Mechanism of Roughness Generation in Resists
Toru Yamaguchi and Hideo Namatsu
Device Physics Laboratory
Ā@Resist patterns less than 10 nm wide are indispensable for the fabrication of nano-structures, such as single-electron transistors. To form these patterns with high accuracy, we must reduce the edge roughness of resist patterns. We have already clarified that the roughness is generated through a dissolution process, which we call aggregate extraction development. Polymer aggregates 20-30 nm in size are naturally contained in resist films. These aggregates are extracted one by one during development. They appear on the pattern surface, causing roughness. The aggregate extraction development was found to strongly depend on the size of developer molecules. Thus, we can suppress it by using a developer molecule of optimum size.
Ā@Figure 1 shows atomic force microscope (AFM) images of a resist surface during development. Surface morphology differs largely with varying developer molecule size. For hexyl acetate, which is usually used for high-resolution pattern formation, many aggregates appear on surface [Fig. 1(c)]. For ethyl acetate, however, hardly any aggregates are observed [Fig. 1(a)].
Ā@In the dissolution of resist polymer, penetration of developer molecules into polymer matrix is a limiting step. Developer molecules penetrate through voids, so-called free volume, which are regions not occupied by polymer molecules. Therefore, the solubility of polymer aggregates is determined by the relative relationship between the size of voids in aggregates and molecular size of the developer. For a large developer molecule, such as hexyl acetate, polymers surrounding the aggregates dissolve first because the developer molecules easily penetrate the surrounding polymers, but they have difficulty penetrating the aggregates themselves. This is due to the difference in void size between the surrounding polymers and aggregates. As a result, the aggregate extraction development occurs and the surface becomes rough. For a small developer molecule, such as ethyl acetate, on the other hand, the aggregates themselves dissolve because developer molecules can easily penetrate them. As a result, the dissolution proceeds at the molecular level, not by aggregate extraction, and the surface becomes flat. For butyl acetate, the situation is in between because its size is between that of ethyl and hexyl acetate. These results indicate that there is an optimum developer molecule size that will reduce resist roughness by suppressing aggregate extraction development.
 T. Yamaguchi et al., Appl. Phys. Lett. 71 (1997) 2388.
 T. Yamaguchi et al., Proc. SPIE 3333 (1998) 830.
Fig. 1. AFM images of resist surface during development
Developer: (a) ethyl acetate, (b) butyl acetate, (c) hexyl acetate