An efficient oxygen activation route for improved ammonia oxidation through an oxygen-permeable catalytic membrane

Article abstract of DOI:10.1002/cctc.201400048

Upon using a reactant/oxygen mixture as co-feed in partial oxidation, adsorbed surface molecular oxygen species can cause low selectivity. We propose a concept different from the conventional co-feed partial oxidation process in packed-bed reactors. In this new configuration, the activation of oxygen is separated from the catalytic oxidation by using an oxygen-permeable membrane to suppress the formation of nonselective surface molecular oxygen species. A continuous flux of lattice oxygen through the membrane allows highly selective partial oxidation. In the oxidation of ammonia to NO, the NO selectivity was improved from 77 to 95 % if a La_0.6Sr_0.4Co _0.2Fe_0.8O_3-δ oxygen-permeable catalytically active membrane was used at 850 °C instead of a co-feed fixed bed reactor. Got to get some activated oxygen through: By using an oxygen-permeable membrane, the activation of oxygen is separated from the catalytic oxidation, which suppresses the formation of nonselective surface molecular oxygen species that can cause low selectivity. A continuous flux of lattice oxygen through the membrane allows highly selective partial oxidation. As a result, the selectivity for NO in the oxidation of ammonia is improved.

Full text of DOI:10.1002/cctc.201400048

CHEMCATCHEM
COMMUNICATIONS
DOI: 10.1002/cctc.201400048
An Efficient Oxygen Activation Route for Improved
Ammonia Oxidation through an Oxygen-Permeable
Catalytic Membrane
Zhengwen Cao,[a] Heqing Jiang,*[b] Huixia Luo,[a, c] Stefan Baumann,[d]
Wilhelm A. Meulenberg,[d] Hartwig Voss,[e] and Jꢀrgen Caro*[a]
Upon using a reactant/oxygen mixture as co-feed in partial oxi-
dation, adsorbed surface molecular oxygen species can cause
low selectivity. We propose a concept different from the con-
ventional co-feed partial oxidation process in packed-bed reac-
tors. In this new configuration, the activation of oxygen is sep-
arated from the catalytic oxidation by using an oxygen-perme-
able membrane to suppress the formation of nonselective sur-
face molecular oxygen species. A continuous flux of lattice
oxygen through the membrane allows highly selective partial
oxidation. In the oxidation of ammonia to NO, the NO selectivi-
ty was improved from 77 to 95% if a La0.6Sr0.4Co0.2Fe0.8O3–d
oxygen-permeable catalytically active membrane was used at
8508C instead of a co-feed fixed bed reactor.
for the oxidation of ammonia. A series of less expensive
oxides, especially cobalt oxide (Co3O4) and perovskite structure
oxides, such as Ca- or Sr-substituted LaMnO3, LaCoO3, and
LaFeO3, have been reported to show promising catalytic per-
formance towards ammonia oxidation.[2] Considering that N2
from air represents approximately 70% of the total flow in the
current plants, Pꢁrez-Ramꢂrez et al. proposed a novel process
for the oxidation of ammonia by using a lanthanum ferrite
based oxygen-permeable membrane reactor, in which the
oxygen for the ammonia oxidation was separated from air and
in situ supplied through the membrane; moreover, N2 in the
air flow was completely kept on the other side of the mem-
brane. By using a La0.8Sr0.2FeO3–d membrane as the reactor for
the oxidation of ammonia, high NO selectivities (95–98%)
were obtained only upon feeding very limited amounts of am-
monia (<0.5 cm3 minꢀ1 NH3 in a total flow of 130 mLminꢀ1 at
1100 K).[3] To obtain a larger oxygen permeation flux, Sun et al.
recently employed a Ba0.5Sr0.5Co0.8Fe0.2O3–d membrane with high
oxygen permeability for the oxidation of ammonia.[4] However,
the NO selectivities were relatively low (<70%). It is, therefore,
desirable to study the catalytic process on the membrane sur-
face to improve the membrane-based oxidation of ammonia
further.
Extensive research efforts have been devoted to improve the
Ostwald process (4NH3 +5O2!4NO+6H2O), as this step
covers approximately 90% of the cost of nitric acid produc-
tion.[1] For many years, nitric acid producers relied on platinum
group metal alloy gauzes as catalysts that show very good ac-
tivity and 94–96% NO selectivity with the formation of N2 and
N2O as byproducts. However, the utilization of Pt–Rh gauzes is
expensive and is linked to N2O emission and platinum loss in
the form of volatile oxides. These facts stimulated research to-
wards the development of new catalysts and new processes
During ammonia oxidation on metal oxide catalysts, follow-
ing the concept of Mars and van Krevelen, the participation of
lattice oxygen[5] was found to play a very important role in the
formation of NO.[6] The formation of lattice oxygen by reoxida-
tion of the oxygen-depleted metal oxides by gaseous oxygen
has been suggested as follows [Eq. (1)]:[7]
[a] Z. Cao, Dr. H. Luo, Prof. Dr. J. Caro
Institute of Physical Chemistry and Electrochemistry
Leibniz University of Hannover
Callinstrasse 3A, Hannover 30167 (Germany)
E-mail: juergen.caro@pci.uni-hannover.de
O2ðgÞ ! O2ðadsÞ ! ðO2ꢀðadsÞ ! O22ðꢀadsÞÞ !
ð1Þ
[b] Prof. Dr. H. Jiang
2ꢀ
2ꢀ
ð2 OꢀðadsÞ ! 2 O Þ ! 2 O
ðadsÞ
ðlatticeÞ
Key Laboratory of Biobased Materials
Qingdao Institute of Bioenergy and Bioprocess Technology
Chinese Academy of Sciences
No.189 Songling Road, 266101 Qingdao (China)
E-mail: jianghq@qibebt.ac.cn
First, oxygen molecules adsorb on the oxide surface, and the
adsorbed O2 becomes reduced and is incorporated into the
metal oxide as lattice oxygen O2ꢀ, as shown in Figure 1a. Many
efforts have been taken to identify the role of various oxygen
species. It is accepted that ammonia can be directly oxidized
by lattice oxygen to NO,[8] and a low selectivity for NO is relat-
ed to high concentrations of adsorbed molecular oxygen spe-
cies that react with adsorbed ammonia to form nitrogen.[9]
Very recently, Biausque et al. reported that one key route to
form N2 in the oxidation of ammonia over perovskite LaCoO3
consisted in the reaction of adsorbed ammonia with adsorbed
oxygen species such as peroxide (O22ꢀ) and superoxide (O2ꢀ)
[c] Dr. H. Luo
Department of Chemistry, Princeton University
Princeton, NJ 08544 (USA)
[d] Dr. S. Baumann, Dr. W. A. Meulenberg
Forschungszentrum Jꢀlich GmbH
Institute of Energy and Climate Research
Leo-Brandt-Strasse, 52425 Jꢀlich (Germany)
[e] Dr. H. Voss
BASF SE
Ludwigshafen 67056 (Germany)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201400048.
ꢃ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1190 – 1194 1190

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
Figure 2. SEM micrographs of a cross-section of a fresh LSCF membrane.
To verify our concept, the oxidation of ammonia in the
membrane reactor with oxygen supplied as lattice/atomic
oxygen was compared with the direct feeding of gaseous
oxygen (co-feed mode) by using the same LSCF membrane as
the catalyst at 8508C (Figure S1, Supporting Information). In
the co-feed mode, if the oxygen and ammonia are jointly fed
to the porous support side of the LSCF membrane, the LSCF
membrane acts only as a contact catalyst providing a large sur-
face. However, in the membrane mode, oxygen for the oxida-
tion of ammonia is separated from air by the LSCF membrane
layer. We show the results of this comparison in Figure 3. The
Figure 1. Ammonia oxidation over a) a conventional packed-bed perovskite
catalyst and b) a perovskite membrane. In a), oxygen as a co-feed becomes
incorporated into the oxide catalyst and adsorbed molecular oxygen species
are present at the early stage of incorporation. In b), for which a membrane
is in operation, lattice oxygen species and atomic oxygen species are pres-
ent at the early stage of oxygen release.
species.[10] These experimental findings led us to the assump-
tion that a high selectivity for NO could be achieved if the
only oxygen species involved in the ammonia oxidation was
lattice oxygen or atomic surface oxygen species.[11] However,
this concept is difficult to implement, as the formation of
weakly bound molecular oxygen species is the first and un-
avoidable step in the activation of oxygen in conventional cat-
alytic packed-bed reactors with an oxygen/ammonia mixture
as co-feed.[12]
To minimize the formation of surface molecular oxygen spe-
cies, herein we developed a concept to separate the activation
of oxygen from the catalytic process through an oxygen-per-
meable membrane.[13] As illustrated in Figure 1b, O2 is reduced
to lattice oxygen O2ꢀ on the air side of the membrane, which
is locally separated from the catalytic oxidation of ammonia
where oxygen is consumed. As lattice oxygen, the oxygen ions
diffuse through the membrane to the permeate side at which
a lower oxygen partial pressure exists. However, given that the
first oxygen species present on the ammonia side before
oxygen release is lattice oxygen, a proper consumption of the
lattice or atomic oxygen species by the oxidation of ammonia
may effectively suppress the formation of nonselective surface
molecular oxygen species. Consequently, a high selectivity for
NO can be expected.
Figure 3. NO selectivity in the oxidation of ammonia for the two different
operation modes at 8508C in a LSCF membrane reactor. The porous support
layer of the asymmetric LSCF membrane was facing the NH3/O2 mixture in
the co-feed mode or facing pure NH3 in the membrane mode (membrane
mode, air side: Fair =100 cm3 minꢀ1; ammonia side: Fall =100 cm3 minꢀ1
,
F
F
=1.6 cm3 minꢀ1
F
Ne =1 cm3 minꢀ1, FHe =balance; co-feed mode:
NH3
Ne =1 cm3 minꢀ1, FNH =1.6 cm3 minꢀ1, FO +FHe =97.4 cm3 minꢀ1). The inset
3
2
shows the NO selectivity for the two different modes if supplying
1.6 cm3 minꢀ1 oxygen.
This work demonstrates that the use of an oxygen-perme-
able membrane with sufficient surface area can supply catalyti-
cally selective lattice oxygen or atomic surface oxygen species
and suppress the formation of nonselective molecular oxygen
species. A newly developed asymmetric ultrathin perovskite
LSCF (La0.6Sr0.4Co0.2Fe0.8O3–d) membrane was used.[14] As shown
in Figure 2, a thin (25 mm), dense LSCF layer is supported by
a mechanically stable porous LSCF layer that is approximately
800 mm thick. The LSCF material was selected as the oxygen-
permeable membrane, as it combines high permeability with
an intrinsic catalytic activity towards the oxidation of ammo-
nia.[2,15] This, as a rule, is a prerequisite for use in membrane re-
actors.
highest NO selectivity obtained in the membrane mode after
reasonable optimization of the reaction parameters was 95%,
whereas the maximum NO selectivity in the co-feed mode was
only 77%. Notably, the amount of permeated oxygen in the
membrane reactor mode and the co-feed fixed-bed reactor
mode was the same. Given that the operating temperature for
both reactors was also the same, the strength of the bonding
of oxygen to the surface was also comparable. The only differ-
ence between the co-feed mode and the membrane mode
was the way in which oxygen was supplied: In the co-feed
ꢃ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1190 – 1194 1191

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
fixed-bed reactor, adsorption and reduction of gaseous oxygen
and the oxidation of ammonia take place in the same compart-
ment. As we can expect from the reduction of oxygen shown
in Figure 1a, in the fixed-bed co-feed reactor, adsorbed molec-
ular oxygen species can react with ammonia to form undesired
nitrogen. In contrast, in the membrane mode the activation of
oxygen is separated from the catalytic reaction, and therefore,
the formation of nonselective surface molecular oxygen spe-
cies can be suppressed. The NO selectivity as a function of the
fed oxygen was also tested in a small range for the co-feed
mode, as shown in Figure 3, but the NO selectivity obtained in
the membrane reactor was always higher than that in the co-
feed mode.
In an additional experiment, instead of the porous support
layer, the dense layer of the LSCF membrane was exposed to
the ammonia side. Figure 4 shows the NO selectivity and the
Figure 5. Asymmetric oxygen-permeable membrane with a) a porous surface
and b) a smooth surface facing the ammonia side of the membrane reactor.
surface oxygen species (molecular).[16] Therefore, there are
more lattice oxygen and atomic surface oxygen species that
can be consumed in selective oxidations before molecular
oxygen species are formed. In contrast, if the dense membrane
layer faces the ammonia oxidation side, the relatively high con-
centration of permeated oxygen ions can combine to form
weakly bound oxygen species that then desorb as molecular
oxygen from the dense LSCF side, as shown in Figure 5b; this
enhances the formation of the undesired nitrogen byproduct
as discussed above. With the help of the residual partial pres-
sure of oxygen at the outlet of the ammonia side (Figure 6), it
might be helpful to further reveal the dominating step in both
modes. If the dense side faces ammonia, the oxygen partial
Figure 4. NO selectivity and ammonia conversion in the oxidation of ammo-
nia as a function of the concentration of ammonia in the feed in the two dif-
ferent operation modes of the LSCF membrane reactor at 8508C with the
porous support layer to ammonia and the dense layer to the ammonia side
(air side: Fair =100 cm3 minꢀ1; ammonia side: Fall =100 cm3 minꢀ1 FNH =1.3–
3
3.0 cm3 minꢀ1, FNe =1 cm3 minꢀ1, FHe =balance).
conversion for the ammonia oxidation reaction plotted versus
the concentration of ammonia in the feed gas for these two
operation modes: Either the dense LSCF or the porous LSCF
support layer was facing ammonia. If the ammonia oxidation
took place on the surface of the dense LSCF layer, a low NO se-
lectivity (<10%) was found, whereas the selectivity was ap-
proximately 90% if ammonia oxidation took place on the
porous LSCF support. Although the ammonia conversions
were comparable for these two operation modes (Figure 4),
the NO yield was much lower on the dense LSCF layer than on
the porous LSCF layer that was used as a large contact surface
for the oxidation of ammonia. Given that the surface morphol-
ogy—porous versus dense—was the major difference, the sur-
face area appears to be the determining factor for the per-
formance of the membrane reactor. Relative to a dense sur-
face, as shown in Figure 5a, the porous layer provides a much
higher surface area. The permeated oxygen ion (atomic) can
be sparsely distributed through the porous network, and this
leads to sufficient suppression of the formation of nonselective
Figure 6. The residual oxygen partial pressure on the ammonia oxidation
side as a function of the concentration of ammonia in two different opera-
tion modes, that is, porous and dense side, of the asymmetric membrane at
the reaction side. 1 bar=100 kPa.
ꢃ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1190 – 1194 1192

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
pressure is clearly higher; this is indicative of a higher concen-
tration of weakly bound surface molecular oxygen species. In
this case, the recombination of the lattice oxygen probably
occurs prior to the reaction between the lattice oxygen and
ammonia, and thus, ammonia is mainly oxidized to nitrogen.
This finding further explains the low selectivity for NO if the
dense side faces ammonia.
selective molecular oxygen species was suppressed. The large
surface provided by the porous perovskite layer diluted the
permeated oxygen ions (lattice oxygen) and suppressed the
formation of weakly bound molecular surface oxygen species,
which thus led to a high selectivity for NO. In the oxidation of
ammonia, 95% NO selectivity and 81% ammonia conversion
were achieved in a catalytic membrane reactor with an asym-
metric perovskite La0.6Sr0.4Co0.2Fe0.8O3–d oxygen-permeable cata-
lytically active membrane at 8508C. The oxygen used in the se-
lective oxidation of ammonia was taken from the perovskite
membrane, and the membrane was reoxidized on the air side.
In the classical Mars and van Krevelen mechanism, this reoxida-
tion takes place by gaseous oxygen present in the co-feed.
This work may open the door for various selective oxidation
processes and a detailed analysis of their mechanisms.
Under the assumption that the good performance of the
membrane reactor in the ammonia oxidation reaction benefits
from the specific oxygen supply and the surface morphology
of the membrane, that is, the manner in which the oxygen is
supplied, the two oxygen-permeable pervoskite materials,
LSCF and Ba0.5Sr0.5Co0.8Fe0.2O3–d (BSCF), were compared by
using comparable asymmetric membranes (Figure 7). Given
that the BSCF membrane showed a higher oxygen transport
Acknowledgements
We kindly thank the European Union through the FP7 NASA-OTM
project (grant agreement no. 228701) for financial support. H.J. is
thankful for the support of the Recruitment Program of Global
Youth Experts of China.
Keywords: ammonia · membrane reactors · membranes ·
oxidation · oxygen activation
[1] V. A. Sadykov, L. A. Isupova, I. A. Zolotarskii, L. N. Bobrova, A. S. Noskov,
V. N. Parmon, E. A. Brushtein, T. V. Telyatnikova, V. I. Chernyshev, V. V.
Lunin, Appl. Catal. A 2000, 204, 59–87.
[2] Y. Wu, T. Yu, B.-S. Dou, C.-X. Wang, X.-F. Xie, Z.-l. Yu, S.-R. Fan, Z.-R. Fan,
L.-C. Wang, J. Catal. 1989, 120, 88–107.
[3] a) J. P. Pꢁrez-Ramꢂrez, B. Vigeland, Angew. Chem. 2005, 117, 1136–1139;
Angew. Chem. Int. Ed. 2005, 44, 1112–1115; b) J. Pꢁrez-Ramꢂrez, B. Vige-
land, Catal. Today 2005, 105, 436–442.
Figure 7. NO selectivity in the oxidation of ammonia as a function of the
concentration of ammonia in the feed gas in membrane reactors with asym-
metric BSCF and LSCF membranes (porous support side of the asymmetric
membranes to the ammonia side, T=8508C; air side: Fair =100 cm3 minꢀ1
ammonia side: Fall =100 cm3 minꢀ1, FNe =1 cm3 minꢀ1, FHe =balance).
;
[4] S. Sun, M. Rebeilleau-Dassonneville, X. Zhu, W. Chu, W. Yang, Catal.
Today 2010, 149, 167–171.
[5] a) H. Over, Y. D. Kim, A. P. Seitsonen, S. Wendt, E. Lundgren, M. Schmid,
P. Varga, A. Morgante, G. Ertl, Science 2000, 287, 1474–1476; b) A.
Bueno-Lꢄpez, K. Krishna, M. Makkee, J. A. Moulijn, J. Catal. 2005, 230,
237–248; c) J. L. Gunjakar, T. W. Kim, H. N. Kim, I. Y. Kim, S. J. Hwang, J.
Am. Chem. Soc. 2011, 133, 14998–15007.
[6] R. Dittmeyer, J. Caro in Handbook of Heterogeneous Catalysis (Eds.: G.
Ertl, H. Knçzinger, F. Schꢀth, J. Weitkamp), Wiley-VCH, Weinheim, 2008,
S. 2221.
rate, a higher ammonia conversion could be realized, but LSCF
showed a significantly higher selectivity for NO. Although a se-
lectivity for NO of approximately 90% was achieved in the
LSCF reactor, the best selectivity obtained in the BSCF reactor
was approximately 75%. This finding is easily understood, as
BSCF exhibits a faster surface exchange rate owing to the fact
that it has a higher oxygen vacancy mobility than LSCF.[17]
Therefore, a higher concentration of nonselective molecular
oxygen species is expected on the membrane surface of BSCF
than on that of LSCF, because the recombination of lattice
oxygen is easier on BSCF, which results in a higher amount of
active peroxide and superoxide ions. This leads to a decrease
in the selectivity for NO at the expense of an increase in the
formation of nitrogen. Evidently, the dominating oxygen spe-
cies delivered by the membrane play a big role in the final
product distribution.
´
[7] A. Bielanski, J. Haber, Catal. Rev. Sci. Eng. 1979, 19, 1–41.
[8] Y. Kosaki, A. Miyamoto, Y. Murakami, Bull. Chem. Soc. Jpn. 1979, 52,
617–618.
[9] a) B. Bahrami, V. G. Komvokis, M. S. Ziebarth, O. S. Alexeev, M. D. Amiri-
dis, Appl. Catal. B 2013, 130–131, 25–35; b) E. V. Kondratenko, J. Pꢁrez-
Ramꢂrez, Appl. Catal. A 2005, 289, 97–103.
[10] G. Biausque, Y. Schuurman, J. Catal. 2010, 276, 306–313.
[11] Y. Xu, A. V. Ruban, M. Mavrikakis, J. Am. Chem. Soc. 2004, 126, 4717–
4725.
[12] a) R. Merkle, J. Maier, Angew. Chem. 2008, 120, 3936–3958; Angew.
Chem. Int. Ed. 2008, 47, 3874–3894; b) J. Pꢁrez-Ramꢂrez, E. V. Kondraten-
ko, J. Catal. 2007, 250, 240–246; c) J. Pꢁrez-Ramı’rez, F. Kapteijn, K.
Schçffel, J. A. Moulijn, Appl. Catal. B 2003, 44, 117–151; d) E. V. Kondra-
tenko, H. Wang, V. A. Kondratenko, J. Caro, J. Mol. Catal. A 2009, 297,
142–149.
In conclusion, we presented a novel route to activate
oxygen for a better catalytic performance in the selective oxi-
dation of ammonia to NO by using an oxygen-permeable per-
ovskite membrane, in which the formation of catalytically non-
[13] a) C. S. Chen, S. J. Feng, S. Ran, D. C. Zhu, W. Liu, H. J. M. Bouwmeester,
Angew. Chem. 2003, 115, 5354–5356; Angew. Chem. Int. Ed. 2003, 42,
5196–5198; b) J. Sunarso, S. Baumann, J. M. Serra, W. A. Meulenberg, S.
Liu, Y. S. Lin, J. C. Diniz da Costa, J. Membr. Sci. 2008, 320, 13–41; c) Z. P.
ꢃ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1190 – 1194 1193

CHEMCATCHEM
COMMUNICATIONS
www.chemcatchem.org
Shao, H. Dong, G. X. Xiong, Y. Gong, W. S. Yang, J. Membr. Sci. 2001,
183, 181–192; d) H. Q. Jiang, H. H. Wang, S. Werth, T. Schiestel, J. Caro,
Angew. Chem. 2008, 120, 9481–9484; Angew. Chem. Int. Ed. 2008, 47,
9341–9344; e) H. Jiang, Z. Cao, S. Schirrmeister, T. Schiestel, J. Caro,
Angew. Chem. 2010, 122, 5790–5794; Angew. Chem. Int. Ed. 2010, 49,
5656–5660; f) Y. Liu, X. Zhu, M. Li, H. Liu, Y. Cong, W. Yang, Angew.
Chem. 2013, 125, 3314–3318; Angew. Chem. Int. Ed. 2013, 52, 3232–
3236; g) H. H. Wang, Y. Cong, W. S. Yang, Chem. Commun. 2002, 1468–
1469.
[15] S. Baumann, J. M. Serra, M. P. Lobera, S. Escolꢅstico, F. Schulze-Kꢀppers,
W. A. Meulenberg, J. Membr. Sci. 2011, 377, 198–205.
[16] C. Batiot, B. K. Hodnett, Appl. Catal. A 1996, 137, 179–191.
[17] a) R. M. L. Wang, Y. A. Mastrikov, E. A. Kotomin, J. Maier, J. Mater. Res.
2012, 27, 2000–2008; b) L. Wang, R. Merkle, J. Maier, ECS Trans. 2009,
25, 2497–2505; c) F. S. Baumann, J. Fleig, H. U. Habermeier, J. Maier,
Solid State Ionics 2006, 177, 3187–3191; d) H. J. M. Bouwmeester, H.
Kruidhof, A. J. Burggraaf, Solid State Ionics 1994, 72, 185–194.
[14] J. M. Serra, J. Garcia-Fayos, S. Baumann, F. Schulze-Kꢀppers, W. A. Meu-
lenberg, J. Membr. Sci. 2013, 447, 297–305.
Received: January 15, 2014
Published online on March 18, 2014
ꢃ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 1190 – 1194 1194