Cross-sectional characterization of the conversion layer formed on AA2024-T3 by a lithium-leaching coating

This work focuses on the cross-sectional characterization of the protective conversion layer formed on AA2024T3 by lithium-leaching from a polyurethane coating in a corrosive environment. The layer shows a multi-layered arrangement comprising nanoscopic local phases. Transmission electron microscopy (TEM) and complementary high-resolution secondary ion mass spectroscopy (SIMS) were employed to observe the cross-sections of the entire layer formed at different locations of a 1-mm-wide scribe, in terms of morphology, structure and chemical composition. The conversion layer was comprised of two ubiquitous sublayers; a thin dense layer (i.e. 150 nm) adjacent the alloy substrate and a porous layer. The former represents an amorphous lithium-containing pseudoboehmite phase, Li-pseudoboehmite, whereas the latter is composed of amorphous and crystalline products; an outer columnar layer merely seen on the peripheral region is also crystalline. Through a sandwich structure and the d(003) basal spacing, the crystalline phases were identified as Li-Al layered double hydroxide. Although lithium was found uniformly spread within different regions, the local phases with no/low concentration of lithium were revealed with energy filtered TEM and confirmed with SIMS analysis.


Introduction
A high strength-to-weight ratio is a key characteristic that has made aluminum alloy (AA) 2024-T3 attractive for aerospace applications. However, its high susceptibility to localized corrosion because of a heterogeneous microstructure is a downside of this alloy [1][2][3]. For many decades, hexavalent chromium-containing compounds have been applied reliably for conversion surface treatments and active protective coating pigmentation [4][5][6]. However, international health and safety legislation is imposing restrictions on their utilization, demanding the development of new corrosion inhibitive systems with reduced ecological footprints. This requires researchers either to establish new corrosion mitigation strategies or introduce new inhibitor compounds, providing equal or better active protection as compared to the existing chromated systems [7,8].
In the early 1990s, lithium salts dissolved in alkaline solutions were reported to stabilize aluminum hydroxide through lithium intercalation, showing a corrosion inhibitive character for aluminum and its alloys [9][10][11]. In 1996, Drewien et al. proposed an alkaline lithium carbonate-lithium hydroxide bath, which favored the hydrotalcite formation over aluminum hydroxides, as a promising alternative to chromate conversion coatings [12]. Later, Visser and Hayes patented the use of lithium-salts as a leachable corrosion inhibitor when incorporated in organic coatings to provide active corrosion protection [13]. Visser and his co-workers demonstrated that a protective layer was generated from these lithium-leaching coatings. Thereafter, the studies were focused on the characterization, formation mechanism and corrosion protection properties of the conversion layer formed in artificially-damaged areas from lithium-leaching model coatings using chemical and electrochemical characterization techniques [14][15][16].
In an earlier work, Visser et al. reported that leaching of lithiumsalts from the coating matrix can establish an alkaline environment (pH 9-10) within the defect area, facilitating the formation of a conversion layer on AA2024-T3 [17]. In fact, gradual leaching can induce a ranging local electrolyte chemistry in the scribe, resulting in the formation of different conversion layer morphologies with exposure time. Furthermore, the crystallographic orientation of grains at the surface of the underlying metal alloy microstructure plays a role as well [17]. In terms of composition, the results suggested formation of an Li-Al layered double hydroxide (LDH) structure over an aluminum hydroxide gel [17]. The layer, albeit morphologically diverse, is capable to protect the aluminum substrate in aggressive environments like neutral salt spray (NSS) conditions [16]. The morphology of the conversion layer shows a multi-layered structure composed of two ubiquitous layers; a dense layer adjacent the alloy substrate and a porous layer. A columnar outer layer is also observed in the peripheral region of the scribe as a consequence of higher local pH and lithium concentration [14]. Electrochemical impedance spectroscopy studies revealed that the inner dense layer is the main contributor to the overall corrosion protection provided by the conversion layer while the porous and columnar layers contribute only to a minor extent [18].
Although previous studies have shown that lithium plays a determinative role in the layer formation, the full characterization of the layer has been challenging as lithium is hard to detect within such conversion layers. Lithium has very low characteristic X-ray energy and this causes limitations to detect it even with windowless energy dispersive spectroscopy (EDS) detectors on TEM. Auger electron spectroscopy (AES) can detect lithium, however it cannot be used for the current system as the lithium peak overlaps with the binding energy of aluminum oxide. X-ray photoelectron spectroscopy (XPS) analysis is restricted by a low sensitivity (about 1%wt detection limit) toward lithium and additionally a low lateral resolution to differentiate the dissimilar phases of a surface area. Visser et al. have used XPS, detecting 4.5-4.8 at.% lithium just at the surface of the conversion layer, that was associated with Li-Al LDH [17]. However, lithium might be present also in the inner parts of the layer either below the XPS detection threshold or removed while sputtering the top part, hence remaining undetected. Recently, Marcoen et al. employed time-of-flight secondary ion mass spectrometry (ToF-SIMS) to follow the generation of the protective conversion layer within a 1-mm-wide scribe as a function of Li-leaching time in a corrosive environment [19]. They reported that initially a hydrated aluminum oxide surface layer with a little incorporated lithium is formed upon exposure, which laterally grows as a result of further Li-leaching. Subsequently, a columnar deposit, with likely an LDH structure, can nucleate when the lithium concentration surpasses a certain threshold in the coating defect. Nevertheless, all the time-resolved information is collected from the near-top surfaces which provide no cross-sectional chemical insights into the mature layer. In spite of the mobile character of lithium to the electron beam [20,21], energy filtered transmission electron microscopy (EF-TEM) technique can be used to map lithium at a high resolution. Following that analysis approach, Liu et al. [14] detected lithium in different parts of the conversion layer with EF-TEM. However, owing to the difficulties preparing the TEM specimens out of different locations in the deep scribe, the cross-sections of the entire layer were not studied to observe the lithium distribution. In addition, high resolution TEM (HR-TEM) examinations to structurally characterize the local phases were not conclusive, likely caused by the destructive influence of the used electron beam on the structure. That is why the lithium conversion layer which spatially comprises many sub-micron local phases has not completely been described in terms of composition and structure, yet. This information is of great importance to fully elucidate the formation mechanism of the conversion layer within the scribe.
This work is aimed to fully characterize the protective layer on a morphological, structural and chemical level. Therefore, TEM is utilized to study the protective conversion layer generated from lithiumleaching coatings as this technique allows to produce chemical and structural images at nanometric and atomic level. The analytical TEM investigations including HR-TEM imaging, selected area electron diffraction (SAED) patterns, EF-TEM and EDS analysis are carried out on the cross-sections of the entire layer obtained by a dedicated focused ion beam (FIB) sample preparation procedure. Complementarily, crosssectional elemental maps of the layer are acquired through high-resolution SIMS. The findings obtained in this study reveals spatial distribution of lithium within different nanoscopic parts where the local phases are structurally characterized as well.

Materials and methods
The coating was formulated as a white opaque primer using the compounds listed in Table 1 [22]. Polyurethane was used as the polymer matrix as its chemistry shows no interaction with discrete particulate ingredients in the primer preventing interference with the leachability of the lithium carbonate corrosion inhibitor [23]. The lithium-salt loaded coatings have a total pigment volume concentration (PVC) of 30 vol%, comprising 15 vol% inorganic pigments and fillers and 15 vol% lithium carbonate, respectively. The same procedure as reported already elsewhere was used for the coating preparation [14]. AA2024-T3 sheets, provided by Arconic, with a thickness of 0.8 mm were used as the substrate. The substrate was pre-treated by a tartaric sulfuric acid (TSA) anodizing process using a potential of 15 V for 25 min at 43°C, leading to an anodic film of 3-3.5 μm, according to aerospace application requirements (AIPI 02-01-003). Coatings were applied with a high volume low pressure (HVLP) spray gun followed by a 1-h flash-off at 23°C and curing at 80°C for 16 h. The dry coating thickness was 20 to 25 μm. The coated panels, with dimensions of 7 cm × 7 cm, were damaged intentionally (St. Andrews cross) with a carbide steel milling blade, creating 1-mm-wide scribes penetrating 100-150 μm into the metal. Prior to placing the panels in the salt spray chamber, their backs and sides were masked with a polyester tape. The panels were exposed to a 5% (w/w) NaCl neutral salt spray (NSS) environment (ASTM-B117) for 168 h. After the exposure, the samples were rinsed with flowing deionized water for 2 min in order to quench the process and remove any chloride residues.
For TEM investigations, the thin cross-sections of the entire layer were fabricated successfully from the regions of interest by means of a Thermo Fisher Helios G4 FIB and lift-out procedure. Prior to gallium ion milling, top-view appearances of the conversion layer were first examined using 2-keV secondary-electron scanning electron microscopy (SE-SEM) observations. To protect the conversion layer from gallium damage, a layer of carbon was initially deposited with the electron beam, followed by Pt deposition with the Ga ion beam (see Supplementary Fig. 1). A Tecnai F20ST/STEM 200 kV was used for typical bright-field TEM imaging and scanning transmission electron microscopy/energy dispersive spectroscopy (STEM/EDS) elemental mapping. A Titan cubed Cs-corrected 300 kV, equipped with a direct electron (DE-16) camera, was employed for conducting high resolution TEM (HR-TEM) imaging at a low dose rate, i.e. 19 e/Å 2 s. To map lithium within the conversion layer, the Li K-edge was energy-filtered by inserting 6-eV slits before and after the corresponding edge (EF-TEM jump-ratio approach) at either 80 or 300 keV. The identical procedure was used to filter the Al L 2,3 -edge with an onset at ΔE = 73 eV. To do so, the same electron dose rate as HR-TEM was used but the exposure time increased up to 60 s for acquiring sufficient counts. High-resolution secondary ion mass spectroscopy (SIMS) imaging was performed in a helium ion microscope (HIM) (Zeiss, Peabody, MA, USA) coupled with a mass spectrometer [24][25][26]. To avoid edge effects during the SIMS analysis, the cross-section samples were embedded in Wood alloy. The SIMS images were acquired using 20 keV Ne + primary ion with a probe current of 3 pA, leading to a probe size in the range 20-40 nm. The raster size was from 5 × 5 µm 2 to 10 × 10 µm 2 with a matrix of 512 × 512 pixels 2 . The sample stage was biased to +500 V to allow the detection of positive secondary ions. The parallel detection capability of the SIMS spectrometer enables the detection of several masses simultaneously; the secondary ions detected were 7 Li + , 24 Mg + and 27 Al + .

Results and discussion
In general, the layer can be recognized by two predominant topview morphologies; as shown in Fig. 1. The layer formed in the peripheral area (Region I) appears with a petal-like top view whereas the middle region (Region II) appearance shows globular features. A mixed morphology is observed normally while passing from the peripheral to the middle region (see Supplementary Fig. 2). In fact, the conversion layers are cross-sectionally different in morphology, including various local phases that are yet to be fully identified.

Cross-sectional characterization of the layer formed in the peripheral region
A typical cross-sectional morphology of the layer formed at the peripheral region of the scribed area is revealed by TEM micrography, as shown in Fig. 2a. The thickness of the layer is approx. 1 µm. It should be noted that the outer region of the scribe is the area where a higher alkalinity and lithium concentration is probable due to the leaching process from the coating [19]. The cross-section reveals a multi-layered morphology which is comprised of three distinct sublayers; a 150-nmthick dense layer with roundish porosities on the substrate, a 700-nmthick columnar layer on the top and a thin porous layer in between.
HR-TEM observations of the columnar part of the layer reveal a local sandwich structure [27], directly referring to a layered double hydroxide (LDH) structure which can be characterized through the d (003) basal spacing of its hexagonal structure (see Fig. 2b) [28][29][30]. The d (003) basal spacing corresponds to the stacking sequence symmetry; however, this spacing is highly dependent on the type of intercalated anions [31]. In Fig. 2b, in areas with lattice spacing, numerous crystal imperfections are observed (circled in Fig. 2b), leading to varied values of the d spacing (i.e. 7.4-8.2 Å). In order to extract further structural information, attempts were made to record magnified images in a low dose condition, but the layer was rapidly evolving due to removal of interlayer water with the electron beam (see Supplementary Fig. 3). The observed lattice spacing corresponds well to the d (003) spacing in hydrotalcite (i.e. 7.62 Å) reported by Drewien et al. [12] and Hou et al. [36]. However, the chloride intercalation is also feasible as it can reportedly give rise to a slight increase in the d spacing (i.e. 7.8 Å) [36]. For the current system, the formation of LiAl 2 -CO 3 [36]; as depicted in the inset in Fig. 2b. Please note that the coated panels were exposed to NSS environment and a minor amount of chloride (~1.3 At %) is detected within the layer with EDS analysis (see Supplementary  Fig. 6). In addition, lithium in the LDH sheets might partially be replaced with magnesium that exists in the conversion layer (~1.2 At%); this might cause a rise in the d spacing as well [36]. To avoid any confusion, they are all defined as Li-Al LDH in this work. A closer look at the porous layer shows that the porous structure is comprised of amorphous and crystalline products (Fig. 2c). Measuring the d spacing indicates the presence of Li-Al LDH at some parts of the porous layer, but in a dissimilar morphology than the columnar layer. In addition, the d (003) spacing can also be recognized in the local Fast Fourier Transform (FFT) of different parts in the image. The dense layer represents an amorphous structure, revealing no interference spots in the FFT of the image (see Fig. 2d). However, care should be taken not to turn the amorphous layer into a nano-crystalline one by exposure to a high-dose electron beam, as shown in Supplementary Fig. 4.
To map lithium, electron energy loss spectroscopy (EELS) spectra were initially collected from the layers formed in the peripheral and middle region; Li K-edge clearly shows up as illustrated in Supplementary Fig. 5. Energy filtered TEM (EF-TEM) was used to map lithium and aluminum; other elements including Mg, Cl, C and O are also detected within the layer with EDS analysis and presented in Supplementary Fig. 6. Due to the presence of an extended Plasmon peak, jump-ratio EF-TEM imaging in which the intended element is mapped through dividing the Li K-edge filtered image by the pre-edge filtered image was performed instead of normal elemental mapping with two pre-edge slits [26]. In addition, STEM/EELS mapping was also not feasible, because of perforation of the layer by the electron beam while collecting the spectrum at the required dwell time. The EF-TEM results are shown in Fig. 3. Comparing the Li and Al maps in Fig. 3a, lithium is uniformly distributed throughout the conversion layer, in particular within the dense layer. However, the local aluminum hydroxides with no/low concentration of lithium are also revealed in the top part of the layer, indicated by the yellow arrows in Fig. 3a. A closer  Fig. 4 shows structural TEM analysis of the conversion layer established in the middle region of the scribe. Compared to the peripheral region, the middle-region layer comprises an inner barrier layer with a heterogeneous morphology and a significantly thicker porous layer (ca. 400 nm) where the columnar part is missing (see Fig. 4a). It should be noted that the conversion layer formed in the middle area (approx. 500 nm) is thinner than that formed in the peripheral region of the scribe. HR-TEM imaging of the porous part of the conversion layer evidences the crystalline local phases within this sublayer, shown in Fig. 4b. Although a smaller d spacing has been measured (i.e. 7.2 Å), but it still implies the presence of Li-Al LDH. In addition, the appearance of the spots in the selected area electron diffraction (SAED) pattern shows that the porous layer is composed of co-existing crystalline and amorphous phases. Verified by the collected SAED pattern, the dense layer is structurally amorphous, as observed for the peripheral region, where roundish 5-20 nm porosities are also seen (see Fig. 4c). The porosity size increases with the distance from the substrate; the porosities in the porous layer are irregular in shape.

Cross-sectional characterization of the layer formed in the middle region
EF-TEM was used to see the distribution of lithium in the layer of the middle region. As shown in Fig. 5a, lithium is spread uniformly within the dense and porous layer similar to the peripheral region. Recently in a work, Visser et al. detected lithium throughout the 1-mm scribe, albeit at lower concentrations in the middle region [38]. However, the local lithium-poor products can hardly be distinguished in the EF-TEM lithium and aluminum maps which are the projection of the whole thickness. Moreover, EF-TEM maps are mainly intensity-based images with moderate detection limit and therefore inadequate to map elements present at low concentration. Thus, a high-sensitivity surface analysis is beneficial to determining the regions in the protective layer with high and low lithium concentration. For this reason, high resolution SIMS analysis was performed on a cross-section of the protective layer in the middle region. Because of possible artefacts that may arise near the peripheral areas (e.g. uneven electric field for secondary ion extraction), the SIMS analysis was limited to the middle region of the scribe. An overlay of 27 Al + and 7 Li + SIMS images is presented in Fig. 5b, revealing a higher lithium signal in the inner part of the conversion layer. The presence of lithium is detected rather intensely and uniformly in the inner part of the layer whereas partly observed in the top part. Fig. 5c shows the line profiles along the yellow arrow in Fig. 5b. As can be seen, Al and Li concentrations decrease toward the top part where the layer has a higher level of porosity. In addition, a maximum Li/Al ratio of~0.05 is estimated within the dense amorphous layer close to the substrate (note the right-side y-axis of Fig. 5c). This shows that the amorphous inner layer products contain lithium, as qualitatively observed in the EF-TEM maps. On the other hand, the presence of the products with less lithium can be an indication of lithium concentration variation at different stages at which various local phases can exist.

Overall cross-sectional and lateral conversion layer formation mechanism
The leaching of lithium carbonate occurs upon exposure of the coated and scribed panels to NSS, establishing a lateral concentration gradient of lithium and pH within the scribe (see Fig. 6) [14,17]. Presumably, the Li concentration and pH are also diverse in the direction perpendicular to the surface which can result in the formation of a multi-layered morphology and chemistry of the conversion layer at different locations of the scribe [38,39]. To establish a Li-Al conversion layer, lithium becomes available through the corrosion inhibitor leaching while aluminum is provided by the direct dissolution of the substrate. Scanning ion-selective electrode technique (SIET) measurements by Visser et al. [17] showed that the pH within the scribe rapidly goes up to 9.5 after 5 min of the exposure and stabilizes at approx. 9.7; Fig. 3. Bright field TEM images and the corresponding EF-TEM maps of aluminum and lithium obtained at 300 keV; (a) the entire layer formed in the peripheral region; the yellow arrows indicate lithium-free region, (b) at columnar layer/porous layer interface and (c) an individual column in the top layer. Li K-edge and Al L 2,3 -edge were filtered by putting 6-eV slits before and after the corresponding edges (jump-ratio technique), respectively. such alkalinity is high enough to trigger active dissolution of the substrate as the aluminum oxide passive layer would be undermined and soluble AlO 2 − would be stable. The lithium then starts to intercalate, resulting in the formation of stable corrosion products like Li-Al LDH. Through this simultaneous dissolution/precipitation process, the formation of the conversion layer may occur in the alkaline environment. For instance, the peripheral layer (Region I) appears in a columnar morphology as this region endures higher pH at which Li-Al LDH is merely stable [38]. Comparatively, in the middle part (Region II) a lower dissolution/precipitation ratio is expected as a result of a lower alkalinity, leading to the formation of a less porous layer that contains lithium free aluminum hydroxides. As schematically shown in Fig. 6, closest to the substrate, a ubiquitous dense layer in which roundish nano-porosities are evident is formed. Visser et al. proved that this layer is the main contributor of the corrosion protection [18]. Considering all structural and chemical characterizations into account, the amorphous dense inner layer can be Li-pseudoboehmite in agreement with what Visser et al. identified previously using the reference samples [17]. Pseudoboehmite is known as an aluminum compound with the chemical composition AlO(OH); however, it is finely crystalline boehmite that contains a high amount of water [40]. The porous middle layer is quite diverse in chemical composition and morphology and characterized as a mixed layer of coexisting lithium free/containing amorphous and Li-Al LDH crystalline products. The porous outer layer formed in the middle part is twice thicker than that in the peripheral region. Considering the chemical composition of the layer, the lithium-free phase is mainly aluminum hydroxide whereas the amorphous Li-containing part can be Li-pseudoboehmite. The presence of lithium-free products can be an indication of local pH and Li concentration variation in which various corrosion products can remain stable. Establishment of a columnar morphology is directly ascribed to the solution chemistry experienced by the peripheral regions. According to the chemical composition of the layer, [LiAl 2 (OH) 6 ](CO 3 ) 0.5 ·nH 2 O) is considered to be the most prominent product in the columnar layer, intercalation of Cl − in the anionic interlayer spacing can also lead to the formation of [LiAl 2 (OH) 6

Conclusions
In this study, the conversion layer generated on AA2024-T3 from a Li-leaching coating was fully characterized. This was achieved through successful analytical TEM and SIMS studies over the cross-sections of the entire layer formed at different locations of the scribe. It was shown that the conversion layer, regardless of the location in the scribe, is composed of an inner dense layer and a porous layer. The peripheral region which experiences higher pH and lithium concentrations during the exposure to a corrosive environment comprises an additional outer columnar layer. Lithium was found uniformly spread over the different sublayers, although the local phases with no/low concentration of lithium were also detected. The dense layer, approx. 100-150 nm thick, was specified as amorphous Li-incorporated aluminum hydroxides with a high amount of trapped water (named Li-pseudoboehmite). It was   shown that the columnar morphology is composed of single-crystal pillars of Li-Al LDH. This is a product of CO 3 2− , H 2 O and even Cl − intercalation between the d (003) planes which contains Al 3+ , Li + and OH − , leading to a sandwich structure. Crystal imperfections were also observed within the structure. The porous layer, shown to be thicker in the middle region, consists of co-existing amorphous and crystalline compounds. The crystalline products are Li-Al LDH while the amorphous products may comprise different concentration of lithium. Crystallization of the amorphous dense layer and removal of interlayer water from the Li-Al LDH structure were observed under the electron beam but could competently be avoided through a low-dose imaging.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.