Effect of Support and Promoter on Activity and Selectivity of Gold Nanoparticles in Propanol Synthesis from CO2, C2H4, and H2

Article abstract of DOI:10.1021/acscatal.6b00590

Direct propanol synthesis from CO₂, H₂, and C₂H₄ was investigated over TiO₂- and SiO₂-based catalysts doped with K and possessing Au nanoparticles (NPs). The catalysts were characterized by scanning transmission electron microscopy and temperature-programmed reduction of adsorbed CO₂. Mechanistic aspects of CO₂ and C₂H₄ interaction with the catalysts were elucidated by means of temporal analysis of products with microsecond time resolution. CO₂, which is activated on the support, is reduced to CO by hydrogen surface species formed from gas-phase H₂ on Au NPs. C₂H₄ adsorption also occurs on these sites. In comparison with TiO₂-based catalysts, the promoter in the K-Au/SiO₂ catalysts was found to increase CO₂ conversion and propanol production, whereas Au-related turnover frequency of C₂H₄ hydrogenation to C₂H₆ decreased with rising K loading. The latter reason was linked to the effect of the support on the ability of Au NPs for activation of C₂H₄ and H₂. The positive effect of K on CO₂ conversion was explained by partial dissolution of potassium in silica with formation of surface potassium silicate layer thus inhibiting formation of potassium carbonate, which binds CO₂ stronger and therefore hinders its reduction to CO.

Full text of DOI:10.1021/acscatal.6b00590

Research Article
pubs.acs.org/acscatalysis
Effect of Support and Promoter on Activity and Selectivity of Gold
Nanoparticles in Propanol Synthesis from CO2, C2H4, and H2
Sardor A. Mavlyankariev, Stefan J. Ahlers, Vita. A. Kondratenko, David Linke,
and Evgenii V. Kondratenko*
Leibniz-Institut fur Katalyse e.V. an der Universitat Rostock, Albert-Einstein-Str. 29 a, D-18059 Rostock, Germany
̈
̈
S
* Supporting Information
ABSTRACT: Direct propanol synthesis from CO2, H2, and C2H4 was investigated over TiO2- and SiO2-based catalysts doped
with K and possessing Au nanoparticles (NPs). The catalysts were characterized by scanning transmission electron microscopy
and temperature-programmed reduction of adsorbed CO2. Mechanistic aspects of CO2 and C2H4 interaction with the catalysts
were elucidated by means of temporal analysis of products with microsecond time resolution. CO2, which is activated on the
support, is reduced to CO by hydrogen surface species formed from gas-phase H2 on Au NPs. C2H4 adsorption also occurs on
these sites. In comparison with TiO2-based catalysts, the promoter in the K−Au/SiO2 catalysts was found to increase CO2
conversion and propanol production, whereas Au-related turnover frequency of C2H4 hydrogenation to C2H6 decreased with
rising K loading. The latter reason was linked to the effect of the support on the ability of Au NPs for activation of C2H4 and H2.
The positive effect of K on CO2 conversion was explained by partial dissolution of potassium in silica with formation of surface
potassium silicate layer thus inhibiting formation of potassium carbonate, which binds CO2 stronger and therefore hinders its
reduction to CO.
KEYWORDS: nanoparticles, Au, propanol, CO2, hydroformylation, TAP reactor
Cs2O.8−11 Those studies also elucidated mechanistic aspects of
product formation. Propanol was only formed when CO2,
1. INTRODUCTION
Fossil carbon-containing materials are the primary source of
C2H4, and H2 were present in the feed, whereas CO and H2O
energy and the main feedstock for the chemical industry for the
were the only products in the absence of C2H4. Irrespective of
production of a wide range of commodity chemicals. However,
the promoter, propanol is formed according to the following
the amount of such raw materials is limited, and their above
mechanistic scheme. CO2 initially reacts with H2 to CO and
transformations are typically accompanied by formation of
H2O via the reverse water−gas shift (RWGS) reaction followed
CO2. The scientific community now agrees that an increased
by C2H4 hydroformylation with CO to propanal. The latter is
concentration of CO2 in the atmosphere comes from burning
further hydrogenated to propanol. CO2 reaction with C2H4 to
fossil fuels. This gas is deemed to be the major contributor to
global warming due to the greenhouse effect.1,2 Therefore,
utilization of CO2 for green synthesis of fuels and useful
chemicals is of enormous environmental and economic interest.
propanol in the absence of H2 is thermodynamically not
possible.
Usage of CO2 instead of CO enables to operate under less
toxic conditions as compared to the industrial way of propanol
There are already successful examples of commercially available
formation through hydroformylation of C2H4 with CO and H2.
technologies for one-step CO2 conversion to methanol or
methane.3,4 Other approaches under investigation are CO2
Propanol could be used as an alternative to alcohol-based fuel
conversion into formic acid5,6 or into synthetic fuels via
stocks; it has a higher specific energy density than methanol
and ethanol.12 Furthermore, it can be easily converted into
Fischer−Tropsch synthesis.7
Recently, our group demonstrated that CO2 in the presence
of H2 and C2H4 can be converted into n-propanol (propanol)
with 100% selectivity with respect to CO2 over TiO2-supported
catalysts with Au nanoparticles (NPs) and doped by K2O or
Received: February 26, 2016
Revised: April 1, 2016
© XXXX American Chemical Society
3317
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
propylene, which is one of the core commodity chemicals. The
major drawback of the above CO2-based approach is low
propanol selectivity (<4.5%) with respect to C2H4 due to high
catalyst activity for nondesired C2H4 hydrogenation to C2H6. It
is, however, well-documented that the rate of olefin hydro-
genation over supported catalysts with Au NPs can be tuned by
metal loading and average NPs size.13−15 For example, Sermon
et al.14 showed for Au/SiO2 catalysts that the 1-pentene
hydrogenation rate related to overall Au amount decreased with
rising metal loading (0.05 to 5 wt %); the worst hydrogenation-
active catalyst was Au(5 wt %)/SiO2, possessing Au NPs of
about 23 nm. Chou et al.15 systematically investigated micelle-
encapsulated Au NPs supported on TiO2, ZnO, ZrO2, and SiO2
in C3H6 hydrogenation and found that Au/TiO2 with the
smallest Au NPs of 8 nm showed the highest hydrogenation
activity. For liquid-phase cyclohexene hydrogenation over Au/
SiO216 with Au NPs ranging from 10 to 20 nm, the catalyst with
the smallest NPs was the most active one. In addition, our
recent studies on CO2 conversion into propanol in the presence
of C2H4 and H2 established that the nondesired C2H4
hydrogenation to C2H6 over Au/TiO2 catalysts could be
partially suppressed by the usage of promoters like K2O8 or
Cs2O.9 Unfortunately, the conversion of CO2 decreased with
promoter loading, too.
The objective of this study was to elucidate the effects of
support and promoter on the activity and selectivity of
supported Au NPs in the conversion of CO2, C2H4, and H2
into propanol. Our specific aims were (i) to increase propanol
selectivity with respect to ethylene, that is, to reduce
nondesired ethylene hydrogenation to ethane and (ii) to
improve CO2 conversion without significant loss of high
propanol selectivity with respect to this feed component. To
this end, we explored the potential of SiO2-supported catalysts
containing similarly sized (around 5 nm) Au NPs but different
amounts of K for their activity and selectivity in the target
reaction. For comparative purposes, K−Au/TiO2 catalysts
promoted by either potassium nitrate8 or potassium silicate
were also investigated. Fundamentals of CO2 activation were
studied by means of CO2 temperature-programmed desorption
tests and CO2 pulse experiments in the temporal analysis of
products reactor. The latter technique was also applied for
analyzing C2H4 adsorption. To determine the size of Au NPs,
the used catalysts were characterized by high-angle annular dark
field (HAADF) and annular bright field (ABF) scanning
transmission electron microscopy (STEM).
overnight, and calcined in air at 573 K for 4 h. Doping of
calcined samples with K was performed via incipient-wetness
impregnation method with an aqueous solution of KNO3
(>99%, Merck). Hereafter, drying and calcination steps were
repeated yielding XK−Au/SiO2 with ″X″ representing the
particular K weight concentration.
Additionally, two batches of supported catalysts based on
TiO2 anatase (BASF, SBET of 58 m2 g−1) were prepared. They
differed in the source (KNO3 vs K2SiO3) of K and the method
of its deposition on TiO2. For a first batch, Au was initially
loaded onto TiO2 as described in ref 10. Thereafter, Au/TiO2
was incipient-wetness impregnated with a mixture of KNO3 and
K2SiO3 (SiO2/K2O = 2.5:1 wt %, Alfa Aesar) with the portion
of K2SiO3 being 10%, 50%, or 100% gaining an overall K
content of 3 wt %. These titania-supported catalysts were dried
at 353 K overnight, and calcined in static air at 573 K for 4 h.
Such catalysts are abbreviated as XK-YK(Si)-Au/TiO2, where
″X″ and ″Y″ stand for the particular K weight concentration
from KNO3 and K2SiO3, respectively.
To prepare the second batch of catalysts, we first
impregnated TiO2 anatase with K2SiO3 dissolved in concen-
trated ammonia (pH of solution 11−11.5) according to an
incipient-wetness method. The nominal K loading was 1 wt %.
In a next step, the obtained TiO2−K2SiO3 precursor was dried
at 353 K and further impregnated with an aqueous solution of
HAuCl4·xH2O to achieve a loading of Au of 2 wt %. After an
additional drying step, a concentrated ammonia solution was
added to the Au-loaded TiO2−K2SiO3 precursor for an
incipient wetness deposition of Au(OH)3, held impregnated
for 10 min, and washed with water under filtration for chlorine
removal. Hereafter, Au/TiO2−K2SiO3 was dried overnight at
353 K and then doped with K through an incipient-wetness
method using an aqueous solution of KNO3 to obtain 1, 3, and
5 wt % K. After the last doping step, all the titania-supported
catalysts were dried at 353 K overnight and calcined in static air
at 573 K for 4 h. These catalysts were called XK−Au/TiO2−
K2SiO3 where ″X″ stand for the particular K weight
concentration from KNO3. Finally, all as-prepared precursors
were pressed and sieved to yield particles of 250−450 μm.
The weight content of Au and K in the fresh but calcined
catalysts was determined by inductively coupled plasma optical
emission spectrometry (ICP-OES) using a Varian 715 emission
spectrometer.
The Brunauer−Emmet−Teller specific surface areas (SBET
)
were determined from nitrogen adsorption−desorption iso-
therms collected at 77 K on BELSORP-mini II (BEL Japan,
Inc.).
2. EXPERIMENTAL SECTION
2.1. Preparation and Characterization of Au-Contain-
ing Catalysts. K−Au/SiO2 materials with the nominal Au
loading of 2 wt % and different (1−4 wt %) amounts of K were
prepared by sequential deposition−precipitation and incipient
impregnation methods, similar to the procedure for K−Au/
TiO2 catalysts.8 SiO2 (Grade 646, Davisil, SBET of 301 m2 g−1)
was initially calcined in air at 773 K for 8 h. Two grams of
calcined SiO2 were suspended in 200 mL deionized water, and
then the appropriate amount of HAuCl4·xH2O (41.1 wt % Au,
Chempur) was added. The suspension was stirred at 343 K for
1 h and afterward cooled to room temperature. Ammonia (25%
aqueous solution, Roth) used as a reducing agent was added to
the aqueous suspension until a pH value of around 10−10.5
was reached. This pH range was chosen according to the results
of Somodi et al.17 After an additional 10 min of stirring, Au/
SiO2 precursor was filtrated, thoroughly washed, dried at 353 K
The size of Au NPs on the surface of the catalysts was
determined by high-angle annular dark field scanning trans-
mission electron microscopy (HAADF-STEM) employing an
aberration-corrected JEM-ARM200F microscope operated at
200 kV and equipped with a JED-2300 energy-dispersive X-ray
spectrometer (EDXS, JEOL) for elemental analysis. Scanning
transmission electron microscopy (STEM) operations were
done using a Cs-corrector CESCOR (CEOS). Samples were
deposited on a holey-carbon-supported copper grid of mesh
size 300. Particle size distributions of supported Au NPs were
calculated from at least 150 particles with ImageJ software.18
Powder X-ray diffraction (XRD) analysis was carried out
using an X’Pert Pro (Panalytical, Almelo) at the 2θ range of 5−
80° and using a silicon standard. Phase composition of the
samples was identified with the program WINXPow by
3318
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
STOE&CIE with inclusion of the Powder Diffraction File
PDF2 of the ICDD (International Centre of Diffraction Data).
Reduction and/or desorption behavior of preadsorbed CO2
on the catalysts were probed by temperature-programmed
reduction measurements with H2 (H2-TPR). Each fresh catalyst
(80 mg) was loaded into a fixed-bed tubular quartz reactor and
heated up to 523 K with 10 K min−1 in a flow of 5 vol % H2 in
Ar. Afterward, the flow was replaced by a flow of 10 vol % CO2
in Ar for 2 h. The reactor was cooled in the same flow to ca.
300 K. After purging in Ar, it was heated in a flow of 5 vol % H2
in Ar up to 1173 K with 10 K min−1 and held at this
temperature for 20 min. The feed components and the reaction
products were analyzed by an online quadrupole mass
spectrometer (Pfeiffer Vacuum OmniStar 200).
2.2. Continuous-Flow Catalytic Tests. Catalytic tests
were performed at 473 and 523 K and 2 MPa using an in-house
developed setup equipped with 50 continuous-flow fixed-bed
reactors operating in parallel. CO2 (4.5, Air Liquide), H2 (5.0,
Air Liquide), C2H4 (3.0, Linde), and N2 (5.0, Air Liquide) were
the reactants and used without further purification. Sub-
sequently, 300 mg of each catalyst (fraction of 250−450 μm)
was placed in stainless-steel tube reactors (ID = 4 mm) with 3-
fold weight of SiC for dilution and heated up to 523 K in a N2
flow (6.67 mL min−1 per reactor). Hereafter, the N2 flow was
replaced by a flow (6.67 mL min−1 per reactor) of a reaction
feed CO2/H2/C2H4/N2 = 4:2:1:1 for 24 h. Then, a CO2/H2/
C2H4/N2 = 1:1:1:1 mixture was fed to the reactors at 473 and
523 K. The contact time was kept constant at 45 g min l−1.
Steady-state data were collected after at least 10 h stabilization.
The feed components and the reaction products were analyzed
by an online GC (Varian CP-3800) equipped with FID (HP-
Plot Q) and TCD (HP Plot Q and Molsieve 5A) detectors.
Yield of propanol/propanal (eq 1) was calculated on the CO2
or C2H4 basis taking into account that one oxygenate molecule
is formed from one CO2 and one C2H4 molecule. Conversion
of both C2H4 and CO2 (eq 2) was calculated as the sum of
yields of products from the corresponding feed components.
Equation 3 was used for calculating product selectivity.
mg) was packed between two layers of quartz particles (sieve
fraction of 250−450 μm) into a quartz reactor (ID = 6 mm,
length = 40 mm) within the isothermal zone of the reactor. It
was heated at ambient pressure to 523 K with a ramp of 10 K
min−1 in a flow of H2 and Ar fed with rates of 5 and 15 mL
min−1, respectively. After exposition to this mixture for 1 h, the
reactor was evacuated to ca. 10−5 Pa followed by cooling to 473
K. Hereafter, single pulse experiments with CO2/Ar = 1:1,
C2H4/Ar = 1:1 or H2/D2/Ar = 1/1/1 mixtures were performed.
CO2 (4.5), Ar (5.0), H2 (5.0), and C2H4 (3.0) were supplied by
Linde, whereas D2 was purchased from CK Special Gases Ltd.
An overall pulse size was approximately 1015 molecules or Ar
atoms, respectively. At this pulse size, gas-phase collisions are
suppressed (Knudsen diffusion regime),20 so purely heteroge-
neous reaction steps were analyzed. For each atomic mass unit
(AMU), the pulses were repeated 10 times and averaged to
improve the signal-to-noise ratio. The feed components and
reaction products were quantitatively analyzed by an online
quadrupole mass spectrometer (HAL RC 301, Hiden
Analytics). The following AMUs were used for mass-
spectrometric analysis: 44 (CO2), 28 (CO and CO2), 26
(C2H4), 4(D2), 3(HD), 2(H2) and 40 (Ar). The concentration
of feed components and reaction products was calculated from
the areas of signals recorded at the respective AMUs using
standard fragmentation patterns and sensitivity factors. The
fragmentation patterns and respective sensitivities of feed
components and reaction products were determined from
separate calibration experiments using a reactor filled with
quartz particles (rinsed in HNO3; 250−355 μm), which were
inert for the reactions studied.
3. RESULTS AND DISCUSSION
3.1. Physico-Chemical Catalysts Characterization.
Table 1 presents the weight concentrations of Au and K,
Table 1. Catalysts and Their Selected Physicochemical
Properties
a
b
c
SBET
/
SBET
/
dAu NPs
/nm
catalysts
Au/SiO2
Au/wt % K/wt % m2 g−1 m2 g−1
n − n
̇
̇
i
i,0
Y(i, f) =
X(f) =
× 100%
1.5
1.6
1.6
1.6
1.5
1.7
1.6
1.6
1.6
-
277
261
260
234
223
-
225
222
156
53
4.5
5
n
̇
(1)
(2)
f,0
1K−Au/SiO2
2K−Au/SiO2
3K−Au/SiO2
4K−Au/SiO2
2.7K-0.3K(Si)-Au/TiO2
1.5K-1.5K(Si)-Au/TiO2
3K(Si)-Au/TiO2
1K−Au/TiO2−K2SiO3
0.9
1.8
2.7
3.6
2.7
2.6
2.8
0.9
4.0
4.5
4.5
5
Y(i, f)
∑
i
34
46
Y(i, f)
X(f)
S(i, f) =
× 100%
-
45
6
(3)
-
47
3
55
57
4 and
23
In these equations,
n
and
n
are the molar flows of feed
i
̇
̇
f
components and reaction products, respectively. Subscript “0”
refers to an inlet molar flow. Turnover frequency (TOF) was
calculated according to eq 4:
3K−Au/TiO2−K2SiO3
5K−Au/TiO2−K2SiO3
1.5
1.5
2.6
48
41
46
42
4
-
4.6
c
a
b
As-prepared catalysts. Spent catalysts. Size determined by HAADF-
F × X(f) × c(f, F) × N
STEM images of spent catalysts.
A
TOF =
m(cat) × Vm × N
(4)
Au atoms
where F is the total feed flow, c(f,F) the concentration of feed
component, Vm the molar volume, m(cat) the mass of catalyst,
and NAu atoms the amount of surface Au atoms calculated
according to ref.19
2.3. Transient Experiments. Mechanistic aspects of CO2
and C2H4 interactions with K−Au/TiO2 and K−Au/SiO2
catalysts as well H/D exchange were studied in the temporal
analysis of products (TAP) reactor.20,21 The fresh catalyst (70
SBET of fresh and used catalysts, and the mean size of Au NPs
on used catalysts. For SiO2-supported catalysts, the K loading
was close to the nominal one, while the Au loading was around
1.6 wt % instead of the intended 2 wt %. The specific surface
area of fresh K−Au/SiO2 catalysts slightly decreased from 277
m2 g−1 to 223 m2 g−1 with an increase in the K loading from 0
to 4 wt %. However, the surface area significantly declined after
catalytic tests. This effect became more pronounced with rising
3319
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
K loading. For example, the area of the fresh 4K−Au/SiO2
catalyst decreased from 223 m2 g−1 to only 34 m2 g−1, whereas
the corresponding values for 1K−Au/SiO2 were 261 m2 g−1
and 222 m2 g−1. Such changes may be caused by partial
dissolution of the dopant in the support resulting in formation
of thermodynamically more stable K2SiO3 compared with K2O
and K2CO3. TiO2-based catalysts possessed lower BET surface
areas, which did not depend on K content.
The size of supported Au NPs and their distribution on the
surface of spent catalysts were analyzed by HAADF-STEM.
Compared with the previously investigated K−Au/TiO2
catalysts,8,9 no visible effect of K loading on the size of Au
NPs could be established for the K−Au/SiO2 catalysts. The
nanoparticles on the surface of the latter materials were in the
range between 4 and 5 nm. The representative images and the
size distribution profiles are shown in Figure S1 in the
Supporting Information (SI). To check if the size changed
under reaction conditions, we also analyzed fresh 3K−Au/SiO2,
which significantly lost its surface area after catalytic tests
(Table 1). The surface of this material was populated by Au
NPs of 4 nm (Figure S2). In other words, the nanoparticles did
not change their size under reaction conditions. Due to the low
contrast between K and Si, it is difficult to reveal where the
promoter is located. However, our previous studies with Cs−
Au/TiO2 catalysts demonstrated that Cs is homogeneously
distributed on the surface of both support and Au NPs.9 No
significant changes in Cs distribution were observed after
reaction. Taking into account similar chemical nature of Cs and
K, it can be expected that reaction-induced changes of K
distribution are also minimal.
3.2. Steady-State Catalytic Performance. Propanol and
CO were the main products formed from CO2, whereas C2H4
was converted into propanol and C2H6. Propanal was detected
in traces owing to high catalyst activity for hydrogenation of
this aldehyde to propanol. No significant changes in catalytic
performance were observed within 170 h on stream under
different reaction conditions (Figure S3 in SI). This is in
agreement with the results of TEM analysis proving stability of
Au NPs against sintering. Since K strongly affects catalyst
performance (Figure 1), which does not, however, alter with
time on stream, promoter leaching under reaction condition
can be practically excluded.
Figure 1 shows CO2 and C2H4 conversion (eq 2) at 473 and
523 K as well as propanol selectivity calculated with respect to
each feed component (eq 3). We shall start with the discussion
of conversion data. Au/SiO2 showed some activity for CO2
conversion only at 523 K (Figure 1a). The conversion
increased when the catalyst was promoted with K; the higher
the promoter content, the higher was the conversion (Figure
1a). The highest conversion of around 5% was obtained over
4K−Au/SiO2 at 523 K. When comparing the above activity
data with those previously reported for K−Au/TiO2 materi-
als,8,9 it becomes obvious that in contrast to Au/SiO2, the K-
free Au/TiO2 catalyst converted CO2 into CO and further into
propanol but this activity decreased with rising K content.
These results strongly suggest the importance of the interplay
between the support, promoter, and Au NPs for CO2 activation
and propanol formation.
Figure 1. Conversion of (a) CO2 and (c) C2H4 as well as selectivity to
propanol calculated with respect to (b) CO2 and (d) C2H4 over K−
Au/SiO2 catalysts at 473 K (closed symbols) and 523 K (open
symbols). Reaction conditions: CO2/H2/C2H4/N2 = 1:1:1:1, 2 MPa,
modified contact time of 45 g min l−1.
reported by us for K−Au/TiO2 catalysts.8,9 The latter materials
showed, however, significantly higher conversion of C2H4.
Au/SiO2 did not produce propanol, while this product was
formed over all K-promoted catalysts. The highest alcohol
selectivity with respect to CO2 of 100% and 70% was obtained
over 1K−Au/SiO2 at 473 and 523 K, respectively, and slightly
decreased with an increase in K loading. The propanol
selectivity with respect to C2H4 was lower because of high
catalyst activity for nondesired C2H4 hydrogenation to C2H6.
This selectivity increased from 0 to 14% with an increase in K
loading (Figure 1d). It is important to highlight that previously
tested K−Au/TiO2 materials8,9 performed with significantly
less selectivity; the selectivity did not exceed 4.2%.
Fundamental origins of the support effect on propanol
selectivity are thoroughly discussed in sections 3.4 and 3.5.
To clarify the effect of the support and the loading on propanol
formation and catalyst activity, we investigated interactions of
CO2 and C2H4 with K−Au/SiO2 and K−Au/TiO2. The results
are presented and discussed in the following sections.
3.3. CO2 and C2H4 Interaction with K−Au/SiO2 and K−
Au/TiO2. The type of CO2 and C2H4 adsorption (reversible vs
irreversible) was established from analysis of CO2/Ar = 1:1 and
C2H4/Ar = 1:1 pulse experiments at 473 K in the TAP reactor
according to the approach described in literature.20 The
experimental data were transformed into a dimensionless
form for discriminating between the pure diffusion process
(the dimensionless Ar response) and reversible/nonreversible
interactions of reactive gases. Figure 2 compares the
dimensionless responses of CO2 and Ar over K−Au/SiO2
with 0−3 wt % K and K−Au/TiO2 with 0−2 wt % K.8 For
Au/SiO2, there is no visible difference between the CO2 and Ar
responses (Figure 2a). In other words, CO2 does not practically
interact with the catalyst but simply diffuses similarly to Ar.
This statement agrees with the results of catalytic tests in
Section 3.2, where no CO2 conversion was observed at 473 K.
The dimensionless CO2 responses obtained in the experi-
ments with K-free Au/TiO2 and all K-doped catalysts are
The conversion of C2H4 over K−Au/SiO2 decreased from
33% to 7% at 473 K and from 72% to 21% at 523 K with an
increase in K loading from 0 wt % to 4 wt %, respectively
(Figure 1c). A similar effect of the promoter was previously
3320
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
Figure 2. Dimensionless responses of CO2 and Ar after pulsing of a CO2/Ar = 1:1 mixture over (a) Au/SiO2, (b) 1K−Au/SiO2, (c) 3K−Au/SiO2,
(d) Au/TiO2, (e) 1K−Au/TiO2, and (f) 2K−Au/TiO2 at 473 K.
located below the corresponding Ar responses and cross them
(Figure 2). This is an indication of reversible CO2 adsorption.
One can see in Figure 2 that the CO2 response tailing (nonzero
CO2 signal at high dimensionless times) increases with rising K
loading in K−Au/SiO2 and K−Au/TiO2. According to ref 20,
this is due to the fact that (i) the strength of CO2 interaction or
(ii) the number of CO2 adsorption sites increase with rising K
loading. Importantly, the tailing is stronger for the K−Au/TiO2
catalysts than for their SiO2-based counterparts. This difference
between these two catalyst systems is explained by stronger
CO2 adsorption over the former materials.
Compared with CO2, C2H4 interacts with the catalysts
significantly weaker, and the strength of this interaction does
not practically depend on K loading. This conclusion was
derived from the analysis of the dimensionless responses of
C2H4 and Ar; these responses do not significantly differ from
each other (Figure S4 in SI).
Further insights into CO2 adsorption and reduction of
adsorbed CO2 species to CO were derived from H2-TPR
experiments performed with catalysts, on which CO2 had been
preadsorbed at 523 K. Irrespective of K loading, CO2 desorbed
from K−Au/SiO2 in three temperature regions (Figure 3);
between 323 and 473 K, between 473 and 673 K, and above
673 K with a desorption maximum at around 743 K. The
amount of desorbed CO2 increased with rising K loading
(Figure 3a). CO2 on the surface of Au/SiO2 was reduced to CO
only above 673 K (Figure 3c), while CO was formed between
373 and 573 K over all K−Au/SiO2 catalysts. The amount of
formed CO increased with K loading, thus supporting the
positive effect of K on CO2 conversion mentioned in section
3.2.
In coincidence with the results of transient and catalytic tests,
K−Au/TiO2 catalysts accumulated significantly higher amounts
of CO2 than K−Au/SiO2 (Figure 3a,b). The CO2 desorption
from the former materials took place in three regions with
maxima at around 383, 600, and 800 K. The amount of CO2
desorbed at the two latter temperatures significantly decreased
upon impregnation of Au/TiO2 with 1 and 2 wt % K. When K
loading was increased to 3 wt %, the most CO2 desorbed at 473
Figure 3. Temperature profiles of CO2 desorption and CO production
on as-prepared (a,c) K−Au/SiO2 and (b,d) K−Au/TiO2 catalysts
doped with (gray ) 1 wt % K, (- - -) 2 wt % K, ( - ) 3 wt % K,
( ) 4 wt % K or () without K.
K, whereas the peak at 383 K completely disappeared. Based on
our previous FTIR studies,8 such changes in CO2 desorption
behavior with rising K loading can be result of formation of
ionic bidentate carbonates over 3K−Au/TiO2. The highest
amount of CO formed over this catalyst was observed at
around 650 K. The K−Au/TiO2 catalysts with 1 and 2 wt % K
produced CO at 433 and 503 K, respectively. The shift in the
temperature of maximal CO formation indicates a decrease in
the rate of CO generation with rising K loading, thus
supporting the results of catalytic steady-state tests (Figure 1).
3321
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
3.4. Effect of Support and Promoter on CO2
Conversion. Our previous mechanistic studies with TiO2-
based catalysts showed that CO2 adsorbed on the support in
the vicinity of Au NPs is reduced to CO by hydrogen surface
species.8 Gas-phase H2 and C2H4 are activated on Au NPs. As a
consequence, propanal/propanol formation takes place over
such particles, too. The same mechanistic picture should be
valid for K−Au/SiO2 as discussed below. The fact that the CO2
conversion was zero over Au/SiO2 at 473 K but increased upon
doping with K proves that CO2 adsorption takes place on basic
sites created by the promoter homogeneously distributed on
the surface of catalysts. As C2H4 was hydrogenated over both
nondoped and doped Au/SiO2, Au NPs should be responsible
for activation of this olefin.
Despite the above similarities between K−Au/TiO2 and K−
Au/SiO2 with respect to activation of CO2 and C2H4, these two
catalytic systems differ in the effect of K loading on propanol
formation. Importantly, the K−Au/TiO2 catalysts converted
CO2 to propanol in the absence of the promoter but lost their
activity with rising K content. In contrast, Au/SiO2 did not
produce propanol, but the activity strongly increased with rising
K loading. Such different behavior should be related to the
catalyst ability for CO2 reduction to CO with the latter being
required for ethylene hydroformylation to propanal, which is
the precursor of propanol.8−10
The above difference between K−Au/TiO2 and K−Au/SiO2
can be explained when assuming that promoting Au/SiO2 with
K results in formation of K2SiO3, which does not decompose to
K2O upon calcination12 and cannot be converted to K2CO3
under reaction conditions due to thermodynamic reasons. This
process would explain the negative effect of promoter on the
SBET of K−Au/SiO2 (Table 1). Such silicate cannot be formed
on the surface of K−Au/TiO2, where K is stabilized as K2O and
K2CO3 in the presence of CO2. Indeed, our in situ FTIR studies
revealed formation of different carbonate species upon catalyst
contact with reaction feeds.8 Thus, basic properties of K2SiO3
and K2O/K2CO3 should determine the activity of K−Au/SiO2
and K−Au/TiO2 for CO2 conversion to CO with the latter
reacting with C2H4 to propanal in a separate step. However, our
XRD analysis did not reveal any crystalline K2SiO3 phase
(Figure S5 in SI).
To verify the hypothesis about the effect of potassium silicate
on catalytic performance, we prepared three additional 3K−
Au/TiO2 catalysts using a mixture of KNO3 and K2SiO3 for
impregnation of Au/TiO2. Importantly, all these catalysts
possessed 3 wt % of K, but the ratio of KNO3/K2SiO3 was
varied to achieve 10%, 50%, or 100% of the promoter from the
silicate. To distinguish between the different K sources in these
catalysts, we call them as XK-YK(Si)-Au/TiO2, where ″X″ and
″Y″ stand for the particular weight concentration of K from
KNO3 and K2SiO3 respectively. The SBET of these catalysts is
around 46 m2 g−1 (Table 1). The representative images and the
size distribution profiles of Au NPs are shown in Figure S6
(SI).
Figure 4. Conversion of (a,b) CO2 and (c,d) its selectivity to propanol
over (a,c) K−K(Si)-Au/TiO2 and (b,d) K−Au/TiO2−K2SiO3 at 473
K (closed symbols) and 523 K (open symbols). Reaction conditions:
CO2/H2/C2H4/N2 = 1:1:1:1, 2 MPa, modified contact time of 45 g
min l−1.
to CO in the presence of hydrogen species formed from gas-
phase H2 over Au NPs. It is also worth mentioning that the
selectivity to propanol with respect to CO2 did not practically
depend on the content of K2SiO3 and was around 100% and
between 40% and 60% at 473 and 523 K, respectively (Figure
4c). Owing to high CO2 conversion, the obtained propanol
yield of 3.5% is higher than 2.4% previously reported by us for
K−Au/TiO2 catalysts prepared without K2SiO3.8,9
For further analyzing the collective influence of potassium
silicate and oxide, we synthesized three additional catalysts as
follows. Initially, TiO2 was impregnated with K2SiO3 to gain a
support containing 1 wt % K. Hereafter, the modified support
TiO2−K2SiO3 was loaded with Au using an aqueous solution of
HAuCl4. So-prepared Au/TiO2−K2SiO3 was additionally doped
with KNO3 (1, 3, or 5 wt % K). According to our ABF-STEM
analysis, the silicate covers the surface of TiO2 as a thin layer
(about 1 nm), where Au NPs are located (Figure 5).
Histograms of Au NPs for the corresponding catalysts are in
Figure S8.
Analogously to K−Au/SiO2, the conversion of CO2 over K−
Au/TiO2−K2SiO3 increased with K loading (Figure 4b). This
effect is particularly pronounced for tests at 523 K. It is
important to highlight that the conversion values are higher
than those obtained over K−Au/TiO2 loaded with the same
amount of K.8,9 This is due to the fact that CO2 is weaker
adsorbed on K2SiO3 than on K2O and therefore easily reduced
to CO on Au NPs, which are actually located on the silicate
(Figure 5). However, in contrast to K−Au/TiO2, the selectivity
to propanol over K−Au/TiO2−K2SiO3 decreased from 80% to
25% at 473 K and passed over maximum of 45% at 523 K with
an increase in K loading from 1 wt % to 5 wt % (Figure 4d).
Nevertheless, the propanol yield over K−Au/TiO2−K2SiO3 was
Figure 4a shows that the conversion of CO2 in propanol
synthesis tests increases with an increase in K2SiO3 content.
However, the number of CO2 adsorption sites decreased with
this content as concluded from CO2 desorption tests (Figure
S7 in SI). On the basis of these two results, we suggest that
replacing K2O by K2SiO3 inhibits formation of ionic potassium
carbonates under reaction conditions and consequently the
catalyst activity for CO2 conversion increases because CO2
activated over less basic potassium silicate can be easily reduced
3322
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
Figure 6. Turnover frequencies of C2H4 conversion as a function of K
content in K−Au/TiO2 (squares) and K−Au/SiO2 (circles) catalysts
at 473 K (closed symbols) and 523 K (open symbols). Reaction
conditions: CO2/H2/C2H4/N2 = 1:1:1:1, 2 MPa.
also interacts with Au NPs, thus affecting their performance.
The most likely explanation for the K effect might be a
repulsion of electronic densities between the promoter and the
π-bonding of ethylene, thus resulting in inhibition of ethylene
adsorption.
Figure 5. ABF- and HAADF-STEM images of (a,c,d) 1K−Au/TiO2−
K2SiO3 and (b) 3K−Au/TiO2−K2SiO3. A thin layer of K2SiO3 on
TiO2 surface is indicated by black arrows. Images (c,d) illustrates Au
NPs located on K2SiO3.
The discrepancy between the K−Au/SiO2 and K−Au/TiO2
systems in terms of K effect on the TOF values could be
explained by the fact that the promoter influences not only
ethylene adsorption but also activation of hydrogen on Au NPs.
Our previous study on H/D exchange demonstrated that
promoting Au/TiO2 with K accelerates generation of adsorbed
H species,8 whereas such an effect was not observed over the
K−Au/SiO2 catalysts tested in the present study. From a
kinetic viewpoint, the rate of ethylene hydrogenation depends
on the coverages by ethylene and hydrogen species. Bearing
this in mind, the independence of C2H4-TOF on K loading in
K−Au/TiO2 can be explained as follows. The negative effect of
K on ethylene adsorption is compensated by the positive K
effect for generating hydrogen species. Because such a positive
effect of K was not determined for the K−Au/SiO2 catalysts,
the TOF values of ethylene conversion over these materials
decreased with K loading due to the negative effect of the
promoter on ethylene adsorption (Figure 6).
Another important conclusion from Figure 6 is the effect of
the supporting material on the TOF values for ethylene
conversion; K−Au/TiO2 catalysts showed significantly higher
activity than their SiO2-based counterparts. Because the activity
of Au/SiO2 was at least 2 times lower than that of Au/TiO2, it
can be concluded that, in addition to K, the support also
influences the intrinsic activity of surface Au atoms. The
importance of the kind of support for hydrogenation activity of
Au NPs is further illustrated by the results shown in Figure 7.
This figure shows the conversion of ethylene and the selectivity
to propanol over K−Au/TiO2−K2SiO3 and K−K(Si)-Au/TiO2
catalysts differing in the location of K2SiO3 and Au NPs. As
discussed in section 3.4, TiO2 in K−Au/TiO2−K2SiO3 is
covered by a layer of K2SiO3, and Au NPs are situated on this
layer; contrarily, the nanoparticles are located on the surface
TiO2 and decorated by K2SiO3 in K−K(Si)-Au/TiO2 materials.
One can see in Figure 7a,b that the K−K(Si)-Au/TiO2 catalysts
have a higher hydrogenation activity than K−Au/TiO2−
K2SiO3. On the basis of these results, one can assume that
higher than over K−Au/TiO2 as a result of easier CO2
activation on the former catalysts. A possible reason for the
negative effect of K loading on propanol selectivity is given in
the following section after discussing the factors affecting
ethylene activation.
3.5. Mechanistic and Kinetic Origins of the Effect of
Support and Promoter on C2H4 Hydrogenation. It is in
general accepted that the activity of various Au-containing
catalysts for hydrogenation of olefins and aldehydes increases
with a decrease in the size of supported Au NPs.13,22,23 This
effect is ascribed to facilitated adsorption and activation of
hydrogen (i.e., smaller NPs have higher number of low-
coordinated Au atoms on edges and corners). Such atoms
adsorb and dissociate H2 easier than Au atoms on
terraces.13,24,25 Alternatively, the size of Au NPs influences
their electronic structure, viz., narrowing of the d-band and its
shift closer to the Fermi level.23,26 As a consequence of this
electronic effect, molecules like H2, CO, and O2 dissociate
easier.
To take into account the effect of Au NPs size on catalyst
activity for ethylene conversion, we calculated an apparent
turnover frequency (TOF) of ethylene consumption over K−
Au/SiO2 and K−Au/TiO2 catalysts. To this end, an overall
concentration of surface Au atoms in these catalysts was
estimated from the size of Au NPs as suggested by Abe et al.19
for Ru/TiO2 catalysts. The TOF values obtained at 473 and
523 K are presented in Figure 6. The activity of K−Au/SiO2
strongly decreased with rising K loading, while for K−Au/TiO2
it was practically independent of the promoter content at 473 K
and slightly decreased with K content at 523 K. Because K−
Au/SiO2 catalysts possessed similarly (4−5 nm) sized Au NPs
(Table 1), the promoter should directly influence particle
activity. As demonstrated in our recent study on Cs−Au/TiO2,
the promoter is even located on the Au NPs.10,11 On the basis
of the chemical similarity between K and Cs, we suggest that K
3323
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
4. CONCLUSIONS
K−Au/SiO2 catalysts were demonstrated to efficiently produce
propanol from CO2 and C2H4 in the presence of H2. They
outperform previously developed K−Au/TiO2 in terms of
propanol yield and propanol selectivity with respect to
ethylene. Compared with TiO2, Au NPs supported over SiO2
reveal lower intrinsic activity for the undesired C2H4 hydro-
genation to C2H6. This catalyst property is additionally affected
by K promoter in the K−Au/SiO2 catalysts; the higher the
loading, the lower the Au-related TOF values. Support redox
and promoter basic properties were suggested to influence this
catalyst performance.
It was also established that the promoter affects the rate of
CO2 reduction to CO and thus indirectly propanol production.
In contrast to the TiO2-based catalysts, promoting the Au/SiO2
catalysts with K facilitates CO2 conversion. The main reason for
the distinctive effect of the promoter on this catalyst property is
related to the kind of compound, which is formed from the
promoter under reaction conditions. K is present mainly as
potassium carbonate species on the surface of K−Au/TiO2.
Their reduction to CO by hydrogen species formed upon
activation of gas-phase H2 on Au NPs is inhibited with an
increase in K loading because of formation of stable ionic
carbonates. Due to thermodynamic reasons, in addition to
K2CO3, K2SiO3 species are stabilized on the surface of K−Au/
SiO2, which adsorb CO2 less strongly. As a consequence, the
conversion of CO2 to CO increases with rising promoter
loading because the concentration of such sites increases.
Figure 7. Conversion of C2H4 and selectivity to propanol on (a,c)
3K−Au/TiO2 containing mixed source of dopants and (b,d) K−Au/
TiO2−K2SiO3 at 473 K (closed symbols) and 523 K (open symbols).
Reaction conditions: CO2/H2/C2H4/N2 = 1:1:1:1, 2 MPa, modified
contact time of 45 g min l−1.
ASSOCIATED CONTENT
■
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acscatal.6b00590.
the presence of potassium silicate layers on TiO2 appreciably
reduces C2H4 hydrogenation ability of Au NPs.
HAADF-STEM images and Au NPs distributions,
selected catalytic data, results of CO2 desorption tests,
and XRD data (PDF)
Thus, the location of K2SiO3 (i.e., between Au NPs and TiO2
vs on or at Au NPs) appears to be an important factor
governing the hydrogenation activity of Au NPs. This
conclusion is supported by the fact that K−Au/TiO2−K2SiO3
catalysts lose their activity with rising K loading in a similar way
as determined for K−Au/SiO2 (Figure 1). Moreover, for these
both catalytic systems, propanol selectivity increases with the
dopant loading (Figure 7d and Figure 1d), thus further
stressing the effect of support on selective and nonselective
ethylene transformations, which can be explained as follows.
According to Fujitani et al.,27 Hartfelder et al.,28 and Panayotov
et al.,29 an irreducible support like alumina or silica is less prone
to accommodate the resulting electrons compared to reducible
supports like titania. As a consequence, Au NPs on irreducible
supports show in general low activity for generation of
adsorbed hydrogen species from gas-phase H2.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: Evgenii.kondratenko@catalysis.de (E.V.K.).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Dr. Marga-Martina Pohl for carrying out
ABF- and HAADF-STEM characterization. This work was
financially supported by the Leibniz-Gemeinschaft under the
research grant SAW-2011-LIKAT-2.
In summary, the present study revealed that the interplay
between Au NPs, support, and the source of K promoter are
the key parameters for developing new catalysts for CO2
conversion into propanol in the presence of H2 and C2H4.
When taking into account the fact that nondesired C2H4
hydrogenation to C2H6 in the present study was suppressed
without losing propanol productivity when comparing with
previous literature data,8−10 it is expected that catalytic
performance can be further improved by a proper catalyst
design. To this end, deep understanding of factors governing
the kinetics of CO2, H2, and C2H4 activation is still required.
REFERENCES
■
(1) Hobbs, P. V. Introduction to Atmospheric Chemistry; Cambridge
University Press: Cambridge, U.K., 2000.
(2) Xiaoding, X.; Moulijn, J. A. Energy Fuels 1996, 10, 305−325.
(3) Olah, G. A. Angew. Chem., Int. Ed. 2013, 52, 104−107.
(4) Kondratenko, E. V.; Mul, G.; Baltrusaitis, J.; Larrazabal, G. O.;
Perez-Ramirez, J. Energy Environ. Sci. 2013, 6, 3112−3135.
(5) Federsel, C.; Boddien, A.; Jackstell, R.; Jennerjahn, R.; Dyson, P.
J.; Scopelliti, R.; Laurenczy, G.; Beller, M. Angew. Chem., Int. Ed. 2010,
49, 9777−9780.
(6) Zhang, Z.; Hu, S.; Song, J.; Li, W.; Yang, G.; Han, B.
ChemSusChem 2009, 2, 234−238.
3324
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325

ACS Catalysis
Research Article
(7) Robert, W. D.; Dennis, R. H.; Frederick, W. W.; Heather, D. W.,
Catalytic CO2 Hydrogenation to Feedstock Chemicals for Jet Fuel
Synthesis Using Multi-Walled Carbon Nanotubes as Support. In
Advances in CO2 Conversion and Utilization; American Chemical
Society: Washington, DC, 2010; Vol. 1056, pp 125−139.
(8) Ahlers, S. J.; Bentrup, U.; Linke, D.; Kondratenko, E. V.
ChemSusChem 2014, 7, 2631−2639.
(9) Ahlers, S. J.; Kraehnert, R.; Kreyenschulte, C.; Pohl, M.-M.;
Linke, D.; Kondratenko, E. V. Catal. Today 2015, 258, 684−690.
(10) Ahlers, S. J.; Pohl, M.-M.; Radnik, J.; Linke, D.; Kondratenko, E.
V. Appl. Catal., B 2015, 176−177, 570−577.
(11) Ahlers, S. J.; Pohl, M. M.; Holena, M.; Linke, D.; Kondratenko,
E. V. Catal. Sci. Technol. 2016, 6, 2171−2180.
(12) Lide, D. R. CRC Handbook of Chemistry and Physics, 84th ed.;
CRC Press: Boca Raton, FL, 2003.
(13) Kartusch, C.; van Bokhoven, J. Gold Bulletin 2009, 42, 343−348.
(14) Sermon, P. A.; Bond, G. C.; Wells, P. B. J. Chem. Soc., Faraday
Trans. 1 1979, 75, 385−394.
(15) Chou, J.; Franklin, N.; Baeck, S.-H.; Jaramillo, T.; McFarland, E.
Catal. Lett. 2004, 95, 107−111.
(16) Mukherjee, P.; Patra, C. R.; Ghosh, A.; Kumar, R.; Sastry, M.
Chem. Mater. 2002, 14, 1678−1684.
̋
́ ́
(17) Somodi, F.; Borbath, I.; Hegedus, M.; Tompos, A.; Sajo, I. E.;
́
Szegedi, A.; Rojas, S.; Fierro, J. L. G.; Margitfalvi, J. L. Appl. Catal., A
2008, 347, 216−222.
(18) Collins, T. J. BioTechniques 2007, 43, S25−S30.
(19) Abe, T.; Tanizawa, M.; Watanabe, K.; Taguchi, A. Energy
Environ. Sci. 2009, 2, 315−321.
(20) Gleaves, J. T.; Yablonskii, G. S.; Phanawadee, P.; Schuurman, Y.
Appl. Catal., A 1997, 160, 55−88.
(21) Per
160−169.
́
ez-Ramírez, J.; Kondratenko, E. V. Catal. Today 2007, 121,
(22) Bus, E.; Prins, R.; van Bokhoven, J. A. Catal. Commun. 2007, 8,
1397−1402.
(23) van Bokhoven, J. A. Chimia 2009, 63, 257−260.
(24) Brodersen, S. H.; Grønbjerg, U.; Hvolbæk, B.; Schiøtz, J. J.
Catal. 2011, 284, 34−41.
(25) Zanella, R.; Louis, C.; Giorgio, S.; Touroude, R. J. Catal. 2004,
223, 328−339.
(26) Bond, G. C.; Sermon, P. A. Gold Bulletin 1973, 6, 102−105.
(27) Fujitani, T.; Nakamura, I.; Akita, T.; Okumura, M.; Haruta, M.
Angew. Chem., Int. Ed. 2009, 48, 9515−9518.
(28) Hartfelder, U.; Kartusch, C.; Makosch, M.; Rovezzi, M.; Sa, J.;
van Bokhoven, J. A. Catal. Sci. Technol. 2013, 3, 454−461.
(29) Panayotov, D. A.; Yates, J. T. J. Phys. Chem. C 2007, 111, 2959−
2964.
3325
DOI: 10.1021/acscatal.6b00590
ACS Catal. 2016, 6, 3317−3325