Thermal-Stable Carbon Nanotube-Supported Metal Nanocatalysts

(1-8) In contrast to porous carbons, most of the CNT surface is accessible to metal .... The liquid products were analyzed by GC (Agilent 6820) equipp...
0 downloads 0 Views 8MB Size
ARTICLE pubs.acs.org/Langmuir

Thermal-Stable Carbon Nanotube-Supported Metal Nanocatalysts by Mesoporous Silica Coating Zhenyu Sun, Hongye Zhang, Yanfei Zhao, Changliang Huang, Ranting Tao, Zhimin Liu,* and Zhenduo Wu Beijing National Laboratory for Molecular Sciences, Institute of Chemistry, Chinese Academy of Sciences, Beijing100190, P.R.China

bS Supporting Information ABSTRACT: A universal strategy was developed for the preparation of high-temperaturestable carbon nanotube (CNT) -supported metal nanocatalysts by encapsulation with a mesoporous silica coating. Specifically, we first showed the design of one novel catalyst, Pt@CNT/SiO2, with a controllable mesoporous silica coating in the range 1139 nm containing pores ≈3 nm in diameter. The hollow porous silica shell offers a physical barrier to separate Pt nanoparticles from contact with each other, and at the same time the access of reactant species to Pt was not much affected. As a result, the catalyst showed high thermal stability against metal particle agglomeration or sintering even after being subjected to harsh treatments up to 500 °C. In addition, degradation in catalytic activity was minimized for the hydrogenation of nitrobenzene over the catalyst treated at 300 °C for 2 h. The scheme was also extended to coat porous silica onto the surfaces of CuRu@CNT and the resultant catalyst thereby can be reusable at least four times without loss of activity for the hydrogenolysis of glycerol. These results suggest that the as-prepared nanostructured CNT-supported catalysts may find promising applications, especially in those processes requiring rigorous conditions.

1. INTRODUCTION Their remarkable structure-dependent properties, high surface areas, and excellent chemical and physical stabilities make carbon nanotubes (CNTs) attractive for use as heterogeneous catalyst supports.18 In contrast to porous carbons, most of the CNT surface is accessible to metal immobilization, which allows better contact between the reactant and the catalytic component. Significantly, the interactions between the active phase and nanotubes, in addition to the specific electronic and geometric structures of CNTs, could modify the reaction rate and reaction pathway.915 As an example, a very high selectivity for cinnamyl alcohol (up to 92%) was observed on Ru/CNT at an 80% conversion of cinnamaldehyde. This contrasts with only 2030% cinnamyl alcohol selectivity on Ru/Al2O3 with a similar Ru dispersion.10 Not coincidentally, complete ring saturation of polycyclic aromatic hydrocarbons was achieved under mild hydrogenation conditions over Rh/CNT, which otherwise could not be obtained with commercially available Rh nanocatalysts.13 More recently, an unprecedentedly high activity and selectivity were demonstrated over Pt/CNT for the hydrogenation of nitrobenzene to aniline.14 Despite the fact that CNT-supported metals have been widely investigated and have shown potential implications in catalysis, catalyst nanoparticles (NPs) deposited on nanotubes are unstable due to their high surface energies and the weak interaction between CNTs and NPs. Thus they tend to coalesce or sinter to minimize their chemical potential even at moderate temperatures. In essence, particle growth accompanied by a corresponding loss of active surface areas often leads to the deterioration of r 2011 American Chemical Society

catalytic activity and hence poor long-term stability. From this context, it is in high demand to improve the stability, recyclability, and selectivity of CNT-supported catalysts, especially to make them sustainable for high operating temperature utility.15 Current strategies to stabilize colloidal particles usually involve the use of organic species to restrain them from coming into contact with each other. Unfortunately, the organic capping molecules decompose typically above 300 °C; as a result, the NPs can deform and aggregate, which inevitably imposes unfavorable limitations for use in medium- and high-temperature reactions. It has been recently realized that metal NPs can be effectively stabilized against coalescence by encapsulation within metal oxide layers.1621 SiO2 has been widely applied as the oxide shell primarily due to its biocompatibility, excellent chemical stability, and versatile functionalization chemistry, coupled with the fact that silicon alkoxides are readily available and easy to work with. An example of this concept is that Pt NPs encaged in mesoporous SiO2 were stable and can withstand calcination in air at 750 °C.17 Similarly, porous-silica-coated individual Au particles and Fe3O4/SiO2/Au composites both demonstrated good stability owing to the porous shell for the catalytic liquid-phase reduction of 4-nitrophenol by NaBH4.19,20 Despite the stabilization of colloidal particles, few works have concerned the stabilization of CNT-supported catalysts, which thus presents a nontrivial challenge, particularly for the sake of Received: December 23, 2010 Revised: March 20, 2011 Published: April 11, 2011 6244

dx.doi.org/10.1021/la105087f | Langmuir 2011, 27, 6244–6251

Langmuir technological interest.22 Herein, we developed a simple, straightforward, and versatile method to coat porous SiO2 layers onto the exteriors of CNT-supported catalytic NPs with the aid of cetyltrimethylammonium bromide (CTAB). CTAB molecules can not only exceptionally stabilize carbon nanotubes or CNTbased hybrid composites in the aqueous phase but also act as an organic template for directing the formation of a mesoporous silica coating via the base-catalyzed hydrolysis of tetraethyl orthosilicate (TEOS) and subsequent condensation of silicate onto the surfactant surface. We first showed its applicability for a controllable coating of Pt@CNT with mesoporous silica, as both Pt and Pt@CNT are presently among the most frequently studied catalytic materials with remarkable properties. As far as we are aware, this is the first design of one novel catalyst, catalytic NP@CNT/mesoporous SiO2, that appears aesthetically intriguing for use especially in rigorous processes. The precursor mass ratio of TEOS-to-Pt@CNT (mTEOS-to-Pt@CNT) was tuned in an attempt to manipulate the thickness of the porous layer. The stability effect of the porous layer was evaluated in terms of resistance against metal particle sintering at a series of high temperatures, and comparisons were also made with the nonencapsulated Pt@CNT exposed to analogous treatments. To check whether the as-prepared Pt@CNT/SiO2 is catalytically active, we chose the hydrogenation of nitrobenzene (NB) as a model reaction to test its performance. In addition, the activity of the Pt@CNT/SiO2 and Pt@CNT was compared after it was subjected to heating at 300 °C for 2 h in both cases. In order to convincingly justify the significance of the designed catalytic systems for high-temperature processes, the hydrogenolysis of glycerol was carried out at 210 °C over porous-silica-coated CuRu@CNT obtained by applying an analogous protocol. The stability of the catalyst was studied after four reuses.

2. EXPERIMENTAL SECTION 2.1. Materials. All chemicals were of analytical grade and were purchased from Sinopharm Chemical Reagent Beijing Co., Ltd., and used as supplied. High-purity (95%) multiwalled carbon nanotubes (MWNTs) were purchased from Shenzhen Nanotech Port Co., Ltd., and used without further treatment. MWNTs were prepared by the catalytic decomposition of CH4 with La2NiO4 as a catalyst precursor. To remove the catalyst, CNTs were purified via being dispersed in 1 M HNO3 solution for 6 h, followed by filtering and washing with distilled water several times. Transmission electron microscopy (TEM) characterization reveals that the outer diameters and lengths of MWNTs were 4060 nm and 112 μm, respectively. 2.2. Coating of Pt@CNT with SiO2. The Pt@CNT sample was originally prepared by following the route described in our previous work.14 In a typical procedure to synthesize Pt@CNT/SiO2, 1 mg of Pt@CNT was added into 1 mL of CTAB aqueous solution at 10 mg 3 mL1 and then subjected to 3 h of sonication at ambient temperature in a sonic bath with a nominal power output at 100 W (KQ 100, Kunshan Instruments). Afterward, 8 mL of distilled H2O and 1 mL of NaOH aqueous solution (0.4 mg 3 mL1) were subsequently dropped into the dispersion slowly under bath sonication, followed by heat treatment at 60 °C for 20 min. In the following step, some certain amount of TEOS/ethanol (v/v = 1/4) solution was injected into the dispersion under magnetic stirring. After this, the dispersion was heated at 60 °C for 24 h to allow complete hydrolysis of TEOS and subsequent condensation of silicate. The yielded mixture was ultracentrifuged, and the collected precipitate was first washed repeatedly with absolute ethanol and distilled water and then vacuum-dried at 60 °C for 6 h.

ARTICLE

2.3. Synthesis of CuRu@CNT and Coating of CuRu@CNT with SiO2. The CuRu@CNT sample was prepared by following a

similar route described in our previous work.14 Specifically, 1 mg of MWNTs was initially dispersed in 20 mL of ethanol by 2 min tip sonication (Vibra Cell CVX, 500 W, 20 kHz, 20% amplitude). Subsequently, 1 mL of the precursor [Cu(NO3)2 3 3H2O þ RuCl3 3 3H2O] dissolved in ethanol at a certain concentration was dropped into the above suspension, followed by the dropwise addition of 1 mL of a NaBH4 ethanol solution (its concentration was 4 times higher than that of the precursor) under tip sonication. The obtained mixture was unltracentrifuged, and the collected precipitate was first washed repeatedly with absolute ethanol and distilled water and then vacuum-dried at 60 °C for 6 h. The yielded CuRu@CNT was ready for silica coating according to the procedure shown in section 2.2. 2.4. Characterization. TEM observations were performed on a transmission electron microscope (JEOL JEM-2010) operated at 200 kV equipped with an energy-dispersive X-ray spectrometer (EDS). X-ray diffraction (XRD) was carried out on a D/MAX-RC diffractometer operated at 30 kV and 100 mA with Cu KR radiation. Nitrogen adsorptiondesorption measurements were made on a Micromeritics ASAP 2020 instrument to determine the Brunauer EmmettTeller (BET) surface area and BarrettJoynerHalenda (BJH) pore size distribution. Note that the samples were first degassed in vacuum at 350 °C for 10 h prior to N2 adsorption at 195 °C. 2.5. Catalytic Activity Test. 2.5.1. Nitrobenzene Hydrogenation over Pt@CNT/SiO2. The same catalyst testing procedure was followed for each run. Typically, the catalyst (1.28 μmol of Pt) and NB (10 mmol) were loaded into a high-pressure stainless steel reactor; the reactor was then sealed and flushed with H2 three times to remove the air inside. Subsequently, the reactor was moved to a water bath set at 60 °C and H2 was introduced up to 2 MPa. The hydrogenation reaction was carried out with magnetic stirring (625 rpm). The H2 pressure was kept constant by replenishing H2 as the reaction proceeded. The reaction sample, after removal of the catalyst and the water by ultracentrifugation, was analyzed by gas chromatography (GC) (Agilent 4890D) with a capillary column and a flame ionization detector (FID). The turnover frequency (TOF) of NB conversion was defined as moles of NB converted per moles of Pt per hour. 2.5.2. Hydrogenolysis of Glycerol over CuRu@CNT/SiO2. The as-made CuRu@CNT/SiO2 catalysts were first treated at 300 °C for 4 h in H2 prior to each run of reactions to allow the complete reduction of copper and ruthenium precursors. Typically, the catalyst (3.56 μmol of CuRu) and glycerol (0.2 g) were loaded into a 7 mL autoclave reactor; the reactor was then sealed and flushed with 1 MPa H2 three times before the reaction. Subsequently, the reactor was moved to an air bath set at 210 °C and H2 was introduced up to 7.5 MPa. The reaction was carried out with magnetic stirring. After some period, the autoclave was cooled and the gases were collected. The liquid products were analyzed by GC (Agilent 6820) equipped with a PEG-20 M capillary column and a FID. The gas product was measured by GC (Agilent 4890) equipped with a packed carbon molecular sieve column and a thermal conductivity detector (TCD). For each run of recycling experiments, the catalyst was separated by centrifugation and washed with ethanol three times and then vacuum-dried at 60 °C for 18 h.

3. RESULTS AND DISCUSSION Representative transmission electron microscopy (TEM) images for Pt@CNT and the as-synthesized Pt@CNT/SiO2 are shown in Figure 1. As can be clearly seen from Figure 1b, the external surfaces of Pt@CNT were fully coated with a smooth and quasi-uniform SiO2 layer, which was confirmed by EDS analysis (inset of Figure 1b). Additionally, there were hardly any free SiO2 particles in the TEM views. Closer examination by 6245

dx.doi.org/10.1021/la105087f |Langmuir 2011, 27, 6244–6251

Langmuir

ARTICLE

Figure 1. TEM observations for (a) 5 wt % Pt@CNT and (b, c) 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 11.2 (the loading of SiO2 is around 76.3 wt %). The inset in panel b is the EDS profile taken for the sample shown in the image. The inset in panel c displays a high-resolution TEM image for the catalyst.

Figure 2. Schematic illustration of mesoporous silica coating onto a metal-decorated CNT, wherein a longitudinal section of the individual CTAB-wrapped Pt@CNT is given for more clarity.

high-resolution TEM (Figure 1c) revealed the mesoporous structure of the silica layer. Most of the pores were perpendicular to the surface of CNTs, allowing for reactant diffusion to the Pt NPs. We noted that the mesoporous silica preferably tended to form on the circumference of nanotubes rather than NP surfaces, probably owing to the stronger affinity of CTAB molecules for CNTs via van der Waals forces and hence the evolution of CTABsilicate organization and orientation along nanotubes, as illustrated in Figure 2. The pore size of the shell was approximately 3 nm by TEM estimation, which was reasonably bigger than the Pt NPs (≈ 2.1 nm) in 5 wt % Pt@CNT/SiO2. Careful inspections for different parts of the sample by TEM showed that fine Pt NPs remained exclusively on nanotubes in Pt@CNT/ SiO2 with high dispersion quality resembling that of Pt in the nonencapsulated Pt@CNT (Figure 1a). More importantly, in most cases each Pt NP was ideally encapsulated within one pore channel of the silica shell, implying that the coverage of Pt active sides or edges by the silica wall can be minimized. Such a configuration seems intriguing because not only can the catalytic NPs be prevented from aggregating by encapsulation within silica layers but reactant species were readily accessible to the catalyst through the thin and hollow porous structure. The silica coating thickness can be well controlled by monitoring the precursor mass ratio of TEOS-to-Pt@CNT. The histogram of the coating thickness distribution was extracted by directly measuring 30 individuals from the TEM view, whereby the mean thickness could be estimated. Figure 3 shows typical silica thickness distribution histograms for the catalysts prepared with varying mass ratios of TEOS-to-Pt@CNT. Moreover, the average thickness of SiO2 was plotted as a function of TEOS-toPt@CNT mass ratio and displayed in Figure 4a. It could be witnessed that the coating thickness scaled monotonically in an approximately linear manner with the TEOS-to-Pt@CNT mass

Figure 3. Histograms of the SiO2 thickness distribution for the catalysts with varying mTEOS-to-Pt@CNT: (a) 5.5, (b) 7.4, (c) 13.0, (d) 29.8. The inset in each graph is its corresponding TEM image for the catalyst.

ratio at mTEOS-to-Pt@CNT e 11.2, beyond which the thickness fell below the value derived from the linear dependence. This occurrence may be due to the fact that almost all SiO2 coated onto the surfaces of Pt@CNT at low TEOS-to-Pt@CNT mass ratios, while some mesoporous silica formed free from the Pt@CNT at high concentrations of TEOS addition, as confirmed by TEM imaging (Figure 4b). Specifically, surfactants bonded with nanotubes through strong van der Waals type interactions, which induced a higher local concentration of surfactants on CNT surfaces than in bulk solution.23 This did facilitate the heterogeneous nucleation of silicate preferentially on the nanotube surfaces, where silicate polyanions evolved from the hydrolysis and condensation of TEOS interacted with positively charged CTAB moieties via Coulomb forces. As such, the formation of free unbound silica particles originating from the homogeneous nucleation process was suppressed at low TEOS concentrations. Nevertheless, the homogeneous nucleation of silicate became more favorable with increasing levels of TEOS addition, as was reflected by the negative deviation of coating thickness from the linear dependence critically at mTEOS-to-(Pt/CNT) > 11.2 (Figure 4a). Despite this, we found that removal of CTAB 6246

dx.doi.org/10.1021/la105087f |Langmuir 2011, 27, 6244–6251

Langmuir

ARTICLE

Figure 4. (a) Thickness of SiO2 coating as a function of TEOS-to-5 wt % Pt@CNT mass ratio. The dashed line is drawn for a guide to the eye. (b, c) TEM images for 5 wt % Pt@CNT/SiO2 prepared (b) without and (c) with removal of bulk CTAB before addition of NaOH at mTEOS-to-Pt@CNT = 27.9. The inset in panel c displays a magnified TEM image for an individual sample. (d) XRD patterns of 5 wt % Pt@CNT; 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 5.5 (the loading of SiO2 is around 61.5 wt %); and Pt@CNT with the same Pt loading as 5 wt % Pt@CNT/SiO2 after exposure to 200 °C for 1 h in N2 .

from the bulk solution prior to the addition of NaOH enabled us to restrain the homogeneous nucleation of silicate, whereby more silicate nucleated on the nanotube surfaces and denser silica coating was formed, as shown in Figure 4c. The X-ray diffraction (XRD) pattern of 5 wt % Pt@CNT/SiO2 was very similar to that of 5 wt % Pt@CNT, where the strongest diffraction peak at 26.5° and less-intense peaks at 43.16°, 44.6°, 54.2°, and 77.7° were attributable to the (002), (100), (101), (004), and (110) reflections typical of MWNTs, respectively (Figure 4d).12 It is worth noting that both profiles showed the absence of either Pt or platinum oxide reflections. For comparison, XRD measurements were also conducted on a reference sample with an equivalent Pt loading to 5 wt % Pt@CNT/SiO2 that was prepared by blending 5 wt % Pt@CNT with pure CNTs and being exposed to heating at 200 °C for 1 h in N2. Surprisingly, an apparent diffraction peak at 39.75° was identified, which agrees well with the (111) plane of a face-centered cubic (fcc) structure for Pt (JCPDS file 04-0802) (top curve in Figure 4d). Application of Scherrer’s equation to the Pt (111) reflection gave a rough estimate of 2.8 nm for the average Pt particle diameter (dXRD) in the sample. Obviously, the absence of Pt diffraction peaks was unlikely due to the low metal loading in 5 wt % Pt@CNT/SiO2. Instead, in view of the essentially crystalline nature of Pt in the catalyst, this occurrence plausibly resulted from maintenance of the high dispersion quality of fine Pt NPs on CNTs without particle aggregation after formation of the silica coating, in line with prior TEM observations. Occasionally, some SiO2 double helices wrapped around Pt@CNT and tilted by some angles (dependent on nanotube diameters) relative to the tube axis were observed by TEM. Given both theoretical envisioning and experimental imaging regarding

the supermolecular self-assembly of surfactants at solidaqueous interfaces and on CNTs,24,25 it is highly probable that the helices were evolved from the half-cylinder CTABsilicate organization and orientation along metal-decorated CNTs. Note that only a small fraction of SiO2 helix coating occurred in the resulting catalysts, which may strongly suggest that most CTAB molecules were oriented perpendicularly to the metal-decorated nanotube surfaces, forming cylindrical micelles during the synthesis process. The porosity of the silica structure was further confirmed by nitrogen sorption analysis. Shown in Figure 5A are typical nitrogen adsorption/desorption isotherms of pure CNTs and 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 5.5. The CNTs showed type II isotherms in the BDDT classification, where the small adsorption in the relative pressure (P/P0) range e0.01 arose from some microporosity and/or formation of the first N2 adsorption layer on the surface of nanotubes; the slowly increasing adsorption in the mid-range of P/P0 was due to the multilayer adsorption; and the sharp adsorptiondesorption hysteresis loop at P/P0 > 0.8 correlated with the mesopores via a capillary condensation mechanism.26 By contrast, in addition to adsorption from the CNT contribution, Pt@CNT/SiO2 isotherms exhibited a more prominent increasing adsorption in the P/P0 range 0.20.3, which unambiguously indicated the mesoporous character of the silica layer. The average BJH (Barrett JoynerHalenda) pore diameter (Figure 5B) calculated from the desorption branch of the isotherm was determined to be about 3.0 nm, consistent with TEM observations. The BET (BrunauerEmmettTeller) surface area and single-point total pore volume were 87.5 m2 3 g1 and 0.22 cm3 3 g1 both of which were significantly larger than those of pristine CNTs centered at 6247

dx.doi.org/10.1021/la105087f |Langmuir 2011, 27, 6244–6251

Langmuir

ARTICLE

Figure 5. (A) N2 adsorptiondesorption isotherms of pure CNTs and 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 5.5. (B) Pore size distribution of 5 wt % Pt@CNT/SiO2 calculated from the desorption branch of the isotherms.

36.8 m2 3 g1 and 0.16 cm3 3 g1, respectively. In all likelihood, a similar shape of isotherms was obtained for samples prepared with higher TEOS-to-Pt@CNT mass ratios despite the improvement of BET surface areas due to the increasing loading of mesoporous SiO2 (Figure S1). As expected, the BET surface area increased from 265.1 to 325.9 m2 3 g1 as the mass ratio of TEOSto-Pt@CNT was raised from 9.3 to 13.0. One could argue that the migration of NPs and thus their aggregation may be induced by bath sonication over a long time scale such as 3 h. However, this is unlikely here as evidenced by the finding that no NPs departed from the nanotube surfaces, coupled with the fact that variation of NP sizes during the coating process was negligible. For example, the mean size of Pt in 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 5.5 was 2.1 ( 0.04 nm, virtually identical to that of Pt (2.0 ( 0.05 nm) within error for 5 wt % Pt@CNT prior to the silica coating. We believe that several relevant aspects are responsible for this phenomenon despite the lack of direct evidence by in situ observation of the NP microstructures throughout the sonication process. On one hand, the one-dimensional (1D) structure of nanotubes and the strong affinity of Pt for CNTs made NPs readily anchored on the surfaces of nanotubes, irrespective of the fact that smoothing or deformation of particles might be induced by intense shock waves from the collapse of bubbles during cavitation.27 Given the approximations as shown in the literature,28 the critical (i.e., minimum) velocity (vc) required to melt the Pt particle was estimated to be ≈808.4 m/s by pffiffiffi ð1Þ vc ¼ 2½CðTm  Tb Þ  L1=2 where C is the specific heat [130 J/(kg 3 K) for Pt], Tm is the melting temperature (2045 K for Pt), Tb is the bath temperature (300 K), and L is the heat of fusion (1.00  105 J/kg for Pt). If the shock wave pressure is taken to be 1 MPa, particle size R ≈ 2 nm, average viscosity η = 7.225  104 Pa 3 s, mean interparticle distance λ ≈ 5R (estimated from TEM observations), and Δt ≈ λ/v1, then the velocity of interparticle collisions (v) driven by the shock waves could be roughly estimated by adapting the equation:28 v  ðPR=6ηÞ½1  expðð  9ηΔtÞ=2FR 2 Þ

ð2Þ

Here Δt is the duration of the shock wave acceleration or collision time, v1 is the speed of sound in the liquid (for water, 1500 m/s), and F is the density of the particle (for Pt, 21.4  103 kg/m3). In this scenario, v was calculated to be approximately 0.1 m/s, far below vc, which suggests that particle collisions resulting from the ultrasonic irradiation did not lead to the occurrence of local particle melting and consequent agglomeration.

Alternatively, CTAB molecules assembled around CNTs enabled stabilization of NPs and hence inhibited their mobility along nanotubes as well as their further growth in size. To address whether the encagement of metal NPs deposited on CNTs within a porous SiO2 shell can prevent particles from sintering at high temperatures, we conducted TEM studies for Pt@CNT/ SiO2 after it was subjected to thermal treatments at varying temperatures for 2 h in N2. Shown in Figure 6a,ce are TEM images for the catalyst treated at 200, 300, 400, and 500 °C, respectively. Insets in each image are its corresponding metal NP size distribution histograms, extracted by directly measuring 100 NPs on CNTs from the TEM view. We noticed that the integrity of the Pt@CNT/SiO2 was preserved where Pt NPs were well distributed on the nanotube surfaces even upon heating the sample as high as 500 °C. The expression of Pt surfaces may be retained after the high-temperature treatments, as has been shown in a previous study where the Pt core encaged within a silica shell continued to be faceted after calcination at 550 °C.17 Shown in Figure 7 is the plot of NP sizes versus thermal treatment T for 5 wt % Pt@CNT/SiO2. It is noteworthy that the NP size (≈ 2.1) showed lack of change at T e 300 °C. In the cases of heat treatments at 400 and 500 °C, only a small variation occurred in the average sizes of Pt, corresponding to 2.3 ( 0.04 and 2.6 ( 0.05 nm, respectively. This is reflective of the fact that the unique silica shell in Pt@CNT/SiO2 serves as an effective spacer to inhibit coalescence of NPs at high temperatures. The slight increase in particle size may stem from the coalescence of very small Pt NPs encaged in one silica shell at high temperatures. In sharp contrast, a significant shift of particle sizes toward higher values and broadening of the size distribution occurred upon heating the nonencapsulated catalyst, 5 wt % Pt@CNT, at 200 °C for 2 h in N2 (Figures 6b and 7). In particular, more than a 2-fold increment in particle sizes was indeed observed for the catalyst treated at 500 °C (Figure 6f and Figure 7). These undesirable behaviors were presumably due to the migration and subsequent coalescence of NPs in close proximity to CNTs without silica encapsulation upon exposure to thermal treatments. The shape of Pt became rounder after heating, probably due to the melting of adjacent particles on nanotubes. Although isolation of metal catalyst particles by hollow mesoporous silica coating does indeed allow stabilization of the catalyst against sintering, another two relevant questions need to be answered: one is whether substantial mass transfer limitations occurred, and the other is whether activity of the catalyst after high-temperature treatments was retained. To this end, liquidphase hydrogenation of nitrobenzene was investigated over the as-synthesized Pt@CNT/SiO2 under solvent-free conditions since Pt behaves as a kind of robust catalyst for this 6248

dx.doi.org/10.1021/la105087f |Langmuir 2011, 27, 6244–6251

Langmuir

ARTICLE

Figure 6. TEM images for 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 5.5 treated at (a) 200, (c) 300, (d) 400, and (e) 500 °C for 2 h in N2 and for 5 wt % Pt@CNT treated at (b) 200 and (f) 500 °C for 2 h in N2. Insets in each image are the Pt NP size distribution histograms.

Figure 7. Plots of Pt size versus treatment T for 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT at 5.5 and 5 wt % Pt@CNT catalysts. The dotted line is drawn for a guide to the eye.

standard reaction. We found that 5 wt % Pt@CNT/SiO2 with mTEOS-to-Pt@CNT = 5.5 gave a high turnover frequency (TOF 50 200 h1) and aniline (AN) selectivity (98.8%), comparable to those of 5 wt % Pt@CNT (TOF 66 900 h1, AN selectivity 100%).14 This is an indication that the mesoporous silica shell did not significantly affect the diffusion and transport of reactant and product molecules. We suppose that there are two interrelated explanations for the observed result. On one hand, the thickness of the SiO2 shell is rather thin (≈10 nm), enabling the fast access of reactant species to the embedded catalytic NPs. The

promising “rattle-type” structure where the size of the metal NPs is sufficiently less than that of the porous shell, among other things, facilitates the release of Pt active sites free from the silica wall covering. More importantly, 5 wt % Pt@CNT/SiO2 showed little loss of catalytic activity by