What's New and Different for EUV Masks?

[Daniel Nenni]

Tom Faure from GF does a nice overview of the EUV value proposition, challenges, and the GF planned implementation. This is from the eBeam initiative. Tom came to GF from the IBM Semiconductor acquisition and is an expert in this field, absolutely.

[video=youtube;_UxDGH8uslM]https://www.youtube.com/watch?v=_UxDGH8uslM[/video]

About the eBeam Initiative
The Initiative provides a forum for educational and promotional activities regarding new design-to-manufacturing approaches that help reduce mask costs for semiconductor devices based on electron beam (eBeam) technologies.

 

[Fred Chen]

The new aspect of EUV illumination is that there is different shadowing across the slit. This cross-slit shadowing variation cannot be corrected by mask OPC alone because it's a 3D effect not represented by 2D shapes; the source illumination shape (weights at different positions) also needs to be changed for source-mask optimization (SMO). This is further aggravated by aberration variations across slit. This means different illuminations at different slit positions, but this can only be carried out by different exposures.

 

[paulhylander]

When the presenter talks about data challenges (near the end of the presentation), he mentions he doesn't think they will be problems. It was unclear to me what these challenges are. Does anybody understand this part?

 

[Fred Chen]

When the presenter talks about data challenges (near the end of the presentation), he mentions he doesn't think they will be problems. It was unclear to me what these challenges are. Does anybody understand this part?

He is referring to EUV-specific effects: flare, slit corrections, proximity corrections from ebeam scattering in the multilayer, as well as EUV assist features, and the preparation for multi-beam tools.

Since this is a mask-centered talk, the source optimization has been ignored completely, but SMO is definitely a must, as already mentioned by ASML, IMEC, etc. OPC has always been dependent on illumination angular distribution and it's now known it varies across the slit as well, the illumination must be included not only to compensate shadowing but also aberrations. Significant nm-scale aberrations are also thermal in origin (mirror heating), so OPC cannot address these real-time effects.

 

[Fred Chen]

Another perspective of difference for EUV masks is the pellicle. I will update about that shortly.

 

[Fred Chen]

Mask vertical displacement, flatness, local slope

An issue that has been remaining relatively quiet all this time is EUV mask flatness, but this can be generalized to consistent EUV mask feature vertical placement.

View attachment 21878
Due to the off-axis 6-degree illumination, the EUV mask feature must be consistently placed vertically as it loaded onto the chuck. For a +/-1 nm wafer position tolerance, the allowed vertical range is roughly +/-40 nm. Besides the repeatability between different mask loads, another source of variability comes from deformation on the chuck itself, as well as blank thickness variations up to over 70 nm (~2 nm pattern shift on the wafer).

View attachment 21880
Bending also produces different reflected angles, whose differences are magnified 4x by the scanner, resulting in another shift. This translates to a very tight mask flatness spec.

View attachment 21884
We might anticipate a special procedure necessary for loading different EUV masks, to check for the pattern shift.

References:

- T. Schmoeller, T. Klimpel, I. Kim, G. F. Lorusso, A. Myers, R. Jonckheere, A. M. Goethals, K. Ronse, Proc. of SPIE vol. 6921, 69211B (c) 2008 SPIE.
- ISDP12 Keynote: B. J. Lin, Scope and Limit of Lithography to the End of Moore's Law
- S. Yoshitake, H. Sunaoshi, S. Tamamushi, S. Mitsui, M. Ogasawara, EUV Mask Flatness Requirements: E-beam Mask Writer Supplier Perspective.
- X. Chen, C. Turley, J. Rankin, T. Brunner, A. Gabor, Proc. of SPIE vol. 10143, 101531F (c) 2017 SPIE.

 

[Fred Chen]

At the 2018 SPIE Advanced Lithography conference earlier this year, there was a paper on heating of EUV masks causing deformations. For a 90 WPH tool case at 250 W (~35 mJ/cm²), the temperature rise approached 10 deg C, but deformation was tens of nm. Heavier absorption materials such as Ni instead of Ta aggravated the situation. This new issue (actually related to the general thermal aberrations issue) will make solving 3D effects and stochastic effects more difficult.

Reference: "Thermomechanical changes of EUV mask and absorber dependency", Chung-Hyun Ban, Sung-Gyu Lee, Eun-Sang Park, Jae-Hun Park, Hye-Keun Oh, Proc. SPIE 10583, Extreme Ultraviolet (EUV) Lithography IX, 1058322 (c) 2018 SPIE.

Throughput vs power/dose given in I. Fomenkov et al., "Light sources for high-volume manufacturing EUV lithography: technology, performance, and power scaling," Adv. Opt. Tech. 6, 175 (2017).