A review—Pitting corrosion initiation investigated by TEM

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1455

2. Initial dissolution in stainless steels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1456

2.1. 316F stainless steel [24,25] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1456

2.2. Super 304H stainless steel [28]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1458

3. Atomic-scale origins of pitting corrosion in 2024Al alloy [48-51] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1459

4. Conclusions and perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1463

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1464

References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1464

1. Introduction

Passivity, extremely retarding the general corrosion of the materials, has been regarded as the key to our metals-based civilization [1]. The localized corrosion is well known to be one of the most serious issues confronting the passive metal, and thus clarifying the mechanism of localized corrosion has been a fundamental issue in corrosion field.

Pitting corrosion, as a typical localized corrosion type, is one of the most common and catastrophic causes of failure of metallic structures [[2], [3], [4], [5]] and has attracted corrosion community striving to study [2,3,[6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19]]. In spite of the large number of past studies, controversies still exist, especially in the mechanism at the pitting initiation stage. Actually, clarifying the pitting initiation mechanism is to answer “where, why and how the pitting initiates”.

As we know, the pitting corrosion always begins with the local breakdown of passive film [20] which induced by the heterogeneity or discontinuity of the passive film. The nature of the ultrathin passive film (usually 1-3 nm) heavily depends on the structural characters of the beneath matrix. The pitting corrosion is generally observed to be associated with inclusions, second phase or precipitates. These macro-defects induce the discontinuity of the passive film or/and their potential is differ from that of the surrounding matrix, which induces many miniature galvanic cells or “micro-galvanic cell” that could drive local electrochemical dissolution. Usually, corrosion is recognized and understood at a relatively larger scale as above mentioned. However, infinitely close to the very initial stage, corrosion dissolution is believed to begin with departure of some atoms from the lattice position under the effect of the electrolyte. In a similar way, heterogeneities in chemistry and structure at the atomic level are expected to induce the initially preferential corrosion at atomic scale. Indeed, mesoscale and nanoscale theories of corrosion dissolution are now, more than ever before, relatively well understood, whereas there are significant knowledge gaps in our understanding of corrosion processes at atomic length scales. That is mainly due to the difficulty encountered in obtaining precise experimental information with enough spatial resolution. Deciphering this link between microstructure and corrosion at the atomic scale will surely provide clear atomistic insights on the mechanism of the corrosion process.

It is evident that tracking the accurate sites where initial dissolution occurs as well as the propagation of the dissolution by means of multi-scale (from nano-meter scale to atomic scale) characterization is key to clarify the pitting initiation mechanism.

Transmission electron microscopy (TEM) is capable of providing high spatial and energy resolution, which has tremendous potential in resolving the origin of pitting corrosion. Particularly, the incoherent high angle annular dark field scanning TEM (HAADF-STEM) imaging mode from high-angle scattered electrons yields strong atomic number (Z) contrast [21], in response to the local variability of chemical composition and/or thickness contribution of the different species. So the local dissolution should lead to a darker contrast in such a mode. This makes it possible for monitoring the initial corrosion dissolution and establishing the link between the corrosion initiation sites and the corresponding structural characters. The recently developed aberration-corrected TEM not only features ultra-high spatial resolution but also eliminates the delocalization effect, which is very suitable for investigating interface-induced phenomenon.

Here, we will review our recent progresses in this issue by summarizing the results in three representative materials of 316F, and Super 304H stainless steel as well as 2024-Al alloy, using in situ ex-environmental TEM technique.

2. Initial dissolution in stainless steels

Stainless steels are used in diverse applications for their superior general corrosion resistance. Nevertheless, they are susceptible to pitting corrosion when they are utilized in a chloride containing environment.

Why stainless steels pitting.--Inclusions induced structural heterogeneity

In the steel-making process, impurity element of sulphur is hard to be removed thoroughly, even though the modern advanced steel-making technique can well control the amount of sulphur. Sulphur in stainless steel is usually present in the form of manganese sulphide (MnS). Under consideration of some special purpose, such as free-cutting property, high sulphur content and thus high volume of MnS is introduced artificially to improve the lubricating effect. So, the sulphide inclusions in form of MnS are more or less ubiquitous in austenitic stainless steels. In the past several decades, MnS has been specially focused since it is found that the macro pits always occur at MnS and it is generally believed that the pitting events result from the local dissolution in MnS inclusion [2,6,7,9,[11], [12], [13],15,16]. Apart from the sulphide, oxide inclusions also exist more or less. Different from sulphide, oxide inclusion induces the dissolution of passive film rather than dissolved itself. Meanwhile, some micro-defects, such as dislocations and grain boundaries, also lead to the heterogeneity in properties of passive film and thus contributing to the pitting corrosion to some extent. Whatever, the inevitably heterogeneity or discontinuity in the passive film makes the stainless steel susceptible to pitting corrosion.

Where pitting initiates.--Local dissolution of MnS or chromium-depleted zone

The breakdown of the passive film and very initiation of the pitting process is probably the least understood aspect of the pitting phenomenon. Of the existing mechanisms, we are bold to classify them as two divergent schools of thought: One advocates that pitting originates from the local dissolution of MnS, which is generally believed. The other postulates that pitting initiates at the chromium-depleted zone adjacent to the sulphide inclusion [14,17].

Where the local dissolution of MnS begins, or what feature in structure triggers the initial dissolution, is the usual concern. Reasonably, the interface of MnS/matrix is proposed to be the active site and trigger the initial dissolution [11]. However, MnS inclusions were occasionally found to be dissolved in the interior zone [12,19], which indicates that the interface itself is not the absolute inducement of local dissolution. Furthermore, MnS inclusions behave heterogeneous activity since only some inclusions are usually found to dissolve in the initial corrosion stage. Therefore, the origins of local dissolution in a single MnS inclusion and the heterogeneous activity between varied MnS inclusions are the major concerns and poorly understood. Previous work even proposed that the activity of MnS is dependent on the size wherein MnS inclusions larger than 0.7 μm behave active and below 0.7 μm behave inactive [10,22,23]. Also, some researchers remarked that the activity is correlated with the purity of MnS inclusions. Only sulfur-containing inclusions were attacked, whereas Cu-containing MnS inclusions were unattacked [9].

From the metallurgical point of view, Williams and Zhu [14] proposed that chemical changes in and around MnS inclusions are a mechanism for the pitting initiation. Following this proposal, Ryan et al. [17] carried out a chemical composition analysis by means of secondary ion mass spectroscopic analysis of areas selectively sputtered by a focused ion beam. They reported a significant reduction in the Cr/Fe ratio of the steel matrix around MnS particles, and proposed that the chromium-depleted zones are susceptible to high-rate dissolution that “triggers” pitting. However, they cannot provide any experimental evidence to support their idea.

Evidently, deciphering the structural origins of the preferential dissolution sites as well as the heterogeneous activity of MnS inclusions would make the debate come to an end.

2.1. 316F stainless steel [24,25]

A commercial hot-rolling 316F type austenite stainless steel with high sulphur content used in our study contains a large number of needle-shaped MnS inclusions with dimensions 20-50 μm in length and 0.3-1.0 μm in width (Fig. 1(a)). We find that a “single-grained” MnS inclusion in the steel is compositionally and structurally inhomogeneous. Fine octahedral precipitates of spinel MnCr₂O₄, with dimensions of several tens of nanometers, are dispersedly distributed in the MnS inclusions (Fig. 1(b) and (d)), generating local MnCr₂O₄/MnS nano-galvanic cells. In situ ex-environmental TEM studies indicate that the MnS initially dissolves at the MnCr₂O₄/MnS interface in the presence of salt water, and the dissolution gradually spreads outwards, leaving a pit around the MnCr₂O₄ octahedron (Figs. 1(c), (e), (f), 2 and 3 ).

The dissolution of MnS was proposed to occur chemically or electrochemically or both [7,26,27]. The preferential site where dissolution occurs is actually resulted from the integrated effect of the multi-galvanic cell comprised of MnS/matrix (covered with passive film) and MnS/MnCr₂O₄. Thermodynamically, the standard electrode potential (SEP) difference of MnS/MnCr₂O₄ is larger than that of MnS/matrix (covered with passive film), yielding a greater corrosion tendency at interface of MnS/MnCr₂O₄. Dynamically on the other hand, on the anode (MnS), MnS dissolution occurs, while the oxygen reduction reaction (ORR) takes place on the cathode (MnCr₂O₄). The ORR is proposed to be the rate-determining step in the neutral corrosion system, and thus the rate of MnS dissolution reaction can be assessed by evaluating the rate of ORR occurring on the MnCr₂O₄ oxide surface. By means of large-angle tilting experiments and 3D tomography in a TEM, the MnCr₂O₄ nano-particles are visualized to have specific geometric shape, which was identified to be an octahedron enclosed by eight triangles (Fig. 4). Such {111} surface has alternatively metal-terminal or oxygen-terminal surface configuration (Fig. 5, Fig. 6). First-principles calculations clarified that the nano-octahedron, enclosed by eight {111} facets with metal terminations, are much more energetically favoured for ORR than O-terminated ones, yielding more reactive in catalyzing the MnS dissolution.

Vividly speaking, MnS inclusions are in an identical over-all environment comprised of the surrounding passive film-covered matrix. In this case, any additional micro-galvanic would trigger the preferential dissolution in MnS, just like the effect of nano-octahedron MnCr₂O₄.

MnS inclusions embedded with nano-oxide is not prevalent in other types of stainless steels. Then a concern is that, if the MnS is free of oxide particles, where the preferential dissolution site is. Combining bright-field (BF) and high angle annular dark field (HAADF-STEM) imaging, we have found that the preferential dissolution occurs at the surface emergences of the dislocations with both edge and screw characters (Fig. 7) [25]. With aberration-corrected HRTEM and quantitative strain-analysis techniques, the SEP is found to drop sharply in the region with $\widetilde{1}$ nm radius around the dislocation core induced by the strain (Fig. 8), indicating a large dissolving trend in this area.

Our recent progress in the preferential dissolution of MnS, closely related with the pitting initiation, uncovers the structural origins at atomic scale. The dissolution is not at the MnS/matrix interface as generally believed, but at the sites where the defects-induced additional galvanic effect is more pronounced, such as the more efficient ORR on the metal-terminated MnCr₂O₄ (cathodically), or the strain-induced SEP reduction at dislocation core (anodically). Undoubtedly, by means of in-situ ex-environmental method in TEM, we have answered the issues “where, why and how the pitting initiates” and provided a new basis for understanding pitting corrosion of stainless steels.

2.2. Super 304H stainless steel [28]

Some newly-developed types of austenitic stainless steels, with superior properties for some particular applications, are not featured by the sulphide inclusions any longer. Taking Super 304H stainless steel (Super 304H SS) as an example, it is developed by the addition of alloying elements Cu, Nb and N on the basis of 18Cr-8Ni stainless steel [[29], [30], [31], [32]], is being widely used in ultra-super critical fossil boilers because of its superior high temperature strength [33], oxidation resistance and steam corrosion resistance [34]. The main feature of Super 304H SS is precipitation of fine Cu-rich phases dispersedly distributed in the austenitic matrix, derived from the presence of the alloying element Cu [31]. Evidently, the mechanism based on the local dissolution of MnS inclusions is not applicable to the Super 304H SS, and the contribution of Cu-rich phase to its superior corrosion resistant properties should be particularly focused.

The effects of copper-alloying in corrosion resistance of stainless steels have been presented in numerous publications [[35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46]]. Usually, the addition of copper to ferritic, austenitic or duplex steels improves the resistance to uniform corrosion in sulfuric acid [[35], [36], [37],45], whereas shows positive [38,39,41,43] or negative effects to the local corrosion [40,42,44]. The effect of the alloying element copper has been implied to be ascribed to the alteration in the microstructure and its dissolution as well as re-deposition process. Actually, Cu has long been recognized as a mysterious alloying element. For example, Cu oversaturation in antibacterial stainless steels yields Cu in form of either a solid solution and/or a nanoscale Cu-rich phase, which plays a key role in the antibacterial action [47]. The design is mainly based on the mechanism that elemental Cu in either form can slowly dissolve into the electrolyte from the stainless steel. Such preferential dissolution of Cu in form of Cu-rich phases is also a key feature of the initial corrosion stage.

With the application of HAADF-STEM imaging combined with energy dispersive spectroscopy (EDS) analysis, we have clarified the effect of the Cu-rich phase and solid-solution state copper on the corrosion behavior of the Super 304H SS through monitoring the structural and compositional evolutions with the corrosion process. The Cu-rich phase particle is preferentially dissolved at the initial corrosion stage (Fig. 9). The “tiny pits” left by the dissolution of nano-scaled Cu-rich particles are oxidized and bordered by a newly formed spinel FeCr₂O₄ oxide ring (Fig. 10), which is a passive film-like oxide in composition, yielding the dissolution not to extend further into the steel matrix. The solid-solution state copper is enriched in the matrix immediately beneath the oxide film and is dissolved through the film when contacting the electrolyte. It is implied that the pristine discontinuous passive film abridged by Cu-rich phase particles becomes continuous when the Cu-rich phase particles are dissolved, which would be a factor contributing the superior corrosion resistance.

3. Atomic-scale origins of pitting corrosion in 2024Al alloy [48-51]

In addition to stainless steels, another category of widely-used structural materials is Al alloys, which are also prone to undergo localized corrosion. For stainless steels, the pitting events usually are correlated with MnS inclusions. Comparatively, the pitting of Al alloys is derived from their heterogeneous microstructure that consist of hardening precipitates and dispersoids which also contributes their superior strength property [52,53]. Contrasting to the simpleness in stainless steel, there exist a large number of second phase particles with varied types and sizes in Al alloys, which renders more difficult to clarify the origins of pitting events in Al alloys.

Al-Cu-Mg alloys (2××× series commercial alloys) are widely used in aerospace and other industrial applications. In the last a few decades, researchers have been striving to study the pitting of Al-Cu-Mg alloy using various techniques [[54], [55], [56], [57], [58], [59]]. Among various second phase particles present in 2024Al alloy, attention is usually paid to the influence of S-Al₂CuMg phase which is believed to dissolve preferentially as the initiation sites of pitting [[60], [61], [62], [63], [64], [65], [66]]. The role of the precipitates and dispersoids in the pitting initiation period is generally analyzed based on the “electrode potential” theory. The intermetallic compound S-Al₂CuMg phase is electrochemically active with respect to matrix and dissolved initially as anode with Mg and Al selective dissolution. With the development of Mg and Al dealloying, the remnant becomes noble and tends to behave as a cathode which results in the dissolution of the matrix around the S phase and consequently pitting occurrence [60,62,67]. The trend of potential evolution with dissolution in S phase has been experimentally evidenced by SKPFM technique [58,68]. However, the S phase particles are found to feature a great heterogeneity in the electrochemical activity between particles and in the interior of single particle. In the chloride-containing electrolyte, some S particles are susceptible to dissolution, while others are electrochemically inactive. Moreover, the dissolution of a so-called active S particle is usually in an inhomogeneous manner [58,68]. Evidently, the idea of “volta potential” or “electrode potential” respect to matrix is incapable to clarify the cause resulting to that heterogeneity in the electrochemical corrosion activity. Structural differences between particles and in the interior of single S phase particle could be responsible for the heterogeneity of activity, since structure determines property. However, the link between the activity and local structure as well as chemistry has not yet been established because of the fact that the widely used analytical approaches failed to provide the structural and chemical information with required spatial/chemical resolution. The intrinsic mechanism is not clarified yet which limits our understanding on the pitting initiation of Al-Cu-Mg alloys.

With the application of in situ ex-environment transmission electron microscopy, we have identified the initial site, at an atomic scale, of pitting corrosion in an Al-Cu-Mg alloy (Fig. 11). Fig. 11(b) is the HAADF image of an S phase particle while Fig. 11(c) is the same particle that has experienced an immersion in 0.5 mol/L NaCl for 15 min. The S-Al₂CuMg phase is compositionally and structurally inhomogeneous (Fig. 11(b)). A large number of nano-sized approximants of the decagonal quasicrystal, which is determined to be Al₂₀Cu₂Mn₃ (Fig. 12), are embedded in most S particles. The dissolution of S phase is seen to be strongly localized with some small pits. They are distributed randomly and always feature a nano-core with a different contrast. It is of great interest to find that each core appears to correspond to one nano-sized Al₂₀Cu₂Mn₃ nanoparticle embedded in the S phase. We also find that the S phase particles free of or with few Al₂₀Cu₂Mn₃ nanoparticles usually do not suffer local dissolution. Therefore, it is the Al₂₀Cu₂Mn₃ approximant that is responsible for the heterogeneous activity between particles and in the interior of single particle. Namely, the preferential dissolution occurs at the periphery of the Al₂₀Cu₂Mn₃ approximant. The S phases with Al₂₀Cu₂Mn₃ approximant behave more active, while those free of or with few nano-particles behave inactive at the early stage of corrosion.

With further reducing the corrosion time, we have directly observed that the dissolution of Al₂₀Cu₂Mn₃ approximant is the initial stage prior to the S phase dissolution (Fig. 13(a)). The dissolution of Al₂₀Cu₂Mn₃ triggers S phase dissolved and enlarged (Fig. 13(b)). The local dissolved area in S phase turns to be cathodic, which in turn accelerates the further dissolution of the adjacent Al₂₀Cu₂Mn₃ (Fig. 13(c) and (d)). This process could be regarded as a “self-catalyzing” like process. The prominent effect of Al₂₀Cu₂Mn₃ phase is to indirectly influence the localized corrosion behavior of Al alloy through increasing the electrochemical activity of S phase, other than directly impact the Al matrix, which is one of the main reasons why the Al₂₀Cu₂Mn₃ phase has been neglected in electrochemistry in the past a few decades.

We also find the electrochemical activity of different Al₂₀Cu₂Mn₃ particles is varied. As shown in Fig. 14(a) and (b), in an S phase particle suffering immersion in NaCl for 25 min, some Al₂₀Cu₂Mn₃ particles marked with black arrows are dissolved severely, around which S phase is also decomposed. Moreover, the Al matrix at the interface S phase/Al is dissolved as well. Some other Al₂₀Cu₂Mn₃ particles labeled as I, II, III, IV, nevertheless, were unattacked, which indicates that Al₂₀Cu₂Mn₃ particles behave different activity. In Fig. 14(c), I-IV shows the zoom-in images of the four undissolved Al₂₀Cu₂Mn₃ particles in Fig. 14(b). It can be seen that these particles feature only simple parallel twins. By observation of numerous Al₂₀Cu₂Mn₃ particles in S phase, some other undissolved Al₂₀Cu₂Mn₃ particles free of twinning were also observed. Meanwhile, we find a large number of Al₂₀Cu₂Mn₃ particles in Al matrix which are slightly dissolved, typically shown in Fig. 15. All these particles feature multiple twins and are preferentially dissolved in the vicinity of the intersected zone of the prism twin plates or at the center of multi-parallel shaped twin plates. The difference in the configuration of twins existing in Al₂₀Cu₂Mn₃ phase is responsible for the heterogeneous activity. In other words, those with multiple twins behave active and are dissolved preferentially in the vicinity of the junction zone, while those free of twins or with simple parallel twins are inactive and undissolved even though dissolution occurs in some S phase particles.

Further reducing the corrosion degree and improving the resolution of observation, we designate an experimental procedure to facilitate close observation of the initial dissolution of the Al₂₀Cu₂Mn₃ phase at the atomic scale and thereby attempt to establish an atomic-scale understanding of the link between microstructure and corrosion [49]. We briefly immersed the 2024Al alloy in neutral 0.5 mol/L NaCl electrolyte, allowing enough time for very slight dissolution in the Al₂₀Cu₂Mn₃ phase and then monitored the structural evolution using the HAADF-STEM technique in the TEM. We observe that curled streak with brighter contrast frequently exists at the twin boundary. The bright streak is determined to be enriched in Cu, with a large number of defects [51], as also shown here in Fig. 16. Initial dissolution just occurs near the Cu-rich streak, leaving some small spots with slightly darker contrast behind (Fig. 16(a) and (f)). Meanwhile, the copper-rich products form at the ring sites bordering the spots (Fig. 16(a) and (f)). The result confirms, as widely believed, that it is at the atomic scale that dissolution ensues. Preferential dissolution evidently originates from atomic-scale heterogeneities in chemistry and microstructure within the Al₂₀Cu₂Mn₃ phase. The bright curled streak, formed by the segregation of Cu at atomic-scale defects, and its adjacent zone constitute a galvanic couple, in which the nobler Cu-rich zone triggered preferential dissolution of the adjacent zones. We term this system an “Atomic-scale galvanic Cell”, akin to the concept of the micro-galvanic cell. In other words, the location where the selective dissolution begins is determined by the atomic scale galvanic cell.

4. Conclusions and perspectives

In this project, we have performed various techniques under TEM to provide nano- and even atomic-scale information on the initial site where initial dissolution occurs, and to clarify the structural origins of the preferential dissolution. The heterogeneity in electrode potentials induced by structural defects is found to be the most fundamental driving-force triggering the initial corrosion. By means of multi-scale TEM characterizations together with some theoretical calculations, the physical insights on the pitting initiation are ultimately clarified at the atomic scale, such as the metal-terminated surface configuration of octahedral MnCr₂O₄ and dislocation emergency in stainless steels, and the elemental segregation at the twin-boundary of Al₂₀Cu₂Mn₃ decagonal approximant in the Al alloy.

Transmission electron microscopy is capable of providing local information on morphology, crystallography, and chemical composition. The recently developed aberration-correction in the optical column makes the spatial resolution of a TEM up to atomic scale. The era has come in which we have great opportunity to review some classic problems in materials science and to validate the correctness of some well-established hypotheses and theoretical models. As exampled in this paper, pitting initiation happens at the atomic scale, which can be readily captured by atomic mapping in a TEM. To systemically track further development of a pitting, experimental observation supplemented with a time-scale is necessary. Thus, in-situ investigation under a corrosive environment should be performed during which microstructural evolution could be monitored in a dynamic mode. The atomic-scale information on pitting initiation and development may certainly benefit material scientists and engineers to work out suitable approaches for improving the pitting resistance through composition design and control of the processing technology.

Acknowledgements

This work was supported financially by the National Natural Science Foundation of China (Nos. 51771212 and 11327901) and the Innovation Fund in IMR (No. 2017-ZD05).

The authors have declared that no competing interests exist.

---