Sie sind auf Seite 1von 18

/

ELSEVIER

APPLIED
CATALYSS
I
A:GENERAL

Applied Catalysis A: General 151 (1997)443-460

Ruthenium catalysts for ammonia synthesis at high


pressures: Preparation, characterization, and
power-law kinetics
F. Rosowski, A. Hornung, O. Hinrichsen, D. Herein, M. Muhler *,
G. Ertl
Fritz-Haber-lnstitut der Max-Planck-Gesellschaft, Faradayweg4-6, D-14195 Berlin (Dahlem), Germany
Received 16 June 1996; revised 7 August 1996; accepted 7 August 1996

Abstract
Supported Ru catalysts for NH 3 synthesis were prepared from Ru3(CO)I2 and high-purity
MgO and AIEO 3. In addition to aqueous impregnation with alkali nitrates, two non-aqueous
methods based on alkali carbonates were used to achieve alkali promotion resulting in long-term
and high-temperature stable catalysts. For the reliable determination of the Ru particle size, the
combined application of H 2 chemisorption, TEM and XRD was found to be necessary. The
power-law rate expressions were derived at atmospheric pressure and at 20 bar which were shown
to be efficient tools to investigate the degree of interaction of the alkali promoter with the Ru
metal particles. The following sequence with respect to the turnover frequency (TOF) of NH 3
formation was found: C s 2 C O 3 - R u / M g O > C s N O 3 - R u / M g O > R u / M g O > R u / K - A 1 2 0 3 >
R u / A I 2 0 3. The Cs-promoted R u / M g O catalysts turned out to be more active than a multiplypromoted Fe catalyst at atmospheric pressure with an initial TOF of about 10 -2 s - l for the
non-aqueously prepared C s 2 C O 3 - R u / M g O catalyst at 588 K. The strong inhibition by H 2 was
found to require a lower molar H2:N 2 ratio in the feed gas than 3:1 in order to achieve a high
effluent NH 3 mole fraction. The optimum ratio for C s 2 C O 3 - R u / M g O at 50 bar was determined
to be about 3:2, resulting in an effluent NH 3 mole fraction which was just a few percent lower
than that of a multiply-promoted Fe catalyst operated at 107 bar and at roughly the same
temperature and space velocity. Thus, alkali-promoted Ru catalysts are an alternative to the
conventionally used Fe catalysts for NH 3 synthesis also at high pressure.

Keywords: Ruthenium; Non-aqueous preparation; High pressure

NH 3

synthesis; Kinetics

* Corresponding author.
0926-860X/97/$17.00 Copyright 1997 Elsevier Science B,V. All rights reserved.
PH S0926- 860X(96)00304-3

444

F. Rosowski et al./ Applied Catalysis A: General 151 (1997) 443-460

1. Introduction

Alkali-promoted Ru-based catalysts are expected to become the second


generation NH 3 synthesis catalysts [1 ]. Based on the Kellogg Advanced Ammonia Process (KAAP), the 600 t N H 3 / d a y Ocelot NH 3 synthesis plant started to
produce NH 3 in 1992 using promoted Ru catalysts supported on carbon [2]. The
Ru-based catalysts permit milder operating conditions compared with the magnetite-based systems, such as low synthesis pressure (70-105 bar compared with
150-300 bar) and lower synthesis temperatures, while maintaining higher
conversion than a conventional system [2].
It is known that chlorine acts as severe poison in NH 3 synthesis over Ru
catalysts [3-5]. The addition of chlorine to chlorine-free precursors was found to
create about six activated adsorption sites for H 2 per chlorine adatom [3].
Hence, in recent kinetic studies, chlorine-free Ru complex compounds like
Ru3(CO)I2 [6-8] or Ru(NO)(NO3) 3 [9] were used in the preparation. Aika et al.
[8] demonstrated that MgO is one of the most effective supports for Ru. The
Cs-promoted Ru on MgO ( C s N O 3 - R u / M g O ) prepared from chlorine-free
Ru3(CO)I2 was found to be as highly active for NH 3 synthesis as CsNO3-Raney
Ru or as K-promoted Ru on activated carbon [8].
Alkali-promoted catalysts are usually prepared by impregnation with aqueous
solutions of alkali metal salts such as CsNO 3 [8]. Datye et al. [10] showed that
aqueous impregnation of MgO may lead to severe morphologic changes due to
the partial transformation to the hydroxide phase. The present study reports on
two novel water-free methods using alkali carbonates for the preparation of
alkali-promoted Ru catalysts supported on MgO and A1203 to achieve long-term
and high-temperature stability. The first method uses a solution of Cs2CO 3 in
ethanol for the impregnation of R u / M g O resulting in a significantly higher
catalytic activity than the conventionally prepared C s N O 3 - R u / M g O catalyst.
The second method employs a K2CO3-pretreatment of the y-alumina support
prior to the impregnation with the solution of Ru3(CO)12 in THF. The pretreatment was carried out at 1153 K corresponding to the preparation temperature of
mixed oxides of the /3-alumina type. The structure of /3-alumina consists of
spinel-type (AljlOl6)+ blocks separated by mirror planes in which mobile K
ions are located together with oxygen anions bridging the blocks. Compounds
with the /3-alumina structure are usually non-stoichiometric having nominal
compositions of e.g., K/A1 = 1 / 6 [11]. Busca et al. [12] recently performed a
surface characterization of Ba-/3-alumina which is applied as support for gasturbine combustion catalysts due to its high thermochemical stability. Their
results provide evidence for a predominant basic character of the Ba-/3-alumina
surface [12]. Moggi et al. [6] carried out a pretreatment of the surface of
T-A1203 with KOH followed by calcination in air at 773 K for 8 h.
In the present study, information about the particle size distribution was
obtained from H 2 chemisorption, transmission electron microscopy (TEM) and

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

445

X-ray diffraction (XRD). It is shown that the combined application of all three
techniques is necessary for a reliable assessment of the Ru metal particle size. A
modified volumetric sorption set-up was used which allows H 2 chemisorption
and conversion measurements to be carded out in the same U-tube to avoid
transfer artifacts due to the formation of hydroxides and carbonates in air.
In our laboratory, a systematic study is in progress aiming at a detailed
understanding of the catalytic phenomena involved in the synthesis of ammonia
on ruthenium. In spite of the industrial importance, relatively few studies in the
catalytic literature deal with the kinetics of NH 3 synthesis over supported Ru
catalysts [9,13-17]. These studies focus on the influences of PNn3' PN2 and Pn:
on the rate of NH 3 formation mostly at atmospheric pressure. Alkali promotion
was found to decrease the inhibition by NH 3 significantly and to give rise to a
strong inhibiting effect of H e [9,15,16]. On multiply-promoted Fe catalysts, H 2
has a positive reaction order due to high coverages of adsorbed atomic nitrogen
(N-* ) [18]. The reaction order for N 2 was found to be essentially in unity for
both Ru and Fe catalysts indicating that the dissociative chemisorption of N 2 is
the rate-determining step (rds). To our knowledge, reaction orders for N 2 and
H : at high pressures have not been published yet for supported Ru catalysts.
In his review about "Alternative Non-iron Catalysts," Tennison [1] provides
some kinetic data at high pressure illustrating that a lower PHJPN2 ratio than 3
is favorable for Ru-based catalysts under industrial synthesis conditions without,
however, specifying the active metal area. Recently, Kowalczyk et al. [17]
reported the results of NH 3 synthesis rate measurements at high pressures in
which the active metal area was not specified either. In the latter study, it was
observed that a K-promoted Ru catalyst supported on carbon is much less
sensitive to high NH 3 partial pressures and consequently more active than a
triply-promoted Fe catalyst under these conditions [17]. It is the main intention
of the present study to present both kinetic data sets obtained at pressures up to
50 bar and the results of Ru metal area measurements from which turn-over
frequencies (TOF) are derived. Moreover, power-law rate expressions r = kapp "
P~H3"P~:'P~: were determined both at atmospheric pressure and at 20 bar.
The kinetic data set is intended to serve as the basis for a detailed microkinetic
analysis of NH 3 synthesis kinetics [ 19,20] following the concepts by Dumesic et
al. [21].

2. Experimental

2.1. Preparation
The catalysts were prepared from high purity A1203 (99.99%, Johnson
Matthey) or MgO (Puratronic, 99.996% metals basis, Johnson Matthey) and
Ru3(CO)12 (Johnson Matthey, 99%) by wet impregnation in a rotary evaporator

446

F. Rosowski et al. /Applied Catalysis A: General 151 (1997)443-460

following the procedures in Refs. [8,6,22]. The achieved metal loading was
about 5 wt.% Ru.
In both cases, 1.5 g of the catalyst support were heated in high vacuum at 773
K for 6 h and then dispersed in a solution of 0.157 g Ru3(CO)~2 in 60 ml
THFab , for 4 h at room temperature. The impregnation was carried out under
inert gas atmosphere to avoid contact with air and moisture. After evaporating
the solvent at 313 K, the slightly orange powder was pressed at a pressure of 1.6
MPa into cylindrical pellets which were subsequently crushed and sieved. The
250-800 ~zm sieve fraction was slowly heated in high vacuum up to 723 K to
decompose the carbonyl precursor yielding dark gray grains.
The alkali-promoted catalysts C s N O 3 - R u / M g O and CsNO3-Ru/AI203 were
obtained by impregnating the supported Ru catalysts R u / M g O or Ru/A1203
with solutions of CsNO 3 (99.99%, Strem) in a mixture of acetone-water -- 9 / 1
(vol./vol.). Cs2CO3-Ru/MgO was prepared non-aqueously by impregnation
with a solution of Cs2CO 3 (ultrapure, Johnson Matthey) in absolute ethanol.
After stirring the suspension for 3 h, the solvent evaporated and the catalysts
were dried in vacuum. The atomic ratios were Cs:Ru -- 1:1 for C s N O 3 - R u / M g O
and Cs2CO3-Ru/MgO and Cs:Ru = 3:1 for CsNO3-Ru/AI203. All catalysts
were hygroscopic and were stored in a desiccator.
The R u / K - A I 2 0 3 catalyst was prepared by heating a thoroughly ground
mixture of 1.385 g A1203 (99.99%, Johnson Matthey) and 0.176 g K2CO 3
(analytical grade, Merck) in a platinum crucible at 1158 K in air for 15 h
following the procedure developed by Scholder and Mansmann [11 ] to produce a
ternary oxide of the /3-alumina type with a stoichiometry of K 2 0 . 1 1 A1203.
The reaction temperature was chosen to be close to the melting point of K2CO 3
at 1164 K. At temperatures above 1273 K, y-Al~O 3 transforms into ce-A1203
[23]. Subsequently, the wet impregnation with Ru3(CO)j2 was carried out as
described above.

2.2. Characterization of the catalysts


The Ru metal area was determined by volumetric H 2 chemisorption in the
quartz U-tube of an Autosorb 1-C set-up (Quantachrome) following the procedure described in Ref. [24]. Prior to chemisorption, the catalysts were reduced
by passing 80 cm 3 (STP)/min high-purity synthesis gas (Pn2/PN2 = 3 / 1 ) from
a connected feed system through the U-tube and heating up to 673 K for
alkali-promoted catalysts or up to 773 K for alkali-free catalysts with a heating
rate of 60 K / h . The NH 3 mole fraction in the effluent was analyzed at
steady-state by a non-dispersive infrared detector (BINOS, Fisher-Rosemount)
to determine the catalytic activity. The BET area was measured by static N 2
physisorption in the same set-up.
The TEM morphologies were analyzed in a Siemens Elmiskop 102 instrument at 100 keV. Samples were deposited from a slurry in hexane onto

F. Rosowski et al. /Applied Catalysis A: General 151 (1997)443-460

447

carbon-coated Cu grids and predried before insertion. The XRD data were
collected in transmission with a Stoe Stadi P instrument using monochromated
Cu K~I radiation and a position-sensitive detector. The full width at half
maximum (FWHM) was determined by fitting pseudo-Voigt line profiles.

2.3. Kinetic measurements


The kinetic experiments were carried out in an all stainless steel microreactor
flow system with three gas lines (Ar, N 2, H 2) which were operated at pressures
up to 100 bar. The gases had all a purity of 99.9993%, and were further purified
by means of a self-designed guard reactor [25]. The flows were controlled by
electronic mass flow controllers and the temperatures by microprocessor controllers. The reactor consisted of a glass-lined stainless steel U-tube with an
inner diameter of 3.8 mm which was placed in a copper block oven. The
effluent mole fraction of NH 3 was determined by a non-dispersive infrared
detector (BINOS, Fisher-Rosemount) which was calibrated by using a reference
gas mixture (Linde). Limitations by heat or mass transport were avoided by
using 138 mg of the 250 txm-800 p~m sieve fraction for the kinetic experiments
resulting in bed heights of < 15 nun. The reduction was carried out in a flow of
80 cm 3 ( S T P ) / m i n synthesis gas with a heating ramp of 60 K / h up to 673 K.
The absence of poisoning by oxygen-containing compounds was tested as
described in Ref. [25]. It was possible to run steady-state NH 3 synthesis at
temperatures as low as 500 K. All catalysts were tested for a period of 24 days
at least.

3. Results and discussion


3. I. Characterization of the catalysts
The results of the BET measurements are summarized in Table 1. As
expected, y-A120 3 turned out to be the chemically and mechanically more stable
support with a higher surface area than MgO. Only as a result of aqueous alkali

Table 1
Results of the BET measurements after drying the supports in high vacuum at 773 K for 6 h, after
impregnation with Ru3(CO)12 and NH 3 synthesis at 773 K, and after impregnation with Ru3(CO)12, thermal
decomposition in vacuum, alkali impregnation and NH 3 synthesis at 673 K
Sample

BET area (m2/g)

Sample

BET area (m2/g)

MgO
Ru/MgO
CsNO 3-Ru/MgO
CszCO 3-Ru/MgO

51.6
25.0
23.0
21.0

A1203
Ru/Al203
CsNO 3-Ru/Al203
Ru/K-A1203

110.0
104.4
70.0
91.0

448

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

impregnation, the surface area was found to decrease significantly from 104
m 2 / g to 70 m2/g. For MgO, 52 m 2 / g were observed after drying in high
vacuum at 773 K. For R u / M g O , however, the specific area was found to be
only 25 m2/g after NH 3 synthesis at 773 K. After the strong initial loss in
surface area, alkali impregnation led to a further decrease by only 2 m2/g. Aika
et al. [8] observed a similar decrease in surface area from 90 m 2 / g to 36 m 2 / g
for a 2 wt.% R u / M g O catalyst as a result of aqueous promotion with CsNO 3 in
the molar ratio of Cs:Ru = 1:1. High surface areas of MgO are obviously not
stable under typical catalytic conditions.
The H 2 chemisorption results are summarized in Table 2. The H 2 monolayer
capacities were used to derive Ru metal dispersions and mean particle sizes
assuming spherical particles. On both MgO and A1203, the impregnation with
Ru3(CO)12 resulted in mean particle sizes of about 2 nm after NH 3 synthesis at
773 K. It is remarkable that essentially the same Ru metal areas were obtained
on MgO and A1203 inspite of the largely differing BET areas of the supports.
For R u / M g O , a specific Ru metal area of 14.4 m 2 / g was obtained initially after
NH 3 synthesis at 673 K. Increasing the NH 3 synthesis temperature to 773 K
was found to decrease the metal area to 12.9 mZ/g equivalent to a particle size
of 1.9 nm. For the R u / M g O catalyst impregnated with the aqueous solution of
CsNO 3, the amount of chemisorbed hydrogen was found to be reduced by about
a factor of two. Aika et al. [8] also observed a decrease from 136 txmol H z / g to
104 p~mol H z / g for a 5 wt.% R u / M g O catalyst as a result of impregnation
with an aqueous CsNO 3 solution in the molar ratio of Cs:Ru = 1:1. These
observations raise the question whether the loss in metal area is due to site
blocking as suggested by Nwalor and Goodwin [9] for K - R u / S i O 2 or due to an
increase in particle size.
The TEM micrographs shown in Fig. 1 reveal that the particle size distribution of R u / M g O (Fig. 1A) is fairly homogeneous with an average value of
about 2 nm in good agreement with the result obtained by H 2 chemisorption.
The Ru metal particles are seen to be evenly distributed over the MgO particles
which have a cubic morphology to some extent. However, the TEM micrograph
Table 2
Results of the H2 chemisorption measurements after NH 3 synthesis based on H:Ru = 1:1. The NH 3 synthesis
was run at 773 K with R u / M g O and R u / A I 2 0 3 , and at 673 K with all alkali-promoted catalysts. The mean
particle size was calculated assuming spherical particles
Catalyst

H 2 monolayer
(p~mol H e / g )

Metal area
(me/g)

Dispersion
(%)

Mean particle
size (nm)

Ru/MgO
Ru/A1203
CsNO 3 - R u / M g O
CsNO 3-Ru/AI203
Cs2CO 3 - R u / M g O
Ru/K-AI203

130
118
69
100
57
128

12.9
11.7
6.8
9.9
5.6
12.6

53
48
28
41
23
52

1.9
2.1
3.6
2.5
4.3
1.9

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

449

Fig. 1. TEM micrographs after NH 3 synthesis: Ru/MgO (A, upper image), Cs2CO3-Ru/MgO (B, lower
image), CsNO3-Ru/MgO (C, upper image), and CsNO3-Ru/AI203 (D, lower image). The original
magnification was 300000.

of C s N O 3 - R u / M g O (Fig. 1C) displays a rather non-uniform particle size


distribution with particle sizes ranging from 3 nm to 10 nm. The largest particles
appear to be hexagonally shaped. Fig. 1D shows a TEM micrograph of
CsNO3-Ru/A1203. The contrast of the metal particles in Fig. 1D is significantly lower compared with the MgO-based catalysts. However, the absence of
larger Ru particles as seen in Fig. 1C for CsNO3-Ru/MgO can clearly be
confirmed. The estimated particle size of about 2-3 nm demonstrates that the
Ru/A1203 catalyst was not significantly affected by the impregnation with
C s N O 3 in agreement with the minor increase in particle size from 2.1 nm to 2.5
nm derived from H 2 chemisorption.
The increase in size of the Ru metal particles on MgO due to Cs impregnation
is further corroborated by the XRD results shown in Fig. 2. The diffraction
pattern of R u / M g O after NH 3 synthesis at 773 K (upper trace in Fig. 2) reveals
only the presence of highly crystalline MgO. This is in agreement with the Ru
particle size of about 2 nm determined by means of H 2 chemisorption and TEM
since such a particle size lies below the detection limit of the diffractometer. For
the C s N O a - R u / M g O catalyst, an additional weak diffraction pattern is observed which originates from Ru metal. From the widths (FWHM) of the
diffraction peaks, the dimensions of the Ru crystallites are derived applying the
Scherrer equation to the three most intense peaks. The (100), (002) and (101)

F. Rosowski et al. / Applied Catalysis A: General 151 (1997) 443-460

450

15

Ru/MgO

XRD

(200)

~o 10
(220)

Y
[

CsNO 3 - Ru /

MgO

15

10

"E

,loo, i/

(110)

(102) .)
I

30

40

5O
20

60

70

Fig. 2. XRD patterns after NH 3 synthesis of Ru/MgO (upper pattern) and of CsNO 3R u / M g O (lower pattern).

dimensions listed in Table 3 are in good agreement with the TEM results and
indicate rather isotropic Ru crystallites. The XRD patterns of the 3/-A1203
supported catalysts confirmed the poor crystallinity of the support in agreement
with the TEM micrograph shown in Fig. 1D. Diffraction peaks due to Ru were
not detected for Ru/A1203, CsNO3-Ru/AI203, and Ru/K-A1203.
In order to avoid the evident disadvantages of aqueous alkali impregnation for
Ru/MgO, Cs2CO 3 was chosen as alkali precursor, which is soluble in ethanol
Table 3
XRD results: Ru crystallite dimensions of the CsNO 3 - R u / M g O and

Cs2CO3-Ru/MgO catalysts after NH 3

synthesis obtained by applying the Scherrer equation


Diffraction peak
(hkl)
100
002

101

CsNO 3 - R u / M g O
D (nm)

Cs2CO 3 - R u / M g O
D (nm)

10
8
9

9
6
6

F. Rosowski et al. /Applied Catalysis A: General 151 (1997)443-460

451

contrary to C s N O 3. To our knowledge, Cs2CO 3 has not yet been used for this
purpose presumably due to its high decomposition temperature of 880 K. The
reduction of Cs2CO 3 under NH 3 synthesis conditions below 773 K should occur
only in close contact with Ru metal particles which can provide atomic
hydrogen. This assumption is supported by a significantly lowered onset temperature of C s N O 3 reduction in the presence of Ru powder or RuC13 as observed
by Aika et al. [26].
The H 2 chemisorption results shown in Table 2 indicate that apparently the
non-aqueous impregnation with Cs2CO 3 led to even larger Ru particles than the
aqueous process with CsNO 3. The TEM micrograph of the Cs2CO3-Ru/MgO
catalyst shown in Fig. 1B indeed reveals a non-uniform Ru particle size
distribution. However, a considerable fraction of the 2-3 nm Ru particles is seen
to be still present after Cs2CO 3 impregnation as confirmed by the comparison
with the TEM micrograph of R u / M g O (Fig. 1A). Furthermore, the mean size of
the larger Ru particles detected by XRD is found to be considerably smaller as
shown in Table 3. Thus, both TEM and XRD results indicate a significantly
higher Ru metal area for the Cs2CO3-Ru/MgO catalyst compared with the
C s N O 3 - R u / M g O catalyst. The H 2 chemisorption at room temperature may be
blocked by the presence of Cs2CO 3 species thus yielding only a lower limit to
the actual Ru metal area. This conclusion is further supported by the significantly higher catalytic activity of the Cs 2CO3-Ru/MgO catalyst compared with
the C s N O 3 - R u / M g O catalyst as will be discussed further.
For the Ru/K-A1203 catalyst, a BET surface area of 91 m 2 / g (Table 1) was
observed after NH 3 synthesis. The Ru metal area of 12.6 m 2 / g is even
somewhat higher than the Ru metal area of Ru/AI203. The XRD patterns of the
T-A1203 support, the Ru/A1203 and the Ru/K-A1203 catalysts were found to
be essentially identical within the experimental resolution. In order to prepare
crystalline K-/3-alumina, higher amounts of K 2 C O 3 (K:A1 = 1:6) and longer
periods of calcination are necessary which, however, would presumably lead to
a significant loss in specific surface area.
3.2. Kinetics at atmospheric pressure

The results of the initial steady-state conversion measurements at atmospheric


pressure are summarized in Fig. 3. The catalysts with MgO as support were
found to have a significantly higher catalytic activity than the ones with A1203
as support. On both supports, alkali promotion increased the catalytic activity
strongly. The non-aqueously prepared catalyst CszCO3-Ru/MgO turned out to
be almost twice as active than the CsNO3-Ru/MgO catalyst prepared by
aqueous impregnation. The R u / K - A I 2 0 3 catalyst was found to be nearly as
active as CsNO3-Ru/AI203. The activity of the Cs2CO3-Ru/MgO catalyst
exceeded that of a multiply-promoted Fe-based catalyst significantly. Based on
114 ~ m o l / g Ru surface atoms (Ru:H -- 1:1) and a NH 3 synthesis rate of 1.23

452

F. Rosowski et al./Applied Catalysis A." General 151 (1997) 443-460

\
0.8

/,,'~

o~

m/

I::

=
~

'/

~f-"~"

0.6
~=o

~=

"1-

Ru / MgO
CsNO3- Ru / AI203

\
A
" s ~

/
/

Cs2CO 3 - RU / MgO
CsNO 3 - RU / MgO

D\
_

Ru / K - AI20 3
Ru/AI20 3

\~

thermodynamic

equilibrium

0.2

$
0

550

600

650
700
Temperature / K

750

Fig. 3. NH 3 mole fractions in the reactor effluent observed subsequent to reduction for 138 mg of catalyst
using a total flow of 40 c m 3 (STP)/min with XH[XN2 ~ 3 : 1 at atmospheric pressure. Traces A - F (from
bottom to top) were obtained with R u / A I 2 0 3 , R u / K - A 1 2 0 3 , C s N O 3 - R u / A I 2 0 3 , R u / M g O , C s N O 3 R u / M g O and Cs2CO 3 - R u / M g O , respectively. Trace G was obtained with a multiply-promoted Fe catalyst
(KMI, Haldor Topscie). The corresponding NH 3 equilibrium mole fractions at atmospheric pressure are
displayed as dashed line.

txmol/(s g) at 588 K using a synthesis gas flow of 40 cm 3 (STP)/min, a TOF


of about 1 0 - 2 S-1 is derived for C s 2 C O 3 - R u / M g O
which is one of the highest
values reported so far in the literature.
The values of the TOF at atmospheric pressure and at 20 bar derived from the
steady-state kinetic measurements after at least 14 days on stream are summarized in Table 4. For the calculation of the TOF, the Ru metal areas specified in
Table 2 were used. The alkali-promoted catalysts were found to deactivate to
some extent initially. It is evident that alkali promotion is mandatory to obtain
highly active NH 3 synthesis catalysts. Comparing the TOFs at 588 K reveals
that the difference between the least active R u / A I 2 0 3 catalyst and the most
active C s 2 C O 3 - R u / M g O catalyst is almost two orders of magnitude.
Fig. 4 displays the effluent NH 3 mole fraction (XNH~) as a function of
temperature for the feed gas compositions XH=:XN2:XAr = 1 : 1:2 (trace A), 3:1:0
(trace B), and 1:3:0 (trace C). Traces C and B illustrate the enhancing effect of a
higher mole fraction of N 2 and the inhibiting effect of a higher mole fraction of
H 2 in the feed gas, respectively. All reaction orders and apparent activation
energies derived from the steady-state kinetic experiments following the analysis
given in Ref. [15] are summarized in Table 5. The reaction orders of NH 3 were
determined by varying the synthesis gas flow between 40 c m 3 (STP)/min and
160 cm 3 (STP)/min. The reaction orders of N 2 and H 2 were derived by varying

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

453

Table 4
Turn-over frequencies obtained from steady-state kinetic measurements after at least 14 days on stream using a

_
surfaceX
synthesis gas composition of XH2.XN2
-- 3:1. The Ru metal areas ztnRu
) were derived from H 2 chemisorption (Table 2) using a stoichiometry of H:Ru = 1:1. The TOF values have been calculated according to

TOF =

XNH3"Q
22414cm 3 ( S T P ) / m o l . n~Uurfac~

Catalyst

P (bar)

Q (cm 3 (STP)/min)

Ru/MgO

1
20
1
20
1 (initial)
1
20
1
20
1 (initial)
1
20
50
1

120
120
120
120
40
120
120
120
120
40
120
40
100
120

Ru/AI203
CsNOs-Ru/MgO

CsNO 3 - R u / A I 2 0 3
CseCO 3 - R u / M g O

Ru/K-AI203

TOF ( s - 1) at
588 K

623 K

673 K

7.5.10 -4
1.1.10 3
1 / 0 . 1 0 -4
2.5-10 -4
5.3.10 -3
2.9.10 -3
3.0.10 -3
1.7- 10 4
2.8.10 -4
1.0.10 -2
6.6.10 3
5.8.10 -3
5.8.10 -3
1.3.10 -4

1.6.10-3
3.0.10-3
2.5.10 -4
7.5.10 -4
equi.
8.3.10 -3
1.0.10 -2
5.9.10-4
8.3.10 -4
equi.
1.3.10 -2
1.9.10 -2
1.9.10 2
2.9.10 -4

3.7.10- 3
8.8.10-3
6.5.10 -4
2.5-10- 3
equi.
equi.
4.8.10-2
2.8- 10-3
3.6.10 -3
equi.
equi.
8.9.10-2
1.1 10-3

the H2:N 2 ratio between 3:1 and 1:3 using a total flow of 120 cm 3 (STP)/min
with Ar as balance gas. Both determinations were carded out in the temperature
range specified in Table 5 ensuring the measurements to be performed in the
kinetically controlled regime far from equilibrium [27].

CsNO 3 - Ru / MgO
1 bar
120 cm3(STP) / min

E
o. 3000
Q..

"='~o

H2:N2:Ar
1:1:2

.OO0o
_

f
r4/

Z~

I\

E
z

1000

:::
UJ

500

550

600

650

Temperature / K
Fig. 4. Dependence of the NH 3 effluent mole fraction on the feed gas composition observed for CsNO 3Ru/MgO. Trace A was obtained with XH2:Xr%:XAr = 1:1:2, trace B with XH2:XN2:XAr = 3:1:0, trace C with
XH2:XN2:XAr = 1:3:0, respectively, using a total flow of 120 cm 3 (STP)/min.

454

F. Rosowski et al. / Applied Catalysis A: General 151 (1997) 443-460

Table 5
Power-law exponents r = kapp'PNH3"PN," P~t2 and apparent activation energies as a function of the total
pressure determined in the given temperature range using a total flow of 120 cm 3 ( S T P ) / m i n . The latter were
derived from Arrhenius plots at constant flow rate. The accuracy of the determination of the power-law
exponents and of the apparent activation energy is about _+0.1 and _+5 k J / m o l , respectively
Catalyst

P (bar)

T range (K)

a ( N H 3)

/3(N 2)

y ( H 2)

Ea ( k J / m o l )

Ru/MgO

1
20
l
20
1
20
1
20
1
1
I

513-603
573-663
593-663
573-688
498-570
550-630
543 608
573-663
493-543
588-633
558-633

- 0.3
- 0.3
-0.4
-0.5
0.0
0.0
0.0
0.0
0.0
0.0
-0.2

0.8
1.0
0.9
0.9
0.7
0.8
0.7
0.9
0.7
1.0
0.9

- 0.3
- 0.5
-0.1
-0.3
- 0.7
- 0.9
-0.6
- 0.6
-0.6
-0.5
-0.3

69
78
70
76
96
109
103
101
88
125
105

Ru/AlzO 3
CsNO 3 - R u / M g O
CsNO3-Ru/AI203
CseCO3-Ru/MgO
R u / K - A I 2 0 3 , Tm~x = 673 K
R u / K - A I 2 0 3 , T,nax = 773 K

The effect of alkali promotion on the power-law kinetics is threefold: first,


the reaction order for NH 3 is changed to essentially zero, second, the inhibiting
effect of H e is stronger, and third, the apparent activation energy is higher by
more than 20 kJ/mol. The reaction order of N 2 is found to be close to 1.0
reflecting that the dissociative chemisorption of N 2 is the rate-determining step
of NH 3 synthesis for all catalysts under all experimental conditions studied.
Contrary to the results obtained by Aika et al. [15] and Baris et al. [16], the
reaction order for H e was negative for all catalysts investigated. The positive
reaction order for H 2 reported by Aika et al. [15] and Baris et al. [16] for
Ru/A1203 and R u / M g O may be due to the presence of chlorine originating
from RuC13 used for the catalyst preparation.
The reaction orders indicate a rather low fractional coverage of atomic
nitrogen (O N) under synthesis conditions which is further decreased in the case
of alkali-promoted catalysts. On Fe-based catalysts, the catalytic activity is
significantly reduced at high XNH3 leading to high ON blocking further N 2
dissociation [18]. This effect is obviously absent on Ru surfaces. The interaction
of N 2 with supported Ru catalysts has been studied recently in great detail
[28,29].
In our recent publication about the interaction of N 2 with R u / M g O , it was
assumed that a small fraction of the total Ru metal area dominated the rate of
NH 3 synthesis, and that such promoted sites might originate from the interaction
with the MgO support [29]. Comparing the power-law rate expressions of
R u / M g O and R u / A l z O 3 listed in Table 5 reveals that the rate of NH 3
synthesis over R u / M g O is somewhat less inhibited by NH 3 and somewhat
stronger inhibited by H 2. However, the differences are rather small and do
therefore not allow to identify R u / M g O as alkaline-earth promoted catalyst.
Following the approach by Holzman et al. [14], Arrhenius plots at constant XNH3

F. Rosowski et aL / Applied Catalysis A: General 151 (1997)443-460

455

were used to derive apparent activation energies Uax from conversion measurements at atmospheric pressure. For Ru/A1203, a value of 100 kJ/mol was
obtained with 150 ppm < XNH3 < 600 ppm which is significantly higher than the
value of 70 kJ/mol obtained at constant flow. For Ru/MgO, a value of 68
kJ/mol was obtained with 500 ppm <XNH3 < 1250 ppm which is almost
identical to the value of 69 kJ/mol obtained at constant flow. Thus, the different
values of E~ x point out that the kinetics of NH 3 synthesis over Ru/MgO and
Ru/A1203 differ more than indicated by the reaction orders. Shiflett and
Dumesic [4] used the determination of Ea~x for various supported Ru catalysts to
elucidate the role of the support and the effect of chlorine in terms of electronic
interactions.
According to the three criteria derived from the power-law kinetics, both
non-aqueously prepared catalysts belong among the alkali-promoted catalysts
(Table 5). This implies the presence of mobile alkali species on both catalyst
surfaces, originating either from the reduction of Cs2CO 3 or from the K2CO 3pretreated "y-AI203 support, which are able to diffuse onto the Ru metal
surfaces. Hence, alkali promotion can also be brought about by an alkali doping
of the support thus avoiding another impregnation of the supported Ru catalyst.
Since the power-law kinetics of Cs2CO3-Ru/MgO and CsNO3-Ru/MgO
are virtually the same, the higher rate of NH 3 formation per weight of the
former catalyst must result from a higher specific number of Cs-promoted active
sites under synthesis conditions, inspite of the apparently lower specific Ru
metal area determined by H 2 chemisorption at room temperature. Thus, the
different TOF values reported for the two C s - R u / M g O catalysts in Table 4
should be considered an artifact arising from the Ru area determination by H 2
chemisorption as already suggested by the comparison of the TEM, XRD, and
H 2 chemisorption results in the preceding section. For the reliable determination
of the active Ru metal area, N 2 chemisorption turned out to be a valuable tool
[19,29].

3.3. Long-term and high-temperature stability


For the industrial application of alkali-promoted Ru catalysts, the long-term
and high-temperature stability is of great importance. Since the presence of the
alkali promoter is essential for high conversions, the stability of the Ru catalysts
is correlated with the stability of the alkali promoter. After more than 100 h on
stream, the non-aqueously prepared catalysts were heated up to 773 K for 4 h in
a flow of 40 c m 3 (STP)/min synthesis gas initially followed by periodic
temperature changes. Fig. 5 displays the resulting effluent NH 3 mole fractions
and the corresponding temperature profiles observed for Cs2CO3-Ru/MgO
(left half) and Ru/K-AI203 (right half) as a function of time on stream. At 773
K, the value of 1400 ppm for XNH3 is determined by the thermodynamic
equilibrium for both catalysts. Hence, XNH~ passed through a maximum when

456

F. Rosowski et al. / Applied Catalysis A: General 151 (1997) 443-460

Q.

9000

Cs2CO 3 -

Ru / M g O

8000

2000

6000
O

5000

1500

4000

3~ 3000
Z

1000

2000

~ 1000
W
0

Ru / K - A I 2 0 3

2500

".=
7000
O

3000

500

i. . . .

i
LI

J. . . .

'l

773

773
723

723
ir

523
tl,llJ,I,

,iJ,i

,ll,,Jlltl,

100 104 108 112 116 120 124

~"1

673

673

E
I-

588

144148152156160164168

Time on stream / h
Fig. 5. Influence of heating to 773 K on the NH 3 effluent mole fractions (solid lines) as a function of time on
stream observed for C s 2 C O 3 - R u / M g O

(left) and for R u / K - A I 2 0 3

(right) using a total flow of 40 cm 3

(STP)/min with xH2:xN2 = 3:1 at atmospheric pressure. The corresponding temperature profiles are shown as
dashed lines.

heating both catalysts up to 773 K. The decrease of XNH3 from 4750 ppm to
3600 ppm observed at 588 K for the C s 2 C O 3 - R u / M g O catalyst after 8 h at 773
K indicates that the catalyst indeed deactivated to some extent due to the heat
treatment. However, after the second heat treatment, no further changes of the
catalytic activity were found until the end of the run after 42 days on stream. For
R u / K - A 1 2 0 3, the deactivation from XNU3 = 1350 ppm to 1190 ppm observed
at 673 K was less strong, and the catalytic activity was found to be stable
already after the first heat treatment at 773 K.
The reaction orders of R u / K - A I 2 0 3 derived both prior and subsequent to
ammonia synthesis at 773 K (Table 5) reveal that the degree of interaction of the
K + O promoter with the Ru particles changed significantly. The increased
inhibition by NH 3, the decreased inhibition by H 2 and the decrease in apparent
activation energy indicate a transition from an efficiently alkali-promoted state
to a less promoted state. It is plausible to assume that the K + O promoter was
removed from the Ru metal surfaces at 773 K by desorption and diffusion to the
support. Hence, only the interface between the Ru metal particles and the
K-doped A1203 support should be promoted corresponding to R u / M g O . This
assumption is supported by the similar reaction orders of R u / M g O and R u / K A120 3 after ammonia synthesis at 773 K. The greater affinity of Cs + O to the
acidic Al203 support is presumably also the reason why the promotion with Cs
is much more efficient on the basic MgO support (Fig. 3). This conclusion is
supported by the results of recent X P S / I S S measurements [30].

F. Rosowski et al. / Applied Catalysis A: General 151 (1997) 443-460

457

3.4. Kinetics at high pressure


The reaction orders of N 2 and H 2 for alkali-promoted catalysts were found to
compensate each other almost completely in the kinetically controlled temperature regime both at atmospheric pressure and at 20 bar. Thus, an increase in total
pressure from 1 bar to 50 bar (traces A-D in Fig. 6) with a constant synthesis
gas composition of XH2:XN2= 3:1 did not lead to a significant increase in
conversion for C s N O 3 - R u / A 1 2 0 3 at low temperatures.
The analysis of the power-law kinetics therefore raises the question whether
alkali-promoted Ru catalysts can be applied under industrial high-pressure
synthesis conditions. Due to the inhibiting effect of XH2, and the accelerating
effect of XN~, it is obvious that the 3:1 ratio of XH2:XN2has to be decreased in
order to obtain higher rates of NH 3 formation. However, non-stoichiometric
feed-gas compositions are less favorable because of thermodynamic reasons.
Hence, the Xla2:XN2 ratio was varied between 5:95 and 75:25 at 50 bar using
C s 2 C O 3 - R u / M g O to find a compromise between kinetics and thermodynamics.
The resulting steady-state XNH3 as a function of temperature using a total flow
of 100 cm 3 ( S T P ) / m i n is shown in Fig. 7. The ratio of XNH:XN: = 5:95 (trace
A) leads to a significant rate of NH 3 formation at temperatures as low as 520 K.
When the temperature was higher than about 570 K, XNH~ was controlled by
thermodynamic equilibrium leading to a decrease in XNH~. The maximum
conversion was achieved at 720 K by using a ratio of XH2:XN2= 60:40 (trace G)

3.5
--.

CsNO 3 - Ru / AI20 3

40 cm3(STP) / min

._~_o2 . 5

p = 50 bar

[]

p=20

,',

p = 5 bar

p = 1 bar

o
E
"-r
z

1,5

//~D
~

bar

o.5

A
-

I~"1
600

650

I
700

750

Temperature / K
Fig. 6. Dependence of the NH 3 effluent mole fraction on the feed gas composition observed for CsNO 3 R u / A I 2 0 3 as a function of total pressure, Traces A - D (from bottom to top) were obtained at 1 bar, 5 bar, 20
bar, and 50 bar, respectively, using a total flow of 40 cm 3 ( S T P ) / m i n with XH2:XN2= 3:1.

458

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

Cs2CO 3 -

Ru / MgO

F ',

'

50 bar
1 0 0 cma(sTP)/min

H2 : N2
75:25

60 : 40

~7
O

50:50
40 : 60
30 : 70

20:80

[]

lo:9o

8 6

=73
z

'
D ~
, , ~

".

x /
,
".,

",,

C -.....
/
A/
B

~~ ~

",,

~-

"- ..

"..

"

540

570

600
630
660
Temperature / K

690

720

750

Fig. 7. Dependence of the NH 3 effluent mole fraction on the feed gas composition observed for Cs2CO 3R u / M g O at 50 bar using a total flow of 100 cm 3 (STP)/min. Traces A - H (from bottom to top) were obtained
with XHz:XN2, ratios of 5:95, 10:90, 20:80, 30:70, 40:60, 50:50, 60:40 and 75:25, respectively. The
corresponding NH 3 equilibrium mole fractions at 50 bar are displayed as dashed lines.

equal to 3:2. Correspondingly, lower values than 3:1 can be found in patents by
the M.W. Kellogg Company [31,32].
Finally, the activity of Cs2CO3-Ru/MgO under these optimized conditions
is compared in Table 6 with the kinetic data published by Nielsen [33] using a
multiply-promoted Fe catalyst (KMIR). It is evident that a few percent lower
effluent XNH3 can be obtained at about the same space velocity and temperature

Table 6
Experimental high-pressure data observed for the Fe-based KMIR catalyst [33] and the present Cs2CO 3R u / M g O catalyst. For the KMIR, the catalyst bed density was 2.5 g / c m 3, and the catalyst weight was 3.125
g resulting in a catalyst volume in the reactor of 1.25 cm 3. The space velocity (SV) is based on the inlet flow
rate Q at standard temperature and pressure (STP) and on the catalyst volume. For the CszCO 3 - R u / M g O
catalyst, the catalyst volume in the reactor was 0.142 cm 3 using 0.138 g. Both reactors used were shown to
operate as isothermal plug-flow reactors. The rate of ammonia synthesis under these conditions was not limited
by transport phenomena as shown by Nielsen [33]
T (K)

p (bar)

SV ( h - ] )

Q (cm a (STP)/min)

Xn2:XN2

XNH3 (%)

Catalyst

723
718
723
696
723
718

107
50
107
50
107
50

11600
11531
16000
18 450
47100
46123

221
25
305
40
899
100

3:1
3:2
3:1
3:2
3:1
3:2

13.9
10.2
13.2
9.7
9.8
7.8

KMIR
Cs2CO 3 - R u / M g O
KMIR
Cs2CO 3 - R u / M g O
KM1R
Cs2CO 3 - R u / M g O

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

459

conditions with, however, a pressure of 50 bar only instead of 107 bar necessary
for the Fe-based KMIR catalyst.

4. Conclusions
The preparation of Ru catalysts from Ru3(CO)12 and high-purity supports
resulted in long-term stable catalysts for NH 3 synthesis. For the reliable
assessment of the Ru particle size, the combined application of H 2 chemisorption, TEM and XRD was found to be necessary yielding about 2 nm as average
Ru particle size on MgO and A1203. Impregnating R u / M g O with aqueous
solutions of CsNO 3 was observed to cause a significant increase of the Ru
particle size contrary to Ru/A1203 with essentially unchanged Ru particle sizes.
The non-aqueous impregnation of R u / M g O with Cs2CO 3 in ethanol was found
to be less detrimental to the Ru metal dispersion resulting in the most active
catalyst with an initial TOF at 588 K of about 10 -2 s -1.
On R u / M g O , promotion with Cs was found to result in a higher increase in
TOF than on Ru/A1203 presumably due to a stronger degree of interaction
between the Cs promoter and the acidic A1203 support. The determination of
the apparent activation energy at constant xN~ 3 provided additional evidence
that the higher TOF of R u / M g O compared with the TOF of Ru/A1203
originates from the promotion by the alkaline earth support. Unpromoted
catalysts are hardly active for NH 3 synthesis, whereas the catalytic activity of
efficiently Cs-promoted catalysts exceeds that of a multiply-promoted Fe catalyst at atmospheric pressure.
The rate of NH 3 formation over all catalysts was found to be inhibited by H 2.
The power-law rate expressions derived at atmospheric pressure and at 20 bar
provided three criteria for the effect of alkali promotion in addition to the
increase in TOF by almost two orders of magnitude: First, the reaction order of
NH 3 is changed to essentially zero, second, the reaction order of H 2 is more
negative, and third, the apparent activation energy determined at constant flow is
increased.
These criteria can be used as tools to elucidate the degree of interaction of the
alkali promoter with the Ru metal particles. According to the power-law
kinetics, pretreating A1203 with K2CO 3 prior to impregnation with Ru3(CO)12
resulted in a promoted catalyst which changed to a less promoted state when
heating up to 773 K in synthesis gas.
Due to the inhibiting effect of H 2, a lower XH2:XN2 ratio has to be used under
high pressure NH 3 synthesis conditions. For C s 2 C O a - R u / M g O at 50 bar, the
optimum ratio was found to be around 3:2, leading to an effluent XNH3 which
was just a few percent lower than that of a multiply-promoted Fe catalyst
operated at 107 bar with roughly the same space velocity at nearly the same
temperature. Hence, the alkali-promoted Ru catalysts can indeed be developed
into a low-pressure alternative to the conventional Fe catalysts.

460

F. Rosowski et al. /Applied Catalysis A: General 151 (1997) 443-460

Acknowledgements
The authors benefited from discussions with K.-I. Aika, B. Fastrup, S.R.
Tennison and R. Schl~Sgl and are grateful to W. Mahdi, J. Schiitze, M.
Wesemann, U. Klengler, and N. Pffinder for technical assistance, and to Haldor
TopsCe A / S for supplying the iron catalyst.

References
[1] S.R. Tennison, in: Catalytic Ammonia Synthesis, 1st edn., (Ed.) J.R. Jennings (Plenum, New York, 1991)
p. 303.
[2] T.A. Czuppon, S.A. Knez, R.V. Schneider III and G. Worobets, Chem. Engineering, March 1993,
presented at the 1993 AICHE Ammonia Safety Symposium, Sept. 1993, Orlando, Florida 100, 3 (1993)
19.
[3] K. Lu and B.J. Tatarchuk, J. Catal., 106 (1987) 166.
[4] W.K. Shiflett and J.A. Dumesic, Ind. Eng. Chem. Fundam., 20 (1981) 246.
[5] S. Murata and K.-I. Aika, Appl. Catal. A, 82 (1992) 1.
[6] P. Moggi, G. Predieri, G. Albanesi, S, Papadopulos and E. Sappa, Appl. Catal., 53 (1989) LI.
[7] S. Murata and K.-I. Aika, J. Catal., 136 (1992) 110.
[8] K. Aika, T. Takano and S. Murata, J. Catal., 136 (1992) 126.
[9] J.U. Nwalor and J.G. Goodwin Jr., Top. Catal., 1 (1994) 285.
[10] A.K. Datye, A.D. Logan and N.J. Long, J. Catal., 109 (1988) 76.
[11] R. Scholder and M. Mansmann, Z. Anorg. Allg. Chem., 321 (1963) 246.
[12] G. Busca, C. Cristiani, P. Forzatti and G. Groppi, Catal. Lett., 31 (1995) 65.
[13] P. Moggi, G. Albanesi, G. Predieri and G. Spoto, Appl. Catal. A, 123 (1995) 145.
[14] P.R. Holzman, W.K. Shiflett and J.A. Dumesic, J. Catal., 62 (1980) 167.
[15] K. Aika, M. Kumasaka, T. Oma, O. Kato, H. Matsuda, N. Watanabe, K. Yamazaki, A. Ozaki and T.
Onishi, Appl. Catal., 28 (1986) 57.
[16] H. Baris, M. Glinski, J. Kijenski, A. Wokaun and A. Baiker, Appl. Catal., 28 (1986) 295.
[17] Z. Kowalczyk, S. Jodzis and J. Sentek, Appl. Catal. A: General, 138 (1996) 83.
[18] L.M. Aparicio and J.A. Dumesic, Top. Catal., 1 (1994) 233.
[19] O. Hinrichsen, F. Rosowski, M. Muhler and G. Ertl, Chem. Engng Sci., Proc. of the 14th Int. Symposium
on Chem. React. Engineering, 51 (1996) 1683.
[20] O. Hinrichsen, F. Rosowski, A. Hornung, M. Muhler and G. Ertl, J. Catal., (1996) in press.
[21] J.A. Dumesic, D.F. Rudd, L.M. Aparicio, J.E. Rekoske and A.A. Trevifio, The Microkinetics of
Heterogeneous Catalysis (ACS Professional Reference Book, Washington, DC, 1993).
[22] H. Kntizinger, Y. Zhao, B. Tesche, R. Barth, R. Epstein, B.C. Gates and J.P. Scott, Faraday Discuss.
Chem. Soc., 72 (1982) 53.
[23] K. Eross-Kiss, S. Gal, M. Szerenyi and E. Pungor, Periodica Polytechnica, 20 (1976) 13.
[24] R.A. Dalla Betta, J. Catal., 34 (1974) 57.
[25] B. Fastrup and H.N. Nielsen, Catal. Lett., 14 (1992) 233.
[26] K. Aika, K. Shimazaki, Y. Hattori, A. Ohya, S. Ohshima, K. Shirota and A. Ozaki, J. Catal., 92 (1985)
296.
[27] A. Ozaki, H. Taylor and M. Boudart, Proc. Roy. Soc. A, 258 (1960) 47.
[28] O. Hinrichsen, F. Rosowski, A. Hornung, M. Muhler and G. Ertl, J. Catal., (1996) in press.
[29] F. Rosowski, O. Hinrichsen, M. Muhler and G. Ertl, Catal. Lett., 36 (1996) 229.
[30] M. Muhler, F. Rosowski, U. Wild, Z. Kowalczyk and G. Ertl, J. Phys. Chem., (1996) in press.
[31] P.J. Shires, J.R. Cassata, B.G. Mandelik and C.P. Van Dijk, U.S. Patent 4,479,925 (1984) Oct. 30.
[32] G.S. Benner, J.R. Le Blanc, J.M. Lee, H.P. Leftin, P.J. Shires and C.P. Van Dijk, U.S. Patent 4,568,532
(1986) Feb. 4.
[33] A. Nielsen, An Investigation of Promoted iron Catalysts for the Synthesis of Ammonia, 3rd edn.,
(Gjellerup, Copenhagen, 1968).

Das könnte Ihnen auch gefallen