Plasma-Induced Damage in 3D Structures behind Device Scaling K. Eriguchi, Y. Takao, and K. Ono Kyoto University, JAPAN Acknowledgements: This work was partly supported by JSPS and STARC project in Japan. Plasma R Si Project SHANa Damage O Ar 0.543 nm 2014.05.12 PESM Koji Eriguchi 1
Outline Outline 1. Introduction – Plasma-Induced Damage (PID) 2. A Scenario for PID in Scaled 3D Structures 2-1 PID Models – Straggling & Sputtering 2-2 Molecular Dynamics Simulation 2-3 Electronic State of Defect 3. PID Prediction in 3D Structure 4 . Summary 2014.05.12 PESM Koji Eriguchi 2
Plasma-Induced Damage (PID) Performance Plasma Reliability Yield Variability PID naturally does not scale!! 2014.05.12 PESM Koji Eriguchi 3
PID behind the scaling Auth, VLSI 2012. Mistry, IEDM 2007. Plasma-Induced Physical Damage (PPD) Power consumption increase & Operating speed down!! 2014.05.12 PESM Koji Eriguchi 4
Previous reports on PID Presence of defects S. A. Vitale and B. A. Smith: JVST B 21 (2003) 2205. H. Kokura et al., Proc. Symp. Dry Process, 2005, p. 27. T. Ohchi et al.: Jpn. J. Appl. Phys. 47 (2008) 5324. N. Yasui et al.: Proc. Symp. Dry Process (2007) 195. 2014.05.12 PESM Koji Eriguchi 5
Outline Outline 1. Introduction – Plasma-Induced Damage (PID) 2. A Scenario for PID in Scaled 3D Structures 2-1 PID Models – Straggling & Sputtering 2-2 Molecular Dynamics Simulation 2-3 Electronic State of Defect 3. PID Prediction in 3D Structure 4 . Summary 2014.05.12 PESM Koji Eriguchi 6
PID Range Theory – Planar Device Project Range Stopping Power Project Range Stopping Power dE Plasma ion n S ( E ) n T ( E , p ) 2 pdp ( n e , T e , n ion ) 0 d ion 0 ion dx 0 E ion T: energy transfer Potential-model-dependent p: impact parameter Gate low energy limit : reduced energy E ion 1 s ( ) 0 . 5 B n dam ( x ) n (Wilson model: ~ 0.3) Si Range R p p Straggling Si substrate x x N. Bohr: Mat. Fys. Medd. K. Dan. Vidensk. Selsk. 18 (1948). J. Lindhard et al.: Mat. Fys. Medd. K. Dan. Vidensk. Selsk. 33, 1 (1963). G. Moliere: Z. Naturforschung A2, 133 (1947). W. D. Wilson et al.: Phys. Rev. B 15, 2458 (1977). 2014.05.12 PESM Koji Eriguchi 7
PID Range Theory – Planar Device Plasma ion Si ion radical Mask Damaged layer R A ( E ) Etching p ion ion (Eriguchi et al.) M M 2 Si ion R p p 3 M M R p Si ion (LSS Theory) ex. Si Recess R d p V p p Si 2014.05.12 PESM Koji Eriguchi 8
PID Range Theory – "Straggling" Plasma ion E ion Damaged layer Mask Etching Gate Defect R p & p R p p Si sub. L “Lateral straggling” (a) Planar (1D) (b) Fin-structured (3D) 2014.05.12 PESM Koji Eriguchi 9
PID Range Theory – "Straggling" Mask Etching R A ( E ) (Eriguchi et al.) p ion ion M M 2 Si ion R (LSS Theory) p p 3 M M Si ion K (Furukawa et al.) L p R p p M M 2 L Si ion B ( E ) L ion ion 3 M M Si ion “Lateral straggling” (1) Lateral straggling depends on M ion , E ion , and Si–ion potential. (2) Sidewall etching mechanism is governed, not only by direct ion impact & deposition, but also by THIS STRAGGLING, L ! Damaged layer thickness ~ R p + p (planar), L (3D) 2014.05.12 PESM Koji Eriguchi 10
PID Model – 3D Device Sputtering A E E Plasma sp sp ion th ion Mask ion Si radical L R Lateral straggling Si L p p 2014.05.12 PESM Koji Eriguchi 11
Outline Outline 1. Introduction – Plasma-Induced Damage (PID) 2. A Scenario for PID in Scaled 3D Structures 2-1 PID Models – Straggling & Sputtering 2-2 Molecular Dynamics Simulation 2-3 Electronic State of Defect 3. PID Prediction in 3D Structure 4 . Summary 2014.05.12 PESM Koji Eriguchi 12
Simulation scheme Classical Molecular Dynamics (Ohta and Hamaguchi, JVST 2001.) Si–Si, Cl, O: Stillinger–Weber Phys. Rev. B, Vol.31, No.8, (1985), pp. 5262-5271 Noble gases: Wilson et al. Phys. Rev. B, Vol.15, No.5, (1977), pp. 2458-2468 Interstitial ions Surface layer Local defect dumbbell Si Defect analysis by Si Si Ne Density Functional Theory Starting material 2014.05.12 PESM Koji Eriguchi 13
MD prediction results in Fin Interstitial dumbbell Sputtering Sputtering dumbbell Lateral straggling Interstitial Lateral straggling Si Ar Cl Ar ion 100 eV (normal) Cl ion 100 eV (normal) 2014.05.12 PESM Koji Eriguchi 14
Defect creation in Fin Counting the defects in this region one by one in accordance with the bond order and length. 400 F 300 Ne Counts Ar 200 Xe 100 Cl Kr Br 0 100 200 300 400 500 0 100 200 300 400 500 0 E ion (eV) E ion (eV) An ion with lighter mass and higher incident energy larger damage An ion with lighter mass and higher incident energy larger damage 2014.05.12 PESM Koji Eriguchi 15
Defect creation in "Fin" Fin region 50 eV F 100 eV 30 400 Cl Defect counts Damage layer 200 eV Br 300 400 eV Incident particle counts 25 200 20 100 0 15 25 “fin” Particle counts fin sputtering 10 20 15 body 5 10 5 straggling 0 0 0.0 1.0 2.0 3.0 4.0 5.0 F Cl Br Distance from fin edge (nm) E = 200 eV Both "sputtering" and "straggling" are responsible for PID in 3D. Both "sputtering" and "straggling" are responsible for PID in 3D. 2014.05.12 PESM Koji Eriguchi 16
Outline Outline 1. Introduction – Plasma-Induced Damage (PID) 2. A Scenario for PID in Scaled 3D Structures 2-1 PID Models – Straggling & Sputtering 2-2 Molecular Dynamics Simulation 2-3 Electronic State of Defect 3. PID Prediction in 3D Structure 4 . Summary 2014.05.12 PESM Koji Eriguchi 17
Electronic structure of defect Original super-lattice structure Damage DFT: 6-31G with PBEPBE & PBC Displaced Si Density-of-state (arb. u.) 1.2 0 40 80 120 Band gap energy (eV) 3.0 2.0 He interstitial 0.8 Energy (eV) 1.0 dumbbell 0.0 0.4 Ar interstitial -1.0 Displaced Si -2.0 No damage 0.0 0.0 0.2 0.4 0.6 0.8 -3.0 Distance from lattice site (A) 2014.05.12 PESM Koji Eriguchi 18
Experimental evidence Eriguchi IEDM 2008 / Kamei Thin Solid Films 518 (2010) / Nakakubo AVS 2011 10 20 Shield Defect density: n dam (cm -3 ) Sample V g n dam 10 19 CV Stage 400 kHz 10 18 Control PC 13.56 MHz Hg no bias SiO 2 Si S.P.100W Ar plasma 10 17 0.0 5.0 10.0 15.0 20.0 (V 1/2 ) V V plasma dc Defect Density ~ 10 18 - 10 19 cm -3 2014.05.12 PESM Koji Eriguchi 19
Outline Outline 1. Introduction – Plasma-Induced Damage (PID) 2. A Scenario for PID in Scaled 3D Structures 2-1 PID Models – Straggling & Sputtering 2-2 Molecular Dynamics Simulation 2-3 Electronic State of Defect 3. PID Prediction in 3D Structure 4 . Summary 2014.05.12 PESM Koji Eriguchi 20
PID prediction in Fin-structure Assumption: (1) PID is modeled by the present scheme. (2) Surface damaged layer was stripped off. Typical fin size (3) Latent-defect creation was uncorrelated. ~5 nm ~10 nm ? ~30 nm ~20 nm One in ~ every MD PID Range Theory two snapshots n n exp( n ) MD Typical PID: dam dam P ( n 0 ) n ( x ) dx ( E ) n dy MD dam ion ion n ! Ar 200 eV MD x n p ~ 10 11 cm -3 Experiment in planar structure Present Models n dam ~ 10 18 - 10 19 cm -3 M M A E E 2 Si ion sp sp ion th B ( E ) L ion ion 3 M M Si ion 2014.05.12 PESM Koji Eriguchi 21
Outline Outline 1. Introduction – Plasma-Induced Damage (PID) 2. A Scenario for PID in Scaled 3D Structures 2-1 PID Models – Straggling & Sputtering 2-2 Molecular Dynamics Simulation 2-3 Electronic State of Defect 3. PID Prediction in 3D Structure 4 . Summary 2014.05.12 PESM Koji Eriguchi 22
Summary Plasma-induced damage in 3D structures were discussed. (1) A new PID model was proposed on the basis of (A) lateral straggling at the etched surface and (B) bombardment of sputtered species at the sidewall. (2) A model prediction and MD simulations suggest that both the lateral straggling and the sputtered particle bombardment will become responsible for PID in scaled 3D structures. One should revise the views of plasma etching at the sidewall because the lateral PID is no longer negligible in ultimately scaled 3D devices. 2014.05.12 PESM Koji Eriguchi 23
Recommend
More recommend