DFT Studies on the Allylation of Styrene Oxide Catalyzed by Indium Nanoparticles (InNPs) †

: Allyl–indium species are known to smoothly react with electrophiles as epoxides. The intrinsic tension of these cyclic systems makes them very reactive starting materials, giving access to more complex products through regioselective ring-opening. In this work, indium nanoparticles (InNPs) were prepared by fast reduction of indium(III) chloride with lithium sand and a catalytic amount of 4,4 (cid:48) -di-tert-butylbiphenyl. InNPs were employed as catalysts for the ring-opening al-lylation reaction of styrene oxide with allyl or prenylbromide. All the possible reaction paths for the formation of the different alcohols were modeled applying DFT calculations, achieving good correlation between the experimental and computational results.


Introduction
Epoxides are extremely interesting electrophiles in terms of their properties and reactivity. They are part of many natural products as well as numerous compounds of biological importance. In organic synthesis, organometallic compounds provide a source of nucleophilic carbon centers that can react with electrophiles to form new carbon-carbon bonds. In particular, allyl-indium species react with various electrophiles, such as carbonyl compounds, to generate homoallylic alcohols [1]. Epoxides present great reactivity due to the intrinsic tension of the cyclic system and can react with organometallic compounds, giving rise to the formation of ring-opening products to give molecules that are larger and more complex. For this reason, the general aspects of epoxide synthesis and ring-opening reactions have been extensively reviewed [2].
The allylation of different epoxides with indium metal has been reported [3,4]. Yadav et al. [3] described a novel and highly efficient protocol for the allylation of terminal epoxides using metallic indium and allyl bromide in THF to obtain the corresponding bishomoallylic alcohols (Scheme 1).

Introduction
Epoxides are extremely interesting electrophiles in terms of their properties and reactivity. They are part of many natural products as well as numerous compounds of biological importance. In organic synthesis, organometallic compounds provide a source of nucleophilic carbon centers that can react with electrophiles to form new carbon-carbon bonds. In particular, allyl-indium species react with various electrophiles, such as carbonyl compounds, to generate homoallylic alcohols [1]. Epoxides present great reactivity due to the intrinsic tension of the cyclic system and can react with organometallic compounds, giving rise to the formation of ring-opening products to give molecules that are larger and more complex. For this reason, the general aspects of epoxide synthesis and ring-opening reactions have been extensively reviewed [2].
The allylation of different epoxides with indium metal has been reported [3,4]. Yadav et al. [3] described a novel and highly efficient protocol for the allylation of terminal epoxides using metallic indium and allyl bromide in THF to obtain the corresponding bishomoallylic alcohols (Scheme 1).   In our research group, we have been working on the development of methodologies based on the use of metal nanoparticle catalysts (MNPs). The epoxidation of a variety of olefins has recently been reported with good to excellent yields and high selectivity using a new heterogeneous catalyst system composed of cobalt nanoparticles (CoNPs) supported on MgO with tert-butyl hydroperoxide (TBHP) as the oxidant [5].
Based on the aforementioned, we studied the allylation reaction of styrene oxide with different allyl bromides catalyzed by InNPs under different reaction conditions. With the aim to explain the experimental observations, we performed a computational study applying DFT methods with the Gaussian09 software [6]. Based on the experimental and computational results, we propose a reaction mechanism for the allylation reaction of styrene oxide catalyzed by InNPs.

General
All moisture-sensitive reactions were carried out under a nitrogen atmosphere. Anhydrous tetrahydrofuran was freshly distilled from sodium/benzophenone ketyl. Other solvents used were treated prior to use via standard methods. All starting materials were of the best available grade (Aldrich (St. Louis, MI, USA), Merck (Darmstadt, Germany) and were used without further purification. Preparative plate chromatography was performed with Macherey-Nagel silica gel P/UV254 with CaSO 4 for thin-layer chromatography with a fluorescence indicator. Thin-layer chromatography (TLC) was performed on precoated silica gel plates (Merck 60, F254, 0.25 mm).

Instrumentation and Analysis
Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker ARX-300 spectrophotometer (Billerica, MA, USA) using CDCl 3 as the solvent. Mass spectra (EI) were obtained at 70 eV on an Agilent 7890B gas chromatograph coupled with an Agilent 5977A mass-selective detector (MSD) (Agilent, Santa Clara, CA, USA) equipped with HP-5MS capillary column (30 m × 0.25 mm inner diameter, 0.25 m film thickness).

Reaction between Styrene Oxide and Allyl Bromide Catalized by InNPs
The InNP catalyst was prepared following the procedure previously reported in our group [1]. On 0.66 mmol of InNPs generated in-situ, 1 mmol of allyl bromide dissolved in 1 mL of ACN was added. For the Grignard-type reaction, the solution was allowed to react for 1 h with continuous stirring. After this time, 0.33 mmol of styrene oxide dissolved in 1 mL of ACN was added and the reaction was heated under ACN reflux for 48 h. At this time, the reaction was finished by adding 2 mL of 1M HCl and was elaborated using multiple extraction with dichloromethane. The organic phase was dried over anhydrous MgSO 4 and filtered and analyzed using GC-MS. For the Barbier-type reaction, once the InNPs were generated, the allyl bromide and styrene oxide were added at the same time. In both cases, other conditions were analyzed, and we worked at THF reflux and with ACN and THF at room temperature.

Computational Procedure
The conformational analysis for the structures was performed with the functional PM6. The geometries of the most stable structures were refined with the B3LYP functional and the LANL2DZ pseudopotential [7][8][9]. The characterization of the stationary points was verified with a complete optimization of the minima (or transition states) by the absence of (or the presence of a single) negative frequencies. The figures were built by using Avogadro software [10].

Results and Discussion
Based on a scientific work published in our research group [1], we carried out a study of the reaction and the mechanism of the allylation of styrene oxide catalyzed by InNPs with allyl and prenyl bromide. According to bibliographical references, the formation of allyl-In species [11], which react with the epoxide to give the corresponding alcohols, is assumed. This epoxide allowed us to carry out a regiochemical study of the process, since the nucleophilic attack can occur on the carbon atom attached to the aromatic ring (C-I) or on the less substituted carbon atom (C-I ) (Scheme 2).
allyl-In species [11], which react with the epoxide to give the corresponding alcohols, is assumed. This epoxide allowed us to carry out a regiochemical study of the process, since the nucleophilic attack can occur on the carbon atom attached to the aromatic ring (C-I) or on the less substituted carbon atom (C-I′) (Scheme 2). Scheme 2. Nucleophilic attack on the epoxide.

Grignard-Type Reaction between Styrene Oxide and Allyl Bromide Catalyzed by InNPs
After 48 h, the reaction performed in THF at room temperature exhibited the formation of the by-products 2-chloro-1-phenylethanol and 1-phenylethane-1,2-diol in ratios of 45-55%, respectively. The reaction in ACN at room temperature exhibited the same byproducts ratios of 30-70%, respectively. The first could arise from the opening of the epoxide after attack by the chloride ion (from lithium chloride) and the second would be generated after the attack by the hydroxyl ion (from lithium hydroxide). These by-products indicate that the allylation was not favored under these reaction conditions [12].
In view of the aforementioned results, to favor the allylation reaction, we carried out the same reactions at the reflux of THF and ACN. At the end of these reactions, an allyl alcohol was observed as the only product. It is worth mentioning that, after 5 h of reaction, phenylacetaldehyde was observed using CG-EM.
Yadav et al. [3] inform us that an erroneous assignment in the 1 H-NMR spectrum of the aforementioned alcohol, which according to the authors would be 1-phenylpent-4-en-1-ol, comes from the allyl attack on the carbon atom C-I′ (Scheme 2). Hirashita et al. [11] reviewed the Yadav et al. published results, finding doubtful the signal assignment at 2.75 ppm, and re-assigns it as the benzyl protons of the 1-phenylpent-4-en-2-ol, this alcohol coming from the direct allylation of phenylacetaldehyde. Phenylacetaldehyde is quickly formed (we detected it in both reactions) as a consequence of the benzylic C-O bond opening of the epoxide under the reaction conditions, and gives rise to the formation of the aforementioned homoallylic alcohol after carbonyl allylation [1] (Scheme 3). Taking into account that in their works, Ranú [12] and Hirashita [11] suggested that the allyl-indium species formed under these reaction conditions would have Lewis acid characteristics, favoring the formation of the aldehyde from the epoxide, we considered reducing the ratio of the allylating species in the reaction. We were using an allyl bromide:InCl3:epoxide ratio of 1.5:1:0.33 (mmol), and so we ran a reaction with an allyl bromide:InCl3:epoxide ratio of 1:0.66: 0.33 (mmol), at reflux of ACN, which gave as its only product 1-phenylpent-4-en-2-ol, the homoallylic alcohol from the allylation of phenylacetaldehyde.
As a new experimental approach and based on the work published by Oh et al. [4], we decided to change the order of the addition of the reagents and work under Barbier conditions. Scheme 2. Nucleophilic attack on the epoxide.

Grignard-Type Reaction between Styrene Oxide and Allyl Bromide Catalyzed by InNPs
After 48 h, the reaction performed in THF at room temperature exhibited the formation of the by-products 2-chloro-1-phenylethanol and 1-phenylethane-1,2-diol in ratios of 45-55%, respectively. The reaction in ACN at room temperature exhibited the same by-products ratios of 30-70%, respectively. The first could arise from the opening of the epoxide after attack by the chloride ion (from lithium chloride) and the second would be generated after the attack by the hydroxyl ion (from lithium hydroxide). These by-products indicate that the allylation was not favored under these reaction conditions [12].
In view of the aforementioned results, to favor the allylation reaction, we carried out the same reactions at the reflux of THF and ACN. At the end of these reactions, an allyl alcohol was observed as the only product. It is worth mentioning that, after 5 h of reaction, phenylacetaldehyde was observed using CG-EM.
Yadav et al. [3] inform us that an erroneous assignment in the 1 H-NMR spectrum of the aforementioned alcohol, which according to the authors would be 1-phenylpent-4-en-1-ol, comes from the allyl attack on the carbon atom C-I (Scheme 2). Hirashita et al. [11] reviewed the Yadav et al. published results, finding doubtful the signal assignment at 2.75 ppm, and re-assigns it as the benzyl protons of the 1-phenylpent-4-en-2-ol, this alcohol coming from the direct allylation of phenylacetaldehyde. Phenylacetaldehyde is quickly formed (we detected it in both reactions) as a consequence of the benzylic C-O bond opening of the epoxide under the reaction conditions, and gives rise to the formation of the aforementioned homoallylic alcohol after carbonyl allylation [1] (Scheme 3).
allyl-In species [11], which react with the epoxide to give the corresponding alcohols, is assumed. This epoxide allowed us to carry out a regiochemical study of the process, since the nucleophilic attack can occur on the carbon atom attached to the aromatic ring (C-I) or on the less substituted carbon atom (C-I′) (Scheme 2). Scheme 2. Nucleophilic attack on the epoxide.

Grignard-Type Reaction between Styrene Oxide and Allyl Bromide Catalyzed by InNPs
After 48 h, the reaction performed in THF at room temperature exhibited the formation of the by-products 2-chloro-1-phenylethanol and 1-phenylethane-1,2-diol in ratios of 45-55%, respectively. The reaction in ACN at room temperature exhibited the same byproducts ratios of 30-70%, respectively. The first could arise from the opening of the epoxide after attack by the chloride ion (from lithium chloride) and the second would be generated after the attack by the hydroxyl ion (from lithium hydroxide). These by-products indicate that the allylation was not favored under these reaction conditions [12].
In view of the aforementioned results, to favor the allylation reaction, we carried out the same reactions at the reflux of THF and ACN. At the end of these reactions, an allyl alcohol was observed as the only product. It is worth mentioning that, after 5 h of reaction, phenylacetaldehyde was observed using CG-EM.
Yadav et al. [3] inform us that an erroneous assignment in the 1 H-NMR spectrum of the aforementioned alcohol, which according to the authors would be 1-phenylpent-4-en-1-ol, comes from the allyl attack on the carbon atom C-I′ (Scheme 2). Hirashita et al. [11]   Taking into account that in their works, Ranú [12] and Hirashita [11] suggested that the allyl-indium species formed under these reaction conditions would have Lewis acid characteristics, favoring the formation of the aldehyde from the epoxide, we considered reducing the ratio of the allylating species in the reaction. We were using an allyl bromide:InCl3:epoxide ratio of 1.5:1:0.33 (mmol), and so we ran a reaction with an allyl bromide:InCl3:epoxide ratio of 1:0.66: 0.33 (mmol), at reflux of ACN, which gave as its only product 1-phenylpent-4-en-2-ol, the homoallylic alcohol from the allylation of phenylacetaldehyde.
As a new experimental approach and based on the work published by Oh et al. [4], we decided to change the order of the addition of the reagents and work under Barbier conditions. Taking into account that in their works, Ranú [12] and Hirashita [11] suggested that the allyl-indium species formed under these reaction conditions would have Lewis acid characteristics, favoring the formation of the aldehyde from the epoxide, we considered reducing the ratio of the allylating species in the reaction. We were using an allyl bromide:InCl3:epoxide ratio of 1.5:1:0.33 (mmol), and so we ran a reaction with an allyl bromide:InCl3:epoxide ratio of 1:0.66: 0.33 (mmol), at reflux of ACN, which gave as its only product 1-phenylpent-4-en-2-ol, the homoallylic alcohol from the allylation of phenylacetaldehyde.
As a new experimental approach and based on the work published by Oh et al. [4], we decided to change the order of the addition of the reagents and work under Barbier conditions.

Barbier-Type Reaction between Styrene Oxide and Allyl Bromide Catalyzed by InNPs
We carried out the reaction at reflux of ACN, and after 48 h, the homoallylic alcohol 1-phenylpent-4-en-2-ol was the only reaction product.

Grignard-Type Reaction between Styrene Oxide and Prenyl Bromide Catalyzed by InNPs
Oh et al. [4] carried out a selectivity study as a function of the allyl bromide employed as the allylating species and vinyl epoxide, achieving the highest selectivity towards the Chem. Proc. 2022, 12, 33 4 of 8 expected alcohol when using prenyl bromide. Based on that, we considered this new experimental approach, since it is known that prenyl bromide is less reactive than allyl bromide towards carbonyl allylations [1].
Under reflux of ACN, after 48 h, the reaction did not give the expected alcohols; however, two products were observed: 2-chloro-1-phenylethanol (80%) and an oxidized allylic by-product.
We carried out the reaction at reflux of ACN, and after 48 h, the homoallylic alcohol 1-phenylpent-4-en-2-ol was the only reaction product.

Grignard-Type Reaction between Styrene Oxide and Prenyl Bromide Catalyzed by InNPs
Oh et al. [4] carried out a selectivity study as a function of the allyl bromide employed as the allylating species and vinyl epoxide, achieving the highest selectivity towards the expected alcohol when using prenyl bromide. Based on that, we considered this new experimental approach, since it is known that prenyl bromide is less reactive than allyl bromide towards carbonyl allylations [1].
Under reflux of ACN, after 48 h, the reaction did not give the expected alcohols; however, two products were observed: 2-chloro-1-phenylethanol (80%) and an oxidized allylic by-product.

Barbier-Type Reaction between Styrene Oxide and Prenyl Bromide Catalyzed by InNPs
When we carried out the reaction under reflux of THF, after 24 h, the starting epoxide was recovered. At reflux of ACN, after 22 h, phenylacetaldehyde was detected, and after it was, after 48 h, 2-chloro-1-phenylethanol and 3,3-dimethyl-1-phenylpent-4-en-2-ol, the homoallylic alcohol from the carbonyl prenylation (Scheme 4), were the main products observed (90 and 10%, respectively). We worked in ACN at 50 °C, and although in this case, no formation of phenylacetaldehyde was observed, 2-chloro-1-phenylethanol was the main product obtained.

Computational Study
To start with computational modeling studies following the proposed mechanism for the allylation of carbonyl compounds published by our research group [1], as can be seen from Scheme 5, we considered that, at first, the epoxide is adsorbed on the surface of the InNPs of the allyl-In species to give the initial complex (reactive complex-RC). Then, a cyclic Zimmerman-Traxler type transition state (TS) is formed, and the π electrons of the allylic system attack one of the carbon atoms of the epoxide, giving rise to a new C-C bond through a γ-coupling. By the adding of diluted HCl, the double bond of the alkene is regenerated, giving rise to the final product (product complex-PC). Since the epoxide is not symmetric, we can obtain two alcohols from the allylic attack on each of the carbon atoms (C-I and C-I') of the oxo-cyclic system. We worked in ACN at 50 • C, and although in this case, no formation of phenylacetaldehyde was observed, 2-chloro-1-phenylethanol was the main product obtained.

Computational Study
To start with computational modeling studies following the proposed mechanism for the allylation of carbonyl compounds published by our research group [1], as can be seen from Scheme 5, we considered that, at first, the epoxide is adsorbed on the surface of the InNPs of the allyl-In species to give the initial complex (reactive complex-RC). Then, a cyclic Zimmerman-Traxler type transition state (TS) is formed, and the π electrons of the allylic system attack one of the carbon atoms of the epoxide, giving rise to a new C-C bond through a γ-coupling. By the adding of diluted HCl, the double bond of the alkene is regenerated, giving rise to the final product (product complex-PC). Since the epoxide is not symmetric, we can obtain two alcohols from the allylic attack on each of the carbon atoms (C-I and C-I ) of the oxo-cyclic system. We report below the theoretical results obtained from computational calculations for the direct allylation on each carbon atom at the oxo-cyclic structure of styrene oxide, C-I (Scheme 6) and C-I' (Scheme 7). To simplify the reactive allylic species, an auxiliary bromide atom was used as a ligand to obtain a neutral allylic indium species. We report below the theoretical results obtained from computational calculations for the direct allylation on each carbon atom at the oxo-cyclic structure of styrene oxide, C-I Chem. Proc. 2022, 12, 33 5 of 8 (Scheme 6) and C-I (Scheme 7). To simplify the reactive allylic species, an auxiliary bromide atom was used as a ligand to obtain a neutral allylic indium species. Scheme 5. Mechanism of formation of both alcohols.
We report below the theoretical results obtained from computational calculations for the direct allylation on each carbon atom at the oxo-cyclic structure of styrene oxide, C-I (Scheme 6) and C-I' (Scheme 7). To simplify the reactive allylic species, an auxiliary bromide atom was used as a ligand to obtain a neutral allylic indium species. Scheme 6. Representative structures (top) and potential energy profile (kcal/mol) at B3LYP/LanL2DZ level for the formation of 2-phenylpent-4-en-1-ol (bottom).

Scheme 6.
Representative structures (top) and potential energy profile (kcal/mol) at B3LYP/LanL2DZ level for the formation of 2-phenylpent-4-en-1-ol (bottom). As can be seen in Schemes 6 and 7, from a thermochemical point of view, the formation of both alcohols would be feasible, since they are exothermic processes (28.2 and 33.2 kcal/mol, respectively). However, if we perform a kinetic analysis, the formation of both products would be ruled out due to the high activation barriers: 28 kcal/mol for 1phenylpent-4-en-1-ol and 16.2 kcal/mol for 2-phenylpent-4-en-1-ol.
On the other hand, as can be seen from Scheme 8, based on the experimental observations, we computationally modeled the formation of phenylacetaldehyde and its subsequent allylation. The oxo-cyclic opening promoted by the allyl-indium species to give Scheme 7.
As can be seen in Schemes 6 and 7, from a thermochemical point of view, the formation of both alcohols would be feasible, since they are exothermic processes (28.2 and 33.2 kcal/mol, respectively). However, if we perform a kinetic analysis, the formation of both products would be ruled out due to the high activation barriers: 28 kcal/mol for 1-phenylpent-4-en-1-ol and 16.2 kcal/mol for 2-phenylpent-4-en-1-ol.
On the other hand, as can be seen from Scheme 8, based on the experimental observations, we computationally modeled the formation of phenylacetaldehyde and its subsequent allylation. The oxo-cyclic opening promoted by the allyl-indium species to give the aldehyde requires 12.8 kcal/mol, a lower barrier than those discussed above. Furthermore, the formation of phenylacetaldehyde is thermodynamically highly favored, since 22 kcal/mol is released in the process. InNPS-catalyzed carbonyl allylation is a chemically and energetically very favorable process, demonstrated by the low activation barrier (5.4 kcal/mol) and the observed exothermicity (−13.1 kcal/mol). From these results, we can say that the calculation methodology used was a good approximation for these processes, since it allows us, from both kinetic and thermodynamic points of view, to show that the epoxide prefers to give rise to phenylacetaldehyde, and this aldehyde would be very easily allylated, in agreement with the experimental results.

Proposed Mechanism
Based on the stoichiometry of the reaction, the experimental observations and DFT studies, we propose a plausible reaction mechanism for the studied transformation. As can be seen from Scheme 9, the InNPs would be obtained from the indium salt by the action of the reducing system; then, the addition of allyl bromide could lead to the formation of an allyl-indium intermediate, which by reaction with the epoxide would give the phenylacetaldehyde. This activated carbonyl compound could give rise to the corresponding homoallylic alcohol through a Zimmerman-Traxler-type transition state.
From these results, we can say that the calculation methodology used was a good approximation for these processes, since it allows us, from both kinetic and thermodynamic points of view, to show that the epoxide prefers to give rise to phenylacetaldehyde, and this aldehyde would be very easily allylated, in agreement with the experimental results.

Proposed Mechanism
Based on the stoichiometry of the reaction, the experimental observations and DFT studies, we propose a plausible reaction mechanism for the studied transformation. As can be seen from Scheme 9, the InNPs would be obtained from the indium salt by the action of the reducing system; then, the addition of allyl bromide could lead to the formation of an allyl-indium intermediate, which by reaction with the epoxide would give the phenylacetaldehyde. This activated carbonyl compound could give rise to the corresponding homoallylic alcohol through a Zimmerman-Traxler-type transition state.
Based on the stoichiometry of the reaction, the experimental observations and DFT studies, we propose a plausible reaction mechanism for the studied transformation. As can be seen from Scheme 9, the InNPs would be obtained from the indium salt by the action of the reducing system; then, the addition of allyl bromide could lead to the formation of an allyl-indium intermediate, which by reaction with the epoxide would give the phenylacetaldehyde. This activated carbonyl compound could give rise to the corresponding homoallylic alcohol through a Zimmerman-Traxler-type transition state. Scheme 9. Proposed reaction mechanism for the reaction between styrene oxide and allyl bromide catalyzed by InNPs.

Conclusions
In(0) nanoparticles have been evaluated as catalysts for the allylation of styrene oxide with allyl and prenyl bromide at different reaction conditions. In most of the reactions, we observed the formation of phenylacetaldehyde through the opening of the benzylic C-O bond of the epoxide. The allylation of this aldehyde gives rise to 1-phenylpent-4-en-2-ol as the final product of the reaction. Through a theoretical study with DFT methods, we modeled the complete process, achieving a very good agreement between the experimental and computational results. In addition, we propose a possible reaction mechanism that would explain the oxo-cyclic opening to give the aldehyde, as well the allylation of the phenylacetaldehyde to give rise to the homoallylic alcohol under the studied reaction conditions.