Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation

G Model

ARTICLE IN PRESS

CATTOD-9771; No. of Pages 11

Catalysis Today xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Catalysis Today journal homepage: www.elsevier.com/locate/cattod

Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation Fang Jin a,b , Yuanxin Wu a , Shaowen Liu a , Tsung-Han Lin b , Jyh-Fu Lee c , Soofin Cheng b,∗ a Key Laboratory for Green Chemical Process of Ministry of Education, School of Chemical Engineering and Pharmacy, Wuhan Institute of Technology, Wuhan 430073, China b Department of Chemistry, National Taiwan University, Taipei 10617, Taiwan c Research Division, National Synchrotron Radiation Research Center, Hsinchu 300, Taiwan

a r t i c l e

i n f o

Article history: Received 5 May 2015 Received in revised form 13 August 2015 Accepted 20 August 2015 Available online xxx Keywords: MWW Porosity Ti coordination Propylene epoxidation

a b s t r a c t Two types of novel catalysts consisting of gold nanoparticles (NPs) supported on the Ti-incorporated MWW structures were compared. One is Ti-YNU-1 with 0.67 nm micropores between expanded 2DMWW structures, and the other is the titanosilicate pillared MCM-36 (Si/Ti-MCM-36) with 2 nm mesopores. The influences of pore structure and Ti coordination environment on the particle size of anchored Au NPs and their catalytic performances in propylene epoxidation with mixtures of H2 and O2 were investigated. The catalyst materials were characterized with Electron Probe Micro Analysis, X-ray diffraction, nitrogen sorption, UV–visible diffuse reflectance spectroscopy, transmission electron microscopy, and X-ray absorption spectroscopy. The coordination environment of Ti(IV) in Ti-YNU-1 is mainly isolated tetrahedral (Td ), while that in Si/Ti-MCM-36 was found to be a mixture of Td and octahedral (Oh ). The Au loading was proportional to the amount of Ti incorporated in the porous silica. The particle size of Au NPs in the Si/Ti-MCM-36 support was a little larger than that in the Ti-YNU-1 support, and it is attributed to that the highly isolated Td Ti(IV) in Ti-YNU-1 is favorable for attaining the smaller gold NPs, while the higher loading of Ti in Si/Ti-MCM-36 anchored more Au and resulted in the formation of larger Au particles. On the other hand, the mesopores in Si/Ti-MCM-36 facilitated the diffusion of the produced PO in pore structure, inhibited the further reaction for side product formation, and moderated the deactivation of catalysts. © 2015 Elsevier B.V. All rights reserved.

1. Introduction The development of an efficient catalyst for the production of propylene epoxide (PO) is still an industrialization challenging work, since Haruta and coworkers [1] discovered for the first time that propylene epoxidation could be achieved with hydrogen and oxygen on-site generating H2 O2 over gold nanoparticles supported on TiO2 in 1998. PO is a very important feed stock in chemical industry and widely used to produce a variety of derivatives such as polyurethane, propylene glycol, polyester resins and surfactants. The direct expoxidation of propylene with co-reactants hydrogen and oxygen demonstrates an obvious advantage than the conventional ones, such as chlorohydrins, organic peroxide processes, and the direct epoxidation of propene with H2 O2 over TS-1. The new process does not suffer from generating either

∗ Corresponding author. E-mail address: [email protected] (S. Cheng).

environmentally unfriendly chlorinated compounds, co-products with lower demand and/or market values or expensive cost of the oxidant H2 O2 and its handling problems [2]. Therefore, the direct epoxidation of propylene to PO has gained considerable attention in the past seventeen years. For the direct propylene epoxidation over Au–Ti catalysts, it is generally agreed that the size of Au particles and the nature of the titaniumcontaining supports are important factors to the activity of the catalysts. Various titanosilicates, including TiO2 /SiO2 [3], Ti-TUD [4], Ti-SBA-15 [5], Ti-MCM-41 [6], Ti-MCM-48 [7], Ti-MWW [8], Ti-␤ [9], TS-2, TS-1 [10,11], and alkaline treated TS-1 [12] have been investigated for propylene epoxidation. Chowdhury et al. [13] declared that Au particles supported on anatase TiO2 of 2–5 nm in size were responsible for epoxidation, whereas particles smaller than 2 nm were responsible for hydrogenation to form propane and those larger than 5 nm were responsible for propylene combustion. Later, Haruta’s groups proposed that the dominant gold active sites in Au/TS-1 for the PO reaction might be 55 gold atom clusters of ca. 1.4 nm diameter [14]. Recently, Feng et al.

http://dx.doi.org/10.1016/j.cattod.2015.08.041 0920-5861/© 2015 Elsevier B.V. All rights reserved.

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model CATTOD-9771; No. of Pages 11

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

2

[15] identified the size-dependent activity of Au nanoparticles (2–5 nm) deposited on the exterior surface of TS-1 by employing uncalcined TS-1. The catalyst with Au nanoparticles about 2.5 nm was reported to have the best catalyst performance. Although many studies have addressed the issue, the assignment of the gold active sites is still under debate. Moreover, the effect of pore size of support on the epoxidation is much interesting for the researcher. The MCM-22 zeolite (IZA code MWW) is first identified by Leonowicz et al. [16] in 1994 to have the crystalline structure formed by vertically aligning lamellar zeolite intermediate, and the material was patented in 1990 by Rubin and Chu [17] in Mobil. The zeolitic MWW structure comprises two independent 10-membered ring (MR) channels and 12-MR cups on the crystal exterior, which lamellar structure exhibits advantages of high flexibility in varying the pore structures by different post-treatments [18]. In 1988, Bellussi et al. [19] reported the synthesis of a borosilicate zeolite (ERB-1), which is actually isostructural with the MCM-22 aluminosilicate [20]. Furthermore, Wu et al. [21] and Fan et al. [22] synthesized the titanosilicate zeolites, Ti-MWW and TiYNU-1, respectively, using ERB-1 as the precursor. Ti-YNU-1 has the same basic building unit as Ti-MWW with 3D MWW micropore structures but an expanded interlayer window from 10 to 12-MR and an enlarged 2.59 A˚ interlayer spacing [23,24]. Ti-incorporated MWW structures have attracted scientists because of their catalytic properties in partial oxidation reactions [18]. Recently, a new approach for preparing Ti incorporated MCM-36 material (Si/TiMCM-36) was developed in our laboratory by intercalating the ERB-1 with titanosilicate pillars [25]. The generated Si/Ti-MCM-36 has pillared 2 nm mesopores and higher surface area than Ti-YNU-1. This new development provides a very good chance to find the effect of pore structure on the supported Au catalysts on direct propylene epoxidation by comparing the performances of Au/Si/Ti-MCM-36 and Au/Ti-YNU-1, which are both of MWW structure.

relative weight ratio of TPAOH:CTMABr:H2 O = 1:4:330. The mixture was sealed in a flask at 373 K for 16 h. The swollen ERB-1 material recovered by filtration, washing with distilled water, and drying under vacuum was dispersed in a mixture of tetraethylorthosilicate (TEOS) and TBOT with TEOS/TBOT molar ratio of 100:1, and 160:1, then the mixture was stirred at 353 K for 25 h. The material was then filtered and dried at ambient condition. The hydrolyzation of TBOT and TEOS was carried out by suspending the dried solid in an aqueous solution at 313 K for 6 h at pH 9 adjusted by1 M NH3 ·H2 O. Finally, the sample was calcinated at 723 K for 3 h in nitrogen and at 812 K for 6 h in air (heating rate of 2 K min−1 ). The resultant materials were designated as xSi/Ti-MCM-36, where x is the Ti/Si molar ratio in the pillaring gel. Gold was deposited on the supports by a deposition– precipitation (DP) method using sodium hydroxide. Approximately, 0.1 g of HAuCl4 –3H2 O (Alfa Aesar, 99.99%) was dissolved in 25 mL of D.I. water stirring at 313 K, and a 0.1 M NaOH was added to target the final pH of the gold solution to pH 6. The gold solution was nearly colorless, indicating that most of the chloride ligands were replaced by the hydroxyl ligands [28,29]. The neutralized gold solution was further stirred for 2 h at 313 K. After this, 0.35 g of the support was added, and the suspension was stirred. Then 0.1 M NaOH was added to control the pH 7 for about 1 h, and the mixture was stirred for another 2 h at the same temperature and pH value. The solid was filtered and washed with 500 mL of D.I. H2 O, dried under ambient for 12 h. The prepared catalyst precursors were stored in sealed amber bottles at 277 K. Before catalytic performance they were packed into the reactor to reaction. The sample was designated mAu/x-Ti-YNU-1 or mAu/xSi/Ti-MCM36, where m represents a nominal content of m wt% of gold on this catalyst. The m = 1.5 and 4.5 is corresponding to the Au3+ precursor concentration in the DP solution around 1.1 × 10−3 M and 3.3 × 10−3 M.

2.2. Characterization 2. Experimental methods 2.1. Preparation of materials The hydrothermal synthesis of borosilicate MWW structure, i.e., ERB-1 precursor (ERB-1(P)) was initially prepared following the synthesis method reported by Millini et al. [26] with the ratio of Si:B:piperidine (PI):H2 O = 1:0.75:1.4:19. Titanosilicate Ti-YNU-1 and Ti-MWW were prepared by re-hydrothermal synthesis method described in the literature [22,23,27]. This process includes sequential steps for the synthesis of ERB-1 precursor, deboronation with 2 M HNO3 aqueous solution, and post-hydrothermal treatment at 443 K under a rotating condition for 5 days, with highly deboronated ERB-1 zeolite, tetrabutylorthotitanate (TBOT), piperidine (PI) and water in molar composition of Si:Ti:PI:H2 O = 1.0:0.01:1:15. The as-synthesized precursors were then refluxed in 2 M HNO3 solution at 373 K for 6 h with solid to liquid ratio of 1/50 and further calcined at 823 K for 10 h to obtain a lamellar titanosilicate with a structure analogous to the MWW precursor and denoted as x-Ti-YNU-1, where x is the Si/Ti molar ratio in the pillaring gel. Furthermore, another 6 h acid treatment of 100-Ti-YNU-1 with 2 M HNO3 solution at 373 K resulted in the sample A-Ti-YNU-1. An attempt for synthesis of high Ti contents Ti-YNU-1 was also carried out with similar procedure as that 100-Ti-YNU-1, with the post-hydrothermal molar ratio of Si:Ti:PI:H2 O = 1.0:0.03:1:15. The generated sample is denoted as 33-Ti-YNU-1. The synthesis of Si/Ti-MCM-36 started with a swelling step with ERB-1 in the presence of tetrapropylammonium hydroxide (TPAOH) and cetyltrimethylammonium bromide (CTMABr) with a

X-ray powder diffraction patterns were recorded using a PANalytical X’Pert PRO diffractometer with a Cu-K␣ radiation ˚ operated at 40 mA and 45 kV. N2 sorption isotherms ( = 1.5418 A) were measured at 77 K on a Micrometerics TriStar 3000 analyzer after vacuum pretreating the samples at 473 K for 8 h. Total specific surface areas (SBET ) were calculated using the Brunauer–Emmett–Teller (BET) method, and the total pore volume (Vtotal ) was evaluated from N2 uptake at a relative N2 pressure of 0.99. The t-plot was employed to evaluate the volume of micropores (Vmicro ). The corresponding pore size distribution (PSD) was determined using the desorption branches of the isotherms by Barrett–Joyner–Halenda (BJH) method. Diffuse reflectance (DR) UV–vis spectra were recorded using a Hitachi U-3310 spectrometer equipped with an integrating sphere detector, and BaSO4 was the reference. The elemental analyses were determined using Electron Probe Micro Analysis (EPMA) in JEOL JXA-8200 Electron Probe. The EPMA results are determined by Wavelength Dispersive Spectrometer (WDS), which can give an accurate quantitative analysis of element. Transmission electron microscopy (TEM) photographs were obtained from a JEOL JEM-1200EX II Transmission Electron Microscope and Hitachi H-7100 Transmission Electron Microscope. AuLIII -edge X-ray absorption spectra (XAS) were collected using the fluorescence mode at the beam lines 17C of the National Synchrotron Radiation Research Center (NSRRC) facility in Hsinchu, Taiwan. Standard operating condition was 1.5 GeV and 350 mA. The photon energy was guided using a fixed-exit double-crystal Si(1 1 1) monochromator and calibrated with a metallic Au foil (LIII edge, 11,919 eV).

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model

ARTICLE IN PRESS

CATTOD-9771; No. of Pages 11

F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

3

Table 1 The physico-chemical properties of the ERB-1, Ti-YNU-1, MCM-36 and their corresponding Au loaded samples. Sample ERB-1 100-Ti-YNU-1 100Si/Ti-MCM-36 160Si/Ti-MCM-36 4.5Au/33-Ti-YNU-1 4.5Au/100-Ti-YNU-1 4.5Au/A-Ti-YNU-1 1.5Au/100-Ti-YNU-1 1.5Au/100Si/Ti-MCM-36 4.5Au/100Si/Ti-MCM-36d 1.5Au/160Si/Ti-MCM-36d a b c d e

Si/Ti molara e

Na 130 30 94 45 130 160 130 30 30 94

Si/B molara

Aua loading (%)

SBET (m2 g−1 )

Vtotal (cm2 g−1 )

Vmicro b

Vmesco c

57 ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞ ∞

– – – – 0.32 0.28 0.13 0.12 0.19 0.84 0.14

449 554 909 739 Na 400 Na 416 640 522 459

0.31 0.5 0.55 0.5 Na 0.48 Na 0.44 0.45 0.4 0.35

0.18 0.2 0 0.04 Na 0.15 Na 0.15 0 0 0

0.13 0.3 0.55 0.46 Na 0.34 Na 0.29 0.45 0.4 0.35

Elemental analysis from EPMA. Vmicro for ERB-1 and Ti-YNU-1 from t-plot and that for MCM-36 is calculated by Vtotal − Vmeso. Vmeso for the ERB-1 and Ti-YNU-1 is calculated by Vtotal − Vmicro and that for MCM-36 from t-plot. Spent catalyst. Na = not applicable.

The X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) data were processed using the programs of IFEFFIT [30]. The EXAFS spectra with k3 -weighted functions were Fourier transformed (FT) over a photoelectron momentum (k) range of 2.5–12.1 A˚ −1 . Structural parameters were extracted from the first coordination shell, which was isolated by inverse transforming over a nonphase-corrected ˚ The spectra were then radial distance range (r-space) of 1.8–3.4 A. fitted to the EXAFS equation [31]. The amplitude reduction factor (S0 2 ) value of 0.86 were determined from the reference standard (unirradiated Au foil) and kept constant for fitting all spectra. The interatomic distance R, the coordination number of first nearest neighbor (1NN), the difference of the Debye–Waller factor from the reference ( 2 ), and the correction of the threshold energy (E) were treated as free parameters during the fitting. Fitting analysis in both k1 - and k3 -weighted Fourier transforms was applied in order to obtain a unique set of coordination number (CN) and  2 parameters. The photoelectron scattering-path amplitudes and phases were calculated by ab initio using FEFF8 [32] for a face-centered-cubic (fcc) structure of Au with a first nearest neigh˚ The size dependent average CN of atoms bor distance of 2.885 A. in a 12-fold coordinated   spherical particle can be estimated by 3 CN = 12 · 1 = 2D R1NN , where R1NN is the nearest neighbor distance and D is the average diameter of nanoparticles (NP) [33–36].

2.3. Catalytic reaction A flow quartz tubular microreactor (inner diameter 6 mm, wall thickness = 1.5 mm) equipped with an axial quartz thermocouple well (2 mm O.D.), which allowed monitoring of the catalyst bed temperature, was used to determine the catalytic performance of the different catalysts prepared. Typically, the epoxidation of propylene was carried out under atmosphere pressure, using 0.25 g of catalyst without dilution and a gas feed rate of 28 cm3 min−1 in total, with a feed concentration of 10/10/10/70 vol.% of He (99.9%), O2 (99.9%), H2 (99.9%), and C3 H6 (99.8%). The corresponding space velocity was about 7000 cm3 h−1 g−1 cat. Before reaction, the catalyst was pretreated with a mixture of 10 vol% H2 in He (28 cm3 min−1 ), followed by 10 vol% O2 in He (28 cm3 min−1 ) at 523 K for 0.5 h, after which the temperature was set to 473 K under He flow and the reactants were introduced. Reaction products were analyzed online by one gas chromatograph (Shimadzu GC-14B), containing a CP-wax 52CB and Carboxen-1000 column in two separate channels, with a flame ionization detector (FID) and a thermal conductivity detector (TCD), respectively. The CP-wax 52CB capillary column (0.53 mm × 50 m) was used to detect oxygenates (i.e., acetaldehyde (AA), PO, acetone

(ACE), propionaldehyde (PA), acrolein (ACR)), respectively, whereas the Carboxen-1000 column were used to analyze H2 , O2 , and CO2 and propylene, respectively. The propylene conversion, PO selectivity, H2 conversion, and H2 selectivity were defined as follows: Propylene conversion (XC3H6 ) = mol of (oxygenates + CO2 /3)/mol of propylene in feed. Product selectivity (SN ) = mol of corresponding product/mol of (oxygenates + CO2 /3). N = PO, or oxygenates side product, i.e., AA, ACE, PA, and ACR. H2 conversion (XH2 ) = mol of H2 reacted/moles of H2 in feed. H2 selectivity (SH2 ) = mol of PO/moles of H2 reacted. 3. Results 3.1. Elemental analysis The ERB-1 discovered by Bellussi et al. [37] is actually an borosilicate isostructural with the MWW structure. The EPMA analysis results in Table 1 show that for the YNU system, the 2 M HNO3 aqueous solution treatment can expel almost all the boron from ERB-1. The 100-Ti-YNU-1 and 33-Ti-YNU-1 samples prepared by post-hydrothermal treatment exhibit a final Si/Ti molar ratios close to the target loadings. Further acid treatment of 100-Ti-YNU-1 decreases the Ti content. The pillared 100Si/Ti-MCM-36 also does not contain boron anymore. Moreover, the post-hydrothermal synthesized Ti-YNU-1 materials have lower titanium contents in the final products, in comparison with those of the pillared MCM-36 materials. Table 1 also summarizes the Au loadings of the supported Au catalysts. For Au-loaded Ti-YNU-1 samples, the actual Au loading decreases with the concentration of the Au precursor used in DP solution. Au loading of 0.28 wt% was obtained with 3.3 mM Au3+ solution (4.5Au/100-Ti-YNU-1) and 0.12 wt% loading with 1.1 mM Au3+ solution (1.5Au/100-Ti-YNU-1). When using identical concentrations of Au precursor for DP, the Au loading on xTi-YNU-1 was found to change with the Ti content in the support. For the Ti-YNU-1 samples prepared with same 3.3 mM concentrations of Au precursor, the Au loading decreases slightly with the decrease of Ti content in Ti-YNU-1, i.e., 0.32% Au for 4.5Au/33-Ti-YNU-1, 0.28% for 4.5Au/100-Ti-YNU-1, and 0.13% for 4.5Au/A-Ti-YNU-1. The A-Ti-YNU-1 has the lowest Ti content among the three Ti-YNU-1 samples after post-treatment of 100-TiYNU-1 with 2 M HNO3 . This is consistent with the results reported by other groups that higher Ti content leads to higher Au loading [8,38]. However, it was noticed that to load Au on Ti-YNU-1

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model CATTOD-9771; No. of Pages 11

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

4

Fig. 1. (A) Small angle and (B) wide angle XRD patterns of (a) ERB-1(P), (b) 33-TiYNU-1, (c) 100-Ti-YNU-1, (d) A-Ti-YNU-1, (e) 100Si/Ti-MCM-36 and (f) 160Si/TiMCM-36.

support was more difficult than on Si/Ti-MCM-36. Au is almost not loaded on Ti-YNU-1 while 0.12 wt% can be loaded on Si/Ti-MCM-36 with 0.59 mM Au3+ solution (not shown in Table 1). Even the Au3+ precursor concentration is increased to 3.3 × 10−3 M, the actual Au loadings 0.32 wt% on Ti-YNU-1 samples is still much lower than 0.84 wt% on Si/Ti-MCM-36. It is attributed to that Ti-YNU-1 has lower Ti contents than Si/Ti-MCM-36, and the presence of Ti has been reported to be helpful in capturing the Au precursor during the DP process, which is originated from a certain affinity interaction between Ti species and Au precursor [39]. 3.2. X-ray diffraction The small- and wide-angle XRD patterns of the samples are shown in Fig. 1(A) and (B), respectively. The ERB-1 precursor shows the diffraction (0 0 1) at about 3◦ which is attributed to the stacking of layered MWW structure along the c-direction with a distance ˚ The XRD pattern of 100-Ti-YNU-1 appears to be similar to of 27 A. that of the lamellar precursor. The retention of (0 0 1) and (0 0 2) peaks at 3◦ and 6.4◦ indicates the preservation of original space between the MWW sheets of the ERB-1(P) after a sequential acid, post-hydrothermal, and calcination treatment. This phenomenon is consistent with that reported in the literature that the interlayer space in Ti-YNU-1 is actually enlarged by about 2.5 and 0.6 A˚ relative to those of interlayer condensed 3D Ti-MWW and the lamellar precursor, respectively [23]. The cell expansion of Ti-YNU-1 was explained by the generation of two additional T sites in the interlayer space between the MWW sheets and the formation of a new pore window of 12 MR, which is distinct from the interlayer 10 MR of the 3D-MWW structure. On the other hand, the positions of intralayer diffractions (hk0), such as the prominent narrow peaks (1 0 0), (2 2 0), and (3 1 0) at about 7.1◦ , 25.1◦ and 26.1◦ , in the pillared samples Ti-YNU-1 and Ti-MWW are conspicuously unchanged, in comparison with those of ERB-1(P). These results indicate that the 2D-MWW structure in the ab plane remains invariant during the pillaring processes. Further acid treatment of 100-Ti-YNU-1 has little influence on the XRD pattern, indicating that A-Ti-YNU-1 still keeps the slightly expanded layered structures. However, for the 33-Ti-YNU-1 the (0 0 1) diffraction disappears, and the (0 0 2) diffraction shifts to 7.0◦ and merges with the (1 0 0) peak at 7.1◦ , which is the characteristic peak of the 3D Ti-MWW. These results are in good agreement with the report in literature that Ti-YNU-1 could only be prepared with low Ti contents, and further increase of the Ti content in Ti-YNU-1 led to the formation of Ti-MWW structure [40].

Fig. 2. XRD patterns of (a) 100-Ti-YNU-1, (b) 4.5Au/A-Ti-YNU-1, (c) 4.5Au/100Ti-YNU-1, (d) 4.5Au/33-Ti-YNU-1, (e) 1.5Au/100-Ti-YNU-1 and (f) 1.5Au/100Si/TiMCM-36.

The xSi/Ti-MCM-36 samples show the (0 0 1) reflection shifts ˚ down to around 2◦ , corresponding to an increased d-spacing of 44 A. Meanwhile, the (0 0 2) diffraction at 6.5◦ of the precursors almost disappears. These results indicate that a successful expanding the inter-layer space of the ERB-1 with generated 20 A˚ mesopores [25]. After loading Au on Ti-YNU-1, all the (0 0 2) peaks at ca. 6.4◦ for the Ti-YNU-1 samples shift to higher position and merge with the (1 0 0) peak as shown in Fig. 2. That indicates the shrinkage of interlayer space and probable conversion of Ti-YNU-1 structure to 3D Ti-MWW. However, the incorporation of Au did not significantly change the structure of Si/Ti-MCM-36 framework. Moreover, the diffraction peaks of Au metal at 38.2◦ and 44.4◦ are not seen in the XRD patterns of Au loaded Ti-YNU-1, Ti-MWW and 100Si/Ti-MCM36 (Fig. 2), implying that the Au particle sizes are probably too small to be visible in the XRD patterns. 3.3. N2 -sorption measurements The nitrogen sorption isotherms of ERB-1, Ti-YNU-1 catalysts are shown in Fig. 3. While those of 100Si/Ti-MCM-36 and Au loaded catalysts are shown in Fig. 4A, and the related pore size distribution (PSD) curves calculated from the desorption branches of isotherms using BJH method [41] are shown in Fig. 4B. The calculated SBET , Vtotal and Vmesop are given in Table 1. The sorption isotherm of ERB-1 corresponds to a Type I isotherm, which is the characteristic of microporosity. The isotherm of x-Ti-YNU-1 basically exhibits a

Fig. 3. N2 adsorption/desorption isotherms of (a) 100-Ti-YNU-1, (b) 4.5Au/100-TiYNU-1, (c) 1.5Au/100-Ti-YNU-1 and (d) ERB-1.

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model CATTOD-9771; No. of Pages 11

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

Fig. 4. (A) N2 adsorption/desorption isotherms (B) BJH pore size distribution of (a) 100Si/Ti-MCM-36, (b) 1.5Au/100Si/Ti-MCM-36, (c) spent1.5Au/100Si/Ti-MCM-36 and (d) ERB-1.

type I curve due to its microporous nature. However, a very small H2-type hysteresis loop appears at relative pressure larger than 0.8, suggesting the existence of mesopores probably coming from the stacking of the particles. On the other hand, the MCM-36 samples show Type IV isotherms with an abrupt increase in the adsorption volume at relative pressures of 0.2–0.4, indicating the presence of mesopores. That range of capillary condensation corresponds to mesopores of average diameter about 20 A˚ in the PSD profiles shown in Fig. 4(B). The broad distribution of pore diameters infers the pillaring spaces are not very uniform. The introduction of Au nanoparticles on Ti-YNU-1 and Si/TiMCM-36 resulted in an obvious decrease of the N2 uptakes in the isotherms. The SBET , Vtotal and Vmesop were found to decrease markedly for Si/Ti-MCM-36, but to a less extent for Ti-YNU-1. Since XRD studies showed that the mesoporous structures of Si/Ti-MCM36 were not destroyed during Au loading processes, the decrease in surface area and pore volume should be due to the partial pore blocking by Au nanoparticles. 3.4. UV–vis diffuse reflectance spectroscopy The Ti-incorporated supports with different Si/Ti ratios were examined with DR UV–vis analysis in order to understand the coordination geometry of the Ti(IV) cations. The spectra of Ti-YNU-1 and Si/Ti-MCM-36 materials are shown in Fig. 5. The Ti-YNU-1 samples prepared by the post-hydrothermal and acid treatment have only one UV band at around 220 nm. The 100Si/Ti-MCM-36 has two peaks at 220 and 260 nm. According to Chen et al. [42], the 220 nm band is assigned to charge transfer from O2− to Ti4+ of the tetrahedral (Td ) coordination, while the 260 nm band is attributed to charge transfer from O2− to Ti4+ in octahedral (Oh ) coordination. Therefore, the Ti in the Ti-YNU-1 samples are all assigned to the Ti(IV) sites in Td environment, while the Ti in the Si/Ti-MCM-36 contains both the Td and Oh coordination state. After supporting Au on Ti-YNU-1 and Si/Ti-MCM-36 by the DP method, the materials have the color of white. However, the used catalysts after PO synthesis reaction display different colors, such as blue, pink and brown, and it is a result of the presence of Au nanoparticles (NPs) of different amounts and particles sizes. These results indicate that Au was reduced to metallic state during the pretreatment and catalytic reaction, although it was originally deposited on the supports in the hydroxide or oxide form. Only the sample with larger amounts of the Au loadings, i.e., the 4.5Au/100Si/Ti-MCM-36 has an obvious absorption band in the visible region around 500–600 nm as shown in UV–vis spectra Fig. 6.

5

Fig. 5. DR UV–vis spectra of (a) 33-Ti-YNU-1, (b) 100-Ti-YNU-1 and, (c) A-Ti-YNU-1, (d) 100Si/Ti-MCM-36 and (e) 160Si/Ti-MCM-36.

This characteristic absorption band was due to surface plasmon vibrations of gold, and the wavenumbers are dependent on the gold particle size [43]. All other Au loaded samples exhibit none characteristic absorption band in the visible region around 500–600 nm, which indicates the loaded Au NPs is invisible by the UV–vis diffuse reflectance spectroscopy. 3.5. Au LIII -edge XAS spectroscopy EXAFS analysis of the Au LIII -edge XAS spectra of spent catalysts was carried out to determine the CNs of Au atoms and thus to calculate the sizes of the Au NPs. The Fourier transforms (FT) of Au LIII -edge k3 -weighted EXAFS spectra of the catalysts Au/xTi-YNU-1and Au/100Si/Ti-MCM-36 are shown in Fig. 7, and the k3 -weighted EXAFS spectra and the curve fitting results are summarized in Figs. S2 and S3 (supporting information). These results are compared with the spectrum of the Au foil. The refined fitting parameters are given in Table 2. The obtained coordination number can be related to a mean Au particle size by assigning a geometrical model for the Au particle structure. Assuming spherical particles and using models of fcc Au clusters [5,44], the mean particle sizes for the samples are calculated and listed in Table 2.

Fig. 6. DR UV–vis spectra of (a) 4.5Au/A-Ti-YNU-1, (b) 4.5Au/100-Ti-YNU-1, (c) 4.5Au/33-Ti-YNU-1, (d) 1.5Au/100-Ti-YNU-1, (e) 1.5Au/100Si/Ti-MCM-36, (f) 4.5Au/100Si/Ti-MCM-36 and (g) 1.5Au/160Si/Ti-MCM-36.

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model

ARTICLE IN PRESS

CATTOD-9771; No. of Pages 11

F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

6

Table 2 Refined fitting parameters from EXAFS analysis of the first coordination shell of Au LIII -Edge for different Au/TiMCM-36 samples and Au/x-Ti-YNU-1compared with Au foil ˚ r = 0−6.0 A). ˚ reference materials (with k3 weighting: k = 2.5–12.1 A; Sample

Ave.CN

R (Å)

S0 2 (Set)

 2 (Å2 )

E (ev)

Particle size estimate (nm)

Gold foil 4.5Au/33-Ti-YNU-1 4.5Au/100-Ti-YNU-1 4.5Au/A-Ti-YNU-1 1.5Au/100-Ti-YNU-1 1.5Au/100Si/Ti-MCM-36 4.5Au/100Si/Ti-MCM-36 1.5Au/160Si/Ti-MCM-36

12 8.6 7.3 7.2 5.45 7.9 9.88 6.89

2.86 2.84 2.83 2.81 2.86 2.82 2.86 2.82

0.86 0.86 0.86 0.86 0.86 0.86 0.86 0.86

0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01

4.62 5.6 5.48 4.83 4.34 4.72 5.99 5.28

– 1.48 1.41 1.04 0.79 1.25 2.43 0.99

CN, R, S0 2 ,  2 and E are the coordination number, Au Au bond distance, amplitude reduction factor, Debye–Waller factor and energy correction number, respectively. Due to correlation problems the Debye–Waller factors reported for the samples were extracted from fitting the foil data and were not refined. ˚ for  2 , ±20%; and for E, ±20%. The errors in the final parameters are expected to be for CN, ±10%; for R, ±0.02 A;

The spectra and calculated results of a series of catalysts of 100Ti-YNU-1 with the decreasing of Au loading from 0.87 wt%, 0.28 wt% to 0.12 wt% shows that the smaller amount of Au supported, the smaller Au NPs generated, as indicated by the weaker peak intensities and the correspondingly smaller coordination numbers. Further proof for these observations can be seen by comparing the first shell near-neighbor distances, where a shorter average 1NN Au–Au distance was observed in the catalysts with relative lower actual Au amount [5]. For the catalyst of 4.5Au/x-Ti-YNU-1, the Au-Au FT magnitudes peak intensities increase with increasing Ti content. This proof indicates that the lower Ti incorporated sample has the smallest gold particles, which is consistent with the variation tendency of the calculated CN and R value. With similar amount of actually loaded Au content, the Au NPS in 1.5Au/100-Ti-YNU-1 is much smaller than that of 1.5Au/100Si/Ti-MCM-36. It has been reported that the Au particles are smaller on average in the hydrothermally prepared Ti-SBA-15 than in the corresponding grafted sample, because the grafted sample contains more Ti content in Oh coordination [5]. Similar reason may explain the difference of size between Au NPs supported on Si/Ti-MCM-36 and Ti-YNU-1, because the former samples contain more Oh coordination Ti than the latter.

Fig. 7. EXAFS of the Au LIII -edge phase-uncorrected Fourier transform spectra of the (k)*k3 weighted in R space after applying a window around the first shell from 1.6 to ˚ (a) Gold foil, (b) 4.5Au/33-Ti-YNU-1(vertical value multiple −1), (c) 4.5Au/1003.6 A. Ti-YNU-1(vertical value multiple −1), (d) 4.5Au/A-Ti-YNU-1(vertical value multiple −1), (e) 1.5Au/100-Ti-YNU-1, (f) 1.5Au/100Si/Ti-MCM-36, (g) 4.5Au/100Si/Ti-MCM36 and (h) 1.5Au/160Si/Ti-MCM-36(vertical value multiple −1).

3.6. Transmission electron microscopy Fig. S4 (supporting information) shows the TEM images of samples, namely Au supported on Ti-YNU-1 with different Ti contents prepared by NaOH as the neutralization agents. The size distribution of the gold NPs analyzed by DigitalMicrographDemo software are also shown. The series catalyst of 4.5Au/x-Ti-YNU-1 exhibit that the average sizes of gold NPs decrease with the decrease of titanium loading, as shown in Figs. 8(a–c). The average gold NPs sizes are about 2 nm in 33-Ti-YNU-1, and 1.5 nm in 100-Ti-YNU1 and A-Ti-YNU-1. These results are consistent with the literature reports in [3,5]. With the comparison of Fig. 8(a–d) and (e and f), it is also noticed that the Au NPs supported on Ti-YNU-1 have smaller average particle size than those supported on Si/Ti-MCM-36.

3.7. Catalytic performance The Au/x-Ti-YNU-1 and 1.5Au/100Si/Ti-MCM-36 catalysts were applied for the direct propene epoxidation with O2 and H2 in a plug-flow system for about 6 h. A rapid deactivation was observed in the first 2 h, and then a stable performance followed (Fig. 9). Therefore, the catalytic activities at time on stream (TOS) of 10 min and the average in 6 h are summarized in Table 3. The catalytic activities are expressed in terms of propene conversion (XC3H6 ), PO selectivity (SPO ) and side products selectivities, PO yield, hydrogen conversion (XH2 ), and hydrogen selectivity (SH2 ). Two PO formation rates are shown. One is the specific rate or yield per hour based on per kg of catalyst and are called PO formation rate. The other PO rate is turnover Rate (TOR) based on per gram of gold, and that can demonstrate the Au efficiency [3]. The catalysts 4.5Au/x-Ti-YNU-1 exhibit tendency that the actual Au content and the corresponding NPs size increase with the increase of Ti content. The catalytic activity data, such as PO formation rate, TOR, XC3H6 and XH2 increase steadily as the Si/Ti ratio raises from 45, 130 to 160 for the 4.5Au/x-Ti-YNU-1 catalysts. However, the SPO and SH2 values reach the maxima at Si/Ti ratio of 130. The TOR results indicate that catalyst 4.5Au/A-Ti-YNU-1 with the lowest Au loading and smallest NPs size in the three Au/Ti-YNU-1 samples is the most active catalyst. The further decrease of the Au loading amount and Au NPs causes the decrease of catalytic performance as shown in the catalytic results of 1.5Au/100-Ti-YNU-1. Over the 100Si/Ti-MCM-36 supported Au catalysts, the maximum Au efficiency and SH2 reach 14.6 gPO h−1 gAu −1 and 9.7% over 1.5Au/100Si/Ti-MCM-36 catalyst, and the PO formation rate over this catalyst (27.7 gPO h−1 kgcat −1 ) is close to that over 4.5Au/A-TiYNU-1 (30.0 gPO h−1 kgcat −1 ). In the reaction temperatures of 433–498 K, it is found in Table 3 that the XC3H6 and XH2 increase but SPO and SH2 decrease with increasing the reaction temperature over almost all the catalysts. The time on stream curves in Fig. 10 show that the long-term PO

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model

CATTOD-9771; No. of Pages 11

Catalyst

45

Au wt%

1.5Au/100-TiYNU-1

130

0.12

30

0.19

1.5Au/100 Si/Ti-MCM-36

30

0.84

0.14

(9.3) 6.0 4.1

(2.8) 1.8 1.3

(0) 0 0

(12.6) 22.3 19.2

(16.7) 26.7 38.1

(11.1) 4.3 3.9

(28.5) 15.5 15.2

(10.2) 5.5 5.4

(9.7) 16.6

(0) 0

(26.2) 29.2

(36.2) 30.7

(5.4) 3.2

(30.0) 15.0

(23.1) 11.5

(21.8) 18.2

(6.1) 9.4

(0) 0

(14.7) 14.2

(14.2) 12.3

(8.0) 8.8

(17.4) 16.4

(14.5) 13.0

(82.2) 86.6 90.7 67. 5

(1.4) 0.2 0.3 2.6

(0) 0 0 0.9

(0) 0 0 0

(16.3) 13.1 9.0 29.1

(18.5) 11.4 9.0 17.7

(9.7) 12. 8 16.7 6.2

(27.7) 21.4 20.5 16.9

(14.6) 11.3 10.8 8.9

(0.9) 0.7 0.6 0.6

(84.9) 93.0 94.5 79.4

(0) 0 0 5.0

(4.6) 0.9 1.1 3.6

(0) 0 0 0

(10.2) 6.0 4.3 12.0

(19.6) 14.5 14.8 18.8

(3.8) 4.7 3.5 2.4

(13.9) 10.3 7.9 6.8

(1.7) 1.2 0.9 0.8

(2.2) 1.0 0.8

(43.6) 41.4 57.0

(18.5) 5.7 4.6

(0) 0 0

(0) 0 0

(38.1) 47.0 38.4

(28.3) 18.4 13.9

(3.4) 2.0 2.6

(14.9) 6.1 6.0

(4.8) 2.0 1.9

ACE

CO2

(0) 19.3 15.6

(35.9) 24.8 25.9

(0) 0 0

(1.9) 1.0 1.0

(2.3) 1.6 2.1

(80.1) 63.1 48.6

(0) 0 5.17

(7.4) 14.7 27.1

498

(2.0) 1.0

(3.3) 2.0

(59.3) 49.4

(4.9) 4.8

498

(1.1) 1.1

(2.0) 1.9

(57.4) 58.3

453

(1.8) 1.4 1.3 1.1

(2.2) 1.6 1.5 1.6

(0.9) 0.7 0.5 0.4 (1.0) 0.4 0.4

453 433 473

1.5Au/160Si/Ti- 94 MCM-36

(4.2) 2.1 2.1

ACR

(51.6) 33.3 49.6

433 473 4.5Au/100 Si/Ti-MCM-36

(14.4) 20.3 12.5

PA

(1.2) 1.2 0.6

523 0.13

(12.5) 22.7 8.9

PO (0.6) 0.39 0.27

0.28

160

TOR (gPO h−1 gAu−1)

498 0.32

4.5Au/A-TiYNU-1

PO rate (g h−1 kg ca−1 )

Propene conv. (%)

498 130

H2 select. of PO (%)

PO yield (%)

473 4.5Au/100-TiYNU-1

H2 conv. (%)

T (K)

453 433

Selectivity (%)

Reaction condition: 0.25 g catalysts, 433–498 K, space velocity: 7000 mL min−1 gcat −1 ; PO: propylene oxide, PA: propanal, ACR: acrolein, ACE: acetone. The catalytic performance data in the bracket are collected after 10 min of reaction, the other data are from the average of reaction, the other data are from the average of 6 h of reaction.

ARTICLE IN PRESS

4.5Au/33-TiYNU-1

Si/Ti atom ratio

F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

Table 3 Catalytic performance over different catalysts of Au/x-Ti-YNU-1 and Au/100Si/Ti-MCM-36.

7

G Model CATTOD-9771; No. of Pages 11

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

8

Fig. 8. The size distributions of gold nanoparticles of (a) 4.5Au/33-Ti-YNU-1, (b) 4.5Au/100-Ti-YNU-1, (c) 4.5Au/A-Ti-YNU-1, (d) 1.5Au/100-Ti-YNU-1, (e) 1.5Au/100Si/TiMCM-36 and (f) 4.5Au/100Si/Ti-MCM-36 from TEM analysis.

formation rate of samples 4.5Au/100-Ti-YNU-1, 4.5Au/A-Ti-YNU-1, 1.5Au/160Si/Ti-MCM-36 and 1.5Au/100Si/Ti-MCM-36 as a function of reaction temperature and reaction time. Although there was an initial decay in the PO formation rate in the first few hours, based on the PO formation rate the optimal reaction temperature is 498 K over Ti-YNU-1 supported Au catalysts, while that for the Si/TiMCM-36 supported Au catalysts is a lower temperature of 453 K. The other initial and average PO formation rates for different samples at different temperatures are available in Table 3. The optimal reaction temperature for Au/x-Ti-YNU-1 is 45 K higher than that for Au/xSi/Ti-MCM-36. 4. Discussion 4.1. Effect of support on the Au loadings and particle size for PO synthesis The Au particle sizes estimated from the 1NN coordination numbers of Au LIII -edge EXAFS suggest that the average Au particles are ranged from 1 to 6 nm, which are slightly lower than the

average sizes determined from TEM images. It has been reported that the Au NPs with size smaller than 1 nm are invisible in the TEM images [11,14,28]. In comparison, EXAFS is a more sensitive technique to determine the average size of poly-dispersed Au particles, especially toward diameter smaller than 1 nm [4]. Indeed, despite of the higher sensitivity of EXFAS technique toward smaller particles, TEM and EXAFS results of Au particle sizes appear to be in accordance and some correlation between each other. The Au NPs size is considered to be one of the critical factors to influence the catalytic performance in propylene oxidation [4]. Gold is believed to firstly catalyze the reaction of O2 /H2 to form hydrogen peroxide, which then migrates to Ti(IV) sites on the supports to form titanium hydroperoxide species, and the latter react with propylene to form epoxide [1,3,45,46,47]. For 4.5Au/xTi-YNU-1 catalysts, the actual Au loading and the corresponding Au NPs size decrease with the Ti content in the supports. The PO formation rate and Au efficiency of 4.5Au/x-Ti-YNU-1 catalysts increase in the order of 4.5Au/33-Ti-YNU-1 < 4.5Au/100-Ti-YNU1 < 4.5Au/A-Ti-YNU-1. According to the calculated NPs size from EXAFS in Table 2, Au NPs size of the most efficient catalyst in this

Fig. 9. The PO formation rate as a function of reaction temperature and reaction time for the samples (A) 4.5Au/100-Ti-YNU-1 and 4.5Au/A-Ti-YNU-1; (B) 1.5Au/160Si/TiMCM-36 and 1.5Au/100Si/Ti-MCM-36.

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model CATTOD-9771; No. of Pages 11

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

9

are a little smaller than that supported in Si/Ti-MCM-36. From the UV–vis spectra in Fig. 5 it can be found that Si/Ti-MCM-36 materials contain small amount of Oh Ti, while the Ti species in Ti-YNU-1 are absolutely Td Ti. It can be concluded that the highly isolated Td Ti centers in Ti-YNU-1 are favorable for attaining the smaller gold NPs, while the higher Ti contents in Si/Ti-MCM-36 can anchor more Au and result in the formation of larger particles. 4.2. The effects of supports on PO synthesis reaction temperature

Fig. 10. Time on stream relative formation rate of PO (PO formation rate/initial PO formation rate) during a 6 h catalytic performance for the (a) 1.5Au/100Si/Ti-MCM36 under 453 K and (b) 4.5Au/A-Ti-YNU-1 under 498 K.

series is 1.04 nm. On the other hand, it has been confirmed that for the 1.5Au/xSi/Ti-MCM-36 series with different Si/Ti ratio, the most active catalyst 1.5Au/100Si/Ti-MCM-36 has Au NPs size of about 1.25 nm, which is in between 0.99 nm of 1.5Au/160Si/Ti-MCM-36 and 1.66 nm of 1.5Au/40Si/Ti-MCM-36. In consideration of catalysts 1.5Au/160Si/Ti-MCM-36 and 4.5Au/A-Ti-YNU-1, having close Au loadings (0.14% vs. 0.13%) and particle sizes (0.99 nm vs. 1.04 nm), the catalytic activity of the former is far inferior to the latter. These phenomena are the evidence that the most appropriate size of average Au NPs for propylene epoxidation varies with the supports, and that supported on Ti-YNU-1 is a little smaller than that supported on Si/Ti-MCM-36. Many studies have addressed the Au particle size effect, yet the issue is still under debate. Gold nanoparticles with sizes in 2–5 nm were originally assigned as the active sites for the PO reaction in Au/TiO2 and other Au/Ti-based oxides by Haruta’s groups [1,48]. On the other hand, Delgass’s group [10,11,28,49] suggested that Au clusters smaller than 1 nm inside the nano-channels of TS-1 could contribute significantly to epoxidation reaction. They also suggested that the sub-nanometer Au particles, not visible by TEM, in hydrothermally synthesized Ti-SBA-15 are most active species [50]. Based on the density functional theory (DFT) calculation, Barton and Podkolzin [51] concluded that gold clusters of 0.7 nm (corresponding to Au13 clusters) supported on microporous TS-1 zeolite are the most active in producing the hydroperoxyl intermediate. Oyama et al. [4] reported that very small Au particles (about 1 nm) supported on mesoporous Ti-TUD synthesized by hydrothermal reaction are the most active for epoxidation. On the other hand, Nijhuis’s group [3,5] reported that Au NPs supported on different supports had different optimum sizes to give best activity in direct epoxidation of propylene. The Au average NPs in the range of 1.5–2.8 nm supported on Ti-SBA-15 prepared by Ti grafting method [5] and Au NPs of sub-nanometer to 6 nm supported on the Ti-grafted silica [3] are most active. Yang et al. [52] found that the 1.5 nm Au NPs in Au/TiO2 supported on SBA-15 prepared by grafting method are most active in their investigation. It is interesting to find that the most active Au NPs are larger in size when Ti in the supports coexisted in Oh and Td coordinations, most of which is incorporated in the mesopores by grafting method. In contrast, the active Au NPs are smaller in the supports containing absolutely isolated Td Ti, which are almost prepared by the hydrothermal synthesis. These results are in consistence with our observation that the most active Au NPs supported on Ti-YNU-1

The coordination environment of incorporated Ti in the supports has great influence on the loading and particle size of the anchored Au, as well as its catalytic performance. The optimal reaction temperature of Au/x-Ti-YNU-1 is 45 K higher than that of Au/xSi/Ti-MCM-36. Similar phenomenon of the change in catalytic temperature of Au on different supports for PO formation was also reported in the literature [2]. The propylene epoxidation reaction should be carried out at temperatures below 373 K over the Au/TiO2 catalysts, because PO is favor to further oxidization to ACR, CO2 and H2 O at higher temperatures. However, when replacing TiO2 with highly dispersed TiO2 on SiO2 surfaces or TS-1, the operating temperature could be raised to 423−473 K [2]. The difference of the optimal reaction temperature between Au/x-Ti-YNU-1 and Au/xSi/Ti-MCM-36 catalysts is evidence for the influence of Ti species in the supports on the catalytic performance. It can be deduced that the catalyst with Ti of sole Td coordination is favored at higher reaction temperature for PO formation than that with Ti of Oh coordination also coexisted. 4.3. The effects of supports on catalyst stability for PO synthesis The Si/Ti-MCM-36 is a titanosilicate-pillared MWW with stacking layers structure, 2 nm mesopores and high SBET about 739–1118 m2 g−1 . It has been confirmed that the Ti can be well distributed in the silica matrix of the Si/Ti-MCM-36 without formation of TiO2 cluster by controlling the Si/Ti ratio [25]. After loading the Au NPs on these supports by DP method, the resulted catalysts still maintained the 2 nm mesopore and high SBET (600–640 m2 g−1 ), even after the catalytic reactions (550–459 m2 g−1 ). Ti-YNU-1 is a two atom-expanded 2D-MWW structure with 0.67 nm micropores and large SBET (554 m2 g−1 ) [53], and Ti is incorporated by post-hydrothermal treatment of de-borated ERB1. This method seriously constrains the amount of Ti incorporated in the structure. Our results confirmed this conclusion. The Si/Ti ratio of 130 was like the maximum loading of Ti in Ti-YNU-1, confirmed by of the XRD, N2 sorption and UV–vis characterization in Figs. 1, 3 and 5. Post acid treatment of the 100-Ti-YNU-1 can further remove Ti and reserve the expanded structure. However, the attempt to synthesize Ti-YNU-1 with higher Ti contents generated the 33-Ti-YNU-1, which is actually a condensed 3D Ti-MWW structure, as indicated by the XRD spectra in Fig. 1. After loading the Au NPs on the x-Ti-YNU-1 by DP method, the expanded structure seems to contract, evidenced by the shift of (0 0 2) peak to the higher position and merging with the (1 0 0) peak (Fig. 2). Nevertheless, all the expanded Ti-YNU-1 support, the shrunk structure of Ti-MWW and the Au anchored catalysts belong to the microporous structures. It is considered that the larger pores have the advantages of easy diffusion of the reactants and rapid escape of the produced PO during the reaction. Therefore, less likely to have the catalysts deactivated by coking. Here, the two types of Ti incorporated MWW supports of different pore sizes are compared for their catalytic performances and stabilities. It was found that the best PO formation rate of 1.5Au/100Si/Ti-MCM-36 (28.9 gPO h−1 gcat −1 ) is close with that of 4.5Au/A-Ti-YNU-1 (30.0 gPO h−1 gcat −1 ). However, if comparing the Au efficiencies, the TOR of 1.5Au/100Si/Ti-MCM-36

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model CATTOD-9771; No. of Pages 11 10

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

(14.6 gPO h−1 gAu −1 ) is much lower than that of 4.5Au/A-Ti-YNU-1 (23.1 gPO h−1 gAu −1 ), due to the higher Au loading in the former. On the other hand, the 1.5Au/100Si/Ti-MCM-36 catalyst has relatively higher stability than 4.5Au/A-Ti-YNU-1, as shown in Fig. 9, and that is probably due to that the mesopores of the former catalyst facilitate the PO diffusion. 5. Conclusions Based on the ERB-1 precursor, the Ti incorporated expanded MWW structure of microporous material Ti-YNU-1 and mesoporous material Si/Ti-MCM-36 can be synthesized as the supports for the catalyst. The Au nanoparticles can be supported on these supports as an efficient catalyst for gas-phase propylene epoxidation with H2 and O2 . Although the further Au loading with DP method causes the contraction of the expanded layer structure of Ti-YNU-1, but the mesoporous structure of Si/Ti-MCM-36 is still preserved even after catalytic reaction. The Ti loading in Si/TiMCM-36 are both Oh and Td coordinated Ti, while the synthesized Ti-YNU-1 samples only contained isolated Td Ti. The properties of the loaded Au NPs indicate that the Au loading was proportional to the amount of Ti incorporated in the Ti-YNU1 catalysts. Moreover, less amounts of Au and smaller NPs were loaded on Ti-YNU-1 than on Si/Ti-MCM-36. It is interesting to find that the highly isolated Td Ti in Ti-YNU-1 is favorable for attaining the smaller gold NPs. While the higher incorporated Ti in the Si/TiMCM-36 can anchor more Au and resulted in the formation of larger particle. The corresponding catalytic performance proved that the optimal reaction temperature for Au supported on Ti-YNU-1 is 498 K, while that for Ti-MCM-36 is 453 K, this difference is attributed to that the Oh contained catalyst is more favor to catalytic PO formation at lower temperature than that of the high Td coordinated Ti dispersed sample. The best PO formation rate of Au/100Si/TiMCM-36 (28.9 gPO h−1 gcat −1 ) was close with that of Au/A-Ti-YNU-1 (30.0 gPO h−1 gcat −1 ). However, the mesopores in 1.5Au/100Si/TiMCM-36 probably could facilitate the diffusion of the produced PO in the porous structure and exhibited higher stability. Therefore, the property of supports has great influence on the anchored Au and the corresponding catalytic performance. Acknowledgments The financial supports from Ministry of Science & Technology, Taiwan (NSC101-2113-M-002-012-MY3, NSC102-2811-M-002136) are gratefully acknowledged. F. Jin also acknowledges the supports of Natural Science Foundation, China (NSFC 21306143) and Scientific Research Foundation for the Returned Overseas Chinese Scholars, Education Ministry, China. Acknowledgment is also extended to Prof. H.C. Lin and Mr. C.Y. Kao, Ms. Y.Y. Yang and C.Y. Chien of Instrumentation Center, National Taiwan University for the assistances in EPMA and EM experiments. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cattod.2015.08. 041. References [1] T. Hayashi, K. Tanaka, M. Haruta, Selective vapor-phase epoxidation of propylene over Au/TiO2 catalysts in the presence of oxygen and hydrogen, J. Catal. 178 (1998) 566–575. [2] M. Haruta, J. Kawahara, Epoxidation of propylene with oxygen–hydrogen mixtures, in: S.T. Oyama (Ed.), Mechanisms in Homogeneous and Heterogeneous Epoxidation Catalysis, Elsevier B.V., Netherland, 2008.

[3] J. Chen, S.J.A. Halin, E.A. Pidko, M.W.G.M. Verhoeven, D.M.P. Ferrandez, E.J.M. Hensen, J.C. Schouten, T.A. Nijhuis, Enhancement of catalyst performance in the direct propene epoxidation: a study into gold–titanium synergy, ChemCatChem 5 (2013) 467–478. [4] J. Lu, X. Zhang, J.J. Bravo-Suárez, K.K. Bando, T. Fujitani, S.T. Oyama, Direct propylene epoxidation over barium-promoted Au/Ti-TUD catalysts with H2 and O2 : effect of Au particle size, J. Catal. 250 (2007) 350–359. [5] E. Sacaliuc, A.M. Beale, B.M. Weckhuysen, T.A. Nijhuis, Propene epoxidation over Au/Ti-SBA-15 catalysts, J. Catal. 248 (2007) 235–248. [6] A.K. Sinha, S. Seelan, T. Akita, S.M. Tsubota, M. Haruta, Vapor phase propylene epoxidation over Au/Ti-MCM-41 catalysts prepared by different Ti incorporation modes, Appl. Catal. A Gen. 240 (2003) 243–252. [7] B.S. Uphade, T. Akita, T. Nakamura, M. Haruta, Vapor-phase epoxidation of propene using H2 and O2 over Au/Ti-MCM-48, J. Catal. 209 (2002) 331–340. [8] Y. Ren, L. Xu, L. Zhang, J. Wang, Y. Liu, M. He, P. Wu, Selective epoxidation of propylene to propylene oxide with H2 and O2 over Au/Ti-MWW catalyst, Pure Appl. Chem. 84 (2012) 561–578. [9] M. Haruta, B.S. Uphade, S. Tsubota, A. Miyamoto, Selective oxidation of propylene over gold deposited on titanium-based oxides, Res. Chem. Intermed. 24 (1998) 329–336. [10] W.-S. Lee, M. Cem Akatay, E.A. Stach, F.H. Ribeiro, W. Nicholas Delgass, Enhanced reaction rate for gas-phase epoxidation of propylene using H2 and O2 by Cs promotion of Au/TS-1, J. Catal. 308 (2013) 98–113. [11] W.-S. Lee, M. Cem Akatay, E.A. Stach, F.H. Ribeiro, W.N. Delgass, Gas-phase epoxidation of propylene in the presence of H2 and O2 over small gold ensembles in uncalcined TS-1, J. Catal. 313 (2014) 104–112. [12] J. Huang, T. Takei, T. Akita, H. Ohashi, M. Haruta, Gold clusters supported on alkaline treated TS-1 for highly efficient propene epoxidation with O2 and H2 , Appl. Catal. B. Environ. 95 (2010) 430–438. [13] B. Chowdhury, J.J. Bravo-Suárez, M. Daté, S. Tsubota, M. Haruta, Trimethylamine as a gas-phase promoter: highly efficient epoxidation of propylene over supported gold catalysts, Angew. Chem. Int. Ed. 45 (2006) 412–415. [14] J. Huang, E. Lima, T. Akit, A. Guzmán, C. Qi, T. Takei, M. Haruta, Propene epoxidation with O2 and H2 : identification of the most active gold clusters, J. Catal. 278 (2011) 8–15. [15] X. Feng, X. Duan, G. Qian, X. Zhou, D. Chen, W. Yuan, Insights into size-dependent activity and active sites of Au nanoparticles supported on TS-1 for propene epoxidation with H2 and O2 , J. Catal. 317 (2014) 99–104. [16] M.E. Leonowicz, J.A. Lawton, S.L. Lawton, M.K. Rubin, MCM-22: a molecular sieve with two independent multidimensional channel systems, Science 24 (1994) 1910–1913. [17] M.K. Rubin, P. Chu, Composition of Synthetic Porous Crystalline Material, Its Synthesis and Use, U.S. Patent No. 4954325, 1990. [18] U. Díaz, Layered materials with catalytic applications: pillared and delaminated zeolites from MWW precursors, Chem. Eng. N. Y. 2012 (2012) 1–32. [19] G. Bellussi, G. Perego, M.G. Clerici, A. Giusti, Synthetic, Crystalline, Porous Material Containing Oxides of Silicon and Boron, EP293032, 1988. [20] R. Millini, G. Perego, J.W.O. Parker, G. Bellussi, L. Carluccio, Layered structure of ERB-1 microporous borosilicate precursor and its intercalation properties towards polar molecules, Microporous Mater. 4 (1995) 221–230. [21] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, Hydrothermal synthesis of a novel titanosilicate with MWW topology, Chem. Lett. 29 (2000) 774–775. [22] W.B. Fan, P. Wu, S. Namba, T. Tatsumi, Titanosilicate that is structurally analogous to an MWW-type lamellar precursor, Angew. Chem. Int. Ed. 43 (2004) 236–240. [23] J.F. Ruan, P. Wu, B. Slater, O. Terasaki, Structure elucidation of the highly active titanosilicate catalyst Ti-YNU-1, Angew. Chem. Int. Ed. 44 (2005) 6719–6723. [24] L.L. Wang, Y.M. Liu, Y. Wang, W. Xie, M.Y. He, P. Wu, Post-conversion of MWW lamellar precursors into metallosilicates with large porosity, Stud. Surf. Sci. Catal. 174 (2008) 24. [25] F. Jin, C.-C. Chang, C.-W. Yang, J.-F. Lee, L.-Y. Jang, S. Cheng, New mesoporous titanosilicate MCM-36 material synthesized from pillaring layered ERB-1 precursor, J. Mater. Chem. A 3 (2015) 8715–8724. [26] R. Millini, G. Perego, W.O. Parker Jr., G. Bellussi, L. Carluccio, Layered structure of ERB-1 microporous borosilicate precursor and its intercalation properties towards polar molecules, Microporous Mesoporous Mater. 4 (1995) 221–230. [27] P. Wu, T. Tatsumi, T. Komatsu, T. Yashima, A novel titanosilicate with MWW structure. I. Hydrothermal synthesis, elimination of extraframework titanium, and characterizations, J. Phys. Chem. B 105 (2001) 2897–2905. [28] W.-S. Lee, L.-C. Lai, M. Cem Akatay, E.A. Stach, F.H. Ribeiro, W.N. Delgass, Probing the gold active sites in Au/TS-1 for gas-phase epoxidation of propylene in the presence of hydrogen and oxygen, J. Catal. 296 (2012) 31–42. [29] F. Moreau, G.C. Bond, A.O. Taylor, The influence of metal loading and pH during preparation on the CO oxidation activity of Au/TiO2 catalysts, Chem. Commun. (2004) 1642–1643. [30] B. Ravel, M. Newville, Athena, Artemis, Hephaestus: data analysis for X-ray absorption spectroscopy using IFEFFIT, J. Synchrotron Radiat. 12 (2005) 537–541. [31] E.A. Stern, Theory of the extended X-ray-absorption fine structure, Phys. Rev. B 10 (1974) 3027–3037. [32] A.L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Real-space multiple-scattering calculation and interpretation of X-ray-absorption near-edge structure, Phys. Rev. B 58 (1998) 7565–7576.

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041

G Model CATTOD-9771; No. of Pages 11

ARTICLE IN PRESS F. Jin et al. / Catalysis Today xxx (2015) xxx–xxx

[33] S. de Panfilis, F. d’Acapito, V. Haas, H. Konrad, J. Weissmüller, F. Boscherini, Local structure and size effects in nanophase palladium: an X-ray absorption study, Phys. Lett. A 207 (1995) 397–403. [34] A.I. Frenkel, C.W. Hills, R.G. Nuzzo, A view from the inside: complexity in the atomic scale ordering of supported metal nanoparticles, J. Phys. Chem. B 105 (2001) 12689–12703. [35] A. Jentys, Estimation of mean size and shape of small metal particles by EXAFS, Phys. Chem. Chem. Phys. 1 (1999) 4059–4063. [36] A.Y. Anatoly, I. Frenkel, C. Cooper, R. Vasic, Modeling the structure and composition of nanoparticles by extended X-ray absorption fine-structure spectroscopy, Annu. Rev. Anal. Chem. 4 (2011) 23–39. [37] G. Bellussi, G. Perego, M.G. Clerici, A. Giusti, Synthetic, Crystalline, Porous Material Containing Oxides of Silicon and Boron, 1988. [38] J. Lu, X. Zhang, J.J. Bravo-Suárez, T. Fujitani, S.T. Oyama, Effect of composition and promoters in Au/TS-1 catalysts for direct propylene epoxidation using H2 and O2 , Catal. Today 147 (2009) 186–195. [39] A.K. Sinha, S. Seelan, S. Tsubota, M. Haruta, Catalysis by gold nanoparticles: epoxidation of propene, Top. Catal. 29 (2004) 95–102. [40] X.H. Shen, W.B. Fan, Y. He, P. Wu, J.G. Wang, T. Tatsumi, Epoxidation of alkenes and their derivatives over Ti-YNU-1, Appl. Catal. A Gen. 401 (2011) 37–45. [41] E.P. Barrett, L.G. Joyner, P.P. Halenda, The determination of pore volume and area distributions in porous substances. I. Computations from nitrogen isotherms, J. Am. Chem. Soc. 73 (1951) 373–380. [42] S.-Y. Chen, C.-Y. Tang, J.-F. Lee, L.-Y. Jang, T. Tatsumi, S. Cheng, Effect of calcination on the structure and catalytic activities of titanium incorporated SBA-15, J. Mater. Chem. 21 (2011) 2255–2265. [43] J.C. Martínez, N.A. Chequer, J.L. González, T. Cordova, Alternative metodology for gold nanoparticles diameter characterization using PCA technique and UV–VIS spectrophotometry, Nanosci. Nanotechnol. 2 (2012) 184–189.

11

[44] F. Farges, G.E. Brown Jr., J.J. Rehr, Ti K-edge XANES studies of Ti coordination and disorder in oxide compounds: comparison between theory and experiment, Phys. Rev. B 56 (1997) 1809–1819. [45] T.A. Nijhuis, T. Visser, B.M. Weckhuysen, Mechanistic study into the direct epoxidation of propene over gold/titania catalysts, J. Phys. Chem. B 109 (2005) 19309–19319. [46] L. Cumaranatunge, W.N. Delgass, Enhancement of Au capture efficiency and activity of Au/TS-1 catalysts for propylene epoxidation, J. Catal. 232 (2005) 38–42. [47] W.-S. Lee, M. Cem Akatay, E.A. Stach, F.H. Ribeiro, W. Nicholas Delgass, Reproducible preparation of Au/TS-1 with high reaction rate for gas phase epoxidation of propylene, J. Catal. 287 (2012) 178–189. [48] C. Qi, J. Huang, S. Bao, H. Su, T. Akita, M. Haruta, Switching of reactions between hydrogenation and epoxidation of propene over Au/Ti-based oxides in the presence of H2 and O2 , J. Catal. 281 (2011) 12–20. [49] B. Taylor, J. Lauterbach, W.N. Delgass, Gas-phase epoxidation of propylene over small gold ensembles on TS-1, Appl. Catal. A Gen. 291 (2005) 188–198. [50] T.A. Nijhuis, E. Sacaliuc-Parvulescu, N.S. Govender, J.C. Schouten, B.M. Weckhuysen, The role of support oxygen in the epoxidation of propene over gold–titania catalysts investigated by isotopic transient kinetics, J. Catal. 265 (2009) 161–169. [51] D.G. Barton, S.G. Podkolzin, Kinetic study of a direct water synthesis over silica-supported gold nanoparticles, J. Phys. Chem. B 109 (2004) 2262–2274. [52] C.-H. Liu, Y. Guan, E.J.M. Hensen, J.-F. Lee, Y. Chia-Min, Au/TiO2 @SBA-15 nanocomposites as catalysts for direct propylene epoxidation with O2 and H2 mixtures, J. Catal. 282 (2011) 94–102. [53] W. Fan, P. Wu, S. Namba, T. Tatsumi, Synthesis and catalytic properties of a new titanosilicate molecular sieve with the structure analogous to MWW-type lamellar, J. Catal. 243 (2006) 183–191.

Please cite this article in press as: F. Jin, et al., Effect of Ti incorporated MWW supports on Au loading and catalytic performance for direct propylene epoxidation, Catal. Today (2015), http://dx.doi.org/10.1016/j.cattod.2015.08.041