X-H bond activation on Cr(III),O sites (X = R, H): Key steps in dehydrogenation and hydrogenation processes

Article abstract of DOI:10.1021/acs.organomet.6b00744

We synthesized isolated Cr(III) sites on SiO₂-Al₂O₃ and Al₂O₃ by grafting and subsequent controlled thermal treatment of Cr(OSi(OtBu)3)3(THF)2 and Cr(Al(OiPr)4)3 on alumina. CrOx/Al₂O₃ was obtained from incipient wetness impregnation of Al₂O₃ with CrO3 in H2O followed by calcination, as carried out for the synthesis of industrial Cr-based dehydrogenation catalysts. These materials were characterized by IR, EPR, XAS, and by the adsorption of the probe molecules CO and pyridine, and the results were compared to previously reported isolated Cr(III)/ SiO₂. All of these materials were active in propane dehydrogenation at 550 °C, where higher TOFs were obtained for Cr(III)/SiO₂-Al₂O₃ and Cr(III)/Al₂O₃ than for CrOx/Al₂O₃ and Cr(III)/SiO₂. For mechanistic studies the reverse reaction, propene hydrogenation, was studied. Here, the order of reactivity was opposite that of dehydrogenation, with CrOx/Al₂O₃ and Cr(III)/SiO₂ being more active in hydrogenation than Cr(III)/SiO₂-Al₂O₃ and Cr(III)/Al₂O₃. Kinetic analysis and labeling studies with D2 showed that the rate law is in all cases first order in H2 partial pressure but had different dependence on propene partial pressure from catalyst to catalyst. We found small normal kinetic isotope effects of 1 ≤ KIE ≤ 2, activation enthalpies up to 40 kJ mol-1, and large negative activation entropies between -100 and -194 J K-1 mol-1 for the different Cr catalysts. We also performed parahydrogen-induced polarization (PHIP) experiments, which showed that H2 addition to propene proceeds, at least in part, via a pairwise mechanism. The experimental data for propene hydrogenation suggests adsorption/ desorption pre-equilibria of H2 (or D2) and propene followed by H2 activation and insertion of propene. DFT computations for propane dehydrogenation and propene hydrogenation on Cr(III) on a periodic alumina model show that a mechanism involving X-H activation (X = R, H) is energetically feasible with activation enthalpies and entropies that are comparable to the experimentally determined values.

Full text of DOI:10.1021/acs.organomet.6b00744

Article
pubs.acs.org/Organometallics
X−H Bond Activation on Cr(III),O Sites (X = R, H): Key Steps in
Dehydrogenation and Hydrogenation Processes
Murielle F. Delley,† Marius-C. Silaghi,† Francisco Nunez-Zarur,†,‡ Kirill V. Kovtunov,§,∥
̃
Oleg G. Salnikov,§,∥ Deven P. Estes,† Igor V. Koptyug,§,∥ Aleix Comas-Vives,*,†
and Christophe Coperet*,†
́
†ETH Zurich, Department of Chemistry and Applied Biosciences, Vladimir-Prelog-Weg 1-5, CH-8093 Zurich, Switzerland
̈
̈
‡Instituto de Química, Universidad de Antioquia, Calle 70 No. 52-21, Medellín, Colombia
§International Tomography Center, 3A Institutskaya St., 630090 Novosibirsk, Russia
∥Novosibirsk State University, Pirogova St. 2, 630090 Novosibirsk, Russia
S
* Supporting Information
ABSTRACT: We synthesized isolated Cr(III) sites on SiO2−Al2O3 and Al2O3 by
grafting and subsequent controlled thermal treatment of Cr(OSi(OtBu)3)3(THF)2
and Cr(Al(OiPr)4)3 on alumina. CrOx/Al2O3 was obtained from incipient wetness
impregnation of Al2O3 with CrO3 in H2O followed by calcination, as carried out for
the synthesis of industrial Cr-based dehydrogenation catalysts. These materials were
characterized by IR, EPR, XAS, and by the adsorption of the probe molecules CO
and pyridine, and the results were compared to previously reported isolated Cr(III)/
SiO2. All of these materials were active in propane dehydrogenation at 550 °C, where
higher TOFs were obtained for Cr(III)/SiO2-Al2O3 and Cr(III)/Al2O3 than for
CrOx/Al2O3 and Cr(III)/SiO2. For mechanistic studies the reverse reaction, propene
hydrogenation, was studied. Here, the order of reactivity was opposite that of
dehydrogenation, with CrOx/Al2O3 and Cr(III)/SiO2 being more active in
hydrogenation than Cr(III)/SiO2-Al2O3 and Cr(III)/Al2O3. Kinetic analysis and
labeling studies with D2 showed that the rate law is in all cases first order in H2 partial pressure but had different dependence on
propene partial pressure from catalyst to catalyst. We found small normal kinetic isotope effects of 1 ≤ KIE ≤ 2, activation
enthalpies up to 40 kJ mol−1, and large negative activation entropies between −100 and −194 J K−1 mol−1 for the different Cr
catalysts. We also performed parahydrogen-induced polarization (PHIP) experiments, which showed that H2 addition to propene
proceeds, at least in part, via a pairwise mechanism. The experimental data for propene hydrogenation suggests adsorption/
desorption pre-equilibria of H2 (or D2) and propene followed by H2 activation and insertion of propene. DFT computations for
propane dehydrogenation and propene hydrogenation on Cr(III) on a periodic alumina model show that a mechanism involving
X−H activation (X = R, H) is energetically feasible with activation enthalpies and entropies that are comparable to the
experimentally determined values.
comparison to analogous silica-supported systems.1 However,
INTRODUCTION
■
one drawback of alumina is that it can catalyze cracking and
coking side reactions due to its surface acidity, leading to
catalyst deactivation. Due to the endothermic nature of alkane
dehydrogenation these reactions have to be conducted at high
temperatures (>500 °C). This together with the cracking and
coking side reactions make it even more difficult to elucidate
the operating mechanism on a molecular level. Industrially,
doping with alkali ions (such as K) poisons these acidic sites
and thereby improves performance. As is often the case in
heterogeneous catalysis, the Catofin catalyst contains only a
small fraction of active sites, making investigation of the nature
The high demand for light olefins and its expected significant
growth in the next decades have put pressure on existing
processes.1 Common techniques used to produce olefins, such
as steam cracking and fluid catalytic cracking (FCC) of
naphtha, are energetically demanding and display low selectivity
toward specific olefins. In addition, the use of shale gas and
light alkanes in crackers has decreased the production of
propene. This has led to an increased interest in alkane
dehydrogenation as an effective route for the direct production
of light olefins, especially propene. Since the discovery of Cr2O3
as a dehydrogenation catalyst,2 CrOx/Al2O3also known as
the Houdry or Catofin catalysthas been one of the most
commonly employed propane dehydrogenation catalysts.3 As a
high-surface-area material, γ-alumina is the most widely
employed support, since the corresponding Cr catalysts often
perform better in dehydrogenation of ethane, for instance, in
Special Issue: Hydrocarbon Chemistry: Activation and Beyond
Received: September 22, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
of the active sites and the reaction mechanism difficult.4
Dehydrogenation activity is thought to be associated with Cr3+
surface ions undergoing a C−H bond activation of propane
across a Cr−O bond, followed by β-H transfer and release of
H2, though this is still controversial (Scheme 1).5 A propane
dehydrogenation mechanism involving a heterolytic C−H
activation has also been proposed for single-site catalysts
based on Fe, Co, Zn, and Ga.6
the mechanism of the reverse reaction, alkene hydrogenation, is
easier because of the favored thermodynamics and the use of
lower operating temperatures. To elucidate the mechanism in
hydrogenation, classical approaches such as isotopic labeling
and kinetic studies have been used.7 Parahydrogen-induced
polarization (PHIP) can give insight into the pathways of
hydrogen transfer similar to that of isotopic labeling.8 PHIP is a
powerful technique that can help distinguishing between
pairwise and nonpairwise hydrogenation mechanisms, even
for heterogeneous catalytic systems.9
Previously, we reported an effective protocol to obtain well-
defined Cr(III) sites on silica using a molecular approach.10
These Cr(III) sites are active in ethylene polymerization and
propane dehydrogenation. Here, we use a similar approach to
prepare isolated Cr(III) sites on silica−alumina and on alumina
and compare these materials to the previously studied Cr(III)/
SiO2 system and the industrial system based on CrOx/Al2O3.
We investigate their reactivity in both propane dehydrogen-
ation and propene hydrogenation and perform kinetic
evaluation and PHIP experiments in order to obtain further
understanding of the operating mechanism. The experimental
results are supported by periodic DFT calculations of propane
dehydrogenation and propene hydrogenation on Cr(III)
surface sites included in an alumina slab model.
Scheme 1. Propane Dehydrogenation, Often Proposed To
Proceed via C−H Activation, β-H Elimination and Release
of H2 on Cr−O Sites (Red), and Propene Hydrogenation as
the Microscopic Reverse Reaction (Blue)
RESULTS
■
Synthesis and Characterization. Cr(III)/SiO2-Al2O3. We
prepared isolated Cr(III) sites supported on alumina by
following a similar synthetic protocol developed for the
All of these factors make the study of the mechanism and
active sites of dehydrogenation difficult. In contrast, studying
Scheme 2. Preparation of Materials: (a) Reaction of Cr(OSi(OtBu)3)3(THF)2 with SiO2‑700 Followed by Thermal Treatment
Leading to Cr(III) on Silica (1) (Previously Reported);10b (b) Reaction of Cr(OSi(OtBu)3)3(THF)2 with Al2O3‑700 Followed by
Thermal Treatment Leading to Cr(III) on Silica−Alumina (2); (c) Reaction of Cr(Al(OiPr)4)3 with Al2O3‑700 Followed by
Thermal Treatment Leading to Cr(III) on Alumina (3); (d) Incipient Wetness Impregnation of Al2O3 with CrO3 in H2O,
Drying, and Calcination Leading to Industrial-Like CrOx/Al2O3 (4)
B
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Table 1. Conversions, Activities, and Selectivities of Catalysts 2−4 in Propane Dehydrogenation at 550 °C and 2.5 bar in
Comparison to Previously Reported Data for Catalyst 1
catalyst
conversion, %
activity
a
selectivity, %
C2H6
2 h
TOF, h−1
b
TON
20 h
C3H6
2 h
CH4
2 h
C2H4
2 h
b
0 h
2 h
0 h
2 h
110c
2
10
5
3
3
19
51
60
26
5
28
34
31
84
528
640
409
72
70
66
88
14
16
17
7
2
11
10
13
2
4
3
3
3
4
3
4
16
17
a
b
TOF in (mol C3H6) (mol Cr)−1 h−1. Initial point measured 6 min 30 s after injection of the gas mixture.
synthesis of isolated Cr(III) sites on SiO2 (1), which involved
the grafting of Cr(OSi(OtBu)3)3(THF)2 on a partially
dehydroxylated support followed by thermal treatment at 400
°C under vacuum (Scheme 2a).10b Cr(OSi(OtBu)3)3(THF)2 in
benzene was reacted with alumina that had previously been
partially dehydroxylated at 700 °C, Al2O3‑700. This gave a pale
blue material along with traces of HOSi(OtBu)3. The resulting
solid was treated at 400 °C for 3 h under vacuum to yield the
pale blue material 2 (Scheme 2b). Material 2 contains 0.21 wt
% Cr and 0.16 wt % Si (Cr/Si = 0.7) according to elemental
analysis, suggesting the formation of silicates on alumina as
previously observed for the corresponding Rh and Ir
5.13 and 4.00 and at g = 2.01 and is consistent with the
presence of Cr(III) (Figure S6 in the Supporting Information).
All the data suggest the formation of Cr(III) dispersed on
silica−alumina, as the result of the formation of surface silicates
upon thermolysis of the −OSi(OtBu)3 ligands of the precursor
Cr(OSi(OtBu)3)3(THF)2.
Cr(III)/Al2O3. In order to generate Cr(III) sites in a pure
alumina environment, we also investigated the grafting and
subsequent thermolysis of Cr(Al(OiPr)4)3. Cr(Al(OiPr)4)3 was
synthesized according to a literature procedure14 and fully
1
characterized via H NMR and EPR as well as XANES and X-
ray diffraction (Figures S7−S9 in the Supporting Information).
X-ray diffraction reveals a distorted-octahedral structure of
Cr(Al(OiPr)4)3 with κ2-coordinated Al(OiPr)4 ligands. Cr(Al-
(OiPr)4)3 was reacted with Al2O3‑700, followed by thermal
treatment under vacuum at 400 °C (Scheme 2c). This gave the
pale blue material 3 with 0.18 wt % Cr and no Si according to
elemental analysis. Analysis of the products released during the
thermal treatment by GC and 1H NMR shows the formation of
propene. The infrared spectrum of 3 displays bands between
3620 and 3800 cm−1 and 1400−1600 cm−1, indicating different
types of Al−OH15 as well as a lack of C−H bands (Figure S10
in the Supporting Information). CO adsorption leads to an IR
band at 2196 cm−1, similar to the blue-shifted band observed
for 2 (Figure S2 in the Supporting Information). The
adsorption of pyridine on 3 also leads to IR bands very similar
to those for 2 (Figures S3 and S4 in the Supporting
Information), suggesting that the Lewis acidities of the Cr(III)
sites in 2 and 3 are similar and that they both contain Brønsted
acidic sites. XANES of Cr(Al(OiPr)4)3 and the material after
grafting and post-treatment (3) have nearly identical edge
energies and each has a low-intensity pre-edge feature at 5989.0
eV which is consistent with the retention of the chromium
oxidation state +3 (Figure S9 in the Supporting Information).13
The near-edge feature shifts slightly from 6004.5, to 6007.0 and
6008.0 eV from Cr(Al(OiPr)4)3 to the material after grafting to
3. Overall the XANES spectrum of 3 is similar to the spectrum
of 2. X-band CW EPR of 3 is very similar to the EPR spectrum
of 2 with two broad signals at g = 5.17 and 4.00 and a sharp
signal at g = 2.01 (Figure S11 in the Supporting Information).
The results from CO and pyridine adsorption, XANES, and
EPR show that 2 and 3 have similar spectroscopic signatures
pointing to similar Cr(III) sites.
complexes.11 Analysis by GC and H NMR shows that 9.7
1
equiv of isobutene, 3.9 equiv of THF, and 0.007 equiv of
tBuOH per Cr are released during the thermal treatment,
higher than the expected 9 equiv of C4 compounds and 2 equiv
of THF per Cr. These observed differences could be due to a
large error in the measurement (ca. 1 equiv per Cr) associated
with the low Cr loading. Material 2 does not show any C−H
vibrations in its IR spectrum, consistent with the release of all
organic residues, as already observed for the material prepared
on silica. However, the bands at 1590, 1560, 1548, 1479, and
1395 cm−1 point to the formation of silica−alumina (Figure S1
in the Supporting Information). Upon contact with CO, an IR
band at 2199 cm−1 is observed, consistent with CO
coordination to a Lewis acidic Cr(III) center (Figure S2 in
the Supporting Information). This CO band is between the two
CO vibrations previously observed for Cr(III) on silica at 2202
and 2188 cm−1 corresponding to two different sites.10b
Adsorption of pyridine on 2 gives bands in the aromatic C−
H stretching (3000−3170 cm−1) and bending (1653, 1622
(sh), 1614, 1577, 1561 (sh, also present in neat 2), 1548 (sh,
also present in neat 2), 1491 and 1450 cm−1) regions (Figures
S3 and S4 in the Supporting Information). These bands
indicate the presence of pyridine bound to Lewis acidic centers,
presumably the Cr(III) and Al centers, and pyridine bound to
Brønsted acidic sites, probably originating from the presence of
silica−alumina.11,12 Pyridine adsorption on 1 does not show
bands at 1653 or at ca. 1550 cm−1, consistent with a lack of
Brønsted acidic sites for the silica-supported material. The
expected band at 1614 cm−1 is slightly red shifted to 1611 cm−1,
which could indicate a slightly lower Lewis acidic Cr(III) center
in 1 than in 2 (Figures S3 and S4). The X-ray absorption near
edge (XANES) spectrum at the Cr K edge of 2 displays an edge
energy consistent with Cr(III) in comparison to the edge
energy of the molecular precursor Cr(OSi(OtBu)3)3(THF)2
(Figure S5 in the Supporting Information).10b,13 XANES also
presents a low-intensity pre-edge feature at 5989.0 eV and a
near-edge feature at 6007.5 eV, similar to the previously
reported catalyst 1.10b X-band EPR of 2 shows signals at g =
CrOx/Al2O3. We also prepared an alumina-supported Cr
oxide, 1.5 wt % CrOx/Al2O3 (4), by impregnation of Al2O3
with CrO3 in H2O and drying at 60 and 120 °C, followed by
calcination under a synthetic air flow at 550 °C (Scheme 2d),
similar to what is done in industry. Infrared spectroscopy of the
CO-exposed material shows a band at 2208 cm−1, which is
slightly more blue shifted with respect to free CO than in 2 or 3
C
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Table 2. Rate Law, Rate Constants, Kinetic Isotope Effects, and Eyring Parameters for Catalysts 1−4 in Propene Hydrogenation
in Continuous Flow at 150 °C (unless Noted Otherwise)
catalyst
rate law
k[H2]
kH (×10−3
)
kD (×10−3
)
KIE, kH/kD
ΔH⧧, kJ mol−1
ΔS⧧, J K−1 mol−1
1
2
3
33(1)
16(1)
2.0(2)
−7(1)
20(4)
−194(2)
−160(10)
−100(20)
−175(4)
−145(2)
k[H2][C3H6]
k[H2]
5.00(1)
6(1)
38(1)
0.422(3)
4.32(2)
6.2(5)
1.157(5)
1.0(3)
40(10)
0(2)
4 (150 °C)
4 (50 °C)
k[H2]
26(4)
0.352(2)
1.5(2)
k[H2][C3H6]
1.20(1)
24.9(7)
(Figure S2 in the Supporting Information; IR of CrOx/Al2O3
can be found in Figure S12 in the Supporting Information).
Infrared spectroscopy of pyridine adsorbed on 4 shows bands at
1622 (sh), 1614, 1577, 1491, and 1450 cm−1, as in the case of 2
and 3, consistent with pyridine adsorption to Lewis acidic Cr
sites, but in contrast to 2 and 3, bands at 1653 or at ca. 1550
cm−1 are lacking, pointing to the absence of a significant
amount of Brønsted acidic sites (Figures S3 and S4 in the
Supporting Information). The edge energy in the XANES of 4
is shifted to higher energies in comparison to 2 and 3 and
shows an intense pre-edge feature at 5991.5 eV (Figure S13 in
the Supporting Information). These features are similar to the
XANES signature of CrO3, which shows that the majority of Cr
sites in 4 are probably in a formal oxidation state of +6.
However, X-band CW EPR of 4 contains an intense dispersive
signal at g = 1.97 (Figure S14 in the Supporting Information),
suggesting the formation of some reduced Cr, possibly Cr(III)
sites, during the material synthesis.
Propane Dehydrogenation. The reactivity of 2−4 in
propane dehydrogenation was evaluated in a continuous flow
reactor (10 mL min−1, 20% in Ar, 2.5 bar) at 550 °C and
compared to what we previously observed for 1.10c The output
gases were analyzed via GC-FID and GC-TCD and used to
calculate conversion, activity, and selectivity for propane
dehydrogenation (Table 1; see also Figures S15−S23 in the
Supporting Information).
The industrial-like CrOx/Al2O3 catalyst 4 gives higher
conversions than Cr(III)/SiO2-Al2O3 (2) and Cr(III)/Al2O3
(3), in line with the significantly higher Cr loading of 1.5 vs
0.21 and 0.18 wt % Cr (Table 1). In terms of activity per mole
of Cr, 3 and 2 are the most active propane dehydrogenation
catalysts followed by 4 and 1. Catalysts 3 and 2 show similar
high initial TOFs of 60 and 51 (mol C3H6) (mol Cr)−1 h−1,
respectively, which decrease to 34 and 28 (mol C3H6) (mol
Cr)−1 h−1 over the first 2 h, allowing cumulative TONs of 640
and 528 over 20 h. Catalyst 4 displays lower initial activity of 26
(mol C3H6) (mol Cr)−1 h−1, which increases over the first 2 h
to 31 (mol C3H6) (mol Cr)−1 h−1 and then steadily decreases,
leading to a TON of 409 over 20 h. In terms of selectivity,
material 4 gives propene with the highest selectivity of 88%,
while 2 and 3 display lower propene selectivities of 70% and
66%, respectively. In all cases the main product is propene,
along with methane, ethane, ethylene, and coke as byproducts.
While we attempted to estimate the amount of coke formed by
measuring the amount of hydrogen in the gas stream, it was
within the experimental error of the analysis and could not be
quantitatively determined. The lower selectivity of 2 and 3 in
comparison to that of 4 is consistent with the larger amount of
Brønsted acidic sites, known to catalyze cracking,1,16 as
suggested by probing the acidity with pyridine (vide supra).
Overall, catalysts 2 and 3 behave very similarly in terms of
conversion, activity, and selectivity in propane dehydrogen-
ation, consistent with similar spectroscopic signatures: EPR and
XANES spectra and vibrational frequencies of adsorbed CO
and pyridine. In comparison to previously reported data for
propane dehydrogenation on Cr(III) on silica (1),10c the three
silica−alumina or alumina supported catalysts (2−4) are
significantly more active. This is presumably due to the
increased Lewis acidity of the Cr(III) sites on 2−4 in
comparison to 1 (vide infra). In terms of selectivity, 4 is
more selective than 1, while 2 and 3 display similar or slightly
lower selectivities for propene than 1, consistent with the
presence of Brønsted acidic sites.
Propene Hydrogenation. Catalysts 1−4 were also tested
for propene hydrogenation at 150 °C in a continuous flow
setup (total flow 30 mL/min, 2 bar) with a gas flow consisting
of propene, H2 or D2, and Ar. The gas flow composition was
systematically varied, in order to measure the rate dependence
on each mixture component. For each gas mixture GC analysis
of the output mixture was performed in triplicate. From these
measurements TOFs and an experimental rate law could be
obtained. Rate constants were calculated from this rate law, and
a comparison of the rate constants obtained with H2 vs D2
produced a kinetic isotope effect (KIE) (see Table 2 and
Figures S24−S48 in the Supporting Information). Due to the
high conversions obtained with catalyst 4, the same experiment
was conducted at the lower temperature of 50 °C. High
conversions were also obtained with catalyst 1 at 150 °C.
The TOFs derived from the measurements were linearly
dependent on the partial pressure of H2 (or D2) for all tested
catalysts, giving rate laws that are first order in PH2 (or PD2); see
Figures S24−S43 in the Supporting Information for TOF vs
partial pressure plots. We only observed a linear dependence on
the partial pressure of propene (PC3H6) for catalysts 2 and 4 at
50 °C. In other cases, the rate law was zero-order in propene.
This is probably due to saturation behavior caused by strong
binding of propene. It could also be caused by competing
mechanistic pathways with different propene dependences. At
lower propene pressure, the TOF dependence on PC3H6 can, in
fact, be reasonably well fit with a Langmuir-type behavior (TOF
= kobsKbindPC3H6/(1 + KbindPC3H6)), while the behavior at higher
pressure could be explained with a competing pathway
involving the binding of a second propene molecule to a Cr
center (see Figures S39−S43). The rate constants for propene
hydrogenation (and the activity) for the different catalysts
follows the order 4 ≈ 1 > 3 ≈ 2: i.e., 4 and 1 are faster than 3
and 2. For all of these Cr catalysts, we observed a normal KIE
that is relatively small (1 ≤ KIE ≤ 2). The highest KIE was
observed for Cr(III)/silica (KIE = 2.0(2)), which is
significantly higher than that for Cr(III)/alumina (KIE =
1.0(3)). A small normal KIE as observed for Cr might arise
from the interplay of an inverse equilibrium isotope effect
(EIE) of H2 chemisorption coupled with a normal KIE for
propene insertion. This may indicate that H2 activation is more
rate-determining in case of Cr(III)/alumina, while for Cr(III)/
silica other mechanistic steps, such as insertion of propene for
D
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Table 3. Selected Results Obtained in Propene Hydrogenation with Parahydrogen using Catalysts 1−4 in ALTADENA
a
Experiments Corresponding to the Highest Observed Pairwise Contributions at a Flow Rate of 5.1 mL s−1
catalyst
temp, °C
conversion, %
signal enhancement
% pairwise
% pairwise (corrected)
1
2
3
4
250
550
450
550
0.6
0.2
0.2
1.2
3
112
80
0.02
0.81
0.57
0.74
0.12
4.5
3.2
102
4.2
a
Given are temperature, conversion, signal enhancements calculated for CH3 protons of propane and lower estimates of percentages of pairwise
hydrogen addition calculated using signal enhancements as uncorrected and corrected values for the effect of incomplete adiabaticity of the transfer
of hyperpolarized propane to the 7.1 T field and its relaxation during this transfer.9e The complete data set can be viewed in the Tables S1−S5 in the
Supporting Information.
instance, get more limiting. However, the interpretation of
isotope effects of reactions with H2/D2 are complicated by
contributions from the loss of rotational and translational
entropy upon reaction.17 This makes it possible for the isotope
effect to cross from inverse (<1) to normal (>1) just by
changing the temperature and does not necessarily indicate a
change in mechanism. Thus, an isotope effect of any
mechanism could be 1.0 just by virtue of the temperature at
which the measurement was made and not by a change in
mechanism.
Catalysts 1−4 were also tested for propene hydrogenation in
a continuous flow setup (total flow 30 mL/min, 2 bar with 23
mL/min of Ar, 5 mL/min of H2, 2 mL/min of propene) as a
function of temperature (50−250 °C). The measured GC data
for each temperature were analyzed for TOFs and rate
constants. The Eyring parameters ΔH⧧ and ΔS⧧ in Table 2
were derived from the corresponding Eyring plots, where ln(k/
T) vs 1/T was modeled by a linear fit (Figures S44−S48 in the
Supporting Information). The obtained ΔH⧧ values are
generally close to 0 or are relatively small positive values up
to 40(10) kJ mol−1, consistent with a mechanism that involves
bond breaking. For catalysts 2 and 3 we found higher activation
enthalpies of 20(4) and 40(10) kJ mol−1 in comparison to the
activation enthalpies of 1 and 4 at 150 °C of −7(1) and 0(2) kJ
mol−1. Large negative activation entropies ΔS⧧ of −100 to
−194 J K−1 mol−1 were obtained in all cases, consistent with a
multimolecular reaction.
least some of the hydrogenation events take place via a pairwise
mechanism (Figure S49 in the Supporting Information). As a
lower limit, we estimate that 0.1, 4.5, 3.2, or 4.2% of
hydrogenation events using catalysts 1−4, respectively, are
pairwise. At the same time, the actual contribution of pairwise
hydrogen addition may be much higher, since it is systemati-
cally underestimated due to paramagnetically induced nuclear
spin relaxation by Cr that cannot be quantified.
DFT Calculations. We also evaluated dehydrogenation of
propane and the reverse reactionthe hydrogenation of
propeneon Cr(III) supported on alumina by first-principles
calculations. We constructed alumina-supported Cr(III) sites
20
from the 110 termination of γ-Al2O3 by substituting one of
the Al (AlIII, AlIVb, and AlIVa) by a Cr(III) site (Figure 1).
Substitution of the AlIII site is the most favored configuration
with respect to AlIVa and AlIVb by 96 and 71 kJ mol−1,
respectively (Table S6 in the Supporting Information).
We evaluated the dehydrogenation of propane according to
the mechanism depicted in Scheme 1, which starts via the
heterolytic C−H activation of propane and forms a Cr(III)−
Parahydrogen-Induced Polarization. Parahydrogen-in-
duced polarization (PHIP) experiments can give valuable
information about the mechanism of a hydrogenation
reaction.8,9 If a hydrogenation reaction is carried out using
parahydrogen-enriched H2 and if signal enhancement from the
polarization effect is observed by NMR in the obtained product,
H2 must be added, at least partially, via a pairwise mechanism.
We performed PASADENA (parahydrogen and synthesis allow
dramatically enhanced nuclear alignment)18 and ALTADENA
(adiabatic longitudinal transport after dissociation engenders
nuclear alignment)19 propene hydrogenation experiments on
the materials 1−4 using p-H2-enriched hydrogen gas at
temperatures of 130−550 °C and at gas flow rates from 0.7
to 5.1 mL/s. Highest conversions were generally observed for
the lowest flow rates. For the observation of a PHIP effect,
though, high gas flow rates are often necessary to reduce the
loss of hyperpolarization by relaxation to thermal equilibrium
during gas transfer from the reactor to the NMR spectrometer
in ALTADENA experiments. The results for propene hydro-
genation at selected temperatures and flow rates are shown in
Table 3 (the complete data set can be viewed in Tables S1−S5
in the Supporting Information). A PHIP effect can be observed
with all of the Cr catalysts reported here, which suggests that at
Figure 1. Single-site model of CrIII substituted at the AlIII site on the
(a) dehydroxylated s0 and (b) hydroxylated s1 (3.0 OH/nm2) 110
surface of γ-Al2O3. Single-site models of CrIII sites substituted at the
(c) AlIVb and (d) AlIVa sites of the 110 surface of γ-Al2O3. Cr, Al, H,
and O atoms are depicted in blue, purple, white, and red, respectively.
E
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
propyl and an O−H bond. This step is followed by a β-H
elimination, which forms a Cr−H bond and propene.
Generation of H2 from Cr−H and the O−H group formed
during the first step can complete the catalytic cycle. The
reverse elementary steps lead to the hydrogenation of propene
to propane. We evaluated the reactivity for the surface sites
where Cr occupies the AlIII and the AlIVb positions, since they
are the two most Lewis acidic and reactive sites of the 110
surface of γ-Al2O3;20a,b they are the most reactive toward H2,
CH4, and CH3X activations (X = F, OCH3).20e,21
Reactivity of the CrIII-O2a Site. s0 Surface. For the propane
dehydrogenation reaction on the CrIII-O2a site of the s0 surface
(Figure 1a), we will describe the calculated Gibbs free energies
at 550 °C. The electronic energies for propane dehydrogen-
ation on the s0 surface are reported in the Table S7 in the
Supporting Information, together with the enthalpic (H),
entropic (S), and Gibbs free energies (G) at 550 °C.
The activation of the C−H bond of propane by the adjacent
O2a oxygen of the CrIII site goes via a transition state that is
typical of a σ-bond metathesis step,22 with an O−H−C propane
angle equal to 161° and a C−H distance equal to 1.420 Å
(Figure 2). The C−H activation forming the Cr−propyl
Figure 3. (a) Gibbs free energy profile at 550 °C (in kJ mol−1) for the
dehydrogenation of propane on CrIII‑IVb sites supported on γ-Al2O3.
(b) Gibbs free energy profile at 150 °C (in kJ mol−1) for the
hydrogenation of propene on CrIII‑IVb sites supported on γ-Al2O3.
Figure 2. Transition-state structure corresponding to the C−H
activation of propane on the CrIII-O2a site. Cr, Al, H, C, and O atoms
are depicted in blue, purple, white, gray, and red, respectively. The C−
H distance is given in Å.
kJ mol−1. At this temperature, the subsequent steps to finally
reach propane are less energetically demanding. The rate-
limiting step has an activation enthalpy (ΔH⧧) equal to 93 kJ
mol−1, whereas the activation entropy (ΔS⧧) is equal to −130 J
K−1 mol−1.
intermediate has a Gibbs energy barrier (ΔG⧧) equal to 245 kJ
mol−1 (at 550 °C) with respect to initial reactants, in a reaction
step endoergic by 141 kJ mol−1 (Figure 3a).
s1 Surface. For a 110 termination comprising an additional
water molecule with an OH bridging the AlIVa centers and the
proton on the O2a atom (Figure 1b), the substitution of the
AlIII site is still preferred, but now the difference in energy with
respect to AlIVa and AlIVb is reduced to 29 and 59 kJ mol−1,
respectively (Table S6 in the Supporting Information). Since
CrIII is the most favorable site, we evaluated the propane
dehydrogenation also on a partially hydroxylated γ-Al2O3
surface with a hydroxyl coverage equal to 3.0 OH/nm2 (s1).
We evaluated the analogous propane dehydrogenation
mechanism as described above for the case of the s1 surface
(denoted s1c in previous studies).20e We found that the
energetics of the reaction do not significantly change with
respect to dehydrated surface s0: The C−H activation is now
slightly more demanding, with a Gibbs free energy equal to 262
kJ mol−1 (Figure 3a and Table S9 in the Supporting
Information). The β-H transfer and the H2 formation are
slightly more favored with respect to the dehydrated surface by
39 and 13 kJ mol−1, respectively.
The subsequent β-hydride transfer that forms the Cr−H
bond and releases propene has a relative energy barrier equal to
ΔG⧧ = 133 kJ mol−1 (+274 kJ mol−1 with respect to initial
reactants) in a step exoergic by 43 kJ mol−1, located 98 kJ
mol−1 with respect to initial reactants. In order to close the
catalytic cycle, the hydride and the proton bonded to Cr and
the O2a centers respectively react in an exoergic step by 83 kJ
mol−1 that releases H2 with a relative energy barrier ΔG⧧ of 120
kJ mol−1. Overall, the propane dehydrogenation is slightly
endoergic at 550 °C (by ΔG = 15 kJ mol−1) and the highest
points in the energy profile correspond to the transition states
for C−H activation of propane and β−H elimination, located at
245 and 274 kJ mol−1, respectively, with respect to initial
reactants. We previously reported a Gibbs free energy barrier
for the initial propane activation at 550 °C equal to 239 kJ
mol−1 for CrIII on SiO2 using a cluster model.10c
For propene hydrogenation, which is the reverse of the
previously described pathway, we calculated the enthalpic,
entropic, and Gibbs free energies at 150 °C (Figure 3b and
Table S8 in the Supporting Information). The main difference
to the pathway for dehydrogenation is that, at this temperature
(150 °C), the highest point in the energy profile corresponds to
the H2 activation, with a Gibbs free energy barrier equal to 148
In the case of the s1 surface, propene hydrogenation at 150
°C has an overall Gibbs free energy of activation equal to 136 kJ
mol−1, which corresponds to the H2 activation step (Figure 3b).
This step has an activation enthalpy (ΔH⧧) and entropy (ΔS⧧)
equal to 81 kJ mol−1 and −130 J K−1 mol−1, respectively. All the
enthalpic, entropic, and Gibbs free energies at 150 °C for this
F
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
pathway are reported in Table S10 in the Supporting
Information.
This formation of the Cr−propyl on the CrIVb site is also the
most favorable one, with a reaction energy which is endoergic
by 108 kJ mol−1 with respect to initial reactants. The β-hydride
transfer on the CrIVb site forming the Cr−H and release has a
relative Gibbs free energy barrier of +105 kJ mol−1, being
located at 213 kJ mol−1 with respect to initial reactants. This
step is exoergic by 41 kJ mol−1. Finally, the H2 coupling
between the Cr−H and the O−H hydrogens has a relative
Gibbs energy barrier equal to 87 kJ mol−1, in a structure having
an energy equal to 154 kJ mol−1 with respect to initial reactants.
Overall, the CrIVb-O2b site is the most reactive of all the
evaluated sites, since the transition states for C−H activation of
propane and β-H elimination are located at 193 and 213 kJ
mol−1 with respect to initial reactants.
For the case of the propene hydrogenation at 150 °C on the
CrIVb-O2b site the three transition states that lead to propane
have similar relative Gibbs energy barrier heights (Figure 3b).
The H2 activation and the insertion of the propene into the
Cr−H bond are associated with Gibbs energies equal to 86 and
83 kJ mol−1, whereas the formation of propane by transfer of
the hydrogen to the propyl takes a value of 93 kJ mol−1. The H2
activation step has an activation enthalpy (ΔH⧧) and entropy
(ΔS⧧) of 31 kJ mol−1 and −130 J K−1 mol−1, respectively, with
respect to initial reactants. All the enthalpic, entropic, and
Gibbs free energies at 150 °C for this pathway are reported in
the Table S14 in the Supporting Information.
Reactivity of the CrIII-O3a Site. s1 Surface. If the O3a site is
involved in the C−H activation step (Figure 1b), the Gibbs free
energy barrier of the C−H activation of propane is equal to 297
kJ mol−1, and thus it is more demanding than for the O2a
center (Figure 3a and Table S11 in the Supporting
Information). Moreover, the reaction energy is more endoergic
with 192 kJ mol−1 with respect to initial reactants in
comparison to 158 kJ mol−1 for the C−H activation on the
O2a center also on the s1 surface. The subsequent β-hydride
transfer has a relative Gibbs free energy barrier of +127 kJ
mol−1, with the transition state located at 319 kJ mol−1 with
respect to initial reactants. In this case this step is ca. +100 kJ
mol−1 more demanding than for the route via the O2a center.
The formation of Cr−H and propene is exoergic by 64 kJ
mol−1. Reaction between the Cr−H and the O−H releases H2
in a process with a very low energy barrier equal to only 7 kJ
mol−1. This is in line with the previous results which show that
the CH4 and H2 activation on AlIII centers and nonadjacent
oxygen atoms is very favorable when the surface is hydroxylated
with a hydroxyl group bridging the AlIVa centers and a
protonated O2a center (s1 surface).20c,e,23 Overall, for the
pathway involving the O3a site the highest point in the energy
profile corresponds to the β-hydride transfer transition state,
which is located at 319 kJ mol−1 with respect to initial reactants.
This is thus a more energetically demanding pathway than the
pathway involving C−H activation on the O2a center.
Propene hydrogenation at 150 °C on the CrIII-O3a site has
an overall Gibbs free energy of activation equal to 180 kJ mol−1
(Figure 3b). In contrast to the pathway via the O2a center, the
most energy demanding step corresponds here to the propene
insertion step with an activation Gibbs free energy (ΔG⧧) of
180 kJ mol−1 and an activation enthalpy (ΔH⧧) and entropy
(ΔS⧧) of 52 kJ mol−1 and −302 J K−1 mol−1, with respect to
initial reactants. All the enthalpic, entropic, and Gibbs free
energies at 150 °C for this pathway are reported in Table S12 in
the Supporting Information.
DISCUSSION
■
As previously developed for Cr(III)/SiO2 (1), we prepared
Cr(III)/SiO2-Al2O3 (2) and Cr(III)/Al2O3 (3) via grafting of
the molecular precursors Cr(OSi(OtBu)3)3(THF)2 and Cr(Al-
(OiPr)4)3 on alumina dehydroxylated at 700 °C followed by a
thermal treatment under vacuum. We also prepared CrOx/
Al2O3 (4) via incipient wetness impregnation and calcination as
commonly carried out in industry.4a CO adsorption revealed
the presence of Lewis acidic Cr centers in all four materials with
an increased Lewis acidity for Cr in 2−4 in comparison to 1, as
demonstrated by the larger blue-shift of their ν(CO)’s. Pyridine
adsorption also showed that Brønsted acidic sites are present in
2 and 3. XANES and EPR show that 2 and 3 contain Cr(III)
sites, while 4 contains mainly Cr(VI) along with some reduced
Cr sites. All materials 1−4 are active in propane dehydrogen-
ation to form propene and H2; as well as methane, ethane,
ethylene, and coke as byproducts. The activity in propane
dehydrogenation decreases in the order 3 > 2 > 4 > 1, on the
basis of their activity at 0 h (TOF). The greater activity of the
silica−alumina or alumina supported materials 2−4 in propane
dehydrogenation over 1 correlates with the increased Lewis
acidity of the Cr sites in 2−4, as evidenced by CO adsorption.
These materials also catalyze propene hydrogenation, at much
lower temperatures. The activity and the rate constants in
propene hydrogenation vary as follows: kH(4) ≈ kH(1) > kH(3)
≈ kH(2), thus showing that the order in activity changes in the
forward versus reverse reaction. While 2 and 3 are the least
active catalysts in propene hydrogenation, they are the most
active in propane dehydrogenation in comparison to 1 and 4.
The increased Lewis acidity of 2 and 3 vs 1 would favor a
mechanism involving a C−H heterolytic splitting, as is
commonly proposed for the mechanism of propane dehydro-
genation (see Scheme 1). It is likely that this increased Lewis
acidity also favors propene adsorption, which could thus lead to
propene inhibition in propene hydrogenation and thereby a
decreased propene hydrogenation activity. This is consistent
Reactivity of the CrIVb-O2b Site. s0 Surface. The activation
of the C−H bond of propane by the CrIVb-O2b site (Figure 1c)
has a Gibbs energy barrier equal to only 193 kJ mol−1 at 550 °C
(Figure 3a and Table S13 in the Supporting Information). This
is the most feasible C−H activation of all the evaluated surface
sites. The C−H distance at the transition state is equal to 1.396
Å, whereas the O−H−C propane angle takes a value equal to
147° (Figure 4).
Figure 4. Transition-state structure corresponding to the C−H
activation of propane on the CrIVb-O2b site. Cr, Al, H, C, and O atoms
are depicted in blue, purple, white, gray, and red, respectively. The C−
H distance is given in Å.
G
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
with the observed dependence of propene hydrogenation
activity on propene partial pressure, which exhibits Langmuir-
type saturation behavior. The data for catalyst 3 suggest a
particularly high adsorption equilibrium constant for propene,
in part, probably due to a large heat of adsorption for olefins on
the alumina support.24 However, the observed reversal in the
order of activity (in comparison with the rates of propane
dehydrogenation) may also be due to increased catalyst
deactivation in propane dehydrogenation or due to a
temperature effect. Despite the fact that these reactions are
technically the microscopic reverse of each other, the rates and
equilibrium constants for each mechanistic step may differ
slightly for the forward vs the reverse reaction due to the
difference in temperature at which propane dehydrogenation vs
propene hydrogenation was measured (550 °C vs 150 °C).
This would be the case if propene inhibition were occurring in
catalysts 2 and 3, since at higher temperatures desorption is
favored. In propane dehydrogenation, catalysts 2 and 3 are less
selective for propene than 4, in line with the presence of
Brønsted acidic sites, which can catalyze cracking.
The study of the reverse reaction, propene hydrogenation,
allows the use of lower reaction temperatures, which prevents
coking to a large extent and simplifies kinetic and mechanistic
studies. All Cr materials follow a rate law in propene
hydrogenation that is linear in the partial pressure of H2 and
in the case of catalysts 2 and 4 (at 50 °C) also linear in the
partial pressure of propene with a Langmuir-type saturation
behavior at higher pressures. Comparison of the reactivity with
H2 vs D2 gives in all cases a small positive kinetic isotope effect.
Eyring analysis gives activation enthalpies close to 0 or with
relatively small positive values and large negative activation
entropies. All kinetic data are consistent with adsorption/
desorption pre-equilibria of H2 (or D2) and propene followed
by H2 activation and insertion of propene. To obtain more
mechanistic insight, we also conducted propene hydrogenation
experiments using parahydrogen. The observation of para-
hydrogen-enhanced signals for propane by NMR for all
catalysts shows that hydrogen is at least 0.1% (1), 4.5% (2),
3.2% (3), and 4.2% (4) added via a pairwise mechanism. The
actual percentage for a pairwise mechanism may be much
higher, since paramagnetic Cr centers are expected to
measurably accelerate nuclear spin relaxation and thus
significantly lower the signal enhancement observed for the
product. In fact, the observation of a signal enhancement using
a paramagnetic catalyst is in itself quite remarkable.
CONCLUSION
■
We have shown that well-defined Cr(III)/SiO2−Al2O3 and
Cr(III)/Al2O3 can be prepared via a molecular approach and
that these materials are active in propane dehydrogenation and
propene hydrogenation. These Cr catalysts display higher
activity in propane dehydrogenation than Cr(III)/SiO2,
consistent with the increased Lewis acidity of the Cr(III)
sites, which can favor initial C−H heterolytic splitting of
propane. In comparison to the industrial catalyst CrOx/Al2O3,
Cr(III)/SiO2-Al2O3 and Cr(III)/Al2O3 are more active but
suffer from a decrease in propene selectivity, probably because
of the presence of Brønsted acidic sites known to catalyze
cracking. In contrast, the propene hydrogenation activity of
Cr(III)/SiO2-Al2O3 and Cr(III)/Al2O3 was lower than that of
Cr(III)/SiO2 and CrOx/Al2O3. This could be associated with
propene inhibition as the result of the presence of the stronger
Lewis acid sites for the first two catalysts. Detailed kinetic
studies suggest activation enthalpies close to zero or small
positive values and large negative activation entropies as well as
small positive kinetic isotope effects. Experiments using
parahydrogen-induced polarization evidence that H2 addition
occurs at least in part in a pairwise fashion: i.e., both H atoms of
a H2 molecule end up in the same propane molecule. First-
principles calculations on a periodic alumina model show that
the most active sites in dehydrogenation and hydrogenation are
four-coordinate Cr(III) sites, in contrast to what is expected. All
experimental findings combined with DFT computations are
consistent with a mechanism involving X−H heterolytic
activation (X = R, H), followed by β-hydride transfer or
propene insertion and release of H2 and propene or propane,
respectively, with an excellent agreement of the experimental
and computational activation parameters.
EXPERIMENTAL SECTION
■
General Considerations. All manipulations were performed
under an inert gas atmosphere using Schlenk (Ar), glovebox (Ar),
and high-vacuum techniques (10−5 mbar). Benzene was dried over
Na/benzophenone and distilled prior to use. Pentane was passed over
two columns of activated alumina for drying and then degassed.
Propane and propene were passed through BASF Cu catalyst and
activated 4 Å molecular sieves prior to use. H2 and D2 were passed
through activated 4 Å molecular sieves prior to use. Elemental analyses
were performed by the “Mikroanalytisches Labor Pascher” in
Remagen, Germany. Infrared spectra were recorded on a Nicolet
6700 FT-IR spectrophotometer and on a Bruker Alpha transmission
IR spectrometer in transmission mode. EPR was measured at X band
(9.5 GHz EPR frequency) on a Bruker EMX spectrometer at 110 K.
X-ray absorption near edge spectra (XANES) at Cr K edge were
measured at the SuperXAS beamline at the Swiss Light Source (SLS,
Paul Scherrer Institute, Villigen, Switzerland), a third-generation
synchrotron operating at a 2.4 GeV electron energy and a current of
400 mA. The SuperXAS beamline is positioned on one of three
superbent ports. The incident beam was collimated by a Si-coated
mirror at 2.8 mrad, monochromated using a double-crystal Si(111)
monochromator, and focused with a Rh-coated toroidal mirror (at 2.8
mrad) down to 100 × 100 μm with a beam intensity of (4−5) × 1011
ph/s. Calibration of the beamline energy was performed using Cr
reference foil (Cr K edge position at 5989.0 eV). XANES were
measured in fluorescence mode using a silicon drift Ketek detector.
The samples were packed in quartz capillaries (0.01 mm wall
thickness, 1.5 mm outer diameter, Hilgenberg GmbH) under an inert
atmosphere, sealed with Apiezon vacuum grease and wax, and stored
in glass tubes under argon that were opened just before the
measurements. Two scans were measured and merged for
heterogeneous materials.
DFT calculations for propane dehydrogenation (and
propene hydrogenation) on Cr(III) sites on a periodic alumina
model show that a mechanism involving X−H bond activation
(X = R, H) is energetically feasible for both propane
dehydrogenation at 550 °C and for propene hydrogenation at
150 °C. For propene hydrogenation, the most active site for
Cr/Al2O3 is the CrIVb-O2b surface site. On this site the H2
activation, propene insertion, and propane formation have
similar relative Gibbs energy barrier heights, which take values
equal to 86, 83, and 93 kJ mol−1 respectively. The activation
enthalpies (ΔH⧧) and entropies (ΔS⧧) of the initial H2
activation are equal to 31 kJ mol−1 and −130 J K−1 mol−1
with respect to initial reactants, which are in excellent
agreement with experiment (ΔH⧧ = 40(10) kJ mol−1 and
ΔS⧧ = −100(20) J K−1 mol−1 for catalyst 3, Cr(III)/Al2O3).
The data suggest that H2 activation is likely the rate-limiting
step (see the Supporting Information).
Cr(III)/SiO2 (1) was prepared as described previously.10b
H
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
Alumina-700. ALU C (Evonik) was compacted with H2O and
heated to 250 °C for 6 days. The compacted alumina was pestled and
sieved to obtain particles with a diameter of 0.25−0.45 mm. Then the
alumina was calcined under a flow of synthetic air at 500 °C (ramp 300
°C/h, 500 °C for 18 h) and evacuated under high vacuum (10−5 mbar)
for 1 h. A 1 g portion of the calcined alumina was then placed into a
quartz reactor, evacuated (10−5 mbar), and treated at 700 °C for 18 h.
The obtained Al2O3‑700 was stored in a Ar glovebox, and an IR
spectrum was collected in the glovebox.
PHIP Experiments. For the hydrogenation of propene over
catalysts 1−4 (150−250 mg) in ALTADENA experiments a mixture of
propene and p-H2 with 1/4 ratio was used (no balance gas; total
pressure 1 atm). Hydrogen gas was enriched with p-H2 up to 50% at
liquid nitrogen temperature (77 K). The sealed tubes with the catalysts
were opened in a glovebox under an argon atmosphere. The catalysts
were placed in the middle of the 1/4 in. o.d. stainless steel reactor
between two pieces of fiberglass tissue. The reactor was closed with
two Swagelok valves (all of these operations were carried out in the
glovebox). The reactor temperature was varied from 150 to 550 °C.
The reaction mixture was supplied to the reactor and then to the NMR
tube placed inside the NMR spectrometer through a Teflon capillary.
The gas flow rate was varied from 1.3 to 8.0 mL/s. For propene
hydrogenation in a high magnetic field (PASADENA experiment), 30
mg of Cr catalyst was placed at the bottom of a 10 mm screw-cap
NMR tube under Ar. Then the tube was positioned inside the NMR
magnet. The gas mixture was supplied to the catalyst through a Teflon
capillary extended to the bottom of the NMR tube. The NMR tube
was heated to 130 °C. The gas flow rate was varied from 1.4 to 5.7
mL/s.
Computational Details. For the description of Al2O3, we used a
model with nonspinel sites occupied, on the basis of the simulated
dehydration of boehmite.20b,25 Upon substitution of AlIII centers by
Cr, we used spin-polarized density functional theory (DFT)
calculations under periodic boundary conditions using the Perdew−
Wang (PW91) implementation26 of the generalized gradient
approximation (GGA) for the correlation and exchange energy
functional, using the VASP code (version 5.2).27 The projected
augmented wave (PAW)28 method was adopted for describing
electron−ion interactions. The climbing nudge elastic band method
(CI-NEB)29 was used to determine transition states. The (110) surface
of γ- or δ-Al2O3 was based on a previous established model,20b which
was described by a 8.07 × 8.40 Å unit cell and eight layers (unit
formula Al32O48). The inner-slab distance is ca. 24 Å. The Brillouin
zone integration was performed with a 3 × 3 × 1 k-point grid
generated by the Monkhorst−Pack algorithm. In order to reproduce
the properties of extended surfaces, the bottom two layers were kept
fixed during the calculation at bulk coordinates, while the top layers
were allowed to relax. For the s1 surface, we considered the previous
unit cell with one OH bridging two Al centers and one additional
proton on the O2a center (corresponding to an OH coverage equal to
3.0 OH/nm2). This s1 surface was referred to as s1c in a previous
publication from our group.20e
Preparation of 2. Al2O3‑700 (506 mg, 0.09 mmol of AlOH) was
treated with a solution of Cr(OSi(OtBu)3)3(THF)2 (109 mg, 0.11
mmol, 1.2 equiv per AlOH) in dry benzene. The mixture was stirred
for 2.5 h in the glovebox and then filtered and washed with benzene (2
× 5 mL) and pentane (5 mL). The resulting material was dried under
high vacuum (10−5 mbar), giving a pale blue material. FeCp2 (27.9 mg,
0.15 mmol) was added to the washings as an internal standard, and an
1
aliquot was taken for H NMR analysis. The obtained material (392
mg) was then placed in a reactor, which then was evacuated under
high vacuum (10−5 mbar). The material was heated to 300 °C for 1 h
(ramp 5 °C/min) and then to 400 °C for 3 h (ramp 5 °C/min), giving
the pale blue material 2. The trapped volatiles were analyzed by GC
1
and H NMR.
Preparation of 3. Al2O3‑700 (540 mg, 0.10 mmol of AlOH) was
treated with a solution of Cr(Al(OiPr)4)3 (100 mg, 0.12 mmol, 1.2
equiv per AlOH) in dry benzene. The mixture was stirred for 5 h in
the glovebox and then filtered and washed with benzene (4 × 5 mL)
and pentane (5 mL). The resulting material was dried under high
vacuum (10−5 mbar), giving a pale blue material. FeCp2 (31.3 mg, 0.17
mmol) was added to the washings as an internal standard, and an
1
aliquot was taken for H NMR analysis. The obtained material (352
mg) was placed in a reactor, which then was evacuated under high
vacuum (10−5 mbar). The material was heated to 300 °C for 1 h
(ramp 5 °C/min) and then to 400 °C for 3 h (ramp 5 °C/min), giving
the pale blue material 3. The trapped volatiles were analyzed by GC
1
and H NMR.
Preparation of an Industrial-Like 1.5 wt % CrOx/Al2O3
Catalyst, 4. To compacted and sieved Alu C (1.0092 g) was added
a 0.4 M solution of CrO3 in distilled H2O (0.80 mL). The mixture was
stirred under vacuum (Schlenk line) first at room temperature for 10
min, then at 60 °C for 9 h, and then at 120 °C for 4 h. The orange
material was transferred to a flow reactor and heated to 550 °C for 5 h
(300 °C/h) under a synthetic air flow. Then, the flow reactor was
evacuated under high vacuum (10−5 mbar) giving a brown alumina, 4.
Continuous Flow Propane Dehydrogenation Experiments.
Typically 70 mg of Cr catalyst was mixed with SiC (2.5 g) and placed
in a continuous flow reactor. Argon was flowed through the reactor
(10 mL/min) while the reactor was heated to 550 °C. At this
temperature, a mixture of propane and Ar (10 mL/min, 1/4 propane/
Ar, 2.5 bar, gas hourly space velocity 353 h−1) was passed through the
reactor and the output gases were monitored by GC with a flame
ionization detector (FID) and a thermal conductivity detector (TCD)
for 24 h.
Continuous Flow Propene Hydrogenation Experiments.
Typically 200−450 mg of Cr catalyst was mixed with SiC (2.5 g)
and placed in a flow reactor. Argon was flowed through the reactor (30
mL/min, 2 bar) while the reactor was heated to 150 or 50 °C. When
this temperature was reached, the reactor was closed and bypassed
with the initial gas mixture of Ar (23 mL/min), H2 (5 mL/min), and
propene (2 mL/min). A background GC was measured. Then the
reactor was opened and the gas mixture was flowed through the
reactor. The gas mixture was systematically varied while the total gas
flow was held constant at 30 mL/min: first the propene flow was held
constant and the H2 flow was systematically varied, then the propene
flow was held constant, while varying a D2 flow, and then the propene
flow was varied while holding the H2 flow constant. Then the flow
mixture was set to the initial values of Ar (23 mL/min), H2 (5 mL/
min), and propene (2 mL/min) and the temperature was systemati-
cally increased from 150 °C to a maximum of 250 °C with 25 °C
intervals. At each new reaction mixture or temperature, several GCs
were measured of the output mixture and analyzed.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.organo-
met.6b00744.
Infrared spectra, XANES spectra, EPR spectra, plots of
conversion or activity or selectivity vs time of propane
dehydrogenation for catalysts 2−4, kinetic data and
Eyring plots for propene hydrogenation for catalysts 1−
4, PHIP results for catalysts 1−4, NMR spectra of PHIP
experiments for catalysts 1−4, and DFT computational
data (PDF)
Crystallographic data (CIF)
AUTHOR INFORMATION
Corresponding Authors
*A.C.-V.: tel, +41446334152; e-mail, comas@inorg.chem.ethz.
■
ch.
*C.C.: tel, +41446339394; e-mail, ccoperet@inorg.chem.ethz.
ch.
ORCID
Murielle F. Delley: 0000-0001-6450-5540
I
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(9) (a) Koptyug, I. V.; Kovtunov, K. V.; Burt, S. R.; Anwar, M. S.;
Hilty, C.; Han, S.-I.; Pines, A.; Sagdeev, R. Z. J. Am. Chem. Soc. 2007,
129, 5580−5586. (b) Kovtunov, K. V.; Beck, I. E.; Bukhtiyarov, V. I.;
Koptyug, I. V. Angew. Chem., Int. Ed. 2008, 47, 1492−1495.
(c) Kovtunov, K. V.; Zhivonitko, V. V.; Skovpin, I. V.; Barskiy, D.
A.; Koptyug, I. V. Parahydrogen-Induced Polarization in Heteroge-
neous Catalytic Processes. In Hyperpolarization Methods in NMR
Spectroscopy;Kuhn, T. L., Ed. Springer: Berlin, Heidelberg, 2013; pp
123−180. (d) Kovtunov, K. V.; Truong, M. L.; Barskiy, D. A.;
Salnikov, O. G.; Bukhtiyarov, V. I.; Coffey, A. M.; Waddell, K. W.;
Koptyug, I. V.; Chekmenev, E. Y. J. Phys. Chem. C 2014, 118, 28234−
28243. (e) Barskiy, D. A.; Salnikov, O. G.; Kovtunov, K. V.; Koptyug,
I. V. J. Phys. Chem. A 2015, 119, 996−1006. (f) Balu, A. M.; Duckett, S.
B.; Luque, R. Dalton Trans. 2009, 26, 5074−5076.
Francisco Nunez-Zarur: 0000-0002-9244-9328
Igor V. Koptyug: 0000-0003-3480-7649
Notes
̃
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors thank Dr. Kim Larmier (ETH Zurich, Switzerland)
and Dr. Olga V. Safonova (Paul Scherrer Institute, Switzerland)
for assistance with X-ray absorption spectroscopy and Florian
̈
Allouche (ETH Zurich, Switzerland) for help with X-ray
̈
diffraction. M.F.D. thanks ETH Zurich and the Scholarship
̈
Fund of the Swiss Chemical Industry for financial support.
K.V.K., O.G.S., and I.V.K. thank the Russian Science
Foundation (grant #14-13-00445) for supporting the experi-
ments on catalytic hydrogenation with parahydrogen. The
Swiss National Foundation (SNF) (Ambizione project
PZ00P2_148059 to A.C.-V.) and SCCER Heat and Energy
Storage are also acknowledged for financial support.
(10) (a) Conley, M. P.; Delley, M. F.; Siddiqi, G.; Lapadula, G.;
Norsic, S.; Monteil, V.; Safonova, O. V.; Coperet, C. Angew. Chem., Int.
Ed. 2014, 53, 1872−1876. (b) Delley, M. F.; Nunez-Zarur, F.; Conley,
M. P.; Comas-Vives, A.; Siddiqi, G.; Norsic, S.; Monteil, V.; Safonova,
O. V.; Coperet, C. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 11624−
11629. (c) Conley, M. P.; Delley, M. F.; Nunez-Zarur, F.; Comas-
́
̃
Vives, A.; Coper
M. F. Chimia 2015, 69, 168−171.
(11) Heroguel, F.; Siddiqi, G.; Detwiler, M. D.; Zemlyanov, D. Y.;
Safonova, O. V.; Coperet, C. J. Catal. 2015, 321, 81−89.
(12) (a) Parry, E. P. J. Catal. 1963, 2, 371−379. (b) Valla, M.;
Stadler, D.; Mougel, V.; Coperet, C. Angew. Chem., Int. Ed. 2016, 55,
1124−1127.
́
et, C. Inorg. Chem. 2015, 54, 5065−5078. (d) Delley,
REFERENCES
■
́
(1) Sattler, J. J. H. B.; Ruiz-Martinez, J.; Santillan-Jimenez, E.;
Weckhuysen, B. M. Chem. Rev. 2014, 114, 10613−10653.
́
(2) Frey, F. E.; Huppke, W. F. Ind. Eng. Chem. 1933, 25, 54−59.
(3) (a) Coperet, C. Chem. Rev. 2010, 110, 656−680. (b) Weckhuysen,
B. M. Phys. Chem. Chem. Phys. 2003, 5, 4351−4360. (c) Jackson, S. D.;
Stair, P. C.; Gladden, L. F.; McGregor, J. Alkane Dehydrogenation
over Vanadium and Chromium Oxides. In Metal Oxide Catalysis;
Wiley-VCH: Weinheim, Germany, 2008; pp 595−612.
(4) (a) Weckhuysen, B. M.; Schoonheydt, R. A. Catal. Today 1999,
51, 223−232. (b) Weckhuysen, B. M.; Wachs, I. E.; Schoonheydt, R.
A. Chem. Rev. 1996, 96, 3327−3350.
(5) (a) Olsbye, U.; Virnovskaia, A.; Prytz, Ø.; Tinnemans, S.;
Weckhuysen, B. Catal. Lett. 2005, 103, 143−148. (b) Weckhuysen, B.
M.; Bensalem, A.; Schoonheydt, R. A. J. Chem. Soc., Faraday Trans.
1998, 94, 2011−2014. (c) Derossi, S.; Ferraris, G.; Fremiotti, S.;
Garrone, E.; Ghiotti, G.; Campa, M. C.; Indovina, V. J. Catal. 1994,
148, 36−46.
(6) (a) Hu, B.; Schweitzer, N. M.; Zhang, G.; Kraft, S. J.; Childers, D.
J.; Lanci, M. P.; Miller, J. T.; Hock, A. S. ACS Catal. 2015, 5, 3494−
3503. (b) Das, U.; Zhang, G.; Hu, B.; Hock, A. S.; Redfern, P. C.;
Miller, J. T.; Curtiss, L. A. ACS Catal. 2015, 5, 7177−7185. (c) Hu, B.;
Getsoian, A.; Schweitzer, N. M.; Das, U.; Kim, H.; Niklas, J.;
Poluektov, O.; Curtiss, L. A.; Stair, P. C.; Miller, J. T.; Hock, A. S. J.
Catal. 2015, 322, 24−37. (d) Estes, D. P.; Siddiqi, G.; Allouche, F.;
Kovtunov, K. V.; Safonova, O. E.; Trigub, A. L.; Koptyug, I. V.;
́
(13) (a) Pantelouris, A.; Modrow, H.; Pantelouris, M.; Hormes, J.;
Reinen, D. Chem. Phys. 2004, 300, 13−22. (b) Berry, A. J.; O’Neill, H.
S. C. Am. Mineral. 2004, 89, 790−798.
(14) Singh, J. V.; Jain, N. C.; Mehrotra, R. C. Synth. React. Inorg. Met.-
Org. Chem. 1979, 9, 79−88.
(15) (a) Rascon, F.; Wischert, R.; Coperet, C. Chem. Sci. 2011, 2,
1449−1456. (b) Coper
́
et, C.; Comas-Vives, A.; Conley, M. P.; Estes,
D. P.; Fedorov, A.; Mougel, V.; Nagae, H.; Nunez-Zarur, F.; Zhizhko,
́
̃
P. A. Chem. Rev. 2016, 116, 323−421.
(16) Kogelbauer, A.; Lercher, J. A. J. Catal. 1990, 125, 197−206.
(17) (a) Janak, K. E.; Parkin, G. J. Am. Chem. Soc. 2003, 125, 6889−
6891. (b) Janak, K. E.; Parkin, G. J. Am. Chem. Soc. 2003, 125, 13219−
13224. (c) Parkin, G. Acc. Chem. Res. 2009, 42, 315−325.
(18) (a) Bowers, C. R.; Weitekamp, D. P. Phys. Rev. Lett. 1986, 57,
2645−2648. (b) Bowers, C. R.; Weitekamp, D. P. J. Am. Chem. Soc.
1987, 109, 5541−5542.
(19) Pravica, M. G.; Weitekamp, D. P. Chem. Phys. Lett. 1988, 145,
255−258.
(20) (a) Digne, M.; Sautet, P.; Raybaud, P.; Euzen, P.; Toulhoat, H. J.
Catal. 2002, 211, 1−5. (b) Digne, M.; Sautet, P.; Raybaud, P.; Euzen,
́
P.; Toulhoat, H. J. Catal. 2004, 226, 54−68. (c) Wischert, R.; Coperet,
́
Coperet, C. J. Am. Chem. Soc. 2016, 138, 14987−14997.
C.; Delbecq, F.; Sautet, P. Angew. Chem., Int. Ed. 2011, 50, 3202−3205.
(d) Wischert, R.; Coperet, C.; Delbecq, F.; Sautet, P. Chem. Commun.
2011, 47, 4890−4892. (e) Wischert, R.; Laurent, P.; Coperet, C.;
Delbecq, F.; Sautet, P. J. Am. Chem. Soc. 2012, 134, 14430−14449.
(7) (a) Burwell, R. L.; Littlewood, A. B.; Cardew, M.; Pass, G.;
Stoddart, C. T. H. J. Am. Chem. Soc. 1960, 82, 6272−6280.
(b) Littlewood, A. B.; Burwell, R. L. J. Am. Chem. Soc. 1960, 82,
6287−6288. (c) Cardew, M.; Burwell, R. L. J. Am. Chem. Soc. 1960, 82,
6289−6291. (d) Iwasawa, Y.; Sasaki, Y.; Ogasawara, S. J. Chem. Soc.,
Chem. Commun. 1981, 3, 140−142. (e) Iwasawa, Y. J. Mol. Catal. 1982,
17, 93−104. (f) Khodakov, Y. S.; Makarov, P. A.; Delzer, G.;
Minachev, K. M. J. Catal. 1980, 61, 184−191. (g) Wittgen, P. P. M.
M.; Groeneveld, C.; Janssens, J. H. G. J.; Wetzels, M. L. J. A.; Schuit,
G. C. A. J. Catal. 1979, 59, 168−175. (h) Wittgen, P. P. M. M.;
Groeneveld, C.; Zwaans, P. J. C. J. M.; Morgenstern, H. J. B.; van
Heughten, A. H.; van Heumen, C. J. M.; Schuit, G. C. A. J. Catal.
1982, 77, 360−381.
(21) (a) Comas-Vives, A.; Schwarzwalder, M.; Coper
́
et, C.; Sautet, P.
J. Phys. Chem. C 2015, 119, 7156−7163. (b) Comas-Vives, A.; Valla,
M.; Coper
̈
́
et, C.; Sautet, P. ACS Cent. Sci. 2015, 1, 313−319.
(22) (a) Balcells, D.; Clot, E.; Eisenstein, O. Chem. Rev. 2010, 110,
749−823. (b) Waterman, R. Organometallics 2013, 32, 7249−7263.
(23) Joubert, J.; Salameh, A.; Krakoviack, V.; Delbecq, F.; Sautet, P.;
́
Coperet, C.; Basset, J. M. J. Phys. Chem. B 2006, 110, 23944−23950.
(24) (a) Grinev, V. E.; Khalif, V. A.; Aptekar, E. L.; Krylov, C. V. Bull.
Acad. Sci. USSR, Div. Chem. Sci. 1981, 30, 1338−1341. (b) Salameh,
A.; Baudouin, A.; Soulivong, D.; Boehm, V.; Roeper, M.; Basset, J.-M.;
(8) (a) Eisenberg, R.; Eisenschmid, T. C.; Chinn, M. S.; Kirss, R. U.
Parahydrogen-Induced Polarization and Polarization Transfer in
Hydrogenation and Oxidative Addition Reactions. In Homogeneous
Transition Metal Catalyzed Reactions; American Chemical Society:
Washington, DC, 1992; Vol. 230, pp 47−74. (b) Duckett, S. B.; Wood,
N. J. Coord. Chem. Rev. 2008, 252, 2278−2291. (c) Duckett, S. B.;
Sleigh, C. J. Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 71−92.
́
Coperet, C. J. Catal. 2008, 253, 180−190.
(25) Krokidis, X.; Raybaud, P.; Gobichon, A.-E.; Rebours, B.; Euzen,
P.; Toulhoat, H. J. Phys. Chem. B 2001, 105, 5121−5130.
(26) Perdew, J. P.; Chevary, J. A.; Vosko, S. H.; Jackson, K. A.;
Pederson, M. R.; Singh, D. J.; Fiolhais, C. Phys. Rev. B: Condens. Matter
Mater. Phys. 1992, 46, 6671−6687.
J
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX

Organometallics
Article
(27) (a) Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 15−
̈
50. (b) Kresse, G.; Furthmuller, J. Phys. Rev. B: Condens. Matter Mater.
̈
Phys. 1996, 54, 11169−11186.
(28) (a) Blochl, P. E. Phys. Rev. B: Condens. Matter Mater. Phys. 1994,
̈
50, 17953−17979. (b) Kresse, G.; Joubert, D. Phys. Rev. B: Condens.
Matter Mater. Phys. 1999, 59, 1758−1775.
(29) (a) Henkelman, G.; Uberuaga, B. P.; Jon
2000, 113, 9901−9904. (b) Henkelman, G.; Jon
2000, 113, 9978−9985.
́
sson, H. J. Chem. Phys.
sson, H. J. Chem. Phys.
́
K
DOI: 10.1021/acs.organomet.6b00744
Organometallics XXXX, XXX, XXX−XXX