Observations of aluminum pit initiation using in situ liquid cell transmission electron microscopy

Abstract

Pit initiation has been observed with respect to the underlying metal microstructure. The experimentally observed pit location of several hundred pits with respect to the nearest grain boundary is compared to the expected distribution of pit nuclei in a Monte Carlo simulation of stochastic pit nucleation. The comparison of the experimental and simulated data shows that pit nucleation is biased towards nucleating near grain boundaries and in small grains. This is a substantial deviation from prevailing assumption in the literature that pit nucleation is stochastic, and in this work we attribute the bias towards pitting at the grain boundaries to the grain boundary density increasing the reactivity of small grains, and near-boundary regions.

It is observed that the corrosion morphology changes from fractal-like pitting across the whole surface to ultra-dense pitting under electron radiation localized to the irradiated region. The interaction of the beam with the liquid, oxide, and metal components of the corrosion cell is experimentally isolated to discern the cause for this change. There is no correlation found between the corrosion rate of the aluminum film for varying incident electron energy and therefore no evidence of displacement damage mechanisms active in the oxide film or metal. In addition, enhanced etching of the aluminum film is observed after irradiation in the liquid cell without electrolyte and subsequent corrosion by electrolyte exposure, suggesting the mechanism of corrosion acceleration is not from the water radiolysis products interacting with the film. This corrosion morphology change is then attributed to dissolution of the passive film by radiolysis, because it is seen by depth profile using Auger electron spectroscopy that the irradiated region has 1-2 nm remaining oxide film, while the unirradiated region shows the initial 15 nm of oxide film.

Relatively little is known about the mechanism of pit initiation, or factors which may predict pitting location in pure metals. Such knowledge could enable corrosion mitigation in numerous industries. This thesis is targeted at answering how the passive oxide film breaks down to initiate a pit, and determining if pit initiation location on pure metal surfaces is a stochastic process. This was done using a liquid cell holder for the transmission electron microscope (TEM) in order to observe high purity films of aluminum corroding in real time under potentiostatic and potentiodynamic polarization control in chloride-containing electrolyte. A planar three-electrode corrosion cell was designed, fabricated, and characterized for use with the liquid cell stage.

Pitting corrosion is a pernicious form of localized corrosion, occurring unpredictably in time and space across the surface of metals with passive or barrier oxides. Passivation is found as a naturally formed oxide layer on aluminum, aluminum alloys, and other common structural materials such as titanium and steel. Typically this passivation layer serves as a diffusion barrier to prevent corrosive reactions at the metal/electrolyte interface, however, breakdown in the passivation layer can lead to extremely localized attack, known as a pit. Passivity breakdown is typically associated with the chloride anion, ubiquitous in seawater, rainwater, and soil. Pit growth is autocatalytic once stable, and can rapidly penetrate materials. This creates difficult-to-detect flaws in the material which can become crack initiation sites, or holes. At best these corrosion pits are detected with rigorous inspection and person-hours—at worst, they can lead to catastrophic disasters such as oil pipeline spills and offshore rig explosions.