Rafael L. Oliveiraa, Pedro K. Kiyoharab and Liane M. Rossi*a
aInstituto de Química, Universidade de São Paulo, 05508-000, São Paulo, SP, Brazil. E-mail: lrossi@iq.usp.br; Fax: +55 11 3815 5579; Tel: +55 11 30912181
bInstituto de Física, Universidade de São Paulo, 05508-090, São Paulo, SP, Brazil
First published on 29th October 2009
We present the magnetic separation approach to facilitate the recovery of gold nanoparticle (AuNP) catalysts. The use of magnetically recoverable supports for the immobilization of AuNPs instead of traditional oxides, polymers or carbon based solids guarantees facile, clean, fast and efficient separation of the catalyst at the end of the reaction cycle. Magnetic separation can be considered an environmentally benign separation approach, since it minimizes the use of auxiliary substances and energy for achieving catalyst recovery. The catalyst preparation is based on the immobilization of Au3+ on the surface of core–shell silica-coated magnetite nanoparticles, followed by metal reduction using two different methods. AuNPs were prepared by thermal reduction in air and by hydrogen reduction at mild temperature. Interestingly, the mean particle size of the supported AuNPs was similar (ca. 5.9 nm), but the polydispersity of the samples is quite different. The catalytic activity of both catalysts in the aerobic oxidation of alcohols was investigated and a distinct selectivity for benzyl alcohol oxidation was observed.
Gold in the bulk form has been regarded as being chemically inert towards chemisorption of reactive molecules such as oxygen and hydrogen. Consequently, bulk gold was considered to be an uninteresting metal from the point of view of catalysis. However, the catalytic properties of gold are revealed when the size is reduced to few nanometers, particularly with dimension less than 10 nm.1 The overall performance of gold nanoparticle (AuNP) catalysts highly depends on the size and shape of the nanoparticles, the structure and properties of the catalyst support and the gold–support interface interactions.2 A variety of interesting catalytic properties of AuNPs in both oxidation and reduction reactions has been reported.1-3 Despite this, little attention has been paid to possible difficulties arising from size reduction, and the challenges of separating the very small particles, most of the time in a colloidal equilibrium, from the products. In this regard, AuNPs have been dispersed on solid supports to increase the catalyst stability and to facilitate the catalyst recovery.3 As the size of the support decreases, separation using physical methods, such as filtration or centrifugation, becomes a difficult and time-consuming procedure. Simple filtration is inefficient to accomplish product isolation in systems comprised of nanoparticles stabilized in solution or very thin solids. In these circumstances, an alternative that has drawn attention lately is the use of magnetically recoverable solid supports. Magnetic separation is easy, fast, clean and a very promising recovery tool for isolation of thin solids from the reaction medium. It can be considered an environmentally benign separation approach, since it minimizes the use of auxiliary substances and energy for achieving catalyst recovery. Magnetic supports, mainly composed of superparamagnetic nanoparticles, can be attracted to relatively low static magnetic field strengths and are easily recovered, since no residual magnetization is observed when the magnetic field is removed. The immobilization of homogeneous and NP-based catalysts on magnetically separable solid supports has been shown to be a powerful tool for catalyst recovery and recycling.4-7
We report herein the preparation of AuNPs supported on a magnetic solid that can be easily recovered from solution by applying an external magnetic field, while keeping good control over different parameters, such as particle size, size distribution and catalytic activity. The oxidation of alcohols was used to probe the catalytic activity of the catalyst. Catalytic oxidations are green alternatives to stoichiometric oxidations with permanganate, manganese dioxide and chromium(VI) reagents, major sources of waste, particularly in fine chemicals and pharmaceuticals manufacture.8
Fig. 1 UV-Vis spectra of ethanolic suspensions of Fe3O4@SiO2-NH21 (a), Au(0) catalyst 4 (b) and Au(0) catalyst 3 (c). |
Scheme 1 Step-by-step preparation of magnetically recoverable AuNPs. |
The morphology of the supported AuNPs were examined by transmission electron microscopy (TEM), shown in Fig. 2a and b. The support is comprised of magnetic cores (Fe3O4∼ 10 nm) spherically coated with silica (spheres of ∼ 40 nm in diameter). The mean diameter of the supported AuNPs in both catalysts was estimated at 5.9 nm with polydispersity of 18% for catalyst 4 and 32% for catalyst 3. The AuNPs are fairly well-distributed over the support in both catalysts; however, the particle size distribution of catalyst 4 is sharper if compared to catalyst 3, as can be seen in the size distribution histograms adjusted to lognormal functions presented in Fig. 2c–d. The SPR band of catalyst 4 appears at λmaxca. 520 nm, which agrees with the higher monodispersity of the AuNPs. Moreover, the λmax red shift of about 45 nm and the broadened optical spectrum of catalyst 3 can be realized by aggregation of small particles, light scattering and lower overall quality of the supported nanoparticles.9
Fig. 2 TEM images of catalyst 3 (a) and catalyst 4 (b). Histogram showing particle size distribution and lognormal fitting of the AuNPs of catalyst 3 (c) and the AuNPs of catalyst 4 (d). |
The distribution and composition of each nanoparticle of the nanocomposite was revealed by energy dispersive spectroscopy (EDS) analysis of the image shown in Fig. 3. The core nanoparticles were found to be exclusively composed of Fe atoms (Fig. 3b), and the darker nanoparticles of Au atoms (Fig. 3c), through a detailed EDS analysis of the image areas shown in box (b) and box (c), respectively. The EDS analysis shown in Fig. 3a corresponds to the whole nanocomposite which contains Fe, Au and Si (Cu from sample holder).
Fig. 3 HRTEM image and EDS analysis of the AuNP magnetically recoverable catalyst 4 in different areas of the image: (a) corresponds to the whole area, (b) corresponds to the iron oxide particle shown in box b and (c) corresponds to the gold particle shown in box c. |
The magnetic properties of the Fe3O4@SiO2 nanocomposite were investigated elsewhere.7 It displays excellent magnetic properties characterized by negligible coercivity and remanence, and very high saturation magnetization MS at room temperature (MS∼ 69 emu g−1 at 70 kOe). These magnetic properties describe a superparamagnetic material that responds to an external magnetic field but does not remain magnetized when the magnetic field is removed. Such materials can be easily concentrated from the solution with a magnet and immediately redispersed after removing the magnetic field, since the absence of remanence prevents aggregation of particles.
The catalytic oxidation of benzyl alcohol was used to probe the activity and recovery of the supported AuNPs in catalysts 3 and 4. The first attempt to catalyze the oxidation of benzyl alcohol using our supported AuNPs was performed under atmospheric pressure of O2 and no products (aldehyde or ester) were observed (Table 1, entry 1). It was also observed that in the absence of base, the catalyst was not efficient in catalyzing the selective oxidation of alcohols, even at 100 °C and 5 atm O2 (Table 1, entry 2). The conversion increased to 75% (95% selectivity) as long as a small amount of K2CO3 was present (Table 1, entry 3). As it can be seen in Fig. 4, there is a strong dependence of the oxidation reaction with the O2 pressure for both catalysts 3 and 4. Catalyst 3 is less active and requires higher pressure for the substrate conversion, but the selectivity was maintained in the pressure range studied. On the other hand, catalyst 4 is more active, reaching 100% conversion at 3 atm of O2, but the high activity is accompained by the drop of selectivity to benzaldehyde to ca. 50%. The oxidation reaction catalyzed by catalyst 3 is highly selective to benzaldehyde, found as the main product. It is also known that benzaldehyde can be further oxidized into benzoic acid, and, finally, benzyl alcohol and benzoic acid would interact and produce benzyl benzoate.10 In fact, benzyl benzoate was found as the only by-product in the oxidation of benzyl alcohol by catalysts 3 and 4.
Entry | Catalyst | Substrate | P O2 (atm) | Conversion (%)b | Selectivity (%)b | |
---|---|---|---|---|---|---|
Aldehyde or ketone | Ester (homocoupling) | |||||
a Reaction conditions: 1.5 mmol of substrate, 2 mL of toluene, 50 mg of catalyst (1.8 μmol Au), 70 mg K2CO3, 100 °C, reaction time 6 h.b Determined by GC-MS.c Bubbling O2.d Without base. | ||||||
1 | 3 | Benzyl alcohol | —c | 0.0 | — | — |
2d | 3 | Benzyl alcohol | 5 | 2.2 | 99 | 1 |
3 | 3 | Benzyl alcohol | 5 | 75 | 95 | 5 |
4 | 3 | Benzyl alcohol | 6 | 84.3 | 95 | 5 |
5 | 4 | Benzyl alcohol | 3 | 100 | 47 | 53 |
6 | 3 | 1-Phenyl ethanol | 6 | 100 | 100 | 0 |
7 | 4 | 1-Phenyl ethanol | 3 | 100 | 100 | 0 |
8 | 3 | Cyclohexanol | 6 | 31 | 100 | 0 |
9 | 4 | Cyclohexanol | 6 | 42 | 100 | 0 |
Fig. 4 Pressure dependence of the catalytic performance of supported AuNPs: catalyst 3 (circle) and catalyst 4 (square). Reaction conditions: 1.5 mmol of benzyl alcohol, 50 mg catalyst, 70 mg K2CO3, 2 mL toluene, 100 °C, 6 h. |
Secondary alcohols such as 1-phenyl ethanol and cyclohexanol (Table 1, entries 6–9) were also oxidized to the corresponding ketones with 100% selectivity. In all experiments, the catalyst was recovered by using a permanent magnet and the products were collected with a syringe and analysed by GC-MS.
Recycling experiments were examined for the catalytic oxidation of benzyl alcohol at 5 atm of O2 and 100 °C using catalysts 3 and 4. After the first reaction, the catalyst was recovered in the reactor wall by using a permanent magnet, the products were collected with a syringe, and the recovered catalyst was washed with ethanol (2 mL) inside the reactor, and dried under vaccum. The catalyst was reused in successive oxidation reactions by adding new portions of solvent, base and substrate. This procedure was repeated four times, however the supported AuNP catalyst 3 lost 50% of the initial conversion, considering the same reaction time. After the 4th cycle, the conversion of benzyl alcohol was 35% but the selectivity to aldehyde was maintained as shown in Fig. 5. The deactivation process of catalyst 4 is more severe than catalyst 3 under similar experimental conditions. The activity of 100% in the 1st run drops to less than 10% in the 4th run.
Fig. 5 Conversion and selectivity of the catalytic oxidation of benzyl alcohol to benzaldehyde using the magnetically recovered AuNP catalysts 3 and 4 (conditions similar to entry 3, Table 1). |
The low reusability of the catalyst is a point of concern. One of the key factors that must be considered is the possibility that the active metal would leach into the reaction mixture, especially in liquid phase oxidation reactions, because of the possible dissolution of metals by the reactants and, particularly, the carboxylic acid-type (by)products, thereby leading to catalyst deactivation. To address this possibility, the isolated products obtained after magnetic separation were analysed by atomic absorption analysis (ICP OES). Results have shown that negligible Au leaching of our catalysts occurs (less than the limit of quantitation of 2.5 ppb of Au). This demonstrates the high affinity of gold and our magnetic support, the high performance of the magnetic separation in the present catalytic system and, as a first conclusion, indicates that the catalyst deactivation is not related to metal loss. Another key factor to be considered is the stability of the catalyst support and nanoparticles when subjected to the reaction conditions. TEM images and EDS analysis of the spent catalysts (see ESI†, Fig. S1) and TEM image of the catalyst support subjected to basic reaction conditions (see ESI†, Fig. S2) provided supporting evidence for the catalyst deactivation under reuse. First, the morphology of the catalyst support (core–shell Fe3O4@SiO2) has changed; the silica layer was strongly affected by the presence of base in the reaction medium and the catalyst support lost the spherical morphology, which caused aggregation. Second, large Au particles (∼80 nm) were revealed in the TEM image of spent catalyst, which suggests the coalescence of the small, active gold particles into larger, less active or inactive particles. The growth of the Au particles, above the catalytic limit of about 10 nm,1 can also be a consequence of the changes occurring in the catalyst support during the reaction. Therefore, the use of base associated with a silica-based catalyst support should be the main reason for the catalyst deactivation. Another point that might be relevant concerns the presence of potassium as detected by EDS analysis. This is also a consequence of the use of base that can cause metal surface contamination.
It is also important to mention that some catalytic activity still remains even after the 4th reuse (ca. 50% for catalyst 3), most probably because small particles are still present, as indicated by the TEM image.
The catalysts reported here are very efficiently recovered by magnetic separation, with negligible Au leaching to the solution. This separation procedure is an efficient and green alternative to the conventional separation techniques used for the recovery of solids, such as filtration and centrifugation, since it is fast, clean, easy to scale-up and constitutes a waste minimization and low energy consumption procedure.
The use of base, K2CO3, was necessary to attain good catalytic activities in the oxidation of alcohols by our AuNPs catalysts, but seems to be responsible for the catalyst deactivation under reuse due to degradation of the catalyst support, which causes metal particles to grow beyond the catalytically active limit. The literature contains alternatives for the use of base in such reactions, such as the use of basic and redox supports3q,10 or bimetallic Au–Pd alloys.2f Additional experiments are in progress in our laboratory in order to synthesize and charaterize magnetic catalysts comprised of supports and/or metal particles more resistant to base or, hopefully, active in the absence of base.
Footnote |
† Electronic supplementary information (ESI) available: TEM and EDS analyses of the spent catalyst. See DOI: 10.1039/b916825g |
This journal is © The Royal Society of Chemistry 2010 |