Understanding Acid Diffusion in Chemically Amplified Photoresists


Michael D. Stewart, Gerard M. Schmid, Sean D. Burns, & Hoang Vi Tran

Problem and Motivation

The current mainstay of the semiconductor industry is deepultraviolet (DUV) wavelength lithography using photoresists designed on the principle of chemical amplification. Chemically amplified (CA) photoresists have played a major role in semiconductor microlithography over the last decade. All foreseeable future lithography technologies (e.g. EUV, SCAPLE) will likely also require CA resists. It is important to have a clear understanding of how these CA resists function so that both resist designand processing can be improved to meet the challenges of sub-100 nmlithography.
The Willson Group has individual projects studying almost all aspects of CA resist processing from exposure to development, but the work discussed in this section is on the very important post exposure bake (PEB)step.
In a chemically amplified photoresist, the solubility-switching chemistry necessary for imaging is not caused directly by the exposure; rather exposure generates a stable catalytic species that promotes solubility-switching chemical reactions during a subsequent baking step. The term “chemical amplification” arises from the fact that each photochemically generated catalyst molecule can promote many solubility-switching reaction events. The apparent quantum efficiency of the switching reaction is the quantum efficiency of catalyst generation multiplied by the average catalytic chain length. The original exposure dose is “amplified” by the subsequent chain of chemical reaction events. The catalytic chain length for a catalyst can be very long (up to several hundred reaction events) giving dramatic exposure amplification.
Chemical amplification is advantageous in that it can greatly improve resist sensitivity, but it is not without potential drawbacks. For instance as a catalyst molecule moves around to the several hundred reactions sites, nothing necessarily limits it to the region that was exposed to the imaging radiation. There is a potential trade-off between resist sensitivity and imaging fidelity. Figure 1 below demonstrates the problem. The catalyst molecules are depicted as acids, which is the typical catalyst type.
1. Generation of Latent Acid
The CA photoresist is exposed through a photomask, generating acid catalyst in the exposed regions.
2. PEB Step
The latent acid image generated in the first step is converted in to an image of soluble and insoluble regions by raising the temperature of the wafer, which allows chemical reactions to occur. Some acid migrates out of the originally exposed region causing "CD bias" problems.
3. Image Development
After baking the image is developed with a solvent. The developed feature width is larger than the nominal mask dimension as the result of acid diffusion from exposed into the unexposed regions. (In this case imaging is shown as positive tone.)
Figure 1: The Acid Diffusion Problem

For much of the history of CA resists this trade-off was of little concern as the catalyst diffusion distances were insignificant relative to the printed feature size, but as feature sizes have decreased, the diffusion distances have remained roughly the same and catalyst diffusion has emerged as a significant concern.
Our work is geared towards developing a fundamental understanding of acid diffusion so that:
  1. Process conditions can be found to minimize diffusion
  2. New resist materials/formulations can be developed to limit intrinsic diffusion
  3. Lithography simulators can model catalyst migration as realistically as possible

Experimental

Our group uses several methods to study diffusion in photoresists. Equipment used includes infrared spectrometers,UV spectrometers, scanning electron microscopes, and spectroscopicellipsometers amongst other things. We also have access to the microlithography facilities of SEMATECH located here in Austin.