Environmentally Assisted Cracking Measurements in Structural Aluminum Alloys Under Accelerated Test Conditions

Environmentally assisted cracking (EAC) in metals is a result of the combined influence of stress (applied or residual), environment, and a susceptible microstructure. EAC in high-strength aluminum alloys such as the 5xxx and 7xxx series is known to result from a combination of electrochemical processes and hydrogen embrittlement. The EAC mechanism is still not fully understood, particularly for 7xxx series alloys that are not susceptible to sensitization.¹  Further advancement in understanding is inhibited by lack of EAC testing and sensing methods that are representative of atmospheric corrosive conditions where EAC of marine and aerospace aluminum structures are most commonly observed.

Most investigations of EAC of aluminum have been conducted under immersion conditions in an aqueous electrolyte. However, it is known that cyclic wetting and drying in humid atmospheric conditions represents the most aggressive corrosion environment for metals.^2-4  It has also been shown that intermittent wetting and drying aggravates EAC in aluminum.^5-6  Immersion testing, while more simple to control and characterize, is not necessarily representative of conditions that put aluminum structures at the highest risk for EAC. Oxygen, reactive halogens, and free radicals are likely to be present in higher concentrations within thin film electrolytes formed under atmospheric conditions leading to more aggressive corrosion. Exposure to humid air is known to facilitate increased hydrogen uptake and embrittlement.^7-8  In addition, conventional, advanced EAC characterization methods typically applied to immersion studies, such as slow strain rate tensile testing and direct current potential drop (DCPD), are only effective for well-controlled laboratory studies. Because of the instrumentation and equipment required, they are not well suited for EAC characterization in more representative conditions such as those present in accelerated test chambers (e.g., salt fog) or in outdoor exposure trials.

In order to advance understanding of EAC of aluminum alloys under the most aggressive and representative atmospheric environments, a new EAC sensor/test method is needed that is simple to deploy and allows quantification of EAC with direct correlation to dynamic environmental conditions (i.e., wetting and drying). This work seeks to address this gap using a sensor that can be applied to not only immersion conditions, but also a wide array of atmospheric exposures and test regimes. The present effort uses a novel crack growth sensor for continuous, in situ measurements of EAC propagation rates in AA5083 (UNS A95083⁽¹⁾) and AA7075 (UNS A97075) under varying atmospheric conditions. The goal of this study and development effort is to demonstrate the capability of the sensor to evaluate the effect of corrosive conditions and load interaction effects on EAC in accelerated atmospheric tests. The latest in the understanding of EAC of high-strength aluminum alloys, particularly AA5083, is presented and summarized to frame where such a sensor would be applicable for progression of the state-of-the-art in EAC research. From an industry perspective, measurements made with the EAC sensor can be used to identify significant relationships between environmental parameters and EAC failure processes. By doing so, new test procedures can be established that will assist in improving knowledge of EAC mechanisms and accelerating the development, qualification, and adoption of EAC-resistant materials and protective coatings.

This section will overview the operating principle of the EAC sensor, details about the fracture specimen fabrication, and the specific accelerated test protocols used during this study.

An EAC sensor tailored for aluminum samples and atmospheric testing was used for the present study. The sensing system developed to support this work was intended to provide relatively low-cost material characterization capability suitable for both in-lab and outdoor exposure applications, the latter requiring ultra-low power consumption for extended battery operation. High-resolution crack-sensing strategies such as DCPD⁹  are well-suited for monitoring small-scale crack growth in fracture specimens, but currently are strictly laboratory-based techniques requiring specialized equipment. The EAC sensor uses a tensile-loaded circumferentially notched specimen, following after the work of Ibrahim and others who demonstrated the ability to produce the required plane-strain conditions within smaller samples compared to other fracture mechanics specimen geometries.¹⁰  As an extension of this earlier work, the EAC sensor observes crack initiation and propagation in real time by continuously comparing the tensile load in the sample with the initial preload value. This measurement strategy has been described in detail previously for monitoring aqueous hydrogen embrittlement of high-strength marine fastener materials.¹¹  The same basic operating principle is used for the aluminum-based environmentally assisted crack monitoring system for atmospheric applications.

Primary sensor components include circumferentially V-notched tensile fracture mechanics sample, preload nut, ported load frame, load cell, and battery-powered instrumentation electronics contained within an IP68-rated enclosure (Figure 1). The basic dimensions of the tensile sample are 98 mm length, Ø14.2 mm major diameter, and Ø10 mm minor diameter at the notch root. The load frame is ported to permit communication between the environment and the notched portion of the sample. The frame is made of AA5083 material to minimize galvanic interaction with the tensile sample and mitigate differential thermal expansion effects on the crack depth measurement. Loaded components within the assembly are designed for high rigidity in order to approach a constant-strain loading condition, with the intent to maximize the sensitivity between the tensile force (measured quantity) and the sample crack depth (desired quantity). Further details about the relationship between load and crack depth are provided elsewhere.¹¹ 

Fracture specimens were machined from AA5083-H116 and AA7075-T651 plate with notches oriented to produce crack planes in the S-L to S-T direction. The 5xxx series alloy samples were sensitized at 150°C until the degree of sensitization (DoS) was within the range of 47 mg/cm² to 49 mg/cm² per ASTM Standard G67-99 nitric acid mass loss test (NAMLT).¹²  New fracture samples were utilized for each test performed. Fracture sample notch tips were polished with 0.3 μm alumina suspension, and all sensor components were thoroughly cleaned prior to testing. AA5083 alloy samples were preloaded to an initial stress intensity between 8 MPa√m and 11 MPa√m, whereas the AA7075 samples were loaded to a higher initial value of 14.4 MPa√m to ensure crack propagation in the nonsensitized material. Each test concluded when the tensile sample had fully fractured.

The accelerated test environments selected for this study included: cyclic RH, ASTM B117,¹³  ASTM G85 A5,¹⁴  and General Motors 9540P.¹⁵  The specific parameters within each test are given below.

An AA5083-H116 sample with DoS 47 mg/cm² and crack plane oriented in the S-L to S-T direction was loaded in the EAC sensor and subjected to a custom cyclic RH exposure test. Testing was performed in a Thermotron 8200^† environmental test chamber. Each 24-h RH cycle began with a 3 h initial hold at 40% RH. RH was increased from 40% to 90% over the next 3 h and then held at 90% for 6 h. The RH was decreased to 40% over the next 9 h and held for an additional 3 h at 40%. Temperature was held constant at 30°C. A 0.5 M sodium chloride salt solution was sprayed on the crack tip at the beginning of a RH cycle, approximately every 75 h of testing. The test was repeated using an AA7075-T651 sample to demonstrate applicability of the method on a different alloy that was not susceptible to sensitization but still vulnerable to EAC.

An AA5083-H116 fracture sample with crack plane oriented in the S-L to S-T direction and DoS 49 mg/cm² was placed in the EAC sensor and exposed continuously in an Auto Technology cyclic corrosion chamber to ASTM B117 salt fog (5% sodium chloride). Temperature was held constant at 35°C. Crack velocity remained low and relatively constant during ASTM B117, so RH was then ramped down until fracture of the sample occurred.

An AA5083-H116 fracture sample with crack plane oriented in the S-L and S-T direction and DoS 49 mg/cm² was placed in the EAC sensor and exposed to an ASTM G85 A5 environment in the accelerated corrosion testing chamber. This test included cyclic exposure of the sensor during 1 h of drying at room temperature and 1 h of fogging at 35°C, with a solution containing 0.05% sodium chloride and 0.35% ammonium sulfate by mass. A modified ASTM G85 A5 experiment was then performed by changing the steps from 1 h of drying and 1 h of fogging to 2 h of drying at room temperature and 2 h of fogging at 35°C in order to provide more time for sensor thermal stabilization. The same solutions were utilized during the standard and modified tests.

An AA5083-H116 fracture sample with crack plane oriented in the S-L and S-T direction and DoS 49 mg/cm² was placed in the EAC sensor and exposed to a GM 9540P environment in the accelerated corrosion testing chamber. This test included a cyclic exposure of alternating wet and dry cycles with varying temperature of the sensor with a solution containing 0.9% sodium chloride, 0.1% calcium chloride, and 0.25% sodium bicarbonate by mass. An iso-temperature GM 9540P experiment was also performed at a constant temperature of 35°C to reduce thermal influences on the crack depth measurement. The same solutions were utilized for both sets of experiments.

Experimental results and discussion are partitioned according to the specific tests conducted. For the cyclic relative humidity test, both AA5083 (sensitized) and AA7075 alloys were evaluated as a comparative study. The remainder of the testing focused on sensitized AA5083 as the primary material of interest.

Crack depth and RH are plotted as a function of time for the last five RH cycles of the test where most of the cracking occurred (Figure 2). The crack depth data were filtered with a moving average method and differentiated with respect to time to produce crack velocity as a function of time (Figure 3). The crack velocity plot is separated into five 24-h RH cycles. Each cycle showed a peak in the crack velocity, representative of the increase in crack rate during the dry off portion of the RH cycle. The peaks increased in magnitude with each subsequent RH cycle and increasing crack depth. This increase most likely occurred because the stress intensity at the crack tip increased as the crack propagated. Lower crack velocity values were associated with the high humidity portion of the RH cycle.

Crack velocity as a function of RH is plotted for the last five RH cycles of the test (Figure 4). Little corrosion occurred during the ramp-up portion of the RH cycle. RH cycle 4 is the only cycle that displayed increasing crack velocity as the RH increased toward 90%. However, there was a sharp decrease in crack velocity immediately before reaching 90% RH. During the transition from wetting to drying, all five cycles showed an increase in crack velocity that occurred between 78% and 85% RH. The crack velocities continued to increase to a maximum crack velocity around 55% to 68% RH. Each cycle’s crack velocity then decreased as the RH decreased to 40%. It should be noted that the crack velocities remained above zero for the entirety of the cycle, including at low RH equal to or slightly below the efflorescence relative humidity of sodium chloride (43±3% RH).¹⁶ 

Crack depth and RH are plotted as a function of time, where the last five RH cycles of the test are plotted to more easily show the crack depth response as it relates to changes in RH (Figure 5). The crack depth data were differentiated with respect to time to produce crack velocity. The crack velocity plot is separated into two 24-h RH cycles, representing 204 h to 252 h (Figure 6). Each cycle showed a peak in the crack velocity representative of the increase in crack rate during the dry off portion of the RH cycle. The peaks increased in magnitude with each subsequent RH cycle and increasing crack depth. This increase most likely occurred because the stress intensity at the crack tip increased as the crack propagated. Lower crack velocity values were associated with the high humidity portion of the RH cycle. Slightly negative crack velocity values are an artifact of temperature variation effects on the sensor instrumentation.

Minimal crack growth occurred during the ramp-up portion of the RH cycle. During the transition from wetting to drying, all cycles showed an increase in crack velocity that occurred between 76% and 78% RH, similar to AA5083-H116 measurement results. Crack velocities continued to increase to a maximum crack velocity around 65% RH. Each cycle’s crack velocity then decreased as the RH decreased to 40%. It should be noted that the crack velocities remained above zero during the entirety of the cycle, including at low RH equal to or slightly below the efflorescence relative humidity of sodium chloride.

There was no significant crack depth increase during 330 h of B117 salt fog exposure (Figure 7). At 330 h, a power interruption disabled the salt fog and the RH dropped accordingly, but with no appreciable change in cracking response. However, the crack depth (and velocity) increased dramatically after the salt fog was restarted and shut down a second time, causing the chamber RH to decrease below 80% (~410 h). The cracking continued until final fracture occurred even as the RH decreased and was held constant at 35% for 15 h.

Minimal crack depth increase was also observed during the replicate B117 salt fog experiment. Again after salt fog ceased and the RH was decreased, crack depth increased dramatically and the sample fractured after 120 h of testing (Figure 8). B117 salt fog was initiated just prior to final fracture to determine if rewetting the sensor would arrest crack growth, visualized by the sharp rise in humidity at 119 h. By the time the humidity ramped up, the sample had already entered into ductile overload and was not able to be arrested.

This first ASTM G85 A5 test cycle comprised a 1 h fog at room temperature and a 1 h dry off at 35°C, repeated for the duration of the study (Figure 9). During the first 13 h of the first ASTM G85 A5 test, minimal change in crack growth was observed. The cyclic response during this period was correlated with the change in chamber air temperature. The temperature compensation algorithm’s effectiveness was limited as a result of the rapidly changing cycles relative to the long thermal time constant of the sensor assembly.

After 13 h of testing, the crack depth began to increase at a decreasing rate until a slight inflection point was observed at approximately 37 h. After 37 h of testing, crack depth continued to increase but at an accelerating pace. After 51 h of testing, four small step changes of 0.06 mm to 0.16 mm were observed prior to final fracture occurring after 54 h of testing. The small step changes prior to final fracture are assumed to be representative of ductile overload of remaining AA5083-H116 ligaments.

As mentioned, the temperature compensation strategy implemented in the sensor’s data acquisition for the first ASTM G85 experiment is only capable of correcting for longer-term transients, typical of natural outdoor atmospheric temperature changes. Extremely quick transients or step changes in temperature that occurred during the 1 h drying and fogging cycles resulted in a peak in the crack depth data, convoluting crack velocity values calculated as the derivative of the crack depth with respect to time. Therefore, longer 2 h holds were used in the replicate experiment and both the dry off and fog portions of the modified ASTM G85 A5 test were performed at 35°C. Despite changes to the ASTM G85 A5 test, the accelerated corrosion testing chamber’s temperature varied several degrees when switching between dry off and fog portions of the test cycle, resulting in a cyclical, systematic response clearly evident in the first 54 h of testing (Figure 10). After 54.5 h, the crack depth began steadily increasing to 0.57 mm at an average rate of 19.6 nm/s. After 62.6 h of testing, a sharp increase of 0.51 mm was observed and followed by a flatter region, representative of lower crack velocities. This cycle of a steep increase in crack depth followed by a flatter region occurred two more times, beginning after 63.9 h and 66.2 h of testing, and correlates to an increase in crack depth of 0.79 mm and 0.86 mm, respectively. The small step changes prior to final fracture are assumed to be representative of ductile overload of remaining AA5083-H116 ligaments. After 69 h of testing and only approximately 15 h of cracking, the AA5083-H116 V-notched sample fully fractured.

Crack velocity values were calculated by differentiating crack depth with respect to time. The resulting crack velocity was plotted as a function of relative humidity to study the dependence of crack velocity on cyclic humidity (Figure 11). Three (4-h) cycles are shown, representing 12 h of data beginning after 57 h of testing. The negative crack velocity values and peaks at 100% RH are again a result of temperature variation occurring when switching between dry off and fog portions of the test cycle. Minimal cracking occurred during the wetting cycle as RH was increasing. During the drying cycle, crack velocity increased to a single maximum that shifted from 31% to 37% to 48% RH as crack depth increased. The highest crack velocity of 610 nm/s occurred during the drying portion of the final RH cycle at 48%. The peaks increased in magnitude with each subsequent RH cycle and increasing crack depth. This increase most likely occurred because the stress intensity at the crack tip increased as the crack propagated (Figure 12).¹⁷  For all three cycles, crack velocity was halted once RH dropped below 22% RH.

During the initial 24 h of the first GM 9540P test, minimal change in crack growth was observed (Figure 13). Crack depth deviations observed during this time were assumed to be related to temperature change effects on the sensor. After 24 h of testing, a large step change correlating to 0.48 mm of crack growth was observed during the transition from drying to wetting in the test cycle. This result was not consistent with previously observed cracking during the drying cycle. Minimal change in crack growth was observed during sustained wetting and drying portions of the test cycle. This trend of minimal crack growth followed by a large step change in crack growth (ranging from 0.35 mm to 1.10 mm) during the transition from drying to wetting was observed on all eight 24-h GM 9540P test cycles until final fracture occurred after 192 h of testing. During the transition from drying to wetting, crack velocity increased to a maximum value at very low RH, approximately 30% to 40%. All cycles demonstrated that cracking was predominately shut down at 50% RH and higher.

A modified GM 9540P test was conducted next, where temperature set-point was constant at 35°C, although chamber temperatures still fluctuated by 1°C to 2°C during RH transients. Temperature was maintained to isolate crack response as a function of RH. During the first 94 h of testing, minimal change in crack growth was observed. Slight deviations observed during this time were related to the inherent temperature changes as the sample was wetted and dried. After 94 h of testing, a pattern of cracking was observed for each 24 h cycle of the GM 9540P iso-temperature test (Figure 14). This pattern remained consistent until final fracture occurred after 384 h of testing. During the transitions from wetting to drying, a large increase in crack growth (up to 0.56 mm) was observed. Crack growth rates decreased as the drying period continued before a small step change increase in crack depth (~0.08 mm) occurred during the transition from drying to wetting. Cracking was then observed to rapidly arrest when transitioning to periods of high RH.

The resulting crack velocity was plotted as a function of relative humidity to study the dependence of crack velocity on cyclic humidity (Figure 15). Four 24-h cycles are shown, representing test hour 234 to test hour 331 of the previously discussed GM 9540P iso-temperature testing. Test cycles were labeled in chronological order such that “Cycle 1” and “Cycle 4” correlate to the first and last 24 h of the test window, respectively. Negative crack velocity values are assumed to be a result of temperature variation effects. Because the test was run under ambient temperature conditions, some variation occurred as a result of HVAC system switching and test cycle period switching (e.g., wetting to drying).

Crack velocity values increased during the transition from wetting to drying, and reached a maximum value at 61% to 66% RH. The magnitude of the crack velocity peaks increased as the fracture sample’s crack depth increased, as observed with the other tests. Cracking rates then decreased as the drying cycle progressed. Because the drying cycle was also run at ambient temperatures, RH was only controlled to mid-50% levels. No increase in crack rate was observed during the transition from drying to wetting until > 90% RH, when a small step change in crack growth occurred. Cracking was then quickly arrested during the remaining wetting period.

Great progress in the understanding of different aspects of aluminum EAC, particularly for sensitized 5xxx series alloys under immersion conditions, has been made in the last several years. In order to push the state-of-the-art forward, the EAC sensor developed and demonstrated in this work is well suited to characterization of high-strength aluminum in a range of corrosion environments (including a range of applied potentials if desired¹¹ ), applied stress intensities, and sensitization levels, and for fast and short crack growth speeds of in-plane cracking. It is worthwhile to summarize the state-of-the-art in EAC of aluminum alloys and generally compare previously reported results with measurements from this study, as well as to highlight areas for advancement.

The Al-Mg alloys (5xxx series) are non-heat-treatable and processing steps such as alloying and work hardening provide requisite strength (through solid solution strengthening and cold working, respectively).^18-21  To control grain size, manganese, chromium, and titanium are added and specific combinations of cold work and annealing are applied to retard recrystallization and decrease recrystallized grain size.^18,20,22  At annealing temperatures, aluminum has a high solubility for magnesium (14.9 wt% at 450°C). At room temperature the solubility decreases substantially to 1.7 wt%.²⁰  In a quenched 5xxx alloy with greater than 3.0 wt% Mg having a super-saturated solid solution α matrix, exposure to temperatures as low as 50°C for long periods of time causes Mg diffusion to low energy sites such as grain boundaries.²³ 

The β phase (Al₃Mg₂) eventually precipitates at grain boundaries, leading to sensitization and the corresponding increased susceptibility to intergranular attack.²¹  The extent of precipitation of the β phase is dependent upon composition, grain boundary characteristics, temperature, and exposure time (with higher temperature facilitating more rapid precipitation).^24-25  Conflicting results regarding the importance of low-angle grain boundaries (less than 15°) to β precipitation exist, but it is agreed that β formation is favored along high-angle grain boundaries with precipitate area increasing with misorientation angle.^24-25  The DoS is directly coupled to β precipitation and is likewise influenced by alloying, exposure temperature, and time at temperature. DoS is conventionally quantified by NAMLT per ASTM G67.¹²  High DoS has been correlated with approximately 50% coverage of grain boundaries, whereas low DoS has been metallurgically associated with only discrete β.^24,26-27  It has recently been shown that continuous grain boundary β phase does not exist even at high DoS; rather, the shape, number, spacing, and size of the β phase change with DoS.²⁸  Current studies have brought into question the viability of using DoS for predicting resistance to intergranular corrosion (IGC) by showing that materials with the same DoS exhibited different IGC morphology and IGC propagation rates in the same chloride environment.^29-30  A phosphoric acid etch is used as a qualitative assessment of IGC susceptibility for alloys of intermediate DoS and is described by Lim, et al.²² 

The β precipitates are anodic to the bulk α matrix in near-neutral chloride salt solutions, creating a susceptible IGC path.²¹  At the alloy open-circuit potential, anodic polarization of the β phase leads to its preferential dissolution along grain boundaries and spreading of IGC.^26-27  The dissolution of the β phase generates metal cations that hydrolyze water causing local acidification and high chloride concentration within occluded regions such as pits or IGC cracks. The change in chemistry may decrease the breakdown potential and can increase the β dissolution rate.²⁶  The development of a critical local environment favors steady state IGC growth. Changes in this environment may stifle or even arrest the process.²²  Cathodic reduction of hydrogen ions in the low pH solution leads to hydrogen uptake within the metal.³¹  Highly sensitized AA5083-H131 (DoS of 49 mg/cm²) was found to exhibit maximum IGC rates of up to approximately 6 nm/s (within the measurement resolution demonstrated by the EAC sensor) by Lim, et al.,²²  in the L orientation at approximately −0.8 V_SCE in 0.6 M NaCl (pH 8.3). The interested reader is directed to recent literature reviews on sensitization and IGC of 5xxx alloys compiled by Lim, et al.,²²  and Zhang, et al.³² 

Intergranular stress corrosion cracking (IGSCC) is also known to occur in sensitized 5xxx alloys and has been studied extensively, primarily in immersion conditions. The proposed mechanism for IGSCC in sensitized Al-Mg is related to selective grain boundary β dissolution described earlier leading to boundary embrittlement by atomic hydrogen.^33-39  The effect of hydrogen in EAC of Al-Zn-Mg-Cu (7xxx series) alloys has also been broadly investigated and accepted in various environments.^8,40-44  Many studies of IGSCC in 5xxx alloys have focused on high-temperature exposure sensitization (above 150°C) where sensitization occurs rapidly by formation of near-continuous β films at grain boundaries. Sensitization at lower temperatures and long times causes discrete β precipitation.^28,45  Hydrogen environment assisted cracking (HEAC) is a likely mechanism for IGSCC between separated β particles as IGC resulting from anodic dissolution of β particles is insufficient to explain observed EAC rates in 5xxx alloys.^33-35 

Crane, et al., investigated the effect of low-temperature sensitization on the kinetics of IGSCC propagation in AA5083 in the S-L orientation.³¹  Crane observed that IGC growth occurs, at rates consistent with bold surface IGC fissure penetration without stress, at an occluded fatigue precrack tip under low stress intensity in sensitized AA5083-H131 immersed in NaCl solution. Low-temperature sensitization was reported to result in severe IGSCC above a critical DoS level of 9 mg/cm² to 12 mg/cm². The threshold stress intensity and Stage II IGSCC rate were correlated with mass loss for sensitization at 60°C to 100°C. The dependence of IGSCC on sensitization in AA5083-H131 was reported by Crane to be consistent with the coupled mechanism of β and matrix dissolution leading to hydrogen production and intra-β/α boundary hydrogen embrittlement. IGSCC was observed to occur above a critical DoS when sufficient crack acidification was established for hydrogen production. Very rapid crack growth (10³ nm/s to 10⁴ nm/s at applied K of 15 MPa√m and DoS of 50 mg/cm² measured using DCPD methods in immersion condition, similar to measurements and applied conditions using the EAC sensor) was speculated to be a result of increased hydrogen diffusivity related to a dislocation-trap-free zone between grain boundary β precipitates.³¹  In a study of inhibition of stress corrosion cracking (SCC) of sensitized AA5083, Seong, et al., suggested that the mechanism of inhibition of IGSCC was mainly a result of control of the cathodic reaction (hydrogen uptake and embrittlement) and not anodic dissolution.⁴⁶  This conclusion was reached after observation that polarization of sensitized AA5083 above the breakdown potential of β aggravated IGSCC (using slow strain rate testing in NaCl immersion) even in the presence of an inhibitor and polarizing below the β breakdown potential in the absence of inhibitor arrested IGSCC. The sensor developed in this work would be highly applicable to extending these qualitative conclusions by quantifying IGSCC rates. In other work, Crane, et al., shed further light on the importance of hydrogen embrittlement of 5xxx series alloys as related to critical crack solutions.⁴⁷  Cracking and occluded crack electrochemistry measurements confirmed that cation production from dissolution leading to crack to acidification and hydrogen uptake is necessary for embrittlement of α segments between β particles. When dissolution was eliminated by cathodic polarization or minimized by low DoS, IGSCC was mitigated. The authors also found that β-free AA5083-H131 is susceptible to IGSCC when the crack was acidified by anodic polarization or alteration of the bulk environment composition.⁴⁷  The crack sensor developed in this work is able to accommodate crack growth rate measurements on the same order of magnitude at similar DoS and applied stress intensities and is well suited to advancement of the understanding of AA5083 IGSCC. In particular, as the majority of work has been conducted in immersion conditions, the state-of-the-art can be advanced using the described EAC sensor in relevant atmospheric environments. Additionally, coated fracture specimens placed in the sensor could be used to accurately quantify inhibitor effects or sacrificial coatings on IGSCC growth rates in representative test conditions.

Other authors have conducted EAC studies of sensitized 5xxx series alloys using double-cantilever beam (DCB) specimens with intermittent salt solution applied to the crack tip using a dropper.^6,48-49  This effectively creates a cyclic wet-dry condition. Crack velocities are recorded manually throughout testing. Bovard measured crack growth in AA5083 sensitized to 44 mg/cm² of the order of 1,000 nm/s with applied stress intensities (K_I) ranging from 10 MPa√m to 18 MPa√m. At a stress intensity of 8 MPa√m, crack velocities were an order of magnitude lower at ~100 nm/s.⁶  Cormack observed EAC velocities in AA5456-H116 sensitized to approximately 60 mg/cm² on the order of 3 nm/s to 30 nm/s for applied K_I of 10 MPa√m to 14 MPa√m.⁴⁸  Gao recorded crack growth rates in AA5083 DCB specimens using ultrasonic techniques ranging from 300 nm/s to 700 nm/s in 1 M NaCl at an applied K of ~11 MPa√m.⁴⁹  The demonstrated EAC sensor is able to resolve EAC rates on the same orders of magnitude and can apply the necessary stress intensity ranges, for similar DoS levels. Going beyond the reported testing, the EAC sensor can advance future studies by supporting in situ measurements of crack velocity in dynamic atmospheric tests rather than simple, and possibly nonrepresentative immersion/drying tests.

High-strength aluminum alloys such as AA7075 in the peak-aged temper (T6) are known to be susceptible to SCC.^50-51  AA7075-T6 exhibits accelerated cracking in immersion in NaCl solutions and distilled water. Cracking is also observed in conditions of high water vapor and in inert environments following pre-exposure to environments containing water.^52-54  Because cracking can occur without an aqueous phase present to solvate ions and in inert environment after exposure to water (with pre-exposure effect diminishing at a rate that corresponds to desorption of hydrogen), it is presumed that the alloy absorbs hydrogen when exposed to water and hydrogen is a likely cause of cracking.^52,55-58  It still remains to be resolved if this is the responsible mechanism for SCC in ambient aqueous solutions or in thin film electrolytes.

Although further work is required in thin film electrolytes, immersion testing has shed light on 7xxx EAC processes. Cooper, et al., conducted targeted SCC studies of AA7050 under immersion conditions to carefully observe the effects of crack potential and chemistry on SCC rates.^7,59  The crack tip environment in 7xxx alloys can vary significantly from the bulk solution. Aluminum ion production by anodic dissolution leads to acidification of the occluded solution within the crack by hydrolysis in the same fashion as for 5xxx described earlier.⁷  Chloride ingress from the bulk solution proceeds to balance the positive charge buildup from increased aluminum cation concentrations. The result is a low pH, high chloride solution within the crack with restricted mass transport assisting in maintaining the severe chemistry.⁷  Cooper observed that when the potential within a crack decreased below that of the externally applied potential, dissolution within the crack was favored and chloride concentration increased along with an eventual decrease in pH. This change corresponded with a high rate of SCC crack growth. The authors also injected a simulated solution into the crack that was representative of the aggressive low pH, high chloride solution and again observed a rapid increase in SCC growth.⁷  These conditions are conducive to hydrogen embrittlement and analysis of crack wakes has revealed hydrogen uptake during aqueous crack growth testing in similar conditions.^60-61 

Additional quantitative EAC studies with AA7075-T651 have been conducted using both immersion and humid atmosphere exposures with crack velocity measurements that are resolvable by the demonstrated EAC sensor in this work (10 nm/s to 13 nm/s) at similar stress intensities. Nguyen, et al., conducted immersion tests in NaCl solutions using DCB specimens and measured crack velocities ranging from 10 nm/s to 20 nm/s. The authors did not quantify the applied K. They measured alterations in crack solution chemistry and found that local changes in concentration of dissolved metal ions and pH within the occluded crack contributed to EAC. Nguyen concluded that anodic dissolution was the primary contributory mechanism.⁶²  Shastry, et al., using DCB specimens, measured crack velocity of AA7075-T651 to be 7 nm/s to 10 nm/s at an applied K of about 5.5 MPa√m, under immersion conditions. Their study was focused on the metallurgical aspects of AA7075 and found that increased grain boundary solute concentration of Mg, Zn, and Cu resulting from changes in solution heat treatment were likely responsible for increased EAC susceptibility.⁶³  Cooper and Kelly measured EAC velocities ranging from 10 nm/s to 90 nm/s in under-aged material of AA7050 and roughly 4 nm/s in peak-aged material using wedge-open loaded specimens, in immersion conditions. Applied stress intensities were reported as 14 MPa√m to 15 MPa√m at the start of the test and 9 MPa√m to 12 MPa√m at the conclusion. Cooper and Kelly concluded that anodic dissolution contributed significantly to Stage II crack growth in aqueous chloride solution by generating a critical crack tip solution chemistry, but its effect on EAC was of the same order of magnitude as hydrogen embrittlement.⁵⁹  Scully reported plateau EAC crack velocity for AA7075-T651 at 15 MPa√m to 20 MPa√m in humid air environments to be approximately 10 nm/s.⁴² 

The increased cracking rates upon drying steps measured during this work may be a result of increased electrochemical activity during drying. High corrosion rates in metals during transients from wet to dry conditions under atmospheric exposure have long been known.⁶⁴  This observation has been rationalized on the basis of accelerated oxygen reduction, as the electrolyte thickness is reduced in the drying phase. Mansfeld and Kenkel found maximum corrosion current in a galvanic couple under atmospheric conditions to occur at a point just before complete drying, again attributed to increased oxygen reduction with decreasing solution layer thickness.⁶⁵  Development of critical chloride concentrations leading to onset of pitting has also been attributed to the wet to dry transient.⁶⁶  It is therefore feasible that anodic dissolution and corresponding cathodic reactions (namely hydrogen generation and uptake leading to hydrogen embrittlement)⁸  are accelerated under the thin film developed during drying and may be responsible for the observed increase in EAC rates for AA5083 and AA7075 in this work. The occluded geometry of the crack and capillary effects may also allow electrolyte to exist at RH below where the rest of the boldly exposed surface has fully dried.

• Testing of the atmospheric environmentally assisted cracking sensor under various accelerated environmental conditions has demonstrated that it is effective for observing small scale crack propagation with sufficient resolution (1.0 nm/s) to obtain several velocity measurements across multiple cycles.

• Test results of AA5083-H116 under cyclic RH, ASTM G85 A5, and GM 9540P tests have shown strong dependence of crack velocity on cyclic humidity. Consistent results were obtained for AA7075-T651 under cyclic RH testing, demonstrating that the phenomenon is not limited to sensitized alloys.

• The most severe environmental conditions observed occurred during the drying cycle, and highest crack rates were observed at moderate to low RH. Assuming an electrochemical-based mechanism for crack growth, this is consistent with observations of increased corrosion severity during wet to dry transients in atmospheric conditions reported in the literature.

This material is based upon work supported by the SERDP/ESTCP program via contract W912HQ-09-C-0042. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of SERDP. The authors thank team members on this effort including James Dante, Manager of the Environmental Materials Performance Division at the Southwest Research Institute, Francine Bovard, Technical Specialist, and James Moran, Senior Staff Engineer, at Alcoa. Discussions with Robert Kelly, Professor of Engineering at the University of Virginia, are also gratefully acknowledged.