Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The transition from stress corrosion cracking to corrosion fatigue in AA-7075 and AA-8090 Rechberger, Johann 1990

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

Item Metadata

Download

Media
831-UBC_1990_A1 R42.pdf [ 16MB ]
Metadata
JSON: 831-1.0078570.json
JSON-LD: 831-1.0078570-ld.json
RDF/XML (Pretty): 831-1.0078570-rdf.xml
RDF/JSON: 831-1.0078570-rdf.json
Turtle: 831-1.0078570-turtle.txt
N-Triples: 831-1.0078570-rdf-ntriples.txt
Original Record: 831-1.0078570-source.json
Full Text
831-1.0078570-fulltext.txt
Citation
831-1.0078570.ris

Full Text

THE TRANSITION FROM STRESS CORROSION CRACKING TO CORROSION FATIGUE IN AA-7075 AND AA-8090 By JOHANN RECHBERGER Dipl. Ing. ETH, Swiss Federal Institute of Technology, 1982 M.A.Sc, The University of British Columbia, 1986 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Metals and Materials Engineering) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 1989 © Johann Rechberger, 1989 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of M e t a l s and M a t e r i a l s F . n g i n p P r i n E The University of British Columbia Vancouver, Canada Date /s/C%Z2m4^ Sff? DE-6 (2/88) ii Abstract The effect of crack tip strain rate (CTSR) on environmentally assisted cracking was studied for alloys AA-7075 (Al-Zn-Mg-Cu) and AA-8090 (Al-Li-Cu-Mg) in the artificially aged condition. Fatigue pre-cracked double cantilever beam (DCB) specimen were employed with the crack plane parallel to the rolling plane. The cracking behaviour under monotonic and cyclic loading conditions was investigated in aqueous sodium chloride solutions with and without additions of sodium chromate as a corrosion mhibitor. CTSR values were described in terms of K-rate AK/At (ie. dK/dt) as a measured average over the loading period of a fatigue cycle. This allowed a comparison with CTSR's of rnonotonically increasing load or constant load tests. At frequencies < 1 Hz, the load was applied with a triangular wave form. A high frequency of 30 Hz was obtained by sinusoidal loading. Expressed as K-rate, CTSR values were varied over 7 orders of magnitude from 10"5 MPaVm/s to 102 MPaVm/s. Stress intensities investigated were mainly around region II values with respect to SCC K-log(da/dt) behaviour. At low K-rates, real time crack velocities (da/dt) measured under monotonic slow loading or constant load conditions were comparable to crack velocities obtained with cyclic loading experiments. As the K-rate was increased from low values, typical of constant load experiments, the real time crack velocities decreased. This was caused by plasticity induced crack growth retardation effects and a decrease in crack tip film rupture events during the unloading part of a cycle. The crack propagation rate decreased until minimal crack advance increments per cycle were dictated by mechanical parameters iii acting on a hydrogen embrittled crack tip region. Under monotonic loading conditions region II crack velocities were not influenced by an increase in K-rate which was explained with a mass transport controlled cracking process. Tests with alloy 7075 at intermediate K-rates and a high R-ratio of 0.78 allowed a crack tunnelling mechanism to operate. This overcame the plasticity induced crack growth retardation and, therefore, cracks propagated at the same rates as during low K-rate tests where no retardation phenomena were encountered. Scanning electron microscope investigations revealed a striated intergranular fracture surface of alloy 7075 if tested at K-rates above the transition value to K-rate independent crack propagation rates. Individual striations could be matched on opposing fracture surfaces and the striation spacing corresponded to the average crack propagation increment per cycle. The striations, therefore, were formed as part of the crack advance during every fatigue cycle. At the lower K-rates no striations were present but micro tear ridges could be found on the intergranular fracture facets indicating that dissolution processes alone did not cause the intergranular crack advance. Alloy 8090 did not reveal significant changes in fractography over the entire K-rate range investigated, except at the highest K-rates where small interlocking steps could be detected on some opposing transgranular fracture surfaces. In general, however, the crack path at all K-rates was mainly intergranular with dimpled fracture facets. Alloy 8090 exhibited a high resistance to SCC with fatigue pre-cracked DCB specimen. Therefore, to obtain crack velocity values with low K-rate monotonic loading tests very long test durations would have been necessary. iv It is concluded that the transition from intergranular SCC to intergranular CF occurs at a critical K-rate. Below the critical K-rate crack velocities are not increased by cyclic loading. Instead crack growth retardation effects can result in lower real time crack velocities than those typical for constant load tests at comparable stress intensities but much lower K-rates. Table of Contents Abstract ii List of Tables , vii List of Figures viii Abbreviations xii Symbols xiii Acknowledgements xiv 1 Introduction 1 2 Objective 5 3 Literature Review 6 3.1 High Strength Al-Zn-Mg Alloys 6 3.1.1 SCC - Behaviour 7 3.1.2 CF-Behaviour 10 3.2 High Strength Al-Li alloys 13 3.2.1 SCC - Behaviour 14 3.2.2 CF - Behaviour 16 3.3 Effects of Cyclic Frequency and Strain Rate 17 3.4 Inhibitors for Environment Assisted Cracking 21 3.5 SCC - Fractography of High Strength Al-Alloys 22 3.6 CF-Fractography of High Strength Al-Alloys 23 3.7 The Superposition Model 27 3.8 The Crack Tip Strain Rate Parameter 29 4 Experimental 36 ^.1 Material 36 4.1.1 Chemical Composition 36 4.1.2 Mechanical Properties 37 4.1.3 Microstructure 39 4.2 Specimen Design 41 4.3 Specimen Precracking 43 4.4 Specimen Compliance 44 4.5 Monotonic Slow Loading Experiments (Rising-K Experiments) 47 4.6 Bolt Loading and Constant Load Experiments 50 4.7 Cyclic Loading Experiments 51 4.8 Test Environments, Cell Arrangement 57 4.9 Scanning Electron Microscopy 59 5 Results 61 5.1 Monotonic Loading Experiments (Alloy AA-7075) 61 5.1.1 Slow Loading (Rising-K) Experiments (AA-7075) 61 5.1.2 Bolt Loading and Constant Load Experiments (AA-7075) 76 5.2 Monotonic Loading Experiments (Alloy AA-8090) 80 5.2.1 Slow Loading (Rising-K) Experiments (AA-8090) 80 5.2.2 Bolt Loading and Constant Load Experiments (AA-8090) 81 5.3 Cyclic Loading Experiments (AA-7075 and AA-8090) 82 5.3.1 Effect of Cyclic K-Rate (Alloy 7075) 85 vi 5.3.2 Effect of Cyclic K-Rate (Alloy 8090) 94 5.4 Electrochemical Potential Measurements 97 5.5 Fractography Alloy-7075 103 5.5.1 Intergranular Fracture Regions 106 5.5.1.1 Low K-rate Intergranular Fractography 107 5.5.1.2 High K-rate Intergranular Fractography 110 5.5.1.3 Effect of R-Ratio and 1 ^ on the Striated Fracture Path 113 5.5.2 Fractography of Transgranular Steps 116 5.5.2.1 Transgranular steps at Low Cyclic K-rates 116 5.5.2.2 Transgranular Steps at High Cyclic K-rates 116 5.5.3 Fractography of Regions With Larger Precipitates 120 5.5.4 The Crack Front Profile, Tunnelling 121 5.5.5 Fractography of Tests in Dry Air 124 5.6 Fractography Alloy-8090 125 5.6.1 Intergranular Fracture Regions 126 5.6.2 Fractography of Transgranular Steps 128 6 Discussion 130 6.1 Cyclic Loading Alloy 7075, R=0.78 130 6.1.1 Cycling at the Lowest Rates 131 6.1.2 Interaction Between Crack Velocity and the Plastic Zone at Low Cyclic K-rates 133 6.1.3 Crack Tunnelling at Intermediate Loading Rates 138 6.1.4 Loading Rates (K-rates) Producing Crack Retardation 139 6.1.5 Cycling at the Highest Loading Rates 139 6.1.6 Summary of Cyclic Loading Alloy 7075, R=0.78 142 6.2 Cyclic Loading Alloy 7075, R=0.31 143 6.3 Cyclic Loading Alloy 8090, R=0.75 145 6.4 Monotonic Loading Experiments 146 6.4.1 Slow Rising Load Tests 146 6.4.2 Bolt Loading and Constant Load Tests 149 6.5 Fractography of AA-7075 152 6.5.1 Effect of Larger Precipitates 153 6.5.2 Low K-rates 153 6.5.3 High K-rates 154 6.6 Fractography of AA-8090 155 7 Summary and Conclusions 157 8 References 160 Appendix I -. 170 Appendix II 171 Appendix HI 194 Appendix IV 195 List of Tables vii Table Page 4.1 Chemical composition of AA-7075 and AA-8090 36 4.2 Mechanical properties of AA-7075 and AA-8090 38 4.3 Fatigue precracking of AA-7075 43 4.4 Fatigue precracking of AA-8090 44 4.5 Compliance polynomials for the DCB specimens 46 5.1 Results from slow monotonic loading tests AA-7075, 3.5wt% NaCl 63 5.2 Crosshead displacement rate and loading rate 64 5.3 Results from bolt loading tests AA-7075,3.5wt% NaCl 76 5.4 Results from constant load tests AA-7075, 3.5wt% NaCl 80 5.5 Results from slow monotonic loading tests AA-8090, 3.5wt% NaCl 81 5.6 Results from bolt loaded and constant load tests AA-8090, 3.5wt% NaCl 82 5.7 Results from cyclic loading at different K-rates AA-7075, R=0.78, 3.5wt% NaCl 92 5.8 Results from cyclic loading at different K-rates AA-7075, R=0.78, 3.5wt% NaCl + 0.01M Na2Cr04 92 5.9 Results from cyclic loading at different K-rates AA-7075, R=0.78, dry air 93 5.10 Results from cyclic loading at different K-rates AA-7075, R=0.31, 3.5wt% NaCl and dry air 93 5.11 Results from cyclic loading at different K-rates AA-8090, R=0.75, 3.5wt% NaCl and dry air 94 List of Figures Figure Page 1.1 Typical SCC K r v plot 2 3.1 Typical CF logAK-log(da/dN) plot 11 4.1 Grain structure of AA-7075 and AA-8090 40 4.2 DCB specimen geometry for AA-7075 42 4.3 DCB specimen geometry for AA-8090 42 4.4 Cyclic loading test arrangement 53 4.5 The triangular load wave form 54 4.6 The sinusoidal load wave form 55 4.7 The positive sawtooth load wave form 56 4.8 The square load wave form 56 4.9 Test cell arrangement for the Hounsfield tensometer 58 4.10 Test cell arrangement for the Sonntag fatigue machine 58 4.11 Sample sectioning for SEM crack tip observations 60 5.1 Typical plot of crack length versus time for a slow monotonic loading test (Specimen A13, K-rate = 1.3xl0'5 MPaVm/s) 62 5.2 Rising-K experiments at a K-rate of ~2.5xl0'6 MPaVrn/s 66 5.3 Rising-K experiments at a K-rate of ~8.5xl0"6 MPaVrn/s 67 5.4 Rising-K experiments at a K-rate of ~1.3xl0"5 MPaVrn/s 68 5.5 Rising-K experiments at a K-rate of ~2.6xl0"5 MPaVm/s 69 5.6 Rising-K experiments at a K-rate of ~3.7xl0'5 MPaVm/s 70 5.7 Rising-K experiments at a K-rate of ~4.5xl0"5 MPaVm/s 71 5.8 Rising-K experiments at a K-rate of ~6.8xl0"5 MPaVm/s 72 5.9 Rising-K experiments at a K-rate of ~ 1.1 x 10"4 MPaVrn/s 73 5.10 Rising-K experiments at a K-rate of ~1.4xl0"4 MPaVm/s 74 5.11 Crack velocities obtained under different monotonic slow loading conditions 75 5.12 Crack velocities obtained under bolt loading conditions with different amounts of crack advance 78 5.13 Crack velocities obtained under slow loading and bolt loaded conditions 79 5.14 CF tests at 30 Hz with a sinusoidal load wave form, plot of log (AK)-log(da/dN) 83 ix 5.15 Comparison of CF and SCC crack velocities for different values 84 5.16 Real time crack velocities at different cyclic K-rates for R=0.31 87 5.17 Cyclic crack propagation rates at different cyclic K-rates for R=0.31 88 5.18 Real time crack velocities at different cyclic K-rates for R=0.78 89 5.19 Cyclic crack propagation rates at different cyclic K-rates for R=0.78 90 5.20 Crack velocities for different K-rates obtained under monotonic and cyclic loading 91 5.21 Real time crack velocities at different cyclic K-rates for R=0.75 95 5.22 Cyclic crack propagation rates at different cyclic K-rates for R=0.75 96 5.23 Anodic polarization curves for AA-7075 and AA-8090 in 3.5wt% NaCl with and without chromate corrosion inhibitor 100 5.24 Corrosion potential fluctuations at a K-rate of 10"3 MPaVrn/s, AA-7075, 3.5wt%NaCl 101 5.25 Corrosion potential fluctuations at a K-rate of 5\10A MPaVm/s 101 5.26 Corrosion potential fluctuations at a K-rate of 10"4 MPaVm/s 101 5.27 Corrosion potential fluctuations at a K-rate of 5X10"4 MPaVrn/s, a) beginning of experiment, b) after several hundred hours 102 5.28 Typical SEM micrograph displaying the four major types of fractography: A, B, C and D AA-7075, Specimen A76, Mag.: 400x, MCPD bottom to top 104 5.29 Transition from CF to overload failure AA-7075, Specimen A76, Mag.: 150x, MCPD bottom to top 105 5.30 Transition from fatigue pre-crack produced in air to a CF crack produced in 3.5wt% NaCl AA-7075, Specimen A76, Mag.: 150x, MCPD bottom to top 106 5.31 Front view of intergranular crack path with transgranular steps AA-7075, Specimen A72, Mag.: 450x 107 5.32 SEM micrograph of rippled grain boundary with small subgrains, polished and etched surface normal to T direction AA-7075, Mag.: lOOOx 108 5.33 Faint subgrain boundaries within larger grains, polished and etched surface parallel to the rolling plane AA-7075, Mag.: 440x 109 ' 5.34 SEM micrograph of matching micro tear ridges on opposing intergranular fracture surfaces AA-7075, Specimen A74, Mag.: 13000x, R=0.78, K-rate=7.8xl03 MPaVm/s 110 5.35 SEM micrograph of intergranular striations AA-7075, Specimen A76, Mag.: 3000x, R=0.78, K-rate=2.2xl02 MPaVm/s, MCPD bottom to top I l l X 5.36 SEM micrograph of opposing fracture surfaces with interlocking ridges (arrowed) AA-7075, Specimen A76, Mag.: 6800x, R=0.78, K-rate=2.2xl02 MPaVm/s, MCPD bottom to top 112 5.37 SEM micrograph of opposing crack tip fracture surface with striations produced during overload AA-7075, Specimen A76, Mag.: 5300x, R=0.78, K-rate=2.2xl02 MPaVm/s, MCPD bottom to top 112 5.38 SEM micrograph of opposing intergranular surface with fatigue striations AA-7075, Specimen A89, Mag.: 8000x, R=0.31, K-rate=5.9xl02 MPaVm/s, MCPD bottom to top 114 5.39 Intergranular striations AA-7075, Specimen A89, Mag.: 4000x, R=0.31, K ^ l 1 MPaVm, K-rate=5.9xl02 MPaVm/s, MCPD bottom to top 115 5.40 Intergranular striations AA-7075, Specimen A86, Mag.: lOOOOx, R=0.78, ^ = 1 1 MPaVm, K-rate=l.6xl02 MPaVm/s, MCPD bottom to top 115 5.41 SEM micrograph (low Mag.) of transgranular step AA-7075, Specimen A74, Mag.: 3100x, R=0.78, K-rate=7.8xl0'3 MPaVm/s, MCPD bottom to top 117 5.42 SEM micrograph (high Mag. of Fig. 5.41) of opposing fracture surfaces on a transgranular step with matching dimples and precipitates (arrowed), Mag. 19000x 117 5.43 SEM micrograph of ridges and faint lines (arrowed) on transgranular steps AA-7075, Specimen A73, Mag.: 2500x, R=0.78, K-rate=3.1xl0'1 MPaVm/s, MCPD bottom to top 118 5.44 SEM micrograph of a polished and etched plane normal to the T direction with faint subgrain boundaries AA-7075, Mag.: 1500x 119 5.45 SEM micrograph of opposing fracture surfaces on a transgranular step with fine striations (arrowed) AA-7075, Specimen A73, Mag.: 3500x, R=0.78, K-rate=3.1xl0"1 MPaVm/s, MCPD bottom to top 121 5.46 Stepped crack front with precipitates on a polished plane normal to the crack propagation direction AA-7075, Specimen A72, Mag.: 800x, R=0.78, K-rate=3.0xl0"1 MPaVm/s 122 5.47 SEM micrograph of the tunnelling crack front AA-7075, Specimen A82, Mag.: HOOx, R=0.78, K-rate=5.1xlO-4MPaVm/s 123 5.48 Schematic drawing of crack front tunnelling 124 xi 5.49 Intergranular, stepped crack path on a polished plane normal to the crack propagation direction AA-8090, Specimen Li-25, Mag.: 175x, R=0.75, K-rate=3.1xlO-3MPa>/rn7s 125 5.50 SEM micrograph of dimpled intergranular fracture surface AA-8090, Specimen Li-22, Mag.: 2000x, R=0.75, K-rate=1.9xl02 MPaVm/s, MCPD bottom to top 126 5.51 SEM micrograph of opposing fracture surfaces at a grain boundary triple point with precipitates on both fracture halves AA-8090, Specimen Li-22, Mag.: 9000x, R=0.75, K-rate=1.9xl02 MPaVm/s, MCPD bottom to top 127 5.52 SEM micrograph of pronounced slip steps at a grain boundary AA-8090, Specimen Li-22, Mag.: 8000x, R=0.75, K-rate=1.9xl02MPaVm/s 127 5.53 Grain boundary separation produced by tensile overload in air, slip lines are confined to one fracture surface whereas precipitates (arrowed) are located on the opposing surface AA-8090, Mag.: 3500x 128 5.54 SEM micrograph of transgranular steps which connect the intergranular fracture regions AA-8090, Specimen Li-26, Mag.: 450x, R=0.75, K-rate=1.9xl02 MPaVm/s, MCPD bottom to top 129 5.55 SEM micrograph (high Mag. of Fig.5.54) of interlocking steps on a transgranular fracture region Mag.: lOOOOx 129 6.1 Crack advance during a single load cycle experiment with specimen A93 132 6.2 Crack advance during a single load cycle experiment with specimen A91 133 6.3 Crack tip coordinates 134 6.4 Size of crack tip plastic zone for 1^=17.5 MPaVm and KIllill=13.5 MPaVm 135 6.5 Movement of the crack tip plastic zone with crack advance for specimen A93 136 6.6 Movement of the crack tip plastic zone with crack advance for specimen A91 136 6.7 Decrease of stress intensity due to crack propagation in bolt loaded specimen 151 6.8 K-rates obtained with bolt loaded and single cycle decreasing load tests 152 Abbreviations xii AA-7075: Aluminum Association designation for a Al-Zn-Mg-Cu alloy AA-8090: Aluminum Association designation for a Al-Li-Cu-Mg alloy CF: Corrosion fatigue COD: Crack opening displacement CTOD: Crack tip opening displacement CTSR: Crack tip strain rate DCB: Double cantilever beam AKu,: Threshold stress intensity range for CF K I S C C : Threshold stress intensity for SCC K-rate: Average change in stress intensity over time (dK/dt) LET: Linear elastic theory MCPD: Macroscopic crack propagation direction PFZ: Precipitate free zone R-ratio: Ratio between minimum and maximum cyclic stress intensity S, L, T: Directions with respect to rolling plane: S-short transverse, T-transverse, L-longitudinal SCC: Stress corrosion cracking ST, LT, SL: Crack propagation directions V S C E : Potential with respect to a saturated calomel electrode V S H E : Potential with respect to a standard hydrogen electrode Symbols xiii a: Crack length Aa: Crack advance C: Specimen compliance 8: Crack tip opening displacement de/dt: Strain rate dK/dt: K-rate E: Elastic modulus AK: Cyclic stress intensity range AKa,: Threshold stress intensity range for CF Kj: Opening mode crack tip stress intensity K i s c c : Threshold stress intensity for SCC K ^ : Maximum stress intensity during cyclic loading K^: Minimum stress intensity during cyclic loading v: Poissons ratio P: Load o"y: Yield strength t: time v: Real time crack velocity (da/dt) Acknowledgements xiv I would like to express my sincere gratitude to my supervisor, Dr. Desmond Tromans, for his friendly guidance and encouragement during this work. Thanks are also extended to other faculty and staff members, and to fellow graduate students for their generous help. My wife, Virginia, has given me a great amount of support and exhibited great patience throughout this project. She also deserves a special thanks for typing this thesis. Financial support has been provided by a Killam Predoctoral Fellowship for two years. Additional financial assistance was given by the Natural Science and Engineering Research Council and by the Department of Metals and Materials Engineering for Teaching Assistance. The author is very grateful for these contributions. 1 1 Int roduct ion Stress corrosion cracking (SCC) of metals is the time-dependent propagation of sub-critical cracks in the conjoint presence of a tensile load and a corrosive environment, where the environment is frequently of an innocuous nature. The phenomenon is particularly troublesome in sophisticated structures where high strength/weight ratios are critical elements in the overall design (eg. airframes), because severe load limitations have to be imposed in order to minimize the problem. Much effort has been applied, and continues to be expanded, in the development of more SCC-resistant alloys and in understanding the factors controlling crack propagation. Many studies have been conducted on SCC of high strength Al-alloys in aqueous solutions and humid atmospheres. The experimental techniques have ranged from simple load - failure time studies on smooth specimens [1] to loading rate effects (slow strain rates) on the cracking susceptibility of smooth specimens [2] and, more recently, to fracture mechanics test methods where crack propagation rates are studied as a function of the opening mode crack tip stress intensity factor (K,) [3]. Usually, crack propagation behaviour exhibits three regions as is increased. At lower Kj levels, above a threshold stress intensity for detectable crack propagation (K I S C C), the crack velocity (v) increases with increasing Kj (region I). This is followed by a K,-independent crack velocity plateau (region II). Finally, as the stress intensity for unstable crack propagation (KIC) is approached, Kj-dependent cracking reappears (region III). The K r v behaviour is summarized in Figure 1.1 where, following conventional practice, is plotted against log(v). The region II crack velocity appears to be independent of whether tests are 2 conducted under rising Kj conditions (constant load) or decreasing K x conditions (constant crack opening displacement). Figure 1.1 Typical SCC K r v plot Log (Crack Velocity) i Region III / Region II / j /Region 1 K ISCC K Stress Intensity C Although the fracture mechanics approach has proven to be a useful technique for studying SCC, it is recognized that stress intensity alone is not a sufficient parameter to describe SCC behaviour. Slow loading rate tests on smooth specimens in many metal/environment systems have shown that strain rate has a strong influence on SCC behaviour [4]. Usually, the occurrence of SCC in such tests is confined to a narrow range of strain rates. However, it is difficult to define the strain rate at a crack tip and there is no general agreement on such a definition in the literature. One possible approach, 3 proposed and utilized in the present study, is to recognize that the crack tip strain must be dependent on K x and that the rate of change of K x with time (dKj/dt) must influence the crack tip strain rate. The parameter dKi/dt (described simply as "dK/dt" or "K-rate" in this study) then becomes the important variable in SCC that combines the effect of both stress intensity and strain rate. There are several possible methods for examining the effect of dK/dt on crack propagation. One technique is to test specimens of different crack lengths at the same initial Kj. Propagation of the cracks will produce different dK/dt values. However, practical consideration suggest that the range of dK/dt values will be somewhat limited by this procedure. A second method is to load cracked specimens at different rates and monitor the resulting crack growth. Again, there are practical limitations. High loading rates will produce small periods of crack growth, with inevitably small increments of crack advance, and measurement of crack velocity will be difficult. A third method is to use repetitive cyclic loading of cracked specimens. A wide range of dK/dt values may be obtained by changing the loading frequency over several orders of magnitude. Sufficient time for crack advance to occur at high loading rates may be achieved by testing over many load cycles. A particular advantage of the cyclic loading approach to a study the effect of dK/dt on SCC is that it offers the possibility of investigating the relationship and transition between SCC and corrosion fatigue (CF). The phenomenon of CF involves the time-dependent sub-critical propagation of cracks under the simultaneous action of cyclic (fluctuating) loads and a corrosive environment. At extremely low frequencies the CF situation approaches SCC. Furthermore, it is well established that CF crack propagation rates obtained under high frequency loading conditions are several orders of magnitude higher than values obtained under monotonic loading conditions [5]. This suggests that 4 above some critical loading rate, region II SCC propagation rates can be no longer independent of dK/dt (ie. no longer independent of Kx). To date, although many separate studies have been reported on SCC and CF, linking of CF and SCC has been based mainly on theoretical grounds, because very few data have been available from experiments that represent the intermediate (transition) process. The present study was directed towards an investigation of the transition between SCC and CF in two high strength aluminum alloys, using the parameter dK/dt as the means of uniting the two phenomena. The alloys chosen were the well established Al-Zn-Mg-Cu alloy AA-7075 and the new generation Al-Li-Cu-Mg alloy 8090. The Li-containing alloy has lower density and higher specific modulus of elasticity than the conventional Al-alloys [6]. The environment assisted cracking behaviour of the Li-containing alloy is less well established than that of the AA-7000 series of alloys. 5 2 Objective The major objective of the study was to investigate the transition and relationship between CF and SCC of the high strength Al-alloys, AA-7075 and AA-8090, in aqueous environments. In particular, the parameter, dK/dt, was defined and employed that was compatible with both SCC and CF processes. It was used, together with fracture mechanics testing techniques conducted under widely different loading rates (including cyclic loading), to establish possible K-rate regimes for different cracking mechanisms. High magnification matching stereo fractography was used to aid interpretation of the cracking behaviour. A secondary objective was to compare the crack propagation behaviour in monotonic and single load cycle experiments with that observed under constant load and constant crack opening displacement conditions in order to define limits for an accelerated SCC test method. 6 3 Li terature Review 3.1 H igh Strength A l -Zn-Mg Alloys The high strength 7000 series aluminum alloys have been gradually introduced since 1940 [7,8]. However, despite their attractive tensile properties, they were not an immediate success, because of their unsatisfactory resistance to stress corrosion cracking (SCC) and low fracture toughness. Only extensive metallurgical research has led to the modern high strength aircraft alloys. Important factors were the decrease in iron and silicon content [9,10], in order to avoid brittle intermetallic compounds such as (Fe,Mn,Cu)ALj, Al7Cu2Fe or Mg2Si, and an overaging heat treatment to improve SCC resistance in the short transverse direction [11]. The 7000 series aluminum alloys belong to the aluminum-zinc-magnesium or the aluminum-zinc-magnesium-copper system. They are used primarily in the heat treated and aged tempers. Highest strength levels can be obtained in the T6 temper with tempering temperatures around 130° C. This produces GP zones and partly coherent MgZn2 precipitates on {111} A 1 planes. At longer aging times, or higher temperatures, MgZn2 becomes incoherent or is replaced by Mg 3Zn 3Al 2 and is often found at grain boundaries [12,13]. The addition of copper serves mainly for solid solution strengthening. Small amounts of chromium or zirconium help to retard recrystallization and grain growth during solution treatment, which leads to improved fracture toughness [14]. Newer developments on the 7000 series include rapid solidification and mechanical alloying procedures. These techniques decrease the size of microstructural elements like intermetallic compounds and increase the homogeneity of the alloy. However, these powder metallurgical alloys are more expensive. 7 3.1.1 SCC - Behaviour Crack propagation tests are usually performed in the most critical short transverse direction (SL) of plate material because SCC in aluminum alloys is almost exclusively intergranular. In this orientation, the crack plane is parallel to the rolling plane and propagation is in the rolling direction. (See Appendix I for the designation of testing orientations). A review on the relationship between composition, thermal treatment, microstructure and stress corrosion cracking has been given by Sprowls et al. [1]. Speidel [11,15] and Hyatt [3] conducted extensive stress corrosion studies on high strength 7000 series Al-alloys. In aqueous sodium chloride solutions, the reported region II crack velocities were in the range of 10"9 m/s up to 10'5 m/s, depending on alloy composition, heat treatment and environment. The peak aged 7075 alloy gave region U crack velocities of about 10"8 m/s in the stress intensity range of 13 to 20 MPaVni. The threshold stress intensity, Kjgcc, was found to be about 5 MPaVrn in aqueous chloride solutions. Most tests were performed with bolt loaded double cantilever beam (DCB) specimens. Also, several other researchers [16,17,18] obtained essentially the same results with the peak aged 7075 alloy. Most reported data have been produced under free corrosion conditions. Gest et al. [18] and Speidel [11] investigated the effect of applied electrochemical potential on SCC. Cracking rates did not increase by more than a factor of two at potentials more anodic than the free corrosion potential. Polarization of more than 100 mV in the cathodic direction, however, decreased cracking rates significantly. 8 Anodic dissolution of the grain boundary region was one of the first mechanisms proposed for SCC of the high strength aluminium alloys. Electrochemical investigations by Dix [19] showed that continuous grain boundary precipitates in naturally aged Al-Mg alloys (5000 series) were anodic with respect to the Al-Mg solid solution. Therefore, in the presence of an electrolyte, selective corrosion along the grain boundaries could occur. In Al-Cu alloys (2000 series) the Cu-depleted regions along the grain boundaries were anodic both with respect to the grain interior and with respect to the grain boundary precipitates. This also localized the corrosive attack to the grain boundary region. In SCC sensitive tempers of the Al-Zn-Mg-Cu alloys the grain boundary precipitates are hot attacked by selective corrosion and it is also suggested that the depletion of copper along the grain boundaries creates an anodic dissolution path [1]. Composition changes during aging procedures can change the electrochemical potential differences between precipitates, precipitate free zones and the matrix, and influence the severity of grain boundary attack in these alloys. Therefore, in some temper conditions 7000 series aluminum alloys are found immune to intergranular corrosion in the unstressed condition. Nevertheless, they still show a strong susceptibility towards SCC [11]. An electrochemical dissolution model alone may therefore, not be sufficient to describe the SCC susceptibility of these alloys. With a more stress oriented approach it is suggested that the local deformation pattern controls the SCC susceptibility [20]. Small coherent or partly coherent particles, as found in the peak aged condition, can be sheared by dislocations and promote planar slip. This results in dislocation pile ups and stress concentrations at the grain boundaries. Overaging causes precipitates to grow and become incoherent which prevents them from being sheared by dislocations. The result is less planar slip. Stress concentrations at grain boundaries due to planar slip can cause rupture of the protective oxide film and lead to localized corrosive attack by dissolution processes [21]. This SCC crack propagation 9 model is usually referred to as slip-dissolution model. Important parameters in the slip-dissolution process are the crack tip strain rate and the kinetics of the passive film formation [22]. If passivation rates are slow and strain rates high, crack tip blunting can occur. This is caused either by crack tip plastic deformation or due to strong dissolution not only at the crack tip but also on the crack sides and will eventually lead to crack arrest. High passivation rates on the other hand allow only very minimal crack advance per film rupture event. At high strain rates the cracking rate controlling reaction can change from film rupture/passivation rate control to liquid diffusion control because of the high demand for reactants at the crack tip and the removal of solvated species. This concept of a transition in rate determining steps as the crack tip strain rate is increased has been discussed by Ford [22]. Early studies by Gest et al. [18] and Scamans et al. [23] have shown that high strength aluminum alloys can be embrittled by cathodic charging experiments or pre-exposure to water vapour saturated air. This led to the conclusion that hydrogen embrittlement may also make an important contribution to SCC [24,25,26]. Different mechanisms such as adsorption or decohesion models have been formulated that relate hydrogen to the cracking process. They have been summarized and discussed by Louthan et al. [27] and Latanision et al. [28]. In the adsorption model hydrogen reduces the surface energy necessary to form a crack and thus lowers the fracture stress [29]. In the decohesion model hydrogen is absorbed into the metal and accumulates at specific sites like inclusions or regions of high triaxial stresses. The high accumulation of hydrogen then lowers the atom-atom bond strength [30] or recombination of hydrogen atoms causes high local gas pressures [31]. To date no clear consensus has emerged on the exact mechanisms of SCC in high strength aluminum alloys. 10 3.1.2 CF - Behaviour Crack propagation behaviour under CF conditions is usually reported in terms of the effect of the cyclic stress intensity, AK, on cracking, where AK = K ^ - K ^ . Crack rates are most frequently reported in terms of the average crack growth increment per load cycle, da/dN. Crack propagation rates are reported less frequently in terms of real time growth, da/dt. It is conventional practice to present CF data as a plot of logAK vs. log(da/dN), as shown in Figure 3.1. Such plots exhibit three stages of crack growth behaviour somewhat analogous to SCC, except that stage II is not a crack rate plateau independent of AK. In fact at intermediate AK ranges where stage II is observed, a linear relationship is often obtained between logAK and log(da/dN). Such observations suggest a stage II power law relationship of the form da/dN = C(AK)m, where C and m are experimental constants. This was first recognized by Paris et al. [32] and is frequently referred to as Paris' Law. Over the years many different expressions have been suggested to relate da/dN with AK. A long list of formulae describing crack propagation under cyclic loading conditions has been collected by Sudarshan et al. [33]. According to Speidel [34] and McEvily et al. [35], environment enhanced fatigue crack growth curves may deviate significantly from those obtained under inert conditions. This is particularly true if the alloy shows a substantial SCC sensitivity in the same environment. In most cases the largest effect of an aggressive environment is to increase fatigue crack rates in stages I and II. Figure 3.1 Typical CF log(AK )-log(da/dN) plot 11 Crack Propagation Rate Log (da/dN) Stage III Stage II / y / X / y / y I y (da/dN) = C(AK) m m 1 Stage I / Stress Intensity Range Log (AK) Extensive work on the 7000 series aluminum alloys has been done by Speidel [34]. He found exclusively transgranular cracking if the specimen were not tested in the short transverse direction with the loading direction perpendicular to the rolling plane. Aqueous solutions almost always showed acceleration of fatigue cracks as compared to an inert environment. The frequency dependence of cyclic cracking rates da/dN was very small except if CF tests were performed in the SL direction. In the LT direction, aqueous salt solutions gave on the average about 5 times faster growth rates as compared to air if tested at intermediate cyclic stress intensity ranges AK. The cracking rate again dropped by a factor of about 3 to 4 when tested in vacuum [36,37]. Similar results on 7075 alloys 12 have been obtained by many researchers [35,38,39,40,41,42,43]. In general, the real time cracking rates, da/dt, in salt solutions vary from about 10"7 to 10"6 m/s at intermediate AK ranges, if tested in the LT direction. Fatigue tests with the loading direction being normal to the rolling plane (SL and ST crack directions) were performed by Speidel [34], Holroyd et al. [5] and Khobaib et al. [38]. In these testing directions, the effect of environment is more pronounced and the cyclic threshold stress intensity, AK^, is lower. Further major differences compared to other testing directions lay in the intergranular crack path and the dramatic increase of crack propagation per cycle, da/dN, if the cyclic frequency is lowered. A second important variable in fatigue is the load-ratio or R-ratio, where R = K ^ i n / K ^ . An increase in the R-ratio leads usually to higher cyclic crack propagation rates at comparable AK levels [44] and to a decrease in cyclic threshold stress intensity [39]. The effect of R-ratio on da/dN is more pronounced in alloys with a higher sensitivity to SCC because R affects the time within each load cycle during which the stress intensity is above some threshold level where SCC can occur [35]. The influence of electrochemical potential on cyclic cracking rates, da/dN, seems to be very small in the 7075 alloy at potentials near the free corrosion potential and at more anodic values. [34,45,46]. At more cathodic potentials cyclic cracking rates have been reported to decrease [34,46]. However, this effect was not confirmed by Stoltz [45], who found no relation between da/dN and potential. 3.2 High Strength A l - L i alloys 13 Commercial Al-Li alloys were initially developed in Germany in 1924. In the 1950's, Alcoa recognized the potential of Li-containing alloys for aerospace applications because of their lower density and higher specific elastic modulus. [47,6]. In the mid seventies, aluminum companies worldwide intensified their investigations and Al-Li alloys were designed for substitution of the Aluminum Assn. 2000 and 7000 series. Targets were a density reduction of 10% and a stiffness increase of 10%. In recent years, four special conferences were organized which dealt exclusively with the various aspects of alurmnum-hthium alloys [48,49,50,51]. In the 1980's Alcan International introduced new generation Al-Li-Cu-Mg-Zr alloys with the trade names "Lital-A", "Lital-B" and "Lital-C". Lital-A is now registered with international agreement as 8090 alloy [52,53]. Other companies like Alcoa [54] and Pechiney [55] have also developed Al-Li alloys with similar composition and properties. The second phase microstructure in the precipitation hardened Al-Li-Cu-Mg-Zr alloys is rather complicated and has been the subject of intensive research [56,57,58,59,60]. Most of the rmcrostructural studies have been originally presented in one of the last four International Al-Li Conference Proceedings [48,49,50,51]. In the quarternary Al-Li-Cu-Mg alloys, coprecipitation of 5'(Al3Li), S'(Al2CuMg) and T^AljCuLi) phases occurs. At long aging times, the equilibrium phases S (Al2CuMg), T 2 (Al^CuLij) and Al 2MgLi appear preferentially at grain boundaries and a 5' precipitate-free zone (PFZ) can be observed at the grain boundaries. The 8 (AlLi) phase is rarely found in Mg containing alloys [58]. In zirconium-bearing alloys, additional intermetallic phases such as Al3Zr or Al3(Li,Zr) appear. Nucleation of the S' and Tj phase are enhanced by the presence of dislocations. Consequently, plastic deformation 14 prior to aging results in a finer distribution of these precipitates [61]. Noble et al. [62] found that the effect of lithium on the electronic structure gives rise to the increased elastic modulus. Hereby the main modulus increase comes from lithium in solid solution and to a lesser degree from the 8' phase. The relatively low ductility observed in many Al-Li alloys has been attributed to the grain boundary segregation of tramp elements [63,64,65], planar slip and shear bands [66,67,68,69,70,71], formation of PFZ's [72,73,74] and grain boundary precipitates [52,75]. Planar slip can occur because the coherent 8' precipitates are easily cut by dislocations. This is more pronounced in Al-Li alloys than, for example, in Al-Cu or Al-Zn-Mg alloys [66]. An increase in the volume fraction of 8' intensifies planar slip and decreases fracture toughness [67,68]. The reason is related to strain localization which nucleates voids and cracks at grain boundaries. Additions of dispersoid forming elements like Zirconium result in a more homogeneous deformation substructure and improved ductility [76]. Planar slip is also reduced if sufficient amounts of incoherent S and T1 phases are present [72,77]. Toughness and strength can also be increased by mechanical alloying techniques whereby fine carbides and oxides are embedded in the material [78,79]. 3.2.1 SCC - Behaviour The general corrosion behaviour of the Al-Li-Mg-Cu-Zr alloys in aqueous salt solutions is comparable to the conventional 2000 and 7000 series aluminum alloys. However, this is only true as long as the formation of the 8 (Al-Li) phase is suppressed [80]. For conventional (Li-free) high strength aluminum alloys it is well established that 15 overaging can substantially increase the resistance to SCC [1]. Overaging treatments of the Al-Li alloys, on the other hand, can result in 8 phase formation which has a detrimental effect on the general corrosion resistance. Rinkler et al. [81] studied the effect of heat treatment on the SCC behaviour of an AlLiCu alloy and found excellent SCC resistance in the peak aged condition. Region II plateau velocities were below 10"9 m/s. It was concluded that strain localization at grain boundaries due to planar slip does not necessarily promote SCC. It has been suggested that dissolution of grain boundary precipitates like T x and 8 is the determining factor for SCC [81,82]. Ahmad [83] and Vasud6van et al. [84] also investigated several aging conditions and found with the 8090 alloy that region II crack velocities ranged between 10"10 and 10"9 m/s for all tempers tested. Similar cracking rates are reported by Pizzo et al. [85] and Gray [86]. Christodoulou et al. [87] obtained region JJ cracking rates of 10"9 m/s with a binary Al-Li alloy in acidic sodium chloride solutions. Lumsden et al. [88] used the slow strain rate technique with alloy 8090. A high susceptibility to SCC was measured in various aging conditions. Anodic polarization even increased the susceptibility. Bavarian et al. [89] used smooth U-bend specimen to test several Li containing alloys. They observed intergranular attack with cracking rates being slower than in a 7075-T6 alloy. Craig et al. [90] did extensive work on the 8090 alloy in carbonate/bicarbonate solutions. These environments gave very low SCC resistance especially if polarized to the active to passive transition potential. In sodium chloride solutions, cracking of specimens was only observed under alternate immersion conditions. This effect was attributed to the development of a locally alkaline crack environment when only small volumes of electrolyte are present. Overall, it can be summarized that the Li-containing Al alloys seem to possess superior SCC resistance as compared to conventional high strength aluminum alloys. However, in very specific environmental or loading conditions the contrary is true. 16 3.2.2 CF - Behaviour Coyne et al. [91] found that an increase in the elastic modulus of Li-containing aluminum alloys improves the resistance to subcritical crack growth at low stress intensities. This can be explained via the crack opening displacement (COD), which is inversely proportional to the elastic modulus and influences the fatigue crack growth [92]. Vasudevan et al. [93] found that with an AlLiCu alloy only a small difference in fatigue crack growth rates could be detected between moist air (90% relative humidity) and dehumidified helium (3 ppm moisture). Also, an AlLiMn alloy tested in air (3 ppm moisture) and distilled water gave the same crack growth rates in both environments. It was suggested that due to the highly reactive nature of lithium with moisture, contents of less than 3 ppm already cause the damaging environmental interactions. However, the crack growth rates of the tested Al-Li alloys are still lower than crack growth rates measured with comparable Li-free aluminum alloys in the same environments [93]. Venkateswara et al. [94] studied cracking rates of short and long cracks in an AlLiCuZr alloy. They found that in air the fatigue crack growth is highly anisotropic and depends on the orientation of the fracture plane. This is caused by the different morphology of the crack path, depending on the grain orientation. Different grain orientations cause different magnitudes of crack tip shielding from crack deflection and result in roughness induced crack closure. Lowest crack closure effects were noticed with the loading direction normal to the rolling plane, which resulted in the fastest 17 cracking rates. In general, crack propagation rates were slower than with 7000 series alloys. This was explained with the highly deflected and branched crack path in the Li containing alloys. With short cracks the deflected crack path does not play a significant role and roughness induced crack closure effects are much less pronounced. Peters et al. [95] tested the alloy 8090 in NaCl solutions and found that it exhibited superior behaviour as compared to Li-free alloys. This was also attributed to crack closure effects. The superiority of the 8090 alloy was mainly noticeable at low R-ratios where crack closure effects play a more significant role. Closure was not attributed to thick oxide films but rather to the rough fracture surface and fretting debris deposits. However, the fact that threshold values (AK^ in an aggressive environment were higher than in air could support crack closure caused by corrosion products. Crack closure mechanisms caused by the highly deflected and branched crack path or increased corrosion products are supported by other researchers [96,97,98,99,100]. They also attribute this effect to the generally better CF behaviour of Li containing Al-alloys. 3.3 Effects of Cyclic Frequency and Strain Rate Holroyd et al. [5] investigated the interfacial region between CF and SCC of the 7017 T651 aluminum alloy. Their incentive was stirred by the non-applicability of the superposition model for certain types of alloys. (The superposition model is described in Chapter 3.7). Their tests were performed in natural seawater at the free corrosion potential. The frequency range of 70 Hz down to 0.1 Hz was investigated by applying a triangular load wave form to precracked compact tension specimen. Real time cracking 18 rates (da/dt) in stage II at 0.1 Hz were about an order of magnitude higher than cracking velocities under static conditions. The authors suggested that by further lowering the test frequency one gradually would approach cracking velocities of a static test. Intergranular CF at the lower frequencies was fractographically indistinguishable from intergranular SCC. At higher frequencies mainly striated ductile transgranular fracture was observed. Speidel [34] investigated frequency effects on the high strength aluminum alloy 7079 T651 which showed very high SCC propagation rates in a saturated aqueous NaCl solution. Sinusoidal load wave forms with the loading direction normal to the rolling plane were used at frequencies ranging from 10'3 to 10 Hz. The occurrence of an intergranular fracture path, together with the linear dependence of cyclic crack growth rate log(da/dN) with the inverse of frequency (log(f)), led him to the conclusion that CF was not taking place but, instead, "stress corrosion cracking under cyclic loads". Endo et al. [101] performed cyclic experiments on smooth bending specimens of AA 7075-T6 in aqueous sodium chloride solutions. They applied a trapezoidal stress wave form at frequencies ranging from 0.3 to 3X10"4 Hz. They found that the fracture appearance changed from striated transgranular fracture at the higher frequencies to intergranular SCC-like fracture at lower cycling rates. It was suggested that fatigue damage occurs at higher frequencies during the period of changing stress whereas at lower frequencies stress corrosion "damage" occurs during the time interval of maximum strain. When comparison was taken with other materials the trend was observed that a strong frequency dependence of CF was paralleled by a high susceptibility to SCC. Magnin et al. [102] studied the behaviour of weldable AA 7020 T4 alloys in the initial stages of CF and SCC. Smooth tensile specimen were tested at different monotonic and cyclic strain rates. A threshold strain rate was observed above which no SCC occurred whereas CF could be produced at all investigated strain rates. If CF cracks 19 were initiated below the threshold strain rate for SCC they were intergranular, whereas cracks initiated above the threshold strain rate were transgranular. Crack initiation by SCC at the lower frequencies produced a pronounced reduction in fatigue lifetime. Crack initiation at higher frequencies was thought to be caused by plastic strain localization due to fatigue and local passive film breakdown. Limited data by Krupp et al. [40] on precracked high strength aluminum alloys showed that a decrease in frequency brings an increase in cyclic cracking rates. They also support the theory that a greater sensitivity to SCC causes a more pronounced effect of cyclic frequency on fatigue crack propagation. Wei [103] performed fatigue tests with 7075 in humid air. He found a very small decrease in cyclic cracking rates if the frequency was increased from 0.5 Hz to about 60 Hz. Slow strain rate tensile tests on high strength aluminum alloys were performed by Holroyd et al. [2] and Khobaib et al. [104] in aqueous NaCl solutions and humid atmospheres of S0 2 and N0 2. A steady increase of SCC susceptibility was found for the 7000 series alloys with decreasing strain rate. The 2000 series alloys on the other hand showed a maximum sensitivity at a strain rate near 10'V1. Ugiansky et al. [105] found a maximum sensitivity at a strain rate near lO 'V 1, for a 7000 series alloy but mentioned that this maximum was decreased to lower strain rates if the alloys were tested in less sensitive heat treatment conditions. Similar results were obtained by Buhl [4], but the maximum SCC sensitivity in aqueous NaCl solutions for the 2000 and 7000 series Al-alloys was near 10'6s"!. 20 Pizzo et al. [106] investigated an AlLiCu alloy by the slow strain rate technique and found a steady increase of SCC susceptibility with decreasing strain rate. Braun et al. [107], on the other hand, found no influence of strain rate on the cracking behaviour of the 8090 alloy in aqueous NaCl. Several studies on frequency effects in stainless steels have been performed by Ford et al. [108], Andresen [109] and Hudak et al. [110] at elevated temperatures in different aqueous solutions. Although it is not clear whether these results are relevant to the behaviour of Al-alloys, it was found that real time crack propagation decreased with decreasing frequency. However, below a certain frequency crack velocities (da/dt) stayed more or less constant The same trend was seen by Parkins [111] with an Al-Ni bronze in seawater. Ford [22] also gave a schematic presentation of a smooth transition from SCC to CF with increasing crack tip opening displacement rates. It was also explained why higher monotonic strain rate tests might obscure the environmental effects, due to mechanical blunting. Therefore, it may be more appropriate to study environmental interactions by cyclic loading at strain rates comparable to high monotonic strain rates. Selines et al. [112] studied effects of the cyclic stress wave form on CF crack propagation of a high yield strength 7075 alloy. They found that environmental effects occurred only during periods when crack opening plastic deformation was occurring as during the increasing stress part of a cycle. However, they studied cyclic stress levels where K,,,^ was below K B C C where the stress corrosion component of crack advance was negligible during constant load periods. Similar results were obtained by Barsom [113] with high strength steels. 3.4 Inhibi tors for Environment Assisted Cracking 21 A wide range of inhibitors that control SCC for many alloys was presented by O'Dell et al. [114]. Several chromate compounds are listed for the protection of aluminum aerospace structures. These are often incorporated into primer coatings. Agarwala et al. [115] studied a multicomponent inhibitor system of dichromate, nitrite, borate and molybdate in mineral spirit on 7075 aluminum alloys. He found a 3 fold decrease in region II SCC crack velocities and almost a 2 fold improvement in the threshold value KJSCC. SCC tests were carried out in a humid atmosphere with chloride and inhibitor solution being applied periodically to the notch of the DCB specimen. The effect of a borax nitrite-based inhibitor on CF of 7075 was investigated by Khobaib et al. [38]. Compared with cracking rates in sodium chloride solutions, the addition of the inhibitor decreased cracking rates to the same level as obtained by fatigue in air. Also, the fractographic features changed to those typical for fatigue in air, which produced ductile striations as compared to brittle striations in inhibited NaCl solutions. (The different striation types will be discussed in Chapter 3.6). Even the addition of inhibitor to a growing crack in an aggressive environment could immediately change the cracking rates. Comparative studies by Khobaib et al. [38] on the dissolution rate of 7075 also showed a marked effect of their inhibitor in NaCl solutions. Inhibitors of general corrosion do not necessarily always prevent SCC. In carbon steels for example, nitrates, carbonates and phosphates inhibit general corrosion, but the same species also promote SCC [116]. Stoltz [42] who employed a nitrate solution as a corrosion inhibitor for alloy 7075 in aqueous NaCl solutions found a reduction in CF crack growth rates by a factor of 10 to 22 the level observed in air. Agarwala [117], reporting on CF of 7075, stated that multicomponent inhibitors are often necessary to control the complex electrochemical reactions of environmental cracking. He listed several possibilities, including the reduction of anodic dissolution, buffering of pH, inhibition of pitting attack, reduction of hydrogen adsorption into the metal and enhanced hydrogen recombination reactions in the solution. He studied the effect of dichromates, molybdates, nitrites and borates in acidic chloride solutions and mixtures thereof. A more mechanistic treatment of corrosion inhibition in the context of environment sensitive cracking is given by Parkins [116]. 3.5 SCC - Fractography of High Strength Al-Alloys Fracture surfaces produced by SCC are usually intergranular in the high strength aluminum alloys. However, transgranular cleavage-like fracture has been observed under certain embrittling conditions [118,119]. A list of fee materials that can undergo transgranular SCC has been compiled by Pugh [120]. Also, at very high stress intensities, ductile transgranular fracture is observed. A fractographic investigation by Helfrich [121] on a 7075 alloy showed that the fracture surface is composed of intergranular and transgranular regions. Intergranular fractures were featureless and were attributed to SCC. Transgranular regions were often associated with large precipitates and were caused by mechanical rupture of uncracked ligaments. 23 McEvily et al. [122] and Nelson et al. [119] observed transgranular SCC and striation-like markings on the fracture surfaces of Al-Zn-Mg alloys. The striations were attributed to a discontinuous cracking mechanism, probably caused by hydrogen embrittlement phenomena. Crack arrest markings were also found by Scamans [123,124] and Hermann [125] on intergranular fracture surfaces of Al-Zn-Mg alloys. The intergranular fracture was brittle and also attributed to a hydrogen embrittlement mechanism. Lynch [126] studied the mechanistic aspects of hydrogen embrittlement and found that the cleavage-like appearance of brittle facets on fracture surfaces actually occurred by very localized plastic deformation. This localized deformation is caused by chemisorption of species (eg. H^) which reduce the strength of interatomic bonds in the metal lattice. Lynch often observed very small dimples on cleavage-like fracture surfaces. Very little has been reported on the SCC fractography of Al-Li alloys. Braun et al. [107] reported slip steps on intergranular fracture surfaces. Additionally, in an inert environment dimples were observed on the intergranular facets. In aggressive environments, they observed intergranular corrosion on the fracture surface along grain boundaries and subgrain boundaries. Other researchers also report intergranular SCC fracture in the Li-containing alloys [82,83]. 3.6 CF - Fractography of High Strength Al-Alloys Fatigue striations are the most typical features on fatigue fracture surfaces in stage II. Aluminum alloys usually show well developed fatigue striations [44]. However, if an 24 aggressive environment is present these features are often less pronounced. Many models have been suggested for the formation of striated fracture surfaces during fatigue. They have been summarized in several review articles [127,128,33]. Early models of fatigue assumed that the material ahead of the crack tip work-hardened progressively until ductility was exhausted in the plastic zone and then an incremental crack advance took place [129]. Cyclic tension-compression experiments by Laird and Smith [130] later suggested that plastic deformation is responsible for striated crack propagation. Their work showed that the crack tip is blunted during loading and resharpened on unloading (compression) by reversible plastic deformation. Details of the plastic blunting have been investigated by Neumann [131], who studied single crystals. He found that crack blunting and resharpening occurred by coarse slip on slip planes that pass through the immediate neighborhood of the vertex of the crack tip. During a single fatigue cycle, more than one pair of slip planes were thought to be active. Pelloux [132] described an alternating shear model for fatigue crack extension in polycrystalline materials. It was suggested that differences in fatigue behaviour in vacuum and air could be explained by their model. Reversed slip was not possible in air due to oxidation of slip planes. Consequently, slip on the unloading cycle had to take place on a new slip plane leading to the formation of a fatigue striation. In vacuum reversed slip was possible because oxidation did not occur. Therefore, fatigue striations were not usually found on the fracture surface. A slightly different approach was taken by Tomkins et al. [133]. In aluminum it was observed that plastic deformation was localized in two symmetrical narrow shear bands inclined at 45° to the crack plane. During the loading part of a fatigue cycle the crack was progressively blunted by shear flow at the crack tip on two symmetrical slip 25 planes. When strain hardening restricted further plastic flow on two opposing slip planes, fracture occurred at the centre of the blunted crack. Upon further strairiing, plastic flow occurred on a new pair of slip planes. This process was repeated until the maximum load was reached. During unloading, plastic flow on the same slip planes occurred in the opposite direction. The result was a striated fracture surface where several striations were produced during a single fatigue cycle. This process was described as typical for stage II cracking. Work by Forsyth [134] revealed for the first time two different types of striations on a fracture surface of an Al-Zn-Mg alloy. They were distinguished as "ductile" and "brittle" striations. Ductile striations were found on planes essentially perpendicular to the applied stress. The brittle striations on the other hand were lying on crystallographic facets that were frequently inclined to the general fracture plane. It was found that ductile striations were more likely formed at higher stress levels, whereas brittle striations were favoured by corrosive environments, low cyclic stresses and low frequencies [134]. Wanhill [118] studied high strength aluminum alloys in regard to ductile and brittle striations. It was found that brittle striation spacings were about three times that of ductile striation spacings and the facets lay on {100} planes. Also, Pelloux [135] found {100} to be the fracture planes in high strength Al alloys. It was suggested that chemisorption could explain the cleavage-like crack advance. Laird [128] suggested that slip on {111} planes could cause crack propagation on a plane that is symmetrically contained by the operative slip planes i.e. {100} and {110}. The cleavage steps over relatively short distances were explained by adsorption induced phenomena which produced damaged zones ahead of the crack tip. Arguments against a cleavage mechanism were that cleavage fracture normally occurred on {111} planes, for example in liquid mercury, and that cleavage fracture was usually favoured by higher strain rates. 26 Laird [128] and Bowles et al. [136] found earlier that the fracture surface itself lay on {100} or {110} planes but segments of striations were made up of {111} planes as found with etch pits on single striation flanks. With extensive etch pitting experiments on 316-L stainless steel, Fong et al. [137] recently proved that in CF cleavage-like fracture on {111}, {110} and {001} planes occurs by microscale slip on {111} planes. This fracture mode is typical for stage I CF behaviour. Lankford [36] related fatigue striation spacings in 7075 alloys to dislocation band spacings, crack tip opening displacement (CTOD) in a moist air environment. At stage I stress intensities (AK), the calculated average crack advance increment per cycle, da/dN, could not be related to CTOD or dislocation band spacing but was considerably smaller. Therefore, it was concluded that crack growth did not proceed cycle by cycle but occurred in increments after numerous cycles. This was related to an accumulated damage mechanism and brittle crack extension on {110} planes. The effect of the moist air environment was thought to decrease the cycles required to cause the critical crack tip "damage" (cyclic plastic deformation) for an increment of brittle crack extension. This explained why da/dN values in moist air were higher than values obtained in vacuum. Also, Nix and Flower [138] investigated micromechanisms of fatigue crack growth in an AlZnMgCu alloy. Tests were performed in a moist air environment. Fatigue striations were observed on cleavage-like crack planes which were identified as {110} planes. It was suggested that each fatigue striation was formed by a cleavage-type crack extension through a finite hydrogen embrittled zone ahead of the crack tip followed by plastic blunting as the crack emerged from the embrittled zone. Further plastic 27 deformation during unloading produced slip steps on one side of the fatigue striations. Meyn [139] also noticed slip steps on striations produced at high stress intensities where extreme local plastic deformation took place. Most of the mechanistic studies on fatigue crack propagation in aluminum alloys have been conducted in very innocuous environments like air. In more aggressive environments detailed fracture surface features are often destroyed by the corrosive action of the medium. It also has to be pointed out that CF testing of aluminum alloys almost exclusively produces transgranular fracture and there are very few data on intergranular fatigue. Aluminum lithium alloys show intergranular cracking if stressed in the short transverse direction [140]. In other testing directions mainly transgranular fracture has been reported. Crystallographic cracking has been found on slip planes at lower stress intensities (AK), together with secondary intergranular cracks [97,141,142]. Ohrloff [97] noticed that in inert environments fracture occurred in each grain on a single slip plane whereas in NaCl solutions many fracture facets were associated with different slip planes. Microvoids observed along grain boundaries led Jata et al. [142] to believe that a ductile PFZ was present. 3.7 The Superposition Model The shape of a CF real time crack growth rate log(da/dt) versus K ^ , curve shows some similarities to a SCC crack velocity log(da/dt) versus K curve under sustained loads. In both cases three distinct regions of cracking, behaviour can be found. This and other facts have led Wei et al. [143] to suggest that a division of the two cracking mechanisrns is quite artificial. They, therefore, suggested that fatigue in an aggressive environment (CF) is simply composed of a mechanical fatigue component and an environmental component. Therefore, the fatigue crack growth rate in an aggressive environment is the sum of growth rates produced by these two components. Expressed algebraically: Integration of Equation 3.1 is taken over one cycle and incorporates the effects of loading wave form and frequency through K(t). The superposition model was also successfully applied to high strength aluminum alloys by Speidel [34]. However, he suggested that strongly frequency-dependent CF tests must be categorized in a new group called stress corrosion under cyclic loading because "true" CF mechanism were thought to be frequency independent. Gerberich et al. [144] modified Wei's model. They incorporated stress intensity range (AK), R-ratio and a proposed threshold stress intensity value for CF. Good agreement was found for high strength aluminum alloys and steels. (Eq. 3.1) where; : Corrosion Fatigue Holroyd et al. [145] were critical about the applicability of a very theoretical superposition model, especially if SCC cracking rates in a corrosive environment are very different from fatigue crack propagation rates in an inert environment. Another 29 important fact is that CF can occur in environments where stress corrosion effects are negligible. They suggested that the problem should be examined from a different perspective whereby the inert fatigue component is a competitive member in the CF process rather than an additive one. Several other empirical formulations of superposition models found in the literature were summarized by Sudarshan et al. [33], with the comment that some experimental data will always fit a certain model. 3.8 The Crack T ip Strain Rate Parameter With the slow strain rate technique the SCC susceptibility of a specific material/environment system at a given strain rate is determined by measuring different variables. Examples are the reduction in cross sectional area after failure, the number of stress corrosion cracks formed on the specimen surface or the depth of surface cracks. The rising load test technique with a precracked specimen on the other hand has been mainly used to determine K I S C C threshold values for a specific material/environment system. With both techniques it is possible to apply strain rates to a specimen which compare to strain rates during loading or unloading cycles of low frequency fatigue tests. Therefore these techniques can be applied to investigate the interface between SCC and CF. It seems practical to use the strain rate as a typical parameter because the frequency becomes rather meaningless for SCC tests. Because the strain distribution around crack tips is complicated, many simplified models have been suggested to quantify the crack tip strain rate parameter. Several relations are presented in the following text. More extensive reviews are given by Parkins [146] and Lidbury [147]. Scott and Truswell [148] modeled the crack tip area by defining the crack tip opening displacement 8 as the gage length undergoing local tensile loading. Based on 30 linear elastic fracture mechanics theory (LEFM), the following relationship was given: where; K = Stress intensity 8 = Crack tip opening displacement E = Elastic modulus cy = Yield Strength Ai = Numerical constant Under the assumption that uniform straining occurred over the whole gage length (8), the crack tip strain rate was described as 8 = A t • K 2 E - G y (Eq. 3.2) d e = l d8 dt ~ 8 dt (Eq. 3.3) For a triangular load wave form, Equation 3.3 led to an average crack tip strain rate of 31 where; R = R-ratio = K m i n / K m A I T = Loading time Tomkins and Wareing [149] based their strain rate estimations on a model of localized flow bands within the plastic zone. Their bands were symmetrically inclined at 45° to the crack plane. The stress within these bands was assumed to be independent of the distance (r) ahead of the crack tip, and the strain e(r) varied as the inverse of r. This was experimentally verified by Lankford et al. [150]. Therefore, the strain (£,) at the elastic/plastic boundary was inversely proportional to the size of the plastic zone (p), as shown in Equation 3.5. (Eq. 3.5) where; B t = Numerical constant It follows from LEFM that P = (Eq. 3.6) where; B 2 = Numerical constant 32 At a location (r) ahead of the crack tip the strain was described as: B 2 -o 2 -a 1 (Eq. 3.7) The strain rate at location (r) was then given by Equation 3.8 de B 2 a i da (Eq. 3.8) ——.-.zo.-— oj r dt For a triangular load wave form the average strain rate at location (r) was: T - £ y a2, r 1 B 2 a l (Eq. 3.9) where T = Loading time The formulation in Equation 3.9 shows some of the complexity of the problem since the strain rate is dependent on the location relative to the crack tip. According to Lankford et al. [150] this formula could only be applied to calculate strain rates at locations more than approximately 1 L i m ahead of the crack tip. Lidbury [147] explained how the calculations of Scott et al. [148] could be modified with the Tomkins model [149] of shear bands. Instead of using the crack tip opening displacement as a gage length, the total width (W) of the active flow band was incorporated in a shear displacement calculation based on the idea that shear displacements account for the crack opening. The shear strain rate dy/dt became dy_JL_d5 dt ~ W dt (Eq. 3.10) 33 For a triangular wave form, the resulting shear strain rate was given by Equation 3.11 1 1 D 3 . , dy v dt , W T ECT average " A (KM-K*) (Eq.3.11) where; D 3 = Numerical constant Hudak et al. [110] used experimental techniques to measure crack tip strain rates based on optical displacement field measurements on precracked specimens of stainless steels. Based on the formulation of strain rate as: de Se 5K 5e 8a ^ „ , „ v — = -^'-Z- + T-T- (Eq. 3.12) dt 8K 8t 8a 8t M they found that the second term in the Equation 3.12 could be neglected (ie. the strain rate contribution from crack advance). By fitting an equation to their experimental data, they formulated the strain rate as de — = 2 - F - H - A K H - f (Eq.3.13) dt where; F and H = Numerical constants f = Frequency of cyclic loading AK = Stress intensity range 34 Essentially the same formula with similar constants was quoted by Ford et al. [151]. Congleton et al. [152] derived an expression for crack tip strain rate based on the macroscopic displacement rate on smooth tensile specimen. In their analysis, they accounted for the number of cracks present on the macroscopic gage length and included a microscopic gage length at each crack as the width of the active flow bands, as proposed first by Tomkins and Wareing [149]. The following expressions of crack tip strain and crack tip strain rate were derived by finite element analysis. The analysis was based on ideally elastic and ideally plastic crack tip behaviour and on an estimated width of shear bands. Crack tip strain: where; Mj and M 2 = Numerical constants n = Number of cracks on the gage length a = Crack length da/dt = Crack propagation rate (Eq. 3.14) Crack tip strain rate: (Eq. 3.15) 35 Parkins [146] took the effort to compare several methods for crack tip strain rate calculations and applied them to a pipeline steel for stress intensity ranges up to 15 MPaVm and a cyclic frequency of about 10"2 Hz. With these specific conditions, the calculated crack tip strain rates varied over several orders of magnitude leaving it very questionable which formula to use. Since most models formulated include some assumptions based on hnear elastic fracture mechanics, a simple descriptive method is proposed in the present work. Recognizing that the stress intensity factor (K) characterizes stresses at the crack tip, which then affect strains, it seems meaningful to choose the stress intensity rate parameter (dK/dt) as the descriptive term for different types of loading experiments. Also, dK/dt allows one to account for crack propagation during an experiment which can become important for very slow cycling experiments. In the following text (dK/dt) will be described as K-rate. 36 4 Experimental 4.1 Mater ia l 4.1.1 Chemical Composition The chemical compositions of alloys 7075 and 8090 are given in Table 4.1. Values are in weight % and represent the nominal composition. No actual chemical analyses of the test materials were performed. Table 4.1 Chemical Composition (wt.%) Zn Li Cu Mg Cr Zr Rest Alloy 7075 5.6 1.6 2.5 0.23 Al Alloy 8090 2.5 1.3 0.7 0.12 Al The alloy 7075 was received as 25 mm thick plate heat treated to the T651 condition (ie. peak aged). The alloy 8090 was obtained as plate material of 46 mm thickness. It was in the solution treated and aged condition. Complete heat treatment details were not known because this material was still under production development by Alcan. Typically, plate material of 8090 is solutionized in a salt bath at 540*C and water quenched. Afterwards the material is stretched by 2-3% and then aged at 190'C for 16 hrs [53]. 37 4.1.2 Mechanical Properties Tensile properties of both test materials were measured in the longitudinal (L) and the short transverse (S) direction. The fracture toughness was only measured in the SL direction with the tensile load applied perpendicular to the rolling plane and crack propagation in the rolling direction. The gage section of tensile specimen was 20 mm but was smaller for tests of alloy 7075 in the S direction. The specimen diameter was always 5 mm. Table 4.2 summarizes the results of all tests. Tensile properties show the results of two specimens whereas the fracture toughness was obtained with a single specimen. For comparison, literature data of the two alloys are also included in Table 4.2 which represent minimum specified values. Because tensile properties of thin plate material in the short transverse direction is usually not specified, these data are not given for alloy 7075. Literature data for alloy 8090 correspond to a tentative heat treatment specification (T-8771) by Alcan. The test material used in this study may have had a different heat treatment because tensile properties in the short transverse direction were lower and the elongation to failure was much smaller. The fracture toughness of both test materials was measured with the same DCB-specimen geometry as used later for SCC and CF experiments. The specimens were tested in laboratory air and the load was applied with an Instron tensile machine. The load versus crack opening displacement curves were analyzed according to ASTM norm E-399 74 for fracture toughness testing. The fracture toughness values obtained were within the validity limits set by the norm for a plane strain fracture toughness K J C . The fracture toughness of alloy 7075 was 23.9 MPaVm with an almost linear load versus COD curve up to K I C . However, already near 21.5 MPaVm small load drops on the load 38 versus displacement curve indicated first signs of cracking. The fracture toughness for alloy 8090 was 15.5 MPaVra The load versus COD curve deviated from linearity above about 12 MPaVm and around 14 MPaVm the first "cracking sounds" could be heard. Table 4.2 Mechanical properties UTS YS e (Elongation) Kic MPa MPa % MPaVm Alloy 7075 L (test material) S 578/578 603/606 526/534 532/547 10.4/10.8 I .S^ .O 2 * 23.9(SL) Alloy 8090 L (test material) S 483/497 303/309 448/467 303/309 1.7/1.8 <0.5 15.5(SL) Alloy 7075 L (literature values3)) 540 470 7 20(SL), 25(TL), 29(LT) Alloy 8090 L (literature values4)) 440 430 390 330 4 2 15(SL), 24(TL), 27(LT) (1) The gage section was 6 mm (2) The gage section was 10 mm (3) From reference [153] (4) From reference [154] For comparison, K I C values were obtained from specimens that contained a SCC precrack (eg. K I C determined by simply raising the load to failure at the end of a SCC test). The K I C values obtained in this manner were comparable to the values listed in Table 4.2. For example, K I C was 21-23 MPaVm for 7075 and near 17 MPaVm for 8090. 39 4.1 J Microstructure The rnicTostructuxe of both materials was examined in three orthogonal planes, including the rolling plane. The specimens were ground with SiC paper down to 800 grit and subsequently polished with diamond paste to a 1 p:m finish. The grain structure was revealed by immersion etching in a mixture of: "Keller's reagent" composed of the following reagents: 2ml HF 3mlHCl 5 ml HN0 3 190 ml H 2 0 Figure 4.1 shows the grain structure of both alloys. Alloy 7075 showed grain elongation parallel to the rolling direction with grain diameters of several hundred micrometers. In the thickness direction the grains had approximate diameters of 5 -10 p.m. Alloy 8090 was fabricated by cross rolling and did not exhibit grain elongation in one specific direction. The partly recrystallized microstructure exhibited grain diameters in the rolling plane in the range of about 2-150 |im. In the thickness direction grain diameters varied between about 2 and 30 urn.. In order to establish the crack path with respect to microstructural features, a specimen was sectioned a few millimeters ahead of the crack front normal to the crack propagation direction. The specimen was mounted in epoxy resin and the remaining ligament ahead of the crack was carefully ground and polished until the crack tip was reached. Then the specimen was etched to reveal the grain structure. 40 41 4.2 Specimen Design Double cantilever beam (DCB) specimen of both materials were machined from the plate material with the crack starters (chevron notch) lying parallel to the rolling plane. The direction of crack propagation was parallel to the rolling direction. (The alloy 8090 did not have a specific rolling direction due to cross-rolling). In the case of alloy 8090, two specimens were cut from the parent plate in the thickness dimension with the notch being at the quarter thickness plane of the plate. After machining all specimens were ground with SiC paper down to 800 grit and subsequently polished with a 1 urn alumina slurry. The specimen geometry of the DCB specimen is shown in Figures 4.2 and 4.3. The holes parallel to the crack plane were for the grips of the loading apparatus. A hole drilled through the specimen normal to the crack plane was threaded and served for bolt loading tests. Hereby two opposing bolts were tightened against each other to a desired crack opening or stress intensity respectively. Tightening of two opposing bolts was also used to break a specimen open for inspection of the fracture surface and crack length measurements after a test was completed. The 7075 alloy specimen had knife edges machined at the loading line. Because specimen of alloy 8090 were narrower knife edges had to be bolted onto the sample 8 mm away from the load line. During the experiments a clip gage was clamped between the knife edges to measure the crack opening. 42 Figure 4.2 DCB specimen geometry for AA-7075 25.4 mm -0 = 4.76 mm -0= 6.35 mm (threaded) 25.4 mm Figure 4.3 DCB specimen geometry for AA-8090 21mm E o 0 = 4.76 mm 0 = 6.35 mm (threaded) 21 mm 43 4.3 Specimen Precracking The crack starters in DCB specimens were sharpened by fatigue pre-cracking on a Sonntag SF-l-U fatigue machine at approximately 30 Hz under cyclic load control. In order to control the maximum stress intensity, the applied cyclic load had to be decreased several times because of the increasing crack length. The average fatigue crack length with the 7075 alloy was 65 mm and with the 8090 alloy 47 mm respectively. Typical stress intensity values during a precracking experiment are given in the Tables 4.3 and 4.4. The crack length during fatigue precracking was measured on the specimen surface. Because a Chevron notch was used as crack starter, crack initiation was forced to the centre of the specimen. This produced a crack front which was symmetrical with respect to the centre line along the specimen. The crack front was only slightly curved with the crack in the centre of the specimen being about 2 mm longer than on the edges. Due to the flat grain structure of both alloys the crack always propagated on a plane parallel to the rolling plane and, therefore, parallel to the specimen surface. Table 4.3 Fatigue precracking of alloy 7075 Crack length intervals AJC^CMPaVm") K^CMPaVm") up to 35 mm 3.5 4.0 35 to 50 mm 3.8 4.9 50 to 60 mm 3.7 5.0 60 to 65 mm 3.2 4.7 44 Table 4.4 Fatigue precracking of alloy 8090 Crack length intervals AK™x(MPaVm) K^CMPaVm) up to 35 mm 5.2 6.3 35 to 40 mm 5.1 6.5 40 to 45 mm 4.5 6.2 45 to 47 mm 4.0 5.7 4.4 Specimen Compliance Based on the Griffith energy criterion for fracture [155] the elastic energy release rate per unit area of crack plane (G) can be described as: B I da J (Eq. 4.1) where; U = Total elastic energy in a plate of thickness B a = Crack length. with U = ^CP2 (Eq. 4.2) 45 where; C = Compliance, C = d/P d = Crack opening displacement (COD) at the load line P = Applied load From Equations 4.1 and 4.2 the energy release rate is described as P2 dC G= — • — (Eq. 4.3) 2B da For plane strain conditions Griffith calculated G as: G = ^ . ( l - v 2 ) (Eq.4.4) E where; K = Stress intensity v = Poissons ratio E = Young's modulus Therefore, from Equations 4.3 and 4.4 K 2 = E P 2 d C 1 2B da (1-v 2) (Eq. 4.5) 46 To obtain the compliance (C) as a function of crack length, experiments with DCB specimen of different crack length were performed. Hereby the crack opening displacement (COD) at the load line was measured as a function of the applied load. For a given crack length a linear relationship existed between COD and applied load. The crack length was determined after the test as the average of 9 data points across the width of the specimen (specimens were broken open and measurements made in the crack plane). For alloy 7075 the compliance was measured for 9 different crack length between 40 mm and 90 mm. For alloy 8090 compliance measurements for 6 different crack length between 40 mm and 60 mm were performed. A second order polynomial was fitted through the data points of compliance versus crack length. The crack length was presented as the dimensionless ratio of (a/L), where (L) is the length of the specimen measured from the load line. The results of (C) as a function of (a/L) for the specimen geometries of alloys 7075 and 8090 are given in Table 4.5. The knowledge of the specimen compliance allowed one to determine the crack length for a measured COD/load ratio. With the derivative dC/d(a/L) it was then possible to calculate the stress intensity for a given crack length and applied load via Equation 4.5. Table 4.5 Compliance polynomial C(a/L), [kN"1] Alloy 7075 1 } C = 0.70596 - 5.0919(a/L) + 13.5445(a/L)2 Alloy 8090 2 ) C = 1.5457 - 7.4241-(a/L) + 11.8434-(a/L)2 (1) Specimen length L = 165 mm (2) Specimen length L = 110 mm The polynomial and the measured data points are displayed in Appendix III on page 192. 47 Because the knife edges on alloy 8090 specimen were located at the top of the sample, the crack opening displacement was not measured directly at the load line. To calculate the crack opening at the load line, it was assumed that the crack opens like a hinge with no opening at the crack tip. The opening along the crack would then be proportional to the crack length. This assumption introduced a small error because of the crack tip opening on the specimen. However, even under the highest stress intensity applied the crack tip opening calculated is smaller than 15 Lim and the resulting displacement error at the load line would be less than 3 Jim which was smaller than the accuracy of the clip gage. 4.5 Monotonic Slow Loading Experiments (Rising-K Experiments) A Hounsfield tensometer was fitted with sets of interchangeable reduction gears and a 12 rph synchronous motor. With this modification, different crosshead displacement rates were possible in the range of 4.7X10"4 um/s up to 2.9xl0"2 Lim/s (1.7 Lim/hour to 104 pm/hour). In the Hounsfield tensile machine a spring beam acted as load cell. Therefore, the crosshead displacement did not translate into an equal amount of crack opening displacement at the DCB specimen but the latter being about 27% smaller. The spring beam used for all tests had a spring constant of 0.271 mm/kN, as determined experimentally. The compliance of a DCB specimen was typically around 1 mm/kN. The load beam was fitted with strain gages which allowed measurement of load changes down to about 5 N. The crack opening displacement on the specimen was measured with a clip gage that had a sensitivity better than 10 \im. The clip gage and load cell were connected to a "Bean-Digital-Strain-Indicator, Model 20". 48 During a slow loading experiment the calibrated output of load and COD versus time was continuously recorded on a chart recorder. The change in crack length due to SCC was obtained from the measured COD/load ratio and the specimen compliance calibration. Compliance measurements of crack advance were limited to measurements >0.15 mm, due to limited sensitivity of the load cell and clip gage. It should be noted that the compliance method was only used to calculate crack advance and not the initial crack length. The initial crack length was determined by optical measurements on the exposed fracture surface after opening the specimen at the end of a test. The position of the initial crack front produced by fatigue pre-cracking was readily visible on the fracture surface. Also, the position of the final crack front produced by overload was measured on the exposed fracture surface. However, this was not true for alloy 8090 where it was rather difficult to distinguish between fatigue pre-crack, stress corrosion crack and overload crack. Therefore, photographs were taken of the fracture surface with strong illumination on one side. The crack advance caused by SCC or CF was then measured on the photograph (about 6x magnification). The crack length on the fracture surface was always measured as the average of 9 data points across the width of the specimen. From the initial and final crack length measurements it was possible to compensate for any drift in the electronic recording of load and COD (which yield a crack length from the compliance calibration) during long term experiments. For comparison and verification, the crack advance during most tests was also measured periodically on the specimen surface with a travelling microscope. Crack length versus time data were fitted to a second order polynomial. For the regression analysis the final data points where the stress intensity was approaching K I C were not included. Also, the initial data points were not included where no cracking was measured. A second order polynomial was chosen because higher order polynomials did not produce significantly better correlation coefficients. The fitted curve was later used 49 to compute crack velocities versus time by differentiation of the second order polynomial. (Note differentiation gives da/dt). From crack length (a) and load (P) stress intensities (K) were calculated via Equation 4.5. This produced (a)-K or time (t)-K curves respectively. The combination of (t)-(da/dt) data with (t)-K data allowed to compute K-log(da/dt) curves. Because SCC of alloy 8090 only occurred at stress intensities near K I C , the compliance method could not be applied to monitor the crack length. The reason being that the load versus COD curve deviated from linearity at the higher stress intensities and, therefore, C was not constant for a given crack length. In some instances the clip gage did not function properly and the crack advance could only be monitored on the specimen surface. To calculate the stress intensity corresponding to the average crack length measured across the specimen width, a constant value was added to the surface crack length measurements in order to account for the curved crack front. The curvature of the crack front was determined after the test on the fracture surface and a value between 1 and 2 mm usually had to be added to the surface crack length in order to obtain the average crack length. From the initial and final crack length measurements on the fracture surface it was obvious that the overall curvature of the crack front did not change significantly during an experiment. Different K-log(da/dt) curves were obtained by conducting rising-K experiments at different loading rates (K-rates). Each such K-log(da/dt) plot was characterized by an average K-rate that was defined as: ' d K l tmax tmin (Eq. 4.6) The definition of the average K-rate by Equation 4.6 is justified because a (t)-K plot was approximately linear for most experiments (except where stated). 50 Note, (da/dt) is called "crack velocity" in order to distinguish between cracking rates under cyclic loading (da/dN) and cracking rates (crack velocities) measured in real time (da/dt). Therefore, in this case the term "velocity" does not represent a vector. 4.6 Bolt Loading and Constant Load Experiments For most bolt loading experiments fatigue pre-cracked DCB specimens were used. Before the start of a test the specimen was loaded in a Hounsfield tensometer to the desired stress intensity. Then the crack opening was measured with a clip gage. After unloading, the specimen was reloaded with two opposing bolts to the same crack opening displacement. In one test the specimen was not fatigue pre-cracked but a pop-in pre-crack was produced by tightening two opposing bolts. The crack advance on the specimen with a pop-in pre-crack was measured at periodic intervals during the experiment (long term experiment >1000 hrs). In order to do this, the specimen was taken out of the test solution for about one minute. Time versus crack length data for specific stress intensity regions were fitted to a second order polynomial and differentiation produced (da/dt) data. The initial stress intensity after producing a pop-in pre-crack was assumed to be at KjC. Because the (t)-K curve for the entire experiment was non hnear it was not possible to assign a single K-rate to the experiment. Instead individual K-rates were calculated for approximately linear sections of the (t)-K curve. Specimen with fatigue pre-cracks were tested for shorter periods and crack advance was not recorded during the test. Only initial and final crack length were measured on 51 the fracture surface after the test. Average crack velocities were simply obtained from the total crack advance and the test duration. The average K-rates were calculated with Equation 4.6. During the constant load experiments the same arrangement was used as during the slow loading tests. The load drop due to cracking was adjusted manually at periodic intervals. This was not difficult because crack propagation rates at the tested stress intensities gave average load decrease values of less than 50 Newtons/day. The crack advance was only measured at the beginning and the end of a test on the fracture surface. K-rates were also calculated with Equation 4.6. In some monotonic slow loading tests very slow loading rates could also lead to approximately constant load conditions, due to the fact that the change in specimen compliance accompanying crack advance offset the slow displacement rate of the crosshead. Therefore, the load on the specimen stayed constant even though the crack opening slowly increased. In these experiments the K-rate decreased steadily during the experiment. Therefore, different average K-rates had to be assigned to different sections of the (t)-K curve. However, in the region where constant load conditions were established also an approximate constant K-rate was measured. 4.7 Cyclic Loading Experiments A Hounsfield tensometer was fitted with a pulley system and connected to a DC motor of variable speed. With this configuration, crosshead displacement rates in the range of 100 Lim/s down to about 0.07 Lim/s (250 Ltm/h) were obtained. This enabled 52 cyclic frequencies to be obtained on the DCB specimen in the range of 5xl0"2 to 2x10s Hz. For some tests a soft spring was mounted in series between the crosshead and the specimen to allow cycling at even lower frequencies. The load beam of the Hounsfield was connected to a lever system which magnified the load beam deflection. Microswitches were positioned to intercept the movement of the lever and to reverse the motor at the desired constant maximum and minimum load levels. This configuration produced a triangular load wave form with a constant loading and unloading rate respectively. Since maximum and minimum load were constant, the stress intensity range AK increased as the crack advanced. However, because the crack advance during a test was usually only a few millimeters, the experiments were conducted close to constant AK conditions. The specimens were mounted in a vertical position and the total crack advance due to CF was measured at the end of the experiment after opening the specimen to expose the fracture surface. From the total crack advance during an experiment, two different cracking rates were calculated. Namely the average real time crack velocity (da/dt), and the cyclic crack propagation rate (da/dN) which represents the average crack advance during each cycle. Figure 4.4 shows schematically the cyclic loading test arrangement. In addition to the slow triangular load wave tests, faster sinusoidal loading tests were performed at approximately 30 Hz on a Sonntag SF-l-U fatigue machine. The tests were also run under load control. The Sonntag tests required the DCB specimens to be mounted horizontally. In a few experiments a positive sawtooth and a square load wave form was applied. The fast unloading during the positive sawtooth experiments was done manually on the Hounsfield and took not more than a few seconds. Also the rapid periodic load changes with the square load wave form were performed manually. Figure 4.4 Cyclic loading test arrangement Micro-Switches (max.& min. load) Spring Beam (Load Cell) DCB-Specimen 54 Figures 4.5 to 4.8 show schematically plots of time versus stress intensity for the different cyclic load wave forms. Figure 4.5 The triangular load wave form Stress Intensity A (Kmax)final (Kmax)initial (Kmin)final (Kmin)initial Time The average stress intensity range (AK) a v e r a g e during a triangular load wave form fatigue test where the change in (AK) was small was obtained via Equation 4.7. (AK), [(Kmax)jnitjal CKmitJinitiaJ £(^ma*)final CKmin)fjnail average (Eq. 4.7) 55 The average cyclic K-rate was calculated as 'dK^ _(AK) a v e r a g e -2-N v. ^ Average (Ocycling where; N = Number of cycles during the test. The average real time crack velocity da _ Aa dt (Ocycling where; Aa = crack advance during the test The cyclic crack propagation rate da _ Aa dN~ N7 (Eq. 4.8) ( Eq. 4.9) (Eq. 4.10) Figure 4.6 The sinusoidal load wave form (Kmax)final (Kmax)initial (Kmin)final (Kmin)initial Stress Intensity A Time 56 For the experiments with sinusoidal cyclic loading, the same definitions were used as for the triangular load wave form. It is clear that this is a very "crude" description of the average loading rate due to the strong non-linearity of the loading wave form. Figure 4.7 The positive sawtooth load wave form Stress Intensity (Kmax)final t i (Kmax)initial (Kmin)final (Kmin)initial Time For the positive sawtooth load wave form the average cyclic K-rate was defined as the slope of the rising stress intensity part of a cycle. (dK/dt), average (AK) (t) average N cycling (Eq. 4.11 ) Figure 4.8 The square load wave form (Kmax)final (Kmax)initial (Kmin)final (Kmin)initial Stress Intensity i i (t)cycling Time The square load wave form was not characterized by a typical K-rate. 4.8 Test Environments, Cell Arrangement 57 Most tests were performed in aqueous sodium chloride solutions which were contained in a cell placed around the specimen. The solution volume was approximately 2 liters and was not changed during a single test. Chemicals used were reagent grade. All tests were performed at room temperature (~23°C). The following test solutions were used: - 3.5 wt.% NaCl; pH = 6.5 - 3.5 wt.% NaCl + 0.01M N^CrC^; pH = 8.4 - 3.5 wt.% NaCl + 0.1M Li 2 C0 3 ; pH = 10.5 (pH adjusted with HC1) - 0.1 M Li 2 C0 3 ; pH = 10.5 (pH adjusted with HC1) Prior to the actual start of the experiments, the specimens were immersed in the test solution for half a day at an initial stress intensity of 4 MPaVm This allowed local electrochemical crack tip environments to be established before loading was started. The electrochemical potential was recorded at the crack tip with a Luggin capillary connected to a saturated calomel reference electrode. The Luggin capillary was placed approximately 1 mm away from the crack tip. The capillary did not need to be repositioned during a test because of the very small crack advance (eg. <3mm). All tests were performed under free corrosion (E,^) conditions without imposing any external potential. Values of E ^ were recorded. However, the potential at the specimen surface may differ from that at the crack front due to localized changes in solution chemistry that occur in occluded sites. Some tests were also performed in dry air. The specimen was enclosed and sealed in a plastic container (bottle) exposing only the grip area. The specimen was stored for 2 58 days in the container before the loading experiment was started. To provide dry air conditions the "bottle" was partly filled with silica gel. Figures 4.9 and 4.10 show the cell arrangement for tests on the Hounsfield tensometer and the Sonntag fatigue machine. On the Sonntag machine, tests in aqueous solutions and dry air were performed with the same cell arrangement For dry air experiments on the Hounsfield also the dry air cell from the Sonntag was used but it was mounted in a vertical position. Figure 4.9 Test cell arrangement for the Hounsfield tensometer Figure 4.10 Test cell arrangement for the Sonntag fatigue machine 4.9 Scanning Electron Microscopy 59 Fracture surfaces were studied by conventional scanning electron microscope (SEM) techniques with a Hitachi S-670 SEM. An exitation energy of 20 keV resulted in the best secondary electron imaging conditions. Prior to examination in the SEM, the specimens were ultrasonically cleaned in an inhibited acid solution for a few seconds. The solution composition of the inhibited acid was [156]: 70 ml Orthophosphoric acid 32 mg chromic acid 130 ml distilled water Matching studies were conducted on opposing fracture surfaces by carefully aligning the two halves of the specimen beside each other and glueing them to the specimen holder. The macroscopic crack propagation direction on the specimen halves was set parallel to the tilt axis of the SEM specimen. This arrangement allowed matching areas to be more easily located, even in the tilted position used for stereo photography. With the 8090 alloy it was additionally necessary to place reference scratches on the fracture surface for clear identification of specific SCC or CF areas. In order to view the crack front of a DCB-specimen by SEM, the sample had to be sectioned in a special way. The procedure used is schematically shown in Figure 4.11. After the CF experiment, the specimen was sectioned with a jewellers saw in the unloaded condition. A plane approximately normal to the crack propagation direction was carefully ground and polished until the crack front was reached. Then the crack was opened by inserting a small wedge between the remaining crack flanks left on one side of the specimen. Figure 4.11 Sample sectioning for SEM crack tip observations Observa t ion (SEM) 61 5 Results 5.1 Monotonic Loading Experiments (Alloy AA-7075) 5.1.1 Slow Loading (Rising-K) Experiments (AA-7075) A typical plot showing the general increase in crack length with time for alloy 7075 in 3.5 wt% NaCl at a K-rate of 1.3xl0"5 MPaVm/s is presented in Figure 5.1. The lower curve corresponds to optical measurements of crack length on the surface and the upper curve shows the average crack length obtained by the compliance method. The difference in initial crack length between the two measurements is due to curvature of the crack front. The fitted 2nd order polynomial regression curves are also shown in Figure 5.1. The curve fit was usually better for the clip gage data (compliance method) than for the fewer surface crack length measurements. The coefficient of correlation for specimen A13 in Figure 5.1 was 0.9994 for the clip gage data and 0.9897 for the surface measurements. (The coefficient of correlation is defined in the appendix and listed together with the correlation coefficients for the curve fit of all other tests) At K-rates < 5x10s MPaVm/s produced by crosshead displacement rates slower than 31 pm/h, the first crack advance was monitored at stress intensities between 8 to 10 MPaVnx Cracking at higher K-rates could not be detected until considerably higher stress intensities were reached. Most rising-K experiments were conducted up to a maximum stress intensity level below K I C . All data from slow loading tests are summarized in Table 5.1. Figure 5.1 Typical plot of crack length versus time for a slow monotonic loading test (Specimen A13, K-rate = 1.3xlO"5MPaVm/s) Slow Loading Test Crack Length (mm) 80 AA-7075, Specimen A13 3.5wt% NaCl + + Clip Gage Data + 1 1 t 1 1 t 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 I i I ) r t H r r t r l 1 1 t 1 • + -l- 1 H - Surface Measurements 78 76 74 72 70 68 66 64 62 60 i r 0 100 200 Time (hrs) i r 300 400 Table 5.1 Results from slow monotonic loading tests AA-7075, 3.5wt% NaCl Specimen Aa Test dK/dt No. duration average [mm] [MPaVm] [mm] [MPaVm] [hrs] [MPaVm/s] A17 64.95 6.1 20.36 13.7 1170 2.3xl0'6 A30 63.94 5.7 4.10 12.8 744 2.6xl06 A34 64.32 2.9 2.83 16.2 450 8.2xl0'6 A67 64.60 2.9 1.23 13.5 340 8.7xl06 A70 64.00 2.9 17.61 17.6 622 8.7xl0"6 A93l> 66.17 5.9 9.64 17.2 558 6.4x10'* A3 74.99 5.1 4.14 19.1 336 1.2xl0'5 A13 65.50 4.6 5.24 19.9 315 1.3xl05 A33 65.34 2.9 1.75 13.7 224 1.3x10s A412) 65.19 13.5 1.26 4.3 170 1.5xl0-5 A43 65.69 3.1 5.70 18.7 332 1.3xl0"5 A50 65.03 2.9 4.47 18.2 160 2.6x10-5 A65 65.92 2.9 1.52 13.8 118 2.6xl0"5 A813' 66.74 3.0 0.83 17.8 153 2.7xl0-3 A55 64.43 2.9 2.01 17.5 109 3.7x10s A56 66.83 3.0 <0.3 13.7 80 3.7xl05 A58 64.87 2.9 0.92 13.5 80 3.7xl0'5 A4 68.25 3.2 10.29 20.3 119 4.4xl0'5 A6 65.37 2.9 3.87 21.9 118 4.5x10'5 A49 64.77 2.9 13.3 62 4.7x10-' A91" 66.29 2.9 1.89 18.0 85 4.9x10 s A15 67.14 3.0 11.63 91.5 6.8xl05 A16 67.66 3.1 4.8 21.4 79 6.9xl0-5 A l l 64.91 3.0 4.09 49 1.2x10-" A22 65.61 2.9 1.23 19.4 40 l.lxlO'4 A36 66.45 3.0 13.6 26 l.lxlO-4 A10 65.00 3.1 3.91 40 1.4xl04 A42 63.76 2.9 0.1 13.5 21 1.4xl0"4 Single Cycle Experiment Unloading half cycle With Inhibitor 64 Considerable crack advance occurred during tests at average K-rates below -lxlO"5 MPaVm/s, which led to an upper limit of the rising stress intensity. This limit was set by the change in compliance of the specimen with increasing crack length and the fact that the loading was controlled via the crosshead displacement rate. Therefore, the K-rate could not be assumed to be approximately constant but decreased as the crack was propagating. In Table 5.2 the different applied crosshead displacement rates are listed together with the corresponding average values of K-rate (dK/dt) for an initial crack length of about 65 mm. Table 5.2 Displacement rate and loading rate Crosshead-displacement rate (dK/dt)average pm/hr MPaVm/s 1.7 -2.5 x lO-** 4.8 -7.5 x 1QT6* 9 1.3 xlO - 5 * 18 2.7 x 10_s 23 3.7 x 10-5 31 4.6 x lO - 5 52 6.9 x 10-5 87 1.1 x lO - 5 104 1.4 xlO"5 * values are calculated during early stages of the experiment where little crack propagation occurred. 65 The results of all rising-K experiments are displayed in Figures 5.2 to 5.10. The three graphs in each figure show crack advance versus time (Aa - 1 ) , stress intensity versus time (K-t) and logarithm of crack velocity versus stress intensity 0og(da/dt)-K). Some graphs show data from several tests which were performed at approximately the same K-rate. In most cases, the crack advance was based on compliance measurements, except for samples A6, A43, A70 and A90 where crack advance was measured optically at the specimen surface because of problems with the clip gage. On the (Aa - 1 ) plots all experiments performed at a specific K-rate are included, even when only the initial and final crack length were measured. These latter tests were used mainly to obtain fracture surfaces corresponding to a specific (K) value and K-rate. A steady increase in crack velocity was measured with increasing stress intensity in all cases and no crack velocity plateaus were observed on the K-log(da/dt) plots.. Curves of K-log(da/dt) obtained with different K-rates are combined in Figure 5.11. All data seem to fall within a wide scatter band indicating no significant influence of K-rate on crack velocities. Figure 5.2 Rising-K experiments at a K-rate of ~2.5xlO"6MPaVrn/s Crack Advance (mm) 20 AA-7075 3.5wt% NaCl A-17 (2.3E-6 MPalrWs) —A— (2.6E-6 MPafrfvs) 400 600 800 Time (hrs) 1.000 1.200 Stress Intensity (MPafm) 25 Crack Velocity (m/s) 1E-07 6 8 10 12 14 Stress Intensity {MPafm) Figure 5.3 Rising-K experiments at a K-rate of ~8.5xlO"6MPa-\/m/s Crack Advance (mm) 20 AA-7075 3.5wt%NaCI A-93 {6.4E-6 UPjOls) A-70 18.7E 6 MPaRVsl —&— A M (S.2E 6 MPatfivsi A-67 (8.7E-6 MPafiSjl 4 * A , i i i , J 1 15 10 200 400 Time (hrs) 600 Stress Intensity (MPafm) 25 400 Time (hrs) 800 Crack Velocity (m/s) 1E-07 5E-08 1E-09 6 8 10 12 14 Stress Intensity (MPafm) Figure 5.4 Rising-K experiments at a K-rate of ~1.3xlO'5MPaVrn/s Crack Advance (mm) 5 AA-7075 3.5wt% NaCl (1.3E-5MPa)nvs) —A 200 Time (hrs) 300 Stress Intensity (MPaffn) 25 Crack Velocity (m/s) 1E-07 6 8 10 12 14 Stress Intensity (MPafrn) Figure 5.5 Rising-K experiments at a K-rate of ~2.6xlO"5MPaVm/s Crack Advance (mm) 5 l AA-7075 3.5wt% NaCl A-50 (2.6E-5 MPafSVs] —A— A-65 (2.6E-5 MPalm/s) A-81 (2.7E-5 MPafm/s) A A A A A ( A81 mlh inhibitor) 50 too Time (hrs) 150 200 Stress Intensity (MPafrri) 25 100 Time (hrs) Crack Velocity (m/s) 1E-07 5E-08 6 8 10 12 14 Stress Intensity (MPafm) Figure 5.6 Rising-K experiments at a K-rate of ~3.7xlO"5MPaVrn/s Crack Advance (mm) 5 AA-7075 3.5wt% NaCl (3.7E-5 ^Pafm/s) (3.7E-5 MPatfWs) A-58 (3.7E-5 MPafm/s) 100 Time (hrs) Stress Intensity (MPafrn) 25 100 Time (hrs) i Velocity (m/s) 6 8 10 12 14 Stress Intensity (MPa|m) Figure 5.7 Rising-K experiments at a K-rate of ~4.5xlO'5MPaVm/s Crack Advance (mm 5 100 Time (hrs) Stress Intensity (MPafm) 25 100 Time (hrs) Crack Velocity (m/s) 1E-07 5E-08 1E-09 Stress Intensity (MPafrri) Figure 5.8 Rising-K experiments at a K-rate of ~6.8xlO"5MPaVm/s Crack Advance (mm) AA-7075 3.5wt% NaCl A-16 (6 9E-5 M P a W s ) —A— A-15 (6.8E-5 MPafm/s) 40 60 Time (hrs) 80 Stress Intensity (MPaffn) 25 Crack Velocity (m/s) 1E-07 5E-08 3E-08 2E-08 1E-0B 5E-09 -3E-09 -2E-09 -1E-09 A-16 —A— 40 60 Time (hrs) 6 8 10 12 14 16 18 Stress Intensity (MPa)rn) Figure 5.9 Rising-K experiments at a K-rate of -l.lxlO^MPaVm/s Crack Advance (mm) 5 AA-7075 3.5wt% NaCl A-II (1.2E-4 MPafWs) —A— A-22 (LIE-4 MPa(m/s) A-36 (1.1E-4 MPa(m/s) A A A A . 40 60 Time (hrs) 80 100 Stress Intensity (MPa(m) 25 40 60 Time (hrs) Crack Velocity (m/s) 1E-07 6 8 10 12 14 Stress Intensity (MPafm) 20 Figure 5.10 Rising-K experiments at a K-rate of ~1.4xlO^MPaVrn7s Crack Advance (mm) 5 i AA-7075 3.5wt% NaCl A-10 (1.4E-4 M P a W s ) (1.4E-4 MPatnVs) 20 40 60 Time (hrs) Stress Intensity (MPafm) 25 40 60 Time (hrs) Crack Velocity (m/s) 1E-07 6 8 10 12 14 Stress Intensity (MPafrri) 20 Figure 5.11 Crack velocities obtained under different monotonic slow loading conditions 75 Slow Loading Tests Crack Velocity (m/s) 1E-07 5E-08 3E-08 2E-08 1E-08 1E-09 A-17 (2.3E-6 MPafm/s) A A-70/A-93 (7.5E-6 MPafm/s) • A-3/A-13/A-43 (1.3E-5 MPafm/s) o A-50 (2.6E-5 MPafm/s) •* A-55 (3.7E-5 MPafm/s) AA-7075 3.5wt% NaCl 6 8 10 12 14 Stress Intensity (MPafm) 16 18 20 76 5.1.2 Bolt Loading and Constant Load Experiments (AA-7075) Very long testing times were necessary, in order to obtain data at low stress intensities, therefore, most experiments were conducted at intermediate and high stress intensities. Table 5.3 summarizes the results of all bolt loading tests. Specimen A-SI had a pop-in precrack and the stress intensity was decreasing, due to crack advance, from K I C down to about 11 MPaVm. All other specimens had fatigue precracks and were bolt loaded to a preselected stress intensity. For these specimens, shorter testing periods were chosen and the covered stress intensity range due to crack advance was shorter. Consequently, it was necessary to use several specimens to cover a larger range of stress intensities. Table 5.3 Results from bolt loading tests AA-7075, 3.5wt% NaCl Specimen ^iniL Aa da/dt dK/dt No average average [mm] [MPaVm] [mm] [MPaVm] [m/s] [MPaVm/s] A5 68.98 21.34 4.66 18.00 1.5xl08 7.8xl0'6 A7 64.88 17.50 7.82 14.10 8.2xl0"9 3.3xl0"6 A8 65.43 11.00 2.25 10.30 2.2xl09 6.9xl0-7 A12 63.94 14.30 3.65 13.00 3.6xl0"9 1.2xl0'6 A26 64.13 8.60 <0.30 8.60 <8.5xlO"u <3xl0"8 A-SI 44.29 12.10 6.30 9.70 2.0xl0"9 7.5xl0'7 (A-SI) 30.39 22.00 20.20 9.70 Figure 5.12 shows a K-log(da/dt) plot containing results from several bolt loaded tests with different amounts of crack advance during a test. At the higher stress intensities crack velocities were slightly faster on specimen with a fatigue pre-crack (A5, 77 A7). At the lower stress intensities no significant difference between the two types of specimen could be measured. It should be noted that specimen A7 was inverted in the solution with respect to the usual testing procedure so that the crack advance occurred upwards. This was done to determine whether gravity had any effect on the movement of the local solution within the crack (eg. as would be expected if a local change in solution chemistry caused density changes and/or trapping of hydrogen gas bubbles occurred). However, data from all tests with a fatigue pre-crack lay within a scatter band and therefore, cracking direction did not appear to influence crack velocities significantly. A comparison of data from bolt loaded tests with those from increasing load tests showed that the average crack velocities of the bolt loaded tests were comparable to velocities measured with increasing load tests. The scatter of test data overlapped, as shown in Figure 5.13 where slow loading tests with K-rates smaller than 1.5xl0"5 MPaVm/s are plotted together with the bolt loading results. Only a few tests were performed under constant load conditions. Specimens A32 and A63 were quickly loaded to the specified load and then the load was only periodically adjusted if a decrease of more than about 20 N but less than 50 N was observed. Specimens A17 and A70, on the other hand, were tested with the slow loading technique and an approximately constant load condition was maintained by the change in specimen compliance with crack propagation, as described earlier (see Section 4.6). The specimen with periodic load, adjustments (<50N) gave lower crack velocities relative to those obtained with the slow loading set-up. However, the difference was small as can be seen in Table 5.4. 78 Figure 5.12 Crack velocities obtained under bolt loading conditions with different amounts of crack advance Bolt Loaded Tests Crack Velocity m/s) 1E-07 5E-08 3E-08 2E-08 1E-08 5E-09 3E-09 2E-09 1E-09 -AA-7075 - A-SI - a(i),a(f):30.4mm,50.6mm • A-5 3.5wt% NaCl - a(i),a(f):69.0mm,75.5mm A-7 a(i),a(f):64.9mm,72.7mm • A-12 -a(i),a(f):63.9mm, 67.6mm o A-8 a(i),a(f):64.4mm,67.7mm A-26 a(i),a(f):64.3mm,64.6mm • • u -a(i):initial crack length a(f):final crack length , I , I , T 1 , 1 , 1 , 1 , 1 , 1 , 1 , 0 2 4 6 8 10 12 14 16 18 20 Stress Intensity (MPafm) Figure 5.13 Crack velocities obtained under slow loading and bolt loaded conditions Slow Loading and Bolt Loaded Tests Crack Velocity (m/s) 1E-07 5E-08 3E-08 2E-08 1E-08 5E-09 3E-09 2E-09 1E-09 AA-7075 3.3wt% NaCl A-17 (2.3E-6 MPatm/s) A A-70/A-93 (7.5E-6 MPa-fm/s) • A-3/A-13/A-43 (1.3E-5 MPafm/s) O A-SI/A-5/A-7/A-12 (Bolt loaded) 6 8 10 12 14 Stress Intensity (MPafm) 20 80 Table 5.4 Results from constant load tests AA-7075, 3.5wt% NaCl Specimen No K-init. Aa da/dt average dK/dt average [mm] [MPaVm] [mm] [MPaVm] [m/s] [MPaVm/s] A321J A631} 64.71 65.99 13.3 17.4 4.12 3.88 14.1 18.3 4.4xl0-9 1.1x10"' 8.5xl0"7 2.6xl0"6 A172) A702) 73.68 74.73 12.2 16.4 11.63 6.88 13.7 17.6 7.7xl09 2.0xl0"8 9.9xl0'7 3.7xl0"6 Load periodically adjusted Constant displacement rate test 5.2 Monotonic Loading Experiments (Alloy AA-8090) 5.2.1 Slow Loading (Rising-K) Experiments (AA-8090) Alloy 8090 exhibited high resistance to SCC in sodium chloride solutions, consistent with the behaviour reported in the literature. With rising-K experiments, it was found that essentially no SCC occurred except at high stress intensities close to K I C . Furthermore, SCC could not be obtained in carbonate solutions when tested at a K-rate of about 4xl0"5 MPaVm/s. Tests in carbonate solutions were performed because Craig et al [90] found fast cracking in this environment with smooth bend specimens. Experiments with faster K-rates were not attempted because a very short time would have been spent in the stress intensity range where SCC occurred, making it impossible to obtain a meaningful crack velocity. 81 A very slow K-rate test at 4.2x10"6 MPaVm/s was performed with specimen Li-16 in sodium chloride solution. The stress intensity was increased from about 2 MPaVm up to 14 MPaVm/s. In this specific test, the crack length was not monitored during the experiment and only the total crack advance was measured after the test on the exposed fracture surface. Assuming that no significant cracking occurred up to 12 MPaVm, an average crack velocity of 5xl0"9 m/s resulted. (This assumption is based on results from long term bolt loading tests where no significant cracking was measured even at 12.9 MPaVm). All slow loading data for alloy 8090 are given in Table 5.5. Table 5.5 Results from slow monotonic loading tests AA-8090, 3.5wt% NaCl Specimen No Kinit Aa Kfinal Test duration dK/dt average [mm] [MPaVm] [mm] [MPaVm] [hrs] [MPaVm/s] Li-2 49.5 2.2 - 17.2 100 4.2 x 10-5 Li-16 50.98 1.9 1.76 13.9 795 4.2 x 10"* Li-4 50.5 2.4 - 17.2 110 3.7 x 10-5 Li-3 48.7 2.0 - 16.5 110 3.7 x 10"5 Specimen Li-2 and Li-16 were tested in aqueous 3.5 wt.% NaCl solution Specimen Li-3 was tested in 0.1MLi2CO3 + HC1 at pH = 10.5 and Li-4 was tested in 3.5 wt.% NaCl + 0. lMLi 2 C0 3 + HC1, pH = 10.5 5.2.2 Bolt Loading and Constant Load Experiments (AA-8090) Very slow cracking rates were obtained with bolt loaded and constant load experiments. These tests produced very small crack growth increments and the relatively 82 rough fracture surface made it difficult to measure the small amount of crack advance. The observation was consistent with the slow loading tests where cracking could only be measured with very low K-rates. When average crack velocities were calculated for specimen tested at stress intensities >12.7 MPaVm, values smaller than 2xl0"10 were obtained. These crack velocities are much smaller than a value obtained with a slow loadingtest. However it was difficult to measure the small increments of crack advance in both cases and therefore, no accurate values of crack velocities could be determined. Table 5.6 lists the crack advance data of these specimen obtained after relatively long testing periods (several hundred hours). Table 5.6 Results from bolt loaded and constant load tests AA-8090, 3.5wt% NaCl Specimen ainit Kinit Aa Test dK/dt No duration average [mm] [MPaVm] [mm] [MPaVm] [hrs] [MPaVm/s] Li-14 48.5 12.7 <0.3 12.7 343 Li-17 47.8 9.4 - 9.4 744 Li-18 49 12.9 0.5 13.0 744 «5xl0' 8 Specimen Li-14 was tested under constant load conditions Specimen Li-17 and Li-18 were bolt loaded. 5.3 Cyclic Loading Experiments (AA-7075 and AA-8090) High frequency (30 Hz) corrosion fatigue tests were performed with both alloys in sodium chloride environments. The alloy 8090 was tested at a R-ratio of 0.75 and the alloy 7075 at R-ratios of 0.78 and 0.31. The stress intensity range investigated was around stage II where striated fatigue fracture occurs and/or Paris behaviour is observed. 83 Both alloys displayed a similar dependence of cyclic cracking rate on the stress intensity range, log(AK), and on the maximum cyclic stress intensity, K ^ . A typical corrosion fatigue plot of log(AK) versus log(da/dN) is shown in Figure 5.14. In order to compare CF data with SCC data, a plot of the maximum cyclic stress intensity (K^J versus real time cracking rates (crack velocities) log(da/dt) is displayed in Figure 5.15. At the lower frequencies, tests were performed with two specific stress intensity ranges (AK) on alloy 7075 and only one stress intensity range with alloy 8090. Figure 5.14 CF tests at 30 Hz with a sinusoidal load wave form, plot of log(AK)-log(da/dN) Crack 1E-05 Propagation Rate, Log(da/dN) (m/cycle) 5E-06 A-89 R=0.31 AA-7075 and AA-8090 3.5wt% NaCl 2E-06 A-86 R=0.78 2E-07 5E-07 1E-06 5E-08 1E-07 Li-26 R=0.77 . . . . A — 2E-08 1E-08 2 3 5 Stress Intensity Range, Log(AK) (MPa/m) 10 Figure 5.15 Comparison of CF and SCC crack velocities for different K , ^ values Crack Velocity (m/s) 1 E-04 1E-05 1E-06 1E-07 1E-08 1E-09 AA-7075 and AA-8090 3.5wt% NaCl A-SI — • — A-89 R=0.31 B A-86 R=0.78 — e — Li-26 R=0.77 0 J L 4 6 8 10 12 14 16 1 Stress Intensity, Kmax (MPafm) 5.3.1 Effect of Cyclic K-Rate (Alloy 7075) 85 The relationship between K-rate and either da/dt or da/dN are shown in Figures 5.16 to 5.19 for alloy 7075 tested in NaCl. As the K-rate was decreased from about 100 MPaVm/s (30 Hz test) to 10"2 MPaVm/s, a linear decrease of the real time crack velocity, log(da/dt), was measured. A decrease of K-rate over four decades was accompanied by a corresponding decrease in crack velocity (da/dt) of approximately four decades. If the cracking rate was expressed as crack advance per cycle (da/dN), only about a four fold increase in cyclic crack propagation rate was measured over the same K-rate range. At a R-ratio of 0.31 (Figure 5.16) crack velocities decreased with decreasing K-rate to a minimum around lxlO'3 MPaVm/s and then slowly increased to a crack velocity typical of constant load or slow loading tests at a K-rate of lxlO"5 MPaVm/s. The behaviour at a R-ratio of 0.78 was more complex (Figure 5.18). Real time crack velocities decreased with decreasing K-rate to a minimum near lxlO"2 MPaVm/s , rose to a maximum at 2xl0"3 MPaVm /s, decreased to a minimum at 2x10"" MPaVm/s and then slowly increased again. Thus R-ratio had a major effect on crack rate behaviour in the K-rate range between ~10'2 and ~10"3 MPaVm/s. Figures 5.16 to 5.19 also contain CF data in dessicated air and a sodium chloride solution that contained a chromate corrosion inhibitor. Cracking rates in air were approximately 10 times lower than in the aqueous solutions at K-rates above 10"1 MPaVrn/s. However, a measurement taken at 10"3 MPaVm/s gave no visible cracking and cracking rates, therefore, were several orders of magnitude smaller. The addition of a corrosion inhibitor had no influence on crack propagation rates at high K-rates. At K-rates below 10"1 MPaVm/s, cracking rates were slower in the inhibited solution by 86 about a factor of 2. No inhibitor tests were done at a R-ratio of 0.31. For comparison, two tests were performed with a value around 14.5 instead of 17.5 MPaVm but the same R-ratio. The two data points (A88 and A92) are shown in Figure 5.18. Even though the crack velocities lay below those obtained at the higher value they appeared to follow the same trend with very low crack velocities around K-rates of 10"2 MPaVrn/s and higher crack velocities at K-rates of 10"3 MPaVm/s. Additionally to the triangular load wave form, single tests were performed with a positive sawtooth and a square wave form at a R-ratio of 0.31. The K-rate with the sawtooth test was 1. lxlO"4 MPaVm/s and a real time crack velocity of 2.3xl0~9 m/s resulted. At a comparable K-rate with a triangular wave form, the crack velocity was l.lxlO"9 m/s. In both cases, crack propagation per cycle was approximately lxlO"4 m/cycle. This indicates that probably no crack propagation occurred during the unloading part of the cycle. The square load wave form gave a crack velocity of 2.9xl0"9 m/s, as compared to 8.7xl0'10 m/s for triangular loading and 4.4xl0"9 m/s for constant load tests at K,,^. Crack velocities during the maximum stress intensity part of the square wave form cycle are, therefore, slightly faster than constant load cracking rates at the same maximum stress intensity. However, the small difference may also be attributed to data scatter. All data from cyclic loading experiments with alloy 7075 are summarized in Tables 5.7 to 5.10. Crack velocities for different K-rates obtained with slow monotonic loading experiments can be compared with the data from cyclic loading experiments. It was shown in the slow loading tests that the crack velocity increased with increasing (K) and a specific value of (da/dt) could be assigned to a particular (K) value during the test. 87 However, during cyclic loading the (K) value changes during the cycle and usually only one average value of (da/dt) is obtainable for the stress intensity range between K,,^ and Figure 5.20 shows results from tests in sodium chloride solution with both monotonic and cyclic loading. The real time cracking rates for the monotonic loading experiments are represented by velocity bars, instead of single data points. The bar describes the range in crack velocities obtained at stress intensities corresponding to and K ^ of the cyclic loading experiments. The bar is not indicative of data scatter, but simply a recognition of the fact that a single unique velocity could not be assigned to each monotonic K-rate test. Figure 5.16 Real time crack velocities at different cyclic K-rates for R=0.31 Cyclic Loading Tests Crack Velocity (m/s) 1E-05 1E-06 1E-07 1E-08 1E-09 1E-10 AA-7075 R=0.31 ,AK=9.5MPa-fm 3.5wt% NaCl B Dry Air O 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 1E+01 1E+02 1E+03 K-Rate dK/dt (MPa/rrT/s) 88 Figure 5.17 Cyclic crack propagation rates at different cyclic K-rates for R=0.31 Cyclic Loading Tests Crack Propagation per Cycle da/dN (m/cycl.) 1E-02 1E-03 1E-04 1E-05 1E-06 1E-07 1 E-.08 3.5wt% NaCl — B — Dry Air o AA-7075 -o-JL J-1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 1E+01 1E+02 1E+03 K-Rate dK/dt (MPa/fn7s) 89 Figure 5.18 Real time crack velocities at different cyclic K-rates for R=0.78 Cyclic Loading Tests Crack Velocity (m/s) 1E-05 1E-06 1E-07 1E-08 1E-09 : AA-7075 - R = 0.78,AK = 3.9 MPafm : ( * R - 0 . 7 7 , A K W 3.3 MPafm) 1E-10 3.5wt% N a C U Na2Cn0 4 3.5wt% N a C l • Dry Air '-Q 3.5wt% N a C l 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 1E+01 1E+02 1E+03 K-Rate dK/dt (MPafm/s) 90 Figure 5.19 Cyclic crack propagation rates at different cyclic K-rates for R=0.78 Cyclic Loading Tests Crack Propagation per Cycle da/dN (m/cycl.) AA-7075 R = 0.78.A.K = 3.9MPa1m~ - 3.5wt% NaCU Na 2Cr04 \ . A . . . S \ .. 3.5wt% NaCl —e— \ \ \ \ Drj^Air -5 - — A -1 i i G © i i i i i 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 1E+01 1E+02 1E+03 K-Rate dK/dt (MPa1m/s) 91 Figure 5.20 Crack velocities for different K-rates obtained under monotonic and cyclic loading Cyclic Loading and Monotonic Loading Tests Crack Velocity (m/s) 1E-07 1E-08 1E-09 1E-10 AA-7075 3.5wt% NaCl Cyclic Loading Monotonic Loading (R = 0.78.AK = 3.9MPap)(Region II velocities) — B — — x — 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 K-Rate dK/dt (MPafm/s) ' Table 5.7 Results from cyclic loading at different K-rates AA-7075, R=0.78, 3.5wt% NaCl Specimen Aa Rtverigo da/dt da/dN dK/dt No average average average [mm] [MPaVmj [mm] [MPaVm] [m/s] [m/cycle] [MPaVm/s] A76 64.99 3.79 1.86 3.90 0.778 4.4x10-* 1.6xl0"7 2.2X102 A61 65.62 3.82 2.33 3.95 0.779 2.5x10-" 7.9xl0"7 2.4x10' A62 65.61 3.83 1.39 3.90 0.778 4.0x10' 7.7xl0"7 4.0xlO'2 A52 65.69 3.83 1.15 3.89 0.779 2.2x10-' 7.6xl0"7 2.2x10* A64 65.27 3.80 0.41 3.83 0.779 7.8xlO"M 6.6xl0"7 9.1X10"3 A59 66.28 3.86 2.88 4.01 0.779 5.8x10' 2.1x10s 2.1x10-' A45 65.01 3.79 5.04 4.06 0.779 8.5x10' 5.6xl0"3 1.2x10'' A66 65.04 3.80 2.81 3.95 0.778 4.7x10-' 4.8xl0"5 7.7x10^  A44 64.71 3.77 1.43 3.85 0.779 2.4xl0'9 5.3xl05 3.5X10"4 A51 63.20 3.69 1.86 3.79 0.779 2.7x10' 1.4x10-* 1.4x10^  A91" 64.68 4.04 0.96 - 0.767 5.4x10-' 9.6x10^  4.5xl0"5 A93" 68.03 3.80 10.61 - 0.779 7.5x10"' l.lxlO 2 5.5X106 88a 67.24 3.39 0.20 3.39 0.769 3.3xlO-10 2.9xl0"7 7.7xl0"3 922) 65.81 3.32 1.38 3.38 0.769 3.1x10' 1.5x10-' 1.4x10-' Single cycle experiment Lower AK value Table 5.8 Results from cyclic loading at different K-rates AA-7075, R=0.78, 3.5wt% NaCl + 0.01M Na2Cr04 Specimen A K ^ Aa AKn„j Revenge da/dt da/dN dK/dt No average average average [mm] [MPaVm] [mm] [MPaVm] [m/s] [m/cycle] [MPaVm/s] A75 64.64 3.77 2.56 3.91 0.779 4.3xl06 1.4xl0"7 2.3xl02 A73 66.08 3.85 6.94 4.21 0.779 3.2x10-" 8.6xl0"7 3.1x10"' A69 65.46 3.81 0.72 3.85 0.779 2.7x10-' 4.0xl0"7 5.2xl0"2 A85 65.39 3.81 0.57 3.85 0.778 4.6xl010 4.0xl0"7 8.7x10° A74 67.60 3.93 0.73 3.98 0.778 1.0x10"' l.lxlO"6 7.8x10"' A68 64.49 3.76 2.40 3.90 0.778 4.4x10' 2.6x10s 1.3x10"' A82 66.50 3.87 0.80 3.92 0.778 7.7xl0"10 1.2x10s 5.1x10^  Table 5.9 Results from cyclic loading at different K-rates AA-7075, R=0.78, dry air Specimen Aa da/dt da/dN dK/dt No average average average [mm] [MPaVm] [mm] [MPaVm] [m/s] [m/cycle] [MPaVm/s] A79 64.79 3.77 1.60 3.86 0.779 8.0xl0"7 2.7xl0"8 2.3X102 A78 67.54 3.92 0.39 3.95 0.779 2.4x10-' 5.2X10-11 3.7x10' A94 66.10 3.85 <0.10 3.85 0.779 <8.3xlff" <4.9xl0"7 1.3xl03 Table 5.10 Results from cyclic loading at different K-rates AA-7075, R=0.31, 3.5wt% NaCl and dry air Specimen A K ^ Aa AKfto.1 R IX«vermge da/dt da/dN dK/dt No average average average [mm] [MPaVm] [mm] [MPaVm] [m/s] [m/cycle] [MPaVm/s] A89 78.91 9.70 2.35 9.93 0.306 2.4x10"3 7.8xl0"7 5.9xl02 A24 66.34 9.44 3.46 9.89 0.306 3.8x10-" 1.9X10"6 3.9x10"' A25 64.09 9.13 2.63 9.49 0.306 8.3x10"' 2.1xl0"6 7.4x10-2 A20 65.53 9.33 1.74 9.56 0.306 1.5x10"' 2.3x10"* 1.2xl0"2 A21 66.62 9.47 1.52 9.68 0.306 1.1x10"' 1.2xl0"3 1.7xl0"3 A31 64.11 9.13 1.78 9.38 0.306 8.7x10'° 3.1x10s 5.2X10-* A39 66.21 9.42 1.94 9.67 0.306 1.1x10"' 1.8X10"4 1.2X10-4 A37 63.70 9.08 1.00 9.22 0.306 9.6xl0"10 1.7x10^  l.lxlO"4 A33° 65.34 9.64 1.75 0.298 2.4x10"' 1.3xl0"5 A412' 65.19 1.26 9.20 0.312 2.1x10"' 1.5xl0"5 A33/413' 65.34 9.64 3.01 9.20 0.305 2.3x10"' 3.0xl0"3 1.4x10s A19"' 66.32 9.43 2.03 9.7 0.306 2.4x10-9 2.0X10"4 1x10^ * A71 5 ) 67.56 9.60 1.60 9.81 0.306 2.9x10"' 1.6x10"* A876> 66.92 9.52 0.94 9.64 0.306 3.1xl0"6 9.4xl0"8 6.3xl02 Single loading half cycle Single unloading half cycle A3 3 and A41 combined to a single full cycle Positive sawtooth load wave form (24 hrs/cycle) Square load wave form (15.2 hrs/cycle) Test in dry air 2) 3) i) 5) 6) 5.3.2 Effect of Cyclic K-Rate (Alloy 8090) 94 At K-rates above 10'1 MPaVm/s, crack propagation per cycle was almost independent of K-rate in the NaCl solution. Therefore, in terms of real time, the crack velocity decreased as the K-rate was decreased. Below about 10"1 MPaVm/s, the cyclic crack propagation rate (da/dN) increased but not as dramatically as observed with the alloy 7075. Real time crack velocities at K-rates slower than 10"2 MPaVm/s were below 10"9 m/s. Cracking rates in dry air were only measured at the highest K-rate and were only about a factor of 3 slower than in the aqueous solution. Since monotonic loading experiments showed stress corrosion cracking only at stress intensities above about 12 MPaVm, it was difficult to compare data from both testing methods because cyclic tests were performed between about 10 and 13 MPaVm. However, it can be assumed that cyclic crack velocities at K-rates below 10s MPaVm/s are smaller than 10"9 m/s. Data for all cyclic loading tests with alloy 8090 are presented in Table 5.11. The dependence of real time crack velocities (da/dt) and cyclic crack propagation rates (da/dN) on the K-rate are displayed in Figures 5.21 and 5.22. Table 5.11 Results from cyclic loading at different K-rates AA-8090, R=0.75, 3.5wt% NaCl and dry air Specimen Aa A K , ^ a^verage da/dt da/dN dK/dt No average average average [mm] [MPaVm] [mm] [MPaVm] [m/s] [m/cycle] [MPaVm/s] Li 22 46.50 3.03 3.03 3.26 0.750 5.1x10-* 1.6xl0-7 1.9xl02 Li 12 47.08 3.08 1.80 3.21 0.750 6.9x10-' 8.7xl0'8 5.0x10-' Li 11 47.24 3.10 1.62 3.21 0.750 3.1x10"' 2.4xl07 8.4X10"2 Li 15 48.86 3.21 1.18 3.29 0.750 1.5x10-' 8.1xl0"7 1.2xl0"2 Li 19 47.40 3.11 1.23 3.19 0.750 5.9x10-'° 9.3xl0-7 4.0xl03 Li 24 46.86 3.06 1.09 3.29 0.750 7.3x10'° 2.7x10* 1.7X103 Li 21 49.20 3.23 1.21 3.32 0.750 5.2x10'° 1.3xl0"5 2.7X10"4 Li 23" 46.86 3.06 1.72 3.19 0.750 1.2x10"* 3.8x10-" 1.9xl02 Tested in dry air 95 Figure 5.21 Real time crack velocities at different cyclic K-rates for R=0.75 Cyclic Loading Tests Crack Velocity (m/s) 1E-05 1E-06 -1E-07 1E-08 1E-09 1E-10 AA-8090 R = 0.75.AK = 3.2MPafm 3.5wt% NaCl B Dry'ATr" O -L 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 1E+01 1E+02 1E+03 K-Rate dK/dt (MParm/s) Figure 5.22 Cyclic crack propagation rates at different cyclic K-rates for R=0.75 Cyclic Loading Tests Crack Propagation per Cycle da/dN (m/cycl.) 1E-02 1E-03 r 1E-04 -1E-05 -1E-06 1E-07 1E-08 AA-8090 R = 0.75.4K = 3.2MPa7m 3.5wt% NaCl B Dry air O 1E-06 1E-05 1E-04 1E-03 1E-02 1E-01 1E+00 1E+01 1E+02 1E+03 K-Rate dK/dt (MPafm/s) 5.4 Electrochemical Potential Measurements .97 Anodic polarization curves of alloys 7075 and 8090 that were obtained by scanning the potential from -1200 mV to -600 mV with respect to a saturated calomel reference electrode showed passive behaviour in the sodium chloride solution. The measured corrosion currents in the passive region were between 10'2 to 10"1 A/m2. Additions of the chromate corrosion inhibitor to the test solution decreased the corrosion currents in the passive region only slightly. Polarization curves of both alloys are shown in Figure 5.23. The free corrosion potential recorded during the SCC experiments in sodium chloride solutions lay between about -750 mV and -820 mV for tests with the 7075 alloy and between -820 mV and -860 mV for tests with the alloy 8090. Fluctuations of the corrosion potential were partly caused by the cleaning of the specimen surface around the crack tip for crack length measurements. Tests in solutions of sodium chloride with additions of sodium chromate gave free corrosion potential values between -760 mV S C E and -780 mV S C E . Only the alloy 7075 was tested in this solution. No significant difference in corrosion potential behaviour was observed between monotonic loading experiments and cyclic loading tests at a R-ratio of 0.78. Consequently, no correlation between individual load cycles and corrosion potential fluctuations were evident. At a R-ratio of 0.31, alloy 7075 showed quite distinct corrosion potential behaviour for different cyclic loading rates. At K-rates higher than 10"2 MPaVm/s the corrosion potential stayed more or less constant during the entire experiment. Fluctuations related 98 to individual loading cycles were smaller than 5 mV. At K-rates around 10"3 MPaVm/s, each load cycle was matched with a well defined potential fluctuation of approximately 30 mV. As the load increased the potential decreased and vice versa as shown in Figure 5.24. The fluctuation were not observed in the early stages of an experiment. They occurred only after a period of approximately 100 hours. Single potential fluctuations could be described as having a trapezoidal shape, i.e. the potential was more or less constant during periods of higher loads and increased only when the load was around the minimum value. At a K-rate of 5X10"4 MPaVm/s, potential fluctuations became larger than at 10"3 MPaVm/s and were often bigger than 50 mV. Also, the shape of the potential fluctuations was no longer as well defined. Even so, maximum values could usually be matched with minimum load values. An example is shown in Figure 5.25. When the K-rate was reduced even lower, fluctuations in potential became irregular and maximum potential values no longer matched minimum load values. They often appeared during the loading or unloading part of a cycle as seen in Figure 5.26. Assuming the presence of crack tip closure, the difference between tests at low and high R-ratios can be explained by a combination of crack closure effects and mixed potential theory. When the load was approaching minimum values at the lower R-ratio the crack tip closed because of the low stress intensity ( K ^ . This then reduced the active anodic area near the crack tip, which reduced the total anodic current relative to the total available cathodic current, and raised the potential to more noble values. Hereby it is assumed that anodic processes occurred mainly at the crack tip, whereas the crack flanks and exterior surface served as sites for the cathodic reaction. At the R-ratio of 0.78, on the other hand, the crack tip did not close (K^,, was higher) and, therefore, no periodic changes in area of anodic or cathodic sites took place. 99 At very slow K-rates the closing of the crack tip was also very gradual and, therefore, no sudden change in size of anodic to cathodic site took place. This then did not produce well defined periodic potential fluctuations. The COD curves in Figures 5.25 and 5.26 support the crack tip closure effect. Near 5,^ the slope of the 8 versus time curve changes (rounded tip at reversal point, arrowed) which indicates restricted closing at the load line which is caused by a closed crack tip above K ^ . This closure was not observed at the beginning of the experiment but only after several hundred hours of cycling (Compare Figure 5.27 a) and b)). Therefore, either corrosion product build-up or the stepped nature of the crack path produced during CF were responsible for the crack closure. The condition for hydrogen evolution via the cathodic reduction of H* ions, 2H* + 2e" = H 2 , is given by the Nernst equation [Eq. 5.1]. At 25° (298K) and one atmospheric pressure the reversible potential, E ^ /H 2 , can be calculated as Er r / H 2 = - 2 - 3 0 3 RT(pH)/F,VS H E (Eq. 5.1) where; R = Gas constant (8.314 JK'1 mole"1) T = Temperature (K) F = Faraday constant (96500 C) V S H E = Potential with respect to the standard hydrogen electrode With 3.5wt% NaCl the pH in the bulk solution was 6.5. The addition of a chromate corrosion inhibitor raised the pH to 8.4. Equation 5.1 shows that hydrogen evolution is thermodynamically possible at potentials below -382 mV S H E at pH 6.5 and below -493 mV S H E at pH 8.4. With respect to the saturated calomel electrode, V S C E , the reversible 100 potential is -626 mV S C E and -737 mV S C E respectively using the conversion mVscE = mV S H E - 244 mV. All potentials recorded under freely corroding conditions were, therefore, below the hydrogen evolution condition defining the stability regime of water. In other words, water reduction could take place with the evolution of hydrogen. This, therefore, lends credence to the wide spread opinion that environment assisted cracking of aluminum alloys in aqueous solutions is in part attributed to hydrogen embrittlement. Figure 5.23 Anodic polarization curves for AA-7075 and AA-8090 in 3.5wt% NaCl with and without chromate corrosion inhibitor Potential Current Density AA-7075: 3.5wt% NaCl AA-7075: 35wt% NaCl • 0.01M Nc^CrO^ (pH= 8.4) AA-8090: 35wt% NaCl Figure 5.24 Corrosion potential fluctuations at a K-rate of 10'3MPaVm/s, AA-7075, 3.5wt% NaCl max mm -700 mV. SCE -800 mV. SCE Figure 5.25 Corrosion potential fluctuations at a K-rate of SxlO^MPaVrn/s max ^min A -V-- 7 0 0 m V S C E 1-800 mV S C f£ 20hrs Figure 5.26 Corrosion potential fluctuations at a K-rate of lO^MPaVrn/s • 7 0 0 m V S C E •800mV, SCE Figure 5.27 Corrosion potential fluctuations at a K-rate of SxlO^MPaVm/s, a) beginning of experiment, b) after several hundred hours 5.5 Fractography Alloy-7075 103 The following four major types of crack fractography (i, ii, iii and iv) were usually present on the SCC and CF fracture surfaces. These regions are labeled as A, B, C and D in Figure 5.28. The macroscopic crack propagation direction (MCPD) is on most micrographs from bottom to top. i) Type A fractography - Large fracture regions lying parallel to the rolling plane with a rippled appearance. These regions occupied the major part of the fracture surface and were identified as being intergranular. ii) Type B fractography - Small featureless regions that did not show the ripples, where fracture may have been of transgranular nature. iii) Type C fractography - Rough steps, usually oriented perpendicular to the rolling plane that connected the intergranular regions where the cracks were propagating on different levels. The crack path within these steps was mainly transgranular. iv) Type D fractography - Areas containing larger precipitates and debris from fractured particles. The crack path was probably both transgranular and intergranular in these regions. • -At lower magnifications, fracture surfaces obtained under the different loading conditions appeared identical. Differences could only be found at higher magnification. A change in fractography from intergranular to striated intergranular was observed when the K-rate was raised above about 10'2 MPaVm/s. This corresponded to the transition from K-rate dependent crack advance per cycle to an almost constant crack advance per cycle at the higher K-rates. 104 Figure 5.28 Typical S E M micrograph displaying the four major types of fractography: A, B, C and D AA-7075, Specimen A76, Mag.: 400x, MCPD bottom to top The fractography produced by overload failure and fatigue precracking was always quite different from fractography produced by either SCC or CF. Overload failure exhibited a very rough fracture surface caused by interfacial separation between the matrix and larger precipitates. Flat featureless regions, typical of the SCC and CF fracture surfaces, could also be found in small areas. Figure 5.29 is a S E M micrograph taken at the tip of a CF crack which was opened after the test by mechanical overload. The transition from CF to mechanical overload is easily recognized. 105 Figure 5.29 Transition from CF to overload failure AA-7075, Specimen A76, Mag.: 150x, MCPD bottom to top Fatigue precracking at low K,,,^ values, on the other hand, produced very flat fracture surfaces consisting of a mixed intergranular/transgranular crack path. Figure 5.30 shows the transition region from a fatigue pre-crack produced in air to a CF crack produced in sodium chloride solution. A faint line indicates the crack front of the fatigue pre-crack. It can also be seen that CF fractography was affected by precipitates which gave rise to a rough fracture surface in some areas. This helped in determining the location of the fatigue pre-crack/CF transition on the fracture surface at low magnification. 106 Figure 5.30 Transition from fatigue pre-crack produced in air to a CF crack produced in 3.5wt% NaCl AA-7075, Specimen A76, Mag.: 150x, MCPD bottom to top 5.5.1 Intergranular Fracture Regions The major part of the fracture surface showed "flat" intergranular fractography. Figure 5.31 is a front view of the crack path on a polished and etched surface, normal to the crack propagation direction. The intergranular crack path is clearly visible. Additionally, one can see transgranular steps that connect segments of intergranular fracture that were propagating on different levels parallel to the rolling plane. The specimen in Figure 5.31 was tested under cyclic loading at a K-rate of about 3x10"1 MPaVm/s. However, specimens tested at much lower K-rates showed the same crack path. 107 Figure 5.31 Front view of intergranular crack path with transgranular steps AA-7075, Specimen A72, Mag.: 450x 5.5.1.1 Low K-rate Intergranular Fractography In general, the intergranular fracture regions showed a very similar appearance up to K-rates of about 10"2 MPaVm/s. However, at the low K-rates and with long test durations, corrosive attack occurred on the fracture surface producing numerous shallow pits and intergranular crevices. Tests performed in sodium chloride with chromate inhibitor exhibited fracture features that were much less affected by corrosion effects. The faint rippled appearance of the intergranular fracture regions visible in Figure 5.28 was verified by SEM micrographs of polished and etched surfaces, that were 108 perpendicular to the rolling plane and parallel to the crack propagation direction. An example is shown in Figure 5.32. It can be seen clearly that the grain boundaries are not flat but show a slight wavy pattern. The deeply etched surfaces also revealed that some of the larger grains were composed of many smaller subgrains of a size comparable to the ripple spacing. Photomicrographs taken on polished and etched planes parallel to the rolling plane also revealed fine subgrain boundaries in some areas. The diameter of these subgrains was only a few micrometers, with small precipitates located at the grain boundaries, as shown in Figure 5.33. Figure 5.32 SEM micrograph of rippled grain boundary with small subgrains, polished and etched surface normal to T direction AA-7075, Mag.: lOOOx ^ «s 109 Figure 5.33 Faint subgrain boundaries (circled) within larger grains, polished and etched surface parallel to the rolling plane AA-7075, Mag.: 440xi In some tests, at a R-ratio of 0.78, high magnification SEM micrographs revealed micro-tear ridges on the intergranular fracture surface, indicating ductile behaviour of the grain boundary region. Precipitates were present on both sides of the opposing matching intergranular fracture surfaces with tear ridges or dimples around them. Often, more than one precipitate was contained in a single dimple. The tear ridges on intergranular fracture surfaces were most pronounced at K-rates near 10'2 MPaVm/s, which were slightly lower than the K-rates necessary to produce intergranular striations. They were also found on fracture surfaces produced under monotonically loaded specimen where stress intensities were near K ^ (corresponding to K ^ of the cyclic loading tests). A typical high magnification SEM micrograph of matching micro-tear ridges on opposing intergranular fracture surfaces is shown in Figure 5.34. 110 Figure 5.34 SEM micrograph of matching micro tear ridges on opposing intergranular fracture surfaces AA-7075, Specimen A74, Mag.: 13000x, R=0.78, K-rate=7.8xl0"3 MPaVm/s i -L_J 5.5.1.2 High K-rate Intergranular Fractography At loading rates above 10"2 MPaVm/s, it was possible to find faint striations on the intergranular fracture surface. Measurements of the striation spacing matched the macroscopic crack advance per cycle of about 0.1 to 0.2 u.m. The height of these striations is thought to be very small because prolonged observation of a specific area with the SEM caused contamination of the specimen surface which then made it difficult to resolve single striations. An example of the striations on an intergranular fracture surface is given in Figure 5.35. In addition to the fine striations, widely spaced ridges oriented at an angle to the striations could also be observed, as shown in the matching pair of opposing fracture surfaces in Figure 5.36. In this case, the matching fracture Ill surfaces proved that the wider ridges were interlocking and, therefore, were produced at the time of initial material separation at the crack tip. The finer striation pattern matched each other on the opposing fracture surfaces, suggesting they were associated with events at the crack tip, but could not be categorized as interlocking steps due to the limited resolution of the photomicrograph. Striations of similar appearance could also be found at the crack tip close to the overload failure. However, the opposing fracture surfaces in Figure 5.37 indicate that these striations do not match exactly. Therefore, they may not have been formed during the initial metal separation process but probably formed afterwards on the two crack flanks due to plastic deformation. Figure 5.35 SEM micrograph of intergranular striations AA-7075, Specimen A76, Mag.: 3000x, R=0.78, K-rate=2.2xl02 MPaVm/s, MCPD bottom to top 112 Figure 5.36 SEM micrograph of opposing fracture surfaces with interlocking ridges (arrowed) AA-7075, Specimen A76, Mag.: 6800x, R=0.78, K-rate=2.2xl02 MPa-v/m/s, MCPD bottom to top Figure 5.37 SEM micrograph of opposing crack tip fracture surface with striations produced during overload AA-7075, Specimen A76, Mag.: 5300x, R=0.78, K-rate=2.2xl02 MPaVm/s, MCPD bottom to top 5.5.1.3 Effect of R-Ratio and K ™ . on the Striated Fracture Path 113 At the R-ratio of 0.31, CF tests were usually performed at a K , ^ value near 13.5 MPaVm, whereas tests with an R-ratio of 0.78, were run at values of 17.5 MPaVm. However, at the highest K-rates investigated (2.2xl02 MPaVm/s for R=0.78 and 5.9xl02 MPaVm/s for R=0.31) experiments at several other AK and K ^ values were conducted with both R-ratios. (These tests produced data for the log(AK)-log(da/dN) and Kmax-log(da/dt) curves in Figure 5.14 and 5.15 on page 83 and 84. From Figure 5.14, it could be seen that cyclic crack propagation rates (da/dN) at a AK of about 4 MPaVm were near 2xl0"7 m/cycle for the R-ratio of 0.78, as well as for the R-ratio of 0.31. Therefore, the striation spacing for both R-ratios was also expected to be the same. SEM investigations of matching fracture surfaces confirmed this and also showed that the appearance of the intergranular striations was very similar for both R-ratios even though the values of 17.5 MPaVm and 5.3 MPaVm were considerably different. An example of a matching set of intergranular striations produced at a R-ratio of 0.31 and a AK of about 4 MPaVm is shown in Figure 5.38. (To be compared with Figure 5.35 at similar AK but R=0.78). At AK values higher than about 5 MPaVm, the crack path changed from intergranular to mainly transgranular. However, this could only be observed with a R-ratio of 0.31 because a R-ratio of 0.78 did not allow such high AK values to be used. The striated fracture surfaces at the high AK values also showed much larger amounts of ductile tear ridges which obscured the matching of individual striations. When a comparison was made between tests performed at the same K ^ values (11 114 MPaVm), but different R-ratios, it was obvious that the striation spacing was considerably larger at the R-ratio of 0.31 than at a R-ratio of 0.78. It was also quite difficult to view the very faint striations produced at the R-ratio of 0.78 because AK was only about 2.5 MPaVm SEM micrographs of intergranular striations produced at the same K ^ value but different AK values are shown in Figures 5.39 and 5.40. (Note difference in magnification). Figure 5.38 SEM micrograph of opposing intergranular surface with fatigue striations AA-7075, Specimen A89, Mag.: 8000x, R=0.31, K-rate=5.9xl02 MPaVm/s, MCPD bottom to top 115 Figure 5.39 Intergranular striations AA-7075, Specimen A89, Mag.: 4000x, R=0.31, K ^ l l MPaVm, K-rate=5.9xl02 MPaVm/s, MCPD bottom to top Figure 5.40 Intergranular striations AA-7075, Specimen A86, Mag.: lOOOOx, R=0.78 ,K^l l MPaVm, K-rate=1.6xl02 MPaVm/s, MCPD bottom to top 5.5.2 Fractography of Transgranular Steps 116 5.5.2.1 Transgranular steps at Low Cyclic K-rates The transgranular steps connecting the intergranular fracture regions displayed a marked difference in appearance depending on the K-rate at which they were produced. At K-rates below about 10"2 MPaVm/s these transgranular regions failed by ductile tearing with dimple formation due to microvoid coalescence. High magnification SEM observations showed that small precipitates of about 0.1 p:m size were associated with dimples, indicating that these particles nucleated the initial microvoids. Interestingly, precipitates were found only on one fracture surface and were not present on the opposing surface. This indicates possible failure along subgrain boundaries where coherency between the 0.1 lira precipitates and matrix was established with one subgrain and not with the other, resulting in different bonding strengths between the particle and the two subgrains. Figures 5.41 and 5.42 show the transgranular dimpled fracture surface at low and high magnifications respectively. The high magnification photograph in Figure 5.42 was a pair of opposing and matching fracture surfaces obtained from a small plateau region lying parallel to the rolling plane. Careful examination of this pair of micrographs shows that the fine precipitates are confined principally to one fracture surface. 5.5.2.2 Transgranular Steps at High Cyclic K-rates At the higher cyclic K-rates, the transgranular steps showed arrays of ridges perpendicular to the plane of intergranular failure and normal to the macroscopic crack 117 propagation direction. The distance between these ridges was around 5 urn. This suggested that they were not formed during every fatigue cycle because the average fatigue crack advance at these loading rates was below 1 p.m/cycle. Figure 5.41 SEM micrograph (low Mag.) of transgranular step AA-7075, Specimen A74, Mag.: 3100x, R=0.78, K-rate=7.8xl0"3 MPaVm/s, MCPD bottom to top Figure 5.42 SEM micrograph (high Mag. of Fig. 5.41) of opposing fracture surfaces on a transgranular step with matching dimples and precipitates (arrowed), Mag. 19000X 118 Figure 5.43 shows a SEM micrograph of a high cyclic K-rate transgranular region (steps). The ridges spaced at about 5 u.m are readily visible on the left half of the picture. Between these ridges very fine lines (arrowed) can be seen which are running parallel to the macroscopic crack propagation direction and perpendicular to the ridges. The spacing between these lines is approximately 1 to 2 u,m. Also visible are numerous corrosion crevices penetrating the intergranular regions normal to the rolling plane. Figure 5.43 SEM micrograph of ridges and faint lines (arrowed) on transgranular steps AA-7075, Specimen A73, Mag.: 2500x, R=0.78, K-rate=3.1xl0"1 MPaVm/s, MCPD bottom to top 119 SEM micrographs taken from a polished and etched specimen revealed that the arrowed fine lines observed in Figure 5.43 probably represent subgrain boundaries and, therefore, are related to the microstructure rather than the fracture process. Figure 5.44 shows an example of an etched micrograph which is taken on a plane normal to the transverse (T) direction. Figure 5.44 S E M micrograph of a polished and etched plane normal to the T direction with faint subgrain boundaries AA-7075, Mag.: 1500x In some instances very fine striations could be detected on the transgranular crack steps. Their spacing was considerably smaller than the fine lines described in Figures 5.43 and 5.44. Usually the striations were resolved clearly on one fracture surface and were very faint on the opposing surface. Thus, attempts to match individual striations on opposing fracture surfaces proved to be very difficult. Furthermore, the striation spacing 120 was about 4 times smaller than the average crack advance per cycle. This suggested that the cychc crack propagation rate on the transgranular steps was lower than the average cyclic crack propagation rate if each fine striation corresponded to one load cycle and each striation corresponded to the position of the local crack front. The smaller spacing is difficult to account for unless the local crack propagation direction was not parallel to the macroscopic direction of cracking. This is likely if the cracks advanced (tunnel) on intergranular surfaces with transgranular cracking occurring laterally on the uncracked ligaments between the tunnels. Transgranular striations are shown in Figure 5.45, which shows a pair of opposing and matching fracture surfaces containing tear ridges spaced at ~5 pm and fine striations spaced at ~0.2 pm. The fine striations were parallel to the macroscopic crack propagation direction. Also visible on each side of the micrographs are the two intergranular fracture planes which were linked by the transgranular, striated steps. Inspection of the widely spaced ridges revealed in some instances that the fine precipitates were confined mainly to one fracture surface, while ductile dimples were more pronounced on the other fracture surface. This is analogous to observations made on transgranular steps produced at low K-rates. 5.5.3 Fractography of Regions Wi th Larger Precipitates Even though detailed SEM investigations were performed on fracture regions with larger precipitates, a few general observations were made. At low K ^ values, the mainly intergranular fracture surface was rather flat and featureless and the crack path did not seem to be influenced by larger precipitates. When was increased the crack tended to follow grain boundaries associated with larger precipitates and produced a 121 stepped fracture surface. However, at the R-ratio of 0.31 an increase in K , ^ also produced increased amounts of transgranular cracking. The transgranular crack path was not influenced by larger precipitates because they were mainly located at the grain boundaries. Figure 5.45 SEM micrograph of opposing fracture surfaces on a transgranular step with fine striations (arrowed) AA-7075, Specimen A73, Mag.: 3500x, R=0.78, K-rate=3.1xl0"1 MPaVm/s, MCPD bottom to top 5.5.4 The Crack Front Profile, Tunnelling This far, the crack profile has only been described normal to the crack plane. The profile of a propagating crack along the crack front was determined by using the specimen sectioning technique described in Chapter 4.1.3. 122 Figure 5.30 on page 106 showed that fatigue precracking at low ^ values produced an irregular crack front on a flat fracture surface. During the CF tests, however, was usually higher than that used for pre-cracking and the crack did not advance on a flat fracture surface any longer but was propagating on different levels producing steps perpendicular to the rolling plane. Figure 5.46 shows the crack front of a specimen tested at a cyclic K-rate of about 3x10"1 MPaVm/s. The specimen was sectioned normal to the macroscopic crack propagation direction and the view of the picture is from the crack tip towards the crack mouth. It appears that the crack front is following larger precipitates where either decohesion between the matrix and precipitates occurred or the precipitates were fractured. It is also apparent that the crack front did not advance uniformly but penetrated further in areas with more precipitates, while areas without precipitates remained uncracked, indicating crack front tunnelling. The distribution of precipitates, together with the rolling structure of the alloy, also led to an uneven, stepped crack front advancing on separate planes parallel to the fracture plane. Figure 5.46 Stepped crack front with precipitates on a polished plane normal to the crack propagation direction AA-7075, Specimen A72, Mag.: 800x, R=0.78, K-rate=3.0xl0"1 MPaVm/s 123 The tunnelling effect was verified by SEM of the crack tip on a sectioned specimen as described in Chapter 4.9 on page 60. Pronounced differences in crack front advance were visible when the specimen was observed in the SEM. Figure 5.47 shows an example of the crack front tunnelling. A schematic drawing of the crack profile is given in Figure 5.48. Figure 5.47 SEM micrograph of the tunnelling crack front AA-7075, Specimen A82, Mag.: HOOx, R=0.78, K-rate=5.1xlQ-4 MPaVm/s 124 Figure 5.48 Schematic drawing of crack front tunnelling 5.5.5 Fractography of Tests in Dry Air Because of very low crack propagation rates at low K-rates, tests in dry air were only performed at the higher K-rates. The fractography of all tests revealed a mainly transgranular fracture surface. The very small striation spacings observed in dry air were very difficult to observe in the SEM. On some transgranular sections striations could only be detected on one side of the fracture pair while on the opposing matching side large numbers of precipitates were present. These observations suggest that the effect of an aqueous environment on crack propagation was magnifested by a mainly intergranular crack path and an accompanying increase in crack propagation rates (as shown in Figures 5.16-5.19). 5.6 Fractography Alloy-8090 125 The fractography of alloy 8090 was quite different from the alloy 7075 and is probably related to differences in microstructure. For example, alloy 8090 has pancake-like grains and is partly recrystallized compared to the elongated grain structure found in alloy 7075. Fracture was predominantly intergranular when tested in air and aqueous environments, independent of cyclic or monotonic loading conditions. Transgranular fracture sections (steps) were found between regions of intergranular failure. These occurred because the crack propagated (tunneled) on different intergranular interfaces and connection between these fracture regions was established by ductile tearing of the intervening ligaments. The mainly intergranular crack path with the transgranular steps is clearly visible in Figure 5.49. Figure 5.49 Intergranular, stepped crack path on a polished plane normal to the crack propagation direction AA-8090, Specimen Li-25, Mag.: 175x, R=0.75, K-rate=3.1xl0"' MPaVm/s 5.6.1 Intergranular Fracture Regions 126 Examination of the intergranular fracture surface revealed ductile dimple fracture on some grain boundary facets as can be seen in Figure 5.50. Matching fracture surfaces of a grain boundary triple point are shown in Figure 5.51. Precipitates could be found on both opposing fracture surfaces. Another typical feature of the intergranular fracture surfaces were pronounced slip steps. Usually they were only found on one fracture surface and not detected on the opposing surface, indicating that they were produced after initial metal separation. An example of pronounced slip steps is given in Figure 5.52. Figure 5.50 SEM micrograph of dimpled intergranular fracture surface AA-8090, Specimen Li-22, Mag.: 2000x, R=0.75, K-rate=1.9xl02 MPaVm/s, MCPD bottom to top 127 Figure 5.51 SEM micrograph of opposing fracture surfaces at a grain boundary triple point with precipitates on both fracture halves AA-8090, Specimen Li-22, Mag.: 9000x, R=0.75, K-rate=1.9xl02 MPa^/s , MCPD bottom to top Figure 5.52 SEM micrograph of pronounced slip steps at a grain boundary AA-8090, Specimen Li-22, Mag.: 8000x, R=0.75, K-rate=1.9xl02 MPaVm/s 128 Another example of grain boundary separation is presented in Figure 5.53 from a tensile test performed in air. The matching pair of opposing surfaces shows dimples and slip lines on one intergranular fracture surface, whereas on the opposing side, numerous small precipitates are visible. Figure 5.53 Grain boundary separation produced by tensile overload in air, slip lines are confined to one fracture surface whereas precipitates (arrowed) are located on the opposing surface AA-8090, Mag.: 3500x 8 9 @ 5.6.2 Fractography of Transgranular Steps With fatigue experiments, the fracture appearance was similar to the monotonic loading tests. No striations could be found on the intergranular fracture surface and the only difference in fractography produced under cyclic loading conditions was found on the transgranular steps that connect the intergranular fracture regions, On some of these steps, tear ridges could be found with spacings of a few microns. Higher magnification 129 micrographs revealed that these apparent tear ridges were constructed of a multitude of small steps running in two different directions. With the aid of stereo imaging, it was possible to identify interlocking steps which were as small as 0.1 |im. An example of the interlocking steps is given in Figures 5.54 and 5.55 which are low and high magnification microphotographs of the same area. Figure 5.54 S E M micrograph of transgranular steps which connect the intergranular fracture regions AA-8090, Specimen Li-26, Mag.: 450x, R=0.75, K-rate=1.9xl02 MPaVm/s, MCPD bottom to top jj^HflHIHK&VlKfJ TP 1 . <-% Figure 5.55 SEM micrograph (high Mag. of Fig.5.54) of interlocking steps on a transgranular fracture region Mag.: lOOOOx V 130 6 Discussion Real time crack velocities measured at K-rates typical of SCC experiments are several orders of magnitude smaller than crack velocities measured under high K-rate conditions typical of CF experiments. Crack velocities at intermediate K-rates, therefore, might be expected to lie between the crack velocities measured during SCC tests and those obtained during CF. However, experimental data have shown that this is not the case and crack velocities at intermediate K-rates are at a minimum, being even lower than velocities measured during low K-rate SCC experiments. Additionally, alloy 7075 tested at a high R-ratio of 0.78, showed increased crack velocities in a narrow K-rate range close to the intermediate K-rates where the lowest velocities were measured. In the following discussion, the experimentally measured K-rate effects on real time crack velocities (da/dt) and cyclic crack propagation rates (da/dN) will be explained. It will be based on models derived from fracture mechanics theory, plus SCC and CF cracking mechanisms. Fractography observations of actual test specimens will be discussed with respect to the varying dependence of crack propagation rates on K-rate. 6.1 Cyclic Loading Alloy 7075, R=0.78 The dependence of crack velocity on K-rates is rather complex in alloy 7075 tested at a R-ratio of 0.78. Crack velocities plotted in Figure 5.20 on page 91 show minima and maxima if the K-rate is varied from 10'5 MPaVm/s to about 10"1 MPaVm/s. Therefore, the individual K-rate regions of crack propagation behaviour will be discussed separately. 131 6.1.1 Cycling at the Lowest Rates The real time crack velocity obtained at the slowest cyclic loading rate (specimen A93, Fig. 5.20) was almost identical to those measured at comparable stress intensities and K-rates under slow monotonic loading conditions, and almost identical to those measured at comparable IK-ratel magnitudes under decreasing load conditions (bolt loaded specimen A-SI, Fig. 5.13). This implies that crack velocities during the loading and unloading parts of the fatigue cycle were similar to each other at low K-rates. The limiting IK-ratel (loading or unloading) below which changes in loading rate have no effect on crack velocities, when compared at similar stress intensities, was deduced to be near ~lxl0"5 MPaVm/s. This deduction follows from the observation that rising K-rate tests produced no significant changes in crack velocity up to lxlO"4 MPaVm/s, whereas increases in the cyclic K-rate beyond lxlO'5 MPaVm/s produced a significant decrease in crack velocity (Fig. 5.20). For example, when the cyclic K-rate was increased from lxlO"5 MPaVm/s to 4X10"4 MPaVm/s, the crack rate decreased by a factor of ~3, as shown in Figure 5.20. The decreasing crack velocity at cyclic K-rates >lxl0"5 MPaVm/s was attributed primarily to a decreasing average crack velocity during the unloading portion of the cycle. This is confirmed by the data in Figure 6.1 (specimen A93) and 6.2 (specimen 91), which show crack advance during a single load cycle at average cyclic K-rates of 5.5xl0'6 MPaVm/s and 4.5xl0"5 MPaVm/s, respectively. Average computed crack velocities during the rising load were found to be relatively unchanged at 8.1xl0"9 m/s and 8.8xl0~9 m/s, at the lower and higher K-rates respectively. However, the average unloading crack velocities in region II between about 17.5 and 13.5 MPaVm were different from each other, being 7.4xl0'9 m/s at an average cyclic K-rate of 5.5xl0"6 132 MPaVni/s and 2xl0"9 m/s at a cyclic K-rate of 4.5x10"5MPaVrn/s. These observations suggest that at even higher cyclic K-rates very little crack advance will occur during unloading at R-ratios of 0.78. A decrease of the average crack velocity during unloading with increasing IK-ratel can be explained if SCC is controlled by a film rupture/dissolution process. Hereby, the local plastic strain accumulation at the crack tip necessary for film rupture events is caused by time dependent creep processes. Therefore, fast unloading rates will decrease the number of film rupture events during an unloading cycle and result in slower average cracking rates. The detailed relationship between cyclic K-rate and crack velocity at low K-rates is expected to be influenced by crack tip plasticity effects. For example, at K,^ , a K-dependent crack tip plastic zone is formed in an elastic matrix. Upon unloading, elastic contraction of the matrix may produce compressive stresses at the crack tip when K ^ is reached, unless the crack propagated beyond the K ^ plastic zone during unloading. The resulting compressive stresses will retard crack advance during loading on the subsequent cycle and give lower average crack velocities. The interaction between low cyclic K-rates and its plastic zone is explored in more detail in the following section. Figure 6.1 Specimen A93 (5.5E-6 MPafrTvs) Crack advance during a single 2 0  load cycle experiment ,8 -Stress Intensity (MPa/m) Figure 6.2 Crack advance during a single load cycle experiment with specimen A91 133 Specimen A91 (4.5E-5 MPafn/s) 10 12 14 16 18 20 Stress Intensity (MPa/m) 6.1.2 Interaction Between Crack Velocity and the Plastic Zone at Low Cyclic K-rates Most text books on fracture mechanics describe methods for calculation of the crack tip plastic zone size. Complex solutions incorporating elastic/plastic theories usually involve numerical methods. To a first approximation, however, the linear elastic theory (LET) can be applied. A detailed description is given in the books by Broek [155] and Kanninen et al. [157] and is summarized below. The stresses at the crack tip under plane strain conditions are: K x ef . 9 . 38^  (Eq. 6.1) K t ef, . e . 3eN a = , — cos- 1 + sm-sin— y Vim 2^ 2 2 , az = v(ax + ay) K r . e e . 36 xIV = -== • sin-cos-cos— X Y V2OT 2 2 2 134 (Eq. 6.2) (Eq. 6.3) (Eq. 6.4) Figure 6.3 Crack tip coordinates For the case of plane strain, the principal crack tip stresses are: Kj ef . e^  o~, = , • cos- 1 + sin-1 V2^ 2^ 2 K, ef . ex C 2 = ^COS2{l-Sml c 3 = 2vK, e - p = - cos-V2OT 2 (Eq. 6.5) (Eq. 6.6) (Eq. 6.7) The stress field described by the LET can be used to estimate the shape of the crack tip plastic zone after introducing a yield criterion. Hence, by using von Mises criterion for yielding 135 to - o^2 + (o2 - a3)2+(a3 - a,)2 = 2a2, (Eq. 6.8) the boundary of the plastic zone can be described. For plane strain conditions the extent of the plastic zone becomes r(0) = K 4710" Y S 2 I sin2 9 + (1 - 2v)2 (1 + cos 6) (Eq. 6.9) In Figure 6.4, the plane strain crack tip plastic zone for two stress intensity values is described. They correspond to the maximum and minimum stress intensities applied for cyclic loading of the alloy 7075 at a R-ratio of 0.78 and a y s = 540 MPa. Figure 6.4 Size of crack tip plastic zone for K.^17.5 MPaVm and KTnin=13.5 MPaVm Crack Tip Plasic Zone Distance (pm) 200 /Kmax \ ^ K m i n Crack •100 0 100 Distance ahead of crack tip (pm) 136 The crack tip plastic zone can be drawn at different time steps during the crack advance of one cycle. Figures 6.5 and 6.6 illustrate this for two different cyclic K-rates where the crack advance during loading arid unloading was measured. The shape of the plastic zone is distorted because of different scales on the x and y axes. However, it can be seen that the crack advance during each cycle is considerably bigger than the dimensions of the plastic zone itself. Figure 6.5 Movement of the crack tip plastic zone with crack advance for specimen A93 (5.5E-6 MPalm/s) Crack Advance with Crack Tip Plastic Zone Unl oa ding 1 Lo adinc J 1 Crack Advance (mm) Figure 6.6 Movement of the crack tip plastic zone with crack advance for specimen A91 (4.5E-5 MPaffTOs) Crack Advance with Crack Tip Plastic Zone Jnioading Crack Advance (mm) 137 If the K-rate was increased to 1.4X10"4 MPaVm/s, as for specimen A51, real time cracking rates decreased to a value of 2.7x10"9 m/s giving a crack advance for one complete cycle of 143 pm The separate amounts of crack advance during loading and unloading could not be measured any longer but it can be assumed that crack advance during unloading was several times smaller than the crack advance during loading. Therefore, the crack tip plastic zone at K ^ will have advanced only litde and will be partly contained in the large plastic zone formed at K ^ . It is, therefore, possible that the overlapping of the plastic zones produced compressive forces around minimum load levels. If the K-rate were to be increased further the crack advance during unloading would become negligible and the plastic zone at the end of a cycle would be contained within the larger plastic zone formed at K^, . Therefore, upon loading during the following cycle, the crack cannot start growing until the compressive forces are overcome, causing a decreased average crack advance during the loading part of a cycle. This is believed to be the reason for the decreasing crack velocity observed in Figure 5.20 as the cyclic K-rate was increased up to ~3.5xl0"4 MPaVm/s. According to the crack tip plasticity model, further increases in K-rate should be accompanied by decreasing crack velocities. This was not the case (see Figure 5.20). The opposite behaviour was observed (ie. increased crack velocities) between K-rates of 3.5X10"4 and 1.2xl0'3 MPaVm/s. Crack tip tunnelling is believed to explain this effect, as described in the subsequent section. 6.1.3 Crack Tunnell ing at Intermediate Loading Rates 138 The uneven tunnelling advance of the crack front, described in Chapter 5.5.4, gave rise to local differences in stress intensity. Crack front sections that had advanced further, (now designated as "long crack sections"), experienced lower K values than at the lagging crack front", (now designated as "short crack sections"), due to crack tip shielding by the uncracked ligament between advancing and lagging crack fronts. Therefore, during the unloading fatigue cycle, the short crack sections experienced higher compressive stresses because of their larger plastic zones at K^. This effect wedged open the long crack sections and delayed, or eliminated, compressive stresses from being formed at long cracks. Consequently, longer crack sections could continue to advance by corrosion fatigue during unloading, even when shorter sections could not advance. Note that this is only applicable above a critical cyclic K-rate, because otherwise the short crack sections will have time to advance beyond the boundary of the K,,,, plastic zone during unloading. Therefore, compressive stresses will not occur. Also, longer crack sections are unlikely to outgrow shorter crack sections to any significant degree, because the increasing K-value on the short crack sections will eventually cause them to advance at higher velocities (stage III velocities) until the difference between advancing and lagging crack front sections is again decreased. Overall, average crack velocities in the intermediate K-rate range near 10"3 MPaVm/s were not retarded because of crack tunnelling effects. 139 6.1.4 Loading Rates (K-rates) Producing Crack Retardation At higher K-rates in the range 1.2xl0"3 to 9.1xl0"3 MPaVm/s, crack velocities decreased by approximately one decade as the K-rate increased. (See Figure 5.20). The unloading time at these faster K-rates was too short for significant crack advance to occur. Therefore, during unloading, even the longer crack sections could no longer outgrow their plastic zones that formed at K^, . This led to compressive crack tip stresses as K ^ was approached, causing crack growth retardation. Hence, crack growth was limited to shorter intervals near K ^ . These effects led to the observed decrease in average crack velocities. (Note it is assumed that a rising K-rate does not decrease the real crack velocity during crack growth, but decreases the crack growth period which then produces a decrease in average velocity). The K-rate regime of 1.2xl0"3 to 9.1xl0'3 MPaVm/s may be best compared with the well known effects of a cyclic overload experiment, where crack growth retardation is also caused by compressive forces acting at the crack tip, due to a large plastic zone formed during the overload cycle. 6.1.5 Cycling at the Highest Loading Rates Crack velocities continued to increase as the K-rates were increased above 9.1xl0"3 MPaVm/s (Figure 5.20). The cyclic crack propagation rate, on the other hand, was almost constant at 8xl0"7 m/cycle up to a K-rate of 3.5xl0_1 MPaVm/s. Because the crack advance per cycle was small, the material ahead of the crack tip underwent considerable cyclic plastic deformation before it was traversed by the crack front. A 140 decrease in cyclic crack propagation rate was observed only at the highest K-rate of 2.2xl02 MPaVm/s, which was obtained under sinusoidal loading where it fell to 1.5xl0"7 m/cycle. Also, the addition of the chromate corrosion inhibitor no longer decreased the crack propagation rates (See Figures 5.18 and 5.19). The result from electrochemical potential measurements have shown that hydrogen embrittlement may be involved in the cracking process. This could explain the lower cracking rate measured at the highest K-rate of 2.2xl02 MPaVm/s because an increasing K-rate decreases the time for hydrogen absorption during each cycle. An estimation of the average hydrogen diffusion distance in a semi-infinite solid with a replenished source can be given as: x = VDH • t (Eq. 6.10) where; D H = Hydrogen diffusion coefficient in AA7075-T6 obtained from permeation studies [5] x = Distance ahead of the crack tip where the hydrogen concentration has decreased to half of its original value at the crack tip For a K-rate of 2.2xl02 MPaVm/s the time interval for a single cycle is 1/30 s. Equation 6.9 predicts an average H-diffusion distance during a load cycle of about 0.1 pm. This is comparable to the average cracking rate of 1.5xl0"7 m/cycle measured at this K-rate. This estimation, therefore, would also support the possibility of a crack advance mechanism that is assisted by hydrogen diffusion ahead of the crack tip. In the K-rate range between 9.1xl0"3 MPaVm/s and 3.5x10"' MPaVm/s, where constant cyclic crack propagation rates were measured, the estimated average hydrogen 141 diffusion distance during a cycle was considerably larger than the crack advance per cycle. Therefore, if hydrogen embrittlement assists CF, the cracking process is not influenced by the length of time available for the diffusion reaction at K rates between 9.1xl0"3 and 3.5xlO"x MPaVm/s. This leads to the conclusion that the embrittling action of hydrogen reaches an upper limit above which further hydrogen diffusion does not alter the crack propagation rates. Instead, mechanical parameters dictate the crack advance per cycle once the hydrogen concentration in the material immediately ahead of the crack reaches a certain concentration. It should be noted that changes in the test environment from aqueous NaCl to air produced changes in the cyclic crack propagation rate. In dry air, the crack advance per cycle at the highest K-rate of 2.2x10* MPaVm/s was only slightly lower than cyclic propagation rates at 3xl0_1 MPaVm/s (See Figure 5.19). In the NaCl environment cyclic cracking rates differed by about a factor of 5 between these two K-rates, which was explained with a lesser degree of hydrogen embrittlement at the higher K-rate of 2.2xl02 MPaVm/s. It can be assumed that in the dry air environment very little atomic hydrogen was available at the crack tip and, therefore, an almost constant crack advance per cycle, independent of K-rate, resulted. A constant crack advance per cycle indicates that mainly time independent mechanical factors controlled the crack advance. At the K-rate of 1.3xl0"3 MPaVm/s, the crack advance was too small to be measured within the time period of the experiment. However, it needs to be pointed out that crack velocities in air at the higher K-rates were only approximately 5 to 10 times lower than in a NaCl solution, whereas at the much lower K-rate of 1.3xl0"3 MPaVm/s the crack velocities differed by a factor of >100. This, therefore, supports a time dependent cracking mechanism at the lower K-rates if an aqueous environment is present. Overall, the high K-rate behaviour appears to be consistent with a hydrogen diffusion model of embrittlement. The decreasing crack propagation rate at the highest K-rates employed is consistent with the limited time available for hydrogen diffusion to 142 enhance cracking. However, it must be recognized that if dissolution processes contributed to crack advance, then there might be insufficient time for dissolution to have a significant effect on crack growth per cycle at high frequencies and the crack propagation rate would decrease at high K-rates, as observed. Thus, while a hydrogen diffusion model is generally consistent with the observed behaviour, it is clearly not a unique explanation of the observations.. 6.1.6 Summary of Cyclic Loading Alloy 7075, R=0.78 At the lowest K-rates investigated crack velocities during the loading and unloading part of a fatigue cycle were not significantly influenced by variations in K-rate. At higher K-rates lower real time crack velocities were caused, both by a decrease in film rupture events during the unloading part of a fatigue cycle and plasticity induced crack growth retardation effects. Cyclic crack propagation rates decreased until minimal crack advance increments per cycle were dictated by mechanical parameters acting on a hydrogen embrittled crack tip region. At intermediate K-rates a crack tunnelling mechanism overcame the plasticity induced crack retardation and cracks propagated at the same rates as during low K-rate tests, where no retardation phenomena were encountered. At the highest K-rates, cyclic crack propagation rates decreased because the embrittling action of hydrogen at the crack tip was limited by the shorter diffusion times per cycle. 6.2 Cyclic Loading Alloy 7075, R=0.31 143 The results of the cyclic loading experiments at a R-ratio of 0.31 in aqueous NaCl were conducted between of ~4 MPaVm/s and of -13.5 MPaVm. However, the slow rising K experiments in the same environment showed no detectable cracking at K values < 10 MPaVm Therefore, it is questionable whether significant cracking occurred throughout the entire fatigue cycle and probable that it occurred only during the period spent at higher stress intensities. At the lowest cyclic K-rate investigated, a loading and unloading half-cycle was performed on two different specimens. The crack advance was practically identical in both cases. Therefore, it can be concluded that absolute IK-ratel values below 10s MPaVni/s do not have a significant influence on crack propagation rates. One test was performed with a positive sawtooth load form with a rising K-rate of lxlO^MPaVm/s. This resulted in twice the average crack velocity when compared with a triangular load wave form with a cyclic K-rate of lxlO"4 MPaVm/s. Hence, recognizing that the period of the sawtooth was half that of the triangular wave form and that the crack advance per cycle was identical in both tests, the results indicated that crack advance occurred principally during the rising load portion of the cycle. Furthermore, essentially similar behaviour was observed at a R-ratio of 0.78, where the unloading crack advance became more and more negligible at cyclic K-rates > 10"5 MPaVm/s until crack front tunnelling occurred. Tunnelling effects caused an increase in crack velocities between K-rates of 2x10^ and lxlO"2 MPaVm/s at a R-ratio of 0.78, whereas an almost constant velocity was 144 observed at these K-rates with a R-ratio of 0.31 (Figure 5.16). This implies that a tunnelling mechanism was not operative in the same K-rate range and could be explained by the fact crack velocities at the lower stress intensities of a fatigue cycle were very small and almost no crack advance occurred. Therefore, even if the shorter crack sections prevented compressive stresses from forming at the tip of longer crack sections during unloading, crack velocities were still too slow for significant crack propagation of the longer crack sections. At K-rates above 10'2 MPaVm/s real time crack velocities increased rapidly and were comparable to those observed at the R-ratio of 0.78. Cyclic crack propagation per cycle was almost constant at 2xl0"6 m/cycle up to a K-rate of 4x10* MPaVm/s. A lower cyclic crack propagation rate was again obtained at the highest K-rate of 6x102 MPaVm/s. Cyclic crack propagation rates measured in dry air were about a factor of 8 lower than those measured in NaCl solution. Compared to cyclic crack propagation rates at the R-ratio of 0.78, crack advance per cycle was ~3 times faster at a R-ratio of 0.31, when tested at K-rates > 10"2 MPaVm/s. However, the overall dependence of cyclic crack propagation rates on K-rate above 10"2 MPaVm/s was very similar to tests performed at the R-ratio of 0.78. Therefore, essentially the same cracking mechanisms may apply and it is concluded that the difference in crack propagation rates is caused by mechanical factors. Hereby, AK plays the important role because it controls the amount of plastic deformation during each load cycle. This is supported by the experiments performed at different R-ratios of 0.31 and 0.78, respectively, but at the same AK value of about 4 MPaVm/s. Essentially the same cyclic crack propagation rate resulted in both tests, as can be seen in Figure 5.14. 6 . 3 Cyclic Loading Alloy 8 0 9 0 , R = 0 . 7 5 145 With the alloy 8090 it was difficult to obtain data at very low cyclic K-rates, because crack propagation due to SCC was very minimal and limited to stress intensities > 12 MPaVm. The K ^ , values during the cyclic loading experiments were around 13 MPaVm and K,,^ values were at about 9.7 MPaVm (which produced a R-ratio of 0.75). Consequently, one can assume that crack propagation rates around the minimum cyclic stress intensity value are negligible. This meant that tests on alloy 8090 at a R-ratio of 0.75 may in part be compared with alloy 7075 tested at a R-ratio of 0.31. At the highest K-rates investigated, alloy 8090 gave approximately the same real time crack velocities as alloy 7075 when tested at the comparable AK value (see Figure 5.14). In the K-rate range from 102 MPaVm/s to 10~2 MPaVm/s crack velocities of alloy 8090 and alloy 7075 decreased in a similar way and both alloys produced crack velocities near 10"9 m/s at 10"2 MPaVm/s. (Note that AK values for alloy 8090 were slightly lower than for alloy 7075). Also, tests in dry air produced the same crack velocities at comparable K-rates, indicating a similar cracking process in both alloys (see Figures 5.19 and 5.20). In alloy 7075 cyclic crack propagation rates at K-rates above 102 MPaVm/s were attributed mainly to mechanically dominated processes. At a AK value of about 4 MPaVm, the crack advance per cycle was around 10"6 m. When limiting hydrogen diffusion effects decreased cracking rates at the highest investigated K-rate, the crack advance per cycle dropped to 10'7 m. Essentially the same value for crack advance per cycle resulted for alloy 8090 if it is assumed that mechanically dominated cracking occurred at K-rates higher than 4xl0"3 MPaVm/s and limiting hydrogen diffusion effects 146 appeared at K-rates above 10"2 MPaVm/s. The difference in K-rate values where hydrogen diffusion affected cracking rates can be explained with different diffusion coefficients in both alloys or a different critical hydrogen concentration necessary in the crack tip region to cause embrittlement. (No diffusion coefficients were available for alloy 8090). Investigations on the fractography confirmed a transition in cracking mechanism at a K-rate of 10"2 MPaVm/s for alloy 7075. Fractography of alloy 8090, however, did not reveal a pronounced difference at different K-rates. Therefore, it was not possible to clearly define a transition from SCC to CF in alloy 8090 partly because no significant SCC occurred in this alloy. 6.4 Monotonic Loading Experiments 6.4.1 Slow Rising Load Tests In the Results section it was shown that no crack advance was recorded below a stress intensity of approximately 8 to 10 MPaVm, due to the difficulties in measuring very small crack advance steps at the lower stress intensities. Also, no accurate crack velocity values could be obtained in this threshold region under faster loading rate conditions. Another factor influencing the crack velocity in the initial stages was the crack path. During fatigue precracking at low cyclic stress intensities, the crack path did not preferentially follow grain boundaries containing large precipitates but was almost perfectly flat across the width of the specimen with intergranular and transgranular regions. When the rising-K SCC test started a new crack front profile was established 147 (primarily intergranular). This, in turn, might have influenced the cracking rate during the early stages, especially at the high K-rates where the time for crack propagation was very short and an unrealistic high K^cc threshold value resulted. Therefore, in order to apply the slow loading technique as a rapid SCC test procedure, these problems have to be considered. Very sensitive crack length recording techniques have to be used to measure region I crack velocities, especially at higher K-rates. In the present work, however, the main interest was on the higher region II velocities which were easier to measure. Within the data scatter (Figure 5.11) it could be concluded that the monotonic loading experiments showed no significant effect of K-rate on region II SCC velocities, even at the highest K-rates used where crack tip blunting might be expected. It is, therefore, concluded that mass transport phenomena in the liquid within the crack dictated the upper limit for crack velocities as the K-rate was increased. This conclusion was reached because slow monotonic K-rates cannot influence mass transport in the liquid phase, although they could influence events in the metal and at the metal/electrolyte interface. No attempt was made to determine whether mass transport control was due to diffusion of either reactants or products to and from the crack tip. Future monotonic loading experiments on SCC at even higher K-rates should be conducted on specimens with a stress corrosion pre-crack instead of a fatigue precrack. The primary reasons being that the total amount of crack advance in region II will be smaller than about 0.2 mm, the test duration will be shorter than a few hours, and changes in crack path morphology during these short testing times will definitely affect the cracking rates measured, particularly if the crack morphology changes from that characteristic of fatigue to that of SCC. 148 In order to compare the results from rnonotonically rising-K experiments with literature data obtained with slow strain rate tests on smooth specimens, a method of estimating the strain rate had to be chosen. Several methods were described in Chapter 3.8. Equation 3.4 described the average strain rate (de/dt) as: (dz) 2,(1) Hereby, the CTOD, produced at K ^ , is considered at a gage section undergoing uniform straining as the stress intensity is increased from K ^ to Kmlix over the time period T. The K-rate, on the other hand, was calculated as ( K ^ - K ^ / T . In the present study, monotonic K-rates between 6xl0"6 to lxlO"4 MPaVm/s were applied. The corresponding strain rates can, therefore, be calculated from the rising-K times (T) between K ^ (13.6 MPaVm) and K ^ (17.5 MPaVm) for a R-ratio of 0.78. At 6xl0"6 MPaVm/s, T was 6.7xl05 s and at lxlO"4 MPaVm/s, T was 4xl04 s. Equation 3.4 then yields strain rates of 7.4xl0"7 and 1.2x10s s"1 respectively. Ugiansky et al. [105] found with alloy 7075, tested in the short transverse direction, a maximum sensitivity to SCC at a monotonic strain rate near 10'V1. The sensitivity to SCC was characterized by the percentage elongation to failure and the reduction of cross-section area. If the strain rate was decreased from 10V1 to 10"7s"\ the percentage elongation to failure decreased by a factor of 4. At strain rates >10'5 s'1 essentially no change in percentage elongation was measured when compared with tests performed in air. Figure 5.20 shows that region II crack velocities obtained under monotonic loading conditions were approximately constant between K-rates of 6.6xl0'6 and lxlO"4 149 MPaVm/s, which corresponds to crack tip strain rates of 7.4xl0'7 to 1.2xl0"5s'\ However, in Figure 5.20, no decreasing trend in crack velocities is visible as the monotonic K-rate increased from 6.6xl0"6 to lxlO"4 MPaVm/s which does not support the results from Ugiansky et al. [105]. Monotonic rising-K tests have to be performed at even lower K-rates in order to allow a more complete comparison with the slow strain rate data obtained on smooth specimen. However, the present results on precracked specimens do not indicate an effect of monotonic K-rate on the cracking behaviour at region II stress intensities. On smooth specimens crack initiation may influence significantly the parameters used to define SCC resistance. The measured region II crack velocities in precracked specimens, on the other hand, are not directly influenced by crack initiation. With alloy 8090 very few slow K-rate data were obtained because of the relatively high resistance to SCC. The fact that cracking was observed only at high stress intensities close to K I C suggests that mechanical factors dominate crack propagation. 6.4.2 Bolt Loading and Constant Load Tests In Figure 5.12 bolt loaded tests with specimens of different initial crack length were compared. At the higher stress intensities, specimen A-SI with a shorter initial crack length produced slightly higher cracking rates. One might expect that under bolt loading conditions the shorter crack length results in higher decreasing K-rates and, therefore, lower crack velocities. In Figure 6.7 the decrease of stress intensity with time is plotted for specimen A-SI with a short crack and specimen A5 and A7 with longer cracks. It appears that the difference in crack length has no significant effect on K-rate (slope of the 150 curve) if compared at the same stress intensity. The difference in measured crack velocities between A-SI, A5 and A7 could, therefore, have been caused by the type of pre-crack used. (Note specimen A-SI had a pop in pre-crack whereas specimen A5 and A7 had fatigue-pre-cracks). Figure 6.7 also contains curves for specimen A91 and A93 where the decreasing K-rate was produced by slow unloading during a single cycle experiment. K-rates were considerably faster as compared to the bolt loaded tests. In Figure 6.8, the decreasing K-rate of region II stress intensities (~13.5 to 17.5 MPaVm) is plotted against measured crack velocities. It can be seen that IK-ratesI smaller than lxlO"5 MPaVm/s produce approximately the same region II crack velocities. At IK-ratesI above 4xl0"5 MPaVm/s, however, smaller crack velocities were measured. The IK-ratel, therefore, has an effect on crack velocities but to produce higher IK-ratesI with bolt loaded specimen very short crack lengths would be necessary. The reason for the decreased crack velocities at higher decreasing K-rates was explained in Chapter 6.1.1 with a decrease in film rupture events at the crack tip. In other words, a bolt loaded specimen with a very short crack may not necessarily display the same crack velocities as a specimen with a longer crack at the same stress intensity. In Chapter 3.6, it was described how different types of constant load tests were performed. Even though these experiments were considered constant load tests, this was not true on a finer scale. With the slow loading set-up variations in load smaller than 20 N occurred over the duration of several days, whereas the constant load test required adjustments of < 50 N almost every day. Even though only very few measurements were obtained with constant loaded tests, it can be seen from Table 5.4 that higher crack velocities were measured with the slow loading set-up (const, displ. rate) as compared to tests with periodic load adjustments. The later tests produced crack velocities almost 151 identical to the velocities measured with bolt loading tests (see Table 5.3). This means that experiments requiring periodic load adjustments can be approximated in part to a series of bolt loading tests, whereas experiments conducted with the slow loading set-up more closely approached a dead load test. Therefore, if the K-rate controls crack velocities by, for example, the frequency of film rupture events at the crack tip, its influence may be seen already on a very fine scale. This means that cyclic loading at very small AK values may already produce differences in crack propagation rates compared to constant load (dead load) tests. However, the few tests done in this study do not allow a conclusive statement to be made. Figure 6.7 Decrease of stress intensity due to crack propagation in bolt loaded specimen Decreasing-K Tests Stress Intensity (MPafm) 24 AA-7075 A-SI (short crack) 22 20 18 16 14 12 10 0 100 200 300 400 500 Time (hrs) 152 Figure 6.8 K-rates obtained with bolt loaded and single cycle decreasing load tests Crack Velocity (m/s) Decreasing-K Tests 2E-08 1E-08 -5E-09 3E-09 2E-09 1E-09 AA-7075 3.5wt% NaCl (Stress Intensity 13.5 - 1 7 5 MPafrri) I I A-Si Bolt loaded A-7 Bolt loaded A-93 Decreasing load cycle A-91 Decreasing load cycle 4 J u _i_L _L 1E-06 2E-06 3E-06 5E-06 1E-05 2E-05 3E-05 5E-05 K-Rate (MPatm/s) 1E-04 6.5 Fractography of AA-7075 The change from K-rate dependent to K-rate independent crack propagation rates occurred near 10"2 MPaVm/s. This transition was paralleled by a pronounced change in fractography. At cyclic K-rates below 10'2 MPaVm/s fractography was typically intergranular and was very similar to the fractography of bolt loaded and constant load SCC-tests. Above cyclic K-rates of 10"2 MPaVm/s striations appeared on the fracture surface, consistent with a fatigue process except for the rather unusual intergranular crack path during CF. 6.5.1 Effect of Larger Precipitates 153 Low K1sax values produced flat fracture surfaces, whereas at higher K , ^ values the crack front propagated on different levels following grain boundaries with large intergranular precipitates (see Figure 5.46). This can be explained with the size of the plastic zone. A larger plastic zone formed at the higher K , ^ values which led to plastic deformation at larger distances above and below the crack tip. (ie. normal to the main fracture plane). Hence, large precipitates lying within the plastic zone were subject to void formation at the precipitate matrix interface, allowing the initiation of intergranular cracks on different levels above and below the crack tip. The average thickness of grains in the S direction was < 50 pm which was comparable to a plastic zone size produced at about 10 MPaVm. 6.5.2 Low K-rates Micro tear ridges associated with precipitates could be found on intergranular fracture surfaces of rnonotonically loaded specimen at stress intensities comparable to of the cyclic loading tests performed at R=0.78. Quite pronounced tear ridges were also observed at cyclic K-rates near 10'2 MPaVm/s. This would support the proposed mechanism of plasticity induced crack growth retardation around 10"2 MPaVm/s, because during each cycle crack advance could only occur around K m a x values. The tear ridges indicate that even though a film rupture/dissolution mechanism may have been operative during crack advance, substantial tearing also occurred during the fracture process. Therefore, intergranular corrosion alone did not cause failure. On the 154 other hand, if hydrogen embrittlement caused a weakening of metal bonds at the grain boundaries, it did not prevent the grain boundary region from undergoing ductile plastic deformation. The addition of a chromate corrosion inhibitor to the sodium chloride test solution helped to preserve the fine tear ridges from subsequent corrosion. They were essentially invisible on fracture surfaces obtained in NaCl solutions without inhibitor because of corrosive attack of the crack flanks. Crack velocities, however, were not considerably reduced by the presence of the inhibitor, suggesting that dissolution processes alone were not controlling the cracking process. 6.5.3 High K-rates At higher K-rates the crack advance per cycle was dominated by mechanical factors, whereby crack tip blunting and re-initiation occurred during every cycle to produce a striated fracture surface. However, it was the aqueous environment that caused the intergranular crack path. This was concluded because tests in dry air produced mainly transgranular failure. The striation spacing matched the average crack propagation rate per cycle which was a clear indication for their formation during every fatigue cycle. The role of the environment is thought to be principally hydrogen embrittlement of the grain boundary region. Localized grain boundary dissolution processes may have resharpened the crack tip every time it was blunted by plastic deformation. The formation of striations on grain boundaries by crack tip blunting processes also indicated the presence of a ductile grain boundary region. 155 The interlocking ridges oriented at an angle to the striations (Figure 5.39) can be explained with shear bands being intersected by the crack front. With respect to Figure 5.39, a displacement had occurred in a shear band which crossed adjacent grains and created a small ridge on the grain boundary. Corrosion crevices on the intergranular fracture surfaces can account for the initiation of cracks propagating perpendicular to the macroscopic crack propagation direction. These produced striated steps connecting crack segments that were propagating on different levels. 6.6 Fractography of AA-8090 Fractography of alloy 8090 revealed very little information on possible cracking mechanisms because fracture surfaces looked essentially identical under the different testing conditions. The large number of dimples associated with grain boundary precipitates suggested that failure was caused by microvoid coalescence due to slip processes being confined to the grain boundary region. Therefore, mainly mechanical factors controlled the fracture process at all K-rates. Hence, the much less pronounced effects of K-rates on crack velocities, as compared to alloy 7075, is also in part reflected by the unchanged fractography of alloy 8090. The only difference in fractography between high and low K-rate tests was manifested on some transgranular fracture sections which were at an angle to the intergranular fracture regions. In analogy to striations on transgranular steps in alloy 7075, fine striations or ridges could sometimes be observed in alloy 8090. (Note that in alloy 8090 no striations were found on intergranular surfaces). The ridges were parallel 156 to the macroscopic crack propagation direction with a spacing of about 0.1 u\m. In contrast to alloy 7075, the striation spacing on the transgranular steps of alloy 8090 was comparable to the macroscopic crack propagation rate. Matching of the striations on opposing fracture surfaces revealed that they were interlocking which indicated that they were associated with events at the crack tip. Interlocking striations on matching fracture surfaces can be an indication for restricted reversible slip processes at the crack tip as explained by Fong. et al. [137]. This would suggest that these striations were formed as part of the cyclic crack advance. However, because these interlocking striations were not found on all transgranular steps it can not be concluded that restricted reversible slip was responsible for the overall crack advance but may have been operative only in some regions. Fractography of specimens tested in dry air and aqueous NaCl solutions was essentially identical. The crack path was always mainly intergranular. However, crack propagation rates in the two media were different by a factor of about 3 (tests in air were only performed at a K-rate of 2x10* MPaVm/s). Even though mechanical factors seem to have dominated crack advance at all K-rates, hydrogen embrittlement may have caused a weakening of the grain boundary region and, therefore, accelerated crack advance in the aqueous environment. 7 Summary and Conclusions 157 Summary: The K-rate parameter allowed the characterization of experiments typical for SCC and CF. By applying a triangular load wave form to DCB specimen, it was possible to compare monotonic slow loading experiments with tests performed under cyclic loading conditions. Additions of a chromate corrosion inhibitor to the sodium chloride test solution helped to preserve the fracture surfaces during long term experiments without significantly influencing cracking rates. A careful study of the fractography by SEM techniques revealed microscopic details that could explain the measured crack propagation behaviour in the different K-rate regions. Alloy 7075 showed a distinct dependence of real time crack velocities on K-rates if tested at a high R-ratio. Alloy 8090, on the other hand, showed a high resistance to SCC with fatigue pre-cracked specimen which made it difficult to investigate the low K-rate region. The use of the K-rate parameter to compare CF and SCC behaviour was unique. The study covered a range of 7 orders of magnitude in K-rate and is the first study to cover such a wide range of crack tip strain rate effects that encompass the spectrum from SCC to CF. The work led to new and unique information on crack tip strain rate effects on SCC and CF. 158 Conclusions: Several conclusions emerged that relate to the behaviour of the 7075 Al-alloy: 1) The transition from intergranular SCC to intergranular CF in alloy 7075 is not gradual but occurs at a specific "critical" K-rate. 2) Below the "critical" K-rate, cyclic crack propagation rates (da/dN) increase dramatically with decreasing K-rate because of a time dependent cracking process typical for SCC. 3) Above the "critical" K-rate, (da/dN) is almost independent of K-rate and the minimal crack advance per cycle is dictated mainly by AK which controls cyclic plastic crack tip deformation. 4) The high K-rate behaviour was consistent with, but not uniquely described by the limited diffusion of hydrogen ahead of the crack tip. 5) At low K-rates real time crack velocities (da/dt) are not increased by cyclic loading. Instead crack growth retardation effects can result in lower crack velocities than those typical for monotonic loading SCC tests. 6) Grain boundary precipitates and elongated flat grains of alloy AA-7075 give rise to a crack tunnelling effect which can overcome crack growth retardation effects at intermediate K-rates. 7) The intergranular crack path at low and high K-rates observed with AA-7075 is dictated by the aqueous environment. However, separation along the grain boundary is not caused by dissolution processes but in part produced by ductile tearing. 159 8) No effect of K-rate on region II SCC crack velocities was observed during monotonic loading. This suggests that region II cracking is controlled by mass transport processes in the liquid phase. 9) The slow monotonic loading technique with fatigue precracked DCB specimen proved to be a useful technique for providing region II SCC data. However, alloys with a high resistance to SCC can only be tested with a very sensitive crack length measuring device. No clear conclusion could be reached about the SCC to CF transition in alloy 8090 because of its high resistance to SCC and the almost identical fractography at all investigated K-rates. 160 8 References 1. D.O. Sprowls and R.H. Brown, "Stress Corrosion Mechanisms for Aluminum Alloys" in Proc. of Conference on Fundamental Aspects of Stress Corrosion Cracking, Ohio State University, R.W. Staehle, A J . Forty, D. van Rooyen, eds., NACE, Houston, TX, 1967, pp. 466-512. 2. N.J.H. Holroyd and G.M. Scamans, "Slow-Strain-Rate Stress Corrosion Testing of Aluminum Alloys" in Environment-Sensitive Fracture: Evaluation and Comparison of Test Methods, S.W. Dean, E.N. Pugh and G.M. Ugiansky, eds., ASTM STP 821, 1984, pp. 202-241. 3. M. V. Hyatt, "Use of Precracked Specimens in Stress Corrosion Testing of High Strength Aluminum Alloys", Corrosion, 1970, vol. 26, pp. 487-503. 4. H. Buhl, "Validity of the Slow Straining Test Method in the Stress Corrosion Cracking Research Compared with Conventional Testing Techniques" in Stress Corrosion Cracking - The Slow Strain-Rate Technique, G.M. Ugiansky, J.H. Payer, eds., ASTM STP 665,1979, pp. 333-346. 5. N.J.H. Holroyd and D. Hardie, "Factors Controlling Crack Velocity in 7000 Series Aluminium Alloys During Fatigue in an Aggressive Environment", Corrosion Science, 1983, vol. 23, pp. 527-546. 6. K. Welpmann, M. Peters and T.H. Sanders, Jr., "Aluminium-Lithium Alloys(I): Metallurgical Fundamentals", Aluminium, 1984, vol. 60, pp. E641-E646. 7. E.H. Dix, Jr., "New Developments in High Strength Aluminum Alloy Products", Transactions ASM, 1945, vol. 35, pp. 130-155. 8. E.H. Dix, Jr., "Aluminum-Zinc-Magnesium Alloys", Transactions ASM, 1950, vol. 42, pp. 1057-1127. 9. C.J. Peel and P.J.E. Forsyth, "The Effect of Composition Changes on the Fracture Toughness of an Al-Zn-Mg-Cu-Mn Forging Alloy", Metal Science, 1973, vol 7, pp. 121-127. 10. G.T. Hahn and A.R. Rosenfield, "Metallurgical Factors Affecting Fracture Toughness of Aluminum Alloys", Metallurgical Transactions, 1975, vol. 6A, pp. 653-670. 11. M.O. Speidel, "Current Understanding of Stress Corrosion Crack Growth in Aluminum Alloys" in The Theory of Stress Corrosion Cracking in Alloys, NATO Scientific Affairs Division, Brussels, 1971, pp. 289-354. 12. H.Y. Hunsicker, "The Metallurgy of Heat Treatment" in Aluminum, K.R. Van Horn, ed., ASM, Ohio, 1967, vol. I, pp. 109-162. 13. M.E. Fine, "Precipitation Hardening of Aluminum Alloys", Metallurgical Transactions, 1975, vol. 6A, pp. 625-630. 14. D.S. Thompson, "Metallurgical Factors Affecting High Strength Aluminum Alloy Production", Metallurgical Transactions, 1975, vol. 6A, pp. 671-683. 15. M.O. Speidel, "Stress Corrosion Cracking of Aluminum Alloys", Metallurgical Transactions, 1975, vol. 6A, pp. 631-651. 16. R.C. Dorward and K.R. Hasse, "Incubation Effects in Precracked Stress Corrosion Specimens from Al-Zn-Mg-Cu Alloy 7075", Corrosion Science, 1979, vol. 19, pp. 131-140. 17. S. Lim and D. Tromans, "The Stress Corrosion Cracking of AA 7075-T651 Aluminum Alloy in Various Aqueous Solutions", Report MME 499, UBC, 1987. 161 18. RJ. Gest and A.R. Troiano, "Stress Corrosion and Hydrogen Embrittlement in an Aluminum Alloy", Corrosion, 1974, vol. 30, pp. 274-279. 19. E.H. Dix, Jr., "Acceleration of the Rate of Corrosion by High Constant Stresses", Trans AIME, 1940, vol. 137, pp. 11-41. 20. M.O. Speidel, "Interaction of Dislocations with Precipitates in High Strength Aluminum Alloys and Susceptibility to Stress Corrosion Cracking" in Fundamental Aspects of Stress Corrosion Cracking, R.W. Staehle, A.J. Forty, D. van Rooyen, eds., NACE, Houston, Texas, 1967, pp. 561-579. 21. D.A. Verrnilyea, "A Theory for the Propagation of Stress Corrosion Cracks in Metals", Journal of the Electrochemical Society, 1972, vol. 119, pp. 405-407. 22. F.P. Ford, "Current Understanding of the Mechanisms of Stress Corrosion and Corrosion Fatigue" in Environment-Sensitive Fracture: Evaluation and Comparison of Test Methods, S.W. Dean, E.N. Pugh and G.M. Ugiansky, eds., ASTM STP 821, 1984, pp. 32-51. 23. G.M. Scamans, R. Alani and P.R. Swann, "Pre-Exposure Embrittlement and Stress Corrosion Failure in Al-Zn-Mg Alloys", Corrosion Science, 197o, vol. 16, pp. 443-459. 24. D. Hardie, N. J.H. Holroyd and R.N. Parkins, "Reduced Ductility of High-Strength Aluminium Alloy During or After Exposure to Water", Metal Science, 1979, vol. 13, pp. 603-610. 25. M.O. Speidel, "Hydrogen Embrittlement and Stress Corrosion Cracking of Aluminum Alloys", in Hydrogen Embrittlement and Stress Corrosion Cracking, G. Gibala and R.F. Hehemann, eds., ASM Ohio, 1980, pp. 271-296. 26. J. Albrecht, A.W. Thompson and LM. Bernstein, "The Role of Microstructure in Hydrogen-Assisted Fracture of 7075 Aluminum", Metallurgical Transactions, 1979, vol. 10A, pp. 1759-1766. 27. M.R. Louthan, Jr. and R.P. McNitt, "The Role of Test Technique in Evaluating Hydrogen Embrittlement Mechanisms" in Effect of Hydrogen on Behavior of Materials, Proceedings of an Int. Conf., Moran, Wyoming, 1975, pp. 496-506. 28. R.M. Latanision, O.H. Gastine and CP. Compeaus, "Stress Corrosion Cracking and Hydrogen Embrittlement: Differences and Similarities" in Environment-Sensitive Fracture of Engineering Materials, Z.A. Foroulis, ed., TMS-AIME, Warrendale Pa, 1977, pp. 48-71. 29. R.B. Heady, "The Petch-Stables Theory of Hydrogen Embrittlement", Corrosion, 1977, vol. 33, pp. 441-447. 30. R. A. Oriani and P.H. Josephic, "Equilibrium Aspects of Hydrogen-Induced Cracking of Steels", ACTA Metallurgica, 1974, vol. 22, pp. 1065-1074. 31. M.R. Louthan, Jr., "Effects of Hydrogen on the Mechanical Properties of Low Carbon and Austenitic Steels", in Hydrogen in Metals, Proc. Int. Conf. on Effects of Hydrogen on Materials Properties and Selection and Structural Design, LM. Berstein and A.W. Thompson, eds., Champion Pa, ASM, 1973, pp. 53-75. 32. P.C. Paris, M.P. Gomez, E.W. Anderson, "A Realtional Analytic Theory of Fatigue", The Trend in Engineering, 1961, vol. 13, pp. 9-14. 33. T.S. Sudarshan and M.R. Louthan, Jr., "Gaseous Environment Effects on Fatigue Behavior of Metals", International Materials Reviews, 1987, vol. 32, No. 3, pp. 121-151. 162 34. M.O. Speidel, "Stress Corrosion and Corrosion Fatigue Crack Growth in Aluminum Alloys", Proc. of the NATO Advanced Study Institute on Stress Corrosion Research, H. Arup, R.N. Parkins, Denmark, 1975, pp. 113-176. 35. A J . McEvily and R.P. Wei, "Fracture Mechanics And Corrosion Fatigue", in Corrosion Fatigue: Chemistry, Mechanics and Microstructure, O. Devereux, A J . McEvily and R.W. Staehle, eds., NACE, Houston, Texas, 1971, pp. 381-395. 36. J. Lankford and D.L. Davidson, "Fatigue Crack Micromechanisms in Ingot and Powder Metallurgy 7XXX Aluminum Alloys in Air and Vacuum", Acta Metallurgica, 1983, vol. 31, pp. 1273-1284. 37. RJ.H. Wanhill, "Fractography of Fatigue Crack Propagation in 2024-T3 and 7075-T6 Aluminum Alloys in Air and Vacuum", Metallurgical Transactions, 1975, vol. 6A, pp. 1587-1596. . 38. M. Khobaib, C.T. Lynch and F.W. Vahldiek, "Inhibition of Corrosion Fatigue in High Strength Aluminum Alloys", Corrosion, 1981, vol. 36, pp. 285-292. 39. A.K. Vasudevan and P.E. Bretz, "Near-Threshold Fatigue Crack Growth Behavior of 7XXX and 2XXX Alloys: A Brief Review", in Proc. Int. Symp. of Fatigue Crack Growth Threshold Concepts, D.L. Davidson, S. Suresh, eds., Met. Soc. of AIME, Philadelphia, 1983, pp. 25-42. 40. W.E. Krupp, D.W. Hoeppner and E.K. Walker, "Crack Propagation of Aluminum Alloys in Corrosive Environments", in Corrosion Fatigue: Chemistry, Mechanics and Microstructure, O. Devereux, A J . McEvily and R.W. Staehle, eds, NACE, Houston, Texas, 1971, pp. 468-483. 41. P.S. Pao, M. Gao and R.P. Wei, "Environmentally Assisted Fatigue-Crack Growth in 7075 and 7050 Aluminum Alloys", Scripta Metallurgica, 1985, vol. 19, pp. 265-270. 42. R.E. Stoltz and R.M. Pelloux, "Inhibition of Corrosion Fatigue in 7075 Aluminum Alloys", Corrosion, 1973, vol. 29, pp. 13-17. 43. Fu-Shiong Lin and E.A. Starke, Jr., "Mechanisms of Corrosion Fatigue Crack Propagation of 7XXX Aluminum Alloys in Aqueous Environments", in Hydrogen Effects in Metals, Proc. 3rd Int. Conf. on Effect of Hydrogen on Behavior of Materials, LM. Berstein, A.W. Thompson, eds., Carnegie-Mellon University, TMS-AIME, 1980, pp. 485-492. 44. D. Broek, "Elementary Engineering Fracture Mechanics", 3rd Edition, Martinus Nijhoff Publishers, The Hague, 1982, pp. 250-262. 45. R.E. Stoltz and R.M. Pelloux, "Mechanisms of Corrosion Fatigue Crack Propagation in Al-Zn-Mg Alloys", Metallurgical Transactions, 1972, vol. 3, pp. 2433-2441. 46. J.E. Dresty and O.F. Devereux, "The Effect of Specimen Polarization on Fatigue Crack Growth Rates in 7075-T6 Aluminum", Metallurgical Transactions, 1973, vol. 4, pp. 2469-2471. 47. H.F. de Jong, "A Survey of the Development, Properties and Applications of Aluminium-Lithium Alloys", Aluminum, vol. 60, 1984, pp. E587-E593. 48. T.H. Sanders, Jr. and E.A. Starke. Jr., eds., Aluminum-Lithium Alloys, Proc. of the First Int. Aluminum-Lithium Conference, Stone Mountain, Georgia, May 1980, TMS-AIME, Warrendale, PA, 1981. 163 49. E.A. Starke, Jr. and T.H. Sanders, Jr., eds., Aluminum-Lithium Alloys JJ, Proc. of the Second Int. Aluminum-Lithium Conference, Monterey, California, April 1983, TMS-AIME, Warrendale, PA, 1984. 50. C. Baker, P.J. Gregson, S.J. Harris and C.J. Peel, eds., Aluminium-Lithium Alloys III, Proc. of the Third Int. Aluminum-Lithium Conference, Oxford, England, July 1985, The Institute of Metals, London, England, 1986. 51. G. Champier, B. Dubost, D. Miannay and L. Sabetay, eds., 4th International Aluminum Conference, June 1987, les Editions de physique, Paris, France. 52. M.A. Reynolds, A. Gray, E. Creed, R.M. Jordan and A.P. Titchener, "Processing and Properties of Alcan Medium and High Strength Al-Li-Cu-Mg Alloys in Various Product Forms", see ref. 50, pp. 57-65. 53. C.J. Peel, B. Evans and D. McDarmaid, "Current Status of UK Lightweight Lithium-Containing Aluminium Alloys, see ref. 50, pp. 26-36. 54. P.E. Bretz and R.R. Sawtell, "Alithalite Alloys: Progress, Products and Properties", see ref. 50, pp. 47-56. 55. LE Roy, R. Mace, D. Marchive, P. Meyer, R. Nossent and F. Schlecht. - "Status Report on the Development of Aluminum-Lithium at Pechiney", see ref. 51, pp. 33-39. 56. G.W. Lorimer, "Precipitation in Aluminium Alloys", Precipitation Processes in Solids, Proc. of a Symp. New York, 1976, K.C. Russel and H.I. Aaronson, eds., TMS-AIME, New York, 1978, pp. 87-119. 57. T.H. Sanders, Jr. and E.A. Starke. Jr., "Overview of the Physical Metallurgy in the Al-Li-X Systems", see ref. 49, pp. 1-16. 58. N. Kanani, S.P. Abeln and G.R. Abbaschian, "Ueber das Aushaertungsverhalten einer AlLiMgZr Legierung", Aluminum, 1985, vol. 61, pp. 358-361. 59. M. Ahmad and T. Ericsson, "Coarsening of 8', Tu S' Phases and Mechanical Properties of Two Al-Li-Cu-Mg Alloys", see ref. 50, pp. 509-515. 60. K. Welpmann, M. Peters and T.H. Sanders, Jr., "Age Hardening Behavior of DTDXXXA", see ref. 50, pp. 524-529. 61. D.T. Markey, R.R. Biedermann and A.J. McCarthy, "Effect of Cold Deformation on Mechanical Properties and Microstructure of Alcan XXXA", see ref. 50, pp. 173-183. 62. B. Noble, S.J. Harris and K. Dinsdale, "The Elastic Modulus of Aluminium-Lithium Alloys", J. Mater. Sci., 1982, vol. 17, pp. 461-468. 63. D. Webster, "Temperature Dependence of Toughness in Various Aluminum-Lithium Alloys", see ref. 50, pp. 602-609. 64. A.K. VasudeVan, A.C. Miller and M.M. Kersker, "Contribution of Na-Segregation to Fracture Behavior of Al-11.4at%Li Alloys", see ref. 49, pp. 181-200. 65. J.A. Wert and J.B. Lumsden, "Intergranular Fracture in an Al-Li-Cu-Mg-Zr Alloy", Scr. Metall., 1985, vol. 19, pp. 205-209. 66. K. Welpmann, M. Peters and T.H. Sanders, Jr., "Aluminium-Lithium Alloys It: Mechanical Properties", Aluminium, 1984, vol. 60, pp. E709-E712. 67. B. Noble, SJ. Harris and K. Dinsdale, "Yield Characteristics of Aluminium-Lithium Alloys", Met. Sci., 1982, vol. 16, pp. 425-430. 164 68. K.V. Jata and E.A. Starke, Jr., "Fatigue Crack Growth and Fracture Toughness Behavior of Al-Li-Cu Alloy", see ref. 50, pp. 247-256. 69. T.H. Sanders, Jr. and E.A. Starke. Jr., "The Effect of Slip Distribution on the Monotonic and Cyclic Ductility of Al-Li Binary Alloys", Acta MetalL, 1982, vol. 30, pp. 927-939. 70. A.K. Vasudevan and S. Suresh, "Microstructural Effects on Quasi-static Fracture Mechanisms in Al-Li Alloys: The Role of Crack Geometry", Mater. Sci. & Eng., 1985, vol. 72, pp. 37-49. 71. J. Dhers, J. Driver and A. Fourdeux, "Cyclic Deformation of Binary Al-Li Alloys", see ref. 50, pp. 233-238. 72. O. Jensrud and N. Ryum, "The Development of Microstructures in Al-Li Alloys", Mater. Sci. & Eng., 1984, vol. 64, pp. 229-236. 73. F.S. Lin, "The Effect of Grain Structure on the Fracture Behavior and Tensile Properties of an Al-Li-Cu Alloy", Scr. MetalL, 1982, vol. 16, pp. 1295-1300. 74. A.K. Vasudevan, E.A. Ludwiczak, S.F. Baumann, R.D. Doherty and M.M. Kersker, "Fracture Bahavior in Al-Li Alloys: Role of Grain Boundary 8', Mater. Sci. & Eng., vol. 72, pp. L25-L30. 75. N.J. Owen, D.J. Field and E.P. Butler, "Initiation of Voiding at Second-Phase Particles in a Quaternary Al-Li Alloy", see ref. 50, pp. 576-583. 76. K.K. Sankaran, J.E. O'Neal and S.M.L. Sastry, "Effects of Second-Phase Dispersoids on Deformation Behavior of Al-Li Alloys", Metallurgical Transactions A, 1983, vol. 14A, pp. 2174-2178. 77. P.J. Gregson and H.M. Flower, "Microstructural Control of Toughness in Aluminium-Lithium Alloys", Acta MetalL, 1985, vol. 33, pp. 527-537. 78. R.T. Chen and E.A. Starke, Jr., "Microstructure and Mechanical Properties of Mechanically Alloyed, Ingot Metallurgy and Powder Metallurgy Al-Li-Cu-Mg • Alloys", Mater. Sci. & Eng., 1984, vol. 67, pp. 229-245. 79. W. Ruch and E.A. Starke, Jr., "Fatigue Crack Propagation in Mechanically Alloyed Al-Li-Mg Alloys", see ref. 50, pp. 121-130. 80. P. Niskanen, T.H. Sanders, Jr., J.G. Rinker and M. Marek, "Corrosion of Aluminum Alloys Containing Lithium", Corrosion Science, 1982, vol. 22, pp. 283-304. 81. J.G. Rinker, M. Marek and T.H. Sanders, Jr., "Microstructure, Toughness and Stress Corrosion Cracking Behavior of Aluminum Alloy 2020", Mater. Sci. & Eng., 1984, vol. 64, pp. 203-221. 82. N.J.H. Holoryd, A. Gray, G.M. Scamans and R. Hermann, "Environment-Sensitive Fracture of Al-Li-Cu-Mg Alloys", see ref. 50, pp. 310-324. 83. M. Ahmad, "Correlation Between Ageing heat treatments, microstructure and stress corrosion properties of Al-Li-Cu-Mg Alloys", see ref. 51, pp. 871-879. 84. A.K. Vasud6van, P.R. Ziman, S.C. Jha and T.H. Sanders, Jr., "Stress Corrosion Resistance of Al-Cu-Li-Zr Alloys", see ref. 50, pp. 303-309. 85. P.P. Pizzo, R.P. Galvin and H.G. Nelson, "Stress Corrosion Behavior of Aluminum-Lithium Alloys in Aqueous Salt Environments", see ref. 49, pp. 627-656. 86. A. Gray, "Factors Influencing the Environmental Behavior of Aluminium-Lithium Alloys", see ref. 51, pp. 891-904. 87. L. Christodoulou, L. Struble and J.R. Pickens, "Stress Corrosion Cracking in Al-Li Binary Alloys", see ref. 49, pp. 561-580. 88. J.B. Lumsden and A.T. Allen, "The Stress Corrosion Cracking Behavior of AlLi Alloy 8090", Corrosion Science, 1988, vol. 44, pp. 527-532. 89. B.B. Bavarian and M. Zamanzadeh, "Localized Corrosion of Al-Li Alloys in Aqueous Environments", Corrosion 88, 1988, paper 386, pp. 1-21. 90. J.G. Craig, R.C. Newman, M.R. Jarrett and N.J.H. Holroyd, "Local Chemistry of Stress-Corrosion Cracking in Al-Li-Cu-Mg Alloys", see ref. 51, pp. 825-833. 91. E.J. Coyne, Jr., T.H. Sanders, Jr. and E.A. Starke, Jr., "The Effect of Microstructure and Moisture on the Low Cycle Fatigue and Fatigue Crack Propagation of Two Al-Li-X Alloys", see ref. 48, pp. 293-306. 92. R.J. Donahue, H. Mcl. Clark, P. Atanmo, R. Kumble and A.J. McEvily, "Crack Opening Displacement and the Rate of Fatigue Crack Growth", Inter. J. of Fracture Mech., 1972, vol. 8, pp. 209-219. 93. A.K. Vasud6van, P.E. Bretz and A.C. Miller, "Fatigue Crack Growth Behavior of Aluminum Alloy 2020", Mater. Sci. & Eng., 1984, vol. 64, pp. 113-122. 94. K.T. Venkateswara Rao, W. Yu and R.O. Ritchie, "Fatigue Crack Propagation in Aluminum-Lithium Alloy 2090: Part U. Small Crack Behavior", Metallurgical Transactions A, 1988, vol. 19A, pp. 563-569. 95. M. Peters, V. Bachmann and K. Welpmann, "Fatigue Crack Propagation Behavior of the Al-Li Alloy 8090 Compared to 2024", see ref. 51, pp. 785-791. 96. R. Tintillier, H.S. Yang, N. Ranganathan and J. Petit, "Near Threshold Fatigue Crack Growth in a 8090 Lithium Containing Al Alloy", see ref. 51, pp. 777-784. 97. N. Ohrloff, A. Gysler and G. Lutjering, "Fatigue Crack Propagation Behavior of 2091 T8 and 2024 T3 Under Constant and Variable Amplitude Loading", see ref. 51, pp. 801-807. 98. F.L. Haddleton, S. Murphy and T.J. Griffin, "Fatigue and Corrosion Fatigue of 8090 Al-Li-Cu-Mg Alloy", see ref. 51, pp. 809-815. 99. T. Magnin, P. Rieux, C. Lespinasse and C. Bathias, "Fatigue Crack Initiation and Propagation Properties of Al-Li-Cu Alloys in Air and in Aqueous Corrosive Solutions", see ref. 51, pp. 817-822. 100. J. Petit, S. Suresh, A.K. VasudeVan and R.C. Malcolm, "Constant Amplitude and Post-Overload Fatigue Crack Growth in Al-Li Alloys", see ref. 50, pp. 257-262. 101. K. Endo and K. Komai, "Effects of Stress Wave Form and Cyclic Frequency on Low Cycle Corrosion Fatigue", in Corrosion Fatigue: Chemistry, Mechanics and Microstructure, O. Devereux, A.J. McEvily and R.W. Staehle, eds, NACE, Houston, Texas, 1971, pp. 437-450. 102. T. Magnin and P. Rieux, "The Relation Between Corrosion Fatigue and Stress Corrosion Cracking in Al-Zn-Mg Alloys, Scripta Metallurgica, 1987, vol. 21, pp. 907-911. 103. R.P. Wei, "Some Aspects of Environment-Enhanced Fatigue-Crack Growth", Engineering Fracture Mechanics, 1970, vol. 1, pp. 633-651. 104. M. Khobaib and C T . Lynch, "Slow-Strain Rate Testing of Al 7075-T6 in Controlled Atmospheres" in Environment-Sensitive Fracture: Evaluation and Comparison of Test Methods, S.W. Dean, E.N. Pugh and G.M. Ugiansky, eds., ASTM STP 821, Philadelphia, 1984, pp. 242-255. 166 105. G.M. Ugiansky, C.E. Johnson, D.S. Thompson and E.H. Gillespie, "Slow Strain-Rate Stress Corrosion Testing of Aluminum Alloys" in Stress Corrosion Cracking - The Slow Strain-Rate Technique, G.M. Ugiansky, J.H. Payer, eds., ASTM STP 665, 1979, pp. 254-265. 106. P.P. Pizzo, R.P. Galvin and H.G. Nelson, "Utilizing Various Test Methods to Study the Stress Corrosion Behavior of Al-Li-Cu Alloys" in Environment-Sensitive Fracture: Evaluation and Comparison of Test Methods, S.W. Dean, E.N. Pugh and G.M. Ugiansky, eds., ASTM STP 821,1984, pp. 173-201. 107. R. Braun and H. Buhl, "Corrosion Behavior of Al-Li-Cu-Mg alloy 8090-T651", see ref. 51, pp. 843-849. 108. F.P. Ford and M. Silverman, "Effect of Loading Rate on Environmentally Controlled Cracking of Sensitized 304 Stainless Steel in High Purity Water", Corrosion, 1980, vol. 36, pp. 597-603. 109. P.L. Andresen, "Environmentally Assisted Growth Rate Response of Nonsensitized AISI316 Grade Stainless Steels in High Temperature Water", Corrosion, 1988, vol. 44, pp. 450-460. 110. S.J. Hudak, Jr., D.L. Davidson and R.A. Page, "The Role of Crack-Tip Deformation in Corrosion Fatigue Crack Growth" in Proc. Int. Symp. on Embrittlement by the Localized Crack Environment, R.P. Gangloff, ed., TMS-AIME, 1983, pp. 173-198. 111. R.N. Parkins, "A Critical Evaluation of Current Environment-Sensitive Fracture Test Methods", in Environment-Sensitive Fracture: Evaluation and Comparison of Test Methods, S.W. Dean, E.N. Pugh and G.M. Ugiansky, eds., ASTM STP 821, 1984, pp. 5-31. 112. R.J. Selines and R.M. Pelloux, "Effect of Cyclic Stress Wave Form on Corrosion Fatigue Crack Propagation in Al-Zn-Mg Alloys", Metallurgical Transactions, 1972, vol. 3, pp. 2525-2531. 113. J.M. Barsom, "Effect of Cyclic Stress Form on Corrosion Fatigue Crack Propagation Below K I S C C in a High Yield Strength Steel" in Corrosion Fatigue: Chemistry, Mechanics and Microstructure, O. Devereux, A.J. McEvily and R.W. Staehle, eds, NACE, Houston, Texas, 1971, pp. 424-436. 114. CS. O'Dell and B.F. Brown, "Control of Stress Corrosion Cracking by Inhibitors" in Corrosion Control by Coatings, H. Leidheiser, Jr., ed., Science Press, Priceton, 1979, pp. 339-348. 115. V.S. Agarwala and J.B. Boodey, "Control of Stress Corrosion Cracking in High Strength Aluminum Alloys" in Environmental Degradation of Engineering Materials III, M.R. Louthan, R.P. McNitt, eds., Sisson, Penns. State University Press, 1987, pp. 341-349. 116. R.N. Parkins, "Inhibition In the Context of Environment-Sensitive Cracking", Corrosion 89, New Orleans, 1989, paper 139, pp. 1-23. 117. V.S. Agarwala, "Modification of Crack-Tip Chemistry to Inhibit Corrosion and Stress Corrosion Cracking in High Strength Alloys" in Proceedings of an Int. Symposium, Embrittlement by the Localized Crack Environment, R.P. Gangloff, ed., The Metallurgical Society of AIME, Philadephia, Pennsylvania, 1983, pp. 405-418. 118. R.J.H. Wanhill, "Formation of Brittle Fatigue Striations", Corrosion, 1975, vol. 31, pp. 66-71. 167 119. J.L. Nelson and E.N. Pugh, "The Occurrence of Transgranular Cleavage-Like Fracture in an Al-Zn-Mg Alloy During Tensile Testing", Metallurgical Transactions A, 1975, vol. 6A, pp. 1459-1460. 120. E.N. Pugh, "Progress Toward Understanding the Stress Corrosion Problem", Corrosion-NACE, 1985, vol. 41, pp. 517-526. 121. W.J. Helfrich, "Technical Note Fractography of Stress Corrosion Cracks in Aluminum Alloy 7075", Corrosion-NACE, 1973, vol. 29, pp. 316-318. 122. A.J. McEvily, Jr., J.B. Clark and A.P. Bond, "Effect of Thermal-Mechanical Processing on the Fatigue and Stress-Corrosion Properties of an Al-Zn-Mg Alloy", Transaction ASM, 1967, vol. 60, pp. 661-671. 123. G.M. Scamans, "Discontinuous Propagation of Stress Corrosion Cracks in Al-Zn-Mg Alloys", Scripta Metallurgica, 1979, vol. 13, pp. 245-250. 124. G.M. Scamans, "Evidence for Crack-Arrest Markings on Intergranular Stress Corrosion Fracture Surfaces in Al-Zn-Mg Alloys", Metallurgical Transactions A, 1980, vol. 11 A, pp. 846-850. 125. R. Hermann, "Environmentally Assisted Fracture of Aluminum Alloys", Corrosion Science, 1988, vol. 44, pp. 685-690. 126. S.P. Lynch, "A Comparative Study of Stress-Corrosion Cracking, Hydrogen-Assisted Cracking and Liquid-Metal Embrittlement in Al, Ni, Ti and Fe-Based Alloys", in Hydrogen Effects in Metals, LM. Bernstein, A.W. Thompson, eds., The Metallurgical Society of AIME, Pittsburgh, Pennsylvania, 1989, pp. 863-871. 127. A.J. McEvily, "On the Quantitative Analysis of Fatigue Crack Propagation" in Fatigue Mechanisms: Advances in Quantitative Measurement of Physical Damage, J. Lankford, D.L. Davidson, W.L. Morris and R.P. Wei, eds., ASTM STP 811, 1983, pp. 283-312. 128. C. Laird, "The Influence of Metallurgical Structure on the Mechanisms of Fatigue Crack Propagation" in Fatigue Crack Propagation, ASTM STP 415, 1967, pp. 131-180. 129. A.K. Head, "The Growth of Fatigue Cracks", Philosophical Magazine, 1953, vol. 44, pp. 925-938. 130. C. Laird and G.C. Smith, "Crack Propagation in High Stress Fatigue", Philosophical Magazine, 1962, vol. 7, pp. 847-857. 131. P. Neumann, "New Experiments Concerning the Slip Processes at Propagating Fatigue Cracks-I", Acta Metallurgica, 1974, vol. 22, pp. 1155-1165. 132. R.M.N. Pelloux, "Crack Extension by Alternating Shear", Engineering Fracture Mechanics, 1970, vol. 1, pp. 697-704. 133. B. Tomkins and W.D. Biggs, "Low Endurance Fatigue in Metals and Polymers", Journal of Materials Science 4, 1969, pp. 544-553. 134. P.J.E. Forsyth, "Fatigue Damage and Crack Growth in Aluminum Alloys", Acta Metallurgica, 1963, vol. 11, pp. 703-715. 135. R.M.N. Pelloux, "Mechanisms of Formation of Ductile Fatigue Striations", Transactions of the ASM, 1969, vol. 62, pp. 281-285. 136. C.Q. Bowles and D. Broek, "On the Formation of Fatigue Striations", International Journal of Fracture Mechanics, 1972, vol. 8, pp. 75-85. 168 137. C. Fong and D. Tromans, "Stage I Corrosion Fatigue Crack Crystallography in Austenitic Stainless Steel (316L)", Metallurgical Transactions, 1988, vol. 19A, pp. 2765-2773. 138. K J . Nix and H.M. Flower, "The Micromechanisms of Fatigue Crack Growth in a Commercial Al-Zn-Mg-Cu Alloy", ACTA Metallurgica, 1982, vol. 30, pp. 1549-1559. 139. D.A. Meyn, "Observations on Micromechanisms of Fatigue-Crack Propagation in 2024 Aluminum", Transactions of the ASM, 1968, vol. 61, pp. 42-51. 140. K.T. Venkateswara Rao, W. Yu and R.O. Ritchie, "Fatigue Crack Propagation in Aluminum-Lithium Alloy 2090: Part I. Long Crack Behavior", Metallurgical Transactions A, 1988, vol. 19A, pp. 549-561. 141. A.K. Vasudevan and S. Suresh, "Lithium-Containing Aluminum Alloys: Cyclic Fracture", Metallurgical Transactions A, 1985, vol. 16A, pp. 475-477. 142. K.V. Jata and E.A. Starke, Jr., "Fatigue Crack Growth and Fracture Toughness Behavior of an Al-Li-Cu Alloy", Metallurgical Transactions A, 1986, vol. 17A, pp. 1011-1026. 143. R.P. Wei and J.D. Landes, "Corrosion Between Sustained - Load and Fatigue Crack Growth in High-Strength Steels", Materials Research and Standards, 1969, vol. 9, pp. 25-46. 144. W.W. Gerberich, J.P. Birat and V.F. Zackay, "On the Superposition Model for Environmentally-Assisted Fatigue Crack Propagation" in Corrosion Fatigue: Chemistry, Mechanics and Microstructure, O. Devereux, A J . McEvily and R.W. Staehle, eds., NACE, Houston, Texas, 1971, pp. 396-408. 145. N.J. Holroyd and D. Hardie, "Corrosion Fatigue of 7000 Series Aluminum Alloys" in ASTM Symposium on Environment-Sensitive Fracture, Evaluation and Comparison of Test Methods, Maryland, April 1982, S.W. Dean, E.N. Pugh, G.M. Ugiansky, eds., ASTM-STP 821, 1982, pp. 534-547. 146. R.N. Parkins, "Factors Influencing Stress Corrosion Crack Growth Kinetics", Corrosion, 1987, vol. 43, pp. 130-139. 147. D.P.G. Lidbury, "The Estimation of Crack Tip Strain Rate Parameters Characterizing Environment Assisted Crack Growth Data" in Proc. of an International Symposium, Embrittlement by the Localized Crack Environment, R.P. Gangloff, ed., The Metallurgical Society of AIME, Philadelphia, Pennsylvania, 1983, pp. 149-172. 148. P.M. Scott and A.E. Truswell, "Corrosion Fatigue Crack Growth in Reactor Pressure Vessel Steels - Measurement and Application" in Structural Integrity of Light Water Reactor Components, L.E. Steele, K.E. Stohlkopf and L.H. Larsson, eds., Appl. Science Publishers, London, 1982, pp. 287-309. 149. B. Tomkins and J. Wareing, "Elevated-Temperature Fatigue Interactions In Engineering Materials", Metal Science, 1977, pp. 414-424. 150. J. Lankford and F.N. Kusenberger, "On Crack Tip Yielding During Fatigue Cycling of a High-Strength Steel", Philas Magazine, 1972, pp. 1485-1490. 151. F.P. Ford and P.L. Andresen, "Stress Corrosion Cracking of Low Alloy Steels in 288°C Water", Corrosion, 1989, paper 498, pp. 1-19. 152. J. Congleton, T. Shoji and R.N. Parkins, "The Stress Corrosion Cracking of Reactor Pressure Vessel Steel in High Temperature Water", Corrosion Science, 1985, vol. 25, pp. 633-650. 169 153. ASM Metal Handbook, Desk Edition, H.E. Boyer and T.L. Gall, eds., ASM, Metals Park, Ohio, 1985, pp. 6.20-6.47. 154. R. Grimes, T. Davis, H.J. Saxty and J.E. Fearon, "Progress to Aluminium-Lithium Semi-Fabricated Products", see ref. 51, pp. 3.11-3.24. 155. D. Broek, "Elementary Engineering Fracture Mechanics", 3rd edition, Martinus Nijhoff Publishers, The Hague, 1982, pp. 115-141. 156. JJ. Dickson, "Fractography: A Tool in Failure Analysis and Fracture Research", de Ecale Poly technique, Montreal, Canada. 157. M.F. Kanninen and C H . Popelar, "Advanced Fracture Mechanics", Oxford University Press, New York, 1985. Appendix I Description of orientations in the rolled plate material 170 Appendix I I 171 Data from slow monotonic loading experiments. Measurements of crack length on the specimen surface and via the compliance method. Specimen A10 Average crack length after fatigue:65.00mm Average crack length after SCC:68.91mm Crack length determined via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0.00 65.00 3.08 5.00 65.00 5.65 10.00 65.00 8.22 15.00 65.00 10.79 20.00 65.09 13.35 65.10 9.67E-09 22.50 65.22 14.64 65.21 1.37E-08 25.00 65.34 15.89 65.35 1.77E-08 27.50 65.51 17.18 65.52 2.18E-08 30.00 65.76 18.42 65.74 2.58E-08 32.50 65.98 19.69 65.99 2.98E-08 33.75 66.10 20.29 66.13 3.19E-08 35.00 66.27 20.89 66.28 3.39E-08 37.50 67.08 21.97 40.00 68.76 22.44 40.25 68.91 22.54 Crack length obtained from surface measurements: 2. Order From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 63.85 3.080 12.25 63.85 9.370 15 63.85 10.790 17 63.85 11.810 19 63.85 12.840 63.91 2.78E-10 20.25 63.96 13.480 63.92 4.14E-09 22.8 64.09 14.790 64.00 1.20E-08 24.5 64.09 15.640 64.09 1.73E-08 25 64.09 15.89 64.12 1.88E-08 25.5 64.09 16.14 64.15 2.04E-08 36.25 65.59 21.43 65.59 5.36E-08 38 65.71 22.06 39.00 65.87 22.25 39.50 66.00 22.34 40.25 67.70 22.54 172 Specimen A l l Average crack length after fatigue: 64.91mm Average crack length after SCC:69.00mm Crack length deterrnined via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0.00 64.91 2.96 5.00 64.91 5.14 10.00 64.91 7.33 15.00 64.91 9.51 20.00 64.92 11.69 25.00 64.95 13.85 64.91 6.03E-09 27.50 65.01 14.91 64.98 9.75E-09 30.00 65.07 15.97 65.08 1.35E-08 32.50 65.18 17.00 65.22 1.72E-08 35.00 65.33 18.00 65.39 2.09E-08 36.25 65.45 18.48 65.49 2.28E-08 37.50 65.58 18.94 65.60 2.46E-08 38.75 65.74 19.40 65.71 2.65E-08 40.00 65.90 19.83 65.84 2.84E-08 42.50 66.20 20.80 45.00 66.54 21.75 49.00 69.00 22.73 Crack length from surface measurements: 2.0rder From a vs. t 2.Order Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(rnm) v(m/s) 0.00 63.60 2.96 12.50 63.60 8.42 18.00 63.60 10.82 20.00 63.60 11.69 25.00 63.65 13.85 31.00 63.80 16.38 63.71 6.65E-09 37.00 63.80 18.75 63.91 1.17E-08 38.75 63.90 19.40 63.99 1.31E-08 41.25 64.23 20.32 64.12 1.52E-08 43.5 64.43 21.180 44.5 64.73 21.56 49 66.8 22.73 173 Specimen A16 Average crack length after fatigue:67.66mm Average crack length after SCC:72.46mm Crack length determined via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 67.66 3.10 10 67.66 5.61 20 67.66 8.12 30 67.66 10.64 40 67.66 13.15 50 67.66 15.66 55 67.66 16.91 67.68 9.62E-09 57.5 67.81 17.56 67.77 1.03E-08 60 67.88 18.21 67.87 1.09E-08 62.5 67.91 18.82 67.97 1.15E-08 65 68.10 19.45 < 68.07 1.22E-08 67.5 68.38 20.04 70 68.80 20.59 72.5 69.05 21.30 75 69.94 21.55 78.7 72.46 21.36 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 65.95 3.10 65.95 1.00E-15 24 65.95 9.13 65.96 1.00E-15 47 66.05 14.90 66.02 3.50E-09 59.75 66.21 18.14 66.25 6.51E-09 71.2 66.59 20.93 66.58 9.21E-09 78.7 70.30 21.36 66.85 1.10E-08 174 Specimen A4 Average crack length after fatigue:68.25mm Average crack length after SCC:78.54mm Crack length deterrnined via compliance measurements: 2. Order From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 68.25 3.21 4.5 68.25 3.84 10 68.25 4.62 15 68.25 5.34 20 68.25 6.09 25 68.25 6.84 30 68.25 7.62 35 68.25 8.41 40 68.25 9.21 45 68.42 10.02 68.40 2.16E-09 50 68.52 10.86 68.46 4.43E-09 55 68.64 11.72 68.56 6.70E-09 60 68.81 12.57 68.70 8.97E-09 65 68.85 13.41 68.88 1.12E-08 70 69.15 14.28 69.11 1.35E-08 75 69.41 15.15 69.37 1.58E-08 80 69.54 16.07 69.67 1.81E-08 85 69.90 16.94 70.02 2.03E-08 90 70.36 17.77 70.41 2.26E-08 95 70.80 18.62 70.83 2.49E-08 100 71.45 19.36 71.30 2.71E-08 105 72.30 20.01 110 73.20 20.75 115 75.27 20.97 119 78.54 20.26 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 66.8 3.21 34 66.81 8.25 66.81 1.21E-09 58.5 67.11 12.31 67.12 5.87E-09 70.5 67.48 14.37 67.43 8.15E-09 75.5 67.61 15.25 67.58 9.10E-09 79.5 67.65 15.97 67.72 9.86E-09 94.5 68.29 18.53 68.33 1.27E-08 95.2 68.42 18.64 68.36 1.28E-08 99.2 68.49 19.24 68.55 1.36E-08 102.5 68.76 19.68 68.72 1.42E-08 175 Specimen A6 Average crack length after fatigue:65.37mm Average crack length after SCC:69.24mm Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 63.20 2.93 5 63.20 3.81 19.5 63.20 6.31 25.5 63.20 7.33 40.5 63.49 9.87 63.50 6.54E-09 43.5 63.51 10.39 63.58 7.00E-09 46.5 63.72 10.87 63.65 7.46E-09 53.5 63.87 12.01 63.85 8.53E-09 65 64.31 13.88 64.24 1.03E-08 73.5 64.61 15.29 64.58 1.16E-08 79 64.85 16.09 64.82 1.24E-08 89.25 65.06 17.73 65.31 1.40E-08 101.5 66.09 19.44 65.97 1.59E-08 113.5 66.49 21.24 118 66.80 21.87 Specimen A91 (One cycle) Average crack length after fatigue:66.29mm Average crack length after SCC:68.38mm 176 Crack length determined via compliance measurements: 2.0rder a vs. t Curvefit t(hrs) a(mm) K(MNmA-3/2) a(mm) From 2.0rder da/dt v(m/s) Loading: 0 10 20 30 40 50 60 70 80 85 66.29 66.29 66.29 66.29 66.82 66.75 67.08 67.80 67.90 68.18 2.89 4.66 6.44 8.21 10.07 11.87 13.66 15.41 17.15 17.99 66.36 66.61 66.90 67.22 67.58 67.97 68.18 6.52E-09 7.50E-09 8.48E-09 9.46E-09 1.04E-08 1.14E-08 1.19E-08 Unloading: 90 67.44 17.55 100 67.58 15.62 110 67.83 13.66 120 68.05 11.76 130 68.29 9.91 140 68.38 8.02 150 68.38 6.31 160 68.38 4.52 165 68.38 3.65 Specimen A91 (continued) 177 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) Loading: 24.1 64.15 7.16 64.17 1.9E-09 32.6 64.26 8.69 64.24 2.8E-09 47.6 64.50 11.44 64.43 4.3E-09 54.6 64.55 12.69 64.55 5.0E-09 72.6 64.90 15.86 64.93 6.9E-09 75.6 64.91 16.38 65.01 7.2E-09 81.1 65.05 17.33 65.16 7.7E-09 85.9 65.46 17.91 65.30 8.2E-09 Unloading: 96.6 65.55 16.27 104.4 65.55 14.76 126.6 65.89 11.28 147.6 65.89 6.72 168.8 65.95 2.99 178 Specimen A55 Average crack length after fatigue:64.43rnm Average crack length after SCC:66.44mm Crack length deterrriined via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 64.43 2.68 5 64.43 3.36 10 64.43 4.05 15 64.43 4.73 20 64.43 5.42 25 64.43 6.10 30 64.43 6.79 35 64.43 7.47 40 64.43 8.16 45 64.43 8.84 50 64.43 9.53 64.38 7.28E-09 55 64.55 10.19 64.51 7.74E-09 60 64.60 10.87 64.66 8.20E-09 65 64.80 11.56 64.81 8.66E-09 70 64.88 12.26 64.97 9.12E-09 75 65.06 12.95 65.14 9.58E-09 80 65.37 13.64 65.31 l.OOE-08 85 65.48 14.28 65.50 1.05E-08 90 65.77 14.95 65.69 1.10E-08 95 65.98 15.62 65.89 1.14E-08 100 66.13 16.30 66.10 1.19E-08 105 66.30 16.94 66.32 1.23E-08 109.2 66.44 17.48 66.51 1.27E-08 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(ram) K(MNmA-3/2) a(mm) v(m/s) 0 63.05 2.68 63.05 1.00E-15 70.2 63.47 12.29 63.45 5.71E-09 78.7 63.67 13.46 63.64 6.72E-09 98.7 64.09 16.12 64.21 9.08E-09 109.2 64.65 17.48 64.58 1.03E-08 179 Specimen A50 Average crack length after fatigue:65.03mm Average crack length after SCC:69.5mm Crack length determined via compliance measurements 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 65.03 2.87 10 65.03 3.86 20 65.03 4.85 30 65.03 5.84 40 65.03 6.83 50 65.03 7.82 60 65.03 8.81 65.03 4.73E-09 70 65.16 9.82 65.16 6.27E-09 80 65.30 10.82 65.30 7.81E-09 90 65.77 11.80 65.77 9.36E-09 100 66.12 12.77 66.12 1.09E-08 110 66.56 13.73 66.56 1.24E-08 120 66.99 14.69 66.99 1.40E-08 130 67.62 15.61 67.62 1.55E-08 140 67.99 16.50 67.99 1.71E-08 150 68.76 17.34 68.76 1.86E-08 160 69.46 18.16 69.46 2.02E-08 160.5 69.50 18.23 69.50 2.02E-08 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 63.85 2.87 13.75 63.85 4.23 38.25 63.86 6.66 68.75 64.82 9.69 64.21 4.01E-09 118.5 64.82 14.55 65.39 9.20E-09 133.75 65.82 15.94 65.94 1.08E-08 157.75 67.14 17.98 66.98 1.33E-08 160.5 67.25 18.23 67.11 1.36E-08 Specimen A13 Average crack length after fatigue:63.50mm Average crack length after SCC:75.82mm Crack length determined via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 65.50 ' 4.61 5 65.50 4.87 10 65.50 5.12 15 65.50 5.38 20 65.50 5.64 25 65.50 5.89 30 65.50 6.15 35 65.50 6.40 40 65.50 6.66 45 65.50 6.91 50 65.50 7.17 55 65.50 7.42 60 65.50 7.68 65 65.50 7.94 70 65.50 8.19 75 65.50 8.45 80 65.50 8.70 85 65.50 8.96 90 65.50 9.21 95 65.50 9.47 100 65.50 9.73 65.58284 1.28E-09 105 65.51 9.98 65.60815 1.53E-09 110 65.54 10.26 65.63792 1.78E-09 115 65.63 10.50 65.67212 2.02E-09 120 65.72 10.74 65.71076 2.27E-09 125 65.82 10.98 65.75385 2.52E-09 130 65.90 11.23 65.80137 2.76E-09 135 65.94 11.48 65.85334 3.01E-09 140 66.03 11.73 65.90975 3.26E-09 145 66.04 11.99 65.97059 3.50E-09 150 66.10 12.25 66.03589 3.75E-09 155 66.18 12.49 66.10562 4.00E-09 160 66.24 12.74 66.17979 4.24E-09 165 66.26 13.01 66.25840 4.49E-09 170 66.32 13.26 66.34146 4.74E-09 175 66.38 13.51 66.42896 4.98E-09 180 66.44 13.76 66.52089 5.23E-09 185 66.56 13.99 66.61727 5.48E-09 190 66.65 14.22 66.71809 5.72E-09 195 66.82 14.48 66.82335 5.97E-09 200 66.96 14.73 66.93306 6.22E-09 205 67.01 14.99 67.04720 6.46E-09 210 67.24 15.24 67.16578 6.71E-09 Specimen A13 (continued) 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 215 67.25 15.49 67.28881 6.96E-09 220 67.36 15.72 67.41628 7.20E-09 225 67.49 15.97 67.54819 7.45E-09 230 67.64 16.21 67.68453 7.70E-09 235 67.79 16.44 67.82533 7.95E-09 240 68.01 16.66 67.97056 8.19E-09 245 68.15 16.87 68.12023 8.44E-09 250 68.29 17.11 68.27435 8.69E-09 255 68.44 17.36 68.43290 8.93E-09 260 68.57 17.59 68.59590 9.18E-09 265 68.76 17.81 68.76334 9.43E-09 270 68.96 18.03 68.93521 9.67E-09 275 69.08 18.26 69.11154 9.92E-09 280 69.32 18.47 69.29230 1.02E-08 285 69.47 18.66 69.47750 1.04E-08 290 69.68 18.88 69.66714 1.07E-08 295 69.85 19.09 69.86123 1.09E-08 300 70.08 19.30 70.05976 1.12E-08 305 70.24 19.49 70.26272 1.14E-08 310 70.46 19.70 70.47013 1.16E-08 315 70.74 19.87 70.68198 1.19E-08 320 70.98 20.07 325 71.16 20.26 330 71.40 20.47 335 71.60 20.69 340 71.96 20.83 345 72.30 20.98 350 72.58 21.14 355 72.85 21.31 360 73.06 21.48 365 73.31 21.60 370 73.73 21.69 375 75.13 21.45 380 75.47 21.51 385 75.82 21.56 Specimen A13 (continued) Crack length obtained from surface measurements: 2. Order From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 63.65 4.610 20 63.65 5.640 47 63.65 7.020 69 63.68 8.140 85 63.71 8.960 92 63.71 9.320 63.69922 1.81E-10 121 63.89 10.790 63.78124 1.39E-09 143 64.22 11.890 63.92770 2.31E-09 158 64.27 12.640 64.06922 2.93E-09 169 64.34 13.210 64.19447 3.39E-09 181 64.34 13.810 64.35183 3.89E-09 205 64.6 14.99 64.73141 4.89E-09 217 64.71 15.58 64.95362 5.39E-09 229 64.9 16.16 65.19746 5.89E-09 253 65.56 17.26 65.75000 6.90E-09 278 66.52 18.38 66.41753 7.94E-09 289.5 66.75 18.86 66.75610 8.42E-09 309 67.6 19.66 67.37558 9.23E-09 325.5 67.82 20.28 333 68.18 20.6 348.5 68.97 21.09 372.5 70.19 21.69 374.5 72.05 21.5 383 74.4 21.54 Specimen A3 Average crack length after fatigue:74.99mm Average crack length after SCC:79.13mm 183 Crack length determined via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 74.99 5.14 50 74.99 7.13 139 74.74 11.08 163 75.13 12.04 75.18 2.66E-09 187 75.54 13.35 75.45 3.58E-09 197 75.85 13.81 75.59 3.96E-09 215 76.26 14.46 75.87 4.65E-09 265 76.45 16.35 76.88 6.56E-09 292 77.38 17.54 77.56 7.60E-09 309 77.81 18.13 78.05 8.25E-09 315 78.17 18.71 78.23 8.48E-09 319 78.43 18.59 78.35 8.63E-09 330 78.83 18.84 78.70 9.05E-09 336 79.13 19.08 78.90 9.28E-09 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 72.40 5.14 50 72.40 7.13 139 72.67 11.08 72.82 2.89E-09 163 73.15 12.04 73.11 3.81E-09 187 73.65 13.35 73.48 4.74E-09 197 73.99 13.81 73.65 5.13E-09 265 74.85 16.35 75.23 7.75E-09 292 75.88 17.54 76.03 8.80E-09 309 76.37 18.13 76.59 9.45E-09 315 76.75 18.71 76.80 9.69E-09 319 77.03 18.59 76.94 9.84E-09 330 77.47 18.84 77.34 1.03E-08 336 77.79 19.08 77.56 1.05E-08 184 Specimen A43 Average crack length after fatigue:65.69mm Average crack length after SCC:71.39mm Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 64.05 3.05 26.5 64.05 4.38 116.5 64.05 8.90 63.87 1.44E-09 146.5 64.10 10.41 64.09 2.53E-09 166 64.10 11.34 64.29 3.24E-09 194.5 64.71 12.72 64.67 4.28E-09 217.75 64.77 13.91 65.07 5.12E-09 268.5 66.39 16.14 66.17 6.97E-09 285.25 66.76 16.88 66.61 7.58E-09 310.65 67.38 17.86 67.35 8.51E-09 331.5 67.70 18.64 68.01 9.27E-09 332 68.20 18.67 68.03 9.29E-09 185 Specimen A70 Average crack length after fatigue:64.00mm Average crack length after SCC:81.61mm Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 65.50 2.87 216 65.98 9.26 66.09 3.09E-09 286.5 67.34 11.14 67.25 6.02E-09 335.5 68.53 12.38 68.49 8.05E-09 479.5 74.50 15.39 74.21 1.40E-08 527.5 76.23 16.37 76.81 1.60E-08 622 83.10 17.61 82.93 1.99E-08 Specimen A93 (One cycle) Average crack length after fatigue:66.17mm Average crack length after SCC:82.03mm Crack length obtained from surface measurements: 2.0rder a vs. t Curvefit t(hrs) a(mm) K(MNmA-3/2) a(mm) ing: 0 65.22 5.93 46.5 65.22 6.90 72.25 65.23 7.50 65.25 95.75 65.23 7.94 65.27 120.25 65.50 8.65 65.35 148 65.59 9.33 65.49 167.5 65.65 9.79 65.63 196.5 65.73 10.48 65.89 222 66.35 11.11 66.18 240 66.46 11.50 66.41 268 66.55 12.21 66.82 289 66.71 12.62 67.17 316.5 68.03 13.39 67.68 336.5 68.20 13.77 68.09 343.5 68.26 13.97 68.24 367.5 69.00 14.50 68.78 393 69.25 14.71 69.42 414 70.04 15.18 69.97 433.5 70.37 15.57 70.52 441 70.53 15.68 70.74 465.5 71.77 16.16 71.49 480.5 72.05 16.30 71.97 504.5 72.86 16.62 72.78 534.5 73.56 16.85 73.85 558 74.86 17.18 74.74 Unloading: 576.5 75.07 17.23 75.03 583 75.36 17.28 75.26 600.5 75.78 16.94 75.84 624.5 76.45 16.29 76.58 633 76.47 15.61 76.83 648.5 77.28 15.40 77.25 672.5 77.77 14.44 77.84 681.5 78.47 14.11 78.04 705.5 78.64 13.52 78.53 729.5 79.09 13.15 78.95 749 79.30 12.29 79.23 769.5 79.59 11.71 79.48 796.5 79.59 10.83 79.72 818.5 79.59 10.30 79.85 847.5 79.59 9.94 79.92 895.5 80.11 8.39 From 2.0rder da/dt v(m/s) 7.07E-11 5.89E-10 1.13E-09 1.74E-09 2.17E-09 2.81E-09 3.38E-09 3.77E-09 4.39E-09 4.85E-09 5.46E-09 5.90E-09 6.06E-09 6.59E-09 7.15E-09 7.61E-09 8.04E-09 8.21E-09 8.75E-09 9.08E-09 9.61E-09 1.03E-08 1.08E-08 9.77E-09 9.55E-09 9.12E-09 8.13E-09 7.76E-09 7.24E-09 6.46E-09 6.17E-09 5.25E-09 4.37E-09 3.72E-09 2.95E-09 2.00E-09 1.20E-09 1.70E-10 Specimen A17 Average crack length after fatigue:64.95mm Average crack length after SCC:85.31mm Crack length deternained via compliance measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(mm) K(MNmA-3/2) a(mm) v(m/s) 0 64.95 6.12 10 64.95 6.21 30 64.95 6.39 50 64.95 6.57 70 64.95 6.75 90 64.95 6.94 110 64.95 7.12 130 64.95 7.30 150 64.95 7.48 170 64.95 7.66 190 64.95 7.85 210 64.95 8.03 64.79 2.52E-09 230 65.32 8.17 64.97 2.66E-09 250 65.51 8.35 65.17 2.81E-09 270 65.74 8.52 65.38 2.95E-09 290 65.80 8.74 65.59 3.10E-09 310 65.99 8.92 65.82 3.24E-09 330 66.15 9.11 66.06 3.38E-09 350 66.38 9.27 66.31 3.53E-09 370 66.56 9.46 66.57 3.67E-09 390 66.80 9.63 66.84 3.82E-09 410 66.92 9.82 67.12 3.96E-09 430 67.12 10.00 67.41 4.11E-09 450 67.33 10.17 67.71 4.25E-09 470 67.64 10.34 68.02 4.40E-09 490 68.01 10.48 68.34 4.54E-09 510 68.30 10.65 68.68 4.69E-09 530 68.73 10.78 69.02 4.83E-09 550 69.12 10.93 69.37 4.98E-09 570 69.50 11.08 69.73 5.12E-09 590 69.72 11.24 70.11 5.27E-09 610 69.91 11.42 70.49 5.41E-09 630 70.30 11.52 70.89 5.56E-09 650 70.95 11.63 71.29 5.70E-09 670 71.70 11.72 71.71 5.84E-09 690 72.17 11.83 72.13 5.99E-09 710 72.76 11.94 72.57 6.13E-09 730 73.18 12.05 73.02 6.28E-09 750 73.68 12.17 73.48 6.42E-09 770 74.20 12.27 73.94 6.57E-09 790 74.60 12.38 74.42 6.71E-09 810 75.06 12.47 74.91 6.86E-09 830 75.74 12.55 75.41 7.00E-09 850 76.09 12.63 75.92 7.15E-09 Specimen A17 (continued) t(hrs) a(mm) 870 76.69 890 77.33 910 78.04 930 78.47 950 79.17 970 79.76 990 80.34 1010 80.90 1030 81.26 1050 81.49 1070 81.85 1090 82.13 1110 82.88 1130 83.65 1150 84.45 1170 85.31 2.0rder a vs. t Curvefit K(MNmA-3/2) a(mm) 12.73 76.44 12.79 76.97 12.87 77.51 12.96 78.06 13.02 78.62 13.08 79.19 13.15 79.78 13.23 80.37 13.40 80.97 13.46 81.58 13.54 82.21 13.64 82.84 13.68 83.49 13.67 84.14 13.70 84.81 13.70 85.49 Crack length obtained from surface measurements: 2,Order a vs. t Curvefit t(hrs) a(mm) K(MNmA-3/2) a(mm) 0 63.55 6.12 97 63.55 6.94 62.37 324 63.66 9.05 64.05 445 64.56 10.13 65.42 493 64.56 10.51 66.05 600.5 66.18 11.34 67.66 649 67.2 11.63 68.47 672.5 69.55 11.73 68.88 696.5 70.58 11.86 69.32 792 72.05 12.39 71.17 816 72.62 12.49 71.67 840 72.62 12.59 72.18 864 73.1 12.70 72.70 936 74.66 12.98 74.35 960 76.11 13.05 74.92 984 76.86 13.13 75.51 1104 76.86 13.67 78.65 1128 77.59 13.67 79.31 1152 78.29 13.70 79.99 1170 80.47 13.70 80.51 1176 82.80 13.70 80.68 From 2.0rder da/dt v(m/s) 7.29E-09 7.44E-09 7.58E-09 7.73E-09 7.87E-09 8.02E-09 8.16E-09 8.31E-09 8.45E-09 8.59E-09 8.74E-09 8.88E-09 9.03E-09 9.17E-09 9.32E-09 9.46E-09 From 2.0rder da/dt v(m/s) 1.35E-09 2.77E-09 3.52E-09 3.82E-09 4.49E-09 4.79E-09 4.94E-09 5.09E-09 5.68E-09 5.83E-09 5.98E-09 6.13E-09 6.58E-09 6.73E-09 6.88E-09 7.63E-09 7.78E-09 7.93E-09 8.04E-09 8.08E-09 Specimen A-SI (Bolt loaded) Average crack length after Pop-in:30.39mm Average crack length after SCC:50.59mm 189 Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(rnm) K(MNmA-3/2) a()mm) v(m/s) 0.0 30.39 22.00 0.9 30.44 21.95 1.5 30.46 21.92 20.5 31.32 21.02 48.1 32.90 19.49 68.9 33.51 18.94 140.9 35.16 17.57 35.26 6.64E-09 165.5 35.72 17.14 35.84 6.48E-09 189.2 36.28 16.73 36.38 6.33E-09 213.7 36.94 16.26 36.93 6.17E-09 235.5 37.51 15.87 37.41 6.02E-09 310.3 39.15 14.82 38.97 5.54E-09 332.0 39.89 14.38 39.40 5.40E-09 377.3 40.45 14.06 40.25 5.10E-09 406.3 40.62 13.96 40.78 4.91E-09 477.5 41.79 13.33 41.98 4.45E-09 497.3 41.96 13.24 42.29 4.32E-09 545.4 42.71 12.87 43.01 4.01E-09 575.1 43.46 12.50 43.43 3.82E-09 646.3 44.29 12.12 44.35 3.35E-09 689.2 45.33 11.67 44.84 3.08E-09 713.4 45.37 11.65 45.11 2.92E-09 738.3 45.45 11.62 45.36 2.76E-09 814.2 46.02 11.38 46.05 2.26E-09 834.2 46.08 11.35 46.20 2.13E-09 857.3 46.18 11.31 46.38 1.98E-09 910.3 46.70 11.10 978.7 47.41 10.83 1,050.0 47.44 10.82 1,073.4 48.23 10.52 1,150.8 48.94 10.27 1,193.3 49.16 10.19 1,217.6 49.20 10.18 1,312.8 49.55 10.06 1,341.4 49.76 9.98 1,361.5 49.92 9.93 1,415.8 50.00 9.90 1,486.8 50.42 9.77 1,529.8 50.52 9.73 1,553.8 50.59 9.71 Specimen A5 (Bolt loaded) Average crack length after fatigue:68.98mm Average crack length after SCC:75.5mm Crack length obtained fro surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(average) K(MNmA-3/2) a(mm) v(m/s) 0 66.93 21.34 66.84 1.49E-08 7.5 66.93 21.34 67.24 1.50E-08 25.5 68.58 20.44 68.23 1.53E-08 34 68.70 20.38 68.70 1.55E-08 58.3 69.86 19.79 70.07 1.58E-08 70.5 70.87 19.29 70.76 1.60E-08 75.2 70.98 19.23 71.04 1.61E-08 79.5 71.50 18.98 71.29 1.61E-08 94.3 71.69 18.90 72.15 1.64E-08 103.3 72.91 18.32 72.69 1.65E-08 118.5 73.64 18.00 73.60 1.67E-08 Specimen A7 (Bolt loaded) Average crack length after fatigue:64.88mm Average crack length after SCC:72.72mm Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(average) K(MNmA-3/2) a(mm) v(m/s) 0 63.75 17.46 63.67 6.93E-09 5 63.75 17.44 63.80 6.90E-09 16.5 63.75 17.40 64.08 6.84E-09 19.5 63.83 17.35 64.16 6.82E-09 22.5 63.85 17.33 64.23 6.80E-09 40.5 64.58 16.92 64.67 6.69E-09 43.5 64.68 16.87 64.74 6.68E-09 53.5 64.79 16.78 64.98 6.62E-09 73.5 65.18 16.54 65.45 6.50E-09 79 65.18 16.52 65.58 6.47E-09 89.7 65.42 16.37 65.83 6.40E-09 101.5 65.71 16.21 66.10 6.33E-09 113.5 66.03 16.03 66.37 6.26E-09 136.5 67.08 15.51 66.88 6.12E-09 209 68.58 14.69 68.42 5.70E-09 281.5 69.45 14.15 69.85 5.27E-09 Specimen A12 (Bolt Loaded) Average crack length after fatigue:63.94mm Average crack length after SCC:67.59mm Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(average) K(MNmA-3/2) a(mm) v(m/s) 0 62.70 14.35 62.50 3.89E-09 10 62.70 14.35 62.64 3.87E-09 18.5 62.70 14.35 62.75 3.85E-09 26 62.73 14.33 62.86 3.83E-09 31.5 62.98 14.24 62.93 3.82E-09 37 62.98 14.23 63.01 3.81E-09 43 62.98 14.23 63.09 3.80E-09 50 63.09 14.19 63.19 3.78E-09 116 64.08 13.81 64.07 3.65E-09 146 64.46 13.66 64.46 3.59E-09 173 64.97 13.47 64.81 3.53E-09 195 65.01 13.46 65.08 3.48E-09 218 65.45 13.30 65.37 3.44E-09 285 66.11 13.06 66.18 3.30E-09 296 66.32 12.98 66.31 3.28E-09 193 Specimen A32 (Const, load) Average crack length after fatigue:64.71mm Average crack length after SCC:68.83mm Crack length obtained from surface measurements: 2.0rder From a vs. t 2.0rder Curvefit da/dt t(hrs) a(average) K(MNmA-3/2) a(mm) v(m/s) 0 63.65 13.28 63.72 1.09E-09 67.5 64.17 13.43 64.12 2.18E-09 91.5 64.40 13.49 64.33 2.57E-09 115 64.72 13.57 64.56 2.96E-09 139 64.84 13.61 64.83 3.34E-09 163 65.06 13.67 65.14 3.73E-09 189.5 65.19 13.72 65.52 4.16E-09 214.25 65.99 13.89 65.91 4.57E-09 235.25 66.32 13.96 66.26 4.91E-09 260.25 66.78 14.07 66.72 5.31E-09 Appendix BLT Compliance data for DCB specimens of alloy AA-7075 and AA-8090 (2nd order polynomial curve fit) C (kN*1 ) 2.8 2.4 2.2 2 1.8 1.8 1.4 -1.2 1 -0.8 0.6 0.4 0.2 0 Compliance AA-7075 • x : Measurements - : Polynomial (Curve fit) 0.3 0.4 0.5 Crack length (a/L) Appendix I V 195 In order to calculate crack velocities from crack length (a)-time (t) data, a second order polynomial was fitted. The coefficient of correlation (r) for each slow loading test is given in Table AJJJ.. Hereby, r is defined as: -V 5? where; Sr = Sum of residuals between data points and fitted curve a = crack length Cj = constants of 2 order polynomial t = time and; St = Sum of squares of residuals between data points and the arithmetic mean S ^ H y i - y f - l y , Table AJJJ Coefficient of correlation (r) for crack lengths measured on the specimen surface obtained via the compliance method: Specimen Surface Clip gage A-17 0.9772 0.9984 A-4 0.9979 0.9956 A-6 0.9936 -A-91 0.9834 0.9754 A-93 0.9979 -A-70 0.9989 -A-13 0.9897 0.9994 A-43 0.9935 -A-3 0.9944 0.9887 A-50 0.9670 0.9992 A-55 0.9929 0.9962 A-16 0.9941 0.9696 A - l l 0.9121 0.9915 A-10 0.9956 0.9993 A-SI 0.9979 -A-7 0.9958 -A-5 0.9937 -A-32 0.9945 -A-12 0.9973 -

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078570/manifest

Comment

Related Items