Topic number 10
Actinometry for understanding PECVD of thin films
from O2/HMDSO plasmas
R. Šmíd1
, L. Zajíčková1
, V. Buršíková1
1
Faculty of Science, Masaryk University, Kotlářská 2, 637 00 Brno, Czech Republic
We applied actinometry for the calculation of dissociation degree in oxygen CCP discharges
used for the deposition from O2/HMDSO plasmas. Dissociation degree exhibited a slight
increase with increasing r.f. power and maximum of 20 % for 5 Pa of oxygen. This relatively
high value was not enough for deposition of SiO2-like films because the HMDSO percentage
in the feed was too high at this low oxygen partial pressure. Rapid decrease of dissociation
degree to 2­4 % for higher oxygen flow rates, i. e.higher pressures resulted in still insufficient
oxidation of film precursors.
1. Introduction
Organosilicon-oxygen mixtures can be used for
plasma enhanced chemical vapor deposition
(PECVD) of inorganic silicon oxide films (SiOx)
as reported many times for tetraethoxysiloxane
(TEOS) as well as for hexamethyldisiloxane
(HMDSO) compounds. Decreasing high percentage
of oxygen in the organosilicon-oxygen feed the
inorganic character of the deposits can be changed
to more organic as shown for example by Valée
et al. [1]. The films prepared without any oxidizing
gas, i. e., from pure monomer or with addition
of an inert gas like argon, are then often referred
as plasma polymers due to their high organic content
[2].
2. Experimental
In this paper we discuss the role of oxygen radicals
and ion bombardment on the deposition of thin
films under various conditions in 13.56 MHz low
pressure glow discharges from oxygen/HMDSO
feed with different concentrations of HMDSO. The
experiments were carried out in a capacitively coupled
(CCP) stainless steel parallel plate reactor.
The bottom electrode, 420 mm in diameter, was
r.f. powered and due to an asymmetric coupling
and different mobility of electrons and ions the
negative self-bias was superimposed over the r.f.
voltage. The bottom electrode was used as a substrate
holder in order to allow an ion bombardment
of the growing films. Gases were fed into
the chamber through an upper grounded showerhead
electrode. The distance between the electrodes
was 55 mm. R.f. power was in the range
50­450 W. For the deposition, the HMDSO reactant
with fixed flow rates of 2 or 4 sccm (QHMDSO)
were used. The oxygen flow rate varied from 0 to
98 sccm (QO2). Thus a HMDSO percentage in the
feed ( = QHMDSO/(QHMDSO + QO2) from 4.8 to
100 % was achieved. Corresponding total pressure
in the reaction chamber prior to the deposition
was in the range 1.2­53 Pa.
Discharges in oxygen/HMDSO and pure oxygen
were studied by optical emission spectroscopy
using a Jobin Yvon TRIAX 550 spectrometer.
Actinometry method described below was applied
to assess dissociation degree, i. e., oxygen radical
concentration. Electron energy distribution function
in the CCP reactor with similar electrode
configuration and power density was determined
by Langmuir probe measurements using r.f. compensated
probe from Scientific Systems Ltd.
3. Results and Discussion
3.1 Film deposition
The optical and mechanical properties as well as
the atomic composition and chemical structure of
the deposited films were investigated in detail in
the previous paper [3]. According to the film mechanical
properties we can divide the depositions
roughly into three groups, (i) deposition of films in
HMDSO-rich discharges ( = 16.7­100 %), (ii) deposition
in intermediate conditions and (iii) deposition
in oxygen-rich discharges ( = 4.8­6.7 %).
Films from the first group exhibited compressive
intrinsic stress, especially at the high power, i. e.,
high d.c. self-bias and besides SiOx, SiH, SiOH
and OH contained high amount of organic groups
such as SiCHx and CHx. The percentage of carbon
and hydrogen in the films were in the range
10­17 at. % and 42­65 at. %, respectively. Refractive
index was higher than tabulated for SiO2
and the films exhibited absorption in the ultraviolet/visible
range. Polymer character of the films
was confirmed by low hardness, low elastic modulus
and viscoelasto-plastic behavior.
Films deposited in the intermediate conditions
had hardness below 5 GPa and were stress-free, especially
for low r.f. power. Intrinsic stress in the
films changed its character to the tensile one for
the films deposited in oxygen-rich discharges. The
film properties, such as refractive index, extinction
coefficient and hardness, approached the properties
of SiO2 but the films still contained some
hydrocarbon impurities. The atomic percentage
of carbon and hydrogen for our lowest HMDSO
concentration  = 4.8 % were in the range 58
and 20­40 at. %, respectively. This is a different
result compared to the deposition from inductively
coupled O2/HMDSO or O2/TEOS plasmas
where pure SiO2-like films were obtained for
 = 10 % [4]. Furthermore, we compared our
oxygen-rich deposits ( = 2 %, QHMDSO = 2 sccm,
r.f. power of 300 W) with the films from the inductively
coupled plasma (ICP) and found their optical
and mechanical properties quite different [5].
The ICP film deposited with the same  and r.f.
power had its refractive index higher that tabulated
SiO2 whereas our CCP films exhibited values
lower or close to SiO2. Comparison of the mechanical
properties revealed even more interesting
differences. The ICP films exhibited compressive
stress and hardness higher than SiO2. The hardness
of the CCP films depended on the r.f. power,
i. e., self-bias on the substrate electrode. A tensile
stress was found in the high hardness CCP films.
It is obvious that oxygen radicals play a key
role in the deposition of SiO2-like films from oxygen/organosilicon
plasmas. They are involved in
dissociative reaction with monomer, oxidation of
organic plasma species and surface oxidation reactions
with adsorbed precursors. Raupp et al. [6]
and Stout et al. [7] suggested that the oxygen
atom flux was a limiting step for their deposition
of SiO2 films from the O2/TEOS CCP discharges.
Therefore, we assumed that the major difference
between the CCP and ICP deposition processes
discussed above was a concentration of oxygen
radicals. In order to prove this hypothesis and
also to understand the changes of atomic composition
for different  we measure the concentration
of oxygen radicals [O] in our experimental conditions
by actinometry.
3.2 Actinometry
Within the actinometric method intensities of two
emission lines, one of the monitored species and
the other of the reference ones, are divided in order
to calculate the specie density. According to
the suggestion in Refs. [8, 9] we used the oxygen
emission line at 844.6 nm (3p3P  3s3S transition)
and argon at 750.4 nm (2p1  1s2 transition)
as the actinometer for determination of [O].
Ratio between argon and oxygen concentrations
in the gas feed was kept as low as 0.01-0.03. It is
assumed that the corresponding excited states are
populated from the ground states by an electron
impact. Additionally, the dissociative excitation
of oxygen has to play a minor role and the calculations
are simplified neglecting a non-radiative
de-excitation by quenching.
The emission intensity IX
ij of a transition
X
i - X
j + hij normalized to a plasma volume
is
IX
ij = [X
i ] hij Aij (1)
where Aij is Einstein transition probability of
spontaneous emission, hij is the energy of each
light quantum and [X
i ] is the concentration of excited
species X
i . The concentration [X
i ] can be
determined from an excitation/de-excitation balance
equation. Neglecting the quenching we can
write for the concentration of oxygen and argon
excited species
[Ar
p1
] = ne
[Ar]kAr
exc
j
AAr
ij
(2)
[O
3P] = ne
([O]kO
exc + [O2]kO
diss)
j
AO
ij
where [Ar], [O] and [O2] are the concentrations
of argon, atomic and molecular oxygen in plasma,
respectively, kAr
exc, kO
exc and kO
diss are the rate constants
for direct excitations to Ar(p1), O(3P)
and dissociative excitation of O(3P), respectively,
j AAr,O
ij represent the sums of all radiative deexcitation
processes for Ar(p1) or O(3P) and ne is
the density of electrons. Thus, the electron density
is excluded from considerations calculating
the ratio of oxygen-to-argon emission line intensities
IO/IAr. This is the principle of actinometric
method. Finally, we can write
IO
IAr
=
750AO
844 j AAr
ij
844AAr
750 j AO
ij
kO
exc[O]
kAr
exc[Ar]
+
kO
diss[O2]
kAr
exc[Ar]
(3)
where the intensity ratio has to be corrected for
an optical response of the measurement system.
The rate coefficients kX
 (X = Ar, O and  = exc,
diss) can be expressed as
kX
 =
2
m

0

E X
 (E) f(E)dE (4)
where X
 (E) are collision cross sections for excitation
or dissociative excitation of argon and oxygen
as function of electron energy E, m is the electron
mass and f(E) is a spherically symmetric electron
energy distribution function (EEDF). Assuming
Maxwell distribution the EEDF can be written as
f(E) =
4
m
2
m
m
2kBT
3/2 
Ee-E/kBT
(5)
where kB is Boltzmann's constant and T is the
electron temperature.
0 10 20 30 40 50 60
0.06
0.08
0.10
0.12
0.5
1.0
1.5
actinometricsignal[a.u.]
oxygen flow rate [sccm]
100 W
200 W
300 W
100 W
200 W
300 W
Figure 1. Actinometric signal given by Eq. (6) in
pure oxygen (solid symbols) and oxygen/HMDSO
(open symbols) plasmas.
In order to calculated the concentration of oxygen
atoms in the plasma, i. e., the dissociation degree
d = [O]/[O2]feed we have to couple Eq. (3)
with the mass conservation law between oxygen
molecules in the gas feed and oxygen atoms and
molecules in the plasma. Similar as in Ref. [10] we
obtain following relation between so called actinometric
signal on the left side and d and rate
constants on the right side of the equation:
1
C
[Ar]
[O2] feed
IO
IAr
=
kO
diss
kAr
exc
+d 2
kO
exc
kAr
exc
-
kO
diss
kAr
exc
(6)
where the constant C, according to Eq. (3), contains
the corrections on different wavelengths of
the emission lines used and on the Einstein transition
probabilities. The actinometric signal measured
in pure oxygen and oxygen/HMDSO discharges
in dependence on oxygen flow rate and
r.f. power is shown in Fig. 1. We can see that
its value is much lower in O2/HMDSO than pure
O2 which can be caused by high depletion of oxygen
radicals in oxidation reactions but also deexcitation
of excited atomic oxygen by quenching.
In pure oxygen and low flow rates, i. e.low pressures,
the signal exhibited a certain increase with
the r.f. power. This trend can be expected from
a simple view of increased energy delivered into
the discharge. However, this effect almost diminishes
for higher flow rates. Granier et al. [10] reported
almost linear increase of the actinometric
signal with the power for the ICP discharge but
the influence of the power in the CCP mode was
not quite distinguishing. It is interesting to notice
that the actinometric signal rapidly decreases
with increased flow rate from the maximum value
at 20 sccm. We think that these results are necessary
to clarify by more experimental points.
0 10 20 30 40 50
0
2
4
6
8
10
12
14
16
18
20
22
dissociationdegree[%]
oxygen flow rate [sccm]
100 W
200 W
300 W
400 W
Figure 2. Dependence of dissociation degree in
pure O2 on the oxygen flow rate and r.f. power
given in the figure legend. The pressure changed
from 1 to 27 Pa.
Combining Eqs. (4), (5) and (6) we can see
that the electron temperature must be known for
the calculation of dissociation degree. Since it was
not possible to measure the EEDF directly in the
"deposition" reactor we carried out these measurements
in another, 13.56 MHz capacitively coupled,
plasma reactor with a similar geometry ("diagnostics"
reactor). In this reactor, the distance between
two electrodes, 100 mm in diameter, was
also 55 mm. Therefore, comparing the electrode
areas we concluded that the r.f. power range 50450
W in the "deposition" reactor matches the
range 5­25 W used in the "diagnostics" one. Assuming
the Maxwell distribution for simplicity we
found that the electron temperature is approximately
3 eV and did not change within the discussed
r.f. power range. Taking into account this
temperature and the cross section data published
in Refs. [11, 12, 13] we calculated the necessary
rate constants using Eqs. (4) and (5). Resulting
values for kO
exc/kAr
exc and kO
diss/kAr
exc, appearing in
Eq. (6), are 3.8 and 0.1, respectively. The dissociation
degree is then plotted in Fig. 2. Its
values for low flow rates seems to be quite high
but we would like to notice that the low partial
pressure of oxygen results in the low concentration
of oxygen atoms (see Fig. 3). Thus, polymerlike
films were deposited under these conditions
(  17 %). Rapid decrease of d for the high
oxygen flow rates, on the other hand, resulted in
the deposition of the films with remaining carbon
impurities.
0 50 100 150 200 250 300 350 400 450 500
0
5
10
15
20
25
atomicconcentration[10
19
m
-3
]
rf power [W]
Q(O2
)=5 sccm
Q(O2
)=10sccm
Q(O2
)=20sccm
Q(O2
)=45sccm
Figure 3. Concentration of oxygen radicals as a
function of r.f. power for different oxygen flow
rates, i. e., oxygen partial pressure.
4. Conclusion
Actinometric method is very useful and simple
tool for understanding the deposition process in
O2/HMDSO discharges where oxygen atoms play
a key role in oxidation reactions. We see a high
depletion of oxygen emission line when HMDSO
is added to the gas feed. The dissociation degree
d in pure O2 exhibited a slight increase with
increasing r.f. power and more complicated functional
dependence on the oxygen flow rate QO2,
i. e., pressure. The d reached about 20 % for
5 Pa (QO2 = 20 sccm) and then rapidly decreased
to 2­4 % for 27 Pa (QO2 = 45 sccm). The low
d for high pressures is then probably the reason
why pure SiO2-like films were not deposited at the
high dilution of HMDSO in oxygen.
Acknowledgments
The present work was supported by the Ministry
of Education of the Czech Republic, under
the projects FRVŠ 2752/2005, 1K05025 and
MSM0021622411.
References
[1] C. Valée, A. Goullet, A. Granier, A. van der
Lee, J. Durand, Marli`ere, J. Non-Cryst.
Solids 272 (2000) 163.
[2] A. M. Wrobel, M. R. Wertheimer, In Plasma
Deposition, Treatment and Etching of Polymers,
ed. R. ďAgostino, Academic Press,
New York 1990, pp. 163­268.
[3] L. Zajíčková, V. Buršíková, V. Peřina,
Macková, D. Subedi, J. Janča, S. Smirnov,
Surf. Coat. Technol. 142-144 (2001) 449.
[4] K. Aumaille, C. Vallée, A. Granier, A. Goullet,
F. Gaboriau, G. Turban, Thin Solid Films
359 (2000) 188.
[5] L. Zajíčková, V. Buršíková, G. Borvon, A.
Goullet, A. Granier, V. Peřina , A. Macková,
Proceedings of ISPC XVI (2003), p.451
[6] G. B. Raupp, T.S. Cale, H.P.W. Hey, J. Vac.
Sci. Technol. B 10 (1992) 37.
[7] P. J. Stout, M. J. Kushner, J. Vac. Sci. Technol.
A 11 (1993) 2562.
[8] D. Pagnon, J. Amorim, J. Nahorny, M.
Touzeau and M. Vialle, J. Phys. D.: Appl.
Phys. 75 (1995) 1856.
[9] M. A. Lieberman, A. J. Lichtenberg, Principles
of Plasma Discharges and Materials Processing,
John Wiley & Sons, New York 1994,
p. 259.
[10] A. Granier, D. Berhomieu, J. P. Deville,
M. Ducarroir, J. Durand, A. Goullet, P. Leprince,
C. Marli`ere, P. Raynaud, Y. Ségui,
G. Turban, V. Vervloet, Thin Solid Films 270
(1995) 109.
[11] V. Puech, L. Torchin, J. Phys. D: Appl. Phys.
19 (1986) 2309.
[12] R. R. Laher, F. R. Gilmore, J. Phys. Chem.
Rev. Data 19 (1990) 277.
[13] M. B. Schulman, F. A. Sharpton, S. Shung,
C. C. Lin, L. W. Anderson, Phys. Rev. A 32
(1985) 2100.