1. TNO TPD TNO Science and Industry, 12 May 20051 Simulation of processing Influence of secondary electron statistics More Moore SP3WP6.4

2. TNO Science and Industry, 12 May 20052 Simulations of processing Rough aerial image in resist represented with white noise PEB acid diffusion represented by ‘blurring’ edge outward Subsequent etch represented by ‘blurring’ edge inward DPS 2004, Leunissen et al.

3. TNO Science and Industry, 12 May 20053 Animation of acid diffusion 0 to 10 nm Simulations of processing High frequency components mitigated by acid diffusion

4. TNO Science and Industry, 12 May 20054 Animation of acid diffusion 0 to 30 nm Simulations of processing Decrease of high frequency components reduces total spectral power

5. TNO Science and Industry, 12 May 20055 Simulations of processing Simulation matches shape familiar from typical experimental results Power Spectrum y ~ x -1.46 1101001000 litho(30nm) white noise edge

6. TNO Science and Industry, 12 May 20056 Animation of etch 0 to 30 nm Simulations of processing Not so much influence on spectral power since high frequency components are already ‘blurred away’

7. TNO Science and Industry, 12 May 20057 Simulations of processing LER decreases during etch, but it depends on amount of LER that is already ‘blurred away’ during acid diffusion

8. TNO Science and Industry, 12 May 20058 h < 20 eV h > 20 eV Statistics of secondary electrons Secondary electron blur Super-Poissonian noise DUV lithography: pure photochemistry EUV lithography: secondary electron generation

9. TNO Science and Industry, 12 May 20059 Statistics of secondary electrons Model predicting SE blur from excitations, ionizations, plasmon loss and elastic scattering (MNE 2004) Mean number of chemical activation events:  AE  3 (S. Tagawa: 1~3 e.g. MNE 2004) "single photon noise" 051015 # chemical excitation events MC simulation Poisson distribution  AE =3.2

10. TNO Science and Industry, 12 May 200510 Statistics of secondary electrons Resist sensitivity according to the ITRS Litho Roadmap EUV: 2 – 15 mJ/cm 2  1.4 – 10  /nm 2 Shot noise induced dose variation can be reduced using higher dose Statistics in chemical activation events leads to “effective dose variations”: For  AE =2, effective dose variations have to be compensated with 50% higher dose

11. TNO Science and Industry, 12 May 200511 Determination of dose induced LER vary dose / line width Need high contrast and good dose control: preferably MET

12. TNO Science and Industry, 12 May 200512 Experimental artifacts Environmental vibrations Source uniformity (temporal) Source uniformity (spatial) Optics / mask errors CD measurement Resist contrast repeat exposures with different illumination time; source characteristics repeat with different masks repeat CD measurements repeat exposures with different resist contrast

13. TNO Science and Industry, 12 May 200513 Procedure to correct for artifacts Line edge Y i,j = A i + B j + E i,j i is index of photo j is index of line within a photo A is measure for variation due to SEM imaging B is measure for variation due to mask error, source uniformity E is measurement error and shot noise E = dose dependent noise + residual noise This elaborate procedure proves necessary at the desired extreme sensitivities, to discriminate true shot noise from experimental artifacts

14. TNO Science and Industry, 12 May 200514 Experimental setup Multiple CDs on mask Multiple gratings Multiple wafer positions Multiple illuminations Multiple photon energies (for SE blur measurement) Multiple CD SEM imaging CD redundancy mask errors multiple illumination wafer / PEB variations

15. TNO Science and Industry, 12 May 200515 Summary Transfer of LER during etch better understood Additional effective “dose variation” due to statistics in chemical activation  possible point of concern Determination of dose induced high frequency LER could be seriously hampered by experimental artifacts