Corrosion and Corrosion Fatigue Predictive Modeling State of the Art Review

Prepared for Mr. Garth Cooke NCI Information Systems Inc. 3150 Presidential Drive, Building 4 Fairborn, OH 45324

Prepared by David W. Hoeppner, P.E., Ph.D. V. Chandrasekaran, Ph.D. FASIDE International Inc. 1146 S. Oak Hills Way Salt Lake City, UT 84108-2026

Submitted as a part of the research program Modeling Corrosion Growth on Aircraft Structure Subcontract Number NCI USAF 9061-008

April 30, 1998

Introduction
The phases of life of a structure may be classified according to the division in the table below. NUCLEATION OR FORMATION OF DAMAGE BY A SPECIFIC, PHYSICAL OR CORROSION DAMAGE PROCESS INTERACTING WITH THE FATIGUE PROCESS IF APPROPRIATE. CORROSION AND OTHER PROCESSES MAY ACT ALONE TO CREATE THE DAMAGE. A TRANSITION FROM THE NUCLEATION STAGE TO THE NEXT PHASE MUST OCCUR. PHASE L1. MICROSTRUCTURALLY DOMINATED CRACK LINKUP AND PROPAGATION (SHORT OR SMALL CRACK REGIME). PHASE L2. CRACK PROPAGATION IN THE REGIME WHERE EITHER LEFM, EPFM, OR FPFM MAY BE APPLIED BOTH FOR ANALYSIS AND MATERIAL CHARACTERIZATION (THE LONG CRACK REGIME). PHASE L3. FINAL INSTABILITY. PHASE L4. Thus, the total life (LT) of a structure is T 1 2 3 4. Figure 1 on the next page presents a depiction of the degradation process. The regions shown, e.g. 1, 2, 3, and 4, illustrate the portion of life, on the abscissa, and the corresponding growth in discontinuity size plotted schematically on the ordinate. This report concentrates on the phases L1 and L2. That is, the corrosion process that results in the generation of a specific form of corrosion generating a specific form of discontinuity that is not necessarily a crack like discontinuity, and the development of short cracks and their propagation. The requirement of the community to come up with design methods to deal with corrosion or other degradation, fatigue, creep, and wear, is essential and some of the elements are depicted in Figure 2. This figure illustrates that most of the quatitative methods that have been developed used the concepts of mechanics of materials with an incorporation of fracture mechanics. The major sections of the report that follows will discuss the following three major areas: Corrosion in Aircraft Structural Aluminum Alloys, Pitting Corrosion, Microstructure and environment effects on short crack behavior of materials. Subsequent to the presentation of the technical issues that relate to the development of models that will assist in the development of estimation of the effects of corrosion on structural integrity some concluding remarks and recommendations are presented.

L = L +L +L +L

The degradation process

4 1 Discontinuity Size A 2 3

Life A= "FIRST" detectable crack 1. Nucleation phase, "NO CRACK" 2. "SMALL CRACK" phase-steps related to local structure (Anisotropy) 3. Stress dominated crack growth, LEFM, EPFM 4. Crack at length to produce instability

Figure 1. A depiction of the degradation process (after Hoeppner-1971,1986

ii

METHODS FOR EACH LIFE PHASE

NUCLEATION Material failure mechanism with appropriate stress/strain life data Nucleated discontinuity (not inherent) type, size, location Presence of malignant D*, H* Possibility of extraneous effects Corrosion Fretting Creep Mechanical Damage "SMALL CRACK" GROWTH Crack Prop. threshold related to structure (micro) Structure dominated crack growth Mechanisms, rate Onset of stress dominated crack growth Effects of R ratio Stress state Environment Spectrum -waveform STRESS DOMINATED CRACK GROWTH Fracture mechanics similitude boundary cond. Data base** Appropriate stress intensity factor Initial D*, H* size, location, type Effects of R ratio Stress state Environment Spectrum -waveform FAILURE (FRACTURE) K LEFM EPFM
etc.

Ic

C.O.D. Tensile/ compressive buckling

t chem T

t chem T

Figure 2-Methods for each life phase (after Hoeppner-1971,1986). NOTE: Initiation as frequently used by the technical community is usually part of the nucleation (or formation), short crack growth, and stress dominated crack growth phase of life. One is never sure however how much of the life is taken up by the traditional use of the initiation concept. To avoid this we have used the term initiation herein only to refer to the beginning of a specific degradation process such as corrosion, fatigue, or initiation of crack propagation. As depicted in Figure 1 what often is referred to as initiation is life to a certain detecable crack size or damage size. This is a critical distinction in that use of first crack detection concepts, or related on condition evaluation terms, forces the designer to think about inspectability and detectability of specific forms of degradation. As well, it is imperative that the technical community develop an understanding of the nucleation and growth phases of degradation processes as is discussed extensively in the following report.

iii

TABLE OF CONTENTS

Introduction ..........................................................................................................................i LIST OF TABLES .............................................................................................................vi LIST OF FIGURES ............................................................................................................vi 1 Corrosion in Aircraft Structural Aluminum Alloys .........................................................1 1.1 Introduction ...................................................................................................................1 1.2 Intergranular and Exfoliation Corrosion .......................................................................6 1.3 Corrosion fatigue...........................................................................................................8 1.4 Corrosion pillowing and its effect on structural integrity of aircraft lap joints.............9 1.5 Pitting corrosion ..........................................................................................................10 1.6 Pitting corrosion fatigue ..............................................................................................12 1.7 Fretting corrosion and fatigue .....................................................................................13 1.7.1 Mechanisms of fretting.........................................................................................14 1.7.2 Fretting fatigue in aircraft joints..........................................................................15 1.7.3 Reduction or prevention methods ........................................................................15 1.8 References ...................................................................................................................19 2 Pitting Corrosion ............................................................................................................23 2.1 Introduction .................................................................................................................23 2.2 Formation of passive films and their growth...............................................................23 2.3 Structure of the passive film in aluminum ..................................................................24 2.4 Pitting potential and induction time ............................................................................27 2.5 Pit growth rate and pit morphology.............................................................................28 2.6 Mechanisms of pit nucleation .....................................................................................29 2.6.1 Adsorption-Induced Mechanisms.........................................................................29 2.6.2 Ion migration and penetration models..................................................................32 2.6.3 Mechanical film breakdown theories-Chemico-mechanical breakdown theories 35 2.7 Pitting corrosion fatigue ..............................................................................................37 2.8 References ...................................................................................................................42 3 Microstructure and environment effects on short crack behavior of materials...........47 3.1 Introduction .................................................................................................................47 3.2 Effect of microstructure on the short crack behavior of materials...........................51 3.3 Stage I crack propagation studies and its relation to "short" crack growth behavior ..62 3.4 Environment effects on "short" crack behavior of materials.......................................65 3.5 Temperature effects on "short" crack behavior of materials .......................................66 3.6 Studies on transition of pit to small crack ...............................................................68 3.7 Conclusion...................................................................................................................68 3.8 References ...................................................................................................................70

iv

Appendix I -- A global view of short crack challenge in materials ...............................80 4 Conclusions and recommendations................................ Error! Bookmark not defined.

LIST OF TABLES Table I: Corrosion and fatigue issues in the US Air Force Aging Aircraft ------------------ 3 Table II Pit Nucleation Theories ------------------------------------------------------------------11 Table III Adsorption-Induced Mechanisms -----------------------------------------------------31 Table IV: Ion migration and penetration models -----------------------------------------------33 Table V Chemico-mechanical breakdown theories --------------------------------------------35 Table VI Pitting Corrosion Fatigue Models -----------------------------------------------------38 LIST OF FIGURES Figure 1. A depiction of the degradation process (after Hoeppner-1971,1986-------------- ii Figure 2-Methods for each life phase (after Hoeppner-1971,1986).------------------------- iii Figure 3. Design method for fretting fatigue in aircraft joints (Hoeppner-1993) ----------18

vi

1 Corrosion in Aircraft Structural Aluminum Alloys 1.1 Introduction Corrosion is an electrochemical reaction process between a metal or metal alloy and its environment [1]# . For corrosion to occur, four conditions must exist viz. an anode, a cathode, an electrolyte, and an electrical path (flow of electrons). The anode and the cathode could be of two dissimilar metals or anodic and cathodic cells could be formed in the same metal alloy because of the potential difference in the constituent chemical elements. Moreover, depending on the availability of oxygen (differential aeration cells) and electrolyte (differential concentration cells) on the surface of the metal alloy, special types of localized corrosion could occur. 2xxx (Al-Cu alloys) and 7xxx (AlZn alloys) series aluminum alloys are commonly used in manufacturing aircraft structural components. Depending upon strength and toughness requirements, different types of aluminum alloys such as 2024, 7075, 7178 are used for commercial and military aircraft fuselage skins, wing skins, and other extrusions and forging such as stringers, and fuselage frames. In general, 2024-T3 is used for skins and 7075-T6 for stringers and frames although many applications of these and other alloys in the 2xxx and 7xxx families exist. Lap or butt splices are the common configuration for longitudinal joints whereas butt joints are for circumferential joints. A common joining method is riveting and in some cases it is in combination with adhesive bonding. In older aircraft, spot welding also can be found. As Wallace and Hoeppner mentioned in their AGARD report on "Aircraft Corrosion: Causes and Case Histories", in the initial stages, corrosion is in the form of filiform or pitting in the interior and exterior of fuselage skins [2]. Moreover, as noted in their report, crevice corrosion between the riveted sheets in fuselage joints is a significant issue and it is usually associated with the trapped small "stagnant solution". Furthermore, depending upon the chemical conditions this could lead to a combination of pitting, galvanic or exfoliation corrosion. As well, it is recognized that fretting corrosion/wear in faying surfaces and within fastener holes plays a role in the corrosion mechanisms within aircraft joints [2]. The process of corrosion may start early in the process of manufacturing and continues when the aircraft enters its service. Therefore, it has been realized that the corrosion prevention and control program (CPCP) should be planned concurrently from the initial design until the aircraft is out of service.

Numbers in parentheses [ ] refer to references at the end of each section.

Many types of corrosion mechanisms such as intergranular, exfoliation, pitting, crevice, fretting, microbiologically influenced corrosion, stress corrosion cracking, and hydrogen embrittlement have been found to occur in aircraft structural aluminum alloys [2]. Moreover, the synergistic effects of corrosion and the loading conditions have been found to initiate the corrosion fatigue failure process and the stress corrosion cracking failure process of aluminum alloy aircraft structural components. As identified, recently, in a report by the National Research Council's National Materials Advisory Board [3], corrosion in aircraft structural joints would result in the following: (i) significant changes in the applied stress because of material loss as well as corrosion product buildup that may cause pillowing or bulging of aluminum alloy sheet, (ii) hydrogen embrittlement that may result in reduced toughness, strength, and ductility of the material, and (iii) increase in fatigue crack growth rates that may severely hamper the planned inspection intervals. These issues have been discussed in workshops presented for the USFAA and UCLA* as well as FASIDE Int. Inc.** workshops since 1971. In addition, Hoeppner has discussed the following other potential effects of corrosion on structural integrity. production of localized stress concentrations that act as crack nucleation sites. change of the structurally significant item (SSI) modification of the fail safety by any of the above. Moreover, recently, an attempt has been made to model loss of thickness due to crevice corrosion growth in a corroded lap joint***. Several metallurgical, mechanical and environmental factors influence the corrosion process in aluminum alloys [4]. Metallurgically induced factors include heat treatment, chemical composition of alloying element, material discontinuities such as the presence of voids, inclusions, precipitates, second phase particles, and grain boundaries as well as grain orientation. Environmental factors include temperature, moisture content, pH, type of electrolyte, and the time of exposure. Aircraft often are exposed to both external and internal environments. External surfaces of the aircraft are exposed to a variety of environments including rain, humidity, acid-rain, deicing fluid, industrial pollutants, hot and cold temperatures, dust, high content of deposits of exhaust gases
*

Aircraft Structural Fatigue, with D.W. Hoeppner et al., Course sponsored by the FAA Structural Integrity of New and Aging Metallic Aircraft, with M. Creager, T.R. Brussat, D.W. Hoeppner, and T. Swift, Short Course by UCLA Extension, Department of Engineering, Information Systems and Technical Management, Los Angeles, CA. ** Practical Considerations in Structural Fatigue and Damage Tolerant Design of New and Aging Aircraft, with D.W. Hoeppner, Course offered by FASIDE International Inc., in conjunction with the Division of Continuing Education, University of Utah, and the Quality and Integrity Design Engineering Center (QIDEC), Department of Mechanical Engineering, University of Utah, Salt Lake City, UT. *** X. Zheng, Corrosion Growth in Lap Joint Crevice, Presented at the First Technical Interchange Meeting on Corrosion Modeling, Organized by NCI, 15-16 October, 1997.

from engines, and salts. In addition, the inside of the aircraft is affected by condensed moisture, spilled beverages, cargo leak, deicing fluid, lavatory seepage, and accumulated water in the fuel tank as well as others. Moreover, aircraft are exposed to wide ranges of environment depending upon their route and geographical location viz. tropical, marine, industrial and rural [2, 16]. In both military and commercial aircraft, internal and external wing structures as well as the fuselage bilge areas and flight control surfaces are found to be most affected by corrosion in a marine and tropical environment [5]. The major causes of corrosion in aging aircraft as observed in Indonesian aging aircraft were found to be due to spillage of toilet liquid, contamination due to spillage or evaporation from the cargo compartment, and contamination due to high humidity [6]. In addition, in these aircraft, corrosion was often found in the area surrounding the cargo compartment, wing structure, and landing gear. The types of corrosion found in these aircraft were of exfoliation, galvanic, filiform, and stress corrosion and among these exfoliation corrosion was found in most cases [6]. Several structural issues such as exfoliation, pitting, stress corrosion cracking, fatigue cracking, fastener corrosion, wear, fatigue and corrosion, delamination and disbonds have been observed in the U.S. Air Force aging aircraft as shown in Table I [3]. For example, in C/KC 135 fleet, crevice corrosion in the spot welded lap joint/doubler and corrosion around the steel fasteners on the upper wing skin have been recognized as significant corrosion issues [7]. In the later case, as was noted, there was a possibility of moisture from condensation or deicing solution trapped around fastener heads forming a galvanic couple. This was observed to result in intergranular attack of the grain boundaries leading to exfoliation [7]. Table I: Corrosion and fatigue issues in the US Air Force Aging Aircraft [3] Type of aircraft C/KC 135 (Tanker aircraft) Issues Corrosion between fuselage lap joints and spot welded double layers, around fasteners in the 7178-T6 aluminum upper wing skins, between wing skins and spars, between bottom wing skin and main landing gear trunnion, between fuselage skin and steel doublers around pilot windows, Stress Corrosion Cracking (SCC) of large 7075-T6 aluminum forging (fuselage station 620, 820, and 960), corrosion and SCC of fuselage station 880 and 890 floor beams, wing station 733 closure rib, and corrosion in the E model engine struts.

Type of aircraft C-141B (Transport aircraft)

3 4

C-5 (Airlifter) B-52H (Bomber)

F-15 (Fighter aircraft)

F-16 (Fighter aircraft)

Issues Widespread Fatigue Damage (WFD) in the fuel drain holes in the lower surfaces of the wings, corrosion and SCC in the upper surface of the center wing, fatigue cracking and SCC around the wind shield, fatigue cracking in the stiffeners in the aft pressure door, SCC in the fuselage main frames, and corrosion in the empennage. SCC of the 7075-T6 aluminum mainframes, keelbeam, and fittings in the fuselage, 7079-T6 fuselage lower lobe and aft upper crown. Cracking in the bulkhead at body station 694, fatigue cracking in flap tracks and in the thrust brace lug of the forward engine support bulkhead, cracking in the side skin of the pressure cabin, aft body skins, and upper surface of the wing. Low-cycle fatigue cracking in the upper wing surface runouts, upper wing spar cap seal grooves, front wing spar conduit hole, upper in-board longeron splice plate holes, corrosion in nonhoneycomb structure including fuselage fuel tank, the outboard leading-edge structure of the wings, and the flap hinge beam. Cracking of the vertical tail attachment bulkhead at fuselage station 479, fuel vent holes of the lower wing skin, the wing attach bulkhead at fuselage station 341, the upper wing skin, fastener problems on the horizontal tail support boxbeam, and the ventral fin.

Type of aircraft A-10 (Attack aircraft)

E-3A (Airborne Warning and Control System)

10

E-8 (Joint Surveillance and Attack Radar System) T-38 Fatigue cracking in the lower surface of the wing, lower (Air training wing skin fastener holes, wing skin access panel holes, command aircraft milled pockets on the lower wing skin, and the fuselage upper cockpit longerons. SCC in the fuselage cockpit upper and lower longerons, fuselage forgings. honeycomb corrosion in the horizontal stabilizer (due to water intrusion), and the landing gear strut door

Issues Fatigue cracking in the wing auxiliary spar cutout of the center section rib at wing station 90, outer panel front spar web at wing station 118 to 126, outer panel upper skin at leading edge. Fatigue cracking in the center fuselage forward fuel cell floor at the boost pump, forward fuselage gun bay compartment, forward fuselage lower longeron and skin at fuselage station 254, and center fuselage overwing lower floor panel stiffeners. Fatigue cracking in the aft nacelle hanger frame, thrust fitting and the engine inlet ring assembly skin/frame. Fatigue cracking in the main landing gear shock strut outer cylinder. Exfoliation corrosion in the 2024-T351 aluminum lower wing skin, 7075-T6 aluminum upper wing at the leading edge, 2024-T3511 aluminum lower front spar cap, 7075-T6 aluminum fuselage bottom skin 2024-T3/7075-T6 aluminum fuselage side skin and beaded pan, and 2024T3511 aluminum horizontal stabilizer upper spar caps. Pitting corrosion in the 9Ni-4Co-0.3C steel wing attach fitting bushing and lug bore, main landing gear fitting attach bolts, 7075-T6 aluminum aft fuel cell aft bulkhead, and 2024-T351 center fuselage upper longeron. SCC in the wing attach bushing flange, and the main landing gear attach bolts. Fatigue and corrosion in the 7178-T6 rudder skins, and spoiler actuator clevis. Exfoliation corrosion in the 7178-T6 upper wing skin, leading edge slats, main landing gear door, fillet flap, fuselage stringer 23, and magnesium parts. Delamination and disbonds in the windows, floor panels, and nose radome core. Wear in the antenna pedestal turntable bearings. "Small" fatigue cracks in fastener holes in the 7075-T6 aluminum stringers, in the 2024-T3 aluminum skins.

Examination of C/KC 135 fuselage lap splices (stiffened aluminum lap joint) revealed that outer skin corrosion was predominantly intergranular and exfoliation [8]. Moreover, extensive cracking was noted at these sites in the outer skin. In addition, extensive "pillowing" with more than 300% change in volume due to corrosion products along the faying surfaces was observed. In the rivet/shank region, severe localized corrosion and intergranular corrosion were observed. The fracture of rivet heads was attributed to high local stress due to environmentally assisted cracking at the junction. As well, in this study, solution samples were collected from selected areas of lap splice joints and the solution analysis showed the presence of several cations such as Al3+, Ca2+, Na+, K+, and Ni2+ and also anions Cl-, SO42-, and NO3-. Subsequent potentiodynamic tests using solution containing these ions led to the belief that dissolution rates could completely penetrate the fuselage outer skin during service life [8]. Thus, in addition to fatigue cracking, different corrosion mechanisms occur in aircraft structures depending upon their location, geometry, exposure to environment, and loading conditions. Research studies conducted within the Quality and Integrity Design Engineering Center (QIDEC) at the University of Utah as well as other related studies are briefly discussed below. 1.2 Intergranular and Exfoliation Corrosion Exfoliation corrosion is believed to be a manifestation of intergranular corrosion. Intergranular corrosion results from either the segregation of reactive impurities or from the depletion of passivating elements at the grain boundaries. This makes the regions at or surrounding the grain boundaries less resistant to corrosion resulting in preferential corrosion. The high strength aluminum alloys such as 2xxx and 7xxx series are highly susceptible to intergranular corrosion [1]. Exfoliation corrosion is a form of intergranular attack that occurs at the boundaries of grains elongated in the rolling direction. The 7xxx series aluminum alloys are particularly less resistant to exfoliation corrosion because during heat treatment (to achieve maximum desirable strength) their constituent elements copper and zinc accumulate at grain boundaries leaving the adjacent region free of precipitates. As aluminum and aluminum intermetallic compounds are highly reactive in the emf series as well as aluminum is anodic to copper in the galvanic series, the resulting galvanic couples cause the grain boundaries to preferentially corrode (intergranular attack). McIntyre and Dow have related the localized corrosion problems in the 7075T7352 fuel tanks of underwater weapon systems to intergranular corrosion [9]. In their study, aluminum alloys 7075 and 6061 were exposed to artificial seawater containing 6

nitrate ions. It was observed that accelerated intergranular corrosion occurred in 7075 alloy. From the test results, they hypothesized that refueling the improperly cleaned fuel tank may cause the propellant in contact with the small quantity of sea water remaining in the fuel tank resulting in the release of nitrate ions from an hydrolysis process leading to reduced pH that may cause the dissolution of the oxide film (localized corrosion). They further hypothesize that corrosion eventually propagates to the bulk regions of the alloy due to intergranular attack by the preferential corrosion of reactive MgZn2 intermetallic compounds located at grain boundaries. This was found to be true for 7075 aluminum alloy but not for 6061 aluminum alloy because the later does not contain either Cu or Zn as alloying element [9]. Reducing the impurities such as iron and silicon as well as heat treatment modifications in aluminum alloys have resulted in increase in the resistant to exfoliation corrosion [10]. For example, overaged 7075-T7 alloy is more resistance to exfoliation corrosion when compared to 7075-T6 alloy. In addition, Rinnovatore showed that in the T6 temper, exfoliation corrosion resistance was found to be greater for forgings produced from rolled bar stock than forgings from extruded bar stock [11]. Moreover, it was shown that rapid quenching from the solution temperature in cold water increased exfoliation corrosion resistance of forgings tempered to T6. Fatigue and exfoliation interactions have been studied. Mills reports that most of the studies have been performed during the last five years on this issue although Shaffer in 1968 reported significant reduction in the fatigue life of exfoliated extruded 7075-T6 spar caps [12]. Moreover, multiple crack nucleation sites were observed in 7075-T651 [13], and 2024-T3 [14] aluminum alloy specimens when the specimens were subjected to exfoliation corrosion and then fatigue tested. Mills found an 88% decrease in the fatigue life of the specimens with prior exfoliation corrosion damage when compared to specimens tested without prior corrosion damage. Chubb et al. showed in their study using panels containing fastener holes that the end grains exposed in the rivet holes would be the potential corrosion sites that could eventually result in multiple site damage. In a recent study [12], experiments were performed to determine the effect of exfoliation on the fatigue crack growth behavior of 7075-T651 aluminum alloy. First the specimens were subjected to prior-corrosion damage using ASTM standard EXCO corrosive solution and then fatigue tested in corrosion fatigue environments of dry air, humid air, and artificial acid-rain. Test results indicated that prior-corrosion damage resulted in higher crack growth rates when tested in dry air as well as in acid-rain environments when compared to uncorroded specimens. Fractographic analysis showed quasi-cleavage fracture close to the exfoliated edge of the specimens tested in all the 7

three environments indicating embrittlement by prior-corrosion. Thus, embrittlement by prior-corrosion was stated to result in accelerated crack nucleation, faster short crack growth, and earlier onset of fatigue phenomena such as multiple-site damage. 1.3 Corrosion fatigue Corrosion fatigue is defined as "the process in which a metal fractures prematurely under conditions of simultaneous corrosion and repeated cyclic loading at lower stress levels or fewer cycles than would be required in the absence of the corrosive environment" [4]. Corrosion acting conjointly with fatigue can have major effects on materials in structures of aircraft. First, corrosion can create discontinuities (pits, cracks, etc) that act as origins of fatigue cracks with significant reductions in life at all stress levels. In crack propagation, corrosion effects are well known to produce accelerated fatigue crack propagation. The combination of aggressive environment and cyclic loading conditions have been observed to accelerate crack growth rates in aluminum alloys. Several mechanisms were proposed to explain the corrosion fatigue process [1]. They are: (i) dissolution of material at the crack tip in corrosive environment, (ii) hydrogen embrittlement in which diffusion of hydrogen (a by product of corrosion process) into the lattice space that could weaken the atomic bonds thereby reducing the fracture energy, (iii) theory of adsorbed ions in which the transport of critical species to the crack tip results in lowering of the energy required for fracture, and (iv) film-induced cleavage in which it is hypothesized that crack speed would increase at the film-substrate interface when the crack grows through the low toughness oxide layer leading to the rupture of the film. In general, corrosion fatigue effects on crack propagation are more pronounced at lower stress intensities whereas at higher stress intensities the crack propagates at such a high rate that the effects of chemical dissolution or localized embrittlement will be negligible. Several parameters affect corrosion fatigue crack propagation rates. For example, crack growth rates increase with increase in the stress intensity range. Also, at lower frequency corrosion fatigue effects will be more severe than at higher frequency because of the time dependent nature of the process. Increase in R value has been found to generally increase corrosion fatigue crack propagation rates. As well, increasing the concentration of corrosive species, lowering the pH, increasing the moisture content, and temperature usually result in more severe effects [4]. The most common corrosion fatigue environment that is simulated in laboratory testing is 3.5% NaCl as it is believed to result in severe general corrosion rates as well as 8

it represents roughly the salinity of sea water. In addition, other environments such as humid air, salt sprays, and artificial acid rain (to simulate industrial pollutants) also are used to characterize corrosion fatigue crack growth behavior of aluminum alloys. As aircraft are exposed to several complex chemical environments both inside and outside, no single environment could simulate the actual condition. Therefore, a few studies used sump tank water that was considered close to a "realistic chemical environment" [15]. The quest for realistic corrosion fatigue environment led Swartz et al. [16] to collect and analyze solution samples from bilge areas, external galley and lavatories of five different airplanes. As a result, a new chemical environment was developed to perform corrosion fatigue crack growth experiments on 2024-T351, 2324-T39, 7075-T651, and 7150-T651 aluminum alloys. For all the alloys studied the fatigue crack propagation rates in synthetic bilge solution were found to be between the dry air and the 3.5% NaCl data. In another study [17], cyclic wet and dry environment was simulated in characterizing the corrosion fatigue crack growth rates in 2024-T351 aluminum alloy. It was hypothesized that during the dry cycle the partial evaporation of the aqueous solution may allow some chemical species to get deposited at the crack tip and then in the wet cycle when the rehydration occurs, corrosion could occur at a greater rate than before. To simulate aircraft service corrosion, fatigue crack growth studies were conducted on service corroded 2024-T3 aluminum panels extracted from a C/KC-135 aircraft [18]. Test results showed that in some cases fatigue crack growth rates were two or three times greater in the corroded material, however, in other cases, there was little difference. It was observed that "the difference in the crack growth rates was due to high variability in the amount of corrosion damage between specimens". 1.4 Corrosion pillowing and its effect on structural integrity of aircraft lap joints Recently, some studies have shown that the increase in stress levels is not only because of the thickness loss due to corrosion but also due to the volume of the corrosion product build-up in a joint [19]. Also, evidences show that lap joints contain "faying cracks" under the rivet heads in the corroded areas. The complexity of this issue as explained by Komorowski et al. is that "the majority of the cracks had not penetrated the outer skin surface and appeared to grow more rapidly along the faying surface creating a high aspect ratio semi-elliptical crack and it is difficult to detect and affects the structural integrity of the joint" [19]. As reported by Krishnakumar et al.[20], the major corrosion product in the lap splices is found to be aluminum oxide trihydrate, an "oxide mix" which has a high molecular volume ratio to the alloy. As the oxide is insoluble, it is found to 9

remain within the joint and in turn is responsible to deform the skins in the joint which usually gives a bulging appearance, commonly termed as "pillowing". Moreover, finite element analysis revealed that for a two layer joint the stresses due to 6% thinning due to corrosion resulted in stress more than the yield strength of 2024-T3 aluminum alloy [21]. In addition, "pillowing induced deformation" was observed on the corroded joints after removal of the rivets and the separation of the skin. Moreover, multiple cracks were found to nucleate from rivet holes. Fracture mechanics analysis has shown that as the pillowing increases, the stress intensity factor for the crack edge along the faying surface increases [22]. On the other hand, the stress intensity factor decreases for the crack edge along the outer surface. Therefore, it was hypothesized that pillowing produces compressive stresses in the rivet area on the outer surface because of the resultant bending stresses. At the same time, high tensile stress is produced on the faying surface resulting in more rapid growth of faying surface cracks in the direction of the row of rivets than through the skin towards the outer surface [22]. 1.5 Pitting corrosion Pitting corrosion is defined as localized corrosion of a metal surface, confined to a point or small area, that takes the form of cavities [4]. Pitting is a deleterious form of localized corrosion and it occurs mainly on metal surfaces which owe their corrosion resistance to passivity. The major consequence of pitting is the breakdown of passivity, i.e. pitting, in general, occurs when there is breakdown of surface films when exposed to pitting environment. Pitting corrosion is so complicated in nature because oxide films formed on different metals vary one from another in electronic conduction, porosity, thickness, and state of hydration [23]. The empirical models that have been developed to understand the pitting process are closely related to the integrity of the metal oxide film. The salient features of the empirical theories related to pit nucleation mechanisms are mentioned in Table II.

10

Table II Pit Nucleation Theories Proposed by Evans et al. [24] (1929-30) Theory Proposed penetration theory. Ability of a chloride ion to penetrate the film was linked to the occurrence of pitting. Halide ions are assumed to be transported from the film-solution interface to the metal-oxide interface either by the application of electric field or exchange of anions. Hoar et al. Assumed the adsorption of anions on the oxide surface as the key [25,26] aspect in the pit nucleation process. (1960s) Proposed ion-migration model that involves activating anions that enter the oxide film lattice without exchange thereby increasing the ionic conductivity of the film resulting in local high anodic dissolution rates and pitting. Proposed mechanical model in which it was assumed that adsorption of anions at the oxide-solution interface lowers the interfacial energy resulting in the formation of cracks in the protective oxide film under the influence of the electrostatic repulsion of the adsorbed anions. Suggested a concept of local acidification of pit as a critical factor in pit growth. Uhlig [27]and Proposed adsorption theory in which at a certain value of the Kolotyrkin [28] potential (pitting potential) the adsorption of aggressive anions on the (1961-1967) metal surface displaces the passivating species such as oxygen. Kolotyrkin suggested that adsorption of anions at preferred sites forming soluble complexes with metal ions from the oxide. Once such species leave the oxide, thinning of the film starts locally increasing the electric field strength which accelerates the dissolution of the oxide. Sato [29,30] Proposed that at a critical potential an internal film pressure exceeds (1971, 1982) the critical compressive stress for film fracture. Considered thinning of film at local sites and suggested that pitting occurs only when a critical concentration of aggressive anions and a critical acidity is locally built up. Macdonald et al. Proposed that metal vacancies may accumulate as a result of the [31] diffusion of metal cations from the metal/film to the film/solution (1981) interface, forming voids at the metal/film interface. When the voids grow to a critical size the passive film will collapse leading to pit growth.

Therefore, nucleation of pits generally involves certain localized changes in the structure and properties of the oxide film. However, propagation of pits is related to the

11

dissolution of the underlying bulk metal. Further discussion on this subject is presented later in this report. 1.6 Pitting corrosion fatigue In corrosion fatigue conditions, as mentioned before, several studies showed greater increase in fatigue crack growth rates compared to baseline fatigue conditions. Although major efforts were expended to understand the crack propagation behavior of materials, a few studies have focused on the crack nucleation stage in the overall fatigue process [32,33,39]. McAdam first suggested that corrosion induced pits might act as stress concentrators from which cracks could form [34]. A large number of chemical or electrochemical factors such as potential, passive film, pH, and composition of environment are found to affect the pitting corrosion fatigue process. As well, mechanical factors such as stress range, frequency, stress ratio (R), and load waveform and metallurgical factors such as material composition, microstructure, heat treatment, and orientation can influence pitting corrosion fatigue process. Nucleation of cracks from corrosion pits were observed by many researchers including the works of Hoeppner [32,33,39], Goto [35] in heat-treated carbon steel, and Muller [36] in several steels. As well, in NaCl environment, lowering of the fatigue life due to the generation of pits in carbon steel [37] and 7075-T6 aluminum alloy [38] was observed under corrosion fatigue conditions. The Linear Elastic Fracture Mechanics (LEFM) concepts are widely used to characterize the crack growth behavior of materials under cyclic stresses in different environmental conditions. It is important to note that both pitting theory and crack growth theory have been used in model development as follows. Pit growth rate theory proposed by Godard is combined with the fatigue crack growth concepts. The time (or cycles or both) to nucleate a Mode I crack from the pit (under cyclic loading) could be modeled using LEFM concepts. Based on this idea, a few models [39-42] were proposed. All of the models assume hemispherical geometry for the pit shape and the corresponding stress intensity relation is used to determine the critical pit depth using the crack growth threshold (Kth) that is found empirically. For a hemispherical pit geometry, these models provide a reasonable estimate for the total corrosion fatigue life. Details of these models are presented later in this report.

12

1.7 Fretting corrosion and fatigue Virtually all mechanical joints of fixed wing aircraft, rotary wing aircraft, engine, gearbox and other electrical and mechanical components of aircraft are susceptible to the occurrence of fretting fatigue. The complexities of fatigue design, either safe-life or damage tolerant, are enough to challenge the most talented technical personnel. Add to that the occurrence of fretting and truly complex phenomena occur. Fretting fatigue was originally discovered in the early part of the 20th century. Since that time a great deal of work has been done to understand the mysteries of fretting fatigue. As well, engineers have devoted a significant effort to studying fretting fatigue to be able to cope with the potentially deleterious effects it may have on the life of engineering components. Fretting is a phenomenon that involves both environmental effects (corrosion in metals) and wear. Thus, it is a complex phenomenon. When it acts with cyclic loading on engineering components the result frequently is to shorten the fatigue life or to lower the fatigue design allowable Historically three approaches have been employed in fretting fatigue studies. One approach is to perform fretting fatigue testing by simulating the contact or fastening conditions. The fretting fatigue test results thus generated are used to determine the fretting fatigue life reduction factor Kff. Although this approach has been to introduce a fretting fatigue life reduction factor into the fatigue design process this is not always a satisfactory procedure. A second approach is to develop the fatigue structural allowables by testing the mechanical joints and including fretting considerations in these tests. Another approach is to use a fretting protection and control plan to focus on potentially blocking the occurrence of fretting or to prevent the propagation of cracks from fretting damage areas. This approach simulates fretting damage that will assist in developing greater understanding of the basic mechanisms of fretting which in turn will lead to the development of standardized fretting fatigue test methods. In addition, this approach will help to find some preventive systems to alleviate fretting fatigue problems on structural components. This mechanistic approach also will be useful to develop inspection system. Some investigators are coupling damage tolerance concepts to fretting studies to assist in more accurate tracking of damage as well as setting more realistic inspection intervals. Therefore, one of the needs for improved fretting fatigue resistance is the development of adequate experimental procedures to simulate fretting fatigue phenomena.

13

1.7.1 Mechanisms of fretting Fretting is believed to rupture the protective oxide film of the metal alloys that are in contact although the influence of oxide film is brief in the subsequent degradation mechanism of fretting. Therefore, it is commonly believed that the first stage of fretting is adhesive contact of the asperities on contact surfaces [43]. Microscopic plastic deformation of the contacting asperities in relative motion may result in the nucleation of cracks. In addition, metal transfer may occur from one surface to another depending on the hardness and the relative displacement of the contacting surfaces. The production of fretting debris and the oxidation of the fresh surface may result in the buildup of oxides on the contacting surfaces. These may result in the formation of pits on the contacting surfaces. These damages may very well accelerate the nucleation of cracks. Depending on the contact stress state the nucleated cracks may result in the early propagation in a rapid manner. Contact stresses are high enough to significantly influence crack propagation. The cracks may propagate at various rates and angles. Subsequent final propagation of cracks may depend on the bulk stress alone rather than on the contact stress. As the crack grows into the material, a depth is reached beyond which the contact stresses have little or no influence and the applied cyclic stresses play a dominant role. Therefore, the fretting process can be divided into four stages viz. nucleation of cracks, early propagation of cracks by contact stress state, final propagation of cracks by bulk stress and instability resulting in the decreased fatigue life. The aforementioned fretting process is dependent on many different variables such as material microstructure, stress state, environment, relative slip amplitude, and contact pressure. The theories related to nucleation of fretting fatigue cracks are usually based on the adhesive contact of the asperities as well as the cracks that form subsurface. Displacements that are "small" on a macroscopic scale are very large when acting on asperities. Relative movement of adhesively bonded asperities results in plastic or nearplastic strains around asperities. The importance of the formation of oxides may be based on their affect on adhesive contacts. Oxides prevent direct metal contact and reduce coefficient of friction by rolling. However, under certain conditions, such as the presence of discontinuities and/or sliding motion, cracks could nucleate subsurface. Dependent upon nucleation and contact stress state it has been observed that after a specific time, fretting contact no longer affects the fatigue life. This implies that nucleation occurs very early. In addition, it should be noted that fatigue life reduction occurs only after a specific amount of fretting damage and also the nucleated crack must be large enough to propagate under bulk stress alone [44]. 14

1.7.2 Fretting fatigue in aircraft joints Fretting fatigue has been observed as a potential degradation mechanism in aircraft structural components [45]. Many cracks have been found to originate at fasteners and on faying surfaces. Cracks may form due to fretting mechanisms from one or more rivet holes leading to the concept of multiple-site damage. Evidence supports that elimination of fretting with adhesives greatly increases fatigue life [46]. Also, it has been observed that cracks appear in lugs much sooner if fretting is present. In addition, failures observed originating at areas of fretting damage, often away from fastener holes. Fig. 3 shows a schematic of the design methods proposed in the past by Hoeppner to minimize the effect of fretting fatigue in aircraft joints. In addition, the methods to alleviate fretting fatigue challenge in aerospace components are provided in the next section. 1.7.3 Reduction or prevention methods The following are some methods that can be employed to reduce the problem of fretting fatigue in aircraft joints [47]. I Design a) prevent all relative motion of the surfaces b) prevent the surfaces from contacting c) surface roughness d) surface hardness e) etc. II Mechanical Methods a) shot peening b) vapour blasting c) bead blasting d) surface rolling e) dimpling These methods include processes that cold-work the surface and introduce residual surface compressive stresses.

15

III Coatings a) hard metal coatings b) soft metal coatings c) polymers IV Lubricants a) solid lubricants b) greases c) oils V Cathodic Protection VI Surface Treatments a) anodize b) ion implants c) sulphidize e) phosphatize f) etc. VII Influence substances used in aircraft joints a) Faying Surface Sealants -ADHESIVE types can significantly increase fatigue life. -FLEXIBLE sealants can decrease fatigue life. -Increasing sealant thickness decreases life. b) Adhesives -Can significantly improve fatigue life even if fretting does not occur. -Curing an adhesive, then assembling the joint can decrease fatigue life due to decreased friction. c) Penetrants -Usually reduce fatigue life but can have no affect or increase life. -Penetrants applied after joint assembly can easily enter joints depending upon the type of fay surface sealant. VIII Palliatives tested in aircraft joints a) Cold working can significantly improve joint fatigue life. 16

b) Solid lubricants escape from the joint or are wiped away. c) Teflon usually decreases fatigue life due to load transfer to fastener shanks. In summary, adhesion based mechanisms of fretting fatigue appear to be most common and fretting contact significantly increases stress at and near the surface. It should be noted that palliative effectiveness is extremely CASE SPECIFIC and only methods which induce surface residual compressive stresses consistently increase fretting fatigue life. Moreover, evidences show that fretting fatigue is a pervasive mode of failure with riveted aircraft joints. Furthermore, methods which reduce fretting in aircraft joints often do not increase the fatigue life as a reduction in coefficient of friction requires more load to be taken by the fastener shank.

17

Figure 3. Design method for fretting fatigue in aircraft joints (Hoeppner-1993)

18

1.8 References 1. Jones, D., Principles and prevention of corrosion, Macmillan Publishing Co., N.Y., 1992, pg. 4. 2. Wallace and Hoeppner, "Aircraft corrosion: causes and case histories", AGARD Corrosion Handbook, Vol. 1, AGARD-AG-278-Vol. 1, 1985. 3. Aging of U.S. Air Force Aircraft, National Research Council (NRC) National Materials Advisory Board (NMAB) report, National Academy Press, Washington,D.C., 1997. 4. Metals Handbook - Ninth Edition, Corrosion, Vol. 13, 1987, Publ: American Society for Metals (ASM), Metals Park, OH. 5. Alvarez, A., Corrosion on aircraft in marine-tropical environments: a technical analysis, Material Performance, May 1997, pp. 33-38. 6. Suyitno, B.M. and Sutarmadji., Corrosion control assessment for Indonesian aging aircraft, Anti-corrosion Methods and Materials, Vol. 44, No. 2, November 1997, pp. 115-122. 7. Groner, D.J., US Air Force aging aircraft corrosion, Current Awareness Bulletin, Structures Division, Wright Laboratory, Spring 1997. 8. Piascik, R.S., Kelly, R.G., Inman, M.E., and Willard, S.A., Fuselage lap splice corrosion, WL-TR-96-4094, Vol. II, ASIP, 1996. 9. McIntyre, J.F. and Dow, T.S., Intergranular corrosion behavior of aluminum alloys exposed to artificial seawater in the presence of nitrate anion, Corrosion, Vol. 48, No. 4, pp. 309-319, 1992. 10. Thompson, J.J., Tankins, E.S., and Agarwala, V.S., A heat treatment for reducing corrosion and stress corrosion cracking susceptibilities in 7xxx aluminum alloys, Materials Performance, June 1987, pp. 45-52. 11. Rinnovatore, J.V., Lukens, K.F., and Corrie, J.D., Exfoliation corrosion of 7075 aluminum die forgings, Corrosion, Vol. 29, No. 9, 1973, pp. 364-372. 12. Mills, T.B., The combined effects of prior-corrosion and aggressive chemical environments on fatigue crack growth behavior in aluminum alloy 7075-T651, Ph.D. Dissertation, University of Utah, 1997. 13. Mills, T.B., The effects of exfoliation corrosion on the fatigue response of 7075T651 aluminum plate, M.S. Thesis, University of Utah, 1994.

19

14. Chubb, J.P., Morad, T.A., Hockenhull, B.S., and Bristow, J.W., The effect of exfoliation corrosion on the fatigue behavior of structural aluminum alloys, in Structural Integrity of Aging Airplanes, 1991, pp. 87-97. 15. Pettit, D.E., Ryder, J.T., Krupp, W.E., and Hoeppner, D.W., Investigation of the effects of stress and chemical environments on the prediction of fracture in aircraft structural materials, AFML-TR-74-183, 1974. 16. Swartz, D.D., Miller, M. and Hoeppner, D.W., "Chemical Environments in Commercial Transport Aircraft and Their Effect on Corrosion Fatigue Crack Propagation", in Estimation, Enhancement and Control of Aircraft Fatigue Performance, Vol. I, J.M. Grandage and G.S. Jost, Eds., The proceedings of the 14th symposium of the International Committee on Aeronautical Fatigue, 3-5 May, 1995, Melbourne, Australia, pp. 353-364. 17. Kramer, J., and Hoeppner, D.W., Effects of Cyclic Immersion in 3.5% NaCl Solution on Fatigue Crack Propagation Rates in Aluminum 2024-T351, Proceedings of the 1995 USAF Structural Integrity Program Conference, WL-TR-96-4093, Vol. II, pp. 1089-1112. 18. Mills, T.B., Magda, D.J., Kinyon, S.E., and Hoeppner, D.W., Fatigue crack growth and residual strength analyses of service corroded 2024-T3 aluminum fuselage panels, Report to Oklahoma City Air Logistics Center and Boeing Defense and Space Group, University of Utah, 1995. 19. Komorowski, J.P., Bellinger, N.C., and Gould, R.W., The role of corrosion pillowing in NDI and in the structural integrity of fuselage joints, in Fatigue in New and Aging Aircraft, Proceedings of the 19th Symposium of ICAF 1997. 20. op-cit ref. 19. 21. Bellinger, N.C., Komorowski, J.P., and Gould, R.W., Damage Tolerance Implications of Corrosion Pillowing on Fuselage lap joints, AIAA journal. 22. Bellinger, N.C., and Komorowski, J.P., Corrosion pillowing stresses in fuselage lap joints, AIAA journal, Vol. 35, No. 3, 1997. 23. Jayalakshmi, M., and Muralidharan, Empirical and deterministic models of pitting corrosion - an overview, Corrosion Reviews, Vol. 14, Nos. 3-4, 1996, pp. 375-402. 24. Evans, V.R., J. Chem. Soc., 92, 1929. 25. Hoar, T.P., The Production and Breakdown of the Passivity of Metals, Corrosion Science, Vol. 7, 1967, pp. 341-355. 26. Hoar and Wood, The sealing of porous anodic oxide films on aluminum, Electrochemica Acta, 7, 1962, pg. 333 20

27. Bohni and Uhlig, H.H., Environmental Factors Affecting the Critical Pitting Potential of Aluminum, Journal of the Electrochemical Society, Vol. 116, 1969, pp. 906-910. 28. Kolotyrkin, Ya, M., Effects of Anions on the Dissolution Kinetics of Metals, Journal of the Electrochemical Society, Vol. 108, No. 3, 1961, pp. 209-216. 29. Sato, N., Electrochemica Acta, 16, 1971, pg. 1683. 30. Sato, N., Anodic Breakdown of Passive Films on Metals, and The Stability of Pitting Dissolution of Metals in Aqueous Solution, Journal of the Electrochemical Society, Vol. 129, No. 2, 1982, pp. 255-260 and 260-264. 31. Lin, Chao, and Macdonald, A point defect model for anodic passive films II. Chemical breakdown and pit initiation, Journal of the Electrochemical Society, Vol. 128, No. 6, 1981, pp. 1194-1198. 32. Hoeppner, D.W., Mann, and Weekes, Fracture mechanics based modelling of corrosion fatigue process, in Corrosion Fatigue: Proceedings of the 52nd meeting of the AGARD Structural and Materials Panel held in Turkey, 5-10 April, 1981. 33. Hoeppner, D.W., Corrosion fatigue considerations in materials selections and engineering design, Corrosion Fatigue: Chemistry, Mechanics, and Microstructure, NACE, 1972, pp. 3-11. 34. McAdam, D.J., and Gell, G.W., Pitting and its effect on the fatigue limit of steels corroded under various conditions, Journal of the Proceedings of the American Society for Testing Materials, Vol. 41, 1928, pp. 696-732. 35. Goto, M., and Nisitani, H., Crack initiation and propagation behavior of a heat treated carbon steel in corrosion fatigue, Fatigue Fracture Engineering Material Structure, Vol. 15, No. 4, 1992, pp. 353-363. 36. Muller, M., Theoretical considerations on corrosion fatigue crack initiation, Metallurgical Transactions, Vol. 13A, 1982, pp. 649-655. 37. Mehdizadeh, P., et al., Corrosion fatigue performance of a carbon steel in brine containing air, H2S, CO2, Corrosion, Vol. 22, 1966, pp. 325-335. 38. Corsetti, L.V., and Duquette, D.J., The effect of mean stress and environment on corrosion fatigue behavior of 7075-T6 aluminum, Metallurgical Transactions, Vol. 5, 1974, pp. 1087-1093. 39. Hoeppner, D.W., Model for Prediction of Fatigue Lives Based Upon a Pitting Corrosion Fatigue Process, Fatigue Mechanisms, Proceedings of an ASTM-NBS-

21

NSF Symposium, J.T. Fong, Ed., ASTM STP 675, American Society for Testing and Materials, 1979 pp. 841-870. 40. Lindley, T. C., McIntyre, P., and Trant, P. J., Fatigue Crack Initiation at Corrosion Pits, Metals Technology, Vol. 9, 1982, pp. 135-142. 41. Kawai, S. and Kasai, K., Considerations of Allowable Stress of Corrosion Fatigue (Focused on the Influence of Pitting), Fatigue Fracture of Engineering Materials Structure, Vol. 8, No. 2, 1985, pp. 115-127. 42. Kondo, Y., Prediction of Fatigue Crack Initiation Life Based on Pit Growth, Corrosion Science, Vol. 45, No. 1, 1989, pp. 7-11. 43. Hoeppner, Mechanisms of fretting fatigue, Keynote Paper, in Fretting Fatigue, ESIS 18, Mechanical Engineering Publications, London, (1994) 3-19. 44. Hoeppner and G.L. Goss, A fretting fatigue damage threshold concept, Wear, 27, (1974) 61-70. 45. Hoeppner and S. Adibnazari, Fretting fatigue in aircraft joints, in Durability and Structural Integrity of Airframes, International Committee of Aeronautical Fatigue (ICAF), (1993) 191-207. 46. Hoeppner, "Fretting of aircraft control surfaces," in AGARD Conference Proceedings No. 161, Specialists Meeting on Fretting in Aircraft Systems, (1974), 1.1-1.7. 47. Hoeppner, S. Adibnazari and M. W. Moesser, Literature review and preliminary studies of fretting and fretting fatigue including special applications to aircraft joints, DOT/FAA/CT-93/2, (1994).

22

2 Pitting Corrosion 2.1 Introduction Pitting is classified as a localized attack that results in rapid penetration and removal of metal at small discrete areas [1]*. An electrolyte should be present for pitting to occur. The electrolyte could be a film of condensed moisture, or a bulk liquid. How and when pitting occurs on a metal depends on numerous factors, such as, type of alloy, its composition, integrity of its oxide film, presence of any material or manufacturing induced discontinuities as well as chemical and loading environment, to name a few. Many metals and their alloys are subject to pitting in different environments. These include alloys of carbon steels, stainless steels, titanium, nickel, copper, and aluminum [2]. In passivated metals or alloys that are exposed to solutions containing aggressive anions, primarily chloride, pitting corrosion results in local dissolution leading to the formation of cavities or holes. The shape of the pits or cavities can vary from shallow to cylindrical holes and the cavity is approximately hemispherical [3]. The pit morphology depends on the metallurgy of the alloy and chemistry of the environment as well as the loading conditions. As observed first by McAdam in 1928, these pits may cause local increase in stress concentration and cracks may nucleate from them [4]. According to Foley [5], pitting corrosion of aluminum occurs in four steps: (1) adsorption of anions on the aluminum oxide film, (2) chemical reaction of the adsorbed anion with the aluminum ion in the aluminum oxide lattice, (3) penetration of the oxide film by the aggressive anion resulting in the thinning of the oxide film by dissolution, and (4) direct attack of the exposed metal by the anion. The susceptibility of a metal to pitting corrosion as well as the rate at which pitting occurs on its surface depends on the integrity of its oxide film. Therefore, a brief overview of the mechanisms of the formation of passive film is discussed below. 2.2 Formation of passive films and their growth The following discussion on the oxide film formation and its growth is extracted from ref. (6).

Numbers in parentheses [ ] refer to references at the end of each section.

23

Early investigators examined the effects of natural waters on metals by placing them outside. One investigator, Liversidge, in 1895, observed that an aluminum specimen, .lost its brilliancy, and became somewhat rough and speckled with grey spots mixed with larger light grey patches; it also became rough to the feel, the grey parts could be seen to distinctly project above the surface, and under the microscope they presented a blistered appearance. This encrustation is held tenaciously, and does not wash off, neither is it removed on rubbing with a cloth [7]. Liversidge proposed that a hydrated aluminum oxide had formed, but did not confirm this with further testing of the layer. He did, however, note that when weighed, the aluminum specimens gained weight with exposure, rather than losing weight [8]. It was later confirmed that the weight gain was due to formation of an oxide film [9]. Although Liversidge suggested the formation of an aluminum oxide film, subsequent investigators proposed other theories to explain the passive behavior of aluminum. Some of these were changes in the state of electric charge on the surface, changes in valence at the surface, and a condensed oxygen layer [10]. The presence of the oxide film on the surface of the metals was proven by Dunstan and Hill in 1911. Through experiments with iron, they determined that the passive film was reduced at 250 F, the temperature at which magnetic iron oxide is reduced. Similar films were found on other metals [10]. Barnes and Shearer attempted to determine the constitution of passive films on aluminum and magnesium in 1908. They determined that aluminum formed hydrogen peroxide when reacting with water and that the passive film consisted of Al2(OH)6 [9]. This was later determined to be incorrect [11]. 2.3 Structure of the passive film in aluminum It later was determined that this film on aluminum consists of an aluminum oxide created when the aluminum comes in contact with an environment. Generally, this film is amorphous; however, under certain circumstances it will develop one of seven crystalline structures: 1. Gibbsite (also called hydrargillite): (-Al2O33H2O) 2. Bayerite: (-Al2O33H2O) 3. Boehmite: (-Al2O3H2O or AlOOH) 4. Diaspore: (-Al2O3H2O) 24

5. Gamma alumina: (-Al2O3) 6. Corundum: (-Al2O3) 7. Combinations of aluminum oxides with inhibitors, for example (2Al2O3P2O53H2O) Gibbsite and diaspore structures are not found during corrosion of aluminum, but are frequently found in bauxite ores. Boehmite, bayerite, gamma alumina, and corundum are sometimes found in the passive layers of aluminum under certain conditions. Additionally, bayerite is frequently found as a corrosion product during pitting of aluminum. Combinations of aluminum oxides with inhibitors are not understood very well in the literature, but it is known that they will combine with oxide layer to form improved corrosion resistance through changing the passive film structure. Changes in the amorphous structure of the oxide film have been studied by several researchers. In one investigation, the passive film formed on the pure aluminum sheet revealed changes in structure with an increase in temperature and oxygen content. Prior to heating, the structure was reported to be amorphous oxide. As the temperature was increased, the amorphous film thickened, formed boehmite, and bayerite. The rate of film formation increased with temperature, and with an increase in oxygen content, intergranular attack began. The researcher suggested the following sequence of events in the formation: boehmite is nucleated at dislocation centers that are at the surface of the amorphous film, it then grows by a diffusion mechanism. During thickening of the boehmite, a process occurs that allows aluminum ions to escape into the solution which results in bayerite growth [12]. Other investigations revealed that aluminum in the molten state will develop an oxide film of gamma alumina which will convert to corundum when exposed to dry air. Aluminum sheet in water at temperatures below 70 to 85 C after long aging will develop a pssive film consisting of bayerite. Boehmite is found on aluminum exposed to water at high temperatures (above 70 to 85 C) [11]. More recently, researchers have found small regions of crystallized -alumina within the amorphous layers created during anodizing [13]. During exposure to air and water, alumina will form a passive film with a duplex structure. The film will consist of two layers, a permeable outer layer, and a protective, nonporous layer next to the metals surface. In the case of an air environment, the protective layer is thicker and the permeable layer is comparatively thin. In the case of an immersion in water, the permeable layer is thicker and the protective layer is thinner. In both cases, the total thickness of the duplex film is the same [11]. The protective layer will quickly reach maximum thickness, with the permeable layer growing slower. The growth rate of each layer depends on a few parameters. In air, 25

it is dependent on temperature; in water, it is dependent on temperature, oxygen content, pH, and the type of ions present in the electrolyte; and in anodization procedures, it depends on electrolyte and applied potential. The film is typically formed on pure aluminum when the pH of the solution is between 4.5 to 8.5 [11]. Other researchers have suggested that the permeable outer layer consists of hexagonal close-packed pores in pure aluminum. The size of these pores will depend on conditions of formation. These conditions of formation are sometimes controlled by sealing processes in an attempt to improve the characteristics of the passive film. In sealing processes, the pores are blocked or made smaller by boehmite or gamma alumina formation, nickel acetate is added to obstruct the pores, or dichromates or chromates can be added to create pores of a different structure [14]. The passive film formed on metals will differ according to the environment in which it forms. Studies done by Seligman and Williams in the 1920s illustrate this difference. In experiments with tap water, the presence or absence of certain impurities caused either the passive film to breakdown and the metal to corrode or, the film will become thick and less susceptible to corrosion. They determined that nitrates and chromates would combine with the passive film and serve to increase resistance of the passive film to localized corrosion [15]. Later studies emphasized this conclusion. One researcher found a film of 55,000 angstroms in distilled water and another found a film of only 4,800 angstroms for the same alloy (AA-1099) immersed in tap water [11]. Additionally, experiments performed by Bengough and Hudson on aluminum in sea water showed that the passive film varied with corroding liquid and with different alloying elements [8]. In a more recent paper, researchers determined that the reaction between aluminum and water takes place in three steps: formation of the amorphous oxide, dissolution of the oxide, and deposition of the dissolved products as hydrous oxide. In the first step, the amorphous oxide layer is formed and grows by the anodic and cathodic reactions present at the water/metal interface. The second step involves a hydrolysis reaction with the surface which depends on temperature, pH and aluminum concentration and the last step is accomplished when the resulting hydroxide is deposited on the surface. The rate at which the film will grow is controlled by the diffusion of water molecules through the existing layers. At temperatures between 50 and 100 C, pseudoboehmite grows on the amorphous oxide. At 40 C, however, bayerite crystallization occurs and with time will overcome the pseudoboehmite [16]. Upon exposure of an air-formed-film to water, the air-formed-film will break down and another film will form that is thicker and contains more water. The rate at 26

which the film is reformed depends on the anions present and the temperature [11]. In more recent work, the water in the aluminum passive film has been stated to be a medium for the mobilization for aluminum cations and deposited anions [17]. In air, the thickness of the passive film is dependent on humidity. In higher humidity, the oxide layer is thicker. The growth rate of the film, however, does not depend on humidity. Rosenfeld et al. found that in high purity air, the growth rate was not changed. However, when small amounts of impurities were added, growth was accelerated in humid air [18]. In addition to impurities, the growth of the film is highly dependent on temperature. Below 200 C, the film will grow only to a few hundred angstroms, above 300 to 400 C, the rate gradually increases, between 400 and 600 C, the film will grow to a thickness of 400 angstroms, at 450 C, the film will crystallize to gamma alumina [11]. 2.4 Pitting potential and induction time According to Smialowska [2], the susceptibility of a metal or alloy to pitting can be estimated by determination of one of the following criteria: characteristic pitting potential, critical temperature of pitting, number of pits per unit area, or weight loss and the lowest concentration of chloride ions that may cause pitting. One of the most important criteria to determine an alloys susceptibility to pitting corrosion is to find the pitting potential, i.e., the potential at which the passive film starts to break down locally. The potential above which pits nucleate is denoted by Ep and the potential below which pitting does not occur and above which the nucleated pits can grow is often indicated by Epp. Once the passive film begins to breakdown, the time it takes to form pits on a passive metal exposed to a solution containing aggressive anions, for example, Cl-, is called the induction time or incubation time [2]. The induction time is meaningful in a statistical sense as it represents the average rate of reaction over the whole surface to produce a measurable increase in current. It should not be considered as the time to form the first pit. This is because micro pits have been observed to form during the induction time [5]. The induction time is usually denoted by . It is measured as the time required to produce an appreciable anodic current at a given anodic potential. It is expressed as 1/ = k (E-Ep), where E is the applied potential and K is a function of

27

Cl- ion concentration [19]. In general, pitting potential decreases with increasing Cl- ion concentration. The most commonly used relation for estimating t is based on an exponential relationship between time and activation energy i.e. 1/ = Ae-Ea/RT; the activation energy needed for pit nucleation can be obtained from an arrhenius plot of log (1/) vs. 1/temperature [5]. As well, Hoar and Jacob [20] have proposed a relationship 1/ = K(Me)m (X-)n to estimate the induction time. Where Me is the metal ion concentration, Xis the halide ion concentration, and m and n are orders of reaction which are determined experimentally. Subsequent to the nucleation of pits it has been observed they grow. The following subsection presents a discussion of pit growth. 2.5 Pit growth rate and pit morphology Godard [21] developed a simple but effective relation based on the experimental data to estimate the rate at which pits grow. The empirical relation he developed was d = K(t1/3). Even though he found this relation when tested using aluminum, it was observed to be true for other materials in different types of water environments. In general, the rate of pit growth depends on several factors such as temperature, pH, properties of passive films, chloride ion concentration, presence of anions and cations in solution, and the orientation of the material [5]. The pit growth can be viewed as a direct interaction of the exposed metal with the environment. Upon observing the geometry of the pits formed on 7075 aluminum alloy in halide solutions, Dallek [22]proposed a pit growth rate expression i - ip = a(t - ti)b in which current was expressed as a function of time. In this expression, i is the dissolution current, ip is the passive current, t is the time, ti is the induction time, a is the constant depending on the halide, and b is the constant depending on the geometry of the pit. From this expression, a plot of log (i - ip) vs. log (t - ti) will give the slope b. Dallek predominantly observed pits of hemispherical shape. However, Nguyen et al [23] have observed hemispherical pits at low potential on 1199 aluminum alloy in chloride solutions and at high potential they observed a porous layer film covered on the pit mouth with orifice at the center. This study indicated the effect of potential on the morphology of pits. Chloride ion concentration also was found to affect the pit morphology. Baumgartner and Kaesche [24] observed that in dilute to medium concentrated solutions, pit morphology was rough whereas at high concentration, pits were found to be smooth and rounded. In addition, a recent study by Grimes [6] showed clearly the 28

effect of loading conditions on the morphology of pits. This study was conducted on 7075 - T6 aluminum alloy in 3.5% salt water under three different loading conditions, viz. zero, sustained and cyclic. It was found that the pits propagated under cyclic loads were three times larger in cross sectional area when compared to those grown under sustained or zero load conditions. Also, it was found that most of the pits originated from the grain boundaries. This study concludes that the effect of both mechanical and chemical environment must be considered in pitting corrosion studies. However, when studying the effect of pitting on the fatigue life of aluminum alloy 7075-T6 in 3.5% NaCl solution, Li [25] found that although the test frequency (5 and 20 Hz) had a pronounced effect on the total corrosion fatigue life, the fatigue test frequency did not have any effect on the pit morphology. On the other hand, Chen et al. [26] have found that the size of the pit from which a crack nucleated was comparatively larger at the lower frequencies and stresses than at higher frequencies and stresses when fatigue tested using 2024-T3 aluminum alloy. 2.6 Mechanisms of pit nucleation In general, pit nucleation mechanisms are classified into three categories. (i) Adsorption-induced mechanisms, (ii) Ion migration and penetration models, and (iii) Mechanical film breakdown theories. 2.6.1 Adsorption-Induced Mechanisms In this section mechanisms of pit nucleation based on the adsorption of aggressive anions at energetically favored sites are discussed. Many researchers including Uhlig et al. [27-29] Hoar et al. [20, 30] and Kolotyrkin [31] have suggested mechanisms related to the ion-adsorption concepts (see Table III). Many of the mechanisms proposed in the literature consider this as a necessary step in the pit nucleation process. Uhlig [27-29] and Kolotyrkin [31] independently proposed that both oxygen and chlorine anions can be adsorbed onto the metal surfaces. When the metal is exposed in air, oxygen is adsorbed by the metal resulting in the formation of passive oxide film. Consequently, a chemical bond is established between the oxygen anion and the metal cation. This process is known in corrosion terminology as chemisorption. Chemisorption results in the formation of a metal-compound that covers the surface of a metal. If aluminum is exposed in oxygen, the resulting compound is aluminum oxide i.e. Al2O3. However, the type of compound that is formed on the metal surface depends on the environment in 29

which the metal is exposed. For example, in the case of salt water, Cl- ions in addition to oxygen are present. When oxygen is adsorbed, passivation of metal occurs whereas if chlorine anion is adsorbed, it dose not result in passivation but breakdown of passivity occurs. As proposed by Kolotyrkin [31], below the pitting potential, metals may prefer to adsorb oxygen and above this critical potential metals may adsorb halides, such as ClThis mechanism is termed competitive adsorption as the presence of different anions will compete with the oxygen to be chemisorbed by the metal. Therefore, at or above the pitting potential, chlorides and other aggressive anions if present, combine with the metal and then diffuse from the metals surface into the solution. Subsequently, it combines with water in solution to form metallic oxides, hydrogen and chloride ions. These chloride ions are attracted to the surface of the metal and the process begins again. It was hypothesized by many researchers that the chloride ions might diffuse to regions of high energy such as inclusion, dislocations and other form of discontinuities. Hoar et al. [20, 30] originally proposed a complex ion formation theory which stated that the formation of Cl- containing complexes on the film-solution interface might lead to a locally thinned passive layer. This was proposed because Cl- -containing complexes are more soluble when compared to complexes formed in the absence of halides. They assumed that a high energy complex is formed when a small number of Clions jointly adsorb around a cation in the film surface which can readily dissolve into solution. This creates a stronger anodic field at this site that will result in the rapid transfer of another cation to the surface where it will meet more Cl- and enter into solution. Experimental support was provided for this concept by Strehblow et al.[32] by conducting investigation on the attack of passive iron by hydrogen fluoride. They found that the breakdown process occurred with complete removal of the passivated oxide layer. It was observed that hydrogen fluoride catalyzed the transfer of Fe3+ and Ni2+ ions from oxide into the electrolyte. As mentioned in a paper by Bohni [33], similar observation was made in another study by Heusler et al. regarding the influence of chloride containing borate and phthalate solutions on the passive film breakdown of iron. Different behavior of Cl- and F- ions in the pit nucleation process was proposed in a model by Heusler et al. Cl- ions were suggested to form only two dimensional "clusters" leading to the localized thinning of the passive layer. However, it was proposed that Fions adsorb homogeneously on the oxide surface thereby promoting a general attack. It should be noted that the proposed models did not take into account material discontinuities such as point "defects", dislocations, inclusions, voids and others. Also, another model based on the concept of an increased probability of "electrocapillary film 30

breakdown" was proposed by Sato [34] (see Table III). Although Sato includes the effect of dislocations in this purely theoretical approach, no experimental evidence was found in the literature to support his model. However, Sato's theoretical model proposed that ntype passive oxide films are more stable than p-type films because of the difference in the band structure of electron levels. From these studies it can be concluded that in addition to chloride anions, other anions such as chromate and sulphate also get adsorbed changing the nature of the compound. In addition, as observed by Richardson and Wood [35], enhanced adsorption takes place at the imperfections or flaws in the oxide film. These discontinuities in the film usually become the sites of anion adsorption. Nilsen and Bardal [36] have observed by measuring the pitting potential of four aluminum alloys (99% pure Al, Al-2.7Mg, Al4.5Mg-Mn, and Al-1Si-Mg) and found that the pitting potential values for the four alloys were within only 25 mv. From this study, they concluded that alloy composition does not directly depend on the adsorption step of the process. Table III Adsorption-Induced Mechanisms Proposed by Uhlig et al. 1950-69 Kolotyrkin 1961 Hoar 1967 Summary Proposed concepts based on either competitive adsorption or surface complex ion formation. Description In competitive adsorption mechanism Cl- anions and passivating agents are simultaneously adsorbed. Above a critical potential Cl- adsorption is favored resulting in the breakdown of passivity. Kolotyrkin suggested that there were critical Cl/inhibitor concentration ratios, depending on the potential above which pitting would occur. Limitations Occurrence of induction times varying with passive film thickness cannot be explained.

31

Proposed by Sato 1982

Summary Proposed a theoretical concept based on the potential dependent transpassive dissolution which depends on the electronic properties of the passive film. The electrochemical stability of a passive film depends strongly on the "electron energy band structure" in the film.

Description Stated that the critical potential above which potential-dependent dissolution of the film occurs will be less noble at the sites of chloride ion adsorption. As a result of the increased dissolution rate above the critical potential, local thinning of the passive films occurs until a steady state is reached. Proposed that the local thinning of the oxide film as a mechanism of pit "initiation" Included the effect of dislocations similar to the influences of Cl- ions.

Limitations Knowledge of the electronic properties of passive films has not been fully understood. Experimental evidence for this mechanism is lacking.

2.6.2 Ion migration and penetration models A few models (see Table IV) were proposed based on either penetration of anions from the oxide/electrolyte interface to the metal/oxide interface or migration of cations or their respective vacancies. This theory is based on the concept that Cl- ions migrate through the passive film and results in breakdown of the film once they reach the metal/film interface. Hoar and Jacob [20] explain that when a critical potential is reached, smaller ions, like Cl-, may penetrate the film under the influence of an electrostatic field which exists across the film. These aggressive anions prefer the high energy regions like grain boundaries and impurities as sites for migration because these regions produce thinner passive film locally. During the migration, the ions either pass through the film completely or they may combine with the metal cation in the midst of the film resulting in the formation of what is called a contaminated film which is a better conductor than the uncontaminated film. This process results in an autocatalytic reaction which encourages more ions to penetrate the film. This hypothesis is supported by some researchers as they have observed a higher concentration of Cl- ions over thin films on the surface of iron as well as that the time to breakdown the film increases with the

32

thickness of the film [37]. It was further hypothesized that Cl- ions first fills anion vacancies on the surface of the passive film and then migrates to the metal/oxide interface. However, other works revealed that the time required for Cl- to penetrate through the film is much longer than the induction time measured experimentally [33]. Later, Macdonald et al. [38] proposed a model in which the growth of the passive film was explained by the transport of both anions (e.g. oxygen ion) and cations (e.g. metal ion). Diffusion of anion from film-solution interface to metal-film interface results in thickening of the film. Cation diffusion from the metal-film interface to the filmsolution interface results in the creation of metal vacancies at the metal/film interface. These metal vacancies usually submerge into the metal itself. However, if the cation diffusion rate is higher than the rate of vacancy submergence into the bulk metal, the metal vacancies will increase leading to the formation of voids at the metal/film interface. This process is known as pit incubation. Subsequently, when the void reaches a critical size, the pit incubation period ends leading to the local rupture of the passive film. This eventually results in pit growth at that local site. Based on this theory, Macdonald et al expressed a criterion for pit initiation as stated below. (Jca - Jm) * (t - ) where, Jca is the cation diffusion rate in the film, Jm is the rate of submergence of the metal vacancies into the bulk metal, t is the time required for metal vacancies to accumulate to a critical amount x is a constant, Also, in this model, the role of halide ion in accelerating the film breakdown by increasing Jca was suggested. The ion penetration and migration theories do not include the effect of mechanical breakdown of the oxide film that may result because of the scratches from which pits can nucleate. Nor is the mechanical breakdown of the oxide film included that results from strain and local cracking of the oxide film.

Table IV: Ion migration and penetration models 33

Proposed by Hoar et al 1965

Macdonald et al 1981

Summary Presented that when the electrostatic field across the film/solution interface reaches a critical value corresponding to the critical breakdown potential , the anions adsorbed on the oxide film enter and penetrate the film. Presented a theoretical model to explain the chemical breakdown of passive film.

Description Favored sites for ion migration are suggested to be high energy regions like grain boundaries and impurities where thinner passive films are produced. If the aggressive ions meet a metal cation, contaminated film is produced that encourages further ions to penetrate the film. Then, this process continues as an autocatalytic reaction. Proposed that metal vacancies may accumulate as a result of the diffusion of metal cations from the metal/film to the film/solution interface, forming voids at the metal/film interface. When the voids grow to a critical size the passive film will collapse leading to pit growth.

Limitations Did not explain the observation that pits often form from mechanical breaks in the oxide film or from scratches.

Surface discontinuities such as grain boundaries etc. were not considered in developing the model. No direct observation of void formation was made. As the measured induction times usually show a large scatter, definite quantitative agreement is difficult to obtain.

In addition Macdonald et al [39] have proposed a "point defect" model for anodic films to calculate Jca for "thin" films on the order of 10-40 A. Also, the "point defect"

34

model could be used to calculate incubation times Although, the "point defect" model was one of the most detailed models proposed, this model has some limitations as mentioned in Table IV. 2.6.3 Mechanical film breakdown theories-Chemico-mechanical breakdown theories Pit nucleation models proposed so far based on the concepts of the chemicomechanical breakdown of films have not included the effect of externally applied stresses (see Table V). Sato [40] showed that a significant film pressure always acted on "thin" films that he attributed to "electrostriction". Sato expressed a relation between the film pressure, thickness and surface tension of the film as follows. p = po + [((-1)2)/8] -/L where p is the film pressure, po is the atmospheric pressure, is the film dielectric constant, is the electric field, is the surface tension, L is the film thickness. According to his hypothesis, both and L have significant influence on film pressure p. Based on this relation, Sato suggested that the adsorbtion of chloride ion significantly reduces the surface tension thereby increasing p. Also, he proposed that when p is above the critical value, the film might break down. In addition, Sato proposed that breakdown of the film occurs when it attains a thickness at which mechanical stresses caused by electrostriction become critical. Therefore, building up of critical stresses in the film could cause pitting.

Table V Chemico-mechanical breakdown theories

35

Proposed by Sato 1971

Summary Proposed a breakdown mechanism for anodic films from thermodynamic considerations.

Sato 1982

Derived an equation for the work required to form a cylindrical breakthrough pore in the passive film.

Description Showed that thin films always contain film pressure due to "electrostriction". Hypothesized that both the surface tension of the film and the film thickness have a significant effect on film pressure. Proposed that adsorption of chloride ions, depending on their concentration, greatly reduces surface tension. Proposed that for a pit nuclei to grow to macroscopic size a critical radius corresponding to a critical pore formation energy must be exceeded.

Limitations Experimental proof is not found.

Experimental proof is not found. Microstructural parameters such as grain boundaries, inclusions that may influence pitting "initiation" were not considered.

In addition to the aforementioned theory, some researchers have observed the influence of mechanically produced discontinuities (such as scratches in the passive film) on the formation of pits along those scratches [41]. If there is a scratch in the passive film that sets up a local anodic site which will eventually be the preferred site for pit to form. This smaller anode/cathode ratio results in higher local potential leading to the nucleation of pits. Other researchers proposed a similar theory that is related to the value of product of the length of the discontinuity and the current density. Assuming a unidirectionally growing pit, if this value exceeds a critical value, the discontinuity such as fissures in the oxide film may form a local area of low pH leading to the formation of pits from them. This happens due to the difference in the pH at the local site (fissure) when compared to the bulk solution. It was proposed that a fissure of size in the order of 10-6 cm could be a limiting condition for this to happen [42].

36

Hoar [30] also assumed that the presence of pores or flaws could mechanically stress and damage the passive films in contact with an aggressive solution. Moreover, Hoar assumed that aggressive anions would replace water and reduce surface tension at the solution-film interface by repulsive forces between particles, producing cracks. In conclusion, there is no full agreement among the researchers regarding the mechanisms of pit nucleation. However, as the pitting process itself is a complex one, the commonly accepted view is that the first step in the pit nucleation process is the localized adsorption of aggressive anions on the surface of the passivated metal. Several experimental studies also have indicated that the preferred sites for the passage of anions through the oxide film are the discontinuities present in an alloy. Such discontinuities are non-metallic inclusions, second phase precipitates, pores or voids, grain or phase boundaries as well as other mechanical damages [2]. These discontinuities eventually may become pit nucleation sites. The aforementioned theories on pit nucleation are based purely on electrochemical concepts. However, the breakdown of surface film is dependent not only on the solution conditions (e.g. pH), and the electrochemical state at the metal/solution interface, but also on the nature of the material as well as the stress state. In addition, the aforementioned pit nucleation mechanisms did not take into account the material parameters such as the microstructural effects, inherent discontinuities such as voids, inclusions, second phase particles as well as the externally applied stress. Moreover, localized corrosion also may take place at slip bands during fatigue loading [43]. Once the pit is formed, the rate of pit growth is dependent mainly on the material, local solution conditions and the state of stress. Cracks have been observed to form from pits under cyclic loading conditions. Therefore, to estimate the total corrosion fatigue life of an alloy, it is of great importance to develop some realistic models to establish the relationship between pit propagation rate and the stress state. Furthermore, pitting corrosion in conjunction with externally applied mechanical stresses, for example, cyclic stresses has been shown to severely affect the integrity of the oxide film as well as the fatigue life of a metal or an alloy. Therefore, to understand this phenomena, some models based on pitting corrosion fatigue mechanisms have been proposed as discussed below.

2.7 Pitting corrosion fatigue

37

The Linear Elastic Fracture Mechanics (LEFM) concepts are widely used to characterize the crack growth behavior of materials under cyclic stresses in different environmental conditions. It is important to note that both pitting theory and crack growth theory have been used in model development as follows. Pit growth rate theory proposed by Godard is combined with the fatigue crack growth concepts. The time to nucleate a Mode I crack from the pit (under cyclic loading) could be modeled using LEFM concepts. Based on this idea, a few models [44-48] were proposed since 1971 (see Table VI). All of the models assume hemispherical geometry for the pit shape and the corresponding stress intensity relation is used to determine the critical pit depth using the crack growth threshold (Kth) that is found empirically. For a hemispherical pit geometry, these models provide a reasonable estimate for the total corrosion fatigue life. However, it is well known that corrosion pit morphology varies widely. Thus, this aspect must eventually be dealt with in LEFM models that attempt to deal with pit growth and the ultimate nucleation of crack(s) from pit(s). As mentioned before, the combined effect of corrosion and the applied cyclic loading have been shown to produce cracks from corrosion pits. In addition, pits have frequently been the source of cracks on aircraft operating in fleets. Depending upon the fatigue loading and corrosion conditions, some studies have shown that the crack nucleation site may change from slip bands to corrosion pits [49]. This observation was made when fatigue tested at reduced strain rates in Al-Li-Cu alloy. Another study also showed an anodic dissolution in slip bands in Al-Li-Zr alloy at high stress levels whereas at low stress levels fatigue cracks nucleated from corrosion pits [50]. Therefore, it was hypothesized that at higher stress levels, conditions are favorable to form cracks from slip bands before the corrosion pit reaches the critical condition to favor the nucleation of crack from it. In addition, a recent study also showed that larger pit was formed at lower stress and frequency. It also was observed in 2024-T3 (bare) aluminum alloy in NaCl solution that once pits formed from the constituent particles, because of the applied cyclic stresses, the pits coalesced, laterally and in depth to form larger pits from which crack was observed to nucleate [51]. Therefore, modeling the transition of a pit first to a short crack and then to a long crack is considered to be important in characterizing the total corrosion fatigue life of a material as discussed in the next section [44, 52, 53].

Table VI Pitting Corrosion Fatigue Models

38

Proposed by 1 Hoeppner (1971 current)

Advantages/Limit ations This model Proposed a model Using a four parameter Weibull to determine provides a fit, fatigue crack growth critical pit depth to reasonable threshold (Kth) was found from estimate for nucleate a Mode I corrosion fatigue experiments for crack under pitting hemispherical the particular environment, corrosion fatigue geometry of the material, frequency, and load conditions. pits. spectrum. combined with the The stress intensity relation for This model is pit growth rate useful to surface discontinuity (half penny estimate the theory as well as shaped crack) was used to total corrosion the fatigue crack simulate hemispherical pit. growth curve fit in fatigue life with a corrosive knowledge of a i.e.) K = 1.1 the kinetics of environment, the Q pitting cycles needed to develop a critical corrosion and where, is the applied stress, a is the fatigue crack pit size that will pit length, and Q is the function of form a Mode I growth. a/2c, Sty. fatigue crack This model did can be estimated. not attempt to Using the threshold determined propose empirically, critical pit depth was mechanisms of found from the stress intensity crack relation mentioned above. nucleation from Then, the time to attain the pit corrosion pits. depth for the corresponding Quantitative threshold value was found using studies of 3 pitting d t = corrosion c fatigue behavior of materials can where, t is the time, d is the pit depth, be made using and c is a material/environment this model. parameter. This model is valid only for the conditions in which LEFM concepts are applicable. Material dependent.

Summary

Description

39

Proposed by 2 Lindley et al. (1982)

Summary Similar to Hoeppners model, a method for determining the threshold at which fatigue cracks would grow from the pits was proposed.

Description Pits were considered as semielliptical shaped sharp cracks Used Irwins stress intensity solution for an elliptical crack in an infinite plate and came up with the relationship to estimate threshold stress intensity values related to fatigue crack nucleation at corrosion pits. i.e.) Kth = ( a)1.13 0.07 a c 1 + 1.47 a c

Advantages/Limit ations The proposed stress intensity relation can be used in tension - tension loading situations where stress intensity for pits and cracks are similar. Critical pit depths for cracked specimens can be estimated using the existing threshold stress intensity values. This model is valid only for the conditions in which LEFM concepts are applicable. Material dependent.

( )
1 2

( )

1.64

where, is the stress range, a is the minor axis, and c is the major axis of a semi-elliptical crack. From the observed pit geometry i.e. for a/c ratio, threshold stress intensity can be calculated. For the corresponding a/c ratio, critical pit depth can be estimated.

40

Proposed by 3 Kawai and Kasai (1985)

Summary

Description

Advantages/Limit ations Using this model, allowable stress in relation to corrosion fatigue threshold as a function of time can be estimated. Material dependent. This model is valid only for the conditions in which LEFM concepts are applicable.

Proposed a model based on estimation of allowable stresses under corrosion fatigue conditions with emphasis on pitting. As corrosion is not usually considered in developing S-N fatigue curves, a model for allowable stress intensity threshold involving corrosion fatigue conditions was proposed.

Considered corrosion pit as an elliptical crack. Based on experimental data generated on stainless steel, new allowable stresses based on allowable stress intensity threshold was proposed. i.e.) all = k all F h max

where, Kall can be determined from a da/dN vs. K plot for a material, hmax is the maximum pit depth, and F is a geometric factor.

41

Proposed by 4 Kondo (1989)

Summary Corrosion fatigue life of a material could be determined by estimating the critical pit condition using stress intensity factor relation as well as the pit growth rate relation.

Description Pit diameter was measured intermittently during corrosion fatigue tests. From test results, corrosion pit growth law was expressed as 2c Cp t1/3 where, 2c is the pit diameter, t is the time, and Cp is an environment/material parameter. Then, critical pit condition (Kp) in terms of stress intensity factor was proposed by assuming pit as a crack. Kp = 2.24 a c Q where, a is the stress amplitude, a is the aspect ratio, and Q is the shape factor. Critical pit condition was determined by the relationship between the pit growth rate theory and fatigue crack growth rates. c = cp (N/f)1/3 where, N is the number of stress cycles, f is the frequency, and 2c is the pit diameter. The pit growth rate dc/dN was developed using K relation as given below.
dc

Advantages/Limit ations The aspect ratio was assumed as constant. Material and environment dependent.

4 1 3 1 2 2 2 4 dN = 3 Cp f Q (2.24 a ) K

dc/dN was determined using experimental parameter Cp. Finally, the critical pit size 2Ccr was calculated from the stress intensity factor relation. i.e.) 2Ccr = (2Q/)( Kp/2.24a)2 2.8 References

42

1. Jones, D., Principles and Prevention of Corrosion, Macmillan Publishing Co., N.Y., 1992, pg. 198. 2. Szklarska, S., Pitting Corrosion of Metals, published by National Association of Corrosion Engineers (NACE), Houston, Texas, 1986, pg. 1. 3. Aluminum Properties and Physical Metallurgy, Ed., J.E. Hatch, American Society for Metals (ASM), Metals Park, Ohio, 1984, pg. 253. 4. McAdam, D.J. and Gell, G.W., Pitting and its Effect on the Fatigue Limit of Steels Corroded Under Various Conditions, Journal of the Proceedings of the American Society for Testing Materials, Vol. 41, 1928, pp. 696-732. 5. Foley, R.T., Localized Corrosion of Aluminum Alloys - A Review, Corrosion, Vol. 42, No. 5, 1986, pp. 277-288. 6. Grimes, L., A Comparative Study of Corrosion Pit Morphology in 7075-T6 Aluminum Alloy, M.S. Thesis, University of Utah, 1996. 7. Liversidge, On the Corrosion of Aluminum, Chemical News, LXXI, 1895. 8. Bengough and Hudson, Aluminum, Fourth Report to the Corrosion Committee, Journal of the Institute of Metals, Vol. 21, No. 1, 1919, pg. 105. 9. Barnes and Shearer, A Hydrogen Peroxide Cell, Journal of physical chemistry, Vol. 12, 1908, pg. 155. 10. Dunstan and Hill, The Passivity of Iron and Certain Other Metals, Journal of the Chemical Society, Vol. 99, 1911, pg. 1853. 11. Godard, The Corrosion of Light Metals, John Wiley and Sons, N.Y., 1967. 12. Hart, The Oxidation of Aluminum in Dry and Humid Oxygen Atmospheres, The Proceedings of the Royal Society, A236, 1956, pg. 68. 13. Thompson, Shimizu, and Wood, Observation of Flaws in Anodic Films on Aluminum, Nature, Vol. 286, 1980, pg. 471. 14. Hoar and Wood, The Sealing of Porous Anodic Oxide Films on Aluminum, Electrochemica Acta, Vol. 7, 1962, pg. 333. 15. Seligman and Williams, The Action on Aluminum of Hard Industrial Waters, Journal of the Institute of Metals, 23, 1920. 16. Alwitt, The Growth of Hydrous Oxide Films on Aluminum, Journal of the Electrochemical Society, Vol. 121, No. 10, 1974, pp. 1322-1328. 43

17. Graedel, Corrosion Mechanisms for Aluminum Exposed to the Atmosphere, Journal of the Electrochemical Society, Vol. 136, No. 4, 1989, pg. 204C. 18. Rozenfield and Danilov, Electrochemical Aspects of Pitting Corrosion, Corrosion Science, Vol. 7, 1967, pg. 129. 19. Broli, A. and Holtan, H., Use of Potentiokinetic Methods for the Determination of Characteristic Potentials for Pitting Corrosion of Aluminum in a Deaerated Solution of 3% NaCl, Corrosion Science, Vol. 13, 1973, pp. 237-246. 20. Hoar, T.P., The Production and Breakdown of the Passivity of Metals, Corrosion Science, Vol. 7, 1967, pp. 341-355. 21. Godard, The Corrosion of Light Metals, John Wiley and Sons, N.Y., 1967. 22. Dallek, S. and Foley, R.T., Propagation of Pitting on Aluminum Alloys, Journal of the Electrochemical Society, Vol. 125, No. 5, 1978, pp. 731-733. 23. Nguyen, T.H. and Foley, R.T., On the Mechanism of Pitting of Aluminum, Journal of the Electrochemical Society, Vol. 126, No. 11, 1979, pp. 1855-1860. 24. Baumgartner and Kaesche, Aluminum Pitting in Chloride Solutions: Morphology and Pit Growth Kinetics, Corrosion Science, Vol. 31, 1990, pp. 231-236. 25. Ma, L., Pitting Effects on the Corrosion Fatigue Life of 7075-T6 Aluminum Alloy, Dissertation, University of Utah, 1994. 26. Chen, G. S., Liao, C., Wan, K., Gao, M., and Wei, R. P., Pitting Corrosion and Fatigue Crack Nucleation, Effects of the Environment on the Initiation of Crack Growth, ASTM STP 1298, W. A. Van Der Sluys, R. S. Piascik, and R. Zawierucha, Eds., American Society for Testing and Materials, 1997, pp. 18-33. 27. Uhlig, H.H., Adsorbed and Reaction-Product Films on Metals, Journal of the Electrochemical Society, Vol. 97, No. 11, 1950, pp. 215c-220c. 28. Bohni and Uhlig, H.H., Environmental Factors Affecting the Critical Pitting Potential of Aluminum, Journal of the Electrochemical Society, Vol. 116, 1969, pp. 906-910. 29. Leckie, H.P. and Uhlig, H.H., Environmental Factors Affecting the Critical Potential for Pitting in 18-8 Stainless Steel, Journal of the Electrochemical Society, Vol. 113, No. 2, 1966, pp. 1262-1267. 30. Hoar, T.P. and Jacob, W.R., Nature, 216, 1967, pg. 1299.

44

31. Kolotyrkin, Ya, M., Effects of Anions on the Dissolution Kinetics of Metals, Journal of the Electrochemical Society, Vol. 108, No. 3, 1961, pp. 209-216. 32. Strehblow, H.H., Titze, B., and Loechel, B.P., The Breakdown of Passivity of Iron and Nickel by Flouride, Corrosion Science, Vol. 19, 1979, pp. 1047-1057. 33. Bohni, H., Localized corrosion, in Corrosion Mechanisms, Ed., F. Mansfeld, Marcel Dekker, Inc., N.Y., 1987, pp. 285-328. 34. Sato, N., Anodic Breakdown of Passive Films on Metals, and The Stability of Pitting Dissolution of Metals in Aqueous Solution, Journal of the Electrochemical Society, Vol. 129, No. 2, 1982, pp. 255-260 and 260-264. 35. Richardson and Wood, A Study of the Pitting Corrosion of Al by Scanning Electron Microscopy, Corrosion Science, Vol. 10, 1970, pp. 313-323. 36. Nilsen, N. and Bardal, E., Short Duration Tests and a New Criterion for Characterization of Pitting Resistance of Al Alloys, Corrosion Science, Vol. 17, 1977, pp. 635-646. 37. Hoar, T.P. and Jacob, W.R., Nature, 216, 1967, pg. 1299. 38. Chao, Lin, and Macdonald, A Point Defect Model for Anodic Passive Films I. Film Growth Kinetics, Journal of the Electrochemical Society, Vol. 128, No. 6, 1981, pp. 1187-1194. 39. Lin, Chao, and Macdonald, A Point Defect Model for Anodic Passive Films II. Chemical Breakdown and Pit Initiation, Journal of the Electrochemical Society, Vol. 128, No. 6, 1981, pp. 1194-1198. 40. Sato, N., Anodic Breakdown of Passive Films on Metals, and The Stability of Pitting Dissolution of Metals in Aqueous Solution, Journal of the Electrochemical Society, Vol. 129, No. 2, 1982, pp. 255-260 and 260-264. 41. Richardson and Wood, A Study of the Pitting Corrosion of Al by Scanning Electron Microscopy, Corrosion Science, Vol. 10, 1970, pp. 313-323. 42. Galvele, Transport Processes and the Mechanism of Pitting of Metals, Journal of the Electrochemical Society, Vol. 123, No. 4, 1976, pp. 464-474. 43. Akid, R., The role of stress-assisted localized corrosion in the development of short fatigue cracks, Effects of the Environment on the Initiation of Crack Growth, ASTM STP 1298, W. A. Van Der Sluys, R. S. Piascik, and R. Zawierucha, Eds., American Society for Testing and Materials, 1997, pp. 3-17. 44. Hoeppner, D.W., Corrosion Fatigue Considerations in Materials Selections and Engineering Design, in Corrosion Fatigue: Chemistry, Mechanics, and 45

Microstructure, O. Devereux, A.J. McEvily, R.W. Staehle, Eds., NACE-2, National Association of Corrosion Engineers,1972, pp. 3-11. 45. Hoeppner, D.W., Model for Prediction of Fatigue Lives Based Upon a Pitting Corrosion Fatigue Process, Fatigue Mechanisms, Proceedings of an ASTM-NBSNSF Symposium, J.T. Fong, Ed., ASTM STP 675, American Society for Testing and Materials, 1979 pp. 841-870. 46. Lindley, T. C., McIntyre, P., and Trant, P. J., Fatigue Crack Initiation at Corrosion Pits, Metals Technology, Vol. 9, 1982, pp. 135-142. 47. Kawai, S. and Kasai, K., Considerations of Allowable Stress of Corrosion Fatigue (Focused on the Influence of Pitting), Fatigue Fracture of Engineering Materials Structure, Vol. 8, No. 2, 1985, pp. 115-127. 48. Kondo, Y., Prediction of Fatigue Crack Initiation Life Based on Pit Growth, Corrosion Science, Vol. 45, No. 1, 1989, pp. 7-11. 49. Rebiere, M. and Magnin, T., Materials Science and Engineering, Vol. A128, 1990, pg. 99. 50. Chen, G.S. and Duquette, D.J., Corrosion Fatigue of a Precipitation-Hardened Al-LiZr Alloy in a 0.5 M Sodium Chloride Solution, Metallurgical Transactions, Vol. 23, 1992, pp. 1563-1572. 51. Chen, G.S. et al., Corrosion, Vol. 52, 1996, pg. 8. 52. Hoeppner, D.W., Mann, and Weekes, Fracture mechanics based modelling of corrosion fatigue process, in Corrosion Fatigue: Proceedings of the 52nd meeting of the AGARD Structural and Materials Panel held in Turkey, 5-10 April, 1981. 53. Ma, L., and Hoeppner, D. W., The Effects of Pitting on Fatigue Crack Nucleation in 7075-T6 Aluminum Alloy, Proceedings of the FAA/NASA International Symposium on Advanced Structural Integrity Methods for Airframe Durability and Damage Tolerance, NASA Conference Publication 3274, Part 1, 1994, pp. 425-440.

46

3 Microstructure and environment effects on short crack behavior of materials 3.1 Introduction As the meaning of the word "short" as related to the formation and early propagation of cracks differs significantly with respect to the conditions of interest (coi), it has been realized by the technical community that there are three basic types of "short"/"small" cracks viz. microstructurally-short-cracks, chemically-short-cracks, and mechanically-short-cracks. This section discusses the mechanisms of short crack formation as influenced by microstructure and environments as well as some studies on crack formation from corrosion pits in the short crack regime. Design of structural components using fracture mechanics concepts requires three basic parameters viz. load or applied stress, stress intensity factor, and discontinuity or crack size. Conventionally damage tolerant design methods consider an initial discontinuity size typically in the order of 1 mm (0.04") while applying fracture mechanics concepts to design damage tolerant components (Potter and Yee, 1983* ). Fatigue crack growth tests in the "long" crack regime as per ASTM E647 are conducted on materials that are to be used to make the damage tolerant parts and together with the assumed initial discontinuity size as well as the appropriate loading and material parameters safe inspection intervals are computed. However, the technical community has realized the significance of the formation and existence of the so called "short" cracks (examples: manufacturing discontinuities like crack(s) from a hole that may result from improper riveting operation, and material microstructural discontinuities in the "short" crack range, typically in the order of grain size, or less) and their growth is "faster" when compared to "long" cracks in the equivalent or even at lower stress intensity range. With the support of experimental studies conducted by several researchers to determine the "short" crack behavior of materials, it can be argued that the current practice of computing inspection intervals of aircraft critical structural parts using the "long" crack
*

References are listed in the alphabetical order of the principal authors at the end of the section.

47

growth data and the initial "flaw" size of 1 mm might lead to unexpected fatigue crack growth behavior resulting in a catastrophic failure. This implies that designing components using damage tolerant concepts may still not be safe when "short" crack behavior is not integrated in fatigue life prediction methods. Moreover, the applicability and the accuracy of the Linear Elastic Fracture Mechanics (LEFM) concepts in predicting short crack behavior was addressed by several researchers in the Specialists Meeting organized by AGARD (AGARD, 1983). "Small" cracks were found in aircraft fuselage riveted lap joints (Schijve, 1992), fastener holes (Potter and Yee, 1983) and in tear-down aircraft wing critical structures (Wood and Rudd, 1983). Although many research studies were conducted, there is still not a clear understanding among the researchers with regard to the usage of terminology such as "small" and "short" as related to the size of the crack as these terms are often mixed up in the literature and this alone exemplifies the complexity of the challenge we have at present. However, many "short" crack researchers agree on the following aspects: There are three basic types of "short" cracks viz. mechanically or physically "short", microstructurally "short" and chemically "short" (McClung, Chan, Hudak and Davidson, 1994). "Short" cracks cannot be modeled using the Linear Elastic Fracture Mechanics (LEFM) concepts although some workers have attempted to convert the "short" crack data to "long" crack, to compare the crack growth behavior in both the regimes and also to evaluate the "effective crack driving force" for "small" and "large" fatigue cracks (Tokaji, Ogawa, Harada and Ando, 1986; Tokaji, Ogawa and Harada, 1987; Davidson, 1988; Hyspecky and Stanadel, 1992; Nicholls and Martin, 1991; Sheldon, Cook, Jones and Lankford, 1981). "Short" cracks often grow "rapidly" when compared to "long" cracks at a lower stress intensity range that is below the "long" crack threshold and also at an equivalent stress intensity range. In some materials, the "short" growth rate is observed to be much

48

"faster" than would be predicted by extrapolating "large" crack data (Lankford, 1982; Lankford, 1985). The material parameters which influence plastic deformation; viz. grain size, grain orientation, texture, work hardening rate, slip band character, local microscopic fracture toughness, inclusion size and content as well as second phase particles have an important role in "short" crack growth (Hoeppner, 1981 and Miller, 1982). Current NDI technologies are not capable of detecting the discontinuities that are in the "short" crack range (Wood and Rudd, 1983). This directly affects the inspection related issue as it is inextricably linked with the damage tolerant design concepts. There is no single parameter that can define the "short crack driving force". Although a few studies have attempted deterministic "short" crack growth prediction methods for physically or mechanically "short" cracks, as the influence of the microstructural variations on the "short" crack behavior of materials is extremely complex, the challenge of incorporating "short" crack methodology into present practice of fatigue life prediction analyses must be treated with probabilistic approaches. This necessitates the study of scatter in the behavior of "small" cracks to understand the physical basis of scatter in the fatigue lives of components or specimens (Goto, 1992 and Goto, 1993). Several researchers have postulated different mechanisms for the behavior of "short" cracks. They are indeed related to the type of "short" cracks. For microstructurally "short" cracks, "crack tip shielding" and "enhanced crack tip plastic strains" are stated to be responsible phenomena (Ritchie and Lankford, 1986). Also, evidences suggest that microstructurally "short" cracks are sometimes obstructed locally by grain boundaries (crack arrest), influenced by non-uniform growth and sometimes may experience higher cyclic plastic strains at the crack tips resulting in "faster" growth (Lankford, 1981, Lankford, 1982, Ritchie and Suresh, 1995). Moreover, "short" cracks may be subjected to "crack deflection" (Suresh, 1983) that may be related to the orientation of each grain.

49

Thus, texture has an important role in determining the behavior of microstructurally "short" cracks. Excessive plasticity is stated to be the mechanism for mechanically "short" cracks. This implies that the assumption of "small-scale yield" that forms the basis of linear elastic fracture mechanics is not applicable. Local crack tip environment (Gangloff, 1985) has been hypothesized as a predominant mechanism for the "faster" propagation of chemically "short" cracks as they are more vulnerable to chemical attack than "long" cracks because of the relative ease of access to the crack tip. For physically "short" cracks, it is hypothesized that the "crack closure effect" that decreases the "crack driving force" for "long" crack propagation may be absent (Schijve, 1986). Furthermore, it is believed that "short" cracks that are usually associated with a limited "wake" are less able to develop the same magnitude of shielding as equivalent "long" cracks at the same nominal stress intensity range (Lankford 1986). Some experimental studies have showed that although the growth rate of microstructurally "short" cracks is "faster" initially, once the crack tip reaches the grain boundary the growth rate either gets reduced or some times completely arrested at the grain boundary (Lankford, 1982, Lankford, 1983, Lankford and Davidson, 1986, Tokaji and Ogawa, 1992, Ritchie and Suresh, 1995). Moreover, the dimension of the crack also is related to the types of crack. Mechanically "short" crack size 'a' is considered to be less than the plastic zone size and microstructurally "short" crack size is related to the grain size (Hoeppner, 1981; Ritchie and Suresh, 1995). As the crack length is "short" compared to the microstructural dimensions, such as the grain size, it has been realized that the assumption that the crack grows in a homogeneous, isotropic continuum is no longer valid. Therefore, the primary focus of this section is to review the present understanding of the influence of microstructure and environment on the behavior of "short"/"small" cracks in structural materials.

50

3.2 Effect of microstructure on the short crack behavior of materials

This section reviews microstructurally "short"/"small" crack studies conducted on materials. The mechanisms of the "short" crack growth behavior as related to the local microstructural variations and crack closure phenomena are discussed in detail. When applying damage tolerance principles to the design of structural components subjected to cyclic loading, a geometry independent parameter K (stress intensity range), is used to characterize the crack growth. This is the basis of Linear Elastic Fracture Mechanics (LEFM). Moreover, it is presumed that cracks would not propagate at a stress intensity below a threshold value and it is usually denoted as Kth. Although it has been realized that there is no geometry independent crack driving force in the "short" crack regime, several attempts have been made to predict "short" crack growth rates including some approaches that are based on crack deflection, crack closure, 'J' integral and some semi-empirical methods (Hoeppner and Cameron; Morris and Buck, 1977, McEvily and Minakawa, 1984, Suresh, 1985, and El Haddad, Dowling, Topper, and Smith, 1980). As microstructure has varying effects in materials, it has been realized that the deterministic way of fatigue life prediction in the "short" crack regime is far from accurate. A need for the development of probabilistic concepts and models to solve the short crack challenges was emphasized by Hoeppner (1983). However, some researchers have proposed models which incorporate microstructural parameters into the "short" crack growth rate equations (Petit and Zeghloul, 1986, Petit and Zeghloul, 1990) A distinction between "small" and "short" crack growth behavior was made by some researchers (Lankford and Davidson, 1986, Breat, Mudry, and Pineau, 1983) by experimental studies. Their studies have shown that the experimentally measured difference in crack closure response between "short" and "large" cracks in A508 steel may be sufficient to explain differences in their crack propagation behavior. From experimental studies conducted in A508 steel, it was observed that da/dN versus K for

51

"short" cracks was only an extension of the "long" crack behavior into the subthreshold regime (Breat, Mudry, and Pineau, 1983). However, "small" cracks are related to a controlling microstructural element, usually the grain size, and these cracks are observed to be distinct from "short" cracks, i.e., through thickness cracks (0.5-2.0 mm) in length. (Taylor and Knott, 1981) have suggested that the "small" crack regime corresponds roughly to microcracks such that 2a = 10 GS where GS is grain size. Also, (Breat, Mudry, and Pineau, 1983) have showed that "short" cracks do not grow much faster at a given K than do "long" cracks. But "short" cracks have a lower threshold, above which they follow roughly the same da/dN versus K curve that a "long" crack would if it were simply to continue growing below its threshold. However, "small" cracks, as hypothesized by these researchers, not only grow below Kth, they also grow much faster than would be predicted by extrapolating long crack results below their threshold. Some studies (Taylor and Knott, 1981, Hicks and Brown, 1984, Lankford and Davidson, 1986) have showed that "small" crack growth converges with the "large" crack curve when plastic zone size is approximately equal to the grain size. This phenomenon has been observed in aluminum alloy, fine grained, coarse grained and single crystal astroloy, and p/m aluminum alloy. In many cases it has been shown that the arrest or retardation of small cracks correlates with the crossing of grain boundaries (Lankford, 1982, Lankford 1985). It also has been postulated that if the plastic zone size is less than the grain size such crossings are infrequent. In comparison with the "large" cracks, cracks are observed to grow in many unfavorable grains simultaneously, hence, the average rate of growth is much lower than that for "small" cracks at equivalent K levels in favorably oriented grains. Although these studies attempted to distinguish "small" and "short" cracks, many of the researchers used the terminology in a mixed manner. Therefore, as the purpose of this report is only to review the present understanding with regard to the fundamental issues pertaining to

52

microstructural and environmental aspects of the fatigue crack formation and growth, this report includes the terms "short" and "small" as they are extracted from the literature. In general, the "faster" growth of microstructurally "small" fatigue cracks has been observed to be associated with second phase particles (Pearson, 1975), inclusion particle clusters or voids (Newman and Edwards, 1988), eutectic colony boundaries (Taylor and Knott, 1981), and grain boundaries (Lankford, 1985). Early in 1975, bending fatigue studies were performed on commercial aluminum alloys viz. aluminum-copper-magnesium (BS L65) and DTD 5050 (aluminumzinc-magnesium) (Pearson, 1975). It was observed that cracks in the order of, or even less than, the grain size grew "faster" than "long" cracks. i.e. the mean crack growth rate in the early stage was observed to be 1.27 X 10-6 mm (5 X 10-8 in.). In this study crack formed at surface inclusions and it was related to previous cold working that was performed on the material. Fracture mechanics approach was used to calculate K values for "short" and "long" cracks because the plastic zone size was observed to be 1/20th of the crack length. Also, it was concluded that the growth rates in the early stage was much "faster" than would be predicted from the "long" crack data. Following this study, in 1976, uniaxial fatigue loads were applied on Aluminum alloy 2219-T851 parallel to the rolling direction in room temperature and it was observed that crack nucleated at the surface at intermetallic inclusions (Morris, Buck and Marcus, 1976). Moreover, it was showed that at the fatigue loads that are less than 0.6 times the yield stress several cracks coalesced to form a macrocrack that lead to the ultimate failure. The most important finding of this study showed that there was significant retardation of microcracks with grain boundaries. Similar observation was made in another study (Lankford, Cook and Sheldon, 1981), in that it was hypothesized that "small" cracks grew "rapidly" during the initial stage and when the crack tip interacted with grain boundaries, the growth rate slowed down. However, it also was formulated that as the crack moved away from the boundary, the crack growth became "faster" again.

53

This behavior was related to local microplasticity in certain preferentially oriented grains (Lankford, 1982). This study was conducted in laboratory air in 7075-T6 (precipitation hardened aluminum) with 60% humidity. It was observed that for the "smallest" crack (2a<40 m), da/dN did not increase monotonically with K. However, there was a decrease in the crack growth rate and reached a minimum in the range 30 m < 2a < 40 m. Some of the cracks were found to be "nonpropagating". It was concluded that "small" cracks could become "large" cracks "when their LEFM plastic zones begin to exceed in size the maximum grain dimension". Similar observation was made when "small" crack behavior was studied in A286 steel in which the "short" crack effect disappeared when the crack-tip plastic zone size became greater than the grain size (Mei and Morris, 1993). This study also supported the hypothesis that peak stress and microstructural effects in addition to the absence of crack closure are some of the factors that influence the "short" crack growth in this material. In another study, it was observed that when the applied stress was sufficiently high the "short" crack growth rate could be sustained and could overcome the microstructural barrier when tested in plain specimens (Pan, De Los Rios and Miller, 1993). Also, in this study, tests on notched specimens (8090 Al-Li alloy) revealed that the extent of notch tip and crack tip plastic zones control "short" crack propagation. It was hypothesized that "a short crack will continue to propagate only if its own plastic zone can sustain growth as the crack tip extends beyond the notch zone". Also, the effects of grain boundary, viz., changing the direction of the crack, temporary stopping of a crack, and forcing the crack to adopt a zig-zag path were observed. In SiC reinforced aluminum alloy (6061), "short" cracks were found to propagate through the SiC particles as the crack front fractured the particles (Kumai, King and Knott, 1990). In another study, the presence of voids in an overaged 2024 aluminum alloy was observed to be responsible for the formation of fatigue "micro cracks" (Sigler, Montpetit and Haworth, 1983). In this study, the density of microcracks (cracks were

54

counted when the size was more than 5 m) was observed at different stress amplitudes during the fatigue life. It was found to be 300/mm2 at 320 MPa, in 771 cycles at failure, but less than 0.5/mm2 at 200 MPa, in 1.5X105 cycles (at failure). This result has very high significance as related to the possibility of microcracks coalescence into macrocracks that may affect the ultimate fatigue life. In nodular cast iron, "short" cracks formed from either the graphite nodules or microshrinkage pores (Clement, Angeli and Pineau, 1984). Crack closure effect was suggested as the mechanism for the "faster" growth of the "short" cracks for the given stress intensity factor. On the other hand, in a medium carbon steel, the "short" crack growth rate was related to the intensity and the extent of plasticity of the crack tip (De Los Rios, Tang and Miller, 1984). Also, it was observed that "short" crack growth decreased or even arrested at ferrite-pearlite boundaries. However, as the stress level was increased to certain value, the two of the arrested cracks at the ends of the ferrite joined up and the resultant crack branched off along the prior austenite grain boundaries. Therefore, it was suggested that the critical fracture occurred when two of the "short" cracks joined and branched that made it possible to propagate into the pearlite. Similar hypothesis can be postulated for a "short" crack nucleating from corrosion pit in aluminum alloys and the stress concentration of the pit may be sufficient for "short" crack to overcome the grain boundary barrier for subsequent propagation. It can be further hypothesized that if cracks form from adjacent pits these "short" cracks can join and may grow "faster" when compared to "microcracks" coalescence resulting from other microstructural heterogeneities. However, there are no experimental data to support this theory and the very possibility of this occurring in a material should be investigated. "Small" surface crack (2 to 1000 m) studies on other materials like aluminumlithium alloy 2090-T8E41 also have shown an accelerated growth at K levels as low as 0.7 MPa m (Venkateswara Rao, Yu and Ritchie, 1988) and this behavior was related to

55

restrictions in the development of "crack tip shielding" resulting from "roughness-induced crack closure". In low carbon steel, when fatigue tested with specimens produced with two ferrite grain sizes of 24 and 84 m, it was found that in fine grained material most of the cracks formed within ferrite grains and in coarse grained material the cracks formed at grain boundaries (Tokaji, Ogawa and Harada, 1986). The most important finding of this study was that the effect of grain boundary strongly depended on "short" crack length. For fine and coarse grained material, when 2c>200 m and 2c>250 m respectively, the grain boundary was not a barrier for the "short" crack propagation and dc/dN increased with increasing crack length. Also, other studies showed that the effect of a grain boundary on the "short" crack growth appeared to be dependent on the orientation of ferrite grains (Lankford, 1985 and Suresh and Ritchie, 1984). In general, it has been recognized that when cracks are of a length comparable to the scale of the microstructure, the growth is greatly affected by the microstructure and the relevance of continuum mechanics is limited. The concept of "microstructural dissimilitude" was proposed to explain this kind of behavior (Chan and Lankford, 1988). During the growth of "short" cracks, if the crack front behaves in a similar fashion when it intersects many grains irrespective of the crack length, these cracks are believed to possess similitude and the stress intensity range (corrected for yielding) can be used to correlate the crack growth rates. However, for "short" cracks, it has been observed that there is no linear relation between da/dN and K. Thus, the challenge becomes so complex as it is highly impossible to predict the potential site of crack nucleation that may be related to the preferred orientation of the grain for the cracks to form. More importantly, because of the greater dependence of the "short" crack challenge on the microstructural variations in a material that is frequently in the order of a grain size, the "short" crack problem is intrinsically statistical in nature (Hoeppner, 1981). Therefore, to improve the resistance of the material microstructure to the nucleation of "short" fatigue

56

cracks, alloy design incorporating "small" randomly oriented grains, and texture to have only a few grains as possible for an easy crack growth under a known loading condition was suggested to be a useful way to deal with this issue (Ritchie and Lankford 1984). Also, the challenge becomes still more complex as the local fracture toughness of each grain is of practical significance as related to the formation and the propagation of cracks in the "short" regime. The effect of processing techniques on the "short" crack behavior was studied in an aluminum-magnesium-silicon alloy (Plumtree and O'Connor, 1991) in strain control. "Short" cracks were observed to form from second phase particles and the growth was impeded at grain boundaries. It was concluded that the extruded alloy with a finer microstructure and smaller second phase particles demonstrated a superior resistance to formation of "short" cracks when compared to squeeze-cast material. "Short" cracks in the order of 3 to 147 m were found to form from Ti3Al hcp alpha phase in a titanium aluminide alloy (Davidson, Cambell and Page, 1991). The results showed that the growth of "small" cracks in titanium aluminide alloy was slower by a factor of 10 to 100 when compared to aluminum alloys. As the grain size has an important role in the behavior of microstructurally "short" cracks, a study was conducted to demonstrate the effect of grain size on the "short" fatigue cracks. Specimens from 7075-T6 were prepared to produce grain sizes of 12 and 130 m. Growth rates of surface cracks in the order of 20 to 500 m were studied during the axial fatigue test in laboratory air of 45 to 60% RH (Zurek, James and Morris, 1982). It was found that the mechanism for the "short" crack growth behavior was dependent on alloy grain size. As discussed so far, many of the published works on the behavior of "short" cracks indicate that grain boundaries impede propagation, resulting in decrease in the growth rate or complete arrest in some cases. This kind of behavior has been modeled by some workers (Eastabrook, 1984, Hobson, 1982, Lankford, 1982). These models predict

57

that increasing grain size will lead to faster crack propagation rates in the "short" crack regime. (Brown, King, and Hicks, 1984) supported this model when they conducted "short" crack studies on a Ni-base superalloy, Astroloy, with grain sizes of 12 and 50 micrometer. They observed slower crack propagation rates in the fine grained material than those in the coarse grain size. Also, (Wagner, Gregory, Gysler, and Lutjering, 1986) showed a similar trend in Ti-8.6Al alloy with grain sizes of 20 and 100 m. In addition, some studies (Hirose and Fine, 1983) reported slower growth in a powder metallurgy aluminum alloy with a fine grain size. However, a different behavior was observed (Taira, Tanaka, and Hoshina, 1979) in a 0.2% C steel with a ferrite/pearlite microstructure. They observed similar growth rates at ferrite grain sizes of 20.5 and 55 m and lower growth rates only when the grain size was reduced to 7.8 m. In contrast to these results are those of (Brown and Taylor, 1984, and Zurek, James and Morris, 1982). Their studies (Brown and Taylor, 1984) in a mill annealed alpha/beta titanium alloy (Ti-6Al-4V) with grain sizes of 4.7 and 11.7 m, could not detect any grain size effect. Also, (Zurek, James and Morris, 1982) in 7075-T6 aluminum alloy, observed a decrease in growth rate with increasing grain size from 12 to 130 m. Thus, much of the published work on the effect of grain size on "small" fatigue crack growth indicates that increasing grain size leads to "faster" crack growth rates. In the microstructurally "small" crack regime, this trend is related to the difference in the "blocking effect" of grain boundaries. This so called "blocking effect" was observed to occur more frequently in fine grained material than in coarse grained material. As demonstrated (Tokaji, Ogawa, Harada and Ando, 1986) in quenched and tempered steel, prior austenite grain boundaries act as barriers to the growth of microstructurally "short" cracks. It was observed in fine and coarse grained materials. The effect of variation of precipitate sizes on the "short" crack behavior was studied (Brown, King and Hicks, 1984) using Astroloy with 50 m grain size. In this study, they investigated a range of gamma prime distributions, achieved by different heat-

58

treatment processes and they found similar "short" crack growth rates in all conditions. Also, similar observation was made in a nickel base super alloy, Waspaloy, with about the same grain size, but somewhat lower gamma prime volume fraction. It also was realized by some investigators that one of the important ways of controlling short crack behavior in steels and titanium alloys was through variations in the distribution and proportions of the phases present. (De los Rios, Tang and Miller, 1984) When tested in a 0.4% C steel with a strongly banded structure with alternate layers of alpha and pearlite, observed cracks nucleating in the alpha and were held up by the alpha/pearlite interfaces and sometimes the propagation was completely stopped after reaching these boundaries. Also, in another study using the same material (De los Rios, Mohamed and Miller, 1985), but with a different microstructure in which the alpha outlined the prior austenite (gamma) grain boundaries, the similar observation was made. In 0.37% C steel (Hoshide, Yamada, Fujimura, 1985), with an alpha/pearlite microstructure in two conditions, air-cooled from 865C and furnace-cooled from 940C, cracks formed in the alpha/pearlite boundaries and stopped when the cracks reached the pearlite. Also, slower "short" crack growth rates were presented in the air-cooled material with the finer distribution of alpha. In one of the earlier studies (Kunio and Yamada, 1979) using martensite steel (alpha prime/ferrite alpha mixtures consisting of about 50 volume % of each), crack formation occurred in the alpha and growth stopped on reaching alpha prime. However, it was found that at higher stress levels these cracks continued to propagate, but until they reached lengths of 400-500 m, their growth was still impeded by regions of alpha prime. From these studies, in steels, it was clearly observed that the presence of harder phases forced the cracks to take a tortuous path by deflecting the crack paths. Similar study was carried out in titanium alloys (Hicks and Brown, 1984). They compared the behavior of a beta processed titanium alloy, IMI 685, with a coarse aligned alpha structure with that of alpha/beta heat-treated IMI 318. The changes in orientation of the alpha plates at the prior beta grain boundaries and the coarse

59

beta grain sizes produced at the high beta heat-treated proved to be the main barriers to "short" crack propagation in the IMI 685. However, in IMI 318 which consisted of regions of primary alpha and transformed beta, cracks formed in the alpha and were impeded by the harder regions of transformed beta, thereby meeting the effective barriers at lengths very much shorter than the prior beta grain size. A five-fold difference in average growth rate was obtained between the two microstructures. Also, another study in IMI 318 and IMI 550 with different heat-treatment processes (Boilngbroke and King, 1986) clearly showed that a finer and harder transformed beta impedes the crack growth more effectively than a coarser transformed beta produced at a slower cooling rate. Moreover, they showed that an average "short" crack growth rates in the beta heat-treated structure are up to an order of magnitude faster than in an alpha/beta heat-treated condition in IMI 318. Similar observations were made in dual phase steels with ferrite and martensite phases, crack growth rates and crack path were strongly affected by martensite phase (Minakawa, Matsuo, and Mcevily, 1982, Dutta, Suresh, and Ritchie, 1984, and Shang, Tzou, and Ritchie, 1987). Somewhat similar research observations also were presented (Tokaji and Ogawa, 1988) in medium carbon steel and dual-phase stainless steel. It also was found that as cracks grew into pearlite phase from ferrite phase in medium carbon steel and austenite phase from ferrite phase in dual-phase stainless steel, crack growth rates showed a marked decrease at phase interfaces. Also, it was observed that cracks tend to grow predominantly within ferrite phase in medium carbon steel. Fatigue crack was also found to form in softer ferrite in CMn steel (de los Rios, Navarro, and Hussain, 1992). Moreover, (Kawachi, Yamada, and Kunio, 1992) showed the possibility of coalescence of "small" cracks in a dual-phase (martensitic-ferritic) carbon steel and as a result of this, cracks increased their length along the matrix-ferrite, by-passing the harder martensite. They also demonstrated that the crack coalescence could be suppressed in this kind of steel by preparing a dual phase

60

microstructure with the matrix ferrite enclosed by the second phase martensite. This resulted in increase in the fatigue strength of this steel. Also, other microstructurally "short" crack studies in different metals viz. low carbon steel (Tokaji, Ogawa, and Harada, 1986), medium carbon steel (Tokaji and Ogawa, 1988), high tensile steel (Tokaji, Ogawa, and Harada, 1987), low alloy steel (Tokaji, Ogawa, Harada and Ando, 1986), aluminum alloy (Tokaji and Ogawa, 1990), and pure titanium (Tokaji, Ogawa, Kameyama, and Kato, 1990) revealed a similar overall growth behavior as crack growth rates were markedly decreased by grain boundaries, triple pints and interfaces between phases depending on the microstructures. These studies also supported the above mentioned research works that large decreases in microstructurally "short" crack growth rate are more frequent in fine grained materials than in coarse grained materials. This was observed to increase the overall fatigue life in the fine grained materials. In fine and coarse grained materials such as in pure titanium (Tokaji, Ogawa, Kameyama, and Kato, 1990), the crack path was observed to be extremely tortuous and it increased with increase in grain size. Therefore, decrease in crack growth rate resulted. It also was shown that decrease in crack growth rate were more frequently observed in fine grained material as compared to coarse grained material. This was attributed to grain boundary and crack deflection. As described in one study (Suresh, 1983), the deflection in crack path might lower the crack driving force. Also, as hypothesized by some researchers (Tokaji and Ogawa, 1992), the tortuous nature of crack path morphology in pure titanium might be associated with planar slip characteristics or fewer slip systems than bcc and fcc metals. Furthermore, they postulated that because of this, as cracks reached grain boundary, large changes of growth direction might occur due to the incompatibility of deformation and it was related to misorientation between two grains. Although extensive research studies on the microstructurally "short" crack behavior were performed on various possible materials, a direct comparison is not possible because of

61

some obvious differences in test conditions viz. stress level, loading type and other test parameters. However, as indicated before, a general behavior of microstructurally "short" cracks is observed in many of the materials bearing a few contradictory results. The effect of stress ratio (R) on microstructurally "short" cracks was studied by some workers (Tokaji and Ogawa, 1990). In aluminum alloy (7075-T6) tested at R=-1 and R=0, faster growth was recorded at R=-1. Also, the results at R=-2 showed the fastest crack growth rates. (Tokaji and Ogawa, 1992) observed stage I facets on the fracture surfaces when tested at R=-1 and R=-2 but not at R=0. They related the growth behavior of microstructurally "short" cracks to the existence of stage I facets. Also, some studies on Stage I crack formation as related to "short" crack growth behavior are discussed in the next section.

3.3 Stage I crack propagation studies and its relation to "short" crack growth behavior

The slip mechanism for the formation of fatigue cracks was first studied by Ewing and Humphrey, Gough and Forsyth among others. Since then, formation of persistent slip bands (PSB) has been recognized as a general phenomenon for the nucleation of fatigue cracks. Forsyth termed the formation and growth of crack in slip bands during the fatigue process as Stage I. Stage I or slip band cracking was related to the range of resolved shear stress on the slip plane (Hoeppner, 1967 and Forsyth, 1969). Also, slip bands produced by cyclic stress were shown to be a series of grooves and ridges and the fatigue deformation mechanism of "slip band intrusion and extrusion" was related to Stage I crack growth mechanism (Forsyth, 1969). However, not all the fatigue cracks form from slip bands. Under favorable conditions of stress and environment these cracks may form on those planes most closely aligned with the maximum shear-stress directions in the component or test specimens. Some "short" crack research studies related the behavior of cracks in the "short" regime to Stage I cracks and they are discussed below.

62

Two high strength and low alloy steel containing V and Nb were fatigue tested to observe the nucleation of cracks and growth of "microcracks" (Kim and Fine, 1982). Strain controlled fatigue tests were conducted in 30-40% humidity air at room temperature. Also, results of crack formation at different mean stress were reported. This study showed that at all stress levels, fatigue cracks were observed to form along persistent slip bands. The number of cycles to nucleate a crack of size 5 m long was observed at 2000 to 3000 magnification and the "microcracks" were always found to be associated with slip bands. (Kwun and Fournelle, 1982) also reported more density of slip bands and "small" cracks for quenched and tempered at a lower temperature when compared to tempering performed at higher temperature in Niobium bearing high strength and low alloy steel. Although the formation of cracks in extruded aluminum alloy X7091 containing Zn-Mg-Cu and Co occurred at grain boundaries at both low and high stresses, the same was not true when the material was subjected to thermomechanical treatment. This resulted in the slip band crack nucleation (Hirose and Fine, 1983). Another study (Kim, Mura and Fine, 1978) in 4140 steel also showed that "microcracks" formed at grain boundaries in as-quenched specimens and at intrusions and extrusions in the tempered specimens. In another study, Tokaji and Ogawa, 1988, observed many straight lines on the facets of medium carbon steel and the direction of these lines were confirmed to be consistent with the slip direction <111> of this material. This indicated that the crack grew along slip planes in a shear mode and it was related to stage I crack growth mechanism. Moreover, they (Tokaji and Ogawa, 1992) argued that the microstructural effect "short" crack growth occurred as a sequence of stage I crack growth. At higher strains, at two different microstructures, when fatigue tested, Ti-6Al2Sn-4Zr-6Mo exhibited nucleation of cracks within the slip bands in Widmanstatten plus grain boundary alpha and equiaxed structures of different alpha particle (Mahajan and

63

Margolin, 1980). However, at low strains the crack nucleation occurred at alpha-beta interface and the resultant "microcracks" linked up and extended. Similar observation was made in another study conducted in 2024 and 2124 aluminum alloy in the T-4 condition (Kung and Fine, 1979). This study was conducted in dehumidified laboratory air, 10% humidity as well as 50% humidity. The loading was applied in tension-tension and also in tension-compression with the direction of stress normal to the long direction of the notch and parallel to the specimens rolling direction. It was found that at high stresses the fatigue cracks formed on "coarse slip lines" in both alloys. However, at low stresses majority of cracks originated from the constituent particles. However, the important finding of this study was the probability of nucleation of a fatigue crack at a constituent particle size normal to the stress direction decreased below 6 m and the crack formation mode was along slip bands which originated from the inclusions. Polycrystalline copper of commercial purity (99.9%) exhibited formation of extrusions and intrusions along persistent slip bands within the grain and also in preferably oriented grain boundaries when tested in constant strain amplitude cyclic loading (Polak and Liskutin, 1990). "Short" crack growth was characterized by PSB nucleation and was related to localization of the cyclic slip at or close to grain boundaries and "the cracks advanced by linking a newly nucleated crack at the tip of the existing crack or in front of it" (Plumtree and O'Connor, 1991) also observed the stage I crack growth to a depth of about 250-350 m in Al-Mg-Si alloy subjected to two different processing techniques viz. extruded and squeeze-cast. They stated that the "short" crack behavior was observed during this portion of fatigue life. Different kinds of crack nucleation mechanisms were observed in nickel base super alloy namely, Waspaloy (Yates, Zhang and Miller, 1993). Four point bending fatigue tests were conducted on Waspaloy specimens and was found that crack formed directly from PSBs, along a twin boundary and also in grains. Moreover, it was observed

64

that in all the three types, cracks formed at 45 to the principal stress axis suggesting the Stage I growth until the crack depth of around 600 m. This study proposed a model for the "short" crack growth behavior in Waspaloy incorporating the microstructural effect as well as the characteristic nature of grain boundary in blocking the growth of "short" cracks.

3.4 Environment effects on "short" crack behavior of materials

The environment (chemical) and temperature effects on the fatigue crack propagation in the "long" crack regime are not well understood and the same is applicable to "short" crack studies. Very few researchers have performed "short" crack studies to evaluate these issues. In an earlier study, very low environmental influence on "small" surface cracks was observed in a 7075-T651 alloy when tested in ambient air and in purified nitrogen (2 ppm water) (Lankford, 1983). However, some studies (Zeghloul and Petit, 1985, Petit and Zeghloul, 1986) have shown a greater effect of environment on the propagation of "short" through cracks grown in a 7075-T651 and T7351 when tested in air and in purified nitrogen and also the test results were compared in vacuum. Moreover, in another study (Petit and Zeghloul, 1990), faster growth rate of "small" surface cracks was observed in ambient air compared to vacuum when 7075-T651 and T7351 were fatigue tested. This behavior was related to water vapor embrittlement as in the case of long cracks. These studies have shown that the growth in ambient air of stage II cracks in a 7075 alloy can be rationalized with that of long and "short" through-section cracks in terms of Keff after correction for local plasticity. However, another study (Petit and Kosche, 1992), showed that the initial propagation in vacuum of "small" surface cracks naturally "initiated" on smooth specimens in 7075-T651 and 7075-T351 alloys is much faster than stage II

65

propagation, and is similar to the intrinsic stage I regime, as identified on Al-Zn-Mg single crystals. More dramatic results were shown in a review paper (Gangloff and Wei, 1986) in which it was concluded that "small" corrosion fatigue cracks in high strength steels when cycled in aqueous hydrogen producing environments grew up to 500 times faster than "long" crack at constant K. Another study (Akid and Murtaza, 1992) showed that environmental assisted "short" fatigue crack growth could influence the propagation of cracks beyond the "crackarresting barriers" such as grain boundaries. Their studies in high strength spring steel using an intermittent fatigue tests in air and NaCl solution clearly showed that strain assisted dissolution caused the transition of stage I crack to stage II at shorter crack lengths. Also, (Boukerrou and Cottis, 1992), another study in a structural steel (BS4340 grade) in 3.5% NaCl and in pitting solution, showed that cracks could nucleate from corrosion pits when cycled at low stresses.

3.5 Temperature effects on "short" crack behavior of materials

At room temperature, "short" cracks were found to form primarily at inclusion particles, and less frequently at grain boundaries in Nickel base super alloy when cyclically loaded in four point bending specimens (Mei, Krenn, and Morris, 1993). However, when tested at elevated temperature (873K) "short" cracks were observed to form from micropores, slip planes and carbide precipitates at grain boundaries in Nickel base super alloy (Okazaki, Tabata and Nohmi, 1990). In this study, Stage I fatigue fracture occurred on the {111} planes. Also, (Stephens, Grobowski and Hoeppner, 1993) slip band cracking was observed in Waspaloy tested at 25C, 500C and crack was observed to form at twin boundaries and slip bands at 700C. Moreover, this study demonstrated the "faster" growth rate of "short" cracks at 500C when compared to 25C

66

and 700C. This was related to diffusive nature of slip bands and "changes in material leading to precipitate coarsening" at 700C. Similar observation of slip band cracking was found in Waspaloy when tested at 19C and 500C (Healy, Grabowski and Beevers, 1991). In addition, "short" cracks formed at coarse carbide particles. The "short" crack growth rate measured at R=0.1 was "faster" at 500C when compared to 19C at an equivalent values of stress intensity range as reported by Stephens, Grobowski and Hoeppner. In another study (Suh, Lee and Kang, 1990), numerous "microcracks" were observed to form at grain boundaries in 304 stainless steel specimens tested at 538C. In general, the mechanisms for the "short" crack behavior at room and elevated temperature as postulated in the above mentioned studies are related to absence of closure effects, heterogeneous microstructure (responsible for statistical scatter in the "short" crack growth rate (Goto, 1993; Goto, 1994)), grain boundary cracking (may result from embrittlement due to stress-assisted grain boundary oxidation during heat treatment), grain orientation (if the grains are favorably oriented the cracks are observed to grow "faster" and if not "short" crack growth either gets slowed down or arrested), crack deflection (may occur within a grain or when the crack passed to another grain), crack tip deflection resulting in "roughness induced crack closure" and subsequent reduction in crack growth rate, Stage I crack growth (Stage I crack growth increases as the crack length increases. However, as the crack tip neared a grain boundary the "short" crack growth rate decreased because of crack deflection due to secondary slip and finally fracture occurring on the {111} planes in fcc).

67

3.6 Studies on transition of pit to small crack

A few small crack studies under corrosion fatigue conditions have been performed to characterize the transition of a pit to a small crack. In 2024 aluminum alloy, Piascik and Willard have shown a three times increase in crack growth rates of small cracks in salt water environment when compared to air. Moreover, their studies clearly have observed the transition of pits formed at the constituent particles to intergranular microcracks and then to transgranular fracture path once the crack reaches the depth of 100 mm. In addition, the increase in small crack growth rates was observed even at very low mode I K (<1 MPa m). As well, Kondo (1989) also observed in two low alloy steels that short cracks from pits propagated at K that is well below the threshold value of a long crack for these materials. In a recent in-situ fatigue study, prior pitted 2024-T351 and 7075-T651 aluminum alloy specimens exhibited faster crack growth rates in the short crack regime when compared to specimens without prior corrosion damage (A. Hoeppner (now A. Taylor) 1996). This study showed that prior corrosion damage did influence the small crack growth rates. It also was observed that the 7075 aluminum alloy specimen had faster crack growth rates compared to the 2024 aluminum alloy specimens. Also, in this study cracks were observed to form from pits on the prior corroded specimens whereas on the specimens without any prior corrosion damage, cracks formed from constituent particles.

3.7 Conclusion

This literature review on microstructural and environmental effects of "short" crack behavior of structural materials has clearly revealed the lack of experimental studies to characterize materials response in the "short" crack regime under fretting fatigue and corrosion conditions. As mentioned in this report, very little data have been published

68

with regard to "short" crack formation mechanisms in corrosive environments of aluminum alloys in aircraft structures. In addition to a few previous studies (Hoeppner, 1971, 1979) in which pitting was modeled statistically with different materials and specimen types, recently, as discussed before in this report, there was a study demonstrating corrosion fatigue induced "short" crack formation from pits (Akid and Murtaza, 1992). Also, recent studies (Ma and D. Hoeppner, 1994, Grimes, 1996, and A. Hoeppner, 1996) have shown that pits form in different shapes depending upon environment and loading conditions in contradictory to general assumption that pits have hemispherical shape. Although this assumption simplifies the modeling part of research (Kondo, 1989), further studies to characterize the formation of cracks from pits in the "short" crack regime must be evaluated as indicated by A. Hoeppner (1996). Apart from these studies the literature search has not found any "short" crack studies to evaluate the formation of cracks from pits and their crack morphologies and paths. Moreover, this may further be aggravated by fretting mechanism(s) in conjunction with fatigue and corrosion.

To conclude this section, a global view of the "short" crack challenge is schematically illustrated in Appendix I.

69

3.8 References AGARD Conference Proceedings No. 328 (1983), Behavior of Short Cracks in Airframe Components, Papers presented at the 55th Meeting of the AGARD Structures and Materials Panel in Toronto, Canada on 19-24 September. R. Akid and G. Murtaza (1992), "Environment assisted short crack growth behavior of a high strength steel", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 193-208. R.K. Bolinbroke and J.E. King (1986), "The growth of short fatigue cracks in titanium alloys IMI550 and IMI318", Small Fatigue Cracks, R.O. Ritchie and J. Lankford, Eds., A Publication of The Metallurgical Society, Inc., pp. 129-144. A. Boukerrou and R.A. Cottis (1992), "The influence of corrosion on the growth of short fatigue cracks in structural steels", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 209-217. J.L. Breat, F. Mudry and A. Pineau (1983), "Short crack propagation and closure effects in A508 steel", Fatigue Fracture of Engineering Materials Structures, 6, pp. 349358 C.W. Brown, J.E. King, and M.A. Hicks (1984), "Effects of microstructure on long and short crack growth in nickel base super alloys", Metal Science, 18, pp. 374-380. C.W. Brown and D. Taylor (1984), "The effects of texture and grain size on the short fatigue crack growth rates in Ti-6Al-4V", Fatigue Crack Threshold Concepts, D.L. Davidson and S. Suresh, Eds, AIME, pp. 433-446. D.W. Cameron (1984), Perspectives and insights on the cyclic response of metal, Ph.D. Dissertation, University of Toronto. K.S. Chan and J. Lankford (1988), "The role of microstructural dissimilitude in fatigue and fracture of small cracks", Acta Metallurgica, pp. 193-206. P. Clement, J.P. Angeli and A. Pineau (1984), "Short crack behavior in nodular cast iron", Fatigue Fracture of Engineering Materials and Structures, 7, pp. 251-265. 70

D.L. Davidson (1988), "Small and large fatigue cracks in aluminum alloys", Acta Metallurgica, 36, pp. 2275-2282. D.L. Davidson, J.B. Campbell, and R.A. Page (1991), "The initiation and growth of fatigue cracks in a titanium aluminide alloy", Metallurgical Transactions, 22A, pp. 377391. E.R. De Los Rios, Z. Tang and K.J. Miller (1984), "Short crack fatigue behavior in a medium carbon steel", Fatigue Fracture of Engineering Materials and Structures, 7, pp. 97-108. E.R. De Los Rios, H.J. Mohamed, and K.J. Miller (1985), "A micro-mechanic analysis for short fatigue crack growth", Fatigue Fracture of Engineering Materials and Structures, 8, pp. 49-63. E.R. de los Rios, A. Navarro, and K. Hussain (1992), "Microstructural variations in short fatigue crack propagation of a C-Mn steel", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 115-132. V.B. Dutta, S. Suresh, and R.O. Ritchie (1984), "Fatigue crack propagation in dualphase steels: effects of ferritic-martensitic microstructures on crack path morphology", Metallurgical Transactions, 15A, pp. 1193-1207. J.N. Eastabrook (1984), "A dislocation model for the rate of initial growth of stage I fatigue cracks", International Journal of Fracture, 24, pp. 43-49. M.H. El Haddad, N.E. Dowling, T.H. Topper, and K.N. Smith, "J-integral applications for short fatigue crack at notches", International Journal of Fracture, 16, pp. 15-30. P.J.E. Forsyth (1969), The physical basis of metal fatigue, Blackie and Son Limited. R.P. Gangloff (1985), "Crack size effects on the chemical driving force for aqueous corrosion fatigue", Metallurgical Transactions, 16A, pp. 953-969.

71

R.P. Gangloff and R.P. Wei (1986), "Small crack-environment interaction: The hydrogen embrittlement perspective", Small Fatigue Cracks, R.O. Ritchie and J. Lankford, Eds., A Publication of The Metallurgical Society, Inc., pp. 239-264. M. Goto (1993), "Scatter in small crack propagation and fatigue behavior in carbon steels", Fatigue Fracture of Engineering Materials and Structures, 16, pp. 795-809. M. Goto (1994), "Statistical investigation of the behavior of small cracks and fatigue life in carbon steels with different ferrite grain sizes", Fatigue Fracture of Engineering Materials and Structures, 17, pp. 635-649. J.C. Healy, L. Grabowski, and C.J. Beevers (1991), "Short-fatigue-crack growth in a nickel-base superalloy at room and elevated temperature", International Journal of Fatigue, 13, pp. 133-138. M.A. Hicks and C.W. Brown (1984), "A comparison of short crack growth behavior in engineering alloys", Fatigue 84, C.J. Beevers, Ed., EMAS, Warley, U.K. pp. 1337. S. Hirose and M.E. Fine (1983), "Fatigue crack initiation and microcrack propagation in X7091 type aluminum P/M alloys", Metallurgical Transactions, 14A, pp. 1189-1197. P.D. Hobson (1982), "The formulation of a crack growth equation for short cracks", Fatigue Fracture of Engineering Materials and Structures, 5, pp. 323-327. D.W. Hoeppner (1967), The effect of grain size on fatigue crack propagation in copper, Fatigue Crack Propagation, ASTM 415, American Society for Testing and Materials, pp. 486-504. D.W. Hoeppner (1981), Estimation of component life by application of fatigue crack growth threshold knowledge in Fatigue, Creep and Pressure Vessels for Elevated Temperature Service, The American Society of Mechanical Engineers, N. Y. pp. 1-83. D.W. Hoeppner (1983), Application of Damage Tolerance concepts to short cracks in safety critical components, in Industrial Applications of Damage Tolerance Concepts, Proceedings of the Twelfth ICAF Symposium, ICAF Doc. No. 1336, pp. 2.1/1 - 2.2/20

72

A.M. Hoeppner now A. Taylor (1996) The effect of prior corrosion damage on the short crack growth rates of two aluminum alloys, M.S. Thesis, University of Utah. T. Hoshide, T. Yamada, and S. Fujimura (1985), "Short crack growth and life prediction in low-cycle fatigue and smooth specimens", Engineering Fracture Mechanics, 21, pp. 85-101. P. Hyspecky and B. Strnadel (1992), "Conversion of short fatigue cracks into a long crack", Fatigue Fracture of Engineering Materials and Structures, 15, pp. 845-854. S. Kawachi, K. Yamada, and T. Kunio (1992), "Some aspects of small crack growth near threshold in dual phase steel", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 101-114. Y.H. Kim, T. Mura, and M.E. Fine (1978), "Fatigue crack initiation and microcrack growth in 4140 steel", Metallurgical Transactions, 9A, pp. 1679-1683. Y.H. Kim and M.E. Fine (1982), "Fatigue crack initiation and strain-controlled fatigue of some high strength low alloy steels", Metallurgical Transactions, 13A, pp. 59-71. S. Kumai, J.E. Kino, and J.F. Knott (1990), "Short and long fatigue crack growth in a SiC reinforced aluminum alloy", Fatigue Fracture of Engineering Materials and Structures, 13, pp. 511-524. C.Y. Kung and M.E. Fine (1979), "Fatigue crack initiation and microcrack growth in 2024-T4 and 2124-T4 aluminum alloys", Metallurgical Transactions, 10A, pp. 603-610. T. Kunio and K. Yamada (1979), "Microstructural aspects of the threshold condition for non-propagating fatigue cracks in martensitic-ferritic structures", Fatigue Mechanisms, J.T. Fong, Ed., ASTM STP 675, pp. 342-370. S.I. Kwun and R.A. Fournelle (1982), "Fatigue crack initiation and propagation in a quenched and tempered niobium bearing HSLA steel", Metallurgical Transactions, 13A, pp. 393-399.

73

J. Lankford, T.S. Cook, and G.P. Sheldon (1981), "Fatigue microcrack growth in nickel-base superalloy", International Journal of Fracture, 17, pp. 143-155. J. Lankford (1982), "The growth of small fatigue cracks in 7075-T6 aluminum", Fatigue Fracture of Engineering Materials and Structures, 5, pp. 233-248. J. Lankford (1983), "The effect of environment on the growth of small fatigue cracks", Fatigue Fracture of Engineering Materials and Structures, 6, pp. 15-31. J. Lankford and D.L. Davidson (1986), "The role of metallurgical factors in controlling the growth of small fatigue cracks", Small Fatigue Cracks, R.O. Ritchie and J. Lankford, Eds., A Publication of The Metallurgical Society, Inc., pp. 51-72. Y. Mahajan and H. Margolin (1982), "Low cycle fatigue behavior of Ti-6Al-2Sn-4Zr6Mo: Part I. The role of microstructure in low cycle crack nucleation and early crack growth", Metallurgical Transactions, 13A, pp. 257-267. R.C. McClung, K.S. Chan, S.J. Hudak, and D.L. Davidson (1994) "Analysis of small crack behavior for airframe applications", FAA/NASA International Symposium on Advanced Structural Integrity Methods for Airframe Durability and Damage Tolerance. A.J. McEvily and Minakawa (1984), "Crack closure and the conditions for crack propagation", Fatigue Crack Growth Threshold Concepts, D.L. Davidson and S. Suresh, Eds., TMS-AIME, pp. 517-530. Z. Mei and J.W. Morris (1993), "The growth of small fatigue cracks in A286 steel", Metallurgical Transactions, 24A, pp. 689-700. Z. Mei, C.R. Krenn, and J.W. Morris (1995), "Initiation and growth of small fatigue cracks in a Ni-base superalloy", Metallurgical Transactions, 26A, pp. 2063-2073. K.J. Miller (1982), "The short crack problem", Fatigue Fracture of Engineering Materials and Structures, 5, pp. 223-232

74

K. Minakawa, Y. Matsuo, and McEvily (1982), "The influence of a duplex microstructure in steels on fatigue crack growth in the near-threshold region", Metallurgical Transactions, 13A, pp. 439-445.. W.L. Morris, O. Buck, and H.L. Marcus (1976), "Fatigue crack initiation and early propagation in Al 2219-T851", Metallurgical Transactions, 7A, pp. 1161-1165. W.L. Morris and O. Buck, (1977), "Crack closure load measurements for microcracks developed during the fatigue of Al 2219-T851", Metallurgical Transactions, 8A, pp. 597601. Y. Murakami and M. Endo, (1983), Quantitative evaluation of fatigue strength of metals containing various small defects or cracks, Engineering Fracture Mechanics, 17, No. 1, pp. 1-15. D.J. Nicholls and J.W. Martin (1991), "A comparison of small fatigue crack growth, low cycle fatigue and long fatigue crack growth in Al-Li alloys", Fatigue Fracture of Engineering Materials and Structures, 14, pp. 185-192. J.C. Newman and P.R. Edwards (1988), "Short crack growth behavior in an aluminum alloy" -- An AGARD cooperative test Programme. M. Okazaki, T. Tabata, and S. Nohmi (1990), "Intrinsic Stage I crack growth of directionally solidified Ni-Base superalloys during low-cycle fatigue at elevated temperature", Metallurgical Transactions, 21A, pp. 2201-2208. J.Z. Pan, E.R. De Los Rios, and Miller (1993), "Short fatigue crack growth in plain and notched specimens of an 8090 Al-Li alloy", Fatigue Fracture of Engineering Materials and Structures, 16, pp. 1365-1379. S. Pearson (1975), "Initiation of fatigue cracks in commercial aluminum alloys and the subsequent propagation of very short cracks", Engineering Fracture Mechanics, 7, pp. 235-247. J. Petit and A. Zeghloul (1990), "Environmental and Microstructural influence on fatigue propagation of small surface cracks", Environmentally Assisted Cracking: 75

Science and Engineering, ASTM STP 1049, W.B. Lisagor, T.W. Crooker, and B.N. Leis, Eds., American Society for Testing and Materials, Philadelphia, pp. 334-346. J. Petit, J. Mendez, W. Berata, L. Legendre, and C. Muller (1992), "Influence of environment on the propagation of short fatigue cracks in a titanium alloy", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 235-250. J. Petit and A. Zeghloul (1986), "On the effect of environment on short crack growth behavior and threshold", The Behavior of Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 163-178. J. Petit and K. Kosche (1992), "Stage I and Stage II propagation of short and long cracks in Al-Zn-Mg alloys", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 135-151. R.S. Piascik and S.A. Willard (1994), The growth of small corrosion fatigue cracks in alloy 2024, Fatigue Fracture of Engineering Materials and Structures, Vol. 17, pp. 12471259. A. Plumtree and B.P.D. O'Connor (1991), "Influence of microstructure on short fatigue crack growth", Fatigue Fracture of Engineering Materials and Structures, 14, pp. 171184. J. Polak and P. Laskutin (1990), "Nucleation and short crack growth in fatigued polycrystalline copper", Fatigue Fracture of Engineering Materials and Structures, 13, pp. 119-133. J.M. Potter and B.G.W. Yee (1983), "Use of small crack data to bring about and quantify improvements to aircraft structural integrity", in Behavior of Short Cracks in Airframe Components, AGARD. R.O. Ritchie and J. Lankford (1986) "Small fatigue cracks: A statement of the problem and potential solutions", Materials Science and Engineering, 84, pp. 11-16.

76

R.O. Ritchie and S. Suresh (1995), "Mechanics and physics of the growth of small cracks", A University of California, Berkely report. J. Schijve (1986), "Fatigue crack closure, observations and technical significance", NLR TR-679, National Aerospace Laboratory NLR, Amsterdam. J. Schijve (1992), "Multiple-site-damage fatigue of riveted joints", International Workshop On Structural Integrity Of Aging Airplanes, Atlanta, 31 March-2 April. J.K. Shang, J.L. Tzou, and K.J. Miller (1987), "Role of crack tip shielding in the initiation and growth of long and small fatigue cracks in composite microstructures", Metallurgical Transactions, 18A, pp. 1613-1627. G.P. Sheldon, T.S. Cook, J.W. Jones, and J. Lankford (1981), "Some observations on small fatigue cracks in a superalloy", Fatigue Fracture of Engineering Materials and Structures, 3, pp. 219-228. D. Sigler, M.C. Montpetit, and W.L. Haworth (1983), "Metallography of fatigue crack initiation in an overaged high-strength aluminum alloy", Metallurgical Transactions, 14A, pp. 931-938. R.R. Stephens, L. Grabowski and D.W. Hoeppner (1993), "The effect of temperature on the behavior of short fatigue cracks in Waspaloy using an in situ SEM fatigue apparatus", International Journal of Fatigue, 15, pp. 273-282. C.M. Suh, J.J. Lee, and Y.G. Kang (1990) "Fatigue microcracks in type 304 stainless steel at elevated temperature", Fatigue Fracture of Engineering Materials and Structures, 13, pp. 487-496. S. Suresh (1983), "Crack Deflection: Implications For The Growth Of Long And Short Fatigue Cracks", Metallurgical Transactions, 14A, pp. 2375. S. Suresh (1985), "Fatigue crack deflection and fracture surface Micromechanical model", Metallurgical Transactions, 16A, pp. 249-260. contact:

77

S. Taira, K. Tanaka, and Hoshina (1979), "Grain size effect on crack nucleation and growth in long-life fatigue of low-carbon steel", Fatigue Mechanisms, J.T. Fong, Ed., ASTM STP 675, Philadelphia, pp. 135-173. D. Taylor and J.F. Knott (1981), "Fatigue crack propagation behavior of short cracks; The effect of microstructure", Fatigue Fracture of Engineering Materials and Structures, 4, pp. 147-155. K. Tokaji, T. Ogawa, and Y. Harada (1986), "The growth of small fatigue cracks in a low carbon steel; The effect of microstructure and limitations of linear elastic fracture mechanics", Fatigue Fracture of Engineering Materials and Structures, 9, pp. 205-217. K. Tokaji, T. Ogawa, Y. Harada, and Z. Ando (1986), "Limitations of linear elastic fracture mechanics in respect of small fatigue cracks and microstructure", Fatigue Fracture of Engineering Materials and Structures, 9, pp. 1-14. K. Tokaji, T. Ogawa, and Y. Harada (1987), "Evaluation on limitation of linear elastic fracture mechanics for small fatigue crack growth", Fatigue Fracture of Engineering Materials and Structures, 10, pp. 281-289. K. Tokaji, T. Ogawa, S. Osako, and Y. Harada (1988), "The growth behavior of small fatigue cracks; the effect of microstructure and crack closure", Fatigue '87, Vol 2, pp. 313-322. K. Tokaji and T. Ogawa (1988), "The growth of microstructurally small fatigue cracks in a ferritic-pearlitic steel", Fatigue Fracture of Engineering Materials and Structures, 11, pp. 331-342. K. Tokaji, T. Ogawa, and T. Aoki, (1990), "Small fatigue crack growth in a low carbon steel under tension-compression and pulsating-tension loading", Fatigue Fracture of Engineering Materials and Structures, 13, pp. 31-39. K. Tokaji and T. Ogawa (1990), "The effects of stress ratio on the growth behavior of small fatigue cracks in an aluminum alloy 7075-T6 (with special interest in Stage I crack growth", Fatigue Fracture of Engineering Materials and Structures, 13, pp. 411-421.

78

K. Tokaji, T. Ogawa, Y. Kameyama, and Y. Kato (1990), "Small fatigue crack growth behavior and its statistical properties in a pure titanium", Fatigue '90, Vol 2, pp. 10911096. K. Tokaji and T. Ogawa (1992), "The growth behavior of microstructurally small fatigue cracks in metals", Short Fatigue Cracks, K.J. Miller and E.R. de los Rios, Eds, Mechanical Engineering Publications Limited, pp. 85-100. K.T. Venkateswara Rao, W. Yu, and R.O. Ritchie (1988), "Fatigue crack propagation in aluminum-lithium alloy 2090: Part II. Small crack behavior", Metallurgical Transactions, 19A, pp. 563-569. L. Wagner, J.K. Gregory, A. Gysler, and G. Lutjering (1986), "Propagation behavior of short cracks in a Ti-8.6Al alloy", Small Fatigue Cracks, R.O. Ritchie and J. Lankford, Eds., A Publication of The Metallurgical Society, Inc., pp. 117-128. H.A. Wood and J.L. Rudd (1983), "Evaluation of small cracks in airframe structures". D.C. Wu (1986), An investigation into the fatigue crack growth characteristics of a single crystal nickel-base superalloy, Ph.D. Dissertation, University of Toronto. J.R. Yates, W. Zhang, and K.J. Miller (1993), "The initiation and propagation behavior of short fatigue cracks in waspaloy subjected to bending", Fatigue Fracture of Engineering Materials and Structures, 16, pp. 351-362. A. Zeghloul and J. Petit (1985), "Environmental sensitivity of small crack growth in 7075 aluminum alloy", Fatigue Fracture of Engineering Materials and Structures, 8, pp. 341-348. A.K. Zurek, M.R. James, and W.L. Morris (1982), "The effect of grain size on fatigue growth of short cracks", Metallurgical Transactions, 14A, pp. 1697-1705.

79

Appendix I -- A global view of short crack challenge in materials

80

4 Conclusions and recommendations The review of the literature clearly show that much progress has been made on modeling the effects of corrosion on material behavior and structural integrity. It is clear that to date the models have centered around characterizing the corrosion and modeling the effects of the corrosion as one or more of the following: section change that affects the area/volume that modifies the stress. nucleation of localized debris that may modify the stress (part of pillowing) that modifies the stress or stress intensity. nucleation of intergranular corrosion that is involved in pillowing that modifies the stress or stress intensity. nucleation of localized corrosion (pitting, fretting, etc.) that modifies the local stress and may ultimately nucleate cracks. production of products of corrosion that produce localized embrittlement effects that may alter the material behavior and produce accelerated crack propagation. All of the above have been reviewed in the preceding sections and lead to the recognition that one of the most pressing issues to be resolved is the actual quantitative characterization of the corrosion in relation to the physical damage state that is underway. Some of this has been accomplished in the past with the efforts of the past at the U. of Utah as discussed in the earlier sections of the report. In addition, the efforts currently underway in other portions of the NCI Information systems effort at U. Of Virginia under the leadership of Dr. Kelly and those at Vanderbilt University under the leadership of Dr. Wikswo and some of the internal NCI efforts may provide additional insight into the characterization issue. From the work of L. Grimes at Utah and the effort underway at U. Virginia, as well as additional efforts at the U. of Utah, the use of the confocal microscope will be of great assistance in characterizing the three-dimensional (3-D) surface damage that results from corrosion of various forms. Some recent work at Lehigh University on the characterization of pits also will be useful. These efforts must be developed further to enhance the models and their development.

94

Even though fracture mechanics based modeling has been extremely useful in modeling the effects of corrosion it has taken many simplifications and, depending on the manner in which the fracture mechanics is used in the model, has resulted in downgrading the real characterization issue and understanding the 3-D nature of the corrosion degradation process. New tools and models will have to be brought to bear on the nucleation and growth of the corrosion with or without load of either sustained (SCC) or cyclic nature (EANC/F)-(Environmentally-assisted nucleation and cracking with fatigue loading) . Furthermore the transitions of corrosion to actual cracks will have to be understood to improve the models that currently exist and any new ones that may be developed. The characterization of chemically dependent short crack propagation and modeling of it will have to be much better understood. The efforts of Dr. Piascik and Dr. Newman and those at Utah will have to be expanded to enhance this area as well as the transition to cracking. One area not addressed in the report is the effect of either prior corrosion and/or concomitant corrosion on either fatigue crack propagation or stress-corrosion cracking. Both of these issues are extremely important to the overall area of model development and consideration should be given to expanding this effort to include a review of those areas and also to assure NCI Information Systems and USAF that adequate efforts are underway to allow the ASIP program managers and others to accurately estimate inspection intervals under the appropriate and accurate conditions of interest.

95