Cavity ring down detection of SiH 3 on the broadband ? 2 a 1′ ← X? 2 a 1 transition in a remote Ar-H 2 -SiH 4 plasma

Article abstract of DOI:10.1016/S0009-2614(00)00795-8

Here we report on the use of the cavity ring down (CRD) technique for the detection of the silyl radical SiH₃ on the broadband ? ²A₁′ ← X? ²A₁ transition around 215 nm. SiH₃ has been detected in a remote Ar-H₂-SiH₄ plasma during hydrogenated amorphous silicon (a-Si:H) thin film growth. The measurements demonstrate the capability of CRD to measure small broadband absorptions in the deep UV in the hostile environment of a deposition plasma. The SiH₃ absorption shows an expected dependence on the SiH₄ precursor flow and correlates well with the a-Si:H growth rate. The observed absorptions correspond with SiH₃ densities in the range 2-13 × 10¹⁸ m^-3, which is at least two orders of magnitude above the estimated SiH₃ detection limit.

Full text of DOI:10.1016/S0009-2614(00)00795-8

2
5 August 2000
Chemical Physics Letters 326 Ž2000. 400–406
www.elsevier.nlrlocatercplett
Cavity ring down detection of SiH on the broadband
3
2
X
2
˜
˜
X A transition in a remote Ar–H –SiH plasma
A A
1
1
2
4
1
M.G.H. Boogaarts , P.J. B o¨ cker, W.M.M. Kessels, D.C. Schram,
M.C.M. van de Sanden)
Centre for Plasma Physics and Radiation Technology (CPS), Department of Applied Physics, EindhoÕen UniÕersity of Technology,
P.O. box 513, 5600 MB EindhoÕen, The Netherlands
Received 26 April 2000; in final form 26 June 2000
Abstract
Here we report on the use of the cavity ring down ŽCRD. technique for the detection of the silyl radical SiH on the
3
2
X
2
broadband A˜ A
X˜ A transition around 215 nm. SiH has been detected in a remote Ar–H –SiH plasma during
1
1 3 2 4
hydrogenated amorphous silicon Ža-Si:H. thin film growth. The measurements demonstrate the capability of CRD to measure
small broadband absorptions in the deep UV in the hostile environment of a deposition plasma. The SiH absorption shows
3
an expected dependence on the SiH4 precursor flow and correlates well with the a-Si:H growth rate. The observed
1
8
y3
absorptions correspond with SiH densities in the range 2–13 = 10
m
, which is at least two orders of magnitude above
3
the estimated SiH detection limit. q 2000 Elsevier Science B.V. All rights reserved.
3
1
. Introduction
radical SiH3 in the plasma could elucidate the cru-
cial plasma chemistry and the assumed dominant
In many low-temperature plasma applications, low
role of SiH in the growth mechanism of a-Si:H thin
3
density Ži.e. short lived. species play a key role, but
films, but not many direct observations of this species
under real plasma deposition conditions exist. Most
detailed studies on SiH3 in a plasma reactor are
performed with mass spectrometry where state of the
art techniques are used to enable neutral radical
detection w1,2x. The sensitivity of mass spectrometry
often their role is not established in detail. The study
of these species in plasmas calls for a sensitive
technique that is applicable in the hostile plasma
environment. The technique should be non-intrusive
and preferably also in situ and remote at the same
time. In the plasma enhanced chemical vapour depo-
sition of hydrogenated amorphous silicon Ža-Si:H.
16
y3
is very high ŽSiH detection limit - 10
m
. but
3
the technique has no spatial resolution and calibra-
tion is laborious. Furthermore, when in situ it is not
non-intrusive and vice versa, unless an extraction
orifice is used at the position of the substrate.
Opticalrspectroscopic techniques generally are
non-intrusive, remote, and in situ, but very few
for instance, an in situ measurement of the silyl
)
Corresponding author. Fax: q31-Ž0.40-2456442; e-mail:
M.C.M.v.d.Sanden@phys.tue.nl
1
E-mail: M.G.H.Boogaarts@tue.nl
0
009-2614r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S0009-2614Ž00.00795-8

M.G.H. Boogaarts et al.rChemical Physics Letters 326 (2000) 400–406
401
studies on SiH3 in plasmas are reported that apply
such techniques. Most of the experimental data on
the electronic spectroscopy of SiH Žand SiD . cur-
absorption, which is characterized by the decay time
t of the light intensity in the cavity, is related to the
absolute absorption per pass ns L, via the expression
1rctsŽ1yRqns L.rd, where c is the speed of
light. A measurement of the decay time, which is
commonly referred to as the ‘ring-down time’, there-
fore directly provides the absolute absorption in the
cavity, independent of the peak intensity of the light
pulse.
3
3
rently available results from resonance enhanced
multi-photon ionization ŽREMPI. studies in a flow
reactor w3–5x. However, like mass spectrometry, the
REMPI technique is very sensitive, but generally not
in situ. The first electronic transition of SiH is the
3
2
X
2
˜
˜
X A transition, which is optically acces-
1
A A
1
sible in the deep UV around 215 nm w5x. The upper
In this Letter we report on the use of the cavity
ring down technique for the detection of the silyl
˜
electronic A state is predicted to be predissociative
2
X
2
w6x and thus short-lived and non-radiative. Therefore,
unless higher electronic states are probed w3–5x, only
direct absorption techniques can be applied that gen-
erally lack sensitivity. Nevertheless, the first experi-
radical on the broadband A˜ A
X˜ A transition
1
1
around 215 nm in a remote Ar–H –SiH plasma
2
4
that is used for the fast deposition of a-Si:H thin
films. The measurement of broadband absorptions
with CRD is discussed, in which special attention is
given to the implication of the hostile plasma envi-
ronment for the broadband measurement procedure.
Also, the problem of identifying the broadband sig-
nal carrier is discussed.
˜
˜
X transition of SiH
3
mental observation of the A
resulted from direct UV absorption measurements in
a flash photolysis cell w7x, and meanwhile this tech-
nique has even been successfully applied for the
detection of SiH3 in a radio frequency ŽRF. gener-
ated H –SiH plasma under high silyl production
2
4
conditions w8x.
2
. Experimental
Opticalrspectroscopic techniques for the detec-
tion of silyl also have been applied in the infrared
region. Infrared diode-laser spectroscopy has been
applied in a silane glow discharge to study the
spectroscopy and the structure of SiH3 in the elec-
tronic ground-state w9x. A further study employed
infrared diode-laser absorption spectroscopy ŽIRLAS.
A scheme of the experimental set-up is depicted
in Fig. 1. The method and the set-up for fast a-Si:H
for the measurement of SiH in an RF silane plasma
3
w10x, yielding among other things the rotational tem-
perature of the silyl radical. However, even with this
sophisticated Žand complex. technique, in situ optical
detection limits for SiH can not be pushed below
3
1
7
y3
1
0
m
.
The availability of the ‘cavity ring down ŽCRD.’
technique in the last decade w11x, enables an impor-
tant step forward in the in situ detection of key
radicals in low-temperature plasma processing w12–
1
6x. The highly sensitive CRD technique Žfirst
demonstrated by O’Keefe and Deacon in 1988 w17x.,
is based on the measurement of the rate of absorp-
tion of a light pulse confined in a closed optical
cavity. The cavity usually consists of two highly
reflective mirrors with a reflectivity R and a separa-
tion d around the sample medium with density n,
absorption cross section s , and absorption path-
length L. With monochromatic light, the rate of
Fig. 1. Scheme of the experimental set-up for the CRD measure-
ments in a remote Ar–H2 –SiH4 plasma. The cascaded arc plasma
source is operated on a 45 A DC current, a 55 sccs Ar flow, and a
1
0 sccs H2 flow. The downstream injected SiH4 flow is varied.
The distance between plasma source and substrate is 35 cm. The
ring-down cavity samples the plasma beam 11 cm from the
substrate. The cavity length is 1.02 m.

4
02
M.G.H. Boogaarts et al.rChemical Physics Letters 326 (2000) 400–406
deposition using an expanding thermal plasma gener-
ated by a cascaded arc have already been described
in detail elsewhere w18–20x. In the experiments re-
is frequency-doubled in a BBO-crystal, resulting in a
light pulse around 215 nm with an estimated band-
width of 0.2 cm . The light leaking out from the
second mirror Žon the top., is detected with a PMT
ŽHamamatsu R928.. An interference bandpass filter
y1
ported here the plasma is generated by a 45 A DC
discharge in a premixed argon-hydrogen flow Ž55
standard cubic centimeter per second Žsccs. Ar, 10
with a central transmission wavelength close to the
light pulse wavelength is used behind the second
mirror to reduce the exposure of the PMT to continu-
ous plasma and background light. A similar filter is
used in front of the first mirror to suppress the
residual fundamental 430 nm dye-laser light and to
assure a low light intensity level in the ring-down
cavity. The PMT signal is sampled by a digital 500
Msrs, 100 MHz oscilloscope ŽTektronix TDS340A.,
averaged on board over several laser pulses, and the
resulting transient is processed by a PC, where the
characteristic ‘ring-down time’ t is determined from
a weighed least-squares fitting of the logarithm of
the data.
sccs H .. The plasma expands from a nozzle into a
2
low-pressure Ž0.20 mbar. vessel. The chemical activ-
ity in the plasma beam emanating from the nozzle
comes mainly from atomic hydrogen. The silane
precursor is injected into the plasma beam via a ring
that is mounted 5 cm downstream from the nozzle.
At this point, silane is dissociated and the silyl
radical is mainly produced by hydrogen abstraction
w21x:
HqSiH4 SiH qH
3
2
kf5P10y17 m s at 1000 K
3
y1
.
Ž
.
Ž .
1
The distance between plasma source and substrate is
5 cm. The chosen experimental conditions corre-
3
3. Results and discussion
spond to the optimum conditions for which device
quality a-Si:H is deposited at rates up to 10 nmrs
w22x.
In order to identify the absorbing species, we first
performed a spectral scan. Fig. 2 shows the absorp-
2
X
2
The ring-down cavity samples the plasma beam
tion of SiH on the A˜ A
X˜ A transition as it
3
1
1
1
1 cm from the substrate. The stable optical cavity is
set up by two plano-concave highly reflective dielec-
tric mirrors ŽLaseroptik GmbH., that are a distance d
s 1.02 m apart. The mirrors have a 1 inch diameter,
a radius of curvature rsy1 m, and a specified
minimum transmittivity Ts0.2% around 218 nm
Žimplying a maximum reflectivity Rs0.998.. The
mirrors are mounted on flexible bellows and directly
seal the plasma reactor. To protect the mirrors from
the hostile plasma environment, a small Ar flow can
be maintained from the mirrors into the vessel
through 5 mm diameter diaphragms that are posi-
tioned in front of the mirrors on the axis of the
cavity w14x.
The light pulse, that is coupled into the cavity via
the lower mirror, is generated by a pulsed tunable
dye-laser ŽSirah Laser- und Plasmatechnik GmbH,
Precisionscan-D. that is pumped with a 10 Hz
Nd:YAG-laser ŽSpectra-Physicsr Quanta-Ray DCR-
2
X
2
Fig. 2. The absorption of SiH3 on the A˜ A
X˜ A transition
1
1
around 220 nm as measured with cavity ring down in an expand-
ing Ar–H –SiH plasma Ž10 sccs SiH4 flow, '.. The absorp-
tion data are the net result of CRD measurements with and
without SiH4 flow at each wavelength value. The data are
compared to those reported by Lightfoot et al. Ž(, scaled verti-
cally. w7x.
2
4
1
1. at 355 nm. The dye-laser has a bandwidth of 1.7
y1
pm Žf0.1 cm
at 430 nm. and is operated on a
coumarin dye around 430 nm. The dye-laser output

M.G.H. Boogaarts et al.rChemical Physics Letters 326 (2000) 400–406
403
is measured in the expanding Ar–H –SiH4 plasma
effect on the mirror reflectivity. In nearly all mea-
surements we found the ring-down time t0 of the
‘empty’ cavity to decrease on a longer time scale
Žhalf an hour., and when the mirrors were kept under
high vacuum conditions for a prolonged time Ža day.
2
with a 10 sccs silane flow. For clarity, we plotted a
reduced data set. The complete data set covers the
full range of our optical set-up Žf13 nm, mainly
limited by the cavity mirrors. with a resolution of at
least 0.1 nm without revealing any structure. This is
consistent with the observations by other groups who
also found a very broad absorption feature w7,8x. For
the reflectivity was found to improve again, some-
times up to 0.985. This effect has also been observed
by other groups that applied CRD detection in a
plasma w15x. As a direct consequence, a procedure
for determining the baseline cavity loss in a broad-
band absorption spectrum, as proposed by Romanini
et al. w23x, where the ‘zero absorption baseline’ is
measured afterwards by repeating the same scan
without absorbing medium, is not possible in our
case. The ability to separately determine the baseline
in a broadband Žor congested. CRD spectrum that is
recorded in a plasma does not only depend on the
mechanical stiffness of the set-up or on the proper
mode matching of the laser beam to the cavity, but
mainly on the durability of the mirrors in a plasma
environment. We therefore think that extreme care
has to be taken when this procedure is followed in
the case of application of CRD in a reactive environ-
ment, such as a plasma.
comparison, Fig. 2 also shows the data reported by
Lightfoot et al. w7x. Measuring the SiH3 A˜
X˜
spectrum for different plasma conditions, had no
influence on the shape of the spectrum. Since these
different conditions correspond to different kinetic
temperatures in the plasma, we conclude that, over
the temperature range encountered upon changing
these conditions, the shape of the spectrum is mainly
determined by the predissociation lifetime broaden-
ing and not by the temperature.
Notice that in CRD absolute absorptions are al-
ways determined by measuring the difference in
‘
cavity loss’ 1rct of the cavity with Ž1rctsŽ1yR
qns L.rd. and without Ž1rct sŽ1yR.rd. the
0
absorbing species. In the case of a narrowband ab-
sorption, this is automatically accounted for by
recording a spectral scan, in which the cavity loss of
the ‘empty’ cavity determines the baseline in the
spectrum. Due to the broadband nature of the SiH3
Unfortunately, the absorbing species cannot be
unambiguously identified from a broadband spec-
trum. The absence of structure in the observed spec-
trum at least excludes the interfering absorption of
species that exhibit a line spectrum. It does not rule
out, however, the broadband absorption by other
species that are produced in the Ar–H –SiH
˜
A
˜
X transition, the net absolute absorption had
to be determined for each data point separately. This
was accomplished by measuring the cavity loss at
each wavelength setting both with and without silane
precursor flow. After switching the SiH4 flow,
enough time Žseveral seconds. was allowed for the
2
4
plasma. More in particular, it does not rule out the
cavity loss due to light scattering from large clusters
that are commonly produced in silane plasmas. The
feasibility of CRD detection of dust has already been
demonstrated in an industrial-scale RF plasma reac-
tor w15x. In that case, powder formation was identi-
fied from the time behaviour of the measured cavity
loss. In our case, the absorption by dust is excluded
by the observation that the ring-down time t de-
creases and increases instantly when the silane flow
is switched on and off, respectively. The topic of
dust formation and its possible influence on the
observed absorption will be more extensively dis-
cussed in a subsequent paper.
plasma reactor to stabilize.
The observed cavity loss without silane flow Žthe
‘
baseline absorption’. between 215 and 218 nm cor-
responds to an effective mirror reflectivity Rf0.96.
On a short time scale Ža few minutes., no significant
change in this baseline absorption was observed
when switching off the plasma or switching off both
the plasma and all gas flows. This reflectivity value
is dramatically less than the above mentioned maxi-
mum reflectivity of 0.998. The difference can partly
be explained by the fact that the minimum transmis-
sion value overestimates the maximum reflectivity
when additional refractive mirror losses are not taken
into account. More importantly, despite the protect-
ing measures we took, the plasma seems to have an
Additional evidence that the observed absorption
is due to SiH3 is obtained from a measurement of
the dependence of the absorption on the silane pre-

4
04
M.G.H. Boogaarts et al.rChemical Physics Letters 326 (2000) 400–406
substrate. In addition, the TIMS measurements are
local, while the CRD technique yields line-integrated
results. Considering the complex and laborious cali-
bration of the TIMS technique and the uncertainty in
the radial SiH3 distribution in the CRD measure-
ments, the agreement is quite satisfactory. As men-
tioned before, the effective absorption cross section
is independent of the temperature in the range asso-
ciated with the range of operating conditions in our
plasma reactor, since in this range the shape of the
spectrum is mainly determined by predissociation
lifetime broadening.
Fig. 3. SiH3 absorption and corresponding SiH3 density as a
function of SiH4 precursor gas flow. The measurements are
performed at 215 nm.
The observed high SiH3 densities are strongly
related to the high a-Si:H growth rate in our plasma
reactor. In the end, SiH is a short lived radical that
3
has a much lower density in more conventional
plasma reactors like those used in RF PECVD w1,10x,
and its detection therefore generally requires a high
sensitivity. Based on the measurements presented in
cursor flow ŽFig. 3.. The absorption Žmeasured at
2
15 nm. increases monotonically with increasing
SiH4 flow, as expected for the SiH density that is
3
produced via reaction Ž1.. The reproducibility of the
measurement can be estimated from the spread in
absorption values for the SiH4 flow settings for
which we performed the measurement twice. The
shape of the silane flow dependence convincingly
this Letter we estimate our current SiH detection
3
1
7
y3
limit to be f10
m
, which is mainly determined
by the effective reflectivity of the cavity mirrors. As
mentioned above, we feel that we can improve on
the reflectivity, which would immediately benefit our
agrees with independent SiH measurements that are
current SiH detection limit.
3
3
performed with threshold ionization mass spectrome-
At this point we could try to correlate the ob-
try ŽTIMS. at the position of the substrate w2x. This
served SiH densities in the remote plasma, via the
3
shape and the conclusions that can be drawn from it
regarding the SiH3 production mechanism are dis-
cussed in Ref. w2x.
corresponding plasma conditions, with the a-Si:H
growth rates that have been established in earlier
experiments w22x. In Fig. 4 the silicon growth flux,
In order to know whether the above mentioned
TIMS measurements and the here presented CRD
results also agree quantitatively, we need to correlate
which is the film growth rate corrected for the Si
atomic density of the film, is plotted against the
the measured absorptions with SiH densities. Since
3
radial density profiles of the silyl radical are not
available yet, we assume a homogeneous distribution
of SiH over a 30 cm absorption path length, which
3
is based on the structure of the flow pattern as it
appears from for instance Phoenics-CVD model cal-
culations. Using in addition the absorption cross
y21
2
section ss2.4P10
m at 215 nm w7x, results in
SiH densities Žgiven by the right scale in Fig. 3.
that lie in the range 2–13 P 10
3
1
8
y3
m
. These values
are in agreement with the TIMS measurements within
a factor of two. When comparing the results of the
two techniques, it should be noticed that TIMS has
been performed at the position of the substrate, while
the CRD results are obtained at 11 cm from the
Fig. 4. Correlation between silicon growth flux Žfrom previous
studies w22x. and SiH3 density Žthis study..

M.G.H. Boogaarts et al.rChemical Physics Letters 326 (2000) 400–406
405
SiH density that is measured with CRD at 215 nm.
Acknowledgements
3
The error in the density is estimated from the repro-
ducibility of the CRD measurements. The figure
shows a clear correlation, which is a minimum re-
quirement when silyl would be a dominant contribu-
tor to a-Si:H film growth.
Besides the correlation, a more quantitative argu-
ment can be obtained to prove the dominance of the
silyl contribution to a-Si:H growth. The silicon
We acknowledge H.M.M. de Jong, A.B.M.
H u¨ sken, M.J.F. van de Sande, and J.F.C. Jansen for
their skillful technical assistance. This work is part
of the research programme of the ‘Netherlands
Foundation for Fundamental Research on Matter’
ŽFOM: 98PR1781 and 99PSI03. and it is also finan-
cially supported by the ‘Netherlands Organization
growth flux G can be directly related to the silyl
g
Ž
.
for Scientific Research’ NWO .
density nSiH 3 in front of the surface via the Chantry
expression w1x:
nÕth
s
Gg s
P
,
2
Ž .
References
4
1ybr2
that uses the thermal flux nÕ r4 of particles arriving
th
w1x P. Kae-Nune, J. Perrin, J. Guillon, J. Jolly, Plasma Sources
Sci. Technol. 4 Ž1995. 250.
w2x W.M.M. Kessels, M.C.M. van de Sanden, D.C. Schram, to
appear in J. Vac. Sci. Technol. 18 Ž2000. issue 5 ŽSep.rOct..
w3x R.D. Johnson III, J.W. Hudgens, Chem. Phys. Lett. 141
Ž1987. 163.
at a surface, where s is the sticking probability and
b the total loss probability of SiH3 at the a-Si:H
surface. From the slope in Fig. 4 we obtain the ratio
G rn
f61 mrs. Then, using a thermal velocity
g
SiH 3
3
at 1500 K of Õ f10 mrs and a loss probability
th
w4x R.D. Johnson III, B.P. Tsai, J.W. Hudgens, J. Chem. Phys.
bs0.3 that has been determined in a previous study
w24x, we obtain sf0.2, which is a plausible value
for the sticking probability w24,25x. This shows that
9
1 Ž1989. 3340.
5
w x
R.D. Johnson III, J.W. Hudgens, J. Chem. Phys. 94 1991
5331.
Ž
.
w6x G. Olbrich, Chem. Phys. 101 Ž1986. 381.
the present silyl density can fully account for the
observed silicon growth flux for the conditions that
are used in this study.
w7x P.D. Lightfoot, R. Becerra, A.A. Jemi-Alade, R. Lesclaux,
Chem. Phys. Lett. 180 Ž1991. 441.
w8x H. Toyoda, M. Goto, M. Kitagawa, T. Hirao, H. Sugai, Jpn.
J. Appl. Phys. 34 Ž1995. L448.
w9x C. Yamada, E. Hirota, Phys. Rev. Lett. 56 Ž1986. 923.
w10x N. Itabashi, N. Nishiwaki, M. Magane, S. Naito, T. Goto, A.
4
. Conclusions
Matsuda, C. Yamada, E. Hirota, Jpn. J. Appl. Phys. 29
Ž1990. L505.
w11x A full account of the CRD literature will not be given here.
We demonstrated the applicability of the cavity
ring down technique for the measurement of small
broadband absorptions at deep UV wavelengths. The
CRD technique is found to be well suited for the
detection of small amounts of the short lived silyl
For further reading about CRD one is referred to: Cavity-
Ringdown Spectroscopy: An Ultratrace-Absorption Measure-
ment Technique, K.W. Busch, M.A. Busch ŽEds.., American
Chemical Society, Washington, DC, 1999.
w12x A. Campargue, D. Romanini, N. Sadeghi, J. Phys. D: Appl.
radical SiH in a low-temperature processing plasma.
Ž
.
3
Phys. 31 1998 1168.
To the best of our knowledge, this is the first re-
ported use of CRD for the detection of the silyl
radical. We detected SiH in a remote Ar–H –SiH
w13x E. Quandt, I. Kraemer, H.F. D o¨ bele, Europhys. Lett. 45
Ž1998. 32.
w14x R. Engeln, K.G.Y. Letourneur, M.G.H. Boogaarts, M.C.M.
van de Sanden, D.C. Schram, Chem. Phys. Lett. 310 Ž1999.
3
2
4
2
X
2
˜
˜
X A transition to the
1
plasma on the A A
1
4
05.
˜
predissociative A state around 215 nm, during a-Si:H
thin film growth. The silicon growth flux of the
a-Si:H film correlates with the observed absorption
due to the silyl radical in the plasma, and we showed
w15x F. Grangeon, C. Monard, J.-L. Dorier, A.A. Howling, Ch.
Hollenstein, D. Romanini, N. Sadeghi, Plasma Sources Sci.
Technol. 8 Ž1999. 448.
w16x J.P. Booth, G. Cunge, L. Biennier, D. Romanini, A.
Kachanov, Chem. Phys. Lett. 317 Ž2000. 631.
that the corresponding high SiH densities in the
3
w17x A. O’Keefe, D.A.G. Deacon, Rev. Sci. Instrum. 59
1988
Ž .
1
8
y3
range 2–13 P 10
m
can fully explain the high
2
544.
a-Si:H growth rates in our plasma reactor.
w18x M.C.M. van de Sanden, R.J. Severens, W.M.M. Kessels,

4
06
M.G.H. Boogaarts et al.rChemical Physics Letters 326 (2000) 400–406
R.F.G. Meulenbroeks, D.C. Schram, J. Appl. Phys. 84 Ž1998.
426.
w22x W.M.M. Kessels, A.H.M. Smets, B.A. Korevaar, G.J.
2
Adriaenssens, M.C.M. van de Sanden, D.C. Schram, Mat.
Res. Soc. Symp. Proc. 557 Ž1999. 25.
w19x M.C.M. van de Sanden, R.J. Severens, W.M.M. Kessels,
R.F.G. Meulenbroeks, D.C. Schram, J. Appl. Phys. 85 Ž1999.
w23x D. Romanini, A.A. Kachanov, F. Stoeckel, Chem. Phys. Lett.
1
243.
270 Ž1997. 546.
w20x M.C.M. van de Sanden, W.M.M. Kessels, R.J. Severens,
D.C. Schram, Plasma Phys. Control. Fusion 41 Ž1999. A365.
w21x A. Goumri, W.-J. Yuan, L. Ding, Y. Shi, P. Marshall, Chem.
Phys. 177 Ž1993. 233.
w24x W.M.M. Kessels, M.C.M. van de Sanden, R.J. Severens,
D.C. Schram, J. Appl. Phys. 87 Ž2000. 3313.
w25x A. Matsuda, K. Nomoto, Y. Takeuchi, A. Suzuki, A. Yuuki,
J. Perrin, Surf. Sci. 227 Ž1990. 50.