As the minimum feature size of electronic devices shrinks to less than 0.25 μm, it is critically important that we reduce the defects that occur in lithography processes. Moreover, as the defects to be controlled become ever smaller, this makes them increasingly difficult to detect by conventional fault detection equipment. In order to detect these minute defects in the context of shrinking device geometries, it is essential that we develop a clear understanding of the behavior of micro defects in developer. In principle, there are three ways in which these defects might be dealt with: (1) defects can be prevented from occurring in the first place, (2) defects can be prevented from adhering to the device, or (3) defects can be eliminated after they occur. Our recent work has mainly been concerned with the first and most effective approach of preventing defects from occurring in the first place, and this motivated the present study to investigate the mechanisms by which defects occur. We believe that defects occur in chemically amplified (CA) resists that are insufficiently unprotected at boundary regions between unexposed and exposed areas or in unexposed areas, so that the de-protection reaction in the resist suns to different degrees of completion due to varying exposure doses. In this study we investigate the number of defects in various developers, and determine the size distribution of the defects. Based on analysis of the behavior of defects from their size distribution in develop we conclude that: (1) the size of defects increases when the exposure dose is reduced by appropriate Eops, (2) defects originate in the boundary area between unexposed and exposed areas, and (3) a portion of CA resist polymer that is insufficiently deprotected is dispersed in the developer, coalesces and is deposited in a form that is not very soluble, and is manifested as relatively large particle defects.