modeling plasma processing dischargeslieber/liebpsc120511crop.pdf · • a one-dimensional hybrid...
TRANSCRIPT
University of California, Berkeley PLASMA
MODELING PLASMA PROCESSING DISCHARGES
M.A. Lieberman
Department of Electrical Engineering and Computer SciencesUniversity of California
Berkeley, CA 94720
Collaborators:E. Kawamura, D.B. Graves, and A.J. Lichtenberg, UC BerkeleyC. Lazzaroni and P. Chabert, Ecole Polytechnique, FranceJ. Gudmundsson, Shanghai Jiao Tong U; Science Institute, U. IcelandA. Leblanc, ENS Cachan, FranceJing Zhang, Donghua U, Shanghai, China
Download this talk:
http://www.eecs.berkeley.edu/∼lieber
LiebermanPSC12 1
University of California, Berkeley PLASMA
OUTLINE
• Fast computation of atmospheric pressure rf capacitive discharges
• Fluid model of E-H transition instability in electronegativeinductive discharge
LiebermanPSC12 2
University of California, Berkeley PLASMA
ATMOSPHERIC PRESSURE
CAPACITIVE RF DISCHARGE
LiebermanPSC12 3
University of California, Berkeley PLASMA
MOTIVATION
• Biomedical — example of reactive oxygen species(Review article: H.W. Lee et al, J. Phys. D 44, 053001, 2011)
— Applications to sterilization, cancer cell treatment, bloodcoagulation, wound healing
• Unique materials — example of anatase crystalline TiO2
(Review article: D. Mariotti and R.M. Sankaran, J. Phys. D, 323001, 2010)
(Anatase TiO2: H.G. Yang et al, Nature 453, 638, 2008)
— Applications to photonics crystals, photo/electrochromic devices,gas sensors, spintronic devices, anticancer or gene therapies,solar cells for electric energy or hydrogen production
LiebermanPSC12 4
University of California, Berkeley PLASMA
DISCHARGE CONFIGURATION
• Atmospheric pressure
• He or Ar with trace reactive gases
• 1D plane-parallel geometry (∼0.2–2 mm gap)
• RF-driven (nominal 13.56 MHz)
He or Ar+
Trace gas
~
Reactivespecies
~0.1–1 mm
13.56 MHz
LiebermanPSC12 5
University of California, Berkeley PLASMA
EEPF TIME VARIATIONS
• He/N2 fluid simulation with kinetic (Bolsig+) EEPF calculation
Lowtemperaturetail
Timevariation
(J. Waskoenig, PhD Thesis, Queens U Belfast, 2010)
• Conclusions used in modeling— The EEPF oscillates in time with the rf electron power absorbed— The EEPF is Maxwellian below a break energy Eb ≈ 20 V
(metastable He excitation energy)— The EEPF has a low temperature tail above the break energy
LiebermanPSC12 6
University of California, Berkeley PLASMA
HYBRID DISCHARGE MODEL
• Numerical solution of particle balances for each speciesdnj
dt= Gj − Lj
Gj = volume creation rate (2-body, 3-body and surfaces)Lj = volume loss rate (2-body, 3-body, and surfaces)
• Analytical solutions of— the discharge dynamics (homogeneous model)— the time-varying Te(t)— the effective rate coefficients 〈K〉
• Coupling the analytical and numerical solutions
=⇒ fast solution of the discharge equilibrium
LiebermanPSC12 7
University of California, Berkeley PLASMA
COMPARISON TO FLUID SIMULATION
• He/0.5%O2 (16 species), 1mm gap, 13.56 MHz
• Neutral (left) and charged (right) densities versus power
(open symbols — global model; solid symbols — fluid results, Waskoenig, 2010)
• ⇒ Reasonable agreement of model and fluid simulations40 sec simulation time on fast laptop
LiebermanPSC12 8
University of California, Berkeley PLASMA
E/H (CAPACITIVE/INDUCTIVE) MODE TRANSITION
INSTABILITY IN ELECTRONEGATIVE DISCHARGE
LiebermanPSC12 9
University of California, Berkeley PLASMA
MOTIVATION
• Low pressure inductive reactors for thin film processing
— Example: fabrication of CMOS transistors formicroprocessors/memory
— Inductive reactors often operate near the E/H transitionwith electronegative feedstock gases
— Macroscopic instabilities observed in both commercialand research reactors
LiebermanPSC12 10
University of California, Berkeley PLASMA
E/H MODE TRANSITION
• Plasma resistance Re ver-sus ne as Irf is varied
• A “gap” occurs betweenIrf = 7.5 and 8 A
• Measurements at 10 mTorrCl2 show “gap region”
• Previous measurements (many)and global models (many)indicate instability
• First calculation of E/Hinstability in fluid simu-lations
1e+14 1e+15 1e+16 1e+17
Center n (m )
0
1
2
3
4
5
6
R (o
hm
s)
e-3
Capacitive(E-mode)
e
Inductive(H-mode)
I =7.5A
Gap
I =8Arf rf
15 mTorr Cl2 fluid simulation
0 100 200 300 400 500 600 700 800
P (W)
1e+14
1e+15
1e+16
1e+17
n
(m
)
Malyshev & Donnelly (8/2001) measurements
abs
e-3
Gapregion
LiebermanPSC12 11
University of California, Berkeley PLASMA
BULK-FLUID/ANALYTIC-SHEATH MODEL
• Inductive reactor (Malyshev and Donnelly, 2000–01)
Wafer chuck
(κ = κ )Plasma
φ
spacer
Air
symmetry (r = 0)Center of
z
outer wallsconductingPerfectly
Axisymmetric cylindrical geometry
3 41
Coils
(κ = 1)Sheath
r
p
(κ = 4)
(κ = 1)
Quartz dielectric
2
(κ = 4)Quartz dielectric window
0.21 m
0.18 m
— Electromagnetic field solve— Fluid bulk plasma model— Analytical sheath model— Flow model of reactive gas— Commercial software (COMSOL)
(Kawamura et al, PSST 2011)LiebermanPSC12 12
University of California, Berkeley PLASMA
E/H TRANSITION INSTABILITY
• Example: 2.2 kHz instability in 15mTorr Cl2 at Irf = 7.75 A, show-ing (a) nCl−(t), (b) ne(t), and (c)Te(t)
• At time t1 the discharge enters ca-pacitive mode
• From t1–t2 the discharge is in ca-pacitive mode
• From t2–t3 the discharge makes atransition to inductive mode
• From t3–t4 the discharge is in in-ductive mode
• From t4–t1 the discharge makes atransition back into capacitive mode
0 0.0005 0.001 0.00156e+16
7e+16
8e+16
9e+16
1e+17
1.1e+17
n
(m
)
(a)
Cl–
-3
0 0.0005 0.001 0.00150
2e+15
4e+15
6e+15
8e+15
1e+16
n
(m
)
(b)
e-3
0 0.0005 0.001 0.0015t (s)2
2.1
2.2
2.3
2.4
2.5
T
(V)
(c)
e
t
t
t
t1
2
3
4
t
tt
t
12
3
4
t
t
t t
1
2
3 4
.
.
..
.
.
.
.
.
. . .
LiebermanPSC12 13
University of California, Berkeley PLASMA
Te(r, z) VERSUS t
• Te strongly lo-calized near coil
• t1 (enter E-mode):Te jumps to high-est value
• t2–t4 (E-modefollowed by tran-sition to H-mode):Te decays in time
t = t1 t = t2
t = t3 t = t4
LiebermanPSC12 14
University of California, Berkeley PLASMA
ne(r, z)/nemax VERSUS t
• ne weakly local-ized near coil
• t1 (enter E-mode):ne rapidly de-cays with time
• t3 (enter H-mode):ne rapidly in-creases with time
nemax = 1.29× 1015 m−3 at t1 nemax = 5.73× 1015 m−3 at t2
nemax = 15.6× 1015 m−3 at t3 nemax = 5.86× 1015 m−3 at t4
LiebermanPSC12 15
University of California, Berkeley PLASMA
FLUID AND GLOBAL MODEL COMPARISON
• Intersection of dne/dt = 0 and dn−
/dt = 0 curves ⇒ equilibrium
• Slope dn−
/dne of dne/dt = 0 curve positive ⇒ unstable
1014
1015
1016
1016
1017
n– (
m–3)
ne (m–3)
dne/dt = 0
dn− /dt = 0
Globalmodeltheory
Fluidsimulation.
• Good agreement of fluid calculation and analytical global model
LiebermanPSC12 16
University of California, Berkeley PLASMA
NEUTRAL TIME AVERAGES OVER INSTABILITY
• Neutral species time variations are very small
• Time averages:
Gas Temperature (K) Molar Cl Fraction
• Tg rises to 530 K inside discharge
• Chlorine density varies significantly with radiusLiebermanPSC12 17
University of California, Berkeley PLASMA
SUMMARY
• A one-dimensional hybrid analytical-numerical global model ofatmospheric pressure, rf-driven capacitive discharges was developed
• Coupling analytical solutions of the time-varying discharge andEEPF dynamics, and numerical solutions of the discharge chemistry,allows for a fast solution of the discharge equilibrium
(Lazzaroni et al, to appear in PSST, 2012)
• The E/H transition instability has been found and studied in2D fluid simulations
• The fluid instability dynamics is in good agreement with ananalytical global model
(Kawamura et al, submitted to PSST, 2012)
Download this talk:
http://www.eecs.berkeley.edu/∼lieber
LiebermanPSC12 18