Liquid-Cell Transmission Electron Microscopy of Localized Corrosion of Aluminum

Abstract

While the growth of pits in passive metals exposed to chloride solutions is well understood, the processes associated with the initiation and propagation of stable pits, versus pits that form and then re-passivate, are still a matter of conjecture. This study uses potentiostatic polarization to determine the pitting potential of aluminum as a function of chloride activity at constant pH. In contrast to earlier studies there is a departure from linearity as the solution becomes increasingly dilute, suggesting a new mechanism for the initiation of pits in less aggressive environments. This is shown in Figure 1, where a line has been fit both to the entire dataset, and two lines have been fit to their respective high and low portions of the dataset. The data at low chloride concentrations do not agree with the formulation developed by McCafferty [1] unless fit separately as two lines of different slope. Pitting corrosion is directly observed in situ in a liquid cell transmission electron microscope (TEM) [2]. Figure 2 shows a schematic of our corrosion cell design that we incorporate into this holder. The sample is created using standard electron beam evaporation, sputter deposition, and shadow mask methods. Current is supplied with titanium leads to the aluminum. This configuration allows pitting corrosion to be randomly distributed over the entire aluminum field. The influence of the microscope induced electron dose on the corrosion properties has been investigated and found to be controllable and consistent. We have characterized the thin film aluminum electrochemical behavior by generating Tafel curves within the liquid cell, and found the observations in situ to be largely representative of the bulk behavior. In Figure 3 this comparison is shown for nominally pure aluminum in the bulk case, versus the thin film samples, both in and out of the TEM. At low chloride concentration the curves have similar characteristics, such as similar anodic current slope, but are translated to more noble potentials and higher current densities for pitting behavior. Changes in the open circuit potential between the two thin film samples; in the microscope and external to it, are partially explained by the bias applied by the goniometer. The difference between open circuit potential in the bulk and thin film cases is consistent over several measurements, but the reason for the shift is a topic of further research. Current work is probing the relationship between thickness of the thin film and behavior compared to the bulk. In all cases we observed current transients representative of metastable pitting. We were able to relate crystalline features found in situ with topographic features using atomic force microscopy (AFM), as seen in this group’s earlier work [3]. Coupled with a depth profile from Auger Electron Spectroscopy of the pit this suggests that the ridges observed around the pit cusp in AFM are metallic aluminum. Pit growth is observed in situ, as seen in Figure 4, and suggests a redeposition of aluminum from the pit anolyte as seen in the red-circled features. Redeposition could result in conditions for repassivation by depriving the system of the critical concentration of aluminum ions [4]. Future work will include oxidation of the aluminum film in situ to examine changes from radiolysis by the beam, and effects of chloride on a fully oxidized aluminum film (no remaining underlying aluminum metal). It is expected this will reveal the influence of chloride on the surface-oxide interface and lead to greater detailed understanding of the oxide film’s role in passivation. This work is funded by NSF, under NSF-DMR 1309509. We acknowledge See Wee Chee and Brent Engler for collaboration, R. Dove and R. Planty for technical assistance and facilities within the Nanoscale Characterization Core operated by the Center for Materials Devices and Integrated Systems at RPI.