In this work, acoustic emission (AE) monitoring was correlated with in-situ optical microscopy observation and electron backscatter diffraction (EBSD) measurements to investigate the evolution of a single stress corrosion crack in SUS420J2 stainless steel subjected to chloride droplet corrosion. A single dominant crack evolution was observed to transition from a slow initiation of active path corrosion-dominant cracking to a rapid propagation of hydrogen-assisted cracking. The initiation-to-propagation was concomitant with a significant increase in the number of AE events. In addition, a cluster analysis of the AE features including traditional waveform parameters and fast Fourier transform (FFT)-derived frequency components was performed using k-means algorithms. Two AE clusters with different frequency levels were extracted. Correlated with the EBSD-derived kernel average misorientation (KAM) map of crack path, low-frequency AE cluster was found to correspond with the location of plastic deformation in the propagation region. High-frequency AE cluster is supposed to be from the cracking process. The correlation between AE feature and SCC progression is expected to provide an AE signals-based in-situ insight into the SCC monitoring.

Fig. 5 The AE activity of amplitude and cumulative events and the profile of pit size plus dominant crack length over the time evolution. Here, marked points “a–i” are consistent with the sequences indicated in Fig. 4.

Stress Corrosion Cracking (SCC), caused by the synergistic interaction of corrosive environment and mechanical load, is an important threat to stainless steels. Considering the widespread use of stainless steels, SCC has been a primary cause of many failures of metallic structures and components in a wide range of industries.¹⁾ Thus, the reliable assessment of the service life of the exposed components and structures requires the accurate prediction of initiation and propagation of SCC.

It is clear that SCC requires contributions from both chemical and mechanical factors. Parkins²⁾ firstly put forward the concept of “stress corrosion spectrum” to support a continuous spectrum of SCC mechanism. It is suggested that SCC is predominantly controlled by a changing dominant factor from electrochemical to mechanical depending on the SCC progression. Figure 1 shows a schematic elucidation of a three-stage model for SCC progression which was rearranged based on the references.^2–4) Despite its complex nature due to combined metallurgical, mechanical and environmental effects, SCC in most cases begins with the early localized corrosion⁴⁾ or mechanical defect sites like pitting, intergranular attack (IGA), scratches, or other pre-existing flaws, subsequently develops as a short crack with slow propagation, to finally reach the long crack regime eventually leading to the ultimate failure of the component. The SCC development in the initiation regime is often perceived to be slow and long term and can largely determine the overall life of an exposed component.^5,6) Therefore, an in-situ monitoring of the initiation-to-propagation transition is greatly significant in the lifetime estimation of SCC.

Fig. 1

Schematic diagram of a three-stage model for SCC progression. Rearranged based on Refs. 2–4).

In general, the initiation and propagation of SCC are strongly affected by the electrochemical conditions and the local stress/strain state at the crack tip, and cannot be directly and integrally estimated via traditional electrochemical methods⁷⁾ or mathematical based models⁸⁾ given the difficulties to obtain the experimental input parameters. Acoustic emission (AE) method, a nondestructive testing (NDT) method based on the recording of elastic waves, is known to be highly sensitive to local phenomena such as chemical reactions and physical changes. AE method has long been used to study the steel corrosion like pitting corrosion and SCC.^9–18) Some of the registered sources of AE signals in literature include pitting,^9–13) hydrogen bubble evolution,^9–14) dislocations motion during plastic deformation,¹⁵⁾ cracking of oxide/corrosion products,^14,16) falling-off of surface grain,^14,16) generation and propagation of cracks.^13,15–18) However, so far no attention was paid to combining in-situ observation and AE monitoring for the transition identification during SCC progression, which is the primary concern of the present work.

However, the accelerating SCC tests, recommended by ASTM G129,¹⁹⁾ ASTM G36,²⁰⁾ or NACE TM0177,²¹⁾ basically require the immersion of specimens into a bulk solution at an elevated temperature which is challenging to facilitate AE experiments and in-situ observations. In the present work, a chloride droplet SCC test was improved with the smooth U-bent SUS420J2 stainless steel specimen in humidity to trigger a single crack evolution, thus accommodating the experimental aim. Moreover, electron backscatter diffraction (EBSD) analysis of the crack path was correlated with in-situ observation and AE behavior in time evolution so as to track the features in relation to the initiation and propagation of SCC.

A commercial SUS420J2 stainless steel (provided by Kusayama Unique Special Steel, Japan, C: 0.32 mass%, Si: 0.45 mass%, Mn: 0.54 mass%, Cr: 13.39 mass%, Ni: 0.09 mass%, P: 0.031 mass%, S: 0.005 mass%, Fe: bal.) was used in this study. To improve the corrosion susceptibility, the steel was first treated at 600°C for 1 h and 800°C for 1 h to precipitate the carbides. It was then annealed at 1010°C for 2.75 h and quenched in Argon air cooling. The obtained mechanical properties of the steel were characterized by a 0.2% proof stress of 1201 MPa, an ultimate tensile strength (UTS) of 1679 MPa, an elongation of 0.9% and a Vickers hardness of 611. The specimen was machined from the bulk steel after heat treatment with a dimension of 170.0 mm (L) × 6.3 mm (W) × 1.5 mm (T) and finally polished up to 0.05 µm Al₂O₃ slurry for a mirror finish. Figure 2 shows the microstructure of the steel specimen after being etched using Viella’s reagent for 10 seconds, indicating the martensitic lath with visible prior-austenite grain boundaries. The average gain size was determined as 22.8 µm. Circular and elongated undissolved carbides were observed within martensitic grains, especially near prior-austenite grain boundaries. Further energy-dispersive X-ray spectroscopy (EDX) mapping confirmed the carbides with enrichment in carbon and chromium and depletion in iron, indicating Cr-rich carbides. The specimen was intended to have low toughness, high residual stress and active corrosion path (i.e., grain-boundaries with segregated Cr-rich carbides) to facilitate the crack initiation.

Fig. 2

The microstructure of chemically etched specimen: (a) optical micrograph, (b) backscattered electrons (BSE) SEM image, and (c) EDX maps.

To improve the reliability in extracting and interpreting the AE signals, a localized SCC attack under a single chloride droplet accommodated with a continuous in-situ monitoring by digital microscope was designed with AE measurements. Figure 3 shows the experimental setup. The specimen was bent and fixed in an experimental jig with a curvature radius of 125 mm leading to a constant strain of 0.6% at the apex of the specimen surface. A 1 µL droplet of 1.0 mass% neutral NaCl solution was dropped in the center area of the specimen surface using a microliter syringe. The experimental apparatus was placed in a thermostatic bath (298 K, 99–100% humidity) with pure water. The bath was covered with a transparent glass cap to allow in-situ observation using a digital microscope (VHX-5000, Keyence Corp., Japan). Simultaneously, continuous AE measurement was performed using a custom AE acquisition system developed in our group, CWM.²²⁾ Two high-sensitivity R-CAST sensor systems with a resonant frequency of 200 kHz (M204A, Fuji Ceramics Corp., Japan, 55 dB gain) were positioned 10 mm and 30 mm away from the droplet site to record the AE signals. A threshold of 27 dB and a 100 kHz high-pass filter (HPF) were used as thresholds to isolate AE events from the continuous waveform. Then the target signals were extracted by filtering out the noise outside the corrosion region using the source location method.

Fig. 3

Schematic of the modified SCC test system incorporated with an in-situ digital microscope and AE measurement systems.

A cluster analysis of the extracted AE data was performed using K-means algorithm. To improve the classification of AE signals, the traditional AE parameters (i.e., amplitude, rise time, duration, counts…) together with the FFT frequency components from 102.5 kHz to 996.1 kHz with a resolution of 4.9 kHz were input as AE features into the k-means algorithm. All input data were first normalized with Z-score standardization method.²³⁾ Then principle component analysis (PCA) was conducted to reduce the dimension of the obtained data matrix before the K-means clustering algorithm was applied. The optimum number k of clustering was determined using average silhouette coefficient (SC_ave.).²⁴⁾ The silhouette coefficient for the i-th point, s(i), is defined as below:   

The SCC test was interrupted immediately after the crack growth shifted from a slow crack initiation to rapid propagation to prevent complete failure of the specimen and allow EBSD measurements along the crack path. After testing, the specimen was cut to an approximate length of 20 mm to fit the SEM observation chamber and ultrasonically cleaned with ethanol at room temperature to remove the corrosion products. Next, a fine polishing of the crack area was performed with a colloidal silica suspension of 0.05 µm for 2 hours in order to enable EBSD measurement. EBSD data were collected using a field-emission SEM (JSM7000F, JEOL Ltd., Japan) equipped with a Tex-SEM laboratory detector. A 15 kV acceleration voltage was selected with a step size of 0.15 µm for EBSD scanning. EBSD data were analyzed using OIM (EDAX-TSL) v7 software. The Kernel Average Misorientation (KAM) analysis was used to assess the local plastic deformation around the crack path.²⁵⁾

Figure 4 shows the corrosion evolution of the SUS420J2 stainless steel under the single NaCl droplet. In details, it firstly started with pitting corrosion as captured in Fig. 4(a)–(b). When pit grew to a critical size, as shown in Fig. 4(c), tiny cracks initiated from both sides of the corrosion pit. Subsequently, a characteristic feature of corrosion “atoll” was gradually visualized as the formation of an annular, ring-like surface stains under the droplet. Meanwhile, as shown in Fig. 4(d)–(f), the tiny crack developed to form a single dominant crack, normal to the U-bend load direction, in the central region of the corrosion atoll. Some corrosion microcracks were also observed to be concurrent with the growth of the main crack and were supposed to be IGA. Figure 4(g) shows the first capture of small gas bubbles around the pit mouth that were believed to be hydrogen bubbles in comparison to the experimental conditions for other bubbles.²⁶⁾ This is an important point as shortly after, the dominant crack shifted toward a rapid propagation with increasingly amount of hydrogen bubbling at the mouths of pit, crack and IGA fissures.

Fig. 4

In-situ observation of the crack initiation and propagation of SUS420J2 under the droplet of 1 µL neutral 1% NaCl solution in humidity: (a)–(b) pitting corrosion; (c)–(f) Slow, single dominant crack initiation from a corrosion pit and IGA; (g)–(i) rapid crack propagation.

Figure 5 shows the AE data extracted during the corrosion process with the surface profile of pit size plus dominant crack length over time. Correlated with in-situ observations, three distinct regions were recognized, i.e., stage I of pitting corrosion without AE events recorded, stage II of slow crack growth with moderate AE activity, and stage III of rapid crack propagation with massive AE signals. Partial enlarged details in relation to the transition of pit-to-crack and slow-to-rapid crack growth is shown in Fig. 6. In stage I, corrosion pit was first observed in-situ with a size of approximately 4 µm and grew to a critical size of approximately 32 µm with no AE events detected. Stage II started with a pit-to-crack transition. AE signals were not detected immediately after the crack initiation but after a short time delay which corresponded to an AE-based NDT limit of approximately 60 µm relevant to the present material and environment combination. In stage I, average pit growth rate (PGR) was approximated to be 0.03 µm/s. In stage II, the crack grew slowly with an average crack growth rate (CGR) of 0.04 µm/s and the number of AE events increased moderately. In the late stage II, the hydrogen bubble evolution was observed. After 31 seconds, crack growth shifted to a high CGR of 1.68 µm/s and intense AE activity began to be recorded massively, indicating the start of stage III.

Fig. 5

The AE activity of amplitude and cumulative events and the profile of pit size plus dominant crack length over the time evolution. Here, marked points “a–i” are consistent with the sequences indicated in Fig. 4.

Fig. 6

Partial enlarged details of Fig. 5 to highlight the transitions of pit-to-crack (a) and slow-to-rapid crack growth (b).

Figure 7 shows the detailed morphology of the corrosion with EBSD scanning including inverse pole figure (IPF) and KAM maps of the dominant crack path. First, different regions corresponding to stage I, II, and III can be distinguished. A large number of IGA-induced fissures along the prior-austenite grain boundaries were clearly observed in the central region of corrosion atoll. The corrosion pit seems to initiate from the prior-austenite grain boundaries and many undissolved carbides are precipitated nearby. Second, the IPF map shown in Fig. 7(b) clearly indicated the intergranular characteristics of the crack development. It is likely that the IGA fissures played in facilitating the crack growth. Moreover, from the KAM map shown in Fig. 7(c), some zones of high KAM intensity situated around the beginning area of stage III can be recognized, indicating local plastic deformation along the specific region of crack path.

Fig. 7

(a) The corrosion morphology including corrosion pit, IGA and the single dominant crack being segregated by different regions correlated with in-situ observations; (b) IPF map, (c) KAM map of the crack path.

As shown in Figs. 4 and 7, the dominant crack evolved simultaneously with the development of IGA. In order to validate whether the IGA contributed to the detected AE signals or not, a comparative experiment on an unstressed flat specimen of the same preparation under the same exposure condition, i.e., being exposed to the same droplet of 1 µL neutral 1% NaCl solution and the same humidity, was carried out. The test was conducted for 15 hours but no AE event was detected at all. Figure 8 shows the post observation of the corrosion morphology. Localized corrosion of pitting and IGA were visualized at the central region of a similar corrosion atoll. Undissolved carbides are clearly distributed in the neighboring regions especially at the pit mouth and the grain boundaries. However, no dominant crack was initiated in addition to the IGA fissures and pitting. Implicitly, the pitting and the IGA can be ruled out from the AE sources. The AE signals during SCC progression are supposed to be from the evolution of the single dominant crack.

Fig. 8

(a) The central region including pitting and IGA of the corrosion atoll from a comparative experiment of an unstressed specimen being exposed to the same droplet and humidity until 15 hours with no AE event recorded; (b) Partial enlarged detail.

Figure 9 shows the result of the cluster analysis. From the SC_ave. variation against the number of clusters k, 2 clusters were selected as the best clustering solution. Figure 9(b) shows a frequency comparison near the center of cluster 1 and 2, indicating the distinct feature in frequency level, i.e., high-frequency for cluster 1 and low-frequency for cluster 2. The AE waveforms of two clusters were also compared as shown in Fig. 10. Cluster 1 signal exhibits burst-type waveform with high frequency component, while cluster 2 has continuous-type feature with lower level in the dominant frequency component.

Fig. 9

The result of AE cluster analysis: (a) The SC_ave. upon k number; (b) Comparison of the centers of cluster 1 and 2 based on the scheme of k = 2.

Fig. 10

The AE waveforms (a)–(b) and corresponding FFT spectrums (c)–(d) of cluster 1 and 2.

Figure 11 shows the evolution of the two AE clusters in stage III with the corresponding optical microscopy observation. Owing to a correlative analysis with KAM map, the start of low-frequency cluster 2 was interestingly corresponding to the area of high local strain recognized by high KAM intensity. Taking into consideration the continuous waveform-type, low-frequency cluster 2 seems to be correlated with the plastic deformation along the crack path. Hence, cluster 1 is supposed to be with the cracking process.

Fig. 11

The evolution of low-frequency cluster 2 correlated with the transition area of high local strain recognized by KAM map of the crack path.

SCC is a complex process dictated by various dynamic mechanisms depending on the combination of material and environmental variables. A unified model responsible for SCC evolution is still far from being realized. So far, the most accepted mechanisms in literature^27,28) for SCC can be divided into two main categories. The first one is anodic dissolution model, like active path corrosion (APC) and film rupture models, which attributes the crack growth to anodic dissolution at the crack tip. The effect of mechanical stress is probably to expose the bare metal by rupturing the protective passive film or deposited corrosion products. The second model considers SCC as a mechanical process accelerated by the specific chemisorption like hydrogen that can either reduce the fracture surface energy or induce dislocation emission at the crack tip. In this work, a combination of AE monitoring and in-situ observation is adopted to focus on one single stress corrosion crack evolution in terms of droplet corrosion. The process of SCC is discussed in details below based on in-situ observations.

In the early stage, the undissolved Cr-rich carbides are believed to be the precursor sites of localized corrosion, as shown in Figs. 4, 7, and 8, in relation to the Cr-depleted zones.²⁹⁾ After localized corrosion formed, a corrosion atoll under the droplet, whether stressed or not, becomes visible as shown in Fig. 4 and 8. This is due to the formation of local cell action, i.e., the center of the atoll serves as the anodic area and the perimeter as the cathodic area.^30,31) Anodic dissolution reactions at the centers are:   

Breakdown of passive film and dissolution of metals (Fe, Cr) by reaction (4) and (5) probably contribute to the localized corrosion of pitting and IGA as observed in Fig. 4(a)–(f). The oxygen reduction by reaction (6) occurred with the formation of annular corrosion products. On the other hand, once the corrosion atoll forms, an electric field will be set up and the potential gradient would concentrate Cl⁻ to the central anodic region and migrate out Na⁺ to the outside region, which was previously verified with elemental mapping using EDX analysis.³²⁾ This would render the central anodic region more prone to corrosion and therefore could explain formation of a single corrosion atoll under the single NaCl droplet. This improved SCC test is expected to ensure the reliability in extracting AE signals related to SCC.

The segregation of carbides in the prior-austenite grain boundaries (Fig. 2) provides the active path for IGA independent of the applied stress (Fig. 8). In the stressed specimen (Fig. 4), the dominant crack in stage II seems to grow from the IGA-induced fissures in the atoll center (Fig. 7) and the CGR is relatively low (0.04 µm/s). This is consistent with the active path corrosion (APC)-predominant cracking mechanism that is dependent on the sensitized microstructural features in much lower CGR than hydrogen-assisted cracking.^28,33)

With the corrosion enhanced, some dissolved metal ions can be hydrated by reactions (7) and (8).³¹⁾ Hydrolysis reactions decrease the local pH, which not only promote further metal dissolution but also cause the cathodic hydrogen evolution by reactions (9) as the depletion in oxygen in the crack tip. The atomic hydrogen adsorbed on to the steel surface, H_ad, can either recombine to form molecular hydrogen and release as gas bubble by reaction (10) or diffuse into the steel, H_ab, by eq. (11)^27,28) and assist in cracking.   

The diffusion of hydrogen into BCC steel is known to be extremely rapid even at room temperature.^27,28,33) This can explain that the crack growth shifted toward a rapid propagation (CGR of 1.68 µm/s) of stage III in only 31 seconds after the first capture of hydrogen bubbling (Figs. 4 and 6). The predominant model responsible for SCC in stage III is therefore supposed to be hydrogen-assisted cracking in relation to hydrogen embrittlement (HE) mechanism. From the high-local plastic strain along the crack path in stage III (Fig. 7), the role of hydrogen on the present case may play a role in promoting dislocation emission at the crack tip in relation with the hydrogen-enhance localized plasticity (HELP) mechanism.^34,35)

In a summary, SCC on the present corrosion system started with a chemical-predominant precursor evolution of pitting corrosion (stage I), subsequently developed with APC-predominant slow crack initiation from the pitting and IGA (stage II), thereafter transitioned to a hydrogen-assisted rapid crack propagation predominantly controlled by HE mechanism (stage III).

In the present corrosion system, the AE signals are well recorded with the crack initiation and propagation. However, no AE event is detected in the precursor evolution of pitting corrosion (stage I). In the author’s previous work,^10,11) steel pitting corrosion that was purely controlled by electrochemical method was indeed monitored with the concurrent AE signals. Additionally, the AE behavior was correlated with the morphology of corrosion pit.¹²⁾ Nevertheless, it must be noted that the AE detectability largely depends on experimental conditions, the pit growth rate, the morphology and dimension of pitting. In the electrochemically controlled pitting in acidic chloride solution,^10–12) pitting grows quickly and results into large pit. Even though, there exists a phenomenon of “delay time” for AE detection which corresponds to a critical pit size required for the acoustically activation of pitting corrosion. Different from the pitting in bulk solution, the pitting under neutral chloride droplet develops with the Cl⁻ diffusion and the reactions of local cell and therefore grows more moderately in rate and size than the electrochemically controlled pitting corrosion. This may account for the difficulty in early pitting detection in droplet corrosion with AE signals.

After pit-to-crack transition (stage II), the AE signals are not detected immediately until a certain length (Fig. 6), indicating an AE-based NDT limit to SCC monitoring in the present case. From a comparison between the single dominant crack growth in stage II of stressed specimen (Fig. 4) and IGA in unstressed specimen (Fig. 8), coalescence between the IGA-induced fissures, which serves as the main AE source, seemingly leads to the formation of a single dominant crack normal to the loading direction. The formation of such fissures is confirmed to be silent in the comparative experiment of unstressed specimen (Fig. 8). Suggestively, the AE-based NDT limit is likely linked to the growth of IGA-fissure at the pit mouth (Fig. 4(c)) before the occurrence of fissure coalescence (Fig. 4(d)).

In stage II, the AE activity is quite moderate in amplitude level and number compared to stage III. This is probably the direct response to the distinction between the CGR of stage II and III (0.04 µm/s vs. 1.68 µm/s). This obvious transition in AE evolution may not only play a part in the SCC monitoring, but matters in the discrimination between the APC and HE-predominant cracking mechanisms. Last but not least, attention should be paid to the correspondence between the low-frequency AE cluster 2 and the transition area of plastic deformation. Despite the qualitative correlation, the result indicates that the AE monitoring exhibits great potential in elucidating the SCC mechanism. Moreover, more attention is being paid to improve the cluster analysis to distinguish the AE features between hydrogen bubble evolution, plastic deformation and cracking. Further improvement is expected to be reported in next paper.

Historically, SCC is often divided into the initiation and propagation phases^2–6) (Fig. 1). In most application, the division is arbitrary to a large extent. In this work, based on discussion above, stage I is the chemical-predominant pitting corrosion being reasonably termed as “precursor evolution”. Stage II is probably APC-predominant crack growth in terms of “low rate and moderate AE response” and could be deemed to be “crack initiation”. Stage III is predominantly controlled by hydrogen-assisted cracking in terms of “high rate and intense AE response” and therefore termed as “crack propagation”. Table 1 shows the correspondence between the SCC process and the corresponding AE feature. Although an exact statistical analysis of all experimental data was not included in the aim of this work, distinct AE behavior is related to the different SCC stages. From this point on, the AE monitoring could effectively reflect the initiation-to-propagation transition during SCC progression. It is thus expected that with the featured characteristics of AE behavior upon the selected corrosion system it could benefit the understanding and monitoring of the initiation and propagation of SCC process.

Table 1 In-situ observation and basic features of the detected AE signals during SCC progression.

A simple, repeatable SCC system (material properties, heat treatment, chloride solution droplet, humidity, surface finish, U-bend, etc.) was improved to facilitate a single crack evolution in SUS420J2 stainless steel and simultaneously to accommodate in-situ observation and AE monitoring experiments. The following conclusions were drawn.

This work was partially supported by Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP), “Structural Materials for Innovation” (Funding agency: JST).

• 1)   P.R.  Rhodes: Corrosion 57 (2001) 923–966.
• 2)   R.N.  Parkins: Brit. Corros. J. 7 (1972) 15–28.
• 3)  S.D. Cramer, ASM Handbook, Volume 13A, Corrosion, (ASM International, Materials Park, OH, 2003).
• 4)   A.  Turnbull: Proc. R. Soc. A 470 (2014) 20140254.
• 5)   R.N.  Parkins: Corros. Sci. 29 (1989) 1019–1038.
• 6)   F.P.  Ford: Corrosion 52 (1996) 375–395.
• 7)   G.S.  Frankel: J. Test. Eval. 42 (2008) 517–538.
• 8)   A.  Turnbull: Corrosion 57 (2001) 175–189.
• 9)   K.  Wu,  W.S.  Jung and  J.W.  Byeon: Mater. Trans. 56 (2015) 587–592.
• 10)   K.  Wu,  W.S.  Jung and  J.W.  Byeon: Mater. Trans. 56 (2015) 398–403.
• 11)   K.  Wu,  W.S.  Jung and  J.W.  Byeon: Corros. Sci. 105 (2016) 8–16.
• 12)   K.  Wu and  J.W.  Byeon: Corros. Sci. 148 (2019) 331–337.
• 13)   G.  Du,  J.  Li,  W.K.  Wang,  C.  Jiang and  S.Z.  Song: Corros. Sci. 53 (2011) 2918–2926.
• 14)   A.  Yonezu,  H.  Cho and  M.  Takemoto: Meas. Sci. Technol. 17 (2006) 2447–2454.
• 15)   J.  Xu,  X.Q.  Wu and  E.-H.  Han: Corros. Sci. 73 (2013) 262–273.
• 16)   M.  Shiwa,  H.  Masuda,  H.  Yamawaki,  K.  Ito and  M.  Enoki: Mater. Trans. 55 (2014) 285–289.
• 17)   K.  Ito,  H.  Yamawaki,  H.  Masuda,  M.  Shiwa and  M.  Enoki: Mater. Trans. 51 (2010) 1409–1413.
• 18)   L.  Calabrese,  L.  Bonaccorsi,  M.  Galeano,  E.  Proverbio,  D.D.  Pietro and  F.  Cappuccini: Corros. Sci. 98 (2015) 573–584.
• 19)  ASTM G129-00, Standard Practice for Slow Strain Rate Testing to Evaluate the Susceptibility of Metallic Materials to Environmentally Assisted Cracking, (2013).
• 20)  ASTM G36-94, Standard Practice for Evaluating Stress-Corrosion-Cracking Resistance of Metals and Alloys in a Boiling Magnesium Chloride Solution, (2013).
• 21)  NACE Standard TM0177, Laboratory Testing of Metals for Resistance to Sulfide Stress Cracking and Stress Corrosion Cracking in H2S Environments, (2005).
• 22)   K.  Ito and  M.  Enoki: Mater. Trans. 48 (2007) 1221–1226.
• 23)   I.B.  Mohamad and  D.  Usman: Res. J. Appl. Sci. Eng. Technol. 6 (2013) 3299–3303.
• 24)   P.J.  Rousseeuw: J. Comput. Appl. Math. 20 (1987) 53–65.
• 25)   Z.  Tarzimoghadam,  D.  Ponge,  J.  Klöwer and  D.  Raabe: Acta Mater. 128 (2017) 365–374.
• 26)   S.H.  Baek,  K.  Wu,  H.S.  Shim,  D.H.  Lee,  J.G.  Kim and  D.H.  Hur: Measurement 109 (2017) 18–26.
• 27)   S.H.  Zaferani,  R.  Miresmaeili and  M.K.  Pourcharmi: Eng. Fail. Anal. 79 (2017) 672–703.
• 28)  R.A. Cottis: Stress Corrosion Cracking – Guides to Good Practice in Corrosion Control, (The National Physical Laboratory, Teddington, Middlesex, 2000).
• 29)   S.K.  Bonagani,  V.  Bathula and  V.  Kain: Corros. Sci. 131 (2018) 340–354.
• 30)   P.  Hewitt and  B.S.  Hockenhull: Corros. Sci. 16 (1976) 47–51.
• 31)   S.  Hastuty,  A.  Nishikata and  T.  Tsuru: Corros. Sci. 52 (2010) 2035–2043.
• 32)   S.X.  Li and  L.H.  Hihara: Corros. Eng. Sci. Technol. 45 (2010) 49–56.
• 33)   K.H.  Anantha,  C.  Örnek,  S.  Ejnermark,  A.  Thuvander,  A.  Medvedeva,  J.  Sjöström and  J.  Pan: Corros. Sci. 148 (2019) 83–93.
• 34)   I.M.  Robertson: Eng. Fract. Mech. 68 (2001) 671–692.
• 35)   M.L.  Martin,  M.  Dadfarnia,  A.  Nagao,  S.  Wang and  P.  Sofronis: Acta Mater. 165 (2019) 734–750.