As

CIV

a

ARRAA

KGTAA

1

nchcboAtsfinasdoti

cgt

h0

Catalysis Today 235 (2014) 72–78

Contents lists available at ScienceDirect

Catalysis Today

j o ur na l ho me page: www.elsev ier .com/ locate /ca t tod

tomically well-defined Au25(SR)17/18 nanoclusters deposited on silicaupports for the aerobic epoxidation of trans-stilbene

hristophe Lavenn, Aude Demessence ∗, Alain Tuelnstitut de Recherches sur la Catalyse et l’Environnement de Lyon (IRCELYON), UMR 5256, CNRS/Lyon 1 University, 2 avenue Albert Einstein, 69626illeurbanne, France

r t i c l e i n f o

rticle history:eceived 20 November 2013eceived in revised form 13 February 2014ccepted 25 February 2014

a b s t r a c t

Two atomically well-defined thiolate gold nanoclusters, Au25(SC2H4Ph)18 and Au25(SPhNH2)17, havingthe same gold core but different thiolate ligands have been deposited on silica supports and the obtainedcatalysts have been tested in the aerobic oxidation of trans-stilbene. By properly selecting the supportto maximize cluster/surface interactions, it was possible to prepare materials with small nanoparticles

vailable online 27 March 2014

eywords:old clustershiolate ligandserobic oxidation

and very narrow particle size distributions. Au25(SPhNH2)17@SBA-15 catalyst showed an activity anda trans-stilbene epoxide selectivity comparable to the best heterogeneous catalysts. Elimination of thethiolate molecules by calcination did not significantly modify the particle size but decreased the activity,pointing out the critical role of the ligands in the reaction.

© 2014 Elsevier B.V. All rights reserved.

lkene

. Introduction

Since Haruta’s work on CO oxidation catalyzed by goldanoparticles [1], those materials have been widely used asatalysts in various catalytic reactions such as alcohols and alde-ydes oxidation, alkenes epoxidation, hydrochlorination of ethyne,arbon–carbon bond formation, and so on [2]. When reactions cane performed in the presence of non-polluting oxidants, typicallyxygen from the air, they are particularly important for industry.erobic oxidations can be carried out over supported gold nanopar-

icles [3] as well as colloids [4]. Among these oxidative reactions,elective alkenes epoxidation remains an important challenge inne chemicals area and efficient catalysts are still needed. Goldanoparticles deposited on different supports can efficiently cat-lyze the aerobic epoxidation of alkenes [5–7] but despite extensivetudies on the subject, the understanding of the role of particlesiameter, their surface, the effect of the support and the difficulty tobtain monodisperse particles with a narrow particle size distribu-ions (PSD) makes difficult to rationalize the mechanisms involvedn this reaction.

Over the past decade, atomically well-defined thiolate gold

lusters, defined as Aun(SR)m, where n and m are number ofold atoms and surrounding thiolate-based molecules, respec-ively, have attracted intense research interests due to their

∗ Corresponding author. Tel.: +33 472 445 322; fax: +33 472 445 399.E-mail address: aude.demessence@ircelyon.univ-lyon1.fr (A. Demessence).

ttp://dx.doi.org/10.1016/j.cattod.2014.02.045920-5861/© 2014 Elsevier B.V. All rights reserved.

unique electronic, magnetic and optical properties [8]. These atom-ically and structurally well-defined thiolate gold clusters haveemerged as a new promising catalysts class to understand the rela-tionship between the size, the structure of the nanoparticles andtheir catalytic activity [9]. Indeed, their atomically well definedcomposition and known structure can lead to a precise under-standing of size, structural and electronic effects on the catalyticmechanisms.

In addition, although sulfur species are considered as a poison incatalytic reactions such as the oxidation of styrene [10], it has beenshown that thiolate gold clusters can be active, even in the pres-ence of thiolate groups, for the oxidation of CO [11] or sulfide [12],the reduction of nitrophenol [13] or the selective hydrogenationof ketones and aldehydes [14,15]. For the epoxidation of alkenes,thiolate gold clusters have been tested on styrene [14], and on bothtrans- and cis-stilbene substrates [16]. The conversion of styrene,with Au25(SC2H4Ph)18 clusters as the catalyst, in the presence ofO2, was of ca. 18 and 27% at 24 h, depending on whether clusterswere supported or not on silica [14]. The major product in this reac-tion was the benzaldehyde (selectivity ∼70%). On the contrary, forthe trans-stilbene epoxidation with Au25(SC2H4Ph)18@SiO2 and anexcess of tert-butyl hydroperoxide used as the oxidant, the mainproduct was the trans-epoxide (selectivity 94%) with a 51% conver-sion at 24 h [16].

In the present work, we have studied the effects of the pres-ence of thiolate ligands around gold nanoclusters, their natureand the type of silica on which they are supported in the aer-obic epoxidation of trans-stilbene. Reactions are carried out in

sis To

mhbnahp

ptibcttAbaitaht[

tt

2

2

dmpapn

ih1(s

wspwpc

2

cfd

A1ppt

C. Lavenn et al. / Cataly

ethylcyclohexane at atmospheric pressure, 80 ◦C, using tert-butylydroperoxide (TBHP) as a radical initiator. Two gold clusters haveeen tested, namely Au25(SC2H4Ph)18 with hydrophobic exter-al phenyl groups and Au25(SPhNH2)17 with hydrophilic externalmino functions. The two supports were AEROSIL® R 972, aydrophobic fumed silica and SBA-15, a hydrophilic ordered meso-orous silica.

In liquid phase heterogeneous catalytic reactions involving sup-orted nanoparticles, some of the main parameters that haveo be considered to select the best catalysts are the wettabil-ty of the support in the reaction medium and the interactionetween the nanoparticles and the support to prevent release ofolloids in the solution. Because of their different surroundinghiolate ligands, the two Au25 clusters have been deposited onwo different supports to maximize cluster-support interactions:u25(SPhNH2)17 on SBA-15, to favor hydrogen bond interactionsetween the silanol groups of the silica with the amino functionsnd Au25(SC2H4Ph)18 on Aerosil-R972 to promote hydrophobicnteractions. These interactions are indeed of prime importance forhe stability and recyclability of catalysts. The recent report on thedsorption of Au25(SC12H25)18 on carbon nanotubes through weakydrophobic interaction clearly shows the importance of selectinghe appropriate support according to the clusters external functions17].

Finally, the effect of ligands has been evaluated by comparinghe activity of Au25(SPhNH2)17 on SBA-15 before and after calcina-ion at 400 ◦C.

. Experimental

.1. Materials

N,N-dimethylformamide (DMF, puriss. p.a. ≥99.8% (GC)),ichloromethane (Laboratory reagent, without stabilizer, >99.9%),ethylcyclohexane (ReagentPlus® 99%), methanol (MeOH, puriss

.a.), tetrahydrofurane (THF, Chromasolv®), trans-stilbene (96%)nd tert-butylhydroperoxide solution (TBHP, 70% in water) areurchased from Sigma–Aldrich Company. Ethanol absolute (tech-isolv) is bought from VWR International.

Gold nanoparticles deposited on TiO2 (Au@TiO2) (sample 105-A)s purchased from the World Gold Council [18]. The gold particlesave an average diameter of 3.5 ± 1.5 nm with a gold loading of.5 wt% and the titania support is Degussa P25, a mixture of anatase80%) and rutile (20%). This solid is used as a reference in the trans-tilbene epoxidation reaction [6].

All reagents and solvents are of commercial quality and usedithout further purification. The glassware used in the synthe-

is are cleaned with aqua regia (aqua regia is a very corrosiveroduct and should be handled with extreme care), then rinsedith copious amount of distilled water and dried overnightrior to the use. All reactions are carried out in atmosphericonditions.

.2. Equipments

Gold loading is determined by inductively coupled plasma opti-al emission spectroscopy (HORIBA Jobin Yvon Activa ICP-OES) andocused at 242.79 and 267.59 nm onto the detector. The solids areecomposed in H2SO4 + aqua regia solution + HF and then in HCl.

Standard X-ray diffraction (XRD) is carried out on a Bruker D8dvance A25 diffractometer using Cu K� radiation equipped with a

-dimensional position-sensitive detector (Bruker LynxEye). X-rayattern is recorded between 4◦ and 90◦ (2�) with 0.02◦ steps and 2 ser step. Divergence slit is fixed to 0.2◦ and the detector apertureo 189 channels (2.9◦).

day 235 (2014) 72–78 73

Transmission electron microscopy (TEM) is carried out on a JEOL2010 LaB6 microscope operating at 200 kV. All samples are dis-persed in ethanol with the help of an ultrasonic bath and then adrop is deposited on a copper grid and let to dry under ambientconditions of atmosphere and temperature. Samples 3 and 4 areembedded in an epoxy resin and sectioned with an ultramicrotometo obtain thin slices.

Gas adsorption isotherms for pressures in the range 0.01–1 barare measured by using a Micromeritics ASAP2020 instrument. Pow-ders of the compounds are transferred to a preweighed analysistube, which is capped and evacuated by heating at 100 ◦C underdynamic vacuum for 12 h. The sealed tube is then transferred tothe analysis port of the gas adsorption instrument. For all isotherms,warm and cold free-space correction measurements are performedby using ultra-high purity He gas (UHP grade 5.0, 99.999% purity).The N2 isotherms at 77 K are measured in a liquid nitrogen bath,using UHP-grade gas sources. The specific surface area is calculatedfrom the Brunauer–Emmett–Teller (BET) method and the pore sizedistribution by employing the Barrett–Joyner–Halenda (BJH) modelon the desorption curve of the N2 isotherm.

X-ray photoelectron spectroscopy (XPS) experiments are car-ried out on a Kratos Axis Ultra DLD spectrometer usingmonochromated Al K� X-rays (1486.6 eV, 150 W), a pass energyof 20 eV, a hybrid lens mode and stainless steel sample holder inultra-high vacuum (P < 10−9 mbar). The analyzed surface area is700 �m × 300 �m. Charge neutralization is required for all sam-ples. Shirley background subtraction is performed with the Vision2.2.8 Kratos processing program.

Gas chromatography is carried out on a Shimadzu GC-2014, using a Equity TMS 30 m × 0.25 mm × 0.25 �m columnprogrammed from 60 ◦C to 180 ◦C, injector and FID detectorset at 280 ◦C and 200 ◦C respectively, and He as carrier gas(26.6 mL min−1). External calibration is carried out by inject-ing distinct standard solutions containing trans-stilbene (96%,Sigma–Aldrich), trans-stilbene oxide (99%, Aldrich) and benzylalcohol (99+%, Sigma–Aldrich) in acetonitrile. Internal standard(dodecane) is used to calculate the conversion and yield of thereaction.

2.3. Catalysts preparation

Syntheses of the clusters Au25(SC2H4Ph)18 [19] andAu25(SPhNH2)17 [20] have been done by following the previ-ous reports. Hydrophobic silica (Aerosile® R972) is purchased fromDegussa and is described with an average primary particle size of16 nm and a surface area of 116 ± 2 m2 g−1. SBA-15 is prepared byfollowing a previous described synthesis [21] and Exhibits 8 nmhexagonal channels and a surface area of 570 ± 5 m2 g−1 (Fig. S1).

Cluster deposition is carried out by following a wet impregna-tion. A calculated amount, for a theoretical gold loading of 2 wt%,is dispersed in a minimum of solvent and is added dropwise into astirred suspension of the silica support.

2.3.1. Preparation of Au25(SC2H4Ph)18@SiO2(R972) (1)Au25(SC2H4Ph)18 clusters (7.5 mg; 1 �mol of clusters) are dis-

persed in dichloromethane (5 mL) and are added into a stirredsuspension of the hydrophobic silica Aerosil R972 (251.5 mg) in20 mL of a solution of dichloromethane/ethanol (v:v, 70:30). After24 h, the solvent is evaporated and the solid is washed two timeswith ethanol and recovered by centrifugation (15 min, 8000 rpm).Catalyst is left to dry under air at room temperature to obtain a lightgray solid.

2.3.2. Preparation of Au25(SPhNH2)17@SiO2(R972) (2)Au25(SPhNH2)17 clusters (7.5 mg, 1.1 �mol of clusters) are dis-

persed in dichloromethane (5 mL) with an ultrasonic bath, then

7 sis Today 235 (2014) 72–78

aR(erd

2

dsespaeo

2(

um

2

easg2ttcaos5a

agostovmhapaeta

(taa3

10 20 30 40 50 60 70 80 90

Inte

nsity (

a.u

.)

Angle 2 (° )

Fig. 1. Powder X-ray diffraction patterns of Au25(SC2H4Ph)18@SiO2(R972) (1) (lightgray), Au25(SPhNH2)17@SiO2(R972) (2) (gray), Au25(SPhNH2)17@SBA-15 (3) (black)

4 C. Lavenn et al. / Cataly

dded into a stirred suspension of the hydrophobic silica Aerosil972 (264.0 mg) in 20 mL of a solution of dichloromethane/ethanolv:v, 70:30) and sonicated for ∼1 min. After 24 h, the solvent isvaporated and the solid is washed two times with ethanol andecovered by centrifugation (15 min, 8000 rpm). Catalyst is left tory under air at room temperature to obtain a light gray solid.

.3.3. Preparation of Au25(SPhNH2)17@SBA-15 (3)Au25(SPhNH2)17 clusters (15.6 mg; 2.2 �mol of clusters) are

ispersed in DMF (2.5 mL) and is added dropwise into a stirreduspension of the mesoporous silica support SBA-15 (513.0 mg) inthanol (50 mL). Within a couple of hours, a decoloration of theolution from black to dark brown is observed. After 24 h, the pre-ared catalyst is recovered by centrifugation (15 min, 8000 rpm)nd washed successively two times with DMF and two times withthanol. Catalyst is left to dry under air at room temperature tobtain a light gray solid.

.3.4. Preparation of Au25(SPhNH2)17@SBA-15 calcined at 400 ◦C4)

Calcination is carried on ∼200 mg of compound (3) in an ovennder air. The solid is heated at 400 ◦C with a rate of 2 ◦C min−1 andaintained for 12 h.

.4. Catalytic evaluation

Catalytic evaluation is carried out in a round-bottom flaskquipped of a condenser under optimized conditions for theerobic epoxidation of trans-stilbene [7,22]. Trans-stilbene (sub-trate, 1 mmol, 187 mg), dodecane (standard, 1 mmol, 228 �L),old-based catalyst (Au: 4 �mol), methylcyclohexane (solvent,0 mL, 155 mmol) and tert-butyl hydroperoxide (radical initia-or, 0.05 mmol, 7 �L of a 70% aqueous TBHP solution) are stirredogether at 300 rpm and maintained at 80 ◦C under atmosphericonditions of pressure. In this study, an appropriate amount of cat-lyst is loaded in order to perform the reaction in the presencef the same amount of gold (4 �mol of Au for 1 mmol of trans-tilbene), the weight of the Au25(SR)17/18 clusters are mg (�mol):7.9 (3.88), 33.6 (3.75), 63.2 (3.43) and 56.0 (3.13) for (1), (2), (3)nd (4), respectively.

The reaction is monitored by regular samplings (∼0.2 mL) thatre diluted 40 times in acetonitrile and analyzed immediately byas chromatography. Trans-stilbene conversion and trans-stilbenexide yield are calculated from the ratio of the number of mole oftilbene converted into trans-stilbene oxide formed over the ini-ial number of mole of trans-stilbene introduced at the beginningf the reaction. Selectivity is defined as the ratio of yield over con-ersion. Turnover frequencies (TOF) are defined as the number ofole of trans-stilbene converted per mole of surface gold atom per

our. They are calculated from the conversion rates by taking intoccount the gold dispersion, which is determined from the meanarticle diameter obtained by TEM analyses after the catalytic testsnd by assuming that gold particles adopt a cuboctahedron geom-try. TOFs (Table 2) are calculated from ∼10% to ∼40% conversiono avoid influence of the induction time and the loss of catalyticctivity observed after ∼50% conversion.

Recyclability of catalyst (3) is tested by stopping the reactionheating and stirring shut off, condenser closed) at the end of

he first run (t = 48 h) and started again 119 h later at t = 167 h, bydding both trans-stilbene (187.5 mg, 1 mmol) and TBHP (7 �L of

70% aqueous solution, 0.05 mmol) at 80 ◦C and under stirring at00 rpm.

and Au25(SPhNH2)17@SBA-15 calcined at 400 ◦C (4) (red). (For interpretation of thereferences to color in this figure legend, the reader is referred to the web version ofthe article.)

3. Results and discussions

Both Au25(SC2H4Ph)18 [18] and Au25(SPhNH2)17 [20] clus-ters are synthesized and characterized by electrospray ionizationmass spectrometry, UV–visible spectroscopy, thermogravimetricanalysis and powder X-ray diffraction to confirm their formulaand purity. Because the two clusters have different solubility,Au25(SC2H4Ph)18 is soluble in toluene or dichloromethane andAu25(SPhNH2)17 only in DMF or DMSO; the impregnation on sil-ica supports is done in a mixture of dichloromethane and ethanolfor the first one and in a mixture of DMF and ethanol for thesecond one. After stirring for 24 h at room temperature, the graysolids are recovered by centrifugation (Fig. S2). The impregna-tion of Au25(SC2H4Ph)18 on SBA-15 is also carried out, but thefinal solid is too heterogeneous, in term of segregation of theclusters, to be used as a catalyst. Finally, to study the effect ofthe surrounding thiolate molecules in the trans-stilbene epoxida-tion, the ligands are removed by calcination. However because ofthe presence of methyl groups on its surface, the thermal stabil-ity of Aerosil R972 is limited. Consequently, calcination is onlyachieved on Au25(SPhNH2)17@SBA-15 at 400 ◦C for 12 h underambient atmosphere, above the decomposition temperature ofthe clusters (200 ◦C) [20]. After calcination, the color of the solidchanges from light gray to light pink, showing through the appear-ance of a plasmonic band, that the particles size has increased (Fig.S2). The four catalysts, namely Au25(SC2H4Ph)18@SiO2(R972) (1),Au25(SPhNH2)17@SiO2(R972) (2), Au25(SPhNH2)17@SBA-15 (3) andAu25(SPhNH2)17@SBA-15 calcined at 400 ◦C (4), are characterizedby inductively coupled plasma mass spectrometry (ICP-MS), pow-der X-Ray diffraction, TEM and porosimetry (Table 1).

From the synthesis, the starting gold percentage is 2 wt% andafter washings of the samples the measured gold loadings are 1.23,2.20, 1.07 and 1.10 wt% for (1), (2), (3) and (4) compounds, respec-tively (Table 1). These relatively high loadings suggest interactionsbetween the clusters and the support surface. This may be dueto the formation of hydrogen bonds between the external aminogroups of the clusters and the silanol groups of the silica surface[23] or to the weak hydrophobic interaction between the phenyland methyl groups of the Aerosil R972.

On the powder X-ray diffraction patterns of the four catalysts(Fig. 1), the large peak between 2� = 15 and 30◦ (max. 21–22◦) is

characteristic of amorphous silica. Peaks of the hexagonal organi-zation of the pores of SBA-15 are visible on solids (3) and (4) atlow angle X-ray scattering (Fig. S3). On the three non-calcined cat-alysts (1, 2 and 3), the absence of peaks characteristic of FCC bulk

C. Lavenn et al. / Catalysis Today 235 (2014) 72–78 75

Table 1Gold catalysts structural properties before the catalytic tests.

Gold loading (wt%) Average gold particle diameter (nm) Surface area (m2 g−1) Color at solid state

Au25(SC2H4Ph)18@SiO2(R972) (1) 1.23 1.1 ± 0.3 110 GrayAu25(SPhNH2)17@SiO2(R972) (2) 2.20 <1 ND GrayAu25(SPhNH2)17@SBA-15 (3) 1.07 0.9 ± 0.2 545 GrayAu25(SPhNH2)17@SBA-15 calcined at 400 ◦C (4) 1.10 1.8 ± 0.5 575 Pink

ND, not determined.

Table 2Catalytic properties of gold-based catalysts for the trans-stilbene epoxidation: 25%, 50% and 90% conversion time (t), trans-stilbene epoxide selectivity at half conversion(Sel50%), the turnover frequencies (TOF) and the gold particles mean size measured by TEM after catalytic tests.

t25% (h) t50% (h) t90% (h) Sel50% (%) TOF (h−1) Average gold particle diameter after test (nm)

Au@TiO2 (WGC) 24 80 228 66 4 3.5 ± 1.5Au25(SC2H4Ph)18@SiO2(R972) (1) 48 106 250 74 NA Polydisperse (2–20)Au25(SPhNH2)17@SiO2(R972) (2) 54 106 260 73 5 3.6 ± 0.9Au25(SPhNH2)17@SBA-15 (3)

1st run 6 12 30 65 29 2.2 ± 0.52nd run 3 6 <24 >95 43 2.3 ± 0.6

6

N

gp[44(imoHaiFf

Fa

Au25(SPhNH2)17@SBA-15 calcined at 400 ◦C (4) 6 72 23

A, data not available.

old points out that the clusters, which are characterized by a largeeak, remain very small after the deposition on the silica supports20]. However after the calcination of Au25(SPhNH2)17@SBA-15 at00 ◦C (4), diffraction peaks of bulk FCC gold appear at 2� = 38.1,4.4, 64.8 and 78.0◦ corresponding to the (1 1 1), (2 0 0), (2 2 0) and3 1 1) reflections. The estimation of the particle size by apply-ng Scherrer’s equation on the pattern refined by the Rietveld’s

ethod gives a diameter of around 2.2 nm. This diameter corrob-rates the observed color changing from light gray to pale pink.exagonal arrangement of the channels of SBA-15 still remains

fter calcination (Fig. S3). The presence of the clusters on the sil-ca supports is confirmed by transmission electron microscopy. Inig. 2, particles have a diameter of 1.1 ± 0.3, <1 and 0.9 ± 0.2 nmor (1), (2) and (3) solids, which is close to the diameter of Au25

ig. 2. (a) TEM picture of (1) and (b) its particle size distribution and fit. TEM pictures of (cnd (4) (pink/red). (For interpretation of the references to color in this figure legend, the

58 6 2.2 ± 0.7

clusters (0.98 nm) measured from the resolved structure [24]. TEMpictures also reveal that the clusters are well dispersed on the sur-face of aerosil R972 and that they are mainly located inside thechannels of the mesoporous SBA-15, the pores corresponding tothe light contrast. After calcination of Au25(SPhNH2)17@SBA-15 at400 ◦C, the particle size distribution obtained from TEM picturesshows that the gold particles grow from 0.9 ± 0.2 to 1.8 ± 0.5 nm,which means that around ten Au25 clusters coalesce together. Fromnitrogen isotherms (Fig. 3), the (1), (3) and (4) catalysts have a BETsurface area of 118 ± 1, 546 ± 5 and 576 ± 5 m2 g−1, respectively.

The low loading of the clusters on the supports did not influencetheir surface area that was originally at 116 m2 g−1 for the aerosilR972 and at 570 m2 g−1 for SBA-15 (Fig. S1). In addition the porediameter evaluated by the BJH method on the desorption curve of

) (2), (d) (3) and (e) (4) and (f) particles size distributions and fits of (3) (gray/black)reader is referred to the web version of the article.)

76 C. Lavenn et al. / Catalysis Today 235 (2014) 72–78

0.0 0.2 0.4 0.6 0.8 1.00

200

400

600

800Q

ua

ntity

Ad

so

rbe

d (

cm

³/g

ST

P)

Relative Pressure (P/Po)

Fot

n(

atirtssrtasAc7ddd(bte

FAbcl

0 48 96 144 192 240

0

20

40

60

80

100

Ep

oxid

e y

ield

(%

)

Time (h)

Fig. 5. Epoxide yield in the aerobic epoxidation of trans-stilbene ofAu25(SC2H4Ph)18@SiO2(R972) (1) (light gray), Au25(SPhNH2)17@SiO2(R972)(2) (gray), Au25(SPhNH2)17@SBA-15 (3) first run (full black circles), second run(empty black circles) and Au25(SPhNH2)17@SBA-15 calcined at 400 ◦C (4) (red). (For

ig. 3. N2 adsorption (fill circles) and desorption (empty circles) isotherms carriedut at 77 K of (1) (gray), (3) (black) and (4) (red). (For interpretation of the referenceso color in this figure legend, the reader is referred to the web version of the article.)

itrogen isotherms is still centered at 8 nm as the original SBA-15Fig. S4).

The supported atomically well-defined thiolate gold clustersre then evaluated as catalysts in the aerobic epoxidation ofrans-stilbene in methylcyclohexane initiated by TBHP. All exper-mental conditions (solvent, temperature, amount of catalyst,adical initiator. . .) are based on previous studies carried out onhis reaction with gold nanoparticles as catalysts [7,22]. The trans-tilbene conversion, the trans-stilbene oxide yield, the catalyticurveys and different catalytic data with the four catalysts areeported in Figs. 4, 5, S6–S11 and Table 2. For thiolate gold clus-ers deposited on Aerosil R972, trans-stilbene conversion is slow,nd starts after an induction period of approx. 12 h. Half conver-ion of trans-stilbene is achieved at approximately 106 h for bothu25(SC2H4Ph)18 (1) and Au25(SPhNH2)17 (2) based catalyst; theorresponding TOF are less than 6 h−1 and the selectivity is approx.0% for both catalysts at this conversion. The average particle sizeetermined after the catalytic test shows that particles are poly-isperse, especially for the phenylethanethiol-stabilized clusters,iameters are between 1 and 7 nm for (1) and 3.6 ± 0.9 nm for (2)

Fig. 6a, b and Table 2). This observation is in good accordance withoth the color of the solution changing from light brown to pink andhe decrease of catalytic activity, which profile becomes almost lin-ar after 50% conversion. On the contrary, Au25(SPhNH2)17@SBA-15

0 48 96 144 192 240

0

20

40

60

80

100

trans

-stilb

en

e c

on

ve

rsio

n (

%)

Time (h)

ig. 4. Trans-stilbene conversion of Au25(SC2H4Ph)18@SiO2(R972) (1) (light gray),u25(SPhNH2)17@SiO2(R972) (2) (gray), Au25(SPhNH2)17@SBA-15 (3) first run (fulllack circles), second run (empty black circles) and Au25(SPhNH2)17@SBA-15 cal-ined at 400 ◦C (4) (red). (For interpretation of the references to color in this figureegend, the reader is referred to the web version of the article.)

interpretation of the references to color in this figure legend, the reader is referredto the web version of the article.)

(3) catalyst exhibits a high catalytic activity after a short inductionperiod of less than 2 h; trans-stilbene is fully converted after 48 h ofreaction with a TOF of 29 h−1 and half conversion is achieved in 12 h.This rate is much higher than that observed over the reference cata-lyst from the World Gold Council (3.5 ± 1.5 nm Au NPs supported onTiO2), for which the time necessary to reach 50% conversion is 80 hwith a TOF of 4 h−1. The activity of the Au25(SPhNH2)17@SBA-15catalyst is comparable to the best heterogeneous catalyst reportedto date for this reaction in the literature [25]. This catalyst, whichconsists in 1.8 ± 0.6 nm gold nanoparticles supported on passivatedsilica, converts 50% of trans-stilbene after ∼12 h with a TOF of29 h−1 (TOF has been recalculated from extracted data for com-parison with our results: 25% conversion at 5 h, 45% conversionat 11 h for 2 �mol of Au), and full conversion is obtained after 50 h[25]. The epoxide selectivity over Au25(SPhNH2)17@SBA-15 (3) cat-alyst is 70%, a value comparable to other catalysts of this studyand similar to that obtained over Au@TiO2 from WGC (Table 2).However catalysts (1), (2) and (3) changed color from gray/brownto pink after 3–6 h of reaction, meaning that the induction periodobserved for these three catalysts could be the result of struc-tural modifications of the clusters, in particular coalescence andgrowth (Fig. S5). These catalyst modifications within the reactionmixture are attributed to the instability of gold thiolate clustersin the oxidant medium. Indeed, when Au25(SPhNH2)17 clustersare dispersed in methylcyclohexane with TBHP at 80 ◦C for 2 h toget similar catalytic medium, the clusters change color from darkbrown to dark blue suggesting their decomposition (see SI for theexperimental conditions). X-ray photoelectron spectroscopy on theclusters before and after their dispersion in the oxidant environ-ment confirms this observation with a drastic evolution of theelectronic state of the elements (Figs. S12–S16): (i) the Au 4f7/2binding energy (BE) shifts from 84.7 to 84.0 eV meaning that goldatoms are all reduced to Au(0); (ii) the S 2p BE moves from 163.2 to168.7 eV pointing out that the sulfur atoms are oxidized into sul-fonate anions; [26] and (iii) the N 1s BE goes from 399.6 to 401.6 eVwhich is attributed to the transformation of the primary amine intoammonium cations [27]. The formation of the sulfonate groups isalso confirmed with the position of the O 1s BE at 531.6 eV. TheC 1s BE at 284.5 eV is attributed to aromatic carbon atoms thatremain intact after immersion in the oxidant medium. Finally the

proposed formation of the −O3SPhNH3

+ zwitterion [28] is consis-tent with the XPS elemental quantification (S, 1; N, 0.80, C, 4.78 andO, 3.06). Consequently, these observations suggest that the clusters

C. Lavenn et al. / Catalysis Today 235 (2014) 72–78 77

F . (3), dfi ize dir the a

acwnass

(aitApsiotTsaecabdtaa3ac1tT

ig. 6. TEM pictures of the catalysts after the epoxidation reaction: a. (1), b. (2), crst (gray/black) and the second run (pink/red), e. TEM picture and f. gold particle seferences to color in this figure legend, the reader is referred to the web version of

re not stable in the TBHP/methylcyclohexane medium used for thisatalytic test and that they decompose into bigger Au(0) particlesith no interacting ligand at the surface. Moreover, the gold coordi-ated aminothiophenolates become amoniophenylsulfonates andre no more interacting with gold particles but may form polymersurrounding the gold nanoparticles or interacting with the silicaurface.

TEM pictures of the catalysts after the catalytic reactionFig. 6) show that clusters deposited on Aerosil R972 appears highly polydisperse gold particles, which is due to the weaknteraction between the clusters and the support, so the forma-ion of bigger particles supports their low catalytic activity. Foru25(SPhNH2)17@SBA-15 (3) catalyst, after the catalytic test fewarticles of ∼10 nm are observed but most of the others remain verymall, with a mean size of 2.2 ± 0.5 nm (Fig. 6). The homogeneityn the particle size can be explained by the good initial dispersionf the clusters and the confinement effect of the mesoporous silicahat protects clusters from further sintering in the oxidant medium.he recyclability of the catalyst (3) has been tested by adding trans-tilbene directly in the reaction medium after the first catalytic runt t = 167 h (see Section 2). As can be seen in Fig. 4, the catalyst isven more active during the second cycle; trans-stilbene is fullyonverted after 24 h (half conversion at 6 h) with a TOF of 43 h−1

nd an epoxide selectivity over 95%. Similar results had alreadyeen reported for the epoxidation of trans-stilbene over Au NPseposited on Gd-doped TiO2 supports [29]. TEM pictures and par-icle size distribution (Fig. 6d) show that the catalyst did not evolvefter the second run, with a particle size of 2.3 ± 0.6 nm. This betterctivity of the second run is related to the presence of more than

mmol of methylcyclohydroperoxides in the medium. A test, in medium already containing methylcyclohydroperoxides without

atalyst, shows that the conversion of trans-stilbene is slow (ca..7% h−1 of conversion), but as soon as the catalyst is reinjected,he catalytic activity enhances with a TOF of 42 h−1 (Fig. S17 andable S1) pointing out the crucial rule of the gold-based catalyst.

. gold particle size distributions and fits after the catalytic tests with (3) after thestribution and fit after the epoxidation reaction with (4). (For interpretation of therticle.)

Catalyst recovered after a catalytic test is also reused after washingsand same catalytic activity has been monitored, suggesting thatthere is no catalyst leaching in the solution and that the catalysis isheterogeneous (Fig. S18 and Table S1).

The effect of the ligands on the activity of the catalyst hasbeen evaluated by carrying out the reaction over a calcinedAu25(SPhNH2)17@SBA-15 (4) catalyst. As can be seen in Figs. 4 and 5,the catalyst is less active after calcination, with a half conversiontime of 72 h and a TOF of 6 h−1. TEM experiments carried out afterthe catalytic reaction show that gold nanoparticles did not grow:from 1.8 ± 0.5 nm after calcination at 400 ◦C, the average particlesize moved to 2.2 ± 0.7 nm (Fig. 6e and f). Consequently the lowcatalytic activity of (4) compared to the non-calcined catalyst (3)can only be explained by the absence of organic molecules. Thisobservation indicates that ligands play a positive role in this cat-alytic reaction. Because the average gold nanoparticle sizes of thenon calcined (3) and calcined (4) catalysts are similar after reaction,the difference in activity suggests that the decomposed based-thiolate ligands in (3) into amoniophenylsulfonates could remainafter reaction as polymers around the particles and may have someelectronic effects on the surface gold atoms as it has been observedfor gold clusters stabilized with PolyVinylPyrrolidone [30]. Anotherhypothesis is the formation of Au(I) species resulting from the step-by-step fragmentation of initial clusters [31]. Indeed modificationsof the catalytic sites during the first hours of reaction is reflectedby an induction period of ca. 3 h for catalyst (3), which does notexist in the case of the calcined catalyst (4) (Figs. S9 and S11). Forcatalysts (1) and (2) (Au25(SC2H4Ph)18 and Au25(SPhNH2)17 sup-ported on SiO2(R972), respectively), longer induction periods areobserved and the activity decreases after 30–40% conversion. Thischange in activity could be related to a more severe degradation

of the active catalyst, which is limited on (3) due to a stabiliza-tion by strong ligands/silanol interactions and a confinement effectof the porous support. High activities in the aerobic oxidation oftrans-stilbene in methylcyclohexane have already been observed

7 sis To

ogiilfoadtf

keccfoca

4

saAA1gcaegetattigndtSttt[

A

taa

[

[

[[

[[[[

[[[

[[[[

[

[

[

[[[

8 C. Lavenn et al. / Cataly

ver octylsilane and octylthiolate-stabilized colloidal dispersion ofold nanoparticles ca. 2 nm diameter [32]. For 1.8 ± 0.6 nm octylth-olate gold nanoparticles the conversion of stilbene reached 50%n ∼7 h. These good catalytic properties were also interpreted as aigand effect that brings geometric hindrance and electronic trans-er on the gold nanoparticles surface. Apart from the sterical effectf the ligands, recent theoretical studies have shown that the Autom in the staple motifs is the major active site for styrene oxi-ation in presence of TBHP as oxidant or initiator [31]. Indeed,he Au atom in the staple motifs can change its valence staterom Au(I) (bicoordinated: 2 SR) to Au(III) (tetracoordinated: 2SR + OH + O2) upon O2 adsorption, which is considered as aey step for the activation of O2 molecules, with a lower activationnergy than that of the Au(0) site. It appears from those reports,oupled with this present study on the decomposition of thiolatelusters in TBHP medium, that extensive work has to be done toully understand the stability/instability of the clusters, the effectf the coordinated/uncoordinated species around the gold parti-les surface and the electronic state of gold atoms on the catalyticctivity.

. Conclusions

This study shows that for the aerobic epoxidation of trans-tilbene in methylcyclohexane, it is really important to select theppropriate functionalized clusters and silica supports, becauseu25(SR)17/18 clusters are not stable in oxidant catalytic conditions.u25(SPhNH2)17@SBA-15 is the best association, because (i) SBA-5 is a mesoporous silica with high surface area that leads to aood mass transfer and acts as a confined medium to stabilize thelusters and (ii) the aminothiophenol functionalized Au25 clustersre well dispersed and stabilized inside the pores due to the pres-nce of hydrogen bonds between the external amino and silanolroups. Consequently even if the clusters are not stable in oxidantnvironment and coalesce at the beginning of the catalytic reac-ion, the gold nanoparticles stabilize at a diameter of 2.2 ± 0.5 nmnd remain stable when the catalyst is reused. The presence ofhe thiolate ligands contributes to the high conversion rate of therans-stilbene, and once removed by calcination, the catalyst activ-ty decreases. Atomically well-defined Au25(SPhNH2)17 clusters areood precursors to design appropriate catalysts due to their exter-al amino groups. Further works will be achieved to prevent theecomposition and coalescence of the clusters and understandhe ligands effect on the aerobic epoxidation of the trans-stilbene.ome solutions could be to decrease the density of clusters insidehe pores of SBA-15 in order to prevent their coalescence [17] oro functionalize the silica surface with mercaptosilane derivateso get strong interactions between the support and the clusters33].

cknowledgments

CL thanks the French ministry for his PhD grant. Authors wish tohank N. Cristin and P. Mascunan for elemental analyses, F. Bosseletnd Y. Aizac for PXRD studies and L. Cardenas for fruitful discussionsnd XPS experiments.

[[

[

day 235 (2014) 72–78

Appendix A. Supplementary data

Supplementary data associated with this article can befound, in the online version, at http://dx.doi.org/10.1016/j.cattod.2014.02.045.

References

[1] M. Haruta, T. Kobayashi, H. Sano, N. Yamada, Chem. Lett. (1987) 405;M. Haruta, N. Yamada, T. Kobayashi, S. Iijima, J. Catal. 115 (1989) 301.

[2] A.S.K. Hashmi, G.J. Hutchings, Angew. Chem. Int. Ed. 45 (2006) 7896;A. Corma, H. Garcia, Chem. Soc. Rev. 37 (2008) 2096;Y. Zhang, X. Cui, F. Shi, Y. Deng, Chem. Rev. 112 (2012) 2467.

[3] M. Stratakis, H. Garcia, Chem. Rev. 112 (2012) 4469.[4] T. Tsukuda, H. Tsunoyama, H. Sakurai, Chem. Asian J. 6 (2011) 736.[5] M.D. Hughes, Y. Xu, P. McMorn, P. Landon, D. Enache, A.F. Carley, G.A. Attard,

G.J. Hutchings, F. King, E.H. Stitt, P. Johnston, K.K. Griffin, C.J. Kiely, Nature 437(2005) 1132;S. Bawaked, N.F. Dummer, D. Bethell, D.W. Knight, G.J. Hutchings, Green Chem.13 (2011) 127.

[6] P. Lignier, F. Morfin, S. Mangematin, L. Massin, J.L. Rousset, V. Caps, Chem.Commun. (2007) 186.

[7] K. Guillois, L. Burel, A. Tuel, V. Caps, Appl. Catal. A 415 (2012) 1.[8] M.Z. Zhu, C.M. Aikens, M.P. Hendrich, R. Gupta, H.F. Qian, G.C. Schatz, R.C. Jin, J.

Am. Chem. Soc. 131 (2009) 2490;R. Jin, Nanoscale 2 (2010) 343;J.F. Parker, C.A. Fields-Zinna, R.W. Murray, Acc. Chem. Res. 43 (2010) 1289;H.F. Qian, M.Z. Zhu, Z.K. Wu, R.C. Jin, Acc. Chem. Res. 45 (2012) 1470;T. Tsukuda, Bull. Chem. Soc. Jpn. 85 (2012) 151;Y. Yu, Q. Yao, Z. Luo, X. Yuan, J.Y. Lee, J. Xie, Nanoscale 5 (2013) 4606;S.H. Yau, O. Varnavski, T. Goodson, Acc. Chem. Res. 46 (2013) 1506.

[9] Y. Zhu, H.F. Qian, R.C. Jin, Chem. Eur. J. 16 (2010) 11455;G. Li, R. Jin, Acc. Chem. Res. 46 (2013) 1749;M. Boronat, A. Leyva-Perez, A. Corma, Acc. Chem. Res. (2013), http://dx.doi.org/10.1021/ar400068w.

10] Y.M. Liu, H. Tsunoyama, T. Akita, T. Tsukuda, Chem. Commun. 46 (2010)550.

11] X. Nie, H. Qian, Q. Ge, H. Xu, R. Jin, ACS Nano 6 (2012) 6014;X. Nie, C. Zeng, X. Ma, H. Qian, Q. Ge, H. Xu, R. Jin, Nanoscale 5 (2013) 5912.

12] G. Li, H. Qian, R. Jin, Nanoscale 4 (2012) 6714.13] A. Shivhare, S.J. Ambrose, H. Zhang, R.W. Purves, R.W.J. Scott, Chem. Commun.

49 (2013) 276.14] Y. Zhu, H.F. Qian, M.Z. Zhu, R.C. Jin, Adv. Mater. 22 (2010) 1915.15] Y. Zhu, H.F. Qian, B.A. Drake, R.C. Jin, Angew. Chem. Int. Ed. 49 (2010) 1295.16] Y. Zhu, G. Chen, R. Jin, Y. Sun, Am. J. Chem. 2 (2012) 18.17] S. Xie, H. Tsunoyama, W. Kurashige, Y. Negishi, T. Tsukuda, ACS Catal. 2 (2012)

1519.18] Gold Bull. 36 (2003) 24.19] Z. Wu, J. Suhan, R. Jin, J. Mater. Chem. 19 (2009) 622.20] C. Lavenn, F. Albrieux, G. Bergeret, R. Chiriac, P. Delichère, A. Tuel, A.

Demessence, Nanoscale 4 (2012) 7334.21] F. Kerdi, V. Caps, A. Tuel, Micropor. Mesopor. Mater. 140 (2011) 89.22] K. Guillois, S. Mangematin, A. Tuel, V. Caps, Catal. Today 203 (2013) 111.23] P.D. Lickiss, Adv. Inorg. Chem. 42 (1995) 147.24] M.W. Heaven, A. Dass, P.S. White, K.M. Holt, R.W. Murray, J. Am. Chem. Soc. 130

(2008) 3754.25] D. Gajan, K. Guillois, P. Delichere, J.-M. Basset, J.-P. Candy, V. Caps, C. Coperet,

A. Lesage, L. Emsley, J. Am. Chem. Soc. 131 (2009) 14667.26] F. Raynal, A. Etcheberry, S. Cavaliere, V. Noël, H. Perez, Appl. Surf. Sci. 252 (2006)

2422.27] P.M. Dietrich, N. Graf, T. Gross, A. Lippitz, S. Krakert, B. Schüpbach, A. Terfort,

W.E.S. Unger, Surf. Interface Anal. 42 (2010) 1184.28] M.A. Leonard, P.J. Squattrito, Acta Cryst. C53 (1997) 648.29] V. Mendez, K. Guillois, S. Daniele, A. Tuel, V. Caps, Dalton Trans. 39 (2010) 8457.30] H. Tsunoyama, N. Ichikuni, H. Sakurai, T. Tsukuda, J. Am. Chem. Soc. 131 (2009)

7086.

31] C. Liu, S. Lin, Y. Pei, X.C. Zeng, J. Am. Chem. Soc. 135 (2013) 18067.32] M. Boualleg, K. Guillois, B. Istria, L. Burel, L. Veyre, J.-M. Basset, C. Thieuleux, V.

Caps, Chem. Commun. 46 (2010) 5361.33] S. Das, A. Goswami, M. Hesari, J.F. Al-Sharab, E. Mikmekova, F. Maran, T. Asefa,

Small (2014), http://dx.doi.org/10.1002/smll.201302854.