EUV Resist Performance under High Stray Light Levels: an Interference Lithography Study Roel Gronheid 1, Alan M. Myers 1,2, Frieda Van Roey 1, Harun H. Solak 3, Yasin Ekinci 3, Tom Vandeweyer 1, Anne-Marie Goethals 1 1 IMEC, Leuven, Belgium 2 On assignment from Intel 3 Paul Scherrer Institut, Villigen, Switzerland
Outline Introduction Experimental Method State-of-the-art EUV Resist Performance Effect of Flare on Resist Performance Exposure Latitude and Profiles Line Width Roughness Conclusions imec 2006 2
Introduction Motivation Flare scales with 1/λ 2 and is therefore a very serious issue for EUV Tight roughness specifications on mirrors in MSFR are needed for flare control DUV Out-of-Band radiation also results in increased effective flare imec 2006 3
Experimental Method - The Interferometer beamline Exposure chamber Diffraction Gratings Thin membrane is only optical element: negligible intrinsic flare imec 2006 4
Outline Introduction Experimental Method State-of-the-art EUV Resist Performance Effect of Flare on Resist Performance Exposure Latitude and Profiles Line Width Roughness Conclusions imec 2006 5
State-of-the-art EUV Resist Performance EUV-38 40nm L/S 32.5nm L/S 30nm L/S 25nm L/S 15.5mJ/cm 2 Some EUV materials start to demonstrate sub-32nm resolution!!! imec 2006 6
Outline Introduction Experimental Method State-of-the-art EUV Resist Performance Effect of Flare on Resist Performance Exposure Latitude and Profiles Line Width Roughness Conclusions imec 2006 7
Experimental Method Flare Exposures 800 μm 12 stnd Exposure: Pattern Flare 50nm 45nm 40nm 32.5nm 60 μm shift 30nm 25nm Multiple pitches are simultaneously printed in a single exposure field 0, 10, 20, 30, 40% flare added Note: In this presentation flare is defined as I min /I max *100% Flare is treated as a DC offset to the aerial image imec 2006 8
Calibration of mask efficiency The mask efficiency is determined by measuring the relative intensity of the 1st and 0th order diffracted light with a CCD camera. With this information the required relative dose to obtain a certain flare level can be determined imec 2006 9
Flare Results in Reduced Aerial Image Contrast Interference Printer: Aerial Image is pitch independent Flare = 0% 10% 20% 30% 40% NILS=3.1 NILS=2.6 NILS=2.1 NILS=1.7 NILS=1.3 EUV Scanner (0.25NA and σ=0.5): Aerial Image depends on feature size 50nm L/S, 0% flare 32nm L/S, 0% flare 32nm L/S, 10% flare NILS=7.8 NILS=4.1 NILS=3.0-50 0 50-32 0 32-32 0 32 imec 2006 10
Contrast Demodulation at PSI EUV-16 Flare = 0% 10% 20% 30% 40% 50nm Not Available 45nm imec 2006 11
Prolith Simulations Flare level: 0% 10% 20% 30% No chemical surface contamination Flare level: 0% 10% 20% 30% With chemical surface contamination imec 2006 12
Effect of amine contamination EUV-18 PEB under N 2 flow 50nm L/S 0% flare 30% flare 40% flare PEB under ambient conditions 0% flare 20% flare 40% flare dose 8.39 dose 9.07 imec 2006 13
Exposure Latitude EUV-30 50nm L/S CD (nm) 75 70 65 60 55 50 45 40 35 0% flare 10% flare 20% flare 30% flare 40% flare Image blurring due to acid diffusion??? 30 0 5 10 15 20 dose (mj/cm2) 0 % Flare 10% Flare 20% Flare NILS 50nm 45nm 40nm 3.1 34% 35% 24% 2.6 26% 20% -- 2.1 22% -- -- ~ 10xNILS imec 2006 14
Exposure Latitude EUV-38 NILS 50nm 45nm 40nm 0 % Flare 3.1 31% 33% -- 10% Flare 20% Flare 2.6 21% 22% -- 2.1 -- 17% -- EUV-44 NILS 50nm 45nm 40nm 0 % Flare 3.1 24% 24% 23% 10% Flare 20% Flare 2.6 26% 19% 17% 2.1 15% 13% -- EUV-44 has less maximum contrast, but is also less affected by flare increase imec 2006 15
Profiles for 50nm L/S EUV-30 0% 20% 30% 40% Increased flare level results in top-loss, footing and increased LER The profiles of 50nm L/S patterns remains reasonable up to 20% flare imec 2006 16
Profiles for 50nm L/S EUV-37 0% 20% 30% 40% imec 2006 17
Profiles for 45nm and 40nm L/S EUV-37 45nm L/S 0% 20% 30% 40% 40nm L/S 0% 20% Latitude disappears rapidly for smaller dimensions imec 2006 18
Line Width Roughness EUV-30 16 14 50nm L/S 16 14 45nm L/S 12 12 LWR (nm) 10 8 6 4 2 0 0% Flare 10% Flare 20% Flare 30% Flare : CD on target 0 5 10 15 LWR (nm) 10 8 6 4 2 0 0% Flare 10% Flare 20% Flare 0 5 10 15 Dose (mj/cm2) Dose (mj/cm2) LWR is mainly governed by shot noise statistics and image contrast Lowest LWR is achieved at values above the polymer grain size (2-3nm) No large LWR improvement should be expected from smaller grain size of molecular resists at these image contrast and dose levels There may be improvement due to better material homogeneity Only escape possible by improving chemical contribution to shot noise: increase quantum yield and/or absorbance imec 2006 19
Line Width Roughness EUV-44 LWR (nm) 14 12 10 8 6 4 2 0 Flare 0% Flare 10% Flare 20% 50nm L/S : CD on target 0 5 10 15 20 25 Dose (mj/cm2) 14 LWR (nm) 12 10 8 6 4 2 0 Flare 0% Flare 10% LWR (mj/cm2) 40nm L/S 14 12 10 8 6 4 2 0 Flare 0% Flare 10% Flare 20% 0 5 10 15 20 25 Dose (mj/cm2) 45nm L/S 0 5 10 15 20 25 Dose (mj/cm2) imec 2006 20
Conclusions Current EUV resists can handle very high flare levels relatively well at 50nm half pitch At 40nm half pitch the flare tolerance for exposure latitude and profiles is significantly reduced Likely, at these dimensions acid diffusion induced image blur kicks in LWR increases significantly with increasing stray light for all dimensions These results cannot be quantitatively translated to EUV scanners Flare will depend on mask layout NILS depends on CD and may be higher than for interference imaging when multiple orders are captured in the pupil Trends as seen in this study should be similar on a scanner At 32nm half pitch (NILS=4.1 for 0.25NA; σ=0.5 and 0% flare) probably little flare tolerance can be accepted to maintain imaging performance in terms of Exposure Latitude, Profiles and LWR imec 2006 21
Acknowledgements Part of this work was funded by the European Commission through the More Moore project. imec 2006 22
imec 2006 23