Trans. A, Vol. 24A, 1993, Pp.2751-2762.182. Cayless, R.B.C., “Alloy and Temper Designation Systems for Aluminium and AluminiumAlloys”, in Metals Handbook, 10th edition, Vol. 2, 1990, pp. 15-28.183. “Heat Treating of Aluminium Alloys”, inASM Handbook, Vol. 4, 1991, pp. 841-879.184. Morrison, J., Defence Research Establishment Pacific, Esquimalt, B.C., Canada (1992).185. Bray, J.W., “Aluminium Mill and Engineered Wrought Products”, in Metals Handbook,10th edition, Vol. 2, 1990, pp. 30-6 1.186. Metallography and Nondestructive Testing, Annua1ASTM Standards, 1980, pp. 49.187. Metals Handbook, 1958, Pp. 284.188. ASTM B 399-78a, Plane Strain Fracture Toughness Testing of Metallic Materials, AnnualASTM Standards, 1980, pp. 580-601.References 179189. Tada, H., P.C. Paris and G.R. Irwin, The Strengthening ofCracks Handbook, Del ResearchCorp., Hellertown, PA, 1975, pp. 210-211.190. ASTM B 647-78T, Constant-Load-Amplitude Fatigue Crack Growth Rates Above 108rn/cycle, Annual ASTM Standards, 1980, pp. 749-767.191. Singbeil, D. and D. Tromans, “Caustic Stress Corrosion Cracking of Mild Steel”, Metall.Trans. A, Vol. 13A, 1982, pp. 1091-1098.192. Hardie, D., N.J.H. Holroyd and R.N. Parkins, “Reduced Ductility of High-StrengthAluminium Alloy during or after Exposure to Water”, Metal Science, Vol. 13, 1979, pp.603-610.193. Dickson, J.I., Fractography: A Tool in Failure Analysis and Fracture Research, ÉcolePolytechnique, Montreal, PQ, 1987.194. Williams, D.B., “Practical Electron Microscopy in Materials Science: Electron EnergyLoss Spectrometry in the Analytical Electron Microscope”, Norelco Reporter, Vol. 30,1983, pp. 16-41.195. Williams, D.B. and J.W. Edington, “Microanalysis of Al-Li Alloys Containing Fine W(A13Li) Precipitates”, Phil. Mag., Vol. 30, pp. 1147-1153, 1974.196. Pourbaix, M., Atlas d’Equilibre Electrochemiques, Gauthier-Villars, Paris, 1963.197. Vasudevan, A.K. and R.D. Doherty, “Grain Boundary Ductile Fracture in PrecipitationHardened Aluminium Alloys”, Acta metall. Vol. 35, 1987, pp. 1193-1219.198. Mukhopadhyay, A.K., D.S. Zhou and Q.B. Yang, “Effect of Variation in the Cu:Mg Ratioon the Formation ofT2and C Phases in AA 8090 Alloys”, Scripta Metall. Mater., Vol. 26,1992, pp. 237-242.199. Geng, H. and R. Li, “Effect of Deformation and Recrystallization on Stress CorrosionCracking Susceptibility of 2090 Al-Li alloy”, Scripta Metall. Mater., Vol. 31, 1994, pp.1431-1436.200. Yasuda, M., F. Weinberg and D. Tromans, “Pitting Corrosion of Al and Al-Cu SingleCrystals”, J. Electrochem. Soc., Vol. 137, 1990, pp. 3708-3715.201. Yasuda, M., F. Weinberg and D. Tromans, “Pitting Corrosion of Al and Al-Cu Bicrystals”,J. Electrochem. Soc., Vol. 137, 1990, pp. 37 16-3723.202. Kamat, S.V. and N. E. Prasad, “Fatigue Crack Growth Behaviour of an 8090 Al-Li Alloyunder Mixed Mode I - Mode ifi Loading Conditions”, Scripta Metall. Mater., Vol. 29,1993, pp. 137 1-1376.203. Tromans, D. and B. Mogambedze, private communication, University ofBritish Columbia,1994.References 180204. Guest, R.J. and A.R. Troiano, “Stress Corrosion and Hydrogen Embrittlement in anAluminium Alloy”, Corrosion, Vol. 30, 1974, PP. 274-279.205. Shewman, P.G., Diffusion in Solids, McGraw Hill, New York, 1963, PP. 13-14.206. Ricker, R.E. and D.J. Duquette, “The role ofHydrogen in Corrosion Fatigue ofHigh PurityAl-Zn-Mg Exposed to Water Vapor”, Metal!. Trans. A, Vol. 19A, 1988, pp. 1775-1783.207. Iyer, R.N., H.W. Pickering and M. Zamanzadeh, “Analysis of Hydrogen Evolution andEntry into Metals for the Discharge- Recombination Process” J. Electrochem.. Soc., Vol.136, 1989, pp. 2463-2470.208. Tromans, D., “On the Thermodynamics of Hydride Formation in Al-Li Alloys at 25 °C”,Scripta Metal!. Mater., Vol. 17, 1992, pp. 217-222.209. Webster, D., “Aluminium-Lithium Alloys-- the Next Generation”, Adv. Mater. &Processes, 5/94, pp. 18-24.210. Pitcher, P.D., R.J. Stewart and S. Gupta, “A Study of Reversion Behaviour in 8090 AlloyUsing Small Angle Neutron Scattering and Transmission Electron Microscopy” ScriptaMetal!. Mater., Vol. 26, 1992, pp. 511-516.Appendix 1819 Appendix9.1 Appendix I. Increased stiffness of Al-Li alloysThe stiffness of a simple beam of rectangular cross section under 3-point bending load isgiven by the formula,(Al)where E is the elastic modulus, b is the beam thickness, h is the beam height, and 1 is thesupporting distance. Consider a beam of a conventional Al-alloy (of elastic modulus E1 anddensity p1) and let it be replaced with a beam of an Al-Li alloy (ofelastic modulusE and densityPu) of equal weight. Further, allow the beam height to change while maintaining b and I at theiroriginal sizes. The stiffness (5) of the conventional beam relative to the stiffness (Sn) of theAl-Li beam is then given by,S1 E..(h1j3—= (A2)S E1(h)3With equal weight, p, h1 = p h, therefore,S11 E11(p)3A3S1 E1(p)3For each weight percent lithium added to an aluminium alloy, the density is reducedapproximately 3% and the elastic modulus is increased approximately 6%, i.e., ps/p, 0.97,1.06. Substitution into Equation (A3) leads to S1/S, 1.16, that is, the stiffness can beincreased 16 percent by one percent lithium addition. Alternatively, the stiffness may be keptAppendix 182constant by changing the beam thickness (b), then the weight (W1) of the conventional Al-alloybeam is changed to W for the Al-Li alloy beam, whereW pb p1EA4p1bE (Consequently, each weight percent lithium addition can lead to -8 percent weight saving.9.2 Appendix II. Identification ofgrain boundary precipitate by selected areaelectron diffractionThe grain boundary precipitates in the as-received Al-Li alloy and the re-aged alloy wereinvestigated by selected area electron diffraction during the TEM studies. Figure Al shows thelimited part of a selected area diffraction pattern obtained from a grain boundary precipitate inthe as-received materials, 8090-T877 1. Using the known cameralength, the pattern was indexedand possible plane spacings were calculated. Table Al lists the comparison between thecalculated spacing (d) and angle values with known values of a possible phase from the X-rayPowder Diffraction File It is seen that it is a close match to theLi3CuAI5phase which hasa body-centred cubic structure.Usually, the larger grain boundary particles are referred to in the literature[79,92,118,121,198,199] asT2-phase. Only one article describes T2 asLi3CuA15(hereafterdesignated as (T2)1, most [79,92,118,121,198,199] describe T2 as AI6CuLi3 (hereafterdesignated (T2). Unfortunately, very little crystallographic data is available for (T2); it is“Powder Dffracrion File, International Centre for Diffraction Data, PA, 1991.Glazer, J. and J.W. Morris, Jr., “Thermomechanical Processing of Two-Phase Al-Cu-Li-Zr Alloy”, see ref.74, pp. 191-198.FigureAlSelectedareadiffractionpatternobtainedfromagrainboundaryprecipitate.(a)TEMmicrograph;(b)schematicdiagramshowingthepatternof interest andtheindexing.I(222)2R3(400)2R(22i(222)(400)(222)(a)(b)00Appendix 184reported to have anicosahedral structure, but is not indexed in the latestX-ray PowderDiffractionFile. Some authors [197] reported that the morphology and d-spacings of (T2)are not markedlydifferent from those of the cubic 6 (AILi) phase, which has a well established crystal structure.However, several grain boundary precipitates were found to have a “five-fold” symmetry, asshown in Figure A2. This symmetry is impossible for 6 (A1Li), but such patterns have beenreported for (T2) [198,199]. Therefore, on this basis it is believed that grain boundaryprecipitates of (T2)phase were present.Table Al Comparison of Calculated Spacing and Angle Values with Known Values of aPossible PhaseCalculated and Measured Values Known Values forLi3CuA152Rd = 2L2e =40 nun.A2R1 = 10.0 mm —* d1 = 4.00 d222 = 4.062R=ll.5mm—*d3.46 d2=3.502=10.Omm—*d4.00 d=4.065504224OO5470L JR1,R3 — 22Z2ZZIt is noticed that the concentrations of Cu in the grain boundary precipitates detected bythe STEM + EDX analyses (-7 wt.% in the as-received state and -.16 Wt.% in the re-aged state,FigureA2Selectedareadiffractionpatternobtainedfromagrainboundaryprecipitate,showingfive-foldsymmetry.00Appendix 186as shown in Figure 5.43) are all much lower than the concentrations of Cu inLi3CuA15(-.29wt.%) and in A1CuLi3 (-.26 wt.%). This may be due to the contribution from the Al matrixenclosed in the sampling region of the analysis.9.3 Appendix III. EELS analysesEELS analysis is useful for detecting light elements, such as lithium, by examining theLi K-excitation edge. However, when attempting to detect Li in Al-alloys, complications arisefrom the overlapping of the Li K-edge at .55 eV with the high background tails of the very welldefined Al plasmon loss peaks at 15, 30 and 45 eV, making the analysis particularly difficult[194]. Consistently, no excitation edge was observed in the EELS analyses conducted in thepresent study.Another attempt to conduct compositional microanalysis was made by examining theplasmon peak energy shift from the energy loss spectrum. A schematic diagram of the electronenergy loss spectrum is shown in Figure A3. The average quantum loss E = (1 +E2)canbeen determined from the spectrum, where E1 and E2 are the measured energies of the first andsecond plasmon loss peaks. This energy loss is related to the concentration of the alloyingelement. Williams and Edington [195] reported a relation between the Li atomic concentration(Ca) in binary Al-Li alloys and the plasmon energy loss (E in eV) as, E = 15.3 - 4.0 C, andused the equation to measure the composition profile across a grain boundary. With regard toZeroLossIEnergyLoss1stPlasmonLossPeakE12ndPlasmonLossPeakE201530eVFigureA3Schematicdiagramof anelectronenergylossspectrum.I-’00Appendix 188the other major alloying elements in Al-Li alloy 8090 (i.e. Cu and Mg), similar equations werefound, E = 15.3 - 10 C for Cu and E = 15.3 - 4.4 C for Mg ‘. It is seen that these threealloying elements shift the plasmon peak in the same direction.The E value for pure Al is 15.3 eV and was used to calibrate the equipment. For example,the system gave E values for 99.9995% purity Al that varied between 17.1 eV and 17.7 eVduring repeated measurements at the same location on the Al specimen. Hence, it was notpossible to say with certainty whether a system measurement of 17.1 eV or 17.7 eV correspondedto a E value of 15.3 eV. Therefore, when searching for E shifts due to alloying effects, onlymeasurements outside the 17.1-17.7 eV limits were likely to be meaningful. In practice, plasmonenergy losses measured from the as-received and re-aged Al-Li alloys ranged between 17.1 eVand 16.1 eV, indicating a definite decrease in E of uncertain value. These shifts were in theright direction according to the above equations for Li, Cu and Mg concentrations, but were notuseful for quantitative purposes. Point-by-point measurements along a straight line across agrain boundary were conducted many times. However, no sensible and reproduciblecompositional profile could be drawn from the data obtained. The scattering and the poorreproducibility of the data may be due to several reasons:(i) Large systematic error associated with determining E from the measurements of E1and E2 on the obtained spectra, which were plotted on chart paper. The systematic error wasestimated to be —0.2 eV, and based on equation E = 15.3 - 4.0 C, this means that a change ofLi concentration of less than 5 at.% will not be detectable.v Doig, P. and J.W. Edington, “Low-Tempemture Diffusion in AI-7 wt.% Mg and A1-4 wL% Cu Alloys”, Phil.Mag., Vol. 28, pp. 961-970, 1973.Hibbert, G., J.W. Edington, D.B. Williams and P. Doig, “The Variation of Plasma Energy Loss withComposition in Dilute Aluminium-Magnesium Solid Solutions”, Phil. Mag., Vol. 26, pp. 1491-1494, 1972.Appendix 189(ii) Drift of the specimen during the data collecting time period. It was found quite oftenthat the point of analysis had moved from the intended position during the collection of thespectrum, leading to poor spatial resolution.(iii) The resolution achievable with the ordinary tungsten filament gun equipped with thescanning transmission electron microscope is limited. To obtain a resolution of —0.1 eV (suchas those indicated in the reported work [l95,V]), a field emission gun is required.9.4 Appendix IV. Threshold stress stateThe generafly accepted formulae for the average crack tip plastic zone size along the crackfront under Mode I monotonic loading are,plane strain conditions: 1 (K ‘2r=—I— I (AS)6it Gy)plane stress conditions: 1 (K 2 (A6)2itAccordingly, the corresponding formulae for the average cyclic plastic zone size under ModeI cyclic loading are,plane strain conditions: 1 (x ‘ (A7)plane stress conditions: 1 (x ‘‘— I (A8)21c2a)Appendix 190Consequently, if the onset of fatigue crack propagation at AKth is a slip dominated processthat is associated with a specific size of reversed plastic zone (i.e. a specific amount of slipreversibility), then Equation (A7) and Equation (A8) indicate that for a constant Lry at threshold,F (AK,h)planestrain 1 (A9)L (AKth)pj,jes] —The experimental data for fatigue tests at the high frequency in the A1C13 solution, theother aqueous environments (AQ), and desiccated air (DA) show that,F (AKth)AQ 12 F (AKth)DA 12 r 1 .7MPa .m”2L (/-S-Kth)a1c3j [(Mth)Alcl3j — L 1 .O5MPa .m112 — (A10)Consequently, the reasonably good correlation between Equation (A9) and (AlO) is consistentwith the conclusion that the lower AKth in the AICJ3 is due to the development of a plane stresssituation at the crack tip that is caused by significant S-L splitting effects.9.5 Suggestions for further workBased on the present study of the S-L splitting effects on the T-L orientation fatigue crackgrowth and of the effects of the re-aging treatment on the S-L orientation fatigue crackpropagation, it will be intriguing to investigate the effects of the re-aging treatment on the T-Lorientation corrosion fatigue crack propagation behaviour of Al-Li alloy 8090-T8771 plate.Such a study will determine whether there is any influence of the re-aging treatment on theoccurrence of the S-L splitting phenomenon and the consequent effects on the T-L orientationfatigue crack growth.


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items