Perovskite membranes in ammonia oxidation: Towards process intensification in nitric acid manufacture

Article abstract of DOI:10.1002/anie.200462024

Intensified nitric acid production is possible when NH₃ is oxidized with Ca- and Sr-substituted lanthanum ferrite perovskites in a membrane reactor with NO selectivities of up to 98% (see picture). In this new process atmospheric N₂ is rejected by the oxygen-conducting membrane, expensive noble metal catalysts are unnecessary, and no environmentally harmful N₂O is produced.

Full text of DOI:10.1002/anie.200462024

Communications
Ammonia Oxidation
Perovskite Membranes in Ammonia Oxidation:
Towards Process Intensification in Nitric Acid
Manufacture**
Javier Pꢀrez-Ramꢁrez* and Bent Vigeland
Ammonia is oxidized over PtRh alloy gauzes to form NO as
the first step in the industrial production of nitric acid, a
process that has remained practically unchanged for over 80
years. The reaction typically yields 94–96% NO and 4–6%
[
1]
by-products (N O and N ) at 1073–1223 K. Major draw-
2
2
backs associated with the platinum-based catalysts are:
) high production cost, 2) metal loss in the form of volatile
1
oxides necessitate efficient metal recovery (Pd catchment)
and refining systems, and 3) the production of N O, an
2
environmentally harmful gas. Nitric acid manufacture is the
largest single source of N O in the chemical industry (125 ꢀ
2
6
1
0 t CO -equiv per annum), and the development and
2
implementation of abatement technology for this gas is
[
2]
required. The above aspects have stimulated research
towards replacing noble metals by oxide catalysts for NH3
oxidation. Oxides may offer the advantage of lower invest-
[
2,3]
ment, simpler manufacture, and reduced N O emission.
A
2
vast number of patent applications have claimed promising
performance of spinels or perovskites in the reaction,
[
3]
preferably containing Co, but also Fe, Mn, Bi, or Cr.
Laboratory, pilot, and industrial tests have typically been
carried out in fixed-bed reactors with oxides in the form of
[*] Dr. J. Pꢀrez-Ramꢁrez
Yara Technology Centre Porsgrunn
Catalysis and Nitric Acid Technology
P.O. Box 2560, 3908 Porsgrunn (Norway)
Fax: (+47)2415-8213
E-mail: javier.perez.ramirez@yara.com
Dr. B. Vigeland
Norsk Hydro, Corporate Research Centre, Materials Development
P.O. Box 2560, 3908 Porsgrunn (Norway)
[**] This research was supported by Yara International ASA.
1
112
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200462024
Angew. Chem. Int. Ed. 2005, 44, 1112 –1115

Angewandte
Chemie
particles, pellets, or monoliths. Several key aspects have
prevented the industrial implementation of oxide catalysts:
1
) relatively low NO selectivity (< 90%), 2) rapid deactiva-
tion under relevant reaction conditions, and 3) the lower
optimal operating temperature compared to noble metal
catalysts causes difficulties with the steam balance in a
[
2]
revamped plant.
Herein we present the first lanthanum ferrite-based
perovskite membranes for ammonia oxidation, with which
NO selectivities up to 98% and no N O formation were
2
attained. Our strategy was to combine in a single membrane
reactor the separately reported properties of the above
[
4]
perovskites as oxygen conductors and catalysts for NH3
Figure 2. NO selectivity and O flux versus inlet ammonia flow at dif-
2
[
5]
ferent temperatures for a La0.8Sr0.2FeO membrane.
oxidation. Accordingly, the applied configuration, depicted
3ꢀd
in Figure 1, integrates the separation of O from air at the
2
selectivities in the range of 90–100% can be obtained by
adjusting the inlet NH flow. No N O was formed (< 10 ppm)
3
2
in these experiments, and N was the only N-containing by-
2
product. Ammonia conversion (80–95%) increases with
increasing temperature and decreases at high ammonia flow
rates. The oxygen flux is expected to increase with temper-
ature at a fixed oxygen potential gradient. Under our
experimental conditions, however, the oxygen flux is almost
exclusively controlled by the ammonia flow rate and thus
independent of temperature. The NO selectivity, however, is
dependent on temperature, and moreover exhibits a max-
imum with the inlet ammonia flow. As expected, the NO
Figure 1. NH oxidation to NO with mixed conducting membranes.
3
selectivity decreases in favor of N under conditions of excess
2
feed side by transport through oxygen vacancies in the mixed
conducting membrane, and the reaction of oxygen with
ammonia on the membrane surface at the permeate side. The
surfaces of the membrane at the feed and permeate sides
ammonia as a result of favorable recombination of adsorbed
[
3,8]
NH fragments.
The additional observation of decreasing
x
selectivity at low ammonia flow rates, which gives rise to the
maximum, is attributed to partial oxidation on the hot reactor
2
ꢀ
function as reduction (O !2O ) and oxidation (NH !NO)
walls of NH bypassing the membrane with excess oxygen
2
3
3
catalysts, respectively. This gives rise to a radically intensified
process for nitric acid manufacture, as large amounts of inert
recombining and escaping the membrane surface. The faster
reaction at high temperature explains the decrease in the
maximum NO selectivity when going from 1000 K (98%) to
1333 K (92%). Implicitly, with these membrane materials,
higher NO selectivities than those displayed in Figure 2 would
have been obtained if wall effects could be excluded. The shift
of the NO selectivity maximum to lower ammonia molar flow
with decreasing temperature reflects the reduced oxygen flux
(at constant oxygen potential) at lower temperatures.
N (2/3 of the total flow in todayꢁs plants) are excluded.
2
Important features of perovskite membranes in ammonia
oxidation include oxygen flux J(O ), selectivity to NO S(NO),
2
and chemical stability in reducing and oxidizing atmosphere
at high temperature. These interrelated parameters can be
tailored by tuning the degree of calcium and strontium
substitution in lanthanum ferrite-based perovskites La1 (Sr,
ꢀx
Ca) FeO . A higher degree of substitution x increases the
Figure 3 shows the maximum NO selectivity and corre-
sponding oxygen flux as functions of the degree of substitu-
tion of Ca and Sr in the lanthanum ferrite perovskites at
1200 K. As expected, the oxygen flux is proportional to the
degree of substitution. The maximum NO selectivity (95–
96%) does not depend on the type and degree of substitution
within this range, and this indicates similar catalytic perform-
x
3ꢀd
number of oxygen vacancies d and thus the oxygen flux, but at
[
4,6,7]
the expense of lower chemical stability.
Furthermore, La
substitution by alkaline earth cations in the perovskite
structure is also known to influence the catalytic properties
[
3]
of these mixed oxides. This can be caused by variation in the
charge and/or coordination of 3d cations, changes in surface
chemical composition, and development of microheteroge-
neities at the catalyst surface.
[
5]
ance of these membrane compositions. Wu et al. studied
NH oxidation at 1073 K in a fixed-bed reactor with catalyst
3
The preparation and characterization of the perovskite
powders and the applied procedure for testing the membranes
in the form of dense disks are described in the Experimental
particles and reported optimal NO selectivities (ꢁ 95%) for
x = 0.2–0.7 in La1 Sr FeO . For the membrane reactor
ꢀx
x
3
described herein, the optimum degree of substitution is
determined by a combination of maximum flux with absolute
chemical stability under process conditions. This optimum
degree of substitution of Sr and Ca probably lies in the range
of x = 0.1–0.2.
Section. For a La Sr FeO membrane disk as an illustra-
0
.8
0.2
3ꢀd
tive example, Figure 2 shows the typical dependence of NO
selectivity and O2 flux on the inlet NH3 flow at different
temperatures. In the temperature range investigated, NO
Angew. Chem. Int. Ed. 2005, 44, 1112 –1115
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1113

Communications
mixed with citric acid in a molar ratio of acid to total cations of 3. The
ꢀ
1
resulting solution was slowly heated to 433 K (40 Kh ) to ensure
complete complexation and left overnight at this temperature to dry.
After further heating to 773 K for 5 h in flowing air to remove organic
matter and final calcination at 1173 K for 10 h, all powders were
single-phase perovskites as evidenced by XRD. The particle size,
obtained by laser scattering, was in the range of 0.1–0.5 mm. Dense
membrane disks (> 95% of theoretical density) were made by
uniaxial pressing and sintering at 1573 K. After final grinding and
2
polishing, membrane disks with a diameter of 10 mm (ꢁ 80 mm ) and
a thickness of ꢁ 0.9 mm were obtained. The disks were mounted in a
quartz microreactor, sketched in Figure 4, which was heated to
Figure 3. Maximum NO selectivity (solid symbols) and corresponding
oxygen flux (open symbols) at 1200 K versus the degree of substitution
in Ca- (circles) and Sr-substituted (diamonds) lanthanum ferrite perov-
skites.
The above results demonstrate the potential of Ca- and Sr-
substituted lanthanum ferrite membranes for the high-tem-
perature oxidation of ammonia; the NO selectivities of up to
9
8% are comparable to those of state-of-the-art PtRh
[
1,2]
alloys.
In addition, formation of undesirable N O is totally
2
suppressed. The implementation of an ammonia oxidation
process based on oxygen-conducting membranes would
constitute a major step change in nitric acid production (a
top-ten product in the bulk chemicals industry) and have a
strong impact on the fertilizer industry. Apart from the
Figure 4. Quartz reactor used in the membrane tests.
1333 K for sealing with gold rings. The oxidation of ammonia was
superior NO selectivity, in situ separation of O from air by
2
investigated at 1000–1333 K by feeding an equimolar O /Ar mixture
2
membranes enables extremely compact and intensified pro-
to the feed side, and NH /He mixtures to the permeate side. The inlet
3
duction units, as N represents about 70% of the total flow in
ꢀ1
2
ammonia flow was varied in the range of 0.05–4.5 mlmin (STP),
ꢀ
1
a current plant. The reduction in total flow would allow a
drastic size decrease in key plant units, including the
absorption tower and the tail-gas train, as well as intensifi-
cation in piping. Moreover, energy savings for compression of
with a total flow of 130 ml min (STP). Product gases were analyzed
on-line with a mass spectrometer and a gas chromatograph. Freedom
from leakages was verified by the absence of Ar on the permeate side.
The NO selectivity was obtained from the concentrations at the
reactor outlet according to S(NO) = c(NO)/[c(NO) + 2c(N O) +
2
the NO gas before the absorption step, which requires high
x
2
c(N )]. The oxygen flux J(O ) was determined from the measured
2
2
pressure, further contributes to an improved nitric acid
production process. In summary, a more efficient and
sustainable process can be attained, which is especially
attractive for nitric acid production that is decentralized
from large existing plants.
concentrations of all O-containing species. Stable membrane per-
formance over the typical testing period (10 days) was verified by
periodically repeating measurements under selected temperature/
flow conditions.
Received: September 17, 2004
Published online: January 13, 2005
Our current research for further implementation of
membrane technology in ammonia oxidation focuses on
scaling up membrane disks to monolithic membranes, with
2
Keywords: heterogeneous catalysis · membranes · oxidation ·
which surface areas in the range of 2000–3000 m per cubic
.
perovskite phases · reactive separation
meter of reactor can be obtained, depending on cell size and
[
9]
wall thickness. The reactor design involves the distribution
of ammonia and air in a monolithic structure with square
channels in a chess-board pattern (one channel for ammonia
and the four surrounding channels for air). The significantly
increased surface area to reactor volume ratio of monoliths
will enable application of substantially higher ammonia
concentrations on the permeate side compared to the disk
configuration used in the present study.
[
1] a) S. L. Hansforth, J. N. Tilley, Ind. Eng. Chem. 1934, 26, 1287;
b) E. Wagner, T. Fetzer in Handbook of Heterogeneous Catalysis,
Vol. 4 (Eds.: G. Ertl, H. Knꢂzinger, J. Weitkamp), VCH,
Weinheim, 1997, pp. 1748 – 1761; c) R. J. Farrauto, C. H. Bartho-
lomew, Fundamentals of Industrial Catalytic Processes, Chapman
& Hall, London, 1997, p. 481.
[
2] J. Pꢃrez-Ramꢄrez, F. Kapteijn, K. Schꢂffel, J. A. Moulijn, Appl.
Catal. B 2003, 44, 117, and references therein.
[
3] 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.
Experimental Section
Lanthanum ferrite-based perovskites La1 A FeO (A = Ca, Sr; x =
ꢀx
x
3ꢀd
0
.1–0.2) were prepared by a conventional wet complexation route
[4] Y. Teraoka, H. Zhang, S. Furukawa, N. Yamazoe, Chem. Lett.
1985, 1743.
with citric acid. Standardized solutions (1m) of the metal nitrates were
1
114
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 1112 –1115

Angewandte
Chemie
[
[
[
5] Y. Wu, T. Yu, B. Dou, C. Wang, X. Xie, Z. Yu, S. Fan, Z. Fan, L.
Wang, J. Catal. 1989, 120, 88.
6] J. E. ten Elshof, H. J. M. Bouwmeester, H. Verweij, Solid State
Ionics 1996, 89, 81.
7] Y. Takeda, K. Kajiura, S. Naka, M. Takano in Proceedings of the
3rd International Conference on Ferrites (Eds.: H. Watanabe, S.
Iida, M. Sugimoto), Japan, 1982, 414.
[
[
8] a) J. M. Bradley, A. Hopkinson, D. A. King, J. Phys. Chem. 1995,
99, 17032; b) J. Pꢃrez-Ramꢄrez, E. V. Kondratenko, V. A. Kon-
dratenko, M. Baerns, J. Catal. 2004, 127, 90.
9] a) F. Kapteijn, J. J. Heiszwolf, T. A. Nijhuis, J. A. Moulijn, Cattech
1
2
999, 3, 24; b) A. Stankiewicz, J. A. Moulijn, Ind. Eng. Chem. Res.
002, 41, 1920.
Angew. Chem. Int. Ed. 2005, 44, 1112 –1115
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1115