Stress-Corrosion Cracking Materials Performance and Evaluation Copyright © 1992 ASM International® Russell H. Jones, editor, p 1-40 All rights reserved. DOI: 10.1361/sccmpae1992p001 www.asminternational.org

CHAPTER 1 Mechanisms of Stress-Corrosion Cracking R.H. Jones, Pacific Northwest Laboratories R.E. Ricker, National Institute of Standards and Technology

Stress-corrosion cracking (SCC) is a term used to process because the combined simultaneous in­ describe service failures in engineering materials teraction of mechanical and chemical forces re­ that occur by slow, environmentally induced sults in crack propagation, whereas neither factor crack propagation. The observed crack propaga­ acting independently or alternately would result tion is the result of the combined and synergistic in the same effect. The exact nature of this inter­ interaction of mechanical stress and corrosion re­ action is the subject of numerous scientific inves­ actions. tigations and will be covered in the section Before SCC can be discussed in detail, we "Crack-Propagation Mechanisms" in this chapter. must clearly define the type of loading involved, The stresses required to cause SCC are small, the types of materials involved, the types of envi­ usually below the macroscopic yield stress, and ronments that cause SCC, and the nature of the in­ are tensile in nature. The stresses can be exter­ teractions that result in this phenomenon. The nally applied, but residual stresses often cause term "stress-corrosion cracking" is frequently SCC failures. However, compressive residual used to describe any type of environmentally in­ stresses can be used to prevent this phenomenon. duced or assisted crack propagation. However, Static loading is usually considered to be respon­ this discussion will focus on the normal usage of sible for SCC, while environmentally induced the term as defined below. crack propagation due to cyclic loading is defined One common misconception is that SCC is as corrosion fatigue. The boundary between these the result of stress concentration at corrosion- two classes of phenomena is vague, and corrosion generated surface flaws (as quantified by the fatigue is often considered to be a subset of SCC. stress-intensity factor, K); when a critical value of However, because the environments that cause stress concentration, Kcnu is reached, mechanical corrosion fatigue and SCC are not always the fracture results. Although stress concentration same, these two should be considered separate does occur at such flaws, it does not exceed the phenomena. critical value required to cause mechanical frac­ The term "stress-corrosion cracking" is usu­ ture of the material in an inert environment (/iTscc ally used to describe failures in metallic alloys. <^crit). Precorrosion followed by loading in an However, other classes of materials also exhibit inert environment will not result in any signifi­ delayed failure by environmentally induced crack cant crack propagation, while simultaneous envi­ propagation. Ceramics exhibit environmentally ronmental exposure and application of stress will induced crack propagation (Ref 1) and polymeric cause time-dependent subcritical crack propaga­ materials frequently exhibit craze cracking as a tion. The term "synergy" is used to describe this result of the interaction of applied stress and envi- 2 Stress-Corrosion Cracking

Table 1 Alloy/Environment Systems Exhibiting SCC

Alloy Environment Carbon steel Hot nitrate, hydroxide, and carbonate/bicarbonate solutions High-strength steels Aqueous electrolytes, particularly when containing H2S Austenitic stainless steels. Hot, concentrated chloride solutions; chloride-contaminated steam High-nickel alloys High-purity steam a-brass Ammoniac al solutions Aluminum alloys Aqueous CT, Br", and I~ solutions - Titanium alloys Aqueous CT, Br , and r solutions; organic liquids; N204 Magnesium alloys Aqueous CT solutions Zirconium alloys Aqueous CT solutions; organic liquids; I2 at 350 °C (660 °F)

Table 2 Alloy/Environment Combinations and the Resulting Films That Form at the Crack Tip

Metal or alloy Environment Initiating layer a-brass, copper-aluminum Ammonia Dealloyed layer (Cu) Gold-copper FeCb Dealloyed layer (Au) Acid sulfate Dealloyed layer (Au) Iron-chromium nickel Chloride Dealloyed layer (Au) Hydroxide Dealloyed layer or oxide High-temperature water Dealloyed layer or oxide a-brass Nitrite Oxide Copper Nitrite Oxide Ammonia (cupric) Porous dissolution zone Ferritic steel High-temperature water Oxide Phosphate Oxide (?) Anhydrated ammonia Nitride CO/CO2/H2O Carbide CS2/H2O Carbide Titanium alloys Chloride Hydride Aluminum alloys, steels Various media Near-surface hydrogen ronmental reactions (Ref 2-5). The mechanical to the presence of ammonia in the environment, properties of composites will degrade if exposure and chloride ions cause or exacerbate cracking in to the environment attacks the matrix, the rein­ stainless steels and aluminum alloys. Also, an en­ forcing phase, or the matrix-to-reinforcement in­ vironment that causes SCC in one alloy may not terface; if crack propagation results during static cause it in another. Changing the temperature, the loading, this degradation is SCC. Until recently, it degree of aeration, and/or the concentration of was thought that pure metals were immune to ionic species may change an innocuous environ­ SCC, but it is now known that this is not true (Ref ment into one that causes SCC failure. Also, dif­ 6, 7). Because considerably more research has ferent heat treatments may make the same alloy been conducted on the SCC behavior of metallic either immune or susceptible. As a result, the list alloys, the discussion in this chapter will focus on of all possible alloy/environment combinations SCC of metals and their alloys. that cause SCC is continually expanding, and the Environments that cause SCC are usually possibilities are virtually infinite. Some of the aqueous and can be either condensed layers of more commonly observed alloy/environment moisture or bulk solutions. SCC is alloy/environ­ combinations that result in SCC are listed in Ta­ ment specific; that is, it is frequendy the result of ble 1. a specific chemical species in the environment. In general, SCC is observed in alloy/environ­ For example, the SCC of copper alloys, tradition­ ment combinations that result in the formation of ally referred to as season cracking, is usually due a film on the metal surface. These films may be Mechanisms ofSCC 3

passivating layers, tarnish films, or dealloyed lay­ reviewed briefly. Stress-corrosion cracking ex­ ers. In many cases, these films reduce the rate of periments can be categorized as: general or uniform corrosion, making the alloy Tests on statically loaded smooth samples desirable for resistance to uniform corrosion in Tests on statically loaded precracked sam­ the environment. As a result, SCC is of greatest ples concern in corrosion-resistant alloys exposed to Tests using slowly straining samples aggressive aqueous environments. The exact role More detailed information on these tests can of films in the SCC process is the subject of cur­ be found in the chapter "Evaluation of Stress- rent research and will be covered in the section Corrosion Cracking" and in Ref 8. "Crack-Propagation Mechanisms" in this chap­ Tests on statically loaded smooth samples ter. Table 2 lists several alloy/environment com­ are usually conducted at various fixed stress lev­ binations and the films that may form at the crack els, and the time to failure of the sample in the en­ tip. vironment is measured. Figure 1 illustrates the typical results obtained from this type of test. In Fig. 1, the logarithm of the measured time to fail­ The Phenomenon of SCC ure, tt, is plotted against the applied stress, a ap­ plied, and the time to failure can be seen to Stress-corrosion cracking is a delayed failure increase rapidly with decreasing stress; a thresh­ process. That is, cracks initiate and propagate at a old stress, CTth,i s determined where the time to slow rate (for example, 10" to 10 m/s) until the failure approaches infinity. The total time to fail­ stresses in the remaining ligament of metal ex­ ure at a given stress consists of the time required ceed the fracture strength. The sequence of events for the formation of a crack (the incubation or in­ involved in the SCC process is usually divided itiation time, fin, and the time of crack propaga­ into three stages: tion, tcp). These experiments can be used to Crack initiation and stage 1 propagation determine the maximum stress that can be applied Stage 2 or steady-state crack propagation in service without SCC failure, to determine an • Stage 3 crack propagation or final failure inspection interval to confirm the absence of SCC The characteristics of each of these stages will be discussed in greater detail below. First, how­ ever, the techniques used to measure SCC will be

fi J£ O aH E o £>

Threshold stress Critical fracture Intensity lor SCC stress Intensity (fscc)

Magnitude of the crack tip stress distribution Threshold "applied Fracture [ Stress Intensity factor, K, MPA-nr0-* or ksl-hr0-5 ] Stress Stmts

Applied stress or load Fig. 2 Schematic diagram of typical crack-propaga­ tion rate as a function of crack-tip stress-intensity be­ Fig. 1 Schematic of a typical time to failure as a func­ havior illustrating the regions of stage 1, 2, and 3 tion of initially applied stress for smooth-sample SCC crack propagation as well as identifying the plateau tests velocity and the threshold stress intensity 4 Stress-Corrosion Cracking crack propagation, or to evaluate the influence of ice, most SCC failures occur under constant-load metallurgical and environmental changes on conditions, so that the stress intensity increases as SCC. However, the time required for crack initia­ the crack propagates. As a result, it is usually as­ tion is strongly dependent on a wide variety of pa­ sumed in SCC discussions that the stress intensity rameters, such as surface finish. The presence of is increasing with increasing crack length. flaws that concentrate stress or crevices that alter Typically, three regions of crack-propagation the environment may dramatically change the rate versus stress-intensity level are found during threshold stress or the crack-initiation time (Ref crack-propagation experiments. These are identi­ 9,10). The entire crack-initiation process is pres­ fied according to increasing stress-intensity fac­ ently not well understood. tor as stage 1,2, or 3 crack propagation (Fig. 2). Tests on statically loaded precracked sam­ No crack propagation is observed below some ples are usually conducted with either a constant threshold stress-intensity level, /fiscc This applied load or with a fixed crack opening dis­ threshold stress level is determined not only by placement, and the actual rate or velocity of crack the alloy but also by the environment and metal­ propagation, da/dt, is measured (Ref 11). The lurgical condition of the alloy, and, presumably, magnitude of the stress distribution at the crack this level corresponds to the minimum required tip (the mechanical driving force for crack propa­ stress level for synergistic interaction with the en­ gation) is quantified by the stress-intensity factor, vironment. At low stress-intensity levels (stage K, for the specific crack and loading geometry. As 1), the crack-propagation rate increases rapidly a result, the crack-propagation rate, da/dt, is plot­ with the stress-intensity factor. At intermediate ted versus K, as illustrated in Fig. 2. These tests stress-intensity levels (stage 2), the crack-propa­ can be configured such that K increases with gation rate approaches some constant velocity crack length (constant applied load), decreases that is virtually independent of the mechanical with increasing crack length (constant crack driving force. This plateau velocity is charac­ mouth opening displacement), or is approxi­ teristic of the alloy/environment combination and mately constant as the crack length changes (spe­ is the result of rate-limiting environmental proc­ cial tapered samples). Each type of test has its esses such as mass transport of environmental advantages and disadvantages. However, in serv­ species up the crack to the crack tip. In stage 3, the

100 Inert environment, A » B "~ ... ■ ^ " m Aggressive ^r 8* environment, Jf d

a I o I

Aggressive environment, alloy A 1 1 1 1 1 1 I 10 10-8 io-6 10-4 10-2 10-1° 10-8 io-6 io-« 10-2

1 (a) (b) Nominal strain rate [ s' ] Fig. 3 Strain-to-failure plots resulting from slow-strain-rate testing, (a) Schematic of typical ductility versus strain- rate behavior of two different types of alloys tested by the slow-strain-rate technique, (b) Schematic of the ductility ratio versus strain-rate behavior of two different types of alloys. The ductility ratio is the ratio of a ductility measure­ ment such as elongation, reduction in area, or fracture energy measured in the aggressive environment to that ob­ tained in the inert reference environment. Mechanisms ofSCC 5

I I I ■ I 25 (■) 0>) * 100.0% -O

SOT?** I-

trogenga s > «c 80.0% s CI / I 15

solution/dr y ni l s lo , e 60.0% c M Binary AI-3LI alloy O • Ternary AI-2Li-1Cu aUoy Q ■ s 2 1 S 10 Ductilit y ra i 3.5% NaCI + C03- /HCC>3- , pH-9.3 • ■ 0.5 M NaCI ° O s 5 ^^^^^^^^^^mm*mmtmmUmlmm4m*mm*»mJ>mm+m^m^**± 1 10 100 1000 5 10 15 20 25

Grain boundary aging time, hrs. Fracture strain in dry nitrogen gas, % (a) (b) Fig. 4 Comparison of (a) slow-strain-rate data plotted as a ductility ratio to (b) the same data plotted as an environ­ ment-dependent property versus the environment-independent value of the same property. Source: Ref 13 rate of crack propagation exceeds the plateau ve­ this property, as shown in Fig. 4. In this manner, locity as the stress-intensity level approaches the the strength of the material in the environment, or critical stress-intensity level for mechanical frac­ the "situation-dependent strength" (Ref 10), and ture in an inert environment, K\c (Ref 12). the extent of any environmental effect can be Slow-Strain-Rate Testing. Stress-corrosion visualized simultaneously. However, the applica­ tests can also be conducted by slowly increasing tion of these data to the prediction of actual in- the load or strain on either precracked or smooth service lifetimes is difficult and unreliable. samples. These tests are called constant-exten­ sion-rate tests, slow-strain-rate tests, or straining electrode tests. Usually, a tensile machine pulls a Overview of SCC Mechanisms smooth sample that is exposed to the corrosive Many different mechanisms have been pro­ environment at a low crosshead speed (1(T to posed to explain the synergistic stress-corrosion 10" m/s). The strain to failure in the corrosive en­ interactions that occur at the crack tip, and there vironment and the strain to failure in an inert en­ may be more than one process that causes SCC. vironment can then be plotted against the strain The proposed mechanisms can be classed into rate, as shown in Fig. 3(a), or the ratio of these two basic categories: anodic mechanisms and measurements can be plotted as shown in Fig. cathodic mechanisms. That is, during corrosion, 3(b). The ratio of other tensile-property measure­ both anodic and cathodic reactions must occur, ments, such as reduction in area and ultimate ten­ and the phenomena that result in crack propaga­ sile strength, may be plotted. tion may be associated with either type. The most Frequently, this type of test is used to evaluate obvious anodic mechanism is that of simple ac­ the influence of metallurgical variables, such as tive dissolution and removal of material from the heat treatment, on SCC resistance. This type of crack tip. The most obvious cathodic mechanism experiment yields rapid comparisons. However, is hydrogen evolution, absorption, diffusion, and since the mechanical properties of the samples embrittlement. However, a specific mechanism also vary with the metallurgical condition, such must be able to explain the actual crack-propaga­ evaluations can become difficult. As a result, it tion rates, the fractographic evidence, and the was proposed (Ref 13) that the environment-de­ mechanism of formation or nucleation of cracks. pendent property be plotted versus the inert-envi­ Some of the more prominent of the proposed ronment or environment-independent value of mechanisms will be covered in greater detail in 6 Stress-Corrosion Cracking the section "Crack-Propagation Mechanisms" in plateau region in which the rate of crack propaga­ this chapter, but they usually assume that break­ tion is independent of the applied mechanical ing of the interatomic bonds of the crack tip oc­ stress. That is, a sequence of chemical reactions curs either by chemical solvation and dissolution and processes is required, and the rate-limiting or by mechanical fracture (ductile or brittle). Me­ step in this sequence of events determines the chanical fracture includes normal fracture proc­ limiting rate or plateau velocity of crack propaga­ esses that are assumed to be stimulated or induced tion (until mechanical overload fracture starts by one of the following interactions between the contributing to the fracture process in stage 3). material and the environment: Figure 5 illustrates a crack tip in which crack propagation results from reactions in metal ahead Adsorption of environmental species of the propagating crack. This example was cho­ Surface reactions Reactions in the metal ahead of the crack tip sen because it maximizes the number of possible Surface films rate-limiting steps. Close examination of Fig. 5 reveals that potential rate-determining steps in­ All of the proposed mechanical fracture clude: mechanisms contain one or more of these proc­ esses as an essential step in the SCC process. Spe­ Mass transport along the crack to or away cific mechanisms differ in the processes assumed from the crack tip to be responsible for crack propagation and the Reactions in the solution near the crack way that environmental reactions combine to re­ Surface adsorption at or near the crack tip sult in the actual fracture process. Surface diffusion Surface reactions Absorption into the bulk Bulk diffusion to the plastic zone ahead of Controlling Parameters the advancing crack The mechanisms that have been proposed for Chemical reactions in the bulk The rate of interatomic bond rupture SCC require that certain processes or events oc­ cur in sequence for sustained crack propagation Changes in the environment that modify the to be possible. These requirements explain the rate-determining step will have a dramatic influ­ ence on the rate of crack propagation, while al­ terations to factors not involved in the rate-determining step or steps will have little in­ fluence, if any. However, significantly retarding the rate of any one of the required steps in the se­ quence could make that step the rate-determining one. In aqueous solutions, the rate of adsorption and surface reactions is usually very fast com­ pared to the rate of mass transport along the crack to the crack tip. As a result, bulk transport into this region or reactions in this region are frequently believed to be responsible for determining the steady-state crack-propagation rate or plateau ve­ locity. In gaseous environments, surface reac­ Fig. 5 Schematic of crack-tip processes that may be tions, surface diffusion, and adsorption may be the rate-determining step in environmentally assisted rate limiting, as well as the rate of bulk transport crack propagation. For this illustration, an internal to the crack tip (Ref 14,15). hydrogen embrittlement mechanism is assumed in or­ der to maximize the number of possible rate-deter­ Several different environmental parameters mining steps. are known to influence the rate of crack growth in Mechanisms ofSCC 7 aqueous solutions. These include, but are not lim­ ited to: The magnitude of the applied stress or the Temperature stress-intensity factor Pressure The stress state, which includes (1) plane Solute species stress and (2) plane strain Solute concentration and activity • The loading mode at the crack tip (tension or . pH torsion, for example) Electrochemical potential Alloy composition, which includes (1) Solution viscosity nominal composition, (2) exact composition Stirring or mixing (all constituents), and (3) impurity or tramp element composition Altering any of these parameters may affect Metallurgical condition, which includes (1) the rate of the rate-controlling steps, either accel­ strength level, (2) second phases present in the matrix and at the grain boundaries, (3) erating or reducing the rate of crack propagation. composition of phases, (4) grain size, (5) Also, it may be possible to arrest or stimulate grain-boundary segregation, and (6) residual crack propagation by altering the rate of an envi­ stresses ronmental reaction. It is well known and gener­ Crack geometry, which includes (1) length, ally accepted that the environment at occluded width, and aspect ratio, and (2) crack open­ sites, such as crack tips, can differ significantly ing and crack-tip closure from the bulk solution. If an alteration to the bulk environment allows the formation of a critical SCC environment at crack nuclei, then crack Important Fracture Features propagation will result. If the bulk environment Stress-corrosion cracks can initiate and cannot maintain this local crack-tip environment, propagate with little outside evidence of corro­ then crack propagation will stop. As a result, sion and with no warning as catastrophic failure slight changes to the environment may have a approaches. The cracks frequently initiate at sur­ dramatic influence on crack propagation, while face flaws that either preexist or are formed dur­ dramatic changes may have only a slight influ­ ing service by corrosion, wear, or other processes. ence. The cracks then grow with little macroscopic evi­ In addition to the environmental parameters dence of mechanical deformation in metals and listed above, stress-corrosion crack-propagation alloys that are normally quite ductile. Crack rates are influenced by: propagation can be either intergranular or trans-

(b) Fig. 6 Optical micrographs showing defects on the innersurface of type 304 stainless steel pipe near weld root (a) and near through-crack (b). Both lOOOx Stress-Corrosion Cracking

(a) (b) Fig. 7 Stress-corrosion crack initiating from a corrosion pit in a quenched-and-tempered high-strength turbine disk steel (339Ni-1.56Cr-0.63Mo-0.11V) test coupon exposed to oxygenated, demineralized water for 800 h under a bending stress of 90% of the yield stress, (a) 275x. (b) 370x. Courtesy ofSJ. Lennon, ESCOM, andF.PA. Robinson, University ofWitwatersrand granular; sometimes, both types are observed on initiation by intergranular corrosion. A cold- the same fracture surface. Crack openings and the worked layer and surface burrs, shown in Fig. deformation associated with crack propagation 6(b), can also assist crack initiation. may be so small that the cracks are virtually invis­ Crack Initiation at Corrosion Pits. Stress- ible except in special nondestructive examina­ corrosion cracks can also initiate at pits that form tions. As the stress intensity increases, the plastic during exposure to the service environment (Fig. deformation associated with crack propagation 7) or during cleaning operations, such as pickling increases and the crack opening increases. When of type 304 stainless steel before fabrication. Pits the final fracture region is approached, plastic de­ can form at inclusions that intersect the free sur­ formation can be appreciable, because corrosion- face or by a breakdown in the protective film. In resistant alloys are frequently quite ductile. electrochemical terms, pits form when the poten­ tial exceeds the pitting potential. It has been shown that the SCC potential and pitting potential Phenomenology of were identical for steel in nitrite solutions (Ref Crack-Initiation Processes 16,17). Crack Initiation at Surface Discontinuities. The transition between pitting and cracking Stress-corrosion cracking frequently initiates at depends on the same parameters that control preexisting or corrosion-induced surface fea­ SCC, that is, the electrochemistry at the base of tures. These features may include grooves, laps, the pit, pit geometry, chemistry of the material, or burrs caused by fabrication processes. Exam­ and stress or strain rate at the base of the pit. Ade- ples of such features are shown in Fig. 6; these tailed description of the relationship between were produced during grinding in the preparation these parameters and crack initiation has not been of a joint for welding. The feature shown in Fig. developed because of the difficulty in measuring 6(a) is a lap, which is subsequently recrystallized crack initiation. Methods for measuring short sur­ during welding and could then act as a crevice at face cracks are under development, but are lim­ which deleterious anions or cations concentrate. ited to detecting cracks that are beyond the The highly sensitized recrystallized material initiation stage. The geometry of apit is important could also more readily become the site of crack in determining the stress and strain rate at its base. Mechanisms ofSCC 9

Generally, the aspect ratio between the penetra­ tion and the lateral corrosion of a pit must be greater than about 10 before a pit acts as a crack- initiation site. A penetration to lateral corrosion ratio of 1 corresponds to uniform corrosion, and a ratio of about 1000 is generally observed for a growing stress-corrosion crack. As in the ease of a growing crack, the pit walls must exhibit some passive-film-forming capability in order for the corrosion ratio to exceed 1. A change in the corrosive environment and potential within a pit may also be necessary for the pit to act as a crack initiator. Pits can act as oc­ cluded cells similar to cracks and crevices, al­ though in general their volume is not as restricted. There are a number of examples in which stress- corrosion cracks initiated at the base of a pit by in- tergranylar corrosion. In these circumstances, the grain-boundary chemistry and the pit chemistry 0 10 20 30 40 50 were such that intergranular corrosion was fa­ Stnm Intensity, MPaVfif vored. Crack propagation was also by intergranu­ lar SCC in these cases. (a) Although the local stresses and strain rates at the base of the pit play a role in SCC initiation, there are examples of preexisting pits that did not initiate stress-corrosion cracks. This observation has led to the conclusion mat the electrochemistry of the pit is more important than the local stress or strain rate (Ref 16). A preexisting pit may not de­ velop the same local electrochemistry as one grown during service, because the development of a concentration cell depends on the presence of i £# an actively corroding tip that establishes the an­ a. /^fif ion and cation current flows. Similarly, an inabil­ I W ity to reinitiate crack growth in samples in which 10 5 active growth was occurring before the samples 2 ' - /W were removed from solution, rinsed, dried, and * In reinserted into solution also suggests that the lo­ +1* I cal chemistry is very important. IO* - f I Crack Initiation by Intergranular Corro­ + F«, 0.6 V (SCE), COP * -0.15 V sion or Slip Dissolution. Stress-corrosion crack + • Nl,-1.0 V (SCE), COP =-0.67 V 10"7I ±1 I 1 I initiation can also occur in the absence of pitting 0 10 20 30 40 SO by intergranular or slip-dissolution processes. In­ Straw Intensity, MPaVm tergranular corrosion-initiated SCC requires that (b) the local grain-boundary chemistry differ from the bulk chemistry. This condition occurs in sen­ Fig. 8 Crack-growth rate versus elastic-plastic stress sitized austenitic stainless steels or with the seg­ intensity for iron and nickel tested in 1 N H^S04 at regation of impurities such as phosphorus, sulfur, given cathodic overpotentials (COP), (a) 2 mm (0.08 in.) thick iron and nickel, (b) 10 mm (0.4 in.) thick iron or silicon in a variety of materials. Slip-dissolu- and nickel 10 Stress-Corrosion Cracking

tion-initiated SCC results from local corrosion at experimentally, even though it is not difficult to emerging slip planes and occurs primarily in low- detect the location from which a growing crack stacking-fault materials. The processes of crack has emanated. Furthermore, crack initiation has initiation and propagation by the slip-dissolution not been precisely defined, and it is difficult to de­ process are in fact very similar. termine at what point a pit is actually a crack and Crack initiation can be thought of as occurring when intergranular corrosion becomes inter­ in several stages, including the first few atomic granular SCC. Also, the fracture-mechanics con­ layers that fail due to the transition from short- cept of design assumes preexisting flaws in crack to long-crack behavior. Evidence obtained materials, although these may not be surface by electrochemical (Ref 18) and acoustic-emis­ flaws that can become stress-corrosion cracks. It sion monitoring (Ref 19) of crack initiation in has been demonstrated that the corrosion-fatigue austenitic stainless steel suggests that the process threshold of 12% Cr and 2.0% NiCrMoV steels occurs by the initiation of multiple short cracks could be related to the minimum depth of surface prior to the propagation of a single dominant pits, as shown in Fig. 9. Using a linear-elastic crack. Examples of the electrochemical and fracture-mechanics approach and relating a pit to acoustic-emission transients thought to be associ­ a half-elliptical surface crack, one researcher has ated with the growth and retardation of multiple shown that the critical pit dimension could be ex­ short cracks are given in Fig. 8. Retardation of the pressed by the following relationship (Ref 21): initiated cracks occurs because of variations in grain-boundary sensitization, crack angle, elec­ trochemistry, and so on, with cracks being ar- rested after they grow about one grain diameter. V J Details of the conditions required for a single where AATu, is the corrosion fatigue threshold, F is dominant crack to propagate have not been fully a constant, and Aoo is the alternating surface evaluated, but most likely it is a statistical process stress. A pit could be represented by a half-ellipti­ where there is a finite probability for a crack to cal surface crack because it had intergranular cor­ find a path that does not retard its growth. rosion at the base that caused the pit to have Another concept suggested by Parkins (Ref cracklike characteristics. 20) for gas pipeline SCC initiation involves crack A model for stress-corrosion crack growth coalescence. It has been suggested that crack in­ from pits in brass tested in a nontarnishing ammo- itiation in gas pipeline steels occurs by the fol­ niacal solution was developed (Ref 22). The re­ lowing steps: searchers evaluated the corrosion and stress Penetration of ground water to the pipeline aspects of a pit to develop a model for crack initia­ surface and generation of a critical SCC en­ tion. They assumed that the corrosion conditions vironment in the base of the pit were essentially the same as Initially rapid formation of multiple cracks, those on a flat surface and that initiation required the velocity decreasing with time to a con­ that a critical crack-tip opening displacement be stant value A period of constant crack-growth velocity exceeded. Using linear-elastic fracture mechan­ Conventional stages 1,2, and 3 as described ics, the researchers developed the following rela­ by single-crack behavior, during which time tionship for the time to initiate a crack: crack coalescence leads to final fracture 2 , (Kiscd (-Vm) m 0, Crack-Initiation Mechanisms ^ = *(