MITLibraries Document Services Room 14-0551 77 Massachusetts Avenue Cambridge, MA 02139 Ph: 617.253.5668 Fax: 617.253.1690 Email: docs@mit.edu http://libraries.mit.edu/docs DISCLAIMER OF QUALITY Due to the condition of the original material, there are unavoidable flaws in this reproduction. We have made every effort possible to provide you with the best copy available. If you are dissatisfied with this product and find it unusable, please contact Document Services as soon as possible. Thank you. Pages are missing from the original document. PAGE 18 MISSING HYDROGEN DEGRADATION AND MICROSTRUCTURAL EFFECTS ON THE NEAR-THRESHOLD FATIGUE RESISTANCE OF PRESSURE VESSEL STEELS by ROSENDO FUQUEN-MOLANO M.Sc., Engineer, Massachusetts Institute of Technology (1973) SUBMITTED TO THE DEPARTMENT OF MATERIALS SCIENCE AND ENGINEERING IN PARTIAL FULFILLmENT OF THE REQUIREMENTS FOR THE DEGREE OF MASSACHUSETTS __...-- INSTiT, OF TEUHNULir Ph.D. at the I M7 C £l.,u&.nU.ll UT Tf I I, M M 1. LL'4'. T lUJ DECI 2 9 205 ir rT vvrr l%*5L mpt 4&WL rr .JLJ 6A LIBRARIES June, 1982 ( Massachusetts Institute of Technology, 1982 I7) (7 A Signature of Author Department of Materials Science a Engineering April 30, 1982 Certified by Robert 0. Ritchie Thesis Supervisor Accepted by I-" PrFofessor Regis M. Pelloux Chairman, Department Committee ARCHIVES 2 HYDROGEN DEGRADATION AND MICROSTRUCTURAL EFFECTS ON THE NEAR-THRESHOLD FATIGUE RESISTANCE OF PRESSURE VESSEL STEELS by ROSENDO FUQUEN-MOLANO Submitted to the Department of Materials Science and Engineering on May 11, 1982 in partial fulfillment of the requirements for the Doctoral Degree in Materials Engineering ABSTRACT Safety of pressure vessels for applications such as coal conversion reactors requires understanding of the mechanism of environmentally-induced crack propagation and the mechanism by which process-induced microstructures as in thick section weldments affect the fatigue resistance of the structure. K At low stress intensities near-threshold K (K = - K ), water vapor inthe environment was f8und to p aucem a pronounced effect on the fatigue resistance for SA378-2In 2Cr-1%Mo partial pressures as low as 10 torr. 22 steel the crack propagation rates at high load ratio (R = K -K ) are increased in the presence of water vapor and the oppeSfte effect is observed at low load ratio. It is proposed that water vapor-containing environments give rise to two mechanisms affecting crack growth rates: embrittlement caused by hydrogen produced in the water-metal reaction; and crack closure, enhanced by the increased surface roughness and the wedging action of the oxidation product. The microstructure is proposed to affect crack propagation rates mainly through crack closure induced by the synergistic effect of fracture surface roughness and oxide produced by fretting. Prof. Robert O. Ritchie (Supervisor) -3- ACKNOWLEDGEMENTS I would like to thank Professor Robert O. Ritchie for his patience and advice. Special thanks to Professor Regis M. Pelloux for his understanding, influence and encouragement. Thanks to all my friends at MIT and the ones who have already left. John and Mark have been specially good to us. Many thanks to my family, my wife Ruth and my parents to whom I owe everything. This work was conducted under the support of a DOE project contract number EX-76-A-01-2295. Support was also received for its completion from Union Carbide through the Mechanical Engineering Department. -4- TABLE OF CONTENTS PAGE Title Page 1 Abstract 2 Acknowledgements 3 Table of Contents 4 List of Figures 9 List of Tables 12 Chapter 1: Introduction 13 Chapter 2: Literature Review 16 2.1 Introduction 16 2.2 Fatigue Crack Propagation 17 2.3 -'atigue Crack Propagation 21 2.3.1 Effect of lean Stress on Threshold 24 2.3.2 Effect of Frequency on Threshold 2.3.-3 2.3.4 Chapter 3: in the Near-Threshold Regime ffect of Strength on IThreshold 26 Crack Size Effect on Threshold 27 Fatigue Resistance of 2Ihr-1'.o Section 25 Thick ¥.,eldcnent Nyear-Threshold 30 3.1 Introduction 30 3.2 Literature Review 31 and Flaws in ;veldmients 31 in ,,eldrents 32 3.2.1 Defects 3.2.2 Crack Propagation 3 2.3 klicrostructural ffects on -5- 3.2.4 3.3 3.4 35 The Closure Phenomena 36 Experimental Procedure 41 3.3.1 Materials 41 3.3.2 Fatigue Testing 41 3.33 Crack 46 3.3.4 Test Procedure onitoring 48 3.4. 1 kiicrostructural Analysis 48 Fatigue Crack Propagation 3.5.2 le Ilear-Threshold Fatigue Crack Propagation Crack Propagation at Low Load Ratio: Crack Closure Phenomena 3.5.2.1 3.5.2.2 Chapter 4: 56 62 Discussion 3.5.1 3.6 46 Experimental Results 3.4.2 3.5 Fatigue Crack Propagation 62 66 Effect of Ductility on Closure 67 Iicrostructural Effects on Closure 68 Discussion 74 hydrogen Degradation of the Near-Threshold Fatigue Resistance of Steels 76 4.1 introduction 76 4.2 Literature 77 4.3 cedure Experimental .Fro 82 4.3.1 Iaterials 82 4.3.2 Gas Purifying System 82 4.3.3 Test Procedure 85 n-~eviev. 4.4 -6Experimental Results 4.4.1 4.5 Chapter 5: 87 44.2 Environmental Effects on the Fatigue 91 Resistance at High Load Ratio 4.4.3 Comparison of the Environmental Effects at Low and High Load Ratio 95 4.4.4 Fractography 97 4-4.5 Summary of Results Discussion 108 110 4.5.1 The Role of Load Ratio in Environmental Studies 110 4.5.2 ;ater Vaoor Effect on Fatigue Crack Growth 117 4.5.3 4.6 Environmental Effects on the Fatigue Resistance at Low Load Ratio 87 Environmental Transport to the Crack Tip 120 4.5.4 Physical Adsorption 122 4.5.5 iate of Adsorption 124 4.5.6 Chemisorption 125 4.5.7 Chemisorption of Hydrogen 126 4.5.8 Chemisorption of '.aterVapor 129. 4.5.9 Influence of k-oisture and Load Ratio on Crack Growth Rates 131 Conclusions influence o Temperature on the Environmental Degradation of the Fatigue esistance Near-Threshold 135 137 5.1 Introduction 137 5.2 Literature Review 138 -7- 5.3 5.4 Experimen+t' Procedure 144 5.3-1 Material 144 5.3.2 Temperature Setup 144 5.3.3 Experimental Setup 145 5.3.4 Testing Procedure 145 Experimental Results 148 5.4.1 5.4.2 5.4.3 5.4.4 Temperature Dependence Environmentally-Induced Near-Threshold Resistance 148 Environmental Effects on Crack Growth at 110 152C Apparent Activation Energy of the Environmental Degradation of Fatigue Resistance 156 Sumary of Results 160 5.5 Discussion 161 5.6 Conclusions 171 Chapter 6: Fatigue Crack Propagation in Vacuum and inert Environments 175 175 6.1 Introduction 6.2 Literature 6.3 Experimental Procedure 178 6.3.1 178 6.3.2 6.4 Review aterials Vacuum Chamber Experimental Results 6.4.1 Crack Propagation in Vacuum Environment 176 179 179 179 6.4.2 Crack Propagation in MIoist ir and 181 Vacuum nvironments 6.4.3 Comparison of Crack Propagation in -8- Vacuum with Other Environments at R = 0.80 6.4.4 6.4.5 6.5 References 183 Comparison of Crack Propagation in Vacuum with Other Environments at R = 0.05 184 Fractography 188 Conclusions 191 192 -9- LIST OF FIGURES page Fig. 2.1 Fig. 3.1 Fig. 3.2 Fig. 3.3 Fig. 3.4 Fig. 3.5 Fig. 3.6 Fig. 3.7 Schematic of fatigue crack growth rate(da/dN) versus alternating stress intensity (K) in steels. 20 Location of C.T. fatigue specimens for weld metal and HAZ areas in SA387 weldment. 45 Schematic diagram of electrical potential crack monitoring system 47 Variation through thickness of microstructure, yield strength and threshold stress intensity AK o for SA387-2-22, 184 mm thick base plate. 49 Schematic continious-cooling transformation diagram for 2Cr-l%Mo steel. 51 Optical micrographs of SA387-2-22 weld metal (1% nital etch). 53 Optical micrographs of SA387-2-22 steel, a)Heat Affected Zone (HAZ) metal, b)Interface between base plate and HAZ metal. 54 Variation in microstructure and Vickers hardness in SA387 thick-section weldment showing base metal, heat-affected zone and weld metal areas. Etched in 2% nital. 55 Fig. 3.8 Fatigue crack propagation in SA387 Class 2 Grade 22 weldment, showing behavior in base plate material at R =0.05 and 0.80 for environments of moist air and dry gaseous hydrogen. 58 Fig. Fatigue crack propagation in SA387 Class2 Grade 22 weldment, showing behavior in weld metal at R =0.05 and 0.80 for environments of moist air 60 and dry gaseous hydrogen. 3.9 Fig. 3.10 Fatigue crack propagation in SA387 Class. 2 Grade 22 weldment, showing behavior in the heat-affected zone (HAZ) at R =0.05 and 0.80 for environments of moist air and dry gaseous hydrogen. 61 Fig. 3.11 Fatigue crack propagation in SA387 Class 2 Grade 22 weldment, showing a comparison of base, HAZ and weld metal areas, tested in moist air at R =0.05 and 0.80. 63 -10Fig. 3.12 Schematic showing the effect of grain size on the dislocation structure at the crak tip. a)Small grain size, b) large grain size.(77). Fig. 4.1 Fig. 4.2 Fig. 4.3 Fig. 4.4 Fig. 4.5 Fig. 4.6 Fig. 4.7 Fig. 4.8 Schematic showing the location of C.T. specimens in the SA387 base metal plate. 83 Schematic representation of the environmental chamber system designed to produce an impurity level of less than 1 ppm. (97) 84 Oxide formed on the fatigue in moist air near-threshold; 4x; b) and c) discontinious d)high magnification of the fracture surface. 4.9 fracture surface in a)fracture surface oxide formation; oxide-covered 88 Fatigue crack propagation in SA387 Class 2 Grade 22, showing the effect of water vapor in the near-threshold behavior for air, hydrogen and oxygen environments. 90 Effect of moiSt air and gaseous hydrogen on the crack propagation rates in SA387 Class 2 Grade 22 steel near-threshold at R =0.80. 93 Effect of moist air, dehumidified air and gaseous hydrogen on the crack propagation rates in SA387 Class 2 Grade 22 steel nearthreshold at R = 0.80. 94 Effect of moist air, dehumidified air on the crack propagation rates in SA387 Class 2 Grade 22 steel near-threshold at R = 0.05. 96 Effect of moist air and dehumidified air on the crack propagation rates of the heat-affected zone microstructure of SA387 thick section weldment Fig. 72 near-threshold at R = 0.80. Fatigue crack propagation in SA387 Class 2 Grade 22 steel in environments of molt air (30% relative humidity) and helium gas (138KPa) at R= 0.80 Fig. 4.10 Transition in fracture surface morphology produced by moist and dry air environments. 98 99 101 Fig. 4.11 Characteristic fatigue fracture surface nearthreshold at R =0.05. a) Dehumidified air, b)moist air (30-40 pct. R.H.) K =10 MPaVm. 102 -11Fig. 4.12 Characteristic fatigue fracture surface produced in purified hydrogen environment near-threshold R=0.05. 104 Fig. 4.13 Variation in fatigue fracture morphology produced by changing environment from moist air tohelium near-threshold at R=0.80. 105 Fig. 4.14 Variation in fracture morphology produced by changing environment from dry air to moist air (heat-affected zone metal) 107 Fig. 4.15 Comparison of fatigue crack propagation results of SA387 steel at R =0.80 obtained by different load shedding procedures. 112 Fig. 4.16 Fatigue crack propagation in SA387 Class 2 Grade 22 weldment, showing behavior in base plate material at R = 0.05 and 0.80 for environ114 ments of moist air and dry gaseous hydrogen. Fig. 4.17 Fatigue crack propagation in 300-M steel at R =0.05 and 0.70 for environments of moist air and dry hydrogen. 115 Fig. 4.18 Effect of moist air and hydrogen gas environment on the fatigue crack growth rates in 300-M 116 steel Fig. 4.19Schematic showing some of the sequential processes involved in hydrogen embrittlement from gaseous 119 environments. Fig. 4.20 Schematic showing the effect of crack closure on the effective stress intensity at the crack tip (a), and on the measured fatigue crack propagation at low load ratio. 133 Fig. 5.1 Schematic showing the specimen set up for tests 146 at temperatures higher than ambient. Fig. 5.2 Influence of temperature on the fatigue crack propagation of SA387 Class 2 Grade 22 steel in environment Fig. 5.3 Fig. 5.4 of moist air (30%R.H.) Influence of temperature on the fatigue crack propagation of SA387 Class 2 Grade 22 steel in environment of dry gaseous hydrogen. 149 151 Fatigue crack propagation in a single specimen of SA387 Class 2 Grade 22 steel for environments of moist air (30% R.H.) and dry gaseous hydrogen at 1100 C. 153 -12Fig. 5.5 Fig. 5.6 Fig. 5.7 Fig. 6.1 Effect of water vapor on the fatigue crack propagation rates in SA387 Class 2 Grade 22 steel near-threshold in hydrogen and helium environments at 1100 C. 155 Apparent activation energies of the fatigue crack propagation rates in hydrogen gas environment. 157 Apparent activation energies of the fatigue crack propagation rates in moist air environment. 159 Fatigue crack propagation in SA387 Class 2 Grade 22 steel at R = 0.05 and 0.80 in vacuum environment (2 x 10 Fig. 6.2 Fig. 6.4 Fig. 6.5 Fig. 6.6 Fig. A-1 180 Fatigue crack propagation in SA387 Class 2 Grade 22 steel at R= 0.05 and 0.80 for environments of moist air (-~0% relative humidity) and vacuum Fig. 6.3 torr). (2 x 10 182 torr). Fatigue crack propagation in SA387 Class 2 Grade 22 steel near-threshold showing a comparison of the effect of moist air, hydrogen and helium at R =0.80 with vacuum at R =0.80 and R = 0.05. 185 Fatigue crack propagation in SA387 Class 2 Grade 22 steel near-threshold showing a comparison of the effect of dehumidified and moist air environments at R =0.05 with vacuum at R =0.05 and R =0.80. 187 Characteristic fatigue fracture morphology produced in vacuum environment near-threshold at R =0.05. 189 Characteristic fatigue fracture morphology produced in vacuum near-threshold at R =0.80. 190 Design of (CT) testpiece used for fatigue crack propagation tests. (96) 193 LIST OF TABLES Table 3.1 Table Table 3.2 3.3 Table 5.1.. Table 5.2 .............................. .. . . 42 ......................... . 43 ............................ 65 ............................ 168-169 ............................... !72 -13CHAPTER 1: INTRODUCTION The design of pressure vessels for cost effective refining processes of alternative energy sources such as coal, has raised a number of important questions about materials performance under complex service conditions of dynamic loading and aggressive environments. Coal conversion processes re- quire, for example, massive welded steel pressure vessels, as large as 60 m high, 6 m in diameter with 250-350 mm wall thickness, to operate continuously, economically, and above all safely at high pressures (up to 10 MPa), and temperatures (540-1000°C) in the presence of erosion-producing solid particles and chemically-aggressive atmospheres including H2 0, H2-,H2S, etc. (136). The vessel wall, although insulated from the reaction processes by a refractory lining and an additional stainless steel or Incoloy weld overlay, may be expected to be in contact with these gas mixtures at operating temperatures around 320 to 350 C. (137). Material requirements for pressure vessel include homogeneous, tough, weldable steels having yield strengths in excess of 350 MPa, with good resistance to creep, fatigue, and environmental degradation (i.e., hydrogen embrittlement and hydrogen attack). are the 2Cr-l%Mo Prime candidate for this application low alloy steels. Considerable effort has been directed towards characterizing the fabrication, mechanical and environmental properties of 2Crl-%Mo steels, and background information on the use of such steels in thick -14- sections is available from hydrocarbon processes and nuclear technologies. However, one area which has received relative- ly little attention is the fatigue crack growth of small incipient flaws at growth rates of less than 10-6mm/cycle, approaching the so-called alternating threshold stress intensity &Ko, 0 below which crack remain dormant or propagate at experimentally undetectable rates. In coal conversion service, such growth may arise from high cycle, low amplitude stresses generated from small pressure and temperature fluctuations ("operational transients") which are expected to occur during "steady state" operations as a result of real variations in coal feed rate, "clusters" of pulverized coal, combustion instabilities, and pressure fluctuations in feed and discharge lines for gases. The object of this investigation was to study the fatigue behavior at low crack growth rates of thick section 2Cr-l%Mo ASTM A387 Class 2 Grade 22 (hereafter referred to as SA387) pressure vessel steel in inert, moist air and gaseous hydrogen environments. The study concentrated on the mechanisms by which microstructure and environment affect the growth rates in the near-threshold regime under cyclic loading simulating the service conditions. Such cyclic stresses in large pressure vessels will be superimposed on a "constant" mean operating stress during most of the service life and will occur due to a variety of operating conditions. Low cycle, high amplitude pressure fluctuations are involved -15in start-up and shut-down sequences for operation, or test purposes. A general literature review on the fatigue crack propagation of metals is presented in Chapter 2 and the results of the investigation are presented in selfcontaining Chapters 3 to 6 covering the study of microstructural effects on the fatigue crack porpagation near-threshold (Chapter 3), the hydrogen degradation of the fatigue resistance nearthreshold (Chapter 4), the temperature activation of the environmental degradation (Chapter 5), and the crack propagation in inert and vacuum environments (Chapter 6). -16CHAPTER 2: 2.1 LITERATURE REVIEW INTRODUCTION Cyclic loading is a characteristic of most structural and machine elements. It is also the source of most service failures better known as fatigue fracture failures, caused by initiation and/or propagation of cracks. Fatigue cracks can propagate under nominal loads far below the yield strength of the material due essentially to two factors: the dynamic character of loading supplying the necessary activation stress for crack extension and the stress intensity concentration at the crack front due to crack geometry. Design against fatigue relies on experimentally determined fatigue behavior of the alloys with account taken of factors that may accelerate crack growth such as mean stress, environment, microstructural variations, etc. Safety/critical engineering applications are designed generally with the defect-tolerant approach, which is essentially based on a crack propagation criterion. Flaws and incipient cracks are assumed present due to the fabri- cation and processing in the alloys methods. The life of a component is estimated on the number of cycles necessary for an initial flaw (usually the largest undetected defect) to obtain a critical size where final failure occurs either catastrophically or at the limit load. Fatigue crack propagation is dependent mainly on the -17loading conditions which determine the stress intensity range imposed on the crack front; however, the rate may be altered substantially by such factors as environment, microstructure, mean load and loading spectrum, etc. Consequently, an in- tense effort has been put into characterizing the effect of these parameters on the crack growth rates and numerous results of this kind have been repeated. However, mechan- istic understanding of these phenomena is far from complete. Design of fatigue-resistance alloys requires a clear and qualitative understanding of the mechanisms responsible for the crack propagation as well as the independent and/or synergistic effects of metallurgical and environmental variables on the growth rates. 2.2 FATIGUE CRACK PROPAGATION The application of linear-elastic fracture mechanics for design against fatigue in conditions of limited plasticity has prompted many investigators to describe the crack growth rates da/dN in terms of the stress intensity range AK given by the difference betweenthe maximum and minimum stress intensities in a given cycle, i.e., AK = K - K max min' This has been one of the most significant developments (1-3) in understanding the relative effect on crack growth rate of variables such as microstructure, load ratio, environment, frequency of loading, etc. For most metals and alloys the relationship between da/dN vs. AK is described by a(CoNTIuED o P J9) PAGES (S) MISSING FROM ORIGINAL PAGE 18 MISSIBG -19sigmoidal curve in a log-log scale as shown in fig. 1 (4). Three distinct regimes of crack growth have been indicated: regimes A, B and C, also designated as near-threshold, midgrowth and high growth rate regimes respectively. A power law of the form (5) da/dN = A (K) m (2.1) describes adequately the crack growth behavior in the range of typically 10 regime. 5 - 10- 3 mm/cycle, that is the mid-growth Constants A and m are experimentally determined scaling constants. Most fatigue data have been generated for growth rates above 10-5 mm/cycle, a range of vital engineering application, for determining safe non-destructive inspection intervals, for example. Characterization of crack growth behavior below 10- 5 mm/cycle has become increasingly necessary for the design of components subject to very high frequency, 10 to low amplitude loading, (lifetimes in the range of 10 102 cycles). Subcritical growth of defects at apparently extremely low growth rates may become sources of catastrophic failures. Equation (2.1) underestimates the propagation rates when Kmax, the maximum stress intensity approaches the fracture toughness (KIC, KC) or limit-load failure, and it is generally conservative in describing the near-threshold -20- QI va 1. _ GJ , cn ur c, so _ - 4) · 01f L.uc- 4- as 4 b t U,) _I U r 4) -9 4J*_ t- En · C_ _,V) , Use 0 I 0 0 I al/L/uWW I 'NP/DP -21regime. Other semi-empirical equations have been developed to include the acceleration of growth rates in regime C (6) or to account for the threshold stress intensity (7) or even to represent the entire crack growth behavior (8); however none of these equations have general applicability. Fatigue failure in the mid-growth regime is generally characterized by transgranular, ductile striation mechanism, and there is often little experimentally observed variation of growth rates with microstructure, mean stress, yield strength. In regime C, static fracture modes such as cleavage, intergranular and fibrous fracture occur in addition to striation growth, and the crack growth rates are sensitive to variations in microstructure and mean stress. Characteristics of the near-threshold regime are discussed in the following section. 2.3 FATIGUE CRACK PROPAGATION IN THE NEAR-THRESHOLD REGIME At crack growth rates below 105 mm/cycle the fatigue resistance is influenced mainly by mean stress, microstructure, environment and crack length. rates The measured propagation approach the order of interatomic spacing per cycle, indicating that the crack does not advance evenly along the front. The scale of plasticity approaches the order of the microstructural size scales. Fractography analysis on low carbon steels has -22identified two modes of crack growth as a function of the stress intensity factor K in the near-threshold regime (9): the shear and tensile modes of growth which are characteristic of low and high values of K respectively. A transition from shear to tensile mode was observed at a constant value of K, irrespective or specimen geometry or loading condition. These modes seem to correspond to the Stage I and Stage II growth pattern designated by Forsyth (10). In Stage I, the crack initiates and propagates in a surface slip band due to shear stress; in Stage II, the crack propagates perpendicular to the principal tensile stress direction. The slip system activated in Stage I is roughly parallel to the direction of crack growth and has a traction component at the crack tip due to the tensile stresses perpendicular to the crack direction. Stage I is then a combination of Mode I and Mode II in fracture mechanics terminology. Results on titanium alloy at low growth rates have also indicated the presence of Mode II loading which, by the relative shear displacement in the fracture plane, prevents matching of the fracture surface during unloading, and enhances closure effects (11). Consequently crack propagation in the near-threshold regime is abounded with Mode I Mode II type crack tip displacements. The threshold value AKo for an alloy and the transition from the near-threshold -23.regime to the mid-growth regime are observed as controlled by the level of crack closure, and by the relative size of the plastic deformation zone ahead of the crack with respect to the appropriate microstructural parameter (12). For low load ratio, crack closure becomes the dominant effect nearthreshold; it is determined by the ductility of the material or plastic strain to fracture and (13), the microstructure roughness of the fracture surface (14), and the presence of debris or oxide From product the engineering in the wake of the crack view, (15,16). the characterization of the near-threshold behavior of fatigue crack growth in metals and alloys is of significant importance in order to assess the life of components subjected to high cycle loading. Most of the service life of these components corresponds to the time necessary fr an incipient flaw to grow to sizes detectable by nondestructive techniques or to the time required for an existing crack to attain higher growth rates characteristic of the mid-growth regime. Understanding the mechanisms and parameters that govern the slow crack growth of metals will have a direct implication on materials selection, metallurgical and thermomechanical processing, and environmental protection. The following sections of this chapter relate to some of the most important parameters in this growth regime with -24the exception of microstructural and environmental effects which will be discussed in greater detail in separate chapters. 2.3.1 EFFECT OF MEAN STRESS ON THRESHOLD Crack growth rates are very sensitive to variations of mean stress level characterized by the load ratio R (R = Kmmn min/K max ) from 0 to 0.90. In general the threshold value AKo decreases as the load ratio increases. The load ratio dependence is found to be reduced by increasing temperature (17), inert atmospheres and increasing strength in tempered martensitic steels (18), and it is found to be very small in tests conducted in vacuum for materials such as medium carbon alloy steel (En 24) and titanium alloy Ti-Al-4V (19, 20). The threshold dependence on R has been rationalized in terms of closure effects in the wake of the crack tip. Closure of a crack at positive load occurs due to the permenent deformation imposed on the material before rupture (21). This permanent deformation in the wake of the crack allows contact of the fracture surface before complete unloading. Closure then affects the minimum stress intensity at the crack tip; therefore, the stress intensity range (AK = Kmax- Kin) calculated from externally applied loads could differ from the effective stress intensity range sensed -25- at the crack tip (22). The effects of crack closure and oxidizing environments could be very effectively coupled. Crack closure can lead to fretting of the contacting fracture surfaces which in the presence of aggressive environment causes accelerated oxidation; furthermore, the thickness of the reaction product enhances closure (23,24). 2.3.2 EFFECT OF FREQUENCY ON THRESHOLD The effect of frequency on near-threshold crack growth is directly related to the exposure of the fracture surface to the environment, to the time necessary for reactions to take place and also to the rate of heat generation at the crack tip. Studies on Aluminum 2024-T3 in the frequency range of 324 - 1000 Hz show that threshold decreases with increasing frequency but this observed effect could be coupled with the temperature rise at the crack tip and therefore are not conclusive (25). increasing the frequency In D6ac steel, however, from 100 to 375 Hz resulted in lower near-threshold growth rates for tests in dry argon with no effect in moist air (26). For ferritic steels, such as A533-B and A508 nuclear pressure vessel steels (27), tested in water environments at frequencies between 0.01 and 10 Hz, there is evidence showing marked frequency dependence on -26the crack growth rates above 10 5 mm/cycle and less dependence as growth rates approach threshold. The decrease of frequency effect on crack growth rates near-threshold is possible due to the fact that the rate of generation of fresh metal surface is slow enough that the time of exposure to environment at atmospheric pressure is sufficient to allow adsorption and reactions to take place. 2.3.3 EFFECT OF STRENGTH ON THRESHOLD Fatigue crack propagation inthe mid-growth regime has been shown to be generally unaffected by variations in yield strength (28). In steels, a variation of strength of nearly one order of magnitude changes the growth rates by a factor of 2 or 3 in the mid-growth steels the value of ing strength. regime (29). In low strength Ko is observed to decrease with increas- This effect is enlarged for high- and ultra- high-strength martenistic steels (yield strength between 1000 and 2000 MPa); in this case, however, the controlling measure was the cyclic rather than the monotomic yield stress (30), and increasing the cyclic yield strength increases nearthreshold growth rates and reduces the threshold stress in- tensity Ko Crack growth rates near-threhold decrease with decreas'ing yield strength. the load ratio This effect is strongly dependent (31) and is enhanced for low values on of R; -27however, this observation may involve synergistic effects of microstructure, grain size and crack closure. 2.3.4 CRACK SIZE EFFECT ON THRESHOLD Most experimental studies on fatigue have concentrated on the crack initiation on smooth specimens or crack propagation of relative length through cracks. The intermediate situation, that is, propagation of small cracks, is receiving increasing attention; this is due to the substantial evidence that in some cases small cracks do not follow the crack growth laws derived from large cracks. Furthermore, often the fatigue fracture process is dominated by the initiation and growth of very small cracks. Early work on mild steel, aluminum and copper has indicated that the threshold stress amplitude multiplied by one-third power of the crack length was a constant (32,33, 34), meaning that the threshold stress intensity creases with decreasing crack length. Kth de- More recent work on steels has confimed this variation of AKth with crack length (35). Other investigators have experimentally found critical crack lengths above which the growth rates are reasonably predicted with growth laws derived from long cracks (36,37). A detail study on high strength steel has found that the stress intensity threshold AKth decreased when the crack was shorter than 0.5 mm (38), and the threshold value of -28- the stress 6-th approached the fatigue limit of smooth specimens for very small cracks. The crack length at which the transition occurs has been found to depend on microstructure (39). Several models have been proposed to pre- scribe behavior of small crack growth in terms of the fatigue limit G , the threshold Intensity, a characteristic crack length, and microstructural properties of the material. Smith's model (40), introduced the characteristic crack length a, also designated as the intrinsic crack length, and predicts that: i. For crack lengths shorter than ao, the threshold amplitude is equal to the fatigue limit a ii. ao, then 6th = e For crack length larger than a, (2.2) the threshold intensity is constant and equal to that evaluated for longer cracks a > ao, then Kth = const. (2.3) El Haddad et al. (41) propose that the growth of very small cracks in smooth and notched specimens can be predicted from long crack data using a modified stress intensity factor. They introduce the concept of the effective crack length as the sum of actual crack length and the intrinsic crack aeff - a + a The alternating stress intensity is expressed: AK = F 1O V (a + a ~o)' 1 (2.4 (2.5) -29F 1 is a geometry dependent constant. At threshold: AKth = F1 h V (a + a) The intrinsic crack length, a, is a constant charac- teristic of a given material and material condition, and can be determined from equation (2.5) for the condition of very small cracks, when "a" approaches ao = .1 ( Kth/ zero and F 1 is equal to 1: e) (2.6) A more recent model of Tanaka (42) expresses that the threshold condition is determined by whether the slip band near the crack tip propagates into the adjacent grain or not. The model is based on results of microscopic obser- vations of the deformation zone ahead of the crack (43). The observations indicate the influence of grain size through interference of the grain boundary with the slip system ahead of the crack. is also presented A comparison of previous models in reference 43. The intrinsic crack length a appears to be the mag- nitude of a flaw that would determine the limit at which the stress intensity reaches a value to produce crack growth. The value of a is microstructure and yield strength-depend- ent. It is thought that at such level of scale, the microstructure effect on the micromechanism of deformation ahead of the crack will play a determining role on a. CHAPTER 3: FATIGUE RESISTANCE OF 2CR-l%MO THICK SECTION WELDMENT NEAR-THRESHOLD 3.1 INTRODUCTION In the present search for diversified energy sources and new cost effective refining processes the reliability of materials working in aggressive conditions is a central issue for their feasibility and future use. The materials selection and manufacturing of coal conversion reactors are very important materials-related factors determining the feasibility of intense use of coal in the production of gas and liquid fuel. The expected life of these reactors (15-20 years) and their cyclic loading resulting from operational transients makes is appropriate to use the defect tolerant approach for their design. This criterion assumes that the largest un- detected flaws would, under design loads, cause a crack to propagate at very small growth rates initially of the order of 10-7 mm/cycle or less. Despite crack growth the design life of the vessel must be smaller than the time required for the largest flaw to grow to a critical size. However, characterization of fatigue behavior of steels near-threshold has shown a determining role of microstructure and environment on the propagation rates. This -31behavior is of considerable importance in structures such as pressure vessels which, due to dimensional size and fabrication processes, present variation in volume fraction of component phases with through plate thickness and the unavoidable combination of grain sizes and microstructures in thick section weldments. In particular it is major concern to evaluate the crack propagation resistance of the thick section weldment, where crack initiation can occur at impurity segregated boundaries and flaws caused by the stress relief treatments. This chapter presents results of the investigation of the fatigue crack propagation behavior of the base, the heataffected zone and the weld metal of a 2Cr-l%Mo thick sec- tion weldment in environments of moist air and gaseous purified hydrogen for load ratios of 0.05 and 0.80. 3.2 LITERATURE REVIEW 3.2.1 DEFECTS AND FLAWS IN WELDMENTS Welded structures present preferential sites for crack initiation and propagation at or near the weldment due to the presence of residual stresses, impurity segregation and microstructural variations. Furthermore the post-weld heat treatment has been found to cause, in some instances, intergranular decohesion in the heat-affected zone or weld metal; -32this mode of failure has been designated as stress relief cracking or reheat cracking (44) and it is thought to be produced by the same mechanism that causes underclad cracking under stainless steel weld overlays (45). The susceptibility for reheat cracking has been found to be strongly dependent on the impurity content of the material and often has been correlated to experimentally determined impurity parameters (47,48). relief cracking in 2Cr-l%Mo The study of the stress steel (49) has shown at high temperatures (850 K) intergranular cavitation would nucleate in the MnS particles and at low temperatures( 725°K) grain boundary sliding and intergranular separation could be promoted by phosphorous (P) segregation. Consequently- weldments may contain microcracks or potential crack initiation sites introduced during processing which would make the weldments susceptible locations for fracture failures (46). This points out the importance of studying the crack resistance in weldments and their mechanisms of fracture. Such results are valuable for improving design and welding techniques. 3.2.2 CRACK PROPAGATION IN WELDMENTS Studies on the crack propagation resistance of weldments have concentrated on comparing welding processes (50), -33*assessing the effect of residual stresses (51) and evaluating the susceptibility to stress relief cracking (52). Most of the work has been done at growth rates above 10 - 5 mmt/cycle. The effect of residual stresses on the crack propagation in the weldment has been suggested based on observations of different fatigue crack propagation resistance of the "as weld" and stress relieved conditions of ASTM 514 weldment (53) and the interaction of tensile and compressive residual stresses with the crack propagation in aluminum welded joints (51). Residual stresses can effectively alter crack propa- gation causing in some instances lower fatigue crack propagation resistance in the weldment than in the base metal. During welding processes such as gas-tungsten-arc welding (GTAW), submerged-arc welding (SAW) and shield-metal-arc welding (SMAW) were found to produce weldments of similar fatigue crack propagation resistance (52). It appears that the welding processes do not affect the fatigue behavior of weldments except perhaps when they give rise to significantly different final microstructures. The microstructure has been observed to be a dominant parameter in determining the fatigue resistance of weldments, for example: a stronger tendency for crack branching was observed in the delta ferrite austenite structure of the weld metal than in the fully austenic structure of the 304 -34- stainless steel base metal (52). The large grain size of the heat-affected-zone has been found susceptible to postweld heat treatment intergranular cracking and refinement of this microstructure has resulted in improved crack propagation--resistance of 0.5Cr-0,5Mo-0.25%V steel weldment with 2Cr-l%Mo weld metal at 565 C and sustained load con- dition (54). Studies on VAN-80 steel at room temeprature (55) found superior fatigue crack propagation resistance in the ocicular structures of bainite and pro-eutectoid ferrite of the weld and of Widmanstatten ferrite of the HAZ than the ferriteperlite structure of the base metal. The crack path was found to be strongly affected by microstructure; results of the base metal showed a transition from a structure-sensitive fracture mechanism characterized by intergranular failure at low stress intensities (14-20 MPaVm) to a structure insensi- tive mechanism of fatigue striations at intermediate stress intensities. The results of fatigue resistance of weldments indicate a strong influence of the final state of stresses and microstructure on the crack growth rates; moreover the post-weld heat treatments may introduce flaws in sites of high residual stresses, enhancing crack initiation. -353.2.3 MICROSTRUCTURAL EFFECTS ON FATIGUE CRACK PROPAGATION Designing of better fatigue resistant alloys and assessing the service life of structural components require a clear understanding of the mechanism by which microstructure influences the fatigue crack propagation resistance, and its degradation through the metal-environment interactions. Both topics are currently under intensive research and although substantial results are available describing the phenomenology of the problem, a mechanistic understanding is yet far from complete. The microstructure has been an important parameter for modifying the strength properties of metals either through metallurgical or thermomechanical treatments. The influences of microstructure on the mechanical static properties are well characterized (56): reduction in grain size results in increasing yield strength; solution hardening and precipitate hardening increase yield strength and improve creep resistance. Results on the microstructural effects on the fatigue resistance are less consistent and may vary substantially from the stage of crack initiation to that of crack propagation, and from one material to another. For example, for low carbon steels, the dependence of fatigue strength on grain size has been observed to follow a similar relation to the Hall-Petch equation (57); in copper and aluminum, the fatigue life was found insensitive to grain size whereas in -36alpha brass a reduction in grain size led to increase in life (58). Crack propagation behavior has been found to be affected by microstructural parameters in the regimes of high crack growth rates and low growth rates or near-threshold (59). An increase in crack growth rates was observed with a reduction in grain size for steel (59,60) alpha-titanium (61) and alpha-brass (62) in the near-threshold regime; the effect is enhanced at low values of load ratio R, (R = Kmi n - K mmn max), where crack closure phenomena is dominant, and it is observed in smaller scale at high load ratio (57). Further progress in understanding the influence of microstructure on the fatigue resistance requires detailed analysis of the micromechanism of deformation ahead of the crack, their interaction, and response to microstructural variations. This is especially true at low growth rates when the plastic zone size becomes of the order of the grain size and grain boundaries, precipitates and grain orientation can significantly affect crack path and rates of growth. 3.2.4 THE CLOSURE PHENOMENA The closure phenomena designates the contact and load transfer in the wake of the crack while testing at low positive load ratio. Crack closure prevents the crack tip from sensing the full range of applied load and consequently the -37stress intensity range experienced at the crack tip could differ substantially from nominal calculations on applied loads and crack length measurements. contributing to closure are thought The main factors to be the ductility or plastic strain to fracture, the presence of corrosion products on the fracture surface, the tortuosity of the crack path and the crack tip shear displacements occurring parallel to the plane of fracture. Crack closure phenomena was first popularized by Elber's results (66) on fatigue crack propagation of aluminum alloys in plant stress conditions. For small load ratios, fatigue cracks were fully open only for part of the loading cycle; leading to the conclusions that closure was produced by the permanent tensile plastic deformation left in the wake of the crack which makes the mating fracture surfaces incompatible. Elber introduced the concept of effective stress intensity range to correlate crack growth rates with applied stresses and load ratios. Theoretical analysis of closure in the plane stress condition (67) has supported the existence of the phenomenon and provided justification for the use of the effective stress intensity range. The crack closure phenomena does not appear to be restricted to the plane stress condition; in plane strain and low load ratio, crack closure levels have been found to vary with stress intensity range showing pronounced effects at low -38- values of K and small effects at higher K (43). The transition can be thought to occur when the crack tip opening becomes sufficiently small to enhance the non-matching character of the fracture surfaces. At low crack growth rates ( 10-6 mm/cycle) in the near-threshold regime crack closure effects are significant and well characterized for a wide range of materials (11,12,65). Also, other studies have indicated that the crack closure effect observed at low growth rates are mainly determined by the plane stress portion of crack front, i.e., near the surface of the test specimen (68,69). At high load ratios, the crack is assumed to remain open during the load cycle with closure effects minimized. Reduc- tion in load ratio decreases the crack growth rates and increases the measured threshold value K . The load ratio o effect on threshold has been described in terms of two experimentally determined parameters assumed constant for given material and testing conditions (22): Kth, defined as the driving force in terms of stress intensity level to produce crack growth at threshold, and Kl, the stress inten- sity level necessary to open the crack. When crack closure effects are present: Kcl +ci Kth th = Kmax max = const. AK o + (Kc1 + Kth)(1 - R) (3.1) -39At high values of load ratio: hK o = Kth Kth Kmax = AKth/(l-R) (3.2) This description suggests that at low load ratios, any material or experimental parameter that increases K result in higher measured thresholds. would Such parameters can be the plastic strain to fracture, oxidizing environments, grain size and loading mode. Experimental results at low load ratios have indicated that increasing the yield strength of steels causes a reduction in the threshold value of K o. materials produce higher thresholds Higher ductility K 0, emphasizing the role of plastic strain to fracture on crack closure levels. A model has been proposed to account for the plastic strain to fracture and the cyclic work hardening exponent in calculations of the stress intensity at closure Kcl (13). This model treats essentially the plastic strain contribution to the crack closure level. The environment contribution to the closure phenomena has been suggested based on the increased formation of oxide product on the fracture surface as the growth rates approach threshold at low load ratios. The higher oxidation rates near-threshold are an indication of increasing environmental activity assisted by mechanical fretting of the contact -40- points in the wake of the crack. Recent Auger measurements of the crack face corrosion debris at threshold levels for SA542 Steel (70-)have found oxide thickness to be of the order of 0.2 microns, a significant fraction of the crack tip opening displacement ( 1 m). These findings support the basic concept of oxide wedging thereby producing higher closure and threshold levels in oxidizing environments. This contribution to crack closure is currently designated as oxide-induced crack closure (16). In the present work it is clearly demonstrat- ed that the increased oxidation is caused by the presence of water vapor in the environment; furthermore there is an embrittlement effect as a result of the water vapor-metal reaction which is discussed in Chapter 4. Microstructural influences on crack closure phenomena have been interpreted based onthe following experimental observations: 1) The influence of microstructure on threshold AKo is pronounced at low load ratios and diminishes for high load ratio; note, however, that the plastic zone size for high load ratio is at least one order of magnitude larger than that of near zero load ratio. 2) Fractography analysis of crack growth at low AK indicates a transition from tensile mode growth (tearing and irregular appearance, a smooth version of microvoid coalescence) in the mid-growth regime to that of shear mode growth (planar and steps features of -41- growth along slip systems) at very low K. This observation has been interpreted as Mode II loading contribution on crack growth at low propagation rates. 3. 3 EXPERIMENTAL PROCEDURE 3.3.1 MATERIALS The material tested in this study was 2Cr-l%Mo pressure vessel steel commonly used in petrochemical and electricity generating plants; a 76x150x184 mm section of SA387 Class 2 Grade 22 weld was obtained from Westinghouse R&D Center. The section was machined from a large weldment, prepared by Chicago Bridge and Iron Co., using the submerged arc welding process with Linde 124 flux. The base plates were cast by Lukens Steel Co. The chemical composition and heat treat- ment are shown in Table 3. 1. The weldment was found to contain no defects in violation of the acceptance standard (72). Mean value of the mechanical properties are listed in Table 3.2 3.3.2 FATIGUE TESTING Fatigue crack growth tests were performed on Instron 1350 servohydraulic 49KN testing machines at cyclic frequencies of 5 and 50 Hz, and in load control wave shape. using a sinusoidal Compact tension (CT) specimens were used, ma- chined out of a thick section weld. The orientation was -42Table 3.1 Chemical Composition of SA387 Steel (from Lukens Steel Co.) Heat No. 3596 C Mn Si Ni Cr Mo Cu 0.12 0.42 0.25 0.14 2.48 1.06 0.16 ASTM Specification for 2Cr-1%Mo Min. Max. .27 .63 0.15 .05 1.88 2.62 P S 0.013 0.020 Steel .85 1.15 0.35 0.35 Heat Treatment of As-Received SA387 Steel (section thickness 184 mm) Base Plate Heat Treatment Normalized 5 Temperized 1 hr/in hr at 9540 C accelerated fan cool at 6910 C Post Weld Heat Treatment 15 hr at 5930 C 22 hr at 649°C 18 hr at 6910 C -43- E HH U) a) 910 _S Ll- o0 0 Cd r.. c 0 S3 Cd E a v oa) 0 +I C H 0 CMl co U) a, a) c 0 :3 U c Zcd -I-) (N V) a1) a) Ed a) a) I O 4d o0 + 0 CO U) a) L a) a 0. 0 + or V) c ro UM ,:: w 0) a) E U) U) a) : rn CO co Cd E .I0 03 Cd U ) 07 +1 EH +l 0 Ln CZ a) C Lr) L a) M:I SE E H --a) a * 3 Co E J 4.) '- b0 O Ln + I +1 r, Ln U-T >- C C, -3cM CMj - U) a) CI CO 0 n ur E aoO E Lt) O -. ++ -T + + -44- determined after proper macro-etching of the metal block; for the heat-affected zone macro-etching of the specimen blanks was necessary for accurate orientation of the notch. The location and orientation of the weldment specimens are shown in fig. 3.1. Specimens H1 through H10 were used for crack propagation studies along and across the hear-affected zone metal, similarly specimens Wl-W8 were used for the weld metal. Specimens of the base metal were obtained from the midthickness of a metal block ( Thickness). Testing was done in two environments: laboratory moist air ( 30% relative humidity) and hydrogen gas (138 KPa) at room temperature. An environmental chamber was mounted on the specimen for hydrogen testing. The gas purifying sys- tem consisted of an oxygen catalytic converter, a molecular sieve, and liquid nitrogen trap. Before each test, the gas supply line and chamber were evacuated to a pressure of 1020 mtorr and baked at about 100°C with a heating blower prior to flooding with hydrogen gas. A few tests were per- formed by running the hydrogen from the high purity tank through the molecular sieve and directly to the chamber. This procedure does not remove enough water vapor and impurities from the hydrogen gas as will be discussed in the results section. I -A0- 4 mm ___ _ Location Orientation weld HAZ T-L TS-L weld/ HAZ LS -T Fig. 3.1 Specimen No. WI - W8 HI-H4, H6- H9 H5, HIO Location of C.T. Fatigue Specimens for Weld Metal and HAZ Areas in SA387 Weldment -463.3.3. CRACK MONITORING Crack lengths were continuously monitored using the D.C. electric potential technique (73). This method consists of passing a constant direct current through the ligament of the specimen and measuring the change in electrical potential as crack growth reduces the unbroken ligament. For a given crack length, the ratio of the potential signal to the initial potential across the notch (Va/Vao) can be related through calibration curves to the ratio of crack length plus ligament length (a/W). This technique can measure absolute crack length to an accuracy of approximately 0.1 mm and detect changes in crack length with an accuracy of 0.01 mm. A schematic diagram of D.C. potential drop system is shwon in fig. 3.2 . 3.3.4 TEST PROCEDURE Threshold crack growth rates were obtained using a load shedding procedure, whereby a fatigue crack is initiated at a AK of 16-18 MPa V m and the loads are subsequently reduced by approximately 10% before monitoring crack advance and dis- 47 _ E 0) 0 0O- < CQ It) C L. 0 cC *0 :) O 0 .0 E V 4- C . c D c O= c Ln 3 00CL. U) cn U 4- 0 _ 0. c m C 0 Ca E C *> Q .0 O w O 0)11) L. 0O = '4 0 cs L.~~ . 0 . _ 0 cU .- 0 \ ' E 0 m 4 0Q t 01 L. 0- 0 0 - 00 CUC - _ - - L. 0.0 I-1t- - -- -I- E n U. -48tances corresponding to at least five times the previous maximum plastic zone size. This procedure minimizes crack retardation or arrest caused by the plasticity of the previous load leves. Thresholds were defined by crack monitoring resolution limitations as the highest AK at which no growth was observed in 107 load cycles, i.e., in terms of maximum growth rate of 10- 8 mm/cycle. All tests were initiated at low load ratio (R=0.05) and adjusments for R were made during the first load reductions. The procedure to attain high load ratio was found to affect the threshold value as discussed in Section 3.4. The method used in this study was to maintain constant the value Kmax of test initiation and reduced K as R attains the desired value. 3.4 EXPERIMENTAL RESULTS 3.4.1 MICROSTRUCTURAL ANALYSIS The microstructure of SA387 Class 2 Grade 22 steel was found by optical microscopy to consist of ferrite and bainite combination. The base plate metal, 184 mm thick,has a variation through thickness of the volume fraction of component phases, yield strength and threshold stress intensity shown Kth as in fig. 3.3. The clear grains in micrographs a, b, and c correspond to ferrite, the remaining is bainite phase. The volume fraction of ferrite phase varies from roughly 20% at the surface to -4q- Quarter thickness (T) Surface Location surface Yield strength (MPa) (MPavm) (MPa m) 390 8.3 340 8.6 290 9.0 R = 0.05 , moist air 30% relative humitity. Half thickness ( T) Fig.3.3 Variation through thickness of microstructure, yield strength and threshold stress intensity K for SA387-2-22, 184 mm thick base plate. -50about 40% at the mid-section. The higher percentage of the more ductile ferrite is consistent with lower yield strength: 390 MPa at the surface and 290 MPa at mid-section. The thres- hold stress intensity AK0 was also found to vary from 8.3 EMPaV-m near the surface to 9.0 MPaV-i at the center of the plate(96); consistent with the observation of other investigators that decreasing yield strength or increasing ferrite volume fraction causes an increase in Ko . The variations in microstructure and mechanical properties are due to differences in cooling rates through the thickness during heat treatment. The schematic of cooling curves at a quarter thickness location of.normalized and water-quenched thick section plate are shown on a continuous-cooling transformation diagram determined for 2Cr-l%Mo steel (74) fig. 3.4. Although the investigators caution that the accuracy of the transformation diagram is not sufficient for microstructural predictions, it proves very useful for qualitative interpretation of the microstructures observed in the thick plate and weldment. It may be noted that faster cooling rates as expected in the weldment would result in microstructures consisting of mixtures of fine lower bainite and coarse bainite. For very fast cooling rates, microstructures of martensite and fine bainite would be expected. - 51- X- r v u,c (.3 C C o o 0 0' _ _ o _ u a) _0 u O C .-D _Ln U, E -i LL N c' *O o ccu V_ c. Q -V_ ._ U - o . ._: C ._ 0 E 0 0 0c 0 0 U- :-C -'- u oo a) 0 E 0 -bN 0Cc U 0 O U E ' 0 '.- 0 o 000 CD 3o 000 00 00 - ajnlDDadw1a C~j I,. -52 The microstructure characteristic of the weld metal is bainite as shown in fig. 3.5; a range from coarse to fine bainite is observed resulting from the cooling rates and the thermal effects of subsequent passes on solidified filler during the multi-pass welding operation. The heat-affected zone is a layer of base metal of approximately 4 mm width, with very fine bainite microstructure. The transformation from ferrite-bainite microstructure of the base metal to fine bainite microstructure of the HAZ is caused by the high temperatures and the rapid cooling rates imposed on the metal during welding. The typical micro- structure from base metal to HAZ is shown in fig. 3.6b and represents the extent of the thermal effects due to welding. The variation of Vickers hardness and microstructure across the HAZ is shown in fig. 3.7. The micrographs at different locations are shown to illustrate the correspondence of hardness and microstructure. Fig. 3.7a corresponds to a base metal ferrite-bainite microstructure with Vickers hardness of 156 kg/mm 2. Fig. 3.7b shows the microstructure of the weldment at the transition from base metal to HAZ that is the limit of the zone subjected transformation due to welding. to thermally-induced Fig. 3.7c shows a micro- structure of fine bainite characteristic of the HAZ. The hardness level increases across the HAZ to a region of maximum hardness of 198 kg/mm 2. The microstructure at - 53 - ';·1- T .a 4 · Fig.3.5 Optical micrographs of SA387-2-22 weld metal (1% nital etch). - 5- Fig.3. Optical micrographs of SA387-2-22 steel, a) Heat Affected Zone (HAZ) metal, b) Interface between base plate and HAZ metals. - 55 Relative Location of Microgrophs a " b BASE METAL 200 c d e HEAT AFFECTED ZONE WELD METAL 0 0 190 0 c3c ~~ - 180 ~0 0 0 0 0 70i 0 G W 0 160 CrX I 00 Fin , 0 ) 2 4 I I 6 8 MM. a b C - r-~ d F 3 '7 e arolaton ' n Microstructure and Vickers Hardness in SA387 Thick-section weldment showing bose metal, heat-affected zone and weld metal areas Etched n 2% Nital -56this point is fine lower bainite shown in fig. 3.7d; this region corresponds to the unmixed zone where a small portion of the base metal totally melts and resolidifies without undergoing filler metal dilution. The unmixed zone bounds a partially melted region at the HAZ interface (75,76). These regions attain the highest temperature of the plate and undergo drastic cooling rates as shown by the hardness level and microstructure. 3.4.2 FATIGUE CRACK PROPAGATION Results on the fatigue crack propagation behavior of the base, weld and heat-affected zone metals are presented in this section. These were performed to evaluate the ef- fect of load ratio on the fatigue crack growth resistance in the presence of two environments: moist air (30% rela- tive humidity) and gaseous dehumidified hydrogen (138KPa). The results for the base metal are shown in logarithmic scales of growth rates vs. alternating growth rates vs. alternating stress intensity, fig. 3.8. istics are: there The main character- first, at growth rates above 5x10- 6 mm/cycle is no distinctive vironment or load ratio. dependence of growth rates on en- Crack propagation is only a func- tion of the applied stress intensity range. These results are consistent with the observed negligible influence of load ratio, environment, microstructure, and yield strength -57on the mid-growth regime of the fatigue crack propagation of steels (28). Second, the effect of load ratio on crack growth rates becomes increasingly evident for ultralow growth rates ( 10- 6 mm/cycle) approaching the threshold. The stress intensity at threshold WKth for low load ratio (R = 0.05) is nearly three times larger than that of high load ratio (R = 0.80) for both environments, moist air and hydrogen gas. The magnitude of the laod ratio effect is illustrated by noting that for example in moist air, a crack subjected to an applied AK of 8 MPaV-' would be dormant when the load ratio is 0.05 but it would grow at a rate of 2x10 6 mm/cycle for laod ratio of 0.80. Third, the relative effect of the two environments on the crack growth rates, observable nearthreshold, is emphasized for low load ratio and most importantly it is reversed for the two load ratios of 0.05 and 0. 80. For low load ratio (R = 0.05) hydrogen gas produces higher growth rates than moist air environment near-threshold; however, at high load ratio (R = 0.80) the crack propagation is similar in the two environments and only very near threshold moist air produces slightly higher growth rates than those of hydrogen gas. The environmental effect for high load ratios was confirmed in separate tests and the results are presented and discussed in Chapter 4. The results of the fatigue crack propagation in the -58- (al3/u! )Np/0 · _·_ 0 I ' '0T·__1__ · I L' I0 · __I__· " I · o_·__·_ · I11.111 I 0_______ _ 11111 'o I _ '0 I0 ________ I 'o I ____ I Is I 1 0 I 0 0(D -u -- C 0 a00 >, 0.i 0c 0 0n) 0 a _ U O 0o o 4. U Ua C\J 1 ,o E C' ., '-c o 4U o0 -o 0 >I U -' 4 O ~i -C 0 , EO H t- ("On- U) a) ww 4 ~1 0%.& at a .q . .:- 4 4 ' 44 r-- -..c0~0 : . .4 (D _ 0 z a) a) E * , a, 2 pq C0 o a) I _ o 'n t c I o C 0 _ 0 ri.. ( 0 o.: U o c o -1 'A, z rn a a T L* cJ L) E 1. ;.- 0 C i E I - '0 I t11t I I 11 i I I I I '0 (alA3/WW) NP/OP '31.V 11l 1 I I 'O NOllV9VdOd 11111I T0 I lltl '0 ),3VHD 3nu911J c2 GI E I "E CJ '0 oO~~ LO *4 cr _o - J. O oaO d 0 0 CaO 1. 00 LJ Z CM '0 1L -59weld metal are shown in fig. 3.9 for conditions similar to those of the base metal. The observations made for the base metal apply equally as well for the the weld metal. Note, however, that for low load ratio the threshold values are lower than those obtained for the base metal and the difference in crack growth behavior in moist air and hydrogen gas occurs only near-threshold. The linear regime for high load ratio approaches that of low load ratio in the midgrowth regime. The crack propagation behavior of the heat-affected zone is shown in fig. 3.10. The threhsold values Kth for load ratio of 0.05 continues to decrease compared to base or weld metal. Analysis of the crack propagation behavior at load ratio of 0.05 shows that the environmentally-sensitive region varies for the three microstructures. rates of 10 Crack growth mm/cycle in the mid-growth regime or linear regime of fatigue behavior. The weld metal environmental sensitivity disappears above 2x106 to the knee of the sigmoidal curve. mm/cycle corresponding The heat-affected zone shows similar growth rates for both environments even below the knee. The transition to environmentally sensitive region occurs at growth rates of 2x10 7 mm/cycle. The discussion of the results presented in this chapter focuses on the near-threshold sensitivity of the crack growth rates to variations in load ratio, microstructure and -60(eIAZ/u! '0 I0 0 I0 _, ! I/3 11 I I I I I )NP/DP I0 I0 '0 - I II T-T -T--T I I I co I I0o I T1I o I I I II I I 1 0 I'll, I I 0 I (.0 (JCU 0LO Ia o ~0 'D 0 cE -n0 -"c 0 \ U 0rn 0 U~~~a o LC)~~ O oo1 3~~~~~~~~~~~~~~~~~~~~~~~~~~~1 < < 0 a 0N C0) cn wQ 0 oE I U a 4 0 U~~~~~~~~~~~' U K 0 <-0) 0 0 03H 0) Lii I- C VO .r C) lJ 0-r rod r 0 Li 0 4· ,. . ODZ Si , ' (D 0o, zw E 410 0 )U)0 0 OC 4 4 N -i w 4 cr 0 ~o cr r~ _0 0 InI o ) o U 6 - a, 0 Os . N in~ rp CO -80 c00~· ,',0 Lo LC) 0 !r11 ) 00 -L~ WC o ULL · i z 170 (al/WUW) J I I UIJIJ I I I '0 NP/DP '31Vd JII 111 I n~ W 1f r (0 NOllV9VdOdd '0 AD~V3 o u~~) .C tRI' C~j o~ ..... i o 0 LO -a rOC a3C.~ toZr, 0 .. C a C: O on-G C :0) t Ld H -o II I 0 C3 i/I: LLL I I '0 3g1.3 I L )o -61(l:)3D/U! ) NW OP to 'o 0N- w w - w w f w s s O _ O _- w s s w j'111 i 17 Npp 0 w E s w s E I1 TFill-ll 1 0 I'I 0p 0 E s s w E . I I = P I 1i . . w . I i w I i w E It w s 0 77 CO s II i I I OG O o o OOt. )a . 0)V Cu, 0 0 o 0 Iii U~~~~~ *>0 U~~~~~ 0t',, C S U _ . . O . ' oo E 0E E QiC E 0 ,, CjE %. U - c 9S 0 0 o , 0) q: <3 o 0 q o 0r -C I- ir CDZ 0 .,. cw : I -j 0 o0 `p N) N .oa, U) 10 ..CD I 0 I o:, I L-) I I I I-) II_ N IV, ! 1-1 ! i ; I I C6 c 4 ''o * cr0 o P,r c ° / ,6 QN q: o 0C. 0 - U 0 - I.......... I I J aQ7o (al)/wW) LL oO m0 -- L I 0)< -a) o oo I- .. oc· .. uU) a) LJ ,. 2 U < I) o0 4 cn LUJ IO J. .4 )N c O Zn . ra %raVq Qo 1. I Ld IN 0 - tZ 111i 0 CDo I ! 1 11 . I I= .I 0'0 NP/OP '3IlV |!; ' I 1 1 11 I 1 'To ! N Il I I I I 'I NCXLVOV!dOHdfDVD3 3nD1LV Ct I D00 '0 ._r 11 -62- environment. At low load ratios crack closure phenomena dominates the growth behavior; understanding of the factors affecting closure levels are yet unclear. At high load ratios the crack remains always open and the existing conditions at the crack tip are more predictable from the externally applied loads and environment. t,,;:mvih h ... .... _ ~ S ...... rects are alscusseu he±u VnyX i"S".-...... Environmental efJ affectsthe it closure phenomena; in-detail discussion is presented in Chapter 4. DISCUSSION 3.5 THE NEAR-TiHRESHOLD FATIGUE 3.5.1 The results CRACK PROPAGATION of fatigue crack propagation in the weld- ment show a transition in behavior from the mid-growth II regime to the near-threshold regime. Crack growth rates fI - 6 mm/cycle depend only on the stress intensity above 5x.10 range K; however, lower growth rates show a pronounced sensitivity to load ratio, microstructure and environment. The crack propagt Lion results fo the weldment are shown in fig. 3.11 for moist air environment. tion The transi- rom mid-growth regime to near-threshold is first anlayzed in terms of the load ratio effects, in order to illustrate the relative scales of plasticity and crack tip opening displacements (CTOD). the base metal ( The crack growth rates of thickness), weld and HAZ microstructures (al3/! ) NP/p O .' . . p,- 0 IC) no '0 I ""'' I _ . . I '0 I _ ""'' . . .' -I " '0 o,0 o . . . I"'' .. - I' ___ __'- -~ _-----I"IIIII' 1 ... ' o -G u a n 0o tA - oo_ C.a a.i 0 0n o E _ o, co o 0 ,_ CJ do 0 a . . ._ 0 z a) 'O z 2 U) v' z oc E CUl- . Z~cj~ F-J O N Mr . O0 p 0 *0 - 0) 0- - i~ w \o 0) E 3Eolo -u o (D -- AW.4 usfum11 No O sk f N U')o e )' 0 A a ~p , ( .< O13 _Eco c00 0 0 000 ooEc o r>S O E rn c~ h b_ N 31 r x NJ UY IIIJ1 1 I 'o (alA3/wJ) coo ,0i I o .?IM L. N C\j n W 0o oo o ,Q I I1,1iL o I . I_ llIL T'o LI I 1iii i, ,I - 'o NP/DP '31./V NOIlLV9dOdd >3V03 3n9lLV- - - N oD ' -64begin to differ for the two load ratios at a K = 11 IMPafl. As for the base metal(% thickness), the deviation is observed at EK = 20 MPaV i-. The comparative size of plastic zone and crack tip displacements are shown in Table 3.3. Plastic zone size and CTOD values at the transition point are about 22 times (r = 100 larger num; CTD = 1 for R = 0.80 than for R = 0.05 m). This indicates that crack growth rates become insensitive to load ratio when the corresponding values of r at CTOD for low R reach a critical value. This Y critical value is expected to be determined by the ductility of the material ductility being the dominant parameter enhancing crack closure phenomena. At R=0.05 and AK less than 11 MIPaV--`the CTOD and maximun plastic zone size are less than 1 m and 100 m respectively. For all the three microstructures the cyclic CTOD is 90% of the maximum CTOD meaning that upon unloading, the crack tip would be open about 0.1 m. Such a small amount of CTOD would require near-perfect mating of the fracture surfaces; it is likely that early contact of surface asperities in the wake of the crack do not permit the CTOD to reach its minimum value, consequently the stress intensity range and the load ratio sensed by the crack tip are different than those applied to the specimen. The early contact of the crack surface was first observed in condition of plain stress (66) when plastic strain and distortion are large and closure occurs even at significant values of CTOD. -65- 0 V. N U a N; r o0 .D .O O O O o NN (N( a~ co~~~~c oc ko (N o E U O O (N N1 V U) ) Cd O E--i OII O E dP )) Z:J co o-4 N O· (N U *ro O 0) N 0rl H* U 00 O 0 m v u) 110 *-. C Cl o O II 0 II 0 N L 4-) U -ri (d f PJ II Z% 0 N N ~~~~Q NJl H- 40 ., 4-) O ro -iH CU) 0U N N -W -H U) - q Cd - a, Z o H H - r -i r- H .,I U U N EH a I_ r4 a) U i c4~~~~~~c a N o~~~~~~~~~~~~ cu~~> N a) N U) F ~ Xf o0a VVu n o ro h~~~~ E U . 1)rX0) U) d m H a) _z N < U) Cd U K 00) ·r4 >1 -66In plain strain conditions, as existing in the fatigue crack propagation near-threshold, crack closure results from the fact that the small scale of CTOD increases the sensitivity to irregularities on non-matching parts of the fracture surface. 3.5.2 CRACK PROPAGATION AT LOW LOAD RATIO: CRACK CLOSURE PHENOMENA The fracture process of engineering materials produces permanent distortion in the neighborhood of the crack tip, leaving fracture surfaces with asperities and sites that do not match and come into contact during unloading. The contacting points support load and therefore prevent the crack tip from sensing the full external applied loading cycle. The re- sult is that the prevailing stress intensities at the crack tip are less than those calculated from external loads and aver- age crack length. The load bearing capacity of the asperities set a lower limit for the effective load ratio. Contact of asperities due to ductility, Mode II loading, microstructure and environment prevent the crack from sensing lower stress intensities than those at which fracture surfaces contact and transfer load. The results of crack propagation at low load ratios in the weldment are discussed in terms of the main factors that affect: closure phenomena: ductility, .microstructure, environment,and a Mode II loading component present at very -67low growth rates. 3.5.2.1 EFFECT OF DUCTILITY ON CLOSURE Results on the crack propagation of steels at low load ratio (64) have shown that lowering the yield strength and increasing the ductility produces lower growth rates and higher AKo. It has not been clear whether yield strength, or ductility or microstructure, or a combination of these factors are responsible for the effect. The results of this work indicate that ductility and grain size are dominant in determining closure levels. The effect of ductility is discussed here using the results of crack propagation in SA387 base metal specimens of different ferrite volume fraction and similar grain size, corresponding to differentplate thickness locations as shown in fig. 3.3. The ferrite volume fraction varies from approximately 20% to 25% to 40% for the surface, quarter thickness, and half thickness of the 184 mm thick plate respectively. The threshold value of volume fraction. K is greater for the higher ferrite This is consistent with results on plain carbon steels (64) which show that for the grain sizes above 40-50 m the threshold value increases with increasing ferrite volume fraction. The more ductile phase presents larger plastic elongation to fracture producing higher asperities and thereby enhances closure effects. A model has been proposed (13) to account for the role of ductility -68on crack closure via surface roughness and be able to calculate the opening stress intensity Kcl. This model incorpo- rates only one of the factors that influence closure, the ductility of the material. This effect is currently designated as plasticity-induced crack closure (16). The low load ratio results of the weldment shown in fig. 3.11 are consistent iwth the above observations. The base metal with a ferrite-bainite microstructure produces rougher fracture surface than the homogeneous bainite microstructure of the weld and HAZ. The closure level for base metal is expected to be higher than weld and HAZ metal resulting in lower growth rates and higher threshold value AK 3.5.2.2 MICROSTRUCTURAL EFFECTS ON CLOSURE At low load ratio the near-threshold fatigue crack propagation decreases with increasing grain size, as shown by the results of the weld and the HAZ metal. The threshold value increases even more with the presence of the ductile ferrite phase in the base metal. These results are in qualitative agreement with findings of other investigators (63,65). value of threshold with V-T (43,64), The Kth has been found to increase linearly being the grain size. Optical observations (63) of the spread of the slip band zone near the tip of the crack indicates that: first, smaller than that predicted from continuum mechanics it is -69calculations and second, that the crack propagation becomes structurally sensitive when the magnitude of the slip band zone or cyclic plastic zone size becomes of the order of the grain size . A critical size of the slip band zone has been proposed as possible condition determining threshold (43). The critical size is determined by whether or not the slip band at the crack tip propagates into adjacent grains. Measurement of closure load by compliance and electrical potential techniques for different grain sizes has shown that in face Keff is dependent on grain size (39) or stated otherwise, the closure level is grain size dependent. The closure level can be influenced by grain size through the fracture surface roughness resulting from more or less tortuous crack paths. Assuming that at low stress intensities most of the deformation is accommodated by only a few slip systems and in the extreme only one slip system per grain, it is reasonable to think that the crack propagates along a preferred slip system in a grain up to the neighborhood of the next grain boundary, grain boundaries being the most likely sites for changes of crack path direction according to the orientation of the next grains. Large grain sizes would give rise to rougher fracture surface than fine grain sizes. This is apparently the case observed with the microstructures of the HAZ and the weld metal, HAZ having the finer grain size. -70Low growth rate results in Al-5.7Zn-2.5Mg-l.5Cu alloy at R = 0.1 and Ti-8.6A1 alloy at R = 0.2 obtained in vacuum (77) show consistent decrease in growth rates with increasing grain size. These results were interpreted based on the concept of reversed slip of dislocations within the plastic zone (63). Generation motion and pile up of dislocations at grain boundaries or hard particles occur during the ascending part of the loading cycle; during the unloading the back stress within the pile-up forces the dislocation to move in the opposite direction on the same slip plane until the back stress equals the friction stress of the matrix. This process forces dislocations out of the slip plane to the crack tip surface and results in lower crack propagation rates. The number of dislocation moving backwards depends on the amount of cross slip and activated secondary slip systems. In fine grain ma- terial the high local stress concentrations at the front of the pile-up produce slip in the neighboring grains as well as on secondary slip systems in the same grain. Upon unloading, reverse slip is less likely in fine grain material; the dislocation in the neighboring grains become trapped within the grain while dislocations on the secondary and primary slip become interactive. Consequently, the number of dislocations moving backwards decreases and the crack propagation rates are faster for the smaller grain size -71material. This mechanism is schematically illustrated in fig. 3.12. The effect of grain size on crack growth rates through reverse slip is thought to be more effective at low load ratio where the scale of plasticity few grain diameters. is of the order of a At high load ratios the plastic zone size is large enough with respect to grain size that the averaging or continuum effect characteristic of mid-growth regime is still operating near-threshold. This is consistent with very little variation in threshold of the three microstructures at R = 0.80. Consequently, at low load ratio, the two proposed mechanisms appear additive in their effect on crack growth rates. Fine grain microstructures, by reducing the reverse slip and the tortuosity of the crack path, produce higher crack propagation rates than coarse grain material. These mechanisms are consistent with the results of crack propagation in the weldment microstructures. The lowest threshold is obtained in HAZ which has the smallest grain size of the three microstructures. The base metal, besides having large grain size is a combination of ferrite and bain. ite and gives rise to higher threshold than weld or HAZ metal. At R = 0.80 the near-threshold crack propagation behavior the weldment is nearly independent of microstructure. of -72- I I I PLASTIC I SIZE ZONE I LA I I I 1~~~~~~I I Fig. 3.12 Schematic Showing the effect of grain size on the dislocation structure at the crak tip. a)Small grain size, b) large grain size. (77). -73The mechanisms discussed above concern grain size effect on a single phase microstructure; however, duplex microstructures with phases of different ductility as in the base metal, have been found to affect growth rates near-threshold. The results on SA387 Class 2 Grade 22 steel (78) show that the threshold value increased with increasing ferrite volume Results of other investigators (65) have confirmed fraction. the pronounced effect of relative volume fraction of the phases in duplex microstructures onthe crack growth rates. Microstructural effects on crack propagation rates are significant at low load ratios and near-threshold. These ob- servations lead to the following suggestions: 1 - Microstructure becomes a relevant factor for the growth rates when the plastic zone size reduces to the scale of the microstructural parameter. 2 - There appears to be a transition in crack propagation behavior from microstructurally insensitive at high growth -5 rates ( 10 mm/cycle) to microstructurally sensitive near threshold. This transition is assumed to coincide with the point of deviation of the results from the different microstructures as discussed in 3.5.1. A mechanism to explain these observations is proposed by which a propagating fatigue crack with a plastic zone size of the scale of the microstructure becomes increasingly -74sensitive to the microstructure ahead of the crack tip. The macrocrack propagates in a general orientation determined by the type of loading but the microscopic crack path depends primarily on the microstructure ahead of the crack tip. The crack path then becomes tortuous and gives rise to two effects that result in low measured growth rates, 1 - The net rate of horizontal advance of the crack is reduced by the excursions taken in seeking lower energy input fracture paths. 2 - The tortuosity of the crack paths brings into effect local Mode II type loading. Certain portions of the crack ad- vance by Mode I - Mode II combination. The Mode II loading causes displacements in the plane of the crack enhancing contact of the surface asperities and producing higher closure loads. 3.6 DISCUSSION The presence of second phase particles in -Fe-Ni-Al alloy has been shown to decrease the growth rates by reducing the homogenity in the slip behavior in the plastic zone (65). For ferrite steels grain refinement decreases the threshold value AhK; the volume fraction of ferrite affects threshold depending on grain size (64); duplex microstructures of ferrite/martensite give rise to significantly different -75near-threshold behavior depending on whether martensite phase encapsulates or is encapsulated by the ferrite phase (65). Near-threshold the plastic zone for low load ratios becomes of the order of the grain size and the fracture path or surface roughness is strongly determined by the microstructure. The size of the asperities or roughness is thought to be a relevant factor in closure. The effect becomes signifi- cant when assisted by relative displacements of the fracture surfaces parallel to the plane of the macrocrack (71). The displacements are caused by the Mode II contribution in the near-threshold regime. induced crack closure. The effect is designated as roughness- -76CHAPTER 4: HYDROGEN DEGRADATION OF THE NEAR-THRESHOLD FATIGUE RESISTANCE OF STEELS 4.1 INTRODUCTION The environmental degradation of the fatigue life resis- tance of metals and alloys is a materials problem of great concern in the design of pressure vessels for coal conversion processes. The vessel wall, although insulated from the reaction processes by a refractory lining and additional stainless steel or Incoloy weld overlay, may be expected to be in contact with chemically aggressive atmospheres including H 2 0, H 2 , H 2S, etc. ((36). In general, the increasing engineering requirements for materials to withstand long service life under cyclic loading and aggressive environments has motivated a large effort to study fatigue. The search has been in the characterization of the fatigue response and the understanding of the mechanism of fatigue and fatigue - environment interactions; however, the complexity of the problem has not made possible quantitative evaluation, rather most of the advancement has been made in the phenomenological understanding of the problem. In fatigue the crack initiation and propagation mech- anisms are not sufficiently clear and the fatigue - environment interaction is only known through the acceleration of the initiation or propagation rates. -77In recent years, increased attention has been devoted to the fatigue - environment interaction and to the study of the elementary processes involved in te overall hydrogen degradation of fatigue resistance; such processes are: - transport of environment to the crack tip - physical adsorption to the fracture surface - surface diffusion and molecular dissociation - chemisorption of atomic hydrogen - dissolution of hydrogen in the metal lattice - diffusion of atomic hydrogen - hydrogen embrittlement. Results of environmental degradation of the fatigue resistance of 2Cr-l%Mo in this chapter. steel at low growth rates are presented The degradation of the fatigue resistance in water vapor containing environments was found to be more pronounced than the degradation observed in gaseous molecular hydrogen. Hydrogen embrittlement was proposed to be the basic process reducing the fatigue resistance; the levels of degradation being only indications of the effectiveness of the hydrogen source and relevance of the elementary reactions. 4.2 LITERATURE REVIEW The fatigue crack propagation in metals occurs mostly as a response to loading conditions; however, a driving force for crack growth can result from environment - metal -78interactions. The environment-induced crack growth in some circumstances becomes the determining factor of fatigue life and therefore must be included in assessing fatigue resistance of engineering components. Work as early as 1932 has shown that the fatigue life could be increased by simply pumping away the air around the metal component (79), indicating also the aggressiveness of moist air environment. Several studies have indicated that water vapor is the outstanding aggressive component of moist air; for example, results in aluminum have shown similar fatigue life of specimens tested in moist air, water vapor and hydrogen as compared to those tested in vacuum, dry oxygen and nitrogen (80,81). It was suggested that hydrogen embrittlement was responsible for the reduction in life and that the atomic hydrogen was produced by the surface reaction of water vapor and aluminum. The results on crack propagation have also substantiated the aggressiveness of hydrogen and water vapor containing environments. For example, aluminum under fatigue loading at R = 0.1 was found to show increased crack growth rates in water vapor and moist air environments; this effect was pressure dependent with a lower limit of the water vapor partial pressure below which no deleterious effects were apparent (82). Results of fatigue crack growth in H-ll steel in the -79presence of moist and oxygen containing environments (83) demonstrated that the crack growth rates were increased in the presence of water vapor and hydrogen gas and decreased by oxygen concentrations as low as 0.7 percent by volume. The accelerating effect was interpreted in terms of a hydrogen embrittlement and the retardation effect in terms of preferential adsorption of oxygen producing a surface poisoning effect, which would be an interesting basis for design of hydrogen embrittlement inhibitors. Several studies have sought to characterize the effect of surface adsorption on the mechanical properties of the materials (84,85,86). Earlier work showed the effect of moist air on lowering the microhardness of germanium (87), and the effect of water and moisture on reducing the fatigue life of AISI-4340 steel (88). More recent studies have concentrated on understanding the surface chemistry (89), evaluating reaction kinetics (90) (frequency, temperature), determining modes of environmental-induced crack growth (stress corrosion cracking, corrosion fatigue) (19), and on finding the rate-limiting steps for the mechanisms of environmental interaction. Linear elastic fracture mechanics has provided a basis for evaluating relative aggressiveness of environments in terms of crack growth rates. Studies on the intial reaction of oxygen and water vapor with fresh iron surface showed that in the presence of oxygen -80- .gas the reaction proceeds at a rate about 6 orders of magnitude faster than in the presence of water vapor (85). It was concluded that the water vapor - metal reaction occurs by the initial formation of a physically adsorbed precursor. The precursor reaches a saturation coverage of 80 pct and forms a disordered c(2x2) structure which was determined by low energy electron diffraction technique (89). The prucursor was model- ed as having surface mobility before dissociation and chemisorption. Metal oxide formation and hydrogen embrittlement are likely products of the overal reactions. The temperature effect on the environmental degradation has been studied through observation of the crack propagation rates in dehumidified hydrogen and water vapor environments. The reproducible temperature activated behavior of the growth rates (85,92,93,94) has been interpreted as produced by the temperature dependence of the reaction kinetics. This pro- cedure is frequently used to determine possible rate-limiting steps of the overall environmental effect. Further studies on 2219-T851 aluminum alloy at R = 0.05 and frequency of 5 Hz indicated that the fatigue crack growth rates increased as water vapor pressure increased from 1 to 26.6 Pa (6) (0.0075 - 0.2 torr); a saturation pressure of 26.6 Pa was observed, above which growth rates remain independent of water vapor pressure and are comparable to those obtained in air (40 to 60 pct relative humidity), distilled -81water and 3.5 pct NaCl solution. The crack growth rates ob- served at pressures below the lower limit (1 Pa) were comparable with results obtained in vacuum and dehumidified argon. Similar results had been observed previously (95) on different aluminum alloys at R = -.1 and growth rates -4 below 10 mm/cycle. Studies of near-threshold fatigue crack propagation in SA387 steel (96) (350 MPa yield strength) and in SA 542 steel (97,98) (590 MPa yield strength) have shown that moist air environment produces lower growth rates than hydrogen gas. However, the reverse behavior has been reported on 300-M steel (99) (1500 MPa yield strength) and in Ni-Mo-V steel (100) (661 MPa yield strength). cussed in this chapter These observations are dis- and it is shown the environmental effects are the same in medium and high strength steel. The results presented in this chapter show that water vapor accelerates crack growth rates even with respect to hdyrogen gas. This effect is presumably due to hydrogen embrittlement and it is observable at high load ratio where the closure phenomena is negligible. Hydrogen embrittlement appears to be enhanced by the surface reaction of water molecules as a more effective source of atomic hydrogen than the surface reaction of molecular hydrogen at room temperature. -82- 4.3 EXPERIMENTAL PROCEDURE 4.3.1 IMATERIALS The material for this study was 2Cr-l%Mo SA387 base plate metal; the material specifications are described in section 3.3. Specimens for comparative environmental ef- fects were chosen from the same plate thickness location, as indicated in fig. 4.1. Fatigue testing and crack monitor- ing were done as described in section 3.3.2 and 3.3.3. 4.3.2 GAS PURIFYING SYSTEM A complete schematic of the purifying system is shown infig. 4.2 and detailed description is found in ref. 97. The helium and hydrogen source are rated with impurity levels of less than 3 ppm of moisture, less than 1 ppm of oxigen and less than 0.5 ppm of hydrocarbons. The objective of the gas purifying system was to reduce the total impurity levels of hydrogen and helium gases to the order 1 ppm. This was achieved through the following process: 1 - Oxygen and moisture removal by gas flow through an oxygen catalytic converter and a Matheson purifier Model 461. 2 - Removal of moisture traces by flow of the gas through a cold trap of coil tubing embeded in liquid nitrogen. The system has vacuum and baking facilities to assure cleanness before each test, or between transition from one environment to another. The operating procedure includes: -83\ I .4 Ed* (O e *H O e u r4 o E ., E E UUe 4Jn 9:w (0 Od 0UHm F H It R( u e 0., IQ · rl 4 m CJ)o c, n 01Q C: w .H tt 4 0~ \1 ED -84- L 0.6 -o EUE U4-P Lo 0 SI- S m Gr 4J>c 0 I- V) w z 0 z LI cI 4r E r L 0 L/10 0.0. N 0E LO -851 - Bake out at approximately 100 C during an inert environment pruge. 2 - Evacuation to a pressure of about 5 millitorr. A cryogenic impurity adsorbing pump is backed up with a roughing pump. 3 - Pressurization with the gas for the desired environment. Crack propagation in the desired environment was accomplished by mounting a vacuum tight aluminum chamber on the specimen and maintaining a gas flow rate of about 20 cc (STP)/ min. Dehumidified gases were also obtained by flowing the gas through 3 columns of drierite (CaS04) and 2 columns of phosphorous pentoxide powder (P205). The results were found to be essentially the same using either system. This is not surprising as it is shown in this chapter that water vapor is the gas impurity with greater effect on growth rates due to its fast adsorption and its efficiency as a hydrogen source. 4.3.3 TEST PROCEDURE The characterization of fatigue crack propagation in the various environments was done using two methods. One method makes use of one specimen per environment, the other allowed for change of environment with the same specimen at constant -86&K. The first method is commonly used, however it has the following disadvantages: a) at crack growth rates near- threshold, small variation in rates due to environment are not well perceived and repeatability of observations is made difficult by the time duration of each test; b) each specimen develops a characteristic crack front geometry which generally deviates from the ideal straight crack front. This effect on the growth rates is not yet quantified and can cause confusion with the pure environmental effect. The second method of using one specimen for several environments allows for careful comparison of growth rates at constant AK in the presence of different environments. It permits the check of reversibility of observations and it is sensitive to small differences in effects of nearly equal agressive environments. It also allows for the comparison of crack growth rates in different environments with the same crack front geometry. The second method requires stopping the cyclic loading and to vacuum, bake and flush the environment supply line thoroughly. -874;4 EXPERIMENTAL RESULTS 4.4.1 ENVIRONMENTAL EFFECTS ON THE FATIGUE RESISTANCE AT LOW LOAD RATIO The study of the near-threshold fatigue behavior of several materials has shown increased sensitivity to environmental effects at rates below 10 ratio. mm/cycle and low load The fracture surface of specimens tested in moist air at R = 0.05 show a distinctive oxidized band at low growth rates. moist A typical fracture surface of a specimen tested in air at R = 0.05 is shown in fig. 4.3a illustrating dense oxide band at low growth rates. the The oxide film some- times forms discontinuously as shown on micrographs of weld metal in figs. 4.3b, 4.3c; a high magnification of the oxide deposit on the fracture surface is shown in fig. 4.3d. Auger measurements have corroborated the fact that the oxide film thickness increases as growth rates approach threshold level (70,98). The oxide formation at low load ratio has been proposed to be a direct consequence of surface fretting during the portion of cycle in which crack closure occurs. The thickness of the oxide at the same time enhances closure and gives rise to higher threshold values. Surface chemistry studies have indicated that the oxygen metal reaction proceeds at a faster rate than the water vapor (85) and it was then necessary to determine the component of moist air that produced the increased oxidation rate. 0o 0 l: Ir) I o ooE '- 0 _ 0 . ac 0·l ,oo w 3 -W <|s. .°: . t. C ILI t n0 ( > t`uD 0o . 0 V v ~ . Jt Q V ci IT zw 0 a. 0 -J 4 w 0 'M0 c0 0 m 2 3c Cf w LU U) I- t 0 ' (00 c o a . 0 cja: 1a Z (m c Ib O -89- The results of tests intended to compare the effects of different environments on the crack propagation at load ratio of 0.05 are shown in fig. 4.4. The environments tested were dehumidified oxygen, dehumidified hydrogen, dehumidified air, and moist air. It also shows results obtained in commercially pure hydrogen denoted in the figure as hydrogen. Commercially pure hydrogen is rated at 3 ppm of water vapor and dehymid- hav ty 'fapur2 o1-3 ified hydrogen is calculated to have a purity of 10 to 10 ppm of water vapor. tially Dehumidified environments exhibit essen- the same threshold level of 5.2 MPaF-m - to moist air environment AK = 9.0 MPa Vi/ as contrasted Consequently, m at low load ratio the presence of water vapor decreases crack growth rates producing higher threshold. The water vapor effect on the crack propagation rates near-threshold is highlighted by considering a crack subjected to a AK = 8.0 MPaVi7; according to the results of fig. 4.4 the crack would be dormant in the presence of moist air but it would grow at a rate of 10 ments. mm/cycle in dehumidified environ- The main factors responsible for this phenomena are the increased surface roughness and oxide formation at low growth rates in moist environments; both factors contribute to raising load thereby decreasing the effective stress intensity at the crack tip. The difference between calculated and effective stress intensity range at the crack tip - 0(Ia33/U!) NP/DP 0c c C o 0) c E 0 e. I- .C cr. z x Z zUn - -c 0 vc tZi LU z co t3 c c o. 0 Ž (alID(OuW)NP/OP '3it'V8 NOIIVOVdO d 13)VO 3lOI.L1= -91increases, causing an apparent increase in the material fatigue resistance. The roughness mechanism is substantiated in this work and was observed in other materials (11). The oxide mechanism has been suggested by several investigators (15,19,102) and was quantified and modeled as the oxide-induced crack closure (70). However, results on crack propagation of 300-M steel (99) and Ni-Mo-V steel (100) indicated that purified hydrogen gave rise to higher threshold value than moist air at low load ratio. These results pointed out, among other things, the difficulty of studying the relative environmental and closure effects at low load ratios. Crack closure effects can be produced independently and/or sinergystically by factors such as ductility, fracture roughness, oxide, Mode II loading, etc.; the relative contributions of these factors are not yet well understood. 4.4.2 ENVIRONMENTAL EFFECTS ON THE FATIGUE RESISTANCE AT HIGH LOAD RATIO Testing for environmental studies was done under high load ratio conditions to assure that the crack remained open and exposed to environment during cycling. Variation in crack propagation rates with environment were monitored at constant AK in the same specimen. This method reduced data scatter produced by variations in crack front geometries. -92Partial results of a test performed on base metal 2¼Cr-l%Mo steel are shown in fig. 4.5. These results show that moist air accelerates the crack propagation rates relative to purified hydrogen. The effect is reversible with change of environment and is consistent with results obtained on high strength alloys 300-M and Ni-Mo-V steels. If hydrogen embrittlement is causing the degradation of the fatigue resistance, then the results would indicate that moist air is more embrittling than molecular gaseous hydrogen. The hy- drogen embrittlement process has been considered as produced by atomic hydrogen which suggests that the observed difference of the two environments reflects the relative driving force for the environment - metal reaction, leading to production of atomic hydrogen. In order to evaluate the relative effects of moist air, dehumidified air and hydrogen gas on the crack growth rates near-threshold, a different experiment was performed, and the results are presented in fig. 4.6. Moist air environment produces higher growth rates than dehumidified air or hydrogen gas, as shown by the results at AK = 2.95 MPaVm', AK = 3.0 Iv4PaV -m' and AK = 3.4 MPaVTf; it is clear from the results that water vapor degradates the fatigue resistance near-threshold. It is also shown that hydrogen gas produces higher growth rates than dehumidified air. The acceleration of crack growth rates is assumed to be the result of hydrogen enmbrittlement in the presence of moist air and hydrogen gas -q3 0 n () c o0 O Co 0) ¢. N c0 o,. O c - o00c~ co ,. C -a N z oCo *i O c-- oC - O I 0 d ww HN3-1 N9V8D - q4- c 0 CP PDe 0 O ) K- N U 00 00 0c a) -- N) 4) " U) Z rQ NQ W 0 ) C. @ z TLLI ID~ 3 -Q u C ._~ 11 rc) (. N CQ u N u N W HN31 u N NO1V8d u-i N I, N : N -95environments. The faster growth rates in moist environment than in gaseous hydrogen indicate that the reaction of water vapor on the fracture surface is a more effective source of atomic hydrogen than the reaction of molecular hydrogen on the metal surface. 4.4.3 COMPARISON OF THE ENVIRONMENTAL EFFECTS AT LOW AND HIGH LOAD RATIO The relative effect of moist air, hydrogen and dehumidified air on the fatigue crack growth rates near-threshold at R = 0.80 appeared to be opposite of the results obtained at R = 0.05. Several tests were performed to confirm and evaluate the observed trends. air environment at R = 0.05. A test was conducted in moist The loads were shedded to a AK such that the crack was practically stopped and no appreciable growth was observed in 5x105 cycles; at this point the cyclic load amplitude was reduced to zero, the chamber evacuated, the system flushed with dry air, and the supply lines baked to reduce adsorbed water on the tubing walls. This procedure was common to all environmental changes. The results in fig. 4.7 show that dry air environment causes an increase in crack growth rates by a factor of nearly 5 over those of moist air at comparable K. The effect is reversible with environment and opposite of the results shown in fig. 4.6 for R = 0.80; at high load ratios moist air causes higher growth rates than dehumidified air. -q6- ru a) o 0 U c0 .3 U) C o LD a) OL 0 o cO x oQ "0 NO a 0 C 0C.C - - O _0 OI u, v>o ._r 11 -k v Nv ww H19N31 N N M1tV N -97A separate test to confirm the environmental effect at high load ratio was conducted on the heat-affected zone metal (HAZ) of a SA387 thick section weldment at load ratio of 0.80. The results are shown in fig. 4.8 and clearly indicate that at high load ratio and near-threshold, moist air accelerates growth rates with respect to dry air, the reverse of observations at low load ratio. At R = 0.05: (da/dN)moist air < (da/dN)dry air At R = 0.80: (da/dN)ist air > (da/dN)dry air Finally, a test was performed in helium gas at 138 KPa to compare crack propagation in moist air with inert environments and the results are shown in fig. 4.9. The crack propa- gation rates are higher in moist air than in helium gas throughout the range of stress intensity tested. The condition for studying environmental effect on fatigue crack propagation are best at high load ratio because closure is minimized and the crack remains open and exposed to the environment during cycling. At low load ratio crack closure phenomena becomes important and it is difficult to assess the synergistic effect of closure and environmental aggressiveness. 4. 4.4 FRACTOGRAPHY The fracture surfaces were examined using scanning -98 c C O0N O ar Ul ) co O O r u,, o oo u TO rr~ O O- OJ Q 0r- UE O > CD C C m) Z ww H±DN31 >NdoV~ Qa ) 4- a) 0 OL - ) U) N -( (al/u!) p/opN 0 Ir I l I I 0o 0 O r l LI I Il.I 1 I I I _ _ 0 ~~~~~~~ o 10 _ ______~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ______ _ _ _ . .~ m11111l II I 11111111 I jIl 1I 0 'O __1 __ O llll 11 § ml[ill 0 0 0._ O > N O t) ao~ S Lr o . v 0 _ 0 0 0re) _ O zO 0 ('O ?E 0CJ _ >: > O; 4, I '1 1- Q 00 E 0 F-n' e "0 Zo O a,, F o 0) _ 0O * 'C% QJdQ . E I .O _ I) I LLJ LL N y '0 0 _ 0 ) o n 0 CO rn F- Ed n LL U) cr (N yoC ._ Io 0 n (.tU 0 0 0 T' U 0) LE -C- . L. U- NU II I '0 .-. .r o E a- o, 0 v, t. C/) o 0 a ._. I I I III I I I I _ '0 ~~ _ I ,0 '31V (al3A3/WuW)NP/OP I ~ II I I I ~~~~~~~~~.... III I I I I I I I I I _ ')10 (010 '0 NOIlVOVdOdd >O38D33l911tvJ I II I I r\ TO '0 -100electron microscope and the selected fractography illustrates the following features: 1) The fatigue fracture surface near-threshold produced in moist air at R = 0.05. 2) The changes in fracture morphology due to environment for moist and dehumidified air at R = 0.05 near-threshold. 3) The fatigue fracture surface near-threshold produced in gaseous 4) hydrogen at R = 0.05. The changes in fracture morphology due to environment for moist air and gaseous hydrogen at R = 0.05 near-threshold. 5) for moist The changes in fracture morphology due to environment and dehumidified air at R = 0.80 near-threshold. Fig. 4.10 shows the fracture surface near-threshold just at the transition from moist to dehumidified air environment at R = 0.05 (see the insert micrograph in fig. 4.7). Note the increased surface roughness in moist air and the preferential oxide formation at the ridges of the asperities where contact fretting of the mating surfaces may occur (oxide produces bright areas due to electric charging). ness is a consequence The increased rough- of the microstructural the fracture path in moist air. sensitivity of The change of environment at constant AK causes a drastic change in fracture morphology; dehumidified air produces less fracture roughness and no observable oxidation, as illustrated in the higher magnification fractographs of fig. 4.11. The crack growth rates in - )01- Z 0 o (D 0cr EL O cr a AK=10 MPaV F 3. 4.10 Transifion in fractu re surface morpholo0y procduced y mois clnldry a'r en/ironrnen-.+- R=O.05 AK= I0 MPao/" ' DRY AIR a) MOIST AIR b) R = 0.05 Fig.4.11 Characterist ic fatigue fracture surface near-threshold at R= 0.01I R.H.) a) Dehumified air K=IOMPa,/m-. b) moist air(30-40 pct -103dehumidified air at R = o.05 are faster than in moist air because of the crack closure produced by the surface roughness and oxide in moist air. Fig. 4.12 shows the characteristic fractography produced in purified hydrogen at R = 0.05 near-threshold. The fracture surface presents random intergranular fracture near-threshold presumably of the grain boundaries closely oriented to the overall fracture plane. The intergranular fracture is drastically increased at high stress intensities but nearthreshold the fracture surface resembles that produced in dehumidified air; qualitatively both environments produced similar fracture surface roughness. The threshold in de- humidified air and purified hydrogen was found to be the same (Ko = 5.2 MPa\VFT). The interesting contrast of the fracture surface features produced in moist air and purified helium is shown in fig. 4.13 for R = 0.80 near-threshold. Fig. 4.13a shows the change of fracture appearance with environment. Fig. 4.13b and fig. 4.13c clearly illustrate the difference infracture mode in the two environments. In moist air the fracture surface shows a patch-like appearance; each patch showing a characteristic type of fracture, some showing a cleavage-like fracture along particular plane orientations, thought to correspond to crystallographic planes. The fracture morphology (in helium) shows smooth appearance with no - o4- AK= 12 MPa /i THRESHOLD 4- AK - o = 5.2 M Pao /mF z 0 I- Q0 C- cc C) Fig. 4.lI Characteristic fatigue fracture surface produced in purified hydrogen environment near-threshold. R= 0.05 - 105- C a) E C z 0 O c a) C c C fc a- U _ .J w >, D o 0 -0. O o2- aI O - I {>, "o a O .. 4 a o I E . - C uE o I ./ o I_ a ao cj '4-- a: -d ._ . C V) C -O . E > O 0L C, L._ -106variations in fracture features in areas having diameters larger than 100 in moist m. Note however, that the patches formed air have a diameter thought to be determined of the order of 50 by the anisotropy km and are of the grains. Finally, the variation in fracture morphology caused by the presence of water vapor is shown in fig. 4.14 for the very fine structure of the heat-affected zone metal at R = 0.80. A change from dehumidified air to moist air near-threshold produces an increase in fracture roughness and faster crack growth rates as illustrated in fig. 4.14 and fig. 4.3 respectively. From fractography it is clear that the presence of water vapor in the environment produces an increased fracture surface roughness. posed that water vapor causes embrittlement. It is pro- The embrittle- ment results in increased microstructural sensitivity of the fracture path, producing rough fracture surfaces. The ob- served effect on the crack growth rates are as follows: 1 - at high load ratio the embrittlement in the presence of water vapor produces faster crack growth rates; 2 - at low load ratio two effects are present: first, embrittlement causing acceleration of crack growth rates and increased surface roughness; second, the surface roughness enhances contact of asperities raising the closure load thereby decreasing the effective stress intensity range at the crack tip. The roughness produced by the embrittlement is the main source of contact fretting resulting in enhanced - o01- a) Fig 4.14 Variation in fracture morphology produced by changing environment from dry air to moist air (heat-affected zone metal). -108.oxidation. The presence of the added oxide on the surface may add to the closure effect. 4.4.5 SUMMARY OF RESULTS 1) At low load of 0.05, the presence of water vapor in the environment decreases the crack growth rates near-threshold with respect to dehumidified environments of oxygen, hydrogen and air. The threshold results are: Environment K moist air dehumidified air 9.0 5.2 hydrogen (3 ppm water) dehumidified hydrogen 7.1 5.2 5.1 dehumidified oxygen and dehumidified environments are purity of about 2) 10 MPaVh' calculated to have a to 103 ppm of water. There is a transition in fracture morphology with environment. Moist air increases the fracture roughness with respect to dehumidified environments. Preferential oxide formation is observed at the ridges of asperities where contact may occur during closure. 3) At load ratio of 0.80 near-threshold, water vapor accelerates the crack propagation rates with respect to purified hydrogen and dehumidified air. -109At R = 0.80 near-threshold: (da/dN)moist air > (da/dN)purified hydrogen (da/dN)purified hydrogen > (da/dN)dry Environment A% MP aV--7 moist air 2.9 purified hydrogen 4) air 3.0 The observed near-threshold crack propagation be- havior in moist air at R = 0.05 relative to purified hydrogen and dehumidified air is the reverse of observations at high load ratio (R = 0.80) for SA387 steel. Near-threshold: At R = 0.05: (da/dNmoist air < (da/dN)purified hydrogen (da/dN)moist air < (da/dN)dehumidified air At R = 0.80: (da/dN) moist air > (da/dN)purified hydrogen (da/dN)moist moist air air >(da/dN) dehumidified air 5) The fatigue crack propagation rates in moist air environments at R = 0.80 are higher than in purified helium throughout the range of stress intensities tested (3 - 17 PaV-ii). At R = 0.80: (da/dN)moist air > (da/dN)hlium moist air helium -1106) The increased fracture roughness produced in the presence of water vapor occurs at low and high load ratios. Water vapor produces increased sensitivity of fracture to microstructure. The fractography shows areas of common frac- ture features of the order of grain size with cleavage-like facets indicating a reduction in ductility, that is the embrittlement effect of water vapor. 4.5 DISCUSSION The results presented in this chapter are discussed in terms of the following main guidelines: a) The significance of testing for environmental study under high load ratio conditions; b) the water vapor effect on accelerating crack growth rates; c) 4.5.1 the low R and high R results of moist - dry air. THE ROLE OF LOAD RATIO IN ENVIRONMENTAL STUDIES A significant step in obtaining the results presented in this chapter was the recognition that near-threshold fatigue crack propagation can be significantly influenced by the load ratio as shown in Chapter 3. Testing at low load ratio gives rise to the crack closure phenomena which inhibits the observation of the physical crack tip loading and environmental conditions. Calculated crack tip stress -111intensity range can substantially differ fromthat felt at the crack tip which is responsible for crack growth. Assessment of environmental effects on crack propagation at low load ratio requires careful monitoring of closure load in order to evaluate the effective stress intensity range. In oxi- dizing environments it is also necessary to evaluate the contribution of the added oxide volume to raise the closure load. In consequence environmental and crack closure effects at low load ratio can not be separated without great experimental burden. The experimental study of environmental effect on fatigue crack growth rates was done in most part at high load ratio, R = 0.80, to assure that crack closure was minimized and that the crack tip was always exposed to the environment. Previous to testing at load ratio of 0.80 it was found that the procedure for shedding the loads affects drastically the threshold value and the observed environmental effects. In fig. 4.15 is shown behavior of 2Cr-l%Mo ferent methods. the comparison of crack propagation steel at R = 0.80 obtained by two dif- The procedure found and used in this work which gives the lowest threshold value AKo, consists in initiating crack propagation at low load ratio (R = 0.05) and maintaining the maximum stress intensity constant as R is increased Kmax a= to 0.80. K/(1 - R) = constant - I vr . I 0 I I' a I1 aZ CCE wws '- CD 11111 1 I I 'o '0 '0 'o '0 l i I I NP/op '03A U!) ( '0 12- _I·r_ I 11 · -- _ I TT · l I lll . . l I -1 l. '0 · X- | _ XW l 11111 (~ 0") lll I O o CDIc t 0 0 o o _ 0O0 - 0. t o (D ) CL3 0o _ 0rn o O _ 0 oO 0N _ O o 0 c y-o 'O 0 .o Q ._ - 0 <! Q) 010 _ F-- 0o 0 C (/) o t 0o a) . ., I NLD 0 - .rIn- p2 U, ._ a- _ . ·· o CUo 0 If) I * 0 0 N o 'D 2 0 I0 N n 0 r-. N "Zi oZn- o oo U oz._ .- C- 0 I .U, Q) o >3 O c0 \ E 0 c --c U-) 0 C ui LL E, . . . . o0 ffr) Ccj , 'o 00 U) *1~ 0 a) * * *S . 0 .2 ' CCj N ,w Co 'a 0 (90 N E 0 . I... . .. . . I . . i0 C O ( al . I I....... , I I . . I . ,0 0/ WW) NP/oP '3JLV' I... ... I. I . I I 0o I I 0 ..... I I I I I To NOllVDVdOEd I I I...... I 1 '0 O6 39_liv I I I 0o , -113In the second method, after initiation of crack growth the load ratio is raised from a low value of about 0.05 to 0;.80within a small variation of K forcing Kmax and con- sequently the maximum plastic zone size to increase. In the first method the maximum plastic zone size is maintained small and constant for a wide range of K, preventing crack retardation effects by large plastic zone sizes introduced in the early steps of load shedding. Results reported previously on environmental effects at low and high load ratio for SA387 Class 2 Grade 22 steel (96,103) showed different trends difficult to reconcile. These results are shown in fig. 4.16 and fig. 4.17. Crack propagation rates in moist air are lower than in hydrogen gas for SA387 steel; while the reverse occurs for 300-M steel. The results presented in this chapter for high load ratio are in agreement with the behavior of 300-M and indicate that the high load ratio data of fig. 4.16 are affected by retarding effects of large plastic zone sizes. Comparison can be made with the high load ratio data obtained in this investigation. (Shown in fig. 3.8.) Crack growth rates at R = 0.80 are higher in moist air environment than in hydrogen gas for SA387 steel as shown in the results of this chapter, and are consistent with results obtained in 300-M steel shown in fig. 4.17 and fig. 4.18. This leads to the conclusion that moist air produces a deleterious effect on the crack propagation resistance - 114- (alA3/U! )NP/op 1 0 r- aD '0_·_·_r °o I0 _ · rl_·_ · _ I0 0 I0 I II I fill 1 I 111111 I I0 I I I"' ' I I I 1 I I I 0 X) :0 ) I cc D O =Z 00 , to 0 - C.Q )InM .0: 0 Ua r- o a 0 o U 0 I-r Lo.. -rCt) Z o. O O (%l- ()0 0 Z'o,C O D 0 o LO i - U -o o u c p IC O .[: Nc- 0 o: WVL 0 m., &.c. -- Cl .4 oa C a) E :2 LC z,OL LJ o0 O -J a 0 d o~w N 0 in) 3 .. <0 cr I (h oc I oO 0 *0 o L).-- '~ LO c aC, cr 0 41,0' uE .c En Lo 0 O -nI.) n 0 O ccO ' 00.O a It CQj4 0 0 .4 0 -JI 9V Li 0 4 4 o 0 C 0 ! QD oo U-- 0*LL - 1 a' c LL - I __ ....... '0 I I. & i i -· I II· . 'o · I. I· . I . I . . . I 111 . . I. I. o' (al3,/WW) NP/Dp '.LV . . . . . Ilitil . 0 . . I . I . . I . . Ill . . . I. . . I .......... I, I I · l· · To0o NOLV9VldOdd 13V3 3nf91Vd I -- '' L · '0 C0 - f 15 - (al3A3/U!) NP/DP 0 I I r 0 . _ . rrrlr , _ _______ o1""'1 I . 0 . r . T r . . r- I0 IO .. .... I . . II-V I -I- II . . . I1 0 r . g . . I I I 0 ..... . 111111 I I I O 593 . ..... III II c) o g- 8 0 U*) · I ..... ..... I D o o LO D u>. w ZCP C a) .2 u D D .a - °0CL n . m U C) 0 on 0lz o ro ~t o 0~r o*00 m 0 rp) 0 a ~jU 0N" zo - Z o 0 (n ?0 O aa 0 IO (D _ O 0, 0 a YC >t 10 (V a - ra) O) DO. a _o4 E 2 a .m 1%, 0C - o Q 4 Z o 0 W ~. o a at I">V uI, C: - - 5O C., E Eo -o - a) < o ~O (J c z CY, 0 C:; m(, Lu o= o 0 L~L L z I c: l cr MN Z d .0:.' U3: oPo ~ I 0,Ioo II HO: 0 0000 IC) z c 111 1 1 I1 11 I 0 4· . 0 0 >-OD K) (CH J)(d 0* LU U) 0 - 'J 1>> 0 0_ 1 . . 1, I .. bI" (alA3/ww) .. ,, .. .. 1 11II ,I 1 ,,,i t ,1 1 A I ,I i 0 I LL 111 I _ D I' I0 NP/Dp' 31V8 NOIlV9VdOdd o0 0 >3tO8 3n9l 11v. 'o I0 -116- _ 0 _ I I IL I W) 0 0f O O I.- 0> _ aS h (I) U q 0 - 0 ztr 0 Nl o & 1- 00nL O r) . 0< m4a CE N oo . . X C) w 34 0 Uc u 0 0O 0ac. R dA,c o Y ,i. ( 0 0 [ a 0 L a: z E .b I o4 o 00c wD0-0 - 00 4- 4- O g D 0Z II I·[ uI .u oo 0 0 1I N I_ _ ON Ln O (0 00 N4 (WW) HI9N31 )1:tl N: 0 -117presumably through hydrogen embrittlement. 4.5.2 WATER VAPOR EFFECT ON FATIGUE CRACK GROWTH The results of crack propagation at high load ratio showed that the presence of water vapor in the environment accelerates growth rates with respect to those characteristic of dehumidified air, helium or even gaseous hydrogen. The results show the aggressiveness of water vapor by reducing the fracture resistance and are consistent with reported results on Aluminum alloys (82) and AISI 4340 steel (85). Detailed study of the role of the water vapor partial pressure on the growth rates has indicated the existence of a pressure range where the growth rates are pressure dependent (84,95). The partial pressure effects indicate that the transport of water vapor to the crack tip or availability of water molecules could be rate-limiting steps for the environmentally enhanced crack growth process. The moist air results presented in this chapter were obtained at a relative humidity of 30 pct and partial pressure of water vapor of the order of 100 to 1000 Pa., considerably above the saturation value found in previous studies. In con- clusion, the results are assumed to be independent of water vapor partial pressure in the regime tested. A discussion of possible elemental steps in the environment-metal interaction is presented to qualitatively evaluate main parameters and -118steps controlling the kinetics of the process. Some ele- mental steps are schematically indicated in fig. 4.19 and correspond to: - transport of the aggressive environment to the crack tip - physical adsorption and surface diffusion - molecular dissociation and chemisorption - dissolution and diffusion of the embrittlement element in the metal lattice - embrittlement. The anlysis of the elemental steps and the testing conditions (sections 4.5.3 to 4.5.8), leads to the following observations: 1 - The environmental pressure used in this work assures that the transport of environment in the crack channel is by viscous flow and therefore it is not likely to become the rate-limiting step. 2 - Below a saturation pressure, physical adsorption is a potential rate-limiting step and the crack growth rates are related to testing variables as follows.: (da/dN) = C P/(fVT') C = constant; P = pressure; f = frequency; T = Temperature 3 - The environmentally-induced crack propagation in gaseous hydrogen is proposed to be rate-limited by the dissociation reaction of the molecular hydrogen on the -119- Local Stress t Fracture Crock Tip Region Transport Zone Processes I. Gas Phase Transport 2. Physical Adsorption 3. Dissociative Chemical Adsorption 4. Hydrogen Entry 5. Diffusion Fig. 4.19 Embrittlement Reaction Schematic showing some of the sequential processes involved in hydrogen embrittlement from gaseous environments -120fracture surface. The constant for the rate of dissociation is assumed as an Arrhenius function of temperatures resulting in crack growth rates of the form: (da/dN) hydroge A P exp(-E /kT) 4 - Similarly, the environmentally-induced crack propagation in water vapor-containing environments is proposed to be rate-limited by the dissociation reaction of water on the fracture surface. The crack growth rates are then of the form: (da/dN)moist oist air A2PH 2 H2 4.5.3 ep(-E2/T) 2 2 ENVIRONMENT TRANSPORT TO THE CRACK TIP At high pressure, viscous flow and ordinary gas diffusion of the environment in the crack is expected to be dominant. At low pressures when the mean free path of the gas molecules is large compared to the characteristic crack dimension, Knudsen diffusion or molecular flow is dominant. Using the molecular theory of gases the regime of gas flow is estimated by calculating the Knudsen Number or ratio of the mean free path of the molecules to the characteristic dimension of the channel: Nu = L/a (4.1) Nu = Knudsen Number L = mean free path of molecules a = characteristic dimension of the channel L = 1/( 2 d2 n ) (4.2) -121d = collision diameter n = concentration of the molecules The crack opening is taken as the characteristic dimension. The flow regimes have been experimentally delineated as follows (104): 1) L/a 0.01 : viscous flow 2) L/a >1.00 : molecular flow 3) 0.02 < L/a < 1.0 : transition range For air at 250 C the mean free path in cm is related to the pressure in microns of Hg as L = 5.09/P. Taking the maximum crack opening displacement as the characteristic dimension of the channel, it is found that crack openings larger than 6.67 m and environmental pressure over 1 atmosphere result in viscous flow, however the pressures of the order of 10 torr or below the environment supply to the crack tip proceeds by molecular flow, and transport phenomena is likely to become the rate limiting step. is consistent with the range of pressure This 0.1 to 2 torr for 5070 Aluminum alloy (82,95) where acceleration of crack growth with water pressure is observed; it is also consistent with the saturation pressure of water vapor of 2 torr above which no further crack accelerations occur with increasing pressure. The results of this work were obtained at pressures of 138 KPa (20 psi) for hydrogen tests and 101 KPa (14.7 psi) -122for moist air assuring a viscous regime of flow in the crack channel and eliminating the transport of the environment to the crack tip as a possible rate-limiting step. 4.5.4 PHYSICAL ADSORPTION It is convenient now to discuss other possible rate- limiting steps and the factors affecting them. The rate of physical molecular adsorption or rate of surface coverage can become a limiting step if it is slow compared to the rate of generation of new surface. The variables of improtance can be qualitatively analyzed based on the kinetic gas theory. The rate of adsorption is proportional collision rate on the surface. to the molecular The collision rate is ex- pressed by the following Herz-Knudsen equation (105): Z = 2.64 x 1020 P/VThfcollisions/cm P = gas pressure 2 x sec (4.3) (Pa) M = molecular weight of the gas T = temperature (K) The rate of adsorption is primarily a function of pressure and temperature J ads oLe P/V N-T' (4.4) Jads = rate of molecular adsorption The rate of generation of new surface sites can be expressed as follows: Jc i B(da/dN) f/A (4.5) -123J = rate of production of new surface sites da/dN = crack growth rate (mm/cycle) f = loading frequency A = cross-sectional area of the adsorbed molecule B = width of the specimen Critical conditions can occur when the rate of adsorption equals the rate of surface site generation. ads Js : C P/V-F Jads c = (da/dN)f (4.6) C = constant P = (da/dN)f VT (4.7) A critical pressure can be found for a given set of experimental conditions. Below the critical pressure physical adsorption is a potential rate-limiting step. From this expression is also apparent the importance of the parameter (P/f). P/f = C (da/dN) F' (4.8) Bradshaw and Wheeler (82) showed that the rate of fatigue crack growth at a given K level appeared to depend on the product of water vapor pressure and cyclic load period (1/frequency). Wei (85) has used this parameter in the in- terpretation of his results and has designated exposure. It is the time available for impingement and adsorption. In physical adsorption, the adsorbate is held to the surface by Van der Waals forces and the binding energy is of -124the order of the heat of condensation (i.e., 10 KJ/mole or less). It occurs when a molecule striking the surface is able to transfer most of its translational energy to the substrate; this process becomes less effective at high temperatures. If physical adsorption becomes a rate-limiting step, the effect of temperature on the crack growth rates can be qualitatively obtained from eq. (4.8): (da/dN)=C (P/f)/V-7 (4.9) Increasing temperatures would cause a reduction in the crack growth rates. This behavior was observed in the gaseous hydrogen environment in the temperature range of 65°C to 110°C and it is discussed 4.5.5 in chapter 5. RATE OF ADSORPTION The maximum rate of adsorption at a gas-solid interface is the collision rate of the gas molecules with the adsorbent surface as expressed by eq. (4.3). Considering water vapor (M=18 g/mole) and temperature of 2980K Z= 3.6 P x 1018 collision/cm 2 sec. (4.10) The initial rate of adsorption is limited by the number of sites per unit area. For water vapor the molecular cross sectional area is about 1.7x10-15 cm 2 about 6.1Px10 sticking A site would experience collisions per second-Pascal. coefficient of 0.5 it is possible Assuming a to estimate the time required to adsorbe a monolayer on the bare metal surface -3 t = (1/3P) x 10 sec x Pa. -125The exposure time for a monolayer to be adsorbed is then tP = 3 x 10 -4 Pa - sec. Study of the initial reaction of water vapor with Fe(001) surface (89) has shown that coverage reaches a saturation at 0.80 following an exposure of about 7xlO-4 Pa-sec at 298 K. These results are consistent with the extimated time for a monolayer adsorption shown above. Crack growth rate at a frequency of 50 Hz allows for an exposure time of 2x10-2 sec. when the crack remains open during cycling. is of the order A minimum pressure for monolayer adsorption of P = 1.67 x 10 -2 Pa = 1.25 x 10 -4 torr. Environmental pressure of this level gives rise to molecular flow along the crack and transport of mass to the crack tip could result in a rate-limiting step. The results presented in this chapter were obtained at atmospheric pressure for moist air environment and at 138KPa for hydrogen and helium environments. At these pres- sures the mass transport or physical adsorption are not likely to become rate-limiting steps. 4. 5. 6 CHEMISORPTION The differentiation between physical adsorption and chemisorption is based on the bonding strength between -126adsorbed species and the surface. Chemisorption is consider- ed to be the chemical reaction of the adsorbate molecules on the surface. The binding energy is similar in magnitude to those of primary bonds in free molecules and the heat of chemisorption is of the order of 50 - 400 KJ/mole (105). The environment-metal interaction occurs through a series of elementary ferent rates. reaction steps which proceed at dif- Often the rate of one of these reaction steps is much lower than that of the others and all preceeding andsubsequent steps are in equilibrium. The slowest step is designated as the rate-limiting step of overall environment. reaction. Analysis of the elementary reaction is helpful to study possible rate-limiting steps, to illustrate the role of other experimental variables such as pressure, temperature, etc. and to detect mechanisms of the overal phenomena. 4.5. 7 CHEMISORPTION OF HYDROGEN The complete reaction is divided into elementary reactions, step. of which acts as the rate-limiting The general equation of the reaction of gaseous hy- drogen to solution H2 the slowest (g) = 2H in the metal (sol) The possible reaction steps are: lattice is: -127a) H 2 (g) + 0 = H 2 (ads) b) H2(Ads) + O= c) H(ads) = H(sol) + 0 2H(ads) 0 = free surface site It is assumed that the physical adsorption (step a) occurs fast and that at pressures of 138 KPa it is essentially in equilibrium. The electron transfer and the atomic hydrogen solution into the metal lattice are assumed to occur very fast; the increased solubility of the highly stressed material beneath the surface near the crack tip acts as a sink of atomic hydrogen. The possible rate-limiting step is dissociation of the adsorbed molecular hydrogen, step b. The forward rate of reaction: (dw/dt)f = Kf eH(ads) 0 (4.13) = surface coverage The following analysis is made using the relation of chemical activity and surface coverage derived from statistical mechanics (106,107): ai (4.14) Kiei/E ei = fraction of surface sites occupied by species i er = fraction of vacant surface sites a = chemical activity -128Equation (4.14) is valid under two conditions: 1.0 or 8i Ol. =9 and for step a of equation (4.12) it results in: H2 (ads) a (4.15) aH2(ads) Assuming equilibrium: 415) aH2(ads) : PH 2 (g) The forward rate of dissociation becomes: 2 (dw/dt)f = KP (4.16) 2 The reverse rate of reaction: (dw/dt) = K e2H( r r (4.17) H(ads) From (4.14): 9Hads) Therefore (dw/dt) = aHsOL) 2 K aH(sol)O 2 (4.18) The net rate of reaction (dw/dt)n = (dw/dt)f-(dw/dt)r (dw/dt)n K (K PH n r 2 H2 0- a2 e2 (sol) (4.19) K = Kf/k r 9D assuming if 0 » K P 2 a 1.0, (dw/dt)n )n Kr(K Pif 2H(sol) r (4.20) (4.20) 2 (dw/dt)n = K1P H (4.21) 2 The reaction constant K 1 for therrally activated processes is an Arrhenius function of temperature K = A exp (-Ed/ T); *= Boltzman Constant. (4.22) A is related to the probability of encounter of reactants. Ed is related to the strength of the chemical bonds. The -129- net rate of reaction can be expressed as: (dw/dt)n = A PH2 exp(-Ed/1 T) (4.23) fl 4 2 When dissociation is controlling rate-limiting step, hydrogen embrittlement appears to be a function ofpressure and temperature as indicated in eq. (4.23). Experimentally it has been obtained that the crack growth rates are proprotional to the hydrogen pressure elevated to an exponent n of the order of 0.5 to 1.5 depending on the temperature (92). At room temperature n has been observed by different investigators (92) to have a value of about 1.0. The temperature dependence of crack growth rates exposed to hydrogen have been found to fol-low the form of eq. (4.23). During this work the pressure of the gaseous hydrogen was kept constant. The results of the study of the temperature effect of the crack growth rates in gaseous hydrogen are presented in Chapter 5 with further discussion on this temperature activated process. 4.5. 8 CHEMISORPTION OF WATER VAPOR The general reaction of water vapor with the metal surface is simplified to illustrate the formation of FeO: H20 + Fe = FeO + 2H (4.24) It is also assumed that the transfer of electrons is very fast being that the initial reaction takes place on clean metallic surfaces and abundance of conducting electrons permit fast electron exchange. The general reaction can be -130thought of as occurring in the following steps: = H20O(ads) a) H 2 0(g) + b) H o(ads) + O = H(ads) + OH(ads) c) OH(ads) + = O(ads) + H(ads) d) H(ads) e) O(ads) + Fe = FeO + O (4.25) = H + The physical adsorption of water vapor or, step a, is assumed to occur very fast at pressures above saturation as discussed in section 4.5.5. hydrogen The solution and diffusion of the atomic in the metal occur fast. lattice, step d, is also believed to As far as hydrogen supply for the embrittlement process is concerned it is likely that the other step that limits the process is step b, or the initial dissociation of water vapor. It is assumed in this analysis that step b is the rate-limiting step and other steps are in quasiequilibrium. Note that step e is considered in equilibrium for a monolayer oxide formation only. Multilayer chemisorption would require mass transport through the oxide film rendering step e as possible rate-limiting step, however the presence of thick oxide films (70) on fracture surface suggests the existence of an effective mechanism of clean metal surface exposure such as reverse slip during deformation. -131The resultant net rate of reaction of step b for the initial condition of 8_ (dw/dt)n = Kr(K PH 1.0 is: ao/FeO (4.26) This equation points out the effect of water vapor pressure, the activity of oxigen in the oxide and the activity or concentration of hydrogen in the metal lattice. Again the reaction constants are of the Arrhenius form. Assuming K PH O then a/FeOa (dw/dt) = A2PH ( 0 exp(-E/9 T) (4.27) 2 Further study of the temperature effect on the crack propagation is presented in Chapter 5. 4.5.9 INFLUENCE OF MOISTURE AND LOAD RATIO ON CRACK GROWTH RATES The effect of moist air and dry air on the crack propagation rates appear to be different for low and high load ratio. The results presented in figs. 4.7 and 4.8 showed that for a fixed stress intensity near-threshold the following effects are observed: > R = 0.05: (da/dN)dry air (da/dN)moist air (4.28) R = 0.80: (da/dN)dry air < (da/dN)moist air (4.29) and These observations are explained based on the following -132premises: 1 - The observed effects at high load ratio are independent of crack closure and therefore are true representation of the environmental contribution to crack growth. 2 - The observed behavior at low load ratio is affected by crack closure phenomena and the results are a combination of environment and closure effects. 3 - Crack closure is produced and enhanced by several factors such as: - the amount of plastic strain preceeding fracture or plasticity-induced crack closure - the presence of oxide product on the fracture surface or oxide-induced crack closure - the roughness of the fracture surface created by tortuous crack paths - the Mode II loading component becoming significant at low growth rates. Being that equation (4.29) is considered to represent the true effect, the observed behavior of equation (4.28) for low load ratio remains to be explained. The schematic of fig. 4.20a illustrates the cyclic stress intensity and the difference between the applied stress intensity range and the effective stress intensity range at the crack tip. The cyclic loading conditions applied to the specimen - 33- AKoap closure MOIST AIR DRY AIR (a) da/dN Assumed effective condition at the crock tip rom ck length (do/dN)dry air (do/dN) moist L-- oir KI eff. AK 2 eff. AK opp LK (b) Fig. 4.20 Schematic showing the effect of crack closure on the effective stress intensity at the crack tip (a), and on the measured fatigue crack propagation at low load ratio -134and felt at the crack tip are illustrated for dry air and moist air environment with account of only two contributions to closure and assuming that in dry air the oxide effect is negligible. Notice that at constant applied stress intensity a change of environment from moist to dry air causes an increase in the effective stress intensity at the crack tip. The schematic of da/dN vs. K is shown in fig. 4.20 for low load ratio and shows the trend of the results obtained by calculating the stress intensity K and the assumed effec- tive conditions at the crack tip. Considering a crack propagating in moist air environment at an applied stress intensity range Kapp, the growth rate is indicated as (da/dN)m.a. and the effective stress intensity would be &K 1 as shown in fig. 4.20b. By changing the environment from moist air to dry air at constant applied &K, an increase incrack growth rates is observed. The change of environment causes effective increase in the stress intensity range felt at the tip of the crack from NK 1 to K2 as indicated in figs. 4.20a and 4.20b. The observed increase in growth rates in dry air at low load ratio is a response to the increase in effective stress intensity.Inequality (3.28) occurs because the constant applied stress intensity at low load ratio does not traduce into -135constant effective stress intensity at the crack tip. Consequently it is concluded that in terms of effective stress intensities the moist air environment is suggested tor always produce higher growth rates than dry air independent of load ratio. 4.6 CONCLUSIONS The near-threshold crack propagations in SA387 steel in water vapor containing environments produces a rough surface with cleavage-like facets indicating a reduction in ductility and an increased sensitivity to microstructure, Such features are not observed in dehumidified environments of air, oxygen, hydrogen and helium. The results lead to the conclusion that hydrogen from the water-metal reaction produces embrittlement of the metal and results in a distinctive rough fracture surface. The embrittlement reduces the fatigue resistance and it is observable through acceleration of the crack growth rates at high load ratio when crack closure is minimized. At low load ratio the roughness produced by the embrittlement in moist environments enhances closure through early contact of asperities. Fretting during closure enhances oxidation which at the same time increases the closure load. The effect of closure is to increase the discrepancy between -136effective and calculated stress intensities near-threshold; as a consequence at low load ratio moist environments produce higher threshold than dehumidified environments. Analysis of some elementary reactions of the environmentmetal interaction leads to the conclusion that, at the experimental conditions of this work, the transport of environment in the crack channel corresponds to viscous flow; and that assuming fast hydrogen diffusion in the lattice, the possible rate-limiting steps are: the physical adsorption and the dissociation of the molecular speciesBoth processes are function of the partial pressure of the aggressive environment and are thermally activated. The faster crack propagation in moist environment than in gaseous molecular hydrogen leads to the conclusion that the rate-limiting steps may be different or at least have different activation energy, causing water vapor to be a better source of atomic hydrogen than molecular hydrogen at room temperature. -137CHAPTER 5: INFLUENCE OF TEMPERATURE ON THE ENVIRONMENTAL DEGRADATION OF THE FATIGUE RESISTANCE NEAR-THREHSOLD 5.1 INTRODUCTION The qualitative analysis presented in Chapter 4 showed that possible rate-limiting reaction steps are thermally activated and pressure dependent. This behavior has been confirmed with experimental results (85,92,94) and provides a very useful insight into possible rate-limiting steps. Activation energies measured for the overall environmetal effect on fatigue resistance can be assigned to the ratelimiting step on the assumptionthat the kinetics of the environmentally-induced crack growth are controlled by the slowest step in the reaction. Drastic changes in activation energies are related to changes in rate-limiting step. Note, however, that two or more reaction steps could have similar rates,and additionally, some reaction steps are not thermally activated. In this chapter, results are presented on the crack propagation variations in the temperature range of 25-100°C in environments of moist air and gaseous hydrogen. The measured apparent activation energies are consistent with values reported in the literature. In moist air environment a positive activation energy is found for temperatures near ambient temperature and above 80 C, and it is assumed to correspond to the dissociation of water molecules. However -138from 650 to 80 C a negative activation energy os observed which could reflect low physical adsorption or tendency for molecular hydrogen formation. In hydrogen gas a change of activation energy is observed at a temperature of about 600C indicating a change in rate-limiting step; the lower temperature being the dissociation of molecular hydrogen and at high temperature the physical adsorption on the metal surface. Moist air at 110°C accelerates the crack growth rates and produces lower threshold than hydrogen gas at the same temperature. 5.2 LITERATURE REVIEW The effect of hydrogen on the mechanical properties of steels have been observed since late last century. The char- acterization of the phenomena has been done extensively on hydrogen-charged steel and several theories have been put forward to explain the experimental observations. Embrittlement has been proposed to result from precipitation of molecular hydrogen at internal voids creating high pressures and enhancing their expansion (109,100). This theory does not explain variations in hydrogeninduced fracture with small variations in environmental pressure (82) and although hydrogen blistering is a common phenomena in pressure vessel steels exposed to hydrogen (100), it is not thought to be a significant mechanism in the -139hydrogen-induced fracture. Other investigators have proposed that diffusion of hydrogen to regions of high triaxial stresses enhances the hydrogen- metal interaction and lowers the cohesive strength of the lattice (111,114). This concept was later refined as the "decohesion theory" of hydrogen-induced crack propagation which postulates that sites of high triaxial stresses ahead of the crack lower the chemical potential of dissolved hydrogen allowing for higher concentrations than those attainable at equilibrium by unstressed metal. The abnormally large hydrogen accumu- lation lowers the maximum lattice cohesive force producing higher crack propagation velocities. Another early suggestion was the "stress-sorption theory" that the fracture stress is lowered by a reduction in the surface energy of the material as a direct consequence of hydrogen adsorption at the surface of internal voids or cracks (117). More recently, the effort has concentrated on: acterizing the effect of hydrogen embrittlement of 1) charthe transport mechanisms in the metal lattice by diffusion, dislocation tunneling or grain boundary short circuit paths (118), and 2) studying the role of lattice defects and microstructural features in trapping hydrogen and thereby enhancing or retarding hydrogen supply to potential sites of decohesion -140and fracture (119). A model of hydrogen embrittlement taking into account the micromechanism of lattice transport and binding energy at the traps and based on the generally recognized features that some critical concentration of hydrogen must be reached at potential crack sites for failure to initiate, postulates that (120): Hydrogen embrittlement is directly related to the characteristics of its trap population the dual role of the reversible in particular to traps as sinks or sources of hydrogen to the fracture site which depends on a) the initial state of the material (external versus internal hydrogen), b) the mode of transport through the lattice, diffusion or dislocation, and grain boundary short circuit paths. From the review of the hydrogen embrittlement theories it is clear that each one emphasizes some steps to the process, however all steps may be involved. The physical ad- sorption and chemisorption of hydrogen containing environments on the fracture surface lowers the surface free energy and contributes to lowering the fracture strength. The diffusion of hydrogen in the metal lattice involves necessarily the random walk, dislocation tunneling grain boundary short circuit paths and all of the spectrum of trapping features of the microstructure some of which may allow for recombination to molecular hydrogen. The theories and results con- centrate on the relevance of these features on the embrittlement process, essentially on the transport of atomic hydrogen -141but note that little is known of the final effect of hydrogen on the cohesive strength of the metal. Experimental results of crack propagation of steels in gaseous hydrogen environments have indicated enough similarities with results on hydrogen-charged steels that the two types of embrittlement have been considered as basically the same phenomena (92). Furthermore the deleterious effect of water and water vapor containing environments on the fatigue behavior of steel and aluminum alloys has been suggested as produced by hydrogen resulting from the water-metal reaction (81). The embrittlement is generally assumed as caused by atomic hydrogen in the lattice. The process by which atomic hydrogen is supplied from the environmental-metal interaction widely differs for environments and alloys. For a fixed metal-environment set the rate of atomic hydrogen supply can be altered by testing conditions such as pressure or temperature. Some of the early work on the effect of gaseous hydrogen environment on crack propagation was done under sustained load conditions (85,121). The results showed three stages of response of the crack growth rates against the increasing stress intensity K. In stages I and III of low and high stress intensity respectively, the crack growth rates increased with increasing K. In stage II, the crack growth rates were constant with stress intensity presumably determined by the environmental degradation of the fracture strength. -142Study of the stage II of crack growth in hydrogen gas has shown that the process is thermally activated and a function of pressure raised to an exponent to 1.5 depending on temperature (92). n of the order of 0.5 For a given hydrogen pressure the results on AISI 4130 steel showed that below room temperature the crack propagation rate increased with increasing temperature, the effect reaches a maximum at temperatures about 25-40 C and above this point raising the temperature causes a decrease in the growth rates (92). Characteristic activation energies were -16.32 KJ per mole per g-atom and +21 KJ per g-atom. Very similar results have been found with 18Ni Maraging steels (122), with an activation energy of 16.7+3 KJ per mole for the low temeprature region. However, the drop in crack growth rates with increasing temperature was more rapid than in 4130 steel for the high temperature region. Interpretation of the data was based on the assumption that the effect of gaseous hydrogen on the crack growth rates was controlled by the slowest elementary reaction step of the general hydrogen-metal interaction leading to embrittlement. At low temperatures, dissociation of the adsorbed molecular hydrogen on the metal surface was related to the rate-limiting step. At high temperature the physical adsorp- tion of molecular hydrogen precursor was suggested as the controlling step. -143The proposed rate-limiting steps have been supported by later experimental work. Studies of the adsorption of hydrogen on iron (123) have shown that adsorption decreases with increasing temperature with an activation energy of -5.9t0.6 Kj/mole. Crack propagation at high temperatures without the controlling surface reaction step involving molecular hydrogen showed increased crack growth rates with increasing temperature (94). This condition was created -using a hot tungsten filament to dissociate some of the hydrogen in the gas phase. Previous use of the same tech- nique in the study of hydrogen diffusion had shown that predissociated hydrogen adsorbed more favorably than molecular hydrogen (124). The temperature at which the change of rate-limiting step occurs (the maximum in the crack growth rate vs. 1/T cruve) has been found to shift to higher temperatures as the pressure is increased (125). This however is to some extent expected as the physical adsorption is directly proportional to the gas pressure. The effect of water and water vapor on the crack growth rates has also been fund to be thermally activated (see Table 5.1) but in contrast to the gaseous hydrogen effect, the crack growth rates increased with increasing temperature range. The results presented in this chapter are consistent -144with the observed hydrogen and water vapor effect on crack growth rates. However, these results are obtained at stress intensities far below the environmental threshold KSCC and the environmental effect is observable with low strength alloys as well as high strength alloys. EXPERIMENTAL PROCEDURE 5.3 5.3.1 MATERIAL The material used in this study was 2Cr-l%Mo as described in section 3.3. steel The specimens were selected from the same plate thickness location and the orientation is shown 5.3.2 in fig. 4.1. TEMPERATURE SETUP The study of fatigue crack propagation above room temperature required good temperature control in order to avoid confusion in interpreting variations in electrical potential across the specimen. A temperature change of 1.2°C cuases a variation in the voltage signal equivalent to a crack growth of 0.1 mmi and at very low growth rates the temperature fluctuation could cause voltage variations similar to those produced by crack growth in 106 cycles. The temperature of the specimen was continuously monitored by thermocouples spot welded to the specimen, and-theheating system consisted of copper coils surrounding the specimen and -145.loading grips. Hot silicon oil was circulated through the copper coils and the temperature of the oil was maintained constant. The assembly of specimen, grips and heating coils were enclosed in a fiberglas wool thermally-insulated box. The temperature of the specimen was maintained constant within 0.5°C and a temperature correction of the electrical potential signal was performed in the interpretation of the results. A schematic illustration of the setup used to heat the specimen 5.3.3 is shown in fig. 5.1. EXPERIMENTAL SETUP Hydrogen and helium gases were purified using the system described in section 4.3.2. The desired environment was supplied to an aluminum chamber clamped to the specimen as indicated in fig. 5.1. The chamber-specimen set was vacuum tight to a level of 10 3 torr using rubber gaskets. Wet hydrogen and helium were obtained by bubbling the gas through distilled water provided in a bypass line. Tests in hydrogen and helium gases were performed at 138KPa and the air tests were conducted at atmospheric pressure (3040% relative humidity). 5.3.4 TESTING PROCEDURE The characteristic variation of electrical potential with temperature was determined for the notched specimen. - / ~cn - (. 0 0) V =4) C4 o4In vr, a) 0 I I () I I a) ) I 0., 00 C Cd U) a) ar.l E O1 01 0 U, -) .H V, C-) Cd /7 o C C r- 0 (CD a) a) E C C: -D Ca0 o 6 c o 0 (.) Cr L. E 0 E L 0 , LOr C aE 0 o C, a) (Ac o a) 4-) 0 c 0 U) 44 4-) -147For a current of 30 A and an initial voltage of 2 mv applied to the specimen, the voltage increased linearly with temperature in a proportion of 4 mv per degree centigrade. This relation was confirmed at other crack lengths and in raising and lowering the temperature. Threhsold tests at different temperatures were obtained using one specimen per temperature. The influence of temp- erature on fatigue crack growth rates at fixed &K was obtained on a single specimen. A test at 110 C was performed to study the relative ef- fects of moist air, purified hydrogen and helium, and wet hydrogen and helium on the crack growth rates. This test was initiated in moist air at load ratio of 0.05, the loads were shedded to obtain the stress intensity threshold of moist air, Ko = 7.05 MPaVm'. Once the threshold in moist air was determined, the stress intensity was raised to AK = 10 MPa-m' about 2.5x10 mm/cycle. to obtain growth rates of The test was temporarily stopped by reducing the load amplitude to zero and the environmental supply line was evacuated and then baked, and finally was flushed with purified hydrogen. Following the change of environment, cycling was continued. used for each change of environment. The same procedure was -1485.4 EXPERIMENTAL RESULTS 5.4.1 TEMPERATURE DEPENDENCE ENVIRONMENTALLY-INDUCED NEAR-THRESHOLD FATIGUE RESISTANCE The temperature range used in this study was 25-110 C. The aim was to thermally-activate environmental-metal reactions without altering the mechanical response of the metal due to temperature. Crack growth dependence on temperature within a small range can be assigned to the environmental interaction under the assumption that the deformation mechanism remains unchanged by temperature. A series of tests were performed in moist air environmenta at load ratio fo 0.05. results for temperature 5.2. The fatigue crack propagation for 28, 60, 100°C are shown in fig. The temperature effect is noticeable only for growth rates below 10-5 mm/cycle in the near-threshold regime. This is consistent with the observation made in previous chapters that the environmental effects in the mid-growth any, fall within the scatter band of the data. regime, if It also re- inforces the assumption that in the temperature range tested, only the environmental interaction is activated. The re- sults show that crack growth rates increase with increasing temperature from 28 C to 110°C; consequently higher temperatures give rise to lower stress intensity threshold. The results in therms of threshold values are as follows: - l/4- (ali(A/U!)NP/DP 0r- '0 - I I 0) In I0 If.I. O '1 I o0 I 1 1111 I III 0 bO II ll I I I 111 I I 11111 000 I a) O 0 -00- .p U) 0.lj. 0r0 o0 Ot .I _ 0, 0N' 0? Noc 00 0 e. OW 0 li o- (W U) 0 $::-- <] O ao oI0 cv . *- - C- Ii (O - V o U) ,_ 0 U 0 _- Fll -Ji W C Ccui I I N0 Nc 0--0 otOf I 0) 0 OnN 0 &n~Q C) IT OD@ CD * E (I Ld C'j en W O o 0 cCj EOO L PcNca :..') -o 0)o~ (U( (/·WI i 1 I, '0o CC 0) -0 O 'o (alOO/u.u y, ,,,,,,, .... 1 '7 I --- ) NP/DP '31BV 1 , ,, , 1 ,,,,,, o N1OV9VdOd o ,, 1 I... ,, o X3:V13 31n91VJ 4LI ` L~ CO I0 *0 -150Environment: Temperature Moist Air SK MPaVTi' (C) 28 9.0 60 8.0 110 7.1 Similar sets of tests were performed in purified hydrogen of 138KPa and temperatures at a pressure and 1100 C. hydrogen of 28, 54, 6, The results are shown in fig. 5.3. In gaseous it was found that for the temperatures of 29, 54 and 65°C the crack growth rates increased with temperature but the results at 110 C showed a substantial increase in stress intensity threshold AKo and lower crack growth rates than those of the other temperatures investigated. The results in terms of threshold value are: Environment: Temperature (oC) Gaseous Hydrogen SK IP aV-iT 29 5.5 54 5.1 65 4.8 110 8.3 The threshold hK in gaseous hydrogen at 110 C is higher than the threhsold values found in moist air at 60 and 110°C. The temperature effect diminishes as crack growth rates increase in a fashion similar to that observed in moist air environment. The influence of temperature on the fatigue -/5/- (al33/U! )NP/Dp 0r'- 'n Io I oC I0 TI I I I T T I I 1 1 1 I I I0 0 Y I I I l I '0 11 I I I I I I 0 I Hii 'O I r N ON C0 0 I @ 0to 0 I0 ITI T I I I',- O Ic 0 . - 00h - 0 0IT3 0 00 14) 0IN O S c 0u O . ° 0 O0) 0 HI 0 - Z % CD -o F- In (D F 0 Z _ a, N qD a ko L) rc_ IT C N N o o0 U) > a, 11.0 LLJ -U) 'O~ a Ia, >'2c" mw rr) _) _ 0- f'0 C C ' .. m 03 ~ll3 , IN o Cu, tD - el ll3 1, CJ O C 0 wD '4, Vn lJ t, &r N .... ... . . . .... __L__ | _ _ . I. I I I L 0o (al3/D/Wu) NP/Dp '31V8 1ul ,i ,I1 ,,I I , I ,lI I I _ 'o NOIILV9VdOd . ID I . .. I | iI , . ,o 3XV3 3l91.V_-i . .l A ru 'O -152crack propagation behavior in gaseous hydrogen was further investigated and confirmed in two other separate tests, one carried out at 110 C to contrast the relative effect of environments of moist air, purified hydrogen, wet hydrogen, helium, and wet helium; the other was to investigate the effect of temperature on the growth rates at a constant stress intensity AK. 5.4.2 ENVIRONMENTAL EFFECTS 01? CRACK GROWTH AT 1100 C This test was designed with two goals; the first, to contrast the effect of moist air and hydrogen environments on the fatigue crack propagation, and the second, to investigate the role of the water vapor presence in helium and hydrogen gases. The results of the first part are shown in fig. 5.4 in terms of crack propagation rate vs. K. The test was in- itiated and run down to thresholdin moist air environment. After obtaining a threshold in moist air of 7.05 the loads were increased to a .K of 10MPaiF' PaViim' to obtain a growth rate of 2.5 x 10-7 mm/cycle in moist air. At this point, the cycling load was brought to zero and the environment was changed to purified hydrogen. The crack growth rates obtained in hydrogen were about 8 x 10- 8 mm/cycle or about one third of those in moist air. In terms of crack growth rates, moist air environment was found to be more -/53 - (al:)A/!) NP/oP 0 I_ I__ '0 ~ __ Itlill I0 111111_ _. I · Ir D On _I I '0 I 111111~ '0 I I b Ij ,I ) _l Ill 0 I (w, O (O ro U00 00 _ in CL O 0 "T LoO O 0 .L 0 O _ rn oCY a 0u, . C o, 0c\j _ . , 0 0o 2 in OE _QC I 2 0 I in- zU O) cn0 E % 0gso a f *- - z L.. ._ O0 I- z §'e (D X oL. 0U Lr In -J LLi cJ o o o cuJI O c o ;3 0 , C., po N A, I',J .. a, E-o; >, E 2: (n LL tw 2 I...................... 1 I I II I 'O u_Ec o InO 0 0 -I 0 N 0 n 11111 'o I I Illll1 ' ' ' I_ To (a33/uJwu) Np/op , _ I1,,,,1 " I II 10 I ' I' I I I I o I I I I I 1,,. ......IJ I I I .1 ! ] II '0o '"31. V NOI.VOVdOdd >DV4D 3nfl9vd II J CD tv '010 LI- -154aggressive than gaseous hydrogen at 110 C. value obtained in hydrogen was 8.6 MPafi' 7.05 MPaVim- ' in moist air. The threshold compared to The difference in crack growth rates decreases for higher crack growth rates and it would appear that the two environments would converge to yield similar growth rates at higher K (above 20 MPa m ). In chapter 4 the strong influence of water vapor in crack growth behavior was shown. Water vapor was singled out as the most active component of moist air affecting growth rates near threshold. In this test a comparison of growth rates in purified vs. wet environment was obtained at constant stress intensity K at load ratio of 0.05. The results are shown in!fig. 5.5 for hydrogen and helium gases at stress intensities of 8.6 MPaV-m- and 14.7 MPaVmhi respectively. The crack propagation rates in hydrogen and helium bubbled through distilled water were clearly increased over the propagation rates obtained in the corresponding purified gases. These results show the effective action of water vapor to produce higher velocity of crack advance. For the case of hydrogen environment, the increase in growth rates by introducing water vapor is equivalent to those of purified hydrogen if the 3M4PaV K were increased by -'. Note also that the test was performed at R = 0.05 where a strong role of crack closure phenomena is expected. - 55 - a,) o"0 cnJ cn n 00 rq 0 rr Q cO NO a)O C 0 o co E C- mVW o o .LU _ < O o 03 . NU O > a)O O c a,) o v ww H19N31 )>3V03 - _ 0 LLJC lvI ~Q) -156Addition of water vapor to the purified hydrogen produced increased roughness and oxidation of the fracture surface, and at the same time the crack growth rates were increased. The embrittlement part of the environmental effect appears to completely dominate the closure effects. 5.4.3 APPARENT ACTIVATION ENERGY OF THE ENVIRONMENTAL DEGRADATION OF FATIGUE RESISTANCE The influence of temperature on crack growth rates in gaseous hydrogen and moist air was obtained by testing at constant stress intensity range K at various temperatures within the range of 100°C. The results obtained in purified hydrogen at R = 0.05 are presented in an Arrhenius plot in fig. 5.6. The crack growth rates in hydrogen increase with temperature in the range of room temperature to about 65°C, consistent with the results shown in fig. 5.3. Above 65°C the growth rate de- creases with increasing temperature, confirming the observed behavior in hydrogen at 110°C of high threshold and low growth rates. The crack propagation rates in hydrogen appear to be thermally-activated.. With the assumption that deformation mechanisms remain unchanged in the temperature range tested, the environmental contribution to crack growth would be the thermally activated process and therefore the results - /57 - TEMPERATURE (C) 90 1 10 I z Z I 50 70 I I I I 30 I I h~~~~~~~~~ -6 I x 10 10 -7 7 xlO - 15.5 KJ/rr r_ z0 -7 5x 10 0 ao Hydrogen Gas Environmen A K:7.0 MPao -F- Q_ -7 R = 0.05 3x1O U 2.5 II I II I II I I I 2.6 2.7 2.8 2.9 3.0 3.1 3.2 3.3 I T Fig. ( °K ) oK 5.6: Apparent activation energies of the fatigue crack propagation rates in hydrogen gas environment. -158of fig. 5.6 are assigned to the effect of hydrogen on growth The activation energies rates at different temperatures. estimated from fig. 5.6 are 26.8 KJ/mole for the low temperature range and -15.5 KJ/mole for the high temperature range. The test in moist air was performed at load ratio of 0.80 in order to separate the environemntal effect from the crack closure phenomena, specifically to separate the crack accelerating effect of moist air from the oxide-induced crack closure. to 93 C. The temperature range tested was from 54 The results are shown in an Arrhenius plot in fig. 5.7 indicating thermally-activated growth rates as the net effect reflecting the thermally-activated rate-limiting step of the moist air-metal reaction. The test specimen was used up confirming a trend not reported in the literature of decreasing growth rates with increasing temperatures moist air environment. alternatively in the range of 65 to 80 C in The data points were obtained by stepping up and down in temperature starting from the high temperature. Below 650 C and above 80°C the crack propagation rates increase with increasing temperature. The change in behavior at about 650C is in close agreement with the results obtained in gaseous hydrogen presented in fig. 5.6; however, basic difference is observed at temperatures above 800C a -TEMPER159ATURE TEMPERATURE 100 90 _ -7 '3. io U N U w u I I 80 70 I I 60 __ 50 40 I I I ___ 30 _ I 1 E E z 2.10 51.5 KJ/mole v w1.0 I- -34.2 32 -7 Cr Moist Air Environment AK = 3.2 MFPafiF R= 0.80 : LL r- 2.6 I 27 I 2.8 I I 2.9 3.0 TI- x 103 I 3.1 I I 3.2 3.3 I OK Fig. 5.7: Apparent activation energies of the fatigue crack propagation rates in moist air environment. -160where increasing temperature causes the growth rates to increase in moist air but to decrease in gaseous hydrogen. The activation energies estimated from the graphs are 51.5 KJ/mole and 31.8 KJ/mole for the low and high ends of the temperature range and -34.2 KJ/mole for the intermedrange of 65-800C. iate 5.4.4 SUMMARY OF RESULTS The variations of threshold results of SA387 steel 1) in the temperature range of 28-110°C and load ratio of 0.05 are: Moist Air (30% R.H.) Temperature C (K PaVim' Gaseous Hydrogen (138 KPa) Temperature C (iPaVm'i 28 9.0 29 5.5 60 8.0 54 5.1 110 7.1 65 4.8 110 8.3 2) At room temperature and R = 0.05 moist air environ- ment produces higher threshold than purified hydrogen, however at a temperature of 110°C the opposite results are observed. 3) At 110 C and R = 0.05 wet hydrogen produces higher crack growth rates than pure hydrogen. A similar effect is observed for helium; the effect being pronounced -161near-threhsold. 4) The apparent activation energies of crack growth rates near-threhsold are: Environment Load Ratio Temperature Range Hydrogen gas 0.05 C 28-65 65-110 Moist air 5.5 0.80 Activation Energy KJ/mole 26.8 -15.5 28-65 51.5 65-80 -35.8 80-93 31.8 DISCUSSION The effect of water vapor at room temperature on the near-threshold crack propagation rates has been observed to have the following characteristics: 1 - It produces higher threshold than inert environment or vacuum at R = 0.05. 2 - It increases the growth rates compared to hydrogen gas and dehumidified air at R = 0.80. The low load ratio behavior has been interpreted using crack closure arguments and the oxide-induced crack closure model (70). The high load ratio effect was not known with this material and leads to the mechanism of hydrogen embrittlement in the presence of moist air proposed in Chapter 4. -162At low load ratio the moist air or water vapor containing environments give rise to two mechanisms of opposite effect on the crack growth rates: embrittlement by hydrogen from the dissociation reaction of water causes accelerated crack growth rates; crack closure enhanced by roughness and oxidation of the fracture surface produces apparent lower growth rates and higher threhsold than inert environments. The first mechanism is experimentally supported by the results shown in Chapter 4 obtained at high load ratio where no closure effects are present. The moist air results at low load ratio (R = 0.05) presented in fig. 5.2 showed increasing growth rates with temperature indicating that the embrittlement process is thermally-activated and dominates the retarding effects of crack closure. The corrosion process is also thought to be thermally-activated which would make the oxide-induced crack closure appear thermally-activated; however this effect seems to be very small compared to crack accelerating contribution of hydrogen embrittlement. The results of hydrogen environment presented in fig. 5.3 illustrate two important points: 1) The magnitude of environmental effects on crack growth rates and threshold value K; o. -1632) the temperature activation of environmental inter- action. The first point indicates that change of the aggressiveness of environment with temperature can cause increase in threshold. The increase in threshold (or decrease in crack propagation rates) by raising the temperature 650 C to 110 C in hydrogen is not related to closure from phenom- ena which otherwise would have been observed in fig. 5.2. Consequently caution must be taken with frequently used closure arguments to explain variations in threshold at low load ratio. The difference of hydrogen and moist air effects with temperature indicate that the activated process is the environmental interaction as opposed to deformation mechanisms. The results shown in fig. 5.4 are discussed based on the proposed modeling concept (129) of considering the crack growth rates in deleterious environemnts as composed of the sum of: - crack growth in inert reference environment - environmentally induced crack growth - stress corrosion cracking (da/dN)e = (da/dN)r + (da/dN)cf + (da/dN)scc (5.1) Based on the analysis presented in Chapter 4 it is assumed that the environmentally induced crack growth is surface reaction controlled and that the reaction constant -164is thermally activated. contribution A proposed expression for this is of the form: (da/dN) = (da/dN)cfs (1 - exp(KcPo/2f)) (5.2) P0 = gas pressure in the external environment K c = reaction rate constant f = frequency (da/dN)cf = maximum environmental contribution to growth It is proposed here that the maximum environmentallyenhanced crack growth, (da/dN)cf s is thermally activated of the form (da/dN)cf A (da/dN) s= Ai i exp(-E./RT) (5.3) Where Ei is activation energy of the rate limiting step of the reaction, and Ai is a function of pressure of the environment. The exact form of eq. 5.3 depends on the rate limiting step of the overall reaction. At conditions of saturation and assuming negligible stress corrosion cracking: (da/dN)e = (da/dN)r + Ai exp(-E/RT) (5.4) The form of eq. 5.4 has the following implications that agree with experimental observations: 1 - The increased environmental effect near-threshold results from the fact that the crack growth rates in the reference inert environment become of the order of magnitude -165of the environmentally-induced growth rates. This is il- lustrated using the results presented in fig. 5.4 and considering the relative effect of moist air and hydrogen to remain constant and at saturation. The following esti- mation can be made. i At AK = 10 MPaV-m (da/dN) (da/dN) = 2.5 x 10- (mm/cycle) hydrogen = 8 x 10 -7 (mm/cycle) 1.7 x 10 Relative environmental effect: At (mm/cycle) K = 18 MPaV-m' (da/dN)ir (da/dN) Relative = 1.0 x 10 hydrogen environmental -5 (mm/cycle) = 8.1 x 10 effect -6 (mm/cycle) = 1.9 x 10 -6 (nun/cycle). Even though the environmental effect on the fatigue crack growth remains about the same, its relative effect compared to crack growth induced by the stress condition can become very small in the mid-growth regime and therefore falls within the scatter band of the experimental data. This argument explains the results that crack growth rates in the mid-growth regime appear to be insensitive to environmental effects. 2 - Near-threhsold the crack propagation rates are thermally-activated through the environmental contribution to crack growth. By studying the crack propagation variation -166with temperature, it is possible to gain some understanding of the rate-limiting step of the general reaction and observe changes in these rate-limiting steps with temperature. The decrease in growth rates in hydrogen at 110 C compared to 65°C or room temperature is interpreted as caused by a change in the rate limiting step of the environmental reaction. Moist air or water vapor presence at 110 C is more aggressive than hydrogen gas as shown in fig. 5.4 and fig. 5.5. The variation in temperature alone reverses the relative aggressiveness of moist air and hydrogen at low load ratio. At room temperature hydrogen gas produces higher growth rates than moist air. At about 110 C, the rate limiting step of the embrittlement process appears to be different for the two environments with a positive activation energy for moist air and a negative activation energy for gaseous hydrogen at about 1100 C. The results presented that effectively the sign in fig. 5.6 and in 5.7 indicate of the overall activation energy is different for moist air and hydrogen at temperatures above 80C; however, below this temperature the general trend is similar. here is a breaking-point at about 65°C where the effect of temperature on growth rates reverses sign. This is interpreted as a change in rate limiting step of the environmental reaction. The estimated overall -167activation energies from fig. 5.6 and fig. 5.7 are: Temperature range °C Estimated activation energy Ei KJ/mole Gaseous Hydrogen 26.8 28-65 65-110 -15.5 Moist Air 55-65 51.5 65-80 -34.3 80-110 31.8 These results are compared to results of other investigators summarized in Table 5.1. For hydrogen environment the positive and negative behavior of the apparent activation energy is consistent with results reported in the literature. At low temperatures the apparent activation energy obtained by other investigators range from 16.3 to 28.5 KJ/mole in agreement with the results obtained in this work of 26.8 KJ/mole. These activation energies are often compared with activation energies obtained by studying the single processes such as adsorption, diffusion, etc., in order to identify rate limiting steps. Some results are shown in Table 5.2: -168a) u C) e) a) a, co r- N Ccj -3- a- : O >0)8 > -) ) E - u m N -T 00 O O rn 0 Ln 4-) 3 c C Q~) a) a P: E a, H) a) 0a) C C c 0 O U) 4-) 3 3 Q~'~ O U) U) 0r- CZ a-- mr oo ,~ Lr U) r r 0 S^ a) o0 0---a)Un X n o0 oO O 0 X) X 0 (' -7 a) MH a · r( 'Il0 I0 - X.) N - U) cd 0 I 0 X 4) X N -7 0) 3 n) .1- E 3U) 4-) co U) E O HrH U) co O O O N -7Z a) C CO H - cO O UCO O I o O I O C 0) C L. a: cxd oc 3 a) L 3 O E '-0 1~~ rn +1 C 4-) U) il a, :3 LL, 0 rl E- I a) -A-) Cd :3 "X o -L L. -T I I 0 'I0 0 - OX X.) -7 Lr, 0 I I 0 0 XJ.) X N -I a) C - X Q) 00 a) CO Cd bO L LS 3tm c) O LcN L ~0 a) C a) CO 4--) 4-) 0 u 01 L O - > E1 a =t a0 C -169- _~j CI, _3 Z O^ C C oN > C, 4I a) S -4-l bO - aC -- Q) o c'J O ic _u {U bO - I IC 4- o U a u(D~ (cD~ Io "o a " =3:- V) 0Iaz . I 3 I ~~.O_ r./ C\J0 I [5 0 cO ,- -' ~H H*O3 o ¢ o XE-X x 0 XX O rI- CZx 3o I O X L *OL 3 la0 0 O r) (D ~ bOO 0 (DO -J .O- (~b O X a( bO 0 -170Table 5.2 Process Activation Energy KT/mole Reference - Hydrogen permeation in steels (139-141) 34-38 - Physical adsorption of hydrogen on steel -5.9 (123) 21.3 (137) 27.6 (138) 27.2 (141) 7.5 (141) - Diffusion of hydrogen in A-Ti - Apparent heat of solution of hydrogen - Lattice diffusion of hydrogen Comparison of the results obtained in this work with those shown in Table 5.2 suggests that the rate limitation step for embrittlement process in the low temperature range is the dissociation and solution of hydrogen in the metal lattice near the surface and for the high temperature, the physical adsorption of the gaseous environment. This con- clusion is consistent with the viewpoints of other investigators (92,94,125). The temperature effect on growth rates in moist air can be compared with results obtained from surface reaction studies of water vapor with iron (85,89); the activation enerty reported is 14.7 + 4.3 KJ/mole. The low temperature -171activation energy of 51.5 KJ/mole compares closely to the results obtained in distilled water of 49.8 KJ/mole and suggests that at low temperature the adsorption of water vapor is fast enough to form a multilayer of adsorbed molecules on the metal surface. It is thought that after the adsorption of a few monolayers the metal surface can not distinguish between water and water vapor and the reactions are those of water-metal surface. As the temperature in- creases from 65°C to 800C adsorption decreases and the stability of the adsorbed molecules is also reduced. The de- crease in growth rates with increasing temperature from 60 to 800C is assumed to be caused by a reduction of water molecule surface coverage or an increased tendency for atomic hydrogen recombination or both. For higher temperatures than 80°C, the activation of the water molecule dissociation step is sufficient to supply enough hydrogen for the embrittlement process. Therefore moist air environment - metal reaction at 110°C results in a more efficient source of atomic hydrogen than the molecular hydrogen - metal reaction. 5.6 CONCLUSIONS 1) At low load ratio the moist air or water vapor- containing environment gives rise to two mechanisms of opposite effects on the crack growth rates near-threshold: -1721 - hydrogen embrittlement 2 - crack closure; both are thought to be thermally activated. The results showed increased growth rates with temperature, leading to the conclusion that at high temperatures hydrogen embrittledominates the retarding effect of crack closure, that is the activation energy of the embrittlement process is higher than that of oxide formation for the temperature range tested. 2) Assuming that crack closure in purified hydrogen environment remains constant within the temperature range of 28-1100 C, it is possible to postulate that hydrogen embrittlement is present and is thermally activated. 3) Interpretation of results of crack growth rates near-threshold in moist air and hydrogen gas at low load ratios must include the embrittlement effect of both environments and the roughness and oxide-induced crack closure of moist air. Quantification of these effects was outside of the scope of this work but it is recommended for future research. 4) Using Wei's concept of modeling crack growth rates in deleterious environments as a summation of effects, the corrosion fatigue or environmentally-enhanced fatigue crack growth rate can be qualitatively analyzed: (da/dN)e = (da/dN)r + (da/dN)cf + (da/dN)sc (5.1) (da/dN)cf (5.2) (da/dN)cf,s ( -exp(KcPo/2f)) -173based on the observed temperature-activated behavior of environmentally-induced crack growth rates and on the analysis of the environmental and testing conditions presented in Chapter 4 which indicated conditions of saturation. It is proposed that the corrosion fatigue component of crack growth rates at saturation (da/dN)cf,s is of the form: = Ai exp( -Ei/RT) (5.3) Ei = activation energy of the rate-limiting step of the environmental reaction A i = parameter function of pressure 5) Comparison of the apparent activation energies of crack growth rates at R = 0.05 in gaseous hydrogen with results of studies on single processes lead to the conclusion that the rate-limiting steps may be: at low temper-. ature (below 65°C) the dissociation and solution of hydrogen in the metal lattice and at high temperature the physical adsorption of molecualr hydrogen. It is assumed here that the crack closure level in purified hydrogen remains constant with the temperature range tested. 6) Comparison of the apparent activation energy of crack growth rates at R = 0.80 in moist air from room temperature to 65 C (51.5 KJ/mole) to the activation energy crack growth in distilled water (49.8 KJ/mole) suggests that the same mechanisms may be involved, indicating the high coverage of water vapor on the fracture surface and of -174multilayer adsorption. -175CHAPTER 6: FATIGUE CRACK PROPAGATION IN VACUUM AND INERT ENVIRONMENTS 6.1 INTRODUCTION The characterization of fatigue crack propagation in aggressive environments calls for comparison with base data obtained in vacuum or inert environments such as purified argon or helium. of 10- 6 Testing in vacuum of the order torr brings out the transport process or availabil- ity of aggressive environment as a rate-limiting step of the environmentally-induced crack growth. Moreover, the low coverage of the newly created fracture surface may allow rewelding by adhesion of the metal surfaces upon contact. Inert environments on the other hand are suscep- tible to an impurity content which could cause partial pressures of the aggressive elements to reach critical values for aggressive effect as discussed in Cahpter 4 and experimentally observed in various metals (97,99). In this chapter the results of the fatigue behavior in vacuum are presented for load ratio of 0.05 and 0.80 and are compared to results in purified helium and moist air. The results at 0.80 where closure effects are negligible are compared for moist air, helium and vacuum at room temperature. In vacuum a threshold partial pressure for environment influence is surely obtained but at low load ratio the aggravation of closure or rewelding makes it difficult to -176discern the true environmental contribution, Again it is worthwhile to point out the need to evaluate the environmental effects at high load ratio where closure effects are negligible. The results of tests performed in vacuum at load ratio of 0.05 and 0.80 are presented. A comparison of the relative environmental effect of moist air, helium and vacuum is presented 6.2 at R = 0.80. LITERATURE REVIEW Vacuum environment has been frequenlty used as a ref- erence inert environment in the study of fatigue resistance of metals. The general observed effect is that it increases the fatigue resistance. The beneficial effect of vacuum on fatigue life was reported as early as 1932, by evacuating -3 the testing chamber to a level of 0.5-1x1O torr (79); thereafter many studies have assessed the fatigue life and crack propagation resistance of engineering alloys in high vacuum (105 to 108 is described in reference torr). A review of the early works (130). The general aim has been to compare the fatigue behavior in vacuum with those in ambient air and modified environments in order to isolate the aggressive agents and understand their relative effect on the fatigue resistance. -177The study of low fatigue crack growth rates in Ti-6A1-4V showed 130) that in vacuum at R = 0.35 the fatigue growth rates are microstructurally insensitive and lower than those in moist air by a factor of 2 to 3. in load ratio Furthermore, a variation of 0.12, 0.35 and 0.61 produce very similar fatigue behavior down to growth rates of 10 mm/cycle. Similar investigation on medium carbon, En24 steel (133) showed that the fatigue crack propagation in vacuum was independent of the load ratio even to near-threshold level; in the mid-growth regime the growth rates were increased by a factor of 30 in the presence of moist air. Results on Ni-Cr-V steels (132) in vacuum have indicated similar behavior of reduction in load ratio sensitivity near-threshold. A common observation of these results is that the threshold in vacuum at high load ratio increases from that of moist air by a factor of about 2 and the threshold at R = 0.05 in vacuum is about the same or slightly higher than that obtained in moist air. Research effort has also concentrated on measuring the closure level in vacuum environments. Closure levels have been measured using the electrical potential technique (133,134) and the compliance method (135). Measurements of closure loads with the compliance method and the electrical potential have been found to give similar results in moist air environment (43). In vacuum environment electrical -178potential measurements give higher apparent closure loads than those obtained by the compliance method (134). This is due perhaps to electric conduction through contacting spots which do not necessarily transfer significant loads. Independent of the measuring technique, closure levels are found to be higher in vacuum than in moist air environments (135). Note however that in vacuum, closure at low load ratio can be assisted by the adhesion and rewelding of the contacting surfaces. In moist air on the other hand, closure is enhenced by surface roughness and oxide formed near the crack tip, consequently a relative comparison of the closure levels in these two environments can not be conclusive. A result not clearly explained is the vacuum effect on the threshold value at high load ratio: vacuum increases threshold with respect to moist air, argon or helium environments by a factor of about 2. At high load ratio, however, closure and rewelding are considered negligible compared to those at low load ratio. 6.3 6.3.1 EXPERIMENTAL PROCEDURE MATERIALS The material used in this study was SA387 Class 2 Grade 22 pressure vessel steel described in Section 3.3.1. The specimens were chosen from indicated in fig. 4.1. thickness of the plate as -1796.3.2 VACUUM CHAMBER A Centorr controlled atmosphere vacuum chamber was used and the vacuum was 2 x 106 torr. The chamber can be mounted and fixed to the columns of the Instron testing machine and after proper alignment the pulling rods are connected to the ram and crosshead of the loading fixture. The pulling rods are attached to the vacuum chamber through "O" rings sealed bellows. A diffusion pump backed up by a mechanical roughing pump provide the necessary vacuum. Electrical connections for crack measuring were fed into the chamber through vacuum sealed ports. Vacuum loads were measured by evacuating the chamber with the testing machine in stroke control and were taken into account in the data reduction process. 6.4 EXPERIMENTAL RESULTS 6.4.1 CRACK PROPAGATION IN VACUUM ENVIRONMENT The results in vacuum environment of the fatigue crack propagation behavior in 2Cr-l%Mo steel are shown in fig. 6.1 for load ratio of 0.05 and 0.80. Crack growth rates are lower for low load ratio in the range tested; however, it would appear that at higher crack growth rates there is no significant influence of load ratio on the rates. threshold stress intensities were 6.9 MPaVT2 and 5.1 MPaV hF for R = 0.80. The for R = 0.05 This is the closest range threshold values obtained for the two load ratios in the of -1so0NP/DP (aIDu3/u!) 00- I ·I r '0 _I·r_··· I0 _···_·1 · 1111I I- CD 111 11 ji I · I0 ______ l- · _··__r _ · I 1 ll' I t I I0 10 _ _________ I 1 I 11111 I C 'n CD 10 I I I 0 N (W 0 E 0o 0 IIrI oO C a o 0(D , E !M OL 0 oI_ 1 10 0_ 0c0 . 0r 0rn) C: O o 0o 0CJ o_ 13. U6 I0 o w i <) 0 I a) .D F- Z C Q) zZ 0 U) CM (f) I- 1O i i ~~L~ _) O- I y ,, _ 'o o, .,: i Y , O3 )< ) - 03 C Z Ld LLJ 0 F U) 0 o,) C, (. r l CW W o 0aO i. b I - 0 0 2 - 0N, N . T V. cuO x CC o LL il >:: CM LIIII N O I1ll to III .............. I I I111 t0 (alsA/ww) NP/°p I I ,..-...-. I, Il 'I, ' to ,Ii I . 1,. . .I.,.i. . To . . . . ' I ,,.' ' .' , ,o '31Vd NOIV9VdOt d NODVE3 3n91vl ' I ' L ' r~~l _ - i0 L. -181environments tested in this investigation. It is consistent with the observation that crack growth rates in vacuum tend to be insensitive to load ratio effects. Comparison of the crack propagation behavior in vacuum with those obtained in other environments are aimed to stress the environmental effects at low load ratio. 6.4.2 CRACK PROPAGATION IN MOIST AIR AND VACUUM ENVIRONMENTS The comparison of crack propagation behaviors in vacuum and in moist air environments is of interest given that moist air enhances embrittlement with load ratio. Moist air was found to yield lower threshold than helium or gaseous hydrogen at R = 0.80. This is caused by an embrittlement effect product of the water-metal reaction. Furthermore, at load ratio of 0.05 moist air gives rise to the highest threshold value of the above mentioned environments due to the oxide enhanced crack closure. The results of crack propagation as a function of stress intensity K for vacuum and moist air are presented in fig. 6.2 and indicate the following behavior: 1 - In vacuum the effect of load ratio on threshold is drastically reduced from that observed in moist air environment. The threshold results are: -182- (ala0/u!) 0 II II II O _ O '0 '0 'o .1 - 1 .1 I""" I rT I' .- I I..... . NP/DP '0 'O . I ' ' r I"" '0 --· ---- ' ' 0 '0 I""" III I Tr I - 0 o nA 0- O o o C rc t0 , _ 0 E o O Q 0 0 0Idj _ S o O t _% 0rr> _ 0 i A 0cJ o 0 o o O Id . *_ o a) (I NI -,- 'I db lk Zi Y- h I : . '- · 0 CD 0 X_r tO 0 -J (D .° ) C IE z qb % %" Z Io '0 LO % 0 o _ -W a -~h ao o; o o M. ., c'j li LLJ H -o co 0 I uu 3- 0 O o Ia o 0 oC O~ In ) >^ -Q u a0 ' C cr- IV >0 6- E UL cr o °E N V)u w 0o .. 0 E T - CNI II 0O 0O o C 0 C o07 E ('JI 0 Oh0 PC- 0 ton UY.-L- I I, U Q- o ,, I I, I1f,,, 1, ,,11 1 I ww) NP/P N0 , (alO/b)/Ww) NP/DP '3LV8 NOIV9VdOd I1111 Ml~ 1, I I I-, > 31 o >IDVEI 3D. , 1, I L Vo _-! 'O0 -183Environment Threshold AK (MPaVi') R = 0.05 R = 0.80 Moist air 9.0 2.9 Vacuum 6.9 5.1 the ratio of threshold values at R = 0.05 and R = 0.80 is reduced from 3.10 in moist air to 1.38 in vacuum. 2 - At high load ratio vacuum environment increases the threshold value with respect to moist air. observed at low load ratio: The opposite is vacuum decreases the threshold value with respect to moist air. 3 - At high crack growth rates ( 10-5 mm/cycle), moist air environment produces higher growth rates than vacuum in the entire range tested. 4 - There are crossover points of the moist air low load ratio results with those in vacuum showing that below a critical stress intensity the moist air environment appears to reduce the crack growth rates with respect to the results obtained in vacuum. 6.4.3 COMPARISON OF CRACK PROPAGATION IN VACUUM WITH OTHER ENVIRONMENTS AT R = 0.80 At high load ratio a small amplitude cyclic load is superimposed on a high strain static mean load that maintains the crack open during cycling. Crack closure is nonexistent or negligible and therefore crack propagation variations with environment represent the environmental effects only. -184A comparison of the results in vacuum environment with the high load ratio results in moist air, hydrogen and helium environments is shown in fig. 6.3. The following observations are important: 1 - In vacuum environment tained at 5.1 MPaVTmF at R = 0.80 a threshold is ob- comparatively higher than the threhsold obtained in moist air, helium, or hydrogen of about 3.0 0.1 MPa/mF . 2 - Moist air and hydrogen gas environments produce consistently higher crack growth rates than those obtained in vacuum or helium in the range of stress intensity tested. 3 - At crack growth rates above 10-5 mm/cycle LK 12 MPa--i), ( or a the fatigue behavior in helium environment is essentially the same as that observed in vacuum environment. It is also in this range that the load ratio effect is very small. 4 - The most striking observation is the large difference in fatigue behavior near-threshold resulting from helium and vacuum environments. It is assumed that under these conditions no closure or environmental effects are present and the interpretation of these results may lie in the deformation mechanisms as will be discussed later. 6.4.4 COMPARISON OF CRACK PROPAGATION IN VACUUM WITH OTHER ENVIRONMENTS AT R = 0.05 The fatigue crack propagation behavior at low positive load ratios near-threshold has been observed to be affected - 18S (alA33/u!) 0fl- 1I1I '0 I 0 -r111771 1 1 111111 I I NP/DP 111111 I I.,. 'O 11·11·1 · I I IlI I I' '0 · · ·- 1·1·- · 1. I1 I II 'O I ' I11 l 0i a *_. 0 n O o00 o E I t 0 _ (D 0 '0 ca 0 1 CQ I >1 oo _00-i X 0.. _ - _ n Q :3 0 IT \a O coO 0OJ _ 0 0 a)0 'O C Z E o -o IN a 0(5 _l. ii In V., 0 %. I lx, cr- o E c C C' C _ 'C) 44 0 _ o·oo* - CD -J LLJ N LI !-- o O ~~~oo o .. U IL~~I(c wB 0 c,- o Q a a I. LD U) U) cU: ) ,, D 0) 1I U) (D 0 o N 0 0 O bX _ 0L 0>, w E -c rI c n a C) c) C p 0w > C 0,-o to N O & C) E O '. C E, OE, J E Cf> 0o e QC O 0 0: 0 '0 em .2 ·- CJ ......I I 1 I I 1· A I I I I . ... i ,I i1111 , I lI II ,,,I I 1 ,10 o ,0 N0 0 (al3/ uw) NP/DP'31V NOIlJ/9D:OC:d )>Vt3 1,. . I ............. i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3n 3NlI1~-! 0 -186by the crack closure phenomena. Crack closure and load transfer at discrete points in the wake of the crack before cycling reaches the minimum load produces a reduced ef- fective stress intensity at the crack tip. The overall ef- fect is a contribution of the plastic strain to fracture, the fracture surface roughness determined by the microstructure and environment, the oxide thickness formed on the crack surface and the Mode II loading activated near threshold. In vacuum environment crack closure at low laod ratio is also thought to be affected by rewelding due to adhesion of bare metal surface. Rewelding would increase the load at which the crack opens but would not affect the closure load. The overall effect of reweldingon crack closure or fatigue behavior is not clear. Comparison of results obtained in vacuum with those obtained in moist air, dehumidified air, oxygen and hydrogen at low load ratio is shown in fig. 6.4. The following obser- vations are of particular importance: 1 - Vacuum results at R = 0.05 give higher threshold and lower crack growth rates than dehumidified environments of air, oxigen and hydrogen, however the effect decreases at higher stress intensities indicating that dehumidified environments produce crack growth rates in-between those of vacuum and moist air. -197- 0p,0 __ '0 '0 11111 I I (al,3/U!) '0 ,T liff1I NP/D 'o TrI 0' I0 '0 'o0 TTT Tl · i I l i l 0 l 0 t u - aD 0 ¢C Q 00 CL - 0 0 oo Co o, 0) ocro S0~ 0 0 A6 0 rt) on 0Co I> a 0 0 o r (I< z z af H (r C E 0 Ow Uf) F-LJ o *oE I W OnZ G co -O. oc ro 0Cn o 0 Wi n- tn rc-o oQ 0 ._ .Don C' 0 Oi a, d · oo 0 rr zC- d"C~ In C~ 0 Lrl 4 0oa N Li o. I' C) or o N C CI C)O o06) N1I U) - c C). (3 w >. a L U) LL o 0 "If Q 2 C7 C. o ._ o O L 0 o4a, o 11 -C E 0 V .S S?V-TC N 0 _o~X.. C) C\j E, r o, E E-pE -ro V, E E E oc D m - V .c ar' o- O L -a o.)D > E *CVV, >E C > V O\ o ....... Z C0r.- ,L 1 ,1 I I - C1 1I 1 , I I I I,, l_ ·......i I I I . 1o (al pAO/ wU) NP/DP '31'V NOiV9VdOdd :)V:D 3nDitii o r ,o - I'0 ,I I o r . . . . . . . . . . . . . . . . . . . . . . . . I . ........... C>J '0 OD -1882 - The presence of water vapor has two opposite effects on the crack growth rates at R = 0.05 when compared to other environments: i) at high crack growth rates (or high K), moist air produces higher crack growth rates than vacuum or dehumidified environments. These are however crossover points below which the second effect is apparent; ii) near threshold the crack growth rates in moist air are reduced below those of dehumidified environments and vacuum environment. Moist air gives rise to the highest threshold of the compared environments and would appear to be more beneficial than oxygen, hydrogen and vacuum! 3 - It is also noted that the crack propagation behavior in vacuum at R = 0.80 follows quite closely the trend observed 6.4.5 in dehumidified environments at R = 0.05. FRACTOGRAPHY The characteristic fracture morphology produced in vacuum is shown load ratios in fig. 6.5 and fig. 6.6 for low and high respectively (R = 0.05, R = 0.80). Essentially the fracture surface near-threshold appears very smooth and there is not evidence of microstructurally-determined fracture features (grain size/ 40 pm). At low load ratio (R = 0.05) the fracture surface shows areas where the surface appears flattened as if caused by compressive loads during closure. However, the lower threshold in vacuum than in - 1813- AK= 12 M Pa v/ THRESHOLD AKo= 6.9 MPao / zZ O Q 0 CC Q. o cr U VACUUM R=O.05 Fig. .5 Characteristic fatigue fracture morphology produced in vacuum environment near-threshold at R=0.05 . - - qO- AK= 14 M Pa /' -- ' THRESHOLD AKo=5.1 MPaoJ/ z 0 Cr U VACUUM R= 0.80 Fig. 6b Characteristic fatigue fracture morphology produced in vacuum near- threshold at R=0.80. -191moist air (R = 0.05) is proposed to be the result of a smooth fracture surface, and the reduction of roughness and oxide-induced closure. 6.5 CONCLUSIONS Fatigue crack propagation in vacuum environment pro- duces smooth fracture surface with no apparent microstructural influence. The threshold at low load ratio decreases from that of moist air because closure effects are reduced by eliminating the microstructurally-sensitive, rough fracture surface. At high load ratio, threshold increases and growth rates decrease due to embrittlement; consequently there is a large degree of plasticity. In vacuum a reverse motion of dislocation in slip planes ahead of the crack tip can occur. Dislocations originate at the fracture surface, move and pile up against obstacles during loading; during unloading the back stress of the pile-ups force dislocation back to the free surface until the friction stress of the material is reached. This process reduces "accumulated fatigue damage" resulting in reduced growth rates. -192- REFERENCES 1 Press, Syracuse, 1964, pp. 107 - 132. N.Y., 2 H.H. Johnson and P.C. Paris, Journal Mechanics , vol. 1, No. 3, 1968, p. 3 . 3 R. Con. "Proceedings P. Wei, Syracuse Approach", Interdiciplinary "Fatigue - An Paris, P.C. University of Engineering Fracture Aspects of Stress on the Fundamental National R. W. Staehle, ed., NACE-1, Cracking", Corrosion Association of Corrosion Engineers, Houston, TX, 1969, pp. 104-112. 4 R. Review , Nos. 5 and 6, Metals International Ritchie, O. 1979, p. 205. 5 ASME.D), 6 1963, 85, p. F. 8 11 N. Walker and C.J. Vol. Structures, 1-14. Beevers, 1, 1979, pp. Fatigue of Engineering Materials and 139-148. Fracture S. Purushothaman and J.K Tien, Proc., Conf. "Strength of Metals and ICSMA5, Pergamon Press, New York, Vol. 2, 1979, p. 1267. C.J. Beevers, Metals Science , Vol. 14, 1980, p. 418. P.C. Paris., American 16 of Envrionment and Complex Load History on STP 462, 1970, pp. Y. Nakai, K. Tanaka and T. Nakanishi, Engineering Mechanics, Vol. 15, No. 3-4, 1981, pp. 291-302. Alloys", 15 Fracture Acta Metallurgica , Vol. 11, 1963, p. 703. P.J.E. Forsyth, 14 of 7, 1975, pp. 429-439. 10 13 of Basic A. Otsuka, K. Mori and T. Miyata, Engineering Fracture Mechanics, Vol. 12 Journal Journal International Ratwani, M. ASTM. Life," Engle, R.M. 6, 1970, p. 379. Walker, "Effects E.K. Fatigue 9 and Vol. and ASME.89 D) 1967, p. 459. (Trans. Erdogan Mechanics, (Trans. 528 Kearney V.E. Forman, R.G. Engineering 7 Erdogan, Journal of Basic Engineerinq Paris and F. P.C. R.O. Ritchie, Materials R.J. Bucci, E.T. Wessel, W.G. Clark, and T.R. Mager, Society for Testing S. Suresh, and Technology and Materials, STP 513, 1972, p. 141. and C.M. Moss, Journal of Engineering , vol. 102, 1980, p. 293. -19317 18 P.C. Paris, R.J. Bucci, E.T. Wessel, W.G. Clark and T.R. Mager, American Society for Testing and Materials, STP 513, 1979, p. 106. R.O. Ritchie, Journal of Engineering Materials Technology , (Trans. ASME, H), Vol. 19 20 R.J. Cooke, P.E. Irving, G.S. Booth and C.J. Fracture 7, 1975, p. 69. P.E. Mechanics , Vol. Irving 1974, p. 21 99, 1977, p. 195. and C.J. Beevers, Metallurgical Beevers, Engineering Transactions , Vol. 5, 391. W. Elber, American Society for Testing and Materials, STP 486, 1971, p. 230. 22 R.A. Schmidt and P.C. Materials, 23 Paris, American Society for Testing and STP 536, 1973, p. 79-94. R.O. Ritchie, in Fatigue hhresholds, proc. 1st. Intl. conf.,Stockholm, 3acklund, A. Blam and C. J. Beevers, eds., EMASPubl. Ltd. U.K. 1981. J. 24 25 R.O. Ritchie, R. Fuquen-Molano, Annual Report No. 2 to Fossil Energy Division, D.O.E., Oct. 15, 1980. (MIT Fatigue and Plasticity Laboratory Report No. FPL/R/80/1035). R.A. Schmidt and P.C. Materials, 26 W.H. 1976, p. 29 Bamford and D.M. J. Barson, E.J. 1970, p. 301. T.C. Vol. 30 Lindley, 43, Imhof, C.E. 1976, p. Moon, "Corrosion 79", Prod. NACE Meeting, NACE. and S.T. Richard, Rolfe, and R.O. Eng. Fract. Mech. , Vol. 2, Ritchie, Metall. Met. Form , 268. R.O. Ritchie, Journal of Engineering Materials Technology , (Trans. ASME, H), Vol. Ritchie, 99, p. 195. Metals Science , vol. 31 R.O. 32 N.E. Frost, Proceedings, 1959, pp. 811-836. 33 Testing and 154. March 1979; Houston, TX., 28 American Society for J. Mautz and V. Weiss, American Society for Testing and Materials, STP 601, 27 Paris, STP 536, 1973, p. 79. N.E. pp. Frost, 15-22. Journal 11, 1977, p. 368. Institution of Mechanical Engineers, of Mechanical Engineering 173, Science , 5, 1963, -194of Mechanical Engineering 34 N.E. Frost, Journal pp. 203-210. 35 S. Usami, and S. Shida, Fatigue of Structures , vol. 1, 1979, pp. 471-482. 36 Pearson, S. pp. 235-247. 37 N.E. American Dowling, ASTM-STP 38 Engineering Fracture Society Science , 6, 1964, Engineering , Mechanics for Testing Materials vol. 7, and and 1975, Materials, 637, 1978. Second Proceedings of the and S. Takahashi, H. Kitagawa International Conference on Mechanical Behavior o Materia s, Boston, 1976, pp. 627-631. 39 40 Y. Nakai and K. Tanaka, Proceedings of 23rd Japan Congress on Materials Research, 1980. R.A. Smith, International Journal of Fracture , vol. 13, 1977, RCR, pp. 717-720. 41 42 M.H. El Haddad, T.J. Topper and K.N. Smith, Engineering Fracture Mechanics , vol. 11, 1979, pp. 573-584. K. Tanaka, 43 Nakai and M. Yamashita, Y. , Vol. Fracture International Journal of 17, No. 5, 1981, pp. 519-533. S. Taira, K. Tanaka, M. Hoshina, American Society for Testing and Materials, STP 675, 1979, pp. 35-162. 44 C.F. Meitzner, Welding Research Council Bulletin , 1975, No. 211. 45 A.G. Vinckier and A.W. Pense, Welding Research Council Bulletin 1974, No. 197. 46 K.E. Jones, Conf. Proceedings on "Welding of Creep Steels", Welding Institute, 1970, p. 66. Resistance 47 J.M. Brear and B.L. King, Proceedings of Conference Boundaries, Metals Society, London, 1976 on Grain 48C.J. McMahon, Jr., R.J. Dobbs and D.H. Gentner, Engineering , vol. 37, 1979, pp. 179-186. 49 C.A. vol. 50 Hippsley, J.F. 28, 1980, pp. Materials Science and Knott and B.C. Edwards, Acta Metallurgica, 869-885. L.A. James, Welding Journal, 1973, p. 173-S. Research Supplement , Vol. 52, No. 4, -19 5and R.M. Pelloux, 51 S.M. El-Soudani pp. 144-152. 52 C. Chen, A .W. Thompson and I.M. Bernstein, Transactions A , Vol. 11A, 1980, pp. 1723-1730. 53 M. Parry, N. Nordberg and R.W. Hertzberg, Welding Journal Research Supplement , vol. 51, No. 10, 1972, p. 485-S. 54 D.J. Gooch and pp. 405-411. 55 P.E. Bretz, B.L. Braglia and R.W. Hertzberg, Proc., 5th Bolton Landing Conf. on "Weldments: Physical Metallurgy and Failure E.F. Nippes, H.D. Solomon; Phenomena", eds., R.J. Christoffel, G.E. Co., 1979, pp. 271-283. 56 R.W. Armstrong, 57 Y. Nakai, K. Tanaka, and T. Nakanishi, Engineering Mechanics , vol. 15, No. 3-4, 1981, pp. 291-302. 58 A.W. Thompson and W.A. Backofen, Acta Metallurgica , Vol. 19, 1971, King, B.L. Metallurgical Welding Journal Metals , May 1975, Metallurgical Technology Transactions , vol. , Nov. 1979, 1, 1970, p. 1169. Fracture pp. 597-606. 59 R.O. Ritchie, International Metals Reviews , Nos. 5 and 6, 1979, pp. 205-230. 60 R.K. Priddle, Scripta Metallurgica , vol. 12, 1978, p. 49. 61 J.L. Robinson and C.J. 1973, p. Beevers, Metals Science Journal , Vol. 7,, 153. C.J. 63 E. Hornbogen and K.H. Zum Gahr, Acta Metallurgica pp. 581-592. 64 J. Masounave and J.P. Bailon, Proc. 2nd International Conf. on "Mechanical Behavior of Materials", Boston, 1976, pp. 636-641. 65 K. Minakawa, Y. Matsuo and J. MlcEvily, Metallurgical vol. 13A, 1982, pp. 439-445. 66 W. Elber, Engineering Fracture Mechanics , vol. 2, 1970, pp. 37-46. 67' Beevers, Metal Science , 1980, p. 418. 62 B. BudianskY, Vol. J.W. Hutchinson, 45, 1978, pp. 267-276. Journal , vol. 24, 1976, Transactions of Applied , Mechanics , -196Third International Conference on 68 T.C. Lindley and C.E. Richards, Fracture, Munich, 1973. 69 G. Marci, P.F. Packman, International Journal of Fracture , vol. 16, 1980, pp. 133-153. and R.O. Ritchie, Metallurgical 70 G.F. Zaminski, S. Suresh, Transactions , vol. 12A, 1981. 71 K. Minakawa and A.J. pp. 633-636. 72 D.E. McCabe and J.D. Landes, "Design Properties of Steels for Coal Conversion Vessels," Final Report to EPRI, Research Project 627-1, McEvily, Scripta Metallurgica , vol. 15, 1981, March 1980. 73 G.H. Aronson, and R.O. Ritchie, Journal of Testing and Evaluation, vol. 7, 1979. 74 Du Comite Pour L'Etude Du Fluage Des Metaux Aux "Travaux Burxelles, January, 1962. (Work Temperatures Elevees", I.R.S.I.A., of the committee studying Creep of Metals and Elevated Temperatures; published by the Research in Industry Institute for the Encouragement of Scientific and Agriculture.) Welding Journal, 75 W.F. Savage, E.F. Nippes and E.S. Szekeres, Research Supplement , vol. 55, 1976, p. 260-S. 76 W.A. Baeslack, J.C. Lippold, and W.F. Savage, Welding Journal, Research Supplement , vol. 58, 1979, p.168-S. 77 A. Gyster and G. G. Terlinde, L. Lindigkeit, Metallurgica , vol. 27, 1979, pp. 1717-1726. 78 C.M. Moss, S.M. thesis, M.I.T., 79 H.J. Gough, and D.G. Sopwith, Journal, 1932, p. Lutjering, Acta Sept. 1979. Institute of Metals , vol. 49, 93. 80 N.J. Wadsworth, Internal Stresses and Fatique in Metals , (eds., Rasweiler and Grube), p. 382 (Elseuier, Amsterdam, 1959). 81 T. Broom and A.J. Nicholson, Journal Institute of Metals vol. 89, 1961, p. 183. Bradshaw and C. Wheeler, Applied Materials Research , 1966, 82 F.J. 83 G.G. Hancock and H.H. Johnson, Transactions Society of AIME , vol. 236, 1966, pp. 513-516. of the Metallurgical -19784 R.P. Wei, P.S. Pao, R.G. Hart, T.W. Weir and G.W. Simmons, Metallurgical Transactions , vol. 11A, 1980, pp. 151-158. Transactions, 85 G.W. Simmons, P.S. Pas, and R.P. Wei, Metallurgical vol. 9A, 1978, pp. 1147-1158. 86 A.R.C. Westwood, "Strengthening Mechanisms in Metals and Ceramics, Proc. of the 12th Sagamore Army Research Conference (ed., J.J. Burke, N.L. Reed, V. Weiss), 1966, pp. 407-429. 87 J.H. Westbrook, Environment-Sensitive and Breach, NY, 1966. 88 Environment-Sensitive H. Nichols and W. Rostoker, Behavior , Gordon and Breach, NY, 1966. 89 D.J. Dwyer and G.W. Simmons, Surface pp. 617-632. 90 P. Smith and A.T. Stewart, Metal Science , July 1979, pp. 429-435. 91 R.P. Wei, American Society for Testing and Materials, STP 675, 1979, pp. 816-840. 92 D.P. Willaims and H.G. 1970, pp. 63-68. 93 D.P. Williams and H.G. Nelson, Metallurgical 1972, pp. 2107-2113. Transactions 94 H.G. Nelson, D.P. Williams, and A.S. Transactions , vol. 2, 1971, pp. 953-959. Tetelman, 95 F.J. Bradshaw and C. Wheeler, International Mechanics , vol. 5, 1969, pp. 255-268. 96 C.M. Moss, M.S. Thesis, Sept. 1979, M.I.T. 97 G. F. Zamiski, M. S. Thesis, Aug. 1980 M. I. T. 98 S. Suresh, Ph.D. Thesis, Sept. 1981, M.I.T. 99 J. Toplosky, 100 P. Liaw, S J Hudak Jr.,J . Keith Donald, 14th National Symposimui Fracture Mechanics, 1981. 101 Mechanical Behavior , Gordon Science , vol.&64, Nelson, Metallurgical Transactions Journal 1977, , vol. 18,1970, p. 753. 1, , vol. 3A, Metallurgical of Fracture M.S. Thesis, June 1981, M.I.T. W. Wriedt and R. A. Oriani, Acta Metallurgica, vol. Mechanical 102 -198R.P. Skelton and J.R. Haigh, Materials Science Engineering , vol. 36, 1978, p. 17. 103 J. Toplosky and R.O. Ritchie, Scripta Metallurgica ,vol.15, pp 905- 908, 1981. 104 S. Dushman, Foundations of Vacuum Technique , Second Scientific Edition, J.M. Lafferty, ed., p. 88, Wiley, 1982. 105 F.C. Chemisorption of Gases on Metals , Academic Press, Tomkins, 1978, p. 20. 106 107 W.L. Worrell, and E. T. Turdogan, Trans. AIME, vol.242,1968, p.1673. in Heterogeneous Kinetics at L.S. Darken and E.T. Turkdogan, (G.R. Belton and W.L. Worrel, eds.) Penum Elevated Temperatures Press, New York, 1970, pp. 25-99. 108 S. Suresh, C.M. Moss, and R.O. Ritchie, Transactions Japan Institute of Metals , vol. 21, 1980, p. 481. , vol. 145, 1941, 110 A.S. Tetelman, "Fundamental Aspects of Stress Corrosion Cracking," National Association of Corrosion Engineers, Houston, 1969, p. 464. 111 R.P. Frohmberg, W.J. Barnett, , vol. 47, 1955, pp. 892-925. H.H. Johnson, TMS-AIME 113 E.A. 114 115 Morlet, and A.R. Troiano, Transactions , vol. F.W. Scholler, and A.R. Troiano, Transactions 215, 1959, pp. 1048-1052. A.R. Troiano, Transactions ASM , vol. 52, 1960, pp. 54-80. R.A. p. 116 ASM and A.R. Troiano, Transactions , vol. 212, 1958, pp. 528-536. Stigerwald, TMS-AIME J.G. Sims, Transactions and C. Zapffe pp. 225-237. 112 C. AIME 109 Oriani, Bunsen-Gesellschaft Phys. Chem. , vol. 76, 1972, 848. R.A. pp. Oriani and P.H. Josephic, Acta Metallurgica , vol. 22, 1974, 1065-1074. 117 N.O. Petch and P. Stables, Nature , vol. 169, 1952, pp. 842-843. 118 A.L. West and M.R. Louthan, Jr., 10A, 1979, pp. 119 G.M. Pressouyre 89- 100 . Metaliurgical Transactions , Vol. 1675-1682. and .M. Bernstein, Acta Metallurgica , Vol. 27, p. -199120 G.M. Pressouyre, Acta Metallurgica , vol. 28, 1980, pp. 895-911. 121 J.D. Landes and R.P. Wei, International Journal of Fracture , Vol.49, 1973, pp. 277-293. 122 S.J. Hudak, Jr. and R.P. Wei, Metallurgical Transactions , Vol. 7A, 1976, pp. 235-241. 123 C.M. Ransom and P.J. Ficalora, Metallurgical Transactions , vol. 11A, 1980, pp. 801-807. 124 J.D. 125 R.P. Fast, Philips Technical Review , vol. 7, 1942, p. 74. Gangloff 1977, pp. 126 J.T. Vol. 127 128 and R.P. Wei, Metallurgical Transactions , Vol. 8A, 1043-1053. Ryder and J.P. 92, 1970, pip. Gallagher, Journal of Basic Engineering , 121-125. W.A. Van Der Sluys, Engineering Fracture Mechanics , Vol. 1, 1969, pp. 447-462. H.H. Johnston and A.M. Willner, Applied Materials Research , Vol. 4, 1965, p. 34. 129 R.P. Vol. Wei and G.W. Simmons, International Journal of Fracture, 17, No. 2, 1981, pp. 235-247. 130 P.E. Irwing and C.J. Beevers, Metallurgical Transactions , vol. 5, 1974, pp. 391-398. 131 R.J. Cooke, P.E. Irwing, G.S. Booth and C.J. Beevers, Engineering Fracture Mechanics , vol. 7, 1975, pp. 69-75. 132 A.T. 133 P.E. Irwing, J.L. Robinson, and C.J. Beevers, Engineering Fracture Mechanics , Vol. 7, 1975, pp. 619-630. 134 V. Bachmann and D. Munz, Engineering Fracture Mechanics , vol. 11, 1979, pp. 61-71. 135 J. Petit and C. Gasc, Fatigue of Engineering lM..C. Lafarie-Frenot, Materials and Structures , Vol. 1, 1979, pp. 431-438. 136 T. Scott, "Materials For Coal Liquefaction," Proceedings of Fourth Annual Conference on Materials for Coal Conversion and Utitlization, Oct. 1979, U.S. Department of Energy, Washington, Gaithersburg, D.C., 1979, pp. KI-K59. Stewart, Engineering Fracture Mechanics , Vol. 13, 1980. -200137 138 W. Lochman, "Materials Design Requirements for Coal conversion", Proceedings of Fourh Annual Conference on Materials for Coal conversion and Utilization, Gaithersburg, Oct. 1979, U.S. Department of Energy, Washington, D.C., 1979, pp. 1-31. R.J. Wasilewski and G.L. Kehl, Metallurgia , Vol. 50, 1954, pp. 225-230. 139 W. Beck J., O.M. Bockris, J. McBreen, and L. Nanis, Royal Society (London), Vol. A290, 1966, p. 220. 140 D.I. Phalen and D.A. Vaughan, Proceedings of Corrosion , NACE, Vol. 24, 1963, No. 8, p. 243. 141 Stein, "Gas-Phase Hydrogen Permeation Nelson and J.E. H.G. Through Alpha Iron, 4130 Steel, and 304 Stainless Steel from Less than 100 C to Near 600'C", NASA Technical Note, NASATND-7265, Ames Research Center, Moffett Field, CA., April 1973. 142 S.J. Hudak, Jr., 1976, pp. 235-241. and R.P. Wei, Metallurgical Transactions , vol. 7A,