Controllable Synthesis of Mesoporous Iron Oxide Nanoparticle Assemblies for Chemoselective Catalytic Reduction of Nitroarenes

Article abstract of DOI:10.1002/chem.201504685

Iron(III) oxide is a low-cost material with applications ranging from electronics to magnetism, and catalysis. Recent efforts have targeted new nanostructured forms of Fe₂O₃ with high surface area-to-volume ratio and large pore volume. Herein, the synthesis of 3D mesoporous networks consisting of 4-5 nm γ-Fe₂O₃ nanoparticles by a polymer-assisted aggregating self-assembly method is reported. Iron oxide assemblies obtained from the hybrid networks after heat treatment have an open-pore structure with high surface area (up to 167 m² g^-1) and uniform pores (ca. 6.3 nm). The constituent iron oxide nanocrystals can undergo controllable phase transition from γ-Fe₂O₃ to α-Fe₂O₃ and to Fe₃O₄ under different annealing conditions while maintaining the 3D structure and open porosity. These new ensemble structures exhibit high catalytic activity and stability for the selective reduction of aryl and alkyl nitro compounds to the corresponding aryl amines and oximes, even in large-scale synthesis.

Full text of DOI:10.1002/chem.201504685

DOI: 10.1002/chem.201504685
Full Paper
&
Mesoporous Materials |Hot Paper|
Controllable Synthesis of Mesoporous Iron Oxide Nanoparticle
Assemblies for Chemoselective Catalytic Reduction of Nitroarenes
Ioannis T. Papadas,[a] Stella Fountoulaki,[b] Ioannis N. Lykakis,*[b] and Gerasimos S. Armatas*[a]
Abstract: Iron(III) oxide is a low-cost material with applica-
tions ranging from electronics to magnetism, and catalysis.
Recent efforts have targeted new nanostructured forms of
Fe2O3 with high surface area-to-volume ratio and large pore
volume. Herein, the synthesis of 3D mesoporous networks
consisting of 4–5 nm g-Fe2O3 nanoparticles by a polymer-as-
sisted aggregating self-assembly method is reported. Iron
oxide assemblies obtained from the hybrid networks after
heat treatment have an open-pore structure with high sur-
face area (up to 167 m2 gÀ1) and uniform pores (ca. 6.3 nm).
The constituent iron oxide nanocrystals can undergo con-
trollable phase transition from g-Fe2O3 to a-Fe2O3 and to
Fe3O4 under different annealing conditions while maintain-
ing the 3D structure and open porosity. These new ensem-
ble structures exhibit high catalytic activity and stability for
the selective reduction of aryl and alkyl nitro compounds to
the corresponding aryl amines and oximes, even in large-
scale synthesis.
Introduction
dition, the morphology of the aggregated assemblies is irregu-
lar with limited porosity. In this context, new strategies to
access NP-based architectures with large and accessible porosi-
ty at the nanometer scale are necessary.
The application of transition metal oxide nanoparticles (NPs) in
catalysis has attracted a great deal of attention in the last
years.[1] The use of these nanomaterials offers several advan-
tages, such as unusual catalytic properties and high surface
area-to-volume ratio. Among them, iron(III) oxide in particular
is becoming very popular in organic synthesis due to its high
chemical stability, low cost, and magnetic separability.[2] Exam-
ples include borylation of arenes,[3] Sonogashira cross-cou-
pling,[4] oxidation of styrene[5] and olefins,[6] and hydrogenation
of nitroarenes.[7] More recently, Fe2O3 has been regarded as
a potential photoanode material in visible-light-driven water
splitting owing to its visible-light response (bandgap energy
ꢀ2–2.2 eV) and appropriate valence band edge position
(ꢀ2.5 V vs. NHE at pH 1).[8] In this area, continuous research ef-
forts have been made to synthesize well-defined iron oxide
nanostructures, and so far a large variety of Fe2O3 NPs with
finely controlled size (from 2 to 40 nm) and shape (cubes,
spheres, ellipsoids, and sheets) have been prepared.[9] Unfortu-
nately, implementation of such nanomaterials, especially in cat-
alysis and absorption, is not easy, as it often entails a strong
tendency of NPs to form large agglomerates in solution. In ad-
One interesting approach that has been used with signifi-
cant success in assembling porous networks from colloidal NPs
is polymer templating.[10] This method involves assembly of
soluble inorganic nano building blocks into various mesoscop-
ic structures with the aid of amphiphilic surfactants or block
copolymers. Porous networks of connected NPs hold promise
for applications in catalysis, solar-energy conversion, and size-
selective adsorption and separation.[11] Compared to individual
NPs, 3D porous ensembles of NPs are expected to have advan-
tageous characteristics such as rapid mass transport in the
pore channels of the assembled structure and interfacial trans-
fer of electrons along the framework. Furthermore, aligned
nanostructures of semiconductor nanocrystals may exhibit
new collective properties between adjacent particles that are
not present in the single-component materials.[12] However,
control over the morphology and porosity of NP aggregates at
the mesoscale is not trivial, as more often rapid precipitation
of randomly agglomerated NPs occurs.
We recently demonstrated that ordered mesoporous net-
works of interconnected bismuth ferrite (BiFeO3) NPs can be
prepared.[13] To produce this material, we used polymer-direct-
ed self-assembly of ligand-stabilized BiFeO3 nanocrystals. As
ligand we used 3-aminopropanoic acid, the carboxyl group of
which coordinated to the NP surface while the amino end
group interacted with the polar fragment of the polymer tem-
plate. Herein, we report on a new strategy that affords 3D
highly porous iron oxide architectures through the aggregat-
ing self-assembly (ASA) of ligand-stripped g-Fe2O3 NPs. Com-
pared to previous methods, this route does not require an or-
ganic capping agent for the g-Fe2O3 NPs and occurs in the
[a] Dr. I. T. Papadas, Prof. Dr. G. S. Armatas
Department of Materials Science and Technology
University of Crete
Heraklion 71003 (Greece)
E-mail: garmatas@materials.uoc.gr
[b] S. Fountoulaki, Prof. Dr. I. N. Lykakis
Department of Chemistry
Aristotle University of Thessaloniki
Thessaloniki 54124 (Greece)
E-mail: lykakis@chem.auth.gr
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201504685.
Chem. Eur. J. 2016, 22, 4600 – 4607
4600
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
presence of a nonionic surfactant (i.e., Pluronic P123). The net-
work structure of these new materials, which is made of con-
nected 4–5 nm-sized g-Fe2O3 particles, has high surface area
and uniform pores. Moreover, for the first time, we present re-
sults from thermally induced phase transitions to assess sys-
tematic variation of the crystal structure and properties of the
3D assembled networks. This enriches the family of mesopo-
rous iron oxide assemblies. The high surface area and tunable
electronic and redox properties make these NP assemblies suit-
able for catalytic applications. We show that the pore surface
of these materials is highly active and stable in selective reduc-
tion of aryl and alkyl nitro compounds to the corresponding
aryl amines and oximes with methyl hydrazine or hydrazine.
Indeed, the nitroarene reduction processes operate under mild
reaction conditions (608C), which is a challenging task for
amine synthesis.[14]
Scheme 1. Schematic illustration of the synthesis of mesoporous assemblies
from iron oxide NPs (ASA: aggregating self-assembly, C: calcination, MNAs:
mesoporous nanoparticle assemblies).
Results and Discussion
molecules from the hybrid mesostructure was verified by ther-
mal gravimetric analysis (TGA). The TGA data of the mesopo-
rous sample showed no appreciable weight loss up to 6008C
(Figure S1 in the Supporting Information), which indicated that
the polymer template fully decomposed on heating.
Synthesis and structural characterization
To produce mesoporous assemblies of iron oxide NPs, we used
ASA of ligand-stripped (bare) g-Fe2O3 NPs in the presence of
a block copolymer template. In particular, we first utilized ni-
trosonium tetrafluoroborate (NOBF4) to gently strip the native
ligands (oleyl alcohol) from the surface of preformed g-Fe2O3
The porous structure of g-Fe2O3-MNAs was characterized by
small-angle X-ray scattering (SAXS), X-ray diffraction (XRD), and
À
NPs and charge-stabilized them with BF4 anions. Recently, Mil-
liron et al. used the NOBF4 ligand-exchange strategy to devel-
op NP-based porous thin films of different particle sizes and
compositions.[15] The advantage of using ligand-free NPs is that
their surface OH groups can engage in direct ionic or hydro-
gen-bonding interactions with the polar chains (polyethylene
oxide, PEO) of the polymer template in an evaporation-in-
duced self-assembly process, and thereby result in the forma-
tion of ordered organic/inorganic composites. For the synthe-
sis of mesoporous g-Fe2O3 NP assemblies (denoted g-Fe2O3-
MNAs), BF4-capped g-Fe2O3 NPs dispersed in DMF were added
to a solution of poly(ethylene oxide)-b-poly(propylene oxide)-
b-poly(ethylene oxide) (EO20PO70EO20) triblock copolymer
(10 wt%) in ethanol. We found that, under certain reaction
conditions, including mass fraction of colloidal NPs and evapo-
ration rate of the solvent, the iron oxide NPs coassemble with
the liquid-crystalline block copolymers to produce hybrid
mesostructured NP–polymer composites. The reaction mixture
had
a
copolymer:g-Fe2O3:ethanol:DMF molar ratio of
0.08:1:59.4:14.8. The network assembly reactions between col-
loidal NPs and amphiphilic molecules should be slow enough
to avoid microphase segregation. Thus, the synthesis was car-
ried out at 408C under slow evaporation of solvents (over ca.
7 d). If the evaporation of solvents is performed at a higher
temperature (608C) or within a short period of time (a few
hours), a dark brown product with limited porosity is obtained.
The template was finally removed by thermal oxidation at ele-
vated temperature (3008C) to leave behind the mesoporous
network of g-Fe2O3 NPs with high porosity and uniform pores.
A schematic overview of the synthesis of g-Fe2O3-MNAs is
shown in Scheme 1. The complete elimination of the organic
Figure 1. a) SAXS patterns of mesoporous assemblies (g-Fe2O3-MNAs) (i) and
random aggragates (g-Fe2O3-RNAs) (ii) from g-Fe2O3 NPs (Inset: Guinier plot
for the g-Fe2O3-MNAs sample derived from scattering data. The line is fit to
the data). b) XRD pattern of g-Fe2O3-MNAs. The standard pattern of g-phase
Fe2O3 (JCPDS no. 39-1346) is given for comparison (gray line).
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4601
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
transmission electron microscopy (TEM). Figure 1a shows the
SAXS pattern, in which a weak, but resolved diffraction peak
centered at q=0.57 nmÀ1 (=4psinq/l, where 2q is the scatter-
ing angle) can be readily seen and corresponds to an average
repeat distance (d=2p/q) of 11 nm. Together with TEM images
(see below), the presence of this peak suggests the formation
of NP assemblies with porous periodicity at the mesoscale. The
size of the constituent NPs can be estimated by using the
Guinier approximation (ln[I(q)]/q2Rg2, where Rg is the radius of
gyration)[16] in the low-q region of the scattering plot. The aver-
age size of g-Fe2O3 NPs was thus determined to be 4.2 nm (Fig-
ure 1a, inset), which is very close to the particle size of the
starting materials (ca. 3.8 nm) found by TEM (Figure S2 in the
Supporting Information). In the absence of template, BF4-
capped g-Fe2O3 NPs self-assembled into aggregates with ran-
domly packed morphology. This is manifested clearly in the
SAXS pattern in Figure 1a, whereas the nontemplated g-Fe2O3
sample (denoted g-Fe2O3-RNAs) does not show any scattering
peak in the low-q range.
The wide-angle XRD pattern of g-Fe2O3-MNAs in Figure 1b
demonstrates the presence of a cubic (spinel-type) structure
with P4132 space group, which is identical to that of the pre-
cursor NPs (Figure S3 in the Supporting Information). Within
the XRD detection limits (ꢀ3–5 wt%) no impurity crystal
phase was detected (e.g., a-Fe2O3 or spinel Fe3O4 phase), and
this indicates that the obtained material is single-phase g-
Fe2O3. On the basis of the peak width of the (311) reflection
and the Scherrer equation, we calculated an average grain size
of g-Fe2O3 of about 5.5 nm, which is slightly larger than the di-
ameter of the constituent NPs (ca. 4.2 nm). This variation is in-
dicative of a minimal directional crystal growth of iron oxide
particles along the assembled network, which is due to the an-
nealing process. As a proof of concept, high-resolution TEM
imaging of mesoporous g-Fe2O3-MNAs showed that the pore
walls comprise nanocrystals that fuse together and form
a dense structure (Figure S4 in the Supporting Information).
We surmised that such a close contact between NPs would be
helpful for designing porous assemblies with good structural
stability and high electron-transfer mobility along the grain
boundaries.
Figure 2. a), b) Representative TEM images (inset of a: high-magnification
TEM image) and particle size distribution plot (inset of b) showing an aver-
age g-Fe2O3 particle size of 4.6Æ0.5 nm and c) SAED pattern for the meso-
porous g-Fe2O3-MNAs. d) N2 adsorption/desorption isotherms at À1968C and
the corresponding NLDFT pore size distribution plots (inset) of the g-Fe2O3-
MNAs (i) and g-Fe2O3-RNAs (ii).
Fe2O3-MNAs are of type IV with an H2-type hysteresis loop,
which is typical of interconnected mesoporous systems.[17] The
BET surface area and total pore volume assessed from the ad-
sorption branch of the isotherm are 167 m2 gÀ1 and
0.19 cm3 gÀ1, respectively. Based on the NLDFT model fit to the
adsorption data (assuming cylindrical pores), the pore diameter
of the g-Fe2O3-MNAs sample was estimated to be about
6.3 nm with narrow pore-size distribution (Figure 2d, inset).
Given an average pore-to-pore distance of 11 nm (determined
by SAXS), the difference between the repeat distance and the
pore diameter gives an estimate of the wall thickness of
4.7 nm, which is very close to the diameter of the constituent
g-Fe2O3 NPs (ꢀ4–5 nm). This is consistent with the mesopo-
rous framework comprising a single layer of g-Fe2O3 NPs. In
comparison, the g-Fe2O3-RNAs material obtained by template-
free aggregating assembly of ligand-stripped g-Fe2O3 NPs dis-
plays a distinct N2 adsorption/desorption isotherm that is
a combination of types I and IV with H3-type hysteresis loop
(Figure 2d), typically found for random porous networks with
slitlike pores.[17] Analysis of the adsorption data revealed that
this sample has a BET surface area of about 42 m2 gÀ1 and
a total pore volume of about 0.07 cm3 gÀ1. In addition, the esti-
mated NLDFT pore-size distribution plot (assuming slit-shaped
pores) suggests the formation of NP aggregates with small in-
terstitial voids (ꢀ3.2 nm). Together with the SAXS results, the
remarkable difference in pore morphology noted between the
mesoporous assemblies and randomly interconnected network
suggests that polymer templating dominates the assembly
process to give mesoscopic ensembles of g-Fe2O3 NPs.
Typical TEM images of the g-Fe2O3-MNAs material (Figure 2a
and b) show an extended network morphology with porous
structure. The high-magnification images show that the pore
walls consist of interconnected NPs with an average diameter
of 4–5 nm, which is in good agreement with the particle size
determined by SAXS. The crystal structure of the g-Fe2O3-MNAs
was further examined by selected-area electron diffraction
(SAED). The SAED pattern taken from a small area of the
porous network corroborates the crystallinity noted by XRD,
showing Debye–Scherrer diffraction rings that match the cubic
closed-packed symmetry of g-Fe2O3 (Figure 2c).
Nitrogen sorption measurements showed that the template-
free material has an accessible porous structure with uniform
pores. Figure 2d shows the N2 adsorption/desorption iso-
therms and the corresponding nonlocal density functional
theory (NLDFT) pore size distribution plots of g-Fe2O3-MNAs
and nontemplated analogue g-Fe2O3-RNAs. The isotherms of g-
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4602
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Interestingly, the g-Fe2O3 assemblies can be thermally con-
verted to materials with hematite (a-Fe2O3) and cubic spinel
(Fe3O4) structure while maintaining textural integrity. In gener-
al, solid-state transformation is a viable method for creating
materials with specific crystal structure and morphology. While
this method has been used for the solid–solid transition of iso-
lated NPs[18] and thin films,[15b,19] extrapolation to 3D porous
networks of iron oxide NPs is a challenge. This is because ther-
mal annealing can cause destruction of the pore structure
owing to the large grain growth of assembled nanocrystals, es-
pecially at high temperatures. In successive trial experiments,
we found that transformation of maghemite ferrite into single
hematite (a-Fe2O3) crystals occurs on heating the g-Fe2O3-
MNAs in air at 3508C. We interpret the g!a phase transition
as a coalescence and recrystallization process of g-Fe2O3, in
which thermal annealing causes germination of hematite
phase at the grain-boundary areas of NPs.[20] Subsequent irre-
versible crystal transformation of a-Fe2O3 into spinel Fe3O4
structured assemblies can be achieved by quenching a-Fe2O3-
MNAs in a reducing atmosphere, that is, under H2/Ar (5/95 v/v)
flow at 3008C. Evidence for this was obtained by XRD and in-
frared (FTIR) spectroscopy. The XRD patterns show the pres-
ence of hexagonal a-Fe2O3 phase for a-Fe2O3-MNAs and cubic
spinel-structured Fe3O4 for Fe3O4-MNAs (Figure 3a). In addition,
Scherrer analysis of the XRD peaks [(110) and (104) reflections
for a-Fe2O3-MNAs and (311) reflection for Fe3O4-MNAs] gives an
average grain size of about 6 nm for both a-Fe2O3-MNAs and
Fe3O4-MNAs, which indicates small crystallite grain growth (or
fusion) during phase transition. The FTIR spectrum of mesopo-
rous g-Fe2O3 assemblies shows characteristic absorption peaks
at 442, 558, and 636 cmÀ1, which are assigned to the vibrations
of the FeO groups of the maghemite (g-Fe2O3) structure (Fig-
ure 3b).[18] The spectrum of the a-Fe2O3-MNAs features two
prominent peaks at 471 and 539 cmÀ1, which are attributed to
the Eu and A2u +Eu phonon modes of the hematite (a-Fe2O3)
structure, respectively. For the Fe3O4-MNAs sample the intense
peak around 565 cmÀ1 is assigned to the T1u mode of magnet-
ite.[21] N2 sorption measurements also evidenced that these
new iron oxide assemplies are still mesoporous; the a-Fe2O3-
MNAs and Fe3O4-MNAs have BET surface areas of 162 and
147 m2 gÀ1 and total pore volumes of 0.23 and 0.14 cm3 gÀ1, re-
spectively (Figure 4).
Catalytic activity
The high pore volume and accessible surface area of the as-
sembled mesoporous networks of iron oxide NPs prompted us
to investigate their catalytic properties in the reduction of ni-
Figure 3. a) XRD patterns of mesoporous a-Fe2O3-MNAs (i) and Fe3O4-MNAs
(ii). The standard patterns of hexagonal a-Fe2O3 (JCPDS no. 33-0664) and
cubic spinel structured Fe3O4 (JCPDS no. 65-3107) are also given for compar-
ison (gray lines). b) Transmission-mode FTIR spectra of the primary g-Fe2O3-
MNAs (i) and thermally converted a-Fe2O3-MNAs (ii) and Fe3O4-MNAs (iii) ma-
terials.
Figure 4. N2 adsorption/desorption isotherms at À1968C and the corre-
sponding NLDFT pore size distribution plots (insets) for mesoporous a) a-
Fe2O3-MNAs and b) Fe3O4-MNAs. The NLDFT pore size distribution plots
show pore sizes of 6.4 and 5.6 nm for a-Fe2O3-MNAs and Fe3O4-MNAs, re-
spectively.
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4603
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
troarenes. The chemoselective reduction of nitro to amino
groups is a key process in organic synthesis. For instance,
amino groups can serve as reactive sites for further derivatiza-
tion to produce pharmaceuticals, agrochemicals, and bioactive
compounds.[22] Initially, we optimized the catalytic conditions
by studying the reduction of 4-nitroaniline (1) with different re-
ducing agents and solvents (Supporting Information, Table S1,
entries 1–9). Among the examined conditions, g-Fe2O3-MNAs
catalyst (10 mol% based on 1) with four equivalents of methyl
hydrazine (MeNHNH2) in ethanol afforded 53% conversion of
1 to 4-aminoaniline (1a) in 1 h (Supporting Information,
Table S1, entry 5).[23] No byproducts such as azoxy-, azo-, or 1,2-
diarylhydrazine were detected during the reaction progress by
The present catalyst system could efficiently promote the
synthesis of aryl amines from various substituted nitroarenes. It
was found that the mesoporous g-Fe2O3-MNAs catalyze the re-
duction of nitroarenes 2–13 almost quantitatively to their cor-
responding aryl amines 2a–13a (>94% yields; see Scheme 2).
1
means of H NMR spectroscopy. In stark contrast, no reduction
of 1 was observed under mild hydrogenation with H2 (1 atm)
or other transfer hydrogenation reagents such as hydrosilanes
(1,1,3,3-tetramethyldisiloxane and dimethylphenylsilane) and
boranes (NaBH4 and NH3BH3) (Supporting Information,
Table S1, entries 6–9). In the absence of catalyst no appreciable
reduction of 1 takes place; that is, the reduction process is cat-
alytic in nature. For comparison, use of commercially obtained
iron compounds and bulk g-Fe2O3 solid as catalysts led to neg-
ligible conversions to 1a (Supporting Information, Table S1, en-
tries 10–14), even over a four-hour reaction period. Also,
random g-Fe2O3 aggregates (g-Fe2O3-RNAs) and mesoporous
assemblies from a-Fe2O3 (a-Fe2O3-MNAs) and Fe3O4 (Fe3O4-
MNAs) NPs resulted in lower conversions of 1 (Supporting In-
formation, Table S1, entries 15–17). These results clearly sug-
gest that the high catalytic activity of g-Fe2O3-MNAs is related
not only to the large pore volume being accessible to reac-
tants but also to the crystal phase of the iron oxide NPs, which
provide active facets and edge sites for the reaction.[7b] We fur-
ther found that a small addition of water significantly im-
proved the reactivity of g-Fe2O3-MNAs (see Supporting Infor-
mation, Figure S5 and Table S1, entries 18–20). Presumably,
under these reaction conditions, the nitro groups are protonat-
ed to give more electrophilic substrates[24] that bind in the H
sites of the catalysts (see below), although further studies are
needed to ascertain this effect.
Scheme 2. Reduction of various aryl and alkyl nitro compounds by g-Fe2O3-
MNAs catalyst. The percentages correspond to the yields of isolated prod-
ucts. In parentheses: the corresponding yields of isolated products with hy-
drazine as reducing agent.
Notably, nitroarenes containing additional reducible groups,
such as halo [chloro (5) and bromo (6)], methyl carboxylate (7),
cyano (8), and carbonyl (12), were selectively reduced to the
target amines. Similarly, the reaction process was found to be
inactive towards other easily reducible groups such as lactone
(11) and vinyl (13) substituents under the present catalytic
conditions. In addition, the same reductions were also per-
formed by using hydrazine as reducing agent, which gave sim-
ilar product yields at certain reaction times (Scheme 2, values
in parentheses). These results clearly support the high propen-
sity of the g-Fe2O3-MNAs catalyst to selectively reduce nitroar-
enes to aryl amines. Remarkably, the g-Fe2O3-MNAs could also
catalyze reduction of nitroalkanes 14–17, which do not readily
react,[25] to give the corresponding oximes 14b–17b as the
only products, albeit in lower yields (73–89%), and with an
eightfold excess of MeNHNH2 (Scheme 2). Currently, NPs of
precious metals such as Rh, Ni, Pd, and Au[24b,26] are regarded
as the state-of-the-art catalysts for reduction of aromatic nitro
compounds with hydrazine. However, the high cost and poor
durability (they are moisture-sensitive) of these nanomaterials
pose significant limitations to large-scale application. The iron
oxide assemblies described herein can be considered to be
highly effective, stable (even in water), and inexpensive alter-
native catalysts for widespread use in amine production. To
the best of our knowledge, only one example of an g-Fe2O3-
based catalyst exhibits similarly high catalytic activity to our
material, but it requires an excess of hydrazine and harsh con-
ditions (858C).[7b]
Owing to the 3D structure and magnetic susceptibility of
the g-Fe2O3-MNAs, this catalyst could be readily recycled by
simple filtration or by using an external magnetic field. As
shown in Figure S6 of the Supporting Information, the catalyst
reaches an almost constant activity, giving as high as 99% con-
version to 1a in 2 h even after seven consecutive runs. N2
sorption isotherms of the reused sample show no obvious
changes in the surface area and pore volume (162 m2 gÀ1 and
0.19 cm3 gÀ1; Figure S7, Supporting Information), reflecting ex-
cellent durability and recyclability. Given these encouraging re-
sults, the g-Fe2O3-MNAs catalyst was also tested for possible
large-scale production of anilines from nitroarenes. In particu-
lar, 3.6 mmol of 1 was reduced in the presence of g-Fe2O3-
MNAs (4 mol%) with MeNHNH2 (6 equiv) in MeOH (10 mL).
After completion of the reaction (ꢀ9 h based on TLC analysis),
the catalyst was magnetically separated from the mixture and
the corresponding aniline 1a was purified by column chroma-
tography and isolated in 91% yield.
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4604
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Furthermore, kinetic studies indicated an electronic effect in
the hydrogenation reactions. The results in Figure S8 and S9 of
the Supporting Information reveal that the reactivity of para-X-
substituted nitroarenes [X=4-MeO (2), 4-Me (3), 4-H (4), 4-Cl
(5), and 4-COOMe (7)] is remarkably affected by the nature of
the X substituent, whereby the reduction proceeds faster as
the electron-withdrawing ability of the substituent increases.
For example, 5 (4-Cl) and 7 (4-COOMe) were reduced at
a faster rate than 4 (X=H), as indicated by the relative rate
constant ratios kCl/kH =4 and kCOOMe/kH =11. On the contrary, ni-
troarenes 2 (4-MeO) and 3 (4-Me) with electron-donating sub-
stituents were converted to the corresponding amines 2a and
3a at slower reaction rates with kMeO/kH =0.4 and kMe/kH =0.7.
Hammett-type correlations for the competition of para-X-sub-
stituted nitroarenes [X=MeO (2), Me (3), Cl (5), and COOMe
(7)] versus nitrobenzene (4)[27] gave positive slopes of 1ꢀ1.2
for s+ values and 1ꢀ1.9 for s values (see Figure S10 of the
Supporting Information). These findings clearly suggest that
a negative charge (or hydride transfer) develops in the transi-
tion state, which is better stabilized by electron-withdrawing
substituents. These results are in agreement with previous
mechanistic studies on similar nitroarene reductions. It was
suggested that iron oxide hydroxide/hydrazine complexes,
which are formed by adsorption of hydrazine on the surface of
iron oxide particles, act as hydrogen-transfer centers for the re-
duction of nitroarenes.[14d,28] On the basis of these observa-
tions, it is reasonable to deduce that our hydrazine/g-Fe2O3-
MNAs catalyst system operates through a similar mechanism,
and therefore a transfer hydrogenation process can be pro-
posed.
Scheme 3. Proposed pathways for the g-Fe2O3-MNAs-catalyzed reduction of
nitroarenes to anilines by methylhydrazine or hydrazine (R=H, Me;
Ar=C6H5).
zine or hydrazine by the g-Fe2O3-MNAs catalyst. However, addi-
tional work is necessary to understand the mechanism of the
above catalytic reactions.
Conclusion
We have demonstrated a new versatile method for synthesiz-
ing 3D mesoporous assemblies of iron oxide NPs. The synthetic
scheme, which involves polymer-assisted ASA of ligand-strip-
ped g-Fe2O3 NPs, enables synthesis of porous architectures of
iron oxide NPs with large internal surface area and uniform
pores. We also showed that the constituent iron oxide nano-
crystals could undergo controllable crystal-to-crystal phase
transition (from g-Fe2O3 to a-Fe2O3 and to Fe3O4) through an
annealing process under different conditions, while maintain-
ing the 3D structure and open porosity of the assembled struc-
ture. The porous network of g-Fe2O3 NPs showed impressive
catalytic activity and stability for the selective reduction of vari-
ous aryl and alkyl nitro compounds to the corresponding aryl
amines and oximes. The present approach can provide a gener-
ic synthetic pathway for the rational design and fabrication of
other porous networks of metal oxide NPs. Such high-surface-
area porous assemblies may enable low-cost catalysts, chemi-
cal sensors, and electrochemical energy-storage and energy-
conversion devices.
It is noteworthy that reduction of nitroarenes 2, 3, 5, and 7
in CD3OD gives only the electron-poor N-hydroxylamines as in-
termediate (see Figure S11–S14 of the Supporting Information);
in all cases, no significant amount of the corresponding azoxy-
1
or azoarenes was detected by H NMR spectroscopy. In addi-
tion, an azoxy arene (1,2-bis(4-methoxyphenyl)diazene oxide)
did not react under the same conditions, whereas azobenzenes
react with g-Fe2O3-MNAs to form the corresponding hydrazo-
benzenes after 4 h (Figure S15 and S16 in the Supporting Infor-
mation). From these results, it appears that the condensation
route, which involves formation of azoxy- or azoarene com-
pounds as intermediates,[24b] does not take place in the reac-
tion system presented herein.
More likely, there are two reaction pathways that follow the
direct reduction route. For electron-rich nitroarenes, such as 2
and 3, nitrosoarenes may be formed as intermediate products
that are thermodynamically stable through resonance, which
are rapidly reduced to anilines by methylhydrazine or hydra-
zine (pathway I, Scheme 3). Nitroarenes that in principle form
less thermodynamically stable nitrosoarenes, such as 5 and 7,
do not follow the direct route and rather skip formation of the
nitrosoarene intermediate, leading directly to N-aryl hydroxyla-
mines (pathway II, Scheme 3). A similar mechanistic profile has
also been proposed for the transfer hydrogenation of nitroar-
enes catalyzed by different iron oxide particles.[14a,d,29] In gener-
al, it is possible the nitroarene reduction occurs through iron
hydride species that are formed by activation of methyl hydra-
Experimental Section
Materials
Poly(ethylene glycol)-b-poly(propylene glycol)-b-poly(ethylene
glycol) (Mn ꢀ5800) triblock copolymer (Pluronic P123), oleyl alcohol
(85%), sodium oleate (80%), diphenyl ether (99%), dimethylforma-
mide (99.9%), acetone, dichloromethane (99%), acetonitrile
(99.9%), hexane (95%), toluene (99.7%), and absolute ethanol
were purchased from Aldrich Chemical Co. Nitrosonium tetrafluor-
oborate (NOBF4, 97%) was purchased from Acros Organics. Iron(III)
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4605
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
sulfate hydrate (Fe2(SO4)3·xH2O, ꢁ97%), iron(III) nitrate nonahy-
drate (Fe(NO3)3·9H2O, ꢁ98%), iron(III) citrate (C6H5FeO7), FeCl3
(97%), and g-Fe2O3 were obtained from Sigma-Aldrich. Aromatic
nitro compounds were of high purity and commercially available
from Aldrich. Primary nitro alkanes were prepared by reaction of
the corresponding amines with an excess of m-chloroperbenzoic
acid in CH2Cl2 at room temperature.[30] The products were then ob-
tained in high yield (>90%) by evaporating the solvent under re-
duced pressure. Phenyl-substituted nitro alkanes were synthesized
according to the literature procedure.[31]
were obtained with a PerkinElmer Model Frontier FTIR spectrome-
ter with 2 cmÀ1 resolution.
Catalytic reactions
Reduction of nitro compounds: MeNHNH2 or NH2NH2 (0.8 mmol)
and catalyst (10 mol%) were added to a sealed tube containing
the nitro compound (0.2 mmol) and ethanol (1 mL), and the reac-
tion mixture was heated at 608C for an appropriate time under an
inert atmosphere. The reaction was monitored by TLC, and after
completion the slurry was filtered or centrifuged to separate the
catalyst and washed twice with ethanol or methanol (ꢀ2 mL). The
filtrate was evaporated under vacuum to afford the corresponding
amines in pure form. The 1H and 13C NMR spectroscopic data of
amines 1a–13a were in agreement with those previously report-
ed.[35]
Synthesis of NPs and MNAs
Preparation of ligand-stripped g-Fe2O3 NPs: g-Fe2O3 NPs with an
average diameter of 4 nm were prepared according to the previ-
ously reported procedure.[32] Next, the surface of iron oxide NPs
was modified with NOBF4 by a ligand-exchange reaction.[33] The
BF4-capped g-Fe2O3 NPs were collected by precipitation with tolu-
ene followed by centrifugation, dried under vacuum at 608C for
12 h, and dispersed in DMF to form a stable colloidal solution
(140 mgmLÀ1).
Recycling studies: 4-Nitroaniline (0.3 mmol), catalyst (10 mol%),
methylhydrazine (1.2 mmol), and ethanol (2 mL) were added to
a vial. The reaction mixture was stirred for 2 h at 608C. After each
catalytic run, the catalyst was isolated by filtration or by means of
an external magnetic field, washed with ethanol, and dried under
ambient conditions. Product analysis was based on NMR spectra.
Synthesis of mesoporous g-Fe2O3 NP assemblies: In a typical syn-
thesis, Pluronic P123 triblock copolymer (0.2 g) was dissolved in
ethanol (1.5 mL) at room temperature. Then, colloidal g-Fe2O3 NP
solution (0.5 mL) was mixed with the surfactant solution by vigo-
rous stirring for 2 h. The resulting mixture was then transferred to
an open-top container (20 mL volume) and left at 408C for about
one week under static conditions. This procedure gave a yellow-
brown dry powder. Template removal was achieved by gently heat-
ing the dry product at 3008C for 6 h in air at a heating rate of
Kinetic studies: Nitroarene (0.3 mmol), catalyst (10 mol%), and
methylhydrazine (1.2 mmol) were stirred in methanol (2 mL), and
a 100 mL aliquots of the mixture were taken at appropriate times.
The mixture was centrifuged to separate the catalyst and the fil-
trate was evaporated under vacuum. Then, the corresponding con-
sumptions of nitroarenes were determined directly by integrating
the appropriate proton signals in the 1H NMR spectra. Each reac-
tion was repeated at least three times, and the average values are
depicted in Figure S8 of the Supporting Information.
0.58CminÀ1
.
Reductions in deuterated solvent: Nitroarene (0.2 mmol), catalyst
(10 mol%), methylhydrazine (0.8 mmol), and CD3OD (1 mL) were
added to a vial. The reaction mixture was stirred for an appropriate
time at 608C, and the catalyst was removed by using a magnet.
The relative amounts of the products, intermediates, and the re-
Physical characterization
TGA was performed with a PerkinElmer Diamond system. Thermal
analysis was conducted from 40 to 6008C in air (200 mLminÀ1 flow
rate) with a heating rate of 58C minÀ1. SAXS measurements were
performed with a Rigaku S-MAX 300 high-brilliance system by
using CuKa radiation (80 kV and 40 mA). Measurements were per-
formed by transmission in samples that were gently ground and
held in a quartz capillary tube (inner diameter of 1 mm). The
sample-to-detector distance and center of the beam were precisely
measured by using a Ag behenate standard (d001 =58.38 Š). The
two-dimensional diffraction images were integrated into a one-di-
mensional diffraction pattern, as a function of q, with the Fit2D
program.[34] Scattering data were corrected for dark current and
empty-tube scattering. Wide-angle XRD patterns were collected
with a PANanalytical X’pert Pro MPD X-ray diffractometer equipped
with a Cu (l=1.5418 Š) rotating anode operated at 40 mA and
45 kV in the Bragg–Brentano geometry. TEM images were taken
with a JEOL Model JEM-2100 electron microscope operating at
200 kV acceleration voltage. Samples for TEM analysis were pre-
pared by drying an ethanolic dispersion of the particles on
a holey-carbon-coated Cu grid. Nitrogen adsorption/desorption
isotherms were measured at -196oC by using liquid N2 with a Quan-
tachrome Model NOVA 3200e sorption analyzer. Before analysis,
samples were degassed overnight at 1508C under vacuum (<
10À4 mbar). The specific surface areas were calculated by the BET
method on the adsorption data in the relative pressure (P/P0)
range of 0.05–0.24. The total pore volumes were estimated from
the adsorbed amount at P/P0 =0.98 and the pore size distributions
were obtained from the adsorption branch of isotherms by the
NLDFT method. IR spectra of dried samples pressed into KBr pellets
1
maining starting materials were determined by H NMR spectrosco-
py.
Acknowledgements
This research was supported by the European Union and
Greek Ministry of Education under the ERC grant schemes
(ERC-09) and THALIS project (MIS 377365). I.N.L. acknowledges
the COST action (CM1201) for sponsorship. We thank Dr. E. Ev-
genidou (AUTH) and Dr. A. Katsoulidis (University of Liverpool)
for collecting the LC-MS and SAXS data, respectively.
Keywords: amines · iron oxides · mesoporous materials ·
nanoparticles · reduction
[1] D. Astruc in Nanoparticles and Catalysis, Wiley-VCH, Weinheim, 2008.
[2] A. H. Lu, E. L. Salabas, F. Schüth, Angew. Chem. Int. Ed. 2007, 46, 1222–
1244; Angew. Chem. 2007, 119, 1242–1266.
[3] G. Yan, Y. Jiang, C. Kuang, S. Wang, H. Liu, Y. Zhang, J. Wang, Chem.
Commun. 2010, 46, 3170–3172.
[4] H. Firouzabadi, N. Iranpoor, M. Gholinejad, J. Hoseini, Adv. Synth. Catal.
2011, 353, 125–132.
[5] F. Rajabi, N. Karimi, M. R. Saidi, A. Primo, R. S. Varma, R. Luque, Adv.
Synth. Catal. 2012, 354, 1707–1711.
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4606
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[6] F. Shi, M. K. Tse, M.-M. Pohl, A. Brückner, S. Zhang, M. Beller, Angew.
Chem. Int. Ed. 2007, 46, 8866–8868; Angew. Chem. 2007, 119, 9022–
9024.
[7] a) R. V. Jagadeesh, A.-E. Surkus, H. Junge, M.-M. Pohl, J. Radnik, J.
Rabeah, H. Huan, V. Schunemann, A. Bruckner, M. Beller, Science 2013,
342, 1073–1076; b) X. Gu, Z. Sun, S. Wu, W. Qi, H. Wang, X. Xu, D. Su,
Chem. Commun. 2013, 49, 10088–10090.
[8] K. Sivula, F. L. Formal, M. Grätzel, ChemSusChem 2011, 4, 432–449.
[9] X. Hu, J. C. Yu, Adv. Funct. Mater. 2008, 18, 880–887.
[10] a) S. C. Warren, L. C. Messina, L. S. Slaughter, M. Kamperman, Q. Zhou,
S. M. Gruner, F. J. DiSalvo, U. Wiesner, Science 2008, 320, 1748–1752;
b) I. Vamvasakis, K. S. Subrahmanyam, M. G. Kanatzidis, G. S. Armatas,
ACS Nano 2015, 9, 4419–4426.
[11] Z. Nie, A. Petukhova, E. Kumacheva, Nat. Nanotechnol. 2010, 5, 15–25.
[12] a) Z. Lua, Y. Yin, Chem. Soc. Rev. 2012, 41, 6874–6887; b) C. Schmidtke,
R. Eggers, R. Zierold, A. Feld, H. Kloust, C. Wolter, J. Ostermann, J.-P.
Merkl, T. Schotten, K. Nielsch, H. Weller, Langmuir 2014, 30, 11190–
11196.
[22] S. A. Lawrence in Amines: Synthesis Properties and Applications, Cam-
bridge University Press, New York, 2004.
[23] Comparatively, a lower MeNHNH2 concentration (less than 2 equiv) re-
sults in incomplete reduction of 1, whereas with 3 equiv of MeNHNH2
a prolonged reaction time is needed (16 h).
[24] a) S. Fountoulaki, V. Daikopoulou, P. L. Gkizis, I. Tamiolakis, G. S. Armatas,
I. N. Lykakis, ACS Catal. 2014, 4, 3504–3511; b) P. L. Gkizis, M. Stratakis,
I. N. Lykakis, Catal. Commun. 2013, 36, 48–51.
[25] R. J. Rahaim, R. E. Maleczka, Org. Lett. 2005, 7, 5087–5090.
[26] a) P. Luo, K. Xu, R. Zhang, L. Huang, J. Wang, W. Xing, J. Huang, Catal.
Sci. Technol. 2012, 2, 301–304; b) U. Sharma, N. Kumar, P. K. Verma, V.
Kumar, B. Singh, Green Chem. 2012, 14, 2289–2293; c) S. Dutta, S.
Sarkar, C. Ray, A. Roy, R. Sahoo, T. Pal, ACS Appl. Mater. Interfaces 2014,
6, 9134–9143.
[27] With an excess of methyl hydrazine (more than 4 equiv based on nitro-
arene), the percentage consumption of nitroarene as a function of time
was found to be constant, that is, zero-order dependence on MeNHNH2
concentration. Thus, formal Hammett-type kinetics for the reduction of
the para-X-substituted nitroarenes [X=4-MeO (2), 4-Me (3), 4-H (4), 4-Cl
(5) and 4-COOMe (7)] was studied in the presence of 4 equiv of
MeNHNH2 in methanol.
[13] I. T. Papadas, K. S. Subrahmanyam, M. G. Kanatzidis, G. S. Armatas, Nano-
scale 2015, 7, 5737–5743.
[14] a) D. Cantillo, M. M. Moghaddam, C. O. Kappe, J. Org. Chem. 2013, 78,
4530–4542; b) D. Cantillo, M. Baghbanzadeh, C. O. Kappe, Angew.
Chem. Int. Ed. 2012, 51, 10190–10193; Angew. Chem. 2012, 124, 10337–
10340; c) R. V. Jagadeesh, G. Wienhofer, F. A. Westerhaus, A.-E. Surkus,
M.-M. Pohl, H. Junge, K. Junge, M. Beller, Chem. Commun. 2011, 47,
10972–10974; d) Q. Shi, R. Lu, K. Jin, Z. Zhang, D. Zhao, Green Chem.
2006, 8, 868–870.
[15] a) R. Buonsanti, T. E. Pick, N. Krins, T. J. Richardson, B. A. Helms, D. J. Milli-
ron, Nano Lett. 2012, 12, 3872–3877; b) I. E. Rauda, R. Buonsanti, L. C.
Saldarriaga-Lopez, K. Benjauthrit, L. T. Schelhas, M. Stefik, V. Augustyn, J.
Ko, B. Dunn, U. Wiesner, D. J. Milliron, S. H. Tolbert, ACS Nano 2012, 6,
6386–6399.
[16] O. Glatter, O. Kratky in Small-Angle X-ray Scattering, Academic Press,
New York, 1982.
[17] F. Rouquerol, J. Rouquerol, K. S. W. Sing in Adsorption by Powders and
Porous Solids. Principles Methodology and Applications, Academic Press,
London, 1999.
[28] a) M. Benz, R. Prins, Appl. Catal. A 1999, 183, 325–333; b) M. Lauwiner,
P. Rys, J. Wissmann, Appl. Catal. A 1998, 172, 141–148.
[29] G. Wienhçfer, I. Sorribes, A. Boddien, F. Westerhaus, K. Junge, H. Junge,
R. Llusar, M. Beller, J. Am. Chem. Soc. 2011, 133, 12875–12879.
[30] K. E. Gilbert, W. T. Borden, J. Org. Chem. 1979, 44, 659–661.
[31] G. W. Kabalka, L. H. M. Guindi, R. S. Varma, Tetrahedron 1990, 46, 7443–
7457.
[32] B. H. Kim, N. Lee, H. Kim, K. An, Y. I. Park, Y. Choi, K. Shin, Y. Lee, S. G.
Kwon, H. B. Na, J.-G. Park, T.-Y. Ahn, Y.-W. Kim, W. K. Moon, S. H. Choi, T.
Hyeon, J. Am. Chem. Soc. 2011, 133, 12624–12631.
[33] A. Dong, X. Ye, J. Chen, Y. Kang, T. Gordon, J. M. Kikkawa, C. B. Murray, J.
Am. Chem. Soc. 2011, 133, 998–1006.
[34] F. A. P. Hammersley in ESRF98 A01T: FIT2D V9.129 Reference Manual V3.1,
ESRF Internal Report, France, 1998.
[35] a) U. Sharma, P. Kumar, N. Kumar, V. Kumar, B. Singh, Adv. Synth. Catal.
2010, 352, 1834–1840; b) H. Yang, Y. Li, M. Jiang, J. Wang, H. Fu, Chem.
Eur. J. 2011, 17, 5652–5660; c) E. Vasilikogiannaki, C. Gryparis, V. Kotza-
basaki, I. N. Lykakis, M. Stratakis, Adv. Synth. Catal. 2013, 355, 907–911.
[18] X. Zhang, Y. Niu, X. Meng, Y. Li, J. Zhao, CrystEngComm 2013, 15, 8166–
8172.
[19] T. Brezesinski, M. Groenewolt, M. Antonietti, B. Smarsly, Angew. Chem.
Int. Ed. 2006, 45, 781–784; Angew. Chem. 2006, 118, 795–798.
[20] Y. El Mendili, J.-F. Bardeau, N. Randrianantoandro, F. Grasset, J.-M. Grene-
che, J. Phys. Chem. C 2012, 116, 23785–23792.
Received: November 20, 2015
[21] A. M. Jubb, H. C. Allen, ACS Appl. Mater. Interfaces 2010, 2, 2804–2812.
Published online on February 16, 2016
Chem. Eur. J. 2016, 22, 4600 – 4607
www.chemeurj.org
4607
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim