Effect of the support on the nature of metal-promoter interactions in Ru-Cs+/MgO and Ru-Cs+-Al2O3 catalysts for ammonia synthesis

Article abstract of DOI:10.1007/s10975-005-0153-1

The Ru-Cs⁺/MgO and Ru-Cs⁺/γ-Al ₂O₃ catalysts, which were prepared by an impregnation method using RuOHCl₃ and Cs₂CO₃ as precursor compounds and reduced with H₂ at 450°C, are characterized by X-ray diffraction, high-resolution transmission electron microscopy (with X-ray microanalysis), and X-ray photoelectron spectroscopy (XPS). The Cs ⁺/MgO(Al₂O₃) systems, Ru-Cs⁺ black, and model systems prepared by cesium sputtering onto polycrystalline ruthenium foil are studied as reference samples. It is found that, in the Ru-Cs ⁺/MgO sample, cesium is present as a Cs_{2 + x}O cesium suboxide, which weakly interacts with the support, localized on the surface of Ru particles or near them. In the case of Ru-Cs⁺/γ-Al ₂O₃, cesium occurs as a species that is tightly bound to the support; this is likely surface cesium aluminate, which prevents promoter migration to Ru particles. The Ru-Cs⁺/MgO sample exhibits a considerable shift of the Ru3d line in the XPS spectra toward lower binding energies, as compared to the bulk metal. It is hypothesized that this shift is due to a decrease in the electron work function from the surface of ruthenium because of the polarizing effect of Cs⁺ ions in contact with Ru particles. Based on the experimental results, the great difference between the catalytic activities of the Ru-Cs⁺/MgO and Ru-Cs⁺/γ- Al₂O₃ systems in ammonia synthesis at 250-400°C and atmospheric pressure is explained.

Full text of DOI:10.1007/s10975-005-0153-1

Kinetics and Catalysis, Vol. 46, No. 6, 2005, pp. 891–899. Translated from Kinetika i Kataliz, Vol. 46, No. 6, 2005, pp. 940–948.
Original Russian Text Copyright © 2005 by Larichev, B. Moroz, E. Moroz, Zaikovskii, Yunusov, Kalyuzhnaya, Shur, Bukhtiyarov.
Effect of the Support on the Nature of Metal–Promoter
+
+
Interactions in Ru–Cs /MgO and Ru–Cs –Al O Catalysts
2
3
for Ammonia Synthesis
Yu. V. Larichev*, B. L. Moroz*, E. M. Moroz*, V. I. Zaikovskii*, S. M. Yunusov**,
E. S. Kalyuzhnaya**, V. B. Shur**, and V. I. Bukhtiyarov*
Boreskov Institute of Catalysis, Siberian Division, Russian Academy of Sciences, Novosibirsk, 630090 Russia
*
*
* Nesmeyanov Institute of Organoelement Compounds, Russian Academy of Sciences, Moscow, 119991 Russia
Received January 26, 2005
+
+
Abstract—The Ru–Cs /MgO and Ru–Cs /γ-Al O catalysts, which were prepared by an impregnation method
2
3
using RuOHCl and Cs CO as precursor compounds and reduced with ç at 450°ë, are characterized by X-ray
3
2
3
2
diffraction, high-resolution transmission electron microscopy (with X-ray microanalysis), and X-ray photoelec-
+
+
tron spectroscopy (XPS). The Cs /MgO(Al O ) systems, Ru–Cs black, and model systems prepared by cesium
2
3
sputtering onto polycrystalline ruthenium foil are studied as reference samples. It is found that, in the Ru–
+
Cs /MgO sample, cesium is present as a ës O cesium suboxide, which weakly interacts with the support,
2
+ x
+
localized on the surface of Ru particles or near them. In the case of Ru–Cs /γ-Al O , cesium occurs as a species
that is tightly bound to the support; this is likely surface cesium aluminate, which prevents promoter migration
to Ru particles. The Ru–Cs /MgO sample exhibits a considerable shift of the Ru3d line in the XPS spectra
toward lower binding energies, as compared to the bulk metal. It is hypothesized that this shift is due to a
2
3
+
+
decrease in the electron work function from the surface of ruthenium because of the polarizing effect of Cs
ions in contact with Ru particles. Based on the experimental results, the great difference between the catalytic
+
+
activities of the Ru–Cs /MgO and Ru–Cs /γ-Al O systems in ammonia synthesis at 250–400°ë and atmo-
2
3
spheric pressure is explained.
INTRODUCTION
7]. Although a negative shift of XPS spectra can in fact
suggest an excessive electron density on the supported
metal because of the interaction of the support with the
active component, other reasons, including instrumen-
tal factors, can also be responsible for this effect. More-
over, this explanation is inconsistent with the low activ-
ity of samples based on aluminum oxide promoted with
alkalis, although this modified support is highly basic
Ruthenium catalysts promoted with alkali metal
compounds are highly active in ammonia synthesis at
low pressures (0.1–3.0 MPa) and temperatures (250–
3
50°ë) [1]. The most active catalysts were prepared
with the use of magnesium oxide as a support [2]. The
replacement of MgO by alumina, a more practical sub-
stance, or various carbon materials noticeably
decreased the low-temperature activity of ruthenium in
the synthesis of ammonia [3, 4]. It was found that cata-
lytic activity is related to the basicity of the support: the
more basic the support, the higher the activity of the
catalyst in the reaction of ammonia synthesis [4, 5].
Aika et al. [4] explained this correlation by electron-
density transfer from the basic support (MgO) to ruthe-
nium particles. Because of the appearance of a negative
[5, 8].
It is also well known that the activity of ruthenium
catalysts for ammonia synthesis depends strongly on
the basicity of the promoter and increases in the order
Na < K < Rb < Cs [1, 9]. However, there is no direct
experimental data on the electron effect of the pro-
moter. The occurrence of an interaction between the
promoter and the support and the nature of this interac-
tion are not understood, although it is obvious that such
an interaction can also affect the activity of catalysts.
charge on metal particles, the dissociation of N mole-
cules is facilitated. This dissociation is the rate-limiting
step of ammonia synthesis. This hypothesis was sup-
2
The aim of this work was to study the chemical state
+
ported experimentally based on the decreased binding of a promoter (a Cs compound) and the interaction of
energies (E ) of electrons in the Ru3d core level in this promoter with a catalytically active metal (Ru) and
b
+
+
Ru/MgO samples (279.2–279.5 eV), as compared with a support in the Ru–Cs /MgO and Ru–Cs /Al O sys-
2
3
that of the bulk metal. In contrast to this, normal values of tems with the use of X-ray diffraction (XRD), transmis-
the binding energy of ruthenium metal (280.2–280.5 eV) sion electron microscopy (TEM), and X-ray photoelec-
were found in Ru/Al O (SiO ) and Ru/C samples [4, 6, tron spectroscopy (XPS).
2
3
2
0
023-1584/05/4606-0891 © 2005 MAIK “Nauka /Interperiodica”

8
92
LARICHEV et al.
EXPERIMENTAL
MgO and γ-Al O with particle sizes of 0.25–
washed with distilled water and dried. Thereafter, an
aqueous solution of cesium carbonate, which was taken
in a fourfold excess by volume with respect to black,
was added. The mixture was evaporated to dryness at
2
3
2
0
.50 mm and specific surface areas of 200 and 220 m /g,
respectively (determined by the low-temperature
adsorption of nitrogen), were used as catalyst supports.
Before sample preparation, the supports were evacu-
ated at 450°ë and 0.02 Torr for 2–3 h.
5
0°ë with continuous stirring. The dry residue was
evacuated at 20°ë for 2 h and at 60°ë for 6 h to a resid-
ual pressure of 0.02 Torr and treated with hydrogen
under conditions analogous to those described above.
The resulting sample contained three weight parts of
Ru and one weight part of Cs.
Ru/MgO and Ru/Al O samples were prepared by
2
3
the incipient wetness impregnation of the supports with
a solution of RuOHCl in acetone (of high-purity
3
+
grade). Immediately after impregnation, the solvent
was removed by purging the samples with air at room
temperature up to a loose state. Next, the samples were
evacuated at 20°ë for 2 h and at 60°ë for 2 h to a resid-
ual pressure of 0.02 Torr. Because of the low solubility
of the ruthenium compound in acetone, the impregna-
tion procedure was additionally performed two times.
After the final cycle of impregnation, the samples were
evacuated at 60°ë for 6 h and then reduced in a flow of
The catalytic activity of Ru/MgO and Ru–Cs /MgO
samples in ammonia synthesis was studied in a glass
fixed-bed flow reactor at 200–400°ë and atmospheric
pressure with the use of a stoichiometric mixture of
N + H (the rate of gas supply to the reactor was 10 l/h).
2
2
The gas mixture was purified in order to remove trace
oxygen, moisture, and other impurities. A weighed por-
tion of a catalyst was placed in the reactor away from
air in a flow of the N + H mixture and heated to a spec-
2
2
ç (80 ml/min) in a glass tube reactor. The reactor tem-
2
ified temperature. The ammonia content of a gas flow at
the reactor outlet was determined by measuring
changes in the concentration of a dilute aqueous H SO
perature was increased stepwise to 450°ë for 2–2.5 h;
next, the samples were reduced for 6 h at this tempera-
ture. After the reduction, the samples were cooled to
2
4
solution through which the gas was passed. Tests at the
specified temperature were continued until the catalyst
reached a steady state; thereafter, the reaction tempera-
room temperature in a flow of H , removed from the
2
reactor, and kept under argon.
The cesium promoter was introduced in accordance ture was increased by 50°ë and the steady-state con-
with a published procedure [5], which was modified. A centration of NH in the reactor outlet gas was deter-
3
weighed portion of Cs CO was dissolved in absolute mined once again (the steady-state concentration of
2
3
ethanol at 70°ë, and the solution was added to a NH was calculated by averaging the results of 6–
3
reduced Ru/MgO or Ru/Al O sample thermostated at 10 measurements at a constant temperature).
2
3
the specified temperature (the solution volume added
was approximately equal to the pore volume of the sup-
port). The solvent was distilled off in a flow of argon,
and the sample was evacuated at 20°ë for 2 h and at
High-resolution TEM micrographs were obtained
on a JEM-2010 instrument with a maximum grid reso-
lution of 0.14 nm and an accelerating voltage of 200 kV.
The sample was removed from an ampule filled with
argon immediately before taking the micrographs, rap-
idly ground in air, and suspended in hexane. A drop of
the suspension was applied to a porous carbon film
fixed on copper gauze; thereafter, the solvent was
allowed to evaporate. To obtain the particle-size distri-
bution of Ru (TEM bar graphs) and to determine the
statistically valid average particle size for each particu-
lar sample, the linear dimensions of no less than
6
0°ë for 2 h to a residual pressure of 0.02 Torr. Because
Cs CO is sparingly soluble in absolute ethanol even on
2
3
heating, the impregnation procedure was additionally
repeated three times. After the completion of impregna-
tion, the samples were evacuated at 60°ë for 6 h and
treated with hydrogen. The procedure and conditions of
hydrogen treatment were analogous to those described
above.
According to X-ray fluorescence analysis data, the
concentrations of ruthenium and cesium in
3
00 particles were measured. The elemental composi-
+
+
tion of particles visible to the microscope was deter-
mined by energy dispersive X-ray (EDX) analysis
using an EDAX energy dispersive spectrometer with an
Si(Li) detector with a resolution of 130 eV.
Ru−Cs /MgO and Ru–Cs /Al O samples were the
2
3
same and equal to 4.5 and 4.0 wt % for Ru and Cs,
respectively (the Cs : Ru atomic ratio was 0.8).
+
+
The reference samples of Cs /MgO and Cs /Al O
2
3
were prepared in accordance with a procedure analo-
gous to that described above for the introduction of a
The XRD analysis of samples was performed on an
HZG-4 X-ray diffractometer (CuK radiation with a
α
cesium promoter into Ru/MgO and Ru/Al O samples.
2
3
diffracted beam graphite monochromator) with a scan
step of 0.05°. The size of the coherent-scattering
regions of ruthenium particles was determined from the
broadening of diffraction peaks and calculated with
According to X-ray fluorescence analysis data, the con-
+
+
centrations of cesium in the Cs /MgO and Cs /Al O
2
3
samples were 3.5 and 4.8 wt %, respectively.
+
Ruthenium black promoted with Cs ions was pre- consideration for the instrumental line width using the
pared as described below. The reduction of RuOHCl3 Selyakov–Scherrer formula. Before recording XRD
with formaldehyde in an aqueous alkali solution at patterns, the sample was rapidly ground in air and
8
0°ë [10] afforded ruthenium black, which was wrapped in an organic polymer film to protect the sam-
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

EFFECT OF THE SUPPORT ON THE NATURE OF METAL–PROMOTER INTERACTIONS
893
+
ple from contact with air in the course of measure- Table 1. Results of the catalytic tests of Ru–Cs /MgO in
ammonia synthesis
ments.
The XPS spectra were measured on a VG
Temperature, °C
[NH ] , vol %
[NH ] , vol %
3
eq
3 st
ESCALAB HP electron spectrometer using unmono-
2
3
3
50
00
50
5.40
2.17
0.86
0.44
0.014
0.15
0.67
0.43
chromatized AlK radiation (E = 1486.6 eV; power of
α
hν
2
00 W). The scale of binding energies (E ) was precal-
b
ibrated with reference to the peak positions of Au
Au 4f7/2 (84.0 eV) and Cu 2p3/2 (932.6 eV) core levels
from polycrystalline gold and copper foils, respec-
tively. We used the internal standard method for the cor-
rect calibration of photoelectron lines, in particular, for
the elimination of a charging effect, which occurs in a
study of nonconducting samples. The Mg 2s line with
E = 88.1 eV and the Al 2p line with E = 74.5 eV
400
Note: [NH3]eq is the equilibrium concentration according to pub-
lished data [12]; [NH3]st is the steady-state concentration in
a gas at the reactor outlet. Atmospheric pressure; H /N =
2
2
3
: 1; catalyst weight, 3.39 g.
+
b
b
Table 2. Ruthenium particle sizes in supported Ru–Cs sys-
served as internal standards for samples containing
MgO and Al O , respectively. The value of charging
+
tems and Ru–Cs black according to TEM and XRD data
2
3
was determined as the difference between the measured
Coherent-scatter-
ing region, nm
Sample
〈d〉, nm
dvs, nm
and tabulated values of E ; the XPS lines of the other
b
elements were then shifted by this value. The ratios
between the surface atomic concentrations of elements
in the samples were determined from the integrated
intensities of XPS lines corrected for corresponding
atomic sensitivity factors [11]. Before measuring the
spectra, the samples of supported systems were pressed
into nickel gauze and placed in the preparation chamber
of the spectrometer (in the course of this procedure, the
samples were in air for no longer than 5 min). In the
spectrometer, the samples were additionally treated
with hydrogen under static conditions at 350°ë and a
pressure of 0.1 MPa for 1 h. The samples containing a
+
Ru–Cs /MgO
3.0
2.1
36
9.5
8.3
63
9.0
4.0
47
+
Ru–Cs /Al O
2
3
+
Ru–Cs black
+
+
9
.5 and 8.3 nm for Ru–Cs /MgO and Ru–Cs /Al O ,
2
3
respectively). Ruthenium black promoted with a
cesium compound consisted of much coarser particles
about several tens of nanometers in size.
Figure 1 shows the high-resolution TEM images of
+
+
Ru–Cs /MgO and Ru–Cs /Al O samples. Periodicity,
2
3
which corresponds to Ru–Ru interplanar spacing
d100 = 0.234 nm, d101 = 0.206 nm, etc.), can be clearly
thin film of CsO on the surface of polycrystalline
x
(
ruthenium were prepared immediately in the prepara-
tion chamber of the XPS spectrometer either in a vac-
seen in ruthenium metal particles. A thin layer of a dis-
ordered structure <1 nm in thickness was detected on
ruthenium metal particles in the Ru–Cs /MgO sample
Fig. 1a). This layer was also present on the surface of
–
7
–2
uum (<5 × 10 Pa) or in flowing oxygen (1 × 10 Pa).
+
(
RESULTS
MgO but only in a narrow region (5–10 nm in diameter)
near a metal particle. At the same time, the presence of
+
The catalytic activity of Ru–Cs /MgO and Ru–
+
this layer on the metal and support surfaces was not
Cs /Al O samples was tested in ammonia synthesis at
2
3
+
detected in the Ru–Cs /Al O sample (Fig. 1b). Figure 2
2
00–400°ë and an atmospheric pressure of the 1 : 3
2
3
demonstrates the EDX spectra of surface regions adja-
mixture of N and H . Table 1 summarizes the steady-
2
2
+
cent to ruthenium particles in Ru–Cs /MgO and
state concentrations of NH in a gas flow at the reactor
3
+
Ru−Cs /Al O samples. The spectrum of the
outlet, which were measured in the presence of the Ru–
2
3
+
+
Ru−Cs /MgO sample exhibited a maximum due to
Cs /MgO catalyst at various temperatures. The forma-
cesium along with the signal of ruthenium, whereas a
tion of NH on this catalyst was observed from 250°ë;
3
+
signal of cesium was not detected in the Ru–Cs /Al O
the concentrations of NH in the gas at 350 and 400°C
2
3
3
sample. An analogous pattern was observed in many
surface regions of the test samples. This result allowed
us to assume that, in the Ru–Cs /MgO sample, the layer
that covers ruthenium particles and the support surface
near these particles contained cesium.
were 79 and 100% of an equilibrium concentration,
+
respectively. In the presence of the Ru–Cs /Al O sam-
2
3
+
ple, the concentration of NH in the exit gas was below
3
the determination limit (<0.005 vol %) throughout tem-
perature range examined.
Table 2 summarizes the average linear (〈d 〉) and vol-
The XRD patterns from the supported systems exhib-
l
ume–surface (〈d 〉) particle sizes of ruthenium in Ru– ited only reflections due to ruthenium metal and the cor-
vs
+
+
+
Cs /MgO and Ru–Cs /Al O samples and Ru–Cs
responding support (MgO or Al O ). Table 2 summa-
2 3
2
3
black, as calculated from TEM bar graphs. In the sup- rizes the sizes of the coherent-scattering regions of ruthe-
+
+
ported systems, the major amount of ruthenium parti- nium particles in the Ru–Cs /MgO and Ru–Cs /Al O
2
3
+
cles was small (〈d 〉 ~2–3 nm); the fractions of coarser samples and Ru–Cs black. The size of the coherent-
l
+
particles (of size 5–12 nm), which can also be seen in scattering region found for the Ru–Cs /MgO sample is
TEM micrographs, were approximately equal (〈d 〉 = consistent with the value of 〈d 〉 for ruthenium particles
vs
vs
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

8
94
LARICHEV et al.
Intensity
RuL
CsOx
Ru
CsL
1
2
MgO
5
nm
(
‡)
2
.0
2.5
3.0
3.5
4.0
4.5
5.0
E, keV
Fig. 2. Secondary X-ray (EDX) spectra from (1) Ru–
+
+
Cs /MgO and (2) Ru–Cs /Al O sample surface regions
2
3
adjacent to ruthenium particles, which are shown in Fig. 1.
Al O
2
3
tern exhibited only a weak broadened maximum at
1
4.5°, which can correspond to the group of reflections
due to the stoichiometric cesium suboxide Cs O
4
[JCPDS No. 28-0327]. After keeping the sample in air
Ru
for two weeks, a group of reflections related to cesium
hydrocarbonate CsHCO3 [JCPDS No. 26-0376]
appeared in the diffraction pattern. It is likely that
cesium occurred as a thin film of cesium suboxide in
the initial sample; in contact with air, this cesium sub-
oxide was converted into the hydrocarbonate.
5
nm
(
b)
Figure 5 shows the Ru 3d XPS spectra of the sup-
ported samples after treatment with hydrogen. The val-
Fig. 1. Electron micrographs of the surfaces of (a)
Ru−Cs /MgO and (b) Ru–Cs /Al O samples. Magnifica-
+
+
ues of Ru 3d E for Ru–Cs /MgO, Ru–Cs /Al O ,
+
+
5/2
b
2
3
2
3
+
and Ru–Cs black are equal to 279.0, 280.2, and
80.1 eV, respectively. The value of Ru 3d Eb for the
Ru–Cs /MgO sample is lower than the value of Ru 3d
tion of 800000.
2
5
/2
+
5
/2
+
Eb measured in polycrystalline ruthenium foil
280.2 eV) by 1.2 eV. Previously, Muhler et al. [5]
found from TEM data. The Ru–Cs /Al O sample
2
3
(
exhibited a considerable discrepancy between the val-
observed a shift of lines in the Ru 3d5/2 spectra of
ues of the coherent-scattering regions and 〈d 〉, which
vs
Ru/MgO systems toward lower values of E with refer-
may be due to a great number of defects in ruthenium
particles.
b
ence to the corresponding line in the spectrum of bulk
ruthenium and interpreted this shift as a result of elec-
tron-density transfer from the support to supported
ruthenium particles.
Figure 3 compares the XRD patterns from the
+
Ru−Cs /Al O sample and the original support. The
2
3
supported system exhibited an increase in the signal
However, note that the shift of an XPS line toward
intensity at 2θ = 25°–40°, as compared with Al O . This
2
3
lower E with reference to the line of bulk metal can
+
b
is also characteristic of Cs /Al O . It is believed that
2
3
result from not only the appearance of a negative charge
on the atoms of supported metal particles but also the
occurrence of a relaxation effect and differential charg-
ing, which can occur in systems that consist of phases
with different conductions (in our case, support and
metal phases) [13–15]. We calculated corrections to the
this change was due to the presence of X-ray amor-
+
phous cesium compounds in the Ru–Cs /Al O and
2
3
+
Cs /Al O samples. In particular, the reflections of the
2
3
cesium aluminate Cs Al O [JCPDS No. 23-0882] were
2
2
4
observed in this region of 2θ.
Figure 4 shows the XRD pattern of a freshly reduced values of Ru 3d5/2 Eb in accordance with a published
+
sample of Ru–Cs black. Although the cesium content procedure [13, 15] in order to take into account relax-
of this sample was high (25 wt %), the diffraction pat- ation and differential charging effects. With consider-
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

EFFECT OF THE SUPPORT ON THE NATURE OF METAL–PROMOTER INTERACTIONS
895
Intensity
Intensity
Al O
2
3
+
Ru–Cs /Al O
2
3
#
Cs O
4
+
CsHCO3
+
+
+
+
+
+
2
*
#
*
1
Cs Al O
4
2
2
2
0
30
40
50
5
10
15
20
25
30
35
θ, deg
2
θ, deg
2
+
Fig. 3. Diffraction patterns of the γ-Al O support and the
Fig. 4. Diffraction patterns of Ru–Cs black (1) after reduc-
tion with hydrogen at 450°C and (2) after keeping in air for
two weeks. Reflections due to an organic polymer film in
which the reduced sample was wrapped to protect it from
oxidation with air are marked with asterisks.
2
3
+
Ru–Cs /Al O catalyst.
2
3
ation for these corrections, the values of Ru 3d Eb for
5
/2
+
+
the Ru–Cs /MgO and Cs /Al O samples are 279.7 and
2
3
2
80.6 eV, respectively. Thus, even with consideration
mation of a cesium suboxide like Cs + xO. The treat-
2
for the relaxation and differential charging effects, the
+
ment of the Ru−Cs /Al O sample with hydrogen under
the same conditions did not cause changes in the spec-
trum.
2
3
value of Ru 3d Eb in the Ru 3d spectrum of the
5
/2
+
Ru−Cs /MgO sample remained lower than that in the
spectrum of the bulk metal (∆E = –0.5 eV).
b
+
+
The values of E for Cs /MgO and Cs /Al O sam-
b
2
3
Figure 6 shows the Cs 3d spectra measured in
5
/2
+
+
ples after contact with air were found to equal 724.9
and 725.1 eV, respectively. These values remained
practically unchanged after the treatment of the sam-
Ru−Cs /MgO and Cs /Al O samples both immedi-
2
3
ately after loading into the photoelectron spectrometer
in air and after treatment with hydrogen in the spec-
trometer. Table 3 summarizes the values of Cs 3d5/2 Eb
for various cesium compounds with oxygen. These val-
ues, which were taken from [16–18], can be used for
Intensity
Ru 3d3/2 + C 1s
Ru 3d5/2
the assignment of the values of E in the experimental
b
2
79.0 eV
spectra. Lines in the Cs 3d5/2 spectra measured after
exposure of the samples to air exhibited maximums at
+
7
24.8 and 724.9 eV for Ru–Cs /MgO and
+
Ru−Cs /Al O , respectively. These values of Cs 3d Eb
1
2
2
3
5/2
lie between the values found for the individual com-
pounds Cs O (E = 724.5 eV) and Cs O (E = 725.2 eV).
2
2
b
2
b
Because it is well known [16, 17] that the cesium oxide
Cs O is easily converted into the peroxide Cs O even in
2
2
2
the presence of trace oxygen at a very short exposure
–
6
(
>(5−10) × 10 Torr s), it is believed that cesium
280.2 eV
occurred as peroxide in the tested samples at least after
keeping the samples in air. The treatment of the
+
Ru−Cs /MgO sample with hydrogen in the spectrome-
ter chamber at 350°C and 0.1 MPa resulted in a shift of
the line maximum to E = 725.5 eV. Because cesium
b
and some other metals (such as Ag, Cd, and Ba) exhib-
ited inverse chemical shifts when the value of Eb
increased rather than decreased as the oxidation state of
metal decreased (see Table 3), we can state that the
observed shift of the Cs 3d5/2 line toward higher values
276
280
284
288
292
E , eV
b
+
Fig. 5. Ru 3d
spectra of (1) Ru–Cs /MgO and (2)
5
/2
+
Ru−Cs /Al O samples after hydrogen treatment within the
2
3
spectrometer. The spectra include a C 1s line at 285.0 eV
+
of E was due to the reduction of Cs ions with the for-
from carbon impurities.
b
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

8
96
LARICHEV et al.
+
+
+
+
Cs /Al O
Ru–Cs /MgO
Ru–Cs /Al O
Cs /MgO
2
3
2
3
+
+
0
0
0
0
.045
.040
.035
.030
Ru–Cs /MgO
Ru–Cs /Al O
2
3
7
24.8 eV 725.5 eV
724.9 eV
1
2
7
20 724 728 732
Eb, eV
720 724 728 732
0.025
0
200 400 600
0
200 400 600
Temperature, ë°
+
+
Fig. 6. Cs 3d spectra of Ru–Cs /MgO and Ru–Cs /Al O
Fig. 7. The temperature dependence of surface Cs/Mg(Al)
5
/2
2
3
+
samples (1) before and (2) after hydrogen treatment within
atomic
ratios
in
Ru–Cs /MgO(Al O )
and
2
3
+
the spectrometer.
Cs /MgO(Al O ) samples.
2
3
+
+
ples with hydrogen in the spectrometer chamber (E = the Ru–Cs /MgO and Cs /MgO samples exhibited a
b
7
25.0 and 725.1 eV, respectively).
considerable decrease in the Cs/Mg ratio, whereas the
Cs/Al ratio also changed only slightly with temperature
The line maximum in the Cs 3d spectrum of
5
/2
+
+
in the case of Ru–Cs /Al
2
O
3
and Ru–Al
2
O . The exper-
3
Ru−Cs black measured after exposure of the sample to
imental data can be qualitatively interpreted so that the
cesium promoter possessed certain mobility on the sur-
face of MgO, whereas this mobility was not observed in
the systems based on Al O ; it is believed that, in this
air was observed at E = 725.0 eV; after the treatment
b
of the sample with hydrogen in the spectrometer, the
line shifted to 725.5 eV. Thus, after both oxidative and
reductive treatments, the chemical state of cesium in
2
3
+
case, the promoter is strongly bound to the support sur-
face.
Ru–Cs black was analogous to the state of cesium in
+
the Ru–Cs /MgO sample after corresponding treat-
ments.
Table 4 summarizes the values of Cs 3d5/2 and O 1s
Eb and the surface Cs/O atomic ratios for samples pre-
pared by cesium sputtering onto the surface of poly-
crystalline ruthenium foil in a vacuum or in a flow of
Figure 7 illustrates the results of experiments in the
course of which the Cs 3d and Mg 2s (or Al 2s) spec-
5
/2
+
+
tra of Ru–Cs /MgO(Al O ) and Cs /MgO(Al O ) sam-
2
3
2
3
oxygen. In this case, the values of O 1s E , which are
b
ples were measured as the temperature was increased
characteristic of various cesium compounds with oxy-
gen, can be used for the identification of surface com-
pounds in addition to the values of Cs 3d5/2 Eb. A com-
from room temperature to 620°C at a constant rate of
0
.2 K/s. The Cs/Mg(Al) atomic ratios depending on
temperature were determined from the experimental
spectra. Figure 7 shows that, in all cases, the values of
Cs/Mg(Al) changed only slightly with temperature at
T ≤ 300°ë. As the temperature was further increased,
parison of the values of O 1s E obtained in this work
b
with published data [16–18], which are given in
Table 3, allowed us to conclude that, upon vacuum
sputtering, the surface of ruthenium foil was covered
with a layer of cesium suboxide ës2 + xO, which was
formed by the interaction of cesium metal vapor with
trace oxygen in the spectrometer chamber. Cesium per-
oxide ës O was formed upon cesium sputtering in a
Table 3. Cs 3d and O 1s binding energies (E ) for various
5
/2
b
cesium compounds
2
2
Eb, eV
flow of é . The fact that the surface Cs/O atomic ratio
found for ruthenium foil coated with a layer of Cs2 + x
2
Compound
Csmetal
O
Cs 3d5/2
O 1s
was lower than 2 (Cs/O = 1.9) can be due to the partial
oxidation of cesium suboxide with trace oxygen that
was present in the spectrometer chamber. In the case of
726.1
725.5
725.0
724.5
724.2
724.1
–
Cs2 + xO
531.0
528.2
530.6
532.9
–
Cs O
2
ruthenium foil coated with a layer of Cs O, the devia-
2
Cs O
2
2
tion of the Cs/O ratio from a stoichiometric value
(Cs/O = 0.89 in place of 1) can be explained by the for-
mation of an amount of the superoxide Cs O . The O 1s
Cs O
2
4
CsOH
2
4
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

EFFECT OF THE SUPPORT ON THE NATURE OF METAL–PROMOTER INTERACTIONS
897
Table 4. Cs 3d5 and O 1s binding energies (E ) and ratios between the surface atomic concentrations of cesium and oxygen
/2
b
for model systems prepared by cesium sputtering onto the surface of polycrystalline ruthenium foil
Eb, eV
Sputtering procedure
Cs/O
Assignment
Cs 3d5/2
O 1s
In a vacuum (<5 × 10–7 Pa)
725.8
725.0
531.5
530.5
1.90
0.89
Cs2 + xO/Ru
Cs O /Ru
–
2
In a flow of O (1 × 10 Pa)
2
2
2
spectrum of the Cs O /Ru-foil sample exhibited a com- Ru/MgO free of an alkaline promoter. In this case, it
2
2
ponent with E = 533.5 eV, which is usually attributed was found that this shift was only due to the effect of
b
to Cs O [18], in addition to the main signal at E = differential charging of the sample surface because of
2
4
b
5
30.5 eV.
Thus, the results of an XPS study of model samples
different conductivities of MgO and supported ruthe-
+
nium metal phases. In the case of Ru–Cs /MgO, after
taking into account the effect of differential charging
and the effect of relaxation [13], the negative shift of
the Ru 3d5/2 line with reference to the corresponding
line in the spectrum of bulk metal was retained,
although the value of this shift somewhat decreased
containing films of a cesium compound with oxygen on
the surface of polycrystalline ruthenium foil demonstrated
that Cs 3d lines in the range E = 725.4–725.8 eV, which
5/2
b
were observed in the spectra of reduced supported
+
+
Ru−Cs systems and Ru–Cs black, can be attributed to
(
∆E = –0.5 eV). This result suggests that electron den-
cesium suboxide Cs O (although the partial presence
b
2
+ x
+
sity on ruthenium particles in the Ru–Cs /MgO system
was in reality increased primarily because of the inter-
action of ruthenium with the cesium promoter rather
than the support.
of Cs O also cannot be excluded in this case), whereas
2
Cs 3d5/2 lines in the range E = 724.7–725.2 eV, which
b
were observed in the spectra of the above samples after
oxidation in air, belong to cesium peroxide Cs O .
2
2
A hypothesis on the chemical state of cesium in the
+
Ru–Cs /MgO catalyst can be proposed based on the
DISCUSSION
values of Cs 3d5/2 Eb observed in the XPS spectrum of
In this study, the test materials were the Ru– this sample after treatments with air and hydrogen
+
+
Cs /MgO and Ru–Cs /Al O systems prepared by the (Fig. 5). To attribute these values to particular states of
2
3
same procedure (sequential impregnation of the sup- cesium, we used both published data (Table 3) and XPS
port with nonaqueous solutions of RuOHCl and data in the Cs 3d5/2 and O 1s regions for model samples
3
Cs CO with the high-temperature treatment of the sup- prepared by cesium sputtering onto polycrystalline
2
3
ported compounds with hydrogen) with approximately ruthenium foil (Table 4). An analysis of this set of data
equal concentrations of Ru and Cs. However, these sys- allowed us to conclude that the promoter mainly
tems demonstrated dramatically different catalytic occurred as cesium suboxide ës + xO in the
2
+
+
activities toward ammonia synthesis: the Ru–Cs /MgO Ru−Cs /MgO sample after its high-temperature treat-
catalyst exhibited noticeable activity in this reaction ment with hydrogen. In contact with the sample with
from 250°ë and a near-equilibrium yield of NH was oxygen, ës2 + xO was converted into cesium peroxide
3
reached on this catalyst at 350°C (Table 1), whereas the ës O , which can be reduced with hydrogen once again
2
2
formation of detectable amounts of NH in the presence to ës O. Because the XRD pattern of the
3
2 + x
+
+
of Ru–Cs /Al O was not observed even at 400°C.
Ru−Cs /MgO catalyst exhibited no reflections due to
2
3
ës O and/or ës O phases (Fig. 3) and the high-res-
According to XRD and TEM data, the particle-size
2 + x
2
2
olution TEM images exhibited no bulk structures other
than ruthenium particles and the support, we can
hypothesize that the promoter in this sample did not
form a crystallized phase but occurred only as a surface
compound. Note that, according to XPS data, the alkali
distributions and average particle sizes of ruthenium in
+
+
the Ru–Cs /MgO and Ru–Cs /Al O samples were
2
3
approximately the same (Table 2). Therefore, in this
case, the difference in catalytic activity cannot be
explained by a different dispersity of ruthenium.
+
metal also occurred in the ionic form in the Cs /MgO
At the same time, XPS data exhibited a difference in
reference sample after its contact with air. However,
hydrogen treatment at 350°C and 0.1 MPa did not result
in the formation of cesium suboxide in this case. It is
likely that the presence of ruthenium metal as an activa-
tor of molecular hydrogen is necessary for this; after the
+
the electronic states of ruthenium in the Ru–Cs /MgO
+
and Ru–Cs /Al O samples; in both cases, the elec-
2
3
tronic state of ruthenium was different from the elec-
tronic state of Ru atoms in bulk metal. The Ru 3d5/2 line
+
observed in the spectrum of the Ru–Cs /MgO sample
dissociative chemisorption of H molecules on ruthe-
2
was shifted toward lower values of E by 1.2 eV with
b
+
nium, hydrogen atoms are transferred to Cs ions in one
reference to the spectrum of polycrystalline ruthenium
foil (Fig. 5). An analogous negative shift of the Ru 3d
spectrum, as compared with the spectrum of bulk ruthe-
way or another to reduce them.
High-resolution TEM data obtained with the use of
+
nium metal, was found previously [15] in a sample of EDX indicate that ruthenium particles in the Ru–Cs /MgO
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

8
98
LARICHEV et al.
sample were coated with a layer containing cesium this compound was not reduced upon hydrogen treat-
Figs. 1a, 2). With consideration for the above XPS ment in the spectrometer chamber (Fig. 6). A negative
(
data, it is reasonable to assume that this layer in reduced line shift with reference to the corresponding signal in
or oxidized samples was surface cesium suboxide the spectrum of bulk ruthenium was absent from the Ru
ës2 + xO or cesium peroxide ës O , respectively. The 3d spectrum. The value of Ru 3d Eb for Ru–
2
2
5/2
+
same layer covered MgO surface regions adjacent to Cs /Al O with consideration for differential charging
2
3
ruthenium particles; however, it was invisible in MgO and relaxation effects was 280.6 eV (Fig. 5), which is
surface regions free of ruthenium particles. This obser- typical of ruthenium on oxide supports. On this basis,
+
vation indicates that, in the Ru–Cs /MgO sample, the we can conclude that, in this case, the presence of
promoter was primarily localized on the surface of cesium did not cause a decrease in the work function of
ruthenium particles and tightly bound to them. The electrons from the surface of ruthenium, most likely,
migration of a promoter layer to the support can occur because of the absence of contact between the cesium
only within the limits of small MgO surface regions compound and the surface of ruthenium particles.
adjacent to ruthenium particles. Previously, in a study
The question arises, why did the coating of ruthe-
nium particles with a film of cesium suboxide occur
with the use of MgO as a support and not occur in the
case ofAl O ? This question can be tackled based on an
of model systems prepared by cesium sputtering onto a
+
single crystal of silver metal and Ag–Cs /Al O cata-
2
3
lysts, it was found that the film of cesium suboxide can
be fixed on the surface of silver by the formation ofAg–
Cs bonds [19]. It is reasonable to assume that such an
interaction between the cesium promoter and the tran-
2
3
analysis of the temperature dependence of surface
Cs/Mg(Al) atomic concentration ratios in the test sys-
+
+
tems (Fig. 7). In the Ru–Cs /MgO and Cs /MgO sys-
tems, heating at T > 300°C significantly changed the
surface concentration of cesium, whereas the surface
+
sition metal (Ru) also takes place in the Ru–Cs /MgO
system.
Various mechanisms of the effect of a cesium layer concentration of cesium depended on temperature only
that covers ruthenium particles on the electronic state of slightly over the entire range of test temperatures in the
these particles can be conceived. Only one of them, case of analogous systems based on Al O . This result
2
3
electron-density transfer from the alkaline promoter to suggests that surface cesium suboxide in the Ru–
+
ruthenium particles, has usually been considered for Cs /MgO sample possessed certain mobility, whereas
+
Ru–Cs /MgO systems [4, 5]. The experimental data on cesium was strongly bound to the support in the sys-
the chemical state of the promoter cast doubt on the tems based on Al O . The XRD data (Fig. 3) allowed us
2
3
possibility of this transfer because the promoter is a to assume that, in these systems, cesium occurred as the
cesium compound with oxygen rather than cesium surface cesium aluminate Cs Al O , which resulted
2
2
4
metal. An alternative mechanism implies that cesium from the reaction of Cs CO with Al O ; this reaction
2
3
2
3
ions on the surface of ruthenium particles exert a polar- can occur under conditions of the hydrogen treatment
izing effect, which decreases the work function of elec- of samples at 450°ë. It is well known that alkali metal
trons from the surface of ruthenium. An effect of this aluminates are refractory and that they cannot possess
kind has long been found in many systems of bulk tran- surface mobility at the temperatures of XPS measure-
sition metals surface-coated with films of alkali metal ments (T ≤ 600°ë). In contrast to amphoteric Al O ,
2
3
compounds (for example, cesium–silver S1 photocath- magnesium oxide is a typical base and it can hardly
odes) [18]. It is clear that, in the case of supported sys- react with cesium carbonate or cesium oxides, which
tems where the support is a dielectric (such as MgO or are strong bases. Consequently, cesium suboxide can
Al O ) and, consequently, there is no electric contact migrate over the surface of MgO and come into contact
2
3
between the Fermi levels of the spectrometer and the with ruthenium particles to facilitate a decrease in the
supported metal, a decrease in the work function of work function of electrons from the surface of ruthe-
electrons results in a shift of the metal line in the XPS nium. In the case of Al O , the initial promoter species
2
3
spectrum toward lower values of E . In contrast, in the is bound to oxygen centers to lose the capacity for inter-
b
case of bulk systems where contact between the Fermi action with ruthenium particles.
levels of the spectrometer and the metal is retained, a
change in the work function does not cause a line shift
in the XPS spectrum. Indeed, a negative shift of the
In accordance with the current concepts of the
mechanism of ammonia synthesis on ruthenium cata-
lysts [1, 4, 5], the work function of electrons from the
surface of ruthenium has a significant effect on catalytic
activity. A decrease in the work function facilitates
electron transfer from the surface of ruthenium to the
Ru 3d5
/2
line similar to that observed in the
+
Ru−Cs /MgO sample was absent from the spectrum of
+
Ru–Cs black, although it is likely that direct contact
between ruthenium metal and a cesium compound
occurred in this black.
antibonding molecular orbital of the adsorbed N mol-
2
ecule and thereby facilitates N dissociation, which is
2
The high-resolution TEM micrographs of the the most difficult step of ammonia synthesis. Hence, it
+
Ru−Cs /Al O sample did not exhibit a coating of is believed that the contact of ruthenium particles with
2
3
ruthenium particles with a layer of any compound a film of cesium suboxide, which results in a decrease
(
Fig. 1b). It follows from the Cs 3d spectra that in the work function of electrons, is of importance for
5/2
+
+
cesium occurred as a Cs compound in the sample and the Ru–Cs /support system to exhibit considerable cat-
KINETICS AND CATALYSIS Vol. 46 No. 6 2005

EFFECT OF THE SUPPORT ON THE NATURE OF METAL–PROMOTER INTERACTIONS
899
alytic activity toward ammonia synthesis. This can
4. Aika, K., Shimazaki, K., Hattori,Y., Ohya,A., Ohshima, S.,
+
Shirota, K., and Ozaki, A., J. Catal., 1985, vol. 92,
explain the fact that the Ru–Cs /MgO catalyst, in which
p. 305.
this contact takes place, exhibits activity in ammonia
synthesis, whereas an analogous system based on
Al O , in which there is no contact of this kind because
5
6
. Muhler, M., Rosowski, F., Hinrichsen, O., Hornung, A.,
and Ertl, G., Stud. Surf. Sci. Catal., 1996, vol. 101,
p. 317.
. Larichev, Yu.V., Prosvirin, I.P., Shlyapin, D.A., Shi-
tova, N.B., Tsyrul’nikov, P.G., and Bukhtiyarov, V.I.,
Kinet. Katal., 2005, vol. 46, no. 4, p. 635.
2
3
of the interaction of the promoter with the support, was
found to be inactive in this reaction.
The above assumptions also explain the well-known
+
fact that Ru–Cs /Al O systems can exhibit catalytic
7. Cattania, M.G., Parmigiani, F., and Ragani, V., Surf. Sci.,
2
3
activity in ammonia synthesis but only at a very high
1989, vols. 211/212, p. 1097.
promoter concentration (Cs/Ru atomic ratio ≥ 3.0) [3,
8. Moggi, P., Albanesi, G., Predieri, G., and Spoto, G.,
8
]. It is likely that, at a high promoter concentration, a
Appl. Catal., A, 1995, vol. 123, p. 145.
portion of the promoter remains unbound to Al O and
9. Hikita, T., Aika, K., and Onishi, T., Catal. Lett., 1990,
2
3
enters into contact with ruthenium particles.
vol. 4, p. 157.
1
0. Kobayashi, M. and Shirasaki, T., J. Catal., 1973, vol. 28,
p. 289.
ACKNOWLEDGMENTS
11. Practical Surface Analysis by Auger and X-ray Photo-
electron Spectroscopy, Briggs, D. and Seah, M., Eds.,
Chichester: Wiley, 1983.
This work was supported by the Russian Founda-
tion for Basic Research (project no. 02-03-32681).
Yu.V. Larichev acknowledges the support of the Zama-
raev International Charity Scientific Foundation.
1
2. Kuznetsov, L.D., Dmitrenko, L.D., Rabina, P.D., and
Sokolinskii, Yu.A., Sintez ammiaka (Ammonia Synthe-
sis), Moscow: Khimiya, 1982.
1
3. Bukhtiyarov, V.I., Prosvirin, I.P., and Kvon, R.I., J. Elec-
tron Spectrosc. Relat. Phenom., 1996, vol. 77, p. 7.
REFERENCES
1
1
4. Barr, T.L., Crit. Rev. Anal. Chem., 1991, vol. 22, p. 229.
5. Larichev, Y.V., Moroz, B.L., Prosvirin, I.P., Bukhti-
yarov, V.I., and Likholobov, V.A., Chem. Sustainable
Dev., 2003, vol. 11, p. 155.
6. Hwang, C.C., An, K.S., Park, R.J, Kim, J.S., Lee, J.B.,
Park, C.Y., Lee, S.B., Kimura, A., and Kakizaki, A,
J. Electron Spectrosc. Relat. Phenom., 1998, vol. 88,
p. 733.
1
. Anastasiadi, S.A., Semenova, T.A., Rabina, P.D., and
Kuznetsov, L.D., Nizkotempraturnyi ruteniisoderzha-
shchii katalizator sinteza ammiaka: obzornaya infor-
matsiya (Low-Temperature Ruthenium-Containing Cat-
alyst for Ammonia Synthesis: A Review), Prom–st
Proizvod. Miner. Udobr., Ser.: Azotn. Prom–st, Moscow:
NIITEKhIM, 1988.
1
1
1
7. Ebbinghaus, G. and Simon, A., Chem. Phys., 1997,
2
. Moggi, P., Predieri, G., and Maione, A., Catal. Lett.,
vol. 43, p. 117.
2
002, vol. 79, p. 7.
8. Yang, S.J. and Bates, C.W., Appl. Phys. Lett., 1980,
vol. 36, p. 675.
3
. Rosowski, F., Hornung, A., Hinrichsen, O., Herein, D.,
Muhler, M., and Ertl, G., Appl. Catal., A, 1997, vol. 151, 19. Podgornov, E.A., Prosvirin, I.P., and Bukhtiyarov, V.I.,
p. 443.
J. Mol. Catal. A, 2000, vol. 158, p. 337.
KINETICS AND CATALYSIS Vol. 46 No. 6 2005