Nanoscopic and in-situ cross-sectional observations of Li-based conversion coating formation using liquid-phase TEM

Aluminium alloys used in aerospace, such as AA2024-T3, are highly susceptible to localized corrosion due to complex microstructures. Chromate-based coatings have provided reliable active protection but are toxic and must be replaced. Among chromate-free alternatives, lithium salts are promising both as conversion bath chemicals and, especially, as leachable inhibitors in organic coatings. In defects, leached lithium salts create alkaline conditions (pH 9–11) that destabilize the native oxide, while lithium stabilizes corrosion products to form a protective, multi-layered conversion layer. Although the final structure is known (columnar outer layer, porous middle, dense inner layer), the temporal sequence from nucleation to multi-layer formation remains unclear. This study aims to directly observe, in situ and at nanoscopic resolution, the formation mechanism and sequence of sublayer development of Li-based conversion coatings on AA2024-T3 using LP-TEM with specially prepared sandwiched cross-sectional specimens, complemented by ex situ SEM/TEM on bulk samples.

Prior work has introduced multiple chromate-free systems for AA protection, including trivalent chromium, rare earth salts, Zr/Ti mixed systems, lithium salts, and transition metal oxyanions. Lithium salts have shown promise for both bath conversion and leaching from organic coatings, providing long-range throwing power and inducing alkaline environments in defects. The established morphology of Li-based layers comprises columnar outer Li–Al LDH, a porous intermediate, and a dense inner lithium-pseudoboehmite-like layer; the dense inner layer is most protective. Supersaturation of Al(OH)₄⁻ is critical for precipitation of stable Li-incorporated corrosion products. Earlier studies on bulk systems, electrochemical responses, and cross-sectional characterizations provided composition and morphology but lacked real-time mechanistic observation. LP-TEM offers dynamic imaging of corrosion and deposition, though challenges include specimen stability and beam effects. A sandwich approach with inert TEOS masking can confine reactions and enable cross-sectional in situ observation of conversion processes.

In situ LP-STEM/TEM: AA2024-T3 lamellae (approximately 300–400 nm thick) were prepared by FIB/SEM (Thermo-Fisher Helios G4 with EasyLift). Both sides of each lamella were coated with ~30 nm TEOS to create a sandwiched specimen that confines electrolyte and forces oxidation/dissolution to proceed in-plane from a single clean edge. During mounting on liquid-cell chips, three edges were further masked with TEOS, leaving one clean, Ga-free edge exposed to electrolyte. Chips were plasma cleaned (1 min) for hydrophilicity, assembled into a homemade liquid-cell TEM holder, and sealed with epoxy. The electrolyte (stagnant) was 0.01 M NaCl + 0.01 M Li₂CO₃ (pH 10.6), representative of defect chemistry during lithium leaching. In situ imaging used a Cs-corrected Thermo-Fisher Titan 300 kV TEM in STEM mode, primarily BF-STEM for lower dose and improved SNR, beam current ~0.1 nA (with liquid only), intermittent snapshot imaging over ~1 h to mitigate beam-chemistry changes. Images were acquired at 5000× (2048×2048 px). Drift-corrected frame integration combined three frames at 0.2 μs dwell to reduce point-excitation effects. The specimen was placed at the entrance side of the liquid cell to minimize beam broadening. Complementary in situ HAADF-STEM was used to visualize Z-contrast features (e.g., Cu-rich particles) and early dissolution fronts. Ex situ validation on bulk AA2024-T3: 2 mm thick sheets were mechanically polished (0.5 and 0.05 μm alumina in non-aqueous media), cleaned with isopropanol, and exposed for 7 h to either 5 mL (electrolyte thickness 2 mm; 0.2 mL cm⁻²) or 100 mL (electrolyte thickness 40 mm; 4 mL cm⁻²) of 0.01 M NaCl + 0.01 M Li₂CO₃ (pH 10.6). After exposure, samples were rinsed with distilled water and examined by SEM (15 keV, SE). Cross-sectional lamellae from regions of interest were prepared by FIB lift-out for STEM/EDS (Thermo-Fisher Titan with Super-X EDS) to analyze morphology and elemental distributions. Post-mortem analysis of the sandwiched thin specimens was also conducted by HAADF-STEM/EDS.

- Real-time LP-TEM revealed the sequence of Li-based conversion layer formation on AA2024-T3: initial uniform active dissolution, emergence of needle-like precipitates (columns), development of a porous layer, and eventual formation of a thin dense inner layer adjacent to the alloy. - Temporal evolution (BF-STEM): 0–23 min: active dissolution without conversion products; 23–42 min: continued dissolution (total ~3.48 μm removal up to 42 min); at 42 min: first appearance of needle-like precipitates at the dissolution front, often near grain boundaries; by 47 min: needles averaged ~0.8 μm length; by 55–60 min: a multilayered conversion layer was established (columnar outer, porous middle, dense inner near substrate). - Dissolution kinetics slowed as the layer formed (depth penetration and rates): 0–23 min ~1950 nm (84 nm min⁻¹); 23–42 min ~1530 nm (80 nm min⁻¹); 42–47 min ~300 nm (62 nm min⁻¹); 47–60 min ~700 nm (53 nm min⁻¹). - Early-stage HAADF-STEM showed initial localized dissolution “dents” at the edge (t=5 min) progressing to more uniform dissolution (t=25 min) with an undulating dissolution front indicating transient local anodes. Nanoscopic Cu-rich intermetallic particles (IMPs) acted as cathodic sites and underwent dealloying/self-corrosion, sometimes dissolving completely or leaving bright remnants. - Grain-dependent activity: dissolution accelerated along grain boundaries and varied between grains, consistent with orientation-dependent dissolution rates; needle formation preceded porous layer development. - Post-mortem HAADF-STEM/EDS of sandwiched specimens (1 h exposure) confirmed three sublayers and identified Al and O throughout, with Cu-rich regions in the layer originating from corroded nanometric IMPs. Lithium mapping by EELS was not feasible due to specimen thickness and membrane/mask layers. - Bulk validation: Exposure to a larger solution volume (100 mL; 4 mL cm⁻²) delayed conversion layer formation, yielding rough corrosion without a protective layer after 7 h. A smaller volume (5 mL; 0.2 mL cm⁻²) produced a conversion layer within 7 h, with typical petal-like top-view morphology. Cross sections showed a 50–100 nm columnar outer layer and an underlying 50–100 nm porous layer; Cu-rich regions were detected within the layer. Prior work confirms Li incorporation in both crystalline Li–Al LDH and amorphous phases. - Mechanistic driver: Supersaturation of Al(OH)₄⁻ at the alloy/solution interface, aided by confinement between TEOS layers, triggers precipitation of Li-containing products; as the layer grows, the precipitation/dissolution ratio increases, culminating in a less-porous dense inner layer that slows dissolution.

The study directly addresses the open question of how Li-based conversion coatings nucleate and evolve into a protective multilayer on AA2024-T3. In alkaline Li₂CO₃/NaCl solution (pH ~10.6), the passive film is unstable, leading to active dissolution that preferentially initiates at grain boundaries and Cu-rich IMPs. Confinement by TEOS masks retains Al(OH)₄⁻ near the dissolving front, enabling local supersaturation that nucleates needle-like columns (analogous to Li–Al LDH). Continued dissolution and precipitation increase local supersaturation, causing a porous layer to develop. Ultimately, a dense inner layer forms adjacent to the alloy, decreasing porosity and significantly reducing dissolution, thus enhancing barrier properties. The observed slowing of dissolution rates correlates with progressive layer development. The in situ thin-specimen observations mechanistically map onto bulk behavior. While columns grow predominantly in two dimensions in the sandwiched geometry (needles), bulk systems permit three-dimensional growth and entanglement, slowing dissolution and extending the conversion process. The effect of solution volume on layer formation underscores the role of Al(OH)₄⁻ supersaturation: larger volumes require more substrate dissolution to achieve supersaturation, delaying precipitation. A probable pH gradient away from the surface further favors eventual passivation. The findings thus rationalize prior macroscopic results on lithium-leaching coatings and explain why extended exposure improves corrosion performance via thickening of the dense inner sublayer.

This work establishes, via in situ LP-TEM with sandwiched AA2024-T3 specimens, the nanoscopic, time-resolved mechanism of Li-based conversion coating formation: initial active dissolution, nucleation of needle-like columns upon Al(OH)₄⁻ supersaturation, development of a porous layer, and late-stage formation of a thin dense inner layer that slows further dissolution. Complementary ex situ SEM/TEM on bulk samples validates the morphology, composition (Al, O, Cu-rich regions), and highlights the critical influence of solution volume on achieving supersaturation and timely layer formation. The approach provides direct mechanistic insights and a general framework for studying other complex conversion systems in situ. Future work could quantify lithium distribution and phase evolution during growth (e.g., with optimized EELS or alternative spectroscopy), explore the roles of alloy microstructure and orientation on nucleation/growth kinetics, and systematically vary solution chemistry, flow, and confinement to tune layer architecture and performance.

- In situ LP-TEM primarily provided morphological information; structural and compositional changes of sublayers during formation could not be fully resolved in real time. - Lithium mapping was not achieved in the sandwiched configuration due to specimen/membrane thickness limiting EELS sensitivity; EDS cannot directly detect Li. - Electron beam effects on solution chemistry necessitated low-dose, intermittent imaging; despite mitigation, some beam-induced artifacts cannot be entirely excluded. - The sandwiched thin-specimen geometry and stagnant electrolyte differ from bulk conditions (specimen size, surface-to-volume ratio), potentially affecting kinetics; correlations to bulk were inferred and validated ex situ but not strictly identical. - The electrolyte was stagnant; effects of flow or replenishment on supersaturation and morphology were not addressed.