Synthesis of Platinum–Ruthenium Nanoparticles under Supercritical CO2 and their Confinement in Carbon Nanotubes: Hydrogenation Applications

Abstract

Bimetallic platinum–ruthenium nanoparticles stabilised by pyridine- and monophosphine-based ligands were prepared either in supercritical CO₂ or in THF. TEM analyses evidenced a tendency of the nanoparticles prepared in supercritical CO₂ to agglomerate. Both types of bimetallic nanoparticles were further confined into functionalised multiwalled carbon nanotubes. Upon confinement, PtRu nanoparticles stabilised by phosphine ligands appeared more agglomerated than those stabilised by the pyridine ligand. These materials were applied to cinnamaldehyde hydrogenation. Confined PtRu nanoparticles showed higher catalytic activity and selectivity than unsupported nanoparticles.

Introduction

The application of metallic nanoparticles (MNPs) as catalysts has undergone an exponential growth during the last decade.1–6 Immobilised on solid supports, such as carbon, inorganic oxides (e.g., silica, alumina, titanium dioxide), and polymeric materials, they have been applied to a number of catalytic transformations,7–10 and are easily separated from the organic product for recovery and reuse.11 Notably, the catalytic activity and selectivity could be modified by changing the support. Multiwalled carbon nanotubes (MWCNTs) have proven to be good catalyst supports, owing to their specificity: 1) high catalytic activities owing to their mesoporous nature, which avoids mass transfer limitations;12 2) high activity and selectivity owing to the confinement effect of their inner cavity;13 3) well-defined and tuneable structures;14 and 4) selective growth of nanocarbons by use of catalytic chemical vapour deposition on defined substrates to build nanoreactors.15 MNPs supported on MWCNTs have been employed recently in hydrogenation reactions of organic compounds.16, 17

The chemical preparation of MNPs often involves the reduction of a metal precursor in the presence of a stabilising agent dissolved in a solvent.18 The nature of the solvent influences the size and shape of MNPs, and it can also participate as a stabilising agent. Supercritical carbon dioxide (scCO₂) has recently been used as a medium for the preparation of MNPs19–22 and the dispersion of nanoparticles over a solid support23 or over MWCNTs.24 Carbon dioxide is a nontoxic, relatively inert solvent and a green alternative to toxic organic reaction media. scCO₂ has a high diffusivity and miscibility with gases, favouring mass transfer phenomena. Its critical point is accessible (304 K and 7.38 MPa)25 and its density and dielectric constant, for example, can be modified by changes in pressure and temperature, which is useful for the adaptation of scCO₂ to the solubilisation of different compounds. Furthermore, the shape and size distribution of MNPs can be controlled by use of scCO₂.20–22, 26 To overcome the solubility limitation that results from the low polarity of carbon dioxide, water-in-scCO₂ microemulsions formed with aid of perfluorinated surfactants have been employed in the preparation of nanoparticles.27–29 Perfluorinated ligands were also used, either to increase the solubility of the precursors30 or to stabilise the nanoparticles.31

Fluorinated P-donor ligands (2 and 3 in Scheme 1) were selected to stabilise bimetallic platinum–ruthenium nanoparticles (PtRuNPs) for studies of their behaviour in an scCO₂ medium. Based on previous encouraging results from the preparation of mono- and bimetallic nanoparticles in THF with 4-(3-phenylpropyl)pyridine as a stabiliser (Scheme 1), ligand 1 has been used as a comparison.32, 33 PtRuNPs prepared in both THF and scCO₂ solvents were confined in functionalised MWCNTs to study their catalytic behaviour in the selective hydrogenation of cinnamaldehyde (CAL) to cinnamyl alcohol (COL).

Scheme 1. Synthesis of PtRuNPs (PtRuLx), stabilised by 4-(3-phenylpropyl)pyridine (1) and monophosphine ligands (2, 3) in THF (a) and scCO₂ (b) (1 bar=100 kPa).

Results and Discussion

Synthesis of bimetallic PtRuNPs

PtRuNPs were synthesised through the simultaneous decomposition of [Ru(cod)(cot)] (cod=1,5-cyclooctadiene; cot=1,3,5,7-cyclooctatetraene) and [PtMe₂(cod)] organometallic precursors under a hydrogen atmosphere at room temperature in the presence of ligand 1, 2, or 3 in THF (a) or scCO₂ (b) solvent (Scheme 1). The metal/ligand molar ratio was Pt/Ru/L=1:1:0.5.

The IR spectra of the bimetallic nanoparticles 4b–6b prepared in neat scCO₂ 2400–2800 psi evidence two absorption bands in the CO stretching region at ν=2008–2023 and 1932–1961 cm⁻¹ (Table 1 and Figure S1 in the Supporting Information). For PtRuNPs stabilised by phosphine ligands 2 and 3, the band at higher energy is thinner and more intense than that at lower energy; however, for 4b, both bands are fairly similar in intensity. The formation of CO can be explained by the mediated metal activation of CO₂, which gives CO and O.34 Based on literature data for bimetallic PtRu systems,35 both bands are attributed to linearly bonded CO on metal, with the higher energy band attributable to Pt–CO interactions, in agreement with our previous descriptions of Pt nanoparticles,36 and the other band corresponding to Ru–CO stretching. The difference in intensity corresponds to the unequal contents of ruthenium and platinum, observed in elemental analyses. For 4b, the Ru/Pt ratio is 2:1, whereas for 5b and 6b it is approximately 1.25:1 (see calculated formula in Table 1).

Table 1. Diameter, composition, and IR data for PtRuNPs.

PtRuNP^[a]	Diameter [nm]^[b]	IR data, ν [cm^−1]^[c]	Calculated formula from elemental analysis
[a] See Scheme 1; [b] Determined by TEM; [c] Data in the range of 2100–1900 cm^−1, samples prepared as KBr pellets; [d] Data from Ref. [33].
4a^[d]	2.2±0.7	–	Pt1.15Ru1(1)1.2
5a	1.8±0.6	–	Pt1.4Ru1(2)0.28
6a	0.9±0.1	–	Pt0.8Ru1(3)0.56
4b	2.5±1.1	2008, 1932	Pt0.5Ru1(1)0.5
5b	1.5±0.4	2023, 1967	Pt1.2Ru1(2)0.33
6b	1.3±0.4	2018, 1961	Pt1.3Ru1(3)0.35

TEM analyses were performed for PtRuNPs prepared in both solvents (Figure 1). Those synthesised in THF are small (mean size between 1 and 2.2 nm; see Table 1 and Figure S2 in the Supporting Information for size distribution diagrams) and well-dispersed, mainly for PtRuNPs stabilised by phosphine ligands 5a and 6a. In contrast, PtRuNPs prepared in scCO₂ show a tendency to agglomerate. 4b constitutes a mixture of spherical nanoparticles nonhomogeneous in size (mean diameter 2.5±1.1 nm) and small rods (length ≈20–30 nm) (Figure 1). The phosphine ligand-containing PtRuNPs 5b and 6b gave small nanoparticles (1.5±0.4 and 1.3±0.4 nm, respectively), analogously to experiments in THF. Energy-dispersive X-ray spectroscopy analyses evidenced the presence of both metals in the nanoparticles (Figure S3 in the Supporting Information).

Figure 1. TEM micrographs of bimetallic nanoparticles prepared in THF (4a–6a) and scCO₂ (4b–6b). Scale bar=20 nm for 4a–6a and 5b, 50 nm for 4b and 6b.

Preparation of catalytic materials

Commercially available MWCNTs were functionalised following our previously established procedure.33 Firstly, MWCNTs were segmented and opened by ball milling, to give a length of 0.1–1 μm. They were further oxidised by nitric acid treatment and then successively treated with thionyl chloride and hexadecylamine to introduce amide functions onto the MWCNT surface. These amide-functionalised MWCNTs were used for the confinement of the PtRuNPs prepared in THF or scCO₂ (see above). The confinement of bimetallic nanoparticles 4a–6a and 4b–6b (PtRuLx@MWCNT; L=1–3, x=a or b) was performed successfully by using a simple impregnation procedure (see the Experimental Section), as evidenced by TEM analysis (Figure 2). The particles were located at the ends of the tubes and also in the inner cavity. The particles are not uniformly distributed inside the cavities, showing a tendency to agglomerate.

Figure 2. TEM micrographs of nanoparticles 4–6 confined in functionalised MWCNTs. Scale bar=100 nm for 5a, 6a, and 6b; 200 nm for 4a, 4b, and 5b.

Catalytic hydrogenation

The catalytic performances of PtRuLx and PtRuLx@MWCNT nanocatalysts were evaluated in the selective hydrogenation of CAL as a benchmark reaction (Scheme 2).37 Blank experiments performed at 20 bar of H₂ (1 bar=100 kPa) and 343 K showed that MWCNTs are not active for this reaction. Under the above-mentioned reaction conditions and in the presence of the catalysts, the only products of CAL hydrogenation were cinnamyl alcohol (COL), hydrocinnamaldehyde (HCAL), and hydrocinnamyl alcohol (HCOL). The results obtained with PtRuLx and PtRuLx@MWCNT are reported in Tables 2 and 3, respectively. The turnover frequencies (TOFs) are related either to the total metal introduced or to the amount of platinum.

Scheme 2. Hydrogenation of cinnamaldehyde catalyzed by PtRuLx or PtRuLx@MWCNT.

Table 2. Cinnamaldehyde hydrogenation catalysed by nonconfined nanoparticles, PtRuNP.^[a]

Entry	Catalyst	Pt/Ru ratio [wt. %]	TOF^[b]	Selectivity [%]^[c]
[a] Reactions at 343 K and 20 bar of H2, see Scheme 2; [b] Data at 5 % conversion, TOF=molCOL molPt+Ru^−1 h^−1 (in brackets, TOF=molCOL molPt^−1 h^−1); [c] Selectivity=[molCOL/(molCOL+molHCAL+molHCOL)]×100 (in brackets, the relative ratio of COL/HCAL/HCOL).
1	4a	20:9	0.5(0.9)	55(55:19:26)
2	4b	1:1	0.6(1.8)	21(21:0:79)
3	5a	47:17	0.2(0.3)	99(99:1:0)
4	5b	16:7	0.4(0.7)	58(58:42:0)
5	6a	19:12	0.2(0.4)	66(66:0:34)
6	6b	41:17	0.2(0.3)	80(80:20:0)

Table 3. Cinnamaldehyde hydrogenation catalysed by confined nanoparticles, PtRuNP@MWCNT.^[a]

Entry	Catalyst	t [h]^[b]	Φ [nm]^[c]	TOF [h^−1]^[d]	Selectivity [%]^[e]
[a] Reactions at 343 K and 20 bar of H2, see Scheme 2; [b] Time at 50 % conversion; [c] Mean diameters for PtRuNPs upon confinement; [d] TOF=molCOL molPt+Ru^−1 h^−1 (in brackets, TOF=molCOL molPt^−1 h^−), for metal composition, see Table 2; [e] Selectivity=[molCOL/(molCOL+molHCAL+molHCOL)]×100 (in brackets, the relative ratio of COL/HCAL/HCOL).
1	4a@MWCNT	2.2	3.0	10.4(19.3)	71(71:14:15)
2	4b@MWCNT	2.2	3.0	8.8(25.6)	71(71:20:9)
3	5a@MWCNT	3.0	2.4	8.5(14.5)	67(67:13:20)
4	5b@MWCNT	2.1	5.3	13.5(25.1)	76(76:6:18)
5	6a@MWCNT	5.5	2.9	10.5(23.3)	68(68:16:16)
6	6b@MWCNT	4.7	4.3	8.5(15.1)	72(72:14:14)

The catalytic activities of the unsupported PtRuNPs were very low (TOF<1 h⁻¹; Table 2), irrespective of the nature of the stabiliser or of the solvent used to prepare them. Such poor performances have been previously observed for the catalyst 4a,33 and it seems that neither the particle size (1–2.5 nm) nor the nature of the ligand (pyridine or phosphine) have an influence on the catalytic activity. On such particles, it seems that the presence of the σ-donor ligands studied prevent the adsorption of the reactants. Owing to the low conversions obtained, a discussion on the differences of selectivity appears hazardous.

For confined PtRuNPs, a similar particle size was obtained in the case of 4a@MWCNT or 4b@MWCNT (Table 3, entries 1 and 2), whereas larger sizes were measured for confined 5x@MWCNT and 6x@MWCNT (x=a or b; Table 3, entries 3–6), particularly for those prepared in scCO₂ (5b, 6b; Table 3, entry 3 vs. entry 4 and entry 5 vs. entry 6). This result could be attributed to the surface chemistry between the stabilised nanoparticles and MWCNT walls:33 The affinities of ligands 2 and 3 for the MWCNT walls, higher than that of ligand 1, result in a destabilisation of PtRuNPs by partial ligand decoordination and thus in an enhancement of the sintering. Furthermore, for PtRuNPs that contain labile ligands 2 and 3 prepared in scCO₂, agglomeration appears particularly favoured, owing to the plausible CO adsorption on carbon nanotubes, as reported in some theoretical studies.38

The confinement of PtRuNPs inside MWCNTs produces a drastic increase in the activity; the TOFs obtained for the confined catalysts were found in the range between 8 and 13 h⁻¹. The use of a confined cavity to perform catalysis may impact several properties both at the mesoscopic and the microscopic scale.39 It is possible that high local concentrations of reactants are present inside MWCNTs, helping the displacement of the equilibrium between ligands and reactants. Such a phenomenon would enhance the adsorption of reactants and, in turn, the catalytic activity. Better activities were obtained with ligand 2 relative to 3, owing to the higher σ-donor character of the ligand and thus strength of the metal–phosphorous bond. Regarding selectivity, it has already been shown that the use of bimetallic PtRu catalysts foster high selectivity towards COL.33, 40 In addition, it is known that for CAL hydrogenation, the selectivity to COL increases as the particle size increases.41 The higher selectivity achieved with 5b@MWCNT compared to the other catalytic systems is attributable to the larger particle size observed for this confined system (d_mean=5.3 nm; Table 3, entry 4). The planar CAL molecule cannot adsorb parallel to a flat metal surface because of steric repulsion from the aromatic ring. In addition, the adsorption mode to the different planes of the metallic surface depends on the substrate particle size. Theoretical calculations have shown, furthermore, that aromatic rings that are chemisorbed on a metal surface must lie at a distance exceeding 0.3 nm because there is an energy barrier that prevents a closer approach to the surface.33 As a result of this energy barrier, the CC bond cannot approach the surface as closely as the CO bond and so the latter is hydrogenated preferentially. This steric effect does not operate for smaller particles, whereby both CC and CO functionalities can approach the surface.

Moreover, we investigated the stability of confined PtRuNPs by performing TEM analyses of the catalysts after catalytic runs. Whatever the system investigated, the presence of PtRuNPs in the inner cavity was observed (Figure S4 in the Supporting Information), pointing out the stability of these systems.

Conclusions

We have successfully prepared small (1–2.5 nm) bimetallic platinum–ruthenium nanoparticles (PtRuNPs) stabilised by various ligands in either THF or in scCO₂. Although well-dispersed bimetallic PtRuNPs were obtained in THF, preparation in scCO₂ led to agglomeration. All PtRuNPs were successfully confined within the inner cavity of surface-passivated multi-walled carbon nanotubes (MWCNTs). When monophosphines were used as stabilisers, larger PtRuNPs were observed for confined systems than for unsupported ones. This fact can be tentatively rationalised by consideration of the PtRuNP ligand affinity with the MWCNT walls. The catalytic study has shown the superiority of confined catalysts over unsupported PtRuNPs. The results obtained for confined catalysts regarding activity and selectivity can be attributed to a higher local concentration of reactants and catalyst inside MWCNTs.

Experimental Section

General

Syntheses of bimetallic nanoparticles in THF and catalytic materials supported on MWCNTs were performed under argon atmosphere, using standard Schlenk techniques. Functionalised MWCNTs (Pyrograph III, Applied Science, 98 % purity) were prepared following the methodology previously described.33 All solvents were previously distilled and degassed. Starting chemicals were stored and weighed in a glovebox. Syntheses of bimetallic nanoparticles in scCO₂ were performed in a Parr autoclave (25 mL) with magnetic stirring. The autoclave was equipped with a liquid inlet, a gas inlet, and a thermocouple. An electric heating mantle kept the temperature constant. TEM analyses were performed at the Service Commun Microscopie Electronique, Université Paul Sabatier (TEMSCAN) by using a transmission electron Philips CM12 microscope operating at 120 kV; a JEOL JEM 1011 transmission electron microscope operating at 100 kV for routine images with resolutions of 4 and 5 Å; a slow-scan acquisition Megaview III scanning imaging spectrometer and a JEOL JEM 2100F transmission electron microscope operating at 100 kV for high-resolution images with resolutions of 2 and 3 Å; camera slow-scan image acquisition (Gatan 1 K×1 K) X-ray analysis PGT for light elements detection with resolution of 135 eV. The size distribution and average diameter of nanoparticles were determined directly from TEM images by using Image-J software associated with a Microsoft Excel macro developed by Christian Pradel. Gas chromatography analyses were performed by using a Perkin–Elmer Clarus 500 chromatograph fitted with a flame ionisation detector using anisole as the internal standard and a Stabilawax-DA column (Restec, 30 m×0.25 mm×0.25 μm; temperature range 40–240/250 °C). The injector temperature was 150 °C and the split 0.02 μL. The column head pressure of the helium carrier gas was maintained at 23 psi throughout the analysis. The temperature programme was: 353 K (2 min)/20 K min⁻¹ to 473 K/473 K (7 min).

Synthesis of PtRuNPs 4a–6a in THF

PtRu2a and PtRu3a were prepared by following the methodology described previously for 4a.33 [Ru(cod)(cot)] (71 mg, 0.23 mmol), [PtMe₂(cod)] (75 mg, 0.23 mmol), and the corresponding ligand (0.10 mmol: 60.3 mg of 2 or 109 mg of 3) were dissolved in THF (55 mL) in a Fisher–Porter bottle then pressurised with 3 bar of H₂ at RT and stirred overnight. After evaporation of the solvent, the black solid was washed with pentane (3×10 mL). The product was dried under reduced pressure to yield PtRu2a (102 mg) or PtRu3a (162 mg), which were stored in the glovebox.

Synthesis of PtRuNPs 4b–6b in scCO₂

A 25 mL stainless steel reactor (Parr Instrument Company) was filled with [Ru(cod)(cot)] (71 mg, 0.23 mmol), [PtMe₂(cod)] (75 mg, 0.23 mmol), and the corresponding ligand (0.10 mmol: 24 mg of 1, 61 mg of 2, or 109 mg of 3) in the glovebox. The system was then pressurised under 4 bar of H₂, filled with liquid CO₂, and pressurised to supercritical conditions to reach 2400–2800 psi. The agitation speed was fixed at 1000 rpm. The reaction took place for 24 h at RT, after which the reactor was placed in an ice bath and the pressure released. The bimetallic nanoparticles were then dispersed in THF. The solvent was evaporated under vacuum and the particles were washed with pentane (3×10 mL). The particles were dried under vacuum to yield PtRu1b (44 mg), PtRu2b (35 mg), and PtRu3b (33 mg), which were stored in a glovebox.

Synthesis of PtRuLx@MWCNTs

Functionalised MWCNTs (45 mg) were mixed in a solution of preformed nanoparticles (5 mg) in THF (2 mL) and underwent an impregnation procedure of alternating stirring and ultrasonic treatments (4×30 min), after which the solvent was removed under vacuum.

General procedure for cinnamaldehyde hydrogenation

The PtRuLx@MWCNT catalyst (35 mg) in 2-propanol (25 mL) was first activated in a stainless steel reactor (Top Industrie) under 20 bar of H₂ at 343 K for 2 h under stirring (1000 rpm). After that, a solution of cinnamaldehyde (1 g, 7.6 mmol) in of 2-propanol (25 mL) was transferred to the reactor under argon atmosphere and the hydrogenation reaction takes place under 20 bar of H₂ at 343 K. The reaction was monitored by use of gas chromatography.

Acknowledgements

ANR (Nanoconficat, project no. ANR-05-NANO), CNRS (Centre National de la Recherche Scientifique), Université Paul Sabatier, Ministerio de Ciencia e Innovación (CTQ2007-63510/PPQ and CTQ2010-16676), and Consolider Ingenio CSD2006-0003) are gratefully acknowledged for financial support.