Lithography Simulation Tools Needed for 22nm HP and Beyond. Chris Mack

Transcription

1 Lithography Simulation Tools Needed for 22nm HP and Beyond Chris Mack

2 Slicing the Pie Simulation Tool Characteristics Precision Accuracy Capabilities (speed, features) Simulation Tool Categories Full Physical Models (small area) Approximate Models (full chip) SEMATECH Litho Forum

3 Simulation Precision Precision also called numerical accuracy Trade-off with speed Varies depending on simulation parameters Measured by comparing with slow, small-grid simulation Precision spec same as metrology tool P/T of Since T is about ±10% of nominal CD, required precision is about ±1 2% of CD Ex: 22nm HP, precision is nm SEMATECH Litho Forum

4 Example Image CD 3.5 Error in Image CD (%) Number of Grid Points across Pupil Diameter SEMATECH Litho Forum

5 Simulation Precision Simulation precision of 1% over a wide range of simulation conditions is very challenging for simulation speeds that aren t painful Low NILS cases have worst CD precision No standard protocol for measuring simulator precision How do you do it? SEMATECH Litho Forum

6 Simulation Accuracy Two kinds of accuracy Comparison with analytical solutions (a few special cases) Comparison with experimental data Accuracy vs. Analytical Solutions Mark D. Smith and Chris A. Mack, Methods for Benchmarking Photolithography Simulators, Optical Microlithography XVI, Proc., SPIE Vol (2003) pp M. D. Smith, J. D. Byers, and C. A. Mack, Methods for Benchmarking Photolithography Simulators, Part II, Optical Microlithography XVII, Proc., SPIE Vol (2004) pp Mark D. Smith, Trey Graves, Jeffrey D. Byers, Chris A. Mack, Methods for Benchmarking Photolithography Simulators: Part III, Optical Microlithography XVII, Proc., SPIE Vol (2005) p. p Trey Graves, Mark D. Smith, Chris A. Mack, Methods for Benchmarking Photolithography Simulators: Part IV, Optical Microlithography XIX, Proc., SPIE Vol (2006) p X. SEMATECH Litho Forum

7 Simulation Accuracy Accuracy vs. Experimental Data Problem: How accurate is your experimental data? CD Metrology tools have no specs for accuracy CD SEM vs. other tools shows 5 15 nm offsets, but these offsets are a function of pattern shape (tone, sidewall angle, pitch) Top-down line-end and shape measurements are even more problematic What are the accuracy requirements for simulation? Most aggressive: same requirements as for precision (this makes simulation far more accurate than measurements!) Note: simulation accuracy includes numerical precision errors SEMATECH Litho Forum

8 Simulation Capabilities New features required for new processes Example: PROLITH History First chemically amplified resist model (v ) Phase shift masks (v ) Flex (v ) Off-axis illumination (v ) 2D masks and aerial images (v ) Full 3D simulations (v ) Etch simulator (v ) Immersion lithography (v ) SEMATECH Litho Forum

9 Simulation Capabilities New features required when 2 nd order effects become first order High NA Scalar model (v ) Vector imaging model (v ) Aberrations (v ) Variable diffusivity reaction-diffusion in 3D (v ) Step-and-scan vibrations, aberration averaging (v ) Notch development model (v ) Byers/Petersen chemically amplified model (v ) Base quencher diffusion (v ) Mask topography EMF simulator (v ) Birefringence/Jones Pupil (v ) SEMATECH Litho Forum

10 Requirements for 22nm HP (example/proposal) Precision: 1% Accuracy: 2% Full Physical Models have the added necessity for meeting this accuracy spec over a wide range of parameters without recalibration New Capabilities Needed: Wafer topography for double patterning LER models Metrology simulation Mask error model Resist variation through thickness Is etch or sidewall spacer simulation needed? New freeze/dual tone/2-photon materials? Faster! SEMATECH Litho Forum

11 Summary of Results: Pi06-001ME Calibration and Portability Testing KLA Data Focus Exposure Matrix PROLITH Model RMSE, nm Calibrate 70 nm Line, 140 nm Pitch nm Line, 250 nm Pitch nm Trench, 600 nm Pitch 1.36 Overall Model Calibration RMSE 1.0 nm Exposure Focus Exposure Matrix PROLITH Model RMSE, nm Validate 0.85NA DiP 0.85NA DiP 0.85NA DiP 0.85NA cquad 0.85NA cquad 100 nm Line, 200 nm Pitch 110 nm Line, 250 nm Pitch 150nm Line, 600 nm Pitch 100 nm Line, 200 nm Pitch 110 nm Line, 250 nm Pitch NA cquad 0.75NA Ann 150nm Line, 600 nm Pitch 100 nm Line, 200 nm Pitch NA Ann 0.75NA Ann 110 nm Line, 250 nm Pitch 150nm Line, 600 nm Pitch NA cquad 1.20NA Ann 45 nm Line, 90 nm Pitch 55 nm Line, 110 nm Pitch NA Ann 1.20NA Ann 80 nm Line, 200 nm Pitch 100nm Line, 400 nm Pitch Overall Model Validation RMSE 1.3 nm 15 PROLITH Technical Forum SCJ 2008 SEMATECH Litho Forum

12 Simulation Vs Data - 60nm DHP 1.2NA, 0.80/0.50 Ann, X/Y Pol. 6% AttPSM FOUP AR070012, SLOT 13, Column 0, Row 0 30 mj/sqcm, 0.0µm Focus, Layer Q1 1 st Litho (pi06) Measured CD: 46.2nm Simulated CD: 44.7nm 17 5/14/2008 Robertson, Biafore, Graves & Smith, Rigorous Physical Modeling of a Materials-Based Frequency Doubling Lithography Process, SPIE 2008 SEMATECH Litho Forum

13 Simulation Vs Data 36nm DHP, Calibrated Pi6 FPM 1.2NA, 0.80/0.50 Ann, X/Y Pol. BIM First Pass Measured CD: 37.5nm Simulated CD: 37.7nm 1 st Litho after 1 st Dev 1 st Litho after 2 nd Dev 18 5/14/2008 Robertson, Biafore, Graves & Smith, Rigorous Physical Modeling of a Materials-Based Frequency Doubling Lithography Process, SPIE 2008 SEMATECH Litho Forum

14 Conclusions Simulation precision and accuracy scale with feature size (0.7X per node), while speed must stay constant or improve This improvement doesn t come for free Many new capabilities are needed Several features require fundamental research Who s working on these? SEMATECH Litho Forum