Journal of Catalysis 275 (2010) 99–107

Contents lists available at ScienceDirect

Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat

Olefin hydrogenation by ruthenium nanoparticles in ionic liquid media: Does

size matter?
Paul S. Campbell a, Catherine C. Santini a,*, François Bayard a, Yves Chauvin a, Vincent Collière b,
Ajda Podgoršek c, Margarida F. Costa Gomes c, Jacinto Sá d
a
Université de Lyon, Institut de Chimie de Lyon, UMR 5265 CNRS-Université de Lyon-ESCPE Lyon, C2P2, Equipe Chimie Organométallique de Surface,
ESCPE 43 Boulevard du 11 Novembre 1918, F-69616 Villeurbanne, France
b
Laboratoire de Chimie de Coordination, UMR CNRS, 205, route de Narbonne, 31077-Toulouse cedex 04, France
c
Laboratoire de Thermodynamique des Solutions et des Polymères, Université Blaise Pascal, Clermont-Ferrand, 24 Avenue des Landais, 63177 Aubière, France
d
CenTACat, School of Chemistry and Chemical Engineering, Queen’s University, Stranmillis Road, Belfast, Northern Ireland BT9 5AG, UK

a r t i c l e i n f o a b s t r a c t

Article history: Tailor-made and size-controlled ruthenium nanoparticles, RuNPs, of three distinct sizes between 1 and
Received 1 June 2010 3 nm are generated from the decomposition of (g4-1,5-cyclooctadiene)(g6-1,3,5-cyclooctatriene)
Revised 19 July 2010 ruthenium(0) [Ru(COD)(COT)], under H2 in 1-butyl-3-methylimidazolium bis(trifluoromethanesulpho-
Accepted 21 July 2010
nyl)imide, C1C4ImNTf2, by simply varying experimental conditions. Catalytic hydrogenation of 1,3-cyclo-
hexadiene, CYD, and cyclohexene, CYE, in C1C4ImNTf2, has been used as a probe for the relationship
between size and catalytic performance (activity and selectivity) of RuNPs. To allow comparison between
Keywords:
different reactions, all catalytic reaction mixtures were diligently prepared in order that the parameters
Nanoparticles
Ruthenium
such as substrate/catalyst and substrate/ionic liquid ratio, and therefore, viscosity and mass transport
Ionic liquids factors remained constant. It was found that the catalytic activity increases with the NP size, while high
Size effect selectivity is only observed with the smaller NPs. In addition, the studied RuNPs exhibit a high level of
Selective hydrogenation recyclability with neither loss of activity nor significant agglomeration.
Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction preferentially dissolved in polar domains and non-polar

compounds in non-polar ones [15]. The non-polar organometallic
Transition-metal nanoparticles (NPs) of 1–10 nm in size exhibit complex, Ru(COD)(COT), is expected to be concentrated in the
physicochemical properties intermediate to those of the smallest non-polar domains of ILs. Therefore, the phenomenon of crystal
element from which they can be composed and those of the bulk growth is controlled by the local concentration of Ru(COD)(COT)
material [1–3]. In catalysis, the performance (activity and selectiv- and consequently limited to the size of the non-polar domains.
ity) of NPs is often said to be related to their size, as this controls These play the role of nanoreactors in which the size of ruthenium
the number of corner, edge and face atoms available for adsorption nanoparticles generated in situ can be controlled [13,16,17].
and activation of substrates [4–9]. However, the synthesis of nano- Using NPs formed in situ in ILs directly in catalysis offers the
particles (NPs) with a controlled size in the range of 1–10 nm in opportunity to exploit the distinct physicochemical and solvation
order to corroborate this theory is still a challenging issue [10]. properties of these media, resulting in unique activities and selec-
The use of ionic liquids (ILs) in NP synthesis has recently be- tivities [12,18–21].
come a popular route. The major advantage is that stabilising addi- When comparing the catalytic activity of differently sized nano-
tives such as ligands, polymers and supports are not required. Also, particles, it is important to maintain constant all other possible
we can tune the IL moieties and reaction conditions to obtain variables. Indeed, differences in catalytic activity could have a
monodisperse and catalytically active NPs of controlled size physicochemical origin [22,23], resulting from peculiar solvation
[7,11–13]. This is because imidazolium-based ILs exhibit a 3-D phenomena including specific interactions between the IL and
organisation in the liquid state due to an extended hydrogen-bond the substrate (H bonds, cation-p) [24–26], mass transfer factors
network of ionic channels, coexisting with non-polar domains cre- (viscosity, diffusivity) [22,27], and effects attributed to the highly
ated by the grouping of lipophilic alkyl chains. Consequently, ILs structured nature of ILs [14,28,29].
present specific solvation properties [14]. Polar substrates are It has been shown that it is possible to obtain ruthenium nano-
particles, RuNPs, of differing sizes from the decomposition of Ru
* Corresponding author. Fax: +33 472431795. (COD)(COT) in C1C4ImNTf2, by simply varying the experimental
E-mail address: santini@cpe.fr (C.C. Santini). conditions [16]. In this work, we use the catalytic hydrogenation

0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.07.018
100 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107

of 1,3-cyclohexadiene, CYD, and cyclohexene, CYE, as probes for carbon). Binding energies were determined with an accuracy of
the relationship between size and catalytic performance of tailor- ±0.2 eV. The data were analysed using Casa-XPS (v 2.3.13) employ-
made and size-controlled RuNPs, generated in the ionic liquid, all ing a Shirley background subtraction prior to fitting and a peak
other physicochemical variables being constant. shape with a combination of Gaussian and Lorentzian (30%
Lorentzian). High-resolution spectra were acquired in the region
2. Experimental of Ru 3p as the Ru 3d region overlaps with the C 2p region of the
residual ionic liquid.
2.1. Materials and methods
2.5. Preparation of catalytic experiments
All operations were performed in the strict absence of oxygen
and water under a purified argon atmosphere using glovebox (Jac- In Table 1 are collected all data concerning the studied hydroge-
omex or MBraun) or vacuum-line (Schlenk) techniques. The ionic nation reaction. In columns 1–3, size and dispersion values of
liquid, C1C4ImNTf2 [30], and the complex, [Ru(COD)(COT)] [31], (Ru0), (Ru25) and (Ru50) are reported.
were synthesised as reported. The halide content of the ionic liquid Each solution of NP was produced from the decomposition of
was under 200 ppm (E.A.) and water under 5 ppm (limit of Karl 43.0 mmol L1 solution of Ru(COD)(COT) as described in Section
Fischer titration). Elemental analyses were performed at the CNRS 2.2 and shown in column 4.
Central Analysis Department of Solaize. For each different temperature of decomposition, a different
1-Methylimidazole (>99%) was purchased from Aldrich and dis- size of NP is obtained: at 0 °C, 1.1 nm (Ru0); at 25 °C, 2.3 nm
tilled prior to use. Chlorobutane (>99%, Aldrich) and lithium bis(tri- (Ru25); and at 50 °C, 2.9 nm (Ru50), shown in columns 1 and 2.
fluoromethanesulfonyl)imide (Solvionic) were used without The value of dispersion, D, describing the ratio of surface sites
further purification. Rus to total number of ruthenium atoms, varies with the NP size
and is given for each size of NP in column 3. Using this and the
known concentration of ruthenium (43.0 mmol L1), it is possible
2.2. Catalyst synthesis
to calculate for each size of catalyst the concentration of Rus (col-
umn 5). For example, for Ru0 (1.1 nm), we have a dispersion of 82%
A solution of (g4-1,5-cyclooctadiene)(g6-1,3,5-cyclooctatri-
and therefore a Rus concentration of 35.2 mmol L1.
ene)ruthenium(0) Ru(COD)(COT) (43 mmol L1) in the ionic liquid
To prepare the catalytic mixtures, we must have the same num-
1-butyl-3-methylimidazolium bis(trifluoromethylsulphonyl)imide
ber of catalyst sites. For this reason, we dilute the most concen-
(C1C4ImNTf2) was transferred under argon to a glass autoclave,
trated solutions, i.e. Ru0 and Ru25, to match the least
the temperature of which was controlled with aid of a thermostatic
concentrated, i.e. Ru50, by the addition of the appropriate amount
bath (0, 25, 50 or 75 °C). Once stabilised, the argon atmosphere was
of pure IL.
evacuated and replaced with molecular hydrogen (4 bar) without
Ru50 has a concentration of Rus of 18.5 mmol L1. A 5-mL sam-
stirring. The yellow solution turned black over time (up to 3 days)
ple of this solution therefore contains 9.52  105 mol of Rus.
as RuNPs were generated releasing cyclooctane (COA) as the only
Ru25 has a concentration of Rus of 22.9 mmol L1. A 3.74-mL
by-product. The resulting solutions were treated under dynamic
sample of this solution is therefore taken, containing
vacuum during a period of 6 h to remove all H2 and cyclooctane.
9.52  105 mol of Rus, and then diluted with 1.26 mL of pure IL
The black solutions could then be stored under argon atmosphere
to make a 5-mL solution.
with long-term stability (at least 6 months – no precipitation, coa-
Ru0 has a concentration of Rus of 35.2 mmol L1. A 2.58-mL
lescence or agglomeration – verified by TEM).
sample of this solution therefore contains 9.52  105 mol of Rus
diluted with 2.42 mL of pure IL to make a 5-mL solution (columns
2.3. Determination of particle size by TEM 6–8).
To each of these 5-mL catalyst/IL solutions, weighing 7.0 g (col-
Transmission electron microscopy (TEM) experiments were umn 9), is added 0.78 g of CYD (column 10). This gives catalytic
performed directly in the IL media. A thin film of RuNP solution mixtures with constant substrate/catalyst ratios of 105 (column
in IL was deposited on a carbon film supported by a copper grid. 11) and constant substrate/IL ratios of 0.59 (column 12) ensuring
Conventional TEM micrographs were obtained at the Centre Tech- identical viscosities and a single phase.
nologique des Microstructures, Université Claude Bernard Lyon 1, Using solutions prepared as described, the reaction is carried
Villeurbanne, France, using a Philips 120 CX electron microscope out in parallel in several 0.5-mL batches under 1.2 bars of pure
with acceleration voltage of 120 kV. Size distribution histograms molecular hydrogen, which are stirred and heated with the aid of
were constructed from the measurement of at least 200 different a thermostatic carousel, to ensure identical reaction conditions.
nanoparticles assuming a near spherical shape and random orien-
tation. High-resolution electron micrographs were obtained at the 2.6. Catalytic tests
‘‘TEMSCAN” centre of the Université Paul Sabatier Toulouse 3, Tou-
louse, France, using a JEOL JEM 200CX electron microscope with Catalytic solutions were made as described in Section 2.5 in a
acceleration voltage of 200 kV. glove box and left stirring for 12 h in a closed system to ensure
homogeneity. Aliquots of 0.5 mL were transferred to identical
2.4. XPS Schlenk tubes containing cross-shaped magnetic stirrer bars. The
argon atmosphere was removed, and the solution was degassed
X-ray photoelectron spectroscopy was performed in a Kratos in vacuo whilst cooling in liquid nitrogen (196 °C). For reactions
Axis Ultra DLD spectrometer, using a monochromated Al Ka X- at 30 °C, six of these Schlenk tubes were placed in a thermostatic
ray with a pass energy of 20 eV and a coaxial charge neutraliser. carousel to ensure identical temperature and stirring conditions.
The base pressure in the analysis chamber was better than After 30 min, when the temperature had stabilised, the Schlenk
5  108 Pa. XPS spectra of Ru3p, C1s, Si2p and O1s levels were tubes were opened to 1.2 bars of H2. After t minutes, a Schlenk tube
measured at a normal angle with respect to the plane of the sur- was isolated and opened to air, releasing the H2 atmosphere thus
face. High-resolution spectra were corrected for charging effects quenching the reaction. The solution was entirely dissolved in
by assigning a value of 284.6 eV to the C1s peak (adventitious 10 mL of acetonitrile containing a 1 M concentration of toluene.
 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107 101

Table 1
Calculations for the composition of the catalytic systems. Column 1 – name of catalyst, column 2 – average RuNP diameter measured by TEM, column 3 – calculated dispersion,
column 4 – initial Ru(COD)(COT) concentration, column 5 – consequent Rus concentration, column 6 – volume of IL/RuNP solution, column 7 – volume of pure IL added, column 8
– consequent number of moles of Rus in the 5 mL mixture, column 9 – mass of IL, column 10 – mass of substrate, column 11 – substrate/catalyst ratio, column 12 – substrate/IL
ratio.

1 2 3 4 5 6 7 8 9 10 11 12
d D [Ru] (mmol L1) [Rus] (mmol L1) Vol. IL–RuNP Vol. IL pure (mL) Rus/105 (mol) m (IL/g) m (CYD/g) CYD/Rus CYD/IL
(nm) (%) solution (mL)
Ru0 1.1 82 43.0 35.2 2.42 2.58 9.52 7.0 0.78 105 0.59
Ru25 2.3 53 43.0 22.9 3.74 1.26 9.52 7.0 0.78 105 0.59
Ru50 2.9 43 43.0 18.5 5.00 – 9.52 7.0 0.78 105 0.59

The composition of the mixture was determined by gas-phase CYD may be partially hydrogenated with high selectivity by molec-
chromatography using toluene as the internal standard. ular catalysts due to the reduced miscibility of cyclohexene (CYE)
in the medium [27,32,33]. Full hydrogenation would lead to cyclo-
2.7. Product quantification hexane (CYA), Scheme 1.

The products were quantitatively analysed by gas chromatog- 3.2. Solubility

raphy on a HP-6890 chromatograph equipped with a flame ioni-
sation detector (FID) and a HP-1 (crosslinked methylsiloxane) The solubilities of the substrate and potential products may
column (L: 30 m, int: 0.32 mm, film thickness: 0.25 lm). The play an important role in the activity and/or selectivity of the sys-
injector and detector temperature was 270 °C, and the injection tem. For example, selective hydrogenation of butadiene to butenes
volume was 1 lL. The programme was as follows: initial temper- has been performed by Dupont’s group in ionic liquids due to the
ature 70 °C for 13.5 min; ramp 40 °C/min to 250 °C, hold 2 min. difference in solubility of the partially hydrogenated product
[34]. The same group has also described the possibility of extract-
ing cyclohexene during benzene hydrogenation using this solubil-
2.8. Density and viscosity
ity difference [35]. For this reason, solubilities of CYD, CYE and CYA
are measured. It is found that the solubility of the hydrogenated
The mixtures of IL and CYD at different compositions were pre-
products (6 ± 1% wt – CYE, 4 ± 1% wt – CYA) is much lower than
pared gravimetrically following the procedure already described
that of CYD (12 ± 2% wt); therefore, the medium may tend to a
[27]. The viscosity of the mixture was measured at 298.15 K
biphasic system during the course of the reaction. As a result, the
(controlled to within ±0.005 K and measured with the accuracy
collection of aliquots from a single batch would render inaccurate
better than ±0.05 K) using a rolling-ball viscometer from Anton
results. Consequently, each point recorded in this work corre-
Paar, model AMVn [27]. The overall uncertainty of the viscosity
sponds to a separate experiment, quenched after time t by opening
is estimated as ±2.0%. The densities of the mixtures, necessary to
the reaction vessel to air, thus releasing the hydrogen and dissolv-
calculate the viscosities, were measured in an Anton Paar vibrating
ing the catalytic system entirely in a 1 M solution of toluene in ace-
tube densimeter model 512 P, at 298.15 K (measured by a cali-
tonitrile for gas-phase chromatography.
brated PRT with an accuracy of ±0.02 K). The overall uncertainty
of the density is estimated as ±0.01%.
3.3. Viscosity

2.9. Solubility Thermophysical properties of the reaction medium such as den-

sity and viscosity may also influence the catalytic performance. We
To measure the solubility, 1 mL of the substrate was stirred have recently demonstrated that reaction kinetics in IL media are
with the ionic liquid in a closed system at 298.15 K for 12 h and highly dependent on the mobility of molecules [27]. Consequently,
then left to settle for a further 2 h. A 0.1-mL sample of the ionic identical concentrations of substrate must be used in each case in
liquid phase was weighed, and its composition was determined order to maintain constant viscosity and eliminate effects due to
by GC using the procedure described in Section 2.6. Tests were mass transport. Furthermore, knowledge of the viscosity is very
repeated four times for each substrate to guarantee reproducibility. important from engineering point of view as it plays a major role
in stirring, mixing and pumping processes. The densities and vis-
3. Results and discussion cosities of the pure C1C4ImNTf2 and those of the mixtures with
CYD were measured at different molar ratio CYD/IL (R) at 25 °C
3.1. Choice of substrate and atmospheric pressure. The results are presented in Table 2.
As can be seen, the viscosity of the mixtures of CYD in IL varies
The substrate investigated is the conjugated diene, 1,3-cyclo- greatly with the concentration of CYD. From the Stokes–Einstein
hexadiene (CYD). It has been shown that in ionic liquid media,

Table 2
Density, q, and viscosity, g, of CYD–IL mixtures of different compositions. xIL = molar
fraction of IL, R = molar ratio CYD/IL.

H2 R xIL q (g cm3) g (m Pa s)
+ 0.000 1.000 1.4376 ± 0.0001 48.5 ± 0.4
0.100 0.909 1.4202 ± 0.0001 44 ± 1
0.200 0.833 1.3999 ± 0.0001 37.0 ± 0.4
0.300 0.769 1.3874 ± 0.0003 33.3 ± 0.3
CYD CYE CYA 0.397 0.716 1.3718 ± 0.0001 31.0 ± 0.3
0.498 0.667 1.3597 ± 0.0001 24.8 ± 0.3
Scheme 1. 1,3-Cyclohexadiene and its hydrogenation products.
102 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107

relation, the diffusion coefficient, D, which reflects to the mobility 160

of molecules, varies inversely with g,
140 T = 0 °C
kT
D¼ ð1Þ T = 25 °C
6pgr s 120
T = 50 °C
100 T = 75 °C

Frequency
3.4. Catalyst characterisation
80
It has been previously demonstrated that the size of RuNPs
60
generated from the decomposition of [Ru(COD)(COT)] under H2
[36], may be governed by the degree of self-organisation of the 40
imidazolium-based ionic liquid in which they are formed: the
more structured the ionic liquid, the smaller the size [16]. Follow- 20
ing previously described methods, RuNPs are synthesised at 0 °C,
0
25 °C, 50 °C and 75 °C in an attempt to obtain a selection of mono-
1 2 3 4 5
disperse sizes of RuNP in the same IL.
Size/ nm

3.4.1. TEM Fig. 2. Comparative size distribution histograms for RuNPs prepared in C1C4Im NTf2
Analysis of the suspensions obtained by transition electron at different temperatures.
microscopy allows the determination of the sizes generated:
1.1 ± 0.2 nm, 2.3 ± 0.3 nm, 2.9 ± 0.4 nm and 3.1 ± 0.7 nm, for RuNPs interplanar distances match with the hcp crystalline phase of Ru
generated at 0 °C (Ru0), 25 °C (Ru25), 50 °C (Ru50) and 75 °C (Ru75), (see Supplementary information). For Ru0, only a larger NP of
respectively, Fig. 1. As can be seen from the TEM image of Ru75 and 2 nm is observed by HREM, probably due to the difficulty in
the consequent size distribution histogram Fig. 2, the size of these observing the smallest NPs with limited contrast although may
NPs does not vary significantly compared to those of Ru50 although be indicative of a lower degree of crystallinity in very small RuNPs,
a poorer size control (wider distribution) is apparent. For this rea- as already observed by reverse Monte Carlo simulations.[37].
son, these NPs are not used in catalytic tests. High-resolution elec-
tron microscopy reveals the crystalline nature of the RuNPs formed 3.4.2. XPS
through elucidation of the crystal planes. The Fourier transform In order to establish the oxidation state of the RuNPs, X-ray
images of the HREM have been exploited and indicate that the photoelectron spectroscopy (XPS) is performed. Due to the weak

Fig. 1. Transition electron micrograph of RuNPs and high-resolution electron micrograph examples showing crystallinity for Ru0 (top left), Ru25 (top right), Ru50 (bottom left)
and Ru75 (bottom right).
 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107 103

Ru 3p1/2 Ru 3p3/2 Ru 3p1/2 Ru 3p3/2

Ru 3p3/2

Ru 3p1/2

490 480 470 460 450 490 480 470 460 450 490 480 470 460 450
Binding Energy / eV Binding Energy / eV Binding Energy / eV

Fig. 3. XPS spectra of the Ru 3p region after NP filtration onto SiO2, experimental data and fitted peaks. Left Ru0, middle Ru25 and right Ru50.

100%
concentration of the solution and the penetration limit of the
90%
X-rays in the solution (maximum depth of 10 nm), no peaks corre-
sponding to Ru binding energies are observed when the analyses 80%
are carried out directly on the RuNP/IL solutions. Samples are 70%

Dispersion
therefore prepared by filtering the RuNPs onto silica under inert 60%
atmosphere and eliminating as much IL as possible. The resulting
50%
spectra of the Ru 3p region are depicted in Fig. 3. It is clear that
in each case, fine peaks are observed, indicating the presence of 40%
only one Ru species. The low 3p3/2 binding energy observed in each 30%
case, 460.3 eV, and doublet separation of 22.2 eV correspond clo- 20%
sely to metallic zero-valent ruthenium, often reported with a
10%
3p3/2 binding energy of around 461 eV [38]. The small difference
may be attributed to the presence of small crystallites, which tend 0%
0 0.5 1 1.5 2 2.5 3 3.5
to exhibit lower binding energies than bulk metal. Indeed, as re-
cently shown for AuNPs [39], the d band narrows with decreasing Diameter / nm
particle size and shifts towards the Fermi level. Fig. 4. Curve of dispersion D against mean diameter of crystalline hcp RuNPs.

3.5. Rus concentration

in several 0.5-mL batches under 1.2 bars of pure molecular hydro-
Maintaining constant the initial ratio of substrate to catalyst is gen, which are stirred and heated with the aid of a thermostatic
imperative. In NP catalysis, as in heterogeneous catalysis, only the carousel, to ensure identical reaction conditions.
atoms at the surface (Rus) take part in reaction. The dispersion (D)
presents the ratio between surface atoms, Rus, and the total num-
3.7. Catalytic activity
ber of atoms, RuT, (D = Rus/RuT) and varies with the size of the NPs,
smaller particles of course having a larger percentage of surface
It can be seen in Table 3 (Experiments 1–3) that the largest NPs
atoms. The different dispersion values must therefore be taken into
(2.9 nm) are the most active in the hydrogenation of CYD. In a sim-
account for each size of nanoparticle formed.
ilar fashion, the three sizes of RuNP are tested in the catalytic
Ruthenium is known to exhibit a hexagonal close-packed crys-
hydrogenation of cyclohexene (CYE) to cyclohexane (CYA) to
tal structure, with the following lattice parameters: a: 270.59 pm,
establish whether a difference in activity is also apparent in the
b: 270.59 pm, c: 428.15 pm, a: 90°, b: 90°, c: 120° [40]. Using these
case of a monoene. The results tabulated in Table 4 show that no
parameters, SYBYL software can be applied to extrapolate the lat-
substantial difference in activity is observed unlike the case of
tice until the measured diameters in order to model the structure
the conjugated diene CYD.
of the different size NPs, assuming crystallinity. It is seen that crys-
In the case of in the catalytic hydrogenation of cyclohexene
talline hexagonal close-packed RuNPs would adopt a truncated
(CYE) to cyclohexane (CYA), the TOF value for Ru0 is comparable
hexagonal bipyramid form, with two symmetric hexagonal faces
to that reported by Roucoux’s group with RuNPs (2.5 nm,
(0 0 0 1) and 12 irregular and uneven trapezoid faces (1 0 (1) 1)
TOF = 34) in water in the presence of cyclodextrins [42], although
[41]. From these findings, a curve of D with respect to diameter
clearly here the experimental conditions are entirely different.
can be plotted and then used to estimate D for each size of nano-
According to literature results, the catalytic activity of NPs
particle, Fig. 4.
depends on their size and generally reaches a maximum for those
of around 3 nm [43,44]. Here, only a size effect on activity was
3.6. Catalysis mixture preparation observed for the case of CYD. The fact that the hydrogenation of
CYD is faster with larger NPs can be related to two factors:
In Table 2 (Section 2.6) are collected all data concerning the
studied hydrogenation reaction. The experimental conditions are (1) Larger NPs present the appropriate number of neighbouring
established to ensure identical concentrations of Rus (column 8) surface sites to facilitate the p-bond activation of the conju-
and substrate (columns 9 and 10), permitting as a result both a gated system [45].
constant substrate/catalyst ratio (column 11) and a constant sub- (2) Through this p-bonding activation similar to benzene, 1,3-
strate/IL ratio (column 12) hence constant viscosity. Using solu- cyclohexadiene would lose part of its resonance energy
tions prepared as described, the reaction is carried out in parallel and react more readily [41].
104 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107

Table 3
Conversion, selectivity, turnover number and turnover frequency for the hydrogenation of CYD at 30 °C. Experiments 1–3 using 1.2 bar H2. Experiments 4 and 5 using 4 bar H2.

Experiment number Catalyst (nm) Pressure H2 (bar) Conversion at 90 min (%) Selectivity CYE (%) TON TOF (h1) Size after catalysis (nm)
1 Ru0 (1.1) 1.2 66 ± 5 97 70 ± 5 46 ± 3 1.3 ± 0.4
2 Ru25 (2.3) 1.2 75 ± 5 86 79 ± 5 53 ± 3 2.1 ± 0.5
3 Ru50 (2.9) 1.2 83 ± 5 80 87 ± 5 58 ± 3 2.7 ± 0.5
4 Ru0 (1.1) 4.0 57 ± 5 92 59 ± 5 40 ± 3 –
5 Ru50 (2.9) 4.0 73 ± 5 80 77 ± 5 51 ± 3 –

Table 4 the hydrogenation of 1,3-cyclohexadiene on metallic surfaces

Conversion, turnover number and turnover frequency for the hydrogenation of CYE at [46,47], and thus, rapid consecutive hydrogenation of both double
30 °C under 1.2 bar H2.
bonds leading to the fully hydrogenated CYA may be envisaged. In
Experiment Catalyst Conversion at 90 min TON TOF Fig. 6 are represented simplified SYBYL models of CYD molecules
number (nm) (%) (h1) coordinating to the surface of perfectly crystalline RuNPs of calcu-
6 Ru0 (1.1) 61 ± 5 64 ± 5 43 ± 3 lated average diameter 1.3 nm and 2.8 nm. This illustrates nicely
7 Ru25 (2.3) 64 ± 5 67 ± 5 45 ± 3 the greater facility of planar coordination to faces of the larger
8 Ru50 (2.9) 67 ± 5 70 ± 5 47 ± 3
NPs and could explain the lower selectivity of the larger RuNPs de-
spite identical reaction conditions. Likewise, in the hydrogenation
The coordination of monoenes such as CYE does not necessitate of 1,3-butadiene or 1-hexyne, the selectivity of small NPs towards
large surfaces, explaining the less pronounced size effect in this 1-butene or 1-hexene versus butane or hexane is still higher than
case. that of larger NPs [45–48]. In this work, in the hydrogenation of
CYD, the selectivity in CYE versus CYA drops from 97% to 80% when
the RuNP size increases from 1.1 to 2.9 nm.
3.8. Catalytic selectivity Our hypothesis is based on idealised particle shape, which is not
likely to exist in reality. Nonetheless, it is widely accepted that
In the hydrogenation of CYD, CYE is obtained as the major prod- large NPs are more likely to present larger open facets where pla-
uct. Interestingly, the selectivity for CYE diminishes with increas- nar p-coordination of diene substrates can occur, whereas small
ing NP size. Indeed, for Ru0, selectivity for CYE is 100% at low NPs are often reported to be amorphous, therefore presenting no
conversion and only slightly diminishes at high conversion (97%). open facets, making this planar coordination even less likely [41].
In contrast, for Ru50, the hydrogenation is unselective even at In studies of CO hydrogenation on RhNPs, the difference in reac-
low conversion, Fig. 5. tivity with size was related to the increasing probability of finding
Assuming highly crystalline particles with hcp structure, a par- step sites with increasing NP size [49–51]. However, for RuNPs of
ticle of diameter 1.1 nm would have the vast majority of catalytic less than 3 nm, as reported here, calculations have shown that such
surface atoms occupying vertex or edge positions. Such vertex step sites are not likely to exist [52].
ruthenium atoms Ruv, which under H2 atmosphere are ligated by The highly selective hydrogenation of 1,3-cyclohexadiene to
hydrides, may coordinate one C@C double bond of CYD. The prod- cyclohexene has also been performed with PdNPs in organic and
uct of the subsequent hydrogenation is CYE, which must undergo a IL media [33,53,54]. The high selectivity results from the intrinsic
second coordination to give the fully hydrogenated CYA. Similarly, properties of Pd metal and its small NP size [55].
for a larger particle of average diameter 2.9 nm assuming high
crystallinity and an hcp structure, it is evident that most of the cat- 3.9. Hydrogen effect
alytically active surface ruthenium atoms are found in facial posi-
tions, Ruf. Indeed, here, such crystallinity has already been According to the literature, the hydrogenation of olefins in ILs is
observed by HREM, Fig. 1. Ruf may hydrogenate the olefin using often biphasic in its nature [56–59], due to the poor solubility of
the mechanism previously discussed, but due to the planar hydrogen and olefins in these media [60–63]. Therefore, the diffu-
arrangement of Ruf, another mechanism may be envisaged involv- sion process of the substrate or H2 may limit the rate of
ing the double coordination of the diene, as generally found during hydrogenation.
To find out whether H2 is in fact a rate-limiting reagent, exper-
iments varying the H2 pressure are performed. Increasing H2 pres-
sure to 4 bars in the case of Ru0 and Ru50 is seen to affect neither
the activity nor selectivity in a substantial manner, Table 3, Exper-
iments 4 and 5. This is similar to results reported by Dupont’s
group who observed that the reaction rate does not depend on
the H2 pressures in C1C4ImBF4 [64].
In parallel, it is generally reported that a higher H2 concentra-
tion should influence the activity of the large rather than small par-
ticles, as the H2 storage capacity is related to the particle volume;
therefore, large particles may experience an increase in the avail-
ability of subsurface hydrogen [44].
In reality, little difference in activity is observed in either case.
This proves that the rate is not dictated by the availability and
adsorption of H2, in agreement with observations of labile surface
hydrides on the NP surface [17], but by the mobility and absorption
of the substrate [26]. Indeed, it is highly likely that the surface of
Fig. 5. Selectivity for cyclohexene as a function of conversion for the three different the NPs is already saturated with adsorbed H2 at the temperature
sizes of RuNPs. of the reaction [17].
 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107 105

Fig. 6. SYBYL representations of CYD coordinated to the face of highly crystalline RuNPs of mean diameter 1.3 nm (left) and 2.8 nm (right).

Fig. 7. Transition electron micrographs of RuNPs after CYD hydrogenation for Ru0 (left), Ru25 (middle) and Ru50 (right).

Table 5
Recycling of the catalyst Ru0. Conversion, selectivity, turnover number and turnover frequency for the hydrogenation of CYD at 30 °C under 1.2 bar H2. Products removed under
vacuum after each run and analysed by GC.

Experiment number Cycle Conversion at 90 min (%) Selectivity CYE (%) TON TOF (h1) Size after catalysis (nm)
1 1st 66 97 70 46 1.3 ± 0.4
9 2nd 73 95 76 51 –
10 3rd 69 94 73 49 –
11 4th 68 89 71 47 –
12 5th 64 86 67 45 –
13 6th 65 86 68 45 1.8 ± 0.5

3.10. Recycling the most selective catalyst, Ru0, recycling experiments are per-
formed, by extracting in vacuo and quantifying the volatiles after
TEM images of the reaction medium after hydrogenation, Fig. 7, each 90-min run. More CYD is then added for hydrogenation. From
show that the average size measured (Table 3) does not differ the results, Table 5 and Fig. 8, it can be seen that both the activity
greatly from the original size, in accordance with the stability of and selectivity remain high after five recycles, diminishing only
RuNPs in ILs under molecular hydrogen [17]; however, the size dis- slightly with each run. This small decrease of course is attributable
tribution is larger, probably as an effect of stirring [65]. to the gradual coalescence of the NPs, leading to a diminution in
The apparent resistance to coalescence of the NPs means that the number of active surface sites and larger, less selective NPs.
they may be tested for their recyclability. Consequently, using Indeed, TEM images obtained of the NPs after all recycling
106 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107

55 100 References
TOF

Selectivity for cyclohexene / %

Selectivity for cyclohexene [1] M. Valden, X. Lai, D.W. Goodman, Science 281 (1998) 1647.
50
90 [2] A.T. Bell, Science 299 (2003) 1688–1691.
[3] G. Schmid, Nanoparticles: From Theory to Application, Wiley-VCH, Weinheim,
45 2004.
[4] K. Lee, M. Kim, H. Kim, J. Mater. Chem. 20 (2010) 3791–3798.
TOF / h-1

80
[5] A.Z. Moshfegh, J. Phys. D: Appl. Phys. 42 (2009) 233001. 233032.
40 [6] G.A. Somorjai, J.Y. Park, Top. Catal. 49 (2008) 126–135.
[7] H. Boennemann, K.S. Nagabhushana, in: B. Corain, G. Schmid, N. Toshima
70 (Eds.), Metal Nanoclusters in Catalysis and Materials Science: The Issue of Size
35 Control, Elsevier B.V., Amsterdam, pp. 21–48.
[8] H. Bönnemann, K.S. Nagabhushana, R.M. Richards, in: D. Astruc (Ed.),
60 Nanoparticles and Catalysis, Wiley-VCH, Weinheim, 2008, pp. 49–92.
30
[9] H. Tada, T. Kiyonaga, S. Naya, Chem. Soc. Rev. 38 (2009) 1849.
[10] D. Astruc, Nanoparticles and Catalysis, Wiley-VCH, Weinheim, 2008.
25 50 [11] J. Dupont, J.D. Scholten, Chem. Soc. Rev. 39 (2010) 1780–1804.
1 2 3 4 5 6 [12] J. Dupont, Nanoparticles and Catalysis, Wiley-VCH, Weinheim.
[13] T. Gutel, C.C. Santini, K. Philippot, A. Padua, K. Pelzer, B. Chaudret, Y. Chauvin,
Cycle J.-M. Basset, J. Mater. Chem. 19 (2009) 3624–3631.
[14] A.A.H. Padua, M.F. Costa Gomes, J.N.A. Canongia Lopes, Acc. Chem. Res. 40
Fig. 8. Evolution of TOF and selectivity with catalyst recycling. (2007) 1087–1096.
[15] J.N. Canongia Lopes, M.F. Costa Gomes, A.A.H. Padua, J. Phys. Chem. B 110
(2006) 16816–16818.
experiments showed that the NPs undergo coalescence to attain an
[16] T. Gutel, J. Garcia-Anton, K. Pelzer, K. Philippot, C.C. Santini, Y. Chauvin, B.
average diameter of 1.8 ± 0.5 nm, approaching the size of Ru25 and Chaudret, J.-M. Basset, J. Mater. Chem. 17 (2007) 3290–3292.
of course presenting a similar selectivity. [17] P.S. Campbell, C.C. Santini, D. Bouchu, B. Fenet, K. Philippot, B. Chaudret, A.A.H.
Padua, Y. Chauvin, Phys. Chem. Chem. Phys. 12 (2010) 4217–4223.
[18] C. Vollmer, E. Redel, K. Abu-Shandi, R. Thomann, H. Manyar, C. Hardacre, C.
4. Conclusion Janiak, Chem. Eur. J. 16 (2010) 3849–3858. S3849/3841-S3849/3834.
[19] E. Redel, J. Kraemer, R. Thomann, C. Janiak, J. Organometal. Chem. 694 (2009)
1069–1075.
In this work, the catalytic hydrogenation of 1,3-cyclohexadiene, [20] M.H.G. Prechtl, J.D. Scholten, J. Dupont, J. Mol. Catal. A 313 (2009) 74–78.
CYD, and cyclohexene, CYE, in C1C4ImNTf2, was used as a probe for [21] B. Leger, A. Denicourt-Nowicki, H. Olivier-Bourbigou, A. Roucoux, Inorg. Chem.
47 (2008) 9090–9096.
the relationship between size and catalytic performance of RuNPs.
[22] P. Wasserscheid, T. Welton, Ionic Liquids in Synthesis, Wiley-VCH, Weinheim,
Firstly, tailor-made, size-controlled and zero-valent RuNPs 2008.
(1–3 nm) were generated from the decomposition of [Ru(COD) [23] H. Olivier-Bourbigou, C. Vallee, Multiphase Homogeneous Catalysis, Wiley-
(COT)] under H2 in C1C4ImNTf2, by varying the experimental condi- VCH, Weinheim, pp. 413–431.
[24] J. Dupont, P.A.Z. Suarez, R.F. De Souza, R.A. Burrow, J.-P. Kintzinger, Chem. Eur.
tions. RuNPs were fully characterised in situ by TEM and HREM to J. 6 (2000) 2377–2381.
determine their sizes and demonstrate their degree of crystallinity [25] J. Lachwa, I. Bento, M.T. Duarte, J.N.C. Lopes, L.P.N. Rebelo, Chem. Commun.
and ex situ by XPS to verify their zero oxidation state. (2006) 2445–2447.
[26] T. Gutel, C.C. Santini, A.A.H. Padua, B. Fenet, Y. Chauvin, J.N. Canongia Lopes, F.
Secondly, all catalytic reaction compositions were carefully cal- Bayard, M.F. Costa Gomes, A.S. Pensado, J. Phys. Chem. B 113 (2009) 170–177.
culated in order that all parameters except particle size were main- [27] P.S. Campbell, A. Podgorsek, T. Gutel, C.C. Santini, A.A.H. Padua, M.F. Costa
tained constant, i.e. Rus concentration, CYD/Rus ratio and CYD/IL Gomes, F. Bayard, B. Fenet, Y. Chauvin, J. Phys. Chem. B 114 (2010) 8156–8165.
[28] A. Mele, G. Romano, M. Giannone, E. Ragg, G. Fronza, G. Raos, V. Marcon,
ratio, the latter governing solvation phenomena and mass transfer Angew. Chem. Int. Ed. 45 (2006) 1123–1126.
factors (viscosity and diffusivity). It was found that for catalytic [29] A. Triolo, O. Russina, H.-J. Bleif, E. Di Cola, J. Phys. Chem. B 111 (2007) 4641–4644.
hydrogenations of CYD and CYE, the activity of catalyst increases [30] L. Magna, Y. Chauvin, G.P. Niccolai, J.-M. Basset, Organometallics 22 (2003)
4418–4425.
with the NP size in agreement with the literature results on heter-
[31] P. Pertici, G. Vitulli, Inorg. Synth. 22 (1983) 176–181.
ogeneous catalysts. In contrast to activity, in the hydrogenation of [32] Y. Chauvin, S. Einloft, B.H. Olivier, Ind. Eng. Chem. Res. 34 (1995) 1149–1155.
CYD, the selectivity for CYE versus CYA drops from 97% to 80% [33] J. Huang, T. Jiang, B. Han, H. Gao, Y. Chang, G. Zhao, W. Wu, Chem. Commun.
(2003) 1654–1655.
when the RuNP size increases from 1.1 to 2.9 nm.
[34] A.P. Umpierre, G. Machado, G.H. Fecher, J. Morais, J. Dupont, Adv. Synth. Catal.
Both results, activity and selectivity, are in agreement with a 347 (2005) 1404–1412.
mechanism involving a p-bond activation and a double coordina- [35] E.T. Silveira, A.P. Umpierre, L.M. Rossi, G. Machado, J. Morais, G.V. Soares, I.J.R.
tion of diene substrates, necessitating several neighbouring surface Baumvol, S.R. Teixeira, P.F.P. Fichtner, J. Dupont, Chem. Eur. J. 10 (2004) 3734–
3740.
atoms only found in facial positions on the larger NPs. Further- [36] K. Philippot, B. Chaudret, C.R. Chimie 6 (2003) 1019–1034.
more, these RuNPs show a high level of recyclability with neither [37] N. Bedford, C. Dablemont, G. Viau, P. Chupas, V. Petkov, J. Phys. Chem. C 111
loss of activity nor significant agglomeration. (2007) 18214–18219.
[38] R. Nyholm, N. Martensson, J. Phys. Chem. 13 (1980) L279.
To conclude, in olefin hydrogenation by ruthenium nanoparti- [39] J.A. Van Bokhoven, J.T. Miller, J. Phys. Chem. C 111 (2007) 9245.
cles in ionic liquid media, both reactivity and selectivity are signif- [40] V.A. Finkel, M.I. Palatnik, G.P. Kovtun, Phys. Met. Metall. 32 (1971) 231.
icantly dependent on the nanoparticle size. [41] R.A. Van Santen, Acc. Chem. Res. 42 (2009) 57–66.
[42] A. Denicourt-Nowicki, A. Ponchel, E. Monflier, A. Roucoux, Dalton Trans. 48
(2007) 5714–5719.
Acknowledgments [43] D.Y. Murzin, Chem. Eng. Sci. 64 (2009) 1046–1052.
[44] A. Binder, M. Seipenbusch, M. Muhler, G. Kasper, J. Catal. 268 (2009) 150–155.
[45] N. Semagina, A. Renken, L. Kiwi-Minsker, J. Phys. Chem. C 111 (2007) 13933–
P.S.C. acknowledges the Ph.D. grant attributed by the Ministère 13937.
de l’Enseignement Supérieur et de la Recherche, France, and the EU [46] M. Saeys, M.-F. Reyniers, M. Neurock, G.B. Marin, Surf. Sci. 600 (2006) 3121–
transnational access programme. A.P. thanks the post-doctoral 3134.
[47] W.L. Manner, G.S. Girolami, R.G. Nuzzo, J. Phys. Chem. B 102 (1998) 10295–
grant by the project ANR CALIST. This work has been funded by 10306.
CNRS and ANR (ANR Project CALIST, ANR-07-CP2D-02-03). [48] J. Silvestre-Albero, G. Rupprechter, H.-J. Freund, J. Catal. 240 (2006) 58–65.
[49] H. Arakawa, K. Takeuchi, T. Matsuzaki, Y. Sugi, Chem. Lett. 13 (1984) 1607.
[50] T. Hanaoka, H. Arakawa, T. Matsuzaki, Y. Sugi, K. Kanno, Y. Abe, Catal. Today 58
Appendix A. Supplementary data (2000) 271.
[51] S. Zhou, H. Zhao, D. Ma, S. Miao, M. Cheng, X. Bao, Z. Phys. Chem. 219 (2005) 949.
[52] J. Gavnholt, J. Schiotz, Phys. Rev. B 77 (2008) 035404/035401–035404/035410.
Supplementary data associated with this article can be found, in [53] M.V. Seregina, L.M. Bronstein, O.A. Platonova, D.M. Chernyshov, P.M. Valetsky,
the online version, at doi:10.1016/j.jcat.2010.07.018. J.r. Hartmann, E. Wenz, M. Antonietti, Chem. Mater. 9 (1997) 923.
 P.S. Campbell et al. / Journal of Catalysis 275 (2010) 99–107 107

[54] J. Huang, T. Jiang, H. Gao, B. Han, Z. Liu, W. Wu, Y. Chang, G. Zhao, Angew. [60] P.J. Dyson, G. Laurenczy, C.A. Ohlin, J. Vallance, T. Welton, Chem. Commun.
Chem. Int. Ed. 43 (2004) 1397–1399. (2003) 2418–2419.
[55] G. Ertl, H. Knozinger, J. Weitkamp (Eds.), Handbook of Heterogeneous [61] J. Jacquemin, M.F. Costa Gomes, P. Husson, V. Majer, J. Chem. Thermodyn. 38
Catalysis, Wiley-VCH, Weinheim, 1997. (2006) 490–502.
[56] J. Dupont, G.S. Fonseca, A.P. Umpierre, P.F.P. Fichtner, S.R. Teixeira, J. Am. [62] J. Jacquemin, P. Husson, V. Majer, M.F. Costa Gomes, Fluid Phase Equilib. 240
Chem. Soc. 124 (2002) 4228–4229. (2006) 87.
[57] P.J. Dyson, Appl. Organomet. Chem. 16 (2002) 495–500. [63] J. Jacquemin, P. Husson, V. Majer, M.F. Costa Gomes, J. Solution Chem. 36
[58] P.J. Dyson, in: a.M.T.J. Mc Cleverty (Ed.), Comprehensive Coordination (2007) 967–979.
Chemistry II, Elsevier, Amsterdam, pp. 557–566. [64] P. Migowski, D. Zanchet, G. Machado, M.A. Gelesky, S.r.R. Teixeira, J. Dupont,
[59] P.J. Dyson, D. Zhao, Multiphase Homogeneous Catalysis, Wiley-VCH, Phys. Chem. Chem. Phys. 12 (2010) 6826–6833.
Weinheim, pp. 494–511. [65] D. Li, R.B. Kaner, J. Am. Chem. Soc. 128 (2006) 968–975.