Paper PS+SE-MoA4
Plasma Smoothing of Extreme Ultraviolet Photoresist: LWR Reduction at 30nm Half Pitch
Monday, October 31, 2011, 3:00 pm, Room 201
Session: |
Advanced FEOL / Gate Etching II |
Presenter: |
Efrain Altamirano-Sanchez, IMEC, Belgium |
Authors: |
E. Altamirano-Sanchez, IMEC, Belgium A. Pret Vaglio, IMEC, Belgium R. Gronheid, IMEC, Belgium D. Marc, IMEC, Belgium W. Boullart, IMEC, Belgium |
Correspondent: |
Click to Email |
Over the past years a tremendous amount of effort has been put on Extreme Ultraviolet lithography (EUVL) for printing the 16nm node. In 2010 imec’s EUV alpha demo tool printed for the first time a 16nm node SRAM cell using state-of-the-art EUV photoresist (PR). In order to meet the line width roughness (LWR) requirements for the 16nm node, we have investigated plasma smoothing techniques on 30nm half pitch lines after exposure.
In this contribution we will report our findings on EUV PR plasma treatment (PT) using H2, Ar and HBr for LWR reduction; and subsequent, in-situ PR encapsulation for preserving the improved LWR during subsequent pattern transfer. Currently we are investigating the vacuum UV role during the H2 PT using MgF2 windows. In parallel, we are characterizing a PR resist encapsulation carried out in-situ (in the etching chamber). The encapsulating layer is characterized through various analytical techniques, such as: XPS, Ellipsometry, mass metrology and TEM. These characterizations will provide understanding of how the H2 plasma improves the LWR and of how the PR encapsulation preserves the PR pattern allowing a straight HM patterning profile.
This study was carried out on 300 mm silicon wafers with the following patterning stack, from top to bottom: 50 nm PR/20 nm under layer (UL)/15 nm SiOC/40 nm amorphous carbon layer (ACL). The dry etching was carried out in a ICP like reactor from Lam research (Kiyo C reactor TCPTM)
Power spectral density (PSD) analyses have shown that Ar and HBr plasmas do not improve the LWR of EUV PR as they do on 193i PR. Using the correct conditions in the ICP reactor, H2 plasmas can improve LWR by ~30% without CD bias; in other words no PR reflow was detected.
After the H2 plasma treatment, the pattern was transferred into the UL using a CH2F2/CF4/O2 gas mixture. This chemistry provides on the one hand a high PR passivation but on the other hand induces a PR pattern degradation, which is translated into higher LWR. A novel alternative for maintaining the improved LWR after H2 PT was to deposit in-situ (in the ICP reactor) a silicon containing layer that encapsulates the PR and preserves the improved LWR when the pattern is transferred into the UL, the SiOC and the ACL.