View
213
Download
0
Category
Preview:
Citation preview
FACTORS AFFECTING THE SEALING EFFICIENCY OF LOW-k DIELECTRIC SURFACE PORES USING
SUCCESSIVE He AND Ar/NH3 PLASMA TREATMENTS
Juline Shoeba) and Mark J. Kushnerb)
a) Department of Electrical and Computer EngineeringIowa State University, Ames, IA 50011
jshoeb@eecs.umich.edu
b) Department of Electrical Engineering and Computer ScienceUniversity of Michigan Ann Arbor, Ann Arbor, MI 48109
mjkush@umich.edu
http://uigelz.eecs.umich.edu
October 2009*Work supported by Semiconductor Research Corporation JULINE_GEC09
Low-k Dielectrics
Modeling of Porous Low-k Sealing
Sealing Mechanism
Surface Site Activation by He plasma pre-treatment
Sealing by Ar/NH3 Treatment
Sealing Efficiency Dependence
Porosity and Interconnectivity
Treatment time and Pore Radius
Concluding Remarks
AGENDA
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
POROUS LOW-k DIELECTRIC
Metal interconnect lines in ICs run through dielectric insulators.
The capacitance of the insulator contributes to RC delays.
Porous oxides, such as C doped SiO2 (with CHn
lining pores) have a lower dielectric constant which reduces the RC delay.
Porosity is 0.5. Inter-connected pores open to surface may degrade k-value by reactions with plasma species.
Ref: http://www.necel.com/process/en/images/porous_low-k_e.gif
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
GOALS AND PREMISES OF SEALING MECHANISM To prevent the degradation of low-k
materials pores open to the surface should be sealed.
He plasma followed by NH3 plasma treatment has been shown to seal pores in SiCOH.
He+ and photons break Si-O bonds and remove H atom from CHn lining pores
Ref: A. M. Urbanowicz, M. R. Baklanov, J. Heijlen, Y. Travaly, and A. Cockburn, Electrochem. Solid-State Lett. 10, G76 (2007).
Plasma Treatment
Time (s)
Function
He 200 Surface
Activation
NH3 30
( Post-He)
Sealing
Subsequent NH3 exposure seals the pores by adsorption reactions forming C-N and Si-N bonds.
Reaction mechanisms and scaling laws for He-NH3 plasma sealing of porous SiCOH have been developed based on results from a computational investigation.
Experimental results by others were used to build the mechanism.
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
MODELING : LOW-k PORE SEALING
Hybrid Plasma Equipment Model (HPEM)
Plasma Chemistry Monte Carlo Module (PCMCM)
Monte Carlo Feature Profile Model (MCFPM)
Energy and angular
distributions for ions and
neutrals
He PLASMA
Porous Low-k
Coils
WaferSubstrate
Metal
Plasma
Ar/NH3 PLASMAS
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
MONTE CARLO FEATURE PROFILE MODEL (MCFPM)
The MCFPM resolves the surface topology on a 2D Cartesian mesh to predict etch profiles.
Each cell in the mesh has a material identity. (Cells are 4 x 4 ).
Gas phase species are represented by Monte Carlo pseuodoparticles.
Pseuodoparticles are launched towards the wafer with energies and angles sampled from the distributions obtained from the PCMCM.
Cells identities changed, removed, added for reactions, etching deposition.
PCMCM
Energy and angular distributions for ions
and neutrals
HPEM
MCFPM
Provides etch rateAnd predicts etch
profile
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
80 nm wide, 80 nm thick porous SiCOH.
Average pore radius: 0.8-1.1 nm
Process integration steps:
1. Ar/C4F8/O2 CCP: Etch 10:1 Trench
2. O2 ICP: Remove CFx polymer and PR.
3. He ICP: Activate Sites
4. NH3 ICP: Seal Pores
INITIAL LOW-k PROFILE, PROCESS INTEGRATION
University of MichiganInstitute for Plasma Science & Engr.
Hard Mask
Porous Low-kSiCOH
Si
JULINE_GEC09
AR/O2 PLASMAS: POLYMER REMOVAL AND CH3
GROUP ETCHING
Polymer Sputtering Ar+(g) + P(s) P*(s) + Ar(g)
Ar+(g) + P*(s) CF2(g) + Ar(g)
O+(g) + P(s) CF2(g) + O(g)
O2+(g) + P(s) CF2(g) + O2(g)
Polymer Etching O (g) + P(s) P*(s) + O (g)
O+ (g) + P*(s) COF(g) + O (g)
O2+ (g) + P*(s) COF(g) + O2(g)
CHn Group Etching M+(g) + CHn(s) CHn-1(s) + H(g) + M(g)
O(g) + CHn-1(s) CO(g) + Hn-1(g)
O+(g) + CHn-1(s) CO(g) + Hn-1(g)
O2+ (g) + CHn-1(s) CO(g) + Hn-1O(g)
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
SURFACE ACTIVATION IN He PLASMA
He+ and photons break Si-O bonds and remove H from CH3 groups.
Bond Breaking He+(g) + SiO2(s) SiO(s) + O(s) + He(g)
He+(g) + SiO(s) Si(s) + O(s) + He(g)
He+(g) + P(s) P*(s) + He(g)
hν + SiO2(s) SiO(s) + O(s)
hν + SiO(s) Si(s) + O(s)
Activation He+(g) + CHn(s) CHn-1(s) + H(g) + He(g)
hν + CHn-1(s) CHn-2(s) + H(g)
He+(g) + CHn(s) CHn-1(s) + H(g) + He(g)
hν + CHn-1(s) CHn-2(s) + H(g)
Reactive sites assist sealing in the subsequent Ar/NH3 treatment.
University of MichiganInstitute for Plasma Science & Engr.
JULINE_GEC09
SEALING MECHANISM IN Ar/NH3 PLASMA
N/NHx species are adsorbed by activated sites forming Si-N and C-N bonds to seal pores.
Further Bond Breaking M+(g) + SiO2(s) SiO(s) + O(s) + M(g)
M+(g) + SiO(s) Si(s) + O(s) + M(g)
N/NHx Adsorption NHx(g) + SiOn(s) SiOnNHx(s)
NHx(g) + Si(s) SiNHx(s)
NHx(g) + CHn-1 (s) CHn-1NHx(s)
NHx(g) + P*(s) P(s) + NHx(s)
SiNHx-NHy/CNHx-NHy compounds seal the pores where end N are bonded to Si or C by Si-C/Si-N
NHy(g) + SiNHx(s) SiNHx-NHy(s)
NHy(g) + CHn-1NHx(s) CHn-1NHx-NHy(s)
University of MichiganInstitute for Plasma Science & Engr.
JULINE_GEC09
CCP for trench etch.
Ar/C4F8/O2 = 80/15/5 40 mTorr, 300 sccm 10 MHz 5 kW
CFx polymers deposited on side-walls efficiently seal the open pores.
Depending on the application polymers:
Removed by O2 plasmas (followed by other pore sealing process)
Maintained for pore sealing
Ar/C4F8/O2 CCP TRENCH ETCH
Animation Slide-GIF
Hard Mask
Porous Low-kSiCOH
Si
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
He PLASMA PRE-TREATMENT
Ion density: 3.8 x 1010 cm-3.
Porous low-k exposed for up to 300 s to the plasma.
250V substrate bias assisted ablating H and Si-O bond breaking.
Conditions: He, 10 mTorr, 300 W ICP, 250V Bias
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
PORE-SEALING BY SUCCESSIVE He AND NH3/Ar TREATMENT
Surface pore sites are activated by 200s He plasma treatment.
Successive 30 s NH3 treatment seals the pores forming Si-N and Si-C bonds.
Initial Surface Pores
University of MichiganInstitute for Plasma Science & Engr.
Animation Slide-GIF
He Plasma Site Activation
Ar/NH3 Plasma Pore Sealing
JULINE_GEC09
POLYMER REMOVAL: Ar/O2 PLASMA TREATMENT
Oxygen Ion density: 3.8 x 1010 cm-3.
56 s exposure to plasma.
250V substrate bias assisted in removal of polymer deep in trench.
Conditions: Ar/O2, 10 mTorr, 300 W ICP, 250V Bias
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
POLYMER REMOVAL FROM SIDEWALLS
Keeping the mask, the trench was exposed to Ar/O2 plasmas for polymer removal.
Ion bombardment activates/removes the polymers. CHn groups are also etched during this process due to the presence of oxygen species.
Neutral oxygen and Ar+ activate polymer while ions remove it by etching or sputtering reactions
Ar+(g) + P(s) P*(s) + Ar(g)
Ar+(g) + P*(s) CF2 + Ar(g)
O(g) + P(s) P*(s) + O (g)
O+(g) + P(s) CF2 + O (g)
O+(g) + P*(s) COF + O(g)
University of MichiganInstitute for Plasma Science & Engr.
Animation Slide-GIF
Hard Mask
Porous Low-kSiCOH
Si
JULINE_GEC09
POLYMER REMOVAL & CH3 DEPLETION Polymer removal in
Ar/O2 plasmas may also remove CH3 groups as O diffuses into the porous network.
Ar/O2 plasma exposure sufficient to remove all polymer also results in CH3 removal.
An optimum strategy may be to leave traces of polymer in recesses (that is, shorter Ar/O2 plasma exposure) to prevent CH3 removal.
University of MichiganInstitute for Plasma Science & Engr.
Hard Mask
Porous Low-kSiCOH
Si
JULINE_GEC09
SEALING: WITH POLYMER REMOVAL
Mask was removed to expose tops of the low-k material.
He plasma activated the tops and the polymer free side-walls of the trench.
Successive NH3 plasma exposure seals the top and side-walls through C-N and Si-N bonding.
University of MichiganInstitute for Plasma Science & Engr.
Si
Animation Slide-GIF
Activation
Sealing
Activation Sealing
JULINE_GEC09
SEALING EFFICIENCY: PORE RADIUS
Sealing efficiency decreases with increasing pore size dropping below 75% > 1.0 nm.
Sidewall efficiency is less sensitive to pore radius due to some unremoved polymer.
C-N and Si-N are “first bonds.”
Sealing requires N-N bonding, which has limited extent.
Too large a gap prevents sealing.
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
SEALING: TREATMENT TIME DEPENDENCE
Without He plasma treatment, Ar/NH3 plasmas seal only 40% of pores.
Sealing efficiency increases with He treatment time for 200s, then saturates.
200 s treatment breaks nearly all side-wall/surface Si-O bonds and activates most CH3 groups.
Sealing efficiency of pores increases for 20s of Ar/NH3
treatment, then saturates – all dangling bonds on the surface are passivated.
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
SEALING EFFICIENCY: ASPECT RATIO
Sealing efficiency on sidewalls decreases with increasing aspect ratio.
Ability for ions to penetrate into features to activate sites decreases with larger aspect ratio.
Sealing improves with longer treatment but some sites are shadowed and will not be sealed.
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
CONCLUDING REMARKS
Integrated porous low-k material sealing was computationally investigated
Ar/C4F8/O2 Etch Ar/O2 Clean He Pretreatment Ar/NH3 sealing
Si-N and C-N bonds formed by adsorption on active sites followed by one N-N bond linking C or Si atoms from opposite pore walls.
Pore sealing efficiency is independent of porosity and interconnectivity, while dependent on both He and NH3 plasma treatment time.
The sealing efficiency degrades when with pore radius > 1 nm and with increasing aspect ratio.
University of MichiganInstitute for Plasma Science & Engr.JULINE_GEC09
Recommended