Reactivity of atoms adsorbed on catalytic surfaces under plasma
exposure
D. Marinov1, D. Lopatik2, M.Huebner2, O. Guaitella1, C. Corbella3, A. von Keudell3, T. de los Arcos3, J. Röpcke2,
A. Rousseau1
1
LPP, Ecole Polytechnique, UPMC, Université Paris Sud-11, CNRS, Palaiseau, France
2
INP, Felix-Hausdorff-Str. 2, 17489 Greifswald, Germany
3
Research Department Plasmas with Complex Interactions, Ruhr-University Bochum, Universitätsstr. 150, 44780
Bochum, Germany
Abstract. We study chemisorption and reactivity of nitrogen atoms on quartz.
Strongly bound N atoms grafted by low pressure nitrogen plasma are detected
and quantified due to their ability to recombine with O atoms producing gas
phase NO. Dynamic of N chemisorption is studied by varying the duration of N2
plasma pretreatment in the range 6∙10-3÷5∙104 s. The role of chemisorbed
nitrogen atoms for surface catalyzed recombination of N was studied with the use
of isotopic 14N - 15N exchange in 15N2 plasma. XPS analysis of quartz samples
exposed to N2 plasma showed that SiOxNy layer containing up to 29 % of N
atoms is formed on initial SiO2.
Keywords: plasma-catalyst interaction, adsorbed atoms, air pollution control
1. Introduction
Interaction of molecular plasmas with dielectric
surfaces is of great importance for a number of
applications. Heterogeneous processes (such as
surface atomic recombination) control the density of
atoms and serve as a source of new molecules
production in low pressure plasmas [1,2]. In case of
atmospheric pressure plasma applications for air
pollution control stabilization of oxygen atoms on
catalytic surfaces increase their effective life-time
for pollutant oxidation [3,4].
It is believed that chemisorbed atoms play a key role
in surface catalyzed reactions in plasmas; however,
their properties are still not well understood. The
aim of this work is to study adsorption and reactivity
of chemisorbed N atoms on quartz surface. We use
the approach developed in [5,6,7] which consists in
treating the surface with N2 plasma followed by
checking the reactivity of the atoms strongly bonded
on the surface after the pretreatment.
2. Experiment
2.1 Experimental set-up
Surface pretreatment was performed with a
radiofrequency discharge generated in a 2cm inner
diameter, 60cm full length discharge tube. The 40cm
removable central part of the reactor was made of
quartz or pyrex while two terminations 10 cm each
were pyrex-made (see figure 1).
RF gen.
Matching unit
Laser
Detector
removable section
pump
gas inlet
Figure 1. Experimental set-up.
Two identical copper ring electrodes were located
outside the tube and were driven symmetrically with
a SAIREM 13.56 MHz generator through a pushpull matching circuit. The discharge power was
deduced using the subtractive method based on the
RF voltage, incident and reflected power
measurements. The visible length of the discharge
column was 50cm and it occupied entirely the
removable part of the reactor.
Time evolution of NO concentration was measured
in-situ using tunable IR diode laser spectrometer.
The absorption line at 1900.52 cm-1 (NO(X23/2)
R(6.5), v=0→v=1 transition) was used. Absolute NO
concentration was deduced from the frequency
integrated absorption coefficient using the transition
parameters from HITRAN database. Concentrations
down to 1012 cm-3 could be detected in a single pulse
without accumulation with 5 ms time resolution.
2.2 Experimental procedure
It is worth noting that with the present approach
one can detect only stable adatoms that survive on
the surface during several minutes after plasma
exposure. Furthermore, only those atoms that react
producing NO during relatively short O2 probe
discharge pulse are accessible for diagnostics. For
example N atoms diffusing into the bulk material are
not detectable with this method. In the further
discussion when speaking about adsorbed N we will
always imply stable nitrogen atoms that are reactive
under O2 plasma exposure.
3. Results and discussion
3.1 Adsorption sites for N on quartz surface
Figure 2 shows the evolution of NO concentration in
a probe O2 discharge after standard N2 plasma
pretreatment of the reactor walls. Two curves
correspond to quartz and pyrex made central
discharge tube sections. Characteristic time of nitric
oxide production is 100 milliseconds for both
materials. The estimation of the initial density of
Nads can be done based on the total number of
produced NO molecules at the end of the probe
pulse and the surface-to-volume ratio of the reactor,
giving 3·1013 cm-2 and 3.5·1013 cm-2 for quartz and
pyrex respectively.
The fact that both pyrex and quartz yield very close
amount of produced NO proves that adsorption sites
on quartz are not related to impurities. The main
impurities of the quartz used in the present work are
Al, Ca, K, Na, Ti and their total content doesn’t
exceed 50 ppm. If these inclusions were indeed the
only chemisorption sites for N atoms, pyrex
containing few percents of different admixtures (B,
Na, Al, K) would posses much higher density of
adsorption sites. So the active sites on silica surface
necessarily originate from the SiO2 structure.
8x10
13
6x10
13
4x10
13
2x10
13
pyrex
NO (cm-3)
The inner surface of the discharge tube was
pretreated by flowing discharge in pure nitrogen
(p=0.53 mbar, Ppl=15W), which left chemisorbed
nitrogen atoms on the reactor walls. Then the reactor
was evacuated during 10 minutes and filled with
oxygen. Nitric oxide production induced by
recombination of Nads with gas phase O atoms was
observed when a single pulse O2 discharge (“probe
discharge”, p=0.53 mbar, Ppl=15W) was ignited in
the closed reactor. The initial coverage of Nads was
deduced from the total amount of NO produced [5].
quartz
0
0.0
0.2
0.4
0.6
0.8
1.0
t (s)
Figure 2. NO production in O2 probe discharge after standard
N2 plasma pretreatment of quartz (—■—) or pyrex (—▲—).
3.2 Adsorption kinetics of N on quartz surface
According to existing models, surface atomic
recombination proceeds in two steps [1,8]. First
atoms are irreversibly trapped by surface active sites
and then they recombine with gas phase or
physisorbed atoms. In [8] V. Guerra demonstrated
that on clean silica-like surface chemisorption sites
become populated in millisecond time scale after the
beginning of plasma exposure following single
exponential saturation law: Nads~N0(1-exp(-t/ads)).
In order to verify if model predictions adequately
describe the behavior of chemisorbed nitrogen atoms
on quartz surface, the adsorption kinetics of N atoms
was studied. First the reactor was treated by flowing
discharge in Ar during 1 hour to obtain a clean
surface free of adsorbed atoms. Then standard N2
plasma pretreatment was performed for time
duration ranging from 6 milliseconds up to 5 hours.
After the pretreatment the reactor volume was
evacuated and then NO production in O2 probe
discharge was followed. The resulting density of
chemisorbed N atoms was deduced from the
maximum measured NO concentration.
In Figure 3 the density of chemisorbed N atoms is
plotted in a semi logarithmic scale as a function of
the pretreatment duration. In the inset the same data
are depicted in linear scale.
As it follows from figure 3, experimentally observed
evolution of [Nads] is slower than predicted by the
model. After 100 ms pretreatment concentration of
adsorbed nitrogen atoms attains only 20% of its
maximum value. Moreover the fractional surface
coverage is not described by exponential time
dependence but rather by a logarithm of the
pretreatment duration. Therefore we conclude that
experimentally observed chemisorption kinetics is
3x10
by
existing
models
even
13
Nads, cm-2
(2)
2x10
13
(1)
0
1x10
2000
3.3 XPS analysis of pretreated surface.
4000
13
0
0.01
0.1
1
10
imagine N+ ions that are adsorbed and then
neutralized on the surface. Furthermore it is known
that N2 plasma treatment leads to the formation of a
few nanometer thick SiOxNy layer on the surface of
SiO2 thin films [10]. As a consequence, creation of
adsorption sites may be related to the changes in the
composition and stoichiometry of the initial SiO2
surface.
100
1000 10000
pretreatment duration, s
Figure 3. The density of chemisorbed N atoms on quartz tube
surface as a function of the pretreatment duration
It is evident that the basic assumptions made in
the models should be modified. Two hypotheses
may be proposed in order to explain observed
results:
1. “Slow adsorption on existing sites.” One can
assume that the sticking probability for
chemisorption sites is less than 1. At the same time
non-exponential time dependence of the fractional
surface coverage may result from the spectrum of
adsorption probabilities for different sites (similar to
the distribution of adsorption energies proposed by
Donnelly et al. [9]).
2. “Creation of adsorption sites by N2 plasma.” The
first hypothesis premises on the fact that adsorption
sites are initially present on the surface after Ar
plasma treatment. However, one can suggest that the
density of adsorbed atoms is controlled not by the
adsorption processes but by the availability of free
adsorption sites. So the long timescale of N
adsorption could arise from the modification of the
surface by N2 plasma and re-creation of adsorption
sites that were probably suppressed during Ar
plasma cleaning.
Experiments with pretreatment of clean
quartz surface by flowing N2 afterglow showed that
adsorption of nitrogen atoms alone is negligible and
direct plasma exposure is needed to graft N to the
surface. The role of energetic plasma particles (ions,
photons, excited species) may consist in creation of
new adsorption sites and cleaning of existing sites
by removing adsorbed species (for example
adsorbed water molecules). In addition impinging
ions themselves may form Nads. For example one can
In order to investigate how N2 plasma modifies SiO2
surface, an ex-situ study of pretreated quartz samples
was performed. Samples were placed in the active
zone of the discharge for 6∙10 1 (Sample 1), 1.5∙103
(Sample 2) and 2∙10 4 (Sample 3) seconds. Figure 4
shows survey spectra of the sample 3 that was
treated during 2∙104 s in the same conditions as on
figure 3.
1
Intensity (a.u.)
not reproduced
qualitatively.
O1s
N1s
Si2s
Si2p
C1s
0
600
400
200
binding energy (eV)
Figure 4. Survey XPS spectrum of quartz sample pretreated by
N2 plasma during 2∙104 s
A N1s peak at 400 eV is clearly observed,
Concentration of atomic nitrogen in the surface layer
was determined for 3 samples: 1){N] <DL
2)[N]=13% 3)[N]=29%. Therefore two slopes
marked in the inset on figure 3 correspond to
adsorption on native SiO2 and on SiOxNy layer
respectively. Slow variation of the density of
adsorbed N after 100 seconds of exposure may be
explained the change of the composition of the
surface layer.
It worth mentioning that the density Nads determined
using NO production is significantly lower than the
result of the XPS analysis. This means that most of
the nitrogen atoms that compose SiOxNy layer are
very stable and do not react with O atoms. Only a
small fraction of nitrogen atoms adsorbed on this
layer can produce NO in recombination with O.
3.4 Do Nads indeed participate in surface atomic
recombination?
Here we verify if chemisorbed N atoms play the role
of active sites for recombination as it is usually
presumed [1,8]. We have shown that Nads may
recombine with O atoms producing NO. In order to
“visualize” recombination reaction of Nads with gas
phase N atoms 15N2 isotope was used.
The quartz surface initially pretreated in a standard
way by N2 plasma with known density of adsorbed
14
N atoms (3∙10 13 cm-2) was exposed to a pulsed
discharge in 15N2. Each discharge pulse was
producing [15N] ~3∙1014 cm-3 according to model
calculation. Eventually all these atoms were lost due
to surface recombination giving 1∙1014 cm-2
recombination reactions per pulse. If adsorbed 14N
indeed play a role of active recombination centers,
15
N should recombine with them and occupy
liberated adsorption sites.
13
3x10
15
Nads, cm-2
14
N
N
13
2x10
3. Conclusions
Adsorption of N atoms on quartz and pyrex has been
studied. It was shown that stable N atoms may be
grafted to quartz surface. Nads recombine with O
atoms producing NO when exposed to O2 plasma.
The maximum density of Nads was found to be
3∙1013cm-2. It was demonstrated that the rate of N
chemisorption under plasma exposure is much
slower than predicted by existing models. Using
XPS analysis of SiO2 samples exposed to N2 plasma
it was shown that a SiOxNy layer containing up to
29% of N is formed on initial SiO2. However only a
small fraction of N atoms adsorbed on this layer is
reactive under O plasma exposure. Using 15N2
isotopic substitution it was proven that strongly
bound nitrogen atoms are not efficient as active sites
for surface N recombination. Therefore we come to
a conclusion that 3 types of nitrogen atoms on quartz
may exist:

Very stable N that form SiOxNy.

Stable adsorbed N (reactive towards O).

N weakly bonded on the active sites that are
responsible for surface catalyzed recombination.
References
13
1x10
0
0
100
200
300
400
500
600
15
number of discharge pulses in N2
Figure 5. Surface densities of 14N and 15N as a function of the
number of plasma pulses in 15N2.
As it follows from figure 4 15N gradually replace
14
N, whereas the total density of adsorbed atoms
stays constant. That means that both isotopes occupy
the same adsorption sites. If we compare now the
absolute number of 14N replaced by 15N with the
estimation of the number of recombination reactions
of 15N, we get that they differ by more than 2 orders
of magnitude. Therefore, most of 15N recombine
without replacing 14N, in other words strongly bound
14
N atoms are not the main sites for surface
recombination of 15N. Some other active site for
surface recombination of N should exist. Atoms
adsorbed on these sites are not thermodynamically
stable at room temperature and hence they are not
detected in the present experiments.
[1] Kim Y and Boudart M 1991 Langmuir 7 29993005
[2] van Helden J, Zijlmans R, Schram D and Engeln
R 2009 Plasma Sources Sci. Technol. 18 025020
[3] F Thevenet, O Guaitella, C. Guillard, E Puzenat,
G Stancu, J Ropcke and A Rousseau I.J.PEST Vol
1., No. 1 (2007)
[4] U Roland, F Holzer and F-D Kopinke Catalysis
Today 73, 315-323 (2002)
[5] D Marinov, O Guaitella, A Rousseau and Y
Ionikh J. Phys. D: Appl. Phys. 43 (2010)
[6] O Guaitella, C Lazzaroni, D Marinov, and A
Rousseau Appl. Phys. Lett. 97, 011502 (2010)
[7] O Guaitella, M Hübner , S Welzel, D Marinov, J
Röpcke and A Rousseau Plasma Sources Sci.
Technol. 19 (2010) 045026
[8] V Guerra and J Loureiro Plasma Sources Sci.
Technol. 13 (2004) 85–94
[9] V. M. Donnelly, J. Guha and L. Stafford J.Vac.
Sci. Technol. A 29(1), Jan/Feb 2011
[10] Seino T, Matsuura T and Murota J 2002 Surf.
Interface Anal. 34 451–455