Measurement of Plasma Parameters in an Inductively
Coupled Plasma Reactor
Hiroshi Sasaki, Kenichi Nanbu and Masayoshi Takahashi
Institute of Fluid Science, Tohoku University, Sendai, Japan 980-8577
Abstract. The objective is to deepen the understanding of basic plasma properties in a high-density inductively coupled plasma(ICP) reactor. The experiment using a Langmuir probe, which is made of tungsten
wire having length of 2.3 mm and diameter of 0.65 mm, is carried out to determine the characteristics of
argon plasma. The axial and radial distributions of the plasma density, electron temperature, and plasma
potential are measured. It was found that the electron density is in the range of 1.0 x 1010 to 1.0 x 1011cm~3,
electron temperature 4 ~ 6 eV, and plasma potential 35 ~ 45 V, respectively. These plasma parameters are
spatially nonuniform in the axial direction; they depend on the distance between the plasma source and the
measuring point. Next, etching of polycristalline silicon wafer by chlorine plasma is studied. The results
obtained in this study indicate the effects of rf coil power, gas pressure, and substrate rf bias power on the
distributions of the etch rate.
INTRODUCTION
The size of microscopic features on electronic devices is becoming smaller and smaller whereas the size of
wafers is becoming larger and larger to get more chips from one wafer. Increase in wafer size requires opening
new fabrication facilities [1]. Plasma etching technology is highly expected to be much more advanced. Until
quite recently, the development of etching technologies is mainly conducted based on experiences and limited
measurements of basic plasma properties. However, in a near future such a stance will be confronted with a
difficulty in the design of advanced etching apparatuses.
Gas inlet
Rotary orifice
:
power
M ^ Rf
k
RF coil ———*O
(8 turns)
Q
o
i
«*——
-
x*
Langmuir S
o ^
m
probe
«— \
200
0
D
or-
r
.—k,
1
Z
1
]/
•\A .-:.»~
Grid
ring
TABLE 1. Measuring positions. The origin z = 0 is at the top of the diffusion
chamber, whose height H is 400 mm.
Port No.
11
10
8
6
4
2
z(mm)
200
210
230
250
270
290
z/H
0.500
0.525
0.575
0.625
0.675
0.725
'
To vacuum pump
FIGURE 1. Schematic of ICP reactor.
CP585, Rarefied Gas Dynamics: 22nd International Symposium, edited by T. J. Bartel and M. A. Gallis
© 2001 American Institute of Physics 0-7354-0025-3/01/$18.00
262
In this work, an inductively coupled plasma, whose density is high even at lower gas pressure, is studied
experimentally to clarify how radicals formed in the plasma diffuse and are transported to the wafer. In order
to completely solve this problem, not only the experiment but also numerical analyse based on accurate particle
modeling [4,6] may be necessary at the same time. First, we constructed an inductively coupled plasma reactor
and examined the basic properties of argon plasma [7], Next, we started studying the etching properties of
polycrystalline silicon wafer by chlorine plasma. The effects of rf coil power, substrate rf bias power, gas
pressure, and gas flow rate on the etch rate and its distribution are examined. A microwave interferometer and
a surface profiler(Tencor) are used to measure the electron density and the etch rate, respectively.
EXPERIMENTAL
The schematic of the inductively coupled plasma reactor is shown in Fig.l. The origin of the coordinate
system is at the center of the entrance to the diffusion chamber. Let the r-axis be in the radial direction and
z-axis be in the direction of the downward normal at the entrance. The reactor consists of a plasma source
chamber (z < 0), diffusion chamber (z > 0), pumping system, and radio-frequency (rf) power source. The
plasma source chamber is made of a quartz tube having inner diameter of 200 mm and height of 400 mm.
The diffusion chamber is made of a stainless steel having inner diameter of 348 mm and height of 400 mm.
The pumping system is driven by a turbo-molecular pump (210^/sec) and an oil-rotary pump (162^/min).
Before starting measurements the diffusion chamber is evacuated up to pressure o f 2 . 0 x l O ~ 5 Pa. The working
pressure is in the range of 0.2 - 2.0 Pa. The quart tube is wound by 8-turns helical coil. The diameter of the
helical coil is 242 mm and the length is 274 mm. The coil is cooled by water to suppress the heating. The
maximal input rf power of the coil is 500 W and the frequency is 13.56 MHz. Argon or chlorine gas is fed to
the plasma source chamber through a rotating disk with many orifices which works for supplying gas uniformly
from the top of the source chamber. The rotating disk has diameter of 198 mm, thickness of 5 mm, and 1139
latticed orifices with diameter of 1 mm. It is rotated at a desired speed clockwise or counterclockwise. Maximal
speed is ±32 rpm. An annular aluminum plate is installed between the diffusion chamber and the entrance of
the pumping system to realize a spatially uniform pumping. On this plate 24 holes of diameter of 10 mm are
drilled.
A movable substrate holder has diameter of 156 mm. It can be moved vertically by ±50 mm from the standard
position (z = 250 mm). In this work the holder position is fixed at z = 300 mm. The diffusion chamber has
11 ports for insertion of the Langmuir probe and a pair of view-ports for a microwave interferometer. There is
still one port for a quadrupole mass spectrometer. The Langmuir probe (PMT Inc.) made of tungsten wire has
length of 2.3 mm and diameter of 0.65 mm. This probe has an rf choke-coil for removing the noise in the I-V
curve due to the radio frequency. The plasma potential, electron density, and electron temperature are obtain
from the I-V curve. The probe can be quickly moved by compressed air to an arbitrary radial location in the
chamber. The 20 data in the radial direction (r < 300 mm) can be obtained in one shot of the probe. The
required time is 200 msec. The vertical distributions of the plasma parameters are obtained by changing the
probe inserting port. The positions of measuring points z and the positions normalized by diffusion chamber
height(jy = 400 mm) z/H are given in Table 1. In the experiment, gas pressure is in the range of 0.5 - 2.0 Pa,
gas flow rate 5 - 100 seem, and rf coil power 100 - 500 W with 13.56 MHz. The bias power for the substrate
holder applied for chlorine plasma is in the range of 30 - 100 W with the same frequency. Gas pressure control
is necessary to examine the effects of pressure on the plasma parameters. A conductance gate valve is used for
this purpose at a constant flow rate(20 seem). The effects of the flow rate on the plasma parameters are also
examined in the range of 20 - 100 seem.
RESULTS AND DISCUSSION
A. Argon plasma
First we present the data for argon ICP. The electron density 7Ve, electron temperature Tej and plasma
potential Vp are plotted as a function of gas pressure in Fig.2. The conditions are: rf coil power is 300 W,
263
gas flow
flow rate
rate 20
20 seem,
sccm, and
and the
the measuring
measuring position
position zz =
= 230 mm and r~0
r∼0 . As is shown in Fig.2, the electron
gas
temperature
decreases
with
increasing
pressure
from
5.5
eV
at
0.5
Pa
to 4.3
temperature decreases with increasing pressure from
at
4.3 eV
eV at
at 2.0
2.0 Pa.
Pa. The
The decrease
decrease in
in the
the
electron temperature
temperature is
is due
due to
to inelastic
inelastic collisions
collisions at
at higher
higher gas
gas pressure. The electron density increases with
electron
−33
10
10
cm−33 at
at 0.5
0.5 Pa
Pa to
to 8.7 ×
cm
at 2.0
increasing the
the gas
gas pressure
pressure from
from 6.0
6.0 x
× 1010
increasing
cm~
x 10
1010
cm~
2.0 Pa. The electron density is
is
lower than
than that
that we
we expected.
expected. One
One of
of the
the reasons
reasons for
for lower
lower plasma
plasma density
density is that
that the
the measured
measured position
position is far
lower
from the
the plasma
plasma source
source chamber.
chamber. Next,
Next, let
let us see the
the plasma potential. It decreases as
from
as the
the pressure
pressure increases.
well-known that
that lower
lower plasma
plasma potential and
and lower
lower electron
electron temperature are desirable in etching processes
ItIt isis well-known
[2]. In
In this
this sense,
sense, higher
higher gas
gas pressure appears to
to be recommendable. Of course, the gas pressure
[2].
pressure should
should be
be as
as
low as
as possible
possible for successful
successful high
high aspect
aspect ratio etching.
etching. Figure 3 shows the
the plasma potential, electron density,
low
and electron
electron temperature
temperature as
as aa function
function of the
the rf coil
coil power
power for
for 20
20 seem
sccm and
and 1.0 Pa at z/H
and
z/H = 0.575. The
power isis the
the one
one applied
applied to
to the
the matching
matching network;
network; reflected
reflected power
power is nearly 0 W. The plasma potential
rfrf power
increases with
with the
the rfrf coil
coil power
power from
from 37
37 V
V at
at 100 W to
to 47 V
V at
at 500 W. Such high potential is undesirable in
increases
10
10
−33
cm−33 at
at 100 W to
cm
at
etching. The
The electron
electron density
density increases
increases from
from 4 ×
etching.
x 1010
cm~
to 99 ×
x 10
1010
cm~
at 500
500W.
W. The
The electron
electron
temperature also
also increases
increases slightly
slightly from
from 4.6 eV at
at 100 W to 5.2 eV at 500
temperature
500 W
W as
as the
the rf
rf coil
coil power
power increases.
increases.
T (eV), N (×10
V (V)
p
e
p
e
V (V)
e
e
T (eV), N (×10
1.0
1.5
Gas pressure (Pa)
-3
-3
cm )
cm )
10
10
0.50
2.0
0
FIGURE
FIGURE 2.
2. Electron
Electron density
density N
Nee,, electron
electron
temperature
and plasma
plasma potential V
Vpp vs
temperature T
Tee,, and
gas
gas pressure.
pressure.
100
200 300 400
RF power (W)
500
0
600
FIGURE
3. Plasma potential VVpp, electron
FIGURE 3.
density N
Tee vs
Nee,, and electron temperature T
rf
rf power.
T (eV), N (×10
V (V)
e
p
p
e
V (V)
e
e
T (eV), N (×10
- 35
0.65
0.70
0.75
0
FIGURE
FIGURE 4.
4. Axial
Axial distributions of plasma
potential
V
,
electron
Nee,, and
potential Vpp , electron density N
electron
electron temperature
temperature T
Tee..
20
40
60
80
100
120
Gas flow rate Q. (seem)
Normalized axial position Z/H
FIGURE 5. Plasma potential VVpp, electron
temperature T
Tee, and electron density
density N
Nee vs
vs
gas flow rate.
264
-3
0.60
cm )
-3
cm )
10
10
0.55
- i
The axial distributions of the electron density, electron temperature, and plasma potential are shown in Fig.4.
The axial distributions
of the
temperature,
and plasma
shown
in Fig. 4.
Experimental
conditions are
2.0 electron
Pa, 400 density,
W, andelectron
20 sccm.
The normalized
axial potential
position,are
z/H,
is measured
Experimental
are of
2.0the
Pa,diffusion
400 W, and
20 seem.
The
normalized
axial
is measured
downward
from conditions
the entrance
chamber.
As is
shown
in Fig.4,
theposition,
electronz/H,
density
decreases
downward
fromthe
theaxial
entrance
of the
diffusion
chamber.
11
−3 As is shown in Fig. 4, the electron
10
−3 density decreases
with
increasing
position
from
1.0 × 10
ncm 3 at z/H = 0.575 to 2.3 × 10 10 cm 3 at z/H = 0.725.
with
increasing
the
axial
position
from
1.0
x
10
cm"
at
z/H
=
0.575
to
2.3
x
10
cm~
at z/H
= wall
0.725.
The main reason of this decrease is that electrons which reached the chamber wall by diffusion
on the
at
The
main
reason
of
this
decrease
is
that
electrons
which
reached
the
chamber
wall
by
diffusion
on
the
wall
the phase when the oscillating plasma potential is lowest. The plasma potential decreases from 37 V to 33atV
thethe
phase
when temperature
the oscillating
plasma from
potential
is lowest.
TheFigure
plasma
potential
fromflow
37 rate
V to on
33the
V
and
electron
increases
4.3 eV
to 5.4 eV.
5 shows
thedecreases
effect of gas
and the electron temperature increases from 4.3 eV to 5.4 eV. Figure 5 shows the effect of gas flow rate on the
plasma potential, electron density, and electron temperature at z/H = 0.725 for 2.0 Pa and 500 W. The plasma
plasma potential, electron density, and electron temperature at z/H — 0.725 for 2.0 Pa and 500 W. The plasma
potential is nearly constant in the range of 20 – 100 sccm. The electron temperature and electron density are
potential is nearly constant in the range 10
of 20 - 100 seem. The electron temperature and electron density are
also
nearly constant, 5.3 eV and 3.5 × 10 cm−3 , respectively. We can see from Fig.5 that the change in the
also nearly constant, 5.3 eV and 3.5 x 1010cm~3, respectively. We can see from Fig. 5 that the change in the
gas
versus the
the normalized
normalized
gasflow
flowrate
ratedoes
doesnot
notaffect
affectthe
theplasma
plasmaparameters
parameters very
very much.
much. The
The electron
electron density
density versus
radial
position,
r/R,
(R=
78
mm:
radius
of
substrate
holder
)
are
shown
in
Fig.6.
Experimental
conditions
radial position, r/R, (R= 78 mm: radius of substrate holder ) are shown in Fig. 6. Experimental conditions
are
0.5
Pa,
100
W,
20
sccm,
and
z/H
=
0.500
−
0.725.
Each
profile
is
relatively
uniform
except
the
one at
at
are 0.5 Pa, 100 W, 20 seem, and z/H = 0.500 — 0.725. Each profile is relatively uniform except the one
z/H
=
0.725.
This
location
is
close
to
the
substrate
at
z/H
=
0.750.
A
good
uniformity
is
due
to
a
large
z/H = 0.725. This location is close to the substrate at z/H = 0.750. A good uniformity is due to a large
coefficient
coefficientnear
nearthe
thecenter
centerofofthe
thechamber
chamber in
inthe
the case
case of
of low
low pressure
pressure discharge
discharge [3].
[3] . It
It isis seen
seen that
that the
the electron
electron
density
at
z/H
=
0.725
rapidly
decreases
for
r/R
<
−0.8.
We
suppose
that
this
is
caused
by
an
asymmetrical
density at z/H — 0.725 rapidly decreases for r/R < -0.8. We suppose that this is caused by an asymmetrical
pumping
from z/H
z/H =
= 0.500
0.500 to
to 0.625,
0.625,
pumpmgofofgas.
gas.The
Theelectron
electrondensity
densityappears
appearsto
to be
be nearly
nearly the
the same
same in
in the
the range
range from
but
the axial
butasa^the
axialdistance
distanceincreases
increasesmore,
more,the
the electron
electron density
density decreases
decreases gradually.
gradually.
111
0.5F 'a, 100W 20sccm
i oow, ;Osccm
O.SPa
L«
I
' ' ' j ' ' '
;
e
p
Plasma potential V (V)
\
-3
Electron density N (cm )
10'
fc
o
cu
w
o
10
10
:
D
:
Q
x
o| o
tf%P[
D
D 0
c
X
X
x
fo
S
ta
y
10
-1.2
i i i
, , ,
, , ,
-0.80 -0.40
0.0
X
0
X
0
^o a
o V
< x
a
n:z/H40.500 :
O: =5.525 :
V:
=5.575
A:
=0.625
O: =5.675 '
x: ( ^p.725 (
0.40
0.80
o
o
<
^
35
%
tn
«
30
^
25
?n
1.2
-1.2
Normalized radial position r/R
FIGURE6.6.Radial
Radialdistributions
distributionsofofelectron
electron
FIGURE
density.
density.
40
PQ
___
, , , 1, , , 1 , . ,
A
'
D: z/H|=0.500 0:
=K).525 :
V:
=0.575 :
A:
=0.625 I
=0.675 O:
x:
=0.725 :
-0.80 -0.40
0.0
0.40 0.80
Normalized radial position r/R
1.2
FIGURE 7.
7. Radial
Radial distributions
distributions of
FIGURE
of plasma
plasma
potential.
potential.
Figure 7 shows the radial distributions of the plasma potential under the same experimental conditions as
Figure 7 shows the radial distributions of the plasma potential under the same experimental conditions as
that for Fig.6. The plasma potential is about 35 - 40 V in the range from z/H — 0.500 to 0.725. As the axial
that
for Fig.6.
The plasma
potential
about 35keeping
– 40 V the
in the
range from
= 0.500
to 0.725.
axial
distance
increases,
the potential
levelisdecreases
uniformity.
Thez/H
radial
distributions
of As
the the
electron
distance
increases,
the
potential
level
decreases
keeping
the
uniformity.
The
radial
distributions
of
the
electron
temperature under the same conditions are shown in Fig.8. The electron temperature is 5 - 6 eV in the range
temperature
under
same
are shown
in Fig.8.
The
electron
temperature is 5 – 6 eV in the range
of z/H = 0.500
—the
0.725.
Theconditions
distributions
are uniform
in the
radial
direction.
of z/H = 0.500 − 0.725. The distributions are uniform in the radial direction.
The disk with the latticed orifices is rotated clockwise or counterclockwise at speed of ±20 rpm to see the
The disk
with theonlatticed
orifices
is rotatedFigure
clockwise
or counterclockwise
at speed
of rpm
±20 and
rpm0 to
see We
the
effects
of rotation
the plasma
parameters.
9 shows
the electron density
for ±20
rpm.
effects
of
rotation
on
the
plasma
parameters.
Figure
9
shows
the
electron
density
for
±20
rpm
and
0
rpm.
We
see that the electron density distributions are very flat and uniform. It is reasonable to conclude that there
seeis that
the
electron
density
distributions
are
very
flat
and
uniform.
It
is
reasonable
to
conclude
that
there
no effect of the rotation in the range of this disk speed. The radial distributions of the plasma potential
is and
no effect
of temperature
the rotation are
in the
range
of this
speed.
The radial
distributions
of the plasma
electron
shown
in Figs.
10 disk
and 11,
respectively.
Experimental
conditions
are thepotential
same as
and
electron
temperature of
arethe
shown
in Figs.10
respectively.
Experimental
as
Fig.9.
The distributions
plasma
potentialand
and11,
electron
temperature
are nearlyconditions
flat, i.e. 38are
V the
and same
1.5 eV,
Fig.9.
The distributions
of the
plasma
and electron
temperature
are say
nearly
i.e. 38 give
V and
eV,
respectively.
Again there
is no
effect potential
of the rotation.
Consequently,
we can
thatflat,
Figs.9-11
the1.5
error
respectively.
Again
there is no Note,
effect however,
of the rotation.
we did
cannot
sayimprove
that Figs.9–11
giveuniformity
the error
bound for our
measurements.
that the Consequently,
effect of rotation
the spatial
bound
for the
our latticed
measurements.
Note, however,
that the effect
of realizing
rotation the
did uniformity.
not improve the spatial uniformity
because
orifice already
played a sufficient
role for
because the latticed orifice already played a sufficient role for realizing the uniformity.
B. Silicon etching by chlorine plasma
B. Silicon
by etching
chlorine
plasmaof the silicon wafer by chlorine plasma. The diameter of the wafer is
We nowetching
discuss the
properties
We now discuss the etching properties of the silicon wafer by chlorine plasma. The diameter of the wafer is
265
12
I
-
-3
O.SPa, 1 0OW, 2Osccm
o 10
§
Electron density N (cm )
10
e
Electron temperature T (eV)
152.4 mm. The wafer is set on the water-cooled substrate holder at the position of z/H = 0.625. The rf bias
152.^mm. The wafer is set on the water-cooled substrate holder at the position of z/H = 0.625. The rf bias
with the frequency of 13.56 MHz is applied to the holder. Figure 12 shows the relation between the etch depth
with the frequency of 13.56 MHz is applied to the holder. Figure 12 shows the relation between the etch depth
and etching time near the center of the wafer. The etching conditions are as follows: Chlorine pressure is 0.5
and ©tching time near the center of the wafer. The etching conditions are as follows: Chlorine pressure is 0.5
Pa,
Pa,RFC
RFCcoil
coilpower
power500
500W,
W,gas
gasflow
flowrate
rate2020scam,
scam,and
andrfrfsubstrate
substratebias
biaspower
power 100
100W.
W.The
The etch
etchdepth
depth increases
increases
linearly
with
time.
The
distributions
of
the
etch
rate
of
Si
wafer
are
shown
in
Fig.13.
The
distance
linearly with time. The distributions of the etch rate of Si wafer are shown in Fig. 13. The distance rr from
from the
the
center
where xy−plane
wafer. Measuring
Measuring points
points
centejjofofthe
thewafer
waferalong
alongthe
thex−and
x—andy−axis
y—axisisisthe
the abscissa,
abscissa,where
xy—plane is
is on
on the
the wafer.
are
= 0 0mm,
pressure is
is 0.5
0.5 Pa,
Pa, rf
rf coil
coil power
power 500
500 W,
W,
arer ruz=
mm,±20
±20mm,
mm,and
and ±40
±40 mm
mm in
in the
the two
two directions
directions3.. Chlorine
Chlorine pressure
2.0Pa 400W,2 Osccm, : :20rpm, :j/H=0.500
o
X
6
X
71
04
CO
4
CD
D: z/H= .0.500
| O:
=0.525
I V:
=p.575
. ____ L-A- _ -JQ...6.2.5-1 O:
=10.675
2
5
"
, , ,
-1.2
, , ,
-0.80
co
v 6
TJS> Vi-< >y rjy
Q
fa
A
y
^rr"^ ^oX°<> ^o Y^
c
-0.40
0.0
0.40
0.80
,*,«,^a
1Q .
C/j["~
TS~
X_QyX._.
^
Q
En
. . . 1 *=. ,=P-725,
, , ,
I
-
X: clockwist
O: counterc lockwise
A: 0 rpm
e
8
:
-
10
10
,
-1.2
1.2
,
,
, , ,
, , ,
-0.80 -0.40
0.0
, , .
0.40
1.2
0.80
Normalized radial position r/R
Normalized radial position r/R
FIGURE8.8.Radial
Radialdistributions
distributionsofofelectron
electron
FIGURE
temperature.
temperature.
FIGURE 9.
9. Effect
Effect of
of orifice
orifice rotation
rotation on
FIGURE
radial distributions
distributions of
of electron
electron density.
density.
radial
PQ
ID
50
10
2.0P£ , 400W, feOsccm, ±20rpm, z/H=0.5 00
fafc
p
45
;
x: clockwise
0: counterc ockwise
A: 0 rpm
e
Plasma potential V (V)
-
Electron temperature T (eV)
0
CD
.
.
x
4
°
a
8
\
^
A
5x
CD
X;
30
-1.2
fa
-0.80
, , ,
-0.40
0.0
. , ,
, , ,
0.40
, , ,
0.80
1.2
O: counterc ockwise
A: 0 rpm
-
4
^ A A ^x ^
A ^ K A X Q LX
_O~ — — -^ '-o —— 0-^ I----X....Q...: •--•o— •••••
AX
A
X
-
2
:
' , , , 1 , , , 1 , , ,
0
-1.2
-0.80
-0.40
0.0
0.40
0.80
1.2
Normalized radial position r/R
Normalized radial position r/R
FIGURE 10. Effect of orifice rotation on
FIGURE
10. Effectofofplasma
orifice rotation
radial distributions
potential.on
radial distributions of plasma potential.
!
P
35
, , ,
2.0Pi, 400W, ?0sccm, j± 20rpm, z/H=0.5 30
6
CO
AX A> ; AX A
« « x .x
0^ 0
°
!.«* 0
CQ
;
FIGURE 11. Effect of orifice rotation on
FIGURE
11. Effectofofelectron
orifice temperature.
rotation on
radial distributions
radial distributions of electron temperature.
and gas flow rate 10 seem. The etching time is five minutes. It is clear that the distributions of the etch rate
anduniform.
gas flow Figure
rate 1014
sccm.
The
time
is five
is clear that
the distributions
etchthe
rate
are
shows
theetching
changes
in the
etchminutes.
rate andItsubstrate
peak-to-peak
voltage of
Vppthe
with
rf
rf
are
uniform.
Figure
shows
the changes
the increasing
etch rate and
substrate
pp with
coil
power. The
etch14rate
decreases
rapidlyinwith
the coil
power peak-to-peak
in the regime voltage
of the rfVcoil
powerthe
less
coil
power.
Thebeyond
etch rate
decreases
with increasing
the coil
in thethe
regime
of the rf
less
than
300 W,
which
the etchrapidly
rate decreases
very slowly
withpower
increasing
coil power.
Oncoil
thepower
contrary,
than
W, beyond
which the
etch rate
decreasesin
very
theand
coilthen
power.
On the contrary,
the 300
substrate
peak-to-peak
voltage
is unchanged
the slowly
regimewith
less increasing
than 300 W
Vpp decreases
rapidly
the
substrate
peak-to-peak
is unchanged
thethese
regime
less than
and thenofVthe
with
a further
increase ofvoltage
the power.
Reflectionin on
suggests
that300
theWdecrease
etch raterapidly
is due
pp decreases
with
a further
increase
the power.
Reflectionvoltage.
on theseNext,
suggests
that thebetween
decrease
the rate
etchand
ratethe
is gas
due
to the
reduction
of theofsubstrate
peak-to-peak
the relation
theofetch
topressure
the reduction
of in
theFig.15.
substrate
relation
theis etch
rate rf
and
gas
is shown
The peak-to-peak
experimental voltage.
conditionsNext,
are asthe
follows:
Gasbetween
flow rate
10 seem,
coilthe
power
pressure
shown inbias
Fig.15.
The
conditions
as follows:
is 10 with
sccm,increasing
rf coil power
500 W, issubstrate
power
50experimental
W, and etching
time fiveare
minutes.
The Gas
etch flow
rate rate
increases
gas
500 W, substrate bias power 50 W, and etching time five minutes. The etch rate increases with increasing gas
266
2500
300
0.5P|J, 500W , 20scc m, bias 100W,
O.SPa, 500W, 1 0$ccm, bias SOW
__
1500
X
Etch rate (Å/min)
i
2000
Etch depth (Å)
, , , , , , , , , 1 , , , , , , , , ,
_^_
>J
H 200
.—x —.
1000
500
>:
X
<
250
O: x-direction
D: y-direction
|
o
t
,
8
§ ;
0"
Q
ID
*
!
|
4
6
0
8
100
-50
12
10
Time (min)
FIGURE 12. Relation between etch depth and
FIGURE
12. Relation between etch depth and
etching time.
etching time.
150
0.0
Radial position r(mm)
50
iURE 13. Distributions of the etch rate
FIGURE
wafer. 13. Distributions of the etch rate
of Si wafer.
w
a
- 200
100
200 300 400
RF coil power (W)
500
600
ID
PQ
Etch rate (Å/min)
pp
Substrate V
Etch rate (Å/min)
(V)
1000
0.5
1
1.5
2
Gas pressure (Pa)
FIGURE 14. Changes in etch rate and
FIGURE 15. Relation between etch rate and
substrate peak-to-peak
rf
gas pressure.
FIGURE
14. Changesvoltage
in etch with
rate and
FIGURE
15. Relation between etch rate and
coil
power.
substrate peak-to-peak voltage with rf
gas pressure.
coil power.
pressure. Roughly, the etch rate shows a linear increase. When the gas pressure becomes higher, the substrate peak-to-peak
voltage
duea tolinear
the higher
impedance
of the
results
in larger
etch
pressure.
Roughly, the
etch increases
rate shows
increase.
When the
gasplasma,
pressurewhich
becomes
higher,
the subrate. The
radial distributions
of the due
etch to
rate
shown
as a function
gas pressure
Fig. 16.
Whenetch
the
strate
peak-to-peak
voltage increases
theare
higher
impedance
of theofplasma,
which in
results
in larger
gas pressure
is lower
than 1.0 of
Pa,the
theetch
distributions
are nearly
same, oftherefore
it caninbeFig.16.
said that
there
rate.
The radial
distributions
rate are shown
as a the
function
gas pressure
When
theis
no effect
of gas
pressure.
when
it is higher
1.0the
Pa,same,
the etch
rates increase
gas
pressure
is lower
than However,
1.0 Pa, the
distributions
arethan
nearly
therefore
it can bewith
saidincreasing
that theregas
is
pressure,
and
the
distributions
of
the
etch
rate
are
not
uniform.
This
is
due
to
the
collisions
among
the plasma
no effect of gas pressure. However, when it is higher than 1.0 Pa, the etch rates increase with increasing
gas
particles and
in the
Figure
17etch
shows
theareetch
as a This
function
of to
thethesubstrate
bias power.
Gas
pressure,
thechamber.
distributions
of the
rate
not rate
uniform.
is due
collisionsrfamong
the plasma
pressure
is
0.5
Pa,
rf
coil
power
500
W,
and
gas
flow
rate
20
seem
in
this
case.
The
etch
rate
increases
linearly
particles in the chamber. Figure 17 shows the etch rate as a function of the substrate rf bias power. Gas
as the rf isbias
reason
that
energy
of ions
incident
wafer
the
pressure
0.5 power
Pa, rf increases.
coil power The
500 W,
and isgas
flowtherate
20 sccm
in this
case. on
Thethe
etch
rateincreases
increaseswith
linearly
rf bias
and hence
the probability
of issurface
reaction
larger with
the wafer
bias power.
Finally,
the
as
the rfpower
bias power
increases.
The reason
that the
energybecomes
of ions incident
on the
increases
with the
change
in the and
electron
with timeofissurface
measured
when becomes
etching islarger
goingwith
on. the
Thebias
electron
density
during
rf
bias power
hencedensity
the probability
reaction
power.
Finally,
the
etching in
is the
measured
using
the with
microwave
Figure
18 isshows
in the electron
change
electron
density
time isinterferometer.
measured when
etching
goingthe
on.change
The electron
density density
during
with
time
for
various
gas
pressure.
Etching
time
is
five
minutes,
rf
coil
power
500
W,
gas
flow
rate 10
seem,
etching is measured using the microwave interferometer. Figure 18 shows the change in the electron
density
substrate
rf
bias
power
50
W,
and
pressure
is
in
the
range
of
0.5
2.0
Pa.
As
is
shown
in
Fig.18,
the
electron
with time for various gas pressure. Etching time is five minutes, rf coil power 500 W, gas flow rate 10 sccm,
density decreases
with increasing
pressure
theofelectron
density
slightly
changes
with the
time.
Next,
substrate
rf bias power
50 W, and gas
pressure
is inand
thealso
range
0.5 – 2.0
Pa. As
is shown
in Fig.18,
electron
the
relation
between
the
electron
density
and
the
rf
coil
power
is
plotted
in
Fig.19.
It
is
clear
that
the
electron
density decreases with increasing gas pressure and also the electron density slightly changes with time. Next,
the relation between the electron density and the rf coil power is plotted in Fig.19. It is clear that the electron
267
Q
, , , , , , , , , _
: x: Z.OPa, V: O.SPa,
400
: O: "1 .SPa, O: O.SPa,
:A:1.0Pa
350
:
300
200
150
100
:
x
x
o
x
!^^__^
i
§
u
1400
I
:
=
_ __^__^^__^__^^
250 :
D
x
500W, lOsccm,
bias SOW
Etch rate (Å/min)
Etch rate(Å/min)
_ , , , , , , , , , j
4
l
x
. _____
i
__
Q:
°
s
V
i
:
i
: . . . . . . . . . ! . . . . . . . . . :
-50
50
0
100
Radial position r(mm)
&
Q
Q
FIGURE 16.
16. Radial
Radial distributions
distributions of
FIGURE
of etch
etch
rate
of
Si.
rat^ of Si.
xx^xxxx) ( X X X X )
o
0
O ooooo
ooooooooooooooo
Q
.T
XI O.SPa
500W,|10sccm, bias 5|OW,
Oi 1 .OPa
Ail.SPa
z=165lmm
I
Vf Z.OPa,
1
2
3
I
!
I
?
:
I
" io9
ooo~ !
10
600
;xxxx> cxxxx><xxxxxxxxx> cxxxx>k
-ooooc ) OOOOC)OOO° o O O °O( >oooo©
W
-3
-3
XA
Electron density (cm )
Electron density (cm )
w
s
500
oio10
10
ID
Q
400
FIGURE
FIGURE 17.
17. Etch rate
rate of
of Si
Si vs
vs rfrf bias
bias power.
power.
10
04
300
rf bias power (W)
a
0
200
V
"
0.5P;a,10scc n, bias|50W,
z=16 5 mm
]O:400 /v
A: 300^ V
:
108
c)
4
Time (min)
1
2
3
4
5
6
Time (min)
FIGURE 18.
18. Relation between electron density
FIGURE
and gas pressure.
and
FIGURE
FIGURE 19.
19. Relation
Relation between
between electron
electron density
density
and rf coil power.
density increases with rf coil power. The electron density changes with time. Figures 18
18 and
and 19
19 show
show that
that the
the
state of
of plasma is stable in our reactor.
state
CONCLUSIONS
CONCLUSIONS
Basic plasma properties in the ICP
ICP reactor were systematically examined for argon discharge plasma using
the
in chlorine
chlorine plasma
plasma were
were studied.
studied.
the Langmuir
Langmuir probe. Also, etching properties of polycrystalline silicon wafer in
We
of plasma parameters on the gas pressure,
pressure, the
the rf
rf power,
power, and
and the distance from
from
We clarified
clarified the
the dependence of
the
of the plasma potential, electron temperature,
temperature, and
and electron
electron density
density
the plasma source. The radial distributions of
were
pressure increases,
increases, the
the electron
electron density
density increases
increases but,
but, the
the plasma
plasma potential
potential
were very uniform.
uniform. As the gas pressure
and
increase with increasing
increasing
and electron temperature decrease. Both the electron density and plasma potential increase
the
we used the inlet disk
disk with the latticed
the rf
rf coil
coil power
power but
but the
the electron temperature does not change. If we
orifice,
uniform plasma.
plasma. The
The gas
gas flow
flow rate
rate has
has no
no
orifice, there is no need to rotate the orifice to realize a spatially uniform
effect
of 20
20 –- 100
100 sccm.
seem. As
As the
the distance
distance between
between the
the plasma
plasma source
source
effect on the plasma parameters in the range of
and
plasma potential
potential decrease,
decrease, whereas
whereas the
the electron
electron
and the
the measuring point increases, the electron density and plasma
temperature
temperature increases slightly.
268
As for chlorine plasma, the etch depth increases linearly with etching time. The electron density increases
with the source (rf coil) power. Although the etch rate increases with the biasing power, it decreases with the
rf coil power. The distribution of the etch rate of Si wafer was uniform in the ICP reactor.
REFERENCES
1. Flamm,D.L. and Herb,D.K., Plasma Etching: An Introduction, Eds.Manos,D.M. and Flamm,D.L., New York: Academic Press, 1989.
2. Hopwood,J., Guuarnieri,C.R., Whitehair,S.J., and Cuomo,J.J., Langmuir probe measurements of a radio frequency
induction plasma. J. Vac. Sci. Technol All(l), 152-156(1993).
3. Hopwood, J., Review of inductively coupled plasmas for plasma processing. Plasma Sources Sci. Technol., 1, 109116(1992).
4. Nanbu,K., Morimoto,T., and Suetani,M. Direct Simulation Monte Carlo analysis of flows and etch rate in an inductively coupled plasma reactor. IEEE Trans.Plasma Sci. 27, 1379-1388(1999).
5. Nanbu,K., Suetani,M., and Sasaki,H. Direct Simulation Monte Carlo(DSMC) modeling of silicon etching in radiofrequency chlorine discharge. Computational Fluid Dynamics Journal. 8, 257-265(1999).
6. Nanbu,K., Nakagome,T., and Kageyama,J. Detailed structure of the afterglow of radio-frequency chlorine discharge.
Jpn. J.Appl.Phys. 38, L951-L953 (1999).
7. Sasaki,H., Nanbu,K., and Takahashi,M. Basic plasma properties in a high-density ICP etching apparatus. Proc. 17th
Symp. Plasma Processing (Nagasaki, Japan), 125-128(2000).
269