“Imprinting” Catalytically Active Pd Nanoparticles onto Ionic‐Liquid‐Modified Al₂O₃ Supports

Abstract

Going against tradition: Uniformly distributed Pd nanoparticles on imidazolium-ionic-liquid-modified Al2O3 surfaces are prepared by a top down approach by using a new sputtering chamber. The hydrogenation of 1,3-cyclohexadiene is used to probe the surface properties of these new Pd nanoparticles. A new sputtering chamber that allows the constant mixing of the solid support during sputtering by using an electro-magnetic oscillator was developed for the generation of metal nanoparticles (top down approach) uniformly distributed over the solid supports. By using this new chamber, small and well-distributed Pd nanoparticles (Pd-NPs) of 2.6 or 4.3 nm were produced over Al2O3 and new imidazolium ionic liquids covalently supported on Al2O3 by simple sputtering from a Pd foil. The Pd-NPs uniformly distributed over the solid supports display comparable catalytic performances in the hydrogenation of 1,3-cyclohexadiene and 1,3-cyclooctadiene to that achieved by using a catalyst prepared by conventional chemical methods (bottom up approach). Mechanistic and labelling studies show that the hydrogenation of 1,3-cyclohexadiene catalysed by Pd-NPs occurs via meta-stable π-allyl intermediates, characteristic of homogeneous-like catalytically active sites, and disproportionation through the outer-sphere mechanism. The ratio of hydrogenation to disproportionation is dependent on the Pd-NP size, and the disproportionation products are more pronounced with small NPs because of the higher affinity of dienes for this size of particle. Therefore, the surface structural features of small Pd-NPs facilitate the arrangement of diene molecules in the correct geometry for the transfer of H from the donor to the acceptor sites, typical of poly-metallic catalytically active sites. It is proposed that the reaction of 1,3-cyclohexadiene under H2 can be used to probe the homogeneous/heterogeneous nature of supported metal NPs. It is demonstrated for the first time that highly active and selective nano-catalysts were obtained by using the sputtering-deposition technique (top down approach) and this opens a new window of opportunity for the preparation of size-controlled metal NPs with clean surfaces. Transition metal nanoparticles (M-NPs) are currently used as effective catalysts for hydrogenation processes in which homogeneous and traditional heterogeneous catalysts are less efficient.1, 2 Modern soluble M-NPs are commonly synthesised by chemical methods, which involve the reduction of metallic salts or the decomposition of organometallic complexes in the presence of stabilising agents.1–5 Recently, the use of sputtering-deposition techniques was revealed as an interesting alternative for the synthesis of soluble M-NPs with clean surfaces under mild conditions with easy control of their size and shape by appropriate tuning of the sputtering conditions.6–15 This approach allows the rapid synthesis of M-NP catalysts by the deposition of metals on support surfaces or the fast generation of M-NPs in solution by using low volatility liquids such as silicones, triglycerides, and ionic liquids (ILs). This method has been widely applied for the formation of thin (nano-sized) films onto supports;16 however, it is limited as it is extremely difficult to prepare uniformly distributed M-NPs on solid catalytic supports.17 Among various types of agents for stabilising soluble M-NPs, ILs18 are extensively employed because of their high efficiency in hydrogenation processes.4 Recently, the use of supported ionic liquid phases (SILPs) is receiving growing interest as the IL surface area is increased relative to its volume and the substrate can readily diffuse to the catalysts, overcoming the mass-transfer limitations produced by the high viscosity of the IL.19–22 The SILP system uses much smaller amounts of IL than those required under liquid–liquid bi-phasic conditions, and this behaviour is of crucial importance for reactions that involve gases with poor solubility in ILs (e.g., H2 and CO). Therefore, catalysis that uses a SILP as a support combines the positive effects of ILs on the catalytic performance with the high stability of the catalysts, good diffusion of the reactants and easy product separation displayed by heterogeneous catalysts. The most widely used ILs in SILP applications are imidazolium ILs, which combine the high solvation power of polar species with their weak coordination strength, making them suitable for reactions with electrophilic catalysts.21, 22 Silica is the most widely used support for the immobilisation of M-NPs owing to its high surface area and the presence of OH, which facilitates the anchoring of stabilising agents, whereas the use of other supports has been seldom reported.23 To the best of our knowledge, the use of Al2O3-based supports, which offer high stability at high pH, in a SILP system has not been explored until now. We report the synthesis and characterisation of imidazolium ILs covalently immobilised on Al2O3 surfaces. Pd-NPs were uniformly distributed on these supports by using a sputtering-deposition technique. Furthermore, the hydrogenation of 1,3-cyclohexadiene was used to probe the surface properties of these new SILP Pd-NPs. The hydrogenation of 1,3-cyclohexadiene was chosen as a benchmark reaction as it is a structure-sensitive reaction24 and its selectivity (hydrogenation vs. disproportionation) can be used as a fingerprint for the homogeneous or heterogeneous nature of the Pd-NPs surface. Moreover, the partial hydrogenation of π-conjugated systems, such as 1,3-dienes, is a process of growing interest because of the formation of high-added-value alkenes conventionally used as building blocks in organic synthesis.25 To obtain uniformly distributed M-NPs in both solid and liquid substrates, a new sputtering chamber has been developed by our group. This new chamber contains an electro-magnetic oscillator (with variable controlled frequency), which allows the constant movement of the conical flask that contains the support (Figure 1).26, 27 Schematic representation of the sputtering-deposition chamber used for the immobilisation of Pd-NPs onto Al2O3 supports. The M1 support was synthesised by the reaction of 1-methyl-3-(trimethoxysilylpropyl)imidazolium chloride (1) with the hydroxyl groups of the Al2O3 surface and further ion-exchange treatment according to modified literature procedures28 was performed to obtain the M2–M4 supports that contained NTf2−, PF6− and BF4− counter-ions, respectively (Scheme 1). Synthesis of M1–M4. 13C cross-polarisation magic angle spinning (CP-MAS) NMR spectra displayed four sets of upfield broad peaks (10, 25, 36 and 52 ppm), which could be assigned to the propylene and methylene functions, whereas the resonances at 123 and 136 ppm were assigned to the three imidazolium C atoms (Figure 2). Thus, these NMR signals confirmed the presence of the unaltered imidazolium ILs in M1–M4. No signals attributed to the methoxy leaving groups (65 ppm) of the silane function were observed, which confirms that the trimethoxysilane group effectively reacted with the hydroxyl groups of the Al2O3 surface under our reaction conditions. 13C CP-MAS NMR spectra of a) M1, b) M2, c) M3 and d) M4. FTIR analysis of M1–M4 showed CN stretching vibration bands of the imidazolium rings and CH stretching vibration bands of the alkyl groups at 1635 and 2950 cm−1, respectively (Figure 3). Thus, these IR signals also confirm the presence of the unaltered imidazolium ILs in M1–M4. Furthermore, the FTIR spectra of M2–M4 display signals attributed to NTf2− (1058, 1145, 1220 and 1350 cm−1), PF6− (741 and 841 cm−1) and BF4− (763, 1039 and 1061 cm−1), respectively, which confirm the efficiency of the Cl−-exchange procedure.29, 30 FTIR spectra of a) M1, b) M2, c) M3 and d) M4. N2 physisorption analysis revealed type-IV-like isotherms with pore diameters of 6.6–7.1 nm for the unmodified Al2O3 and M1–M4 (Table 1). The M1–M4 supports displayed lower surface areas (SBET) and pore volumes than unmodified Al2O3, which indicate that the ILs filled the pores. Similar behaviour was observed for the ILs covalently supported on SiO2.31 Elemental analysis revealed that M2–M4 contained slightly lower amounts of IL than M1, which suggested that the IL was removed from these supports during the Cl−-exchange step (Table 1). Support SBET [m2 g−1] Pore volume [cm3 g−1] Pore diameter [nm][a] Organic content [mmolIL g−1][b] Al2O3 195 0.47 7.1 – M1 134 0.31 6.9 0.46 M2 127 0.28 6.6 0.36 M3 125 0.30 7.0 0.39 M4 80 0.20 7.1 0.38 The effect of Cl− exchange on the support structure was investigated by 29Si CP-MAS NMR analysis (Figure 4). The spectrum of M1 displays typical signals of Si atoms attached to one C atom and to the Al2O3 surface at −46 and −55 ppm assigned to the T1 (20 %) and T2 (80 %) species, respectively.32 After Cl− exchange, a new signal appeared at −65 ppm in the spectra of M2–M4, which was attributed to the formation of T3 and suggests that re-arrangement processes operate under the reaction conditions of the Cl−-exchange step. This behaviour could be ascribed to the formation of HF by the hydrolysis of the F-based anions.33 This behaviour was much more pronounced for the supports that contained PF6− and BF4− (40 and 50 % T3, respectively), which decomposed more easily than NTf2− (10 % T3; Figure 4). Therefore, M1 and M2 should be appropriate supports for the deposition of Pd-NPs by sputtering as the IL is much more uniformly distributed on their surfaces than on the surfaces of M3 and M4. 29Si CP-MAS NMR spectra of a) M1, b) M2, c) M3 and d) M4. Al2O3, M1 and M2 were selected as suitable materials for the deposition of Pd-NPs by the sputtering technique at 100 W for 1.5 min. These supported catalysts were analysed by using inductively coupled plasma optical emission spectroscopy (ICP-OES), which determined Pd contents of 0.11±0.01, 0.10±0.01 and 0.12±0.02 wt. %, respectively. TEM analysis revealed the formation of small NPs (2.6±1.3, 2.7±1.1 and 4.3±0.6 nm, respectively) uniformly distributed on the supports (Figure 5). TEM images and histograms of a) Pd/Al2O3, b) Pd/M1, c) Pd/M2 and d) Pd*/M2. It is clear that similar Pd-NP sizes were obtained on Al2O3 and M1, whereas M2 displayed larger NPs. This behaviour suggests that the size of the NPs is controlled by the nature of the anions, and larger NPs are obtained by increasing the non-polar domains of the anions, akin to M-NPs prepared in pure ILs.13 Indeed, the Pd-NPs dispersed in pure BMI⋅NTf2 (BMI=1-butyl-3-methylimidazolium) by sputtering under the same reaction conditions and Pd*/M2 synthesised by the H2 reduction of Pd(acac)2 (acac=acetylacetonate) displayed similar sizes (4.3 and 5.1 nm, respectively). These new nano-catalysts prepared by sputtering techniques were first tested in the selective hydrogenation of 1,3-cyclohexadiene under 4 bar of H2 at 40 °C as this reaction is commonly used as a benchmark test for the evaluation of the activity and selectivity of Pd-NPs used as hydrogenation catalysts.34–39 The catalysts synthesised by sputtering had quite high activities (ca. 15.8–20.1 s−1; Table 2, entries 1–5). These catalysts formed cyclohexene (4) with high selectivity (up to 95 %) together with significant amounts of benzene (3; ca. 5–11 %) as a by-product (which originates from the hydrogen disproportionation of the diene; Table 2, entries 1–3; Figures 6 and 7), which is not commonly reported in the literature as attention is usually only paid to the partial hydrogenation selectivity (cyclohexene vs. cyclohexane formation rate).34–39 This behaviour was much less pronounced for Pd-NPs supported on M2, probably because of their larger sizes compared to those supported on Al2O3 and M1 (Table 2, entries 1–3 and Figures 6 and 7). This is a strong indication of the higher diene affinity for the smaller NP surfaces, which may allow competition between the hydrogenation and disproportionation pathways.40, 41 The catalytic activity was affected by the reaction temperature, with activities of up to 20.1 s−1 obtained at 60 °C with small variations in the product selectivity (Table 2, entries 3–5). Hydrogenation of 1,3-cyclohexadiene: t [s] vs. selectivity [%] for cyclohexene catalysed by Pd/Al2O3 (▴) and Pd/M1 (▪) and t [s] vs. 1,3-cyclohexadiene conversion [%] catalysed by Pd/Al2O3 (▵) and Pd/M1 (□). Hydrogenation of 1,3-cyclohexadiene: t [s] vs. selectivity [%] for cyclohexene with Pd/M2 (▴) and with Pd*/M2 (▪) and t [s] vs. 1,3-cyclohexadiene conversion [%] catalysed by Pd/M2 (▵) and with Pd*/M2 (□). Entry[a,b] Catalyst Conversion [%] (t [h]) 3 [%] 4 [%] 5 [%] TOF [s−1][c,d] 1 Pd/Al2O3 100 (0.13) 11 87 2 19.1 2 Pd/M1 100 (0.17) 11 88 1 18.2 3 Pd/M2 100 (0.75) 5 95 0 15.8 4[e] Pd/M2 100 (0.75) 6 92 2 13.2 5[f] Pd/M2 100 (0.50) 6 93 1 20.1 6[g] Pd*/M2 100 (0.75) 9 90 1 14.2 For comparison, Pd-NPs of similar sizes supported on M2 were synthesised by H2 reduction of Pd(acac)2 precursors to yield Pd*/M2 (Figure 5). Pd*/M2 was tested in the hydrogenation of 1,3-cyclohexadiene and produced a slightly lower activity and higher benzene selectivity than those of Pd/M2 (Table 2, entry 3 vs. 6). However, analysis of the reaction kinetics revealed that Pd*/M2 displayed an incubation period probably caused by the less clean NP surfaces obtained by conventional methods (Figure 7). The substrate scope was investigated by performing the hydrogenation of 1,3-cyclooctadiene catalysed by Pd/M2 at 40 °C, which produced exclusively cyclooctene with a turnover frequency (TOF) of up to 5.9 s−1 (Figure 8). Hydrogenation of 1,3-cyclooctadiene catalysed by Pd/M2: t [s] vs. selectivity [%] for cyclooctene (▴) and t [s] vs. 1,3-cyclooctadiene conversion [%] (▵). For comparison, Pd(acac)2 and Pd-NPs dissolved in BMI⋅NTf2 were also tested as catalysts for the same reaction (IL/Pd ratio=15, 1,3-cyclohexadiene/Pd ratio=4500, 4 bar H2, 40 °C) and lower 1,3-cyclohexadiene conversions were obtained (up to 10 % after 24 h for the homogeneous system and up to 75 % after 3 h for the IL-soluble NPs). These results demonstrate that the true active species are Pd-NPs and their immobilisation onto Al2O3 improved their catalytic activity with no effect on their product selectivity. Mechanistic studies were performed to elucidate the reaction pathways involved in the hydrogenation and disproportionation of 1,3-cyclohexadiene catalysed by Pd/M2. First, the reaction kinetics from using H2 and D2 were compared, and it was found that a lower activity was obtained with D2 (TOFs of up to 15.8 and 11.5 s−1 for H2 and D2, respectively), which indicates a small kinetic isotopic effect under the reaction conditions employed (Figure 9) probably as a result of the involvement of H−/D− transfer in two steps of the hydrogenation pathway (Scheme 2). Hydrogenation of 1,3-cyclohexadiene with Pd/M2: t [s] vs. conversion [%] for cyclohexene with H2 (▴) and with D2 (▪). Proposed mechanism for the Pd-NP-catalysed hydrogenation of 1,3-cyclohexadiene. Furthermore, a higher formation of 3 (up to 8 %) and lower selectivity to 4 (up to 90 %) were obtained by using D2 as a reactant (Figure 9). GC–MS analysis of the reaction products revealed the formation of [D2]cyclohexene (6 and 7) and [D4]cyclohexane (8) and no incorporation of D in the benzene ring. NMR analysis (1H, 2H, 13C, COSY and HSQC) confirmed the formation of 8 and that there was no incorporation of D in the benzene ring. The 2H NMR spectra performed with relaxation delays of 1–10 s displayed two broad signals at 1.95 and 1.56 ppm assigned to 6 and 7 together with a signal attributed to 8 at 1.37 ppm (Figure 10). 2H NMR spectra of the reaction products obtained in the Pd/M2-catalysed reduction of 1,3-cyclohexadiene with D2. By analysing the integral ratio between these two signals, we estimated a 6/7 ratio of 3.5:1. The 1H NMR spectra, also performed with relaxation delays of 1–10 s, displayed signals attributed to 3 (7.30 ppm) and 8 (1.37 ppm) together with three broad signals at 5.62, 1.94 and 1.55 ppm assigned to a mixture of 4, 6 and 7 (Figure 11). Analysis of the integral ratios between the signals at 1.94 and 1.55 ppm with respect to the signal at 5.62 ppm gave values of 1.44 and 1.61, which confirms the formation of two [D2]cyclohexene isomers (6 and 7). Quantification of the reaction products by using GC analysis together with analysis of the 1H and 2H NMR integral ratios and their comparison with a standard sample of non-deuterated cyclohexene allowed us to estimate the composition of the product distribution as 8 % of 3, 8 % of 4, 64 % of 6, 18 % of 7 and 2 % of 8. 1H NMR spectra of the reaction products obtained in the Pd/M2-catalysed reduction of 1,3-cyclohexadiene with D2. The 13C NMR spectra displayed four upfield signals attributed to the cyclohexene structural isomers (6 and 7; 126.9, 126.8 and 126.7 ppm) and 3 (128.0 ppm; Figure 12). attributed to the C atoms with D with of were observed at and ppm together with signals attributed to non-deuterated C atoms at and These the of two one for the reduction that the species that results from H2 and a H transfer of the 1,3-cyclohexadiene The H for the reduction of the 1,3-cyclohexadiene probably the formation of meta-stable π-allyl (Scheme 2). This of pathway that π-allyl is observed in the hydrogenation of by Pd catalyst 13C NMR spectra of the reaction products obtained in the Pd/M2-catalysed reduction of 1,3-cyclohexadiene with D2. The disproportionation of the diene most probably occurs by a step in which one H atom is from one diene to as amounts of non-deuterated 3 and 4 (Scheme are observed under D2. Moreover, non-deuterated 3 and 4 were in the disproportionation of 1,3-cyclohexadiene performed by using Pd/M2 with D2. the activities found for the catalyst with or with H2 were similar (TOFs of up to and respectively), which suggests that H2 is not involved in the reaction mechanism (Figure However, the catalyst with D2 displayed a lower activity of up to which indicates the presence of kinetic isotopic there is no incorporation of D in the the isotopic effect may be ascribed to the in the NP structure by the of H2 or D2. The presence of such species may structural that may or may not facilitate the arrangement of diene molecules in the correct geometry for the transfer of H from the donor to the acceptor sites. Indeed, such structural were suggested and more it has been demonstrated that H2 and species on the surface of small M-NPs catalytic Proposed mechanism for the Pd-NP-catalysed disproportionation of 1,3-cyclohexadiene. of 1,3-cyclohexadiene with Pd/M2: t vs. conversion [%] with H2 H2 (▪) and with D2 The synthesis and characterisation of imidazolium ILs covalently supported on Al2O3 are of Pd-NPs on these supports can be and easily by simple sputtering from a Pd by using our new sputtering chamber. This chamber allows the constant mixing of the solid support during sputtering by using an electro-magnetic The nano-catalysts prepared by this technique are uniformly distributed over the solid supports and display comparable catalytic performance in the hydrogenation of 1,3-cyclohexadiene and 1,3-cyclooctadiene to those achieved by using a similar catalyst prepared by using conventional It has been demonstrated for the first time that highly active and selective nano-catalysts were obtained by using the sputtering-deposition technique and this opens a new window of opportunity for the preparation of size-controlled metal NPs with clean surfaces. Mechanistic studies show two the hydrogenation and disproportionation of 1,3-cyclohexadiene catalysed by Pd/M2 via meta-stable π-allyl of homogeneous-like catalytically active disproportionation through the outer-sphere of poly-metallic catalytically active sites. Therefore, it is to that the reaction of 1,3-cyclohexadiene under H2 can be used to probe the homogeneous/heterogeneous nature of supported metal NPs. Moreover, the of significant amounts of disproportionation products on the hydrogenation of is a strong indication of the presence of small Pd-NPs. Indeed, the ratio of hydrogenation to disproportionation is dependent on the Pd-NP size, and disproportionation products are more pronounced with small NPs because of the higher affinity of dienes for this size of particle. were performed by using standard techniques under and were according to standard The Al2O3 used in this was by The IL 1 was prepared according to a Elemental analysis of the ILs immobilised on the support surfaces was by using a The N2 isotherms of the supports, at 100 °C under for 3 were obtained by using a The surface areas were determined by using the and the pore size was obtained by using the 13C and 29Si NMR of M1–M4 were performed by using a at the FTIR spectra were obtained by using a with a of 4 with TEM were prepared by the of a of solution under an onto a The TEM were performed by using a at GC were performed by using an GC with a 40 GC–MS were performed by using a with a 40 °C) that employed an of 2H, 13C COSY and analysis of the obtained by D2 reduction of 1,3-cyclohexadiene catalysed by Pd/M2 were performed by using a at the The incorporation of D in the reaction products was by the 1H NMR spectra with these obtained from a standard sample and by performing 2H NMR 10 1 was dissolved in and to Al2O3 The was at 90 °C under and for The Al2O3 was and to yield on the of 1 on the M1 an of or was dissolved in and to M1 to exchange the The were for The were and to yield M2, and M4. a Al2O3, M1 or M2 was in a conical flask and the chamber. The chamber was and its to a of 4 and the support was for 4 the chamber was under a sputtering of 4 by The supports were by the flask at a frequency of 24 The Pd was onto the support at 100 W power for 1.5 min. After the the chamber was with and the was and under for their further characterisation and a M2 and the appropriate of a solution of Pd(acac)2 in were to a The was with 4 bar of H2 and to 75 After 5 the solution which confirmed the formation of Pd-NPs. The reaction was under these conditions for 3 h to the reduction of the Pd the was and After of the reaction Pd*/M2 was by simple a a solution of the appropriate (10 dissolved in was to a that contained the appropriate of catalyst The was with 4 bar of H2 at the were from the reaction mixture 2 min. After the reaction the was to and The conversion and selectivity were determined by GC analysis of the We to and for support for this