OVERVIEW _ _ _ _ __
Crack Growth Resistance Russell H. Jones
INTRODUCTION
Materials design and development for resistance to stress corrosion cracking depends upon an understanding of the crack-growth mechanism. The factors controlling stress corrosion cracking may be classified as environmental, material and mechanical. This overview examines the state of the art in stress corrosion cracking and reviews such considerations as phenomenological influences, material chemistry and microstructure, mechanical factors. Also explored are materials design and future considerations.
___ .... \......
.
" one 2
\-~~~
Cracking Zones
Zone'
Figure 1. Schematic of corrosion current versus electrochemical potential showing zones at which stress corrosion cracking is likely to occur.
32
There is a significant understanding of the phenomenological factors which affect see, and, in some cases, there has been a demonstrated ability to control see. Our understanding of the mechanisms and, in particular, the development of reliable analytical expressions of see is much sketchier. A description of the factors controlling see can be conveniently divided into environmental, material, and mechanical. Environmental factors include the electrochemistry of see such as pH, potential, water chemistry effect. Material factors include the bulk, and microchemical and microstructural effect. Mechanical factors include yield strength, fracture toughness, stress, stress intensity and strain rate dependence of cracking. ENVIRONMENTAL FACTORS
A complete description of see must treat both the thermodynamic requirements and kinetic aspects of cracking. Thermodynamic conditions for anodically assisted see include the requirement that dissolution or oxidation of the metal and its dissolution in the electrolyte must be thermodynamically possible and that a protective film, such as an oxide or salt, must be thermodynamically stable. The first condition becomes a requirement because, without oxidation, crack advance by dissolution would not result. That a process is controlled by anodic dissolution does not indicate that the total crack extension is the sum of the total number of coulombs of charge exchanged at the crack tip. There are processes in which crack advance is controlled by anodic dissolution but in which the total crack length is greater than can be accounted for by the total charge transfer. However, it is important to note that, even if crack advance results from a brittle crack jump process, if the brittle crack advance process is initiated and controlled by anodic dissolution, the crack growth rate will be a function of the current density. A thermodynamic requirement of simultaneous film formation and oxidation of the underlying material led to the identification of critical potentials for the presence or absence of see. An example of these critical potentials is given in Figure 1 for a passive film-forming material such as stainless steel. Zones 1 and 2 are those in which transgranular stress-corrosion crack (TGSee) growth is most likely to occur. TGSee occurs in zone 1 because the material is in transition from active corrosion to passive film formation such that the simultaneous conditions for film formation on the crack walls and corrosion at the crack tip are met. A similar condition exists in zone 2, with the added factor that these potentials are at or above the pitting potential so that cracks can be initiated by pitting. Intergranular stress corrosion cracking OSee) occurs over a wider range of potentials than those shown for zones 1 and 2 because chemical inhomogeneities at the grain boundary produce a different electrochemical response relative to the bulk material. Therefore, passive crack walls and active crack tips can result over the potential range from zone 1 to zone 2. Identification of critical potentials for see has led to the use of electrochemical methods to assess stress-corrosion susceptibility. eritical potentials for see can also be related to potential-pH stability diagrams (Pourbaix diagrams), since these diagrams describe the conditions at which film formation and metal oxidation will occur. An example for mild steel is given in Figure 2, where see is associated with potentials and pHs at which phosphate, carbonate, or magnetic films are thermodynamically stable while the species Fe + 2 and HFeO- 2 are metastable. The effect of many environmental parameters such as pH, oxygen concentration, and temperature on the thermodynamic conditions for see can be related to their effect on the potential-pH diagrams (Figure 2) or on the material potential relative to the various stability regions. Application of potential-pH diagrams for identifying specific conditions at which see will occur is limited by a number of factors such as the availability of these diagrams for complex solutions and for the temperatures of interest and the substantial deviation of the chemistry of a crevice or crack from the bulk solution chemistry. Also, the electrode potential at the crack tip can differ from the free surface of the material. These differences arise from the need for diffusion of metal ions away from the actively corroding crack tip, migration of anions into the crack, reactions along the crack walls, convection of the electrolyte in the JOURNAL OF METALS. December 1987
crack, and, in some cases, by gas bubble formation induced potential drops. Efforts to measure the local crack tip chemistry and potentials are restricted by crack sizes, which can be substantially smaller than 1 J.Lm. Therefore, mathematical modeling of transport within cracks and the resulting crack tip chemistry, reactions, and potentials has been actively pursued in recent years. A knowledge of the thermodynamic conditions at which see can occur is insufficient without a corresponding understanding of the kinetics of crack growth, because the life of a component may be adequate if the crack growth rate is sufficiently slow, even though see is thermodynamically possible. As in the thermodynamic conditions for see, environmental parameters such as potential, pH, oxygen concentration, temperature, and crack tip chemistry have a strong effect on the crack growth kinetics. The crack tip reactions and rate-determining steps controlling the crack growth rate are specific to alloy-environment combinations. Further, crack growth rate depends on the crack advance process even though it is controlled by electrochemical reactions. For the case of a crack growing by anodic dissolution alone, the total crack advance is EI..function of the total charge transfer at the crack tip, while the crack velocity is a function of a crack tip current density. For a crack growing with mechanical fracture, the total crack advance may exceed the total charge transfer at the crack tip, but the crack velocity may still be controlled by the crack tip current density. A limiting velocity can be described for a crack advancing under pure anodic dissolution by the following faradaic relationship: da _ iaM dt - zFp
Fe l
0 '--Fe'
HFeO,
o
Figure 2. Relationship between pH/potential conditions for stress corrosion cracking of mild steel in various environments. (after Parkins)
(1)
where ia = anodic current density of a bare surface, M = atomic weight, z = valence, F = Faraday's constant, and p = material density. Parkins l has shown this relationship between the bare surface current density and the crack propagation rate is applicable to a wide variety of materials, as presented in Figure 3. This relationship assumes that the crack tip is maintained in a bare condition while the crack walls are relatively inactive. A number of factors can reduce the crack velocity below that given by Equation 1 and in Figure 3. The most widely examined crack growth retardation process is that resulting from the crack tip being covered by a film for some fraction of time. The process of crack growth in the presence of a film at the crack tip has been described by several mechanisms, such as slip dissolution and passive film rupture. Other factors that can reduce the crack growth rate below that given by Equation 1 are limits in the diffusion rate of species into and out of the crack tip, crack deflection away from the principal stress, and changes in the local material chemistry. A mechanical fracture process in see can produce crack velocities exceeding that given by Equation 1 by some magnification factor, which can be as large as 100. The mechanism of brittle fracture induced by see is thought to involve the formation of a corrosion product at the crack tip in which a cleavage crack can initiate and propagate some depth into the ductile substrate. As mentioned earlier, this process would produce crack lengths that exceed those accounted for by the total charge transfer/metal oxidation process but would be dependent on the crack tip corrosion rate since the formation rate of the corrosion product would depend on the corrosion rate. Early observations and proposals of this process were made by Edeleneau and Forty,2 Pugh and co-workers,a-s and more recently by Newman and Sieradzki. 6 While the electrochemical conditions controlling the brittle see process have not been carefully cataloged, the main issue for many systems is whether cracking can be completely described by anodic dissolution or whether a mechanical fracture process is involved. The greatest uncertainty involves the possibility of a transition between anodic and mechanical see process which is affected by electrochemistry, material chemistry, or mechanics. MATERIAL CHEMISTRY AND MICROSTRUCTURE
The relationship among material chemistry, microstructure and see is equally as complex as the relationship between the environment and see. Bulk alloy composition can affect passive film stability and phase distribution (e.g., er in stainless steel), minor alloying elements can cause local changes in passive filmforming elements (e.g., e in stainless steel causing sensitization), impurity elements can segregate to grain boundaries and cause local differences in the corrosion rate (e.g., P in nickel or nickel-based alloys), and inclusions can cause local crack tip chemistry changes as the crack intersects them (e.g., MnS in steel). Also, alloys can undergo dealloying, which is thought to be a primary method by which brittle see initiates. Material chemistry and microstructure effects on IGSee generally can be divided into two categories: grain boundary precipitation or grain boundary segregation. Examples of grain boundary precipitation effects include carbide JOURNAL OF METALS. December 1987
14 pH2~ t.
•
o . 6 ..
CSt...... HO J CS1"'~OH
•
F... 'tlC$I.....
•
C Stnlll"l CO~MCo.ll
~l
C) "I Sf 5t... '" NlVC-, C Sf'" OlIo CO CO, ",0
I '•.
i
£
o )(
Aj7"", ...
~
o ........... NH~
l> /
..
'0
~ U I ••
•
'_1
0.-
.~ l1li
'.
I•
s·
Figure 3. Relationship between average stress corrosion crack velocity and the total oxidation rate for a variety of materials and environments. (after Parkins)
33
BOO rl----------------------------------,
C
700
600
OQ15wt%C
500
001
01
1.0
10
100
Time, h
Figure 4. Schematic time-temperatureequisensitization curves for Type 304 SS with various carbon contents. (after Bruemmer)
34
precipitation in austenitic stainless steels and nickel-based alloys, which causes a depletion of chromium adjacent to the grain boundary and intermetallic precipitation in aluminum alloys, which are anodically active. Grain boundary segregation of impurities such as P, S, C, and Si can produce a grain boundary that is up to 50% impurity within a 10-20 A-thick region. These impurities can alter the corrosion and mechanical properties of the grain boundary and hence can cause cracking by anodic dissolution and perhaps mechanical fracture. The IGSCC of austenitic stainless steel is primarily dependent on the nature of the chromium-depleted zone, which is generally explained by the depletion of a passive film-forming element along a continuous path through the material. The stress-corrosion susceptibility and crack growth rate of stainless steel can be described by the degree of sensitization (DOS) as measured by corrosion tests such as the Strauss or electrochemical potentio-kinetic reactivation (EPR) tests. Quantitative comparisons between susceptibility as measured by the presence or absence of IGSCC in a SCC test or the crack growth rate and the DOS or chromium depletion parameters have been successful in cases where sufficient data have been available, but these correlations are limited to specific alloys, environments, and stress conditions. An example of the time/temperature/DOS curves at several carbon concentrations is shown in Figure 4 from results of Bruemmer et al. 7 The curves represent the conditions necessary for the development of a constant DOS with the effect of bulk carbon concentration on the time needed to develop a sensitized microstructure. Therefore, the most common method for reducing the possibility of developing a sensitized microstructure is to reduce the carbon concentration or to control the thermal history of the material. Given that the material is sensitized, control of the environment and stress conditions can be used to reduce the crack propagation rate. The cause of IGSCC in nickel-based alloys is more complex than in austenitic stainless steel. Chromium carbide precipitation and chromium depletion occur in nickel-based alloys such as alloy 600 as they do in austenitic stainless steels, but a clear connection between the presence of this microstructure and IGSCC has not been made. Carbon solubility is considerably lower and chromium diffusion is faster in nickel-based alloys than in austenitic stainless steels. Thus, the nose in the carbide precipitation or sensitization curves shown in Figure 4 shifts to shorter times and higher temperatures. Significant heat-to-heat variations in the chromium depletion of Alloy 600 have been observed for a given heat treatment. This variation is attributed to the sensitivity of Alloy 600 to carbon concentration and to the variability in mill-anneal heat treatment conditions. Chromium depletion is thought to accelerate cracking in oxygenated water but has not been identified as a controlling factor in deaerated water or caustic environments at 300°C. The rapid carbide precipitation kinetics in nickel-based alloys allows the development of a "healed" microstructure where carbide growth is complete, and the chromium profile is eliminated with relatively short heat treatment times. Aluminum alloys are also noted for the occurrence of IGSCC in aqueous environments, including NaCI solutions. The details of the IGSCC process in aluminum alloys are very complex and vary with alloy composition, but some features can be summarized. Grain boundary precipitation has been identified as a contributing factor to IGSCC in aluminum alloys, and galvanic effects between the precipitates and matrix are considered important. In some cases, the precipitates are anodic and, in others, cathodic to the matrix. Peak-aged materials that give maximum hardnesses and microstructures in which slip bands are produced upon staining are considered vulnerable to SCC. Overaged structures are considered less susceptible. The mechanism of crack growth is thought to be a combination of local anodic dissolution and hydrogen embrittlement. Whether crack extension is by anodic dissolution or by hydrogen embrittlement remains an open question. In aqueous solutions, however, the primary source of hydrogen is anodic dissolution. Ferritic steels exhibit IGSCC in hot nitrate, caustic, carbonate and a variety of other environments. The presence of IGSCC is dependent on the electrochemical potential, as IGSCC is predominant at potentials in the active-passive transition. Early studies of this effect identified carbon segregation as the primary element of concerns where the carbon atoms were said to provide suitable imperfection sites for adsorption of nitrate in an adsorption-induced crack growth mechanism. More recently,9.10 phosphorus segregation has been associated with IGSCC of iron alloys in nitrate and caustic solutions. Grain boundary concentrations as low as 2 to 3 at.% have been sufficient to alter the passivity of iron in hot nitrate solutions. A complexity of SCC in ferritic steels is the susceptibility of such steels to IGSCC and to hydrogen-induced subcritical crack growth. The temperatures and electrochemical potentials at which IGSCC and cathodic hydrogen-induced subcritical crack growth occurs are generally not the same. Stress corrosion tends to dominate at temperatures above about 50°C and at potentials in the active-passive transition. Hydrogen effects are predominant at temperaJOURNAL OF METALS. December 1987
Table I. Initiating Layers for Discontinuous Stress-Corrosion Cracking
Metal or Alloy a-Brass, Cu-Al Au-Cu Fe-Cr-Ni a-Brass Copper Ferri tic Steel
Ti Alloys • Al Alloys, Steels
Environment Ammonia FeCl a Acid Sulphate Chloride Hydroxide H-T Water Nitrite, Etc. Nitrite Ammonia (Cupric) H-T Water Phosphate Anh. Ammonia CO/C0zfH20 CSzfH20 Chloride, etc. Various
Initiating Layer De-Alloyed Layer (Cu) De-Alloyed Layer (Au) De-Alloyed Layer (Au) De-Alloyed Layer (Ni) De-Alloyed Layer or Oxide De-Alloyed Layer or Oxide Oxide Oxide Porous Dissolution Zone Oxide Oxide (?) Nitride Carbide Carbide Hydride Near-Surface Hydrogen
tures below 50°C, at more cathodic potentials and lower pHs. Grain boundary impurity segregation of phosphorus, silicon, sulfur, and nitrogen has been reported in austenitic stainless steels. 11 However, no direct effect on IGSCC has been identified in high-temperature water. Phosphorus segregation has been shown to cause intergranular corrosion in highly oxidizing solutions, and impurity segregation of phosphorus and perhaps silicon has been suggested as a primary factor in irradiation-assisted stress corrosion cracking (IASCC), which occurs in the oxidizing in-core environment of light-water reactors.12 Phosphorus segregation can apparently contribute to IGSCC of austenitic stainless steels in high-temperature water if the carbon concentration of the alloy is lower than 0.002%. At this low concentration, there is virtually no sensitization, so the phosphorus segregation effect is observed. Numerous metallurgical factors affect TGSCC, including crystal structure and anisotropy, grain size and shape, dislocation density and geometry, yield strength, composition, stacking fault energy, ordering, and phase composition. Also, some of these factor's effects on TGSCC are related to the corrosion behavior of the alloy, which can be understood from potential-pH diagrams or polarization curves. Alloying effects on slip planarity are a key metallurgical factor in TGSCC. Planar slip occurs in alloys with low stacking fault energy, alloys containing ordered phases, or alloys exhibiting short- or long-range ordering. The consequences of planar slip on TGSCC have been explained by the slip dissolution model proposed by Staehle et al.,13 in which the passive film is ruptured by the emergence of a slip step. In high-chloride environments, evidence was presented that preferential corrosion occurred along the high dislocation density plane created by planar Slip.14 A number of crack growth processes were suggested based on the planar slip-localized corrosion process (e.g., control of crack advance solely by anodic dissolution of the slip plane, brittle fracture of the corrosion product or tarnish film formed along the localized corrosion path, and the tunneling process by which corrosion along the slip plane branches out into tunnels accompanied by mechanical fracture of the remaining ligaments). A significant difficulty associated with the slip-dissolution mechanism is the nature of the TGSCC fracture surfaces, which are generally not on the slip planes and have cleavage-like features. For instance, for admiralty brass tested in aqueous NH3 and Al + 5.5Zn + 2.5Mg tested in aqueous NaCI, the primary fracture facets are (110) planes, while for stable austenitic stainless steel tested in aqueous MgCl2 at 155°C, the primary facets are on (100) planes. There is also evidence that crack advance in brass occurs in a discontinuous manner.5 Several mechanisms for the development of a cleavage crack in ductile fcc alloys have been presented, but a definite correlation has not emerged. One concept that has received considerable attention is that proposed by Newman and Sieradzki,6 in which rapid crack advance that begins in a brittle film at the crack tip induces cleavage fracture of the ductile material ahead of the crack tip. The brittle films that Newman and Sieradzki regard as the initiating layers are given in Table I and include a dealloyed layer, oxide, nitride, carbide, or hydride. A dealloyed layer acting as a cleavage crack initiator is suggested for brass, Cu-Au, and FeCr-Ni alloys. MECHANICAL FACTORS
There is more commonality in the mechanical aspects of stress corrosion between IGSCC and TGSCC and between various materials than there is with environmental and metallurgical aspects. As pointed out earlier, many of the environmental or metallurgical factors are specific to a given material environJOURNAL OF METALS. December 1987
35
ment combination such that a metallurgical factor may be significant in one environment, but not in another. Threshold stress intensities and stresses, the presence of a stress-independent crack growth regime, and a dependence of crack growth rate on strain rate are features common to a variety of environmentally induced crack growth processes and a variety of materials. Anodic and cathodic controlled processes show many of these common features. The stress intensity dependence of environmentally induced subcritical crack growth is shown schematically in Figure 5 along with a relationship for the linear elastic mode I stress intensity. The crack velocity-stress intensity relationship during sec of a wide variety of materials is characterized by a stage I regime in which the crack velocity has a strong stress intensity dependence and a stage II regime in which the crack velocity has a weak stress intensity dependence. The following relationship is frequently used to express the stage I crack velocitystress intensity relationship: da/dt = A Km where A is a constant, K is the stress intensity and m is the stress intensity exponent. Values of m for a variety of alloys and test environments have been reported l5 and range from 7-24. A crack growth rate-strain rate relationship for stress corrosion cracking has been presented as an integral part of passive film models of stress corrosion such as that proposed by Ford and Andresen. l 6 Their results for IGSee of Type 304 SS in high temperature water is shown in Figure 4 where the crack growth rate 1
being proportional to (E2). The strain rate dependence has been explained by Ford as a result of the passive film rupture rate and the subsequent increased anodic corrosion rate at the exposed crack tip. At high strain rates, it is expected that FUTURE TRENDS
Much is known about the phenomenology of SCC in a number of common engineering matenals, including ferritic steels, austenitic stainless steels, nickel-based alloys, aluminum alloys, and titanium alloys. This phenomenological knowledge includes envLronmentaL, microstructural, microchemical, and mechanicaL conditions which contribute to SCC. DetaLis about the mechamsm of cracking in each of these alloy systems in varlOUS environments are much sketchier. Therefore, future research must continue to probe for an understanding of the physical processes which cause SCC. This understanding is required for the design and development of materiaLs which can perform In corroswe environments and for the development of reliabLe life predictive capabilities. Other areas of future research and development In SCC includes crack InLtzatLon modeLing and analysis, crack Up chemistry modeling and analysis, crack length effects on SCC, JASCC, and evaluation of SCC in new materials such as monalithic aluminides and ceramics and metal, aLuminide and ceramic matrix composites. SCC frequently initiates at pre-exlsting or corrosion-induced features. These features may include grooves, laps, or burrs resulting from fabncatlon processes, cleaning operattons, such as pickling or corrosion at incluSLOns or grain boundaries. The transition between pitting and cracking is dependent on the same parameters that control SCC (i.e., the eLectrochemistry at the base of the pit, pit geometry, ma-
36
terial microstructure and chemlstry, and stress and strain rate at the base of the pit). A detailed description of the relationship between these parameters and crack Initiation has not been deveLoped because of the difficulty in measuring crack initiatlon. Therefore, improved tools to measure the transitLon from localized corroSLOn at an incLusion, grain boundary, etc., to an active stress corrosion crack are needed before the phenomenology is understood or quantitative modeLs can b developed. Techniques which are under development include electrochemLcal noise, surface reslstivity, and acoustic emission. EvaLuation of these techniques is in the early stages and their limits for early crack detection are not known, but they all show promlse for helping the development in this cntical area. A key to understanding the transition from localized corrosion to active SCC lS a knowledge of the local eLectroLyte chemistry. The small dimensions of pits and cracks precLudes use of micro-electrodes to determine the local conditions. Crack tip chemistry modeLs can be used to evaluate the effects of crack length and geometry, material lnLcro-chemistry and electrolyte on the pH, potentiaL and crack tip corrosion rate. Turnbull and ThomaS l have developed a steady-state mass transport model considering dlffusLOn and migration for steels with crevices in solutions of 3.5% NaCL. Anodlc dissolution, hydrolysis of the ferrous LOns and cathodic reduction of hydrogen ions and water were considered In the
model and were assumed to take plac, at both the crack tLP and waLLs. TheoreticaL predictions of the potential and pH were found to be in reasonabl agreement wlth values determined expenmen tally with micro-electrodes in slmuLated cracks. Crack length effects have also been consLdered uSing a slmLiar model by Gangloff and TumbuLl.2:l Danielson, Oster and Jones 2 :1 have applied the methodology of Turnbull and Thomas to intergranular craclls Ln mckel with segregated phosphorus or sulfur. In these cases, the crack wall chemistry is variable and was shown to affect the crack tip corrosion rate and hence, the crack growth rate. Under the conditions evaluated, phosphorus at the intergranular crack tip is oxidized and dissolves m the electrolyte and the crack walls are passivated nickel. Sulfur lS very surfac active and remains on the crack walls which keeps them actively corroding at actwe-passive potentials while sulfur is converted to sulfate at transpassive potentials which allows the crack waLLs and crack tip to passivate. Th computer model predicted that nelther of these conditions support crack growth, which agreed with experimentaL results. Future research in crack tIP chemIStry modeling will be useful for ldentifymg the cntical conditions which cause SCC. Comparison to measured crack tIP chemistries or measured experimental materials responses is necessary to calibrate the crack tip chemistry models. Once calibrated, they
JOURNAL OF METALS. December 1987
mechanical fracture will occur faster than SCC so that da/dt in Figure 6 would approach zero at strain rates exceeding 10-4S-1. The decrease in crack velocity with decreasing strain rate results from the increased time that the crack tip is in the passive condition with low anodic activity. STRESS CORROSION CRACKING MECHANISMS
Significant progress has been made in the last decade toward understanding the environment-induced crack growth processes in metallic materials. These mechanisms can be sub-divided into three basic categories: dissolution models, mechanical fracture-ductile fracture, and mechanical fracture-brittle fracture. Despite a better understanding of the physical steps involved in environmentinduced crack growth processes, analytical descriptions of the environmental, material, and stress dependence of crack growth are still relatively incomplete. This difficulty in describing environment-induced crack growth stems from the complexity of the phenomenon. Dissolutton models are among the earliest (and easily the most common) explanations for environment-induced crack growth. Dissolution-controlled SCC is defined as one in which the total crack length can be accounted for by the total anodic charge transfer at the crack tip. Crack growth processes in which the crack velocity is a function of the current density, but where the total crack length exceeds the total anodic charge transfer, are considered mechanical fracture models. The simplest anodic dissolution-controlled crack growth process occurs when there is no passive film at the crack tip, the metal ion is very soluble in the electrolyte and the pH is effectively buffered so that the crack tip environment closely resembles the bulk environment as proposed by Parkins. 1 In this
will be useful for probing a variety of conditions not easily attainable in the laboratory. These comparisons may be done on simulated cracks, or, perhaps, smaller micro-electrodes will be developed which allow the direct analysis of real stress-corrosion cracks. Life prediction for sec is a relatively new field which has recently received attention by Ford and Andresen. u In their work, Ford and Andresen have attempted to use combined phenomenological and mechanistic modeling to predict lifetimes of austenitic and ferritic steels in high-temperature water. Parkins25 has also been concerned with the overall issues in sec and how this information can be translated into improved materials performance in corrosive environments. Life prediction based on stress-corrosion modeling is not currently used in design since most designs involving corrosion are based only on qualitative sec data. If a material shows susceptIbility to sec in an accelerated type stress corrosIOn test, it is considered unsuitable for a parizcular application. However, a component can be designed with a finite stress corrosion crack growth rate much like designing with fatigue crack growth rates for design life. Therefore, it is expected that future developments in life-prediction methodology for sec will occur with further development ofanalytic descriptions of sec processes. There is currently a significant interest in the effects of both gamma and neutron radiation on the sec of materials. This interest stems from the potentwl use of metallic canister ma-
terials for long-term storage of nuclear wastes, in-core materials in light water reactors, and future water-cooled fusion reactors. sec of in-core components in LWRs has been observed in austenitic stainless steels and nickelbased alloys. Explanations for this phenomenon Include irradiation enhanced segregation of impurities such as phosphorus, sulfur, and silicon, grain boundary or near-grain boundary microstructural changes and for changes in toughness from irradiation hardening. A neutron fluence threshold of about UP' n lcm~ at 300'C has been identified for this phenomenon, which is required for the onset of sec. The role of gamma radiation, on the other hand, is thought to accelerate crack growth, but may not be necessary for cracking to occur. Future research in this subject must include direct measurement of the microstructural and microchemical changes occurring at the fluence threshold and the role of radiolysis on the stress corrosion process. A knowledge of the mechanisms of cracking would help in the development of more resistant alloys since all austenitic- and nickelbased alloys tested to date have shown very similar susceptibility. Another key aspect of materials design for sec performance includes evaluation of the sec behavior of new materials such as intermetallic compounds, composites, etc. Evaluation of the sec behavior of these materials is In progress and the results of these evaluatIOns are expected to identify new ways to modify materials behavior for
JOURNAL OF METALS. December 1987
sec performance. For instance, an evaluation of the sec of Mg-Sle by Evans2 f1 suggests that fiber or whIsker reinforcement can reduce the sec growth rate significantly, perhaps by reducing the crack tip opening rate. However, it is obvious that the combination of reinforcement and matrix must be chosen carefully for applications in corrosive environments as galvanic effects or localized corrosion at the matrix'reinforcement interface could lead to sec growth rate for composites which exceed the monolithic material. In high-temperature materials such as ceramics and ceramic composites, there is a need to evaluate the dynamic crack growth behavior of materials in corrosive environments. Also, the mechanisms of environmentally enhanced crack growth are relatively unknown since much of the environmental effects data in ceramics is in the form of corrosion effects on fracture toughness. Also, for ceramic composites, there is a need to identify the effects of environmental interaction with the reinforcement and its interface with the matrix. Advanced structural ceramics are currently being developed for their fracture resistance and strength at ambient temperature and, in some cases, at elevated temperatures and for their stress-rupture properties. Frequently, these materials will have to perform in a high-temperature corrosive environment. Both phenomenological and mechanistic information on environment effects are needed.
37
o
ll~
Stage II ..........
o
<>
"
Q.
]
ec:
K
~
~
Ao
J ;c
~
~"
a = total tensile stress C = crack length '\ A, B = geometrical constants
C>
i
.3
Catastrophic
I
Failure
'-+-Subcritical Crack Propagation--+-...K,C Stress-Intensity
Figure 5. Schematic of typical stress intensity-stress cG:rosion crack velocity curve showing threshold for cracking and stages I. /I and Ill.
model, the crack velocity simply scales with the bare surface current density (Figure 3), and has been suggested to be valid for TGSCC and IGSCC. For IGSCC, this model is conceptually very simple when impurities are segregated to the grain boundaries, making them anodic to the surrounding grains. For TGSCC, this simple model presents more problems because of the absence of a preferred corrosion path. Conditions which cause the formation of a passive or protective film on an alloy can alter the crack velocity-bare surface current density relationship given by Parkins. While the crack velocity may still be proportional to the bare surface current density and the total crack length equal to the total anodic charge transfer, the crack velocity may be substantially retarded by the presence of a passive or protective film at the crack tip. In this case, crack extension occurs by periodic rupture of the film with anodic dissolution causing crack extension during the interval between film rupture and repair. Film rupture may occur by the emergence of a slip step in TGSCC of low stacking fault energy materials, or by fracture of the film by the crack tip stresses and strains. Details about the film rupture rate and repassivation rate and whether a continuous steady state process or a discontinuous crack extension process occurs remains an area of uncertainty. Ductile fracture models of SCC include the corrosion tunnel model by Silcock and Swann 17 and surface adsorption enhanced plasticity as proposed by Lynch. 18 In the corrosion tunnel model, the tunnels are thought to be flat corrosion slots which interconnect by shear fracture. Hence, crack extension occurs by a combination of corrosion and mechanical fracture with the shear ridges running parallel to the crack growth direction. While this geometry is consistent with observation of TGSCC, the plane on which corrosion tunnels should form and the fracture surface are not consistent. Concepts of adsorption enhanced plasticity leading to crack extension have been proposed by Lynch 18 and used to link liquid metal embrittlement and SCC. Details of the crack extension process have not been clearly worked out, but Lynch proposed that it occurs by alternate slip in conjunction with microvoid formation ahead of the crack tip. Basically, the concept relies on adsorption enhancing the nucleation of dislocations from the surface which results in increase strain at the crack tip for a given stress. The role of the environment is to enhance the nucleation of surface dislocations with the plasticity, causing microvoids and fracture at crack tip stresses and strains less than would occur in the absence of the environment. Brittle mechanical fracture models of SCC include those in which the crack propagates through a thick tarnish film which forms at the crack tip, sits idle while a new film forms and then proceeds to propagate through the next layer. An alternate model-the film induced cleavage model-was first proposed by Edeleneau and Forty 2 and considers that the crack propagates by cleavage into the ductile metal ahead of the brittle film. It has been proposed by Sieradzki that this could occur if the crack achieves a velocity within the brittle film such that the crack traverses the film-matrix interface before dislocation nucleation can result in crack tip blunting. This mechanism is consistent with many of the fractographic features and events occurring during TGSCC while many of the other models are only consistent with a few of these observations. MATERIALS DESIGN AND DEVELOPMENT
10- 5
10- 6
10- 7
10- 8
10 -91
I
10- 8
I
'0-7
I
10- 6
10-5
I
10- 4
Log Strain Rate, S - 1
Figure 6. Schematic of stress corrosion crack growth rate dependence on strain rate for Type 304 55 in high temperature water. 38
Examples of materials design and development for SCC are either non-existent or very rare. However, there have been a few examples of a material being selected for a given application and exhibiting SCC or an alloy development or selection processes being initiated to identify materials with greater SCC resistance. The development of INCONEL Alloy 690 for nuclear steam generators is such as example. 19 An increase in the chromium concentration of this alloy relative to its predecessor, INCONEL Alloy 600, produced an alloy which exhibited SCC resistance in highly oxidizing environments and in the presence of various water impurities. In another example, the IGSCC of Type 304 SS used for nuclear reactor piping has been greatly minimized by the use of low-carbon, nucleargrade Type 316 SS. This example does not represent an alloy development effort as much as a materials selection for SCC performance. SCC also includes crack growth by cathodically produced hydrogen. This form of SCC is a problem in deep sour gas wells where the presence of hot CI- and H 2 S are the primary chemical species involved in SCC. While the HASTELLOY alloy C-276 exhibits good SCC performance in this high-temperature environment, crack growth was observed at room temperature when the material was galvanically coupled to carbon steel. Since the materials in sour gas wells must perform at room temperature as well as elevated temperatures, the development of an alternative alloy for this application was undertaken. 2o The development of long-range order and the increase in yield strength was associated with the room temperature hydrogen-induced crack growth observed in HASTELLOY alloy C-276. By increasing the Fe concentration from 5 wt.% to 15-20 wt.% and by JOURNAL OF METALS. December 1987
making minor modifications in the eo, w, Mn, and e concentrations, the performance of these new materials were developed to be superior to e-276 in cathodic hydrogen at room temperature. It is clear from these examples that materials design and development for see resistance is dependent on understanding the crack growth mechanism. Since see is controlled by numerous variables, an empirical approach is time consuming, costly, and risky. Therefore, one would expect materials designed for see resistance and the development of improved mechanistic understanding and analytical models of see to advance sequentially. Thus, enhanced research on see mechanisms should lead to improved see resistant alloys. References 1. RN. Parkins, Br. Corrosion J., Vol. 14 (1979), p. 5. 2. C. Edeleneau and AJ. Forty, Phil. Mag., Vol. 46 (1960), p. 521. 3. J.A Beavers and E.N. Pugh, Metall. Trans. A., Vol. 11 (1980), p. 809. 4. M.T. Hahn and E.N. Pugh, Corrosion, Vol. 36 (1980), p. 380. 5. E.N. Pugh, Atomistics of Frocture, RM. Latanision and J.R. Pickens, Eds., Plenum Press, New York, 1983, p. 997. 6. RC. Newman and K. Sieradzki, Chemistry and Physics of Frocture, RM. Latanision and RH. Jones, Eds., Martinus Nijhoff Publishers, Dordrecht, 1986, p. 597. 7. S.M. Bruemmer, L.A Charlot and D.G. Atteridge, "Evaluation of Welded and Repair.Welded Stainless Steel for Light Water Reactm··(LWR) Service," NUREG/CR~3918, December 1984. 8. L. Long and H. Uhlig, J. Electrochem. Soc., Vol. 112 (1964), p. 1965. 9. J. Kuppa, H. Erhart and H. Grabke, Corr. Sci., Vol. 21 (1981), p. 227. 10. N. Bandyopadhyay, R.C. Newman and K. Sieradzki. Proc. 9th International Congress on Metallic Corrosion, Toronto, Canada, 1984. 11. A Joshi and D.J. Stein, Corrosion, Vol. 28 (1972), p. 321. 12. R.H. Jones, Proc. 2nd International Symposium on Enuironmental Degradation of Materials in Nuclear Power SystemsWater Reactors, American Nuclear Society, 1985, p. 173. 13. RW. Staehle, Corrosion, Vol. 26 (1970), p. 451. 14. P.R. Swann, Corrosion, Vol. 19 (1963), p. 3. 15. R.H. Jones, M.J. Danielson and D.R. Baer, Proc. 20th National Fracture Mechanics Symposium, Lehigh University, June 23-25, ASTM, 1987. 16. F.P. Ford, "Mechanisms of Environmental Cracking in Systems Peculiar to the Power Generation Industry," EPRI NP-2589, Electric Power Research Institute, Palo Alto, California, 1982. 17. J.M. Silcock and P.R. Swann, Environment-Sensitive Fracture of Engineering Materials, Z.A. Foroulis, Ed., TMSAlME, Warrendale, PA, 1979, p. 133. 18. S.P. Lynch, J. Mater. Sci., Vol. 20 (1985), p. 3329. 19. J.R. Crum and R.C. Scarberry, J. Mater. Energy Systems, Vol. 4 (1982), p. 125. 20. AI. Asphahani and H.M. Tawainey, Corrosion and Corrosion Protection, R.P. Frankenthal and F. Mansfield, Eds., The Electrochemical Society, Inc., Pennington, NJ, 1981, p. 154. 21. A Turnbull and J.G.N. Thomas, J. Electrochem. Soc., Vol. 129 (1982), pp. 1412-1422. 22. RP. Gangloff and A Turnbull, Modeling Environmental Effects on Crock Growth Processes, R.H. Jones and W.W. Gerberich, Eds., TMS, Warrendale, PA, 1986, p. 55. 23. M.J. Danielson, C. Oster and R.H. Jones, J. Electrochemical Society, 1987, in press. 24. F.P. Ford and P.L. Andresen, Proc, 3rd International Symposium on Environmental Degradation of Materials in Nuclear Power Systems-Water Reactors, 1987, TMS, Warrendale, PA, in press. 25. R.N. Parkins, Proc. Mechanics and Physics of Crack Growth: Application to Life Prediction, Mater. Science Engineering, in press. 26. J.T. Evans, Acta Metall., Vol. 34 (1986), p. 2075.
ABOUT THE AUTHOR _ _ _ _ __
Russell H. Jones received his Ph.D. in metallurgy from the University of California, Berkeley, in 1971. He is currently technical leader of the Metals Research Group of the Materials Sciences Department of BattelleNorthwest in Richland, Washington. Dr. Jones is also a member of TMS. If you want more information on this subject, please circle reader service card number 52. ~ FATIGUE CRACK· GROWTH ~
OV'R
I ~:.,~~~!§®1 I
llVVITED +
ublished monthly by PThe Metallurgical Society, the acts as an authoritative literature resource,
.Journal of Metals reporting on scholarly work and elucidating the state·of·the·art processing, fabrication, and design of ordered solids. Written at a B.Sc. level. .Journal of Metals appeals to the academic and industrial audience alike and strives to facilitate the transfer of technology from the laboratory to the marketplace.
A
uthors wishing to publish a paper in the .Journal of Metals should first obtain the JOM Author Kit and a copy of the current Technical Emphasis Calendar. These tools will guide the author in determining the format, length, and timing of a submission.
T
o receive the JOM Author Kit and Technical Emphasis Calendar, call or write: .Journal of Metals The Metallurgical Society 420 Commonwealth Drive Warrendale, Pennsylvania 15086 (412) 776-9070
The patented KRAK-GAGE® is a thin-film bondable DC transducer for measuring fatigue crack growth, based on the DC potential drop technique. KRAK·GAGES ® are available in nine sizes for crack lengths of 5 mm to 200 mm. System output is 10 VDC with the FRACTOMAT ®, a 2-channel readout and control instrument. Conventional adhesive bonding is used for metallics and non· metallics, specimens, compo· nents and structural elements. Send for literature, applications, test data, pricing and international user list. TTl DIVISION, HARTRUN CORPORATION 88 RIBERIA ST., ST. AUGUSTINE, FL32084 TEl: (904) 824-8145 TELEX: 29{)796
FRACTURE MECHANICS
Circle no. 8 on reader service card.