Characteristics of C4 F8 plasmas with Ar, Ne, and He additives
for SiO2 etching in an inductively coupled plasma „ICP… reactor
Xi Li,a) Li Ling, Xuefeng Hua, and Gottlieb S. Oehrleinb)
Materials Science and Engineering and Institute for Research in Electronics and Applied Physics,
University of Maryland, College Park, Maryland 20742-2115
Yicheng Wang
National Institute of Standards and Technology, Gaithersburg, Maryland 20899-8113
H. M. Anderson
Department of Chemical and Nuclear Engineering, University of New Mexico,
Albuquerque, New Mexico 87131
共Received 19 June 2003; accepted 25 August 2003; published 14 October 2003兲
We have characterized the effect of adding Ar, Ne, and He noble gases to C4 F8 inductively coupled
plasmas for SiO2 etching. The systematic variation of their ionization potentials, metastable energy
levels, and mass of the dominant ion in C4 F8 /X% discharges 共X⫽He, Ne, or Ar兲 containing a high
percentage of the noble gas provides a means to evaluate the relative importance of gas phase and
surface processes in the etching of SiO2 . The total ion flux, ion composition, FC deposition and
etching rates, and composition of the surface reaction layer formed on Si or SiO2 in these discharges
show systematic differences as a function of process parameters because of the different electron
impact ionization thresholds for Ar, Ne, and He gases, and differences in the mass of the dominant
ion for discharges containing a large proportion of the noble gas. For given experimental conditions
共600 W and 20 mTorr for most of this work兲, Ar addition gives rise to the largest ion current density,
and He to the smallest ion current density. When the noble gases are added to C4 F8 , the Ar⫹ ion flux
contributes the largest percentage and He⫹ the smallest percentage of the total ion flux for the same
⫹
dilution of C4 F8 with the noble gases. Ion compositional analysis shows that CF⫹ , CF⫹
3 , and CF2
are the dominant ionic fluorocarbon 共FC兲 species 共in order of importance兲, and that they show
similar trends as a function of added Ar, Ne, and He percentage. The fluxes of the more highly
⫹
dissociated C⫹ , F⫹ , and SiF⫹
x /COFx ions are greater when Ne and He are added to C4 F8 than for
Ar addition. Ion energy distributions of all ions are displaced to lower energies for C4 F8 /Ar
discharges as compared with C4 F8 /Ne or C4 F8 /He plasmas. Infrared laser absorption spectroscopy
was used to determine absolute densities of neutral CF, CF2 , and COF2 radical species as a function
of gas composition. The densities of CF2 and COF2 radical species were enhanced when Ne and He
were added to C4 F8 in comparison to Ar addition. Fluorocarbon deposition rates on unbiased Si
substrates were measured and greater for C4 F8 discharges with Ne or He additives than when Ar was
added. Upon rf biasing of the substrates, the ion energies required for etching to take place are
lowest for C4 F8 /Ar discharges, followed by C4 F8 /He and C4 F8 /Ne discharges. A comparison of
x-ray photoelectron spectra of SiO2 surfaces exposed to C4 F8 /X% discharges at ⫺10 V self-bias
voltage shows pronounced differences in the chemical bond distribution of fluorocarbon films which
can be explained by differences in momentum transfer to the surface and the associated bond
breaking of deposited fluorocarbon layers on the surface. A comparison of SiO2 and Si etching rates
in C4 F8 /X% discharges shows that for C4 F8 /Ne discharges containing more than 70%Ne the
highest SiO2 /Si but lowest SiO2 /resist etching rate ratio is obtained. © 2003 American Vacuum
Society. 关DOI: 10.1116/1.1619420兴
I. INTRODUCTION
Fluorocarbon 共FC兲 plasmas produced using inductively
coupled plasma sources have been extensively studied with
regard to their use for plasma etching of SiO2 and low dielectric constant films.1– 6 Gas mixtures of fluorocarbon
gases like c-C4 F8 or C4 F6 with noble gases are frequently
used for SiO2 etching.7,8 We previously showed that Ar addition to C4 F8 has profound consequences on gas phase and
surface processes due to the strong increase of the plasma
a兲
Electronic mail: [email protected]
Electronic mail: [email protected]
b兲
1955
J. Vac. Sci. Technol. A 21„6…, NovÕDec 2003
density that results from Ar addition, leading to strongly improved SiO2 /Si etch rate ratios for C4 F8 /Ar discharges containing a high percentage of Ar 共above 60% Ar兲.9 The composition of the ion flux and the ion energy distribution of
C4 F8 /Ar discharges are also strongly modified by the addition of Ar to C4 F8 . 10
In this work we have determined the gas phase of
C4 F8 /X% 共X⫽He, Ne, or Ar兲 discharges and surface chemical characteristics of SiO2 and Si substrates in contact with
these discharges. The pronounced differences in electronimpact ionization thresholds between Ar 共15.76 eV兲,11,12 Ne
共21.54 eV兲,13 and He 共24.59 eV兲14,15 are expected to lead to
0734-2101Õ2003Õ21„6…Õ1955Õ9Õ$19.00
©2003 American Vacuum Society
1955
1956
Li et al.: Characteristics of C4F8 plasmas
1956
FIG. 2. Schematic diagram of the inductively coupled plasma reactor. The
location of the ion sampling system and the Langmuir probe are indicated.
FIG. 1. Comparison of electron impact energy thresholds of important processes for Ar, Ne, He, and C4 F8 .
differences in the electron energy distribution, the total ion
current, ion and radical composition of these discharges, and
also to differences in ion–surface interactions. A more detailed comparison of the electron impact ionization energy
thresholds for Ar,11,12 Ne,13 He,14,15 and C4 F8 共Ref. 16兲 gases
is given in Fig. 1, along with important thresholds for electron impact dissociation processes for C4 F8 . The figure
shows that of the noble gases, Ar has the lowest metastable
energy and ionization threshold 共15.76 eV兲 levels. The large
difference in ionization threshold energies between Ar and
Ne or He is expected to lead to important differences in the
electron energy distribution and ion energies and for gas
mixtures, e.g., C4 F8 /X% discharges containing a large percentage of the noble gases X, and differences in their chemical makeup. The dominant neutral dissociation products of
C4 F8 are C2 F4 and CF2 . The position of the metastable energy for He, Ne, and Ar relative to the dissociation and ionization energy thresholds of C4 F8 and C4 F8 derived fragments is also very important and shows some important
changes as we move from Ar to He 共see Fig. 1兲. According to
Font et al.,16 direct production of CF⫹
2 by electron impact of
C4 F8 discharge requires the highest ionization energy. Font
et al.16 also point out that the dominant ion due to electron
impact of C4 F8 is C2 F⫹
4 , and the next most important ion in
C4 F8 discharges is CF⫹
2 . Both dissociative ionization of
C4 F8 and direct ionization from C2 F4 are responsible for
this, whereas CF⫹
2 is primarily produced by direct ionization
of CF2 .
The differences in electron-impact ionization thresholds
and ion mass for the C4 F8 /X% discharges are in turn expected to produce different fluorocarbon deposition and FC,
SiO2 , and Si etching rates, providing the opportunity to produce insights as to the importance of certain gas phase and
plasma–surface interaction mechanisms in determining the
outcome of these plasma etching processes. We will show
that the total ion flux, ion composition, ion energy distribution, the FC film deposition rate, the Si and SiO2 etching
rates, and the composition of the surface reaction layer
J. Vac. Sci. Technol. A, Vol. 21, No. 6, NovÕDec 2003
formed on Si or SiO2 in these discharges show systematic
differences as a function of the amount of Ar, Ne, and He
added to C4 F8 共for a given set of process parameters兲 which
can be explained by the different electron impact ionization
thresholds and ion masses for Ar, Ne, and He gases.
II. EXPERIMENTAL SETUP AND PROCEDURES
A. Inductively coupled plasma source
The rf inductively coupled plasma 共ICP兲 reactor of planar
coil design used in this work is shown in Fig. 2. The locations of a Langmuir probe and a combined ion energy
analyzer-mass spectrometer are also shown. For a description
of experimental details on this setup the reader is referred to
our previous publications.9,17,18 In the present work, we studied C4 F8 /X% 共X⫽He, Ne, or Ar兲 discharges and varied systematically the percentage of He, Ne, or Ar between 0% and
100%. Most of the experimental data were obtained at 20
mTorr pressure, 600 W rf source power, and a total gas flow
of 40 sccm. These ‘‘standard’’ process conditions define a
point in parameter space that, in combination with rf biasing
of the substrates, is highly suitable for selective plasma etching of SiO2 over Si for the current inductively coupled
plasma reactor.9,17,18 The substrate electrode was separately
biased using 3.4 MHz rf power.
B. Diagnostic techniques
The ion current density was measured employing a Langmuir probe. The probe tip of the cylindrical Langmuir probe
was biased at ⫺100 V to prevent deposition of fluorocarbon
films on the metal surface and collect ions in the ion saturation regime. The tip of the Langmuir probe was located
about 7 mm above the center of the wafer. Additional ion
saturation current measurements were performed at the location of the mass spectrometer sampling orifice.
Compositional analysis of the ion flux and ion energy
analysis were performed, measured using a combined ion
energy analyzer-mass spectrometer10,19,20 that samples ions
through an orifice. The ion energy distributions measured in
this manner are essentially ion-flux energy distributions.21
1957
Li et al.: Characteristics of C4F8 plasmas
FIG. 3. Measured ion saturation current densities for 20 mTorr Ar and Ne
plasmas as a function of rf source power using a Langmuir probe. The data
were obtained either at 共a兲 the wafer center or 共b兲 the ion sampling position
共edge of plasma兲.
Infrared laser absorption spectroscopy was used to determine absolute densities of neutral CF, CF2 , and COF2 radical species as a function of the amount of Ar, Ne, and He
added to C4 F8 using procedures described previously.9,22
Measurements of FC deposition rates, FC, SiO2 , and Si
thin film etching rates, and characterization of the modified
surface layers were performed using real-time He–Ne ellipsometry. When the SiO2 etching rate was determined as a
function of self-bias voltage, which provides a measure of
the maximum ion energy, differences in the threshold energy
at which SiO2 etching took place were observed for the
C4 F8 /X% discharges. To provide insight as to the differences in surface composition near this threshold energy,
x-ray photoemission spectroscopy was performed with SiO2
surfaces etched at a constant self-bias voltage of ⫺10 V in
C4 F8 /X% 共X⫽He, Ne, or Ar兲 discharges.
III. RESULTS AND DISCUSSION
A. Ion saturation current densities
The ion saturation current densities of 100% Ne and Ar
discharges measured using a Langmuir probe are shown in
Fig. 3 as a function of rf power at 20 mTorr processing
pressure. The tip of the Langmuir probe was positioned either 7 mm above the wafer center or at the ion sampling
position 共see Fig. 2兲. The measured ion current density
showed a roughly linear increase for both noble gases. At
both sampling positions a higher ion current density was
obtained with Ar. For instance, for a 600 W Ar discharge the
ion current density was 42 mA/cm2 at the wafer center for
Ar, and 23 mA/cm2 for Ne. The higher ion current densities
of Ar plasmas relative to Ne discharges can be explained by
the lower electron impact ionization threshold of Ar as comJVST A - Vacuum, Surfaces, and Films
1957
FIG. 4. 共a兲 Measured ion saturation current densities at the wafer center for
C4 F8 /X% discharges 共X⫽He, Ne, or Ar兲 as a function of gas composition.
共b兲 The ratio of the ion saturation current at the edge of the plasma 共ion
sampling position兲 over the ion saturation current at the wafer center as a
function of gas composition.
pared to Ne. Figure 3 does not show data for 100% He discharges, since we found it very difficult to maintain stable
He discharges for these experimental conditions in our ICP
apparatus. On the other hand, when only 10% of C4 F8 was
added to He, stable discharges could be easily maintained
using these conditions. The ion current densities of pure Ne
and He discharges should be very similar because of the
similarity of the electron impact ionization thresholds, and
the fact that the ion currents measured with C4 F8 /Ne and
C4 F8 /He discharges containing a high percentage of the
noble gas were nearly identical.
Figure 4共a兲 shows the measured ion current densities
共ICD兲 of C4 F8 /Ar, C4 F8 /Ne, and C4 F8 /He plasmas as a
function of the percentage of the noble gas added at the
wafer center. The ICD increases strongly with noble gas percentage. For Ar/C4 F8 discharges the ICD is always the highest, and for Ne/C4 F8 and He/C4 F8 plasmas the ICD is nearly
identical.
Figure 3 showed that the ICD near the edge of the discharge 共location of the orifice of the ion sampling system兲
was lower than at the wafer center. This shows that the ion
density drops off radially. We also calculated the ratio of the
ICD at the ion sampling position 共edge of the discharge兲 to
the ICD at the center of the discharge. This is shown in Fig.
4共b兲, and provides information of the uniformity of the ion
current in the C4 F8 /X% discharges as a function of gas composition. The highest ratio 共85%兲 of the ICD at the ion sampling position over the ICD at the wafer center was measured
when 60% of the noble gases were added to C4 F8 /X%. We
find better ion current uniformity with Ne and He gas additives than with Ar addition to C4 F8 if a high percentage of
the noble gases was used.
1958
Li et al.: Characteristics of C4F8 plasmas
FIG. 5. Mass spectra of the ions produced in pure C4 F8 , 10% C4 F8 /90%Ar,
10% C4 F8 /90%Ne, and 10% C4 F8 /90%He discharges. Conditions: 20
mTorr pressure and 600 W rf source power.
B. Ion composition, ion flux, and ion energy
distribution measurements
The previous section showed that the different electron
impact ionization thresholds of He, Ne, and Ar produce different ion current densities in discharges fed with C4 F8 /noble
gas mixtures. In this section we describe the results of measurements aimed at determining differences in ion composition and ion energy distribution 共IED兲 function for C4 F8 /X%
共X⫽He, Ne, Ar兲 discharges. Data on ion composition, ion
fluxes, and ion energy distributions for C4 F8 /Ar discharges
have previously been reported.10
Figure 5 shows mass spectra of the ions produced in pure
c-C4 F8 共a兲, 90%Ar/C4 F8 共b兲, 90%Ne/C4 F8 共c兲, and
90%He/C4 F8 共d兲 discharges at 20 mTorr pressure and 600 W
rf power. For a pure C4 F8 discharge, the dominant fluorocarbon ion is CF⫹ followed by CF⫹
3 . For a 90%Ar/C4 F8 discharge, the dominant ion is Ar⫹ , and the dominant fluorocarbon ion is CF⫹
3 . Other ions producing significant ion
⫹
⫹
⫹
⫹
fluxes include CF⫹
2 , CF , C , C2 F5 , and C2 F4 . For the
90%Ne/C4 F8 and 90%He/C4 F8 discharges the dominant ions
are Ne⫹ and He⫹ , whereas the dominant fluorocarbon ions
and CF⫹ , respectively. As compared to
are CF⫹
3
90%Ar/C4 F8 discharges, a much higher density of C⫹ , F⫹ ,
⫹
⫹
Si⫹ /CO⫹ , SiF⫹ /COF⫹ , SiF⫹
2 /COF2 , and SiF3 ions is produced in 90%Ne/C4 F8 or 90%He/C4 F8 plasmas. This should
reflect differences in the electron energy distribution function, but may also be due to the higher energies of the Ne and
He metastables produced in these discharges. In addition, a
greater capacitive coupling component and erosion of the
quartz coupling window23 should be the cause of the rela⫹
⫹
tively high density of SiF⫹ /COF⫹ , SiF⫹
2 /COF2 , and SiF3
ions measured for 90%Ne/C4 F8 or 90%He/C4 F8 plasmas
J. Vac. Sci. Technol. A, Vol. 21, No. 6, NovÕDec 2003
1958
FIG. 6. Percentage contribution of individual ions to the total ion flux vs
percentage of Ar, Ne, or He added to C4 F8 . Other discharge conditions: 20
mTorr pressure and 600 W rf source power.
共FC deposition takes place on the unbiased Si or SiO2 substrates in contact with the plasma, and they cannot be the
source of these products兲.
From the ion spectra, the percentage contribution of each
ion to the total ion flux can be calculated and provides a
rough picture on the variation of the composition of the total
flux as a function of gas mixture. This is shown in Fig. 6, and
only includes the major ions for the C4 F8 /X% discharges as
a function the percentage Ar 关Fig. 6共a兲兴, Ne 关Fig. 6共b兲兴, and
He 关Fig. 6共c兲兴 added at the standard process conditions of 20
mTorr pressure and 600 W rf power. Figure 6 shows that
CF⫹ ions contribute 40% of the total ion flux for a C4 F8
discharge, and that this contribution monotonically decreases
with the proportion of noble gas added. CF⫹
3 ions contribute
a little bit more than 20% of the total ion flux for pure C4 F8 ,
and initially slightly increase in relative importance as the
Ar, Ne, or He are added, and subsequently decrease. The
percentage contribution of CF⫹
2 to the total ion flux is initially less than 10%, and nearly doubles when up to 60% Ne
or He is added to C4 F8 , but then decreases above 60% noble
gas additive. For Ar/C4 F8 the relative contribution of CF⫹
2
to the total ion flux is nearly constant. The relative importance of Ar⫹ to the total ion flux at 90% noble gas addition is greater than the contribution of Ne⫹ or He⫹ which
can be explained by the magnitude of the electron impact
ionization threshold relative to other possible processes
shown in Fig. 1.
From the measured ICD and the ion mass spectra, the ion
current density for each ion can be calculated. This information is shown in Fig. 7 for the major ionic species as a function of the percentage Ar, Ne, and He added to C4 F8 共20
mTorr pressure and 600 W rf source power兲. The Ar⫹ , Ne⫹ ,
He⫹ ion current densities and the total ion flux increase with
1959
Li et al.: Characteristics of C4F8 plasmas
1959
FIG. 7. Individual ion fluxes vs percentage of Ar, Ne, or He added to C4 F8 .
Other discharge conditions: 20 mTorr pressure and 600 W rf source power.
Ar, Ne, and He addition, respectively. The C⫹ , F⫹ , and
Si⫹ /CO⫹ ion fluxes increase more quickly with Ne and He
⫹
⫹
addition than with Ar addition. The C⫹
3 , CF , CF2 , and
⫹
CF3 ion fluxes increase when up to 60% of the noble gases
are added, and decrease for all three noble gases when more
than 60% are added. The increase of the total ion flux can be
explained by channeling the applied rf power into ionization,
rather than dissociation as the proportion of the atomic gas in
the gas mixture increases. The initial increase of the fluorocarbon related ions reflects the increase of the plasma density
as a low percentage of each of the noble gases is added to
C4 F8 . The decrease of the fluorocarbon ion flux above 60%
noble gas addition is explained by the limited C4 F8 gas supply for these conditions. The behavior of the fluxes of heavy
⫹
⫹
ions like C2 F⫹
4 , C3 F3 , and C3 F5 is also consistent with this
picture, i.e., the small increase as 20% noble gas is added,
and a rapid decrease for higher levels of noble gas addition
can be explained by the increased plasma density and the
limited availability of the C4 F8 gas. A more detailed analysis
of the change of individual ion fluxes with noble gas addition, and differences from one noble gas to the other, e.g., the
behavior of the CF⫹
2 ion flux in Ne/C4 F8 and He/C4 F8 discharges, which is different from that of Ar/C4 F8 discharges,
requires a more detailed comparison of the experimental data
with the energy level diagram shown in Fig. 1, but will not
be undertaken here.
⫹
Figure 8 shows CF⫹ , CF⫹
2 , and CF3 ion energy distributions measured with 90%Ar/C4 F8 , 90%Ne/C4 F8 , and
90%He/C4 F8 discharges. All measured IED are relative to
ground and show the characteristic saddle-shape. Both the
highest and the lowest ion energies are lower for Ar addition
than for Ne or He addition. The electron energy distribution
determines the plasma potential, which will determine the
highest ion energies which will be measured by the grounded
ion energy analyzer. The differences of the IEDs shown in
Fig. 8 are explained by the lower inelastic collision energy
thresholds of Ar relative to He and Ne, and the resulting
changes of the electron energy distributions.
JVST A - Vacuum, Surfaces, and Films
⫹
FIG. 8. Measured CF⫹ , CF⫹
2 , and CF3 ion energy distributions for 90% Ar,
Ne, or He added to C4 F8 . Other discharge conditions: 20 mTorr pressure
and 600 W rf source power.
C. Neutral fluorocarbon species measurements
The density of CF, CF2 , and COF2 radicals was obtained
using infrared laser absorption spectroscopy during fluorocarbon deposition 共no rf bias兲. Figure 9 shows the densities
FIG. 9. Partial pressures of CF, CF2 , and COF2 calculated from the IR
absorption spectra obtained with C4 F8 /Ar, C4 F8 /Ne, and C4 F8 /He discharges as a function of Ar, Ne, and He percentage added. Other conditions:
40 sccm total gas flow, 20 mTorr pressure, 600 W inductive power, no rf
bias 共fluorocarbon deposition兲.
1960
Li et al.: Characteristics of C4F8 plasmas
1960
FIG. 10. Fluorocarbon deposition rate vs percentage of Ar, Ne, or He added
to C4 F8 . Other discharge conditions: 20 mTorr pressure and 600 W rf source
power.
FIG. 11. Fluorocarbon etching rate vs percentage of Ar, Ne, or He added to
C4 F8 . Other discharge conditions: 20 mTorr pressure, 600 W rf source
power, and ⫺100 V self-bias voltage.
of CF, CF2 , and COF2 calculated from the absorption spectra
obtained with C4 F8 /Ar, C4 F8 /Ne, and C4 F8 /He plasmas
operated at the standard conditions 共40 sccm gas flow, 20 m
Torr pressure, and 600 W inductive power兲 as a function of
percentage noble gas added. The densities of CF, CF2 , and
COF2 decrease with increasing Ar, Ne, and He content. The
CF2 partial pressure is about 50 times greater than that of CF
for a C4 F8 plasma. The relative importance of CF increases
as the noble gas percentage is increased. Overall, because of
the scatter in the data it is difficult to make any definitive
statements on a dependence of the FC radical densities on
the type of noble gas added, although it appears that the CF
species density is higher for C4 F8 /Ar discharges than for
C4 F8 /Ne or C4 F8 /He discharges. The density of COF2 species is higher for Ne or He addition to C4 F8 if the percentage
of the noble gas is greater than 50%. The COF2 is produced
by the etching of the reactor coupling window due to the
capacitive coupling effect. The COF2 radical density data are
consistent with the ion analysis data described above.
by applying an rf bias. The etching rates increase gradually
with Ar, Ne, and He addition up to 60%–70% of the gas
mixture. The FC etch rates decreased when the Ar, Ne, and
He content was increased above 70%. The initial increase of
the fluorocarbon etching rate is explained by the boost of the
ion current as a result of noble gas addition to C4 F8 . The
etch rate decrease at high noble gas proportions is due to a
lack of C4 F8 and the associated changes in ion and neutral
composition.
D. Fluorocarbon deposition and etching
Figure 10 shows the fluorocarbon deposition rate as a
function of the percentage of Ar, Ne, and He added to C4 F8
共standard process conditions, no rf bias applied兲. The fluorocarbon deposition rate is always higher with the Ne or He
additives than with the Ar additive. The observed overall
trend towards lower fluorocarbon deposition rates for all gas
mixtures as the proportion of the noble gases is increased is
expected because of the reduced supply of C4 F8 molecules to
the reactor. The consistent differences in fluorocarbon
growth rates between He, Ne, and Ar gas additives appear
mechanistically interesting and warrant further studies.
Figure 11 shows the corresponding fluorocarbon etching
rates at ⫺100 V self-bias voltage. For these experiments initially a fluorocarbon film was deposited without rf bias. The
same film was then etched back using the same gas mixture
J. Vac. Sci. Technol. A, Vol. 21, No. 6, NovÕDec 2003
E. Effects of adding Ar, Ne, and He to C4 F8 on SiO2
and Si etching
1. SiO 2 etching
Figure 12 shows the measured SiO2 etching rate as a
function of self-bias voltage for C4 F8 /80%Ar, C4 F8 /80%Ne,
and C4 F8 /80%He gas mixtures for discharges produced using 600 W inductive power, 20 mTorr pressure, and a total
gas flow of 40 sccm. At low self-bias voltage, net fluorocarbon deposition is observed. As the self-bias voltage is increased, the FC deposition rate is reduced, and for self-bias
voltages more negative than ⫺25 V SiO2 etching takes place
for all gas mixtures. Figure 12 shows that for our conditions
the etching threshold is lowest with Ar 共⫺2 V兲 and is highest
with Ne 共⫺12 V兲. The etching threshold is very close for Ne
and He gas additive. The differences in the etching energy
thresholds between Ar and Ne or He should at least in part be
due to changes in the mass of the dominant ion, i.e., Ar⫹ ,
Ne⫹ , or He⫹ . The efficiency of the energy transfer to the
SiO2 layer is higher for Ar⫹ than for Ne⫹ or He⫹ . In addition, the stoichiometry of the deposited FC layers needs to be
considered in the etching and the contribution of FC ions,
since for low ion energies steady-state etching of the oxide
occurs through a fluorocarbon layer that is present on the
SiO2 surface.24 The F/C ratio of deposited fluorocarbon films
is slightly higher for Ar 共1.37兲 as compared with He 共1.35兲
and Ne 共1.28兲. This may explain why the etching threshold
1961
Li et al.: Characteristics of C4F8 plasmas
1961
FIG. 12. SiO2 etching rate vs self-bias voltage for C4 F8 /80%Ar,
C4 F8 /80%Ne, and C4 F8 /80%He discharges. Other conditions: 20 mTorr
pressure and 600 W rf source power.
with C4 F8 /He is slightly smaller than for C4 F8 /Ne. It has
also been observed that the SiO2 etching yield increases with
the average ion F/C ratio.25 The average ion F/C ratio calculated from Fig. 7 is 0.56 共Ar兲, 0.60 共Ne兲, and 0.85 共He兲,
respectively. It is clear that the higher SiO2 etching rates
measured for C4 F8 /Ar relative to C4 F8 /Ne or C4 F8 /He are
primarily caused by the higher ion current densities seen for
C4 F8 /Ar 共see Fig. 4兲. This is also consistent with the etching
yield data reported below.
2. Variation of the SiO2 , Si, and resist etching rate
with percentage Ar, Ne, and He added to C 4F 8
The dependence of the SiO2 , Si, and resist etching rates
at a fixed self-bias voltage of ⫺100 V on gas composition in
C4 F8 /Ar, C4 F8 /Ne, and C4 F8 /He gas mixtures at 20 mTorr
pressure and inductive powers of 600 W is shown in Fig. 13.
This etching condition is at a point where the SiO2 etching
rate increases only slowly with ion energy 共see Fig. 12兲. The
SiO2 etching rate increases up to 60% noble gas addition for
all additives 关see Fig. 13共a兲兴. Above that the SiO2 etch rate
decreases, more strongly for He than for the other noble
gases. Interestingly, the maximum SiO2 etching rate at 60%
noble gas addition is the same for all noble gases. The SiO2
etch rates above the etch rate maximum are basically the
same for Ne and Ar, even though the total ion current differs.
This shows that in this regime the etching rate is limited by
the neutral reactant supply.
Silicon etching rates are shown in Fig. 13共b兲. The Si etching rate is greater with Ar addition than for Ne addition. At
70% and 80% noble gas addition, the Si etching rate is greatest for the He containing discharges. The latter result is possibly due to the increase of the oxygen level in the discharge
due to the increased capacitive coupling and erosion of
quartz coupling window in the He containing discharge 共see
Figs. 5 and 9兲.
Resist etching rates are shown in Fig. 13共c兲. The resist
etching rates increase as the proportion of the noble gas is
JVST A - Vacuum, Surfaces, and Films
FIG. 13. Comparison of SiO2 , Si, and resist etching rates as a function of
percentage of Ar, Ne, and He added to C4 F8 discharges. 20 mTorr pressure,
600 W rf power, and ⫺100 V self-bias voltage.
raised, and initially does not depend on the noble gas added.
At 60% and 70% noble gas addition, higher resist etching
rates are measured for Ne and He gas addition than for Ar
addition. For Ne/C4 F8 discharges containing 70% and 80%
Ne, resist etching rates near 300 nm/min are measured.
From the etching rates, the SiO2 /Si and SiO2 /resist etch
rate ratios were obtained as a function of the percentage of
Ar, Ne, and He added to C4 F8 and are shown in Fig. 14. The
highest SiO2 /Si etch rate ratio 共⬃8兲 was obtained with
80% Ne/C4 F8 discharges. On the other hand, the SiO2 /resist
etch rate ratio using 80% Ne/C4 F8 discharges was only
slightly greater than 1, whereas for 80% Ar/C4 F8 it was
about 4.
SiO2 and Si etch yields (SiO2 or Si removed per incident
ion兲 were calculated as a function of the percentage of Ar,
Ne, and He added to C4 F8 and are shown in Fig. 15. For
SiO2 the etch yield increases as the percentage of noble gas
in C4 F8 is increased, reaches a maximum at 60%, and subsequently decreases rapidly 关Fig. 15共a兲兴. Also for Si the etch
yield is highest in the range of 20%– 60% noble gas addition,
and the values for He, Ne, and Ar are similar 关Fig. 15共b兲兴. At
80% noble gas addition, both the SiO2 and Si etch yields are
higher for C4 F8 /Ne and C4 F8 /He discharges than for
C4 F8 /Ar. In part this should be explained by the higher average ion F/C ratio measured for the He and Ne containing
discharges.
For fluorocarbon discharges, the SiO2 /Si etching selectivity correlates with the thickness of the steady-state fluorocarbon layer thickness on Si.5 Figure 16 displays the ellipsometrically measured thickness of the transparent FC/SiFx
layer formed on Si during plasma etching. The addition of
Ne to C4 F8 leads to the thickest steady-state fluorocarbon
1962
Li et al.: Characteristics of C4F8 plasmas
1962
FIG. 16. Comparison of measured FC steady-state films thickness on Si for
C4 F8 /Ar, C4 F8 /Ne, and C4 F8 /He discharges vs noble gas percentage
added to C4 F8 . 20 mTorr pressure, 600 W rf power, and ⫺100 V self-bias
voltage.
FIG. 14. Comparison of SiO2 /Si and SiO2 /resist etch rate ratios as a function of percentage of Ar, Ne, and He added to C4 F8 discharges. 20 mTorr
pressure, 600 W rf power, and ⫺100 V self-bias voltage.
layers on Si, especially at higher percentages of Ne additive.
This is consistent with the high SiO2 /Si etch rate ratio seen
for C4 F8 /Ne discharges. The addition of He to C4 F8 leads to
the thinnest steady-state fluorocarbon layers on Si, consistent
with the lower SiO2 /Si etch rate ratio seen for C4 F8 /He
discharges.
F. SiO2 surface analysis
To obtain information on changes in surface chemistry as
a result of adding Ar, Ne, and He gases to C4 F8 discharges,
FIG. 15. Etch yields of SiO2 and Si as a function of percentage of Ar, Ne,
and He added to C4 F8 discharges. 20 mTorr pressure, 600 W rf power, and
⫺100 V self-bias voltage.
J. Vac. Sci. Technol. A, Vol. 21, No. 6, NovÕDec 2003
XPS analysis was performed with Si 共no rf bias, FC film
deposition兲 and SiO2 when an ⫺10 V self-bias was applied.
Without rf bias, 300 nm thick FC films were deposited in
either pure C4 F8 , C4 F8 /80%Ar, C4 F8 /80%Ne, or
C4 F8 /80%He plasmas at 20 mTorr pressure and 600 W rf
power 共Fig. 17兲. The CF2 component shows the highest intensity for the FC film produced in the C4 F8 discharge. The
spectra of the deposited fluorocarbon film produced in
C4 F8 /80%Ar, C4 F8 /80%Ne, and C4 F8 /80%He plasmas are
very similar. The integrated C (1s) intensity is greater for the
C4 F8 discharge produced film than for FC films formed in
C4 F8 /80%Ar, C4 F8 /80%Ne, or C4 F8 /80%He discharges.
To examine surface chemistry differences near the SiO2
etching threshold for C4 F8 /80%Ar, C4 F8 /80%Ne, or
C4 F8 /80%He plasmas, SiO2 films were exposed to these discharges at ⫺10 V self-bias voltage for 1 min at the ‘‘stan-
FIG. 17. X-ray induced C 1s photoelectron emission spectra obtained with
passively deposited 共no rf bias voltage兲 fluorocarbon layers 共300 nm thick兲.
The FC films were deposited using 600 W inductive power, 20 mTorr discharges fed with either C4 F8 , C4 F8 /80%Ar, C4 F8 /80%Ne, or C4 F8 /80%He.
The electrons were collected at 90° relative to the sample surface.
1963
Li et al.: Characteristics of C4F8 plasmas
1963
SiO2 etching was lowest for Ar addition, followed by He and
Ne. The SiO2 /Si and SiO2 /resist etch rate ratios also depend
on the type of noble gas added to C4 F8 . Most of the observations can be explained by differences in the electron structure of the noble gases 共inelastic electron collision threshold
energies兲 and by the noble gas ion mass.
ACKNOWLEDGMENT
The authors gratefully acknowledge support of this work
by the Department of Energy under Contract No. DEFG0200ER54608.
C. B. Zarowin and C. W. Williams, Appl. Phys. Lett. 11, 47 共1967兲.
J. P. Goldsborough, E. B. Hodges, and W. E. Bell, Appl. Phys. Lett. 8, 137
共1966兲.
3
J. Keller, J. C. Forster, and M. S. Barnes, J. Vac. Sci. Technol. A 11, 2487
共1993兲.
4
J. Hopwood, C. R. Guarnieri, S. J. Whitehair, and J. J. Cuomo, J. Vac. Sci.
Technol. A 11, 152 共1993兲.
5
T. E. F. M. Standaert, M. Schaepens, N. R. Rueger, P. G. M. Sebel, J. M.
Cook, and G. S. Oehrlein, J. Vac. Sci. Technol. A 16, 239 共1998兲.
6
H. Hayashi, T. Kurihara, and M. Sekine, Jpn. J. Appl. Phys., Part 1 35,
2488 共1996兲.
7
M. Matsui, T. Tatsumi, and M. Sekine, J. Vac. Sci. Technol. A 19, 1282
共2001兲.
8
M. Matsui, T. Tatsumi, and M. Sekine, J. Vac. Sci. Technol. A 19, 2089
共2001兲.
9
X. Li, L. Ling, X. Hua, M. Fukasawa, G. S. Oehrlein, M. Barela, and H.
Anderson, J. Vac. Sci. Technol. A 21, 284 共2003兲.
10
A. N. Goyette, Y. Wang, M. Misakian, and J. K. Olthoff, J. Vac. Sci.
Technol. A 18, 2785 共2000兲.
11
Handbook of Chemistry and Physics, 53rd ed., edited by R. C. Weast
共CRC, Cleveland, OH, 1972兲.
12
L. Minnhagen, J. Opt. Soc. Am. 63, 1185 共1973兲.
13
K. Harth, J. Ganz, M. Raab, K. T. Lu, J. Gaiger, and H. Hotop, J. Phys. B
18, L825 共1985兲.
14
K. S. E. Eikema, W. Ubachs, W. Vassen, and W. Hogervorst, Phys. Rev. A
55, 1866 共1997兲.
15
S. D. Bergeson, A. Balakrishnan, K. G. H. Baldwin, T. B. Lucatorto, J. P.
Marangos, T. J. McIlrath, T. R. O’Brian, S. L. Rolston, C. J. Sansonetti, J.
Wen, N. Westbrook, C. H. Cheng, and E. E. Eyler, Phys. Rev. Lett. 80,
3475 共1998兲.
16
G. I. Font, W. L. Morgan, and G. Mennenga, J. Appl. Phys. 91, 3530
共2002兲.
17
T. E. F. M. Standaert, M. Schaepens, N. R. Rueger, P. G. M. Sebel, J. M.
Cook, and G. S. Oehrlein, J. Vac. Sci. Technol. A 16, 239 共1998兲.
18
X. Li, M. Schaepkens, R. E. Ellefson, L. C. Frees, N. Mueller, N. Korner,
and G. S. Oehrlein, J. Vac. Sci. Technol. A 17, 2438 共1999兲.
19
Y. Wang and J. K. Olthoff, J. Appl. Phys. 85, 6358 共1999兲.
20
J. K. Olthoff and Y. Wang, J. Vac. Sci. Technol. A 17, 1552 共1999兲.
21
M. V. V. S. Rao, R. J. Van Brunt, and J. K. Olthoff, Phys. Rev. E 54, 5641
共1996兲.
22
M. Schaepkens, I. Martini, E. A. Sanjuan, X. Li, G. S. Oehrlein, W. L.
Perry, and H. M. Anderson, J. Vac. Sci. Technol. A 19, 2946 共2001兲.
23
M. Schaepkens, N. R. Rueger, J. J. Beulens, X. Li, T. E. F. M. Standaert,
P. J. Matsuo, and G. S. Oehrlein, J. Vac. Sci. Technol. A 17, 3272 共1999兲.
24
N. R. Rueger, J. J. Beulens, M. Schaepkens, M. F. Doemling, J. M. Mirza,
T. E. F. M. Standaert, and G. S. Oehrlein, J. Vac. Sci. Technol. A 15, 1881
共1997兲.
25
X. Li, M. Schaepkens, R. E. Ellefson, L. C. Frees, N. Mueller, N. Korner,
and G. S. Oehrlein, J. Vac. Sci. Technol. A 17, 2438 共1999兲.
1
2
FIG. 18. Carbon 1s electron emission spectra obtained with an SiO2 film
exposed to C4 F8 /80%Ar, C4 F8 /80%Ne, or C4 F8 /80%He discharges at ⫺10
V self-bias voltage for 1 min. Other conditions: 600 W inductive power, 20
mTorr pressure, 600 W, 40 sccm total gas flow. The electrons were collected
at normal emission relative to the surface.
dard’’ conditions 共see Fig. 18兲. In this case SiO2 etching was
observed for C4 F8 /80%Ar, the SiO2 etching rate was negligible for C4 F8 /80%He, and FC film deposition was seen for
C4 F8 /80%Ne 共see Fig. 12兲. To prepare these surfaces, initially 150 nm of SiO2 was etched using a self-bias voltage of
⫺100 V. Subsequently, the self-bias voltage was reduced to
⫺10 V and the SiO2 surface exposed for 1 min. High resolution C 1s electron emission spectra obtained after the
above plasma processes are shown in Fig. 18. With the Ar
additive, there is only a thin FC layer. The C–CFx is of the
highest intensity. With the Ne additive, a thicker FC layer is
observed, and the CF2 component shows the highest intensity. The F/C ratios on the SiO2 layer with Ar, Ne, and He
additives are 0.84, 1.31, and 1.29, respectively.
IV. CONCLUSIONS
Important differences in the electrical and chemical characteristics of inductively coupled C4 F8 /X% discharges 共X
⫽He, Ne, or Ar兲 have been observed. We find that the ion
current density is highest for C4 F8 /Ar discharges, and is explained by the lower electron impact ionization thresholds
relative to Ne and He. Differences in the inelastic electron
collision threshold energies also influence ion and neutral
composition, and ion energetics of these discharges. Surface
processes and processing rates are affected as well, in particular for discharges containing a very high proportion of
the noble gas. For instance, the measured FC deposition and
etching rates are higher with Ne and He additives than with
Ar. For SiO2 , Si, and resist etching interesting differences
were observed. For instance, the threshold energy to induce
JVST A - Vacuum, Surfaces, and Films