Porous anodic oxide films are produced when reactive metals such as Al and Ti are electrochemically oxidized in baths that dissolve the oxide. Much research in porous oxide based devices has been stimulated by the self-organized hexagonally ordered pore arrays found for certain anodizing conditions.1 However, the process of pore formation and self-ordering is not yet understood. Morphology evolution accompanying pore initiation has been characterized during, for example, constant current anodizing of Al in 0.4 M H3PO4.2 A morphological instability of the initially planar barrier oxide occurs at 10-20 nm thickness, resulting in a disordered pattern of small pores spaced at about 20 nm. The barrier oxide thickness then continues to increase up to about 150 nm, at which the self-ordered pattern of pores spaced at 230 nm initiates. The initial instability at 10-20 nm is apparently controlled by the interplay of dissolution and electrical migration.3 A tracer study recently found evidence that plastic flow of oxide is involved in the transition to the self-ordered pore pattern.4 It had been previously shown that stress builds up at a constant rate during the regime of barrier oxide growth, suggesting elastic loading of the anodic film.5,6 The rationale for the transition from elastic to plastic oxide behavior are yet unclear.The present work related the evolution of the stress distribution in the barrier oxide layer to the initiation of the self-organized pore array. Al samples were anodized at constant current density in 0.4 M H3PO4. Through-thickness profiles of the in-plane stress in the oxide were revealed for the first time, by in situ monitoring of stress change during open circuit dissolution following anodizing.7 Oxide morphology evolution during anodizing was statistically characterized by Fourier transformation of SEM images.SEM images showed that the morphological instability initiated at an oxide thickness of 20 nm, and produced a stable surface roughness pattern with a length scale of 20 nm. Prior to the instability, stress measurements showed that compressive stress in the oxide was evenly dispersed through the oxide thickness. However, after the instability, the stress profile became concentrated within a layer of about 20 nm thickness adjacent to the oxide-solution interface. At greater depths, the oxide was stress-free, revealing that the compressive stress is generated by absorption of oxygen ions (and possibly phosphate ions) at the solution interface. With increasing barrier oxide thickness, the stress level within the stress-accumulating layer became increasingly compressive. This behavior continued until the moment of self-ordered pore initiation, when both the barrier layer thickness and the integrated oxide stress rapidly decreased to steady-state values. Both the morphology change and stress transient can be attributed to the relaxation of elastic stress due to the onset of plastic flow. Therefore, the stress measurements reveal that key role of oxide flow in pore initiation, in agreement with the aforementioned tracer study.4 The present experiments suggest that the self-ordered pore array initiates due to plastic yielding of the oxide. The compressive stress buildup to the yield stress is possible because of the initial morphological instability, which seems to establish a layer of constant thickness within which compressive stress accumulates during barrier oxide growth. Without this layer, stress would be dispersed through the thickness of the growing film, and the stress level would not change during as the barrier oxide thickness increased. The reason for localized stress accumulation will be explored using a mathematical model for stress evolution during barrier oxide growth.8 AcknowledgmentSupport was provided by the National Science Foundation CMMI-100748.