Water-gas shift catalysts based on ionic liquid mediated supported Cu nanoparticles

Article abstract of DOI:10.1016/j.jcat.2010.09.019

The sorptive and catalytic properties of alumina supported Cu nanoparticles coated with a thin film of 1-butyl-2,3-dimethyl-imidazolium trifluoromethane sulfonate (BDiMIm) for low-temperature water-gas shift have been explored. For uncoated catalysts, the rate per gram catalyst passed through a maximum with increasing concentration of oxygen on the Cu surface. For reduced catalysts, the presence of oxygen facilitates the dissociation of water, while in excess it decreases the reaction rates due to limiting the reactant concentration. Catalysts coated with ionic liquid showed a higher turn over frequency for the water-gas shift reaction at low temperatures compared to uncoated catalysts and to the best commercial systems, which is attributed to a higher concentration of water in the proximity of the active sites and to the facile decomposition of carboxyl intermediates by the interaction with the ionic liquid. In addition, the presence of the ionic liquid reduces the sorption strength of CO leading to a better balance of the reactants at the surface.

Full text of DOI:10.1016/j.jcat.2010.09.019

Journal of Catalysis 276 (2010) 280–291
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Water–gas shift catalysts based on ionic liquid mediated supported Cu nanoparticles
⇑
Richard Knapp, Sonja A. Wyrzgol, Andreas Jentys , Johannes A. Lercher
Lehrstuhl II für Technische Chemie, Technische Universität München, Lichtenbergstr. 4, D-85747 Garching, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 3 May 2010
Revised 17 September 2010
Accepted 19 September 2010
Available online 25 October 2010
The sorptive and catalytic properties of alumina supported Cu nanoparticles coated with a thin film of 1-
butyl-2,3-dimethyl-imidazolium trifluoromethane sulfonate (BDiMIm) for low-temperature water–gas
shift have been explored. For uncoated catalysts, the rate per gram catalyst passed through a maximum
with increasing concentration of oxygen on the Cu surface. For reduced catalysts, the presence of oxygen
facilitates the dissociation of water, while in excess it decreases the reaction rates due to limiting the
reactant concentration. Catalysts coated with ionic liquid showed a higher turn over frequency for the
water–gas shift reaction at low temperatures compared to uncoated catalysts and to the best commercial
systems, which is attributed to a higher concentration of water in the proximity of the active sites and to
the facile decomposition of carboxyl intermediates by the interaction with the ionic liquid. In addition,
the presence of the ionic liquid reduces the sorption strength of CO leading to a better balance of the reac-
tants at the surface.
Keywords:
Ionic liquids
Supported catalysts
Copper nanoparticles
EXAFS
XANES
In situ IR spectroscopy
Low-temperature water–gas shift
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
species can also be formed by CO2 hydrogenation acting as stable
spectator species. The elementary steps are compiled in Eqs. (2)–
The water–gas shift reaction (Eq. (1)) allows adjusting the CO to
H2 ratio in synthesis gas. Since the reaction is exothermic, the for-
mation of hydrogen is favored at low temperatures and, therefore,
the lower limit for the CO concentration achievable decreases with
decreasing temperature. For example, the exit stream after the
water–gas shift reaction at approximately 200 °C contains less
than 0.5% CO [1].
(6) [5]. Water and CO adsorb on the catalyst surface, and the disso-
ciation of the adsorbed water is reported to be the reaction with
the first significant energy barrier. Its height depends on the Cu
particle size and varies between 86 and 135 kJ molꢀ1 being about
seven times higher than the barrier to desorb molecular H2O
[4,5]. Thus, from the energetic point of view, the abstraction of
hydrogen from water seems to be the rate-controlling step [5].
The nature of the transition state and the energy barrier depend
on the catalyst and its surface properties. Adsorbed CO and OH re-
CO þ H2OꢀCO2 þ H2
D
Ho ¼ ꢀ41 kJ=mol
ð1Þ
*
act to COOH and the carboxyl species decompose to the products
In industrial implementations, the water–gas shift reaction is
performed in two steps, a high-temperature and a low-tempera-
ture reaction. The low-temperature water–gas shift reaction
(LTWGS) is performed at temperatures between 210 and 250 °C
over catalysts based on copper supported on zinc oxide. These cat-
alysts are pyrophoric and deactivate in the presence of liquid water
and by thermal sintering. Moreover, a clean feed is needed in order
to avoid deactivation by sulfur and halides. Since the reaction is
reversible, the rate of the CO conversion is reduced by the presence
of the products CO2 and H2 [2].
Two different reaction mechanisms, the associative and the sur-
face redox mechanisms have been discussed for LTWGS [3–10]. Re-
cent publications show that in the reaction pathway of the
associative mechanism *CO reacts with *OH groups to a formyl
(carboxyl) species (*COOH) [5,11]. Under these conditions, formate
H2 and CO2 desorbing in the last step from the catalyst surface. The
decomposition of *COOH is the second step with a high energy bar-
rier. The calculated potential energy diagram and the formed inter-
mediate species are depicted in Fig. S1.
H2OðgÞ þ ꢁꢀH2Oꢁ
ð2Þ
ð3Þ
ð4Þ
ð5Þ
ð6Þ
H2Oꢁ þ ꢁꢀ ꢁ OH þ Hꢁ
COðgÞ þ ꢁꢀꢁCO
ꢁOH þ ꢁCOꢀꢁCOOH þ ꢁ
2ꢁCOOHꢀCO2ðgÞ þ H2 þ 2ꢁ
The second reaction mechanism described in the literature is the
surface redox mechanism [7]. The first step is the adsorption of
water on the catalyst surface and its dissociation to oxygen and
⇑
Corresponding author. Fax: +49 8928913544.
E-mail address: andreas.jentys@ch.tum.de (A. Jentys).
0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.09.019

R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
281
2.2. Preparation of the supported catalysts
Metal cluster
Ionic liquid film
Support
Before use, the series of alumina supports was dried at 200 °C
for two hours. Cu was loaded onto the supports (5 wt.% metal) by
incipient wetness impregnation using an aqueous solution of cop-
per(II) nitrate trihydrate. After impregnation, water was removed
by freeze drying. The materials were calcined at 300 °C for 3 h
and at 450 °C for 4 h in synthetic air and subsequently reduced
at 215 °C for 3 h in hydrogen flow.
For preparing supported ionic liquid catalysts, the correspond-
ing Cu/Al2O3 catalysts were added to a solution of BDiMIm dis-
solved in methanol. The suspension was stirred at room
temperature for 10 min, and the volatile components were slowly
removed by freeze drying to give a free flowing black powder (Cu/
BDiMIm/Al2O3).
Fig. 1. Catalyst concept with metal clusters (1–2 nm) immobilized in thin films of
ionic liquid (2–5 nm).
hydrogen, followed by a bond formation between oxygen and ab-
sorbed CO. The rate-determining step depends on the partial pres-
sure of the reactants [8]. Hydrogen and CO2 adsorbed on Cu(1 0 0)
surfaces can form formate species, which block the active sites of
the catalyst. These formate species can undergo hydrogenation to
methanol, but since their surface concentration is low under indus-
trial conditions, the formation of methanol is not very important [9].
It is generally assumed that both reaction pathways may occur
under low-temperature water–gas shift conditions over copper-
based catalysts and as the reaction is reported to be structure sen-
sitive, the particle size should play a major role in the associative
mechanism [10].
The BET surface areas, pore volumes and pore radii of the
Al2O3 supports are listed in Table 1.
c-
2.3. Characterization
Modifying the local concentration of reactants and products at
the active sites, e.g., by using an environment with different solu-
bilities of the gases, appears to be an option to enhance the activity
of the catalysts. In the particular case, we intended to improve the
availability of water in the water–gas shift reaction by the presence
of the ionic liquid. Supported ionic liquid catalysts [12–15] are a
special variant of supported liquid phase catalysts [16] in which
the catalytically active components are immobilized in the ionic li-
quid. Due to the low vapor pressure of the ionic liquid, these cata-
lysts can be readily applied in gas phase reactions [17]. The
interactions of the ionic liquid with the support as well as with
the active component have been studied showing ordering phe-
nomena and domain formation of the ionic liquid [15,18,19]. Cata-
lysts with immobilized ionic liquids were successfully used in
reactions such as hydroformylation of olefins [12], achiral hydroge-
nation [20], the Heck reaction [21], hydroamination [13,14] and
hydrogenation [15]. A schematic drawing of a supported ionic li-
quid catalyst with metal nanoparticles as the active component
immobilized on an alumina support is depicted in Fig. 1.
The current contribution focuses on the synthesis and charac-
terization of catalysts based on ionic liquid (1-butyl-2,3-di-
methyl-imidazolium trifluoromethane sulfonate) mediated Cu
nanoparticles as catalysts for the low-temperature water–gas shift
reaction. The correlation between the oxidation state of the metal
and the reactivity in the low-temperature water–gas shift reaction
as well as the enhanced turn over frequencies of the coated system
at lower temperatures is addressed using in situ XAFS and IR spec-
troscopy in combination with investigating the activity for the
low-temperature WGS reaction. Furthermore, the influence of the
specific surface area of the support on the copper particle size
and the reduction degree are discussed.
The copper content of the supported catalysts was determined
by atomic absorption spectroscopy using a UNICAM 939 spectrom-
eter. The concentration of ionic liquid adsorbed on the surface was
determined by elemental analysis.
IR spectra were measured on a Bruker IFS 88 spectrometer
using a vacuum and a flow cell in transmission mode. The spectra
were recorded at a resolution of 4 cmꢀ1 in the region from 4000 to
400 cmꢀ1. The samples were pressed into self-supporting wafers
(of ca. 5 mg) and activated in vacuum for 1 h at 120 °C or in He flow
at 120 °C for 1 h. The reduction was carried out at 215 °C and
0.8 bar hydrogen for 60 min for the vacuum experiments and in a
hydrogen/helium mixture at 215 °C for 60 min for the experiments
under flow conditions. CO, CO2 and D2O were dosed at pressures of
0.01, 0.05, 0.1, 0.5, 1.0 and 1.5 mbar (note that for CO adsorbed on
the coated catalyst only adsorption at the four latter pressures is
reported) and room temperature. For the reaction, a gas mixture
of 2.5 mL minꢀ1 CO and 2.5 mL minꢀ1 H2O diluted in 20 mL minꢀ1
He was used.
The X-ray absorption spectra were collected at the beamline X1
at HASYLAB, DESY, Hamburg, Germany and at the beamline BM26A
at the ESRF, Grenoble, France. The Si(1 1 1) double-crystal mono-
chromator was detuned to 60% of the maximum intensity to min-
imize the intensity of the higher harmonics in the X-ray beam. The
catalysts were pressed into self-supporting wafers (ca. 150 mg),
and the X-ray absorption spectra were collected at the Cu K edge
(8979 eV) in He flow at liquid N2 temperature for EXAFS analysis
and at room temperature or under in situ conditions during activa-
tion to 250 °C in a H2/He mixture and reaction (150 °C with a CO,
He and water flow with the same composition as used for the
in situ IR experiments). Please note that prior to the in situ activity
tests, the samples were activated at 215 °C. The XAFS data were
analyzed using the Six Pack software [22]. For EXAFS analysis, the
scattering contributions of the background were removed from
2. Experimental
2.1. Materials
Table 1
BET surface areas, pore volumes and pore radii of used support materials.
The ionic liquid 1-butyl-2,3-dimethyl-imidazolium trifluorome-
thane sulfonate (BDiMIm) (99%) with a maximum water and halide
content of 48 and 103 ppm, respectively, was provided by Solvent
Innovation GmbH. Copper(II) nitrate trihydrate (99%) and metha-
Support
BET surface area
Pore volume
Pore radius
(nm)
(m2 gꢀ1
)
(mL gꢀ1
)
Al2O3-54
Al2O3-99
54
99
0.30
0.51
0.55
0.52
0.74
0.52
0.98
0.30
21.9
12.0
20.7
13.4
19.5
9.7
7/500 (bimodal)
3.5
nol (99.8%) were obtained from Aldrich. The
c-alumina supports
Al2O3-100
Al2O3-149
Al2O3-151
Al2O3-214
Al2O3-257
Al2O3-360
100
149
151
214
257
360
with pore volumes between 0.3 and 0.98 mL gꢀ1 and the Shift-
Max-240 catalyst were provided by Süd Chemie AG. All chemicals
were used as received. CO (purity 4.0), CO2 (purity 4.5), Ar (purity
5.0), H2 (purity 5.0) and N2 (purity 5.0) were purchased from
Westfalen.

282
R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
Table 2
the X-ray absorption by a third-order polynomial function. The
oscillations were weighted with k2 and Fourier-transformed within
the limit k = 3.5–12 Åꢀ1. The local environment of the Cu atoms
was determined from the EXAFS using the phase shift and ampli-
tude function for Cu–Cu and Cu–O calculated including multiple
scattering processes (FEFF version 8.40) [23,24].
For the analysis of the XANES, the spectra were normalized to
unity using the XANES dactyloscope software [25]. The position of
the edge was calibrated using the spectra of a Cu reference-foil
measured simultaneously.
To further confirm the accessible copper surface area obtained
by the XAFS measurements, N2O chemisorption experiments were
performed using the reactive frontal chromatography method [26].
For each experiment, 500 mg catalyst (activated in situ at 250 °C for
60 min) was used.
Apparent energies of activation for the uncoated catalysts.
Catalyst
Apparent energy of
activation (kJ molꢀ1
Error (kJ molꢀ1
)
)
Cu/Al2O3-54
Cu/Al2O3-99
Cu/Al2O3-100
Cu/Al2O3-149
Cu/Al2O3-151
Cu/Al2O3-214
Cu/Al2O3-257
Cu/Al2O3-360
53
47
50
49
49
52
45
43
0.2
0.2
0.3
0.1
0.1
0.2
0.2
0.6
reaction conditions used for the data reported in the literature (H2
to CO ratio of ca. 5 and a CO2 to CO ratio of 1.1). For this reason, we
also compared our catalysts to a commercial low-temperature
WGS catalyst (see Supplementary material).
2.4. Catalytic activity
On the basis of the turnover frequencies, the activity decreased
with increasing surface area of the support, since the accessible
copper surface area increased (vide infra). The variation of the par-
ticle size of the catalyst pointed the absence of interparticle diffu-
sion limitations. A slight kinetic isotope effect was observed when
D2O was used instead of H2O (see Supplementary material), lead-
ing to an approximately 16% lower rate of reaction.
The low-temperature water–gas shift reaction was studied at
temperatures between 160 and 250 °C and a pressure of 2 bar.
The gas composition used was p(H2) = 0.53 (75%), p(CO) = 0.05
(8%), p(CO2) = 0.08 (13%) and p(N2) = 0.03 (4%), at a steam to gas ra-
tio of 3–7 (total gas flow 40 mL minꢀ1). A fixed bed reactor filled
with 200 mg catalyst (diluted with SiC) was used. The products
were analyzed using a Shimadzu GC-2014 gas chromatograph.
For comparison, also a commercial catalyst (ShiftMax-240) was
also tested at temperatures between 160 and 250 °C. The activity
of the catalysts is either presented as turn over frequency or pre-
sented as rate per gram catalyst to evaluate the performance of
the catalysts (all catalysts had the same copper loading). The equa-
tions used for the kinetic analysis at differential and integral reac-
tion conditions are presented in Supplementary material.
3.2. Structural and electronic properties of uncoated catalysts
To determine the electronic structure of copper, the uncoated
catalysts were reduced in situ at 250 °C and examined by X-ray
absorption spectroscopy (XAS) as shown in Fig. 3. The fraction of
oxidized Cu varied sympathetically with the specific surface area
of the support. The degree of reduction of the Cu/Al2O3 catalysts
was calculated from a linear combination of the XANES of Cu0,
CuI and CuII references (see Supplementary material), and it was
found to vary between 95 and 66% (see Table 3).
3. Results
3.1. Activity of the uncoated catalysts
The local environment of the Cu atoms with respect to the num-
ber of neighbors (N), the distance between them (r) and the statis-
The turn over frequencies of the uncoated Cu/Al2O3 catalysts in
the low-temperature water–gas shift reaction as function of tem-
perature are shown in Fig. 2, the apparent energies of activation
are given in Table 2.
The turn over frequencies for the water–gas shift reaction were
determined at feed compositions close to industrial conditions (H2
to CO ratio of ca. 9 and a CO2 to CO ratio of 1.6). The TOF for a CuO/
Al2O3 catalyst of 0.8 ꢂ 10ꢀ6 mol mꢀ2 sꢀ1 at 200 °C reported [27] is
higher than the TOF observed for the Cu/Al2O3-149 catalyst at
200 °C (0.3 ꢂ 10ꢀ6 mol mꢀ2 sꢀ1), which resulted from the different
tical order (Debye-Waller factor r
2) calculated from the EXAFS as
well as the zero energy correction (E0) are shown in Table 3. The
particle size was determined from the average coordination num-
ber of the nearest metal neighbor atoms assuming a cuboctahedral
geometry [28]. As indicated by XANES, the Cu–O concentration in-
creases with increasing specific surface area.
A particle size of 1.1 nm for the Cu clusters, which can be re-
lated to a dispersion of 0.82, was determined for the catalysts with
the highest activities in the low-temperature WGS reaction (Cu/
Al2O3-149 and Cu/Al2O3-151). In contrast, the catalysts with the
lowest activities either consisted of larger Cu particles (2.1 nm)
not indicating oxygen neighboring atoms (Cu/Al2O3-54) or con-
sisted of smaller copper clusters (0.7 nm) with a coordination
number higher than 1 for oxygen neighbors (Cu/Al2O3-360).
1.0
0.0
-1.0
-2.0
-3.0
-4.0
-5.0
3.3. Catalytic activity of ionic liquid coated copper catalysts
A comparison of turnover frequencies and the rate constants for
the water–gas shift reaction between coated and uncoated cata-
lysts is shown in Fig. 4. Catalysts based on Al2O3-54, Al2O3-149
and Al2O3-257 were tested. With increasing temperature, the turn
over frequency of the coated catalysts decreased indicating a neg-
ative apparent energy of activation. We can exclude that this is
only an effect of deactivation, as activity test for the coated cata-
lysts at 160 and 200 °C indicated higher activity at the lower reac-
tion temperature (see Supplementary material). (Note that we
observed deactivation with time on stream, due to a sintering of
the copper nanoparticles. Also, see the results from XAFS.)
1.85
1.95
2.05
1/T 1000 [1/K]
2.15
2.25
2.35
Fig. 2. Ln of TOF for uncoated copper catalysts between 160 and 250 °C (4 Cu/
Al2O3-54, } Cu/Al2O3-99, ꢁ Cu/Al2O3-100, d Cu/Al2O3-149, s Cu/Al2O3-151, N Cu/
Al2O3-214, j Cu/Al2O3-257 and h Cu/Al2O3-360).

R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
283
0
-1
-2
-3
-4
2.05
2.10
2.15
2.20
2.25
2.30
2.35
1/T 1000 [1/K]
g
f
e
3.0
2.0
1.0
d
c
b
a
0.0
8.97
8.99
9.01
9.03
Energy [keV]
-1.0
-2.0
Fig. 3. XANES of uncoated copper catalysts ((a) Cu/Al2O3-54, (b) Cu/Al2O3-99, (c)
Cu/Al2O3-149, (d) Cu/Al2O3-151, (e) Cu/Al2O3-214, (f) Cu/Al2O3-257 and (g) Cu/
Al2O3-360.) after reduction at 250 °C.
2.05
2.10
2.15
2.20
2.25
2.30
2.35
1/T 1000 [1/K]
At lower temperatures (below 180 °C), the coated copper cata-
lysts showed higher turnover frequencies as well as higher rate
constants for the CO conversion compared to the uncoated cata-
lysts. A comparison of the coated catalyst with the highest activity
in the water–gas shift reaction (Cu/BDiMIm/Al2O3-257) and a com-
mercial catalyst (ShiftMax-240) is given in Supplementary mate-
rial. It is interesting to note that the coated catalyst was five
times more active (on the basis of the TOF) at 160 °C, although
the metal loading of the commercial catalyst was ten times higher.
Fig. 4. Ln of TOF (upper part) and ln of rates (lower part) for ionic liquid coated (h
Cu/BDiMIm/Al2O3-54, Cu/BDiMIm/Al2O3-149, Cu/BDiMIm/Al2O3-257) and
uncoated (d Cu/Al2O3-257, j Cu/Al2O3-54, N Cu/Al2O3-149) copper catalysts.
4
s
the average concentration of Cu0 was determined to increase for
example from 40% for the uncoated catalyst to 51%, 55%, 62% and
66% for the catalyst coated with 5, 10, 20 and 30 wt.% BDiMIm,
respectively.
The catalysts used for the comparison of coated and uncoated
systems (based on Al2O3-54, Al2O3-149 and Al2O3-257) were also
examined by in situ XAS experiments. After the reduction of the
catalyst at 215 °C, XAS was measured at liquid nitrogen tempera-
ture for the EXAFS analysis. Then, the sample was heated to
150 °C under He flow. Fig. 5 shows the spectra of Cu/Al2O3-149
in He at 150 °C, after dosing CO at 150 °C for 30 min, after 60 min-
utes under reaction conditions (CO to H2O ratio 1:2) and in He after
the reaction at 30 °C. When CO is added to the system, the oxida-
tion state is lowered and a further reduction takes place when the
sample is additionally exposed to water by the hydrogen gener-
ated. The structure of the copper nanoparticles after the reaction
was determined by EXAFS analysis.
3.4. Characterization of the electronic and structural properties of the
ionic liquid coated catalysts under reaction conditions
The XANES collected during the reduction of an uncoated cata-
lyst as well as of four BDiMIm coated (5, 10, 20 and 30 wt.%) copper
catalysts in H2 flow at 210 °C are complied in Fig. S7. In general, a
higher concentration of ionic liquid present on the catalysts led to
a higher degree of reduction of the Cu clusters, whereas for the un-
coated catalyst XANES data indicates that the Cu particles are oxi-
dized in part. Using a linear combination of the XANES of Cu0 and
CuI references for the quantitative analysis of the XANES structure,
Table 3
Results from XAS analysis for uncoated copper catalysts after reduction at 250 °C.
Al2O3-54
Al2O3-99
Al2O3-149
Al2O3-151
Al2O3-214
Al2O3-257
Al2O3-360
rCuO (Å)
–
0
–
–
–
0
–
–
1.84
0.6
0.003
1.7
1.88
0.6
0.004
6.2
1.89
0.8
0.003
7.7
1.87
1.0
0.005
5.3
1.91
1.2
0.004
6.6
NCu–O
r2
E0 (eV)
rCuCu (Å)
2.53
9.4
0.006
4.5
0.56
2.1
95
2.53
9.2
0.007
4.4
0.59
2.0
91
2.53
7.1
0.012
2.8
0.81
1.1
83
2.53
7.0
0.011
3.4
0.82
1.1
84
2.53
6.0
0.009
2.6
0.89
0.9
76
2.53
5.8
0.011
1.5
0.90
0.8
72
2.53
5.5
0.010
3.6
0.92
0.7
66
NCu–Cu
r2
E0 (eV)
Dispersion
Particle size (nm)
Degree of reduction (%)

284
R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
With Cu/Al2O3-149, the degree of reduction is higher compared
to the other two tested uncoated supports. The XANES of the
coated catalysts show the same oxidation state after treatment at
reaction conditions and indicate an oxidation state closer to 0
(95% compared to 87%) than the uncoated catalysts.
The results of the EXAFS analysis before and after the treatment
under reaction conditions are compiled in Table 4 (uncoated sys-
tems) and Table 5 (coated systems) (note that the catalysts were
reduced at 215 °C before reaction and not at 250 °C as the catalysts
shown in Table 3, leading to lower states of reduction).
For the uncoated catalysts, the number of copper–oxygen
neighbors is reduced to 60–75% after and the treatment at reaction
conditions, while the number of copper–copper neighbors in-
creases indicating a sintering of the metal particles.
For the coated catalysts, copper–oxygen neighbors were not
found after treatment at water–gas shift reaction conditions. The
largest particles (d = 2.7 nm) after reaction can be found for the
catalyst based on Al2O3-149. Note that the copper particle size
was larger than with the uncoated catalysts.
d
c
b
a
3.5. Adsorption of CO, CO2 and D2O followed by IR spectroscopy
Adsorption of CO, CO2 and D2O was followed by IR spectroscopy
in order to understand the interactions of the reactants with the
active sites. Note that D2O was used instead of H2O to avoid
absorption bands in the region of the perturbed OH vibrations
resulting from the hydrogen bonding interaction between the OH
groups of the support and ionic liquid [15].
8.97
8.99
9.01
9.03
Energy / keV
Fig. 5. XANES spectra of Cu/Al2O3-149 after reduction (a), in CO flow (b), under
reaction conditions (c) and after reaction (d) (note that the catalysts were reduced
at 215 °C before reaction and not at 250 °C as the catalysts shown in Fig. 3).
The IR spectra of CO adsorbed on Cu/Al2O3 (a) and CO adsorbed
on Cu/BDiMIm/Al2O3 (b) are shown in Fig. 6. For the uncoated cat-
alyst, the adsorption of CO led to a band at 2088 cmꢀ1 assigned to
CO linearly adsorbed on Cu0. For the coated catalyst, this band was
smaller and shifted to 2117 cmꢀ1. In both cases, adsorbed CO was
not observed after evacuation.
The averaged oxidation state of the coated catalysts after the
reduction step is lower than for the uncoated systems. In presence
of the reactants, the coated catalysts were also reduced, but the dif-
ferences in the XAS are not as distinct as for the uncoated catalysts.
The XANES for the uncoated and coated catalysts after the reac-
tion can be found in Supplementary material (Figs. S8 and S9).
The adsorption of CO2 on Cu/Al2O3 (shown in Fig. 7) led to bands
at 1657 cmꢀ1
,
1228 cmꢀ1 (bridged bidentate carbonate),
Table 4
Results of EXAFS analysis for uncoated copper catalysts before and after reaction at 150 °C.
Cu/Al2O3-54
Cu/Al2O3-149
before
Cu/Al2O3-257
before
after
after
before
after
rCuO (Å)
1.97
0.8
0.003
21.4
2.53
6.2
1.97
0.6
0.003
13.8
2.54
7.8
1.91
1.0
0.005
8.6
2.53
5.2
0.009
1.9
1.92
0.6
0.005
8.8
2.54
8.1
0.007
3.9
1.91
1.1
0.005
9.8
2.53
4.6
0.010
2.8
1.95
0.6
0.004
12.1
2.54
7.7
NCu–O
r2
E0 (eV)
rCuCu (Å)
NCu–Cu
r2
0.008
3.5
0.006
4.2
0.007
3.8
E0 (eV)
Dispersion
Particle diameter (nm)
0.88
0.9
0.77
1.3
0.93
0.7
0.73
1.4
0.94
0.6
0.74
1.2
Table 5
Results of EXAFS analysis for coated copper catalysts before and after reaction at 150 °C.
Cu/BDiMIm/Al2O3-54
Cu/BDiMIm/Al2O3-149
before
Cu/BDiMIm/Al2O3-257
before after
before
after
after
rCuO (Å)
1.94
0.5
0.006
9.3
2.54
9.0
0.006
4.8
–
–
–
–
1.99
0.3
0.001
17.5
2.54
9.4
0.006
5.2
0.54
2.2
–
–
–
–
1.93
0.6
0.006
6.4
2.54
9.0
0.006
4.9
–
–
–
–
NCu–O
r2
E0 (eV)
rCuCu (Å)
2.54
9.7
0.006
4.6
0.51
2.4
2.54
10.0
0.006
4.6
0.46
2.7
2.54
9.7
0.006
4.1
0.51
2.4
NCu–Cu
r2
E0 (eV)
Dispersion
Particle diameter (nm)
0.63
1.8
0.63
1.8

R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
285
(a)
(a)
Increasing CO2
pressure
0.6 A.U.
Increasing CO
pressure
0.2 A.U.
a
a
b
b
1800
1700
1600
1500
1400
1300
1200
2200
2150
2100
2050
2000
Wavenumber / cm-1
Wavenumber / cm-1
(b)
(b)
Increasing CO2 pressure
0.6 A.U.
0.02 A.U.
a
b
Increasing CO
pressure
a
b
2200
2150
2100
2050
2000
1800
1700
1600
1500
1400
1300
1200
Wavenumber / cm-1
Wavenumber / cm-1
Fig. 6. IR absorption spectra of CO adsorbed on copper supported on alumina (a)
and CO adsorbed on Cu/BDiMIm/Al2O3 (b) (spectrum b after evacuation).
(c)
0.6 A.U.
1429 cmꢀ1 (ionic carbonate) and 1600 cmꢀ1, 1391 cmꢀ1 (carboxyl).
Bands characteristic for the formation of bridged bidentate carbon-
ate (1637 cmꢀ1), carboxyl (1570 cmꢀ1) and an ionic carbonate spe-
cies (1442 cmꢀ1) were only observed on Cu/BDiMIm/Al2O3. The
region below 1400 cmꢀ1 was dominated by changes of the absorp-
tion bands of the trifluoromethanesulfonate anion of the ionic li-
quid (vide infra). The concentration of CO2 remaining on the
sample after evacuation was higher for the coated catalyst. It
should be noted that CO2 was also adsorbed on the parent support.
The observed bands were attributed to bridged bidentate carbon-
ate species (1650 and 1228 cmꢀ1) and an ionic carbonate species
Increasing CO2
pressure
a
b
1800
1700
1600
1500
1400
1300
1200
Wavenumber / cm-1
Fig. 7. IR absorption spectra of CO2 adsorbed on copper supported on alumina (a),
CO2 adsorbed on Cu/BDiMIm/Al2O3 (b) and CO2 adsorbed on Al2O3 (c) (spectrum b
after evacuation) (please note that in (b) the region below 1400 cmꢀ1 is not
depicted, since the spectra would be dominated by changes of bands attributed to
CF3SOꢀ3 ).
at 1435 cmꢀ1
.
Fig. 8 compiles the IR spectra after adsorption of D2O. A broad
band around 2570 cmꢀ1 (symmetric and asymmetric stretching
vibrations) with a feature at 2715 cmꢀ1 (due to the formation of
surface OD groups) for Cu/Al2O3 and around 2615 cmꢀ1 for Cu/BDi-
MIm/Al2O3 was observed. (Note that due to hydrogen bonding, the
bands for the symmetric and antisymmetric stretching vibrations
were broadened resulting in only one band around 2570 cmꢀ1
[29].) After evacuation, surface OD groups appeared at 2635,
2710 and 2745 cmꢀ1 for the uncoated catalyst, and OD-hydrogen
bonds were observed at 2625 cmꢀ1 for the coated material. It
should be noted that the maximum concentration of adsorbed
D2O was two times higher for Cu/BDiMIm/Al2O3. The adsorption
of D2O on Al2O3 showed identical result to the adsorption on Cu/
Al2O3 indicating that the observed features are characteristic of
adsorption on the support.
around 2330 and 2360 cmꢀ1 correspond to CO2 in the gas phase.
The band at 1645 cmꢀ1 is assigned to OH-bending vibrations of
water. The band at 1574 cmꢀ1 (appearing at temperatures higher
than 170 °C) indicates a carboxyl species, and the bands at 1460
and 1420 cmꢀ1 indicate the formation of monodentate and ionic
carbonate species, respectively. With increasing temperature, the
concentration of adsorbed CO and water was reduced, whereas
the concentration of carbonates and gas phase CO2 increased. Purg-
ing the catalysts with He at 150 °C after the reaction led to almost
complete removal of adsorbed CO with a significant concentration
of carbonates remaining on the surface.
Fig. 10 shows the difference IR spectra at reaction conditions for
Cu/Al2O3-149, Cu/Al2O3-257 and Cu/Al2O3-54. (Note that the bands
attributed to CO in the gas phase were subtracted from the spec-
tra.) The position and intensity of the band assigned to CO on
Cu0 varies for the three catalysts. The smallest concentration of ad-
3.6. In situ infrared spectroscopy during reaction
Fig. 9 shows the surface species during the course of an exper-
iment with an uncoated catalyst. After admitting CO at 150 °C the
band at 2088 cmꢀ1 indicated CO adsorbed on Cu0. The concentra-
tion of adsorbed CO was reduced as soon as 10 vol.% water vapor
was added to the reactant gas stream. The two bands appearing
sorbed CO was found for Cu/Al2O3-54 showing
a band at
2110 cmꢀ1. With Cu/Al2O3-257, the band for monodentate CO
was found at lower energy (2086 cmꢀ1) and an additional band

286
R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
(a)
0.5 A.U.
a
Increasing D2O
pressure
a
b
b
2900
2700
2700
2700
2500
2300
2100
1900
1700
c
Wavenumber / cm-1
2400
2200
2000
1800
1600
Wavenumber / cm-1
(b)
0.5 A.U.
Fig. 10. Difference IR spectra at in situ conditions for Cu/Al2O3-54 (a), Cu/Al2O3-149
(b) and Cu/Al2O3-257 (c).
Increasing D2O
pressure
54, over Cu/Al2O3-149 to Al2O3-257 with a ratio of 1:1.8:2.0. The
intensity of the band attributed to the OH-bending remained con-
stant for all three catalysts.
a
The comparison between the uncoated (Cu/Al2O3-149) and a
coated sample (Cu/BDiMIm/Al2O3-149), shown in Fig. 11, indicates
that only a small concentration of CO is adsorbed on the surface of
the ionic liquid coated catalyst (when purging with He the band at
2110 cmꢀ1 results from CO adsorbed on Cu). It should, however, be
emphasized that also the intensity of the band for OH-bending
vibrations of molecular water is smaller for the coated catalyst.
The bands at 1485 and 1570 cmꢀ1, observed only on the coated
catalyst, indicate the formation of carbonate and carboxyl species
on Cu/BDiMIm/Al2O3-149. Also, experiments with Cu/BDiMIm/
Al2O3-54 and Cu/BDiMIm/Al2O3-257 were performed. It was ob-
served that the band attributed to antisymmetric SO3 stretching
of trifluoromethanesulfonate (the anion of the ionic liquid) ion
pairs was increased for Cu/BDiMIm/Al2O3-257 under reaction con-
ditions (also see the following remarks on the interactions of the
reactants with anion of the ionic liquid). In contrast to the un-
coated catalyst, the intensity (20% for Cu/BDiMIm/Al2O3-149 and
30% for Cu/BDiMIm/Al2O3-257) of the band attributed to the OH-
bending vibrations increases with surface area for the coated
catalysts.
b
2900
2500
2300
2100
1900
1700
Wavenumber / cm-1
(c)
0.5 A.U.
Increasing D2O
pressure
a
b
2900
2500
2300
2100
1900
1700
Wavenumber / cm-1
Fig. 8. IR absorption spectra of D2O adsorbed on copper supported on alumina (a),
D2O adsorbed on Cu/BDiMIm/Al2O3 (b) and D2O adsorbed on Al2O3 (c) (spectrum b
after evacuation).
To investigate the influence of the reactants on the ionic liquid,
the absorption region of the trifluoromethanesulfonate anion of
two samples with a ionic liquid coating of 5 and 20 wt.% is
compared in Figs. S10 and S11, respectively. In both cases, the IR
Increasing
temperature
a
a
b
b
c
2400
2200 2000 1800
1600
1400 1200
Wavenumber / cm-1
Fig. 9. Difference spectra of in situ IR experiment with uncoated Cu catalyst:
2.5 mL minꢀ1 CO (a), temperature ramp with CO and water (b) and spectrum after
exposure at reaction conditions (c).
2200 2100 2000 1900 1800 1700 1600 1500
Wavenumber / cm-1
at 2000 cmꢀ1 was observed, which is attributed to CO adsorbed in a
bridged mode. The intensity of the CO bands increased from Al2O3-
Fig. 11. Comparison of the difference spectra of a coated (a) and an uncoated (b)
copper catalyst.

R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
287
spectra before reduction, after reduction, in a flow of CO, at reac-
12
10
8
tion conditions (10% water) and after reaction are displayed.
For the catalyst with a loading of 5 wt.% ionic liquid, bands were
observed at 1171 cmꢀ1 (antisymmetric CF3), 1225 cmꢀ1 (symmet-
ric CF3), 1242 cmꢀ1 (antisymmetric SO3 stretching of trifluoro-
methanesulfonate ion pairs) and 1275 cmꢀ1 (antisymmetric SO3
deformation vibrations). After reduction of the catalyst, the bands
attributed to the antisymmetric SO3 stretching of trifluorometh-
anesulfonate ion pairs and to the antisymmetric SO3 deformation
vibrations decreased in intensity, and a new band was observed
at 1330 cmꢀ1. At the same time, the bands attributed to CF3 vibra-
tions were shifted to lower wavenumbers. Upon flowing CO over
the catalyst, only small changes were observed in the spectrum,
but after adding water the spectrum changed appeared to be very
similar to the spectrum before reduction. This indicates that the
presence of water seems also to induce changes in the interaction
of the ionic liquid with the metal surface.
6
4
2
0
0.0
0.5
1.0
1.5
2.0
2.5
Cu particle size [nm]
Fig. 12. Calculated maximum number of copper (N) and oxygen neighbors (j)
assuming cuboctahedral geometry and the observed numbers of neighbors for
oxygen (h) in dependence of copper particle size (oxygen is only located on the
surface of the catalyst).
The spectra of the catalyst with an ionic liquid loading of
20 wt.% showed four bands before admission of H2. Antisymmetric
and symmetric CF3 vibrations were observed at 1154 cmꢀ1 and at
1222 cmꢀ1, the antisymmetric SO3 stretching of trifluoromethane-
sulfonate ion pairs and the antisymmetric SO3 deformation vibra-
tions at 1240 and 1275 cmꢀ1. A band at 1330 cmꢀ1 appeared
again after reduction and disappeared after exposing the catalyst
to CO/H2O. In contrast to the catalyst with the lower loading, the
other parts of the spectra were hardly influenced by changing
the gas atmosphere.
The spectra of the catalyst with a higher loading of ionic liquid
imply that only the direct interaction with the surface can be ob-
served. The changes of the oxidation state of copper and the inter-
action of the surface species with the ionic liquid can only be
observed by the appearance and disappearance of the band at
1330 cmꢀ1. The changes of the other bands are overlaid by the
vibrations of molecules in the bulk phase.
of the support. Cu/Al2O3-54 and Cu/Al2O3-99 were nearly fully re-
duced, while a significant fraction of oxidized Cu was observed
with the other catalysts. In line with the increase in the white line,
EXAFS showed that the number of oxygen neighbors increased in
parallel to the surface area of the support confirming the presence
of a higher fraction of oxidized Cu. As the copper particles size de-
creased with increasing specific surface area of the support, the re-
sults show that the fraction of oxidized Cu increases, as the particle
size decreases.
Because the Cu–Cu distances in the particles were identical
with those of metallic Cu, the oxygen atoms are concluded to be lo-
cated at the surface of the copper particles. Fig. 12 shows the max-
imum number of oxygen neighbors possible at the surface
assuming that the oxygen atoms are located in the tetrahedral
and octahedral vacancies of the copper surface. A visualization of
the location of O on the surface of Cu(1 1 1) and (1 0 0) terminated
particles are shown in Fig. 13. A cluster containing 147 Cu atoms
has 6 (1 0 0) and 8 (1 1 1) surfaces with 16 and 10 Cu atoms on
each surface, respectively. On each (1 0 0) surface 9 oxygen atoms
can be coordinated to the vacancies, whereas for the (1 1 1) sur-
faces a maximum of six oxygen atoms can be coordinated leading
to 102 oxygen atoms that can be coordinated to a Cu cluster with
4. Discussion
4.1. State of the catalysts
XANES indicated that the fraction of oxidized Cu in the un-
coated catalysts increased in parallel to the specific surface area
Fig. 13. Model of ideal cuboctahedral Cu cluster with oxygen atoms in tetrahedral and octahedral vacancies.

288
R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
oxygen on the Cu surface is lower on the larger particles than on
3.0
2.5
2.0
1.5
1.0
0.5
0.0
the smaller, the increase in wavenumber of adsorbed CO indicates
stronger dipole–dipole interactions of CO adsorbed on larger parti-
cles. However, we cannot rule out strong bonding on defect sites of
the smaller particles. This is supported by the fact that for the Cu/
Al2O3-257 catalyst bridged bonded CO was found indicating an
even stronger bond to the surface compared to the other catalysts.
Note that this is in good agreement with the fact that the observed
apparent energy of activation was lower for the catalyst with the
highest surface area. Corrected for the number of free accessible
surface atoms of Cu, the concentration of adsorbed CO was identi-
cal on all three samples investigated (see Supplementary material).
The linear correlation passes, however, not through the origin indi-
cating that a certain constant fraction of the reduced Cu is blocked
by other species or is not accessible for CO, e.g., the interface be-
tween the particle and the support.
0
100
200
300
400
Surface area [m²·g-1]
Fig. 14. Particle size in dependence of support surface area (h Cu/Al2O3-360, j Cu/
Al2O3-257, N Cu/Al2O3-214, d Cu/Al2O3-149, s Cu/Al2O3-151, } Cu/Al2O3-99 and 4
Cu/Al2O3-54).
In addition to the bands of bridged bidentate carbonates and io-
nic carbonates, which were also observed with the parent support,
carboxylic species were observed on the uncoated copper catalyst
indicating that in the presence of carbon–metal bonds. As this car-
boxyl species is also observed in the in situ experiments, we con-
147 atoms. This results in a maximum number of oxygen neigh-
bors of 2.5 and a Cu–Cu coordination number is 9.0. For a given
copper particle size, the number of oxygen neighbors found exper-
imentally were 20–30% of the theoretically determined maximum
coordination numbers. Thus, we conclude that at most about one-
third of the copper surface atoms were blocked by oxygen. It
should be noted at this point that oxygen on the copper surface
plays an important role for the structure the nanoparticles adopt,
if exposed to water–gas shift conditions (vide infra).
Because the fraction of the potential interface between the me-
tal particle and the oxide support is much smaller than the fraction
of the oxidized Cu, we conclude the oxygen atoms on the surface
are located to a large extent on the accessible part of the copper
nanoparticles. Please note that accessible copper surface area
was estimated from the results of the EXAFS analysis, which per-
fectly agreed with the N2O chemisorption experiments. (Results
are presented in Supplementary material.)
The presence of large copper particles observed for supports
with a lower specific surface area (see Fig. 14) is attributed to
the fact that during the preparation procedure the local concentra-
tion of the copper ions is higher in the case of a low surface area
support (8.8 Cu2+ per nm2) compared to a high surface area sup-
port (1.3 Cu2+ per nm2). This facilitates agglomeration of the pre-
cursor or the metal atoms in formation process.
Compared to the uncoated catalysts, the IL-coated catalysts
show a higher degree of Cu reduction after treatment in hydrogen.
As the concentration of the supported ionic liquid increases, the
fraction of metallic Cu increased. This is attributed to the high sol-
ubility of water, formed during reduction as well as the low solu-
bility of oxygen in the polar ionic liquid. With the coated
catalysts also the copper particle size was higher than with un-
coated catalysts (see Table 2). It indicates that the ionic liquid in-
creases either the mobility of the precursor salt or the mobility
of the Cu particles as they are formed indicating that the restruc-
turing of nanoparticles is facilitated by the ionic liquid.
*
clude that the reaction involves the formation of COOH group on
the surface, which is a strong indication that the reaction follows
the associative mechanism [5,11].
Because the IR spectra after of D2O on the uncoated copper cat-
alyst and the parent support were identical, we conclude that
water is adsorbed in its majority on the support and that the
adsorption is not influenced by the relatively small concentration
of copper nanoparticles. However, water must also adsorb on the
Cu particles, as its dissociation on Cu is the prerequisite for the cat-
alytic chemistry. Because the dissociation of water, which is facil-
itated by the presence of adsorbed oxygen [30], is generally
considered to be the rate-determining step, the concentration of
OH groups on Cu is assumed to be small.
With coated catalysts, the band of adsorbed CO on Cu covered
with IL at 2117 cmꢀ1 appeared at higher wavenumber than the
corresponding band with uncovered Cu particles. The lower cover-
age in the case of IL covered Cu particles (in line with the low activ-
ity of CO dissolved in the ionic liquid in accordance with the
literature [31–33]) allows us to exclude CO dipole coupling as
cause for this blue shift. Because the oxygen coverage of the IL cov-
ered particles after reduction with H2 at 215 °C is minimal (judged
from the XAFS data), the blue shift is tentatively attributed to
markedly weakened CO adsorption in the presence of the ionic li-
quid due to the competitive interaction of the ionic liquid with the
Cu surface.
While the concentration of CO2 adsorbed was higher for the
coated catalyst after evacuation, the principal states of adsorbed
CO2 on the uncoated and coated catalysts were concluded to be
identical. The downward shift of the bands of the bridged bidentate
carbonate and carboxylic species together with the changes in the
bands of the trifluoromethanesulfonate species with the coated
catalyst results from the interaction of the ionic liquid with chem-
isorbed CO2. The higher concentration of bidentate carbonate indi-
cates that chemisorbed CO2 is stabilized by the ionic liquid, which
is in good agreement with the strong interaction of CO2 with the
CF3-group (vide infra) [34–36].
D2O adsorption showed that the concentration of adsorbed
water was two times higher for the coated catalyst, because the io-
nic liquid [37] increases the activity of water during the water–gas
shift reaction. It is interesting to note that despite the stronger
interactions, the IR stretching band of adsorbed water was broader
in the case of the uncoated catalyst than with the IL covered sam-
ples indicating a less diverse environment in the latter case.
The strong interactions between the reactants and the ionic li-
quid are also reflected in the changes of the IR bands of the IL
Also, the observed Cu–O distances for the coated and uncoated
samples correlate perfectly with the degree of oxidation, as a large
distance is found for the more oxidized samples. Note that the Cu–
O distance in Cu2O is around 1.85 Å, whereas the distance in CuO is
1.95 Å.
4.2. Sorption of reactants
For the uncoated catalysts, the in situ IR spectra reflect the dif-
ferent size of the copper particles. The wavenumber of the band
attributed to CO adsorbed on Cu0 was shifted from 2082 cmꢀ1
(Cu/Al2O3-149) to 2110 cmꢀ1 (Cu/Al2O3-54). As the fraction of

R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
289
-
CF3SO3 groups. Note that mainly the anion of the ionic liquid is
interacting with the surface [15]. The change from CuI/II to Cu0 after
the reduction leads to a decrease of the band at 1275 cmꢀ1 attrib-
uted to antisymmetric sulfonate vibration and the appearance of a
band at 1330 cmꢀ1 attributed to this vibration of a species with a
stronger S@O bond. This increases the bond strength of two sulfo-
nate oxygen atoms with the surface, which in turn increases the
strength to the remaining bonds to the oxygen and the carbon
atom. Indeed, the bands corresponding to CF3 vibrations are shifted
to lower wavenumbers indicating weaker CAF bonds as electrons
2.00
1.80
1.60
1.40
1.20
1.00
0.80
0.60
0.40
0.20
0.00
from Cu0 are pushed into the CF3 antibonding * orbital [38]. Thus,
r
we conclude that the trifluoromethanesulfonate is adsorbed side-
ways and normal to the surface with two oxygen atoms bound to
Cu. After admitting CO and water, however, the spectrum turned
back to its state before reduction indicating that the interaction
0
100
200
300
400
Surface area support / m2g-1
Fig. 16. TOF (mol CO per m2 of accessible Cu) in dependence of the support surface
area.
of CO with Cu is stronger (despite its reduced strength in the pres-
-
ence of the IL) than the interaction with CF3SO3
.
accessible copper surface area on the activity. The turnover fre-
quencies for the five uncoated catalysts are compared in Fig. 16,
which shows that the TOF decreases with increasing support sur-
face area. The degree of reduction and also the particle size in-
creased during the treatment of the materials with CO and
during the water–gas shift reaction. XANES shows that the degree
of reduction increases after flowing CO over the catalyst for 30 min.
Under these conditions, the formation of bidentate CO species can
be observed by IR spectroscopy, which can be related to a change
of the shape of the nanoparticles induced by the reaction of CO
with surface oxygen. It is also remarkable that Cu is further re-
duced by admission of water to the stream of CO, which can be
attributed to the oxygen aided dissociation of water forming two
hydroxyl groups quickly reacting with adsorbed CO to carboxylic
surface species. The fact, that the size of the copper particles as
well as the number of oxygen neighbors was similar for all un-
coated samples after reaction, (note that the catalytic activity
was different) implies that not only the size of the clusters is of
importance for the reactivity, but it also suggests that the structure
of the nanoparticles and the concentration of highly uncoordinated
sites are important. Note in this respect that copper nanoparticles
containing defect sites are claimed to lead to a higher CO conver-
sion activity [39]. Thus, it appears that the structure of the copper
particles before the reaction determines the reactivity of the cata-
lysts. In this context, we speculate that for the catalysts having a
high fraction of metallic Cu (Cu/Al2O3-99 and Cu/Al2O3-54) less de-
fect sites can be formed by removing oxygen leading to less active
catalysts. For the nanoparticles itself, the turnover frequency de-
creased with the particle size. IR spectroscopy indicated that the
smaller particles interact stronger with CO (see the formation of
bidentate CO in Fig. 10c), which blocks the metal sites for further
reaction.
4.3. Water–gas shift catalysis
For the uncoated catalysts, the highest rate (per gram catalyst)
in the low-temperature water–gas shift reaction was observed for
Cu supported on both Al2O3-149 and Al2O3-151 with a fraction of
about 20% of the Cu atoms being oxidized. Fig. 15 shows that the
rate decreases with higher and with lower degrees of reduction.
The copper particle size for these catalysts determined from EXAFS
was 1.1 nm and in average 0.6 oxygen neighbors for the Cu atoms
were observed (compared to maximally 3.3 oxygen neighbors if
the entire surface of the cluster is occupied with oxygen).
While we have used a comparison of the rate constants to eval-
uate the performance of the catalysts (all catalysts had the same
copper loading), we would also like to address the influence the
120
110
100
90
80
70
Increasing surface area
60
50
60%
70%
80%
90%
100%
Degree of reduction
120
110
100
90
A kinetic isotope effect (kH/kD) of 1.2 was observed for the un-
coated catalyst. This confirms that the steps of the reaction mech-
anism involving an abstraction of hydrogen (abstraction of
hydrogen from water and the decomposition of *COOH) are in-
volved in the rate-determining step.
80
To compare the activities of coated and uncoated catalysts three
catalysts were chosen, i.e., Cu/Al2O3-54, Cu/Al2O3-149 and Cu/
Al2O3-257, being examples for the optimum catalyst state as well
as one for a too high and one for a too low degree of reduction
for the uncoated materials. In case of the coated catalysts, however,
XAFS confirmed that the oxidation state and the copper particle
size of all catalysts were similar. The catalytic activity increased
with increasing specific surface area of the support leading to a
lower averaged thickness of the IL film. On the one hand, this leads
to changes in the interactions of the ionic liquid with the reactants.
Also, the amount of dissociated water on the surface of the catalyst
70
Increasing surface area
60
50
0
10
20
30
40
50
60
70
Accessible Cu surface area [m2g-1]
Fig. 15. Activity of uncoated copper catalysts in dependence of the degree of
reduction (upper part) and in dependence of accessible Cu surface area (lower part)
(h Cu/Al2O3-360, j Cu/Al2O3-257, N Cu/Al2O3-214, d Cu/Al2O3-149, s Cu/Al2O3-
151, } Cu/Al2O3-99 and 4 Cu/Al2O3-54).

290
R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
was higher for the materials with a thinner film. This can be ex-
plained by the fact that the concentration of water near to the sur-
face was higher compared to the materials with a thicker film,
where the water molecules are distributed throughout the film.
At temperatures below 180 °C, the coated catalysts were more
active than the uncoated materials. Due to the higher activity,
the formation of carboxyl species was observed at lower tempera-
tures compared to the uncoated systems, also indicating that these
species are stabilized by the ionic liquid. With increasing temper-
atures, the catalytic activity for the coated catalysts decreased
leading to an apparent negative energy of activation. We tenta-
tively attributed this to the decreasing solubility of the reactants
in the ionic liquid with increasing temperatures. IR experiments
showed that the concentration of adsorbed CO at higher tempera-
tures is more reduced for the coated compared to the uncoated cat-
alysts. Note that also the enhanced particle growth for the coated
catalysts can be accounted for the lower activity at higher
temperatures.
-120
-125
-130
-135
-140
-145
-150
-155
b
a
aCO(T, mCO) = 0.002
0.00
0.01
0.02
0.03
0.04
0.05
Activity CO
Fig. 17. Chemical potential of CO in the vapor phase (a) and of CO dissolved in a
supported ionic liquid (b).
The question arises at this point as to why the coated catalysts
are more active than the uncoated ones and why the activity drops
with increasing reaction temperature. The activity of catalysts in
the water–gas shift reaction is determined by the concentration
of the reactants at the active sites and by the intrinsic activity of
the catalyzing metal. Because the coated catalysts have an environ-
ment different from the gas phase at the surface, also the concen-
trations/activities of the educts in the ionic liquid have to be
considered. This is especially important for CO, as the solubility
of CO in ionic liquids is known to be low [32].
ꢀ230 kJ/mol indicating that 2.75 times more water than copper
is present at equilibrium conditions. This is also in good agreement
with the IR results, as the amount of observed adsorbed water two
times higher for the coated catalyst, due to the interactions of the
water molecules with the ionic liquid [37]. As pointed out earlier,
the energetic barrier to abstract an H from H2O is about seven
times higher than the barrier to desorb water from the catalyst sur-
face [5], therefore, the increased concentration of water adjacent to
the active sites will lead to a higher rate since the probability to
cleave the OAH bond is higher.
At equilibrium conditions, Henry’s law can be applied to deter-
mine the activity of CO in the ionic liquid:
Not only the interaction of H2O, but also the interaction of the
intermediate carboxyl species with the ionic liquid benefits the
reaction, because the decomposition of *COOH (abstraction of H
KH;COðT; pÞ ꢃ aCOðT; mCOÞ ¼ fCOðT; pÞ
ð7Þ
*
from COOH) is the second high energy barrier of the water–gas
with Henry’s constant of CO in the ionic liquid kH,CO(T, p), the activ-
ity of CO in the ionic liquid aCO(T, mCO) and the fugacity of CO in the
vapor phase fCO(T, p). kH,CO in the supported ionic liquid is 54 MPa
(determined from CO adsorption experiments on coated catalysts
with and without copper, see Supplementary material). This is in
excellent agreement with the published value of kH,CO (55 MPa)
[32]. Note that the presence of copper increases the CO uptake by
a factor of 30, due to the interaction of CO with copper.
The dependence of the chemical potential of CO on the fugacity
of CO is given by:
shift reaction. As the ionic liquid interacts with CO2 [34–36], it is
conceivable that also these carboxyl species interact with the ionic
liquid and facilitate the abstraction of hydrogen. Formate species,
formed from CO2 and H, are assumed to block the active sites
and can reach considerable surface coverages [5]. The presence of
the ionic liquid weakens their adsorption strength leading to less
blocked active sites.
Consequently, the presence of the IL leads to a higher concen-
tration of the reacting species by weakening of all interactions with
the surface and at the same time by adjusting the activities in the
liquid phase above the catalyst. A higher concentration of oxygen
facilitates the dissociation of water, but as the oxygen concentra-
tion increases the concentration of adsorbed CO must decrease cre-
ating so a maximum in activity for catalysts with alumina supports
of intermediate specific surface area.
lCO
¼
log þ RT ꢃ lnðfCO=poÞ
ð8Þ
gas
with the fugacity dependent chemical potential
lCOgas, the standard
chemical potential of CO in the gas phase log (ꢀ137 kJ/mol [40]),
the general gas constant R, the temperature T and the standard
pressure po.
When dissolved in the ionic liquid the chemical potential of CO
changes to:
5. Conclusions
The detailed characterization by in situ spectroscopy methods of
the interactions of ionic liquids with supported metal nanoparti-
cles leads to detailed insight into the reactivity of the catalysts in
the low-temperature water–gas shift reaction.
lCOsol
¼
los þ RT ꢃ lnðaCOðT; mCOÞÞ
ð9Þ
with the standard chemical potential of CO in the fluid phase los
At equilibrium conditions, ( G = Rimi i = 0) lCOgas is equal to
COsol. Thus, by combining Eqs. (7)–(9), los can be calculated to
.
D
ꢃ
l
l
The copper particle size and, thus, the reduction degree vary
sympathetically with the specific surface area of the support. The
copper particle sizes as well as the fraction of unreduced Cu and
of adsorbed oxygen atoms determine the catalytic activity in the
water–gas shift reaction. Overall, the TOF of the water–gas shift in-
creases with decreasing particle size. The presence of adsorbed
oxygen on the one hand increases the rate as it reduces the barrier
for water dissociation. At the same time, the surface coverage with
CO increases. If CO is bound too weakly (Cu/Al2O3-54) or too
strongly (Cu/Al2O3-257) to the active site, the reactivity in the
water–gas shift reaction is decreased. In the former case, the water
be ꢀ122 kJ/mol for the supported ionic liquid. By comparison lCO-
with log (see Fig. 17), the activity of CO at equilibrium condi-
sol
tions is aCO(T, mCO) = 0.002 for the supported ionic liquid.
This shows that as expected the activity of CO in the supported
ionic liquid is low, in perfect agreement with the results obtained
by IR spectroscopy. The concentration of adsorbed CO is, thus, re-
duced to approximately 4%, when the catalyst is coated with ionic
liquid.
The same considerations can be made for H2O. Using kH,H2O(T,
p) = 0.05 [41] and
l
og = ꢀ228 kJ/mol [42], los can be calculated to

R. Knapp et al. / Journal of Catalysis 276 (2010) 280–291
291
[5] A.A. Gokhale, J.A. Dumesic, M. Mavrikakis, J. Am. Chem. Soc. 130 (2008) 1402.
[6] T. Shido, Y. Iwasawa, J. Catal. 140 (1993) 575.
[7] C.V. Ovesen, P. Stoltze, J.K. Norskov, C.T. Campbell, J. Catal. 134 (1992) 445.
[8] C.V. Ovesen, B.S. Clausen, B.S. Hammershoi, G. Steffensen, T. Askgaard, I.
Chorkendorff, J.K. Norskov, P.B. Rasmussen, P. Stoltze, P. Taylor, J. Catal. 158
(1996) 170.
[9] P.B. Rasmussen, P.M. Holmblad, T. Askgaard, C.V. Ovesen, P. Stoltze, J.K.
Norskov, I. Chorkendorff, Catal. Lett. 26 (1994) 373.
[10] G.C. Wang, L. Jiang, Z.S. Cai, Y.M. Pan, X.Z. Zhao, W. Huang, K.C. Xie, Y.W. Li, Y.H.
Sun, B. Zhong, J. Phys. Chem. B 107 (2003) 557.
[11] Q.L. Tang, Z.X. Chen, X. He, Surf. Sci. 603 (2009) 2138.
[12] A. Riisager, R. Fehrmann, S. Flicker, R. van Hal, M. Haumann, P. Wasserscheid,
Angew. Chem., Int. Ed. 44 (2005) 815.
[13] O. Jimenez, T.E. Muller, C. Sievers, A. Spirkl, J.A. Lercher, Chem. Commun. 2974
(2006).
[14] C. Sievers, O. Jimenez, R. Knapp, X. Lin, T.E. Muller, A. Turler, B. Wierczinski, J.A.
Lercher, J. Mol. Catal. A: Chem. 279 (2008) 187.
[15] R. Knapp, A. Jentys, J.A. Lercher, Green Chem. 11 (2009) 656.
[16] C.P. Mehnert, E.J. Mozeleski, R.A. Cook, Chem. Commun. 3010 (2002).
[17] T. Welton, Coord. Chem. Rev. 248 (2004) 2459.
[18] R. Atkin, G.G. Warr, J. Am. Chem. Soc. 127 (2005) 11940.
[19] C. Sievers, O. Jimenez, T.E. Muller, S. Steuernagel, J.A. Lercher, J. Am. Chem. Soc.
128 (2006) 13990.
[20] A. Wolfson, I.F.J. Vankelecom, P.A. Jacobs, Tetrahedron Lett. 44 (2003) 1195.
[21] H. Hagiwara, Y. Sugawara, K. Isobe, T. Hoshi, T. Suzuki, Org. Lett. 6 (2004) 2325.
[22] S.M. Webb, Phys. Scripta T115 (2005) 1011.
[23] A.L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Phys. Rev. B 58 (1998) 7565.
[24] A.L. Ankudinov, J.J. Rehr, Phys. Rev. B 62 (2000) 2437.
[25] K.V. Klementiev, XANES Dactyloscope, Freeware: <http://www.desy.de/~klmn/
xanda.html>.
[26] O. Hinrichsen, T. Genger, M. Muhler, Chem. Eng. Technol. 23 (2000) 956.
[27] N.A. Koryabkina, A.A. Phatak, W.F. Ruettinger, R.J. Farrauto, F.H. Ribeiro, J.
Catal. 217 (2003) 233.
[28] R.E. Benfield, J. Chem. Soc., Faraday Trans. 88 (1992) 1107.
[29] B. Redlich, Ph.D. Thesis, Universität Hannover (Hannover), 1998.
[30] G.C. Wang, S.X. Tao, X.H. Bu, J. Catal. 244 (2006) 10.
[31] J. Jacquemin, P. Husson, V. Majer, M.F.C. Gomes, Fluid Phase Equilib. 240
(2006) 87.
[32] J. Kumelan, A.P.S. Kamps, D. Tuma, G. Maurer, Fluid Phase Equilib. 228 (2005)
207.
[33] I. Urukova, J. Vorholz, G. Maurer, J. Phys. Chem. B 109 (2005) 12154.
[34] C. Cadena, J.L. Anthony, J.K. Shah, T.I. Morrow, J.F. Brennecke, E.J. Maginn, J. Am.
Chem. Soc. 126 (2004) 5300.
adsorption is more difficult on the nearly oxygen-free catalysts,
while on the oxygen-rich materials the strong interaction with
CO leads to a lower concentration of water. In the presence of
the IL, CO is adsorbed weakly improving the competitiveness for
the adsorption of water.
Catalysts coated with the ionic liquid showed a higher turn over
frequency for the water–gas shift reaction at low temperatures
compared to uncoated and commercial systems. This is attributed
due to a higher concentration of water in the proximity of the ac-
tive sites and due to the interaction of the ionic liquid with reaction
carboxyl intermediates facilitating the decomposition of these spe-
cies. The observed concentrations of the reactants on the ionic li-
quid coated catalysts are in good agreement with this
interpretation.
The present results indicate potential to develop easily manage-
able and highly active catalysts with a lower metal loading for low-
temperature water–gas shift superior in activity to the present
commercial systems.
Acknowledgements
The project is funded by the BMBF (promotional reference
03X2012F). The authors are grateful to Max-Buchner-Stiftung for
partial support. The authors acknowledge fruitful discussions in
the framework of the network of excellence IDECAT. The authors
would like to thank HASYLAB, Hamburg, Germany and the ESRF
in Grenoble, France for providing beam time at station X1 and
BM26 for XAFS experiments. Xaver Hecht and Martin Neukamm
are thanked for the experimental support. Furthermore Solvent
Innovation GmbH and Süd Chemie AG are acknowledged for pro-
viding chemicals.
Appendix A. Supplementary material
[35] X.C. Zhang, F. Huo, Z.P. Liu, W.C. Wang, W. Shi, E.J. Maginn, J. Phys. Chem. B 113
(2009) 7591.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2010.09.019.
[36] T. Seki, J.D. Grunwaldt, A. Baiker, J. Phys. Chem. B 113 (2009) 114.
[37] L. Cammarata, S.G. Kazarian, P.A. Salter, T. Welton, Phys. Chem. Chem. Phys. 3
(2001) 5192.
[38] M.G. Miles, G. Doyle, R.P. Cooney, R.S. Tobias, Spectrochim. Acta A A25 (1969)
1515.
[39] C.S. Chen, J.H. Lin, T.W. Lai, B.H. Li, J. Catal. 263 (2009) 155.
[40] E. Wieberg, Die chemische Affinität, second ed., de Gruyter, Berlin, New York,
1972.
[41] J.K. Shah, E.J. Maginn, J. Phys. Chem. B 109 (2005) 10395.
[42] D.R. Stull, H. Prophet, JANAF Thermochemical Tables, Nat. Bur. Stand. (US)
(1971).
References
[1] D.L. Trimm, Appl. Catal., A 296 (2005) 1.
[2] Y. Amenomiya, G. Pleizier, J. Catal. 76 (1982) 345.
[3] T.M. Yureva, Kinet. Catal. 10 (1969) 862.
[4] J.A. Rodriguez, P. Liu, J. Hrbek, J. Evans, M. Perez, Angew. Chem., Int. Ed. 46
(2007) 1329.