Feature Profile Evolution during Shallow Trench Isolation (STI) Etch in Chlorine-based Plasmas

Transcription

1 1 Feature Profile Evolution during Shallow Trench Isolation (STI) Etch in Chlorine-based Plasmas Presentation November 14, 2005 Jane P. Chang and John Hoang Department of Chemical and Biomolecular Engineering University of California, Los Angeles Plasma

2 2 Motivation Feature Scale Modeling Combining accurate descriptions of plasma fluxes to quantitatively predict the feature profile evolution during etching/deposition processes Enabling process development by shorting the experimental time and cost Feature scale model can be coupled to tool scale (e.g. Prof. Graves, UCB) Feature scale model can be coupled with PIC/MC model (Prof. Lieberman, UCB) Shallow Trench Isolation (STI) An enabling technology over local oxidation of (LOCOS) since the 0.18 µm node A lower temperature process avoiding annealing used for thermal oxidation A promising technology for even smaller dimensions with properly developed lithography, etch, and gap-fill technology Plasma

3 3 Critical Issues in Feature Scale Models Accurate reaction kinetics model Systematic beam measurement Carefully designed design of experiments Robust incorporation of competing surface mechanisms Etching vs. deposition Specular vs. non-specular scattering Elemental balance on etching surfaces Realistic etching profiles for validation SEM Plasma

4 Competing Mechanisms during Etching Processes 2 e 2 e e 2e 2 4 e 2 e bulk plasma e- e- e - e - e- 2 e - } 2 E sheath e _ mask Remove thin film directionally Profile evolution affected by etching and deposition Plasma O 2 poly- oxide

5 Multiple Beam Apparatus Mass Spectrometer Poly- or oxide thin film Etchants (, 2 ) TC Ions ( ev Ar, ) Inhibitors (C, 2 ) Reflectance Time He-Ne Laser Independent variation of ionic and radical fluxes Controllable flux levels within an order of magnitude of what typically used in high density plasma processes Etching Yield = f ( Ion, Etchant, Inhibitor, E ion, φ ion... ) Plasma

6 Energy Dependence ( and ) 4 ion-enhanced 75eV / etching yield with ( Eion Eth ) Dotted lines are Etching Yield eV / model fits, as detailed later 1 35eV / Plasma Flux Ratio

7 Etching Yield by vs. 2 ( 55 ev ) 3 milar etching yield at sufficient 55eV / high flux Much higher etching yield with atomic chlorine at Etching Yield eV / 2 low flux ( higher flux ratio ) Flux Ratio Plasma or 2

8 Angular Dependence Result ( 35eV and fluxes ) 1.5 φ 1.2 Normal Etching Yield Lower etching yield at high off-normal ion incident angle Plasma Flux Ratio 70 o off-normal

9 Angular Dependence Result ( 35eV /, Flux Ratio = 200) 1.2 polynomial fit Maximum yield at normal ion incident angle Yield (φ) = c (φ) * Yield (φ = 0 o ) φ Etching Yield physical sputtering Plasma Ion incident angle φ (degree from normal)

10 Surface Kinetic Model Chlorination : Sorption of Chlorine ion: Ion-enhanced etching: s ( g ) ( s) c(φ) ( g ) ( s) c(φ)β ( s) 4 ( s) 4( g ) 4 ER θ Y = c( φ ) β θ 1 Y c( φ ) β θ = s R c( φ ) = s R c( φ ) 4 c( φ ) β 1 = c( φ ) β 1 c( φ ) β 4 s R c( φ ) 4 1 s R Plasma I β E ion s

11 Energy Dependence and Effect of Redeposition ( and ) 4 75eV / 1.2 e Etching Yield eV / 35eV / 80 ev (Lam TCP) Etching Yield / = 120 with 2 alone with Flux Ratio Flux Ratio 2 Good agreement with the etching yield measurement in Lam TCP gnificant reduction in etching yield due to redeposition Plasma

12 4 Selectivity Poly- vs. Oxide (100 ev Ar and ) Angular Dependency φ PR poly oxide Poly Poly Yield Oxide Oxide Etching selectivity ~ 30 Flux Ratio Ar 0.0 Plasma Ion incident angle φ (degree from normal) Distinct angular dependencies between etching of poly and oxide

13 Ar // Ar //O 2 Ar Ar 4 \/ \/ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / O O O O O O O O O O O O incorporation caused by knock-on and spontaneous reaction with Etching scales with surface energy deposition Plasma more confined to the top surface due to the lack of spontaneous reaction with O 2 Etching scales with sputtering of and O

14 Etching of Patterned Polysilicon Wafers Line width: 0.5 µm and 0.35 µm Photo-resist: Apex-E ( Deep-UV resist ) Orientation w.r.t. and beams: Un-shadowed Shadowed Plasma

15 Feature Profile etched with 35eV / ( / flux ratio = 200 ) PR poly- PR poly- O 2 O 2 Un-shadowed Shadowed Unique etching structure suitable for profile modeling confirmation Plasma

16 Monte Carlo mulation of Surface Evolution Source plane mulation domain: o A grid size (~ mixing length) Transport of species in grid-length steps Surface Reaction: Reaction probability based on model Elemental balance Multiple interaction possible PR poly- Surface advancement Grid cells removal and/or addition Rigorous incorporation of all the physics and chemistry Computationally robust and straightforward O 2 Plasma

17 Monte Carlo mulation of Surface Interactions Surface Reactions: 4 ( g ) ( g ) ( s) 4 ( s) 4( g ) ( s) ( s) 4 Scattering: Ions: specular scattering Neutrals: diffusive scattering Surface composition determined by elemental species balance Surface reactions occur with measured probabilities Plasma

18 Etching of Patterned Polysilicon Wafers ( 35eV and, Flux Ratio = 200) recombination probability of on mask, shadowed orientation Unique etching structure modeled by simulator Plasma

19 Effect of Ion Scattering oxide poly- oxide Non-directional ion distribution (10 o FWHM) Trenching formation due to ion scattering Plasma

20 Effect of Deposition/Redeposition photoresist C 2 poly- oxide Non-directional ion distribution (10 o FWHM) Trenching formation reduced due to deposition/redeposition of etching products Plasma

21 21 ITRS dictates stringent conditions for optimal trench isolation as minimum feature size decreases Positive trench tapering angles desired to avoid sharp recesses leading to poly wrap-around Smooth sidewalls needed for less physical and electrical damage Round bottom corners to minimize stress and avoid voids in gapfill tx 1 (nitride ) tx 2 (top ) tx 3 (bot ) SEM Measured Parameters D 1 D 2 D 3 D 4 STI Etching Process Nitride SWA Total Depth top SWA bot SWA PR nitride oxide licon Desired Properties: D 4 > D 2 /2 Shallow Trench Isolation (STI) Isolation stack dewall oxidation and deposit trench oxide Pattern nitride and strip PR CMP planarization θ nitride = 90º arctan[(d 1 -D 2 )/2/tx 1 ] θ top = 90º arctan[(d 2 -D 3 )/2/tx 2 ] θ bot = 90º arctan[(d 3 -D 4 )/2/tx 3 ] Trench etch Strip nitride and remove pad oxide SWA: sidewall angle Adapted from ITRS 2003 Thermal Films Supplemental Plasma Recess < 0.1 D 2 Curvature: r Nitride top = r bottom = 0.1 D 2

22 22 AMAT DP SII Reactor Setup I outer I inner Coil Power W s W s Parameters examined for STI etch Chamber Pressure (mtorr) Source Power (W s ) Wafer bias (W bias ) 2 N 2 O 2 Pressure DC ratio = I outer /I inner 2 flowrate (sccm) N 2 flowrate (sccm) Substrate Bias W bias O 2 flowrate (sccm) Plasma

23 Fractional Factorial DOE x N y etch DOE etch DOE Pressure (mt) W s (W) W b (W) BS He (sccm) CF 4 (sccm) Ar (sccm) ID CHF 3 (sccm) 23 Plasma Pressure (mt) W s (W) W b (W) DC ratio ID factors, 2 levels, and 16 experiments performed for both etch targets 2 (sccm) N 2 (sccm) O 2 (sccm) Nitride etch: pressure was determined to be the statistically significant effect etch: pressure and DC ratio had statistically significant effects

24 24 Source plane Periodic boundary conditions Mask (N x ) licon Monte Carlo mulation n or 4 mulation domain: Å grid size Ions specularly scatter Neutrals diffusively scatter Surface composition determined by elemental balance Surface reactions occur with measured probabilities Transport of species in grid lengths Surface reaction: reaction probability based on kinetic models elemental balances multiple interactions possible Surface advancement: grid cells removed and/or added Rigorous incorporation of all physics and chemistry Computationally robust and straight-forward Plasma

25 25 Etching Chlorination: Sorption of Chlorine ion: Ion-enhanced etching: 2 Deposition: Oxygenation: Sputtering: Sorption of sputtered : Recombination of chlorine: Surface Reaction Kinetics Model s 0 (1 ζ ζ O ) ( g ) ( s) c( φ) ( g) ( s) 4 4 c( φ) β ( s) ( s) 4( g) 3 0 s 2 2 2( g) ( s) ( s) so 0 (1 ζ ζo) O () g () s Y SP () s ( g) s0 ( g) ( s) r 2 ( g) ( s) ( g) O Plasma (implemented) (implemented) (implemented) (in development) (in development) (in development) (in development) (implemented) Nitride Etching Implemented only etch selectivity and angular dependence due to limited kinetic measurements during the etching of nitride in CF 4 /Ar/CHF 3 plasma Experimentally determined reaction kinetics model enables predictable feature profile evolution

26 26 Elemental Balance in Cells Source plane n Periodic boundary conditions Mask (N x ) N O licon provides available sites Etchant such as leads to the formation of volatile products Reactant such as O leads to the oxidation thus changing the etching characteristics Deposition of 2 and will add sites Plasma

27 27 Major Enhancements in mulation Implemented ion etching yield dependence as a function of ion energy Implemented ion energy distribution function (to be enhanced with real experimental or plasma simulation results) Collaboration with Graves and Lieberman Etching Yield (/) number of ions eV 195eV 155eV 115eV 75eV 55eV eV Flux Ratio (/ ) Gas cell Solid interface cell Solid noninterface cell Least squares method normals Implemented sloped mask Determined surface normals using least squares regression fit to center of cells considered effective rounding of corners Plasma Ion energy (ev)

28 28 mulation Details Parameters affecting profile evolution: plasma chemistry ( 2, HBr, O 2, ) plasma composition (, 2,, O, ) electron temperature and distribution (T e and EEDF n i, n n, ) substrate bias (W s E ion ) substrate temperature (T sub ) Baseline Conditions: initial aspect ratio: 0.55 ion angular distribution (IAD) FWHM: 5.3º ion energy distribution (IED) FWHM: 23.5 ev ion energy: 200 ev neutral to ion ratio: 100 selectivity of nitride to : 33.3 Plasma

29 29 mulator Capabilities Deposition Spontaneous Etching Micro-Trenches Bowing Effect of Selectivity Effect of Mask Angle Plasma

30 30 Comparison of mulation with Experiments DOE DOE pressure (mtorr) Ws (W) Wb (W) DC ratio (sccm) N2 (sccm) O2 (sccm) milar plasma densities Substrate bias governs the etch depth DOE DOE pressure (mtorr) Ws (W) Wb (W) DC ratio (sccm) N2 (sccm) O2 (sccm) High density versus low density plasmas Plasma composition controls profile evolution mulation on-going Plasma (significantly different sidewall slope could be due to a change in plasma composition)

31 31 Future Goals Determine plasma gas phase chemistry experimentally Validate the simulation results with specially planned additional experiments Enhance current simulator by including feature charging and simultaneous deposition and etching Correlate plasma operating parameters to simulation input profiles to allow a more direction comparison of the simulation results to experiment outcomes Special Acknowledgements: Helena Stadniychuk and Andrey Zagrebelny at Cypress Acknowledgment Funded by Advanced Micro Devices, Applied Materials, ASML, Atmel, Cadence, Canon, Cymer, Cypress, DuPont, Ebara, Hitachi Global Storage Technologies, Intel, KLA-Tencor, Mentor Graphics, Nikon Research, Novellus Systems, Panoramic Technologies, Photronics, Synopsys, Tokyo Electron, and the UC Discovery Grant. Plasma