Chapter Eleven Corrosion Fatigue

Home , Corrosion fatigue, Fatigue (material)

S.A. Shipilov Department of Mechanical and Manufacturing Engineering, University of Calgary, Canada

Abstract

Corrosion fatigue is generally recognized as an important phenomenon that can lead to unexpected cracking behaviour and failure of structures under certain conditions. Such conditions depend on the specific combination of material, cyclic loading and environment of concern that in turn represent the metallurgical, mechanical and physicochemical (electrochemical in aqueous solutions) components of the corrosion-fatigue problem, respectively. Since the earliest research made during the First World War, significant progress has been made, especially in the last four decades, in understanding the corrosion-fatigue phenomenon. However, despite the progress made, researchers are still far from solving many problems related to corrosion fatigue. At this time, no effective method for preventing corrosion-fatigue failures has been identified, and it is not possible to predict which combinations of material and environment will result in intensive corrosion fatigue under in-service conditions. The theory of corrosion fatigue is far from comprehensive. Little is known about the basic mechanisms of corrosion-fatigue crack propagation, and even less is known about the mechanisms by which the environment can accelerate such crack growth. The purpose of this paper is to outline the present state of knowledge and current controversies concerning the phenomenon of corrosion-fatigue crack growth and its mechanisms.

1 Introduction

The subject of corrosion fatigue of metallic engineering materials has attracted considerable attention from researchers throughout the world beginning in the mid-1910s when the first research on record was carried out [1]. This means that

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) doi:10.2495/1-85312-836-8/11 330 Advances in Fatigue, Fracture and Damage Assessment the history of corrosion fatigue research is a part of the history of engineering of the 20th and 21st centuries. The problem may have had humble beginning in observations made of breakages in the Royal Navy’s paravane towing cables during the First World War [1] but today corrosion fatigue is acknowledged as one of the most important areas of engineering research. Since the early 1960s, corrosion fatigue, along with stress-corrosion cracking (SCC), has been responsible for many, if not most, service failures in a wide variety of industries [2]. The economic and humanitarian aspects of corrosion-fatigue failures have led to considerable scientific and engineering efforts directed at the understanding and prevention of such failures. Since the earliest research [1,3– 8], significant progress has been made, especially in the last four decades, in understanding corrosion-fatigue phenomena. Nevertheless, despite the efforts that have been exerted and the progress that has been made, researchers are still far from solving many problems related to this mode of materials failure. At this time, no effective method for preventing corrosion-fatigue failures has been identified, and it is not possible to predict which combinations of material and environment will result in intensive corrosion fatigue under in-service conditions. The theory of corrosion fatigue is far from comprehensive. Up to now, no firm consensus exists on the crack-growth mechanism in any one material–environment combination, a concern that is central to answering questions related to the possibility, mode and rate of corrosion-fatigue cracking. Even less is known about the mechanisms by which the environment can accelerate such crack growth. In this paper, some areas of corrosion fatigue—including corrosion-fatigue crack growth (FCG)—in which the author has had both a direct interest and conducted research, are discussed. The author is fully conscious of many deficiencies of this discussion. First, because of the complexity of the phenomenon, a full and reliable interpretation might not be possible in a single chapter. Secondly, only a limited number of references could be quoted, and as such some important aspects of the problem could not be discussed deeply enough. Thirdly, because of the large number of variables influencing corrosion FCG, the generalizations presented here may not have universal validity. The first part of the chapter entitled “Principles of corrosion-fatigue testing and research” is devoted to considering: the terminology used in corrosion-fatigue- related research and publications; the early history of corrosion-fatigue research from 1917 when the first research [1] was carried out up to 1947 when Evans and his colleagues [9,10] summarized the work on the electrochemistry of corrosion fatigue carried out up to that time; and the fracture-mechanics technology in corrosion-fatigue research. The purpose of the second part of the chapter entitled “Mechanisms for corrosion-fatigue crack propagation” is to briefly review and evaluate experimental results and mechanistic models for corrosion FCG. This review outlines the present state of knowledge and current controversies concerning corrosion FCG mechanisms. Since this chapter is primarily concerned with the acceleration of corrosion-fatigue cracks in aqueous solutions, milder environments such as dry gases and air will not be discussed in detail. Information on SCC needed to clarify our understanding of corrosion

FCG behaviour in metals is given in some sections but should not be considered as complete and/or exhaustive.

2 Principles of corrosion-fatigue testing and research

2.1 Corrosion-fatigue-related terminology

“Corrosion fatigue” is a term used to describe the phenomenon of the gradual accumulation of damage (cracking and eventual fracture) in material under the combined interactions of external chemical environment and fluctuating (or cyclic) stress with its microstructure, where the environment does not have to be corrosive [2,11]. In this definition, the word “combined” should be emphasized since neither cyclic stress in inert environments nor aggressive environmental attack applied separately produces the same damaging results as their conjoint action. That is, a pre-corroded specimen does not necessarily show appreciable reduction in fatigue life (for example, when it is tested in mild environments such as dry gases and air), nor does pre-fatiguing in an inert/mild environment visibly increase the corrosion rate of metals. Also, it should be emphasized that most textbooks and handbooks, including those that were recently published [12,13] (in contrast to the above definition), require that the external environment that causes this type of failure be corrosive (deleterious) or aggressive. That is, according to the early published definitions [12,13], the term “corrosion fatigue” conjures up the notion of the severe disintegration of the material through chemical attack, accompanied by fatigue-crack initiation and growth. In reality, however, a relatively innocuous environment such as, for example, ambient air and atmospheric moisture can greatly accelerate the cracking process without producing visible corrosion in the commonly accepted sense. This fact is too well known [11] to require detailed discussion, but in order to avoid any misunderstanding related to the terms “corrosion fatigue,” “SCC” and “fatigue” that are used in this chapter, a little necessary information should be given. According to Ref. [2], SCC as a phenomenon is a cracking process caused by the conjoint action of a corrosive environment and nominally static or slowly increasing tensile stress. The tensile stress may be either applied service stress or residual stress resulting from casting, welding or other fabrication processes. [Note: Usually technical journals use SCC as a general term to classify all modes of materials fracture due to environmental (usually aqueous solutions of various compositions) factors.] Also, SCC can be defined as a special case of corrosion FCG in a metal exposed to a specific corrosive environment at the stress ratio (R = minimum/maximum load) equal to 1 [14,15]. Although pure SCC can be obtained in a laboratory when a specimen is tested under sustained or monotonically increasing stress, it cannot be realized under realistic operating conditions because most real engineering structures are exposed to service stresses of varying amplitudes that usually are a mixture of stochastic and deterministic components [16]. Despite this obvious fact, much effort and

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 332 Advances in Fatigue, Fracture and Damage Assessment material expenditures have been directed at studying SCC and much less is known about corrosion-fatigue crack initiation and propagation. In 1984, Speidel [17] summarized environments causing SCC in engineering alloys. Table 1, prepared by him, shows a variety of these environments. Most of them are aqueous solutions of various compositions that represent a wide spectrum of modelling and, sometimes, technological environments. Our concern here is an “inert” environment such as pure (distilled) water. According to the table, pure water causes SCC in certain alloys of iron (carbon steels and austenitic stainless steels), aluminium, copper, nickel, magnesium, titanium or uranium and possibly others. That is, pure water is known as a SCC agent, particularly when the strength of an alloy is high [17]. These data could lead to the erroneous conclusion that just about all materials are susceptible to SCC in pure water. This would make it impossible to operate any plant with steam and water cycles or even in a rainy outdoor environment. Fortunately, and it is thus necessary to emphasize, SCC occurs only in certain limited ranges of temperature, potential, stress, stress intensity factor, strain rate, time, and alloy composition and strength [17,18]. The number of environments that cause corrosion-fatigue failures are practically unlimited. If SCC occurs in alloys only in specific environments (Table 1), almost any external chemical environment differing from absolute (high) vacuum must be considered as “aggressive,” i.e. accelerating both fatigue- crack(s) initiation and propagation, if fluctuating loading is present. Alloys subject to corrosion are subject to corrosion fatigue in any corrosive environment. Gough and Sopwith [19], as early as 1932, showed that the fatigue strengths of several metals were higher in vacuum than they were in normal indoor atmosphere. For most metals, air contributes quite strongly to increasing fatigue-crack propagation and this fact has been well known since the 1960s [20–23]. Most fatigue data obtained to date in inert gases were treated as reference data for more aggressive environments [24,25]. However, Shimojo and his colleagues [26] in 2000 studied the FCG behaviour in titanium at room temperature in an environment described as either vacuum (4 × 10–5 Pa) or high- purity inert gas such as helium (He), neon (Ne), argon (Ar), krypton (Kr) and xenon (Xe) of 1 × 105 Pa and found that inert gases were not completely inactive. These gases increased FCG rates and changed the fracture-surface appearance as compared to those in vacuum. That is, summarizing the data in the literature, it is fairly certain that both fatigue-crack initiation and fatigue-crack propagation are affected by normal air environment and even by pure inert gases. This means that the majority of fatigue failures that were studied in the literature over the last 150 years (beginning in the 1850s [27]) were, in fact, corrosion-fatigue failures, since only fatigue occurring in high vacuum could be termed as “pure fatigue”. Thus, the terms “corrosion fatigue” and “SCC” are not semantically exact as many environments that are not corrodents in the ordinary sense can induce intensive cracking of metals [11,17–23,26]. Contrastingly, other environments

Table 1: Environments causing SCC [17].

Carbon steelsa Austenitic Al alloys Cu alloys Ni alloys stainless steels – – – H2O + NO3 I H2O + Cl H2O + Cl H2O + NH3 H2O + HF – – H2O + OH I H2O + OH H2[SiF6] – H2O + CN T 3– – – – – H2O + PO4 T H2O + F H2O + Br H2O + OH H2O + Cl 2– – – – H2O + SO4 T H2O + Br H2O + I H2O + H2O + OH 2– 2– H2O + CO3 I H2O + SO N2O4 – amines Polythionic acid H2O + CO-CO2 T Polythionic acid HNO3 – citrates Chromic acid H2O + H2S T H2O + H2S Organic liquids – tartrates Acetic acid – H2O + FeCl3 T H2O + H2SO3 Liquid Hg, Ga H2O + NO3 Molten caustics – Organic liquids T Liquid I, Te, Al Moist air H2O + NO2 Steam 2– Liquid ammonia T H2 gas Moist H2, O2, N2 H2O + SO4 Organic liquids 2– Liquid Zn, Li, H2O + SO3 Liquid Li, Hg, – Bi-Pd H2O + S Sn, Zn, Bi, Pb – H2O + F Air Liquid Hg, Bi, Pb atmosphere Steam

H2O T H2O H2O H2O H2O a The mode of crack propagation: transgranular (T) or intergranular (I).

Mg alloys Ti alloys Zr alloys U alloys Co alloys

Humid air H2, F2, Cl2, Br2 H2, Cl2, Br2, I2 Dry O2, H2, air H2 – Outdoors Humid air H2O + FeCl3 Humid air H2O + NO3 – – – H2O + Cl H2O + Cl H2O + CuCl2 Humid N2, O2, H2 H2O + Cl – – – H2O + Br H2O + Br H2O + HI H2O + Cl H2O + HCl – – H2O + I H2O + I H2O + Br2 H2O + H2SO4 2– – H2O + SO4 CH3OH, CCl4 H2O + I2 H2O + H2S + Cl CH2I2, CHCl3 CH3OH, CCl4 H2O + HNO3 CH3CCl3 CHCl3 H2O + NaOH Freon, ClF5 Organic liquids + I2 Hydrocarbons + Alkanes Fused salts HF + BF3 + H2O Fused salts Hg, Cs, Mg, Cd N2O4 HNO3 (RFNA) H2O + NaOH Hg, Ag, Cd

H2O H2O H2O that are strong corrodents may be more beneficial to engineering structures than ambiguous atmosphere [14,28]. To avoid this misconception, the term “environmentally assisted fatigue cracking” is now preferred [11,15], and the use of the term “corrosion fatigue” in the chapter is to be understood within this context.

2.2 Early history of corrosion-fatigue research: 1917–1947

The discovery of corrosion fatigue during the First World War was made by Haigh of the Royal Naval College, Greenwich when he sought an explanation for the frequent failure of paravane towing ropes, which were kept in a state of

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 334 Advances in Fatigue, Fracture and Damage Assessment vibration whilst exposed to seawater. Those failures were among the first failures to be specifically labelled “corrosion fatigue.” The first published account of laboratory experiments on corrosion fatigue of brasses was given by Haigh in 1917 [1]. He found that “the effects of corrosion and fatigue are mutually associated, and that fatigue is accelerated, and occurs under lower stresses, when the conditions tend to promote corrosion” [1]. Haigh stated at the end of his paper that he had noticed that fatigue was accelerated in mild steel and other metals exposed to corrosion by acids, sal-ammoniac and salt water. In addition, in 1919, Langdon [3] in the discussion on the penetration of iron by hydrogen, stated that the fatigue limit of a 9.5-mm diameter steel rod was reduced by 30–50% after the rod was immersed for five minutes in a 10% solution of sulfuric acid at 50°C. No further attention was paid to the subject until about 1925, when investigations were started both in the USA and in England. The most important work was that by McAdam and his colleagues, first at the U.S. Naval Engineering Experiment Station and then at the National Bureau of Standards [4], and that by Gough and his colleagues at the National Physical Laboratory [8,20]. These researchers laid the foundations of our knowledge of the subject and exerted a profound influence on other workers in the field. McAdam carried out a most comprehensive series of investigations covering many variables, and it was he who introduced the term “corrosion fatigue.” In his first paper on the corrosion fatigue of metals [4], McAdam pointed out that “very slight corrosion when simultaneous with fatigue causes low resistance to fatigue. The damaging effect is greater the harder the steel. … Corrosion fatigue is undoubtedly of more importance in machinery parts subjected to repeated bend or torsion than in parts subjected to repeated axial stress”. In 1926, Lehmann [5] showed that strongly corrosive liquids exhibited a surprisingly small effect on the fatigue resistance when they contained little dissolved air. The importance of oxygen on corrosion- fatigue properties was further demonstrated in work carried out by Binnie [7]. Speller, in 1926, pointed out in discussing one of McAdam’s papers that hydrogen penetration along grain boundaries may play a part in the corrosion fatigue of steel [6]. The 1926 papers [4,5] aroused great interest, and corrosion fatigue has since been the subject of vigorous research. Between 1926 and 1930 (and at later intervals) McAdam carried out a magnificent series of studies on this subject, which was conventionally reviewed by Gough in 1932 [8] and Dorey in 1933 [29]. He published upwards of 20 papers on the subject in the period 1926–1941 and studied relationships between the stress range, time, number of cycles, and endurance limit for a wide variety of materials, including ingot iron, steel, low- alloy steels, stainless irons and steels, nickel, Monel, copper, Cu-Ni alloys, brasses, aluminium bronze, aluminium, duralumin and Al-1¼ Mg alloy. In 1930, Harvey wrote, “At present, the discovery of the “corrosion fatigue” phenomenon is universally recognized as one of the eminent metallurgical achievements of the twentieth century. Within four years of its initial study, corrosion-fatigue has

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 335 been recognized as a vital problem to every engineer whose products are subject to repeated stress” [31]. Thereafter, it became evident that corrosion fatigue is responsible for many kinds of failure in marine propeller shaft and rudders, steering arms and axles of motor vehicles, boilers and super-heater tubes, pump shafts, rods and bodies and other equipment [30]. Due to this fact, the volume of literature on corrosion fatigue that appeared between 1925 and the beginning of the Second World War was considerable. Subsequently, however, interest in the subject has declined somewhat. The number of papers published during the 16-year period 1940– 1955 is of the order of one-half of the number published during the 15-year period 1925–1939 [32]. In 1935, DeForest [33] noted that the majority of fatigue failures in service were found to initiate at pre-existing defects (including various notches resulting from design, such as fillets, screw threads, oil holes or resulting from defects of surfaces such as grinding marks, grooves coming from machining operations, decarburized surfaces, or the accidental bruising received in hardening, or the surface produced by corrosion) and the useful life of the component was primarily dependent upon the rate of FCG. He was the first to present experimental data on the rate of crack growth and on the size of cracks. The most important researches on corrosion fatigue since 1939 were those of Evans and his co-workers at the University of Cambridge on electrochemical aspects of corrosion fatigue [9,10,34]. For example, Gould and Evans, by measuring an electrode potential during corrosion fatigue tests on steel wire in chloride and chromate mixture, found that a sudden drop in potential occurred at a certain stage, indicating breakdown of a protective surface film [34]. Early in the study of corrosion fatigue, it was shown that cracking could be delayed or prevented by cathodic protection by contact with a less noble metal [8]. Evans and Simnad showed that the same result could be produced by applied cathodic potential [9]. In 1947, Evans summarized the work on the electrochemistry of corrosion fatigue carried out at Cambridge up to that time [10].

2.3 Stress–life (S–N) corrosion-fatigue behaviour

Traditionally, beginning with the first research [1], investigations of susceptibility to corrosion fatigue have involved the stressing of smooth cylindrical specimens—by some convenient means, involving bending, axial or torsional stress, usually with cycles of reversed stress—until they failed or survived some pre-determined target number of stress cycles. The tests are carried out on the same types of machine as are used for ordinary fatigue tests, suitably modified to enable the desired corrosive environment to be maintained round the specimen or part of the specimen. Test data from this type of experiment are typically presented in the form of an S–N (stress–life) diagram (Fig. 1), which shows the number of cycles to failure, N, as a function of the cyclic-stress range, S. [Note: For a long time, such curves were labelled as a “Wöhler curve” [27] instead of the now more frequently used the term “S–N

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 336 Advances in Fatigue, Fracture and Damage Assessment curve”.] The same technique and method of results presentation [1,27] are still used today, particularly in the context of engineering qualification tests on components and welded connections. High test frequencies, often much higher than 10 Hz (e.g., 160 Hz [35]), are necessary if the complete S–N curve, including low-stress ranges and high-cyclic lives greater than 106 cycles, is to be defined in an acceptably short period of time. The horizontal portion of an S–N curve represents the maximum stress that the metal can withstand for an infinitely large number of cycles. This maximum stress is known as the fatigue (or endurance) limit. For steels, but not necessarily for other metals, a true fatigue limit exists that is approximately half the tensile strength. Most non-ferrous metals do not exhibit a fatigue limit. Instead, their S– N curves continue to drop at a slow rate at high numbers of cycles. For these types of metals, fatigue strength, rather than fatigue limit, is reported for a specific number of cycles.

Figure 1: S–N curves for fatigue tests in air and in a corrosive environment.

In corrosive environments, the S–N curve never becomes truly horizontal, and there is thus no true corrosion-fatigue limit (Fig. 1); metals can eventually fail under any alternating stress if the number of cycles is sufficiently large. In such cases, corrosion-fatigue strength is determined as a fatigue limit at a specific (prescribed) number of cycles. Figure 1 also shows that the number of stress

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 337 cycles that any metal can resist before failure is always less for corrosion fatigue than for fatigue. Corrosion-fatigue strengths were found to be insensitive to the metallurgical condition, showing no correlation with tensile strength in contrast to that observed in air. If the fatigue limits of steels for bending loading in air are about one-half their tensile strengths, their corrosion-fatigue strengths may possibly be no more than 10% of their fatigue limits. For example, Fig. 2 shows that the fatigue limit in air was much higher for high-strength steel (465 MPa) than for low-strength steel (160 MPa). But in water both steels showed the same, extremely low corrosion-fatigue strength that continued to fall further as the number of load cycles increased from 107 (65 MPa) to 109 (10 MPa) [35]. It is noteworthy that the large effect of water was observed despite the use of high

Figure 2: S–N curves of low-strength steel G 10050 (0.02%C; UTS = 294 MPa) and high-strength steel K 08500 (0.8%C; UTS = 1100 MPa) tested in air and in water (after Speidel [35]). cyclic frequency (30 to 60 Hz) when, as will be shown in Sect. 2.4.3, environmental effects on corrosion FCG rates would be negligible. Corrosion resistance, often specifically pitting resistance, was much more important in determining the corrosion-fatigue strength. If pitting corrosion is allowed, all carbon steels, no matter how high their carbon contents or their tensile strength,

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 338 Advances in Fatigue, Fracture and Damage Assessment will all have the same extremely low corrosion-fatigue strength. Most comprehensive compilations of fatigue strength as a function of alloy strength are reproduced in Fig. 3 [35]. From this figure, it can be concluded that the fatigue strength of carbon steels in air increased from 100 MPa to about 600 MPa with increasing tensile strength from 200 MPa to 1200 MPa, but the corrosion-fatigue strength was equally low for all steels in aerated water (110 MPa) and seawater (40 MPa), i.e. in environments that cause pitting corrosion. Similar observations were made with a large number of alloys, a ranking of which is illustrated in Fig. 4. All materials from the same alloy group exhibit the same low corrosion-fatigue strength, almost independent of composition and strength level. Figure 4 shows that cobalt alloys and titanium alloys permit some of the best corrosion-fatigue strength levels to be obtained. This is one reason why materials from these groups are used not only in modern high technology but also where their corrosion-fatigue resistance is most immediately felt—for example, as implants in the human body [35]. The type of loading (reversed bending of a specimen with R = –1) and its frequency (much higher than 10 Hz) in the tests described above are both not typical for realistic operating conditions. For example, daily pressure fluctuations in a liquid pipeline correspond to cyclic loading with the stress ratio of about 0.6 to 0.8 at a cyclic frequency from about 2 to 5 cycles per day; under other circumstances, the stress ratio can vary from zero to 0.9 and a cyclic frequency can be in the range between 0.7 and 1.3 cycles per day [15]. Moreover, tests of this type do not allow for separating the crack-growth stage, which is of special practical interest—the higher the rate of subcritical crack growth, the shorter the lifetime of a structure that experiences a crack-type defect(s)—from the total corrosion-fatigue lifetime that includes the five stages of material failure: nucleation of crack/defect(s); initiation of crack(s) from the nucleus/defects; growth of small mutual cracks; growth of subcritical crack(s); and the final (brittle) fracture of material [15]. The total duration required for such corrosion-fatigue tests can exceed six months [36]. If several materials are to be tested under various conditions, the time problem starts to be critical. A dramatic decrease in the total time required for evaluating the corrosion-fatigue resistance as well as obtaining more applicable experimental data in comparison with those described above can be reached by using fracture-mechanics technology [11,17,35–37].

3: Figure st tensile of varying of carbon steels strength The fatigue

Figure 4: Corrosion-fatigue strength levels for various alloy systems in aerated salt solutions or seawater (after Speidel [35]).

2.4 Corrosion-fatigue crack growth (da/dN–∆K) behaviour

The process of corrosion fatigue is usually discussed in terms of crack initiation (incubation and nucleation) and crack propagation. It begins with the initiation of a crack(s), and with continued cyclic loading the crack propagates, finally leading to the fracture of a component/specimen or structure. That is, for many materials and under a wide variety of experimental and in-service conditions, unstable fracture may be preceded by a period of stable (subcritical) crack propagation. In this period, crack propagation will cease if the cyclic loading or an external chemical environment is removed; conversely, with continued loading and/or environmental action, crack growth will continue until the

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 341 process is no longer self-limited, i.e. further crack propagation is very rapid and loading is unnecessary. With the development of linear elastic fracture mechanics (LEFM) in the early 1960s and the recognition that FCG rates per cycle, da/dN, could be expressed as a simple function of the stress-intensity factor range, ∆K [38,39], increasing attention has been focused on measuring the rates of corrosion-fatigue crack propagation. The introduction of LEFM technology in corrosion-fatigue research was one of the significant developments in the understanding of corrosion-fatigue phenomena and the utilization of crack-growth data in the design and evaluation of engineering structures.

2.4.1 Fracture mechanics in SCC research The results of Brown [40] and Brown and Beachem [41] were those that pioneered, in the mid-1960s, an extension of LEFM methods to SCC testing and stimulated a great body of subsequent research in SCC and corrosion FCG. Brown and Beachem [40,41] introduced into SCC and fracture-mechanics practice the threshold stress-intensity factor, KISCC, below which SCC does not occur. Since that time, KISCC has been a key parameter characterising susceptibility of a material to SCC. The KISCC parameter is commonly referred to as the SCC threshold, meaning that it is the stress-intensity factor in the opening mode (Mode I) below which SCC will not initiate. KISCC is considered to be a material property for a particular material–environment system. The use of the KISCC parameter based on the relationship between the stress-intensity factor, K, nominal applied stress, σ, and the crack length, a, has practical significance. Once the appropriate value of KISCC has been determined, it can be used for structural design to establish the critical combination of applied stress and defect size below which SCC will not occur for a given material–environment system. When KISCC is known, it can be used to calculate either the allowable crack size in the structure (at the given level of applied stress) or the allowable stress below which SCC will not be expected. Another manifestation of the SCC phenomenon is illustrated schematically in Fig. 5, where the average stress-corrosion crack growth rate, da/dt, is plotted as a function of the applied stress-intensity factor, K. Crack growth rates in susceptible material–environment systems often follow three distinct regions as first indicated by Wiederhorn in 1967 [42]. In Region I, or the linear region at low K with a strongly stress-dependent crack growth rate, the crack growth rate increases with increase in stress-intensity factor. In cases where KISCC is time dependent, its value is determined from Wiederhorn’s diagram as the value corresponding to the crack growth rate of 3 × 10–11 m/s (or roughly one mm/y). In Region II, or the steady-state crack-growth region at intermediate K, the crack growth rate is relatively independent of stress-intensity factor (plateau region). Crack growth rates in plateau region (“plateau crack growth rates”) can vary widely and data from 10–12 to 10–1 m/s were reported [17]. In Region III, or the

Figure 5: A schematic representation of stress-corrosion crack growth rate, da/dt, as a function of stress-intensity factor, K. fast-fracture region, the crack-growth rate again accelerates with increase in stress-intensity factor. The use of the crack-growth data such as those shown in Fig. 5 allows for an estimation of how long it would take to break a specimen/component containing a crack(s) of known length under an applied load. If the environment is inert or if the material is fully resistant to SCC, no crack growth will occur until the critical stress intensity (i.e., fracture toughness, KIC) is reached. If the environment is “aggressive” and the material is susceptible to SCC in this environment, subcritical crack growth will occur at stress- intensity factors below or even far below KIC. It is presumed that once SCC starts, it will propagate at increasing rates following first the Region I curve and then the Region II curve until the stress-intensity factor is sufficiently large that KIC is reached and catastrophic failure is reached. There is a strong effect of yield strength on KISCC and crack growth rates. For example, as Fig. 6 shows, a doubling of the yield strength of AISI 4340 steel

(from 800 to about 1600 MPa) is accompanied by a ten-fold (from 70 to 7 1/2 MPa m ) decrease of KISCC [41]. Figure 7 shows, the “plateau crack growth rate” in low-alloy steel of 1700 MPa yield strength in deaerated water at 100°C was equal to 10–4 m/s and exceeded that (∼10–11 m/s) in a similar steel of 760 MPa yield strength by seven orders of magnitude [17]. The data show that steels with higher yield strength (near 1500 MPa) will result in fast stress-corrosion crack growth whereas steels with yield strength below 1000 MPa may still not be entirely immune to SCC, but the crack growth rates in such steels are many orders of magnitude lower.

Figure 6: Effect of yield strength on the fracture toughness, KIC, and on the threshold stress-intensity factor, KISCC, for AISI 4340 steel (after Brown and Beachem [41]).

Figure 7: Effect of yield strength on stress-corrosion crack growth rates in steels (after Speidel [17]).

2.4.2 Fracture mechanics in corrosion FCG research During the past 40 years, systematic studies of the kinetics of corrosion-fatigue crack propagation in metals have led to an improved understanding of the effects of those mechanical, metallurgical and environmental variables that control the possibility, mode and rate of corrosion-fatigue damage [14,37]. Such an understanding has been developed quantitatively through the use of LEFM [11,17,35–37]. Most corrosion FCG tests are conducted by subjecting a pre- cracked specimen to constant-amplitude cyclic-load fluctuations. An incremental increase of crack length is measured, and the corresponding number of elapsed

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 345 load cycles is recorded. The FCG data are most commonly presented on a log- log plot of da/dN versus ∆K, as shown schematically in Fig. 8 [43]. Also shown in this figure is the general form of the time-dependent (not cycle-dependent) SCC rate, da/dt, as a function of the applied stress-intensity factor and how this can be superimposed on corrosion FCG if the stress-intensity factor in the stress cycle exceeds KISCC. The da/dN–∆K curve is usually divided into three regions (Fig. 8). In an inert environment, the FCG behaviour in Region I exhibits a threshold value, ∆Kth, below which there is no observable crack growth. This threshold occurs at crack growth rates in the order of 10–10 m/cycle or less, as defined in ASTM Standard

E647 [44]. Below ∆Kth, cracks are characterized as nonpropagating fatigue cracks. Microstructure, mean stress, frequency and environment mainly control the Region I crack growth. Region II (Paris region) shows essentially a linear relationship between log da/dN and log ∆K, which corresponds to the equation da/dN = C(∆K)n, (1)

Figure 8: A schematic representation of corrosion FCG rate, da/dN, as a function of stress-intensity factor range, ∆K (after Sprowls [43]).

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 346 Advances in Fatigue, Fracture and Damage Assessment first suggested in 1963 by Paris and Erdogan [39]. Here n and C are constants for a given material and stress ratio. Region II corresponds to stable macroscopic crack growth that is typically controlled by the environment, whereas microstructure and mean stress have less influence on FCG behaviour in Region II than in Region I. Typically, FCG rates in this region are larger than 10–8 m/cycle and smaller than 10–3 m/cycle; that is, the region covers the crack growth rates and stress-intensity factor ranges in which most published FCG studies in controlled environments were carried out. In Region III, the FCG rates are higher than those predicted for Region II as they approach instability, and little useful life in the structure remains before disintegration. This region is controlled primarily by fracture toughness, KC or KIC, which in turn depends on the microstructure, mean stress and environment. In an aggressive environment, the da/dN–∆K curve can be quite different from the pure fatigue curve, depending on the sensitivity of the material to the given environment and the occurrence of static SCC above KISCC. In addition, certain loading factors such as cyclic frequency, stress ratio and loading waveform can have marked effects on the crack-growth curves in the same environment. An applied electrode potential also can have a marked effect on the crack-growth curves. The application of potential can, with all other conditions (specimen, solution, pH, cyclic frequency, stress ratio, loading waveform, temperature, etc.) being equal, accelerate or retard the propagation of the crack by a factor that may range up to many times [14,28,45]. Therefore, a variety of curves can be expected from the broad range of material–environment–loading systems that will produce various corrosion FCG behaviours.

2.4.3 Basic types of corrosion FCG behaviour The schematics in Fig. 9 show three general patterns of corrosion FCG behaviour in aggressive environments compared with crack-growth behaviour in an inert environment. [Note: The first representation of the patterns by McEvily and Wei [46] was based on the maximum stress-intensity factor, Kmax, and later Austen and Walker [47] re-represented these patterns based on ∆K.] The first type of corrosion FCG behaviour represents those material–environment systems in which the environmental effect results from the synergistic interaction between the cyclic plastic deformation (fatigue) and the environment (corrosion). The environmental effect is evidenced by a reduction in the apparent threshold stress-intensity factor range (from ∆Kth to ∆KICFC as in Fig. 8) and an increase in the crack growth rate at the given level of ∆K. This behaviour pattern is referred to as “true corrosion fatigue” [48] and is typical for low- to moderate- strength alloys that are either immune to SCC or exhibit high KISCC and low da/dt. Figure 10 shows an example of this type of corrosion FCG behaviour for an aluminium alloy with high resistance to SCC [49]. The crack growth rates are ranged up to one order of magnitude higher in 0.6 M NaCl solution than those in dry air (Fig. 10).

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 347 (c) (b) rrosion FCG behaviour (after Austen and Walker [47]). and Walker Austen (after behaviour FCG rrosion (a) Figure 9: Schematic representations of basic types of co types of basic representations Schematic 9: Figure

Figure 10: Corrosion FCG rates for 7075-T6 aluminium alloy in 0.6 M NaCl. The data also illustrate the similar behaviour when tested with either a K-increasing (remote load) or K-decreasing (wedge force) loading method (Feeney et al. [49]).

The second type of behaviour in Fig. 9 represents those material– environment systems in which there is a substantial environment-enhanced sustained load crack-growth component. The environmental effect is quite strong above KISCC and is negligible below KISCC. This behaviour pattern is referred to as “stress-corrosion fatigue” [48] and is typical for high-strength steels in water vapour and argon. Figure 11 shows an example of this type of corrosion FCG behaviour for high-strength steel 4340 M tested in distilled water at room temperature [35]. In this example, at a frequency of 10–3 Hz, distilled water increased crack growth rates by five orders of magnitude in comparison with those in vacuum.

Figure 11: Effect of cyclic frequency on the corrosion FCG rate in high-strength steel 4340 M tested in distilled water (after Speidel [35]).

The third type in Fig. 9 represents the corrosion FCG behaviour of a broad range of material–environment systems that exhibit “stress-corrosion fatigue” above KISCC and also “true corrosion fatigue” at all levels of ∆K. The transformation from one type of behaviour to the other can be made by the change in cyclic frequency, stress ratio and electrode potential. For example, Figure 12 shows how increasing frequency shifts the “stress-corrosion fatigue” plateau crack

(a)

(b)

Figure 12: The effects of: (a) increasing frequency (F1 < F2 < F3) at constant R, and (b) decreasing stress ratio (R3 < R2 < R1) at constant frequency on corrosion FCG behaviour (after Austen and Walker [47]).

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 351 growth rate vertically downwards and decreasing the stress ratio shifts the “plateau crack growth rate”horizontally to the right [47]. Thus, increasing frequency and/or decreasing the stress ratio suppresses the appearance of “stress- corrosion fatigue” such that only “true corrosion fatigue” remains and the corrosion FCG behaviour changes from type two (or three) to type one as in Fig. 9. This occurs because FCG is a cycle-dependent process such that Region II governed by (1) is invariant despite a change in frequency and stress ratio. On the other hand, SCC is a time-dependent process such that increasing frequency decreases the “plateau crack growth rate” in terms of da/dN [Figs. 11 and 12(a)].

In addition, the values of ∆KSCC under cyclic loading that correspond to KISCC under static loading and the values of KP that correspond to the onset of the plateau on the da/dN–∆K curve vary with stress ratio in a (1 – R) relationship [47]. If, for a constant ∆K, the stress ratio is increased, the environmental acceleration of FCG is expected to increase also [Fig. 12(b)]. On the other hand, if, for a constant Kmax, the stress ratio is increased, the environmental acceleration of FCG decreases. This is why in the literature there are two mutually exclusive statements: “the corrosion FCG rate increases with increasing stress ratio” and “the corrosion FCG rate decreases with increasing stress ratio.” In order to determine which one of the statements is correct, it is necessary to show which of the two stress-state parameters, ∆K or Kmax, was taken into consideration. It is necessary to note that a complete da/dN–∆K curve (Fig. 9) beginning from 10–10 m/cycle is difficult to obtain due to the long time needed to measure low crack growth rates. Cyclic frequencies of 10 to 100 Hz that are usually used in standard fatigue and FCG tests are too high for studying corrosion FCG behaviour, and frequencies of 0.01 and 0.1 Hz are required to obtain realistic corrosion FCG data. This implies that weeks and even months of test time are needed. Using sensitive methods of crack-growth monitoring such as, for example, direct-current potential drop method that allows for detecting as little as 0.001 mm crack extension [14,28], does not make it easier to take the measurement of low crack growth rates because in this case it is difficult to keep the high resolution of the method for longer than say two hours. Thus, most corrosion da/dN data available in the literature are for more than 10–8 m/cycle that comprises Regions II and III in Fig. 9.

3 Mechanisms for corrosion-fatigue crack propagation

By now, the influence of mechanical, metallurgical, and environmental variables on the corrosion FCG behaviour of a broad range of material–environment systems has been studied in some detail [11,14,17,20–24,26,28,35–37,46–49]. However, the search for prevention methods, including the selection of materials resistant to the failure mode, is complicated by the fact that quite a number of fundamental questions regarding the possibility of corrosion fatigue remain unanswered. Foremost among these questions is the problem of the corrosion FCG mechanism(s) [14,37]. The basic mechanisms, which have generally been

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 352 Advances in Fatigue, Fracture and Damage Assessment proposed to explain corrosion FCG, are similar to those invoked in explaining SCC and began with research in SCC [50]. Two processes—hydrogen-induced cracking (HIC, or “hydrogen embrittlement”) and stress-assisted dissolution (SAD) of metal at the crack tip—are given main consideration as possible mechanisms of corrosion FCG. Choosing correctly from among the various measures—alloying elements, heat treatment, parameters of cathodic/anodic protection, and inhibitors—for the prevention of corrosion-fatigue failures is impossible if there is no way to make a reliable identification and quantitative estimation of the role of each of these two principally different mechanisms such as HIC and SAD in corrosion-fatigue crack propagation. In the case of high-strength low-alloy steels and titanium alloys, for example, no firm consensus exists on the corrosion FCG mechanism in any one material– environment combination. For these materials, some authors propose that HIC plays a dominant role in corrosion-fatigue crack propagation [51–61]. As evidence of HIC they point to the deleterious effect of increasingly negative cathodic potential on corrosion FCG rates [51–59] and the similarity between features of the fracture surfaces produced by crack growth in aqueous solutions and in hydrogen gas [51,58,60,61]. Others provide evidence that SAD is the dominant mechanism for corrosion FCG in high-strength steels [62–64] and titanium alloys [48,65]. In contrast to the emphasis placed on high-strength steels and titanium alloys, there is very little information on the corrosion FCG behaviour of materials such as magnesium alloys that have high strength-to- density ratios [66], although much more effort has been devoted to studying SCC in magnesium alloys [67–76]. The mechanism of SCC of magnesium alloys has been ascribed to either continuous crack propagation as a result of SAD [67,68,72], or discontinuous crack propagation as a result of a series of hydrogen-induced fractures at the crack tip [71,73–76]. Also, SCC of magnesium alloys can involve the repeated cycles of a slow stage of SAD, followed by a fast stage of mechanical fracture [69,70]. One of the main arguments in favour of the SAD mechanism is that SCC of magnesium alloys can be stopped by cathodic polarization and restarted after polarization is removed [67,72]. At least two techniques are preferred among those commonly used for identifying corrosion FCG mechanism(s). The first is the technique based on studying the effect of controlled (applied) potential on the corrosion FCG rates [51–60,62,77]. Fractographic analysis of fracture surfaces is the second most popular technique and is usually employed in conjunction with the first technique [51,58,60,62].

3.1 Controlled-potential technique

In aqueous environments, anodic dissolution of metal at the crack tip and the ingress of hydrogen derived from the environment [78] into the crack-tip plastic zone occur simultaneously during corrosion-fatigue crack propagation. As a result, both SAD and HIC are involved in the acceleration of crack growth in

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 353 comparison to crack growth in vacuum. [Note: Crack growth due to internal hydrogen occurring in materials such as, for example, titanium alloys when testing in vacuum was not included in this analysis.] Depending on mechanical (stress level, mode of loading, cyclic frequency, stress ratio, waveform), metallurgical (chemical composition and microstructure of material, grain- boundary composition, surface condition, specimen orientation) and environmental (composition and concentration of solution, pH, potential, dissolved oxygen, temperature, flow rate) variables, one of the two mechanisms plays a dominant role in crack propagation. The intensity of anodic and/or cathodic reactions in the vicinity of the crack tip, the hydrogen-dependent properties of material and the hydrogen content in the crack-tip zone also determine the controlling corrosion FCG mechanism [14]. Generally, anodic polarization increases metal dissolution and decreases hydrogen generation, whereas cathodic polarization decreases metal dissolution and increases hydrogen generation. This statement identifies the controlled-potential technique [79–81] as a technique for distinguishing between HIC and SAD as possible mechanisms of corrosion FCG by studying the effect of applied potential on the crack growth rate [51–60,62,77]. The basic idea of the controlled-potential technique was proposed by Uhlig in 1950 [79] and then developed by Phelps and Loginow in 1960 [80] to distinguish between HIC and SAD as possible mechanisms of SCC in tests using smooth specimens. They suggested that, for example, if the application of small cathodic current decreases the time to failure, whereas small anodic current increases this time, the SCC mechanism operating in the absence of applied currents would be HIC. Similarly, the open-circuit mechanism is considered SAD if the time to failure increases with the application of small cathodic current and decreases with the application of small anodic current. Although at first glance this appears to be a simplistic approach, it has proved useful on occasions. In 1967, Leckie [81] adopted this approach to studying SCC in pre- cracked specimens loaded statically. In 1971, Barsom [51] and Gallagher [52] pioneered the application of the controlled-potential technique to distinguish between HIC and SAD during corrosion-fatigue crack propagation. In addition, Barsom [51], Meyn [53], and, more recently, Rungta and Begley [62] first examined in greater detail the features of the fracture surfaces produced by corrosion FCG under applied potentials by scanning electron microscopy (SEM) and transmission electron microscopy of plastic/carbon replicas. These examinations were conducted to provide further information concerning the corrosion FCG mechanisms. However, none of the above-mentioned attempts [51,52,60,62], nor numerous attempts of other authors [53–59,77] who employed the controlled-potential technique and fractographic evidence, allowed for the possibility of quantifying the role of any one mechanism in corrosion- fatigue crack propagation.

3.2 Defining the role of HIC and SAD in corrosion-fatigue crack propagation

Recent studies [14,28] were undertaken as a step towards the development of an approach that would allow for the identification of the corrosion FCG mechanism in any material, regardless of its composition and microstructure. An attempt was made to quantify the role of HIC and SAD in corrosion fatigue crack propagation in metals that were of a different base, composition and microstructure. For this purpose, the well-known controlled-potential technique [51–60,62,77] was modified, and the effect of short-term electrochemical polarization on corrosion FCG rates was studied using views developed elsewhere to examine SCC mechanisms [82,83].

Materials tested were martensitic steels 39Ni4MoV (σY = 1540 MPa) and 32Cr3NiMoV (σY = 1600 MPa), titanium alloys VT20 (near-α; σY = 870 MPa) and TS6 (near-β; σY = 1540 MPa), and magnesium alloys VMD10 (Mg-Y-Zn; σY = 240 MPa) and IMV7-1 (Mg-Y-Gd; σY = 290 MPa). Specimens were cycled under tension-tension loading with a symmetrical triangular load waveform at a cyclic frequency of 0.1 Hz. The stress ratio was kept constant between 0.1 and 0.9 during each individual test. Tests were performed at 25 ±0.5°C. Crack extension was monitored continuously by direct-current electrical resistance measurements. With temperature stability ±0.5°C, it was possible to detect as little as 0.001-mm crack extension. In addition, the fracture surfaces were examined by SEM to supplement the corrosion FCG data obtained under applied potential conditions. The experimental approach [14,28] was intended for short-term stimulation of electrochemical reactions within a growing crack. For this purpose, anodic or cathodic potential was applied to the specimen after reaching an initial corrosion FCG rate (da/dN)i corresponding to the open-circuit potential (OCP) at a given Kmax for a time period in which it was possible to register the change in the initial crack growth rate and to measure the crack growth rate (da/dN)p under polarization. Due to the high resolution of the crack-growth monitoring technique, some change in (da/dN)i was observed about 10 to 150 s after polarization application. Time periods needed to measure (da/dN)p rarely exceeded 300 s. Therefore, polarization was removed as soon as (da/dN)p was measured and the system was returned to the potential equal to OCP. After that the experiment was repeated at the new level of Kmax. The range of (da/dN)i was changed from about 5 × 10–9 to above 10–5 m/cycle. The degree of the polarization effect on the corrosion FCG was defined as the ratio of (da/dN)p to (da/dN)i. The approach made possible the observation of a non-single mode effect of cathodic polarization on corrosion FCG rates. An important point was that cathodic polarization accelerated crack growth when Kmax and the corresponding da/dN, exceeded certain well-defined critical values characteristic for a given material–solution system. When Kmax and da/dN were lower than the critical values, the same cathodic polarization, with all other conditions (specimen,

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 355 solution, pH, temperature, frequency, stress ratio, etc.) remaining unchanged, retarded or had no obvious influence on crack growth. For example, Fig. 13 shows the diagram indicating the change in the corrosion FCG rates of 39Ni4MoV steel tested in 0.5 M H2CrO4 solution after cathodic polarization application when Kmax was either lower or higher than 8.3 1/2 1/2 MPa m . In the former case (Kmax = 8.1 MPa m ), the crack was arrested when cathodic potential of –200 mVSHE (versus standard hydrogen electrode, SHE) was applied but after about 210 s, the crack growth rate became equal to the 1/2 initial value (da/dN)i. In the second case (Kmax = 8.9 MPa m ), the same cathodic potential increased the crack growth rate by a factor of 12.7 such that –6 (da/dN)p = 1.2 × 10 m/cycle. The crack-growth acceleration began after a time delay of about 120 s after polarization was applied. After polarization was removed and the system was returned to the potential equal to OCP, the crack growth rate returned to the initial value (da/dN)i. Figure 14(a) shows the relationship between the (da/dN)p/(da/dN)i ratio and the initial crack growth rate –8 –6 (da/dN)i that was changed within the range from 2.5 × 10 m/cycle to 4.5 × 10 m/cycle. As shown in Fig. 14(b), the accelerating effect of cathodic polarization on corrosion FCG rates for 39Ni4MoV steel tested in 0.5 M H2CrO4 solution at –200 mVSHE was more pronounced with higher R. For example, the higher ratios of (da/dN)p/(da/dN)i for this material–solution system were equal to 1.6 at R =0.1 and more than 20 at R = 0.85. The nature of the effect of cathodic polarization on corrosion FCG rates did not change when an acidic solution such as 0.5 M H2CrO4 (pH 0.15) was replaced by the neutral 0.01 M and 0.5 M NaCl (pH 7), weakly alkaline 0.01 M and 0.05 M Na2B4O7 (pH 9.2) or more alkaline 0.03 M Na2HPO4 (pH 10) and 0.1 M NaOH (pH 13). For each of the steel– solution combinations tested, the ranges of Kmax and da/dN where a given cathodic potential either increased or decreased corrosion FCG rates comparable to those at OCP were observed [14]. Figure 15(b) shows that cathodic polarization at –200 mVSHE decreased corrosion FCG rates in VT20 and TS6 alloys tested in 0.6 M FeCl3 at R = 0.8 1/2 1/2 when Kmax values were lower than 63.0 MPa m and 72.6 MPa m , respectively; the corresponding critical crack growth rates were 4.5 × 10–7 m/cycle for VT20 alloy and 6.8 × 10–7 m/cycle for TS6 alloy [Fig. 15(a)]. On reaching these values of Kmax, the same cathodic polarization increased crack growth rates by a factor of 50 in VT20 alloy and by a factor of 70 in TS6 alloy. A strong dependence between (da/dN)p/(da/dN)i ratios and R was observed. For example, for TS6 alloy tested in 0.6 M FeCl3 at –200 mVSHE, the higher ratios of (da/dN)p/(da/dN)i were equal to 5 at R = 0.1, 30 at R = 0.5 and 75 at R = 0.8. The critical value of Kmax depended on the solution chemistry and the ratio of activating/passivating species. For example, for VT20 alloy tested at –200 1/2 mVSHE and R = 0.8, the critical Kmax = 48.5 MPa m in 2 M H2CrO4 + 0.1 M 1/2 NaCl solution, 43.5 MPa m in 1 M H2CrO4 + 0.1 M NaCl solution, and 27 1/2 MPa m in 0.5 M H2CrO4 + 0.1 M NaCl solution.

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 356 Advances in Fatigue, Fracture and Damage Assessment s in 39Ni4MoV steel after cathodic- after steel 39Ni4MoV s in was either (A) lower or (B) higher than the critical value of 8.3 value critical the than higher or (B) (A) lower was either max K (after Shipilov [14]). 1/2 MPa m polarization application when application polarization Figure 13:Diagram fragments indicating the change in corrosion FCG rate FCG corrosion change in the indicating fragments 13:Diagram Figure

Figure 14: Dependence of the (da/dN)p/(da/dN)i ratio for 39Ni4MoV steel tested in 0.5 M H2CrO4 solution on: (a) the initial crack growth rate corresponding to OCP and (b) on stress ratio (after Shipilov [14]).

Figure 16 shows results obtained in testing VMD10 alloy [HRB 39, long transverse (LT) orientation] in 1 M NaOH solution at R = 0.5. Cathodic 2 polarization at –1800 mVSHE (current density ≤ 3 mA/cm ) had no influence on crack growth when Kmax and the corresponding FCG rate da/dN were lower than 13.2 MPa m1/2 and 5.6 × 10–7 m/cycle, respectively. But on reaching these values of Kmax and da/dN, the same cathodic potential accelerated crack growth in comparison to that at OCP. Polarization increased the crack growth rate by a 1/2 factor of 1.5 at Kmax = 13.6 MPa m and the degree of the accelerating effect of 1/2 polarization on crack growth rate increased to 6 at Kmax = 16.4 MPa m .

1/2 Kmax (MPa m )

Figure 15: Dependence of: (a) the corrosion FCG rate, and (b) the (da/dN)p/(da/dN)i ratio on Kmax for titanium alloys in 0.6 M FeCl3 solution (after Shipilov [14]).

The non-single mode effect of cathodic polarization on corrosion FCG rates that was observed in studies [14,28] indicated that, depending on the Kmax value, a given applied cathodic potential either increased by several times or decreased (or had no influence on) crack growth rates comparable to those at OCP could not be explained from the standpoint of HIC or SAD only. On the other hand, the discovered effect provided a reason to conceive a possibility for

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 359 simultaneous realisation of both mechanisms in the crack propagation. Here, the role of each individual mechanism was determined by the Kmax value (in addition to other variables) that was either lower or higher than a certain critical value characteristic for a given material–solution combination. The interpretation of the results [14,28] and the explanation of the non-single mode effect of cathodic polarization on corrosion FCG rates were carried out using views and theories developed elsewhere for understanding crack-growth mechanism(s) in metals preliminary charged by hydrogen or exposed to hydrogen-producing environments such as aqueous solutions or hydrogen gas.

3.2.1 HIC during corrosion-fatigue crack propagation

3.2.1.1 Notion of KHIC, ∆KHIC and (da/dN)cr parameters In some metal– solution systems, even for highly conductive solutions, the local chemistry within the crack may not be the same as the chemistry of the bulk solution [48,84–86].

1/2 Kmax (MPa m )

Figure 16: Corrosion FCG rates for VMD10 alloy in 1 M NaOH solution at OCP of –1150 mVSHE and cathodic potential of –1800 mVSHE (after Shipilov [14]).

However, there was no reason to relate this difference as well as the possible difference between the applied (external) potential and the potential at the crack tip [86,87] with critical values of Kmax observed in studies [14,28]. These differences could not explain why only very small increases beyond the critical Kmax reversed the effect of cathodic polarization and accelerated crack growth instead of retarding it. On the other hand, an increased generation of hydrogen under cathodic polarization and its entrance into metal occurred on the external surface of the specimen [78] and within the growing crack [86,88]. In the case

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 360 Advances in Fatigue, Fracture and Damage Assessment being considered, FCG provided local breakdown of the passive film at the crack-tip surface and produced bare (film-free) metal surface during the opening part of each stress cycle [63–65], e.g., by a mechanism of alternating sliding-off at the crack front [89,90]. This promoted entry of hydrogen into the metal that was transported under the driving force of the stress gradient to the region of highest dilation ahead of the crack where some form of hydrogen embrittlement could take place, presumably, by a reduction in the cohesive strength of the lattice [91,92]. The above arguments as well as the fact that susceptibility to HIC under cathodic polarization is inherent to high-strength steels [93,94], titanium alloys [95–97], and magnesium alloys [71,75], suggested that the accelerated corrosion FCG in the materials tested at cathodic potentials was caused by hydrogen charging and HIC. (Others [51–60,64] have explained the deleterious effect of cathodic polarization on corrosion FCG rates in a similar way.) However, it is known that the stress-intensity factor has only a small influence on the rate of crack-tip hydrogen generation [98] and there is no sharp dependence of the kinetics of hydrogen penetration into metals either on stress or stress-intensity factor [99–101]. Moreover, in studies [14,28], the same cathodic potential that increased crack growth rates when Kmax was higher than a certain critical value, decreased (or had no influence on) crack growth rates when Kmax was lower than the critical value. Most theories of HIC are based upon the original proposal of Troiano [102] about stress-induced diffusion of hydrogen to the region of high tensile triaxiality in the metal matrix ahead of a plastically strained notch or crack. According to this proposal, HIC occurs when sufficient hydrogen has concentrated at this region so that a critical combination of stress state and hydrogen concentration is attained. More recently, Doig and Jones [103] have assumed that for HIC the critical concentration of hydrogen should be achieved over an unspecified distance ahead of the crack determined by the microstructure. Akhurst [104], by analogy with brittle modes of unstable fracture (no hydrogen) in steels [105], has assumed that the critical tensile stress should be exceeded over a microstructurally significant distance ahead of the crack tip for hydrogen-induced slow crack growth to occur. He considered the distances to be in the range 10–100 µm, which covers the range of grain diameters encountered in most steels. Views on the mechanisms of hydrogen-induced FCG in high-strength steels [56,106–109] and titanium alloys [60,110,111] and HIC in magnesium alloys [71,73] incorporated Troiano’s proposal too. For example, Fig. 17(a) shows the visualized distribution of tensile stress as a function of distance ahead of the crack tip for high-strength steel 835M30 (σY = 1350 MPa) in which hydrogen-induced FCG occurs at stress-intensity factors above KISCC [107]. Also indicated by the cross-hatched region are the hydrogen concentration and distance over which this should be achieved for a hydrogen- induced increment of crack growth to occur. As the stress level changes ahead of the crack tip, there is a corresponding change in the contents of hydrogen needed to promote the increment of crack growth at the particular point. A minimum concentration of hydrogen is required at the point where the tensile stress is

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 361 highest [107]. This maximum stress is experienced at a position 2δ ahead of the crack tip [112] where δt is the crack-tip-opening displacement (i.e., total separation distance between crack surfaces at the tip) given by

2 δ t = 0.493K /EσY. (2)

The dashed lines in Fig. 17 represent the possible distribution of hydrogen that may be achieved within the plastic zone as the result of random-walk diffusion processes during contact with hydrogen gas at pressures of 200 kPa and 4.2 MPa. The relative positions of these hydrogen-content profiles show that subcritical crack growth can occur readily even at the lower pressure of hydrogen gas without the need for the additional concentration of hydrogen due to the dislocation transport mechanism [113]. Figure 17(b) shows the situation postulated for lower-strength steels 2% Ni-Cr-Mo-V (σY = 575 MPa) and 708M40 (σY = 726 MPa) when hydrogen-enhanced FCG occurs at stress- intensity factors below KISCC. Because of the lower strength of these steels, the peak of tensile stress is lower than that achieved in high-strength steel 835M30 and it is displaced further from the crack tip. The higher ductility of these steels results in a higher resistance to HIC so that higher hydrogen concentrations are required to accelerate the cracking process than for 835M30 steel. Figure 17(b) shows also that the hydrogen concentration ahead of the crack tip is not enough to provide a critical combination of stress state and hydrogen concentration in 708M40 steel tested in 200 kPa hydrogen gas. However, an increase of the gas pressure from 200 kPa to 4.2 MPa results in the increase of hydrogen concentration to the level required for promoting accelerated crack growth [107]. The concept shown in Fig. 17, though speculative, can help to explain quantitatively the non-single mode effect of cathodic polarization on corrosion FCG rates that was observed in studies [14,28]. As Oriani and Josephic [114] and Gerberich et al. [111,115] proposed, there exists a threshold stress-intensity factor (KTH) the achievement of which is required to cause a stable crack growth in a metal exposed to hydrogen atmosphere. According to Gerberich and Chen [115], the stress-intensity factor equals KTH when the equilibrium concentration of hydrogen (Ceq) ahead of the crack tip achieves the critical level of concentration (Ccr) that is required to cause HIC. However, when the stress-intensity factor is lower than KTH and Ceq < Ccr, equilibrium does not achieve the critical event and no cracking occurs at long time. More recently, Gerberich et al. [111] proposed that by analogy with KTH, there exists a fatigue threshold (Kn = KTH/α2 where α2 > 1) above which hydrogen-induced FCG can occur. This was confirmed experimentally by Suresh and Ritchie [92] who studied the role of (138 kPa) hydrogen gas on FCG rates in several classes of steels. They observed acceleration in crack growth rates (up to T 20 times) when Kmax exceeded a certain well-defined critical value (K max). The T crack growth at Kmax > K max occurred by a predominantly intergranular (IG) T mode, whereas below K max, where da/dN did not depend on hydrogen presence,

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 362 Advances in Fatigue, Fracture and Damage Assessment n (b) eel (after McIntyre [107]). McIntyre (after eel ation and the amount of hydrogen required to promote cracking i cracking promote to of hydrogen required amount and the ation (a) the crack-tip region of (a) high-strength steel and (b) low-strength st and (b) low-strength steel of (a) high-strength region crack-tip the

17: Figure hydrogen concentr of stress, distributions Postulated

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 363 the mode of crack growth was transgranular (TG). Suresh and Ritchie [92] concluded that the acceleration in crack growth rates was related to the onset of hydrogen-induced fatigue-crack propagation. In line with all the above findings, it was reasonable to suggest that Kmax, above which cathodic polarization accelerated the corrosion FCG in studies [14,28], equated the threshold value (like to KTH for HIC in hydrogen atmosphere [111,114,115]) which was required to cause the hydrogen-induced crack growth acceleration during corrosion fatigue. This suggestion is in accord with Fujii and Smith [56] who found that cathodic polarization at –1000 mVSCE (versus saturated calomel electrode, SCE) caused an approximately two-fold acceleration in the corrosion FCG of high- strength steel HY-130 in 0.6 M NaCl solution. The crack-growth acceleration occurred when ∆K exceeded 30–40 MPa m1/2 at R = 0.1 and was attributed to the appearance of HIC [56]. Taking everything stated above into consideration, it was concluded [14,28] that the critical values of Kmax that should be exceeded to cause accelerated corrosion FCG at cathodic potentials due to HIC corresponded to conditions when sufficient hydrogen has concentrated at a certain characteristic distance [103,104] ahead of the crack so that the critical combination of stress state and hydrogen concentration [102] was attained. Therefore, the critical values of Kmax as well as ∆K were regarded as corresponding to the onset of corrosion FCG according to the HIC mechanism and designated as KHIC and ∆KHIC (by analogy with threshold stress intensities KFSCC and ∆KSCC for SCC under cyclic loading [89,109,116,117]). The corresponding values of critical crack growth rates were designated as (da/dN)cr. It is important to note that the conclusion about the dominant role of HIC in corrosion-fatigue crack propagation at Kmax > KHIC (∆K > ∆KHIC) and da/dN > (da/dN)cr. was common for high-strength low-alloy steels, near-α and near-β titanium alloys, and the magnesium alloy VMD10, i.e. for materials of different base, composition and microstructure. This is consistent with basic theories of HIC [91,102,103,115,118,119] that were proposed for materials regardless of their base, composition and microstructure.

3.2.1.2 The effect of stress ratio on HIC during corrosion FCG As mentioned, the accelerating effect of cathodic polarization on crack growth rates at Kmax > KHIC was more pronounced with higher R [Fig. 14(b)]. This testified that the relative contribution of HIC (in comparison with another possible mechanism, e.g. SAD) to corrosion FCG increased with increasing R. The contribution of HIC to crack propagation was maximal under constant loading. For example, cathodic polarization at –200 mVSHE increased the corrosion FCG rate in TS6 alloy exposed to 0.6 M FeCl3 solution by a factor of 5 at R = 0.1, 30 at R = 0.5, and 70 at R = 0.8. Under static loading, the same cathodic polarization caused the growth of the crack, that was “motionless” at OCP, at the –5 rate of over 4 × 10 m/s, while KHIC was lower than KISCC. In accordance with Troiano’s proposal on HIC [102], the above observation was related to the reversible character of stress-induced diffusion of hydrogen into and out of the crack-tip region as the load is reversed. This means that if under static loading

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 364 Advances in Fatigue, Fracture and Damage Assessment hydrogen atoms intend to concentrate in a certain place in the metal matrix ahead of the crack, under cyclic loading hydrogen atoms would fluctuate due to cyclic movement of the triaxial position within a wide range [108,120] expanded with decreasing R. It seemed that the cyclic fluctuation of the triaxial position (Fig. 17) and hydrogen atoms ahead of the fatigue crack [108] made the appearance of HIC comparable with that under static loading more difficult and decreased the relative contribution of HIC to corrosion-fatigue crack propagation with decreasing R.

3.2.1.3 Effects of strength and specimen orientation When testing VMD10 alloy, KHIC decreased as the yield strength of the alloy increased. This finding agreed with general trends toward increasing susceptibility of materials to HIC with increasing strength. However, the strength was not the primary criteria for susceptibility of magnesium alloys to HIC. For example, IMV7-1 alloy after ageing at 220°C for 24 h (applied to increase its susceptibility to HIC) was significantly stronger than VMD10 alloy (HRB 30), but the higher strength of IMV7-1 alloy (HRB 65) did not result in the acceleration of corrosion FCG under cathodic polarization due to HIC. Cathodic polarization always retarded corrosion FCG in this alloy. That is, susceptibility of IMV7-1 alloy to HIC during corrosion fatigue was less evident than in VMD10 alloy [121]. It seemed that susceptibility to HIC depended on a number of variables rather and not only on the strength of magnesium alloys. Among the variables are the chemical composition and microstructure of the alloy, the solution pH, the ratio of activating/passivating species, the electrode potential, which are responsible for kinetics of crack-corrosion processes and the properties of the surface film at the crack tip (including its hydrogen permeability), as well as the Kmax value were of primary importance [14]. In addition, a strong dependence between KHIC and specimen orientation was observed and it showed strong anisotropy of susceptibility to HIC. In particular, for a short longitudinal (SL) specimen KHIC was lower than for a LT specimen by a factor of 2.7 when testing VMD10 alloy in 1 M NaOH solution at –1800 mVSHE and by a factor of 2.9 when testing VMD10 alloy in 2 M H2CrO4 solution at –800 mVSHE; where R = 0.5 [121]. Data on the relationship between specimen orientation of magnesium alloys and their susceptibility to HIC was not found in the literature, but Moccari and Shastry [72] reported a slight effect of specimen orientation on KISCC of magnesium alloy AZ61 tested in 0.6 M NaCl + 0.1 M K2CrO4 solution.

3.2.2 Incubation period for hydrogen-induced corrosion FCG As mentioned, accelerated crack growth at Kmax > KHIC did not begin directly after cathodic polarization was applied but only after some time delay (Fig. 13). Delays of 10 to 150 s were not associated with the time required for the penetration of polarization inside the crack because in highly conductive electrolytes, such as those used in studies [14,28], the change in the crack-tip potential occurs once the potential is applied at the external surface of the specimen. The delays seemed to reflect the kinetics for processes associated with

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 365 an incubation period interpreted by Troiano [102] as the time required for the critical amount of hydrogen to concentrate in the highly strained region ahead of the crack. Broadly speaking, these delays were associated with an incubation period for the beginning of the corrosion FCG according to the HIC mechanism. After the incubation period, the hydrogen-induced corrosion-fatigue crack grew with the steady-state rate (da/dN)p at a given Kmax because cathodic polarization provided steady-state generation of hydrogen and, as a result, hydrogen charging of specimens [87]. After cathodic polarization, i.e. the intense surge of hydrogen, was removed and potential was returned to OCP, crack growth rates in steels and VMD10 alloy returned to initial values (da/dN)i. When testing titanium alloys, crack growth rates after cathodic polarization was removed and potential was returned to OCP significantly exceeded the initial value (da/dN)i for some time. This (residual) effect of cathodic polarization on crack growth rates was due to the greater tendency of titanium and its alloys to the formation of brittle hydrides (particularly under charging conditions) and embrittlement when exposed to a hydrogen-producing environment or gaseous hydrogen [95,96,110,122–128].

3.2.3 Fractography The conclusion about the dominant role of HIC in corrosion-fatigue crack propagation at Kmax > KHIC was supported by fractographic observations. The features that were observed on the fracture surfaces produced by accelerated corrosion FCG in high-strength steels and titanium alloys at Kmax > KHIC were similar to those found for the materials tested under cyclic loading in hydrogen gas or after preliminary hydrogen charging [93–95,124,129,130]. In particular, IG cracking mixed with patches of TG cracking and some evidence of inelastic incremental crack advance occurring along grain boundaries was observed in high-strength steels [Fig. 18(b)]. This is typical for HIC of steels of this type. In general, the hydrogen-induced corrosion FCG in martensitic steels at cathodic potentials occurs intergranularly along prior austenite grain boundaries [51,58,62,131]. Frandsen and Marcus [129] found a good correlation between the acceleration in FCG of high-strength steel HP-9-4- 20 tested in (13 kPa) hydrogen gas relative to vacuum and the proportion of IG cracking on the fracture surface (Fig. 19) They proposed that prior austenite grain boundaries might be a preferred fracture path for HIC due to stress concentration created by blocked slip bands at the boundary [132] or due to the presence of various impurities that acted as hydrogen recombination poisons [133]. It is now widely accepted that high-strength martensitic steels generally exhibit brittle fracture along prior austenite grain boundaries when stressed in hydrogen gas or hydrogen-producing environments [119,133,134]. The tendency to IG cracking is directly related to the concentration of trace impurities like P, S, and Sn [119,135,136] which are known to reduce IG cohesion in iron-based alloys [137].

Figure 18: SEM fractographs of 32Cr3NiMoV steel tested in 0.05 M Na2B4O7 solution at: (a) OCP of –50 mVSHE and (b) at cathodic potential of – 1/2 –7 1000 mVSHE; Kmax = 26 MPa m , R = 0.5, (da/dN)i = 5.4 × 10 –6 m/cycle, and (da/dN)p = 4 × 10 m/cycle. The overall direction of crack propagation from top to bottom of the fractographs approximately (after Shipilov [14]).

(a)

Figure 19: (a) FCG rates in HP-9-4-20 steel tested in vacuum and low-pressure hydrogen, and (b) the increase in crack growth rates and the associated percentage of IG fracture in hydrogen gas. Here Rpr is the reverse yield plastic-zone diameter and rpf is the forward yield plastic-zone radius (after Frandsen and Marcus [129]).

(b)

In addition, McMahon et al. [119,136] demonstrated that alloying elements such as Mn, Si and Ni promote segregation of P and S at austenite grain boundaries. In steels tested in the study [14], all those alloying elements and impurities (except for Sn, which was immediately discernable) were present. It was suspected that they promoted the IG propagation of hydrogen-induced corrosion-fatigue crack in steels at Kmax > KHIC. Corrosion FCG in titanium alloys under cathodic polarization at Kmax > KHIC occurred in a more brittle manner than under open circuit. Figure 20(b) shows that many typical flat cleavage facets with characteristic “river patterns” and fissures perpendicular to the fracture surface and parallel to the main crack- growth direction were evident after accelerated corrosion FCG under cathodic polarization [28]. Hack and Leverant [124] and, more recently, Nakasa and Satoh [95], performed microstructural characterisation and fractographic

Figure 20: SEM fractographs of VT20 alloy tested in 0.6 M FeCl3 solution at: (a) OCP of +750 mVSHE and (b) at cathodic potential of –200 mVSHE; 1/2 –7 Kmax = 68 MPa m , R = 0.5, (da/dN)i = 8.5 × 10 m/cycle, and –5 (da/dN)p = (5 to 9) × 10 m/cycle. The overall direction of crack propagation from top to bottom of the fractographs approximately (after Shipilov [28]). analysis of fracture surfaces of titanium alloys fatigue tested under and after hydrogen charging. Almost complete brittle fracture and numerous cleavage facets on the fracture surface of hydrogen-charged IMI-685 alloy were associated with HIC [124]. The fracture surface of Ti-13V-11Cr-3Al alloy cathodically charged with hydrogen revealed the cleavage fracture with frequent “river patterns” that corresponded to the decrease in the cohesive strength of the cleavage plane of β-phase by the hydrogen solution [95]. The mechanisms that could cause a cleavage of titanium lattice due to the action of hydrogen have been discussed by Meyn [125], and by Powell and Scully [138]. TG cracking was revealed on the fracture surfaces of VMD10 alloy after testing in (0.1–1) M NaOH, 0.02 M Na2HPO4, and (0.5–2) M H2CrO4 solutions

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 370 Advances in Fatigue, Fracture and Damage Assessment in which the non-single mode effect of cathodic polarization on crack growth rates was observed. The morphology of fracture surfaces virtually did not depend on the potential and specimen orientation when Kmax exceeded KHIC. It seemed likely that this was related to the fact that the acceleration effect of cathodic polarization on corrosion FCG rates at Kmax > KHIC was less pronounced in VMD10 alloy than in high-strength low-alloy steels and titanium alloys.

3.2.4 SAD during corrosion-fatigue crack propagation The appearance of two fractographic modes of corrosion FCG at Kmax > KHIC and Kmax < KHIC [14,28] suggested that there might be two different mechanisms of crack growth under these conditions. It seemed that, at Kmax < KHIC the critical combination of stress state and hydrogen concentration has not been achieved ahead of the growing crack. Because of this, although the applied cathodic potential was far below the hydrogen equilibrium potential [139] and the hydrogen content ahead of the crack tip could be higher than under open-circuit conditions, the accelerated crack growth due to HIC did not occur. On the other hand, cathodic polarization retarded corrosion FCG in high-strength steels and titanium alloys at Kmax < KHIC. Although the results were obtained in passivating solutions in which SCC due to SAD was suppressed as evident in the increase of KISCC, it is reasonable to suggest that a SAD-based mechanism played a dominant role in the corrosion-fatigue crack propagation at Kmax < KHIC. This suggestion is supported by the visible displacement of OCP (relative to the OCP of an unloaded specimen) during corrosion FCG as well as its periodical fluctuation with the cyclic loading [14,28]. For example, when testing VT20 alloy in 0.5 M H2CrO4 + 0.1 M NaCl solution, OCP was ennobled slightly at da/dN < 1.3 × 10–7 m/cycle but OCP was displaced cathodically at higher da/dN such that its mean displacement reached 350 mV at da/dN = 10–5 m/cycle. In addition, during the opening part of each cycle, OCP became more negative due to the formation of bare (film-free) metal surface at the crack tip. After reaching the maximum load of stress cycle, OCP returned—due to repassivation—to its previous value displaced relative to the OCP of an unloaded specimen. For VT20 alloy tested in 0.5 M H2CrO4 + 0.1 M NaCl solution, the maximum cyclic fluctuation of OCP during a fatigue cycle was about 30 mV for higher crack growth rates [28]. Independently of the Kmax value, SAD was also the main mechanism of corrosion FCG in titanium alloys when tested in 0.5 M NaCl, (0.1–1) M NaOH with and without NaCl, (0.1–1) M KOH with and without NaCl, and (0.5–2) M H2CrO4. In these solutions, cathodic polarization retarded and even stopped corrosion FCG, whereas anodic polarization accelerated crack growth [28]. It seemed that the crack-growth mechanism in VMD10 alloy tested in (0.1–1) M NaOH, 0.02 M Na2HPO4, and (0.5–2) M H2CrO4 solutions at Kmax < KHIC was not electrochemical or the role of electrochemical processes was not significant in the crack propagation because da/dN data obtained in the solutions were like that of da/dN data in air and cathodic polarization had no influence on crack growth rates for a time up to 1500 s. (da/dN data obtained in air cannot be used

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) Advances in Fatigue, Fracture and Damage Assessment 371 as a reference to compare the effect of aqueous solutions on the FCG behaviour of magnesium alloys because laboratory air is an aggressive environment for magnesium and its alloys in comparison with vacuum [140].) Because cathodic polarization retarded corrosion FCG in IMV7-1 alloy whereas anodic polarization accelerated it, and the polarization effect did not depend on solution composition, applied potential and Kmax value, it is suggested that SAD played a dominant role in the corrosion-fatigue crack propagation of this alloy.

3.2.5 Concluding remarks In studies [14,28], the conclusion that one of the mechanisms played a dominant role in the corrosion-fatigue crack propagation did not deny the possibility that other, less evident, mechanisms were involved. For example, although cathodic polarization retarded and even stopped corrosion FCG in VT20 alloy tested in 0.5 M NaCl solution (and this fact caused difficulty in the acceptance of the HIC mechanism), many features of the fracture surfaces produced when testing VT20 alloy in this solution at OCP were similar to those found for VT20 alloy tested in 0.6 M FeCl3 solution at Kmax > KHIC when HIC was the main mechanism of corrosion FCG [28]. It seemed that the similarity between the features of the fracture surfaces was due to coincidence of crack growth rates (3 × 10–5 m/cycle –5 in 0.5 M NaCl and 5.3 × 10 m/cycle in 0.6 M FeCl3) rather than to a coincidence of crack-growth mechanisms. As demonstrated by Lynch [141], the features of fracture surfaces produced under very different conditions can be quite similar. The observation that the features of fracture surfaces depend on the level of crack growth rates rather than on the crack-growth mechanism suggests that fracture surfaces should not be identified with a crack-growth mechanism without additional evidence, including the effect of controlled (applied) potential on the crack growth rate.

3.3 Location of the fracture-process zone for HIC during corrosion FCG

The results obtained in previous studies [14,28], allowed for a comparison between hydrogen-induced corrosion FCG rates that have been presented as average increments of crack growth per cycle, ∆ap, and critical distances ahead of the growing crack at Kmax ≥ KHIC [142]. These distances included the distance to the location of maximum tensile stress ahead of the crack, the sizes of cyclic and monotonic plastic zones, the depths of hydrogen penetration due to lattice diffusion and dislocation sweeping. For the purposes of the comparison, critical distances ahead of the crack were calculated from the equations of continuum solid mechanics. In particular, as Rice [143] suggested with special reference to HIC, the distance ahead of the crack tip at which the stress peak occurs is equal to approximately twice the crack-tip-opening displacement. Therefore, the distance to the location of the stress peak ahead of the crack corresponding to Kmax was estimated as [112,144]:

2 2δmax ≈ (Kmax) /EσY (3)

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 372 Advances in Fatigue, Fracture and Damage Assessment where δmax is the crack-tip-opening displacement corresponding to Kmax; and E is Young’s modulus. In (3), to approximate cyclic hardening or softening, the cyclic yield strength should be used in place of the monotonic yield strength, σY [37]; at room temperature, however, the cyclic yield strength is approximately equal to σY [145]. Two types of plastic zones are distinguished at the crack tip when the fatigue- crack propagation and fracture-process zone behaviour are considered on a continuum-mechanical basis [146]. These are a monotonic plastic zone denoted 2 as the Kmax zone whose radius is proportional to (Kmax/σY) and a cyclic plastic 2 zone denoted as the ∆K zone whose radius is proportional to (∆K/σY) . The monotonic plastic-zone size, rx, and cyclic plastic-zone size, rxc, directly ahead of the crack tip were estimated as [147]:

2 rx = 0.03(Kmax/σY) (4)

2 rxc = 0.0075(∆K/σY) . (5)

The maximum depth of hydrogen penetration due to lattice diffusion was estimated as:

xD = 4√DHt, (6) where DH is the hydrogen diffusion coefficient (here, at room temperature); and t is a half-cycle time period (i.e., the duration of plastic deformation). For –6 tempered martensitic steels of 39Ni4MoV and 32Cr3NiMoV type, DH ≈ 10 2 –9 2 cm /s [94]. For VT20 alloy, DH ≈ 10 cm /s [148]. For TS6 (near-β) alloy, DH was taken between 10–7 cm2/s and 10–6 cm2/s [96]. Data on the kinetics of hydride formation in a Mg-2%Ce alloy [149] suggested that DH for VMD10 alloy could be equal to ∼ 10–9 cm2/s. It was suspected that during hydrogen-induced FCG in ductile alloys, hydrogen atoms penetrate deeply into the plastically deformed region ahead of the growing crack [108]. In particular, Tien et al. [113,150,151] concluded that hydrogen atoms could penetrate much deeper into the plastic zone through dislocation sweep-in than through hydrogen diffusion. Granting this, the maximum depth of hydrogen penetration due to dislocation sweeping was estimated as:

xc = (DH/kT)(EB/30b)t, (7) where k is Boltzmann constant; T is the absolute temperature; EB is the binding energy of 0.3 eV at room temperature; and b is the Burgers vector [150].

, x

r b c x 0.7 0.7 0.7 (mm) 653.4 653.4 653.4 326.7 , hydrogen-

c D m) x x 2.8 2.8 2.8 89.4 89.4 89.4 63.3 ( µ

x m) r 0.9 1.4 7.9 57.6 70.1 ( µ 183.3 140.1

max m) δ 0.22 0.35 2.11 ( µ 2 53.15 29.59 13.04 26.06 Compared distances Compared distances

m) xc r 0.01 0.01 0.50 1.83 3.60 4.38 8.76 ( µ

p m) a 1.8 4.5 4.0 0.8 7.2 ∆ 90.0 ( µ 200.0 crack advances and critical distances , hydrogen-induced increment of crack growth per cycle; p [142]. a a ) ∆ HIC SHE 200 200 900 200 300 K CP 1800 1000 − − − − − − − > (mV

, the distance to location of stress peak ahead crack; . max x r , hydrogen-penetration depth due to lattice diffusion; and K max D δ x ) ½

max 8.3 10.4 26.0 68.0 67.5 11.6 16.4 K (MPa m R 0.8 0.8 0.5 0.8 0.5 0.5 0.5 ahead of growing crack at

4 4

O 4 3 3 4 B 2 CrO CrO 2 2 CrO 2 Table 2: Comparison between hydrogen-induced (mol/L) 1 NaOH Solution 0.6 FeCl 0.6 FeCl 1 H 0.5 H 0.5 H 0.05 Na is limited by the monotonic plastic-zone size, c x TS6 TS6 VT20 VMD10 VMD10 Material , cyclic plastic-zone size directly ahead of the crack; 2 crack; the of ahead , cyclic plastic-zone size directly 39Ni4MoV 39Ni4MoV CP, applied cathodic potential; SHE, standard hydrogen electrode; hydrogen CP, applied cathodic potential; SHE, standard xc 32Cr3NiMoV Distance

monotonic plastic-zone size directly ahead of the crack; a b penetration depth due to dislocation sweeping. r

3.3.1 Comparing theory and experimental data Table 2 shows some of the results obtained from experiments and calculated using the above equations. The values of Kmax given in Table 2 exceeded the critical values of KHIC by about 5–10% for titanium alloys, 10–25% for steels, and 20–25% for VMD10 alloy. Six possible relationships were discovered from the comparison between ∆ap and critical distances ahead of the growing crack [142]:

(1) ∆ap was lower than the hydrogen-penetration depth due to lattice diffusion, the cyclic plastic-zone size, and the distance to the location of the stress peak ahead of the crack (e.g., for VMD10 at Kmax = 11.6 MPa m1/2),

(2) ∆ap was longer than the hydrogen-penetration depth due to lattice diffusion but lower than the cyclic plastic-zone size, the distance to the location of the stress peak ahead of the crack, and the hydrogen- penetration depth due to dislocation sweeping (e.g., for VMD10 at Kmax = 16.4 MPa m1/2),

(3) ∆ap was longer than the cyclic plastic-zone size and the distance to the location of the stress peak ahead of the crack, but lower than the monotonic plastic-zone size and hydrogen-penetration depth due to lattice diffusion (e.g., for 32Cr3NiMoV steel),

(4) ∆ap was longer than the cyclic plastic-zone size, the hydrogen- penetration depth due to lattice diffusion, and the distance to the location of the stress peak ahead of the crack but lower than the monotonic plastic-zone size and hydrogen-penetration depth due to dislocation sweeping (e.g., for VT20 alloy),

(5) ∆ap was longer than the distance to the location of the stress peak ahead of the crack and monotonic plastic-zone size but lower than the hydrogen-penetration depth due to lattice diffusion (e.g., for 39Ni4MoV steel), and

(6) ∆ap was longer than the distance to the location of the stress peak ahead of the crack, the monotonic plastic-zone size, and the hydrogen- penetration depth due to lattice diffusion but lower than the calculated depth of hydrogen penetration due to dislocation sweeping (e.g., for TS6 alloy) that in fact is limited by the monotonic plastic-zone size [151].

Cases one and two characterized the hydrogen-induced corrosion FCG of

VMD10 alloy. As shown in Table 2, ∆ap = 0.8 µm and rxc = 4.38 µm at Kmax = 1/2 1/2 11.6 MPa m , and ∆ap = 7.2 µm and rxc = 8.76 µm at Kmax = 16.4 MPa m . Always, ∆ap was lower than 2δmax and rxc. The latter shows that the fracture- process zone was within the cyclic plastic-zone (Fig. 21) [152]. It seemed that the hydrogen transport to the fracture-process zone was made possible either by 1/2 lattice diffusion (as at Kmax = 11.6 MPa m ) or by dislocation sweeping (as at 1/2 Kmax = 16.4 MPa m ).

Cases three and five characterized the hydrogen-induced corrosion FCG of steels. Case three was in good coincidence with all models of HIC and hydrogen-induced FCG as regards the location of fracture-process zone within the monotonic plastic-zone (Fig. 22) [152]. Although ∆ap exceeded the distance to the location of the stress peak ahead of the crack, this did not conflict seriously with the HIC models. At first glance, case five did not agree with theoretical

Figure 21: The location of fracture-process zone within the cyclic plastic-zone

when ∆ap < rxc. For VMD10 alloy, ∆ap < xD < rxc < 2δmax < rx at Kmax 1/2 = 11.6 MPa m and xD < ∆ap < rxc < 2δmax < rx at Kmax = 16.4 MPa m1/2 (after Shipilov [152]).

Figure 22: The location of fracture-process zone within the monotonic plastic

zone when ∆ap < rx. For 32Cr3NiMoV steel, rxc < 2δmax < ∆ap < rx < 1/2 xD at Kmax = 26 MPa m and for VT20 alloy, rxc < xD < 2δmax < ∆ap < 1/2 rx at Kmax = 68 MPa m (after Shipilov [152]).

WIT Transactions on State of the Art in Science and Engineering, Vol 1, © 2005 WIT Press www.witpress.com, ISSN 1755-8336 (on-line) 376 Advances in Fatigue, Fracture and Damage Assessment views on HIC and the hydrogen-induced FCG mechanisms because ∆ap > rx (Fig. 23) [152]. But a closer look indicated that this could be explained by the fact that (da/dN)p was too high, so that the growing crack crossed the location of the stress peak and the monotonic plastic-zone size rapidly. Because xD was far longer than ∆ap, the hydrogen transport in steels could have resulted from lattice diffusion. Cases four and six characterized the hydrogen-induced corrosion FCG of titanium alloys. For VT20 alloy (case four), the fracture process zone was within the monotonic plastic zone (∆ap < rx), as in Fig. 22. Case six (TS6 alloy at Kmax = 67.5 MPa m1/2) did not agree with conventional theoretical views on HIC and hydrogen-induced FCG mechanisms. It was impossible to explain why ∆ap exceeded all critical distances ahead of the growing crack including the distance

Figure 23: The location of fracture-process zone beyond the monotonic plastic

zone when ∆ap > rx. For 39Ni4MoV steel, rxc < 2δmax < rx < ∆ap < xD 1/2 at Kmax = 8.3 MPa m and for TS6 alloy, rxc < 2δmax < rx < xD < ∆ap 1/2 at Kmax = 67.5 MPa m (after Shipilov [152]). to the location of the stress peak, the cyclic and monotonic plastic-zone sizes (Fig. 23), and the hydrogen-penetration depth due to lattice diffusion. (For TS6 alloy tested in an “inert” atmosphere, formal plane-strain conditions were maintained at K ≤ 53.4 MPa m1/2 based on a criterion that specimen thickness B 2 ≥ 2.5(KIC/σY) [153]. However, examination of the fracture surface did not reveal the features of plane stress for TS6 alloy tested at Kmax values reached.) The fact that ∆ap < xc was not an acceptable explanation is because in fact the depth xc is limited by the monotonic plastic-zone size [151] that was much lower than ∆ap in TS6 alloy. It seemed that dislocation sweeping provided mainly for hydrogen transport in titanium alloys. Although hydrogen transport by lattice diffusion in the more open bcc β-phase structure is much more rapid than in α- phase structure [154], the evidence for the lattice diffusion of hydrogen in TS6 (near-β) alloy was less convincing.

It was found [142] that the lower the yield strength of material, the better was the correlation between ∆ap (i.e., the real position of the fracture-process zone) and plastic-zone sizes calculated from (4) and (5). For example, for VMD10 alloy with σY = 240 MPa, ∆ap was lower than the cyclic plastic-zone size that was so large due to the lower level of σY. For VT20 alloy with σY = 870 MPa, ∆ap < rx but for TS6 alloy with σY = 1540 MPa, ∆ap > rx. The position of the hydrogen-induced corrosion FCG process zone did not always correlate with plastic-zone sizes. Moreover, there was no correlation between ∆ap and the distance to the location of the stress peak ahead of the crack.

3.3.2 Concluding remarks Six possible combinations of the relationship between crack hydrogen-induced corrosion FCG rates (increments of crack growth per cycle, ∆ap) and critical distances ahead of the crack were observed [142]. Some of these combinations did not agree with well-known and conventional theoretical views on HIC and hydrogen-induced FCG mechanisms. For example, there was no correlation between ∆ap and the distance to the location of the stress peak, 2δmax, corresponding to Kmax ahead of the crack. The position of the hydrogen-induced corrosion FCG process zone did not always correlate with plastic-zone sizes and the lower the yield strength of the material, the better was this correlation. The increment of hydrogen-induced corrosion FCG per cycle could be: lower than the cyclic plastic-zone size (e.g., for VMD10 alloy); larger than the cyclic plastic-zone size but lower than the monotonic plastic-zone size (e.g., for 32Cr3NiMoV steel and VT20 alloy); and larger than the monotonic plastic-zone size (e.g., for 39Ni4MoV steel and TS6 alloy).

3.4 Summary

A new successful method for quantitatively defining the role of HIC and SAD in corrosion-fatigue crack propagation was proposed in recent studies [14,28]. The experimental approach intended to create short-term stimulation of electrochemical reactions within a growing crack. For this purpose, anodic or cathodic potential was applied to the specimen for a time period in which it was possible to register the change in the crack growth rate corresponding to OCP and to measure the crack growth rate under polarization. The time rarely exceeded 300 s. The approach made possible the observation of a non-single mode effect of cathodic polarization on corrosion-fatigue FCG rates. An important point was that cathodic polarization accelerated crack growth when Kmax and the corresponding crack growth rate, da/dN, exceeded certain well- defined critical values characteristic for a given material–solution system. When Kmax and da/dN were lower than the critical values, the same cathodic polarization retarded or had no obvious influence on crack growth. The results and fractographic observations suggested that the acceleration in crack growth under cathodic polarization was due to HIC. Therefore, critical values of Kmax, as well as ∆K and da/dN were regarded as characterizing the onset of corrosion

FCG according to HIC mechanism. They were designated as KHIC, ∆KHIC and (da/dN)cr. HIC was the main mechanism of corrosion FCG at Kmax > KHIC (∆K > ∆KHIC) and da/dN > (da/dN)cr. A SAD-based mechanism played a dominant role in the corrosion-fatigue crack propagation at Kmax < KHIC and da/dN < (da/dN)cr. In addition, SAD was the main mechanism of corrosion FCG in titanium alloys and magnesium alloys in most solutions investigated in which cathodic polarization retarded and even stopped corrosion FCG [14,28]. The method proved applicable to high-strength low-alloy steels, titanium alloys, and the magnesium alloy VMD10, i.e. to materials of a different base, composition and microstructure. This is consistent with basic theories of HIC [91,102,103,115,118,119] that were proposed for materials regardless of their base, composition and microstructure. This method allowed for studying the influence of metallurgical, mechanical and environmental variables on the role of HIC in corrosion-fatigue crack propagation. For example, it was shown that the relative contribution of HIC (in comparison with another possible mechanism, e.g. SAD) to corrosion-fatigue crack propagation increased with increasing R. Under static loading (R = 1), HIC was more pronounced than under cyclic loading (0.1 < R < 0.9). Also, the method [14,28] allowed for a comparison between the increment of hydrogen-induced corrosion FCG per cycle and critical distances ahead of a growing crack were [142]. Critical distances included the distance to the location of maximum tensile stress ahead of the crack, the sizes of cyclic and monotonic plastic zones, the depths of hydrogen penetration due to lattice diffusion and dislocation sweeping. Some of the results that were obtained from this comparative study [142] challenged well-known and conventional (beginning in the late 1950s) theoretical views on HIC mechanisms. It seems that the concept of corrosion FCG mechanism that was proposed in studies [14,28,142] subsequent to devising the above method for quantitatively defining the role of HIC and SAD in corrosion-fatigue crack propagation can serve as a basis for the creation of a theory of corrosion FCG in metals.

4 Acknowledgments

The author acknowledges the assistance of the library staff of the University of Calgary in securing reprints of early 20th century documents relevant to this chapter and the assistance of Enoch Ng of the University of Calgary with the preparation of figures. The financial support of the Natural Sciences and Engineering Research Council (NSERC) of Canada is especially appreciated.

References

[1] Haigh, B.P., Experiments on the fatigue of brasses. Journal of the Institute of Metals, 18, pp. 55–77, 1917.

[2] Shipilov, S.A., Environment-assisted cracking of materials as a significant cause of engineering systems malfunctions. Technology, Law and Insurance, 1, pp. 131–142, 1996. [3] Langdon, S.C., Discussion. Transactions of the American Electrochemical Society, 36, p. 131, 1919. [4] McAdam, D.J., Jr., Stress-strain-cycle relationship and corrosion-fatigue of metals. Proceedings of the ASTM, 26, pp. 224–254, 1926. [5] Lehmann, G.D., The variation in the fatigue strength of metals when tested in the presence of different liquids. Aeronautical Research Committee (Great Britain), Report and Memoranda, No. 1054, 1926. [6] Speller, L.N., Discussion. Proceedings of the ASTM, 26, p. 276, 1926. [7] Binnie, A.M., The influence of oxygen on corrosion fatigue. Aeronautical Research Committee (Great Britain), Report and Memoranda, No. 1244, 1929. [8] Gough, H.J., Corrosion-fatigue of metals. Journal of the Institute of Metals, 49, pp. 17–92, 1932. [9] Evans, U.R. & Simnad, M.T., The mechanism of the corrosion fatigue of mild steel. Proceedings of the Royal Society of London, A188, pp. 372– 392, 1947. [10] Evans, U.R., The electrochemistry of corrosion fatigue. Symposium on the Failure of Metals by Fatigue, University Press: Melbourne, pp. 84–90, 1947. [11] Wei, R.P & Shim, G., Fracture mechanics and corrosion fatigue. Corrosion Fatigue: Mecahnics, Metallurgy, Electrochemistry, and Engineering, eds. T.W. Crooker & B.N. Leis, ASTM SPT 801, ASTM: Philadelphia, pp. 5–25, 1983. [12] Glaesser, W. & Wright, I.G., Mechanically assisted degradation. ASM Handbook, Vol. 13: Corrosion, 9th edn., ASM International: Materials Park, pp. 136–144, 1998. [13] Wang, Y.-Z., Corrosion fatigue (Chapter 13). Uhlig’s Corrosion Handbook, 2nd edn., ed. R.W. Revie, John Wiley & Sons: New York, pp. 221–231, 2000. [14] Shipilov, S.A., Mechanisms for corrosion fatigue crack propagation. Fatigue and Fracture of Engineering Materials and Structures, 25, pp. 243–259, 2002. [15] Shipilov, S.A. & Le May, I., Materials aging and structural reliability of buried pipelines having cathodic protection. In preparation for Engineering Failure Analysis. [16] Schütz, W., Fatigue life prediction – a review of the state of the art. Structural Failure, Product Liability and Technical Insurance, ed. H.P. Rossmanith, Elsevier Science: Amsterdam, pp. 49–60, 1993. [17] Speidel, M.O., Stress corrosion cracking and corrosion fatigue fracture mechanics. Corrosion in Power Generating Equipment, eds. M.O. Speidel & A. Atrens, Plenum Press: New York, pp. 85–130, 1984.

[18] Johnson, H.H. & Willner, A.M., Moisture and stable crack growth in a high strength steel. Applied Materials Research, 4, pp. 34–40, 1965. [19] Gough, H.J. & Sopwith, D.G., Atmospheric action as a factor in fatigue of metals. Journal of the Institute of Metals, 49(2), pp. 93–122, 1932. [20] Dahlberg, E.P, Fatigue-crack propagation in high-strength 4340 steel in humid air. Transactions of the ASM, 58, pp. 47–53, 1965. [21] Hartman, A., On the effect of oxygen and water vapor on the propagation of fatigue cracks in 2024-T3 alclad sheet. International Journal of Fracture Mechanics, 1(3), pp. 167–187, 1965. [22] Van der Sluys, W.A., Effects of repeated loading and moisture on the fracture toughness of SAE 4340 steel. Journal of Basic Engineering, Transactions of the ASME, 87, pp. 363–373, 1965. [23] Hudson, C.M. & Seward, S.K., A literature review and inventory of the effects of environment on the fatigue behavior of metals. Engineering Fracture Mechanics, 8, pp. 315–329, 1976. [24] Yuen, J.L., Schmidt, C.G. & Roy, P., Effects of air and inert environments on the near threshold fatigue crack growth behavior of alloy 718. Fatigue and Fracture of Engineering Materials and Structures, 8, pp. 65–76, 1985. [25] Liaw, P.K., Leax, T.R. & Donald, J.K., Fatigue crack growth behavior of 4340 steels. Acta Metallurgica, 35, pp. 1415–1432, 1987. [26] Shimojo, M., Higo, Y. & Oya-Seimiya, Y., Effects of inert gases on fatigue crack growth and their transportation into subsurface regions in titanium. Metallurgical and Materials Transactions A, 31, pp. 1435– 1441, 2000. [27] Wöhler, A., Berichte über die versuche, welche auf der königl. Niederschlesisch-märkischen eisenbahn mit apparaten zum messen der biegung und verdrehung von eisenbahnwagen-achsen während der fahrt angestellt wurden. Zeitschrift für Bauwesen, 8, pp. 642–652, 1859. [28] Shipilov, S.A., Corrosion fatigue crack growth behavior of titanium alloys in aqueous solutions. Corrosion, 54, pp. 29–39, 1998. [29] Dorey, S.F., Transactions of the Institute of Naval Architects, 75, p. 208, 1933 (quoted in [30]). [30] Evans, U.R., The Corrosion and Oxidation of Metals: Scientific Principles and Practical Applications, Edward Arnold: London, 1961. [31] Harvey, W.E., Zinc as a protective coating against corrosion fatigue of steel. Metals and Alloys, 1, pp. 458–461, 1930. [32] Gilbert, P.T., Corrosion-fatigue. Metallurgical Reviews, 1, pp. 379–417, 1956. [33] DeForest, A.V., The rate of growth of fatigue cracks. Journal of Applied Mechanics, 3, pp. A.23–A.25, 1936. [34] Gould, A.J. & Evans, U.R., A scientific study of corrosion fatigue. Preliminary report on experiments at Cambridge University. Iron and Steel Institute, Special Report, No 24, p. 325, 1939.

[35] Speidel, M.O., Influence of environment on fracture. Advances in Fracture Research (Fracture 81), ed. D. Francois, Vol. 6, Pergamon Press: Oxford, pp. 2685–2704, 1982. [36] Tahon, C. & Clerbois, L., Applications of fracture mechanics to corrosion problems in the chemical industry. Corrosion Prevention in the Process Industries, ed. R.N. Parkins, NACE: Houston, pp. 153–161, 1990. [37] Gangloff, R.P., Corrosion fatigue crack propagation in metals. Environment-Induced Cracking of Metals, eds. R.P. Gangloff & M.B. Ives, NACE: Houston, pp. 55–106, 1990. [38] Paris, P.C., Gomez, M.P & Anderson, W.E., A rational analytical theory of fatigue. The Trend of Engineering, 13, pp. 9–14, 1961. [39] Paris, P.C. & Erdogan, F., A critical analysis of crack propagation laws. Journal of Basic Engineering, Transactions of the ASME, 85, pp. 528– 534, 1963. [40] Brown, B.F., A new stress-corrosion cracking test for high-strength alloys. Materials Research and Standards, 6, pp. 129–133, 1966. [41] Brown, B.F. & Beachem, C.D., A study of the stress factor in corrosion cracking by use of the pre-cracked cantilever beam specimen. Corrosion Science, 5, pp. 745–750, 1965. [42] Wiederhorn, S.M., Influence of water vapor on crack propagation in soda- lime glass. Journal of the American Ceramic Society, 50, pp. 407–417, 1967. [43] Sprowls, D.O., Evaluation of corrosion fatigue. ASM Handbook, Vol. 13: Corrosion, 9th edn., ASM International: Materials Park, pp. 291–302, 1998. [44] Standard Test Method for Measurement of Fatigue Crack Growth Rates, ASTM E647, Vol. 03.01, ASTM: West Conshohocken, pp. 615–657, 2003. [45] Shipilov, S.A., Advanced testing methods for studying the mechanical behavior of materials. JOM, 57(3), pp. 36–42, 2005. [46] McEvily, A.J. & Wei, R.P., Fracture mechanics and corrosion fatigue. Corrosion Fatigue: Chemistry, Mechanics and Microstructure, eds. O. Devereux, A.J. McEvily & R.W. Staehle, NACE: Houston, pp. 381–395, 1972. [47] Austen, I.M. & Walker, E.F., Quantitative understanding of the effects of mechanical and environmental variables on corrosion fatigue crack growth behaviour. The Influence of Environment on Fatigue, The Institute of Mechanical Engineers: London, pp. 1–10, 1977. [48] Speidel, M.O., Blackburn, M.J., Beck, T.R. & Feeney, J.A., Corrosion fatigue and stress corrosion crack growth in high strength aluminum alloys, magnesium alloys, and titanium alloys exposed to aqueous solutions. Corrosion Fatigue: Chemistry, Mechanics and Microstructure, eds. O. Devereux, A.J. McEvily & R.W. Staehle, NACE: Houston, pp. 324–345, 1972.

[49] Feeney, J.A., McMillan, J.C. & Wei, R.P., Environmental fatigue crack propagation of aluminium alloys at low stress intensity levels. Metallurgical Transactions, 1, pp. 1741–1757, 1970. [50] Newman, R.C. & Procter, R.P.M., Stress corrosion cracking: 1965–1990. British Corrosion Journal, 25, pp. 259–269, 1990. [51] Barsom, J.M., Mechanisms of corrosion fatigue below KISCC. International Journal of Fracture Mechanics, 7, pp. 163–182, 1971. [52] Gallagher, J.P., Corrosion fatigue crack growth rate behavior above and below KISCC. Journal of Materials, 6, pp. 941–964, 1971. [53] Vosikovsky, O., Frequency, stress ratio, and potential effects on fatigue crack growth of HY130 steel in salt water. Journal of Testing and Evaluation, 6, pp. 175–182, 1978. [54] Congleton, J., Craig, I.H., Denton, B.K. & Parkins, R.N., Crack growth in HY80 and HY130 steels by corrosion fatigue. Metal Science, 13, pp. 436– 443, 1979. [55] Romaniv, O.M., Voldemarov, A.V. & Nykyforchyn, G.M., Factors in acceleration of crack growth during corrosion fatigue of high-strength steels. Fiziko-Khimicheskaya Mekhanika Materialov, 16(5), pp. 21–27, 1980 (in Russian). [56] Fujii, C.T. & Smith, J.A., Environmental influences on the aqueous fatigue crack growth rates of HY-130 steel. Corrosion Fatigue: Mechanics, Metallurgy, Electrochemistry, and Engineering, eds. T.W. Crooker & B.N. Leis, ASTM STP 801, ASTM: Philadelphia, pp. 390–40, 1983. [57] Komai, K., Kita, S. & Endo, K., Corrosion fatigue crack growth of a high- tension steel in NaCl solution. Bulletin of the JSME, 27(227), pp. 847– 853, 1984. [58] Tong, Z.-S., Li, M.-Q., Feng, B.-X. & Shi, Y., The corrosion fatigue of SiCrMoCuV steel in 3.5% NaCl solution. Corrosion, 41, pp. 121–126, 1985. [59] Murakami, R. & Ferguson, W.G., The effects of cathodic potential and calcareous deposits on corrosion fatigue crack growth rate in seawater for two offshore structural steels. Fatigue and Fracture of Engineering Materials and Structures, 9, pp. 477–488, 1987. [60] Meyn, D.A., An analysis of frequency and amplitude effects on corrosion fatigue crack propagation in Ti-8Al-1Mo-1V. Metallurgical Transactions, 2, pp. 853–865, 1971. [61] Peters, M. Gysler, A. & Lütjering, G., Influence of texture on fatigue properties of Ti-6Al-4V. Metallurgical Transactions A, 15A, pp. 1597– 1605, 1984. [62] Rungta, R. & Begley, J.A., The effect of applied potential on corrosion fatigue crack growth rates of a Ni-Cr-Mo-V turbine disc steel in a room temperature 12M NaOH solution. Corrosion, 35, pp. 544–550, 1979. [63] Jones, D.A., A unified mechanism of stress corrosion and corrosion fatigue cracking. Metallurgical Transactions, 16A, pp. 1133–1141, 1985.

[64] Masuda, H. & Nishijima, S., Application of scratching electrode method for corrosion fatigue. Transactions of National Research Institute for Metals, 29(2), pp. 44–50, 1987. [65] Shimojo, M., Higo, Y. & Nunomura, S., Relation between the amount of fresh bare surface at the crack tip and the fatigue crack propagation rate. ISIJ International, 31, pp. 870–874, 1991. [66] Froats, A., Aune, T.Kr., Hawke, D., Unsworth, W. & Hillis, J., Corrosion of magnesium and magnesium alloys. ASM Handbook, Vol. 13: Corrosion, 9th edn., ASM International: Metals Park, pp. 740–754, 1998. [67] Logan H.L., Mechanism of stress-corrosion cracking in the AZ31B magnesium alloy. Journal of Research of the National Bureau of Standards, 61, pp. 503–508, 1958. [68] Tomashov, N.D. & Isaev, N.I., Investigation of anodic processes during stress corrosion cracking in metals. Doklady Akademii Nauk SSSR, 132, pp. 409–412, 1960 (in Russian). [69] Fairman, L. & West, J.M., Stress-corrosion cracking of a magnesium- aluminum alloy. Corrosion Science, 5, pp. 711–715, 1965. [70] Pugh, E.N., Green, J.A.S. & Slattery, P.W., On the propagation of stress- corrosion cracks in a magnesium-aluminum alloy. Fracture 1969, eds. P.L. Pratt et al., Chapman and Hall Ltd.: London, pp. 387–395, 1969. [71] Chakrapani, D.G. & Pugh, E.N., Hydrogen embrittlement in a Mg-Al alloy. Metallurgical Transactions A, 7A, pp. 173–178, 1976. [72] Moccary, A. & Shastry, C.R., An investigation of stress corrosion cracking in MgAZ61 alloy in 3.5% NaCl + 2% K2CrO4 aqueous solution at room temperature. Zeitschrift für Werkstofftechnik, 10, pp. 119–123, 1979. [73] Meletis, E.I. & Hochman, R.F., Crystallography of stress corrosion cracking in pure magnesium. Corrosion, 40, pp. 39–45, 1984. [74] Stampella, R.S., Procter, R.P.M. & Ashworth, V., Environmentally- induced cracking of magnesium. Corrosion Science, 24, pp. 325–341, 1984. [75] Ebtehaj, K., Hardie, D. & Parkins, R.N., The influence of chloride- chromate solution composition on the stress corrosion cracking of a Mg- Al alloy. Corrosion Science, 28, pp. 811–829, 1988. [76] Lynch, S.P. & Trevena, P., Stress corrosion cracking and liquid metal embrittlement. Corrosion, 44, pp. 113–123, 1988. [77] Murakami, R. & Ferguson, W.G., The effects of a marine environment on the corrosion fatigue crack propagation rate of pure titanium and its weld metal. Fatigue and Fracture of Engineering Materials and Structures, 16, pp. 255–265, 1993. [78] Owe Berg, T.G., Kinetics of absorption by metals of hydrogen from water and aqueous solutions. Corrosion, 16, pp. 198t–200t, 1960. [79] Uhlig, H.H., Action of corrosion and stress on 13% Cr stainless steel. Metal Progress, 57, pp. 486–487, 1950.

[80] Phelps, E.H. & Loginow, A.W., Stress corrosion of steels for aircraft and missiles. Corrosion, 16, pp. 325t–335t, 1960. [81] Leckie, H.P., Effect of environment on stress induced failure of high strength maraging steels. Fundamental Aspects of Stress Corrosion Cracking, eds. R.W. Staehle, A.J. Forty & D. van Rooyen, NACE: Houston, pp. 411–419, 1969. [82] Marichev, V.A. & Rosenfeld, I.L., Investigation of the mechanism of stress corrosion cracking in high strength steels. Corrosion, 32, pp. 423– 429, 1976. [83] Marichev, V.A., Rosenfeld, I.L. & Lunin, V.V., Investigation of the kinetics and mechanism of subcritical crack growth by delayed failure of titanium alloys in and out of corrosive environments. Corrosion, 36, pp. 373–379, 1980. [84] Brown, B.F., Fujii, C.T. & Dahlberg, E.P., Methods for studying the solution chemistry within stress corrosion cracks. Journal of the Electrochemical Society, 116, pp. 218–219, 1969. [85] Taunt, R.J. & Charnock, W., Fluid compositions within fatigue cracks. Materials Science and Engineering, 35, pp. 219–228, 1978. [86] Turnbull, A. & Ferriss, D.H., Mathematical modelling of the electrochemistry in corrosion fatigue cracks in structural steel cathodically protected in sea water. Corrosion Science, 26, pp. 601–628, 1986. [87] Turnbull, A. & Saenz de Santa Maria, M., Predicting the kinetics of hydrogen generation at the tips of corrosion fatigue cracks. Metallurgical Transactions A, 19A, pp. 1795–1806, 1988. [88] Zakroczymski, T., Entry of hydrogen into iron alloys from the liquid phase. Hydrogen Degradation of Ferrous Alloys, eds. R.A. Oriani, J.P. Hirth & M. Smialowski, Noyes Publications: Park Ridge, pp. 215–250, 1985. [89] Dawson, D.B. & Pelloux, R.M., Corrosion fatigue crack growth of titanium alloys in aqueous solutions. Metallurgical Transactions, 5, pp. 723–731, 1974. [90] Pelloux, R.M.N., Corrosion-fatigue crack propagation. Fracture 1969, eds. P.L. Pratt et al., Chapman and Hall: London, pp. 731–739, 1969. [91] Oriani, R.A., A mechanistic theory of hydrogen embrittlement of steels. Berichte der Bunsengesellschaft für Physikalische Chemie, 76, pp. 848– 857, 1972. [92] Suresh, S. & Ritchie, R.O., Mechanistic dissimilarities between environmentally influenced fatigue-crack propagation at near-threshold and higher growth rates in lower strength steels. Metal Science, 16, pp. 529–538, 1982. [93] Hirano, K., Kobayashi, Y., Kobayashi, H. & Nakazawa, H., Hydrogen enhanced fatigue crack growth behavior of high strength steels. Materials, Experimentation and Design in Fatigue (Fatigue’81), eds. F. Sherratt & J.B. Sturgeon, Westbury House: Guildford, pp. 87–96, 1981.

[94] Tau, L., Chan, S.L.I. & Shin, C.S., Hydrogen enhanced fatigue crack propagation of bainitic and tempered martensitic steels. Corrosion Science, 38, pp. 2049–2060, 1996. [95] Nakasa, K. & Satoh, H., The effect of hydrogen-charging on the fatigue crack propagation behavior of β-titanium alloys. Corrosion Science, 38, pp. 457–468, 1996. [96] Young, G.A., Jr. & Scully, J.I., Hydrogen embrittlement of solution heat- treated and aged β-titanium alloys Ti-15% V-3% Cr-3% Al-3% Sn and Ti-15% Mo-3% Nb-3% Al. Corrosion, 50, pp. 919–933, 1994. [97] Venkataraman, G. & Goolsby, A.D., Hydrogen embrittlement in titanium alloys from cathodic polarization in offshore environments, and its mitigation. CORROSION/96, NACE: Houston, paper no. 554, 1996. [98] Gangloff, R.P. & Turnbull, A., Crack electrochemistry modeling and fracture mechanics measurement of the hydrogen embrittlement threshold in steel. Modeling Environmental Effects on Crack Growth Processes, eds. R.H. Jones & W.W. Gerberich, The Metallurgical Society of AIME: Warrendale, pp. 55–81, 1986. [99] de Kazinczy, F., Effect of stresses on hydrogen diffusion in steel. Jernkontorets Annaler, 139, pp. 885–895, 1955. [100] Beck, W., Bockris, J.O’M., McBreen, J. & Nanis, L., Hydrogen permeation in metals as a function of stress, temperature and dissolved hydrogen concentration. Proceedings of the Royal Society of London, A290(1421), pp. 220–235, 1966. [101] Blundy, R.F., Royce, R., Poole, P. & Shreir, L.L., Effect of pressure and stress on permeation of hydrogen through steel. Stress Corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys, eds. R.W. Staehle, J. Hochmann, R.D. McCright & J.E. Slater, NACE: Houston, pp. 636–647, 1977. [102] Troiano, A.R., The role of hydrogen and other interstitials in the mechanical behavior of metals. Transactions of the ASM, 52, pp. 54–80, 1960. [103] Doig, P. & Jones, G.T., A model for the initiation of hydrogen embrittlement cracking at notches in gaseous hydrogen environments. Metallurgical Transactions A, 8A, pp. 1993–1998, 1977. [104] Akhurst, K., A criterion for hydrogen-induced fracture. Advances in Fracture Research (Fracture 81), ed. D. Francois, Vol. 4, Pergamon Press: Oxford, pp. 1899–1907, 1982. [105] Ritchie, R.O., Geniets, L.C.E. & Knott, J.F., Effects of grain-boundary embrittlement on fracture and fatigue crack propagation in a low alloy steel. The Microstructure and Design of Alloys, The Institute of Metals: London, pp. 124–128, 1973. [106] Pao, P.S., Wei, W. & Wei, R.P., Effect of frequency on fatigue crack growth response of AISI 4340 steel in water vapor. Environment- Sensitive Fracture of Engineering Materials, ed. Z.A. Foroulis, The Metallurgical Society of AIME: Warrendale, pp. 565–580, 1979.

[107] McIntyre, P., Hydrogen-steel interactions during cyclic loading. Corrosion Fatigue, eds. R.N. Parkins & Ya.M. Kolotyrkin, London: The Metals Society, pp. 62–73, 1980. [108] Nakasa, K., Takei, H. & Kajiwara, K., Effect of stress wave shape on the crack propagation velocity in cyclic delayed failure. Engineering Fracture Mechanics, 14, pp. 507–517, 1981. [109] Hirose, Y. & Mura, T., Crack nucleation and propagation of corrosion fatigue in high-strength steel. Engineering Fracture Mechanics, 22, pp. 859–870, 1985. [110] Pittinato, G.F., Hydrogen-enhanced fatigue crack growth in Ti-6Al-4V ELI weldments. Metallurgical Transactions, 3, pp. 235–243, 1972. [111] Gerberich, W.W., Moody, N.R., Jensen, C.L., Hayman, C. & Jatavallabhula, K., Hydrogen in α/β and all β titanium systems: analysis of microstructure and temperature interactions on cracking. Hydrogen Effects in Metals, eds. I.M. Bernstein & A.W. Thompson, The Metallurgical Society of AIME: Warrendale, pp. 731–743, 1981. [112] Rice, J.R. & Tracey, D.M., Computational fracture mechanics. Numerical and Computer Methods in Structural Mechanics, eds. S.J. Fenves, N. Perrone, A.R. Robinson & W.C. Schnobrich, Academic Press: New York, pp. 585–623, 1973. [113] Tien, J.K., Thompson, A.W., Bernstein, I.M. & Richards, R.J., Hydrogen transport by dislocations. Metallurgical Transactions A, 7A, pp. 821–829, 1976. [114] Oriani R.A. & Josephic, P.H., Equilibrium aspects of hydrogen-induced cracking of steels. Acta Metallurgica, 22, pp. 1065–1074, 1974. [115] Gerberich, W.W. & Chen, Y.T., Hydrogen-controlled cracking – an approach to threshold stress intensity. Metallurgical Transactions A, 6A, pp. 271–278, 1975. [116] Endo, K., Komai, K. & Matsuda, Y., Influences of stress ratios on cyclic stress corrosion crack growth characteristics of a high-strength steel. Bulletin of the JSME, 24(197), pp. 1885–1892, 1981. [117] Gerberich, W.W. & Yu, W., Hydrogen interactions in fatigue crack thresholds. Fracture Problems and Solutions in the Energy Industry, ed. L.A. Simpson, Pergamon Press: Oxford, pp. 39–50, 1982. [118] van Leeuwen, H.P., The kinetics of hydrogen embrittlement: a quantitative diffusion model. Engineering Fracture Mechanics, 6, pp. 141–146, 1974. [119] McMahon, C.J., Jr., Briant, C.L. & Banerji, S.K., The effects of hydrogen and impurities on brittle fracture in steel. Advances in Research on the Fracture of Materials (Fracture 77), ed. D.M.R. Taplin, Vol. 1, Pergamon Press: New York, pp. 363–385, 1978. [120] Bowles, C.Q. & Schijve, J., Experimental observations of environmental contributions to fatigue crack growth. Corrosion Fatigue: Mechanics, Metallurgy, Electrochemistry, and Engineering, eds. T.W. Crooker & B.N. Leis, ASTM STP 801, ASTM: Philadelphia, pp. 96–113, 1983.

[121] Marichev, V.A. & Shipilov, S.A., Influence of electrochemical polarization on crack growth in corrosion cracking and corrosion fatigue of magnesium alloys. Fiziko-Khimicheskaya Mekhanika Materialov, 22(3), pp. 21–25, 1986 (in Russian). [122] Phillips, I.I., Poole, P. & Shreir, L.L., Hydride formation during cathodic polarization of Ti – I. Effect of current density on kinetics of growth and composition of hydride. Corrosion Science, 12, pp. 855–866, 1972. [123] Orman, S. & Picton, G., The role of hydrogen in the stress corrosion cracking of titanium alloys. Corrosion Science, 14, pp. 451–459, 1974. [124] Hack, J.E. & Leverant, G.R., The influence of microstructure on the susceptibility of titanium alloys to internal hydrogen embrittlement. Metallurgical Transactions A, 13A, pp. 1729–1738, 1982. [125] Meyn, D.A., Effect of hydrogen on fracture and inert-environment sustained load cracking resistance of α-β titanium alloys. Metallurgical Transactions, 5, pp. 2405–2414, 1974. [126] Sastry, S.M.L., Lederich, R.J. & Rath, B.B., Subcritical crack-growth under sustained load in Ti-6Al-6V-2Sn. Metallurgical Transactions A, 12A, pp. 83–94, 1981. [127] Pao, P.S. & Wei, R.P., Hydrogen-enhanced fatigue crack growth in Ti- 6Al-2Sn-4Zr-2Mo-0.1Si. Titanium, Science and Technology, eds. G. Lütjering, U. Zwicker & W. Bunk, Vol. 4, DGM: Oberursel, pp. 2503– 2510, 1985. [128] Clarke, C.F., Hardie, D. & Ikeda, B.M., The effect of hydrogen content on the fracture of pre-cracked titanium specimens. Corrosion Science, 36, pp. 487–509, 1994. [129] Frandsen, J.D. & Marcus, H.L., The correlation between grain size and plastic zone size for environmental hydrogen assisted fatigue crack propagation. Scripta Metallurgica, 9, pp. 1089–1094, 1975. [130] Austen, I.M. & McIntyre, P., Corrosion fatigue of high-strength steel in low-pressure hydrogen gas. Metal Science, 13, pp. 420–428, 1979. [131] Rungta, R., Begley, J.A. & Staehle, R.W., Effect of steam impurities on corrosion fatigue crack growth rates of a turbine disc steel. Corrosion, 37, pp. 682–690, 1981. [132] Robertson, W.D. & Tetelman, A.S., A unified structural mechanism for intergranular and transgranular corrosion cracking. Strengthening Mechanisms in Solids, ASM: Metals Park, pp. 217–252, 1962. [133] Yoshino, K. & McMahon, C.J., Jr., The cooperative relation between temper embrittlement and hydrogen embrittlement in a high strength steel. Metallurgical Transactions, 5, pp. 363–370, 1974. [134] Briant, C.L. & Banerji, S.K., Intergranular failure in steel: the role of grain-boundary composition. International Metals Reviews, 23, pp. 164– 199, 1978. [135] Craig, B.D. & Krauss, G., The structure of tempered martensite and its susceptibility to hydrogen stress cracking. Metallurgical Transactions A, 11A, pp. 1799–1808, 1980.

[136] Bandyopadhyay, N, Kameda, J. & McMahon, C.J., Jr., Hydrogen-induced cracking in 4340-type steel: effects of composition, yield strength, and H2 pressure. Metallurgical Transactions A, 14A, pp. 881–888, 1983. [137] Kameda, J. & McMahon, C.J., Jr., The effects of Sb, Sn, and P on the strength of grain boundaries in a Ni-Cr steel. Metallurgical Transactions A, 12A, 31–37, 1981. [138] Powell, D.T. & Scully, J.C., Stress corrosion cracking of alpha titanium alloys at room temperature. Corrosion, 24, pp. 151–158, 1968. [139] Pourbaix, M., Atlas of Electrochemical Equilibria in Aqueous Solutions. Translated by J.A. Franklin, Pergamon Press: Oxford, 1966. [140] Grinberg, N.M., Serdyuk, V.A., Malinkina, T.I. & Kamyshkov, A.S., Effect of vacuum and low temperature on the fatigue crack propagation in magnesium alloys. Fiziko-Khimicheskaya Mekhanika Materialov, 19(4), pp. 48–54, 1982 (in Russian). [141] Lynch, S.P., Failures of structures and components by environmentally assisted cracking. Engineering Failure Analysis, 1, pp. 77–90, 1994. [142] Shipilov, S.A., Location of the fracture process zone for hydrogen- induced corrosion fatigue crack propagation. Scripta Materialia, 47, pp. 301–305, 2002. [143] Rice, J.R., Mechanics aspects of stress corrosion cracking and hydrogen embrittlement. Stress Corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys, eds. R.W. Staehle, J. Hochmann, R.D. McCright & J.E. Slater, NACE: Houston, pp. 11–15, 1977. [144] Knott, J.F., Models for fatigue crack growth. Fatigue Crack Growth: 30 Years of Progress, ed. R.A. Smith, Pergamon Press: Oxford, 31-52, 1986. [145] Yuen, A., Hopkins, S.W., Leverant, G.R. & Rau, C.A., Correlation between fracture surface appearance and fracture mechanics parameters for stage II fatigue crack propagation in Ti-6Al-4V. Metallurgical Transactions, 5, pp. 1833–1842, 1974. [146] Rice, J.R., Mechanics of crack tip deformation and extension by fatigue. Fatigue Crack Propagation, ASTM STP 415, ASTM: Philadelphia, pp. 247–309, 1967. [147] Hahn, G.T., Hoagland, R.G. & Rosenfield, A.R., Local yielding attending fatigue crack growth, Metallurgical Transactions, 3, pp. 1189–1202, 1972. [148] Nazimov, O.P. & Zhuravlyov, L.N., Hydrogen diffusion in titanium alloys. Izvestiia Vysshikh Uchebnykh Zavedenii SSSR, Tsvetnaia Metallurgiia, 1, pp. 160–161, 1976 (in Russian). [149] Renner, J. & Grabke, H.J., Determination of diffusion coefficients in the hydriding of alloys. Zeitschrift für Metallkunde, 69, pp. 639–642, 1978. [150] Tien, J.K., Richards, R.J., Buck, O. & Marcus, H.L., Model of dislocation sweep-in of hydrogen during fatigue crack growth. Scripta Metallurgica, 9, pp. 1097–1101, 1975.

[151] Tien, J.K., Buck, O., Bates, R.C. & Nair, S., On the role of time delayed hydrogen assisted plastic zone growth on subcritical crack growth in high- strength steels. Scripta Metallurgica, 14, pp. 583–590, 1980. [152] Shipilov, S.A., The increment of hydrogen-induced corrosion fatigue crack growth per cycle: comparing theory and experimental data. Hydrogen Effects on Material Behavior and Corrosion Deformation Interactions, eds. N.R. Moody, A.W. Thompson, R.E. Ricker, G.W. Was & R.H. Jones, TMS: Warrendale, pp. 607–616, 2003. [153] Brown, W.F., Jr. & Srawley, J.E., Plane Strain Crack Toughness Testing of High Strength Metallic Materials, ASTM STP 410, ASTM: Philadelphia, 1967. [154] Nelson, H.G., Phenomenological aspects of hydrogen embrittlement. Environmental Degradation of Engineering Materials, eds. M.R. Louthan, Jr., R.P. McNitt & R.D. Sisson, Jr., Pennsylvania State University: University Park, pp. 83–96, 1987.