Reaction mechanism of ammonia oxidation over RuO2(1 1 0): A combined theory/experiment approach

Article abstract of DOI:10.1016/j.susc.2009.07.025

Combining state-of-the-art density functional theory (DFT) calculations with high resolution core level shift spectroscopy experiments we explored the reaction mechanism of the ammonia oxidation reaction over RuO₂(1 1 0). The high catalytic act

Full text of DOI:10.1016/j.susc.2009.07.025

Surface Science 603 (2009) L113–L116
Contents lists available at ScienceDirect
Surface Science
journal homepage: www.elsevier.com/locate/susc
Surface Science Letters
Reaction mechanism of ammonia oxidation over RuO2(1 1 0): A combined
theory/experiment approach
b
b
c
c
c
b,
A.P. Seitsonen a, , D. Crihan , M. Knapp , A. Resta , E. Lundgren , J.N. Andersen , H. Over
*
*
a IMPMC, CNRS & Université Pierre et Marie Curie 4 place Jussieu, case 115, F-75252 Paris, France
b Dept. of Physical Chemistry, Justus-Liebig-University, Heinrich-Buff-Ring 58, D-35392 Gießen, Germany
c Dept. of Synchrotron Radiation Research, University of Lund, Sölvegatan 14, S-22362 Lund, Sweden
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 29 May 2009
Accepted for publication 21 July 2009
Available online 24 July 2009
Combining state-of-the-art density functional theory (DFT) calculations with high resolution core level
shift spectroscopy experiments we explored the reaction mechanism of the ammonia oxidation reaction
over RuO2(1 1 0). The high catalytic activity of RuO2(1 1 0) is traced to the low activation energies for the
successive hydrogen abstractions of ammonia by on-top O (less than 73 kJ/mol) and the low activation
barrier for the recombination of adsorbed O and N (77 kJ/mol) to form adsorbed NO. The NO desorption
is activated by 121 kJ/mol and represents therefore the rate determining step in the ammonia oxidation
reaction over RuO2 (1 1 0).
Keywords:
Synchrotron radiation photoelectron
spectroscopy
Density functional calculations
Models of surface chemical reactions
Catalysis
Ó 2009 Elsevier B.V. All rights reserved.
Surface chemical reaction
Ruthenium, ruthenium dioxide
Ammonia, ammonia oxidation
The oxidation of ammonia to NO is the first step in the indus-
trial synthesis of nitric acid (cf. Eq. (1)). In general, the oxidation
reaction of ammonia is catalyzed over Rh stabilized Pt gauzes at
temperatures as high as 1000–1200 K with great efficiency and
selectivity [1]. This reaction is known as the Ostwald process and
its microscopic reaction steps have been studied quite thoroughly
[2–4]:
demonstrated that RuO2 is an efficient and selective oxidation cat-
alyst for ammonia, running at a temperature as low as 550 K [8].
In this communication we report on a combined theory/exper-
iment approach exploring the microscopic reaction steps in the
oxidation of ammonia over RuO2(1 1 0); we have applied this inti-
mate interplay of theory and experiments in model catalysis rou-
tinely and successfully over the past 15 years [9]. In the
exhaustive density functional theory (DFT) calculations [10] we
modelled the surface with five tri-layers of RuO2(1 1 0). Consecu-
tive RuO2(1 1 0) slabs were separated by a vacuum region of 16 Å
(supercell approach). The reaction coordinate was taken to be the
distance between the reacting H atom of NHx, x = 1, 2, 3 and the un-
der-coordinated O atom. The transition state of each reaction path-
way and the corresponding activation barrier were searched with a
constrained minimization technique.
These DFT calculations are corroborated by high resolution core
level shift spectroscopy (HRCLS) experiments performed at the
beam line I311 at MAXII in Lund, Sweden [11]. The photon energies
for the measurements of the N1s core levels were chosen to be
665 eV in order to enhance the surface sensitivity without facing
a too intense background in the spectra by secondary electrons;
the total energy resolutions of N1s spectra was set to be
180 meV. All HRCL spectra were recorded at a sample temperature
of 100 K. The experimental HRCL spectra were decomposed into a
small number of Donjiac–Sunjic profiles convoluted with Gaussian
Pt
5
2
2NH3 þ O2
ꢀ
2NO þ 3H2O:
ð1Þ
1000 K—1200 K
There are two major shortcomings in the Ostwald process:
Firstly, the Pt gauzes must be replaced routinely every 6–
10 months since Pt is corroded by the formation of volatile PtO2
under these harsh reaction conditions [5]. Secondly, the reaction
temperature is very high so that the resulting water steam is extre-
mely corrosive and technically difficult to handle.
Alternative NH3 oxidation catalysts consist of transition metal
oxides [6,7], which allow operation at lower temperatures and
which produce less N2O than commercial Pt gauzes. The main dis-
advantages of oxide catalysts are the requirement of using leaner
NH3/air mixtures of <10% of NH3, which may be mitigated by using
micro reactors, and the higher flow resistance. Very recently, it was
* Corresponding authors. Fax: +49 641 9934559 (H. Over).
E-mail address: herbert.over@phys.chemie.uni-giessen.de (H. Over).
0039-6028/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2009.07.025

L114
A.P. Seitsonen et al. / Surface Science 603 (2009) L113–L116
Fig. 1. (A) Ball and stick model of the clean RuO2(1 1 0) surface. Large balls represent oxygen, and small balls represent ruthenium atoms of RuO2(1 1 0). The bridge-bonded
oxygen atoms Obr and Rucus atoms are indicated. Both surface species are one-fold under coordinated with respect to bulk-coordination. O3f indicates 3-fold (bulk-
coordinated) surface O atoms. (B) The adsorption of oxygen onto the stoichiometric RuO2(1 1 0) surface proceeds via the on-top positions above the Rucus atoms. Oot indicates
on-top adsorbed O atoms.
Fig. 2. Illustration of the microscopic reaction steps in the oxidation of NH3 over RuO2(1 1 0). The activation energies (red) are determined by DFT calculations and are given
in kJ/mol. ꢀHdiff means that the abstracted H from NHx is removed from its direct neighborhood by diffusion along the various O species on the surface.
distributions [12]. The assignment of experimental N1s features to
specific chemical entities on the RuO2(1 1 0) surface is based on
DFT-calculated core level shifts which were calculated by remov-
ing half of an electron from the core orbital and performing a
self-consistent calculation of the electronic structure [13]. Thus
the screening effects (‘‘final state”) of the hole are included. The en-
ergy is obtained as the energy difference between the Kohn-Sham
eigenvalue of the core state and the Fermi energy.
Under typical reaction conditions, ammonia oxidation proceeds
in oxygen excess where most of cus sites are occupied by on-top O
atoms (cf. Fig. 1B). With DFT calculations we studied therefore the
reaction pathway of cus adsorbed NH3 on RuO2(1 1 0) with both
on-top O and bridging O species in its direct neighborhood to
which H-atoms of NH3 can be transferred. In Fig. 2, the various
microscopic reaction steps are compiled including the respective
energy barriers
DEact (red) and total adsorption energies (black).
The RuO2(1 1 0) offers two distinct catalytically active sites with
which molecules from the gas phase can interact, i.e. the under-
coordinated Ru atoms (Rucus) and the under-coordinated bridging
O atoms (Obr) (cf. Fig. 1A). For the stoichiometric RuO2(1 1 0) sur-
face, DFT calculations show that ammonia adsorbs initially on-
top of the Rucus sites with its lone pair pointing towards the Rucus
atom. The adsorbed NH3 molecule forms a hydrogen bond to the
adjacent Obr species. The adsorption energy of NH3 is 150 kJ/mol
at low coverages and 113 kJ/mol for high coverages in very good
agreement with a recent DFT study [14]. The hydrogen transfer
from NH3 to the Obr species is endothermic by 39 kJ/mol with an
activation barrier of 73 kJ/mol. The H-atoms can practically not dif-
fuse along the bridging O rows due to the high diffusion barrier of
240 kJ/mol. Therefore, if no on-top O atoms are present on the
surface then the shifted H atom of adsorbed NH3 is localized at
the adjacent bridging O position. The equilibrium of NH3(ad) ?
NH2(ad) + Obr–H on the stoichiometric RuO2(1 1 0) surface is
exceedingly on the NH3 side for temperatures below 500 K, consis-
tent with corresponding experiments in Ref. [8].
The adsorption energy of NH3 on O pre-covered RuO2(1 1 0) is
165 kJ/mol, again in good agreement with a very recent DFT study
[14]. The hydrogen abstraction from NH3 to Obr is activated by
65 kJ/mol, while that to the on-top O is activated by 73 kJ/mol.
The activation barrier for the first H abstraction is already quite
low if one compares this value with a recently found value on
Pt(100) (120 kJ/mol) [4]. The found activation energies on RuO2
suggest that at 90 K dissociation of NH3 is not favorable in contrast
to conclusions drawn from recent HREELS experiments [8]. This
first dehydrogenation step of NH3 on RuO2(1 1 0) imposes the
highest activation barrier among the H stripping reactions of NH3
similar to the case of methane activation [15]. The next dehydroge-
nation step from the amine NH2 to NH is activated by only 29 kJ/
mol independent of whether the hydrogen atom shifts to the next
bridging O or to the on-top O atom.
The last H transfer step from NH to N proceeds even without
any noticeable activation barrier so that the imide species is likely
to be instable on the RuO2(1 1 0) surface. Altogether these calcu-
lated activation barriers suggest that as soon as the first hydrogen

A.P. Seitsonen et al. / Surface Science 603 (2009) L113–L116
L115
Fig. 3b. The RuO2(1 1 0) surface was exposed to 2.0 L of oxygen
at room temperature (0.8 monolayer of on-top O) and subse-
quently exposed to 0.5 L of NH3 at 100 K. The N1s spectra are
shown in Fig. 2b1, indicating two spectral features at 398.8 and
Binding Energy shifts /eV
6
4
2
0
-2
-4
6
4
2
0
-2
-4
b1) 2L O2 at RT
a) 0.4L NH3 at 100K
NH3(1)
+ 0.5L NH3 at 100K
400.5 eV. The N1s at 398.8 eV is assigned to NH3 adsorbed on Rucus
,
NH3(1)
the other to NH3 adsorbed in the second layer, H-bonded to the on-
top O. Experimental and calculated N1s core level shifts are com-
piled and assigned to specific N-containing surface species in Table
1.
NH3(2)
NH3(2)
NH3(n)
Upon annealing the surface to 250 K, a single peak remains at
398.9 eV, i.e. on-top adsorbed NH3(1) in proximity to on-top O.
Part of the NH3 has already desorbed from the RuO2(1 1 0) surface,
a process which continues upon annealing to 300 K. Further
annealing to 350 K transforms part of the NH3-related emission
into an additional emission at 397.4 eV and a small signal at
400.3 eV. The N1s emission at 397.4 eV is assigned to adsorbed
N, while that at 400.3 eV is ascribed to adsorbed NO. The assign-
ment of the experimental N1s emissions to particular NHx species
and NO is based on DFT calculations. Taking the adsorbed NH3(1)
species in the first layer as the reference, DFT calculations deter-
mine that the N1s core levels of NH2 and NH are shifted by 3.0
and 2.9 eV to lower binding energies, respectively. None of these
features are seen in the experimental N1s core level spectra consis-
tent with the above determined reaction barriers. A recent HREELS
study is partly consistent with this finding in that a NH species
could not be identified during the oxidation of ammonia, while a
NH2 surface species was proposed [8]. The calculated N1s core le-
vel of a naked N atom adsorbed on Rucus atom is shifted by 1.7 eV
to lower binding energy in good agreement with the experiment
(ꢀ1.4 eV).
b3) 2L O2 at RT
+ 0.5L NH3 at 100K, flash to 300K
b2) 2L O2 at RT
+ 0.5L NH3 at 100K, flash to 250K
NH3(1)
NH3(1)
N
NH3(2)
b5) 2L O2 at RT
+ 0.5L NH3 at 100K, flash to 400K
b4) 2L O2 at RT
+ 0.5L NH3 at 100K, flash to 350K
NO
N
NH3(1)
NO
x2
x2
404 402 400 398 396
404 402 400 398 396
Binding Energy/eV
Fig. 3. High resolution core level shift spectra of N1s, explaining the oxidation of
ammonia to NO. The intensities are given in arbitrary units, but the same units are
Further increase of the temperature to 400 K leads to a substan-
tial increase of the NO-related N1s feature at 400.3 eV at the ex-
pense of the NH3 emission. This assignment is based on DFT
calculations which determine for on-top adsorbed NO an N1s shift
of 1.2 eV to higher binding energies w.r.t. NH3(1). Obviously, only
above 350 K neighboring N and O are able to recombine on the
RuO2(1 1 0) surface to form adsorbed NO consistent with recent
HREELS experiments [8]. The abstracted H-atoms migrates then
along the on-top O-segments with an activation energy of
0.22 eV; this process is called ꢀHdiff in Fig. 2.
used in all panel. (a) N1s spectrum when 0.4
L of NH3 is exposed to the
stoichiometric RuO2(1 1 0) surface at 100 K. (b). The stoichiometric RuO2(1 1 0)
surface was exposed to 2.5 L of O2 at room temperature followed by dosing 0.5 L of
NH3 at 100 K and then annealed to various temperatures: (b1) 250 K, (b2) 300 K, (b3)
350 K, and (b4) 400 K.
is transferred from an ammonia molecule to the on-top oxygen
covered catalyst surface the further H abstraction proceeds auto-
matically. Therefore reaction intermediates in the form of NHx,
x < 3 are of transient nature and accordingly difficult to detect with
spectroscopy.
The N1s spectrum in Fig. 3a of NH3 adsorbed on the stoichiom-
etric RuO2(1 1 0) surface shows three signals at 398.4, 400.5 and
403.5 eV which are assigned to adsorbed NH3 in the first
(NH3(1)), second (NH3(2)) and multi-layer (NH3(3)), respectively.
These assignments are based on thermal desorption spectra of
NH3 [8]. Only for NH3 of the first layer (NH3(1)), ammonia adsorbs
on-top of the Rucus sites.
According to our DFT calculations, the most active O species in
the recombination of N and O to form NO is the on-top O species
with an activation energy of 77 kJ/mol. This activation barrier is
slightly too low to be compatible with a reaction temperature of
about 350–400 K as observed in Fig. 2b4). However, 400 K is just
the temperature where water desorbs from the RuO2(1 1 0) surface
[16,17]. Therefore, we conclude that the NO production is hindered
by Oot–H groups in the direct vicinity of N. Only when the Oot–H
groups recombine to form water that desorbs at 400 K, nitrogen
can react with the remaining Oot on the surface to form NO. Fully
consistent with this proposed reaction scheme a recent HREELS
study indicated that the majority of NO is produced only when
The DFT-based reaction mechanism of ammonia oxidation is ni-
cely reconciled with the experimental N1s spectra shown in
Table 1
N1s core level binding energies determined by HRCLS. Corresponding N1s core level shifts are compared to DFT-calculated ones with having the N1s of NH3(1) as reference.
Experimental N1s core level energy
(eV)
Experimental N1s core level shift
(eV)
Assignment based ()
DFT-calculated N1s core level shifts
(eV)
398.4
400.5
403.5
398.8
ꢀ0.4
1.7
4.7
NH3 adsorbed on Rucus: first layer (TDS) [8]
NH3 second layer (TDS) [8]
NH3 third layer (TDS) [8]
NH3 adsorbed on-top of Rucus with on-top O
neighbor
–
–
–
0
0.0
397.4
400.3
–
ꢀ1.4
1.5
–
N adsorbed on-top of Rucus (DFT)
NO adsorbed on-top Rucus (DFT)
NH2 adsorbed on-top Rucus (DFT)
NH adsorbed on-top Rucus (DFT)
ꢀ1.7
1.2
ꢀ3.0
–
–
ꢀ2.9

L116
A.P. Seitsonen et al. / Surface Science 603 (2009) L113–L116
the H2O species desorbs at 400 K [8]. The resulting NO molecule is
quite strongly adsorbed on the surface by 191 kJ/mol so that
desorption of NO molecules takes place only above 500 K as re-
cently observed in temperature programmed reaction [8]. Slightly
lower values for the NO adsorption energies were calculated in a
recent DFT study by Hong et al. [18].
References
[1] G. Ertl, H. Knözinger, J. Weitkamp (Eds.), Handbook of Heterogeneous
Catalysis, Wiley, New York, 1997.
[2] W.D. Mieher, W. Ho, Surf. Sci. 322 (1995) 151.
[3] J.M. Bradley, A. Hopkinson, D.A. King, J. Phys. Chem. 99 (1995) 17032.
[4] M. Baerns, R. Imbihl, V.A. Kondratenko, R. Kraehnert, W.K. Offermans, R.A. van
Santen, A. Scheibe, J. Catal. 232 (2005) 226.
In conclusion, the high activity of RuO2(1 1 0) in the ammonia
oxidation towards NO is due to low activation energies for the suc-
cessive hydrogen abstractions. The highest energy barrier among
the successive H abstraction is encountered with the first dehydro-
genation step (73 kJ/mol), resulting in a reaction temperature of
about 300 K to 350 K. The association step of adsorbed N and O
on the RuO2(1 1 0) is activated by 77 kJ/mol which is slightly too
low to be reconciled with a required temperature of 350–400 K
as suggested by Fig. 2b4). Therefore, we suggest that OH groups
in the vicinity of the adsorbed N inhibit the association of N and
O to form NO. The desorption of NO is the rate determining step
for the overall reaction process of NH3 oxidation due to an adsorp-
tion energy of 191 kJ/mol, requiring a reaction temperature of
500 K to prevent product poisoning of the catalyst by NO. Whether
these results are also relevant to practical RuO2-based catalysts
exposing various surface orientations will be the subject of forth-
coming paper where we will present ammonia oxidation experi-
ments of a RuO2 powder catalyst in a flow reactor [19]. For the
catalyzed CO oxidation reaction, at least, it has been shown that
the studies on the (1 1 0) orientation of RuO2 are equally relevant
to other surface orientations of RuO2 [20].
[5] (a) O. Nilsen, A. Kjekshus, H. Fjellvag, Appl. Catal. A 207 (2001) 43;
(b) L. Hannevold, O. Nilsen, A. Kjekshus, H. Fjellvag, Appl. Catal. A 284 (2005)
163.
[6] V.A. Sadykov, L.A. Isupova, I.A. Zolotarski, L.N. Bobrova, A.S. Noskov, V.N.
Parmon, E.A. Brushtein, T.V. Telyatniskova, V.I. Cheyshev, V.V. Lunin, Appl.
Catal. A 204 (2000) 59.
[7] J. Perez-Ramirez, B. Vigeland, Angew. Chem. Int. Ed. 44 (2005) 1112.
[8] Y. Wang, K. Jacobi, W.-D. Schöne, G. Ertl, J. Phys. Chem. B 109 (2005) 7883.
[9] A.P. Seitsonen, H. Over, Surf. Sci. 603 (2009) 1717.
[10] We used generalized gradient approximation (GGA) ‘‘PBE” [J. P. Perdew, K.
Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865], employing the computer
code VASP [G. Kresse, J. Furthmüller, J. Comput. Mater. Sci. 6 (1996) 15] for the
calculations. The electronic wave functions were expanded in plane waves up
to a cut-off energy of 37 Ry, and the projector-augment wave method [P.E.
Blöchl, Phys. Rev. B 50 (1994) 17953; G. Kresse, D. Joubert, Phys. Rev. B 59
(1999) 1758] was used to describe the interaction between the valence
electrons with the nuclei and core electrons. The first Brillouin zone was
sampled by a 4 ꢁ 8 grid of Monkhorst-Pack k points, and the occupation
numbers were broadened with a Fermi-Dirac distribution of width 50 meV.
[11] R. Nyholm, J.N. Andersen, U. Johansson, B.N. Jensen, I. Lindau, Nucl. Instr. Meth.
Phys. Res. A 467 (2001) 520.
[12] S. Donjiac, M. Sunjic, J. Phys. C 3 (1970) 285.
[13] E. Pehlke, M. Scheffler, Phys. Rev. Lett. 71 (1993) 2338.
[14] C.-C. Wang, Y.-J. Yang, J.-C. Jiang, J. Phys. Chem. C 113 (2009) 2816.
[15] J.R. Rostrup-Nielsen, J.H.B. Hansen, J. Catal. 114 (1993) 38.
[16] A. Lobo, H. Conrad, Surf. Sci. 523 (2003) 279.
[17] M. Knapp, D. Crihan, A.P. Seitsonen, E. Lundgren, A. Resta, J.N. Andersen, H.
Over, J. Phys. Chem. C 111 (2007) 5363.
[18] S. Hong, T.S. Rahman, K. Jacobi, G. Ertl, J. Phys. Chem. C 111 (2007) 12361.
[19] Ch. Kanzler, F. Hess, J.-P. Hofmann, H. Over, in preparation.
[20] Y.D. Kim, H. Over, G. Krabbes, G. Ertl, Top. Catal. 14 (2001) 95.
Acknowledgements
We would like to thank the LRZ in Munich for the generous
supercomputing time and the DFG for financial support.