Demetre J. Economou - DOE Plasma Science Center

Demetre J. Economou
Plasma Processing Laboratory
Dept. of Chemical and Biomolecular Engineering
University of Houston
Acknowledgments: V. Donnelly, P. Diomede, H. Shin, W. Zhu
Financial Support:
pp
DoE Plasma Science Center, NSF, VARIAN
64th GEC , November 14-18, 2011, Salt Lake City, Utah
1
Agenda
g
• Introduction / Motivation.
• PIC-MCC simulation of ion energy distributions (IEDs) on
plasma electrodes.
Tailored voltage
g waveforms.
Comparison with data.
• Model
M d l for
f rapid
id calculation
l l ti off IED.
IED
Comparison with data.
• Reducing the FWHM of the IED.
• Application: In-plasma
In plasma photo-assisted
photo assisted etching
etching.
• Summary and Conclusions.
2
Introduction / Motivation
• Control of the energy of ions bombarding a substrate in
contact with plasma is crucial for plasma processing.
• The ion energy must be high enough to drive anisotropic
etching, but not too high to induce substrate damage and/or
loss of selectivity.
selectivity
• As device dimensions approach atomic level, precise control
of the ion energy distribution (not just the average ion energy)
becomes critical.
3
4
Need for Precise Control of IED
Chemical
Sputtering Eth
(I) Low energy ions : No reaction occurs
except for chemical (spontaneous)
etching, if any.
Physical
Sputtering Eth
(II) High energy ions: chemical sputtering
and physical sputtering  not wellcontrolled etching with damage.
damage
(I)
(III)
(II)
(III) Medium energy ions: Only chemical
sputtering occurs. Etching stops as soon
as the
th adsorbed
d b d etchant
t h t reacts
t away (self( lf
limiting reaction)  atomic layer control
• For atomic layer etching, precise control of IED is needed.
• Precise placement of peak ion energy.
• Narrow
Narro width
idth of ion energy
energ distribution.
distrib tion
high selectivity and low surface damage.
Goal and Approach
Goal:
p methodologies
g
gy
Develop
to achieve “tailored” ion energy
distributions1 (IEDs) on plasma electrodes (e.g., nearly
monoenergetic IED with tight FWHM).
Approach:
 Use combination of modeling/simulation and experiments.
 Particle
Particle-In-Cell
In Cell simulation with Monte Carlo Collisions (PIC(PIC
MCC).2
 Simplified model for rapid calculation of IED3.
Diomede, D. J. Economou and V. M. Donnelly J. Appl. Phys., 109, 083302 (2011); X.V. Qin, Y.-H. Ting
and A.E. Wendt, Plasma Sources Sci. Technol., 19, 065014 (2010).
2S. K. Nam, D. J. Economou and V. M. Donnelly, IEEE Trans. Plasma Sci., 35, 1370 (2007); V. Vahedi, G.
DiPeso, C. K. Birdsall, M. A. Lieberman, and T. D. Rognlien, Plasma Sources Sci. Technol. , 2, 261
(1993) JJ. Verboncoeur,
(1993);
V b
M Alves,
M.
Al
V Vahedi,
V.
V h di and
dC
C. K
K. Birdsall,
Bi d ll J.
J Comp.
C
Ph
Phys., 104,
104 321 (1993)
(1993); P.
P
Diomede et al. Plasma Sources Sci. Technol, 14, 459 (2005) .
3P. Diomede, M. Nikolaou and D. J. Economou, Plasma Sources Sci. Technol., 20, 045011 (2011).
1P.
5
IED on a RF Biased Electrode
 i 3s M 1 / 2
(
)

 rff
2 2eVs
i/rf <<1  bimodal IED
i/rf >>1  single peaked IED
i = ion transit time through sheath
rf = period of the sheath E-field
J. Coburn and E. Kay, J. Appl. Phys., 43, 4965 (1972).
6
Tailored Voltage Waveforms
M. M. Paterson, H.-Y. Chen, and A. E. Wendt, Plasma Sources Sci. Technol., 16, 257
(2007).
F. L. Buzzi, Y.-H. Ting, and A. E. Wendt, Plasma Sources Sci. Technol., 18, 025009 (2009).
X. V. Qin, Y-H Ting and A. E. Wendt, Plasma Sources Sci. Technol., 19, 065014 (2010).
P. Kudlacek, R. F. Rumphorst, and M.C. M. van de Sanden, J. Appl. Phys., 106, 073303
(2009).
Ankur Agarwal and Mark J. Kushner, J. Vac. Sci. Technol. A, 23, 1440 (2005).
S. Rauf, J. Appl. Phys., 92, 2984 (2002).
P. Diomede, M. Nikolaou and D. J. Economou, Plasma Sources Sci. Technol., 20, 045011
(2011).
DC “Boundary” Voltage
L. Xu,
L
Xu D.
D JJ. Economou
Economou, V
V. M
M. Donnelly and P
P. Ruchhoeft,
Ruchhoeft Appl.
Appl Phys.
Phys Lett.,
Lett 87,
87 041502
(2005).
H. Shin, W. Zhu, L. Xu, V. M. Donnelly and D. J. Economou, Plasma Sources Sci.
Technol., 20, 055001 (2011).
P Diomede
P.
Diomede, D.
D J.
J Economou and V
V. M
M. Donnelly
Donnelly, J.
J Appl.
Appl Phys.,
Phys 109,
109 083302 (2011).
(2011)
7
PIC MCC Simulation of Pulsed CCP Reactor with DC
PIC‐MCC Simulation of Pulsed CCP Reactor with DC Bias in Afterglow
Argon plasma
VRF = 300 V
RF = 13.56 MHz
p = 10 mTorr
d = 6 cm
10 kHz pulse
50% duty ratio
• Pulsed plasma is sustained in capacitively coupled plasma (CCP) reactor.
reactor
• DC bias is applied on the upper (boundary) electrode in the afterglow to modify the
IED on the lower (substrate) electrode.
8
l
f
h f
l
f
Application of DC Bias in the Afterglow of a Pulsed Plasma
• After plasma power turn off (afterglow), Te and Vp decay rapidly.
•Apply a synchronous positive bias Vdc during specified time window in the
g
afterglow.
• Bias raises plasma potential, influencing the IED on the wafer.
9
f cw and
d Pulsed
P l d Plasma
Pl
/ Bias
Bi
IED for
w/o
12
1.2
12
1.2
(a)
(b)
1.0
Normalized IE
ED
Normalized IED
1.0
0.8
0.6
0.4
0.2
0.8
0.6
0.4
0.2
0.0
0
50
100
150
Ion energy (eV)
(a) IED for continuous wave
(cw) argon plasma w/o bias.
200
0.0
0
50
100
150
200
Ion energy (eV)
(b) IED for pulsed argon plasma
(10 kHz, 50% duty ratio) w/o bias
Relative strength of the low and high energy
features can be controlled by varying the
duty ratio.
10
IEDs with Staircase DC Bias Applied in the Afterglow of
Pulsed Plasma (1)
• Afterglow starts at time t = 50 s.
s
• Additional peaks appear in the IED.
• Peak location can be controlled by the value of the applied bias voltage.
11
IEDs with Staircase DC Bias Applied in the Afterglow of
Pulsed Plasma (2)
IED peak strength at 50 and 100 eV can be controlled by the duration of the respective DC
bias voltage. The relative strength of the other two peaks can be controlled by the duty ratio.
P. Diomede, D. J. Economou and V. M. Donnelly, J. Appl. Phys., 109, 083302 (2011)
12
Comparison of PIC Simulation with Experimental Data (1)
11
0.03
11
4.0x10
11
t = 38 s
3.6x10
simulation
11
3.2x10
t = 28 s
0.02
t = 18 s
0.01
Ion flux (a.u.)
IEDF
experiment
4.4x10
t = 48 s
50-98s
60-98s
70-98s
80-98s
11
2.8x10
11
2.4x10
11
2.0x10
11
1.6x10
11
1.2x10
10
8.0x10
10
4.0x10
0.0
0.00
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Energy (eV)
Energy (eV)
IEDs predicted by the PIC simulation in the afterglow (right) compared to data# (left).
The low energy peak of the data is due to the active glow (not simulated by PIC).
3.0
2.5
Electron temperature in the
afterglow predicted by PIC-MCC (line),
compared to data1 (points).
Te (eV)
20
2.0
1.5
simulation
1.0
experiment
0.5
Pulsed argon plasma, 10
KHz modulation, 20% duty,
14 mTorr, 120 W average
power, 24.4 V DC bias
applied in afterglow
afterglow.
0.0
30
40
50
60
70
Time (s)
80
90
1H.
100
Shin, W. Zhu, V. Donnelly and D. Economou, PSST, 20 055001 (2011).
13
Comparison of PIC Simulation with Experimental Data (2)
1.0
1.0
simulation
0.8
0.6
0.4
02
0.2
0.0
Normalized IED
D
0.8
Normalized IED
D
+
H2
+
H3
simulation
0.6
0.4
02
0.2
0.0
0
50
100
150
Ion energy (eV)
experiment
0
50
100
150
Ion energy (eV)
Hydrogen plasma
VRF = 300 V
RF = 13.56 MHz
p = 50 mTorr
d = 6 cm
experiment
Experiment: D. O’Connell et al. Phys. Plasmas 14, 103510 (2007)
Simulation: P. Diomede et al., submitted
14
Electrode in Contact with Plasma
Bulk Plasma (n0, Te)
Sheath
Electrode
(Target)
Blocking capacitor, Cb
Applied rf, Vrf
15
Equivalent Circuit Model
A. Metze et al., J. Appl. Phys., 60, 3081 (1986).
Vrf
Cb
Subscripts T and G refer to “target” and
“ground” electrodes, respectively.
VT
d
(Vrf  VT )  CT
dt
d
CT (VP  VT )  CG
dt
Cb
IT
CT
VP
IG
CG
d
(VP  VT )  IT  0
dt
d
VP  IT  I G  0
dt
Vd  (VT  V p ) Ions respond to
dVd

a “damped” potential Vd
dt
i
Voltage Vrf is applied through blocking capacitor, Cb.
Gi
Given
n0, Te, Vrf and
d Cb, calculate
l l t VT , Vp, and
d Vd.
16
Ion Energy Distribution
Support (y,y+dy)
Vd(t)
t
0
0
y+dy
y
2

dy
IED  f ( y ) 
1
2

# of points in 0<t  2
such that
Vd (t )  y
1
dVd
d (t )
Vd  Vd (t )  " damped " sheath potential
Sample damped sheath
potential waveform
P. Diomede, M. Nikolaou and D. J. Economou., Plasma Sources Sci. Technol., 20, 045011 (2011).
E. Kawamura, V. Vahedi, M. A. Lieberman and C. K. Birdsall, Plasma Sources Sci. Technol., 8, R45
(1999).
17
IED for a sinusoidal sheath potential
(Forward problem)
p sheath p
Damped
potential
V(t)
g IED
Resulting
t
0
0

f(y)
2
Vm+V0
Vm
|Vm|-V0
0
y  Vd ( t )  Vm  V0 sin  t
Vm  V0 ,
y
0
Vm-V0
Vm  0, 0   t  2
f ( y) 
1

Vm
|Vm|+V0
1
 y  Vm 
V0 1  

V
0


2
18
Voltage Waveform Required to Obtain Desired IED
(Inverse problem)
2×1010 cm-3,
Cb = 500 pF, n0 =
Te = 3 eV,
+
0a
u ((Ar ), AG//AT = 20,
0, i = 1
M = 40
amu
Desired IED is Gaussian with specified
peak and standard deviation.
Vrf = rf voltage before blocking cap
Vd = “damped” sheath potential
VT = target electrode potential
Required voltage waveform Vrf is a
“rectangular” pulse with a slope. Slope is
needed because of blocking capacitor
charging.
P. Diomede, M. Nikolaou and D. J. Economou, Plasma Sources Sci. Technol., 20, 045011 (2011).
19
Comparison of Model Predictions with Experimental Data
Argon Plasma
1.2
1.2
experiment
model
1.0
0.8
Normalized IED
Normalized IED
1.0
0.6
0.4
0.8
0.6
0.4
0.2
0.2
0.0
0.0
0
5
10
15
20
25
30
0
5
10
15
20
25
30
Ion energy (eV)
Ion energy (eV)
IED predicted by the model (right) compared to data# (left).
22
model
20
Plasma potential w/o DC bias
predicted by the model (line),
compared to data# (points).
18
16
Vp (V)
14
12
10
experiment
8
6
Pulsed plasma, 10 KHz
modulation, 20% duty, 14
mTorr 120 W average
mTorr,
power, 24.4 V DC bias
applied in afterglow during
µs.
∆tb = 45-95 µ
4
2
# H. Shin, W. Zhu, L. Xu, V. Donnelly and D. Economou,
0
0
20
40
60
Time (s)
80
100
PSST, 20 055001 (2011).
20
Tailored Bias Waveform: actual vs. model
Actual waveform
X. Qin, Y. Ting and A. E. Wendt, PSST, 19, 065014 (2010)
Model input
21
Tailored Bias Waveform: Comparison of
M d l with
ith Experiment
E
i
t
Model
Experiment
Model Prediction
X. Qin, Y. Ting and A. E. Wendt, PSST, 19, 065014 (2010)
22
Experimental Apparatus
Boundary Electrode
Ion energy analyzer
at Z= 170mm
 Differentially pumped ion energy
analyzer (IEA)
Monochromator
periscope
 Movable Langmuir probe (LP)
Ion energy can be manipulated by applying DC bias to the Boundary
Electrode.
23
Timing scheme for pulsed plasma with synchronous DC boundary voltage in afterglow
Positive DC bias (+24V)
rf power
10kHz Modulation
20µs
ON
80µs
OFF
24
IEDs of Pulsed Plasma with Synchronous DC Boundary Voltage
Boundary Voltage

gy broad p
Low energy
peaks
are due to active glow and
part of afterglow w/o bias;
High energy sharp peaks
are due to DC bias in the
afterglow.

Separation of the peaks
can be tuned by DC bias
value and pressure
0.02

Narrow IED can be
achieved in the afterglow.
0.01

Full width at half maximum
(FWHM) of the IED ranges
from 1.7 to 2.4 eV and
scales with Te.
0.05
0.04
7 mTorr
14 mTorr
28 mTorr
50 mTorr
Due to DC bias
IED
Plasma potential peak
0.03
0.00
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Energy (eV)
25
Te and Vp Change During Bias
 Time-resolved Langmuir probe
measurements
ON
4.0
16
OFF
14
3.5
12
25
2.5
10
2.0
8
Vp(V)
Te(eV))
3.0
1.5
6
10
1.0
4
0.5
2
0.0
0
10 20 30 40 50 60 70 80 90 100
time (s)
0
0
10 20 30 40 50 60 70 80 90 100
time (s)
The smaller the electron temperature and the variation of plasma potential
during the biasing window,
window the sharper the expected ion energy
distribution.
26
Early vs. late afterglow bias
Early vs. late afterglow bias
0 04
0.04
22-60s
32-60s
42-60s
t = 38 s
0.03
IED
t = 28 s
0.02
t = 18 s
0 01
0.01
0.00
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Energy (eV)
• Early afterglow biasing (Vp changes considerably and Te is high) vs. late
afterglow biasing (Vp and Te have decayed to low values).
• FWHM is smaller for late afterglow biasing.
27
IEDs for Different Noble Gases
0.5
0.4
Ar
Kr
Xe
FWHM of the IED follows the
order Ar < Kr < Xe. Te for the
three gases follows the same
order
d iin th
the afterglow.
ft l
IED
0.3
02
0.2
0.1
0.0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Energy (eV)
28
29
Application of Precise Control of IED
 Using a nearly mono-energetic
mono energetic IED with a
narrow width, we investigated ion-assisted
etching at near-threshold
near threshold energies
energies.
 We
W discovered
di
d th
the surprising
i i result
lt th
thatt p-
type silicon etches in chlorine-containing
plasmas
l
even when
h th
the iion energy iis b
below
l
threshold.
H. Shin, W. Zhu, V. M. Donnelly and D. J. Economou, submitted
Setup for etching study
30
Bi
bl
Biasable
boundary
electrode
rf coil for ICP
Faraday
shield
Spectrometer
“p-type”
Si sample
Periscope for
OES
Cooling
water
photodiode
IR laser 1.31µm
• The Si 2881 Å line was used to
monitor etching.
• Etching rate was measured by IR
laser interferometry.
• Ion flux was measured to calculate
the etching yield.
• Load-lock was used for clean and
reproducible environment.
Si Etching vs. Ion Energy
14mTorr
28mTorr
50mTorr
60mTorr
5
700
600
ER(E) at 50mTorr
4
500
3
400
Eth
Y(40eV)=0.41
300
2
Y(30eV)=0.14
200
Y(0eV)=0
1
100
0
0
0
1
2
3
E
4
1/2
5
1/2
(eV )
6
7
8
ER (Å/m
min) @ 500 mTorr
Relaative ER
R or Si e mission
6
31
No spontaneous chemical etching
30nm SiO2 mask
p‐type Si
50mTorr 1% Cl2 Ar pulsed plasma with synchronous bias, 40V.
• p‐type Si is known not to etch by Cl or Cl2 spontaneously [Mogab and Levinstein (1979), Ogryzlo et al. (1990), Flamm (1990)].
• There in no mask undercut.
Th
i
k d
t
• The sub‐threshold etching is NOT due to spontaneous chemical (isotropic) etching.
32
Low Energy (<Eth) Ion Etching ?
Low energy ions can create electrons by Auger neutralization
Grids
G id
VA
VB
Vp
C
Substrate (VC)
-5V
A
B
 Substrate bias was
used to repel ions
Fast neutrals by charge exchange can be
safely ignored at the low pressure of 7
mToor
mToor.
Grids A and B are biased at -5 V
to repel plasma electrons.
33
Low Energy (<Eth) Ion Etching ?
Intensity ((arb. units)
2.5
+30V (I=-3.5mA)
0V (I=-0.1mA)
-30V (I=0.7mA)
2.0
Si
Si
1.5
SiCl
Si
1.0
0.5
2500
2600
2700
2800
2900
Wavelength (Å)
• With -30V substrate bias (ion-assisted etching with Ei=30+Vp), the etching
rate increases considerably over the grounded substrate condition
condition.
• With +30V substrate bias (no ions bombarding the substrate), the etching
rate is the same as with the grounded substrate.
• The sub-threshold etching is due to photons from the plasma glow.
34
35
Further Investigation of Photo‐assisted Etching (PAE)
(b)
• Etching in 97% Ar/ 3% Cl2 CW
plasma
l
(300W)
(300W).
• Same plasma and neutral flux under
the q
quartz roof,, but different light
g
illumination under the opaque vs. the
transparent (>170nm) parts of the
quartz roof.
• Etching depth measured with a step
profiler.
VUV is Responsible for PAE
Ettched deptth (nm)
Opaque (p-Si)
Transparent(p-Si)
160
140
120
100
80
60
40
20
0
1
2
Position (#)
3
•
The p-type Si etching rate under
the opaque roof is much smaller
due to weaker light illumination.
•
The p-type Si etching rate under
the transparent quartz roof (105
Å/min), is only a fraction ((<10%)
10%) of
what was expected (~1200 Å/min)
at the conditions of this experiment
(3% Cl2, 300W)
This implies that photo-assisted etching is dominated by the
photons blocked by quartz, i.e., VUV below 170nm.
36
Summary and Conclusions
1.
2.
2
3.
4.
5.
6.
7.
8.
Tailored voltage waveforms can be constructed to obtain desired
IEDs on a plasma electrode.
PIC-MCC
off IEDs.
PIC MCC simulation
i l ti can provide
id accurate
t predictions
di ti
IED
An equivalent circuit model gives good results much more rapidly.
Nearly monoenergetic IEDs can be obtained using synchronous
DC bias in the afterglow of a pulsed plasma.
The peak value and the FWHM of the IED can be controlled by
varying DC bias and the time window it is applied, operating
pressure, and carrier gas.
The FWHM of the IED scales with the electron temperature.
Using
g monoenergetic
g
IEDs with tight
g FWHM to etch p
p-type
yp silicon
with chlorine at near-threshold ion energies led to the surprising
result that in-plasma photo-assisted etching (PAE) is very
important.
PAE is dominated by VUV photons.
37