Gold nanoclusters prepared from an eighteenth century two-phases procedure supported on thiol-containing SBA-15 for liquid phase oxidation of cyclohexene with molecular oxygen

Article abstract of DOI:10.1016/j.cattod.2017.09.045

The present work describes the incorporation of finely divided and well-dispersed Au species into SBA-15 mesoporous silica functionalized with mercaptopropyl groups, and the catalytic activity of the resulting Au-materials in the liquid phase oxidation of cyclohexene with molecular oxygen at atmospheric pressure. The gold entities have been prepared through a two-phase system described in the eighteenth century involving a solution of gold in a mixture of nitric acid and ammonium chloride (aqua regia) and rosemary oil. Gold nanoclusters, which present a disordered structure and sizes below 3 nm, and even single gold atoms which may be anchored/stabilized through sulphur groups are found immobilized on the mesoporous support, but no gold nanoparticles (with well-defined structure and usually over ～5 nm) are detected. Such observations have been carried out by C_s-corrected STEM-HAADF where the visualization of bright spots is attributable to these individual entities. These materials are active in the oxidation of cyclohexene, but their relative activity is governed by the synthesis conditions. Partial aggregation of the gold entities into nanoparticles has been observed during the reaction, which are mainly responsible for the catalytic activity. The allylic oxidation of the cyclohexene ring over the double bond epoxidation is strongly favored, and at high conversion the two dominants products are 2-cyclohexen-1-one and 2-cyclohexen-1-ol.

Full text of DOI:10.1016/j.cattod.2017.09.045

Catalysis Today xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Catalysis Today
journal homepage: www.elsevier.com/locate/cattod
Gold nanoclusters prepared from an eighteenth century two-phases
procedure supported on thiol-containing SBA-15 for liquid phase oxidation
of cyclohexene with molecular oxygen
a
a
b
a,⁎
J. Agundez , L. Martin , A. Mayoral , J. Pérez-Pariente
a
Instituto de Catálisis y Petroleoquímica (CSIC), Marie Curie, 2, 28049 Cantoblanco, Madrid, Spain
Advanced Microscopy Laboratory (LMA), Instituto de Nanociencia de Aragon (INA), University of Zaragoza, Mariano Esquillor, Edificio I+D, 50018, Zaragoza, Spain
b
A R T I C L E I N F O
A B S T R A C T
Keywords:
Gold nanoclusters
Catalysis
Cyclohexene
Oxidation
The present work describes the incorporation of finely divided and well-dispersed Au species into SBA-15 me-
soporous silica functionalized with mercaptopropyl groups, and the catalytic activity of the resulting Au-ma-
terials in the liquid phase oxidation of cyclohexene with molecular oxygen at atmospheric pressure. The gold
entities have been prepared through a two-phase system described in the eighteenth century involving a solution
of gold in a mixture of nitric acid and ammonium chloride (aqua regia) and rosemary oil. Gold nanoclusters,
which present a disordered structure and sizes below 3 nm, and even single gold atoms which may be anchored/
stabilized through sulphur groups are found immobilized on the mesoporous support, but no gold nanoparticles
Transmission electron microscopy
Allylic oxidation
(with well-defined structure and usually over ∼5 nm) are detected. Such observations have been carried out by
C
s
-corrected STEM-HAADF where the visualization of bright spots is attributable to these individual entities.
These materials are active in the oxidation of cyclohexene, but their relative activity is governed by the synthesis
conditions. Partial aggregation of the gold entities into nanoparticles has been observed during the reaction,
which are mainly responsible for the catalytic activity. The allylic oxidation of the cyclohexene ring over the
double bond epoxidation is strongly favored, and at high conversion the two dominants products are 2-cyclo-
hexen-1-one and 2-cyclohexen-1-ol.
1. Introduction
included a recipe claiming that, after prolonged contact between a so-
lution of gold in aqua regia (consisting in this old formulation in a
The behavior of gold nanoparticles as catalysts for a large variety of
mixture of nitric acid and ammonium chloride) and rosemary oil, the
upper oil layer mixed with alcohol got red colored. This red solution
was called the “potable gold of Mademoiselle Grimaldi” and it was
supposed to have medicinal properties. Recent replication of this recipe
has revealed the presence in the rosemary oil of gold nanoparticles
(which adopted decahedral, icosahedral or twinned fcc structures) of an
average size of 5 nm, disordered gold clusters of < 1 nm and even
isolated gold atoms [7].
Based on these findings, in this paper it is reported the im-
mobilization of the gold species produced by this eighteenth century
recipe on a mesoporous SBA-15 material functionalized with mercap-
topropyl groups and the use of the resulting gold-containing materials
as catalysts for the oxidation of cyclohexene in liquid phase by using
molecular oxygen as oxidant [8,9].
organic chemical reactions have been extensively explored [1–3]. The
catalytic activity and selectivity of these nanoparticles are dependent
upon their size and shape, the specific chemical environment they are
exposed during reaction and, if immobilized on a support, their inter-
action with the carrier. In this context, a key challenge is to prepare
stable gold nanoparticles with controlled size and shape, and several
methods have been developed along the time to meet this objective [4].
One of these is the two-phase Brust-Schiffrin method introduced in
1
994 [5], which allows the in situ preparation of thiolate-stabilized gold
nanoparticles with small size and very narrow particle-size distribution.
It is acknowledged in that publication that the preparation of colloidal
metals in a two-phase system was introduced by Michael Faraday in the
nineteenth century [6], consisting in the reduction of an aqueous gold
solution with phosphorous in carbon disulfide. However, it has been
found that in the 1757 edition of the then well-known French textbook
of chemistry Cours de Chymie by N. Lemery, the editor M. Baron
⁎
Corresponding author.
E-mail addresses: jagundez@icp.csic.es (J. Agundez), amayoral@unizar.es (A. Mayoral), jperez@icp.csic.es (J. Pérez-Pariente).
http://dx.doi.org/10.1016/j.cattod.2017.09.045
Received 10 May 2017; Received in revised form 1 September 2017; Accepted 24 September 2017
0920-5861/ © 2017 Elsevier B.V. All rights reserved.
Please cite this article as: Agundez, J., Catalysis Today (2017), http://dx.doi.org/10.1016/j.cattod.2017.09.045

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
2. Experimental
based on the reduction of HAuCl
4
·3H
2
O (Johnson Matthey) in an aqu-
4
eous solution of polyvinyl alcohol (Sigma-Aldrich) with NaBH (Sigma-
2.1. Synthesis of thiol-containing SBA-15
Aldrich, > 96%). The Au content of this catalysts is 1.2 wt%. It will be
denoted as RefCat.
Propyl-thiol mesoporous SBA-15 was prepared from a gel with
molar composition: 1.0 TEOS:0.111 MPTMS:0.0186 P123:6.42
HCl:180H O, where TEOS stands for tetraethyl orthosilicate (Sigma-
Aldrich, > 99%);MPTMS for 3-mercaptopropyltrimethosysilane
Sigma-Aldrich, 95%); P123 for Pluronic 123, the triblock co-polymer
2.4. Catalytic tests
2
The oxidation of cyclohexene with molecular oxygen was carried
out by following a procedure based on that reported in [8]. The cata-
lytic reaction was carried out in a 50 mL glass three-neck round-bottom
flask provided with an additional vertical neck to adapt a condenser
through which water at 5 °C was circulated to minimize the evaporation
of reagents and products. The flask was immersed in a silicone oil bath
to keep the reaction temperature at 65 °C, measured inside the reaction
mixture by a thermometer inserted in one of the necks. The experi-
mental set-up was not protected against environmental light. The re-
action mixture was formed by 0.049 mol of cyclohexene (4.055 g,
Sigma-Aldrich, 99%), 0.4055 g of octane (10 wt% referred to cyclo-
hexene; Sigma-Aldrich, > 99%), 0.2027 g of a ter-butyl hydroperoxide
solution (TBHP, 5 wt% referred to cyclohexene; ∼5.5 M in decane,
Sigma-Aldrich), 3.041 g of toluene (75 wt% of the cyclohexene; Pan-
(
PEO20PPO70PEO20, m.w. ∼5800 (Sigma-Aldrich); and HCl for hydro-
chloric acid (Panreac, 37 wt%), according to the procedure described in
Ref. [10], as follows: 125 mL of 1.9 M HCl were placed in a 500 mL
plastic bottle provided with a cover having a hole for allowing the in-
sertion of a PTFE (Polytetrafluoroethylene) stirrer blade, and 4 g of
P123 were added. Then, the bottle was heated at 40 °C in a silicone oil
bath, and 8.2 mL of TEOS were added. After 45 min, 764 μL of MPTMS
were added, and the mixture was stirred for 22 h. Then, the mixture
was poured into a stainless-steel autoclave provided with a Teflon liner,
and heated statically at 100 °C for 24 h. After that, the autoclave was
cooled and its content filtered, washed with ethanol and dried at room
temperature overnight. The dried sample was treated with ethanol
(
200 mL of ethanol per g of sample) under stirring in a 1L round-bottom
reac, > 99.5%), and 0.050 g of catalyst. O (1.8 mL/min) was bubbled
2
flask at 90 °C for 24 h in order to remove the surfactant. Following this
procedure, two different batches of SH-SBA-15 precursor materials
were prepared, denoted L1 and L2, respectively.
through the stirred reaction mixture. The catalysts were previously
heated at 100 °C for 1 h in the reaction flask provided with a tube
containing molecular sieve 5A as water trap, cooled down to the re-
action temperature of 65 °C, and then the reagents added. The trap was
removed before the reaction starts. Aliquots of 0.20 mL of the reaction
mixture were taken at given time intervals and analyze by GC in a
Varian CP-300 instrument, by using a FactorFour™ (Varian VF-1 ms)
dimethylpolysiloxane capillary column, 15 m of length and 0.25 mm of
i. d. Octane was used as internal standard. Five reaction products were
identified; two of them result from the addition of oxygen to the cy-
clohexene double bond, namely cyclohexene epoxide and cyclohex-
anediol, and three coming from the allylic oxidation of the cyclohexene
ring: 2-cyclohexen-1-ol, 2-cyclohexen-1-one and 2-cyclohexenyl hy-
droperoxide. Cyclohexene conversion was calculated from the yields of
these five products.
2.2. Synthesis of the gold nanoparticles
Gold nanoparticles were prepared from a two-liquid phase system
according to a procedure described in the Lemery’s textbook [7]. A gold
lump (0.1657 g, Johnson-Matthey, 99.99%) was dissolved under gentle
stirring in 53 g of aqua regia, prepared following this old formulation by
mixing (4:1 w/w) nitric acid (Panreac, 65 wt%) and ammonium
chloride (Sigma-Aldrich, > 98 wt%), heated at 40 °C in a sand bath.
This solution has a 1:320 gold-to-aqua regia weight ratio. After cooling,
the resulting golden yellow solution was placed in a 50 mL decanting
funnel, and then 13.50 g of rosemary essential oil were gently added,
which remains as a top layer over the gold solution. The same proce-
dure was followed with a more concentrated gold solution in aqua regia,
2.5. Characterization techniques
1:64. The rosemary essential oil (supplied by the Spanish company El
Granero Integral (The Integral Barn) has the following chemical com-
position (wt.%) as determined by GC–MS employing a gas chromato-
graph (Agilent 6890) coupled with a mass spectrometer (Agilent
Powder X-ray diffraction studies were done using a PANalytical
X’pert Pro instrument (CuKα radiation). Gold content of the solid was
determined by inductively coupled plasma (ICP-OES) spectrometry
with an ICP Winlab Optima 3300 DV Perkin-Elmer spectrometer.
Thermogravimetric analyses were performed in a Perkin-Elmer TGA7
instrument, in an air flow of 40 mL/min, with a heating ramp from 25
to 900 °C at 20 °C/min. CHNS elemental analyses were obtained in a
LECO CHNS-932 analyser provided with an AD-4 Perkin-Elmer scale.
5
(
2
973N) using
30 m × 0.25 mm × 0.25 μm), heating from 70 to 290 °C at 6 °C/min:
4.9% 1,8-cineole, 21.9% alpha-pinene, 20.91% camphor, 9.06%
a capillary column made of methylpolysiloxane
camphene, 3.81% borneol, 3.34% verbenone, 2.59% myrcene, 2.41%
beta-pinene, 2.01% caryophyllene, 2.00% p-cymene, 1.16% alpha-hu-
mulene, 0.98% bornyl acetate, 0.94% gamma-terpinene, 0.63% 4-ter-
pineol, 0.63% alpha-terpineol, 0.29% alpha-terpinolene, 0.28%
fenchone.
Nitrogen adsorption-desorption isotherms where measured in
a
Micromeritics ASAP 2420 apparatus at the temperature of liquid ni-
trogen (−196 °C). The samples were degassed in situ at 120 °C in va-
cuum for 16 h prior to analysis. Surface areas were determined using
the BET method. The pore volume and the average pore diameter were
calculated by applying the BJH protocol to the adsorption branch of the
isotherm. Diffuse reflectance UV–visible (UV–vis) spectra were re-
corded on a Cary 5000 Varian spectrophotometer equipped with an
integrating sphere with the synthetic polymer spectralon as reference.
Transmission electron microscopy analyses were performed in a
XFEG FEI Titan 60–300 which was operated at 300 kV in scanning
transmission electron microscopy (STEM) mode using a High Angle
Annular Dark Field detector (HAADF). The spherical aberrations were
corrected using a CEOS corrector for the electron probe achieving a
spatial resolution of 0.8 Å. The column was fitted with an EDAX de-
tector for EDS chemical analyses and a Gatan Tridiem Energy Filter
(GIF). The samples were prepared by deeply crushing the powders
obtained using mortar and pestle for several minutes. Subsequently,
2.3. Immobilization of the gold nanoparticles on SBA-15
24 h and 8 days after the addition of the rosemary oil, 3.75 mL of
the organic layer were mixed with 18.75 mL of ethanol, to the resulting
solution 0.500 g of the extracted SBA-15 material were added, and the
mixture was stirred at room temperature for 3 h. After that, the solid
was separate by centrifugation and washed with four portions of 40 mL
each of ethanol. The samples were denoted as L1-1 and L1-8 for those
obtained from L1 precursor and the less concentrated gold solution
after 1 and 8 days, respectively, while L2-1 and L2-8 correspond to
those prepared from L2 and the more concentrated gold solution after 1
and 8 days.
For comparison purposes, a gold catalyst supported on L1 precursor
was prepared by the sol-immobilization method described in [11],
2

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
bonded to mercaptopropyl moieties. Taken into account that only a
small fraction of the total Si atoms would be present at channels in-
terface, the thiol groups would be covering in a very effective way the
channels surface. This aspect would be relevant for binding the gold
nanoparticles upon contacting these materials with the gold solutions.
Moreover, under the reasonable assumption that propyl groups are only
present at channel surface, their density can be roughly determined
from surface area and sulphur content, and it turns to be in the range
2
2
1
.1–1.4 S atoms/nm , or 1 S atom per 0.9 and 0.7 nm , respectively.
The area of thiol head groups of self-assembled alkanethiol layers on
2
flat Au(111) surface has been reported to be 0.214 nm [13], while it
has been calculated as 0.114 nm2 for octadecanethiol-derivatized Au
clusters of 2.2 nm in size modeled as spheres [14]. These values are
much smaller than the surface density of mercaptopropyl groups pre-
sent at the SBA-15 surface, and therefore suggest that they are far apart
each other. However, low average density values would not exclude the
presence of closer mercaptopropyl groups. Nonetheless, the low density
of thiol groups would eventually make feasible the binding and stabi-
lization of very small gold clusters formed by just few gold atoms,
provided they would be present at the solution where they are gener-
ated.
From the chemical analysis data collected in Table 2, it can be seen
that the C/S ratio is around 9, i.e., much higher than that corresponding
to mercaptopropyl moieties, even considering the possibility that some
methoxy groups of the precursor were not hydrolyzed during the
synthesis process. This excess of carbon could be accounted by the
partial covering of the surface with ethoxy groups during the prolonged
treatment of the sample in refluxing ethanol for surfactant removal,
which has been previously observed for thiol-functionalized MCM-41
Fig. 1. XRD patterns of representative samples L1and L1-8.
they were dissolved in EtOH and a few drops of the suspension were
placed onto a holey carbon copper microgrid.
[
15]. However, it could also be attributed to the presence of residual
block copolymer species which were not completely removed by
ethanol extraction [10].
3. Results and discussion
3.1. Thiol-containing SBA-15 precursors
3.2. Gold-containing samples
Two different batches of SBA-15 functionalized with mercapto-
After contacting the SH-SBA-15 materials with the rosemary oil/
ethanol precursor gold solution following the procedure described in
the experimental section, gold was indeed found in the resulting solid,
propyl moieties, denoted as L1 and L2, were synthesized and used to
prepare L1 and L2 gold-containing samples, respectively. XRD of the
representative precursor material L1 (Fig. 1) shows the characteristic
profile of well-ordered SBA-15 materials, and only minor differences
ranging from 0.2 wt% to 2.6 wt% (Table 2). The amount of immobilized
−
metal seems to be governed by the initial concentration of [AuCl
4
]
in
among them in d(100) spacing and a
0
(Table 1) were found. They show
the aqua regia solutions and their contact time with the rosemary oil.
For samples L1, prepared from the less concentrated gold solution, the
the characteristic adsorption-desorption isotherms of SBA-15
N
2
2
(
0
(
Fig. 2), a surface area in the range 540–660 m /g and pore volume
.64–0.82 cm /g, and practically the same pore diameter, 5.8–5.9 nm
3
amount of this metal immobilized in the mesoporous material increased
−
if the contact time between the [AuCl
4
]
solution and the rosemary oil
Table 1). Average pore diameter is clearly smaller than that of 7.0 nm
before the impregnation step of SBA-15 with the oil phase was pro-
longed from 1 to 8 days, while the opposite was observed for samples
L2 prepared from the more concentrated gold solution. Moreover, for
day of contact, the content of gold in the sample increased with its
concentration in the aqua regia solution. For the gold-containing sam-
ples, the SBA-15 structure was preserved (Fig. 1), with nearly no var-
0
iation in a parameter and pore diameter (Table 1 and Fig. 2). However,
both the surface area and pore volume decreased for the two samples
prepared from 8 days, which also show some noticeable differences in
the TG pattern as compared to both the precursor and 1 day samples
reported for pure silica SBA-15 [12] which strongly supports the pre-
sence of mercaptopropyl moieties decorating the pore surface, which
would hence reduce the free pore aperture. Chemical analysis evidences
the presence of sulphur and carbon (Table 2), which indicates the ef-
fective incorporation of mercaptopropyl groups. Moreover, TG (Fig. 3)
shows an abrupt weight loss centered at ∼340 °C, which can be as-
signed to the decomposition/desorption of these groups. From chemical
and TG analysis, the number of thiol groups per Si atoms can be esti-
mated in an approximate way, and the results are collected in Table 2.
It can be seen that nearly 10% of the Si atoms of the materials are
1
(
Fig. 3). The weight loss at T < 140 °C corresponding to the desorption
of water is higher, and a more complex pattern of weight loss starts
already at T > 140 °C, in such a way that four weight losses centered
at 230 °C, 350 °C, 530 °C and 630 °C can be distinguished. A similar TG
pattern has been reported for thiol-containing MCM-41 materials after
Table 1
Structural and textural properties of SH-SBA-15 supports and catalysts.
2
3
Sample
d-Spacing
100(nm)
Unit cell parameter
(nm)
S
BET (m /g)
Vp (cm /g)
Dpnm
d
a
0
2 2
oxidation with H O , where sulfonic and disulfide species were de-
tected [15]. Therefore, in the gold-containing materials prepared from
L1
L2
L1-1
L1-8
L2-1
L2-8
9.7
9.2
9.6
9.5
9.1
8.9
11.2
10.6
11.1
10.9
10.5
10.3
542
660
515
454
637
488
0.64
0.82
0.61
0.55
0.78
0.62
5.9
5.8
5.7
5.6
5.7
5.5
−
rosemary oil that was in contact for 8 days with the [AuCl
4
]
solution,
the original mercaptopropyl groups would have been oxidized in con-
siderable extension.
The absorption spectra of the catalysts in the UV–vis region are
presented in Fig. 4. It can be seen that in none of them the surface
3

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
2
Fig. 2. Representative N adsorption/desorption isotherms of sample L1 (top) and L1-8 catalyst (bottom). Inset: pore size distribution.
plasmon resonance band at ∼ 520 nm characteristic of Au NPs is de-
tected [16], which means that the gold aggregates, if present, should be
smaller than ∼2 nm in size [17]. In the spectra of samples L1-8 and L2-
0
a = 9.12 nm. A closer look (Fig. 5b) of the system allows the ob-
servation of a strong contrast particularly on the walls of the con-
forming channels attributed to the presence of gold due to its higher
scattering factor respect the silicon and oxygen that compose the SBA-
15 [22]. Considering that the images have been acquired using a high
annular dark field detector (HAADF) the contrast obtained is directly
related to the atomic number of the elements through Rutherford
scattering [23]. This approach, especially after the implementation of
spherical aberration correctors, has been widely used for the analysis of
incorporated species down to atomic level into lighter supports, just by
image analysis or combined with spectroscopic measurements [24,25].
Fig. 5c and d show another SBA-15 particle orientated with the pore
channels perpendicular to the electron beam, where the Au entities can
be also identified. Fig. 5d depicts the atomic resolution data, with a
magnified image shown inset, where gold clusters below 2 nm are and
even single Au atoms can be observed.
8, a band of weak intensity is observed at ∼370 nm, accompanied by a
shoulder at ∼ 270 nm. In the L1-1 and L2-1 samples, the band at higher
wavenumber is not observed, while the shoulder is more intense in
sample L1-1. UV–vis bands of thiol-stabilize gold clusters with core
diameter of 0.8 nm are detected in the range 300–450 nm [18], which
are slightly blue-shifted for 1–2 nm size gold clusters capped with meso-
2,3-Dimercaptosuccinic acid [19]. Moreover, UV bands at ∼250 nm
have been attributed to partially charged Au nanoclusters [20].
Therefore, the optical spectral features of the catalysts would be con-
sistent with the presence of thiolate-gold nanoclusters of sub-nanometer
size, or even thiolate-monoatomic gold species [21]. In order to eluci-
date the nature of the gold entities present in these materials, spherical
s
aberration corrected (C -corrected) electron microscopy studies have
been carried out and data corresponding to L2-8 material are depicted
in Fig. 5. Fig. 5a corresponds to a typical image of SBA-15 displaying
the hexagonal arrangement of the pores in a similar to a honeycomb
structure; the FFT shown inset can be indexed assuming p6mm space
s
These C -corrected STEM-HAADF findings are in agreement with the
reported high surface disorder of thiol protected gold particles in the
range 1–4 nm, due to the strong SeAu bond [26]. It can be noticed that
the average S-to-S distance calculated as indicated above is 0.7–0.9 nm,
which would be within the approximate value range of gold clusters
group symmetry, obtaining
a unit cell value of approximately
Table 2
Chemical composition (wt%) and TG results of the SH-SBA-15 supports and catalysts.
Sample
C
H
S
Au
C/Sa
S/Aua
Si/Sa
Weight loss T < 140 °C
Weight loss T > 140 °C.
L1
L2
L1-1
L1-8
L2-1
L2-8
14.7
13.2
–
–
12.5
12.5
3.3
2.8
–
–
3.0
2.9
4.2
3.7
–
–
3.5
3.2
–
–
0.19
2.57
1.54
0.53
9.3
9.4
–
–
9.6
10.5
–
–
–
–
13.9
36.9
8.9
10.4
–
–
–
1.87
1.88
2.23
6.32
4.76
4.71
28.16
26.09
27.41
25.53
24.52
25.44
–
a
Mol ratio.
4

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
Fig. 3. TG/DTG curves of samples: L1 (left); L1-1 and L1-8 catalysts (right).
size as determined by STEM. Moreover, the average S/Au ratio is in the
range 14–37 for L2 samples (Table 2), as a function of the gold content.
A cluster formed by 11 gold atoms would have a size of ∼0.8 nm [18],
which would be therefore linked to one mercaptopropyl surface group
on average at least for the lowest S/Au ratio. On the other hand, the
observed differences in the optical spectra among the four catalysts
suggest that the nature of the gold clusters depends on the specific
regard, it can be noticed the similarities of the UV–vis spectra of the
samples prepared at the same contact time, 1 or 8 days, regardless their
disparities in gold content. Therefore, the immobilization process of the
gold species present in the impregnating rosemary/ethanol solution
does not seem to alter significantly the nature of the gold clusters
present in that solution. According to this, thiol groups anchored on the
SBA-15 surface would capture the gold species present in the solution,
though whether this “fishing” process discriminate among the several
gold species that presumably are present in that solution cannot be said.
The remarkable presence of such small Au entities suggests that the
rosemary oil would act not only as a reductant agent of the gold present
conditions operating in the two-liquid phase system formed by the
−
[
AuCl
4
]
solution in aqua regia and the rosemary oil, where the contact
time between both phases seems to be a more determinant factor than
−
the starting [AuCl
4
]
concentration in the aqueous phase. On this
Fig. 4. UV–vis spectra of samples L1-1 ( ); L1-8 ( ); L2-1 (●); L2-8 ( );
L1-8 after reaction (♦).
5

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
Fig. 5. C
s
-corrected STEM-HAADF images of L2-8
catalyst. a) High-resolution image of the pore system
recorded along the [001] with the FFT shown inset.
b) Magnified region of the edge of the SBA-15 par-
ticle. c) High-resolution image recorded perpendi-
cular to the pore system (FFT is shown inset). d)
Closer inspection of the image shown in c, showing
the presence of the Au entities.
at the aqueous phase, but some of its components would also behave as
efficient capping agents preventing massive gold aggregation, whether
they are naturally present at the oil or they come from the slow reaction
between the upper oil layer and the underneath strongly acid and
oxidant aqueous solution. Additional research would be needed to
elucidate details of the mechanism leading to the formation of the gold
entities in this particular two-liquid phase system, which would help
gaining control on the output of the process.
those of 8-days ones, and also in this case the activity increases slightly
with the gold content, thought the difference in catalytic activity is
smaller than that of gold. These results indicate that the activity per
gold atom shows a great disparity among the two series of catalysts.
Moreover, as practically no differences are observed in the average pore
size among them, which is close to 6 nm, nearly one order of magnitude
larger than the molecular size of cyclohexene or TBHP, the reaction
would not be influenced by eventual differences in the accessibility of
the reagents to the gold active sites. Therefore, the observed variation
in the specific activity (conversion per gold atom) among the several
catalysts would be related to differences in the surface properties of the
gold clusters, which could be governed by their size but also by their
effective charge and the chemical interactions stablished with the sul-
phur-containig groups of the support. It is interesting to notice on this
regard that the two 8-days samples present quite similar UV–vis spectra,
as well as the two 1-day ones, which are nevertheless distinct from
those of the former (Fig. 4), as they do not present the band at ∼370
nm. Therefore, it could be tempted to connect the catalytic activity with
the size and nature of the gold clusters present in the as-prepared cat-
alysts. However, it has been observed that the color of the 8-days cat-
alysts changes during the reaction, being white to pale yellowish before
reaction, to pink and then to purple as the reaction proceeds. This ob-
servation evidences that the particle size of the gold increases in the
reaction medium, as it is confirmed by UV–vis spectroscopy of the L1-8
and catalyst after reaction (Fig. 4), which presents an absorption band
at ∼520 nm. In agreement with this result, gold particles in the range
It can be mentioned that gold nanoparticles of sizes in the range
2
–4 nm immobilized on SBA-15 functionalized with mercaptopropyl
groups have been prepared by one-pot synthesis procedure after the
gold-containing materials were calcined and therefore the thiol groups
removed [21].
3.3. Catalytic tests
The four gold-containing SH-SBA-15 materials described above
were used as catalysts for the liquid phase oxidation of cyclohexene
using molecular oxygen as oxidant at atmospheric pressure, in the
presence of a small amount of toluene as solvent and TBHP as initiator
(
Table 3). The four catalysts (Fig. 6) are active in this reaction, but they
show noticeable differences concerning their activity pattern. Only
catalyst L1-8 shows a conversion level significantly above that of gold-
free thiol-SBA-15 support used as a reference or blank experiment. On
the other hand, the conversion is much higher than the amount of in-
itiator TBHP present in the reaction mixture, hence the oxidation of
cyclohexene necessarily involves the participation of molecular oxygen.
For the two 8-days catalysts, the conversion increases with the gold
content (Table 2). The two 1-day catalysts show lower activity than
5
–20 nm are clearly observed by STEM in this catalyst after reaction
Fig. 7), although gold nanoclusters are still present. Fig. 7 shows the
-corrected STEM-HAADF images also recorded along the [110] and
(
C
s
6

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
Table 3
Conversion and selectivity for selected reaction times.
Catalyst
Conversion% (24 h)
Conversion% (48 h)
Selectivity (24 h)
Selectivity (48 h)
L1
L1-1
L1-8
L2-1
L2-8
L1-8 reused
RefCat
17.6
9.9
29.6
15.7
33.4
28.4
26.5
31.0
39.9
61.1
72.5
32.7
48.2
62.3
35.3
25.6
11.3
8.3
15.8
14.5
38.0
24.6
19.4
34.5
41.4
11.2
4.1
7.5
7.8
4.5
8.7
7.8
0.6
0.5
3.9
4.2
1.0
2.9
3.8
36.1
55.4
9.6
13.9
44.1
8.4
17.5
13.2
26.6
22.1
15.7
28.5
28.8
32.7
22.4
60.0
47.6
30.1
58.5
60.5
10.1
8.0
1.5
4.2
7.1
2.3
2.2
3.6
1.0
2.2
12.1
3.0
2.3
2.4
30.1
12.3
13.8
22.6
40.0
17.8
15.1
12.8
18.6
21.4
6.1
[
001] crystal orientations proving that after reaction the material re-
observed in Fig. 6 that the reaction is faster with this catalyst as com-
pared with L1-8, and the conversion is close to 40% after 24 h, although
it does not progresses further.
tained its original structure with no significant damage (Fig. 7a and b).
From these images it can be clearly appreciated the presence of large Au
nanoparticles of several tenths of nm. A closer look of two particles
This experiment strongly suggests that the higher activity of the L1-
8 catalyst is due to in situ growing of the gold nanoclusters initially
present in the fresh catalyst to form active gold nanoparticles. This
phenomenon would also operate for L2-8 catalyst, although its lower
Au content led also to smaller conversion. It can also be taken into
account that very small clusters and single gold atoms are still present
in the used catalyst, which means that the Au growing process has not
been completed yet in the course of the reaction. In contrast to the 8-
days catalysts, no noticeable color change is observed for the 1-day
catalysts during the reaction, being their activity very low, even that of
L2-1, which contains 1.5 wt% of gold (Table 2). These results point to
differences among the two series of catalysts concerning the interaction
strength between the gold entities and the sulphur-bearing moieties to
which they are linked. This interaction would be much stronger for the
two 1-day than for the 8-days ones, preventing the growing of the gold
nanoclusters toward active nanoparticles in the former. The reason for
these differences could be found in the different chemical state of the
sulphur in both cases. As it has been discussed above, the TG results
(Fig. 3) shows that the thiol groups present in the support are preserved
for the 1-day samples, while they are largely oxidized for the 8-days
ones. Hence, the interaction of the thiol group with the gold entities of
the former would be strong enough to avoid their partial reduction and
growing in the reaction mixture. These differences in the Au-S inter-
action would also be reflected in the UV–vis spectra, where the band
(
Fig. 7c and d) confirms the presence of nanoparticles of 10–20 nm,
which in some cases fit the pore size, in addition to the initial metallic
clusters of around 2 nm. The atomic-resolution data shown in Fig. 7e
and f allow the visualization of the metallic cluster together with iso-
lated single Au atoms. In order to unambiguously determine the pre-
sence of the Au as isolated entities, different intensity profiles were
extracted over three distinct high contrast units. The significant con-
trast changes observed over the 3 signals measured corroborates the
heavier nature of those elements, which in the current case can only
correspond to gold atoms. Although these entities are denoted as iso-
lated gold atoms, the fact that they are anchored onto the mesoporous
support before and after reaction, suggests good physical-chemical
stability probably through the coordination of sulphur, and therefore
the Au single atoms should be part of a bit more complex compound
with sulphur linked to the gold atom.
In an attempt to determine whether the observed increase of gold
particle size is responsible for the higher activity of L1-8 catalyst, a
catalyst was prepared by the sol-immobilization method described in
[
11] by using L1 as support, which has been described to led to Au
nanoparticles when supported on graphite. In our case, gold nano-
particles in the range 2–12 nm are indeed observed, with most of the
particles of 5 nm in size, in agreement with the presence of the char-
acteristic ∼520 nm band in the UV–vis spectrum (not shown). It can be
Fig. 6. Conversion as a function of time for catalysts L1-1 ( ); L1-8 ( );
L2-1 (●); L2-8 ( ); L1-8 reused (◊); blank reaction carried out with
support L1(▲); RefCat (*). Solid lines have been drawn as visual aid only.
7

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
s
Fig. 7. C -corrected STEM-HAADF images of the catalyst L1-8 after reaction. a) Low-magnification micrograph with the channels perpendicular to the electron beam. SBA-15 particle
orientated with the pores perpendicular to the electron beam. c) and d) Closer inspections of the particles shown in 7a and b respectively. e) Ultra-high resolution data extracted from
figure c. f) Image of the edge of the SBA-15 showing isolated Au atoms.
at ∼370 nm is only observed for the two 8-days fresh catalysts. In this
way, the strong Au-SH interaction would not only to prevent the growth
of the gold entities, but would poison in large extension their oxidizing
ability. Therefore, not all the Au present in the L1-8 catalyst would be
active, because gold nanoclusters are still detected in the used catalyst.
On the other hand, it can also be observed in Fig. 6 that the L1-8
catalyst can be reused with no loss of activity. Therefore, once the
nanoparticles are formed in the first run, they keep active for reusing.
Among the five reaction products, two of them result from the ad-
dition of oxygen to the cyclohexene double bond, namely cyclohexene
epoxide and cyclohexanediol, and three are coming from the allylic
oxidation of the cyclohexene ring: 2-cyclohexen-1-ol, 2-cyclohexen-1-
one and 2-cyclohexenyl hydroperoxide. As it can be seen in Table 3, the
allylic oxidation pathway is dominant, and for conversion higher than
transforms most probably into cyclohexanediol which is a secondary
and unstable product. We have not been able to identify the product(s)
coming from cyclohexanediol reaction. The scheme in Fig. 9 shows a
tentative reaction pathway of the aerobic oxidation of cyclohexene
under the reaction conditions prevalent in this work.
4. Conclusions
Gold nanoclusters can be conveniently prepared by a two-liquid
phase system from a gold solution in aqua regia by using rosemary oil as
oil phase and reductant agent according to a technique reported in
eighteenth century. These gold entities can be immobilized in hybrid
mesoporous SBA-15 materials functionalized with mercaptopropyl
groups, where gold nanoclusters and single atoms are found, but no
gold nanoparticles are detected. While the structural features of SBA-15
remain unaffected by the immobilization procedure, some thiol groups
can be partially oxidized in the process, as a function of specific
synthesis parameters. These gold-containing materials are active cata-
lysts for the liquid phase oxidation of cyclohexene with molecular
oxygen in the presence of small amounts of ter-butyl hydroperoxide and
toluene as solvent, though their activity is strongly dependent upon the
specific synthesis conditions of the precursor gold solution. It has been
found that part of the gold nanoclusters present in the fresh catalysts
evolves toward larger gold nanoparticles during the reaction, which are
mainly responsible for the catalytic activity. The growing process of
these nanoparticles is governed by the chemical nature of the sulphur
groups attached to the gold entities in the as made catalysts. The cy-
clohexene is oxidized to two main stable reaction products, 2-cyclo-
hexen-1-one and, 2-cyclohexen-1-ol, with a molar enone/enol ratio of
∼
30%, the selectivity to the corresponding products is nearly 90%. The
predominance of the allylic oxidation is in agreement with what has
been previously reported for 1% Au/carbon catalysts containing
5
–50 nm gold nanoparticles [8]. The selectivity to the reaction products
is plotted in Fig. 8 as a function of the conversion for the seven reac-
tions carried out. For conversion < ∼20%, 2-cyclohexenyl hydroper-
oxide is the major product, followed by cyclohexen-1-one and 2-cy-
clohexen-1-ol. However, the selectivity to this peroxide decreases
rapidly as the conversion increases, and the cyclohexen-1-one is the
dominant product beyond ∼25% of conversion. The enone/enol molar
ratio is always > 1 and it increases with the conversion, reaching a
value of ∼2 as the conversion approaches to ∼40%. From Fig. 8 it can
be concluded that the 2-cyclohexenyl hydroperoxide is a primary pro-
duct of the oxidation of cyclohexene, as well as the cyclohexene ep-
oxide, but the former is much more favored. 2-cyclohexen-1-one is a
secondary and stable product coming from 2-cyclohexenyl hydroper-
oxide, while 2-cyclohexen-1-ol could be a primary + secondary stable
product. Cyclohexene epoxide is a primary but unstable product, which
2
, via 2-cyclohexenyl hydroperoxide as primary but unstable product.
Beside these major products coming from the allylic oxidation of cy-
clohexene, minor oxidation route involves the formation of
a
8

J. Agundez et al.
Catalysis Today xxx (xxxx) xxx–xxx
Fig. 8. Selectivity vs total conversion for 2-cyclohexenyl hydroperoxide
(●); 2-cyclohexen-1-ol (■); 2-cyclohexen-1-one (▲); cyclohexene ep-
oxide (♦) and cyclohexanediol (+). Data collected for the seven reactions
describe in this work at different reaction times.
[
[
7] A. Mayoral, J. Agúndez, I.M. Pascual-Valderrama, J. Pérez-Pariente, Gold Bull. 47
(2014) 161–165.
8] M.D. Hughes, Y.-J. Xu, P. Jenkins, P. McMorn, P. Landon, D.I. Enache, A.F. Carley,
G.A. Attard, G.J. Hutchings, F. King, E.H. Stitt, P. Johnston, K. Griffin, C.J. Kiely,
Nature 437 (2005) 1132–1135.
[
9] S.N. Bawaked, F. Dummer, N. Dimitratros, D. Bethell, Q. He, C.J. Kiely,
G.J. Hutchings, Green Chem. 11 (2009) 1037–1044.
[
10] D. Margolese, J.A. Melero, S.C. Christiansen, B.F. Chmelka, G.D. Stucky, Chem.
Mater. 12 (2000) 2448–2459.
[
11] N. Dimitratos, J.A. Lopez-Sanchez, J.M. Anthonykutty, G. Brett, A.F. Carley,
R.C. Tiruvalam, A.A. Herzing, C.J. Kieley, G.J. Hutchings, Phys. Chem. Chem. Phys.
1
1 (2009) 4952–4961.
12] J.P. Thielemann, F. Girgsdies, R. Schlög, C. Hess, Beilstein J. Nanotechnol. 2 (2011)
10–118.
13] N. Camillone, C.E.D. Chidsey, G. Liu, G. Scholes, et al., J. Chem. Phys. 98 (1993)
234–4245.
14] A. Badia, L. Cuccia, L. Demmers, F. Morin, R.B. Lennox, J. Am. Chem. Soc. 119
1997) 2682–2692.
15] I. Diaz, C. Márquez-Álvarez, F. Mohíno, J. Pérez-Pariente, E. Sastre, J. Catal. 193
2000) 283–294.
[16] S. Link, M.A. El-Sayed, J. Phys. Chem. B 103 (1999) 4212–4217.
[
[
[
[
1
Fig. 9. Reaction scheme representing the most probable pathways for cyclohexene
transformation. A: allylic oxidation; B: double bond oxidation.
4
(
cyclohexene epoxide and cyclohexanediol, both unstable products.
(
Acknowledgements
[
17] M.C. Daniel, D. Astruc, Chem. Rev. 104 (2004) 293–346.
[
18] G.H. Woehrle, M.G. Warner, J.E. Hutchinson, J. Phys. Chem. B 106 (2002)
The authors gratefully acknowledge the Spanish Ministry of
Economy and Competitiveness for the funding through Projects
MAT2012-31127, MAT2016-77496-R (AEI/FEDER, UE). And
Government of Aragon and the European Social Fund under the project
grant number E/26 “Construyendo Europa desde Aragón” 2014–2020.
9
979–9981.
[19] Y. Negishi, T. Tsukuda, J. Am. Chem. Soc. 125 (2003) 4046–4047.
[
[
[
[
20] N. Bogdanchikova, A. Pestryakov, I. Tuzovskaya, T.A. Zepeda, M.H. Farias,
H. Tiznado, O.A. Martyniuk, Fuel 110 (2013) 40–47.
21] P. Wu, P. Bai, Z. Lei, K.P. Loh, X.S. Zhao, Microporous Mesoporous Mater. 141
(2011) 222–230.
22] A. Mayoral, T. Carey, P.A. Anderson, I. Diaz, Microporous Mesoporous Mater. 166
(2013) 117–122.
References
23] D.B. Williams, C.B. Carter, Transmission electron microscopy, A Textbook for
Materials Science, Springer, New York, 2009.
[1] A.S.K. Hashmi, G.J. Hutchings, Angew. Chem. Intl. Ed. 45 (2006) 7896–7936.
[2] A. Corma, H. García, Chem. Soc. Rev. 37 (2008) 2096–2126.
[3] Y. Zang, X. Cui, F. Shi, Y. Deng, Chem. Rev. 112 (2012) 2467–2505.
[4] P. Zhao, N. Li, D. Astruc, Coord. Chem. Rev. 257 (2013) 638–665.
[5] M. Brust, M. Walker, D. Bethell, D.J. Schiffrin, R. Whyman, J. Chem. Soc. Chem.
Commun. (1994) 801–802.
[24] A. Mayoral, R. Arenal, V. Gascon, C. Marquez-Alvarez, R.M. Blanco, I. Diaz,
ChemCatChem 5 (2013) 909.
[25] A. Mayoral, R.M. Hall, R. Jackowska, J.E. Readmanv, Angew. Chem. Int. Ed. 55
(2016) 16127–16131.
[26] M.M. Mariscal, J.A. Olmos-Asar, C. Gutierrez-Wing, A. Mayoral, M.J. Yacaman,
Phys. Chem. Chem. Phys. 12 (2010) 11785–11790.
[6] M. Faraday, Philos. Trans. R. Soc. Lond. 147 (1857) 145–181.
9