Supported Pt Nanoparticles on Mesoporous Titania for Selective Hydrogenation of Phenylacetylene

Semi-hydrogenation of alkynes to alkenes is one of the most important industrial reactions. However, it remains technically challenging to obtain high alkene selectivity especially at a high alkyne conversion because of kinetically favorable over hydrogenation. In this contribution, we show that supported ultrasmall Pt nanoparticles (2.5 nm) on mesoporous TiO2 (Pt@mTiO2) remarkably improve catalytic performance toward semi-hydrogenation of phenylacetylene. Pt@mTiO2 is prepared by co-assembly of Pt and Ti precursors with silica colloidal templates via an evaporation-induced self-assembly process, followed by further calcination for thermal decomposition of Pt precursors and crystallization of mTiO2 simultaneously. As-resultant Pt@mTiO2 discloses a high hydrogenation activity of phenylacetylene, which is 2.5 times higher than that of commercial Pt/C. More interestingly, styrene selectivity over Pt@mTiO2 remains 100% in a wide phenylacetylene conversion window (20–75%). The styrene selectivity is >80% even at 100% phenylacetylene conversion while that of the commercial Pt/C is 0%. The remarkable styrene selectivity of the Pt@mTiO2 is derived from the weakened styrene adsorption strength on the atop Pt sites as observed by diffuse reflectance infrared Fourier transform spectroscopy with CO as a probe molecule (CO-DRIFTS). Our strategy provides a new avenue for promoting alkyne to alkene transformation in the kinetically unfavorable region through novel catalyst preparation.


INTRODUCTION
Selective hydrogenation of alkynes to alkenes is one of the most industrially important reactions (Yang et al., 1998Mastalir et al., 2000;Huang et al., 2007;Studt et al., 2008;Yuan et al., 2009;Liu H. et al., 2012;Sheng et al., 2012;Tüysüz et al., 2013;Zhao et al., 2015Zhao et al., , 2019Xie et al., 2016;Xiao et al., 2017;Zhan et al., 2017). The increase of alkene selectivity not only benefits high quality of the downstream polyene products but greatly improves the alkene yields and profitability Bond, 2006, 2008;McCue and Anderson, 2015;Hu et al., 2018a). Pt-based catalysts still suffer from low selectivity to alkenes especially in higher alkyne conversion windows, despite its high activity for hydrogenation Tang et al., 2017;Wang et al., 2018). In theory, the adsorption and activation of alkynes are favored thermodynamically over Pt compared to alkenes. However, as alkynes approach a high conversion, alkenes become kinetically favorable to adsorb on Pt, leading to a full hydrogenation to alkanes and thus a sharp decrease in alkene selectivity (Yang et al., 2012;McCue and Anderson, 2015). For example, commercial Pd/C or Pt/C catalysts generally produce alkane at 100% alkyne conversion . The lack of control of alkene selectivity at higher alkyne conversions greatly limits alkene yields. As a result, selective hydrogenation of alkynes is generally conducted at low alkyne conversions to generate enhanced yields of alkenes.
The key to solving this problem lies in developing more efficient Pt-group metal catalysts with weakened alkene adsorption strength, which facilitates the alkenes desorption rather than over hydrogenation to alkanes. Compared with strong σ-bonded alkenes, π-bonded alkenes are more favorable to desorb from the active Pt-group metal (Wan and Zhao, 2007;Armbrüster et al., 2010;Mitsudome et al., 2016;Feng et al., 2017). For example, Li et al. found that the π-bonded ethylene was 0.75 eV more facile than the σ-bonded ethylene to desorb from Pd surfaces based on density-functional theory (DFT) calculations (Feng et al., 2017). This in turn greatly favors high ethylene selectivity where 80% ethylene selectivity is achieved, even when acetylene is fully converted. The π-mode alkene adsorption is mainly on atop metal sites (Pei et al., 2017). To improve the alkene selectivity in selective hydrogenation of alkyne, both nanoparticles (NPs) downsizing and alloying strategies have been carried out to enrich the atop sites of Pt-group metal catalysts (Hu et al., 2018b). On the other hand, metal-support interaction has long been employed to tune catalytic properties of catalysts especially the small NPs with strong interactions with supports (Teschner et al., 2008). A reduced hydrogen reduction temperature is generally observed on the small NPs functionalized supports, which greatly benefits hydrogenation activity by virtue of strong hydrogen spillover Wei et al., 2018). The decrease of particle sizes also enables more surface-exposed and low-coordinated active sites in the form of corner or edge sites, potentially favoring weak π-adsorbed alkenes (Kleis et al., 2011;Yang et al., 2013;Wu et al., 2019). Thus, smaller Pt NPs hold great potential to catalyze semi-hydrogenation of alkynes with high catalytic activity and alkene selectivity.
In the current study, we present a highly selective catalyst that contains ultrasmall Pt NPs (2.5 nm), strong Pt-support (Pt-TiO 2 ) interaction, and high surface area (mesoporous framework) for selective hydrogenation of phenylacetylene to styrene. Supporting ultrasmall Pt NPs on mesoporous TiO 2 (Pt@mTiO 2 ) were prepared using an evaporation-induced self-assembly (EISA) method (Figure 1a). For this synthesis, platinum acetylacetonate (Pt(acac) 2 ) was mixed with titanium sources and structure-directing polymers in EISA . Upon annealing, ultrasmall Pt NPs were synthesized and encapsulated within the ordered mTiO 2 support. The ordered mesopores of mTiO 2 provide a powerful nanoconfinement and endow the ultrasmall Pt nanoparticles with excellent size uniformity and thermal stability. The monodisperse Pt NPs with sizes of around 2.5 nm remained stable after calcination at 450 • C. Pt NPs supported on mTiO 2 (Pt@TiO 2 ) exhibited excellent catalytic properties toward selective hydrogenation of phenylacetylene. The selectivity to styrene remained at 100% in the phenylacetylene conversion range of 20-75%. Even at 100% phenylacetylene conversion, Pt@TiO 2 exhibited >80% styrene selectivity. The abundant surface atop Pt sites of the Pt@mTiO 2 accounted for the high styrene selectivity as revealed by CO-DRIFTS. In addition, the specific mass activity of the 2.5 nm Pt@mTiO 2 reached as high as 0.5 mol h −1 per gram of Pt metal, which is 2.5 times higher than that of the commercial Pt/C catalyst (0.2 mol h −1 per gram of Pt metal).

Synthesis of Pt@mTiO 2
Pt@mTiO 2 was prepared using an EISA method. In a typical synthesis, 400 µL of newly prepared TiCl 4 solution (10 % in ethanol, volume ratio) was added to 5.0 mL of PEO-b-PTMSPMA micelles solution (15.0 mg mL −1 in ethanol) under stirring. Two hundred and seventy micro liter of TIPO was then added into the above solution, followed by mild stirring for 30 min. 2.0 mg of Pt(acac) 2 dissolved in 1.0 mL of ethanol was added dropwise where the theoretical mole ratio of Pt to Ti is 0.004 (corresponding to 1.0 wt% Pt loading in the obtained Pt@mTiO 2 ). After sufficient stirring for 30 min, the mixture solution was transferred into a Petri dish to evaporate the solvent at 40 • C for 24 h and 100 • C for another 12 h. After this procedure, the obtained intermediates were allowed for calcination at 450 • C under air for 2 h. A template removal process is further carried out by washing with hot 2.0 M NaOH solution for 30 min at 50 • C. The Pt loading given by EDS measurements is 1 wt%, agreeing well with the initial Pt feed ratio. Through subsequent drying in vacuum at 50 • C overnight, Pt@mTiO 2 was finally obtained. About 3.6 nm Pt NPs on mesoporous TiO 2 is also prepared as a control using similar procedures except that the thermal calcination is carried out at 650 • C for 2 h.

Selective Hydrogenation of Phenylacetylene Catalytic Evaluations
Selective hydrogenation of phenylacetylene over the Pt@mTiO 2 catalysts was carried out by the following procedures. Firstly, 10.0 mg of catalyst was added to 3.0 mL of 1,4-dioxane in a 15 mL processed glass bottle. After 20 min sonication, 100.0 mg (1.0 mmol) of phenylacetylene was further added to the homogeneous suspension and stirred for another 30 min. Before transferring the processed bottle to a 25 mL high-pressure stainless-steel reactor, the magnet is taken out, leaving the reaction happen at static conditions without stirring. After purging with hydrogen several times, the reactor was pressurized with 50 psi (pounds per square inch) hydrogen and reacted for varied times. After the reaction finished, the whole reactor was cooled down in an ice bath. The postreaction suspension was separated by centrifugation at 8,000 rpm. The resultant product was kept in a 5 mL test tube and the proton nuclear magnetic resonance ( 1 H NMR) analysis was carried out using CDCl 3 as solvent to calculate the conversion of phenylacetylene and the selectivity to styrene. The reaction kinetics were measured both for the Pt@mTiO 2 and commercial Pt/C catalysts. The specific reaction rate was determined as the consumption rate of the phenylacetylene (normalized to the mass of Pt metal per hour) by controlling the phenylacetylene conversion in the kinetic region. The specific rate in this work was calculated using the following equation: where r is the specific rate, N is the converted phenylacetylene amount in mole, w is supported metal loading weight (gram), and t is the reaction (minute). The turnover frequency (TOF) is calculated based on the following equation: where N is the converted phenylacetylene amount in mole, w is supported Pt loading weight (gram), M is the molar mass of Pt, D is metal dispersion of Pt and t is the reaction (second). The metal dispersion is calculated according to previous reports (Dominguez-Dominguez et al., 2008;Qiao et al., 2011). The stability tests of the Pt@mTiO 2 and the commercial Pt/C were carried out by recycling the catalyst for 5 consecutive cycles.

Characterization
Scanning electron microscopy (SEM) was conducted on a FEI Nova NanoSEM 450. Bright and dark-field transmission electron microscopy (TEM) was performed on JEM-2100F FETEM. The X-ray diffraction (XRD) patterns were collected on a Bruker D2 Phaser. The small angle X-ray scattering (SAXS) patterns were recorded on a Bruker Nano STAR instrument. X-ray photoelectron spectroscopy (XPS) analysis was conducted on ESCALAB 250 Xi X-ray photoelectron spectrometer with Al Kα radiation. The N 2 sorption measurements were conducted on a Micromeritics ASAP 2020 Surface Area and Porosity Analyzer. 1 H NMR spectroscopy was collected on a Bruker Avance 300 MHz spectrometer. Diffuse reflectance Fourier transform infrared (DRIFT) spectroscopy was carried out using a Nicolet IS50 spectrometer using CO as a probe molecule. The DRIFT cell was filled with 20 mg of catalysts and was treated at 100 • C with nitrogen before CO was introduced.

RESULTS AND DISCUSSION
Mesoscopic and microscopic structures of the Pt@mTiO 2 were thoroughly characterized with SEM and high-angle annular dark field scanning TEM (HAADF-STEM) (Figure 1). The SEM image in Figure 1b shows the mesopores throughout entire mTiO 2 support. Supplementary Figure 1 shows that mTiO 2 displays bulk cubic morphology with a broad side length distribution. The zoomed-in SEM picture in Figure 1c confirms the uniform distribution of mesopores. The average pore size of mTiO 2 is 11.7 ± 2.2 nm as given in the inset of Figure 1b. Since Pt NPs are small, there is no obvious contrast on bright field TEM. The HAADF-STEM technique is further used to identify the ultrasmall Pt NPs. The HAADF-STEM images of Pt@mTiO 2 in Figures 1d-f reveal the well-dispersed Pt NPs. Pt NPs with a higher electron density are brighter under darkfield STEM image while TiO 2 is darker. The average size of Pt NPs is 2.5 ± 0.5 nm as plotted in Figure 1e. The pore size of mTiO 2 is about 11 nm from HAADF-STEM characterization, which agrees well with the above SEM observations. Figure 1f shows energy dispersive X-ray spectroscopy (EDS) mapping images of Pt@mTiO 2 , further confirming fine dispersion of these 2.5 nm Pt NPs in mTiO 2 support. The Pt loading of Pt@mTiO 2 is further determined by EDS measurements. As show in Supplementary Figure 2 and Supplementary Table 1, the Pt loadings obtained by EDS measurements of Pt@mTiO 2 are about 1 wt%, agreeing well with the initial feed ratio (1 wt%). This observation implies no loss of Pt during preparation processes. By comparison, we have also carried out TEM observations of the commercial Pt/C catalyst. As displayed in Supplementary Figure 3, the Pt NPs in commercial Pt/C show a very wide size distribution averaging about 3 nm as reported previously (Cui et al., 2019). The Pt loading of the commercial Pt/C is 5 wt% as received.
To examine the mesoporous structures and crystallographic information of the Pt@mTiO 2 , we carried out small-angle X-ray scattering (SAXS) and wide-angle X-ray diffraction (XRD) characterization. Figure 2A is the SAXS patterns of the Pt@mTiO 2 and the pure mTiO 2 support. These data show a set of strong scattering peaks that are clearly identified and assigned to (111), (311), and (500) reflections of the face-centered cubic (fcc) mesostructure. The average cell parameters were calculated to be 38 nm as displayed in Supplementary Table 2. The SAXS signals are almost the same for Pt@mTiO 2 and mTiO 2 , indicating welldefined mesoporous structures. Figure 2B displayed the wideangle XRD patterns, suggesting the anatase TiO 2 phase for both Pt@mTiO 2 and mTiO 2 (JCPDS 00-064-0863). Unfortunately, due to the ultrasmall size, metallic Pt NPs (JCPDS 04-0802) only showed very broad and weak XRD diffraction (Xiang et al., 2011;Prinz et al., 2014). Benefiting from the well-defined mesoporous structures, the BET specific surface area of the Pt@TiO 2 is 42 m 2 g −1 with an average of pore size of 14 nm as displayed in Figures 2C,D.
Pt@mTiO 2 has multiple advantageous properties, including ultrasmall Pt NPs and mesoporous framework with high surface area, which possibly benefit the selective hydrogenation of alkynes to alkenes. In this manuscript, selective hydrogenation of phenylacetylene was chosen as a probe reaction since the semi-hydrogenation product of styrene is industrially important ( Figure 3A). As a control, commercial Pt/C was also investigated. Pt@mTiO 2 was favorable for catalyzing the hydrogenation of phenylacetylene to styrene, while commercial Pt/C fully hydrogenated phenylacetylene to ethylbenzene. Figure 3B showed the phenylacetylene conversion and the styrene selectivity over Pt@mTiO 2 as a function of reaction time. Phenylacetylene was selectively converted to styrene in a wide phenylacetylene conversion window (20-75%), only when phenylacetylene reached full conversion, styrene selectivity slightly dropped to 80.3%. This result (80.3% styrene selectivity at 100% phenylacetylene conversion) is among the best reported to date as demonstrated in Supplementary Table 3 Kleis et al., 2011;Behrens et al., 2012;Montesano Lopez, 2014;Alig et al., 2019;Han et al., 2019). By comparison, pure mTiO 2 is inactive and shows no activity to convert phenylacetylene in the reaction time range from 7 to 26 h (Supplementary Figure 4). We have also compared the catalytic properties of Pt@mTiO 2 with commercial Pt/C catalysts. Phenylacetylene was fully over hydrogenated to ethylbenzene in the measuring time range from 7 to 26 h with a 100% ethylbenzene selectivity as shown in Supplementary Figure 5. Figure 3C shows the catalytic rates of Pt@mTiO 2 and commercial Pt/C normalized to the Pt mass per unit time. By controlling the phenylacetylene conversion in the kinetic region (10-20% of conversion), the rate characterizes the intrinsic activity of the catalyst. The catalytic rate of Pt@mTiO 2 is 0.5 mol g Pt −1 h −1 , 2.5 times higher than that of commercial Pt/C (0.2 mol g Pt −1 h −1 ). In addition, the TOF value of the Pt@mTiO 2 is calculated to be 4.1 s −1 , which is more than 2 times that of the commercial Pt/C (2.0 s −1 ), demonstrating a higher intrinsic activity. According to previous results, the particle size of Pt NPs is intimately connected with the amount of surface-exposed active sites, which in turn affects the activity (Zhang et al., 2007;Qiao et al., 2011). The smaller sized Pt in Pt@mTiO 2 results in more exposed surface Pt sites compared with commercial Pt/C (average size of 3 nm), which benefits high catalytic activity. Catalytic stability is another concern for practical applications. To test the catalytic stability of Pt@mTiO 2 , we have carried out recycling experiments of the post-reacted Pt@mTiO 2 at 50 • C and 50 psi hydrogen. As shown in Figure 3D, styrene selectivity remains around 80% over the Pt@mTiO 2 while phenylacetylene conversion remains at 100% even after five catalytic recycles, showing excellent catalytic stability. By comparison, styrene selectivity over the commercial Pt/C is 0% for the five consecutive measurements at 100% phenylacetylene conversion as displayed in Supplementary Figure 6. By virtue of the nanoconfinment enabled by mesoporous mTiO 2 supports, the average particle size of the post-reacted Pt@mTiO 2 remains the same without any growth or aggregation as displayed in Supplementary Figure 7.
To reveal the intrinsic nature why Pt@mTiO 2 presented excellent catalytic properties for the selective hydrogenation of phenylacetylene, the surface properties of Pt were carefully studied. We first carried out X-ray photoelectron spectroscopy (XPS) for Pt@mTiO 2 . Figure 4A is the high-resolution Pt 4f XPS spectra of Pt@mTiO 2 and commercial Pt/C. The wellresolved XPS binding energy peaks centered at 71.9 and 75.3 eV of Pt@mTiO 2 are ascribed to metallic Pt 4f 7/2 and Pt 4f 5/2 , respectively. For the commercial Pt/C, the Pt 4f 7/2 and Pt 4f 5/2 shifted to a slightly higher binding energy at 72 and 75.4 eV, respectively. The shift of Pt 4f peaks is indicative of the surface electron density difference of the two samples. 2.5 nm Pt NPs is slightly electron-rich compared to that of the commercial Pt/C. The electron-rich 2.5 nm Pt NPs are more effective to transfer electrons to the π * molecular orbitals of phenylacetylene, which in turn promotes phenylacetylene activation and improves catalytic reactivity (Wan and Zhao, 2007). Both the Pt@mTiO 2 and commercial Pt/C show existence of Pt 2+ species as identified at around 73.1 and 77.1 eV, respectively, which may derive from surface oxidation (Miller et al., 2014). The Pt 2+ /Pt 0 ratio, as measured from the integration areas in the fitting cure, is 0.43 for 2.5 nm Pt NPs. By comparison, this is much lower for commercial Pt/C with a Pt 2+ /Pt 0 ratio of 0.35. This result implies that Pt is more surface oxidized in 2.5 nm Pt NPs, which is further evidenced in Ti 2p XPS spectra. As shown in Figure 4B, a strong Ti 3+ XPS binding energy peak at around 458.2 eV is found in 2.5 nm Pt NPs on mTiO 2 , which is absent in the pure mTiO 2 Ou et al., 2018). These results show that 2.5 nm Pt NPs have a stronger interaction with the mTiO 2 support where Pt NPs are partially oxidized and demonstrate electron back-donation to reduce Ti 4+ to Ti 3+ of mTiO 2 . The stronger metal-support interaction is of great importance in stabilizing small NPs as revealed by a series of reports (Lu and Schüth, 2006). In the current case, the excellent thermal stability of 2.5 nm Pt NPs supported on mTiO 2 may also derive from strong metal-support interactions.
Small NPs afford abundant surface low-coordinated corner or edge sites, which potentially modifies reactant/product adsorption modes that are quite different from that on larger ones (Zelcer and Soler-Illia, 2013). For example, alkenes prefer a weak π adsorption on low-coordinated Pt sites such as corner, edge, or single atom sites of small Pt NPs rather than a strong σ adsorption on continuous Pt lattices of larger Pt NPs. The weakening of alkene adsorption strength greatly favors high alkene selectivity Zhang et al., 2019). To this end, CO-DRIFTS was further carried out to understand the high selectivity to styrene on Pt@mTiO 2 . For the commercial Pt/C, it is very difficult to collect CO-DRIFT spectra because of signal blocking of carbon supports. Thus, we further prepared 3.6 nm Pt supported on mTiO 2 by calcination at a higher annealing temperature (650 • C), which preserved the mesoporous structure but with slightly larger Pt NPs (3.6 nm) (Supplementary Figure 8). Supplementary Figure 9 shows that both the catalytic activity and styrene selectivity of the 3.6 nm Pt NPs on mTiO 2 are much lower than that of the 2.5 nm Pt NPs on mTiO 2 . Figure 5 shows the CO-DRIFTS curve of 2.5 nm Pt NPs and 3.6 nm Pt NPs on mTiO 2 , respectively. The CO adsorption models on 2.5 nm Pt NPs are quite different from that on 3.6 nm Pt NPs. The fitted results show that only the linear CO adsorption mode is observed on the 2.5 nm Pt NPs on mTiO 2 while the 3.6 nm Pt NPs on mTiO 2 display both linear CO adsorption and bridged CO adsorption modes in the wavenumber window between 1,800 and 2,200 cm −1 (Singh et al., 2018). The linear CO adsorption of both 2.5 nm Pt NPs on mTiO 2 and 3.6 nm Pt NPs on mTiO 2 are related to the surface atop Pt sites (Wieckowski, 1999). Due to the Pt coordination differences, the linear CO adsorption bands of the 2.5 nm Pt NPs on mTiO 2 and the 3.6 nm Pt NPs on mTiO 2 are divided into two regions that are referred to as the low-coordination Pt-CO linear adsorption band and high-coordination Pt-CO linear adsorption band centered at around 2,004 and 2,160 cm −1 , respectively (Redina et al., 2015;Singh et al., 2018). The low-coordination Pt-CO linear adsorption to high-coordination Pt-CO linear adsorption ratio of 2.5 nm Pt NPs on mTiO 2 is calculated to be 4.8, which is 6 times higher than that of 3.6 nm Pt NPs on mTiO 2 . This result demonstrates high unsaturated coordination of the Pt sites in 2.5 nm Pt NPs on mTiO 2 due to the smaller Pt NPs size as compared with 3.6 nm Pt NPs on mTiO 2 , which agrees well with the above HAADF-STEM observations. The schematic models of Pt NPs with diameters of 2.5 and 3.6 nm are also constructed, respectively, as displayed in Supplementary Figure 10. Smaller Pt NPs show more abundant surface corner and edge sites, which accounts for the pure linear CO adsorption revealed via CO-DRIFTS measurements. The unique surface structures of the 2.5 nm Pt NPs on mTiO 2 greatly benefit weak styrene adsorption in selective hydrogenation of phenylacetylene through a π adsorption mode, which favors styrene desorption and high styrene selectivity.

CONCLUSION
To summarize, we report the preparation of mesoporous TiO 2 confined ultrasmall Pt nanoparticles with an average FIGURE 5 | Diffuse reflectance infrared Fourier transform spectroscopy with CO molecules of the 2.5 nm Pt NPs on mTiO 2 and the 3.6 nm Pt NPs on mTiO 2 . particle size of 2.5 ± 0.5 nm using an EISA method. Asresulted Pt@mTiO 2 displayed excellent catalytic activity and styrene selectivity in semi-hydrogenation of phenylacetylene. The hydrogenation activity of Pt@mTiO 2 displays a 2.5fold increase over that of the commercial Pt/C. A 100% styrene selectivity is also obtained over the Pt@mTiO 2 in a wide phenylacetylene conversion range of 20-75%. The styrene selectivity is remarkably above 80% even at full phenylacetylene conversion. The excellent styrene selectivity is ascribed to the abundant surface atop Pt sites as given by the CO-DRIFTS. In addition, the Pt@mTiO 2 also shows excellent catalytic stability as demonstrated in five consecutive recycling experiments, which shows great potential for practical applications. Our work is expected to open up new opportunities for developing highly efficient alkene-to-alkene conversion catalysts by novel synthesis.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.