Dehydrogenation of 1-butene with CO2 over VOx supported catalysts

Article abstract of DOI:10.1016/j.apcata.2020.117648

Butadiene (BD) is an important intermediate in chemical industry and is obtained predominantly from fossil sources. In the future new and sustainable BD sources and synthesis routes might be required to meet the rising demands for this compound. Here, the synthesis of BD from 1-butene in the presence of CO₂ was studied using supported vanadium oxide as active compound. Among the tested supports SBA-15 was shown to be the most effective one and a maximum BD yield of 39 % could be achieved. The VO_x/SBA-15 catalysts were characterized by complementary methods to obtain insight about the effect of catalyst synthesis and vanadium loadings on the formation of VO_x surface structures as well as the catalytic performance. CO₂ has a positive impact on the reaction. Coking was considered to be the main reason for the catalyst deactivation and the decreasing BD yield with increasing time on stream.

Full text of DOI:10.1016/j.apcata.2020.117648

Journal Pre-proof
Dehydrogenation of 1-butene with CO2 over VOx supported catalysts
¨
K. Ifflander (Investigation) (Methodology) (Visualization) (Writing -
original draft)<ce:contributor-role>Writing -review and editing), R.
Eckelt (Investigation) (Resources), H. Lund (Investigation)
(Resources), C. Kreyenschulte (Investigation) (Resources), S.
Bartling (Investigation) (Resources), A. Wotzka, N. Steinfeldt
(Supervision) (Writing - original draft) (Writing - review and editing)
PII:
S0926-860X(20)30241-6
DOI:
https://doi.org/10.1016/j.apcata.2020.117648
APCATA 117648
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
21 February 2020
1 May 2020
18 May 2020
Please cite this article as: { doi: https://doi.org/
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
© 2020 Published by Elsevier.

Dehydrogenation of 1-butene with CO2 over VOx supported catalysts
K. Iffländer, R. Eckelt, H. Lund, C. Kreyenschulte, S. Bartling, A. Wotzka, N.
Steinfeldt
Leibniz-Institut für Katalyse e. V., Albert-Einstein-Str. 29a, 18059 Rostock, Germany
Corresponding author:
Norbert Steinfeldt
Tel.: +49 (0)381 1281 319
E-mail: norbert.steinfeldt@catalysis.de
Graphical abstract
Highlights
 Butadiene (BD) can be formed from 1-butene on VOx/SBA-15 in presence of
CO2
 maximum BD yield and selectivity of 39 % and 46 % were achieved
 CO2 atmosphere has a positive effect on BD yield and coke formation
1

 VOx single sites produced more BD per coke equivalent than polymerized or
crystalline VOx structures
Abstract
Butadiene (BD) is an important intermediate in chemical industry and is obtained
predominantly from fossil sources. In the future new and sustainable BD sources and
synthesis routes might be required to meet the rising demands for this compound.
Here, the synthesis of BD from 1-butene in the presence of CO2 was studied using
supported vanadium oxide as active compound. Among the tested supports SBA-15
was shown to be the most effective one and a maximum BD yield of 39 % could be
achieved. The VOx/SBA-15 catalysts were characterized by complementary methods
to obtain insight about the effect of catalyst synthesis and vanadium loadings on the
formation of VOx surface structures as well as the catalytic performance. CO2 has a
positive impact on the reaction. Coking was considered to be the main reason for the
catalyst deactivation and the decreasing BD yield with increasing time on stream.
Keywords: butadiene, carbon dioxide, vanadium, 1-butene, dehydrogenation
1 Introduction
The climate change caused by the greenhouse gas CO2 urges the industries around
the world to scale down their vast CO2 emissions and to develop processes which use
this non-toxic gas for the production of valuable materials [1]. CO2 has already been
successfully applied as a soft oxidant in the dehydrogenation of light alkanes and ethyl
benzene [2-9]. In contrast to other oxidants CO2 does not over-oxidize the educt, its
use inhibits the formation of hot spots, and lowers explosion risks. Moreover, it is
usually reduced to CO which is a valuable C1 building block.
Butadiene (BD) is an important bulk chemical in the synthesis of elastomers and
polymer resins [10]. Until now BD is predominantly extracted from refinery waste gas
and natural gas condensates and the demand for BD is still expected to increase [11].
A more sustainable way of BD production would be the appropriate processing of
biomass derived feedstocks like ethanol [12] and C4 alcohols [13]. The dehydration of
2

1-butanol, for example, yields butene that can be dehydrogenated to BD [14]. Another
substance which is available from biomass is -valerolactone. It can be converted over
acid catalysts with high selectivity to a mixture of butenes and CO2 [15].
In general, two possible pathways for 1-butene dehydrogenation to BD with CO2 are
discussed: an oxidative and a non-oxidative one [16]. The oxidative dehydrogenation
pathway (ODH) results in the formation of BD, water and CO (1). The non-oxidative
dehydrogenation pathway (DH) leads to the formation of BD and molecular hydrogen
(2). The dehydrogenation reactions can be affected by several side reactions: CO2 can
react with hydrogen that is formed during the reaction to form water and CO (RWGS)
(3), 1-butene can isomerize to cis- or trans-butene (4), and polymerization of the olefins
or adsorbed intermediates results in the formation of coke. Both dehydrogenation
pathways are even at 873 K endergonic but the DH is thermodynamically slightly
favored compared to the ODH. The Gibbs free energy of the RWGS at 873 K is
relatively close to zero which indicates that the reaction can also proceed.
(1)
ΔrG0m (873 K) = 19.2 kJ/mol
(2)
ΔrG0m (873 K) = 12.5 kJ/mol
(3)
ΔrG0m (873 K) = 6.7 kJ/mol
(4)
Most attention in the dehydrogenation of 1-butene with CO2 was paid to Fe based
systems [17-20] but Pt [21] and Cr [16] containing catalysts were also investigated.
The catalytically active metal compound was either deposited on the surface of a
carrier material or incorporated into the support. The materials with high specific
surface areas and/or mesopores were favored since they can provide a high number
of catalytic active sites. Well distributed active sites reveal a higher catalytic stability
compared to Fe bulk phases [18, 22]. The presence of Lewis acid sites which could be
3

also provided by the support had also a positive impact on the BD formation [19, 20].
Furthermore, the catalysts should be able to activate CO2 which was previously related
to the oxygen mobility of the catalytic active compound [19]. In addition, the results in
the dehydrogenation reactions of other light alkanes in presence of CO2 using vanadia
supported catalysts [23-26] encourage testing of those materials in the
dehydrogenation of 1-butene in presence of CO2. However, the features that determine
the catalytic performance of VOx surface species are still being debated [27]. The aim
of the present work is to elucidate the effect of the support, V loading, VOx structures,
and selected reaction parameters on BD yield, production rate, selectivity as well as
coke and side product formation.
2 Experimental
2.1 Catalyst preparation.
The syntheses of Al2O3, TiO2 and ZrO2 supports are described in the supplementary
material (SM). SBA-15 was synthesized as follows: 25 g Pluronic® P123 (Sigma
Aldrich) were poured into 600 g deionized water and 120 g HCl (37 %, Fisher
Chemicals) while stirring at RT. When the solution became clear 48 g of tetraethyl
orthosilicate (TEOS, 98 %, Alfa Aesar) were added. Within 1 h the mixture was heated
up to 45 °C and stirred at this temperature for 24 h. Then, stirring stopped, and the
suspension was kept at 90 °C for 3 days. Afterwards, the suspension was cooled down
to RT and filtrated. The obtained solid was washed with 300 mL deionized water and
dried (1 day at RT, additional 4 h at 90 °C). The dry solid was heated up to 550 °C
within 10 h in a muffle furnace and kept at this temperature for 6 h. Finally, 13 g of
SBA-15 were obtained.
Three different methods were applied to deposit VOx species on the SBA-15
surface. Firstly, V2O5 (≥ 99 %, Merck) and oxalic acid (dihydrate, ≥ 99 %, Merck) were
added to 30 mL deionized water. The amounts added were 20.7 mg V2O5 and 43.4 mg
oxalic acid (Aq1.2), 42.5 mg and 88.3 mg (Aq2.4), or 89.4 mg and 185.8 mg (Aq5.4).
The obtained suspension was heated up to 80 °C under stirring. After 30 min of stirring
a clear blue solution was obtained which was cooled down to RT. Then, 800 mg of
SBA-15 were added to the clear solution and the obtained suspension was stirred at
RT for an additional 30 min. Afterwards, the solvent was evaporated under reduced
pressure (temperature did not exceed 70 °C) and the residual solid was dried at 70 °C
4

for 16 h. Finally, the solid material was calcined in a tubular furnace at 600 °C for 5 h
(heating rate 5 K/min) in an oxygen atmosphere (flow rate 20 mL/min). This method
was also applied to deposit VOx on Al2O3, ZrO2, and TiO2.
Another series of catalysts was prepared by impregnation of dried SBA-15 support
with a solution of vanadium(V) triisopropoxide oxide (Alfa Aesar, 96 %). The
impregnation was conducted inside of a glovebox using dry isopropanol as solvent.
The catalysts were named Im2.4, Im3.7, and Im7.5. 1.000 g SBA-15 was impregnated
with a solution of 0.126 mL V-precursor and 1.180 mL isopropanol (Im2.4), 0.266 mL
and 1.040 mL (Im3.7), or 0.596 mL and 0.710 mL (Im7.5). Afterwards, the resulting
materials were dried overnight under Ar flow in a Schlenk tube immersed in an oil bath
heated up to 120 °C. The obtained powders were calcined as previously described.
For grafting of V on SBA-15, 2.0 g of dried vanadyl acetylacetonate (99 %, Acros)
was poured into 800 mL of dry toluene. The suspension was stirred at RT under Ar
overnight and then filtrated. 800 mg of dry SBA-15 were added to the filtrate and stirred
at RT under Ar for 24 h. The suspension was centrifuged and the residue was washed
3 times with 80 mL toluene. Finally, the solid was dried at 80 °C overnight and calcined
as described earlier. This catalyst was named Gr0.6.
2.2 Characterization.
The V amount was determined by ICP-OES analysis. 10 mg of catalyst were
chemically digested in 8 mL aqua regia using a microwave (Multiwave PRO, Anton
Paar). The obtained solution was filled up to 100 mL and measured in a Varian 715-
ES ICP emission spectrometer.
An ASAP 2010 (Micromeritics) porosimeter was used to determine specific surface
area and pore volume of the catalysts. Granulated samples (grain diameter 200–
315 µm) were heated up to 200 °C at a pressure range of 1 – 10-2 mbar for 3 h. The
isotherms were recorded using liquid nitrogen as adsorbate at a temperature of -
196 °C. The Brunauer-Emmett-Teller (BET) method was used to calculate the specific
surface areas. The pore volumes were obtained from the desorption branch of the
sorption isotherm using the Barrett-Joyner-Halenda (BJH) method.
X-ray diffraction (XRD) powder patterns were recorded on a Panalytical X'Pert
diffractometer equipped with a Xcelerator detector using automatic divergence slits
and Cu Kα1 radiation (40 kV, 40 mA; λ = 0.15406 nm). The measurements were
5

performed in 0.0167° steps. The samples were mounted on silicon zero background
holders. The obtained intensities were converted from automatic to fixed divergence
slits (0.25°) for further analysis. Peak positions and profile were fitted with Pseudo-
Voigt function using the HighScore Plus software package (Panalytical). Phase
identification was done using the PDF-2 database of the International Center of
Diffraction Data (ICDD).
Scanning transmission electron microscopy (STEM) measurements were
performed at 200 kV with a probe aberration-corrected JEM-ARM200F (microscope:
JEOL, Japan; corrector: CEOS, Germany) using annular bright field (ABF) and high
angle annular dark field (HAADF) detectors. The microscope is equipped with a JED-
2300 (JEOL) energy-dispersive x-ray-spectrometer (EDXS) for chemical analysis. The
sample was deposed on a holey carbon supported Cu-grid (mesh 300) without any
pretreatment and transferred to the microscope.
UV-Vis diffuse reflection spectra of dehydrated vanadia catalysts were recorded in
diffuse reflectance mode at room temperature by an Ava Spec-2048 UV-Vis
spectrometer (Avantes Inc., Apeldoorn, The Netherlands) equipped with a FCR-
7UV400C-2ME optical probe (Avantes Inc., Apeldoorn, The Netherlands). BaSO4 was
used as white reference material. Each sample was placed in the middle of a quartz
tube with an inner diameter of about 8 mm and fixed with quartz wool. Some of the
samples were diluted with SBA-15. Each sample was dried under a synthetic air flow
(total flow= 100 mL/min) at 350 °C for 2 h. After cooling to 25 °C both ends of the quartz
tube were closed with plugs. Subsequently the UV-Vis spectrum was recorded at the
position of the dried sample.
The differential thermal analysis (DTA) and differential scanning calorimetry (DSC)
were performed by a Netzsch STA STA449 F3 Jupiter system (reference Al2O3) with a
heating rate of 10 K/min from 25 °C to 800 °C. Nitrogen and synthetic air were used
as carrier gases.
Temperature programmed reduction (TPR) was conducted with a BELCAT II
(MicroBEL) equipped with a TCD detector. The samples (125-140 mg VOx/SBA-15,
14 mg V2O5) were pretreated for 1 h at 600 °C in 30 mL/min synthetic air flow. After
cooling them down to RT in air they were purged subsequently in 30 mL/min Ar flow
for 30 min. For the measurement a gas mixture of 5 % hydrogen in Ar (total flow
6

30 mL/min) was applied and the samples were heated with a rate of 10 K/min from 25
to 1000 °C.
2.3 Catalytic tests.
200 mg of the catalyst (grain diameter 200–315 µm) were placed in a quartz tube
(inner diameter 6 mm) that was installed in a furnace. Inside the reactor a second
quartz tube (outer diameter 3 mm) equipped with a thermocouple was placed. The
catalyst was heated up to 600 °C under a nitrogen flow (30 mL/min). When the reaction
temperature was achieved the gas composition was changed from nitrogen to a
mixture of 1-butene (3 mL/min), CO2 (24 mL/min) and He as internal standard
(3 mL/min). The gas mixtures were analyzed by gas chromatography (GC). The
organic compounds were analyzed with a GC2010 (Shimadzu) equipped with a Select
Al2O3 capillary column and flame ionization detector (FID). The permanent gases were
analyzed in a GC2014 (Shimadzu) equipped with two packed columns (PoraPak Q,
mole sieve), a thermal conductivity detector (TCD), a methanizer and a FID.
Conversions (conv.), yields, selectivities (sel.) and carbon balances (CB) were
calculated from the chromatograms.
Conversion of 1-butene (CO2 conversion was calculated in a similar manner, ṅ is the
flow rate in mmol/min)
inlet
ṅ
− ṅoutlet
inlet
1−butene
1−butene
conv. =
1−butene · 100%
(1)
ṅ
Yield of alkane or alkene (i = C1 – C4, Fi is the number of carbon atoms in the molecule,
e.g., FC1 = 1, F1-butene = 4)
ṅoi utlet · Fi
yieldi =
· 100%
(2)
(3)
inlet
ṅ
· F1−butene
1−butene
Selectivity
yield
sel. =
· 100%
conv.
The carbon balance (CB)
7

ṅoi utlet · Fi
∑
i
CB =
· 100%
(4)
inlet
CO2
ṅ
· FCO2 + ṅ1in−lebtutene · F1−butene
The specific BD production rate (sBDP) describes the BD formation per time per mol
V. MV is the molar weight of V (50.94 g/mol) and xV is the mass fraction of V in the
catalyst.
inlet
1−butene
ṅ
· yieldBD · MV
sBDP =
(5)
mcat · xv
The V surface density (SD) was calculated using the Avogadro constant NA and the
specific surface area obtained from BET measurements (SSA).
xV · NA
SD =
(6)
MV · SSA
3 Results
3.1 Catalytic results
3.1.1 Influence of the support material
Firstly, different metal oxide supports (SiO2, Al2O3, ZrO2 and TiO2) were loaded with
VOx (nominal V loading = 1.2 wt-%) and tested in the dehydrogenation reaction. The
BET surface area of the VOx/MOx catalysts ranged between 24 and 690 m2/g
(Tab. S1). Only SiO2 supported VOx (Aq1.2) showed weak reflections of V2O5 but this
was not the case for the other prepared catalysts (Fig. S1, Fig. 8A). As presented in
Fig. 1A all catalysts revealed a relatively high initial 1-butene conversion which peaked
between 73 and 84 %. VOx/ZrO2 and VOx/TiO2 catalysts showed a decline in 1-butene
conversion of over 60 % within the first 30 min of time-on-stream (TOS). In case of
VOx/Al2O3 1-butene conversion decreased from 84 to 28 % within 5 h. The slowest
decline in 1-butene conversion was observed for VOx/SBA-15 (Aq1.2) where
conversion decreased only slightly from 84 to 75 % within 10 h. As shown in Fig. 1B,
the VOx/SBA-15 revealed also the highest initial BD selectivity among all studied
catalysts (38 %) which, however, continuously dropped with increasing TOS. The initial
BD selectivity of VOx/Al2O3 was the lowest (14 %) but increased with increasing TOS
to about 36 %. The highest BD yield was obtained with VOx/SBA-15 (Fig. 1C). Here,
the BD yield decreased within 10 h from 32 to 15 %. The BD yield of VOx/Al2O3
increased within the first 50 min of TOS to the same value as obtained with VOx/SBA-
8

15 (24 %) but a much faster decline in 1-butene yield with increasing TOS was
observed compared to VOx/SBA-15. ZrO2 and TiO2 supported vanadia catalysts
showed an initial BD yield of 24 and 20 %, respectively, which dropped sharply within
30 min to values below 4 %. Because the VOx/SBA-15 catalyst showed the highest 1-
butene conversion and BD yield this kind of material was chosen for further studies.
100
80
60
40
20
0
50
40
30
20
10
0
A
B
SiO2
Al2O3
ZrO2
TiO2
SiO2
Al2O3
ZrO2
TiO2
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
50
C
SiO2
Al2O3
ZrO2
TiO2
40
30
20
10
0
0
100 200 300 400 500 600
t (min)
Fig. 1. A) 1-Butene conversion, B) BD selectivity and C) BD yield versus time on stream using different
VOx/MO catalysts (nominal V loading = 1.2 wt%). Conditions: 200 mg catalyst, 3 mL/min 1-butene,
3 mL/min He, 24 mL/min CO2, T = 600 °C, p = 1.15 bar.
3.1.3 VOx/SBA-15 catalysts with different V loading
The catalytic performance of the different VOx/SBA-15 catalysts is presented in Fig.
2. Catalysts prepared in aqueous solution (Aq series) showed only relatively small
differences in their performance (Fig. 2A). 1-butene conversion decreased within 10 h
from about 85 to 75 % while the BD selectivity dropped from approximately 37 to about
9

21 %. A maximum BD yield of 32 % was achieved at the beginning of the reaction (Fig.
2C). The catalytic stability of these materials was slightly improved when the V loading
increased from 1.2 to 2.4 wt-%. A further increase in V content (up to 5.3 wt-%) did not
affect the catalytic performance. The catalyst prepared by grafting (Gr0.6) showed a
similar catalytic performance to that of the catalysts of the Aq series despite its lower
V content.
100
80
60
40
20
0
100
80
60
40
20
0
A
B
Im7.5, conv.
Im3.7, conv.
Im2.4, conv.
Gr0.6, conv.
Aq5.3, conv.
Aq2.4, conv.
Aq1.2, conv.
Aq5.3, sel.
Aq2.4, sel.
Aq1.2, sel.
Im7.5, sel.
Im3.7, sel.
Im2.4, sel.
Gr0.6, sel.
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
40
30
20
10
0
40
30
20
10
0
C
D
Aq5.3
Aq2.4
Aq1.2
Im7.5
Im3.7
Im2.4
Gr0.6
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
Fig. 2. 1-Butene conversion, BD selectivity (A, B) and BD yield (C, D) versus time on stream for different
VOx/SBA-15 catalysts. Conditions: 200 mg catalyst, 3 mL/min 1-butene, 3 mL/min He, 24 mL/min CO2,
T = 600 °C, p = 1.15 bar.
For the catalysts impregnated using isopropanol (Im series) the initial 1-butene
conversion increased with increasing V loading (Fig. 2B). Moreover, BD selectivity and
yield achieved their maximum between 28 and 53 min TOS before declining
continuously (Fig. 2B and D). With catalyst Im7.5 a maximum BD yield of 39 % and a
10

maximum BD selectivity of 46 % were achieved. Noticeably, this catalyst exhibited also
the strongest deactivation among all studied catalysts. 1-butene conversions, BD
selectivities and BD yields of all seven catalysts are also compared at 53 and 153 min
TOS in Fig. S7. In general, for conversion, selectivity and yield the following trend can
be observed: Gr0.6 < Aq series < Im series. Within the Im series all three values
increase with increasing vanadium loading. The trend is less pronounced in the Aq
series.
The temporal evolution of trans-, cis-, and iso-butene is shown in Fig. 3A for Aq2.4
and in Fig. 3B for Im7.5. At the beginning of the reaction the BD yield on both catalysts
was clearly higher than those of cis- and trans-butenes. With increasing TOS, BD yield
decreased and that of 2-butenes increased. At longer TOS, the latter became the main
products. The isomerization of 1-butene to 2-butenes was also the main reaction when
using pure SBA-15 as the catalyst (Fig. S2). Iso-butene was formed only in low
amounts (< 2 %) on both catalysts.
Fig. 3C and D show the influence of TOS on CO, hydrogen and BD formation. The
highest amounts of CO and hydrogen were always detected directly at the beginning
of the reaction and the amount of CO was higher than that of hydrogen and BD. CO
and hydrogen formation was stronger on Im7.5 than on Aq2.4. With increasing TOS,
CO and hydrogen amounts decreased on both catalysts and CO achieved a similar
flow rate like BD. It is worth noting that Aq2.4 was also tested in the RWGS reaction
(Fig. S8). As shown, hydrogen and CO2 were consumed to form CO and water. The
evolution of minor organic side products (C1-C4) is depicted in Fig. S4C for Aq2.4 and
in Fig. S5C for Im7.5. Initial yields of C1-C3 cracking products are higher on Im7.5 than
on Aq2.4.
11

100
90
80
70
60
50
40
60
50
40
30
20
10
0
100
90
80
70
60
50
40
60
50
40
30
20
10
0
A
B
conv.
trans
BD
cis
iso
conv.
trans
BD
cis
iso
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
3.5
3.5
C
D
BD
BD
H2
H2
3.0
2.5
2.0
1.5
1.0
0.5
0.0
3.0
2.5
2.0
1.5
1.0
0.5
0.0
CO
CO
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
Fig. 3. 1-Butene conversion and yields of BD, trans-butene, cis-butene and iso-butene using A) Aq2.4
and B) Im7.5. Flow of BD, H2 and CO using C) Aq2.4 and D) Im7.5. Conditions: 200 mg catalyst,
3 mL/min 1-butene, 3 mL/min He, 24 mL/min CO2, T = 600 °C, p = 1.15 bar.
Fig. 4 shows the temporal evolution of the carbon balance (CB). The CB was
calculated from the flow rates of C1-C4 products leaving the reactor (see equation 4).
CB values lower than 100% are assumed to be caused by carbon species that stay
inside the reactor, mainly on the catalyst surface, as coke. In general, coke formation
is strongest directly at the beginning of the reaction and decreases with increasing
TOS to a constant value (Fig. 4A and B). For coke formation within 120 min TOS the
following order was observed: Im series > Aq series > Gr0.6. Moreover, within the Im
series, coke amount increases with increasing vanadium loading. In case of the Aq
series this trend was less pronounced.
12

100
98
96
94
92
90
100
98
96
94
92
90
A
B
Im7.5
Im3.7
Im2.4
Gr0.6
Aq5.3
Aq2.4
Aq1.2
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
Fig. 4. Carbon balance versus time on stream for the different catalysts. Conditions: 200 mg catalyst,
3 mL/min 1-butene, 3 mL/min He, 24 mL/min CO2, T = 600 °C, p = 1.15 bar.
Fig. 5 compares the CO2 conversion of all VOx/SBA-15 catalysts which was
expected to be lower than the 1-butene conversion due to the large excess of CO2 in
the reaction feed. In general, CO2 conversion increased with increasing V loading.
Differences in CO2 conversion for the various V loadings were more significant for the
Im series than for the Aq series. Moreover, CO2 conversion obtained with the catalysts
of the Im series at comparable V loadings was higher than those of the Aq series.
Using Im7.5, up to 11 % of CO2 was initially converted. Within 10 h CO2 conversion
dropped continuously reaching a final value between 1 and 3 %.
12
10
8
12
10
8
A
B
Aq5.3
Aq2.4
Aq1.2
Im7.5
Im3.7
Im2.4
Gr0.6
6
6
4
4
2
2
0
0
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
Fig. 5. CO2 conversion of different catalysts versus time on stream. Conditions: 200 mg catalyst,
3 mL/min 1-butene, 3 mL/min He, 24 mL/min CO2, T = 600 °C, p = 1.15 bar.
3.1.4 Influence of selected reaction parameters
13

Catalyst Aq2.4 was chosen to study the effect of water addition, nitrogen
atmosphere, reaction temperature, and hydrogen pre-treatment on 1-butene
conversion and BD yield (Fig. 6). When water was introduced as steam into the feed
(Fig. 6A) both 1-butene conversion and BD yield decreased compared to the
experiment without steam addition but the drop of conversion and yield with increasing
TOS proceeded in both feed compositions at a similar rate. Fig. S3 compares the
formation of CO and hydrogen as well as the selectivity of 2-butenes of the experiments
with and without steam addition. When the educt feed contained water, the amount of
CO formed was significantly lowered while slightly more hydrogen was produced. The
selectivity of 2-butenes was higher when steam was added.
When the reaction was performed in nitrogen instead of CO2 the 1-butene
conversion was slightly higher, however, the BD yield decreased significantly (Fig. 6B).
Furthermore, the yields of iso-butene (Fig. S4A and B) as well as of propene, ethene
and methane (Fig. S4C and D) were significantly increased. Moreover, the amount of
hydrogen was 3 - 4 times higher in nitrogen than in CO2 atmosphere and only traces
of CO were detected (Fig. S4E and F).
With increasing reaction temperature from 575 °C to 620 °C both 1-butene
conversion and BD yield increased (Fig. 6C). A similar behavior was recently observed
for the Cr-SiO2 catalyst and explained by the fact that higher temperatures favor the
dehydrogenation reaction compared to isomerization [16].
14

100
80
60
40
20
0
100
80
60
40
20
0
A
B
CO2, conv.
N2, conv.
no H2O, conv.
with H2O, conv.
CO2, yield
N2, yield
no H2O, yield
with H2O, yield
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
100
80
60
40
20
0
100
80
60
40
20
0
C
D
620 °C, conv.
600 °C, conv.
575 °C, conv.
conv.
red., conv.
yield
620 °C, yield
600 °C, yield
575 °C, yield
red., yield
0
100 200 300 400 500 600
0
100 200 300 400 500 600
t (min)
t (min)
Fig. 6. Influence of selected reaction parameters on 1-butene conversion and BD yield using Aq2.4; A)
3 mL/min H2O steam were introduced in the educt flow, B) CO2 was replaced with N2, C) influence of
reaction temperature, D) the catalyst was pre-reduced for 90 min at 600 °C with a mixture of 6 mL/min
H2 and 30 mL/min N2 before the educt flow was introduced. Conditions: 200 mg catalyst, 3 mL/min 1-
butene, 3 mL/min He, 24 mL/min CO2 (21 mL/min CO2 in case of A, 24 mL/min N2 in case of B).
The hydrogen pretreatment had almost no effect on 1-butene conversion and BD
yield of Aq2.4 as shown in Fig. 6D. Differences in the as-prepared and pre-reduced
catalyst were observed in the case of Im7.5 (Fig. S5). The reduced catalyst showed a
slightly lower initial 1-butene (3 %) (Fig. S5A, B) and CO2 (3 %) conversion (Fig. S6A),
however, initial BD yield increased from 24 to 35 % (Fig. S5A, B). Moreover, the initial
CB was also improved when the catalyst was pre-reduced (Fig. S6B). Both the reduced
and unreduced catalysts achieved their maximum BD yield within 53 min. Between
53 min and 350 min of TOS the BD yield was slightly higher on the unreduced catalyst
than on the pre-reduced one. Furthermore, the pre-reduction decreased both the initial
15

amounts of organic cracking products (Fig. S5C and D) and of CO and hydrogen
(Fig. S5E and F).
40
30
20
10
0
50
40
30
20
10
0
run 1 run 2 run 3 run 4
run 1 run 2 run 3 run 4
B
A
O2
O2
CO2
N2
CO2
N2
0
50 100 150 200 250 300 350
0
50 100 150 200 250 300 350
t (min)
t (min)
Fig. 7. A) BD yield and B) BD selectivity in a regeneration experiment testing Im2.4. Prior to the first
runs the catalyst was regenerated at 600 °C for 30 min with 15 mL/min synthetic air + 15 mL/min N2 and
purged for 10 min with 30 mL/min N2. Prior the 2nd, 3rd and 4th run the catalyst was pre-treated with 24
mL/min CO2 + 3 mL/min He or with 30 mL/min N2 for 30 min (Conditions: 200 mg catalyst, 3 mL/min 1-
butene, 3 mL/min He, 24 mL/min CO2, T = 600 °C, p = 1.15 bar).
Finally, periodic switch experiments were performed to study the effect of different
regeneration methods on the catalytic performance (Fig. 7). Therefore, used Im2.4 was
re-oxidized inside the reactor using a mixture of air and nitrogen. Afterwards, the 1-
butene/CO2 feed was introduced (first run). As shown, BD yield and selectivity reach
their maximum at 23 min (see also Fig. 2). At the end of the first run BD yield and
selectivity were about 32 and 39%, respectively. At this stage it is assumed that
vanadium sites were already partly reduced and that parts of the catalyst surface were
covered by coke.
When the catalyst was regenerated with CO2 for 30 min, initial BD yield and
selectivity in the second run were higher (34 and 41%) than at the end of the first run.
With increasing TOS, BD yield and selectivity dropped continuously. After the second
regeneration with CO2 (third run), the initial BD yield and selectivity increased again
compared to the last value of the previous run, but with increasing TOS both BD yield
and selectivity dropped again. The same trend was observed for the fourth run. As
expected, a regeneration of the catalyst with CO2 cannot stop the deactivation. The
increase in initial BD yield and selectivity after pre-treatment with CO2 might be
16

attributed to a partial oxidation of reduced vanadium species and/or the removal of
carbon species from the catalyst surface by reaction with CO2 [28]. When using
nitrogen for regeneration, the increase in initial BD yield and selectivity was
significantly lower compared to those with CO2 but the drop in BD yield and selectivity
with increasing reaction time was comparable.
3.2 Characterization of the as-prepared catalysts
XRD powder patterns of the fresh catalysts are presented in Fig. 8A. The pattern of
pure SBA-15 showed a broad peak at 2 = 22° that is assigned to the amorphous silica
structure [29]. The samples Aq1.2, Aq2.4, and Aq5.3 reveal additionally reflections of
an orthorhombic V2O5 phase (PDF 00-041-1426) whose intensity increased with rising
V loading. The size of the V2O5 crystallites calculated from the diffraction patterns using
the Scherrer equation ranged between 37 and 52 nm. STEM images of Aq2.4 showed
the coexistence of ordered and disordered SBA-15 domains (Fig. 9A) and confirm the
existence of crystalline vanadia particles which were located outside of the channels
of the SBA-15 support (Fig. 9B). Moreover, HAADF-STEM indicates the presence of
thin non-crystalline polymeric VOx species on the SiO2 surface (Fig. 9D). The catalysts
of the Im series showed beside the reflection at 2 = 22° additional broad peaks at
26.1° and 50.5° whose intensity increased with increasing vanadia loading. These
reflections cannot be assigned to a known VOx phase.
A
B
Aq5.3
Aq2.4
Aq1.2
Im7.5
Im3.7
Im2.4
Gr0.6
SBA-15
Aq5.3
Aq2.4
Aq1.2
Im7.5
Im3.7
Im2.4
Gr0.6
*
*
*
*
'
*
*
*
*
*
*
*
* *
*
*
*
**
'
'
200
300
400
500
600
700
10 20 30 40 50 60 70 80
2  (deg)
wavelength (nm)
Fig. 8. A) XRD powder patterns of the as-prepared VOx/SBA-15 catalysts, * indicates orthorhombic
V2O5, B) normalized UV-Vis diffuse reflectance spectra of different catalysts (catalysts were dried in N2
flow before measurement).
17

UV-Vis diffuse reflectance spectra of the dried catalysts are shown in Fig. 8B. Bands
located between 250-300 nm were assigned to charge transfer transitions of
tetrahedrally coordinated VOx monomers [30]. Absorption around 350 nm is caused by
VOx tetrahedra which are linked in a chain like manner [30]. V2O5 absorbs at
wavelengths above 400 nm [30]. Gr0.6 and the catalysts of the Im series showed
independent of their V loading an absorption maximum below 300 nm. Im7.5 had
additionally some light absorption around 400 nm. For the catalysts of the Aq series
the absorption maxima vary with increasing V loading between 300 nm (Aq1.2) and
320 nm (Aq5.3) which might be caused by the presence of crystalline V2O5 structures
on these catalysts. The percentages of VOx single sites on the surfaces of the catalysts
were assessed by the method of Gao et al. [31] and results are listed in Tab. 1. The
Gr0.6 sample prepared by grafting contained nearly exclusively VOx single sites.
Samples prepared by incipient wetness impregnation using isopropanol as solvent (Im
series) possessed at similar V loadings a higher amount of VOx single sites than the
samples of the Aq series. Because the catalysts of the Aq series contained additionally
crystalline V2O5 particles the percentage of single sites determined by UV-Vis
spectroscopy is less exact [32]. Generally, the population of single sites decreased
with increasing V loading for both series and those of polymeric VOx units and even
V2O5 increased indicated by the bathochromic shift in light absorption.
18

Fig. 9. HAADF-STEM images of catalyst Aq2.4 showing A) ordered and disordered structures of the
support; B) the presence of relatively large vanadia crystals; C) a catalyst particle that possesses D)
polymeric non-crystalline VOx domains.
Nitrogen adsorption and desorption isotherms are presented in Fig. S9A and B. All
isotherms can be described as type IV which is typical for a mesoporous material [33].
The pore size distribution derived from the desorption branch of the isotherms are
presented in Fig. S9C and D. Textural parameters derived from N2-sorption
experiments by BET and BJH methods are listed in Tab. 1. The specific surface area
(SSA) and pore volume (Vp) are influenced by the V loading. The sample Gr0.6 which
was prepared by grafting possessed the lowest V amount and exhibited the largest
SSA (700 m2/g) and highest Vp (0.98 mL/g) among all V containing catalysts. In
general, BET surface area, pore volume, and mean pore diameter decreases with
19

increasing V loading. BET surface areas and pore volumes of Aq catalysts are higher
than those of the Im catalysts. As shown by XRD and TEM a certain fraction of the VOx
species was here located outside of the SBA-15 particles reducing the vanadium
amount deposited on the SBA-15 channel surface. SAXS patterns of fresh catalysts
are presented in Fig. S10A and C. Peak maximum of SBA-15 d100 reflection (2/q) was
located at 9.44±0.03 nm (exception Im7.5: 9.3 nm). The mean wall thickness calculated
by subtracting the mean pore diameter determined by the BJH method from the
dimension of the unit cell (a0 = 2d100/31/2) was slightly higher for samples of the Im
series (5 nm) than for the samples of the Aq series (4.6 nm, pure SBA-15: 4.3 nm). In
accordance with the BET results (lower BET surface area and pore volume) this
indicates a higher V loading in the channels of the Im series.
Tab. 1
General characteristics of the fresh catalysts.
Catalyst
V
(wt-%)a)
-
SSA
(m2/g)b)
789c)
700
Vp
(mL/g)b)
1.14c)
0.98
Xmono
(%)
-
H2 uptake
(mmol/g)d)
-
av. ox.
statee)
-
SBA-15
Gr0.6
Im2.4
Im3.7
Im7.5
Aq1.2
Aq2.4
Aq5.3
0.6
97
72
45
2
0.083
0.457
0.716
1.437
0.202
0.461
1.066
3.59
3.05
3.03
3.04
3.28
3.05
2.95
2.4
601
0.80
3.7
559
0.78
7.5
364
0.64
1.2
690
0.98
77f
42f
2.4
628
0.97
f
5.3
528
0.78
-
a) data obtained from ICP; b) granulated material; c) powder; d) obtained from TPR; e) average V
oxidation state after TPR, calculated from H2 uptake; f) calculated despite presence of V2O5
The TPR profiles of the different catalyst are presented in Fig. 10. The shape of the
TPR profile is strongly affected by the synthesis method while the intensity of the TPR
signal and the position of the peak maximum depend on the V loading. The TPR
profiles of Gr0.6 and those of the Im series showed only one approximately symmetric
reduction peak whose intensity increases with increasing V loading. The Gr0.6 catalyst
possesses a maximum which occurs at 545 °C (Fig. 10A). The peak maxima Im3.7
and Im7.5 shifted with increasing V loading to 585 °C and 600 °C, respectively
compared to the 578 °C of Im2.4. This temperature shift might be caused either due to
20

the rising polymerization degree of the VOx species [34, 35] as already indicated by
the presented UV-Vis results or by a decreasing reduction rate with increasing V
loading [35].
The TPR profiles of the Aq series showed either two maxima or a broad peak with
a shoulder. Aq1.2 showed maxima at 564 °C and 631 °C while Aq2.4 at 571 °C and
666 °C (Fig. 10B). The profile of Aq5.3 exhibited besides the maximum at 645 °C a
shoulder at about 580 °C. The first maxima of Aq1.2 and Aq2.4 and the shoulder of
Aq5.3 are in the same temperature range as the maxima of the Im series, thus, they
might be also attributed to the reduction of isolated and polymeric VOx species. The
maxima at higher temperatures (above 630 °C) might be assigned to the reduction of
more bulky vanadium oxide species [23] which were identified by XRD (Fig. 8A). For
comparison TPR profiles of V2O5 bulk phase was also recorded. As shown, the
reduction of unsupported V2O5 only starts at about 600 °C and its TPR profile
possesses several maxima [36]. They occur at higher temperatures than those
observed in the as-prepared catalysts. The differences in the TPR profiles between the
pure bulk V2O5 and the Aq catalysts indicate that the supported nanosized V2O5
particles are much easier to reduce than bulk V2O5 phases.
The hydrogen consumption of the single catalysts is presented in Tab. 1. For most
catalysts the consumed hydrogen amount corresponds to a reduction of V(V) to V(III)
(assumption: 1mol H2 reduces 1 mol V2O5 to 1 mol V2O3). The larger deviation in case
of Gr0.6 is assumed to be caused by the low V content of this sample.
B
A
Im7.5
Im3.7
Im2.4
Gr0.6
V2O5
Aq5.3
Aq2.4
Aq1.2
200 300 400 500 600 700 800 900 1000 200 300 400 500 600 700 800 900 1000
T (°C) T (°C)
Fig. 10. TPR of A) Gr0.6, Im2.4, Im3.7, Im7.5 and B) Aq1.2, Aq2.4, Aq5.3 and V2O5. The signals of all
VOx/SBA-15 catalysts were normalized to the sample mass.
21

3.3 Characterization of the used catalysts
Comparison of the SAXS profiles of selected fresh and used catalysts showed that
the dehydrogenation reaction had no effect on the SBA-15 structure (Fig. S10B and
D). XRD powder patterns of the fresh and used Aq5.3 and Im7.5 catalysts are
displayed in Fig. 11. While fresh Aq5.3 showed distinct reflections of a V2O5 phase, the
used sample revealed exclusively reflections of a rhombohedral V2O3 phase (PDF 00-
034-0187) with relatively low intensity. In contrast to Aq5.3, neither the fresh Im7.5 nor
the used catalyst showed XRD reflexes of a known vanadium oxide phase. The weak,
broad peaks at 2 = 26.1° and 50.6° visible in the fresh sample disappeared in the
used catalyst. Here, only a very broad unknown reflection at 2 = 43.2° was detected.
A
B
Aq5.3 used
Aq5.3
Im7.5 used
Im7.5
+
+
+
*
+
+
+
*
++
*
*
*
°
*
*
*
* *
* *
*
*
*
*
**
10 20 30 40 50 60 70 80
10 20 30 40 50 60 70 80
2  (deg)
2  (deg)
Fig. 11. A) XRD powder patterns of the fresh and used catalysts A) Aq5.3 (* indicates orthorhombic
V2O5, + indicates reflections of a V2O3 phase) and B) Im7.5., signals denoted with ↓ and ° could not be
assigned to any VOx structures.
STEM images of a used Aq2.4 catalyst are presented in Fig. 12. The images show
coke deposits which were located both on the silica surface and on the surface of the
VOx species. Interestingly, the images of the used catalyst also show small VOx
particles with distinct lattice planes which are not observed on the fresh catalyst (Fig.
12C, Fig. S11). This indicates a rearrangement of the VOx species during the reaction.
Coke deposition was further confirmed by EDX (Fig. S12 and 13). Results from XPS
indicate lower atomic Si/C and V/C ratios in used catalysts than in the fresh ones
(Tab. S2). Moreover, the C1s XP spectrum of the used catalyst Im7.5 shows the
presence of coke and oxygen containing carbon species (Fig. S14) [37]. BET
22

measurements of a used Aq2.4 catalyst reveal a decline in surface area of 140 m2/g
and pore volume of 0.18 mL/g compared to the fresh catalyst which is assumed to be
caused by blocking of channels through coke formation.
Fig. 12. HAADF-STEM images (A, B) and ABF-STEM images (C, D) of used Aq2.4 showing A) ordered
and disordered structures of the SBA-15 support; B) non-crystalline polymeric VOx; C) graphite type
carbon layers and distinct patterns of small VOx crystallites; D) VOx crystals covered with coke.
Fig. 13A shows a plot of coke amount versus BD amount obtained after 10 h TOS for
the different catalysts. The amount of coke deposits increases with increasing amount
of BD formed. When CO2 is replaced by nitrogen coking is much stronger while less
BD is produced. In Fig. 13B the ratio mBD/mcoke is plotted versus the V surface density
(SD). For the catalysts of the Im series this ratio seems to be independent of the SD.
23

Higher values of this ratio were obtained for the Aq series. Here, this value decreased
with decreasing number of isolated vanadium surface sites. Gr0.6 which has nearly
only isolated VOx single sites reveals by far the highest mBD/mcoke ratio. The figures
indicate that coke formation depends on BD formation and both are affected by the
structure of the catalytic active vanadium site.
120
100
80
60
40
20
0
70
60
50
40
30
20
10
0
A
B
Im7.5
Gr0.6
Im3.7
Im2.4
Aq2.4
Aq2.4 (N2)
Aq1.2
Aq5.3
Aq2.4
Im2.4
Aq5.3
Aq1.2
Gr0.6
Im3.7
Im7.5
Aq2.4 (N2)
12
14
16
18
20
22
0
1
2
3
SD (V/nm2)
nBD (mmol)
Fig. 13. A) plot of mcoke versus nBD produced during 10 h TOS. The BD amount was calculated from GC
measurements while the coke amount was derived from TGA measurements of the used catalysts. B)
plot of mBD/mcoke versus V surface density (SD).
3.4 Discussion
The structure of VOx species on SBA-15 is affected by the applied synthesis
method. Samples of the Aq series contained isolated VOx units, two-dimensional VOx
domains and nanosized crystalline V2O5. The formation of the crystalline V2O5 phase
can be explained as follows: A certain amount of the V precursor dissolved in water
could not enter the SBA-15 channels because the precursor species agglomerated or
precipitated outside of the SBA-15 channels. Thus, a bulk phase outside of the
channels was formed and the final calcination converted this phase into nanosized
crystalline V2O5 which was detected by XRD and STEM (Fig. 8A, Fig. 9). The amount
of the VOx phase outside of the channels depends on the concentration of the V
precursor in the aqueous solution. In case of the Im catalysts, the V precursor could
completely enter the channels which lead to a better V distribution. After calcination
V(V) was obtained independent of the particular VOx structure. This assumption is
supported by TPR results where the measured hydrogen consumption corresponds
approximately to a reduction of V(V) to V(III) (Tab. 1) [35].
24

When the fresh VOx/SBA-15 catalyst is exposed to the 1-butene/CO2 reaction feed,
different reactions occur at the same time inside the catalyst bed, namely the oxidative
(ODH) and the non-oxidative dehydrogenation (DH) of butenes, the RWGS reaction,
1-butene isomerization, cracking and coking.
In general, the ODH reaction of alkanes both with O2 and CO2 is assumed to require
a 2-electron reduction of one V(V) to a V(III) or of two- neighboring V(V) to V(IV) and
to involve the activation of the C-H bonds of the alkane which is assumed to be the
kinetically relevant step [27, 35]. VOx of high oxidation state can bear both selective as
well as non-selective oxygen species [28]. Moreover, it was suggested that ODH of
propane with CO2 can run also over partly reduced VOx sites [28]. The gas mixture
leaving the reactor at short TOS (< 30 min) contained beside BD and butenes also a
relatively high amount of CO and cracking products like methane and propene (Fig. S4,
Fig. S5). It was found that coke formation was strongest at the early stage of the
reaction because at that time the lowest CB were obtained (Fig. 4). Especially, non-
selective oxygen species of V(V) oxide seem to be involved in coking and cracking
reactions. This assumption is supported by the comparison of the results of the
oxidized (as-prepared) and the pre-reduced catalyst. The oxidized Im7.5 produced
initially more cracking products and CO (Fig. S5) and revealed a lower initial CB
(Fig. S6B) than the pre-reduced catalyst. All catalysts of the Im series reveal an
increase in BD formation within the first 30 min TOS (Fig. 2D). It is assumed that here
the number of non-selective V(V) oxide species is relatively high leading to a stronger
degradation of C4 hydrocarbons and lower initial BD production. An increase in BD
production at the early stage was not observed when Gr0.6 or the Aq catalyst were
used. On these catalysts side reactions caused by non-selective VOx species seem to
be less strong leading to lower coke amounts (Fig. 13).
With increasing TOS the reduction degree of VOx is expected to increase because
CO2 is not able to oxidize the reduced VOx sites completely to their highest oxidation
state [24, 28, 38]. BD is formed both by ODH as well as by DH. The DH leads to the
formation of BD and hydrogen that can be converted with CO2 to water and CO in a
consecutive reaction (RWGS, Fig. S8). The contribution of reduced VOx sites to the
RWGS reaction has been previously suggested [28]. The RWGS shifts the equilibrium
of the DH to the product side and supports the formation of BD. It is believed that the
role of the DH in BD formation increases with increasing TOS.
25

Although the BD yield decreased continuously over time and coke was formed the 1-
butene conversion remained at a relatively high level (Fig. 2A, B). The reason for that
is the formation of 2-butenes whose yield increased with increasing TOS. At longer
reaction times (TOS > 180 min) 2-butenes became the main product (Fig. 3A, B).
Considering the catalytic performance of the bare support (Fig. S2) it can be noted that
besides SiO2 surface sites also VOx sites are involved in the isomerization reaction.
The reason for the increasing selectivity of 2-butenes with increasing TOS might be
the increasing amount of reduced VOx sites. It is known that V(IV) and V(III) are Lewis
acidic while V(V) does not possess Lewis acidity [34]. Presumably, an increase in the
amount of Lewis acid sites is beneficial for double bond isomerization. When water is
added to the educt feed much less CO is detected in the product stream (Fig. S3A)
while the hydrogen flow increases only moderately (Fig. S3B) indicating that water
inhibits rather CO formation than its consumption via RWGS. This means that the ODH
is more hindered than the DH. Presumably, the water molecule inhibits the re-oxidation
of reduced VOx by reversible adsorption. Therefore, the reduction degree and the
acidity of the VOx is assumed to be higher when water is added. Thus, selectivity of 2-
butenes increases (Fig. 3C). The higher selectivity of cis- compared to trans-butene is
probably due to the thermodynamic preference of the latter.
CO2 conversion is depicted in Fig. 5. The highest conversion was achieved at the
beginning of the experiment. In general, CO2 could be consumed by: 1) re-oxidation of
reduced VOx sites [25, 28], 2) reaction with hydrogen produced by DH (RWGS
reaction) [25, 28] or cracking, 3) reaction with carbonaceous deposits [28, 39]. When
CO2 is replaced with nitrogen, ODH can only occur at the very beginning of the reaction
(Fig. 6B, Fig. S4). Therefore, BD is almost exclusively formed by DH which cannot be
supported by the RWGS. Thus, the BD yield was significantly lower. [28, 39]. Reasons
for the higher 1-butene conversion in absence of CO2 were the increasing number of
cracking products, a stronger coke formation, and a faster transformation rate of 1- or
2-butenes to iso-butene.
26

0.03
0.02
0.01
0.00
0.030
0.025
0.020
0.015
0.010
0.005
0.000
A
B
Gr0.6
Aq1.2
Aq2.4
Aq5.3
Im2.4
Gr0.6
Im3.7
Im7.5
Aq1.2
Im2.4
Aq2.4
Im3.7
Aq5.3
Im7.5
0
1
2
3
0
50
t (min)
100
SD (V/nm2)
Fig. 14. Specific BD production rate (sBDP) versus A) V surface density (SD) for the different catalysts
(TOS = 28 min) and B) TOS.
Fig. 14A reveals the impact of the V surface density (SD) on the specific BD
production rate (sBDP), keeping in mind that not all of the VOx was accessible for the
-1
reactants. The highest sBDP (0.023 molBD·s-1·molV ) was obtained for the catalyst with
the lowest SD (0.1 V/nm2) which bears nearly exclusive isolated vanadium sites
(Tab. 1). When the catalyst contains additional 2D-polymerized or 3D-VOx structures,
the sBDP decreases significantly although the surface density was only marginally
higher. Therefore, it can be concluded that isolated VOx sites are more efficient in BD
formation than two or three-dimensional VOx domains. In the dehydrogenation of
propane under Ar atmosphere isolated VOx revealed a higher catalytic performance
than polymerized and crystalline VOx [40]. Fig. 13B shows that the values of sBDP are
affected by the reaction time, but their relative order does not change. The decrease
of sBDP over time is caused by coke formation and probably also by the increasing
reduction degree of VOx.
Due to the two double bonds, BD is assumed to be a better coke precursor than for
e.g. butene [22]. With increasing BD production, the amount of coke increases, too
(Fig. 13A). Catalysts of the Im series produce relatively high amounts of BD and are
more prone to coke formation than the other catalysts. The way coke is formed is
influenced by the particular VOx structure and the distribution of VOx species on the
catalyst surface. The V surface density of Gr0.6 and Aq1.2 is lowest and the distances
between the active VOx sites are larger compared to the distances between VOx sites
on the other catalysts where more VOx sites (especially more polymerized ones) are
present, as indicated in Fig. 8B, Fig. 10. Thus, on Gr0.6 and Aq1.2 BD molecules are
27

formed on sites which are located relatively far from each other or on VOx domains
which contain only a relatively low number of active VOx sites. The formation of coke
from BD molecules on Gr0.6 and Aq1.2 should occur less readily than on catalysts
with higher VOx surface density or domain size. It can be seen, that much more BD
has to be produced on Gr0.6 and Aq1.2 to obtain the same amount of coke like on the
other catalysts (Fig. 13B). Noteworthy, Sokolov et. al. found VOx single sites to have
the lower coking rates in the dehydrogenation of propane than polymerized VOx and
gave the same explanation for their observations [41].
Finally, the BD formation rate of the VOx/SBA-15 catalyst was compared with those
of other active catalysts which were applied in the dehydrogenation of 1-butene with
CO2 (Tab. S3). As shown, FeOx/CeAlOx [20] showed a slightly higher initial activity
than VOx/SBA-15 but its catalytic stability was lower. Moreover, the VOx/SBA-15
exhibits similar BD formation rate and catalytic stability compared to Cr-SiO2 [16].
4 Conclusion
The catalytic performance of VOx species in the dehydrogenation of 1-butene in
presence of CO2 is affected by the support. VOx species loaded on SBA-15 support
possessing high BET surface area and ordered pore structure showed higher BD
yields than other supports. On VOx/SBA-15 catalysts, BD is formed both by oxidative
(ODH) and non-oxidative (DH) reaction pathway. CO2 supports the dehydrogenation
reaction by oxidation of reduced vanadium species, by removing hydrogen via the
RWGS reaction, and by lowering the formation of coke. The extent of ODH and DH on
BD formation is assumed to depend on the reaction time. With increasing TOS, the
reduction degree of VOx species increases because CO2 is not able to re-oxidize
reduced VOx species completely resulting in BD formation mainly by DH. It is assumed
that reduced VOx species are not only involved in DH but also in the isomerization of
1-butene to 2-butenes. The obtained results suggest that VOx single sites are more
efficient in BD formation than polymerized VOx domains and nanosized crystalline VOx
particles. Beside the reduction of VOx surface sites, coking is assumed to be the main
reason for the catalyst deactivation and the decreasing BD yield with increasing TOS.
Both the BD formation and coke formation are affected by the structure of the VOx
species. In the early stage of the reaction, coke formation was strongest. Here, the
28

reaction of BD or butenes with non-selective oxygen in VOx sites is assumed to be
involved in coke formation.
Author statement
K. I.: Investigation, Methodology, Visualization, Writing- Original draft preparation,
Reviewing and Editing R. E.: Investigation, Resources H. L.: Investigation,
Resources C. K.: Investigation, Resources; S. B.: Investigation, Resources A. W.:
Investigation, Resources N. S.: Supervision, Writing- Original draft preparation,
Reviewing and Editing
Declaration of interests
The authors declare that they have no known competing financial interests or
personal relationships that could have appeared to influence the work reported in this
paper.
5 Acknowledgement
The authors owe thanks to the Leibniz organization for the financial support. Moreover,
thanks go to Prof. Sergey Verevkin for providing the thermodynamic data, Dr. Dominik
Seeburg and Benny Kunkel for recording the UV-Vis spectra as well as to Dr. Nikolaos
Moustakas for helpful discussion.
References
[1] M.R. Allen, D.J. Frame, C. Huntingford, C.D. Jones, J.A. Lowe, M. Meinshausen,
N. Meinshausen, Warming caused by cumulative carbon emissions towards the
trillionth tonne, Nature, 458 (2009) 1163.
[2] M.D. Porosoff, M.N.Z. Myint, S. Kattel, Z. Xie, E. Gomez, P. Liu, J.G. Chen,
Identifying Different Types of Catalysts for CO2 Reduction by Ethane through Dry
Reforming and Oxidative Dehydrogenation, Angew. Chem. Int. Ed., 54 (2015) 15501–
15505.
29