The future of EUVL by Winfried Kaiser, Udo Dinger, Peter Kuerz, Martin Lowisch, Hans-Juergen Mann, Stefan Muellender, William H. Arnold, Jos Benshop, Steven G. Hansen, Koen van Ingen-Schenau Outline Introduction Imaging requirements of future nodes High optics Infrastructure and Technology Summary For public use Seite
Status Alpha Demo Tool nm 4 mj/cm nm 4 mj/cm nm 4 mj/cm nm 4 mj/cm st scanner able to print dense features in single exposure See also: Emerging Lithographic Technologies XII Hans Meiling: Field performance of the EUV alpha demo tools Tuesday, 6 February 8, : PM :5 PM For public use Seite Why EUV? Resolution, Shrink k (nm) 8 AT: 6 XT:4 XT:7i 4 Immersion ASML Product XT:9i Introduction Double patterning NEXT EUV EUV 4 5 6 7 8 9 4 Note: Process development.5 ~ years in advance / updated /7 Year of Production Start* For public use Seite 4
Why EUVL: nm SRAM approaches cell area.6u approx SE limit with.5 and standard design standard layout (optimized).5 more DFM? asym shrink?? more DFM? DPT, PT, 4PT?? Best -way split k=. k=. higher SE or EUV.7 9nm SE.u k relative to.9u target Best 4-way split 8 nm 4 mj/cm nm node target range k=5 k= k=.48 k=.48.5 EUV.6u EUV allows the patterning of complex, dense features in a single exposure step! For public use Seite 5 Outline Introduction Imaging requirements of future nodes High optics Infrastructure and Technology Summary For public use Seite 6
EUV Optics: The future EUV is introduced as a high k technology Node \.5..45 k reduction nm 9.76.7 nm.4.7 6 nm..8 nm..6.7 RES = k λ opportunity EUV will follow the optical path with high and k reduction For public use Seite 7 nm features Dense Lines conventional illumination Dense contacts conventional illumination rtial coherence Partia al coherence.8.7.6.4..8.7.6 best focus...4.5.5.5.5 NILS 75 nm defocus.8.7.6.4....4.8.7.6.5.5.5.5 NILS Par.4.4....4....4 Flare: 6%, WFE: nm rms NILS target: For public use Seite 8 4
nm features Dense Lines conventional illumination Partial coherence.8.7.6.4 best focus.5.5 NILS.8.7.6.4 5 nm defocus.5.5 NILS....4....4 Dense contacts conventional illumination Sigma center σ A.8.7.6.4.5.5.8.7.6.4.5.5....4....4. enables nm patterning with conventional illumination For public use Seite 9 6 nm features Dense Lines Sigm ma center σ A annular illumination.6.8.7 best focus.5 5.5 NILS 5 nm defocus.8.5.7.6 5.5 NILS σ A.4.4....4....4 Dense contacts quasar illumination a center σ Q.8.7.6.5.5.8.7.6.5.5 Sigm σ Q.4.4.....4...4 and even 6nm with off axis illumination For public use Seite 5
nm features Dense Lines Sigma center σ A annular illumination.6 σ A.8.7.4 best focus.5.5 NILS.8.7.6.4 4 nm defocus.5.5 NILS Dense contacts quasar illumination σ Q Sigma center σ Q..8.7.6.4.4.6.5.5..8.7.6.4.4.6.5.5..4.6..4.6 nm can be imaged with >.45 and off axis illumination For public use Seite Outline Introduction Imaging requirements of future nodes High optics Infrastructure and Technology Summary For public use Seite 6
The path to =. Enabling for higher : Slit length 6mm MAG = 4x Larger mirror sizes Stronger aspheres.5. Full field 6 mirror designs can be extended to s around. For public use Seite Apodization is limiting 6M designs center of field (x=) edge of field (x=mm) The larger introduce a high angular load on surfaces which cause significant apodisation effects Balancing of optical and coating design is needed to achieve a (quasi) rotational symmetric apodisation uniform over the field For public use Seite 4 7
>.4 : Way out with central obscuration full field M5 reduced field M5 Central obscuration solves the apodisation issue but limits the field size. Full field designs show big central obscurations. In addition stopping down increases the obscuration ratio. For public use Seite 5 Imaging effect of apodisation and central obscuration., Annular.6-.8 6nm feature size Model pupil: / / Transmission % Area with reduced transmission/obscuration The prime effect of apodisation and central obscuration can be compensated by CD biasing For public use Seite 6 8
Alternative with >.4 8M designs allow unobscured full field systems with. The two additional mirrors cause a reduction of system transmission by at least a factor of. For public use Seite 7 Extending to.7.7 For.7 the only design solutions are obscured 8M systems For public use Seite 8 9
High solution roadmap Solution overview:. 5 7.7 6M 8M unobscured central obscured (smaller fields) There are design solutions for high systems enabling nm and beyond For public use Seite 9 EUVL Roadmap down to nm Res 4 5 nm 6nm nm same lens, enhanced off axis illumination., nm OVL, >wph.4x. +off axis illumination 7nm.5, 4nm OVL For public use Seite
Outline Introduction Imaging requirements of future nodes High optics Infrastructure and Technology Summary For public use Seite EUV Coatings Z-graded multilayers Reflectivity.8 R (-8 )=7%.7.6 R (-8 )=64% R =6%.4 R =58%... period. ML. 6 9 5 8 4 Incidence angle, degrees Layer thickness, nm 8 d 6 4 d Si d Mo 4 6 8 Bi-layer number Z-grading of EUV multilayers improves the angular acceptance on cost of peak reflectivity and therefore system transmission For public use Seite
Progress in flare reduction ades) MSFR [nm rms] (evaluated over 4.6 deca,55,5,45,4,5,,5,,5,,5 MET test mirrors mirror on-axis Set Set Set test mirror (MSFR opt.) AD-tool 6 mirrors off-axis setup POB 6% flare tools Pre production tool 8% flare Production tools 4 5 6 7 8 9 4 5 Status: Flare Level < 8% Figure =.4 nm rms MSFR =. nm rms HSFR =.7 nm rms Target: % Flare (6M) Flare is calculated for a µm line in a bright field Flare will be improved further to secure high imaging performance for smaller feature sizes For public use Seite Impact of CRAO change () Chief Ray Angle on Object side (α) EUV optics has to be non-telecentric on reticle side. Actual standard is 6, this limits the to <.4 (Mag 4x). For larger s the CRAO has to be increased accordingly. α Example: HV mask bias (nm) required H 8 7 6 5 4 5 6 7 8 9 CRAO =.45, annular illumination Mask stack wih 67nm TaN absorber For public use Seite 4
) Impact of CRAO change () Att PSM Stack (nm) 7 Required delta Required bias delta bias HV HV Source: Samsung 7nm TaN (magenta) nm Ru (green) nm Si (blue) 4 MoSi bilayer bias (nm 6 5 4 6 7 8 9 CRAO (deg) nm DL required delta bias HV =.45 Quasar The impact of a change of CRAO is not well understood today in full extent. Our view is that there are no showstoppers for an increase to 9 - allowing full field optics with 5 The actual assumption is that there will be mask stacks where on the effect on imaging by larger CRAOs can be compensated by biasing without critical impact. The only alternative for >.4 would be to go to higher Mag (5x, 6x, 8x) which would limit the usable field size in dimensions with significant impact on productivity or require larger mask sizes. For public use Seite 5 EUV Source power wph (full field optics) Source: Cymer For public use Seite 6
EUV source power vs. resist sensitivity Power [Watt @ IF IB] 6 4 8 6 4 4 Resist sensitivity [mj/cm^] 8 mirror POB 6 mirror POB @ wph To keep the required source power (at intermediate focus in band) in a realistic range, resist sensitivity must target mj /cm For public use Seite 7 Effective etendue for off axis illumination Conventional Annular Quasar 45 σ out σ out σ in σ out σ in σ out =.8 σ in =. σ out =.8 σ in =.6 σ out =. σ in =.6 Setting Conventional Annular Quasar 45 σ.8.6-.8 6-.6-. relative etendue (@fixed slit size) =.5 =. =.45..6..4.7.4 8.8 6.6 Off axis illumination will reduce the effective etendue the illumination system can accept from the source. Therefore the source etendue covering the full source power has to be small enough to avoid any productivity loss for these settings. Since the etendue grows with, in future larger source etendues can be accepted or more aggressive off axis settings can be used. For public use Seite 8 4
Resist effect on imaging relative contra ast loss..8.6.4.. 7 7 7 half-pitch (nm) flare flare + aberrations flare + aberrations + 5nm resist blur flare + aberrations + nm resist blur flare + aberrations + 5nm resist blur : L/S Assumptions: Flare: 4% system, 5% mask trans Aberration: nm RMS Resist treated as a Gaussian (σ) image blur Relative contrast loss calculated based on MTF contrast Resist diffusion has strongest effect on contrast loss. For pattern transfer of very fine features significant improvement of resist blur (diffusion) to finally 5nm are needed See also: Advances in Resist Materials and Processing Technology XXV K. van Ingen Schenau: Photoresist-induced contrast loss and its impact on EUV imaging extendibility Wednesday, 7 February 8 9:4 AM : AM For public use Seite 9 Outline Introduction Imaging requirements of future nodes High optics Infrastructure and Technology Summary For public use Seite 5
Summary: the extendibility of EUVL There are solutions visible for high design to.7. The challenge will be to find full field designs with optimum transmission to enable high productivity. Off axis illumination will allow reduction of k to ~.4. These together will enable the printing of nm dense features in single exposure mode and even beyond. Improvements in polishing and coating technologies are expected to support this progress. Mask technology has to follow the resolution roadmap accordingly. For larger s the CRAO has to be increased and the layer stack to be adapted. The expectation is that neither increase of Mag and no larger masks are needed. Very important are improvements of resists: Resist sensitivity have to target mj/cm to keep the required source power in a realistic target range of some W @ wph tput. To enable pattern transfer of features down to nm the resist blur (diffusion length) has to be reduced to at least 5nm. For public use Seite ACKNOWLEDGEMENT Thanks to the EUV teams at ASML and Carl Zeiss SMT The activities received funding by the European Commission in the project "More Moore" and by various national European governments including the German Federal Ministry of Education and Research in the program MEDEA+. For public use Seite 6