1. AVS 2002 Nov 3 - Nov 8, 2002 Denver, Colorado INTEGRATED MODELING OF ETCHING, CLEANING AND BARRIER COATING PVD FOR POROUS AND CONVENTIONAL SIO 2 IN FLUOROCARBON BASED CHEMISTRIES * Arvind Sankaran 1 and Mark J. Kushner 2 1 Department of Chemical Engineering 2 Department of Electrical and Computer Engineering University of Illinois, Urbana, IL 61801, USA email: asankara@uiuc.edu mjk@uiuc.edu http://uigelz.ece.uiuc.edu *Work supported by SRC, NSF and SEMATECH

2. University of Illinois Optical and Discharge Physics AGENDA  Low dielectric constant materials  Surface reaction mechanism and validation  Fluorocarbon etching of SiO 2 /Si  Ar/O 2 etching of organic polymer  High aspect ratio etching of porous and non porous SiO 2  Integrated Modeling: Ar/O 2 strip of polymer and IMPVD  Concluding Remarks AVS03_AS_02

3. University of Illinois Optical and Discharge Physics LOW DIELECTRIC CONSTANT MATERIALS  The increase in the signal propagation times due to RC delay has brought the focus onto low dielectric constant (low-k) materials (inorganic and organic) AVS03_AS_03  Inorganics such as porous silica (PS) are etched using fluorocarbon chemistries; organics are etched using oxygen chemistries.

4. University of Illinois Optical and Discharge Physics GOAL FOR INTEGRATED MODELING  Plasma processing involves an integrated sequence of steps, each of which depends on the quality of the previous steps. CFDRC_0503_05

5. University of Illinois Optical and Discharge Physics SURFACE REACTION MECHANISM - ETCH  CF x and C x F y radicals are the precursors to the passivation layer which regulates delivery of precursors and activation energy.  Chemisorption of CF x produces a complex at the oxide-polymer interface. 2-step ion activated (through polymer layer) etching of the complex consumes the polymer. AVS03_AS_05  Activation scales as  1/L and the L scales as  1/bias.  In Si etching, CF x is not consumed, resulting in thicker polymer layers.  Si reacts with F to release SiF x.

6. University of Illinois Optical and Discharge Physics SURFACE REACTION MECHANISMS - STRIP AVS03_AS_06  Ar/O 2 is typically used for polymer stripping after fluorocarbon etching and resist removal.  Little polymer removal is observed in absence of ion bombardment suggesting ion activation.  For SiO 2 etching in mixtures such C 4 F 8 /O 2, the fluorocarbon polymer is treated as an organic. Resists are treated similarly.

7. University of Illinois Optical and Discharge Physics MONTE CARLO FEATURE PROFILE MODEL (MCFPM)  The MCFPM predicts time and spatially dependent profiles using energy and angularly resolved neutral and ion fluxes obtained from equipment scale models.  Arbitrary chemical reaction mechanisms may be implemented, including thermal and ion assisted, sputtering, deposition and surface diffusion.  Energy and angular dependent processes are implemented using parametric forms. INTELTALK_AS_17  Mesh centered identity of materials allows “burial”, overlayers and transmission of energy through materials.

8. University of Illinois Optical and Discharge Physics MODELING OF POROUS SILICA  MCFPM may include “two phase” materials characterized by porosity and average pore radius.  Pores are incorporated at random locations with a Gaussian pore size distribution. Pores are placed until the desired porosity is achieved with/without interconnects. AVS03_AS_07  Interconnected structures can be addressed.

9. University of Illinois Optical and Discharge Physics TYPICAL PROCESS CONDITIONS  Process conditions  Power: 600 W  Pressure: 20 mTorr  rf self-bias: 0-150 V  C 4 F 8 flow rate: 40 sccm  The fluxes and energy distributions are obtained using the HPEM. AVS03_AS_08

10. University of Illinois Optical and Discharge Physics BASE CASE ION AND NEUTRAL FLUXES  Self-bias = - 120 V. Decrease in neutral and ion fluxes along the radius have compensating effects. AVS03_AS_09  Ions have a narrow energy and angular distribution, in contrast to neutrals.

11. University of Illinois Optical and Discharge Physics VALIDATION OF REACTION MECHANISM: C 4 F 8  The mechanism was validated with experiments by Oehrlein et al using C 4 F 8, C 4 F 8 /Ar and C 4 F 8 /O 2. 1  Threshold for SiO 2 etching was well captured at self-bias  -40 V. Polymer formation is dominant until the threshold bias  As polymer thins at higher biases, the etching proceeds. AVS03_AS_10 1 Li et al, J. Vac. Sci. Technol. A 20, 2052, 2002.

12. University of Illinois Optical and Discharge Physics VALIDATION: C 4 F 8 /Ar and C 4 F 8 /O 2  Larger ionization rates result in larger ion fluxes in Ar/C 4 F 8 mixtures. This increases etch rates.  With high Ar, the polymer layers thins to submonolayers due to less deposition and more sputtering and so lowers etch rates.  O 2 etches polymer and reduces its thickness. Etch rate has a maximum with O 2, similar to Ar addition. AVS03_AS_11

13. University of Illinois Optical and Discharge Physics PROFILE COMPARISON: MERIE REACTOR AVS03_AS_12  Process conditions  Power: 1500 W CCP  Pressure: 40 mTorr  Ar/O 2 /C 4 F 8 : 200/5/10 sccm V. Bakshi, Sematech

14. University of Illinois Optical and Discharge Physics VALIDATION OF POROUS SiO 2 ETCH MODEL  Two porous substrates  2 nm pore radius, 30% porosity  10 nm pore radius, 58% porosity  Process conditions  Power: 1400 W (13.56 MHz)  Pressure: 10 mTorr  rf self-bias: 0-150 V  40 sccm CHF 3  Etch rates of P-SiO 2 are higher than for NP-SiO 2 due to lower mass densities of P-SiO 2. AVS03_AS_13 Exp: Oehrlein et al, J. Vac. Sci.Technol. A 18, 2742 (2000)

15. University of Illinois Optical and Discharge Physics WHAT CHANGES WITH POROUS SiO 2 ?  The “opening” of pores during etching of P-SiO 2 results in the filling of the voids with polymer, creating thicker layers.  Ions which would have otherwise hit at grazing or normal angle now intersect with more optimum angle. INTELTALK_AS_30  An important parameter is L/a (polymer thickness / pore radius).  Adapted: Standaert, JVSTA 18, 2742 (2000)

16. University of Illinois Optical and Discharge Physics EFFECT OF PORE RADIUS ON HAR TRENCHES AVS03_AS_15  With increase in pore radius, L/a decreases causing a decrease in etch rates.  Thicker polymer layers eventually lead to mass corrected etch rates falling below NP-SiO 2. There is little variation in the taper. 4 nm16 nm10 nm

17. University of Illinois Optical and Discharge Physics HAR PROFILES: INTERCONNECTED PORES INTELTALK_AS_40 60% 100% 0% Interconnectivity

18. University of Illinois Optical and Discharge Physics EFFECT OF PORE RADIUS ON CLEANING AVS03_AS_17  Larger pores are harder to clean due to the view angle of ion fluxes.  Unfavorable view angles lead to a smaller delivery of activation energy, hence lower activated polymer sites. 4 nm16 nm ANIMATION SLIDE  Ar/O 2 =99/1, 40 sccm, 600 W, 4 mTorr

19. University of Illinois Optical and Discharge Physics CLEANING INTERCONNECTED PORES CHEME_AS_19  Cleaning is inefficient with interconnected pores.  Higher interconnectivity leads to larger shadowing of ions. 60%100%0% ANIMATION SLIDE Interconnectivity  Ar/O 2 =99/1, 40 sccm, 600 W, 4 mTorr

20. University of Illinois Optical and Discharge Physics EFFECT OF ASPECT RATIO ON STRIPPING AVS03_AS_19  Cleaning decreases with increasing aspect ratios.  Pores at the top of the trench are stripped better due to direct ions (view angle).  Pores near the bottom see ions reflected from the bottom of the trench and are cleaned better. 3 5 1 ANIMATION SLIDE Aspect Ratio  Ar/O 2 =99/1, 40 sccm, 600 W, 4 mTorr

21. 4 nm 16 nm NP10 nm University of Illinois Optical and Discharge Physics EFFECT OF PORE RADIUS ON Cu DEPOSITION AVS03_AS_20  Surrogate study for seed layer deposition and barrier coating.  Larger pores require longer deposition times for conformal coverage.  This produces thicker bottom and open field films.  Voids are created or initiated by larger pores.

22. University of Illinois Optical and Discharge Physics EFFECT OF INTERCONNECTIVITY ON Cu IMPVD AVS03_AS_21  Interconnected pores need to be sealed to avoid pin- hole formation.  Pore sealing by Cu IMPVD ineffective at larger interconnectivities.  Thicker layers to seal pores produces trench narrowing, which can lead to pinch off. 30% 100% 0%60% Interconnectivity

23. University of Illinois Optical and Discharge Physics CONCLUSIONS  Etching of PS obeys scaling laws as that of SS. Etch rate increases for smaller pores and slows for larger pores (at high porosities).  L/a determines etch rate variation of P-SiO 2. Polymer filling increases the net thickness.  Stripping is inefficient for interconnected pore networks and for larger pores due to the unfavorable view angles for the ion fluxes. Low aspect ratio pores are better cleaned.  Cu IMPVD is non-conformal for closed pore networks with larger pores. Pin-hole formation and trench narrowing is seen for interconnected networks. AVS03_AS_22