Size and Shape Dependence on Pt Nanoparticles for the ...

Catal Lett (2011) 141:914–924 DOI 10.1007/s10562-011-0647-6

Size and Shape Dependence on Pt Nanoparticles for the Methylcyclopentane/Hydrogen Ring Opening/Ring Enlargement Reaction S. Alayoglu • C. Aliaga • C. Sprung G. A. Somorjai

•

Received: 7 April 2011 / Accepted: 12 June 2011 / Published online: 23 June 2011 Ó Springer Science+Business Media, LLC 2011

Abstract Monodisperse Pt nanoparticles (NPs) with well-controlled sizes in the range between 1.5 and 10.8 nm, and shapes of octahedron, cube, truncated octahedron and spheres (*6 nm) were synthesized employing the polyol reduction strategy with polyvinylpyrrolidone (PVP) as the capping agent. We characterized the as-synthesized Pt nanoparticles using transmission electron microscopy (TEM), high resolution TEM, sum frequency generation vibrational spectroscopy (SFGVS) using ethylene/H2 reaction as the surface probe, and the catalytic ethylene/H2 reaction by means of measuring surface concentration of Pt. The nanoparticles were supported in mesoporous silica (SBA-15 or MCF-17), and their catalytic reactivity was evaluated for the methylcyclopentane (MCP)/H2 ring opening/ring enlargement reaction using 10 torr MCP and 50 torr H2 at temperatures between 160 and 300 °C. We found a strong correlation between the particle shape and the catalytic activity and product distribution for the MCP/ H2 reaction on Pt. At temperatures below 240 °C, 6.3 nm Pt octahedra yielded hexane, 6.2 nm Pt truncated octahedra Electronic supplementary material The online version of this article (doi:10.1007/s10562-011-0647-6) contains supplementary material, which is available to authorized users. S. Alayoglu C. Aliaga C. Sprung G. A. Somorjai (&) Department of Chemistry, University of California, Berkeley, CA, USA e-mail: [email protected] S. Alayoglu C. Aliaga C. Sprung G. A. Somorjai Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA Present Address: C. Sprung Department of Chemistry, SMN, University of Oslo, P.O. Box 1033, 0314 Oslo, Norway

123

and 5.2 nm Pt spheres produced 2-methylpentane. In contrast, 6.8 nm Pt cubes led to the formation of cracking products (i.e. C1–C5) under similar conditions. We also detected a weak size dependence of the catalytic activity and selectivity for the MCP/H2 reaction on Pt. 1.5 nm Pt particles produced 2-methylpentane for the whole temperature range studied and the larger Pt NPs produced mainly benzene at temperatures above 240 °C. Keywords Platinum Methylcyclopentane hydrogenation Ethylene/hydrogen probing by sum frequency generation vibrational spectroscopy Shape-controlled nanoparticles High-resolution electron microscopy Microscopy Spectroscopy and general characterisation Colloidal synthesis Preparation and materials

1 Introduction Selectivity is often of key importance to heterogeneous transformations of hydrocarbons [1], and thus achievable by controlling the size and/or shape of nanoparticle catalysts [2]. Structure-sensitive reactions such as pyrolle and furan hydrogenation on Pt [3–6], show strong dependence on particle size and/or shape. The hydrogenation of methylcyclopentane (MCP) is a multi-path and multiproduct reaction, which yields C6 isomers (i.e. hexane and its derivatives) via ring opening [1, 7–9], benzene via ring enlargement and dehydrogenation [10], and shorter chain hydrocarbons (i.e. C1–C5 isomers) via cracking [11, 12]. Pt catalysts have been reported to be selective for the formation of branched hexane derivatives (i.e. 2-methylpentane and 3-methylpentane) [1, 13–16], however, no systematic study of the catalytic activity and product

Size and Shape Dependence on Pt Nanoparticles

selectivity for the MCP/H2 reaction on the size- and shapecontrolled monodisperse Pt nanoparticles has been reported [15–17]. In this paper, we focused on the MCP/H2 reaction over the size-controlled Pt NPs in the range between 1.5 and 11 nm, and the shape-controlled *6 nm Pt NPs with cubic, octahedral, truncated octahedral and spherical morphologies. We have synthesized Pt NPs employing alcohol reduction in the presence of polyvinylpyrrolidone (PVP) stabilizer. We have characterized the size- and shape-controlled Pt NPs using an array of techniques including transmission electron microscopy (TEM), highresolution TEM, sum frequency generation (SFG) vibrational spectroscopy using the ethylene/H2 reaction probe. We have employed the ethylene/H2 reaction to determine the concentration of surface sites that are available for catalysis. Then, we evaluated the catalytic reactivity of well-defined Pt NPs for the MCP/H2 reaction using a partial pressure of 10 torr MCP, 50 torr H2 and 700 torr He, and in the temperature range of 160–300 °C. We found a strong correlation between the surface reactivity and the particle shape. Spherical (high index surfaces) and truncated octahedral ((111) and (100) surfaces) Pt NPs showed higher selectivity towards the branched hexane derivatives. In sharp contrast, Pt nano-octhahedra (110) was exclusively selective towards hexane, and Pt nanocubes (100) mainly led to the formation of cracking products. We observed rather a weak dependence of the product selectivity on the particle size, and the most discernible observation was the low concentration of benzene product over the 1.5 nm Pt particles.

2 Experimental 2.1 Nanoparticle Synthesis The synthesis of Pt NPs with 1.5 and 3.0 nm average sizes were reported previously by our laboratory [18]. The synthesis of spherical Pt NPs with 5.2, 7.2 and 10.8 nm average sizes, 6.2 nm truncated octahedral, 6.3 nm octahedral and 6.8 nm cubic Pt NPs were carried out using standard Schlenk technique under Ar atmosphere using a modified polyol (i.e. ethylene glycol, triethylene glycol, etc.) reduction reaction strategy [19, 20]. The metal precursor salt and capping agent (PVP) were dissolved in a polyol solvent at 80 °C. The solution was ramped to the designated temperature and the reaction was terminated by removing the flask off of the heating mantle or oil bath. The colloidal suspension was allowed to cool down to room temperature, and nanoparticles were precipitated in acetone upon centrifugation and re-dispersed in

915

ethanol. Washed colloidal suspensions of nanoparticles were used to prepare supported catalysts immediately. The synthetic details are described in the supplementary information. 2.2 Catalyst Preparation The catalyst loading was 1.0% Pt by weight. Mesoporous silica supports, namely SBA-15 with average pore sizes of 8.8 nm and MCF-17 with average pore sizes of 20–50 nm, were purchased and used without further treatment (Claytec, inc.). Nanoparticles with average particle sizes smaller than 5 nm were supported in SBA-15. MCF-17 was employed for supporting larger nanoparticles (7.2 and 10.8 nm). Washed nanoparticle colloids in ethanol were added dropwise into mesoporous silica in 50 mL ethanol in a 100 mL erlenmayer flask to yield the desired final Pt loading by weight. The suspension was sonicated for 3 h, and then centrifuged and washed with acetone. Washing/ precipitation cycle was repeated at least three times to remove extra PVP and the reaction byproducts. The final slurry was dried in an oven at 60 °C in air. 2.3 Catalytic Testing 2.3.1 MCP Hydrogenation The catalytic testing was performed using our home-built plug-flow reactor connected to a Hewlett Packard 5890 Gas Chromatogram. A 10% SP-2100 on 100/120 supelcoport packed column in line with a FID detector was used to separate and analyze the reactant, MCP, and the products. The plug-flow reactor was designed such a way that it could be converted to a batch reactor by looping the GC exit to the reactor bed. In order to mix the reactants in the batch mode, a recirculation pump was employed. Mass flow controllers were carefully calibrated using a bubble flow-meter and used to introduce the ultra high purity reactive, carrier and filler gases. Saturated methylcyclopentane (MCP) vapor at room temperature (22 °C) were carried to the reactor using a bubbler. The reactant (MCP) flow was carefully calibrated at different temperatures and partial pressures of He carrier. A total flow of 50 mL/min was used. Total pressure under flow conditions was measured to be 790 torr using a Baratron capacitance gauge located above the reactor bed. Given the calibration curves, MCP pressure was estimated to be 10 torr. H2 partial pressure was varied between 0 and 500 torr. 50–55 mg charges of the catalysts were diluted by quartz with average particle size of 0.4 mm and loaded in the reactor bed. The gas hourly space velocity (GHSV) was calculated to be *60,000 h-1, and the residence time was 0.9 s.

123

916

2.3.2 Ethylene Hydrogenation Reaction for Determination of Active Pt Sites Ethylene hydrogenation was employed to determine the active sites of Pt NP catalysts. 10 torr ethylene and 100 torr hydrogen reacted over 0.1–0.3 mg of 0.025 wt.% Pt at temperatures between 22 and 110 °C. A Hayasep Q packed column was used to separate ethylene from ethane. Prior to ethylene hydrogenation (and the catalyst screening for the MCP/H2 reaction) the Pt NP catalysts were preconditioned as follows. Temperature was ramped to 150 °C under He atmosphere and held isothermal during the course of treatment. The NP catalyst was first exposed to air flow for 30 min. Then, the reactor was flushed with He for 15 min. The NP catalysts were reduced by a flow of 10 vol.% H2 in He for 30 min. Finally, the reactor was cooled down to room temperature and temperature programmed reaction was started. A generic turnover number of 11.9 molecule ethane/Pt/s, which is in perfect agreement with the single crystal and nanoparticle data given in literature, was derived from the measurement of 1.5 nm Pt NPs and used to calculate the surface concentration of Pt for the other nanoparticle catalysts [3]. A plot of TEM-projected surface areas versus ethylene hydrogenation reaction-derived surface areas is given in Fig. S2. Ethylene hydrogenation reaction obeys Arrhenius kinetics with activation energy of *10 ± 0.5 kcal/mol in the temperature window between 21 and 90 °C for all the NP catalysts studied. 2.4 Sum Frequency Generation (SFG) Surface Vibrational Spectroscopy An active/passive mode-locked Nd:YAG laser (Leopard D-20, Continuum) with a 20 ps pulse width and a 20 Hz repetition rate was used in all the SFGVS studies. The fundamental output at 1,064 nm was passed through an optical parametric generation/amplification stage where a tunable IR beam (2,700–4,000 cm-1) and a second harmonic visible (VIS) beam (532 nm) were generated. The IR (100 lJ) and VIS (100 lJ) beams were spatially and temporally overlapped at the bottom surface of the fused silica prism where the nanoparticles had been deposited at angles of incidence equal to 42 and 65°, respectively, with respect to the surface normal. All the experiments were carried out in ppp polarization combination. The generated SFG signal was then collected and sent to a photomultiplier tube. The signal-to-noise ratio was further enhanced by a gated integrator, while the IR beam was scanned through the spectral region of interest. Additional information about the SFGVS system used in this study can be found elsewhere [6]. Experiments were carried out in a temperature-controlled high-pressure cell consisting of a stainless steel

123

S. Alayoglu et al.

chamber on top of which an equilateral fused silica prism (20 9 20 mm) with a LB film of nanoparticles deposited onto its bottom face was attached through a KalrezÒ O-ring and clamped using a TeflonÒ piece. Product gases in the reaction cell were constantly mixed by a recirculation pump. After nanoparticle LB-deposition onto the bottom face of the equilateral fused silica prism (SFGVS experiments) or the surface of a silicon wafer (GC experiments), the nanoparticles were UV-hydrogen treated to remove the PVP capping layer, as described previously. The removal of the capping layer was verified by observing the disappearance of the C–H stretch mode in the SFGVS spectrum acquired before each experiment. 2.5 Characterization of Nanoparticle Catalysts In this study, the size- and shape-controlled Pt NPs were synthesized via a general synthetic route. All nanoparticles were capped with PVP. Diols with different chain lengths (e.g. EG and TEG) were employed to achieve size control of Pt NPs with controlled morphologies between 5 and 11 nm. Using Pt(acac)2 as precursor, it was found that a shorter carbon backbone (or lower the boiling temperature) of the reducing solvent resulted in a larger particle size with polyhedral shapes (i.e. truncated octahedron or sphere). For example, 6.2 nm Pt truncated octahedra were prepared in boiling TEG. For the nano-truncated octahedra sample, HRTEM images in Fig. S8 also suggested the presence of cuboctahedral and icosahedral nanoparticles (about 10–15%). Synthesis in EG with the identical monomer concentrations and reaction conditions as the nano-cubocthedra sample, however, produced 10.8 nm spherical (r = 10.1%) particles. The use of H2PtCl6 salt as precursor, instead of Pt(acac)2, resulted in smaller NPs with spherical shapes. For example, 7.2 nm Pt particles (r = 8.2%) were synthesized in boiling EG using H2PtCl6 as Pt precursor. On the other hand, particle size distributions were broad (i.e. r [ 20%) when reacting H2PtCl6 and TEG. By mixing Pt(acac)2 with *20 mol% H2PtCl6, 5.2 nm spherical Pt NPs with narrow size distributions (r = 13%) were synthesized in boiling TEG. In order to prepare other particle morphologies, a modified polyol reduction strategy was employed. Firstly, the syntheses were carried out at lower temperatures to achieve kinetically stable particle shapes. Second, Pt(acac)2 was used with excess amounts of acetylacetone which was externally added to the reaction solution. For example, reacting 0.05 mmol of metal salt with 1.0 mmol acetylacetone in TEG at 180 °C produced 6.8 nm Pt nano-cubes. Carrying out the reaction at 160 °C via a two-step nucleation/growth protocol yielded to 6.3 nm Pt octahedra. A total of *100 NPs from more than one TEM micrograph


Fig. 1 Particle size histograms of the size-controlled Pt NPs. The table shows the mean sizes and standard deviations

were counted for each Pt sample to create the particle size histograms (Fig. 1 and S1). Figure 2 shows TEM, HRTEM images and particle size histograms of the shape-controlled Pt NPs with spherical (Fig. 2a–c), octahedron (Fig. 2d–f), truncated octahedron (Fig. 2g–i) and cubic (Fig. 2j–l) shapes. Additional HRTEM images for the Pt nano-truncated octahedra are provided in Fig. S8. 2.5.1 SFG Vibrational Spectrosopy Using Ethylene as Probe for the Determination of Crystallographic Shapes of NPs The shape-controlled nanoparticles were further investigated by SFG using ethylene molecules as probe. It was well documented that ethylene has three binding geometries on Pt single crystals [21, 22]. These are namely ethylidyne, –CHCH3; di-r-bonded ethylene, –CH2=CH2–; and p-bonded ethylene, CH2=CH2. Ethylidyne adopts the threefold hollow site binding and di-r-bonded ethylene has the bridging geometry. As a result, ethylidyne favors the threefold atomic arrangement present on the Pt(111) crystal. In fact, a threefold arrangement favors ethylidyne on Pt(111) at 300 K [22]. Second, Pt(100) crystal has the fourfold arrangement of atoms (i.e. tetragonal holes). Thus, Pt(100) crystal could accommodate ethylene on both of the ethylidyne and di-r-bonded ethylene configurations. In accord with this, single crystal adsorption studies carried out at 300 K showed signature peaks of ethylidyne at 2,875 cm-1 and di-r-bonded ethylene at 2,910 cm-1, the former having higher intensity (or surface abundance) [23]. Having this in mind, we investigated the shape-controlled Pt NPs by SFG vibrational spectroscopy in 10 torr C2H4, 100 torr H2 and 650 torr He and at 300 K. Our results are in agreement with the Pt single crystal studies in the case of Pt nano-cubes and Pt nanotruncated octahedra (Fig. 3). Pt nano-cubes exhibited a strong intensity vibrational stretch at 2,875 cm-1 and a

917

minor one at 2,910 cm-1 in their SFG spectrum, as expected from (100) surfaces. Differently, the SFG spectrum of Pt truncated octahedra revealed a much weaker peak at 2,910 cm-1, suggesting the involvement of (111) surfaces along with (100) surfaces. For the Pt octahedra, the 2,910 cm-1 peak due to di-r-bonded ethylene was of comparable strength as the 2,875 cm-1 peak, which ruled out the presence of (111) dominant surfaces. The peak intensities of the 2,875 and 2,910 cm-1 vibrational stretches, in fact, suggested the presence of high index open surfaces, like Pt(110) surfaces. Pt(110) surfaces are subject to restructuring, which makes difficult the SFG investigation of ethylene on open surfaces for a direct proof of our attempted correlation of surface structure and binding of adsorbed ethylene. However, DFT calculations predicted that ethylidyne and di-r-bonded ethylene would have similar binding energies on Pt(211) [24] and Pt(110) [25] crystals. This lack of binding preference could translate into similar surface coverage and in turn, SFG signature peaks of comparable intensities (Fig. 3). We explain the formation of Pt nano-octahedra with 110 surfaces rather than 111 surfaces by an elongation in the growth axis (or basal plane) during the kinetically-controlled synthesis [26] as depicted in Fig. S9. 2.5.2 Ethylene Hydrogenation Reaction for the Determination of Concentration of Surface Pt Sites Ethylene/hydrogen reaction on Pt surfaces is shown to be structure insensitive, meaning that Pt surfaces turnover irrespective of the particle size and morphology. This unique property of Pt surfaces opens up the possibility of employing the ethylene/hydrogen reaction as a means of measuring surface area. The catalytic turnovers could be translated to the concentration of available active surface sites for any given Pt nanoparticle. In order to explore the feasibility of this approach, we measured TOFs for the size- and shape-controlled Pt NPs in 10 torr C2H4 and 100 torr H2 and at 294 K. The results for the size-controlled Pt NPs were plotted in Fig. 4, and suggested a power law with exponent -1 for the plot of surface Pt sites and TEM-projected particle sizes, as expected from the inverse relationship between surface area and size. Furthermore, a plot of particle sizes as determined by C2H4/H2 reaction and sizes determined from TEM revealed a regression line of slope 1.1 (see Fig. S2). This suggests approximately 20 at.% of TEM-projected surface (i.e. 100% * (1.12 - 1)) is not available for the reactants due to PVP (and possibly by-products of the nanoparticle synthesis) binding regardless of the particle size, and illustrates the applicability of this novel and robust technique for the measurement of active surface areas.

123

918

S. Alayoglu et al.

Fig. 2 TEM, HR-TEM images and particle size histograms of 5.2 nm Pt spheres (a–c), 6.3 nm Pt nano-octahedra (d–f), 6.2 nm Pt truncated octahedra (g–i) and 6.8 nm Pt cubes (j–l). Next to the particle size histograms, the model clusters are given to illustrate the particle morphologies presented. The scale bars are 20 nm for TEM images and 2 nm for HR-TEM mages

Fig. 3 SFG vibrational spectra of 6.8 nm Pt cubes (blue), 6.2 nm Pt truncated octahedra (red) and 6.3 nm Pt octahedra (green) in the methyl stretch region. Dashed lines show the position of ethylidyne and di-r-bonded ethylene in increasing wavenumbers, respectively

123

Fig. 4 Plots of TEM-projected particle sizes versus surface Pt per gram Pt catalyst (right axis) and activation energy (left axis). Best fit and power law equation are shown for the plot of surface Pt per gram Pt catalyst versus particle size


919

3 Results 3.1 MCP/H2 Reaction Activity of the Size- and ShapeControlled Pt NPs The size- and shape-controlled Pt NPs were studied for the MCP/H2 reaction using 10 torr MCP and 50 torr H2 at 1 atm and in the temperature window between 160 and 300 °C. The turnover rates were calculated using the surface Pt sites determined from the C2H4/H2 reaction. The apparent activation energies were typically measured in the 160–240 °C range for the formation of C6 isomers (e.g. 2-methylpentane, 3-methylpentane, hexane, etc.) and cracking products (i.e. C1–C5), and 240–300 °C range for the formation of aromatics (e.g. benzene). For the shapecontrolled Pt NPs, the highest turnover rates were obtained for the 6.2 nm Pt truncated octahedra and 5.2 nm Pt spheres. In contrast, 6.8 nm Pt cubes showed the lowest TOFs almost two orders of magnitude lower than the most active particles for all the temperature range and partial pressures studied (Fig. 5). Next to the truncated octahedral and spherical particles, 6.3 nm octahedral Pt performed 3–8 times slower for the MCP/H2 reaction. The apparent activation energies are similar in the case of truncated octahedra and spherical Pt. For the octahedral Pt NPs, the activation energies for the formation of the ring opening products were non-Arrhenius. For the formation of aromatics, the activation energy was *45 kcal/mol on the (110) nano-octahedra. Figure 6 shows the TOFs at 240 and 280 °C, and apparent activation energies for the different reaction products plotted against the particle sizes. As determined from the curves, 1.5 nm Pt particles showed the lowest turnover rates at both 240 and 280 °C, and 2–3 times slower turnovers than the most active 5.2 nm Pt particles. The catalytic turnovers dropped again to slower rates above

Fig. 6 TOFs per hour were plotted against TEM-projected particle sizes (a) A plot of the activation energy versus TEM-projected particle sizes for all three classes of MCP hydrogenation products (b)

5.2 nm Pt (Fig. 6a). The apparent activation energies for all the product types (namely C6 isomers, aromatics and C1– C5) were the highest for 1.5 nm Pt. For the ring opening products (e.g. C1–C5 and C6 isomers), the activation energies in excess of 30 kcal/mol were obtained on the smallest Pt NPs. Similarly, the activation energy for the formation of aromatics was measured to be *45 kcal/mol. In contrast, the energy barriers for the ring opening products were 10 kcal/mol and the lowest in the size range between *3 and *5 nm, and climbed to slightly higher values on the larger Pt NPs (Fig. 6b). The reaction orders were also measured at 240 °C, and typically found to be 0.5 ± 0.3 on MCP. The reaction order on H2 was positive (?) 0.71 ± 0.01 over 5.2 nm Pt spheres and 6.2 nm Pt. Differently, the turnover rates were negative (-) order with respect to H2 over 1.5 nm Pt particles (-0.35) and 6.3 nm octahedral Pt (-0.53)(see Figs. S6 and S7). 3.2 MCP/H2 Reaction Selectivity of the Sizeand Shape-Controlled Pt NPs

Fig. 5 TOFs per hour were plotted against various particle morphologies, namely 6,3 nm Pt octahedra (Oh), 6.2 nm Pt truncated octahedra (TOh), 5.2 nm spherical Pt and 6.8 nm Pt cubes

The product distributions were also evaluated for the sizeand shape-controlled Pt NPs in Fig. 8. In the case of the shape-controlled NPs (6.2 nm truncated octahedra, 6.8 nm cubes, 5.2 nm spheres and 6.3 nm octahedra), Pt nanocubes produced the largest fraction of cracking products at temperatures below 240 °C. At temperatures above 240 °C, cubic and octahedral Pt nanoparticles yielded the

123

920

S. Alayoglu et al.

Fig. 7 % Selectivity toward various reaction products were plotted against temperature for 1.5 nm Pt (a), 7.2 nm Pt spheres (b), 6.3 nm Pt octahedra (c) and 6.8 nm Pt cubes (d)

largest fraction of benzene. At 240 °C, Pt nano-octahedra produced benzene and C6 isomers at equi-moles. Similarly, Pt nano-cubes showed little selectivity for different reaction pathways and produced *35 mol% C6 isomers and *40 mol% benzene at 240 °C (Fig. 8a). The other particle morphologies (spheres and truncated octahedra) were more selective towards C6 isomers. For example, Pt spheres produced 95 mol% C6 isomers and only 5 mol% cracking products (Fig. 8a). Among the C6 isomers produced, hexane was the major product with 40 mol% selectivity for octahedral Pt NPs, in contrast, was below 5 mol% for the other particle morphologies (Fig. 8b). Spherical Pt nanoparticles produced 2-methylpentane *70 mol% selectively. Pt cubes and truncated octahedra, similarly, were *30 mol% selective for 2-methylpentane (Fig. 8b). For the size-controlled Pt NPs, it was found that at 200 °C the major reaction products are the C6 isomers. No benzene was detected at this temperature, even in the batch mode upon prolonged (*2 h) accumulation of products. For example, 5.2 nm Pt particles produced 87 mol% C6 isomers and 13 mol% C1–C5 at 200 °C (Fig. 9a). No significant differences were observed for the size controlled Pt NPs at low temperatures. At temperatures starting from 240 °C and above, the size-controlled Pt NPs produced mainly benzene, in expense of C6 isomers. The cracking products (C1–C5) remained mainly unchanged at higher temperatures for all the Pt NPs. Figure 9b shows the product selectivity at 280 °C for the size-controlled Pt NPs. At 280 °C, 5.2 nm Pt particles produced *65 mol% benzene, and only *20 mol% C6 isomers, whereas 1.5 nm Pt particles exhibited a complete opposite trend with

123

*25 mol% benzene and 60 mol% C6 isomers. To this end, 1.5 nm Pt particles showed higher product selectivity for C6 isomers than any other Pt NPs. Among the C6 isomers, Pt NPs were 71 ± 9 mol% selective for 2-methylpentane at 200 °C (Figs. 7 and S3). At 280 °C, the 2-methylpentane selectivity dropped to 63 ± 6 mol% (Figs. 7 and S3). The C6 isomer selectivity showed a small size dependence for the Pt NPs studied. For 10.8 nm Pt particles, the 2-methylpentane selectivity was 55 mol% at 200 °C compared to an average of 75 ± 3 mol% for the other Pt NPs (Fig. S3d).

4 Discussions 4.1 MCP/H2 Reaction Over the Shape-Controlled *6 nm Pt Catalysts First, we studied the MCP/H2 reaction using 10 torr MCP and 50 torr H2 at 1 atm and at temperatures between 160 and 300 °C over the shape-controlled *6 nm Pt catalysts. The reactivity trends for the shape-controlled Pt NPs strongly supported the structure sensitive nature of the MCP/H2 reaction over Pt. Firstly, 6.2 nm Pt truncated octahedra with (100) and (111) surfaces exhibited the highest TOFs and the lowest activation energies for all classes of products in whole temperature window studied. The catalytic reactivity and product selectivity of the truncated octahedral Pt NPs were similar to those of the spherical Pt NPs of comparable particle sizes. For Pt nanocubes with (100) dominant surfaces, TOFs were the lowest,


921

Table 1 The apparent activation energies for three different reaction pathways over various Pt nanoparticle morphologies and Pt single crystals

C6 Isomersa

–c

13.1

34.4

15.4

34.1

11

44.2

19.5

25.6

27.2

43.5

[30

–

–c

21.3

40.4

19.9

30.5

25

26

Aromatics C1–C5a a

Cube

Spherical (5.2 nm)

Spherical (1.5 nm)

Pt(100)d

Oh

b

TOh

Pt(lll)d

Ea (kcal/mol)

9

Measured typically between 160 and 240 °C

b

Measured typically between 240 and 300 °C

c

Non-Arrhenius behavior observed

d

20 torr MCP and 200 torr H2. see also Ref. [1]

approximately two orders of magnitude lower than the most active Pt NPs, and the activation energies were among the largest (Fig. 6; Table 1). In addition, Pt nanocubes led to the largest fraction of cracking, which was consistent with the results of the single crystal studies [1]. The trends in selectivity towards various products turned out to be very important for different sizes and morphologies of Pt NPs. For instance, 6.3 nm Pt (110) octahedra was the most selective toward hexane (maximum of *80 mol% at 220 °C) and produced the smallest fractions of 2-methylpentane and benzene than any other particle morphology (Figs. 7 and 8). At 240 °C, the 3MP/Hex ratio reached at a minimum of *0.1 (i.e. below the thermodynamic ratio of 0.5) for octahedral Pt NPs. Under identical reaction conditions, 6.8 nm Pt (100) cubes, in contrast, was much more selective toward benzene (45 mol%). Accordingly, Pt nano-cubes produced the smallest fraction of hexane ([1 mol%,) with the largest 3MP/Hex mole ratio calculated. Indeed, the 3MP/Hex mole ratio was *45 (i.e. well above the thermodynamic ratio of 0.5) for Pt nanocubes (see Fig. S4). Notably, Pt nano-cubes led to the largest fraction of C1–C5 products via cracking for the whole temperature range studied. 4.2 MCP/H2 Reaction Over Monodisperse Pt NP Catalysts Between 1.5 and 10.8 nm In case of the size-controlled Pt NPs, the size dependence was observed for the turnover rates and apparent activation energies. For 5.2 nm Pt NPs, TOF was 43 molecule/Pt*h-1 at 280 °C. Moreover, the activation energies for the formation of C6 isomers, aromatics and C1–C5 products were 14, 27 and 20 kcal/mol, respectively. Compared to unsupported Pt(111) single crystal studied under 20 torr MCP and 200 torr H2, TOF was of comparable magnitude. Similarly, low turnover rates for the MCP/H2 reaction over supported (mesoporous-SiO2) Pt catalysts was documented by others.[8, 14–16] Santen and co-workers reported turnover rates in the order of *5 molecule/Pt*h-1 over the 1.8 nm (TEM) Pt particles at H2 to MCP partial pressure

Fig. 8 % Selectivity versus various nanoparticle morphologies were plotted for three operative reaction pathways, namely C6 isomerization via ring opening, benzene formation via ring enlargement/ dehydrogenation and C1–C5 formation via cracking, (a) and for three major C6 isomers, namely 2-methylpentane (2MP), 3-methylpentane (3MP) and hexane (HEX) (b) at 240 °C. The representative nanoparticle morphologies are also shown at the bottom for 6,3 nm Pt octahedra (Oh), 6.2 nm truncated octahedra Pt (TOh), 5.2 nm spherical Pt and 6.8 nm Pt cubes

ratio of 40 and at 270 °C [14]. For a reaction feed of 40 torr MCP and 720 torr H2 and at 280 °C, Anderson et al. found turnover rates of 36 molecule/Pt*h-1 on the 6.7 nm Pt/SiO2 and 68 molecule/Pt*h-1 on the 9.4 nm Pt/ SiO2 [16]. In addition, the activation energies were in good accord with the single crystal studies (Table 1) [1]. For the particle sizes down to *3 nm, the activation energies for the formation of C6 isomers remained mainly unchanged.

123

922

Fig. 9 % Selectivity versus TEM-projected particle sizes were plotted for three major reaction pathways, namely C6 isomerization via ring opening, benzene formation via ring enlargement/dehydrogenation and C1–C5 formation via cracking, at 200 °C (a) and 280 °C (b)

However, benzene and cracking products exhibited progressively decreasing activation energies (Fig. 5). For 1.5 nm Pt NPs, TOF were the lowest (e.g. 19 molecule/ Pt*h-1 at 280 °C), and the activation energies were the highest for the entire classes of products ([30 kcal/mol). The product selectivity at elevated temperatures also revealed the structure sensitive behavior of Pt for the MCP/ H2 reaction, which showed dramatic discrepancies for the 1.5 nm Pt at [240 °C. For example, 7.2 nm Pt particles produced 60 mol% benzene and 35 mol% C6 isomers at 280 °C (Figs. 7b and 9b). Among the C6 isomers produced, 2-methylpentane was *20 mol% selective. In contrast, 1.5 nm Pt particles produced 44 mol% 2-methylpentane and only 33 mol% benzene (Figs. 7a and 9b). The ratios of the ring opening products also highlighted the selective nature of the size-controlled Pt NPs. For example 2MP/ 3MP ratio was 2.1 and 3MP/Hex ratio was 1.2 over 10.8 nm Pt particles, larger than the thermodynamic ratios of 2 and 0.5, respectively. 2MP/3MP ratio progressively climbed to *3.3 and 3MP/Hex ratio to *6.7 with decreasing particle sizes down to 5.2 nm (see Fig. S5). For the MCP/H2 reaction, others also documented the size dependent selectivity toward the formation of branched isomers (especially of 2-methylpentane) over Pt (supported on various metal oxides). In line with our findings, a rather weak correlation on the Pt/SiO2 system was reported by others [27–29]. The mole ratios of various C6 isomers were

123

S. Alayoglu et al.

also in agreement with the findings over Pt single crystal surfaces under similar conditions [1]. However, unsupported Pt single crystal surfaces produced no significant amounts of benzene even at high temperatures (e.g. [1 mol% at 300 °C on Pt(111)), and deviated drastically from the nanoparticles. For our size- and shape-controlled Pt, this catalytic trend toward the formation of benzene had similarities with the studies of Alvarez and co-workers [30]. That aspect of the MCP/H2 reaction highlights the importance of the materials gap between catalysis and surface science, and thus requires further investigation. In our catalytic studies, the ring opening and isomerization were the main reaction pathways operative at low temperatures. However, benzene formation lit off at about 240 °C, and was the major reaction pathway at elevated temperatures. Indeed, it was proposed that C6 isomers, which are the ring opening product of MCP, form cyclohexane via an intra-molecular C–C bond making.[1] As shown in the Scheme 1, benzene is then produced from cyclohexane via dehydrogenation. It was also reported that cyclohexene exclusively dehydrogenates to benzene (i.e. no hydrogenation to cyclohexane) at 1 atm and temperatures above 240 °C in H2 partial pressures up to 500 torr [31]. This explains as to why no cyclohexane (or cyclohexenes) was observed for any of our nanoparticle catalysts. The nanoparticle catalysts show excellent stability against sintering after many regeneration and reaction steps at elevated temperatures up to 300 °C (Fig. 10). However, some morphological changes were observed mainly in the form of rounding of sharp edges in the case of shapecontrolled Pt nanoparticles [32]. Inset in Fig. 10 h shows a representative HR-TEM image of an octahedral Pt nanoparticle with round edges after 10 regeneration and reaction (at temperatures between 160 and 300 °C) cycles. Similarly, supported nanoparticles of Pt truncated octahedra rounded and became spherical after 10 reaction and regeneration cycles. Having said that the geometrical shapes slightly changed in the course of reaction, the shape-controlled nanoparticle catalysts maintained their intrinsic activity and selectivity, suggesting that the crystallographic shapes were indeed intact. At low temperatures, the catalytic activity was restored after the regeneration of catalysts at 150 °C. Following the reactions at elevated temperatures, the regeneration cycle (i.e. oxidation and consecutive reduction at 150 °C for 30 min) could restore the catalytic activity only partially (*30–40% less conversion for any given catalyst).

5 Conclusion In this paper, we studied the hydrogenation of methylcyclopentane over the size-controlled spherical Pt NPs in


923

Scheme 1 Reaction pathways and possible products of methylcyclopentane hydrogenation reaction

Fig. 10 Representative TEM images of the supported Pt nanoparticles before (top) and after (bottom) the MCP/H2 reaction under 10 torr MCP and 50 torr H2 at temperatures up to 300 °C, 1.5 nm

(a and e), 5.2 nm (b and f), 10.8 nm (c and g), and 6.3 nm Pt nanooctahedra (d and h). The inset in (h) shows a representative HR-TEM image

the size range between 1.5 and 11 nm, and *6 nm shapecontrolled Pt NPs with cube, octahedral and truncated octahedral particle morphologies. All Pt NPs were synthesized employing the modified polyol reduction reaction routes. As-synthesized nanoparticles were then characterized using TEM, high-resolution TEM, SFG vibrational spectroscopy using ethylene as probe, and catalytic ethylene hydrogenation reaction. Finally, we evaluated the catalytic properties of the Pt NPs supported in mesoporous

silica towards the MCP/H2 reaction under 1 atm of reactants and at temperatures between 160 and 300 °C. The effect of particle morphologies on the activity and selectivity toward the MCP/H2 reaction was evaluated for the *6 nm Pt with cubic, octahedral, truncated octahedral and spherical morphologies. It was found (1) Pt (100) nanocubes produced the largest fraction of C1–C5 via cracking (maximum of 60 mol% at 200 °C), (2) Pt (110) nanooctahedra produced the largest fraction of hexane

123

924

(maximum of 80 mol% at 220 °C) and the smallest fraction of 2-methylpentane (maximum of *5% at 200 °C), and (3) Pt spheres and truncated octahedra showed higher turnovers than octahedral (*8 times at 240 °C) and cubic (*50 times at 240 °C) nanoparticles for the whole temperature range studied. For the Pt size series, we found (1) ring opening and C6 isomerization is the major reaction pathway and 2-methylpentane was the major C6 isomer produced for the 1.5 nm Pt, (2) differently, ring enlargement and dehydrogenation to benzene was the leading reaction pathway at temperatures above 240 °C for the larger Pt NPs (i.e. [3.0 nm). Acknowledgments This work is funded by Office of Science, Department of Energy. The authors acknowledge support of the National Center for Electron Microscopy, Lawrence Berkeley Lab, which is supported by the U.S. Department of Energy under Contract # DE-AC02-05CH11231. Work at the Molecular Foundry was supported by the Director, Office of Science, Office of Basic Energy Sciences, Division of Material Sciences and Engineering, of the U.S. Department of Energy under Contract # DE-AC02-05CH11231.

References 1. Zaera F, Godbey D, Somorjai GA (1986) J Catal 101:73–80 2. Narayanan R, El-Sayed MA (2004) Nano Lett 4:1343–1348 3. Kuhn JN, Tsung CK, Huang W, Somorjai GA (2009) J Catal 265:209–215 4. Tsung CK, Kuhn JN, Huang WY, Aliaga C, Hung LI, Somorjai GA, Yang PD (2009) J Am Chem Soc 131:5816–5822 5. Kuhn JN, Huang WY, Tsung CK, Zhang YW, Somorjai GA (2008) J Am Chem Soc 130:14026 6. Kliewer CJ, Aliaga C, Bieri M, Huang WY, Tsung CK, Wood JB, Komvopoulos K, Somorjai GA (2010) J Am Chem Soc 132:13088–13095 7. Brandenberger SG, Callender WL, Meerbott WK (1976) J Catal 42:282–287

123

S. Alayoglu et al. 8. Zhuang YP, Frennet A (1999) Appl Catal A Gen 177:205–217 9. Galperin LB, Bricker JC, Holmgren JR (2003) Appl Catal A Gen 239:297–304 10. Bai XL, Sachtler WMH (1991) J Catal 129:121–129 11. Teschner D, Paal Z, Duprez D (2001) Catal Today 65:185–190 12. Samoila P, Boutzeloit M, Especel C, Epron F, Marecot P (2009) Appl Catal A Gen 369:104–112 13. Poupin C, Pirault-Roy L, La Fontaine C, Toth L, Chamam M, Wootsch A, Paal Z (2010) J Catal 272:315–319 14. Vaarkamp M, Dijkstra P, Vangrondelle J, Miller JT, Modica FS, Koningsberger DC, Vansanten RA (1995) J Catal 151:330–337 15. Kramer R, Fischbacher M (1989) J Mol Catal 51:247–259 16. Anderson JBF, Burch R, Cairns JA (1987) J Catal 107:351–363 17. Glassl H, Hayek K, Kramer R (1981) J Catal 68:397–405 18. Rioux RM, Song HJ, Yang PD, Somorjai GA (2005) Abstr Pap Am Chem S 229:U723–U723 19. Fievet F, Lagier JP, Blin B, Beaudoin B, Figlarz M (1989) Solid State Ionics 32–3:198–205 20. Alayoglu S, Eichhorn B (2008) J Am Chem Soc 130:17479–17486 21. Horiuti I, Polanyi M (1934) T Faraday Soc 30:1164–1172 22. Cremer PS, Su XC, Shen YR, Somorjai GA (1996) J Am Chem Soc 118:2942–2949 23. McCrea KR, Somorjai GA (2000) J Mol Catal A Chem 163:43–53 24. Watwe RM, Cortright RD, Norskov JK, Dumesic JA (2000) J Phys Chem B 104:2299–2310 25. Anghel AT, Jenkins SJ, Wales DJ, King DA (2006) J Phys Chem B 110:4147–4156 26. Sun YG, Xia YN (2002) Science 298:2176–2179 27. Hayek K, Kramer R, Paal Z (1997) Appl Catal A Gen 162:1–15 28. Hayek K, Fuchs M, Klotzer B, Reichl W, Rupprechter G (2000) Top Catal 13:55–66 29. Corolleu C, Juttard D, Muller JM, Gault FG, Maire G (1972) J Catal 27:466 30. Alvarez WE, Resasco DE (1996) J Catal 164:467–476 31. Rioux RM, Hsu BB, Grass ME, Song H, Somorjai GA (2008) Catal Lett 126:10–19 32. Rupprechter G, Freund HJ (2001) Top Catal 14:3–14