Crystal phase effect of iron oxides on the aerobic oxidative coupling of alcohols and amines under mild conditions: A combined experimental and theoretical study

Article abstract of DOI:10.1016/j.jcat.2019.06.018

Selective catalytic oxidation using air as the terminal oxidant is an ecofriendly route for the synthesis of fine and commodity chemicals. However, the catalyst generally faces the challenges of the inertness of molecular oxygen, limited substrate scope, poor selectivity, and high cost. Moreover, the toxicity of the catalyst should also be considered when the products are used in pharmaceutical or biotechnological areas. Here, upon investigating the dependence of catalytic oxidation on the crystal phases of iron oxides, we find that naked γ-Fe₂O₃ particles exhibit excellent catalytic activity, selectivity, and stability in a series of imine synthetic reactions. The performance of γ-Fe₂O₃ particles is significantly better than that of α-Fe₂O₃ and Fe₃O₄ under mild reaction conditions, and the γ-Fe₂O₃ catalyst can be separated from the reaction mixture magnetically. Both experimental and theoretical calculation results show that γ-Fe₂O₃ possesses supercapability for oxygen activation. The inverse spinel structure of γ-Fe₂O₃ has abundant cation vacancies, which confers unique electronic properties on surface Fe species. These Fe species tend to transfer electrons to molecular oxygen to form O₂^? or O₂^2? species. These oxygen species are favorable for the dehydrogenation of alcohols, which is responsible for the high activity of γ-Fe₂O₃ in this coupling reaction.

Full text of DOI:10.1016/j.jcat.2019.06.018

Journal of Catalysis 377 (2019) 145–152
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Crystal phase effect of iron oxides on the aerobic oxidative coupling of
alcohols and amines under mild conditions: A combined experimental
and theoretical study
a,
Longlong Geng a,c,1, Wei Jian b,1, Pei Jing a, Wenxiang Zhang a, Wenfu Yan a, Fu-Quan Bai b, , Gang Liu
⇑
⇑
a State Key Laboratory of Inorganic Synthesis and Preparative Chemistry, College of Chemistry, Jilin University, Changchun 130012, China
b Institute of Theoretical Chemistry, Jilin University, Changchun 130012, China
c College of Chemistry and Chemical Engineering, Dezhou University, Dezhou 253023, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Selective catalytic oxidation using air as the terminal oxidant is an ecofriendly route for the synthesis of
fine and commodity chemicals. However, the catalyst generally faces the challenges of the inertness of
molecular oxygen, limited substrate scope, poor selectivity, and high cost. Moreover, the toxicity of the
catalyst should also be considered when the products are used in pharmaceutical or biotechnological
areas. Here, upon investigating the dependence of catalytic oxidation on the crystal phases of iron oxides,
Received 18 November 2018
Revised 31 May 2019
Accepted 11 June 2019
we find that naked
ies of imine synthetic reactions. The performance of
-Fe2O3 and Fe3O4 under mild reaction conditions, and the
reaction mixture magnetically. Both experimental and theoretical calculation results show that
Fe2O3 possesses supercapability for oxygen activation. The inverse spinel structure of -Fe2O3 has abun-
c
-Fe2O3 particles exhibit excellent catalytic activity, selectivity, and stability in a ser-
-Fe2O3 particles is significantly better than that of
-Fe2O3 catalyst can be separated from the
Keywords:
Iron oxides
Imine synthesis
Oxidative coupling reaction
Crystal phase effect
Molecular oxygen activation
c
a
c
c
-
c
dant cation vacancies, which confers unique electronic properties on surface Fe species. These Fe species
tend to transfer electrons to molecular oxygen to form Oꢀ2 or O22ꢀ species. These oxygen species are favor-
able for the dehydrogenation of alcohols, which is responsible for the high activity of
pling reaction.
c-Fe2O3 in this cou-
Ó 2019 Elsevier Inc. All rights reserved.
1. Introduction
tural, and pharmaceutical synthesis [13–20]. Oxidative coupling of
alcohols and amines under mild reaction conditions is currently
being intensively researched for imine synthesis [21–26], with
air and water as the oxygen source and sole by-product. This reac-
tion proceeds in two consecutive steps: oxidative dehydrogenation
of an alcohol to an aldehyde and coupling of the aldehyde to an
imine with an amine [25,26]. One of the most important require-
ments for the catalyst is to avoid polymerization and hydrogena-
tion of the reactive C@N moiety [27–31]. Some catalysts with
strong dehydrogenation ability, such as supported noble metal cat-
alysts [32–35], also have strong hydrogenation ability, resulting in
lower selectivity of the target products. Transition metal oxide cat-
alysts have some advantages in improving the selectivity of imine
synthesis [36–40]. However, because of toxicity concerns and sep-
aration problems, many toxic heavy-metal-mediated catalytic pro-
cesses are not suitable for pharmaceutical or biotechnological
applications [39]. Although iron-based oxides are favorable cata-
Iron is an attractive metal element for catalysis not only
because of its abundant reserves in the earth’s crust, but also
because of its low toxicity to organisms and its environmentally
benign effects [1–6]. As one of the essential elements of the human
body, iron is mainly found in hemoglobin in the form of ions and
plays the role of transporting oxygen [7,8]. However, the activation
and transport capability of oxygen for most iron oxides is not ideal,
because it usually requires relatively high temperature and pres-
sure [9–12]. This limitation restricts the catalytic application of
iron oxides in aerobic oxidation under mild conditions (<100 °C,
1 atm), especially the synthesis of some thermally unstable phar-
maceutical intermediates.
Imines are nitrogen compounds containing the reactive C@N
moiety, which are intermediates widely used in biological, agricul-
lysts, current research shows that iron oxides (mainly
exhibit very low activity in this reaction, mainly due to poor O2
activation ability under mild reaction conditions [20].
a-Fe2O3)
⇑
Corresponding authors.
E-mail addresses: baifq@jlu.edu.cn (F.-Q. Bai), lgang@jlu.edu.cn (G. Liu).
Longlong Geng and Wei Jian contributed equally to this work.
1
https://doi.org/10.1016/j.jcat.2019.06.018
0021-9517/Ó 2019 Elsevier Inc. All rights reserved.

146
L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
Chemisorption and dissociation are the fundamental steps in
flask. The flask was connected to an air balloon and the mixture
was vigorously stirred at 80 °C for 8 h. Aliquots of the products
were taken with a sampling pipe and analyzed with a gas chro-
matograph equipped with a HP-5 column and FID detector. The
product mixtures were further identified by GC-MS.
A hot filtration test was carried out as follows: After 2 h of reac-
tion, the solid catalysts were separated with a Buchner funnel.
Then the mixture of the filtrate was put into the reactor and con-
tinuously reacted under the same conditions (80 °C, air 1 atm)
without any solid catalyst. In the recycling experiment, the used
catalyst was separated from the reactant by an external magnet
and treated at 350 °C for 1 h in air before the next test cycle.
molecular O2 activation, which are essentially associated with
charge transfer from the active sites of the catalyst to the antibond-
ing molecular orbitals of O2. The crystal phase determines the geo-
metric and electronic structure of a catalyst [41–45], which
significantly affects the bulk and surface charge transport, and ulti-
mately reflects the activity and density of the active sites. In this
work, by investigation of the dependence of catalytic oxidation
on the crystal phases of iron oxides, we find that naked c-Fe2O3
particles (with a magnetically separable character) exhibit excel-
lent catalytic activity, selectivity, and stability for a series of imine
synthesis reactions under mild conditions (even at 40 °C, 1 atm).
Through a combination of experimental characterizations and the-
oretical calculations based on density functional theory (DFT), the
crystal structure effect of iron oxides on O2 activation and catalytic
oxidative coupling behavior is clarified. It is shown that the inverse
2.4. Density functional theory calculation details
All first-principles calculations in this study were performed
within the frame of DFT using the plane-wave pseudopotential
approach and implemented by the Vienna Ab Initio Simulation
Package (VASP). A spin-polarized exchange–correlation functional
was described within the PBE generalized gradient approximation
[47,48], and electron–core interactions were used with projected
augmented wave (PAW) pseudopotentials. The reciprocal space
was spanned with a plane-wave basis cutoff of 400 eV. The Bril-
louin zone was automatically generated by the original Monk
horst–Pack grid, and the k-points for structure optimization were
selected as 3 ꢂ 3 ꢂ 1. In all cases of structural optimization, the
energy convergence was set to 1 ꢂ 10ꢀ4 eV, and the force on the
atoms was less than 0.05 eV/Å.
spinel structure of c-Fe2O3 (with abundant cation vacancies) con-
fers unique electronic properties on surface Fe species. These Fe
species tend to transfer electrons to molecular oxygen to form
Oꢀ2 or O22ꢀ species, which is favorable for high activity of
in the aerobic coupling reaction.
c-Fe2O3
2. Experimental
2.1. Preparation of iron oxide with different crystal phases
Fe3O4 was synthesized via a modified ammonia-assisted precip-
itation method [46]. Typically, Fe chloride precursors (Fe2+/Fe3+
molar ratio = 1:2) were dissolved in an ethanol/water (1:1) solu-
tion under nitrogen with constant mechanical stirring for 30 min,
and a 10% ammonia solution was used to adjust the pH of the solu-
tion to 9.0. The black suspension was then stirred for 1 h at room
temperature and another 1 h at 60 °C in an oil bath. After the sys-
tem was cooled to room temperature, the resulting solids were
separated using a magnet, washed with alcohol, and dried at
Fe3d74s1 and O2s22p4 valence electron configuration were
adopted in this study. To more accurately consider the effect of
the strong correlation of Fe on the 3d electrons, a DFT+U strategy
was used to calculate the electron density. This method also helped
to correct the O2 binding errors [49,50]. In addition, according to
the widely approved values in the literature [51-53], the effective
U-values of Fe and O atoms were set to 4.0 and 0.0 eV, respectively.
Integration over the irreducible part of the Brillouin zone was car-
ried out using the linear tetrahedron method with Bloch correc-
tions, which depict electronic structure more precisely.
Considering the Bader analysis, charges in adsorption are described
as numbers of electrons gained or lost in comparison with the neu-
tral atom [54,55].
80 °C in a vacuum oven.
ment of the synthesized Fe3O4 in air at 350 °C for 2 h.
c
-Fe2O3 was obtained via thermal treat-
-Fe2O3 par-
a
ticles were obtained using a procedure similar to that for Fe3O4
nanoparticles. The difference is that iron chloride hexahydrate
(FeCl3ꢁ6H2O) was used as an Fe precursor and the resulting solids
were finally thermally treated at 550 °C for 2 h in a muffle furnace.
The lattice parameters of a-Fe2O3, Fe3O4, and c-Fe2O3 used in all
calculations are summarized in Table S1 in the Supporting
Information. In view of surface and molecular oxygen activation
2.2. Characterization of catalysts
properties, the {0 0 1}, {1 1 1}, and {1 1 1} facets of
a-Fe2O3, Fe3O4,
Scanning electron microscopy (SEM) images were taken on a
Hitachi-SU8020 with an accelerating voltage of 30 kV. The crystal
and -Fe2O3 were adopted, respectively [56–58]. To eliminate the
c
influence of the thickness of the slab atomic layers, all models were
composed of eight layers of Fe and four layers of O and separated
from periodic images by at least 15 Å of vacuum. Concerning the
significance of the adsorption interface, the top five layers were
relaxed and the following seven layers of atoms were fixed in the
optimization process. The adsorption energy can be expressed as
structures were recorded using
diffractometer (XRD) equipped with a CuK
a
Rigaku powder X-ray
radiation source
a
( k = 1.542 nm). Raman spectra were collected on a Bruker RFS 100
Raman spectrometer with an argon laser (532 nm) as an excitation
source. Fourier transform infrared spectra (FTIR) were recorded on a
Thermo Nicolet 6700 FT-IR spectrometer. 57Fe Mössbauer spectra
were recorded by an OIMS-500 Mössbauer spectrum instrument.
D
Eads ¼ EsurfþO ꢀ Esurf ꢀ EO
2
2
57Co (Pd) was used as a
was calibrated against standard
c
-ray radioactive source and the velocity
where EsurfþO ,Esurf , and EO represent the total energy of the
a-iron foil. N2 adsorption–
2
2
adsorbed oxygen molecules, the energy of the surface, and the
gas-phase energy of oxygen molecules, respectively.
desorption isotherms were measured at 77 K, using a Micromeritics
ASAP 2010 N analyzer. Samples were degassed at 423 K for 15 h
before measurements. Magnetic measurements were performed
on a vibrating sample magnetometer (YingPu VSM-400 Instru-
ments) at room temperature under an applied field of 40,000 Oe.
3. Results and discussion
3.1. Preparation and characterization of iron oxide catalysts
2.3. Catalytic test
Considering potential applications, three readily available and
A mixture of benzyl alcohol (1 mmol), aniline (2 mmol), catalyst
(0.3 g), and toluene (10 mL) was added into a 50-mL two-neck
stable iron oxides, hematite (
a-Fe2O3), magnetite (Fe3O4), and
maghemite -Fe2O3), were synthesized and systematically
(c

L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
147
Scheme 1. Illustration of crystal structures of (A) a-Fe2O3, (B) Fe3O4, and (C) c-Fe2O3. Color scheme: pink, oxygen; purple, iron.
investigated. Both crystal structures and the oxidation state of iron
are taken into account in the selection process. Scheme 1 shows
avoid polymerization and hydrogenation of the reactive C@N moi-
ety. Under the same reaction conditions, -Fe2O3 also exhibits
obvious active advantages over a series of reducible and irreducible
oxide catalysts (Table 1). -Fe2O3 can catalyze this reaction at very
c
the crystal structures of a-Fe2O3, Fe3O4, and c-Fe2O3. a-Fe2O3 has
a corundum-type structure with Fe3+ ions occupying two-thirds
of the octahedral sites in a hexagonal close-packed O lattice.
Fe3O4 features an inverse spinel crystal structure of oxide anions
arranged as a cubic close-packed lattice, containing atoms of Fe
c
low temperatures, yielding 68.5% and 39.1% imines at 60 and 40 °C,
respectively (Table 1, entries 5 and 6). Under a 1/1molar ratio of
benzyl alcohol/aniline, the yield of imine is 61.6% (Table 1, entry
7), indicating that a relatively high concentration of aniline could
be suitable for optimizing the reaction process. Calculated from
the initial reaction rate, the apparent activation energy (Ea) of the
(II) and Fe(III) in octahedral and tetrahedral sites.
c-Fe2O3 pos-
sesses a cubic structure consisting of 32 O2ꢀ ions, 21 1/3 Fe3+ ions,
and 2 2/3 vacancies in each unit cell. Restricted by the cubic
close-packed O lattice arrangement, ferric ions distribute over
tetrahedral and octahedral sites which are remaining Fe ions and
vacancies [59].
reaction over
lower than that of Fe3O4 (77 kJ molꢀ1) and
Fig. S2). All these results show that -Fe2O3 is an efficient catalyst
c
-Fe2O3 is about 59 kJ molꢀ1, which is obviously
-Fe2O3 (102 kJ molꢀ1
a
,
c
The test samples of
a-Fe2O3 and Fe3O4 were prepared by a mod-
for imine synthesis from oxidative coupling of benzyl alcohol and
ified precipitation method in which ethanol and ammonia were
used as solvent and precipitant, respectively, to control the nucle-
ation and hydrolysis rate of iron ions. FeCl3 and a stoichiometric
mixture of FeCl3 and FeCl2 are used as starting reagents for the
aniline.
Another advantage of
character. Fig. 2B shows that
agnetic properties. The saturation magnetization (Ms) is 75.5 emu/
g, which is comparable to that of bulk -Fe2O3 materials (Ms = 76 -
emu/g) [65,66], while the remnant magnetization (Mr) is only
0.5 emu/g. This property facilitates separation of -Fe2O3 from
the reaction mixture with an external magnet and good redisper-
sion after the external magnetic field is removed (Fig. 2B, inset).
Upon magnetic separation, a hot catalyst filtration test is carried
c
-Fe2O3 is its magnetically separable
c-Fe2O3 has very strong superparam-
preparation of
a-Fe2O3 and Fe3O4, respectively. Both of the resul-
c
tant iron oxide particles occur together with smaller particles of
size 22–35 nm (Fig. 1A and 1B, SEM images). A large number of
mesopores (about 15–23 nm) form the packing of these small
particles (Fig. S1 and Table S2, N2 adsorption–desorption results).
c
c-Fe2O3 was obtained by thermal treatment above Fe3O4 precursor
in air. The transformation of Fe3O4 to -Fe2O3 is a solid-state oxida-
c
out to investigate the heterogeneous nature of
shows that no subsequent yield of imines is detected in the filtrate,
proving that -Fe2O3 is a heterogeneous catalyst. The stability of
-Fe2O3 is investigated by examining its recyclability in the reac-
c-Fe2O3. Fig. 2C
tion process [60,61], with no obvious morphology change being
observed (Fig. 1C, SEM images). The crystal structures of three iron
oxides are first confirmed by X-ray powder diffraction (XRD)
c
c
(Fig. 1D). The rhombohedral structure of
structure of Fe3O4 and -Fe2O3 can be identified [62]. Taking into
account that -Fe2O3 has the same XRD patterns as Fe3O4, FTIR,
Raman, and 57Fe Mössbauer measurements are further carried
out to confirm the -Fe2O3 phase (Fig. 1E–1G). Combined with
the above characterization results and corresponding fitting analy-
sis (Table S3), this finally allows us to verify the purity of the
obtained catalyst [63,64].
a
-Fe2O3 and the cubic
tion. As displayed in Fig. 2D, after a test of six cycles, the catalyst
still maintains an activity as high as the fresh one. The SEM image
c
c
(Fig. S3) and XRD pattern (Fig. S4) show that used
maintains the morphology and cubic structure of fresh
No crystal transformation from the -phase to the -phase occurs
in the reaction process. These results show that -Fe2O3 is a stable
heterogeneous catalyst for this oxidative coupling reaction.
Additionally, the general application of -Fe2O3 catalyst is
c-Fe2O3 still
c-Fe2O3.
c
c
a
c
c
investigated. At least ten imines are achieved by the oxidative cou-
3.2. Catalytic performance of iron oxides in imine synthesis
pling of different kinds of alcohols and amines. Table 2 shows that
c
-Fe2O3 catalyst is active in the reactions using benzyl alcohol
derivatives with either electron-rich or electron-poor substituent
groups, and excellent imine yields are obtained. -Fe2O3 is also
The catalytic oxidative coupling behavior of these iron oxides
was first investigated in the imine synthesis reaction with benzyl
alcohol and aniline as reagents and air as oxygen source. The imine
is identified using gas chromatography–mass spectrometry.
c
active in catalyzing oxidative coupling of benzyl alcohol with sub-
stituted anilines and obtains high imine yields. Aliphatic amines,
such as cyclohexane and n-butylamine, can also react with benzyl
alcohol to afford the desired imines in good yields by prolonging
the reaction time. These results show that the c-Fe2O3 catalyst
has high universality for imine synthesis with an oxidative cou-
c
-Fe2O3 exhibits much higher activity for this reaction than
-Fe2O3 and Fe3O4 (Fig. 2A). A 91.3% yield can be obtained over
-Fe2O3 after reaction for 8 h at 80 °C. The selectivity to imine is
higher than 99% and almost no by-products except water are
detected. This result clearly shows that -Fe2O3 could effectively
a
c
c
pling route.

148
L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
Fig. 1. SEM images of (A)
a-Fe2O3, (B) Fe3O4, (C)
c
-Fe2O3; (D) XRD patterns, (E) FT-IR spectra, and (F) Raman spectra of different iron oxide catalysts and (G) 57Fe Mössbauer
spectra of c-Fe2O3 particles.
We also demonstrate that the reaction catalyzed by
c
-Fe2O3 is
the most sensitive to oxygen among the three, followed by Fe3O4
(Fig. S8). Fig. 4 shows the optimized configurations of O2 adsorp-
tion on the active facets of a-Fe2O3, Fe3O4, and c-Fe2O3. The corre-
sponding bond length and adsorption energy (Eabs) are listed in
Table S4. It shows that adsorbed oxygen molecules mainly form
an aerobic process. As displayed in Fig. 3A, only a small amount
of imine is detected in relatively inert environments. Benzaldehyde
is an intermediate product of the reaction, which is produced by
oxidative dehydrogenation of benzyl alcohol (Fig. 3B). The benzyl
alcohol dehydrogenation rate for these three samples follows the
FeAO4 bonds on the surface of
a-Fe2O3 (Fig. 4A). Two different
order
c
-Fe2O3 > Fe3O4 >
a
-Fe2O3 (Fig. S5). The Ea of benzyl alcohol
FeAO bond lengths, 1.81 Å and 2.56 Å, are obtained (Table S4),
which indicates that weak interaction is present between Fe spe-
dehydrogenation is 51 kJꢁmolꢀ1 on
c
-Fe2O3 and 78kJꢁmolꢀ1 on
Fe3O4 (Fig. S6). All these trends are consistent with oxidative cou-
pling of alcohol and amine. The following step in the coupling of
benzaldehyde and aniline is a fast reaction, either on
cies and O5. On the surfaces of Fe3O4 and
cules are more favorable to forming identical FeAO bonds. The
bond length is 1.84 Å for Fe3O4 and 1.81 Å for -Fe2O3, respectively
c-Fe2O3, oxygen mole-
c-Fe2O3 cat-
c
alyst or on -Fe2O3, and Fe3O4 catalysts (Fig. 3C and Fig. S7). There-
a
(Table S4). The Eabs results show that the adsorption of oxygen
fore, it can be concluded that oxidative dehydrogenation of
molecules on three iron oxides is all the exothermic processes,
alcohols is the rate-determining step for imine synthesis.
and Eabs follows the order
-Fe2O3 (ꢀ1.58 eV).
Additionally, the OAO bond lengths of adsorbed oxygen are also
quite different on the surfaces of iron oxide catalysts. They are
1.29, 1.39, and 1.41 Å for -Fe2O3, Fe3O4, and -Fe2O3, respectively.
The OAO distance of adsorbed oxygen on -Fe2O3 is obviously
greater than that of O2ꢀ (1.33 Å) and approximately that of O22ꢀ
(1.49 Å) and H2O2 (1.48 Å) [69]. All these results show that the
Fe2O3 surface is very active for oxygen activation, followed by that
c
-Fe2O3 (ꢀ4.3 eV) > Fe3O4 (ꢀ1.84 eV)
>
a
3.3. O2 activation mechanism over different iron oxides
a
c
c
DFT calculations were carried out to understand the O2 activa-
tion mechanism over different iron oxides. Based on the surface
activity, the facets of {0 0 1}, {1 1 1}, and {1 1 1} are considered
c-
as models for
a
-Fe2O3, Fe3O4, and
c-Fe2O3, respectively [56–
58,67,68]. Calculated surface energy results show that
c-Fe2O3 is

L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
149
(A) 100
80
γ-Fe2O3
(B)
Mr=0.5 emu/g
80
Ms=75.5 emu/g
40
Hc=6.7 Oe
60
40
20
0
Fe3O4
0
-40
α-Fe2O3
-80
0
2
4
6
8
-40000
-20000
0
20000
40000
Reaction time (h)
Magnetic Field (Oe)
100
100
(C)
(D)
80
60
40
20
0
Filterition
After
80
60
40
20
0
Before
1
2
3
4
5
6
0
1
2
3
4
5
6
Recycle times
Reaction time (h)
Fig. 2. (A) Time dependence of imine formation from oxidative coupling of benzyl alcohol and aniline over iron oxide catalysts. (B) Magnetic hysteresis loop of the
synthesized -Fe2O3 catalyst. (C) Leaching of -Fe2O3 by continuing the reaction after filtration of the catalyst (the red line indicates the imine yield after the catalysts are
-Fe2O3 in the imine synthesis process. Reaction conditions: benzyl alcohol 1 mmol, aniline 2 mmol, catalyst 0.3 g, toluene 10 mL, 80 °C, air 1 bar.
c
c
removed). (d) Reusability of
c
Table 1
Imine synthesis from benzyl alcohol and aniline over various metal oxides.a
Entry
Catalyst
–
Yield [%]b
Entry
Catalyst
Yield [%]b
1
2
3
4
5
6
7
8
<1c
13
14
15
16
17
18
19
20
21
22
23
MgO
ZrO2
Cr2O3
CuO
TiO2
Bio
Cano
MoO3
WO3
Sb2O3
ZnO
1.3
1.0
<1
<1
<1
<1
<1
<1
<1
<1
<1
a-Fe2O3
13.5
52.5
91.3
68.5d
39.1e
61.6f
29.9
23.6
17.5
3.1
Fe3O4
c
c
c
c
-Fe2O3
-Fe2O3
-Fe2O3
-Fe2O3
NiO
9
Co3O4
CeO2
Al2O3
La2O3
10
11
12
1.4
a
b
c
d
e
f
Reaction condition: benzyl alcohol 1 mmol, aniline 2 mmol, metal oxide 0.3 g, toluene 10 mL, 80 °C, air 1 bar.
Isolated imine yield.
Reaction time 12 h.
Reaction temperature 60 °C.
Reaction temperature 40 °C.
Benzyl alcohol 1 mmol, aniline 1 mmol.
of Fe3O4. Moreover, adsorbed oxygen on the
sesses much greater oxidative capability.
Fig. 5 shows the partial density of states (PDOS) of different iron
oxides before and after adsorption of oxygen. The valence bands of
iron oxides are composed of Fe3d and O2p orbits, and the conduc-
tion bands consist of Fe3d orbits (Fig. 5A–C). The calculated band
c
-Fe2O3 surface pos-
the bottom of the conduction band. After adsorption of oxygen,
the band gap becomes narrower and the Fermi level moves to
lower energy (Fig. 5D–F). The Fe3d orbit exhibits a relatively low
strength and overlap with the p-orbital of the adsorbed O, indicat-
ing that Fe is bonded to the oxygen molecule. O2p orbits of three
iron oxides change after oxygen molecule adsorption, indicating
that charges also transfer between Fe and crystal oxygen during
the adsorption process. For O2p orbits of adsorbed oxygen, three
gaps of
respectively. The Fermi level of
top of the valence band, and the Fermi level of Fe3O4 is close to
a-Fe2O3, Fe3O4, and
c
-Fe2O3 are 1.16, 1.25, and 1.81 eV,
a-Fe2O3 and -Fe2O3 is near the
c
spikes can be observed clearly on
c-Fe2O3 in the range from ꢀ2

150
L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
Table 2
Bader charge analysis is further used to determine the charge
distribution of Fe and O atoms around adsorption sites of different
iron oxides (Table 3). For a pristine model slab, the effective
Oxidative coupling of alcohols and amines to imines over c
-Fe2O3 catalyst.a
Entry
1
Product
Reaction time (h)
8
Yield (%)b
91.9
charges of the Fe atom in
1.52, and 1.21, respectively. The effective charge of O in
is about ꢀ1.20, which is greater than that of Fe3O4 (ꢀ0.96) and
-Fe2O3 (ꢀ0.94). This result shows that the FeAO bond in
Fe2O3 has a greater contribution from covalent bonding than those
in -Fe2O3 and Fe3O4. This indicates that more electrons are shared
between Fe and O atoms in -Fe2O3. After adsorption of oxygen, a
significant change can be observed from the Bader charges of Fe
and O in -Fe2O3. This indicates that it is much easier for -Fe2O3
to transfer electrons to adsorbed oxygen. This trend is followed
by Fe3O4, while little change can be found in -Fe2O3. The elec-
tronic properties of different iron oxides further confirm that
a-Fe2O3, Fe3O4, and
c-Fe2O3 are 1.63,
c-Fe2O3
2
3
4
8
8
8
98.9
93.2
86.5
a
c-
a
c
c
c
5
8
81.2
a
6
7
8
8
95.2
69.1
92.3
c
-Fe2O3 possesses the best ability for oxygen activation.
According to our experimental results and corresponding liter-
12
8
ature [24–26], oxidative dehydrogenation of alcohols is the rate-
determining step for imine synthesis. The subsequent coupling
step of aldehydes with amines occurs easily on three iron oxides
(Fig. 3C and Fig. S7). Tsunami et al. have reported that super oxo-
or peroxo-like species could abstract a b-hydrogen atom from the
alcohol to generate the corresponding aldehydes or ketones [70].
Therefore, the O2 activation capability of catalysts is the key factor
for the whole catalytic process. In our case, the theoretical calcula-
9
12
12
85.4
78.9
10
a
Reaction conditions: alcohol 1 mmol, amine 2 mmol, catalyst 0.3 g, toluene
10 mL, 80 °C, air 1 atm.
tion results identify that c-Fe2O3 possesses favorable properties for
b
The yield of imines was determined using GC-MS.
charge transfer between Fe and O species, which can activate
molecular O2 to form O2ꢀ or O22ꢀ species. These oxygen species
are favorable for the dehydrogenation of alcohols, which is respon-
to ꢀ7 eV, which is different from those of Fe3O4 and
a
-Fe2O3. This
sible for the higher activity of
The O2 activation capability should be essentially ascribed to the
c-Fe2O3 than of Fe3O4 and a-Fe2O3.
indicates that c-Fe2O3 possesses a greater capacity for molecular
oxygen activation.
special inverse spinel structure of c-Fe2O3, which has abundant
100
80
60
40
20
0
(A)
(B)100
(C)100
γ-Fe2O3
80
80
60
40
20
0
60
40
20
0
Benzyl alcohol conversion
Imine yield
Benzaldehyde yield
Catalyst free
0
2
4
6
8
N2
0
60
120
Air
Reaction time (h)
Reaction time (min)
Reaction atmosphere
Fig. 3. (A) Effects of reaction conditions (air or N2) on catalytic performance of
coupling of benzyl alcohol and aniline over -Fe2O3 catalyst. (C) Time dependence of imine formation by coupling of benzaldehyde and aniline with and without
catalyst. Reaction conditions: benzyl alcohol or benzaldehyde 1 mmol, aniline 2 mmol, catalyst 0.3 g, toluene 10 mL, 80 °C, air or N2 1 atm.
c
-Fe2O3 catalyst in imine synthesis. (B) Time dependence of imine synthesis by oxidative
c
c
-Fe2O3
Fig. 4. Optimized configurations of O2 adsorption on the facets of (A)
a-Fe2O3 {0 0 1}, (B) Fe3O4 {1 1 1}, and (C) c-Fe2O3 {1 1 1}. The purple and pink colors represent Fe and O
atoms, respectively. O4 and O5 represent adsorbed oxygen molecules. For clarity, the symmetrical portion of the bottom slab is not shown.

L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
151
Fig. 5. Partial density of states (PDOS) for (A)
a-Fe2O3 {0 0 1} clean surface, (B) Fe3O4 {1 1 1} clean surface, (C)
c
-Fe2O3 {1 1 1} clean surface, (D) O2 adsorption on
a
-Fe2O3
{0 0 1} surface, (E) O2 adsorption on Fe3O4 {1 1 1} surface, and (F) O2 adsorption on
c
-Fe2O3 {1 1 1} surface. Fe is an iron atom in the surface adsorption site and O represents
oxygen atoms around the surface adsorption site (O3, O4, and O5). Dashed vertical lines mark the position of the Fermi level, and the Fermi level is set to 0 eV.
Table 3
Bader charges on Fe, O, and adsorbed O atoms in different iron oxides.
Sample
Fe
O1
O2
O3
O4
O5
a
a
-Fe2O3–slab
-Fe2O3–O2
1.63
1.65
1.52
1.65
1.21
1.62
ꢀ0.94
ꢀ0.97
ꢀ0.96
ꢀ0.94
ꢀ1.20
ꢀ1.12
ꢀ0.94
ꢀ0.96
ꢀ0.96
ꢀ0.94
ꢀ1.19
ꢀ1.17
ꢀ0.95
ꢀ0.93
ꢀ0.97
ꢀ0.95
ꢀ1.21
ꢀ1.16
–
–
ꢀ0.24
ꢀ0.12
Fe3O4–slab
Fe3O4–O2
–
–
ꢀ0.38
–
ꢀ0.37
ꢀ0.28
–
ꢀ0.38
c
c
-Fe2O3–slab
-Fe2O3–O2
cation vacancies. This structure determines the electronic proper-
ties of the surface Fe and O atoms of c-Fe2O3.
20180201068SF) and Open Project of the State Key Laboratory of
Inorganic Synthesis and Preparative Chemistry (201703), and the
Fundamental Research Funds for the Central Universities. L. Geng
thanks Dr. Haixiang Han from Cornell University for the discussion
and improvement of this article and the Natural Science Founda-
tion of Shandong Province for its support (ZR2018LB018).
4. Conclusions
Based on crystal phase regulation and screening, we obtain an
active and robust iron oxide catalyst for oxidative coupling in
imine synthesis.
selectivity under mild conditions, but also manifests the recycling
c-Fe2O3 not only possesses high activity and
Appendix A. Supplementary material
characteristics of magnetic recovery. Suitability for activating
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jcat.2019.06.018.
molecular oxygen is the critical factor for
dehydrogenation of alcohols. Theoretical analysis of surface,
geometric structure, and electronic structure shows that -Fe2O3
possesses excellent capability for molecular oxygen activation.
Compared with -Fe2O3 and Fe3O4, the surface Fe atoms of
c-Fe2O3 in oxidative
c
References
a
[1] H. Rao, L.-C. Schmidt, J. Bonin, M. Robert, Nature 548 (2017) 74–77.
[2] H. Arami, A. Khandhar, D. Liggitt, K.-M. Krishnan, Chem. Soc. Rev. 44 (2015)
8576–8607.
[3] S. Chakraborty, G. Leitus, D. Milstein, Angew. Chem. Int. Ed. 56 (2017) 2074–
2078.
[4] M. Ghosh, K.K. Singh, C. Panda, A. Weitz, M.P. Hendrich, T.J. Collins, B.B. Dhar, S.
S. Gupta, J. Am. Chem. Soc. 136 (2014) 9524–9527.
[5] J.M. Byrne1, N. Klueglein, C. Pearce, K.M. Rosso, E. Appel, A. Kappler, Science
347 (2015) 1473–1476.
[6] J.M. Hoyt, V.A. Schmidt, A.M. Tondreau, P.J. Chirik, Science 349 (2015) 960–
963.
c
-Fe2O3 easily transfer electrons to molecular oxygen to form O2ꢀ
or O22ꢀ species. This work provides molecular-level insights into
the crystal phase effect on catalytic performance in molecular
oxygen activation, and can shed light on developing other metal
oxide catalysts for molecular-oxygen-activation-related reactions.
Acknowledgments
[7] T.H. Maugh, Science 187 (1975) 154–156.
[8] S. Nagatomo, M. Okumura, K. Saito, O. Takashi, Biochemistry 56 (2017) 1261–
1273.
The authors acknowledge support from the National Natural
Science Foundation of China (21473073), the Development Project
of Science and Technology of Jilin Province (20170101171JC,
[9] M.S. Chen, M. Christina White, Science 327 (2010) 566–571.

152
L. Geng et al. / Journal of Catalysis 377 (2019) 145–152
[10] J.Z. Lu, L.J. Yang, B.L. Xu, Q. Wu, D. Zhang, S.J. Yuan, Y. Zhai, X.Z. Wang, Y.N. Fan,
Z. Hu, ACS Catal. 4 (2014) 613–621.
[40] H.P. Zhang, C. Wu, W.B. Wang, J. Bu, F.T. Zhou, B.L. Zhang, Q.Y. Zhang, Appl.
Catal. B Environ. 227 (2018) 209–217.
[11] L. Machala, J. Tucek, R. Zboril, Chem. Mater. 23 (2011) 3255–3272.
[12] H.M. Torres Galvis, J.H. Bitter, C.B. Khare, M. Ruitenbeek, A.I. Dugulan, K.P. de
Jong, Science 335 (2012) 835–838.
[13] M. Holmes, L.A. Schwartz, M.J. Krische, Chem. Rev. 118 (2018) 6026–6052.
[14] Y. Qin, L.H. Zhu, S.Z. Luo, Chem. Rev. 117 (2017) 9433–9520.
[15] A. Kumar, N.A. Espinosa-Jalapa, G. Leitus, Y. Diskin-Posner, L. Avram, D.
Milstein, Angew. Chem. Int. Ed. 56 (2017) 14992–14996.
[16] S.P. Midya, J. Pitchaimani, V.G. Landge, V. Madhu, E. Balaraman, Catal. Sci.
Technol. 8 (2018) 469–3473.
[17] K.Z. Demmans, C.S.G. Seo, A.J. Lough, R.H. Morris, Chem. Sci. 8 (2017) 6531–
6541.
[18] R. He, X. Jin, H. Chen, Z.-T. Huang, Q.-Y. Zheng, C. Wang, J. Am. Chem. Soc. 136
(2014) 6558–6561.
[41] R.T. Chen, S. Pang, H.Y. An, J. Zhu, S. Ye, Y.Y. Gao, F.T. Fan, C. Li, Nat. Energy 3
(2018) 655–663.
[42] R. Ye, J. Zhao, B.B. Wickemeyer, F. Dean Toste, G.A. Somorjai, Nat. Catal. 1
(2018) 318–325.
[43] A. Pratt, L. Lari, O. Hovorka, A. Shah, C. Woffinden, S.P. Tear, C. Binns, R. Kröger,
Nat. Mater. 13 (2013) 26–30.
[44] Z. Zhang, Y. Wang, M. Wang, J. Lü, L. Li, H. Zhang, M. Li, J. Jiang, F. Wang, Chin. J.
Catal. 36 (2015) 1623–1630.
[45] X.L. Mou, B.S. Zhang, Y. Li, L.D. Yao, X.J. Wei, D.S. Su, W.J. Shen, Angew. Chem.
Int. Ed. 51 (2012) 2989–2993.
[46] L.L. Geng, B. Zheng, X. Wang, W.X. Zhang, S.J. Wu, M.J. Jia, W.F. Yan, G. Liu,
ChemCatChem 8 (2016) 805–811.
[47] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865–3868.
[48] B. Hammer, J.K. Nørskov, Adv. Catal. 45 (2000) 71–129.
[49] L. Wang, T. Maxisch, G. Ceder, Phys. Rev. B. 73 (2006) 195107.
[50] R. Bliem, E. McDermott, P. Ferstl, M. Setvin, O. Gamba, J. Pavelec, M.A.
Schneider, M. Schmid, U. Diebold, P. Blaha, L. Hammer, G.S. Parkinson, Science
346 (2014) 1215–1218.
[19] R.D. Patil, S. Adimurthy, Asian J. Org. Chem. 2 (2013) 726–744.
[20] S. Kobayashi, Y. Mori, J.S. Fossey, M.M. Salter, Chem. Rev. 111 (2011) 2626–
2704.
[21] Y. Long, Z. Gao, J. Qin, P. Wang, W. Wu, L. Zhang, Z. Dong, J. Ma, J. Colloid
Interface Sci. 538 (2019) 709–719.
[22] J.L. Song, G.Y. Yu, X. Li, X.E. Yang, W.X. Zhang, W.F. Yan, G. Liu, Chin. J. Catal. 39
(2018) 309–318.
[23] L.L. Geng, J.L. Song, B. Zheng, S.J. Wu, W.X. Zhang, M.J. Jia, G. Liu, Chin. J. Catal.
37 (2016) 1451–1460.
[51] P. Liao, J.A. Keith, E.A. Carter, J. Am. Chem. Soc. 134 (2012) 13296–13309.
[52] P. Liao, E.A. Carter, PCCP 13 (2011) 15189–15199.
[53] M.T. Nguyen, N. Seriani, S. Piccinin, R. Gebauer, J. Chem. Phys. 140 (2014)
064703.
[24] M. Tamura, K. Tomishige, Angew. Chem. Int. Ed. 54 (2014) 864–867.
[25] B. Chen, L.Y. Wang, S. Gao, ACS Catal. 5 (2015) 5851–5876.
[26] L.L. Geng, J.L. Song, Y.H. Zhou, Y. Xie, J. Huang, H.W.X. Zhang, L.M. Peng, G. Liu,
Chem. Commun. 52 (2016) 13495–13498.
[27] Y.J. Song, H. Wang, S.J. Liang, Y. Yu, L.Y. Li, L. Wu, J. Catal. 361 (2018) 105–115.
[28] T. Song, J.E. Park, Y.K.J. Chung, Org. Chem. 83 (2018) 4197–4203.
[29] A. Maggi, R. Madsen, Organometallics 31 (2012) 451–455.
[30] T. Higuchi, R. Tagawa, A. Iimuro, S. Akiyama, H. Nagae, K. Mashima, Chem. Eur.
J. 23 (2017) 12795–12804.
[54] W. Tang, E. Sanville, G. Henkelman, J. Phys. Condens Mater. 21 (2009) 084204.
[55] E. Sanville, S.D. Kenny, R. Smith, G. Henkelman, J. Comput. Chem. 28 (2007)
899–908.
[56] P. Liu, T. Kendelewicz, G.E. Brown Jr, E.J. Nelson, S.A. Chambers, Surf. Sci. 417
(1998) 53–65.
[57] T. Kendelewicz, P. Liu, C.S. Doyle, G.E. Brown Jr., E.J. Nelson, S.A. Chambers,
Surf. Sci. 453 (2000) 32–46.
[58] R.C. Baetzold, H. Yang, J. Phys. Chem. B 107 (2003) 14357–14364.
ˇ
ˇ
[59] L. Machala, J. Tucek, R. Zboril, Chem. Mater. 23 (2011) 3255–3272.
[60] Y.T. Zhang, L.L. Li, W.F. Ma, Y. Zhang, M. Yu, J. Guo, H.J. Lu, C.C. Wang, A.C.S.
Appl, Mater. Interface 5 (2013) 614–621.
[31] G.-J. Chen, H.-C. Ma, W.-L. Xin, X.-B. Li, F.-Z. Jin, J.-S. Wang, M.-Y. Liu, Y.-B.
Dong, Inorg. Chem. 56 (2017) 654–660.
[32] H.X. Li, X.T. Wang, F. Huang, G. Lu, J.L. Jiang, Z.-X. Wang, Organometallics 30
(2011) 5233–5247.
[33] M.S. Kwon, S. Kim, S. Park, W. Bosco, R.K. Chidrala, J. Park, J. Org. Chem. 74
(2009) 2877–2879.
[34] H. Sun, F.-Z. Su, J. Ni, Y. Cao, H.-Y. He, K.-N. Fan, Angew. Chem. 121 (2009)
4454–4457.
[61] Y.H. Ni, X.W. Ge, Z.C. Zhang, Q. Ye, Chem. Mater. 14 (2002) 1048–1052.
[62] L.-S. Zhong, J.-S. Hu, H.-P. Liang, A.-M. Cao, W.-G. Song, L.-J. Wan, Adv. Mater.
18 (2006) 2426–2431.
[63] K.F. Zhao, H.L. Tang, B.T. Qiao, L. Li, J.H. Wang, ACS Catal. 5 (2015) 3528–3539.
[64] F. Jiao, J.-C. Jumas, M. Womes, A.V. Chadwick, A. Harrison, P.G.J. Bruce, J. Am.
Chem. Soc. 128 (2006) 12905–12909.
[35] P. Jing, T. Gan, H. Qi, B. Zheng, X. Chu, G. Yu, W. Yan, Y. Zou, W. Zhang, G. Liu,
Chin. J. Catal. 40 (2019) 214–222.
[36] G. Jaiswal, V.G. Landge, D. Jagadeesan, E. Balaraman, Green Chem. 18 (2016)
3232–3238.
[37] B. Chen, J. Li, W. Dai, L.Y. Wang, S. Gao, Green Chem. 16 (2014) 3328–3334.
[38] B. Chen, S.S. Shang, L.Y. Wang, Y. Zhang, S. Gao, Chem. Commun. 52 (2016)
481–484.
[39] A. Mukherjee, A. Nerush, G. Leitus, L.J.W. Shimon, Y.B. David, N.A.E. Jalapa, D.
Milstein, J. Am. Chem. Soc. 138 (2016) 4298–4301.
[65] J.H. Wang, Y.W. Ma, K. Watanabe, Chem. Mater. 20 (2008) 20–22.
[66] E. Darezereshki, M. Ranjbar, F. Bakhtiari, J. Alloy. Compd. 502 (2010) 257–260.
[67] C.S. Lo, K.S. Tanwar, A.M. Chaka, T.P. Trainor, Phys. Rev. B. 75 (2007) 075425.
[68] M.T. Nguyen, N. Seriani, R. Gebauer, ChemPhysChem 15 (2014) 2930–2935.
[69] Z. Kun, Z.Z. Li, J.W. Jia, X.L. Qun, W.H. Wei, J.Y. Jun, J. Am. Chem. Soc. 135 (2013)
15750–15753.
[70] H. Tsunoyama, N. Ichikuni, H. Sakurai, T. Tsukuda, J. Am. Chem. Soc. 131 (2009)
7086–7093.