Home

Search

Collections

Journals

About

Contact us

My IOPscience

Effects of surface activation on the structural and catalytic properties of ruthenium nanoparticles supported on mesoporous silica

This content has been downloaded from IOPscience. Please scroll down to see the full text. 2014 Nanotechnology 25 045701 (http://iopscience.iop.org/0957-4484/25/4/045701) View the table of contents for this issue, or go to the journal homepage for more

Download details: IP Address: 193.255.248.150 This content was downloaded on 04/02/2015 at 12:50

Please note that terms and conditions apply.

Nanotechnology Nanotechnology 25 (2014) 045701 (10pp)

doi:10.1088/0957-4484/25/4/045701

Effects of surface activation on the structural and catalytic properties of ruthenium nanoparticles supported on mesoporous silica Xianfeng Ma1 , Rui Lin1 , Christopher Beuerle1 , James E Jackson2 , Sherine O Obare3 and Robert Y Ofoli1 1

Department of Chemical Engineering and Materials Science, Michigan State University, East Lansing, MI 48824, USA 2 Department of Chemistry, Michigan State University, East Lansing, MI 48824, USA 3 Department of Chemistry, Western Michigan University, Kalamazoo, MI 49008, USA E-mail: [email protected] Received 24 August 2013, revised 10 November 2013 Accepted for publication 25 November 2013 Published 6 January 2014 Abstract

Using colloid-based methods to prepare supported catalytic metallic nanoparticles (NPs) often faces the challenge of removing the stabilizer used during synthesis and activating the catalyst without modifying the particles or the support. We explored three surface activation protocols (thermal oxidation at 150 ◦ C, thermal reduction at 350 ◦ C, and argon-protected calcination at 650 ◦ C) to activate ruthenium NPs supported on mesoporous silica (MSU-F), and assessed their effects on the structural and catalytic properties of the catalysts, and their activity by the aqueous phase hydrogenation of pyruvic acid. The NPs were synthesized by polyol reduction using poly-N-vinyl-2-pyrrolidone (PVP) as a stabilizer, and supported on MSU-F by sonication-assisted deposition. The NPs maintained their original morphology on the support during activation. Ar-protected calcination was the most efficient of the three for completely removing PVP from particle surfaces, and provided the highest degree of particle crystallinity and a metal dispersion comparable to commercial Ru/SiO2 . Its catalytic performance was significantly higher than the other two protocols, although all three thermally activated catalysts achieved higher activity than the commercial catalyst at the same Ru loading. Post-reaction analysis also showed that the supported catalyst activated at 650 ◦ C retained its morphology during the reaction, which is an important requirement for recyclability. Keywords: colloid-based synthesis, thermal treatment, aqueous phase hydrogenation, polyol reduction, sonication-assisted deposition, pyruvic acid S Online supplementary data available from stacks.iop.org/Nano/25/045701/mmedia

1. Introduction

corner steps has been used to tune the active centers for catalytic, electrocatalytic, and photocatalytic reactions [7–10]. However, conventional methods for preparing supported catalysts often lead to ill-defined morphologies and an uneven distribution of active sites. By contrast, colloid-based schemes, where nanoparticle synthesis and deposition on the support are separated into two sequential events, enable

Supported metallic nanoparticles (NPs) with well-defined size and shape have great potential for heterogeneous catalysis that ensures green chemistry, and for sustainable energy production from renewable materials [1–6]. Control over particle morphologies, exposed crystalline surfaces, and 0957-4484/14/045701+10$33.00

1

c 2014 IOP Publishing Ltd Printed in the UK

Nanotechnology 25 (2014) 045701

X Ma et al

2. Experimental section

independent control of the morphology and composition of active sites and their deposition on supports, thus allowing more rational assessment of mechanistic aspects at the molecular level [11–14]. Still, organic stabilizers used for nanoparticle synthesis are often regarded as having a negative effect on catalyst performance because of their potential to partially block the active sites on the particles [2, 4, 5, 15–17]. Thus, the preparation of supported catalysts by colloid-based methods presents the additional challenges of removing the stabilizing ligands used for synthesis and activating the catalysts without significantly affecting their size and shape, or the support on which they are deposited. Various methods have been explored for removing the stabilizing ligands from the surface of noble metallic nanoparticles (Pt, [15, 18–26]. Pd, [27] Ir, [28] and Au [29–31]) prior to their use for catalytic applications, including plasma treatments, UV–ozone cleaning, and thermal treatments. Ruthenium (Ru) is an important catalyst for many reactions critical to biomass conversion, including hydrogenation, hydrolysis, oxidation, and carbon–carbon coupling [32–36]. Several protocols based on colloid chemistry have been used so far to produce monodisperse Ru nanoparticles, including polyol or hydride reduction of Ru salts such as RuCl3 .nH2 O [37–46] and Ru(NO)(NO3 )3 [33], hydrogen decomposition of Ru(cod)(cot) [47–50] and thermal decomposition of Ru3 (CO)12 [51]. The stabilizers used for nanoparticle synthesis include polymers, amines, acetates, thiols and organosilanes. Typical supports for deposition of the Ru nanoparticles include metal oxides, silica, carbonaceous materials, and membranes [52–54]. The goal of this study was to use a colloid-based route to prepare solid catalysts consisting of monodisperse Ru nanoparticles supported on mesoporous silica (MSU-F), assess the effectiveness of three surface activation protocols for removing the stabilizer from the nanoparticles, and evaluate the effects of each protocol on the structural and catalytic properties of the supported nanocatalysts. The colloidal NPs were synthesized by polyol reduction using poly-N-vinyl-2-pyrrolidone (PVP) as a stabilizer, and supported on MSU-F by sonication-assisted deposition. We then explored three surface activation protocols (thermal oxidation at 150 ◦ C, thermal reduction at 350 ◦ C, and argon-protected calcination at 650 ◦ C) for activating the supported NPs. The structural and surface properties of the catalysts were assessed via several physical and chemical characterizations, including high resolution transmission electron microscopy (HRTEM), Fourier transform infrared spectroscopy (FT-IR), x-ray photoelectron spectroscopy (XPS), thermogravimetric analysis (TGA), physisorption and chemisorption, inductively coupled plasma atomic emission spectroscopy (ICP-AES), and x-ray diffraction (XRD). The catalytic properties of the activated Ru nanoparticles on MSU-F were evaluated by the aqueous phase hydrogenation of pyruvic acid as a model reaction, with the reactivity compared to that of a commercial Ru catalyst supported on silica.

2.1. Materials

Ru (III) acetylacetonate [Ru(acac)3 ] (99%) was purchased from Strem Chemicals (Newburyport, MA), and 1,4butanediol (99%) from Alfa Aesar (Ward Hill, MA). PVP (MW of 55 K), methanol (reagent grade, 99.9%), pyruvic acid (ACS reagent, ≥99.5%), MSU-F, and acetone (ACS reagent, ≥99.5%) were all purchased from Sigma-Aldrich (St Louis, MO). Argon (99.99%) and hydrogen (99.999%) were purchased from Airgas (Lansing, MI). A commercial catalyst (Ru/SiO2 ) was purchased from Kaida Chemical Engineering Co., (Shanxi, China), and was used as a control for comparison. All chemicals were used without further modification. 2.2. Synthesis of colloidal Ru nanoparticles and deposition on mesoporous silica

The supported Ru nanoparticles were prepared in two steps. The nanoparticles were first synthesized by a polyol reduction method according to previous reports using Poly-N-vinyl-2-pyrrolidone (PVP) as the stabilizing ligand (please see details in the supporting information available at stacks.iop.org/Nano/25/045701/mmedia) [37, 42, 55]. The colloidal nanoparticles were then purified by a precipitation–centrifugation method. Typically, acetone (seven times the volume of the colloid) poured into the solution induced a cloudy black suspension. This suspension was separated by centrifugation at 5000 RPM for 6 min. The precipitated Ru nanoparticles were collected, washed once with acetone, and re-dispersed in 20 ml of methanol. Loading of purified Ru nanoparticles onto mesoporous silica (MSU-F) was performed by sonication-assisted deposition. In a typical procedure, 1.6 g of MSU-F support was added to 20 ml of the colloidal Ru in methanol suspension (0.8 mg ml−1 ) and the slurry was sonicated (VWR Ultrasonic, 75 T/120 W/45 kHz) for 3 h at room temperature. The suspension was then centrifuged at 5000 RPM for 10 min, and the solid precipitates washed twice with methanol. The excess methanol was removed by evaporation for 2 h under flowing nitrogen at room temperature, and the residual catalysts were stored in vacuum desiccators. 2.3. Surface activation of Ru catalysts supported on MSU-F

We explored three procedures for removing the capping PVP agent from particle surfaces. In the first, the sample was heated in static air to 150 ◦ C, maintained at this temperature for 12 h, and then cooled to room temperature. We refer to this procedure as the gentle oxidation method, and the resulting catalysts are labeled RuF-150. The second was a thermal reduction protocol, which involved reduction of the sample at 350 ◦ C for 2 h under flowing hydrogen to obtain the catalyst labeled RuF-350. In the third procedure, the sample was heated in flowing argon to 650 ◦ C at a rate of 10 ◦ C min−1 , held at this temperature for 2 h, and then cooled to room temperature to obtain the catalyst labeled RuF-650. We refer to this procedure as the argon-protected calcination protocol. 2

Nanotechnology 25 (2014) 045701

X Ma et al

Instrument Corporation, Norcross, GA). For analysis of RuF-150, the sample was flushed with H2 at room temperature for 30 min, heated at a rate of 10 ◦ C min−1 to 150 ◦ C, and maintained at this temperature for 2 h to reduce the catalyst. This was followed by flushing with helium for 90 min and evacuation for 30 min. The sample was then cooled to 35 ◦ C for analysis. The procedures used for analysis of RuF-350 and RuF-650 were similar, except that the reduction temperature was 350 ◦ C for each.

2.4. Catalyst characterization

We used transmission electron microscopy (TEM) to assess the morphologies of the Ru colloidal nanoparticles. To prepare samples for TEM characterization, the nanoparticles were diluted and deposited on 3 nm carbon-coated copper grids. The grids were first dried overnight in vacuum desiccators at room temperature, and then dried again at 80 ◦ C for about 12 h. The catalysts were suspended in methanol and then deposited onto the grids. TEM characterization was performed on a JEOL 2200FS electron microscope (Tokyo, Japan) equipped with an energy dispersive x-ray (EDX) spectrometer (Oxford Instrument, UK). All characterizations were done at an acceleration voltage of 200 kV. The average diameter was calculated by measuring the sizes of 100 randomly selected Ru nanoparticles from the TEM images. The infrared spectra of the samples were obtained on a Mattson Galaxy FT-IR spectrometer (Mattson Instruments, Madison, WI) using KBr pellets. Thermogravimetric analysis (TGA) was done on a TGA/DSC thermogravimetric analyzer (Mettler-Toledo Inc., Columbus, OH) to determine if any transformation of the catalysts occurred during the thermal treatments. In a typical characterization, about 70 mg of the sample was placed in an alumina sample holder, and heated from room temperature to 700 ◦ C at a rate of 10 ◦ C min−1 under flowing nitrogen (20 ml min−1 ). The Ru content in each sample was determined by ICP-AES (Vista Pro, Varian, Cary, NC). To measure the Ru concentration, 0.5 ml of the colloid was transferred into a screw-capped centrifuge tube and the solvent was evaporated. Then 2.0 ml of aqua regia (a mixture of 1 volume of HNO3 and 3 volumes of HCl) was added to the tube. The mixture was heated to 80 ◦ C and maintained at this temperature for 3 h, followed by centrifugation at 4000 RPM for 6 min to remove trace insoluble materials. The supernatant was diluted with deionized water and analyzed by ICP-AES. A similar procedure was used to determine the weight per cent of Ru in each of the supported catalysts. The atomic composition and oxidation states of the samples were determined by XPS, with all the measurements performed on a Perkin Elmer Phi 5600 ESCA (Waltham, MA) system with a magnesium Kα x-ray source. Samples were analyzed at pressures between 10−9 and 10−8 Torr with a passing energy of 29.35 eV and a take-off angle of 45◦ . The spot size was approximately 250 µm2 . All peaks were referenced to the signature C1s for adventitious carbon at 284.6 eV. The XRD patterns of the samples were identified using an x-ray powder diffraction system (D8 ADVANCE, Bruker AXS Inc., Madison, WI) with a Lynx-Eye detector and Cu Kα radiation. The samples were scanned over 2θ values of 30◦ to 50◦ with a step size of 0.02◦ and a dwell time of 1.0 s. The textural properties of the catalysts were investigated by nitrogen physisorption at 78 K using a Micromeritics ASAP2010 analyzer (Micromeritics Instrument Corporation, Norcross, GA). Prior to each measurement, each sample was degassed under vacuum at 220 ◦ C for about 24 h. The Ru dispersion was determined by H2 chemisorption performed on a Micromeritics ASAP2010C analyzer (Micromeritics

2.5. General procedures for hydrogenation of pyruvic acid to lactic acid

All reactions were conducted in a Parr multi-batch reactor system (Model 5000, Parr Instrument Co., Moline, Illinois) equipped with a coupled magnetic stirring unit. The system contains six 75 ml reactors, each equipped with an individual heating mantle and sampling port. In a typical experiment, the quantity of catalyst required to ensure reasonably consistent Ru loading (see table 2) was pre-reduced in the reactor by purging with nitrogen, heating to 150 ◦ C, charging with hydrogen to 3.4 MPa, and holding for 12 h. After cooling the system to room temperature and flushing with nitrogen, 35 ml of the substrate (20 mM pyruvic acid in water) was charged into the reactor from a sample cylinder under nitrogen pressure. When the reaction temperature stabilized at the target of 45 ◦ C, the reactor was pressurized with hydrogen gas at 0.5 MPa to initiate the reaction. To monitor the progress of the reaction, liquid samples (2 ml) were withdrawn periodically, filtered through a syringe microfilter (0.22 µm), and analyzed by HPLC (Waters Corporation, Milford, MA) equipped with a refractive index (RI) detector and a UV detector at 250 nm. The separation was carried out on a Bio-Rad Aminex HPX-87H column at 323 K. The mobile phase was 0.005 M H2 SO4 in HPLC-grade water at a flow rate of 0.60 ml min−1 . We developed multipoint calibration curves to obtain peak response factors, and used the data to calculate pyruvic acid conversions and product yields, and to do overall material balances. 3. Results and discussions 3.1. Morphology of colloidal Ru nanoparticles

The chemical reduction of salt precursors is a widely used method for preparing colloidal metal nanoparticles. The moderate reducing ability of 1,4-butanediol is a significant advantage in controlling the growth rate of the nanoparticles, compared to strong reducing agents. When combined with stabilization by PVP through steric entrapment and weak binding to particle surfaces via coordination chemistry, the morphology and distribution of nanoparticles can be controlled effectively [2, 56, 57]. The TEM images of PVP-stabilized Ru nanoparticles in figures 1(a) and (b) show that the particles are spherical in shape and well dispersed, with no tendency towards agglomeration. The EDS data in figure 1(c) confirm that the nanoparticles contain Ru, and that there are no other metals present. The histogram in 3

Nanotechnology 25 (2014) 045701

X Ma et al

of the mesoporous silica, thus providing better metal dispersion [21]. The activation of supported nanoparticles is one of the most critical aspects of catalyst preparation, because the capping agent must be removed completely without inducing significant changes in the morphology of the active nanoparticles and without causing any damage to the structure of the support [5]. As already stated, we focused on three procedures to remove the stabilizer from the sample, namely, gentle oxidation at 150 ◦ C, thermal reduction at 350 ◦ C and argon-protected calcination at 650 ◦ C. We should note that we ran the treatment protocols at temperatures equal to or higher than 150 ◦ C, because that was also the temperature we used for the pre-reduction of the catalysts prior to using them for reactions (see section 2). The TEM images of the MSU-F support and the morphology of Ru nanoparticles supported on MSU-F following the three thermal treatments are shown in figure 2. The MSU-F support by itself has a hexagonal cellular mesoporous framework with a cell window size of ∼15 nm and unit cell size of ∼22 nm (figure 2(a)). The Ru nanoparticles were well dispersed in the pores of MSU-F following gentle thermal oxidation, with no signs of agglomeration and no damage to the framework and structure of the support (figure 2(b)). As is evident in figures 2(c) and (d), the supported Ru nanoparticles are of similar morphology to that of the colloidal nanoparticles prior to deposition on the support. The TEM images in figures 2(e) and (f) also show that the Ru nanoparticles remain individually dispersed on the support, with no obvious morphology changes after thermal reduction, and that the ordered structure of the support was also preserved during the treatments. Excellent dispersion and similar morphology were also achieved upon treatment under an inert gas atmosphere at 650 ◦ C, as shown in figures 2(g) and (h). This uniformity in size, shape and size distribution of the catalysts enabled us to do a direct investigation of the effects of the surface activation strategies on the catalytic properties of the supported Ru nanoparticles, without having to also account for the influence of changes in particle morphologies on catalytic performance. FT-IR spectroscopy and TGA analysis were performed on the starting material, as well as after each step of catalyst preparation. The results are summarized in figures 3 and 4, respectively. The infrared spectrum of the pure support (MSU-F) has a strong absorption peak at 1100 cm−1 , corresponding to the characteristic Si–O stretch, and a weak band at 1636 cm−1 , assignable to the hydroxyl groups [59]. Pure PVP has C=O and C–N stretch bands at 1694 and 1674 cm−1 , asymmetric CH2 stretches at 2950 cm−1 for the pyrrolidone ring and at 2922 cm−1 for the polymer backbone, a CH bending band at 1371 cm−1 , and a CH2 scissor band at 1461 cm−1 . The spectrum also contains a series of bands in the 750–1300 cm−1 range, corresponding to the C–C bond in the side ring and backbone, in good agreement with a previous study [60]. RuF-150 has the absorption peaks characteristic of MSU-F and PVP, suggesting that only partial decomposition of the capping PVP agent occurred during the gentle oxidation treatment. The RuF-350 spectrum has the absorption peak characteristic of MSU-F, but the

Figure 1. TEM images and EDS spectra for PVP-stabilized Ru

nanoparticles. (a) TEM image of well-dispersed Ru nanoparticles at a magnification of 100K (scale bar 20 nm); (b) TEM image of Ru nanoparticles at a magnification of 600K, showing very stable nanoparticles (scale bar 2 nm); (c) EDS spectrum showing the elemental Ru content of the nanoparticles at a magnification of 100K; (d) histogram showing size distribution of Ru nanoparticles; the average particle size is 3.5 ± 0.5 nm.

figure 1(d) also shows that the particles have a fairly narrow size distribution, with an average particle size estimated at 3.5 ± 0.5 nm. We also determined from ICP-AES data that 89% of the Ru(acac)3 precursor was converted to colloidal Ru nanoparticles, which is an indication of a very efficient protocol, and an excellent use of the noble metal. 3.2. Structural properties of Ru nanoparticles supported on MSU-F

Ordered mesoporous materials have significant advantages over conventional supports used for the preparation of heterogeneous catalysts. For instance, their pores and cavities can be precisely regulated at the micro- or nanoscale as needed to deposit nanoparticles of different sizes and shapes [2, 58]. Compared to conventional impregnation methods, sonication-assisted deposition of nanoparticles on a high surface area promotes homogeneous incorporation of the metallic nanoparticles into the interior surfaces 4

Nanotechnology 25 (2014) 045701

X Ma et al

Figure 3. FT-IR spectra of MSU-F, PVP, and Ru nanoparticles

following treatment by three different protocols: (a) pure MSU-F support; (b) PVP; (c) RuF-150; (d) RuF-350; and (e) RuF-650. The RuF-150 spectrum shows the absorption peaks characteristic of both MSU-F and PVP, suggesting only partial decomposition of the capping PVP agent. The spectrum for RuF-350 shows the absorption peak characteristic of MSU-F, but the disappearance of peaks at 1371 and 1461 cm−1 is an indication that the PVP capping agent was completely decomposed during calcinations. However, the bands remaining at 1694, 1674 and 2950 cm−1 demonstrate that the decomposed PVP residuals are still attached to the surface of the Ru nanoparticles. The spectrum for RuF-650 shows only the characteristic peaks of pure MSU-F, providing evidence of complete decomposition and removal of the capping polymer agent from the supported catalyst.

PVP residues are still attached to the surface of the Ru nanoparticles. By contrast, the IR spectra of RuF-650 has only the characteristic peaks of pure MSU-F, indicating complete decomposition as well as removal of the capping polymer agent from the supported catalyst. The TGA curves in figure 4 show a weight loss of about 8.0% at 100 ◦ C for pure PVP and about 2.0% for each of the rest of the samples, all of which can be attributed to the evaporation of water and residual organic solvent. The TGA curve for MSU-F shows that it is thermally stable over the entire temperature range, except for a slight weight loss at high temperature due to dehydration of surface –OH groups. The thermal decomposition of pure PVP started at about 350 ◦ C and was complete by the time the temperature reached 480 ◦ C, with about 95% of total weight loss. The residues of decomposed PVP account for 5% of the total sample weight loss. The differential thermal gravimetric (DTG) curve of pure PVP suggests that it undergoes thermal decomposition in two stages. The onset decomposition temperature of RuF-150 is close to that of pure PVP, with the total weight loss reaching a maximum of about 13% at 480 ◦ C, most likely due to the decomposition of the capping agent. It is also worth noting that RuF-150 has only a single decomposition stage, as shown by its DTG curve; this might be the random scission of the polymer backbone. From the RuF-650 TGA curve, it can be deduced that more than 98% of the capping PVP was decomposed by heating at 650 ◦ C under argon. We observed less than 3% weight loss for RuF-350 in the TGA curve, which suggests that most of the capping PVP agent on the

Figure 2. TEM images of pure MSU-F and MSU-F-supported Ru

nanoparticles following treatment by three different protocols. (a) pure MSU-F support at a magnification of 100K (scale bar 20 nm), showing a hexagonal mesoporous structure; (b)–(d) RuF-150 at magnifications of 100K (scale bar 20 nm), 200K (scale bar 10 nm), and 300K (scale bar 5 nm), respectively, showing well-dispersed nanoparticles of the same size as the colloidal Ru particles; (e) and (f) RuF-350 at magnifications of 100K (scale bar 20 nm) and 200K (scale bar 10 nm), respectively, showing well-dispersed particles with no damage to the support following thermal treatment; (g) and (h) RuF-650 at magnifications of 100K (scale bar 20 nm) and 200K (scale bar 10 nm), respectively, also showing no damage to the mesoporous support.

disappearance of the bands for asymmetric CH2 stretch and scissor in the polymer backbone is an indication that PVP was completely decomposed during thermal reduction at 350 ◦ C. However, the remaining bands for the C=O stretch, C–N stretch, and CH bend suggest that the decomposed 5

Nanotechnology 25 (2014) 045701

X Ma et al

Figure 5. Powder XRD patterns of pure MSU-F and supported Ru

nanoparticles subjected to three different thermal treatments. (a) MSU-F; (b) RuF-150; (c) RuF-350; (d) RuF-650. The peak at the 2θ value of 43.5◦ is consistent with the (101) face of hexagonal close packed (HCP) Ru nanocrystals. The intensity of the peak is an indication of the crystallinity of each sample.

Figure 4. TGA and DTG diagrams of MSU-F, PVP, and supported

Ru nanoparticles after thermal treatments at 150, 350 and 650 ◦ C. (1) TGA spectra; (2) DTG profiles. (a) PVP; (b) RuF-150; (c) RuF-350; (d) RuF-650; (e) MSU-F. For each system, the temperature was raised from 25 to 700 ◦ C at a rate of 10 ◦ C min−1 . The TGA diagram shows the weight loss of each sample as the temperature was increased. The DTG diagram shows the differential thermal gravimetric change of each sample. MSU-F is thermally stable over the entire temperature range. The decomposition of pure PVP began at about 390 ◦ C and was complete by the time the system reached 480 ◦ C. The decomposition of RuF-150 is similar to that of pure PVP, but much less pronounced. The relatively negligible changes in the TGA and DTG curves for RuF-350 and RuF-650 are an indication of complete decomposition of PVP during the calcination process. All sample weight losses prior to 390 ◦ C can be attributed to the evaporation of water and residual organic solvent.

Ru nanoparticles was decomposed by calcination at 350 ◦ C under hydrogen. This temperature is less than the low end of the range of decomposition temperatures of pure PVP (350–480 ◦ C), which suggests that interactions among PVP, Ru nanoparticles and MSU-F may promote moderately easier decomposition of PVP. One previous study has also reported that, when used as a stabilizer for synthesis of platinum and rhodium nanoparticles, the PVP stabilizer decomposed at a lower temperature than pure PVP [61]. Powder XRD measurements (figure 5) for all three preparations showed a diffraction peak at 2θ = 43.5◦ , characteristic of the {101} facet of hexagonal close packed crystalline Ru metal [62, 63]. This marker peak is weaker in RuF-150 than in RuF-350 and RuF-650, again confirming incomplete removal of the capping polymer in catalysts treated by gentle thermal oxidation. The increase in marker peak intensity and sharpness reflect the fact that treatments at higher temperatures likely allow more annealing and therefore higher degrees of crystallinity in the RuF-350 and RuF-650 samples. As expected, no characteristic diffraction peaks were observed for the pure support between 2θ values of 30◦ and 50◦ .

Figure 6. XPS spectra of Ru 3p for supported Ru nanoparticles

subjected to three different treatments. (a) RuF-150; (b) RuF-350; (c) RuF-650. The three characteristic peaks from high bonding energy to low in each diagram correspond to Ru(VI), Ru (IV) and metallic Ru (0). The binding energies of the metallic Ru peak in RuF-150 and RuF-350 display a slight negative shift compared to that of standard metallic Ru (461.5 eV); a similar trend was observed for the binding energies of Ru (VI) in RuF-150 and RuF-350. The presence of Ru (IV) and Ru (VI) in all the samples also indicates that the surface atoms of the Ru nanoparticles can be easily oxidized when exposed to air, even under ambient conditions.

To further investigate the effect of thermal treatment on the atomic composition and oxidation states of the catalysts, we performed XPS measurements on all the three samples (figure 6). The XPS spectra of the commercial reference catalyst (Ru/SiO2 ) was also recorded (figure S1 available at stacks.iop.org/Nano/25/045701/mmedia). Figure 6 shows the XPS spectra of the catalysts in the Ru 3p region. The binding energy at about 461 eV was ascribed to Ru in the metallic form. The other two signals at higher binding energies (462.5 and 465 eV) are assignable to Ru (IV) and 6

Nanotechnology 25 (2014) 045701

X Ma et al

Table 1. Material properties of catalysts prepared for this study.

Sample ID

BET surface areaa (m2 g−1 )

Ru contentb (wt%)

Metal dispersionc (%)

N/Ru atomic ratiod

MSU-F Ru/SiO2 RuF-150 RuF-350 RuF-650

503.33 ± 1.20 139.64 ± 0.78 320.70 ± 7.38 579.23 ± 1.48 649.89 ± 1.59

— 1.32 ± 0.02 0.77 ± 0.01 0.89 ± 0.03 0.42 ± 0.01

— 17.55 ± 1.37 2.49 ± 0.88 0.55 ± 0.57 18.65 ± 0.43

— — 7.36 0.00 0.00

a

From N2 adsorption. From ICP-AES. c From H adsorption results. 2 d Based on intensities of Ru 3p and N 1s peaks in XPS analysis. b

Table 2. Catalytic activities of supported Ru nanocatalysts for pyruvic acid hydrogenation.

Sample ID

Catalyst loading (mg)

Ru loadinga (mg)

Conversionb (%)

Reaction ratec (mol PyA gRu−1 h−1 )

TOFd (×10−3 h−1 )

MSU-F Ru/SiO2 RuF-150 RuF-350 RuF-650

5.00 29.0 28.0 24.2 51.1

— 0.383 ± 0.006 0.216 ± 0.003 0.214 ± 0.007 0.214 ± 0.015

5.4 90.8 74.2 73.1 95.4

— 1.70 ± 0.01 2.12 ± 0.01 2.16 ± 0.04 4.56 ± 0.05

— 0.99 ± 0.08 — — 2.48 ± 0.08

a

Calculated from catalyst loading (g) and Ru content (wt%) from ICP-AES measurements on the corresponding catalyst. b Substrate conversion in 2 h. c Initial reaction rate calculated as the consumption of pyruvic acid in mmol per g of Ru per h. d TOF calculated from reaction rate and metal dispersion data from hydrogen adsorption experiments, in units of mmol PyA per mmol of available Ru active sites per h.

to Ru atomic concentrations (N/Ru) from XPS measurements. The surface area of pure MSU-F decreased from 503.3 to 320.0 m2 g−1 for RuF-150, which is the direct result of deposition of Ru nanoparticles inside the pores of the support of the sample. This is confirmed by TEM observations of RuF-150 showing that most of the Ru nanoparticles are located at the inner wall of the support (figure 2(d)). It is also possible that partially decomposed PVP on RuF-150 may have covered some of the pores on the support, leading to an apparently reduced surface area. By comparison, the surface area of RuF-350 increased to 579.0 m2 g−1 , which is likely the result of Ru nanoparticles migrating from inside the pores to the edges of the support, leading to more Ru nanoparticles on the surface (figure 2(e)). The surface area of RuF-650 increased to 649.0 m2 g−1 . This 12% increase in BET surface area for RuF-650 relative to RuF-350 suggests more complete removal of decomposed PVP residues from nanoparticle surfaces. The significant increase in surface area of RuF-650 can also be attributed to the restructuring of the silica framework in MSU-F via bond rearrangements [65]. The Ru content increased from 0.77 wt% in RuF-150 to 0.89 wt% in RuF-350, which is consistent with the removal of the capping agent in the samples. On the other hand, the Ru content in RuF-650 unexpectedly decreased to 0.42 wt%. While we do not fully understand why this occurred, a possible reason is the likely formation of low boiling point compounds from a reaction between surface Ru atoms and residues of decomposed PVP at the high temperature used. It has been reported that Ru can be coordinated with reactive

Ru (VI), respectively. The binding energy of metallic Ru in the reference catalyst (461.4 eV) was close to that of the standard metallic Ru (461.5 eV), while the binding energy of metallic Ru in RuF-150 (460.80 eV) is slightly lower than the standard value. This negative shift in binding energy is likely the result of the incomplete removal of the PVP stabilizer. Strong interactions between the surface atoms of the Ru nanoparticles and the N atoms of the polymer may also transfer electron density from the basic capping agent to the metallic particles. A similar effect was observed in RuF-350. In that case, however, we believe this is more likely the result of surface Ru atoms bonding to residues of decomposed PVP. On the other hand, the binding energy of metallic Ru in RuF-650 (461.18 eV) is very close to that of the pure Ru metal, providing another confirmation that essentially all of the stabilizing polymer and its residues were removed during the thermal treatment at 650 ◦ C under an inert gas environment [64]. A similar trend in binding energy shift for Ru (VI) is observed among the three catalysts. We should also note that the presence of Ru (IV) and Ru (VI) in all the samples and the reference catalyst is an indication of the ease by which Ru nanoparticles can be oxidized when exposed to air, even under ambient conditions, making it necessary to pre-reduce the catalysts prior to using them for hydrogenation reactions. The physical properties of the catalysts are summarized in table 1, including the surface areas determined by physisorption, Ru metal content from ICP-AES, metal dispersion from hydrogen adsorption, and the ratio of nitrogen 7

Nanotechnology 25 (2014) 045701

X Ma et al

organic fragments [63, 66]. It is possible that these complexes can be removed from the solid sample by Ar flowing under high temperature, resulting in the apparent decrease in Ru content. Still, this decrease in metallic content is a very surprising result that we are currently investigating. With partially decomposed PVP surrounding the Ru nanoparticles, RuF-150 has a metal dispersion of approximately 2.5%, based on data from the chemisorption of H2 . This value increased 7.5-fold to 18.6% for RuF-650, which can be attributed to complete decomposition and removal of PVP at 650 ◦ C. However, this value represents just half of the dispersion expected for fully accessible 3.5 nm particles. We envision the NPs as roughly 50% embedded in the MSU-F support, exposing only the other half to solution. On the other hand, the metal dispersion of RuF-350 was 0.55%, which is a significant decrease relative to the value for RuF-150. We believe this result may be an indication that residues of degraded PVP were still attached to the Ru surface atoms, thus blocking the active sites and preventing bonding with H2 . A nitrogen to ruthenium (N:Ru) ratio of 7.36 was calculated from the peak intensities in the XPS spectra for RuF-150. By contrast, the N:Ru values for RuF-350 and RuF-650 were negligible, which is an indication that the N-containing PVP side rings were eliminated during the thermal treatments.

Figure 7. Reaction profiles of PyA hydrogenation over Ru NPs

supported on MSU-F and the commercial catalyst Ru/SiO2 . • MSU-F;  Ru/SiO2 ; N RuF-150; O RuF-350;  RuF-650. The points in each plot are the concentration of the substrate at the corresponding time, and the solid line is the least squares-fitted exponential decay for the data set. The inset is the enlarged view of the reaction profiles in the first hour. Reaction conditions: T = 45 ◦ C; P(H2 ) = 5.0 bar; 35 ml of PyA in water (20 mM). All catalysts were pretreated at 150 ◦ C overnight under H2 pressure of 33.4 bar.

calculation details) than the commercial Ru/SiO2 catalyst (table 2). The reaction rate of RuF-650 is twice as high as RuF-150 and RuF-350. This can be attributed to better access to surface catalytic sites following the complete decomposition and removal of the PVP stabilizer, and to better dispersion of nanocatalysts on the support. The reaction rates of RuF-150 and RuF-350 are higher than for the commercial Ru/SiO2 catalyst, even though they have lower metal dispersions. The fact that metal dispersion does not necessarily predict catalyst activity of in the aqueous phase could be interpreted in terms of the behavior of catalysts in different media. It is possible that the presence of the PVP capping agent and/or its residues on the Ru nanoparticles in RuF-150 and RuF-350 inhibit the active metal from adsorbing significant amounts of gas phase molecules such as hydrogen. However, in aqueous media, the PVP capping agent (which is hydrophilic and could therefore be considered to be in a ‘good solvent’) may swell up and be more ‘brush-like’ around the Ru nanoparticles. This could enable the substrate to diffuse more easily to the catalytic sites, which is consistent with the behavior of other nanoparticles synthesized from platinum group members such as Pt and Rh [56]. An important assessment of catalytic performance is the turnover frequency (TOF). Based on the assumption that all the catalytic sites are accessible for the reactions using commercial Ru/SiO2 and RuF-650, we investigated the reactivity of the two catalysts by calculating the TOF for each catalyst. The TOF for RuF-650 was more than two times the value for Ru/SiO2 (table 2), which indicates that RuF-650 has a higher activity in terms of each active site. This is not a surprising result, given the earlier discussion on reaction rates. We could not do any comparative TOF assessments for RuF-350 and RuF-150 because the calculation generally

3.3. Assessment of the activity and stability of Ru nanoparticles supported on MSU-F for hydrogenation of pyruvic acid

The activity of the supported Ru catalysts was assessed using the aqueous phase hydrogenation of pyruvic acid to lactic acid, using pure MSU-F and commercial Ru/SiO2 as references. The Ru content of the activated and commercial catalysts was used to calculate the weight of sample required to ensure approximately identical quantities of Ru in each sample (within 2%), to eliminate catalyst loading as a significant factor in the results. Thus, the only important difference between samples is the type of thermal treatment. To prevent Ru oxides on particle surfaces from affecting the catalytic activity of Ru NPs [30], we pre-reduced each catalyst prior to the reaction by following the procedures in a previous report [67]. In all reactions, we observed 100% selectivity to lactic acid, with no other product detected. Less than 5% of the pyruvic acid was converted in 7 h with pure MSU-F (figure 7 and table 2), confirming that the reaction is catalytic. By contrast, complete conversion of the substrate was achieved within 3 h for both commercial Ru/SiO2 and RuF-650, while reactions using RuF-150 and RuF-350 were complete in 5 h. The results compare favorably to previous reports on the catalytic performance of Ru nanoparticles for pyruvic acid hydrogenation in ethanol–water at 100 ◦ C and 10 bar hydrogen pressure [68]. These comparable results at a lower temperature and hydrogen pressure give an indication of the potential for much higher catalytic efficiencies with the use of nanoparticles. All the Ru nanocatalysts prepared in this study demonstrated higher reaction rates (see supporting materials, available at stacks.iop.org/Nano/25/045701/mmedia for 8

Nanotechnology 25 (2014) 045701

X Ma et al

assumes that all active catalyst sites are accessible and, as discussed earlier, our data show that both samples have a significant number of active sites that are partially covered by PVP and/or its decomposed residues. The structural stability of Ru nanoparticles on MSU-F support is critical to catalyst function and recyclability, and the best way for this assessment would be the recovery of and recycling of catalysts after each reaction. However, it was difficult to recover all the catalyst after each experiment in the batch reactor. Since the quantity of catalysts used in each initial reaction was quite small, even a small loss during catalyst recovery after each reaction would make it difficult to compare catalyst performance in subsequent cycles of reaction. We expect to resolve this issue once we scale up catalyst production using the protocol presented here, which will enable catalyst recovery and recyclability to become a major focus of our future activities. For this study, we focused on using TEM characterization to investigate the structural robustness of the most efficient of the three surface-activated catalysts (RuF-650). We did this by recovering the catalyst by filtration after the reaction and depositing it on TEM grids. The resulting scans showed that the support (MSU-F) retained the flaky structure during the reaction, with several overlapping plates (figure 8(a)). The mesoporous structure remained the same as before the reaction, with no damage to the framework. The Ru nanoparticles remained evenly distributed on the support, in a similar manner to what we observed prior to the reaction. There were no unattached Ru NPs in the TEM images (figures 8(a) and (b)), suggesting that little or no detachment of nanoparticles from the support occurred during the reaction. The higher magnification images in figures 8(c) and (d) show that the supported Ru nanoparticles remained spherical, uniform and well dispersed on the surface. The fact that the Ru nanoparticles on MSU-F support retained their original uniform morphology, narrow size distribution and excellent dispersion during the reaction provides strong evidence of the structural stability of the nanocatalysts during the liquid phase reaction, and shows its potential for recyclability.

Figure 8. TEM images of Ru nanoparticles supported on MSU-F

(RuF-650) recovered after the hydrogenation of PyA.

in providing complete removal of the stabilizing PVP agent, a high degree of particle crystallinity, and the best metal dispersion among the three catalysts. All the Ru catalysts prepared for this study had higher reactivities for the aqueous phase hydrogenation of pyruvic acid than a commercial Ru/SiO2 catalyst. In particular, the reaction rates and TOF obtained with the RuF-650 catalyst are both more than two times those of commercial catalyst. The supported Ru nanoparticles in RuF-650 also retained their uniform morphology and metal dispersion during the reaction, demonstrating the potential for catalyst recycling. Acknowledgments

This work was supported by the 21st Century Jobs Fund of the Michigan Economic Development Corporation (MEDC). We are grateful to Dr Per Askeland from the Composite Materials and Structures Center (CMSC) at Michigan State University (MSU) for performing XPS characterizations. We also acknowledge the help of Professor Dennis Miller and Dr Lars Peereboom of the MSU Department of Chemical Engineering and Materials Science for their assistance in physisorption and chemisorption experiments and reactor set-up.

4. Conclusions

We have assessed the effects of three surface activation protocols on the structural and catalytic properties of monodisperse Ru nanoparticles supported on mesoporous silica (MSU-F). After each of the activation procedures, the MSU-F support retained its mesoporous structure with no damage to the framework. The supported Ru nanoparticles were also dispersed individually on the support without agglomeration. In addition, the particles maintained a uniform morphology similar to those of the unsupported colloidal Ru nanoparticles. Based on this study, gentle thermal oxidation at 150 ◦ C is not sufficient to completely decompose and remove the capping PVP agent from the nanoparticle surfaces. Thermal reduction at 350 ◦ C leads to effective decomposition of PVP, but leaves residues on the metal surface, likely blocking catalytic sites. Argon-protected calcination was the most efficient surface activation method of the three

References [1] Alonso D M, Bond J Q and Dumesic J A 2010 Green Chem. 12 1493–513 [2] Jia C J and Schuth F 2011 Phys. Chem. Chem. Phys. 13 2457–87 [3] Kleis J et al 2011 Catal. Lett. 141 1067–71 [4] Sonstrom P and Baumer M 2011 Phys. Chem. Chem. Phys. 13 19270–84 [5] Zaera F 2010 J. Phys. Chem. Lett. 1 621–7 9

Nanotechnology 25 (2014) 045701

X Ma et al

[6] Zahmakiran M and Ozkar S 2011 Nanoscale 3 3462–81 [7] Liang M, Wang L, Liu X, Qi W, Su R, Huang R, Yu Y and He Z 2013 Nanotechnology 24 245601 [8] Lindquist N C, Nagpal P, McPeak M K, Norris J D and Oh S 2011 Rep. Prog. Phys. 75 036501 [9] Pel´aez R J, Castelo A, Afonso C N, Borr´as A, Espin´os J P, Riedel S, Leiderer P and Boneberg J Nanotechnology 24 365702 [10] Zhang P, Shi Y, Chi M, Park J, Stucky D G, Mcfarland E W and Gao L 2013 Nanotechnology 24 345704 [11] Nolte P, Stierle A, Jin-Phillipp N Y, Kasper N, Schulli T U and Dosch H 2008 Science 321 1654–8 [12] Tian N, Zhou Z Y, Sun S G, Ding Y and Wang Z L 2007 Science 316 732–5 [13] Feldheim D L 2007 Science 316 699–700 [14] Somorjai G A, Tao F and Park J Y 2008 Top. Catal. 47 1–14 [15] Kuhn J N, Tsung C K, Huang W and Somorjai G A 2009 J. Catal. 265 209–15 [16] Rast L and Stanishevsky A 2005 Appl. Phys. Lett. 87 223118 [17] Zawadzki M and Okal J 2008 Mater. Res. Bull. 43 3111–21 [18] Rioux R M, Song H, Hoefelmeyer J D, Yang P and Somorjai G A 2005 J. Phys. Chem. B 109 2192–202 [19] Rioux R M, Song H, Grass M, Habas S, Niesz K, Hoefelmeyer J D, Yang P and Somorjai G A 2006 Top. Catal. 39 167–74 [20] Kim C and Lee H 2009 Catal. Commun. 10 1305–9 [21] Konya Z, Molnar E, Tasi G, Niesz K, Somorjai G A and Kiricsi I 2007 Catal. Lett. 113 19–28 [22] Singh A and Chandler B D 2005 Langmuir 21 10776–82 [23] Lee I, Morales R, Albiter M A and Zaera F 2008 Proc. Natl Acad. Sci. 105 15241–6 [24] Joo S H, Park J Y, Tsung C-K, Yamada Y, Yang P and Somorjai G A 2009 Nature Mater. 8 126–31 [25] Gehl B et al 2008 Adv. Funct. Mater. 18 2398–410 [26] Aliaga C, Park J Y, Yamada Y, Lee H S, Tsung C-K, Yang P and Somorjai G A 2009 J. Phys. Chem. C 113 6150–5 [27] Crespo-Quesada M, Andanson J M, Yarulin A, Lim B, Xia Y N and Kiwi-Minsker L 2011 Langmuir 27 7909–16 [28] Stowell C A and Korgel B A 2005 Nano Lett. 5 1203–7 [29] Menard L D, Xu F T, Nuzzo R G and Yang J C 2006 J. Catal. 243 64–73 [30] Aßmann J et al 2005 Angew. Chem. Int. Edn 44 917–20 [31] Lopez-Sanchez J A et al 2011 Nature Chem. 3 551–6 [32] Astruc D 2007 Nanopart. Catal. 11 349–88 [33] Koh A C W, Leong W K, Chen L W, Ang T P, Lin J, Johnson B F G and Khimyak T 2008 Catal. Commun. 9 170–5 [34] Braden D J, Henao C A, Heltzel J, Maravelias C T and Dumesic J A 2011 Green Chem. 13 1755–65 [35] Deng W P, Tan X S, Fang W H, Zhang Q H and Wang Y 2009 Catal. Lett. 133 167–74 [36] Yoon S J, Kim Y K and Lee J G 2011 Ind. Eng. Chem. Res. 50 2445–51 [37] Yan X P, Liu H F and Liew K Y 2001 J. Mater. Chem. 11 3387–91 [38] Harpeness R, Peng Z, Liu X S, Pol V G, Koltypin Y and Gedanken A 2005 J. Colloid Interface Sci. 287 678–84 [39] Lu F, Liu J and Xu J 2008 Mater. Chem. Phys. 108 369–74 [40] He B L, Chen Y X, Liu H F and Liu Y 2005 J. Nanosci. Nanotechnol. 5 266–70

[41] Viau G, Brayner R, Poul L, Chakroune N, Lacaze E, Fievet-Vincent F and Fievet F 2003 Chem. Mater. 15 486–94 [42] Joo S H, Park J Y, Renzas J R, Butcher D R, Huang W Y and Somorjai G A 2010 Nano Lett. 10 2709–13 [43] Tsukatani T and Fujihara H 2005 Langmuir 21 12093–5 [44] Yang J, Lee J Y, Deivaraj T C and Too H P 2004 J. Colloid Interface Sci. 271 308–12 [45] Chakroune N, Viau G, Ammar S, Poul L, Veautier D, Chehimi M M, Mangeney C, Villain F and Fievet F 2005 Langmuir 21 6788–96 [46] Katarzyna Morawa Eblagon T V-S, Kerry Yu K M, Ramirez-Cuesta A J and Tsang S C 2010 Int. J. Nanopart. 3 104–22 [47] Pelzert K, Laleu B, Lefebvre F, Philippot K, Chaudret B, Candy J P and Basset J M 2004 Chem. Mater. 16 4937–41 [48] Pelzer K, Vidoni O, Philippot K, Chaudret B and Colliere V 2003 Adv. Funct. Mater. 13 118–26 [49] Jansat S, Picurelli D, Pelzer K, Philippot K, Gomez M, Muller G, Lecante P and Chaudret B 2006 New J. Chem. 30 115–22 [50] Tristany M, Chaudret B, Dieudonne P, Guari Y, Lecante P, Matsura V, Moreno-Manas M, Philippot K and Pleixats R 2006 Adv. Funct. Mater. 16 2008–15 [51] Brink M V, Peck M A, More K L and Hoefelmeyer J D 2008 J. Phys. Chem. C 112 12122–6 [52] Narayanan R and El-Sayed M A 2005 J. Catal. 234 348–55 [53] Campelo J M, Luna D, Luque R, Marinas J M and Romero A A 2009 ChemSusChem 2 18–45 [54] White R J, Luque R, Budarin V L, Clark J H and Macquarrie D J 2009 Chem. Soc. Rev. 38 481–94 [55] Brayner R, Viau G and Bozon-Verduraz F 2002 J. Mol. Catal. A 182 227–38 [56] Pachon L D and Rothenberg G 2008 Appl. Organomet. Chem. 22 288–99 [57] Semagina N and Kiwi-Minsker L 2009 Cat. Rev.-Sci. Eng. 51 147–217 [58] Maki-Arvela P, Hajek J, Salmi T and Murzin D Y 2005 Appl. Catal. A 292 1–49 [59] Zhang F-A, Lee D-K and Pinnavaia T J 2009 Polymer 50 4768–74 [60] Borodko Y, Habas S E, Koebel M, Yang P D, Frei H and Somorjai G A 2006 J. Phys. Chem. B 110 23052–9 [61] Borodko Y, Lee H S, Joo S H, Zhang Y W and Somorjai G 2010 J. Phys. Chem. C 114 1117–26 [62] Yang J and Ying J Y 2009 Nature Mater. 8 683–9 [63] Su F B, Lv L, Lee F Y, Liu T, Cooper A I and Zhao X S 2007 J. Am. Chem. Soc. 129 14213–23 [64] Fang M, Machalaba N and Sanchez-Delgado R A 2011 Dalton Trans. 40 10621–32 [65] Bagshaw S A and Bruce I J 2008 Microporous Mesoporous Mater. 109 199–209 [66] Tulevski G S, Myers M B, Hybertsen M S, Steigerwald M L and Nuckolls C 2005 Science 309 591–4 [67] Chen Y, Miller D J and Jackson J E 2007 Ind. Eng. Chem. Res. 46 3334–40 [68] Luque R and Clark J H 2010 Catal. Commun. 11 928–31

10