Size Regulation and Stability Enhancement of Pt Nanoparticle

Sep 21, 2017 - A novel multiwall carbon nanotube (MWCNT) and polypyrrole (PPy) composite was found to be useful for preparing durable Pt nanoparticle ...
0 downloads 8 Views 4MB Size
Download PDF
Article pubs.acs.org/Langmuir

Size Regulation and Stability Enhancement of Pt Nanoparticle Catalyst via Polypyrrole Functionalization of Carbon-NanotubeSupported Pt Tetranuclear Complex Satoshi Muratsugu,*,† Shota Miyamoto,† Kana Sakamoto,† Kentaro Ichihashi,† Chang Kyu Kim,† Nozomu Ishiguro,§ and Mizuki Tada*,†,‡,§ †

Department of Chemistry, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8602, Japan Research Center for Materials Science (RCMS) & Integrated Research Consortium on Chemical Science (IRCCS), Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Aichi 464-8602, Japan § Element Visualization Team, Materials Visualization Photon Science Group, RIKEN SPring-8 Center, 1-1-1 Koto, Sayo, Hyogo 679-5198, Japan ‡

S Supporting Information *

ABSTRACT: A novel multiwall carbon nanotube (MWCNT) and polypyrrole (PPy) composite was found to be useful for preparing durable Pt nanoparticle catalysts of highly regulated sizes. A new pyrene-functionalized Pt4 complex was attached to the MWCNT surface which was functionalized with PPy matrix to yield Pt4 complex/PPy/MWCNT composites without decomposition of the Pt4 complex units. The attached Pt4 complexes in the composite were transformed into Pt0 nanoparticles with sizes of 1.0−1.3 nm at a Pt loading range of 2 to 4 wt %. The Pt nanoparticles in the composites were found to be active and durable catalysts for the N-alkylation of aniline with benzyl alcohol. In particular, the Pt nanoparticles with PPy matrix exhibited high catalyst durability in up to four repetitions of the catalyst recycling experiment compared with nonsize-regulated Pt nanoparticles prepared without PPy matrix. These results demonstrate that the PPy matrix act to regulate the size of Pt nanoparticles, and the PPy matrix also offers stability for repeated usage for Pt nanoparticle catalysis.

1. INTRODUCTION

Several methods for stabilizing Pt nanoparticles have been reported. Stabilizing reagents such as polymers and dendrimers provide highly stable Pt nanoparticles.9−11 Dendrimers are used to control Pt particle size because they allow for assembly of only a limited number of Pt precursors.6,7,12 However, the presence of organic materials is expected to decrease catalytic activity since the active sites of the Pt nanoparticles can become blocked. Encapsulation of Pt nanoparticles in porous supports such as porous silica,11 porous carbons,13 and carbon nanotubes (CNTs)14,15 is expected to suppress aggregation of Pt nanoparticles and offer recyclability. Pores sometimes alter the catalytic performance of Pt nanoparticles due to interactions between the Pt nanoparticle and pore walls.15

Platinum (Pt) nanoparticles have attracted much attention for use in a wide range of catalytic processes such as oxidation of volatile organic compounds,1 hydrogenation of olefins and aromatics,2 and fuel cell applications.3 To increase catalytic activity and control selectivity, it is necessary to regulate the structure of Pt nanoparticles.2 One effective and widely used method is size control of Pt nanoparticles, and it has been reported that Pt nanoparticles/nanoclusters with sizes of a few nanometers exhibit higher catalytic activity compared with larger Pt nanoparticles.4,5 It has also been shown that even Pt nanoclusters consisting of several Pt atoms have catalytic activity (e.g., in the oxygen reduction reaction).6,7 However, finely sized Pt nanoparticles are prone to aggregation under reaction conditions and lose catalytic activity unless appropriate stabilizing systems are employed.8 © XXXX American Chemical Society

Received: June 21, 2017 Revised: August 30, 2017

A

DOI: 10.1021/acs.langmuir.7b02114 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 1. Schematics of the attachment of 1 on the surface of MWCNT, the grafting of polypyrrole, and the transformation to Pt nanoparticles.

recyclability for a model reaction (N-alkylation of aniline with benzyl alcohol) compared with nonsize-regulated Pt nanoparticles.

We have developed an effective way to enhance the stability of metal catalysts on a support surface.16−19 The functionalization of the support surface to which the metal complexes are attached by using surface matrix overlayers consisting of organic/inorganic polymers not only produces a protective wall near the supported metal complex that prevents the metal species from leaching but also produces a catalytic reaction space that controls the selectivity of the catalytic reaction based on a specific template ligand.16−19 The polymer stacks on the oxide surface in a layer-by-layer manner while maintaining an open space where the reactants and products are able to access the catalyst center at the top of the matrix overlayers. For example, a robust heterogeneous Mn catalyst for selective epoxidation could be prepared by attaching a Mn4 complex to SiO2 and successively stacking SiO2-matrix overlayers around a supported Mn cluster, and the stability of the catalyst was greatly improved when recycling was performed five times in the epoxidation of trans-stilbene.19 However, the current development of a supported metal complex precursor/ polymerized silica composite has been achieved on only SiO2 support surfaces. Herein, we applied our strategy to carbon supports in order to prepare a durable Pt nanoparticle catalyst with size regulation. A new Pt4 tetranuclear complex (1) with a Pt4(μ− OCOCH3)8 core20 and four pyrene moieties was prepared and attached to multiwalled carbon nanotubes (MWCNTs) through noncovalent interaction between pyrene and MWCNT. Organic polymer matrix made of polypyrrole (PPy) were then stacked on Pt4 complex/MWCNT and clusterization of the Pt4 complex/PPy/MWCNT composites proceeded. The sizes of Pt nanoparticles were highly regulated (Figure 1; mean diameter: 1.0−1.3 nm; full-width at halfmaximum (fwhm): 0.3−0.5 nm), compared with Pt nanoparticles prepared from Pt4 complex-supported MWCNT without PPy matrix. It was also demonstrated that the sizeregulated Pt nanoparticles exhibited higher stability and

2. RESULTS AND DISCUSSION 2.1. Attachment of 1 on MWCNT. Attachment of Pyrene-Functionalized Pt4 Complex on MWCNT. Pyrene is known to form a π−π stacking interaction with carbon supports.21−23 Since the π−π stacking interaction is a noncovalent interaction, it is not thought to affect or change the physical properties such as the electrical conductivity of carbon supports.21−23 This interaction has been applied to support metal complexes on CNTs to prepare multifunctional species such as donor−acceptor hybrids,24,25 electrodes,26,27 nanoelectronic devices,28 polymer composites,29,30 and catalysts.31,32 We employed this interaction to attach Pt4 complex to MWCNT. The π−π interaction between pyrene and carbon supports is in equilibrium in the solution phase,26,27,32 and controlling this equilibrium is the key factor for dispersed attachment. 1, the structure of which was fully characterized by NMR, FT-IR, ESI-TOF-MS, elemental analysis, UV−vis, fluorescence cyclic voltammogram, and DFT calculations (for details, see Supporting Information), was added to a MWCNT support (denoted FT9110 when an explicit expression is necessary) in dichloromethane suspension, and the mixture was stirred for a period of time. The mixture was then centrifuged and dispersed by ultrasonic treatment for washing. We employed fixed parameters for the centrifugation, ultrasonic treatment, and other procedures, and only changed the stirring time and the initial amount of added 1. The actual Pt loading after the attachment process was monitored by XRF. We first changed the stirring time, and plotted the actual Pt loading against stirring time (Figure 2A). The Pt loading was between 1.7 to 2.6 wt % when the stirring time was changed from 1 to 24 h, suggesting that equilibrium B

DOI: 10.1021/acs.langmuir.7b02114 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

of the Pt LIII-edge XANES (Figures 3(A), S2) were estimated to be both 1.9, respectively. Although the values were slightly

Figure 2. (A) Reaction time versus the amount of supported Pt plot of 1 on FT9110 obtained by XRF. (B) The amount of added Pt versus the amount of supported Pt plot of 1 on FT9110 obtained by XRF.

was reached rapidly, and we adopted 1 h for the remaining attachment procedures. We next changed the amount of 1 that was added to the suspension, and the actual Pt loading was plotted against the used Pt loading (Figure 2B). The actual Pt loading exhibited a proportional relationship with respect to the used amount of Pt until the used amount of Pt reached 4 wt %. The actual Pt loading saturated at around 4 wt % when more than 4 wt % of 1 was used, suggesting that the maximum Pt loading was around 4 wt % under the present conditions. The calculated maximum loading of 1 was 4.2 wt %, which was estimated by the surface area of FT9110 (174 m2 g−1, Table 1, Figure S1) and the cross-section area of 1, indicating that an almost quantitative attachment of 1 on FT9110 was achieved. The mean Pt loading of Pt4/CNT-1 was 2.0 ± 0.2 wt % and that of Pt4/CNT-2 was 4.0 ± 0.7 wt %, respectively (Table 1). The BET surface area of Pt4/CNT-1 was 163 m2 g−1 (Table 1, Figure S1), which was comparable to that of FT9110 (174 m2 g−1), implying that the aggregation or bundling of MWCNT did not occur during the attachment process of 1 on FT9110. Local Structure of Attached Pt4 in Pt4/CNT. The area of the white line peak at the Pt LIII-edge is known to be relative to the actual oxidation state of Pt species. The Pt oxidation states of Pt4/CNT-1 and Pt4/CNT-2 from the calibration curve analysis

Figure 3. (A) Normalized Pt LIII-edge XANES spectra of (a) Pt(acac)2, (b) 1, (c) Pt4/CNT-1, (d) Pt4/CNT/PPy-1, (e) nanoPt/ CNT/PPy-1, (f) nanoPt/CNT-1, and (g) Pt foil. (B) k3-Weighted Pt LIII-edge EXAFS Fourier transforms at 20 K (k = 30−180 nm−1) for (a) 1, (b) Pt4/CNT-1, (c) Pt4/CNT-2, (d) Pt4/CNT/PPy-1, (e) Pt4/ CNT/PPy-2, (f) nanoPt/CNT/PPy-1, and (g) nanoPt/CNT-1.

smaller than that of Pt(acac)2 and 1 (2.0) (Table 1), the local structures of Pt species in Pt4/CNT-1 and Pt4/CNT-2 were almost similar to that of 1 from the Pt LIII-edge EXAFS (vide infra). This suggests that the formal Pt oxidation states of Pt species in Pt4/CNT-1 and Pt4/CNT-2 are also +2. This value was also supported by Pt 4f XPS of Pt4/CNT-1 (73.1 and 76.4 eV, see section 2.2, Figure 4A). The local coordination structures of 1, Pt4/CNT-1, and Pt4/ CNT-2 were investigated by curve fitting analysis of the Pt LIIIedge EXAFS Fourier transforms (Figures 3B, S2, Table S1). The Pt LIII-edge EXAFS Fourier transforms of 1 was fitted with

Table 1. Characterization Data of Prepared Catalysts by XRF, BET, TGA, XPS, Pt LIII-edge XANES, and Raman Spectroscopy Pt wt % (XRF)

BET surface area /m2 g−1

TGA weight loss/wt % (temp/K)a

FT9110 1 Pt4/CNT-1

----2.0

174 --163

Pt4/CNT-2

4.0

157

Pt4/CNT/PPy-1

1.5

147

Pt4/CNT/PPy-2

3.2

140

nanoPt/CNT/PPy-1

1.8

174

nanoPt/CNT/PPy-2

3.4

169

nanoPt/CNT-1 nanoPt/CNT-2

2.0 3.8

170 173

98 (973) --2.5 (522) 95 (789) 5.1 (517) 89 (791) 21.3 (594) 74.5 (779) 25.4 (542) 70.3 (782) 18.4 (666) 76.4 (778) 14.7 (443, 607) 81.5 (778) -----

sample

a

N 1s XPS binding energy/eV

Pt 4f XPS binding energy/eV (4f5/2, 4f7/2)

Pt oxidation state estimated by Pt LIII-edge XANES (formal Pt oxidation state)

Raman shift (Gband, D-band) /cm−1

-------

----76.4, 73.1

--2.0 (+2) 1.9

1575.7, 1346.5 --1578.7, 1346.7

---

76.4, 73.2

1.9

1576.8, 1344.7

400.2

76.1, 72.9

2.0

1572.1, 1340.8

400.4

76.2, 73.0

1.7

1574.9, 1342.7

400.6

74.8, 71.4

0.5

1576.8, 1348.7

400.5

74.9, 71.5

0.5

1576.8, 1349.7

-----

74.4, 71.1 74.4, 71.3

0.5 0.5

-----

Estimated using the TGA differential curve in Figure S4. C

DOI: 10.1021/acs.langmuir.7b02114 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

When PPy was attached to Pt4/CNT-1 to form Pt4/CNT/ PPy-1, the Pt loading in Pt4/CNT/PPy-1 decreased to 1.5 wt %. If used PPy was ideally polymerized on Pt4/CNT-1 (Pt: 2.0 wt %), the Pt loading of Pt4/CNT/PPy-1 was calculated to be 1.3 wt %. This suggests that approximately 70% of PPy used for the preparation of Pt4/CNT/PPy actually remained on Pt4/ CNT/PPy. Indeed, a significant XPS peak of N 1s at 400.2 eV appeared in Pt4/CNT/PPy-1 (Figure 4B), which is negative in Pt4/CNT-1 and characteristic of pyrrole rings,13,42 supporting the fact that PPy was actually present on Pt4/CNT/PPy-1. The interaction between 1 and MWCNT is important to prevent from leaching of the Pt source during the PPy stacking process. We prepared Pt-loaded MWCNT/PPy (Pt: 2 wt %) by using the conventional Pt salt (Pt(NH3)2(NO3)2). The 2 wt % of the Pt salt was impregnated, and subsequent PPy attachment was performed in exactly a similar way to the preparation of Pt4/CNT/PPy-1. However, a serious decrease in the Pt loading was observed during the polymerization of PPy, because of the leaching of the impregnated Pt species into the solution phase. TGA analysis clearly showed the formation of PPy in Pt4/ CNT/PPy (Figures 5A, S3, S4, Table 1). Two types of weight

Figure 4. (A) Pt 4f XPS spectra of (a) Pt4/CNT-1, (b) Pt4/CNT/ PPy-1, (c) nanoPt/CNT/PPy-1, and (d) nanoPt/CNT-1. (B) N 1s XPS of (a) Pt4/CNT-1, (b) Pt4/CNT/PPy-1, (c) nanoPt/CNT/PPy1, and (d) nanoPt/CNT-1. Intensities of the spectra were normalized by C 1s XPS peak area.

three contributions: Pt−Pt at 0.250 ± 0.001 nm (CN = 2.2 ± 0.3) and two Pt−O contributions at 0.200 ± 0.001 nm (out-ofplane position, CN = 2.0 ± 0.6) and Pt−O at 0.215 ± 0.003 nm (in-plane position, CN = 1.6 ± 0.6), which are similar to the value observed from the X-ray single crystal structure of Pt4 complex (Table S1).33 The fitting parameters for the Pt LIIIedge EXAFS of Pt4/CNT-1 were Pt−Pt at 0.250 ± 0.001 nm (CN = 1.9 ± 0.2), Pt−O (out-of-plane) at 0.200 ± 0.002 nm (CN = 2.7 ± 0.7), and Pt−O (in-plane) at 0.216 ± 0.004 nm (CN = 1.5 ± 0.6), and those of Pt4/CNT-2 were Pt−Pt at 0.250 ± 0.001 nm (CN = 2.2 ± 0.2), Pt−O (out-of-plane) at 0.202 ± 0.003 nm (CN = 2.4 ± 0.5), and Pt−O (in-plane) at 0.218 ± 0.003 nm (CN = 1.5 ± 0.4), respectively, which are similar to the values of 1 (Table S1). These results strongly support maintenance of the Pt4 structure after the attachment of 1 on MWCNT, irrespective of the loading of 1. 2.2. Functionalization of Pt4/CNT with PPy Polymers. Preparation and the Actual Amount of PPy in Pt4/CNT/PPy. Chemical functionalization of CNTs with polymers is an efficient way to produce CNT/polymer composites that are utilized for materials such as conductive films, electrodes and electrocatalysts, supercapacitors, photovoltaic and optoelectronic devices, organic light-emitting diodes, infrared sensors, and solar cells.9,23,34,35 Among them, noncovalent exohedral functionalization of CNTs with polymers is a method for wrapping CNT in polymers without changing the physical properties of CNT.21,23,34,35 We prepared the catalyst precursor composite from Pt4/CNT-1 and PPy by using an in situ chemical oxidative polymerization method in which pyrrole was oxidized by hydrogen peroxide and polymerized by ammonium peroxydisulfate directly in the suspension of Pt4/CNT in water.36,37 We focused on PPy since it not only stacks well on carbon supports38 but also exhibits high electrical conductivity that is expected to enhance the electrochemical performance of carbon supports due to its π-conjugated system,39,40 and the combination of PPy and Pt4/CNT is expected to have potential applications in electrocatalysts.9,41 We refrained from using metal oxidants such as FeCl3 to prevent contamination by additional metal species.

Figure 5. (A) TGA curves of (a) FT9110, (b) Pt4/CNT-1, (c) Pt4/ CNT/PPy-1, and (d) nanoPt/CNT/PPy-1. (B) SEM micrographs of (a) FT9110, (b) Pt4/CNT-1, (c) Pt4/CNT/PPy-1, and (d) nanoPt/ CNT/PPy-1. The complete data are shown in Figure S5.

loss were observed: 376−734 K (center at 594 K) (weight loss: 21.3%) and 778−788 K (weight loss: 74.5%) on Pt4/CNT/ PPy-1. In the case of Pt4/CNT-1, a similar large weight loss was observed at 789 K but the weight loss at around 594 K was negligible (2.5% at 522 K). Therefore, the weight loss at around 780 K was attributed to MWCNT and that at around 594 K was attributed to PPy. Assuming that the 70% of the ideal amount of PPy with a PPy density43,44 of 1.48 g cm3 was attached on Pt4/CNT/PPy-1, the calculated weight ratio of PPy on Pt4/CNT/PPy-1 was 26.5%, which is comparable to the observed value. The small weight loss observed at 522 K in Pt4/ CNT-1 (2.5%) was attributed to the ligands of 1, and this peak was hidden by the weight loss of PPy in Pt4/CNT/PPy-1. (The large weight losses at 779 K in Pt4/CNT/PPy-1 and 789 K in Pt4/CNT-1 were attributed to MWCNT, which has combustion temperatures much lower than that of FT9110 (973 K), probably due to the presence of Pt species, which might act as a catalyst to activate the combustion of MWCNT.) Morphology and Location of PPy in Pt4/CNT/PPy. The change in the morphology of MWCNT before and after the PPy functionalization was measured by SEM (Figures 5B, S5). Many rough features were observed on the MWCNT in Pt4/ CNT/PPy-1, which could be attributed to PPy, while this roughness was seldom observed in Pt4/CNT-1 and FT9110. D

DOI: 10.1021/acs.langmuir.7b02114 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 6. (a) TEM images, (b) particle-size distributions, and (c) k3-weighted Pt LIII-edge EXAFS Fourier transforms (k = 30−180 nm−1; measured at 20 K) of (A) nanoPt/CNT/PPy-1, (B) nanoPt/CNT/PPy-2, (C) nanoPt/CNT-1, and (D) nanoPt/CNT-2.

The PPy cluster36 which was formed by the aggregation of only PPy was not observed in Pt4/CNT/PPy-1. The effect to the conjugation network of MWCNT before and after the PPy functionalization was investigated by Raman spectroscopy (Figure S6). Normalized Raman spectra of Pt4/ CNT/PPy-1, Pt4/CNT-1, and FT9110 are shown in Figure S6. The G-band represents the graphite-like structure of CNTs and the D-band derives from the defects in the graphite-like structure of CNTs.45 The peak positions of the G-band and the D-band of Pt4/CNT-1 were 1578.7 and 1346.7 cm−1, respectively (Table 1), which are similar to those of FT9110 (1575.7, 1346.5 cm−1), indicating that the attachment of 1 to FT9110 did not affect the structure of MWCNT. In contrast, the peak positions of the G-band and the D-band of Pt4/CNT/ PPy-1 were shifted to smaller wavenumbers (1572.1, 1340.8 cm−1). This peak shift might suggest aromatic (π−π) interaction between PPy and MWCNT.46 It is of interest whether PPy actually formed near the Pt4 complex at the surface of MWCNT. The change in intensity in the XPS Pt 4f peaks sensitively reflects the relative position of the polymer matrix and Pt species. When the polymer matrix

form near Pt species, the peak intensity of XPS signals become smaller due to blocking of photoelectrons by the polymer.16−19 If PPy polymerization proceeds separately from the Pt species, the XPS peak intensity is expected to be similar to the sample without polymer matrix. Figure 4A shows Pt 4f7/2 and 4f5/2 XPS spectra of Pt4/CNT-1 and Pt4/CNT/PPy-1. It is to be noted that the intensities of the Pt 4f peaks were observed to decrease after the stacking of PPy. Considering the escape depth of Pt 4f, these results clearly indicated that the supported Pt4 complexes on the surface of MWCNT were, to some extent, surrounded by the produced PPy (Figure 1). Local Structure of Attached Pt4 in Pt4/CNT/PPy. The Pt oxidation state of Pt4/CNT/PPy-1 estimated by Pt LIII-edge XANES analysis was 2.0, which is similar to that of 1 (2.0) and Pt4/CNT-1 (1.9) (Table 1), suggesting that the formal Pt oxidation state of Pt4/CNT/PPy-1 was also +2. Pt 4f7/2 and 4f5/2 XPS peaks for Pt4/CNT-1 were clearly observed at 73.1 and 76.4 eV, respectively, and those for Pt4/CNT/PPy-1 were observed at 72.9 and 76.1 eV, respectively. These results suggest that the actual oxidation states of Pt species in both E

DOI: 10.1021/acs.langmuir.7b02114 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

spectrum analysis were 1576.8 and 1348.7 cm−1, respectively (Figure S6, Table 1), which are similar to those of FT9110 (1575.7 cm−1, 1346.5 cm−1) and Pt4/CNT-1 (1578.7 cm−1, 1346.7 cm−1), suggesting that the reduction of CC double bonds in MWCNT or structural change of MWCNT to amorphous carbon did not proceed. Many rough features were observed on the MWCNT in the SEM micrographs of nanoPt/ CNT/PPy-1 (Figures 5B, S5), which could be attributed to PPy, since the structural change in MWCNT was almost negligible from Raman spectral analysis. Formation of Pt Nanoparticle on nanoPt/CNT/PPy. The formation of Pt nanoparticles in nanoPt/CNT/PPy-1 was confirmed by TEM analysis. Figures 1, 6A, and S8 show TEM images and the particle-size distribution of nanoPt/CNT/PPy1. A number of black dots are observed in the TEM images of nanoPt/CNT/PPy-1 which are not observed in the TEM images of Pt4/CNT/PPy-1, Pt4/CNT-1, and MWCNT (Figure S9), indicating that Pt nanoparticles were formed on nanoPt/ CNT/PPy-1. It is to be noted that the mean diameter of the Pt nanoparticles of nanoPt/CNT/PPy-1 was so small as 1.0−1.2 nm with fwhm of 0.3−0.5 nm (Figure 6). The high-resolution LAADF-STEM image in Figure 1 clearly showed the small Pt nanoparticles on the surface of MWCNT. When the reduction temperature was increased from 573 to 773 and 973 K, the average particle sizes were 1.2 ± 0.5 nm (773 K, Figure S10A) and 1.9 ± 0.5 nm (973 K, Figure S10B), respectively. The mean particle sizes were increased accompanied by the reduction temperatures, but it was still less than 2 nm at 973 K, suggesting the high thermal resistance of the Pt nanoparticles. We also investigated the stability of the Pt nanoparticles on nanoPt/CNT/PPy-1 prepared by the reduction with 573 K. The prepared sample was reduced with hydrogen at 973 K for 1 h, and changes in the particle size were investigated. The average particle size was 1.5 ± 0.5 nm (Figure S10C), which was slightly larger to that of the prepared sample before the reduction at 973 K. However, no drastic change in the Pt particle size (no formation of Pt particles larger than 2 nm) also showed the high stability of the prepared Pt nanoparticles. In the Pt 4f XPS spectrum of nanoPt/CNT/PPy-1, Pt 4f7/2 and 4f5/2 peaks were observed at 71.4 and 74.8 eV, respectively (Figure 4A, Table 1). These were shifted to smaller binding energies compared with Pt4/CNT-1 and Pt4/CNT/PPy-1, with the value suggesting that the oxidation state of Pt species in nanoPt/CNT/PPy-1 was zero, indicating that the Pt species was reduced to be metallic. The Pt LIII-edge XANES (Figures 3A, S11A) spectra also suggested the reduction of Pt in nanoPt/CNT/PPy-1, showing the formation of metallic Pt nanoparticles. The curve-fitting analysis of the Pt LIII-edge EXAFS Fourier transform revealed the local coordination structure of the Pt nanoparticle of nanoPt/CNT/PPy-1: Pt−Pt at 0.273 ± 0.001 nm (CN = 3.5 ± 1.1), Pt−O/N at 0.202 ± 0.001 nm (CN = 0.7 ± 0.3), and Pt−C at 0.221 ± 0.003 nm (CN = 1.3 ± 0.6) (Figures 3B, 6A, S11B,C, Table S1). The Pt− Pt distance was close to the values of Pt metal (0.277 nm), supporting the formation of metallic Pt nanoparticles. The contributions of Pt−O/N and Pt−C are thought to be from the remaining ligand, PPy, or MWCNT. 2.4. Size-Regulating Effect of PPy to Pt Nanoparticles. Size-Regulating Effect to Pt Loading. The particle-size distribution of the Pt nanoparticles (1.0−1.3 nm with fwhm of 0.3−0.5 nm) was remarkably narrow as shown in Figures 6 and S8. We investigated the role of PPy to the size-regulating

Pt4/CNT-1 and Pt4/CNT/PPy-1 were +2,41 supporting the Pt LIII-edge XANES analysis. The curve-fitting parameters for the Pt LIII-edge EXAFS of Pt4/CNT/PPy-1 (Figures 3B, S2, Table S1) were Pt−Pt at 0.250 ± 0.001 nm (CN = 1.5 ± 0.2), Pt−O (out-of-plane) at 0.200 ± 0.002 nm (CN = 2.3 ± 0.4), and Pt−O (in-plane) at 0.216 ± 0.004 nm (CN = 1.3 ± 0.4), which are similar to the values of Pt4/CNT-1. Although the CN value of the Pt−Pt contribution in Pt4/CNT/PPy-1 was slightly decreased, these results indicate that the Pt4 structure was mainly kept even after functionalization of the PPy matrix. Effect of Pt Loading on Pt4/CNT/PPy. Preparation of Pt4/ CNT/PPy-2 was conducted from Pt4/CNT-2 (the doubled Pt amount), and the actual Pt loading of Pt4/CNT/PPy-2 was found to be 3.2 wt % (Table 1), which suggests that approximately 80% of PPy (theoretically predicted value: 2.6 wt %) was actually existed on Pt4/CNT/PPy-2. The SEM micrograph of Pt4/CNT/PPy-2 also exhibited many rough features similar to Pt4/CNT/PPy-1, which could be attributed to PPy, while this roughness was seldom observed in Pt4/CNT2 (Figure S5). The peak intensities of Pt 4f XPS peaks in Pt4/ CNT/PPy-2 were also decreased (Figure S7). The Pt oxidation state of Pt4/CNT/PPy-2 estimated by Pt LIII-edge XANES analysis was 1.7, which is similar to that of 1 (2.0) (Table 1), indicating that the formal Pt oxidation state of Pt4/CNT/PP-2 was also +2. The fitting parameters for the Pt LIII-edge EXAFS of Pt4/CNT/PPy-2 (Figures 3B, S2, Table S1) were Pt−Pt at 0.250 ± 0.001 nm (CN = 1.6 ± 0.2), Pt−O (out-of-plane) at 0.200 ± 0.001 nm (CN = 2.4 ± 0.6), and Pt−O (in-plane) at 0.216 ± 0.002 nm (CN = 1.3 ± 0.6), respectively, which are similar to the values of Pt4/CNT-1, Pt4/CNT-2, and Pt4/CNT/ PPy-1. These results indicate that the functionalization of PPy matrix overlayers would not have a significant effect in distorting the Pt4 structure on MWCNT surface even when the Pt loading was doubled on Pt4/CNT/PPy-2. 2.3. Transformation of Pt4 Complex/PPy/MWCNT Composite into Pt Nanoparticle Catalysts (nanoPt/ CNT/PPy). Structure of PPy in nanoPt/CNT/PPy. The transformation of Pt4(μ−OCOCH3)8 cores into Pt nanoparticles was reported on SiO2 and γ-Al2O3 surfaces, indicating that the Pt4(μ−OCOCH3)8 cores have the potential to produce Pt nanoparticles by the removal of bridging ligands.47 The clusterization of Pt4 complexes to Pt nanoparticles (nanoPt/ CNT/PPy) was conducted by evacuation of Pt4/CNT/PPy at 573 K (ca. 10−2 Pa) for 1 h followed by reduction with hydrogen (40 kPa) at the same temperature for 1 h. The BET surface area of nanoPt/CNT/PPy-1 was 174 m2 g−1, which was larger than that of Pt4/CNT/PPy-1 (147 m2 g−1) (Table 1, Figure S1), suggesting the partial pyrolysis of PPy brought about the porous structure in the PPy matrix.13 The TGA weight loss observed in the region of 430 to 736 K (center at 666 K) in nanoPt/CNT/PPy-1 was 18.4%, which is smaller than that in Pt4/CNT/PPy-1 (21.3%), agreeing with the above results. In the N 1s XPS spectra of nanoPt/CNT/PPy-1 (Figure 4B), a peak at 400.6 eV characteristic of pyrrole ring13,42 was still clearly observed, indicating that PPy mostly remained in its original structure. A shoulder peak observed at 398.1 eV can be attributed to the value of the N atom in the pyridine ring (398.1 eV),13,42 suggesting that a transformation of pyrrole ring to pyridine ring might have occurred during the preparation process of nanoPt/CNT/PPy. The peak positions of the Gband and the D-band of nanoPt/CNT/PPy-1 from Raman F

DOI: 10.1021/acs.langmuir.7b02114 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

nanoPt/CNT-1 was also zero, agreeing the results of Pt LIIIedge XANES (Figures 3A, S11A). The intensities of the EXAFS Fourier transforms of nanoPt/CNT (Figure 6C,D) attributed to Pt−Pt bonds in Pt nanoparticles were much larger than those of nanoPt/CNT/PPy (Figure 6A,B), reflecting the formation of larger Pt particles in nanoPt/CNT. Indeed, the coordination numbers of Pt−Pt bonds on nanoPt/CNT-1 and nanoPt/CNT-2 were 4.8 ± 0.7 at 0.277 ± 0.001 nm and 5.6 ± 0.6 at 0.276 ± 0.001 nm, respectively (Table S1), which were larger than those of nanoPt/CNT/PPy-1 (3.5) and nanoPt/ CNT/PPy-2 (4.0), supporting the idea of larger Pt nanoparticles in nanoPt/CNT without PPy compared to nanoPt/ CNT/PPy with PPy. It should be noted that this Pt size-regulating effect by PPy did not depend on the Pt loading in the range of 2−4 wt %. The CN of Pt−Pt bonds estimated by Pt LIII-edge EXAFS of nanoPt/CNT/PPy-1 and nanoPt/CNT/PPy-2 was much smaller than the value of Pt nanoparticles with the observed diameter in TEM image (1.0−1.3 nm). The smaller CN (Pt− Pt) of nanoPt/CNT/PPy-1 and nanoPt/CNT/PPy-2 may be explained by the raft-like Pt nanoparticle structures of several monolayers. 2.5. Catalytic Performance of N-Alkylation of Aniline. Finally, we investigated the catalytic performance of nanoPt/ CNT/PPy and nanoPt/CNT for N-alkylation of aniline with benzyl alcohol,50 which is a useful reaction to produce a number of agrochemicals, pharmaceuticals, and bioactive molecules, as a model reaction to investigate the effect of PPy polymer matrix overlayers. Both catalysts exhibited good activity for the reaction and the conversion of aniline reached about 80% for nanoPt/CNT/PPy-2 (66 h) and nanoPt/CNT-2 (44 h) in the initial stage, with N-phenylbenzylamine produced as the main product (Table S2). Benzylideneaniline (99.8%; Fe content: