Preparation of Ru nanoparticles supported on γ-Al2O3 and its novel catalytic activity for ammonia synthesis

Article abstract of DOI:10.1006/jcat.2001.3418

Uniform ruthenium metal nanoparticles with an average diameter of 5 nm were prepared and supported on γ-Al₂O₃ up to 6.3 wt%. The ruthenium nanoparticles were obtained by the reduction of RuCl₃ in ethylene glycol. While maintaining their original morphology, more than 99% of the colloidal Ru nanoparticles, obtained were successfully supported on γ-Al₂O₃. The Ru/Al₂O₃ catalyst prepared from a metal colloid had an unusually high activity for ammonia synthesis. The rate of ammonia formation was at least 12 times higher than the rate of nonpromoted Ru/Al₂O₃ catalysts prepared by conventional methods, and at least 3.5 times higher than the promoted Cs, Rb, K, Ce, La, and Sm Ru/Al₂O₃ catalysts. Alumina was not expected to be an ideal support for ammonia synthesis because of its acidity, and thus the promotion by additives was considered necessary. The present investigation proves that alumina can be a suitable support for ammonia synthesis if ruthenium is supported on it as metal nanoparticles. The results point out the possibility that the active sites formed by conventional impregnation methods are significantly contaminated with the supports. In contrast, by keeping the interaction with the support at a minimum, the novel preparation method of the Ru/Al₂O₃ catalyst by Ru colloid deposition gives well-defined metal particles with high catalytic activity.

Full text of DOI:10.1006/jcat.2001.3418

Journal of Catalysis 204, 364–371 (2001)
doi:10.1006/jcat.2001.3418, available online at http://www.idealibrary.com on
Preparation of Ru Nanoparticles Supported on -Al2O3 and Its Novel
Catalytic Activity for Ammonia Synthesis
Akane Miyazaki, ,1 Ioan Balint,† Ken-ichi Aika, and Yoshio Nakano
Department of Environmental Chemistry and Engineering, Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute
of Technology, 4259 Nagatsuta, Midoriku, Yokohama, Kangawa 226-8502, Japan; and †Institute of Physical Chemistry,
Romanian Academy, Spl. Independentei 202, 77208 Bucharest, Romania
Received April 16, 2001; revised September 17, 2001; accepted September 17, 2001
this is still in question, because a raft structure is also sug-
gested (4).
Uniform ruthenium metal nanoparticles with an average dia-
meter of 5 nm were prepared and supported on -Al2O3 up to
6.3 wt%. The ruthenium nanoparticles were obtained by the reduc-
tion of RuCl3 in ethylene glycol. While maintaining their original
morphology, more than 99% of the colloidal Ru nanoparticles, ob-
tained were successfully supported on -Al2O3. The Ru/Al2O3 cata-
lyst prepared from a metal colloid had an unusually high activity
for ammonia synthesis. The rate of ammonia formation was at least
12 times higher than the rate of nonpromoted Ru/Al2O3 catalysts
prepared by conventional methods, and at least 3.5 times higher
than the promoted Cs, Rb, K, Ce, La, and Sm Ru/Al2O3 catalysts.
Alumina was not expected to be an ideal support for ammonia
synthesis because of its acidity, and thus the promotion by additives
was considered necessary. The present investigation proves that alu-
mina can be a suitable support for ammonia synthesis if ruthenium
is supported on it as metal nanoparticles. The results point out the
possibility that the active sites formed by conventional impregna-
tion methods are significantly contaminated with the supports. In
contrast, by keeping the interaction with the support at a minimum,
the novel preparation method of the Ru/Al2O3 catalyst by Ru col-
loid deposition gives well-defined metal particles with high catalytic
c
It is well known that the catalytic activity of a supported
metal is strongly related to the morphology of the particle,
i.e., size and shape (5). However, the conventional prepa-
ration of catalysts, consisting of the impregnation of a sup-
port with an aqueous solution of a soluble metal precursor,
makes it difficult to control the final size and shape of the
supported metal particles. Additionally, it is highly possi-
ble that the support has a great influence on the catalytic
activity of the metal when the catalyst is prepared by im-
pregnation. It has been evidenced that the active phase of
catalysts prepared by impregnation can contain variable
amounts of aluminum as a result of support dissolution, de-
pending on the conditions of impregnation such as time,
temperature, and pH (6, 7). Moreover, very little is known
about the structure of small, supported metal particles and
their interaction with the support.
The conventional Ru/Al2O3 catalysts are known to ex-
hibit quite low activities for ammonia synthesis, and this
has been attributed to the acidity of alumina. The addi-
tion of alkali or lanthanide promoters was reported to be
an effective way of enhancing the catalytic activity (2, 3).
The proposed mechanism of this promoting effect is a con-
tentious issue. Aika et al. (8) explained the enhancement
of the catalytic activity by electronic promotion. According
to the mechanism they proposed, the electrons are trans-
ferred from the support to the ruthenium metal, inducing a
decrease in the ionization potential of ruthenium. The de-
creased ionization potential allows electron transfer from
the metal to the anti-bonding orbits of the nitrogen atom,
reducing the activation energy for dissociative adsorption
of the N2 molecule. On the other hand, LEED and NMR
spectroscopy results could not prove such an electron trans-
fer from alkali promoters to ruthenium (9, 10).
activity.
2001 Elsevier Science
Key Words: metal nanoparticles; Ru; ammonia synthesis; polyol
method; supported catalyst.
1. INTRODUCTION
Ruthenium is known to have one of the highest cata-
lytic activities for ammonia synthesis (1). The conven-
tional ruthenium catalysts are prepared by the impregna-
tion of oxide supports either with an aqueous solution of
RuCl3 3H2O or with Ru3(CO)12 dissolved in tetrahydro-
furan (2, 3). When catalysts are prepared by impregna-
tion of alumina with RuCl3, the metal particles, after dry-
ing, calcination, and reduction, are not uniform in size and
shape. The metal particles obtained through the impregna-
tion with Ru3(CO)12 are modeled as Ru6 clusters. However,
An alternative method to obtain supported catalysts with
well-defined metal particles is the preparation of supported
catalysts from metal colloids. The great advantage of the
colloid method is that it provides relatively monodispersed
metal particles. Recent investigation shows that not only
1 To whom correspondence should be addressed. Fax: 81-45-924-5441.
E-mail: akanem@chemenv.titech.ac.jp.
364
0021-9517/01 $35.00
c
2001 Elsevier Science
All rights reserved

PREPARATION OF Ru NANOPARTICLES
365
the particle size but also the crystal structure of the metal 2.2. Preparation of Ru/Al2O3
nanoparticles can be controlled to some extent by using ap-
The alumina used in this study, supplied by Aerosil Japan
Co., has a surface area of 123.8 m2/g and an average particle
size of 13 nm. A ruthenium colloid obtained by the polyol
method was supported on -alumina by (i) deposition of
the Ru colloid, which was already formed, onto alumina;
and (ii) reduction of ruthenium chloride in the presence of
alumina (on the surface of alumina). In the former case,
the Ru colloid was prepared according to the description
presented in Section 2.1. -Al2O3 (2 g) was then added
to the colloidal solution under vigorous stirring to form a
suspension. In the latter case, 2 g of -Al2O3 was suspended
into 50 ml of an ethylene glycol solution of RuCl3, and
then the temperature of the suspension was increased up
to 453 K. In both cases, ethylene glycol was removed later
by diluting the suspension with 450 ml of 0.3 M NaNO3
aqueous solution. The solid was collected by filtration, and
then dried in air.
The size of the ruthenium particles and their dispersion
were determined by chemisorption of H2 and CO, assum-
ing that each exposed Ru atom could adsorb one hydrogen
atom or one CO molecule. Prior to the chemisorption mea-
surements, the Ru/Al2O3 samples were heated in vacuum
up to 673 K, treated with H2 at 200 Torr for 2 h, evacuated,
and then cooled down to room temperature.
propriate structure-directing polymers for colloid prepara-
tion (11). Additionally, the preparation of supported cata-
lysts from metal colloids can minimize the effect of support
on the metallic active phase. However, there are very few
reports about a preparation method for a stable Ru col-
loid. Hirai et al. (12, 13) made an attempt to prepare a Ru
colloid by refluxing a methanol–water solution containing
Ru salts. Although they succeeded in obtaining stable col-
loid solutions of Rh, Pd, Os, Ir, Ag, and Au by this method,
they could not obtain a Ru colloid. Recently, Vidoni et al.
(14) succeeded in obtaining a stable ruthenium colloid. The
method needs high pressure (3 bar of H2), a special precur-
sor, andanorganicsolvent, andthusitisnotveryconvenient
for general application.
Recently, the polyol method (15, 16) was reported as
new and promising for the preparation of nanoscale metal
powders. This method can be applied for all the metal
salts that are soluble in ethylene glycol. Using the polyol
method, Kurihara et al. (16) succeeded in obtaining nano-
size powders of more than 10 metals including ruthenium.
This method produces a black deposit of metal particles on
the bottom of the reactor vessel. The reduction of metal
ions takes place in three steps: nuclear formation, particle
growth, and aggregation of the particles. The metal pow-
ders obtained by the polyol method are the result of the
formation of metal particles and their aggregation. If the
aggregation is avoided, the polyol method can be applied
for the preparation of ruthenium colloids.
2.3. Catalytic Activity Measurements
The catalytic activity of Ru/Al2O3 obtained by the polyol
method was measured for ammonia synthesis. The catalytic
tests were performed at atmospheric pressure in a stainless-
steel reactor containing 0.4 g of Ru/Al2O3 (6.3 wt% Ru).
Prior to the catalytic tests, the Ru/Al2O3 was pelletized,
crushed, andthensieved. Thefraction, from335to1000 m,
was collected and loaded into the reactor. Before the test,
the sample was reduced in H2 flow at 823 K for 2 h.
The objectives of this paper are (i) to prepare a stable
Ru colloid with a sharp size distribution by using the polyol
method, (ii) to find the optimum experimental conditions
for supporting Ru colloidal particles on -alumina, and
(iii) to test the catalytic activity of the alumina-supported
ruthenium nanoparticles for ammonia synthesis. Addition-
ally, we will discuss the support role played in the formation
of catalytic active phases.
The catalytic activity tests were carried out at a flow rate
of the reaction mixture 60 cm3/min STP (45 cm3/min H2 and
15 cm3/min N2). The rate of ammonia synthesis was mea-
sured in a 638 to 773 K temperature range. The produced
2. METHODS
3
ammonia was trapped by a 2 10 N solution of sulfuric
acid, and the rate of ammonia formation was determined
from the decrease in the conductivity of the solution.
2.1. Preparation of Ru Colloid
The polyol method was applied to the preparation of
a Ru colloid as in the report by Kurihara et al. (16). The
original polyol method for the synthesis of metal powders
was adapted for colloid preparation by decreasing the con-
2.4. Analysis
Several experimental techniques have been used to char-
centration of metal precursor in the solution. First, the acterize the samples. The X-ray powder diffraction (XRD)
RuCl3 nH2O (Wako Clemicals, purity >99%) was dissol- analysis was carried out by using the CuK radiation of a
ved in ethylene glycol. The temperature of the solution was Rigaku Geiger Flex RAD-B diffractometer.
then increased up to 453 K by using an oil bath. The process
Transmission electron microscopy (TEM) with electron
of metal reduction was monitored by a TEM and a UV–vis probe X-ray microanalysis (EPXMA) (Hitachi H8 100)
spectrophotometer. A few milliliters of aliquots were sam- was performed at 200 kV. The liquid samples for TEM
pled periodically from the solution and then cooled down werepreparedbydroppingthesolutionontocarbon-coated
rapidly by using an ice-water bath.
copper-mesh grids (Oken) and drying in air. The solid

366
MIYAZAKI ET AL.
samples were grained well with an agate mortar, suspended
in water, and then supported on the carbon-mesh grid.
UV–vis spectra were measured by a Hitachi U-2000A
spectrophotometer using distilled water as the reference.
The zeta potential was measured with an electrophoretic
light scattering spectrophotometer (Otsuka Electronics
ELS-800) using ethylene glycol as the background solution.
The concentration of Ru was determined by an induc-
tively coupled plasma spectrometer (ICP, Seiko SPS 400).
FIG. 2. XRD pattern of the black precipitate.
3. RESULTS AND DISCUSSION
3.1. Synthesis and Characterization of Ruthenium Colloid
cles. The primary small particles had a uniform ellipsoidal
shape with a diameter of less than 5 nm. Therefore, it can be
concluded that the brownish clear solution obtained after
10 min of heating was a colloidal solution of the ruthenium
nanoparticles.
The process of Ru reduction in ethylene glycol was mon-
itored by TEM and a UV–vis spectrophotometer. Figure 1
shows the time change of the solution color together with
the corresponding TEM images. The color of the initial so-
lution, i.e., ethylene glycol solution of RuCl3, was yellow.
In the course of heating, this color changed to transparent
(after 5 min), pink (after 8 min), brown (after 10 min), and
finally gray. In the early stages of heating, before the so-
lution became brownish, particle formation could not be
observed by TEM. After heating for 10 min, the color of
the solution turned to brown, and the formation of small
particles was observed. EPXMA proved that the particles
were composed of ruthenium, and an electron beam diffrac-
tion pattern indicated a crystalline structure of these parti-
Furtherheatingledthecolloidalparticlestoagglomerate.
Aggregates of around 1 m were observed by TEM after
15 min of heating (Fig. 1 (b)). After 30 min of heating,
a fine black precipitate was formed at the bottom of the
reaction vessel. The XRD pattern of this powder confirmed
the complete reduction of RuCl3 to Ru metal (Fig. 2).
The polyol method was developed to prepare metal pow-
ders. By decreasing the concentration of RuCl3 in ethylene
glycol by about 1/100, we succeeded in obtaining a sta-
ble Ru colloid by the polyol method. The concentration of
RuCl3 in ethylene glycol and the reduction time were other
crucial factors for obtaining a monodispersed and stable Ru
colloid. To avoid the agglomeration process, it was neces-
sary to stop the reduction immediately after the formation
of the small Ru particles.
The UV–vis spectrum of the freshly prepared ethylene
glycol solution of RuCl3 exhibited a broad absorption max-
imum at 398 nm (Fig. 3 (a)). This peak shifted to a shorter
wavelength, 328 nm, after 5–8 min of reduction (Fig. 3 (b)).
At this reduction time the solution was transparent. It is
noteworthy that a new absorption maximum was formed
at 275 nm after 8 min of reduction. In this stage the color
of the solution turned pink (Fig. 3 (c)). The intensity of
the new peak at 275 nm did not change by further heat-
ing, whereas the intensity of the peak at 328 nm decreased
in time (Fig. 3 (d)). Despite our expectation, we could not
find any relationship between the colloid formation and
the rise of the peak at 275 nm. This peak may be due to the
formation of some intermediate species which appeared
during the process of RuCl3 reduction. Various Ru species,
such as [Ru(H2O)6]3+, [Ru(H2O)5Cl]2+, or cis- and trans-
[Ru(H2O)4Cl2]+, have been observed during the reduction
of a commercial hydrated ruthenium chloride with mercury
(17). The formation of one or more of these species could
be the reason for the rise of the peak at 275 nm.
The UV–vis spectrum of the gray colloid showed
broad adsorption continua that extend throughout the
FIG. 1. Schematic time course of the reduction, and TEM observation
of the solution.

PREPARATION OF Ru NANOPARTICLES
367
dispersed Ru nanoparticles on -Al2O3 by adding alu-
mina into the suspension of the already formed Ru colloid
solution.
The other preparation method consisted of the re-
duction of RuCl3 by ethylene glycol in the presence of
alumina. First, RuCl3 nH2O was dissolved in 50 ml of
ethylene glycol, and then 2 g of -Al2O3 was added to
form a suspension. The experimental conditions, such as
temperature and time of reduction, were identical to those
described for the preparation of the Ru colloid (Section
2.1). The color of the suspension changed from pale yellow
to light brown. The suspension was then cooled down
to room temperature. Most of the ethylene glycol was
removed by washing the alumina with a NaNO3 solution.
The solid phase was collected by filtration and dried at
room temperature in vacuum.
The concentrations of Ru in the suspension before re-
duction and in the filtrate were determined by ICP. Table 1
lists the concentrations of Ru in the initial [Ru]initial and
final suspensions [Ru]final. By using these data, the concen-
tration of ruthenium on the solid ({Ru}support, expressed
by weight percent) and the proportion of supported ruthe-
nium, expressed as [Ru]support/[Ru]initial were obtained and
are listed in Table 1. From Table 1, it can be observed that,
by this method, ruthenium was supported on -Al2O3 with
a remarkably high efficiency. The experimental results con-
firmed the high reproducibility of this method.
The Ru/Al2O3 obtained was observed by TEM. The
TEM image shows the presence of many black and dis-
persed particles on the support (Fig. 4a). The EPXMA
measurement proved that the black spots corresponded
to ruthenium particles (Fig. 4b). It can be seen that the
ruthenium particles were uniform in size and shape, and
they were dispersed well on the surface of -Al2O3.
Figure 5 shows the size distribution of the ruthenium parti-
cles, which was obtained by counting more than 200 par-
ticles observed by TEM. The mean particle size of Ru
nanoparticles supported on the alumina was 4.3 nm, and
this corresponded well the average size of the colloidal Ru
nanoparticles (around 5 nm) (Fig. 1).
FIG. 3. UV–vis spectra of solution before and after RuCl3 reduction.
(a) 0 min (before reduction), (b) 5 min, (c) 8 min, (d) 11 min, (e) 15 min.
visible–near-ultraviolet region (Fig. 3 (e)). This spectrum
agreed well with the spectrum of the ruthenium colloid hav-
ing a diameter of 10 nm, which was calculated by means of
the Mie theory using the optical constants of the metal (18).
The reduction process of RuCl3 in ethylene glycol can
proceed in two steps: (i) ligand exchange around the Ru3+
ionsand(ii)reductionofRu3+ toRumetal. WhenaK2PtCl4
solution is reduced by H2 gas, ligand exchange between Cl
and H2O takes place, and this is reflected in the change of
the UV–vis spectra (19). It is highly possible that the color
changefromyellowtotransparent(Fig. 1)wascausedbythe
ligand exchange taking place prior to the reduction step. In
the polyol method, ethylene glycol is not only a solvent but
also a reducing agent. The oxidation of ethylene glycol re-
sultsintheformationofglyoxal(CHOCHO)viaglycolalde-
hyde (HOCH2CHO) and/or glycolic acid (HOCH2COOH)
(20). Our IR measurement did not show the presence of a
The sizes of the supported Ru particles obtained by
chemisorption of H2 and CO were 4.8 and 5.4 nm, re-
spectively. These values agree well with the particle size
obtained from TEM observation (Table 2). In reality, Ru
metal can chemiadsorb up to 2 mol of CO/mol, resulting in
dicarbonilated ruthenium species anchored to alumina.
==
C O group. This suggests that ethylene glycol can be easily
oxidized to glyoxal, whose boiling point is 324 K.
3.2. Supporting Ruthenium Nanoparticles on -Al2O3
Attempts were made to support the ruthenium nanopar-
ticles, which were obtained by the polyol method, on
-
Al2O3. As described under Methods two procedures were
tried supporting before or after reduction of ruthenium.
In the latter case, several difficulties arose. As discussed in
Section 3.1, the reduction of RuCl3 proceeds quickly, and
the solution should be cooled down rapidly after adequate
time of reduction to avoid the agglomeration of the par-
ticles. In practice, we could not obtain uniform and well-
TABLE 1
Ru Concentration at the Initial and Final Stages of the Reduction
[Ru]initial
(mmol/L)
[Ru]final
( mol/L)
{Ru}support
[Ru]support/[Ru]initial
(%)
(Wt%)
16.9
76.7
6.29
99.5

368
MIYAZAKI ET AL.
TABLE 2
Size of Ru Nanoparticles (nm) Measured by Different Methods
H2
CO
TEM
chemisorption
chemisorption
Ru colloid
Ru/Al2O3 (before reaction)
Ru/Al2O3 (After reaction)
5
4.3
—
—
4.8
5.1
—
5.4
—
These results suggest that the polyol method has an ad-
vantage in obtaining supported. Ru catalysts, which have
uniform and well-dispersed particles. A good dispersion
of Ru particles on an 8 wt% Ru catalyst was obtained
from a Ru3(CO)12 precursor (21). However, in practice it
is difficult to obtain highly dispersed Ru particles by a con-
ventional preparation method using RuCl3 as a precursor.
Moreover, it is noteworthy that in the case of the polyol
method, metal loading onto the support can be increased
up to 6.3 wt% or more (reported in the following) without
affecting the particle size and dispersion.
To reveal the supporting mechanism of the Ru nanoparti-
cles on -Al2O3, the zeta potential of the ruthenium colloid
was measured by using an electrophoretic light scattering
spectrophotometer. The sample for this measurement was
prepared without alumina, and ethylene glycol was used
as the reference solution. The pH values of the obtained
colloid were adjusted by adding HCl or NaOH solutions.
Figure 6 shows the zeta potential of a ruthenium colloid
obtained by the polyol method. It can be seen that the
Ru colloid is negatively charged at a pH below 8. When
the ethylene glycol solution of RuCl3 was reduced together
with -Al2O3, the pH value of the resultant solution was 5.
In this case, the Ru colloid was negatively charged as shown
in Fig. 6. On the other hand, the pHZPC of the -Al2O3 used
in our study was reported to be 8 (6). Therefore, the surface
of alumina must be positively charged at pH 5. In conclu-
sion it can be said that the ruthenium particles, formed in
ethylene glycol solution, were supported on -Al2O3 by
electrostatic force. This is also supported by additional ex-
periments using a surfactant, poly(vinyl alcohol). Using this
FIG. 4. (a) TEM image and (b) EPXMA spectrum of 6.3 wt% Ru/
Al2O3.
This was observed especially in the case of very small ruthe-
nium clusters prepared from Ru3(CO)12 (21). On the other
hand, the large ruthenium clusters are less reactive against
CO, chemisorbing up to 1.4 mol of CO/mol of metal (21).
From the results shown in Table 2, it can be said that the
ratio between adsorbed CO/H was around 1 : 1, and this
means that the supported ruthenium nanoparticles have
lower reactivity against CO. Therefore, it was reasonable
to suppose that each exposed Ru atom adsorbed one hy-
drogen atom or one CO molecule.
FIG. 5. Size distribution of ruthenium particles on Al2O3 as deter-
mined from TEM micrographs.
FIG. 6. Zeta potential of ruthenium colloid.

PREPARATION OF Ru NANOPARTICLES
369
FIG. 8. Arrhenius plot for the ammonia synthesis rate over 6.3 wt%
Ru/ -Al2O3.
FIG. 7. Temperature dependence of the rate of ammonia synthesis
over Ru/ -Al2O3 (6.3 wt%). The rates over conventional Ru/Al2O3 cata-
lysts are also shown for comparison.
The rates (TOF) of ammonia synthesis over Ru nanopar-
ticles supported on alumina are shown in the Arrhenius-
type plot in Fig. 8. An apparent activation energy of 76.9 KJ
mol 1 was estimated in Fig. 8 and listed in Table 3. Our re-
sult agrees well with the previously published data. The
apparent activation energies determined for promoted
and nonpromoted Ru/Al2O3 catalysts range between 44
and 101 kJ mol 1 (3).
From the previous results there are two points that are
worthy of note. One is the temperature of highest activity
for ammonia synthesis. The highest activity of the con-
ventional Ru/Al2O3 catalysts was observed at 588 K (3),
whereas the catalyst prepared from the Ru colloid had a
maximum activity at a higher temperature, 723 K. From an
industrial point of view, it is preferable for ammonia syn-
thesis to have a catalyst that is more active at a lower tem-
perature. Thermodynamically, the increase in temperature
is not favorable for ammonia synthesis reaction. Therefore,
it is of great interest to obtain the higher equilibrium con-
versions at lower temperatures.
nonionic surfactant, the concentration of ruthenium on the
solid decreased from 6.3 to 4.7 wt%, although the condi-
tion was the same. Polymers are known to cap the surface
of colloid particles. This phenomenon is known as steric
stabilization (22). By capping poly(vinyl alcohol), the dis-
tance between the Ru metal particles and the surface of
-Al2O3 increased, and therefore the Coulombic force be-
tween them decreased inversely proportional to the second
power of the distance.
3.3. Catalytic Activity for NH3 Synthesis
The activity of the Ru/ -Al2O3 catalyst, obtained as de-
scribed previously, was tested for NH3 synthesis. The cata-
lyst which was prepared by supporting the Ru colloid on
-Al2O3, showed a remarkably high activity for NH3 syn-
thesis. The reaction rates expressed as micromoles per
gram-hour and turnover frequency (TOF) as a function of
temperature are shown in Fig. 7 and Table 3, respectively.
Figure 7 shows that the rate of ammonia synthesis over
6.3 wt% Ru/Al2O3 increased progressively with an increase
in temperature, reaching a maximum at 723 K. Above this
temperature, the rate of the reverse reaction (ammonia de-
composition) increased more than the rate of ammonia
formation, and therefore the overall rate decreased. The
highest reaction a rate of 923 mol g 1 h 1 was observed at
723K. Thereproducibilityateachreactiontemperaturewas
within the range of experimental error ( 25 mol g 1 h 1).
The other point is that the Ru/Al2O3 catalyst prepared
from the Ru colloid showed unusually high activity al-
though it was not promoted. The highest catalytic activ-
ities of the promoted and nonpromoted Ru/Al2O3 cata-
lysts prepared by conventional methods using RuCl3 or
Ru3(CO)12 as precursors together with the activity of
the catalyst prepared from the Ru colloid are shown in
Fig. 7. The reported activity of the nonpromoted conven-
tional Ru/Al2O3 catalysts is very small, ranging from 10
1
to 65 mol g
h
1. It was reported that the nonpro-
moted catalysts prepared from RuCl3 exhibited signifi-
cantly lower activities as compared to those obtained from
Ru3(CO)12.
TABLE 3
The acidity of alumina has been considered the main rea-
son for the low activity of the conventional Ru/Al2O3 cata-
lysts for ammonia synthesis. The addition of alkaline (Cs,
Rb, K) or rare earth (La, Ce, La, Sm) elements to Ru/Al2O3
leads to a significant increase in the catalytic activity (2, 21).
Typically, the activity of the promoted Ru/Al2O3 cata-
Ammonia Synthesis Catalyzed by 6.3 wt% Ru/ -Al2O3 Prepared
by the Polyol Method
4
1
TOF (10
673 K
S )
Dispersion (%)
54
638 K
1.4
723 K
7.6
773 K
4.9
Ea (kJ/mol)
76.9
1
1
lysts ranges from 130 to 250 mol g
h
(Fig. 7). The
4.1
Ru/Al2O3 catalyst prepared from the Ru colloid showed a

370
MIYAZAKI ET AL.
significantly higher activity than that from promoted cata- catalytic activity. In contrast to the impregnation method,
lysts. A notable exception is the K+- promoted Ru/Al2O3 metal colloid deposition onto a support gives metal par-
catalyst, prepared from Ru3(CO)12, whose catalytic activity ticles that are uncontaminated by the support. Therefore,
1
for ammonia synthesis was reported to be 2470 mol g
h
the influence of the support on the metallic active phase is
1
under conditions comparable to ours (0.4 g catalyst, minimized. The Ru/Al2O3 catalyst prepared in this study
60 ml min 1) (21). However, the activity of the conven- is supposed to have Ru metal nanoparticles that do not in-
tionally prepared Ru catalysts is strongly dependent on the teract significantly with the support, and this should be the
conditions of preparation. Slight changes of the prepara- reason for the remarkably high catalytic activity demon-
tion variables result in significant changes in the catalytic strated for ammonia synthesis.
activity.
The differences observed between the Ru/Al2O3 cata-
lysts prepared by the conventional impregnation methods
4. CONCLUSIONS
and the catalyst obtained via colloid deposition raise prob-
A stable Ru metal colloid was obtained by reducing
lems regarding the role that supports play in the forma-
RuCl3 in ethylene glycol. The metal particles in the col-
tion of catalytically active phases. We reported that the
loidal solution had a uniform shape and size of 5 nm. The
support plays an essential role in the formation of the ac-
colloidal Ru nanoparticles (up to 6.3 wt%), maintaining
tive phase(s) when the catalysts were prepared by the im-
their initial morphology, were successfully supported on
-
pregnation method (6, 7). The impregnation process can
be regarded as complex sequences of chemical reactions
taking place at the solid (the support)/liquid (solution of
the metal salt) interface. In the case of Cu and Pd, we re-
vealed that a certain amount of alumina is dissolved dur-
ing the adsorption of the metal precursor even in the neu-
tral pH range. The dissolved aluminium ions together with
Cu2+ or Pd2+ ions can reprecipitate on the surface of the
support, and thus participate in the formation of active
sites.
Al2O3. The Ru/ -Al2O3 prepared from the colloid showed
an unusually high activity for ammonia synthesis. The novel
preparation method of supported Ru catalysts proposed
in this paper has the following advantages: (i) it is rela-
tively easy to control the size and shape of metal particles;
(ii) the interaction between the metal particles and the sup-
port is minimized, and therefore the metal-active sites are
not contaminated by the support.
The variation of the activities of Ru/Al2O3 catalysts pre-
pared by conventional impregnation methods has been at-
tributed to various factors such as size distribution of the
supported metal particles, support influence on the mor-
phology of the ruthenium crystals (23), and calcination and
reduction processes. However, the comparison between the
two preparation methods, suggests the possibility that the
low and unstable activity of Ru/Al2O3 catalysts prepared
by conventional impregnation methods is caused not only
by the acidity of the support but also by the contamina-
tion of the Ru metal by aluminum. It is highly possible
that some aluminum is dissolved from the solid during im-
pregnation, and then participates, together with Ru, in the
formation of the active sites. On the other hand, in the
case of Ru/Al2O3 preparation from the Ru colloid, Ru
interacts with the support not as ions but as metal par-
ticles. The small amount of HCl resulting from the re-
duction of RuCl3 with ethylene glycol seems to have lit-
tle influence on the catalytic activity of Ru/Al2O3. It is
clear that the interaction between the support and the Ru
nanoparticles is minimized by this preparation method,
and therefore the active phase (Ru metal) is not con-
ACKNOWLEDGMENT
A research fund from the Japanese Society for the Promotion of Science
(No. P00136) is greatly appreciated.
REFERENCES
1. Aika, K., Petrotechnology 17, 127 (1994).
2. Murata, S., and Aika, K., J. Catal. 136, 118 (1992).
3. Murata, S., and Aika, K., J. Catal. 136, 110 (1992).
4. Gates, B. C., Chem. Rev. 95, 511 (1995).
5. Armadi, T. S., Wang, Z. L., Green, T. C., Henglein, A., and El-Sayed,
M. A., Science 272, 1924 (1996).
6. Balint, I., Miyazaki, A., and Aika, K., Chem. Mater. 11, 378
(1999).
7. Balint, I., Miyazaki, A., and Aika, K., Chem. Mater. 13, 932
(2001).
8. Aika, K., Ohya, A., Ozaki, A., Inoue, Y., and Yasmuri, I., J. Catal. 92,
305 (1985).
9. Uner, D. O., Pruski, M., Gerstein, B. C., and King, T. S., J. Catal. 114,
530 (1994).
10. Over, H., Bludau, H., Skottke-Klein, M., Moritz, W., and Ertl, G., Phys.
Rev. B46, 4360 (1992).
11. Miyazaki, A., and Nakano, Y., Langmuir 16, 7109 (2000).
12. Hirai, H., J. Macromol. Sci. Chem. A13, 633 (1979).
taminated by aluminum or chlorine ions derived from the 13. Hirai, H., Nakao, Y., and Toshima, N., J. Macromol. Sci. Chem. A13,
727 (1979).
support.
14. Vidoni, O., Philippot, K., Amiens, C., Chaudret, B., Balmes, O., Malm,
During the impregnation process, the metal particles, i.e.,
the active sites of the catalysts, are contaminated more or
less by the supports. In this case the acid or base character of
the supports plays an important role in determining the final
J. O., Bovin, J. O., Senocq, F., and Casanove, M. J., Angew. Chem. Int.
Ed. Engl. 38, 3736 (1999).
15. Fievet, F., Lagier, J. P., Blin, B., Beaudoin, B., and Figlarz, M., Solid
State Ionics 32/33, 198 (1989).

PREPARATION OF Ru NANOPARTICLES
371
16. Kurihara, L. K., Chow, G. M., and Schoen, P. E., Nanostruct. Mater. 5,
607 (1995).
20. Morrison, R. T., and Boyd, R. N., “Organic Chemistry.” Allyn &
Bacon, Needham Heights, MA, 1983.
17. Livingstone, S. E., in “Comprehensive Inorganic Chemistry” (J. C.
Bailar, H. J. Emeleus, R. Nyholm, andA. F. Trotman-Dickenson, Eds.),
Vol. C, Chap. 43. Pergamon Press, Oxford, 1973.
18. Creighton, J. A., and Eadon, D. G., J. Chem. Soc., Faraday Trans 87,
3881 (1991).
21. Moggi, P., Albanesi, G., Predieri, G., and Spato, G., Appl. Catal. A.
Gen. 123, 145 (1995).
22. Everett, D. H., “Basic Principles of Colloid Science.” The Royal Soci-
ety of Chemistry, London, 1988.
23. Jacobsen, C. J. H., Dahl, S., Hansen, P. L., Tornqvist, E., Jensen, L.,
Topsoe, H., Prip, D. V., Moenshaug, P. B., and Chorkendorf, I., J. Mol.
Catal. 163, 19 (2000).
19. Henglein, A., Ershov, B. G., and Malow, M., J. Phys. Chem. 99,
14129 (1995).