A new mechanism for allylic alcohol isomerization involving ruthenium nanoparticles as a ‘true catalyst’ generated through the self-assembly of supramolecular triruthenium clusters

Abstract

The primary objective of this study is to determine the ‘true catalyst’ in an allylic alcohol isomerization reaction involving μ₃-oxo-triruthenium(III) acetate [Ru₃O(OCOCH₃)₆(H₂O)₃][OCOCH₃] as catalyst. This ruthenium-complex was previously presumed to act as a homogeneous catalyst. However, we have confirmed the heterogeneous nature of this catalytic reaction that proceeds through in situ-formed metallic nanoparticles. A new, five-step mechanism, where ruthenium nanoparticles are generated from the supramolecular triruthenium clusters during the reaction and guided the progress of the reaction, has been discovered. In this model, a stepwise reduction of ruthenium-centers within the complex is considered generating ruthenium nanoparticles during the reaction and those steps play a key role for the optimization process. Three autocatalytic steps have been proposed to obtain the best fitted profile for the reaction. The ‘true catalyst’ of the reaction has been identified as Ru⁰ nanoparticles having a certain size and geometry (described as ‘C’). In addition, the smaller sized particles (B) present in the reaction mixture are also believed to show some catalytic activity.

---

Introduction

Isomerization of primary and secondary allylic alcohols to the corresponding aldehydes and ketones is a reaction of vast synthetic value.^1–8 Various transition metal complexes of Ru, Rh, Co, Ni, Mo, Ir, and Pt have been used as catalysts for the isomerization reaction.^8–13 In the current work, the reaction (Fig. 1) using μ₃-oxo-triruthenium(III) acetate¹⁴ as a catalyst (catalyst 1) has been studied in detail to investigate the ‘true catalyst’ in the reaction. This Ru-complex (catalyst 1) was originally believed to function as a homogeneous catalyst (e.g. discrete organometallic complex) in the isomerization process of the allylic alcohol to the corresponding ketone.¹⁰ However, the interesting and unusual reaction profile curve with sigmoidal shape has raised some doubt. The question is: does the allylic alcohol isomerization undergo homogeneous catalysis by the organometallic chemistry mechanisms, or does it undergo heterogeneous catalysis by in situ formed Ru⁰ nanoparticles?

Fig. 1 Schematic representations of allylic alcohol isomerization using μ₃-oxo-triruthenium acetate as catalyst (catalyst 1).

In order to make this distinction, the following studies have been performed. A successful method to distinguish between a “heterogeneous” colloid-nanocluster catalyst and a discrete, “homogeneous” catalyst was developed in 1994 by Yin Lin and Richard G. Finke,¹⁵ which is summarized in Fig. 2. It emphasizes: (1) the early use of TEM as a simple and powerful way to detect soluble nanoclusters; (2) kinetic studies (as catalysis is a totally kinetic phenomenon: explanation of any induction period); (3) quantitative phenomenological tests include the catalyst poisoning experiments using mercury or added ligand, especially if it is performed quantitatively; and (4) the important concept that the identity of the true reaction catalyst should be consistent with all the obtained data. Then, a clear conclusion regarding the nature of catalysis could be obtained by carrying out the above studies. In the case of heterogeneous catalysis by metallic nanoparticles, the mechanism of their self-assembly should be proposed and proved. It took many experimental studies to scrutinize this final mechanism, which can both, describe the reaction profile, as well as fit to all the results obtained from the experimental studies.^15–20 Most acceptable mechanism for nanoclusters formation was proposed by Finke and co-workers as given in Fig. 3.

Fig. 2 A general approach to distinguish homogeneous catalysis and heterogeneous catalysis.

Fig. 3 A general approach for mechanism of metal nanoparticles self-assembly in reductive conditions.

This scheme considers the most important kinetic steps that influence the shape of the experimentally obtained reaction profile. The first step is the nucleation step that causes the formation of small nanoparticles, B, from a discrete complex, A. The second step is the autocatalytic step of nanoparticles surface growth. The atom of metal in the complex contacts with the formed particle B and gives the growth of the nanoparticle. In the third step, the agglomeration to C particles of a bigger size occurs. In the last step, the autocatalytic agglomeration of smaller B particles, with larger C particles to give C-bulk metal, takes place. As a result, a four-step, double autocatalytic mechanism was proposed.

Most of the organometallic compounds that were examined in the literature on their appropriateness for this mechanism had only one metal center,⁸ rarely two.²¹ However, cases of three metallic centers were reported, but the general mechanism did not always fully correspond with the reaction curve, as in the one- or two-center cases.²² This leads to the idea that in these cases the general mechanism needs some modifications in order to match experimental data.

In our system, catalyst 1 has a complicated structure (Fig. 1), with three metallic Ru centers bonded to the oxygen atom. In the case of heterogeneous catalysis through in situ-formed nanoclusters, each Ru³⁺ atom in the complex should be reduced to Ru⁰, and only then self-assembled to Ru⁰ nanoparticles. Thus, it is imperative to first understand the elementary steps in the reduction process of μ₃-oxo-triruthenium(III) acetate. Based on previous studies, which deal with hydrogen reduction of olefins in the presence of complex 1,²³ and hydrogen reduction of complex 1,²⁴ the main principles of a possible reduction process were proposed. During 1978–1979, Fouda and coworkers studied those reactions and explained the results without any consideration of nanoparticles. In their works, the H₂ consumption versus time was measured, and the following order of reactivity towards the reduction within complex 1 was proposed and proved. Initially, only one metal center reacts as if it was a complex of a single metal center, while the other two metals behave as regular ligands. When the first-reduced atom is almost consumed, the other two Ru atoms, which remained untouched, also begin to react. It is important to note that the reduction process mechanism of complex 1 is the same, whether the reduction occurs as a parallel process during another chemical reaction (to yield an active catalyst form),²³ or whether it is a direct hydrogen reduction of complex 1.²⁴ Based on the assumption regarding the reduction order, we tried to make parallel modifications of the general model and scrutinize the new mechanism for self-assembly of nanoparticles that can fit to a supramolecule with several metallic centers (such as catalyst 1).

Experimental

Materials

Ethanol (assay >99.9%) and sodium acetate trihydrate (crystal-assay 100%) were purchased from J.T. Baker, while 1-octene-3-ol (assay 98%) and ruthenium(III) chloride, hydrate was bought from Aldrich. Glacial acetic acid (99.7%) and dichlorotris(triphenylphosphine)ruthenium(II) (99.95%) were supplied from Frutarom Ltd and Alfa Aesar, respectively. All the chemicals were used without additional purification.

Synthesis of catalyst 1

[Ru₃O(OCOCH₃)₆(H₂O)₃][OCOCH₃] was prepared from commercially available ruthenium trichloride trihydrate, according to a procedure based on the one given by Wilkinson and co-worker.^25,26 4 g of RuCl₃·xH₂O, together with 7 g of NaOCOCH₃·3H₂O, were dissolved in a mixture of 75 ml glacial acetic acid and 75 ml of ethanol. The solution was refluxed under nitrogen for approximately 2 h, and the initial reddish-brown color of the solution changed to dark green. The solution was cooled to −30 °C and decanted to separate precipitated sodium chloride and sodium acetate. The decanted solution was then filtered and the filtrate was taken to dryness. The solid thus obtained was the crude oxo-triruthenium(III) acetate complex. The resulting crude acetate complex was dissolved in a minimum amount of ethanol and cooled overnight to −30 °C. Sodium acetate and sodium chloride were precipitated and separated by filtration. The filtrate was taken to dryness and the solid was washed with benzene to remove excess acetic acid. Several extractions with ethanol were performed until no more precipitate was apparent on filtration. This final filtrate was taken to dryness, and after a final benzene wash the product was dried overnight in vacuum over sodium hydroxide pellets at 60 °C. The yield was 3.2 g (80% on the basis of Ru content of RuCl₃·xH₂O).

Reaction conditions

The following standard conditions were used: 5 ml of reaction solvent, ethanol, were placed in a three-necked, round-bottomed flask (25 ml), equipped with reflux condenser and thermometer, and heated to desired temperature. Then, 0.5 mg of pre-prepared catalyst, μ₃-oxo-triruthenium(III) acetate, were dissolved in the reaction solvent. The ReactIR 4000 probe was inserted into the third neck to measure in situ IR spectra of the reaction mixture. The solvent with the dissolved catalyst was used as a background for IR measurement. Then, 4.2 g of reaction substrate (1-octene-3-ol) were added. The reaction took place under reflux conditions, T = 80 °C, until no more product formation was observed (2 h). These experiments were performed with substrate concentrations in the range of 0.91–3.6 M and catalyst concentration in the range of 1.48 × 10⁻⁵ to 10⁻³ M.

Method for on-line monitoring

Mettler Toledo/Applied Systems ReactIR™ 4000, a compact ATR-FTIR instrument designed for real-time, in situ analysis of chemical reaction has been used to monitor the reactions. The reaction measurement was performed at the following conditions: diamond sensor; optical range: 650–1950 cm⁻¹, 2150–4000 cm⁻¹; scans: 32; resolution: 4; backgrounds: catalyst dissolved in solvent/catalyst dissolved in solvent with aliquat 336; profile: start – 1712 cm⁻¹.

Methods for nanoparticles characterizations

TEM (High Resolution Transmission Electron Microscopy) analysis was performed using a TECNAI F20 G² from FEI, USA having point resolution 0.24 nm, line resolution 0.1 nm and limit of information 0.15 nm. Samples were prepared by putting a drop of the solution onto a carbon-coated copper grid and then allowed to evaporate before analyzing. SAED (Selected Area Electron Diffraction) pattern was also collected in the same instrument. Particle size analysis was measured by Dynamic Light Scattering (DLS) technique using Zetasizer Nano-S (λ = 633 nm; Malvern Instruments, UK) with a scattering angle of 173° at 25 °C. Data were collected in 3 repeated measurements (10 scans for each repeat). The average size was reported by the number distribution (statistics) for each measurement.

Methods for mechanism optimization

MacKinetics is an integrated package for modeling chemical reaction kinetics. Version 0.9b runs on any Macintosh computer with a 68020 CPU and math coprocessor. Version 0.9.1b runs on Power Macintosh computers.¹⁶

Hg(0) poisoning test

The Hg poisoning test was performed in ethanol as the solvent; hence, the reaction procedure was the same as in Exp. 1, but with addition of vigorous stirring. At the point where the reaction velocity was nearly at its maximum value, an excess amount of Hg(0) was added to the reaction solution, with continuous stirring (250 rpm) throughout the reaction process, and without interruption of IR measurements. The experiment was performed twice under exactly the same conditions.

[1-octene-3-ol] = 1.6 M; [cat (1)] = 0.115 × 10⁻³ M; [Hg⁰] = 0.1 M.

Results

Curve shape of the reaction

Isomerization of 1-octene-3-ol (substrate; S) to 3-octanone (product; P) has been studied in presence of catalyst 1 (reaction/eqn (1)) using ethanol as solvent. Progress of the reaction has been monitored by the ATP-FTIR spectrophotometer and the obtained plot of the product concentrations vs. time is given in Fig. 4.

Fig. 4 Profile reaction composition in ethanol as a solvent, T = 80 °C, [catalyst 1]₀ = 1.48 × 10⁻⁴ M, [substrate]₀ = 3.6 M, conversion = 1.

From the experimental results we can see that the reaction profile has a sigmoidal shape with well defined induction time which is in fact much longer than the reaction time itself. However, full conversions were achieved in all the experiments. The presence of an induction period indicates that the catalyst 1, in the form in which it was injected to the reaction flask, is not the actual catalyst. It is expected that the catalyst 1 was converted into its active form after the induction period and that active species acts as actual catalyst. It can be noted that the sigmoidal acceleration hints the presence of autocatalysis in reaction (1), which is typical for catalysis by in situ formed nanoparticles.^21,27

Visual evidence

During the reaction propagation a color change of the solution was observed from deep green (the color of catalyst 1) to brown or yellowish-brown, which suggests the formation of Ru⁰ nanoparticles in the solution.²⁸ When the reaction was taking place, the reaction flask was allowed to stand in fume-chamber for a long time (about one month) at room temperature and with no stirring. As a result, the dark brown (or black) solid sediment was precipitated at the bottom of the flask. This colour also applies to colloidal Ru⁰ particles.²⁸

Characterization of catalytic particles

To check the existence of the Ru⁰ nanoparticles and also to measure their sizes we have used TEM, SAED and DLS techniques. In the initial stage of the reaction i.e. during the induction time period before any reaction took place, no particles were found. However, samples taken at the end of the reaction resembles the existence of very tiny particles as shown in the Fig. 5. Fig. 5a reveals the presence of some particles of a nano-scale size (black points in the pictures) with sizes ∼3–6 nm. The approximate size dispersion was measured using a DLS instrument (Fig. 5b). The size range of the observed particles in the reaction solution was 3.7 to 12 nm; average size of the particles was estimated to be 5.5 nm, which is in good agreement with the TEM result.

Fig. 5 (a) TEM and (b) DLS results of the solution taken at the end of the reaction. Conditions: T = 80 °C, [substrate]₀ = 3.1 M, [catalyst 1]₀ = 3 × 10⁻⁴ M, conversion = 1.

We also analyzed the black precipitate obtained on keeping the reaction mixture undisturbed for long time. To determine the crystallographic structures of the lattices in the particles, in atomic scales, an SEAD pattern was captured from those particles. Fig. 6 shows the typical SAED pattern, where the spots can be indexed to the (101), (110), (103) and (112) planes of the crystalline hexagonal closed-packed (hcp) structure (JCPDS # 01-089-4903) of Ru metal.²⁹ This supports the proposition that these particles are Ru⁰ particles, formed in situ during the isomerization reaction.

Fig. 6 SAED pattern of the black precipitate obtained on keeping the reaction undisturbed for long time. The spots corresponding to the lattice planes of are indicated in the figure.

Kinetic studies

We propose that during the isomerization reaction, Ru³⁺ in complex 1 has been reduced to give Ru⁰ atoms, which will then assemble to form nanoparticles. The original organometallic complex is catalytically inactive (presence of induction time) and the true catalyst of reaction (1) is the Ru⁰ nanocluster. Hence, the way of the reduction process of Ru³⁺ in this complex should influence the total profile of reaction (1), as in the case of olefin hydrogenation.²⁴ Therefore, the order of metals reduction should be considered in the putative mechanisms of nanocluster formation.

First, the general mechanism (Fig. 3) was checked in the optimization process, using the MacKinetics program. No good suitability was achieved even after a relatively large number of simulations. Poor visual and calculated (residual number of about 0.3–0.5) suitability was found in all cases: with B as the true reaction catalyst, C as the true reaction catalyst, and both of them, B and C, as catalysts. Representative example is shown in Fig. 7.

Fig. 7 Optimization process results: general mechanism considering (a) B as the true reaction catalyst. T = 80 °C, [catalyst 1]₀ = 2.3 × 10⁻³ M, [substrate]₀ = 3.2 M, conversion = 1, residual ≈ 0.4 and (b) C as the true reaction catalyst. T = 80 °C, [catalyst 1]₀ = 2.3 × 10⁻³ M, [substrate]₀ = 3.2 M, conversion = 1, residual ≈ 0.3.

Second, we decided to make some changes and to try other models. We proposed the new mechanism for nanoparticles self-assembly that contains elementary steps of metal centers reduction, and which probably, more efficiently, describes the experimental profiles.

The first step in this model was considered as it was in the general scheme i.e. the nucleation step.¹⁶ Initially, only one of three metal centers, Ru³⁺ (=A), was reduced to give Ru⁰ (=B) nanoparticles,^23,24 while the other two remained in the oxidation form, a organometallic complex (complex 2) with different structure (X) was formed. Here, complex 1 has been reduced by the solvent (i.e. ethanol) which releases 2 hydrogen atoms, as if they were H₂ (there is no proved common mechanism for this reaction yet), to give corresponding aldehyde – oxidation of alcohols.^30,31 The above step can be represented schematically as follows (step-1):

In the second step, autocatalytic surface growth occurs. In this step, the reduction of the first metal (Ru³⁺) center (A) began in some of the complex 1, prior to the other two metal (Ru³⁺) center in complex 2. Therefore, this step is catalyzed by already formed particles to give Ru⁰ surface growth by one atom (=2*B) and complex 2 (X) (step 2):

After the first two steps, we assumed that there was almost no complex 1 left in the system,^23,24 thus, the two remaining Ru ions in complex 2 begin to undergo reduction by ethanol. The Ru⁰ nanoparticles that are already present in the reaction solution catalyze this process, causing an additional autocatalytic surface growth of these particles by two atoms (for every mole of complex 2, two moles of Ru⁰, =3*B are produced). This is schematically shown in step-3.

The fourth step (step-4) was left as in the common mechanism, i.e. agglomeration of two smaller B nanoparticles to form a nanoparticle of a bigger size, C.¹⁶

In the last autocatalytic step (step 5) agglomeration of nanoparticles of both sizes, B and C, to a large bulk metal particle C, takes place in the same way as in the common mechanism.¹⁶

According to this mechanism there are four types of potential catalysts of isomerization reaction (1): complex 1, complex 2, and Ru⁰ nanoparticles from both B and C sizes. The nature of these probable catalysts is different: complex 1 and 2 are of homogeneous type, while B and C particles are heterogeneous. We performed a simple test in order to understand the nature of reaction (1).

Mercury poisoning test

A well-known method to distinguish a heterogeneous catalysis from a homogeneous one is by adding excess amount of liquid mercury to the examined reaction mixture in comparison to the Ru-content. Hg⁰ is “poison” to heterogeneous catalysts, due to its ability to form an amalgam with the metal catalyst. In the case of heterogeneous catalysis, the addition of Hg⁰ should suppress reaction propagation.^15,32,33

In order to perform this test, the reaction of the same reactants' composition and conditions was carried out in the presence and absence of Hg⁰. In our original experiment (without mercury) full conversion was achieved, as expected (Fig. 8). In the second case, after the induction period was over, when the product formation started and the reaction rate was approximately at its maximum, an excess amount of liquid mercury was inserted directly into the reaction flask. The reaction conversion at this point was 40% (Fig. 8).

Fig. 8 Profile of reaction in the presence and absence of Hg⁰. Reaction conditions: ethanol as solvent, T = 80 °C, vigorous stirring, [substrate]₀ = 1.6 M, [catalyst 1]₀ = 1.5 × 10⁻⁴ M.

From this reaction profile it is clearly seen that there was no product formation following mercury injection (constant conversion – 40%). The addition of Hg⁰ stopped the allylic isomerization reaction (1) in the presence of catalyst 1, thus, the true reaction catalyst was of heterogeneous nature. This test eliminated complexes 1 and 2 from the list of possible catalysts.

The task of identifying which of the particles, B or C, is the true reaction catalyst, is a difficult one. Both of them are Ru⁰ nanoparticles those can give heterogeneous catalysis on their surfaces. The only slight difference is in their size. Hence, it was decided to check both of them from the kinetic appropriation point of view. That means that there are three variations of the proposed mechanism. In mechanism A1, B is the true catalyst (S + B → P + B, k₆), whereas mechanism A2 involves C is the true catalyst (S + C → P + C, k₇). Finally, there might be the situation that two different catalysts (both B and C), involved in the reaction to follow mechanism A3 (S + B → P + C, k₆ and S + C → P + C, k₇). It is known that catalytic activity of nanoparticles is strongly dependent on their size (and shape). Previous experience shows the existence of an optimal size, and geometry of a nanoparticle that gives the best activity and efficiency for specific chemical reactions.^34,35 Nevertheless, nanoparticles of a slightly smaller or bigger size are also catalytically active, but they are weaker and slower. That is why the mechanism with both types of particles (B and C) as catalysts, was also checked. After checking all the possible models by using MacKinetics computer program, the most suitable mechanism was chosen and found that the best way to explain the experimental data (Fig. 9) is by using mechanism A3, where both B and C were acted as catalysts. However, the fitting considering only C as catalyst was also very close to this result, which suggests C to be the most active catalyst. Therefore, we can propose a mechanism to be the correct one which involves nanoparticles self-assembly as active catalyst synthesized in situ from supramolecule μ₃-oxo-triruthenium(III) acetate precursor. Other mechanisms showed higher values of residual numbers and worse visual suitability.

Fig. 9 Optimization process results: conditions: T = 80 °C, [catalyst 1]₀ = 2.3 × 10⁻³ M, [substrate]₀ = 3.2 M, conversion = 1; residual ≈ 0.02, when the rate constants were: k₁ = 5.4 × 10⁻⁵ M⁻¹ min⁻¹, k₂ = 2.123 × 10³ M⁻¹ min⁻¹, k₃ = 6.605 × 10³ M⁻¹ min⁻¹, k₄ = 4.107 × 10³ M⁻¹ min⁻¹, k₅ = 17.1 × 10³ M⁻¹ min⁻¹, k₆ = 1.1 × 10³ M⁻¹ min⁻¹, k₇ = 103.0 × 10³ M⁻¹ min⁻¹.

Approximated sizes of B and C particles

Unfortunately, because of the nature of the materials used, it was impossible to measure the size distribution of the nanoclusters directly by the TEM instrument. Hence, the DLS technique was used to obtain the approximated magnitudes of the produced nanoparticles. The reaction mixture was measured at the end of the induction period and the following results were obtained as shown in Fig. 10a. Two particle populations were observed having the size ranges from ∼0.5–1.7 nm (average size = 1 nm) and ∼5.5–12 nm (average size = 6.5 nm) which can be correlated to the nanoclusters B and C, respectively. In most samples, only the larger size population was detected at the end of the reaction. Hence, that population (and not the smaller one) describes the C range (Fig. 10b).

Fig. 10 DLS measurement results of reaction (1) solution in ethanol as a solvent. (a) Conditions: reaction mixture at the end of induction period, [catalyst 1]₀ = 5.8 × 10⁻⁴ M, [substrate]₀ = 1.5 M, conversion = 1. (b) Conditions: reaction mixture at the end of reaction, [cat]₀ = 1.8 × 10⁻⁴ M, [substrate]₀ = 0.9 M, conversion = 1.

Discussions

According to the studies that have been conducted, the following observations were made: (1) the reaction profiles were of sigmoidal form with observed induction time; (2) the TEM studies, coupled with DLS measurements, gave a visual indication of the existence of some small particles of nano scale at the end of the reaction, which were found to be Ru⁰ particles; (3) the kinetic optimization study supplied the evidence for the complex triple-autocatalytic and multi-step reaction behavior, where an in situ formation of nanoparticles takes place; (4) an indication of the true reaction catalyst (nanoclusters of C type) kinetically confirms the reaction profile and corresponds with all experimentally observed data; (5) the poisoning test using liquid mercury suggests the metallic heterogeneous nature of the reaction catalyst; (6) the color change of the reaction mixture can be used as additional evidence to support the most suitable mechanism and proposed nanoparticles formation by in situ. Therefore, in conclusion, it can be said that the catalytic nature of the allylic alcohol isomerization of 1-octene-3-ol is heterogeneous, and that the true reaction catalysts are in situ formed nanoparticles.

MacKinetics fittings revealed that the mechanism involving C-sized particles as the catalyst (the mechanisms from type A2 and A3) gave the best result. The improvement in the fittings was clearly observed by smaller values of residual number and better visual approximation as well, using C-sized particles as true catalyst in comparison to B-sized particles. However, the best optimization was achieved in type A3 mechanisms involving double-cycle catalysis (by both B and C particles). Nonetheless, comparing the optimized values of the two rate constants k₆ (the step in which B is a catalyst) and k₇ (where C is a catalyst), a great difference has been observed: k₆ ≪ k₇ (by approximately 10²). Therefore the isomerization reaction progresses much faster under the catalysis by C-type nanoparticles than by B-sized particles. Thus we can conclude here that C is the “true reaction catalyst”, since the term used to emphasize the most active form of the catalysts.

Active surface area changes with time when particles grow in size and smaller particles, in general, provide larger active surface area. Despite the lager surface area of relatively smaller B-sized particles, C-sized particles have demonstrated much higher activity in the catalysis of reaction (1). This can be explained by the fact that the formation rates of B-sized particles (k₁, k₂, k₃), as observed in Fig. 9, are relatively lower than its consuming rates (k₄, k₅). This suggests that B-sized particles have shorter life-time than C-sized particles under the reaction conditions, thus C-sized particles showed superior catalytic activities. Another possible explanation for this might be the difficulty of the reaction substrate (1-octene-3-ol) to reach the active catalyst's surface when the particles are too small. For catalysts such as transition metal nanoparticles, the particle surface is composed of different types of sites classified as vertex, edge, and face atoms.³⁶ The active site for the reaction (1) was considered to be one particular type of surface atom, or a group of them.³⁷ Therefore, the fraction of different types of sites, with respect to the total amount of surface atoms, becomes critical for the catalytic activity.^38,39 This fraction is known to be strongly dependent on the size and geometry of the catalyst.⁴⁰ It has been well recognized that the reaction rate per unit of catalyst surface area can vary with particle size. Decreasing particle size, however, will not always result in an increased reaction rate per unit of mass of transition metal.^41–43

Another possible hypothesis is derived from the theory of metal-to-ligand bonds strength.^44,45 For the smaller sized particles, the surface-to-volume ratio is larger, therefore the electrophilic nature is greater and M–L bonds are stronger than for the bigger nanoparticle. Thus, the B-sized particle probably attracts to the free ligands (or substrate; 1-octene-3-ol) present in the reaction mixture much stronger than the C-sized particle, and does not release them easily. In other words, catalytically active sites are neither remained empty nor accessible to the reactants. These adsorbents are unable to prevent the aggregation of B particles to give C particles, but they may affect the catalytic activity of B (“poisoning” of the active surface). An additional influence of this strong binding with B particles is the difficulty of the adsorbed product to leave the surface of the catalyst. This can lower the reaction rate of the catalytic cycle where B is a catalyst (too stable intermediates decrease the reaction velocity – volcano principle). Therefore it has been established that C-type particles (with appropriate size and geometry) were acted as the ‘true reaction catalyst’.

Conclusions

The general conclusion of this research is that the allylic alcohol isomerization of 1-octene-3-ol to 3-octanone, in the presence of μ₃-oxo-triruthenium(III) acetate as a precatalyst, undergoes heterogeneous catalysis through in situ formation of Ru⁰ nanoclusters. Ru⁰ nanoparticles with optimum size and geometry (C-type; average size ∼ 6.5 nm) have been acted as the ‘true reaction catalysts’. The mechanism by which these nanoparticles are self-assembled differs from the common type. There is a special order of reduction of Ru³⁺ centers within the supramolecule that strongly influence the reaction curve and should be considered in the process of proposing mechanism. The most suitable mechanism that we propose supports this idea. Therefore, it can be suggested here that in the common cases of catalysis similar to the catalytic process described above, if the pre-catalyst of interest be a complicated-structured molecule^46–48 from several metallic centers, one should first try to understand the process (order) of their intramolecular reduction and then a suitable kinetic model should be proposed. And by optimizing those mechanisms, the best model that corresponds to all the observed results can be obtained.

Acknowledgements

We acknowledge financial support from the Israel Science Foundation (Grant No. 12/207).

Notes and references

1. K. Tanaka, S. Qiao, M. Tobisu, M. M. C. Lo and G. C. Fu, J. Am. Chem. Soc., 2000, 122, 9870–9871.
2. B. Martín-Matute, K. Bogar, M. Edin, F. B. Kaynak and J. E. Bäckvall, Chem.–Eur. J., 2005, 11, 5832–5842.
3. V. Cadierno, P. Crochet and J. Gimeno, Synlett, 2008, 8, 1105–1124.
4. R. C. van der Drift, E. Bouwman and E. Drent, J. Organomet. Chem., 2002, 650, 1–24.
5. K. Voronova, M. Purgel, A. Udvardy, A. C. Benyei, A. Katho and F. Joo, Organometallics, 2013, 32, 4391–4401.
6. P. Lorenzo-Luis, A. Romerosa and M. Serrano-Ruiz, ACS Catal., 2012, 2, 1079–1086.
7. A. M. Trzeciak and J. J. Ziółkowski, J. Mol. Catal., 1987, 43, 15–20.
8. R. Uma, C. Crévisy and R. Grée, Chem. Rev., 2003, 103, 27–52.
9. R. C. van der Drift, E. Bouwman and E. Drent, J. Organomet. Chem., 2002, 650, 1–24.
10. Y. Sasson and G. L. Rempel, Tetrahedron Lett., 1974, 47, 4133–4136.
11. Y. Sasson, A. Zoran and J. Blum, J. Mol. Catal., 1979, 6, 289–292.
12. L. Bellarosa, J. Díez, J. Gimeno, A. Lledós, F. J. Suárez, G. Ujaque and C. Vicent, Chem.–Eur. J., 2012, 18, 7749–7765.
13. A. Varela-Álvarez, J. A. Sordo, E. Piedra, N. Nebra, V. Cadierno and J. Gimeno, Chem.–Eur. J., 2011, 17, 10583–10599.
14. O. Almog, A. Bino and D. Garfinkel-Shweky, Inorg. Chim. Acta, 1993, 213, 99–102.
15. Y. Lin and R. G. Finke, Inorg. Chem., 1994, 33, 4891–4910.
16. C. Besson, E. E. Finney and R. G. Finke, J. Am. Chem. Soc., 2005, 127, 8179–8184.
17. P. D. Kent, J. E. Mondloch and R. G. Finke, J. Am. Chem. Soc., 2014, 136, 1930–1941.
18. C. Besson, E. E. Finney and R. G. Finke, Chem. Mater., 2005, 17, 4925–4938.
19. B. J. Hornstein and R. G. Finke, Chem. Mater., 2004, 16, 139–150.
20. J. A. Widegren, J. D. Aiken, S. Özkar and R. G. Finke, Chem. Mater., 2001, 13, 312–324.
21. C. M. Hagen, J. A. Widegren, P. M. Maitlis and R. G. Finke, J. Am. Chem. Soc., 2005, 127, 4423–4432.
22. C. M. Hagen, L. Vieille-Petit, G. Laurenczy, G. Süss-Fink and R. G. Finke, Organometallics, 2005, 24, 1819–1831.
23. S. A. Fouda, B. C. Y. Hui and G. L. Rempel, Inorg. Chem., 1978, 17, 3213–3220.
24. S. A. Fouda and G. L. Rempel, Inorg. Chem., 1979, 18, 1–8.
25. A. Spencer and G. Wilkinson, J. Chem. Soc., Dalton Trans., 1972, 1570–1577.
26. P. Legzdins, R. W. Mitchell, G. L. Rempel, J. D. Ruddick and G. Wilkinson, J. Chem. Soc. A, 1970, 3322–3326.
27. L. S. Ott, M. L. Cline, M. Deetlefs, K. R. Seddon and R. G. Finke, J. Am. Chem. Soc., 2005, 127, 5758–5759.
28. W. Chen, D. Ghosh, J. Sun, M. C. Tong, F. Deng and S. Chen, Electrochim. Acta, 2007, 53, 1150–1156.
29. T. W. Hansen, J. B. Wagner, P. L. Hansen, S. Dahl, H. Topsøe and C. J. H. Jacobsen, Science, 2001, 294, 1508–1510.
30. Y. Sasson and J. Blum, J. Chem. Soc., Chem. Commun., 1974, 309–310.
31. B. A. Steinhoff and S. S. Stahl, J. Am. Chem. Soc., 2006, 128, 4348–4355.
32. P. Foley, R. DiCosimo and G. M. Whitesides, J. Am. Chem. Soc., 1980, 102, 6713–6725.
33. G. M. Whitesides, M. Hackett, R. L. Brainard, J. P. P. M. Lavalleye, A. F. Sowinski, A. N. Izumi, S. S. Moore, D. W. Brown and E. M. Staudt, Organometallics, 1985, 4, 1819–1830.
34. T. Matsumoto, M. Ueno, J. Kobayashi, H. Miyamura, Y. Mori and S. Kobayashi, Adv. Synth. Catal., 2007, 349, 531–534.
35. I. T. Horváth, G. Kiss, R. A. Cook, J. E. Bond, P. A. Stevens, J. Rábai and E. J. Mozeleski, J. Am. Chem. Soc., 1998, 120, 3133–3143.
36. D. Astruc, F. Lu and J. R. Aranzaes, Angew. Chem., Int. Ed., 2005, 44, 7852–7872.
37. C. O. Bennett and M. Che, J. Catal., 1989, 120, 293–302.
38. G. C. Bond, J. Catal., 1992, 136, 631–632.
39. R. L. Augustine and P. Techasauvapak, J. Mol. Catal., 1994, 87, 95–105.
40. R. Van Hardeveld and F. Hartog, Surf. Sci., 1969, 15, 189–230.
41. A. Benedetti, G. Fagherazzi, F. Pinna, G. Rampazzo, M. Selva and G. Strukul, Catal. Lett., 1991, 10, 215–223.
42. R. L. Augustine and S. T. O'Leary, J. Mol. Catal. A: Chem., 1995, 95, 277–285.
43. L. Piccolo and C. R. Henry, J. Mol. Catal. A: Chem., 2001, 167, 181–190.
44. E. K. Parks, G. C. Nieman, K. P. Kerns and S. J. Riley, J. Chem. Phys., 1998, 108, 3731–3739.
45. G. Schmid, Chem. Rev., 1992, 92, 1709–1727.
46. C. Amatore and A. Jutand, Acc. Chem. Res., 2000, 33, 314–321.
47. L. A. Adrio, B. N. Nguyen, G. Guilera, A. G. Livingston and K. K. Hii, Catal. Sci. Technol., 2012, 2, 316–323.
48. N. T. S. Phan, M. Van Der Sluys and C. W. Jones, Adv. Synth. Catal., 2006, 348, 609–679.

---