Mesoporous Au-TiO2 nanoparticle assemblies as efficient catalysts for the chemoselective reduction of nitro compounds

Article abstract of DOI:10.1039/c3ta13365f

In this article, we demonstrate novel mesoporous Au-loaded TiO₂ nanoparticle assemblies (Au-MTA) as high-effective catalysts for the selective transformation of nitroaromatics into the corresponding aryl amine products. These materials feature

Full text of DOI:10.1039/c3ta13365f

Journal of
Materials Chemistry A
View Article Online
View Journal | View Issue
PAPER
Mesoporous Au–TiO2 nanoparticle assemblies as
efficient catalysts for the chemoselective reduction of
nitro compounds†
Cite this: J. Mater. Chem. A, 2013, 1,
14311
Ioannis Tamiolakis,a Stella Fountoulaki,b Nikolaos Vordos,c Ioannis N. Lykakisb
a
*
and Gerasimos S. Armatas
In this article, we demonstrate novel mesoporous Au-loaded TiO2 nanoparticle assemblies (Au–MTA) as
high-effective catalysts for the selective transformation of nitroaromatics into the corresponding aryl
amine products. These materials feature a three-dimensional open porous structure consisting of
interconnected uniform gold and TiO2 nanoparticles, large internal surface area (ca. 104–120 m2 gꢀ1),
and narrow mesopores (ca. 7.3–7.6 nm). Au–MTA displays outstanding performance for the selective
reduction of nitro into amine groups using sodium borohydride as a reducing agent under ambient
conditions. We show that both chemoselectivity and hydrogenation activity of the Au–MTA catalysts are
highly related to the Au loading and particle size. As a result the 2% Au–MTA associated with 5 nm
sized Au particles was found to be a prominent catalyst for hydrogenation reactions, providing
exceptionally high selectivity (>96%) and conversion yield (>92%) to the corresponding amines. The
implication of these results is that mesoporous ensembles of gold and TiO2 nanoparticles should be
considered for the synthesis of fine amine chemicals.
Received 25th August 2013
Accepted 26th September 2013
DOI: 10.1039/c3ta13365f
www.rsc.org/MaterialsA
systems offering highly selective synthesis of aromatic amino-
compounds is an ongoing research topic.
Introduction
Aromatic amines are highly valuable chemical intermediates
widely used in the manufacture of pharmaceuticals, polymers,
dyes and cosmetics.1 Traditionally, amines can be synthesized by
transition metal-catalyzed hydrogenation of nitro compounds,
which emerged as one of the most important and, synchro-
nously, challenging tasks in synthetic organic chemistry.2
Although various supported metal catalysts such as Au/CeO2,3,4
Au/Al2O3,5 Pt/Al2O3,6 Au/Fe2O3 (ref. 7) and Au/C8 have been
effectively used to reduce nitroaromatic compounds, a successful
catalytic system with high associated product yield and selectivity
is yet to be achieved. A major challenge that persists in these
reactions is the unavoidable formation of harmful azo- and azoxy-
derivatives, which signicantly reduce the environmental impact
of the amine production.9 Therefore, exploring new catalytic
Gold nanoparticles supported on TiO2 (Au–TiO2) is a unique
catalytic system with exceptionally high activity in hydrogena-
tion or oxidation of alkene, carbonyl and nitro compounds.10,11
The reactivity of these catalysts is related to the nature of the
TiO2 support and the small size of Au particles.12,13 Recent
studies on TiO2-supported noble metal nanoparticles have
shown that gold, unlike Pt and Pd, is particularly effective for
the chemoselective reduction of nitroaromatics possessing
other competing functionalities such as olenic, carbonyl and
amide groups.7,14 Nonetheless, these gold-catalyzed reduction
ꢁ
reactions require certain conditions (100–140 C, 9–25 bar H2)
to proceed, although with moderate yield and selectivity. The
selective propensity of gold to convert from nitroaromatics to
amines was attributed to the preferential adsorption and acti-
vation of the –NO2 group on the gold atoms located at the
metal–support interface.15 In addition, apart from gold nano-
particles, the nature of the support and the chemical interac-
tions between the gold and titania surface were shown to have a
prominent effect on the –NO2 hydrogenation processes.16
Interestingly, TiO2 was shown to be effective in (de)hydrogena-
tion reaction of electron-rich aromatic compounds by facili-
tating both the azobenzene cleavage and aniline dissociation
through the formation of phenyl imide intermediates on under-
coordinated Ti surface sites.17 Recently, hydrogenation
catalysts based on supported bimetallic Au–Pd core–shell and
Pt–Au alloy nanoparticles have been reported.18 Although, these
aDepartment of Materials Science and Technology, University of Crete, Heraklion
71003, Crete, Greece. E-mail: garmatas@materials.uoc.gr; Fax: +30-2810-545197;
Tel: +30-2810-545004
bDepartment of Chemistry, Aristotle University of Thessaloniki, University Campus
54124, Thessaloniki, Greece
cDepartment of Electrical Engineering, Kavala Institute of Technology, 65404 Kavala,
Greece
† Electronic supplementary information (ESI) available: The calculation method
for density of gold edge sites, catalytic results of various Au supported catalysts,
UV-vis spectra of mesoporous MTA and Au–MTA materials and catalytic data,
EDS spectra and N2 adsorption–desorption isotherms of the recycled 2%
Au–MTA sample. See DOI: 10.1039/c3ta13365f
This journal is ª The Royal Society of Chemistry 2013
J. Mater. Chem. A, 2013, 1, 14311–14319 | 14311

View Article Online
Journal of Materials Chemistry A
Paper
nanocomposite materials exhibit a higher hydrogenation rate (Germany). Aldrich anatase was supplied by Sigma-Aldrich
than Au–TiO2 while maintaining the chemoselectivity of gold, (Product no. 637254). The aromatic nitro compounds used as
the selectivity towards aniline is feasible only at very high substrates were of high purity and commercially available from
pressures (>10 atm) of H2, which actually impedes their prac- Aldrich.
tical use.
Recently, we have reported a facile surfactant-assisted Synthesis of the MTA
aggregating method for obtaining mesoporous TiO2 nano-
particle assemblies (MTA) with tunable pore size.19 The pore
The mesoporous TiO2 nanoparticle assemblies (MTA) were
synthesized as follows:19 1 g of Brij 58 (HO(CH2CH2O)20C16H33
)
walls of these materials consist of interconnected anatase TiO2
nanoparticles and exhibit superior catalytic performance for
photooxidation of benzyl alcohols to the conventionally used
TiO2 catalysts, i.e. Degussa P25, Hombikat UV-100 and Aldrich
anatase nanoparticles. Importantly, the enhancement of cata-
lytic activity of MTA is a result of a combination of the accessible
open-pore structure, large surface area and high crystallinity of
TiO2 nanobuilding blocks. These results prompted us to exploit
the pore surface of MTA as a unique host matrix for noble metal
nanoparticles and to develop a new type of mesoporous Au-
loaded TiO2 nanocomposite catalysts. Porous assemblies of this
form can combine the merits of both nanoparticle properties
and mesoporous structures. As the gold and TiO2 nanoparticles
can induce distinctive electronic and chemical properties to the
former, mesoporosity can provide a high concentration of the
surface active sites and promote fast diffusion of reactants and
products during the catalytic reactions.
In the present study, we demonstrate for the rst time novel
mesoporous Au-loaded TiO2 nanoparticle assemblies as high-
effective catalysts for the selective transformation of nitro-
aromatics into the corresponding amines, using sodium
borohydride as the reducing agent. In particular, we utilized
deposition–precipitation of gold nanoparticles on the surface of
MTA mesostructures to prepare a series of mesoporous Au–TiO2
nanocomposite materials with controllable gold particle size,
i.e. ranging from 3 to 10 nm. The obtained materials (Au–MTA)
possess a continuous network of interconnected gold and TiO2
(ꢂ8 nm particle size) nanoparticles and exhibit a large internal
surface area and narrow pore size distribution. More impor-
tantly, we addressed the effect of supported gold particles on
the chemoselective reduction of nitro compounds, showing that
the yields and product composition are strongly related to the
Au loading and particle size. As a result the 2% Au–MTA asso-
ciated with 5 nm sized Au particles is found to be a prominent
catalyst for hydrogenation reactions, providing exceptionally
high selectivity (>96%) and conversion yield (>92%) to the
corresponding amines under ambient conditions. To our
knowledge, this Au-based catalytic system is the most efficient
for the selective reduction of nitro compounds that has been
reported to date.
surfactant was dissolved in 10 mL of anhydrous ethanol with
stirring. Then, 0.30 g of titanium tetrachloride (TiCl4) and 1.60 g
of titanium(IV) propoxide (TPP) were added dropwise to the
surfactant solution with continuous stirring. The mixture was
kept at room temperature for 2 h and then transferred to a Petri-
ꢁ
dish and stored in an oven at 40 C for 7 days, yielding a pale
yellow-colored wet gel. Removal of surfactant between the TiO2
nanoparticles as well as crystallization and interparticle
connection was achieved by drying the gel product at 100 ꢁC for
12 h under vacuum and then heating to 260 ꢁC for 6 h and,
subsequently, to 350 ꢁC for 4 h under an atmosphere of air, at a
0.5 C minꢀ1 heating rate.
ꢁ
Synthesis of the Au–MTA catalysts
The supported gold catalysts were synthesized by the deposi-
tion–precipitation method over the MTA sample. Typically, an
aqueous solution of Au(en)3Cl3 (5 mL) corresponding to the
desired gold loading was adjusted to pH 10 with a 1 M NaOH
solution. Subsequently, 100 mg of MTA was dispersed into the
gold solution with rigorous stirring for 1 h and the resulting
powder was recovered by ltration, washed with water and dried
at 40 ꢁC under vacuum. Reduction of gold species on the surface
of MTA was carried out by suspending 100 mg of Au–MTA solid
in 1 mL of ethanol. To this suspension, 0.5 mL of 0.5 M NaBH4
ethanolic solution was added dropwise and the resulting
mixture was kept under stirring for 30 min. The purple-colored
solid was then recovered by ltration, washed with water and
ꢁ
ethanol, and dried in a vacuum oven at 80 C. A series of x%
Au–MTA catalysts with various Au loadings (x ¼ 0.5, 1, 2, 3 and
5 wt%) was prepared by varying the Au amount in the reaction
mixture. Note that this synthetic method is simple, highly
reproducible and, therefore, suitable for large-scale synthesis of
Au–MTA catalysts; it provides large quantities of Au–MTA
(greater than 1 g) with consistent reactivity. For comparative
studies, the Au–TiO2 (P25) and Au–TiO2 (anatase) materials
were prepared in a similar procedure to 2% Au–MTA, but
depositing 2 wt% gold on commercially available Degussa P25
and Aldrich anatase TiO2 nanoparticles, respectively.
Physical characterization
Small-angle X-ray scattering (SAXS) measurements were per-
formed on a JJ X-ray system (Denmark) equipped with a Rigaku
Method
Materials
˚
Helium-3 detector and a Cu (l ¼ 1.54098 A) rotating anode
Brij 58 surfactant (HO(CH2CH2O)20C16H33, Mn ꢂ 1124), tita- operated at 40 kV and 40 mA. The sample-to-detector distance
nium tetrachloride (99.9%), anatase TiO2 (99.8%) and absolute and the center of beam were precisely determined by calibration
˚
ethanol (99.8%) were purchased from Sigma-Aldrich. Titaniu- with the Ag–behenate (d001 ¼ 58.38 A) standard. Two-dimen-
m(IV) propoxide ($98%) was purchased from Merck. TiO2 sional (2D) diffraction images were integrated into a one-
nanoparticles (P25) were purchased from Degussa AG dimensional (1D) diffraction pattern, as a function of q
14312 | J. Mater. Chem. A, 2013, 1, 14311–14319
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Paper
Journal of Materials Chemistry A
wavevector, with the SAXSGUI soware (Rigaku Innovative mixture was stirred at room temperature for a selected time.
Technologies, Inc.). Scattering data were corrected for dark The reaction was monitored by thin layer chromatography
current and empty tube scattering. The average size of spher- (TLC), and aer completion, the slurry was ltered under
ical-shaped TiO2 particles was determined from the scattering pressure through a short pad of celite and silica gel to withhold
data, using the Guinier approximation: I(q) ¼ f exp(ꢀq2Rg /3) the catalyst with the aid of ethanol or methanol (ꢂ5 mL). The
2
where I is the scattering intensity and Rg is the radius of gyra- ltrate was evaporated under vacuum to afford the corre-
tion, which is related to the diameter of spherical particles by sponding products in pure form. Product analysis was con-
the equation D ¼ 2RgO(5/3).20 The X-ray diffraction (XRD) ducted by 1H NMR and 13C NMR spectroscopy (Brucker
patterns were recorded on a PANanalytical X'Pert Pro MPD X-ray 500 MHz) and the identication of the products was carried out
diffractometer (45 kV and 40 mA) using Cu Ka radiation (l ¼ by comparing the NMR spectra data with commercially avail-
˚
1.5418 A) in Bragg–Brentano geometry. Nitrogen adsorption able pure substances (Sigma-Aldrich Company).
and desorption isotherms were measured at ꢀ196 ꢁC using a
Reusability testing of the catalyst was conducted in the case
NOVA 3200e volumetric analyzer (Quantachrome, USA). Before of the 2% Au–MTA sample through the hydrogenation of
analysis, the samples were degassed at 150 ꢁC under vacuum p-nitrotoluene. A 1 mL mixture of the feeding solution
(<10ꢀ5 Torr) for 12 h to remove moisture. The specic surface (0.1 mmol of p-nitrotoluene and 0.6 mmol of NaBH4) together
area was calculated using the Brunauer–Emmett–Teller (BET) with an appropriate amount of catalyst (2 mol%) was placed
method21 on the adsorption data in the 0.05–0.2 relative pres- into a vial. Each catalytic cycle was stopped aer 2 h and the
sure (P/Po) region. The total pore volumes were derived from the catalyst was collected by ltration, washed with ethanol and
adsorbed volume at P/Po ¼ 0.95 and the pore size distributions dried in an oven at 100 ꢁC for 12 h. The catalyst was then used
were calculated by the nonlocal density functional (NLDFT) without any further processing for the next catalytic run.
method22 based on the adsorption data. Transmission electron
microscopy (TEM) analysis was conducted using a JEM-2100
Results and discussion
Morphology and structural properties
electron microscope (JEOL, Japan) equipped with an energy
dispersive X-ray (EDS) detector (Oxford Instruments) and oper-
ated at an accelerating voltage of 200 kV. Samples were gently
Mesoporous titania with large surface area and narrow pore size
ground, dispersed in ethanol by sonication, and then posi-
distribution was synthesized via a so-templating assembly of
tioned on a carbon-coated Cu grid. Elemental microprobe
analysis was conducted on a JSM-6390LV (JEOL) scanning
electron microscope (SEM) equipped with an Oxford INCA
PantaFET–x3 energy dispersive X-ray spectroscopy (EDS)
detector. Data acquisition was performed at an accelerating
voltage of 20 kV and 60 s accumulation time. Diffuse reectance
UV-vis spectra were obtained on a Perkin Elmer Lambda 950
optical spectrophotometer using BaSO4 as the reectance
TiO2 nanoparticles. Namely, we used evaporation induced
aggregating assembly of titanium compounds, i.e. Ti(OPr)4 and
TiCl4, and surfactant (Brij-58) molecules to build three-dimen-
sional TiO2–surfactant nanocomposite structures. To effectively
remove the surfactant inside the pores, we performed careful
thermal treatment at elevated temperature, so that a continuous
mesoporous network of tightly interconnected TiO2 nano-
particles (MTA) was obtained. Deposition of gold nanoparticles
was achieved by impregnating into the MTA support an etha-
nolic solution of Au(en)3Cl3 under the basic environment, fol-
lowed by reduction with NaBH4. Chemical analysis by energy
dispersive X-ray spectroscopy (EDS) on the Au–MTA samples
standard. Reectance data were converted to absorption (a/S)
using the Kubelka–Munk equation: a/S ¼ (1 ꢀ R)2/(2R), where R
is the reectance and a and S are the absorption and scattering
coefficients, respectively.
indicated compositional data that correspond to the expected
Au loadings by the stoichiometry of reactions (see Table 1). The
Catalytic reactions
Supported gold catalyst Au–MTA (20 mg) was placed in a 5 mL EDS spectra did not show any signal from chloride element,
glass reactor, followed by the addition of ethanol (1 mL), nitro suggesting that such Au(III) chloride species are not incorpo-
compound (0.1 mmol) and NaBH4 (0.6 mmol), and the reaction rated into the products.
Table 1 Textural properties of mesoporous MTA and Au–MTA materials
Au loadinga
(wt%)
dAu
dp
Surface area
Pore volumed
Pore size
(nm)
b
c
Sample
(nm)
(nm)
(m2 gꢀ1
)
(cm3 gꢀ1
)
MTA
—
—
16.7
18.0
18.0
18.1
18.2
18.2
159
120
115
110
106
104
0.28
0.24
0.23
0.23
0.22
0.21
7.6
7.6
7.5
7.4
7.4
7.3
0.5% Au–MTA
1% Au–MTA
2% Au–MTA
3% Au–MTA
5% Au–MTA
0.49
1.07
2.06
2.96
5.04
3.2
4.4
5.2
7.1
9.4
a
Weight percentage of Au loading according to the EDS data. b Average Au particle size (dAu) on the basis of TEM analysis. c Average center-to-center
interparticle distance (dp) calculated by SAXS data. d Cumulative pore volume at a relative pressure (P/P0) of 0.95.
This journal is ª The Royal Society of Chemistry 2013
J. Mater. Chem. A, 2013, 1, 14311–14319 | 14313

View Article Online
Journal of Materials Chemistry A
Paper
corroborates to the anatase phase of TiO2 (JCPDS reference card
no. 21-1272) with an average domain size of crystallites, as
estimated from the peak width of the primary (101) reection
using the Debye–Scherrer equation, of about 9 nm. Meanwhile,
clear evidence of the presence of metallic Au comes from the
X-ray diffraction analysis on Au–MTA samples containing a high
content of gold. In Fig. 1b, the XRD patterns for 3–5% Au-loaded
samples show two weak features along the reections of the
anatase structure at ꢂ44.5ꢁ (200), ꢂ64.7ꢁ (220) and ꢂ77.6ꢁ (311),
indicating that the Au precursor is eventually reduced to yield
metallic gold (JCPDS reference card no. 04-0784) aer treatment
with NaBH4. The absence of diffraction peaks due to the
metallic gold in samples with Au loading less than 2% is
attributed to the low amounts present and the very small grain
size of the gold particles.
Fig. 2a and S1 of ESI† display typical transmission electron
microscopy (TEM) images of the Au–MTA materials. As seen in
these images, extended, sponge-like, porous structures of
assembled nanoparticles were formed. By means of this tech-
nique, the average size of TiO2 nanoparticles was estimated to
be ca. 8 nm. The very similar dimension obtained from the TEM
grain size and XRD crystallite size indicates minimal grain
coarsening of nanoparticles during the thermal process, which
attests to the single-crystal nature of TiO2 particles. The prop-
erty of high crystallinity of TiO2 has important implications for
the catalytic reactions such as redox catalysis and photo-
catalysis. In addition, clear evidence for the crystal structure of
titania nanoparticles was obtained from high-resolution TEM
(HRTEM) and selected area electron diffraction (SAED) analysis.
Representative HRTEM images taken along various local areas
of 2% Au–MTA conrm that the structure of TiO2 is indeed
Fig. 1 (a) Small-angle X-ray scattering and (b) X-ray diffraction patterns of
mesoporous MTA and Au–MTA materials. The inset of panel a: plot of the
Guinier's law (ln(I(q)) ¼ ln(I0) ꢀ Rg2q2/3, where Rg is the radius of gyration) for the
mesoporous MTA sample. The red line is fitting the data.
Results of small angle X-ray diffraction (SAXS) experiments anatase with high crystallinity, consistent with XRD results.
on parent and Au-loaded MTA samples are shown in Fig. 1a. The Close inspection of particles in Fig. 2a reveals the lattice fringes
SAXS patterns are indicative of some degree of long-range order, of TiO2 nanocrystals coincident with the (101) planes of the
displaying a strong scattering peak in the wavevector (q ¼ anatase phase. Meanwhile, the electron diffraction pattern of an
4p sin q/l, where 2q is the scattering angle and l is the wave- ensemble of the mesoporous 2% Au–MTA, in Fig. 2b, corrobo-
length of Cu irradiation) range of 0.35–0.38 nmꢀ1. The angular rates the polycrystalline structure of the sample, showing a
position of these reections coincided with an average center-
to-center interparticle distance (dp ¼ 2p/q) of about 16.7 nm for
MTA and 18.0–18.2 nm for Au–MTA materials (Table 1). These
distances indicate a small expansion in the structure of Au–MTA
as compared with the MTA possibly due to the Au deposition
step. As for the large suppression of the scattering peak with
increasing the gold loading, it could be associated with the
reduction of the electron density between the pores and wall
framework due to the pore lling by the Au nanoparticles.
Assuming that TiO2 nanoparticles are spherical in shape, the
average size of titania particles in MTA can be obtained using
the Guinier approximation23 in the low q region of the scattering
plot. Analysis of the scattering data by the Guinier equation
gives an average particle size of about 10 nm (inset in Fig. 1a),
which is consistent with the TiO2 particle diameter obtained Fig.
from XRD and TEM data (see below).
The wide-angle powder X-ray diffraction (XRD) patterns of
the mesoporous Au–MTA suggested nanocrystalline TiO2
judging from the broad reection peaks in the 2q ¼ 30–70ꢁ
2
(a) Representative transmission electron microscopy (TEM) images,
showing porous ensembles of uniform gold and TiO2 nanoparticles (upper inset:
high-resolution TEM of a single TiO2 nanoparticle), (b) the electron diffraction
pattern indexed to the anatase phase and (c) EDS spectra collected from a thin
area (ꢂ10 nm in width) of the framework for the mesoporous 2% Au–MTA
sample (the copper peaks result from the TEM Cu grid). The red cycles, in panel a,
range (Fig. 1b). The peak position of these reections indicate the areas (I and II) where the EDS analysis data were acquired.
14314 | J. Mater. Chem. A, 2013, 1, 14311–14319
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Paper
Journal of Materials Chemistry A
series of Debye–Scherrer concentric rings that reasonably 535–545 nm range, which is interpreted by the excitation of the
assigned to the anatase phase. surface plasmon resonance (SPR) in nanosized Au particles.
The TEM images also revealed that individual Au nano- There is a slight, yet perceptible, shi in the peak maximum
particles (appeared as dark spots as evidenced by TEM-EDS towards long wavelengths with increasing Au loading. A similar
spectroscopy, see Fig. 2c) are uniformly dispersed and studded shi of the SPR signal has been previously observed for gold
on the surface of TiO2. On the basis of TEM analysis, we supported on TiO2 nanoparticles and was largely ascribed to the
determined the gold particle-size distribution in Au–MTA decrease in Au dispersion.24 Although, a red shi of the plas-
samples and the results can be seen in Fig. 3. The size histo- mon resonance of Au nanoparticles due to their electromag-
grams (based on a count of more than 200 individual nano- netic coupling cannot be ruled out. The sharp optical
particles) demonstrate a strong dependence of gold loading on absorption onset in the energy range 380–390 nm (ꢂ3.2 eV) is
the Au particle size, in which as the metal content increases, the associated with the interband electron transitions in TiO2.
size distribution becomes wider and shis to the larger particle
The mesoporosity of the title materials was determined by
sizes. Namely, the 0.5–2% Au-loaded samples are characterized nitrogen physisorption measurements. Fig. 4a shows the N2
by a quite narrow size distribution of gold particles with an adsorption–desorption isotherms of mesoporous MTA and Au–
average diameter being in the range of 3.2 to 4.4 to 5.2 nm. On MTA samples. All the isotherms feature typical type-IV curves,
increasing further the Au loading (up to 5%), the Au–MTA show according to the IUPAC classication, with a sharp capillary
an incremental growth in the average size of the gold particles condensation step at a relative pressure (P/Po) of 0.7, being
from 5.2 nm to 7.1 nm to 9.4 nm. Notably, the Au particle size characteristic of porous solids with narrow-sized mesopores.
distribution of 5% Au–MTA shows a broad feature at 17.1 nm, The pronounced H2-type hysteresis loop between the adsorp-
which suggests the presence of large gold particles possibly tion and desorption branch is associated with the presence of
located at the surface of the titania aggregates. It can be an interconnected pore system between the nanoparticles.25
observed that the line prole of particle-size distributions is Compared with the parent MTA, the Au–MTA materials show
strongly related to the Au content and it tend to form a log-
normal shape when the Au metal loading is decreased below
1%. However, this behavior might point to the unsuccessful
identication of very small Au particles (less than 1 nm in size)
due to their insufficient contrast with the titania support.
The formation of gold nanoparticles in Au–MTA materials
was also veried by ultraviolet-visible light (UV-vis) diffuse
reectance spectroscopy. The UV-vis spectra of Au–MTA, in
Fig. S2 of ESI,† show an intense optical absorption in the
Fig. 4 (a) Nitrogen adsorption–desorption isotherms at ꢀ196 ꢁC and (b) NLDFT
pore size distributions calculated from adsorption branches for mesoporous MTA
Fig. 3 Histograms of the size distributions of gold particles for mesoporous
Au–MTA materials.
and Au–MTA materials. The isotherms of Au–MTA are offset along the y-axis for
clarity.
This journal is ª The Royal Society of Chemistry 2013
J. Mater. Chem. A, 2013, 1, 14311–14319 | 14315

View Article Online
Journal of Materials Chemistry A
Paper
slightly lower Brunauer–Emmett–Teller (BET) surface areas and and we evaluated the catalytic performance of these materials
pore volumes, as expected due to the incorporation of Au under similar reaction conditions. The evolution of the hydro-
nanoparticles. The Au–MTA were found to have a surface area in genation products indicated that Au–MTA, although afforded
the range of 104–120 m2 gꢀ1 and a total pore volume in the excellent conversion (100%) of p-nitrotoluene, possesses
range of 0.21–0.24 cm3 gꢀ1. The mesoporous MTA exhibited a moderate-to-high yield to the corresponding amine (1a), which
surface area of 159 m2 gꢀ1 with a pore volume of 0.28 cm3 gꢀ1
.
is highly related to the Au particle size. As seen in Fig. 5a, the
These results, in combination with TEM and SAXS data, reveal Au–MTA catalysts associated with Au particles smaller than
that gold nanoparticles are entirely grown within the pores of 4 nm or larger than 7 nm provide a moderate yield of the cor-
the MTA and, more importantly, without blocking the access to responding amine (1a) in 30 min, with a noticeable amount of
the internal surface.
1c and 1d byproducts. In contrast, the 2% Au–MTA catalyst
The pore width in Au–MTA materials was determined by associated with Au nanoparticles of about 5 nm size shows a
using the non-local density function theory (NLDFT) on remarkable increase in hydrogenation activity, giving p-
adsorption data, and was found to be ca. 7.3–7.6 nm with
narrow size distribution (Fig. 4b). Interestingly, the MTA sample
shows a pore size distribution with a main peak at around
7.6 nm associated with a broad shoulder at ca. 9–10 nm. The
pore width of 7.6 nm corresponds to the interparticle spaces
templated by the surfactant micelles, whereas the latter (ca.
9–10 nm) is attributed to the large voids between the partially
interconnected TiO2 nanoparticles. Since the main mesopore
diameter of the MTA does not decrease remarkably with the
incorporation of Au, we assume that the gold nanoparticles are
mainly entrapped within the large interstitial cavities (9–10 nm)
of the TiO2 nanoparticle assemblies. Based on the NLDFT pore
size (7.6 nm) and the SAXS center-to-center interparticle
distance (16.7 nm) of the MTA sample, the average thickness of
pore walls was estimated to be about 9.1 nm, which is very close
to the TiO2 particle diameter. This result is entirely consistent
with the mesoporous walls comprising a single layer of inter-
connected TiO2 nanoparticles. All the textural properties of the
mesoporous Au–MTA are given in Table 1.
Catalytic hydrogenation reactions
To investigate the catalytic performance of Au–MTA, the
reduction of p-nitrotoluene (1) using sodium borohydride as the
hydrogen source was initially studied. In general, reduction of
p-nitrotoluene to the corresponding p-toluidine (1a) also
involves the formation of several intermediates, such as 1,2-di-
p-tolyldiazene oxide (azoxy-, 1b) and 1,2-di-p-tolyldiazene (azo-,
1c) (Scheme 1) (which are formed via the condensation reaction
between the initial intermediates hydroxylamine and nitroso
compounds), whose quantitative evolution could be useful for
exploring the chemoselective activity of the gold catalysts. For
comparison purpose, we also prepared Au/TiO2 (P25) and
Au/TiO2 (anatase) catalysts by precipitating 2% gold on Degussa
P25 (ca. 26 nm particle size, 53 m2 gꢀ1) and Aldrich anatase
(<25 nm particle size, 55 m2 gꢀ1) nanoparticles, respectively,
Fig. 5 Product evolution of p-nitrotoluene reduction after (a) 30 min and (b) 1 h
for supported gold catalysts: (1) 0.5% Au–MTA, (2) 1% Au–MTA, (3) 2% Au–MTA,
Scheme 1 Reduction of p-nitrotoluene (1) over a mesoporous 2% Au–MTA
(4) 3% Au–MTA, (5) 5% Au–MTA, (6) 2% Au/TiO2 (P25) and (7) 2% Au/TiO2
catalyst.
(anatase).
14316 | J. Mater. Chem. A, 2013, 1, 14311–14319
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Paper
Journal of Materials Chemistry A
toluidine in 92% relative yield at quantitative conversion levels amine (1a), even aer 12 h reaction time. According to this
(100%), whereas the residual product observed was azobenzene. reaction scheme, the nitro compounds were rapidly converted
Of particular note, p-nitrotoluene was completely reduced by to nitroso and hydroxylamine intermediates over gold nano-
the 2% Au–MTA within 1 h, yielding p-toluidine as the only particles, which in turn undergo a direct rate-determining
product (see Fig. 5b); no detectable amount of 1b and 1c reduction step to form the corresponding amines.29
compounds was observed in the reaction mixture based on
The superior activity of the 2% Au–MTA for the chemo-
NMR spectroscopy. Therefore, within the 5–7 nm range of gold selective reduction of p-nitrotoluene can be further seen with
particles, there is a signicant enhancement in hydrogenation respect to the product evolution of commercial supported gold
activity for Au–MTA mesostructures, whereas their catalytic nanoparticles. Results from Fig. 5a show that 2% Au/TiO2 (P25)
activity is remarkably reduced with decreasing the Au particle gives only a signicant amount of azo- and azoxy-compounds as
size from ca. 5 to 3 nm. These ndings are in contrast to the reaction products in 30 min, whereas no p-toluidine was
previous studies indicating that gold particles in the 2–3 nm detected in the reaction mixture. For the 2% Au/TiO2 (anatase)
range are more effective for the –NO2 reduction.26,27 Control the conversion yield of p-toluidine was found to be notably
experiments showed no obvious transformation of p-nitro- poorer (23%), although >99% conversion of p-nitrotoluene is
toluene in the absence of catalyst or in the presence of Au–free observed. It should be stressed that the 2% Au/TiO2 (P25)
TiO2 (MTA), conrming that Au nanoparticles are essential for catalyst achieve complete conversion of p-nitrotoluene, with
hydrogenation reactions.
more than 91% selectivity to p-toluidine, but in a prolonged
The chemoselective response of Au–MTA catalysts to amines reaction time (1 h), see Fig. 5b. As compared to the 2% Au/TiO2
can be tentatively associated with the available surface atoms of (P25) and 2% Au/TiO2 (anatase) catalysts, the higher reactivity of
gold nanoparticles at the interface with the titania support. It is Au–MTA should be attributed to the three-dimensional open-
acknowledged that gold nanoparticles have a strong inuence pore structure and large surface area, which allows facile mass
on the chemoselective reduction of nitroaromatics by control- transport throughout the material and adequate accessibility of
ling the accumulation rate of the active intermediates (i.e. reactants to the active sites.
nitroso and hydroxylamine compounds).11,14 To rule out this
Reduction of p-substituted aromatic nitro compounds (1–5)
possibility, we determined the density of edge sites of gold also proceeds effectively with 2% Au–MTA catalysts to the cor-
particles (Au atoms per nm2) for Au–MTA on the basis of the responding amines (1a–5a). Results, in Table 3, show that a
average Au particle size, the loading amount of metallic Au and diverse set of nitro compounds can be converted to the corre-
the specic surface area of the MTA support. More details can sponding amines at high yields and selectivity, regardless of the
be found in the ESI.† These calculations revealed density of electron-donating or electron-withdrawing ability of the
active sites at the metal–support boundaries, and the values substituent group. Specically, when nitrobenzene was subject
were consistent with the specic production rates of 1a amine to hydrogenation it yielded 98% of aniline in 1 h (Table 3, entry
(see Table 2). This suggests that the edge sites of gold nano- 1). Nitroarenes possessing electron-donating groups such as p-
particles have a profound inuence on the hydrogenation nitrotoluene and p-nitrophenol (Table 3, entries 2 and 3) were
activity of Au–MTA. In terms of the activation mechanism, we also reduced to the desired amines in high yield (>94%). Similar
can rationalize that reduction of nitro into amine compounds reduction of 1-chloro-4-nitrobenzene and methyl-4-nitro-
could mainly proceed through a direct hydrogenation route, benzoate compounds containing the electron-acceptor group
alternative to the competitive condensation reaction between (Table 3, entry 4 and 5) gives above 92% of the corresponding
nitroso and hydroxylamine to form an azoxy compound.28 This
behavior is consistent with the experimental observation that a
very small amount of condensation products (i.e. azo- and
azoxy-compounds) were detected among the reaction mixture.
In addition, we found that azoxy- (1b) and azo- (1c) arenes were
relatively difficult to hydrogenate over the 2% Au–MTA catalyst,
as reected in the low yield (ca. 10%) of the corresponding
Table 3 Chemoselective reduction of nitro compounds into amines with NaBH4
catalyzed by mesoporous 2% Au–MTAa
NaBH4 Time (h)/
Entry Reactant
Product
(equiv.) yieldb (%)
1
2
6
6
1/98
Table 2 Density of edge site of Au particles and hydrogenation rate for amine
production over mesoporous Au–MTA catalysts
1.5/99
Density of edge sites
Rate of aminea
3
4
6
6
2/94
2/92
ꢀ1
ꢀ1
Sample
(Au sites 100 gcatal nmꢀ1
)
(mol gcatal hꢀ1
)
0.5% Au–MTA
1% Au–MTA
2% Au–MTA
3% Au–MTA
5% Au–MTA
0.282
0.304
0.440
0.346
0.337
2.8 ꢃ 10ꢀ3
4.1 ꢃ 10ꢀ3
9.2 ꢃ 10ꢀ3
7.8 ꢃ 10ꢀ3
5.8 ꢃ 10ꢀ3
5
6
2/93
a
a
Dened as the number of moles of p-toluidine (1a) produced per gram
In all cases the product selectivity (moles of amine per moles of all
of catalyst per hour.
obtained products) was >96%. b Isolated yield (conversion ꢃ selectivity).
This journal is ª The Royal Society of Chemistry 2013
J. Mater. Chem. A, 2013, 1, 14311–14319 | 14317

View Article Online
Journal of Materials Chemistry A
Paper
amines within 2 h. Notably, we have not observed any secondary
reaction affecting the carboxylic or chloro (dehalogenation)
functionalities under the selected reaction conditions, sug-
gesting the high chemoselectivity of the present Au–MTA/
NaBH4 reduction system. The 2% Au–MTA also demonstrated
higher hydrogenation activity and chemoselectivity for the
reduction of aromatic nitro compounds than does other high
active supported Au catalysts. Table S1 in the ESI† shows the
reaction conditions and reported yield of the respective amines
for various supported Au catalysts (including 2% Au–MTA)
from the literature. The 2% Au–MTA outperforms catalysts like
Au/MgO nanoparticles,30 Au/PNIPA yolk–shell nanoparticles,31
Au-poly-vinylpyrrolidone,32 Au/PMMA beads33 and Au/PAA/PAH
lms.34 These results clearly support the superior catalytic
performance and general applicability of the Au–MTA catalytic
system in hydrogenation of aromatic nitro compounds.
Because of the extended network of interconnected gold
and TiO2 nanoparticles, the Au–MTA catalyst can be easily
separated from the reaction mixture by ltration and could be
reused for the next catalytic reaction. The reusability of the
2% Au–MTA was examined using p-nitrotoluene as a model
substrate. The results, in Fig. S3 of the ESI,† show that the
conversion yield of p-toluidine remains as high as 99% even
aer four successive catalytic runs, reecting very good
reusability. Elemental X-ray microprobe analysis and N2
Acknowledgements
This work was supported by the Greek Ministry of Education
and the European Union, under ERC Starting Grant (MESO-
POROUS-NPs, MIS 374071).
Notes and references
1 P. F. Vogt and J. J. Gerulis, Ullmann's Encyclopedia of
Industrial Chemistry, in Aromatic Amines, Wiley-VCH Verlag
GmbH & Co., Weinheim, 2005.
2 H. U. Blaser, H. Steiner and M. Studer, ChemCatChem, 2009,
1, 210.
3 X. B. Ke, X. G. Zhang, J. Zhao, S. Sarina, J. Barry and
H. Y. Zhu, Green Chem., 2013, 15, 236.
4 M. Makosch, J. Sa, C. Kartusch, G. Richner, J. A. van
Bokhoven and K. Hungerbuhler, ChemCatChem, 2012, 4, 59.
5 X. Huang, X. P. Liao and B. Shi, Green Chem., 2011, 13,
2801.
6 P. Serna, M. Boronat and A. Corma, Top. Catal., 2011, 54,
439.
7 A. Corma and P. Serna, Science, 2006, 313, 332.
8 Y. Choi, H. S. Bae, E. Seo, S. Jang, K. H. Park and B. S. Kim,
J. Mater. Chem., 2011, 21, 15431.
9 R. A. Sheldon, Green Chem., 2007, 9, 1273.
physisorption measurements showed no change in the 10 C. Gonzalez-Arellano, A. Corma, M. Iglesias and F. Sanchez,
chemical composition and the porous structure of the four-
Chem. Commun., 2005, 3451.
times reused 2% Au–MTA sample as compared to fresh 2% 11 M. Stratakis and H. Garcia, Chem. Rev., 2012, 112, 4469;
Au–MTA (see ESI, Fig. S4 and S5†), thereby, indicating high
durability and recyclability.
Y. Zhang, X. J. Cui, F. Shi and Y. Q. Deng, Chem. Rev.,
2012, 112, 2467.
12 V. Kratky, M. Kralik, M. Mecarova, M. Stolcova, L. Zalibera
and M. Hronec, Appl. Catal., A, 2002, 235, 225.
13 X. H. Yan, J. Q. Sun, Y. W. Wang and J. F. Yang, J. Mol. Catal.
A: Chem., 2006, 252, 17.
Conclusions
In conclusion, high-surface-area mesoporous TiO2–supported 14 M. Murdoch, G. I. N. Waterhouse, M. A. Nadeem,
gold nanoparticles with tunable Au particle size (i.e. ranging
J. B. Metson, M. A. Keane, R. F. Howe, J. Llorca and
from 3.2 to 9.4 nm) have been successfully prepared. The
H. Idriss, Nat. Chem., 2011, 3, 489.
obtained Au–MTA materials are constructed from narrow- 15 M. Boronat, P. Concepcion, A. Corma, S. Gonzalez, F. Illas
sized gold and TiO2 nanoparticles and possess a three- and P. Serna, J. Am. Chem. Soc., 2007, 129, 16230.
dimensional open-pore structure with high internal surface 16 F. Moreau, G. C. Bond and A. O. Taylor, J. Catal., 2005, 231,
area (104–120 m2 gꢀ1) and uniform mesopores (ca. 7.3–
105.
7.6 nm). Nitro compounds hydrogenation results suggest that 17 S. C. Li and U. Diebold, J. Am. Chem. Soc., 2010, 132, 64.
Au–MTA can have important implication in synthesis of 18 D. P. He, X. D. Jiao, P. Jiang, J. Wang and B. Q. Xu, Green
aromatic amines. We showed that gold nanoparticles, espe-
cially the edge sites, play an important role in the present 19 I. Tamiolakis, I. N. Lykakis, A. P. Katsoulidis and
catalytic system by controlling the hydrogenation rate of G. S. Armatas, Chem. Commun., 2012, 48, 6687.
intermediates and, thus, the chemoselective conversion of 20 P. W. Schmidt, in Modern Aspects of Small-Angle Scattering,
Chem., 2012, 14, 111.
nitro into aromatic amine compounds. As a result the 2% Au–
MTA catalyst associated with a Au particle size of 5 nm not
Kluwer Academic Publishers, Dordrechr, The Netherlands,
1995.
only gives the highest hydrogenation activity and selectivity 21 S. Brunauer, L. S. Deming, W. E. Deming and E. Teller, J. Am.
(100% conversion, >96% selectivity) among the examined Chem. Soc., 1940, 62, 1723.
materials but also allows the efficient synthesis of aromatic 22 P. I. Ravikovitch, D. Wei, W. T. Chueh, G. L. Haller and
amino compounds under ambient conditions. The use of A. V. Neimark, J. Phys. Chem. B, 1997, 101, 3671.
mesoporous ensembles from gold and TiO2 nanoparticles 23 O. Glatter and O. Kratky, in Small-Angle X-ray Scattering,
could provide a signicant development in the industrial Academic Press, New York, 1982.
production of aromatic amines from nitro compounds in 24 B. R. Panda and A. Chattopadhyay, J. Nanosci. Nanotechnol.,
excellent yields and chemoselectivities.
2007, 7, 1911.
14318 | J. Mater. Chem. A, 2013, 1, 14311–14319
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Paper
Journal of Materials Chemistry A
25 F. Rouquerol, J. Rouquerol and K. S. W. Sing, in Adsorption by 30 K. Layek, M. L. Kantam, M. Shirai, D. Nishio-Hamane,
Powders and Porous Solids. Principles Methodology and
Applications, Academic Press, New York, 1999.
26 F. Cardenas-Lizana, S. Gomez-Quero, H. Idriss and
M. A. Keane, J. Catal., 2009, 268, 223.
T. Sasaki and H. Maheswaran, Green Chem., 2012, 14, 3164.
31 S. Wu, J. Dzubiella, J. Kaiser, M. Drechsler, X. H. Guo,
M. Ballauff and Y. Lu, Angew. Chem., Int. Ed., 2012, 51,
2229.
27 R. Zanella, C. Louis, S. Giorgio and R. Touroude, J. Catal., 32 N. Yan, J. G. Zhang, Y. Yuan, G. T. Chen, P. J. Dyson, Z. C. Li
2004, 223, 328. and Y. Kou, Chem. Commun., 2010, 46, 1631.
28 A. Corma, P. Concepcion and P. Serna, Angew. Chem., Int. 33 K. Kuroda, T. Ishida and M. Haruta, J. Mol. Catal. A: Chem.,
Ed., 2007, 46, 7266. 2009, 298, 7.
29 I. A. Makaryan and V. I. Savchenko, Stud. Surf. Sci. Catal., 34 D. A. Dotzauer, S. Bhattacharjee, Y. Wen and M. L. Bruening,
1993, 75, 2439.
Langmuir, 2009, 25, 1865.
This journal is ª The Royal Society of Chemistry 2013
J. Mater. Chem. A, 2013, 1, 14311–14319 | 14319